Cortical granule exocytosis in C. elegans is regulated by cell cycle components including separase

Oocytes must coordinate specialized meiotic divisions with fertilization and the subsequent transition to the mitotic program. Egg activation comprises a sequence of events that occur in response to calcium transients triggered by fertilization, including cortical granule exocytosis and resumption of the cell cycle. Cortical granules are large vesicles located at the cortex of mature oocytes of many species. Cortical granules secrete a collection of proteoglycans and enzymes that modify the extra embryonic covering of oocytes to prevent polyspermy (Wessel et al.,2001). At fertilization, inactivation of cytostatic factor, which arrests oocytes in metaphase through inhibition of an E3 ubiquitin ligase known as the anaphase-promoting complex/cyclosome (APC/C), allows entry into anaphase and resumption of the cell cycle(Tunquist and Maller,2003).

In the Caenorhabditis elegans gonad, oocytes arrested in prophase I undergo maturation in response to signals from sperm and are subsequently ovulated (Yamamoto et al.,2006). After fertilization in the spermatheca, embryos pass into the uterus and extrude two polar bodies during consecutive meiotic divisions. Eggshell formation in nematodes occurs during meiosis(Chitwood and Chitwood, 1974),such that the first polar body remains outside the permeability barrier,whereas the second polar body does not(Polinko and Strome, 2000). The C. elegans eggshell consists of three layers, which we describe according to the following nomenclature: the outermost vitelline layer present around the mature oocyte; a middle chitinous layer necessary for mechanical strength; and an internal lipid layer providing the permeability barrier(Rappleye et al., 1999; Wharton, 1980). Although the process of eggshell formation in C. elegans is poorly understood, a recent genome-wide screen identified 109 genes likely to be required for eggshell formation (Sonnichsen et al.,2005), some of which are cell cycle regulatory components,suggesting a link between cell cycle regulation and eggshell formation.

Regulation of cell division requires faithful orchestration of a wide diversity of cellular processes. During M-phase, precise regulatory mechanisms ensure the fidelity of chromosome segregation and cell cleavage. During metaphase, a complex signaling pathway known as the spindle-assembly checkpoint (SAC) prevents entry into anaphase before the correct alignment of chromosomes on the metaphase plate(Musacchio and Salmon, 2007). The SAC preserves chromosome cohesion by inhibiting the APC/C(Peters, 2006). At the metaphase to anaphase transition, APC/C ubiquitinates securin, an inhibitory chaperone of the protease separase, leading to its degradation by the proteasome (Nasmyth, 2002). Active separase then cleaves a subunit of the cohesin complex, allowing the poleward movement of chromosomes during anaphase.

Inactivation of the C. elegans ortholog of separase, sep-1, causes chromosome nondisjunction, the osmotic integrity defective phenotype (OID) and cytokinesis failures(Siomos et al., 2001). The osmotic integrity defective phenotype is also caused by inhibition of a number of cell cycle regulatory components(Sonnichsen et al., 2005), and has been suggested to be a non-specific consequence of meiotic division failures. However, loss of other cell cycle genes that cause severe meiotic defects does not give rise to an OID phenotype(Polinko and Strome, 2000),indicating that eggshell deposition is separable from cell division, although these processes may be linked.

We have found that cortical granules form in developing oocytes and are exocytosed, leading to the formation of an impermeable three-layered eggshell in C. elegans. The exocytosis of cortical granules occurs during anaphase I and requires a number of cell cycle components, including separase. During anaphase I, separase localizes to cortical granules and may have a direct role in regulating cortical granule exocytosis. Our observations indicate that events occurring during egg activation are coordinated by cell cycle regulatory controls in C. elegans.

The wild type was the Bristol strain N2, and all strains were cultured using standard techniques (Brenner,1974). Temperature-sensitive strains were maintained at 16°C and shifted to 25°C as L4 animals; all other strains were maintained at 20°C. The following strains were also used: WH0216 {sep-1(e2406)I/hT2[bli-4(e937) let-?(q782) qls48]}, WH0213 {sep-1(e2406)I; unc-119(ed3) III, ruls32 III[GFP::His-58 unc119(+)]/hT2[bli-4(e937) let-?(q782) qls48]}, HY604 mat-1(ye121) I, AZ212 {unc-119(ed3) III, ruls32III[GFP::His-58 unc119(+)]}, WH0327 {unc-119(ed3) III, ojIs23[GFP::SP12 unc119(+)]}, WH0351 {unc-119(ed3) III, ojIs37[GFP::UGTP-1 unc119(+)]}, RT688 {unc-119(ed3) III, pwIs281[GFP::CAV-1 unc119(+)]}, WH0416 {unc-119(ed3) III, ojIs58[SEP-1::GFP unc119(+)]}, WH0438 {unc-119(ed3) III, ojEx64[GFP::SEP-1 unc119(+)]}.

cDNA clones for syn-4 (yk493b10.3), cpg-1 (also known as cej-1 - WormBase) (yk253a11) and cpg-2 (yk1687e03) were subcloned into the L4440 vector for feeding RNAi(Kamath et al., 2001). Separase feeding RNAi was performed as previously described(Siomos et al., 2001), and all other constructs were obtained from the Ahringer library(Kamath et al., 2003). L4 animals were fed at 20°C for 24 hours(Fig. 4C-F), or as otherwise noted. The 2.5 kb ugtp-1 (ZK370.7) coding region was cloned from genomic DNA into the pID3.01 vector for biolistic transformation, as described in (Poteryaev et al., 2005). The 5.2 kb coding sequence of sep-1 was amplified from N2 genomic DNA using Phusion polymerase (NEB, Ipswich, MA, USA) and subcloned into the pJK derivative plasmids of pFJ1.1 (Verbrugghe and White, 2004) that allow for directional cloning, and pie-1-driven expression of N- and C-terminal fusions with several worm-optimized fluorescent proteins including mGFP (A206K mutant). Biolistic transformation was done as previously described(Praitis et al., 2001).

The last 438 amino acids of the sep-1 open reading frame fused to a hexa-histidine tag expressed in bacteria was purified using the Probond protein purification kit as per the manufacturer's protocol (Invitrogen,Carlsbad, CA, USA) under denaturing conditions in 8 M urea. This purified protein was concentrated, dialyzed and used to raise rabbit polyclonal antibodies following standard protocols (Covance Research Products, Berkeley,CA, USA). The resulting immune serum was affinity-purified against antigen that had been gel-purified and immobilized on nitrocellulose paper. Embryo staining was performed as described in Gonczy et al.(Gonczy et al., 1999) to preserve membrane structures. The following dilutions, in PBS+0.05% BSA, were used: α-SEP-1 antibodies, 1:200; α-HIM-10 antibodies, 1:50;rhodamine-lectins from Vector Laboratories, 1:10; tubulin antibody DM1α(Sigma, St Louis, MO, USA), 1:100; Alexa-conjugated secondary antibodies(Invitrogen), 1:200; and TO-PRO-3 Iodide (Invitrogen), 1:500. FM2-10(Invitrogen) was used at 17 μM, FM4-64 was used at 0.02 mg/ml and a fluorescein-conjugated 3000-MW dextran (Invitrogen) was used at 0.5 mg/ml in blastomere culture medium (Shelton and Bowerman, 1996).

OID embryos were mounted in blastomere culture media(Shelton and Bowerman, 1996)by hanging drop to relieve osmotic and mechanical pressures. A toxicity effect(Siomos et al., 2001) was avoided by removing bacteria and the mother carcass. Wild-type meiosis I embryos develop to hatching in these mounting conditions.

Animals were prepared for TEM as described previously(Poteryaev et al., 2005),without gluteraldehyde fixation and also imaged with a Phillips CM 120 (FEI,Hillsboro, OR) at 80 kV, with a Soft Imaging Systems (Lakewood, CO) CCD camera and software, and a Tecnai T12 (FEI, Hillsborough, OR, USA) at 80 kV, with an ES500W CCD camera and software from Gatan (Pleasanton, CA, USA).

Multiphoton laser-scanning microscopy (MPLSM) was done at LOCI(http://www.loci.wisc.edu)with a custom-designed multiphoton laser scanning optical workstation(Wokosin et al., 2003). This system uses a Nikon Eclipse TE300 inverted microscope with a Nikon VC 60X Plan Apo oil-immersion lens (NA 1.2) and 5W mode-locked Ti:sapphire laser(Spectra-Physics-Millennium/Tsunami) with excitation at 890 nm. The intensity data were collected by a H7422 PMT (Hamamatsu) and the transmitted signal was collected by a transmitted photodiode detector (Bio-Rad, Hercules, CA, USA). Image acquisition was done with WiscScan(http://www.loci.wisc.edu/wiscscan/). A Swept Field Confocal system (Prairie Technologies Inc, Middleton, WI, USA)with 488, 514 and 568 nm excitation lines was utilized with a Nikon Eclipse TE300 Microscope, 100× Superfluor Lens NA 1.4 (Nikon, Melville NY) and a Dual-View beam splitter (Roper Scientific, Tucson, AZ, USA) utilizing a FITC/Texas Red emission set. Fixed samples were imaged using a Bio-Rad MRC 1024 confocal microscope (Hercules, CA, USA). To quantify degranulation, MPLSM was done with similar settings (which preserve embryo viability for over 2000 scans) and vesicle fusion events were manually counted. Images were analyzed in ImageJ v1.37p (Abramoff et al.,2004), and results were recounted on several data sets using the point picker plugin. Images were processed in Photoshop v8.0 (Adobe Systems)using Gaussian blur to reduce noise, and levels and curves to enhance contrast of the object of interest.

Statistical calculations of mean, standard deviation and P-value were determined by fitting a one-way ANOVA model with SAS 9.1.3 on measurements gathered from individual embryos.

Cortical granules in oocytes of other organisms readily stain with lectins(Hoodbhoy and Talbot, 2001),which bind sugar molecules of glycosylated proteins. To obtain evidence that C. elegans oocytes contain cortical granules, we stained them with a panel of fluorescent-labeled lectins. Several lectins stained the vitelline layer, cytoplasmic vesicles and 1 μm vesicles at the cortex of oocytes and embryos in meiosis I (see Fig. S1A-F in the supplementary material). Wheat germ agglutanin (WGA), used to identify cortical granules in human oocytes(Talevi et al., 1997), labeled 1 μm vesicles in C. elegans oocytes and meiosis I embryos, but not in older embryos (Fig. 1A-C). Eggshell labeling with WGA appeared brighter in older embryos, suggesting that it recognized a glycoprotein cargo of cortical granules that incorporates into the eggshell. The endoplasmic reticulum (ER) reorganizes into a reticulate structure in the cortex around the time of fertilization(Poteryaev et al., 2005). We found that domains of the reticulate ER were closely associated with cortical granules in clusters during prometaphase I (see Fig. S1G-I in the supplementary material) that were redistributed evenly across the cortex during metaphase I (Fig. 1D-F). Therefore, we conclude that C. elegans contains cortical granules that undergo dynamic trafficking similar to that of other organisms(Sardet et al., 2002).

We generated transgenic animals expressing a GFP fusion with ZK370.7: a protein with sequence homology to transmembrane nucleotide-sugar transporters,predicted to transport UDP-galactose. This putative Golgi marker (hereafter referred to as UGTP-1::GFP) labeled cytoplasmic puncta and large 1 μm vesicles in oocytes and young embryos, which were present for a narrow time window after fertilization. These vesicles were clustered in early prometaphase I, redistributed within the cortex and lost by the end of meiosis I (Fig. 1G-I). We found that UGTP-1::GFP co-localized with WGA-stained vesicles(Fig. 1J-L), indicating that it is packaged into cortical granules. The observation that cortical granules contain UGTP-1::GFP suggests that they may be Golgi-derived. CAV-1::GFP, a GFP fusion to caveolin, was previously shown to undergo trafficking consistent with cortical granule exocytosis (Sato et al., 2006), and also localizes to WGA-stained cortical granules more specifically than UGTP-1::GFP (see Fig. S1J-L in the supplementary material). Therefore, UGTP-1::GFP and CAV-1::GFP are convenient markers for live cell imaging of cortical granule trafficking.

We used transmission electron microscopy (TEM) and observed a unique population of vesicles present in oocytes. Oocytes are fertilized approximately every 23 minutes (McCarter et al., 1999), so that the age of each embryo can be inferred by its position within the uterus (see Fig. 4A). A mature oocyte contained a large number of vesicles closely associated with the reticulated ER (Fig. 2A). These approximately 1 μm vesicles had an unusual morphology reminiscent of the `refringent granules' shown to contribute to eggshell formation in other nematode species(Foor, 1967). Prometaphase I embryos contained reticulate ER interlaced with clusters of heterogeneous vesicles near the cortex (Fig. 2B). Metaphase I embryos contained the 1 μm vesicles dispersed throughout the cortex in close association with ER(Fig. 2C). These results are similar to our observations with lectin staining. We were unable to capture an embryo during degranulation, but we anticipate that the vesicles dock at the plasma membrane and are subsequently exocytosed based on our observations with the light microscope (see below). Consistent with this interpretation, embryos at meiosis II or older did not contain the 1 μm vesicles(Fig. 2D). These observations suggest that the 1 μm vesicles are cortical granules and that they may contribute to eggshell formation.

In attempt to observe the rapid process of degranulation, we imaged embryos using swept field confocal (SFC) microscopy (Prairie Technologies, Middleton,WI, USA) to acquire images every 200 milliseconds. Embryos expressing histone::GFP (Praitis et al.,2001) were mounted with optimized techniques to preserve viability(see Materials and methods), labeled with the plasma membrane dye FM2-10 and imaged by SFC (Fig. 3; see Movie 1 in the supplementary material). Interestingly, exocytosis began shortly after the homologous chromosomes separated during anaphase I in the region near the meiotic spindle (Fig. 3B). Fused vesicles incorporated FM2-10 and appeared as pockets outlined with fluorescence. As anaphase I proceeded, a wave of exocytosis spread across the cortex (Fig. 3B-D,F), and a gap opened between the vitelline layer and the plasma membrane near the polar body (Fig. 3E). The size of the vesicle cavity ranged from 0.2 to 1.5 μm,consistent with the size of cortical granules we observed by lectin staining and TEM.

To analyze secretion, we developed a correlative imaging technique using MPLSM. We simultaneously acquired transmitted bright-field and fluorescence images by MPLSM. Embryos labeled with either FM2-10 or a fluorescent dextran dissolved in the culture medium indicated vesicle-fusion events in the fluorescent image (Fig. 3F),which appeared coincident with cup-shaped plasma membrane invaginations in the bright-field image (see Movies 2 and 3 in the supplementary material). These observations demonstrate that the focal perturbations of the plasma membrane visualized in the bright-field image correlate with secretory events observed with plasma membrane and extracellular media labels and can serve as a marker for exocytosis.

To determine whether the wave of secretion during anaphase I represented the exocytosis of cortical granules, we observed the loss of UGTP-1::GFP-labeled vesicles using correlative MPLSM imaging. Loss of UGTP-1::GFP-labeled vesicles coincided with the appearance of plasma membrane perturbations visible in the bright-field image (see Movie 4 in the supplementary material). We also observed the loss of UGTP-1::GFP-labeled vesicles in the presence of extracellular fluorescent dextran (see Movie 5 in the supplementary material). In this experiment the signal of the extracellular dextran was saturating, whereas the UGTP-1::GFP signal was significantly weaker. Therefore, the exchange of vesicle contents with extracellular buffer after exocytosis resulted in a dramatic increase in fluorescence in the fused vesicle cavity. We also imaged the fusion of CAV-1::GFP-labeled vesicles to the plasma membrane in the presence of the plasma membrane dye FM4-64. The fluorescent signal from CAV-1::GFP is bright enough to image using SFC and capture two channels simultaneously. FM4-64 photobleached quickly and caused a toxicity effect when imaged by SFC with the 488 nm laser (not observed with FM2-10, data not shown), but allowed us to capture short movies. After fusion, the vesicle cavity was immediately labeled with FM4-64 and co-localized with CAV-1::GFP, which remained on the cytoplasmic side of the plasma membrane (see Movie 6 in the supplementary material). These results demonstrate that cortical granule exocytosis occurs during anaphase I.

Cortical granule exocytosis is required for the development of a polyspermy barrier on oocytes of other species(Wessel et al., 2001),therefore we hypothesized that it may contribute to formation of the eggshell in C. elegans. Inactivation of genes that lead to eggshell permeability cause the OID phenotype(Kaitna et al., 2002; Rappleye et al., 1999; Siomos et al., 2001). Many of the OID genes encode proteins either known or predicted to be involved in protein trafficking through the ER and Golgi. A number of cell cycle regulators also give rise to OID when depleted from the embryo. To gain further insight into the mechanisms that regulate cortical granule exocytosis and determine if this process is involved in eggshell formation, we assessed the fate of cortical granules following RNAi depletion of a selection of OID genes representing these functional categories. Each gene was tested in both the UGTP-1::GFP and CAV-1::GFP transgenic lines (see Table 1).

Our results are consistent with the hypothesis that secretion of cortical granule proteoglycan cargo is required for eggshell formation. Previous results showed that depletion of SAR-1 (ZK180.4 - WormBase), a small GTPase required for anterograde transport from the ER, prevented formation of cortical granules (Sato et al.,2006). The syntaxin SYN-4 is required for vesicle fusion and is needed for eggshell formation and cytokinesis(Jantsch-Plunger and Glotzer,1999). syn-4(RNAi) oocytes had abnormal populations of vesicles, sometimes accumulated at the plasma membrane (see Fig. S2E,F in the supplementary material). Eggshell chitin is generated by the chitin synthase,CHS-1 (Zhang et al., 2005), a putative cargo protein. Oocytes and embryos in chs-1(RNAi) animals showed no obvious change in cortical granule trafficking(Fig. 3C). Depletion of the redundant CPG-1 and CPG-2 chitin-binding chondroitin proteoglycans, also putative cargo proteins, and the glycosylation enzyme, SQV-4, reduced the size, but not the trafficking of cortical granules (see Fig. S2J,K in the supplementary material). CPG-1/-2 and SQV-4 may therefore be required to generate a significant mass of the proteoglycan cargo (the protein core and glycosyl side chains, respectively) secreted through cortical granule exocytosis for deposition of the eggshell.

The finding that membrane-trafficking regulators, but not putative cargo proteins, were required for cortical granule trafficking was not surprising. However, it was less clear how cell cycle regulatory genes might affect cortical granules. Therefore, we tested the effect of RNAi-depletion of several cell cycle OID proteins on cortical granule trafficking. First, we analyzed regulators that control the overall timing of cell division. Precocious CDK-1 activation occurs with wee-1.3(RNAi), resulting in premature oocyte maturation (Burrows et al., 2006). wee-1.3(RNAi) oocytes had large aggregates of UGTP-1::GFP-labeled puncta in the cytoplasm, which obscured the cortical granule population (see Fig. S2A in the supplementary material). CAV-1::GFP appeared normal in oocytes depleted in WEE-1.3, although in some cases vesicles moved to the cortex prematurely (see Fig. S2B in the supplementary material). Timely entry into mitosis requires the active CDK complex. Depletion of the primary M-phase CDK, CDK-1, caused retention of cortical granules in arrested embryos (see Fig. S2H in the supplementary material). To test the specificity of these effects, we depleted the CDK CKS-1 subunit by RNAi, which causes cytokinesis failures and delayed exit from mitosis, but not OID (Polinko and Strome,2000). As expected, cortical granule trafficking appeared normal in cks-1(RNAi) embryos (see Fig. S2L in the supplementary material). These data indicate that only a subset of genes that regulate global cell cycle progression impact the trafficking of cortical granules.

A striking correlation exists between inhibiting genes that regulate progress through the metaphase to anaphase transition and the OID phenotype. czw-1 is homologous to the ZW10 spindle assembly checkpoint component(Karess, 2005), implicated in regulation of vesicle trafficking (Hirose et al., 2004). czw-1(RNAi) caused UGTP-1::GFP to localize in enlarged puncta and smaller cytoplasmic vesicles (see Fig. S2C in the supplementary material). CAV-1::GFP was observed in small cytoplasmic vesicles(see Fig. S2D in the supplementary material), indicating that CZW-1 may regulate biogenesis of cortical granules. RNAi of APC/C subunits leads to cell cycle arrest in metaphase I (Shakes et al., 2003) and causes retention of CAV-1::GFP-labeled vesicles(Sato et al., 2006). We noticed that apc-2(RNAi) embryos retained clusters of cortical granules, reminiscent of wild-type prometaphase I embryos(Fig. 4D). The RNAi depletion of the cyk-3 deubiquitination enzyme causes OID and cytokinesis failures (Kaitna et al.,2002), and might disrupt ubiquitin-mediated regulation of the cell cycle or membrane trafficking. cyk-3(RNAi) embryos retained clusters of cortical granules, similar to apc-2(RNAi) (see Fig. S2G in the supplementary material). These results suggest that the redistribution of cortical granules throughout the cortex is an ubiquitin-regulated process linked to the cell cycle and that regulators of the metaphase to anaphase transition also control trafficking of cortical granules.

To gain further insight into the cell-cycle regulation of degranulation, we investigated the role of an important downstream effector of APC/C in cortical granule exocytosis: separase. Inactivation of separase causes cytokinesis failures and delayed progression through mitosis, similar to cks-1(also known as dom-6 - WormBase). However, in contrast to cks-1, separase inactivation also causes OID. Consistent with this phenotype, we found that cortical granules were retained in older embryos after depletion of separase (SEP-1) (Fig. 4E) or its inhibitory chaperone, securin (IFY-1) (see Fig. S2I in the supplementary material). To pursue these observations further, we examined embryos from an APC/C mutant [mat-1(ye121)] and from sep-1(RNAi) animals by TEM. APC/C mutant embryos were arrested in a meiotic state and contained cortical granules(Fig. 5A). sep-1(RNAi)embryos were not arrested but still retained cortical granules(Fig. 5B).

We compared the ultrastructure of the eggshell in wild-type, APC/C mutant and sep-1(RNAi) embryos. In the wild type, oocytes lacked an obvious covering (data not shown), but embryos in meiosis I contained cortical granules and had a single covering that is likely to be the raised vitelline layer (Fig. 5C). In older embryos without cortical granules (i.e. putative post-degranulation), a three-layered eggshell was observed (Fig. 5D), similar to previous results(Rappleye et al., 1999). APC/C mutant (Fig. 5E) and sep-1(RNAi) embryos (Fig. 5F) only had a single covering, similar to immature wild-type eggshells. Therefore, the formation of an impermeable three-layered eggshell requires secretion of cortical granule cargo, a process that is regulated by APC/C, separase and probably other OID genes.

The observation that inactivation of a subset of cell cycle regulatory genes caused the retention of cortical granules suggests that a specific regulatory pathway controls their exocytosis. Separase stood out among these genes because it is an important regulator of anaphase, the time when degranulation occurs. To investigate whether separase regulates degranulation,we measured the extent and kinetics of exocytosis during anaphase I. We monitored the progress of anaphase I by measuring the rate of homologous chromosome separation and counted the number of exocytic events observed in a single focal plane of histone::GFP expressing embryos labeled with FM2-10. In wild-type embryos, secretion begins 92±8 seconds (n=5 embryos)after the chromosomes initiate separation at 20°C, and 72±9 seconds(n=7 embryos) at 25°C. The exocytic wave occurs during the middle of anaphase before the polar body is extruded(Fig. 6C-E), and takes 121±6 seconds to complete (n=7 embryos) at 20°C, and 76±10 seconds (n=7 embryos) at 25°C. We observed 57±7 (n=7 embryos), and 59±7 (n=10 embryos)exocytic events during anaphase I at 20°C and 25°C, respectively(Fig. 6A,B).

We next analyzed the dynamics of degranulation in sep-1(RNAi) and sep-1(e2406) mutant embryos. Animals fed sep-1(RNAi) for 17 to 20 hours at 25°C displayed chromosome nondisjunction and dramatically delayed anaphase completion (Fig. 6B,F-H), consistent with sep-1 playing a role in CDK inhibition as observed in other species(Gorr et al., 2006). In addition, exocytosis was significantly reduced (34±7 events, P<10^-4, n=9 embryos, Fig. 6A,B), the anaphase spindle was disorganized and polar body extrusion failed(Fig. 6F-H). Surprisingly, sep-1(e2406) embryos entered anaphase without severe chromosome nondisjunction after 12 to 14 hours at 25°C(Fig. 6B), but only 26±6 exocytic events were observed (P<10^-4, n=9 embryos) (Fig. 6A,B). The reduced chromosome separation observed in sep-1(e2406) embryos during late anaphase (Fig. 6I,J) may be due to steric hindrance of polar body extrusion by the limited gap between the vitelline and plasma membrane and by the mislocalization of separase (see below). In some sep-1(e2406) embryos, a furrow enclosed the chromosomes at the cortex despite the reduced gap(Fig. 6K). In summary, both sep-1(RNAi) and sep-1(e2406) embryos have a similar reduction of cortical granule exocytosis, but sep-1(RNAi) shows much stronger cell cycle delays, more severe chromosome nondisjunction, spindle and polar body extrusion defects. Therefore, sep-1(e2406) is a hypomorphic allele that is compromised for a subset of the multiple distinct functions of separase.

As a control, we measured cortical granule exocytosis during anaphase in cks-1(RNAi) embryos, which have cell cycle delays and cytokinesis failures similar to those of separase mutants, but are not OID. We also analyzed chs-1(RNAi) embryos, which are OID because they lack chitin in the eggshell (Zhang et al.,2005). In both cases degranulation occurred normally(Fig. 6A). The spindle of cks-1(RNAi) embryos did not undergo the normal rotation and movement observed in the wild type (Yang et al.,2005), the distance that chromosomes separated during anaphase was reduced, and polar body extrusion failed (data not shown). However, after chromosomes separated, a normal wave of cortical granule exocytosis occurred(58±8 events, n=5 embryos, Fig. 6A). chs-1(RNAi)embryos failed polar body extrusion, but cortical granule exocytosis occurred normally (56±8 events, n=3 embryos, Fig. 6A). The plasma membrane became deformed immediately after secretion (data not shown), possibly due to the improper organization of eggshell components in the absence of chitin. Together, these results demonstrate that anaphase failures do not universally inhibit degranulation and that not all OID phenotypes result from secretion failures. The observation that sep-1(e2406) embryos have relatively normal anaphase progression yet reduced cortical granule exocytosis suggests that separase may have a direct role in regulating this process.

To examine further the role of separase in cortical granule exocytosis, we determined the localization of separase by expressing GFP::SEP-1 in oocytes and young embryos (Fig. 7A-Fand see Movie 7 in the supplementary material) and by immunofluorescence with polyclonal antibodies directed against endogenous separase(Fig. 7G-I). During oocyte maturation, cytoplasmic separase rapidly accumulated in the nucleus on homologous chromosomes. Separase also localized to unusual filaments distributed throughout the cortex (Fig. 7A,B,H). Several proteins have been found in similar filaments in meiosis I embryos (Monen et al.,2005), and also in Drosophila oocytes(Gilliland et al., 2007). We found that separase co-localized in filaments with HIM-10(Fig. 8A-C), the homologue of the kinetochore component nuf2 (Howe et al., 2001). Before anaphase onset, separase was lost from these filamentous structures and accumulated on cortical vesicles(Fig. 7C,G). Separase co-localized with WGA-labeled cortical granules during anaphase I(Fig. 8D-F). Separase-labeled vesicles were lost during anaphase I, beginning in the region near the meiotic spindle (Fig. 7G). We demonstrated that these vesicles undergo exocytosis by live cell imaging with correlative MPLSM and double-labeling with fluorescent dextran (see Movies 8 and 9 in the supplementary material). In mitotic embryos, separase localized to centrosomes, chromosomes and a diffuse cloud around the metaphase plate and anaphase spindle (Fig. 7E,F,I).

To obtain further insight into the sep-1(e2406) mutant phenotype,we localized the mutant SEP-1 protein by immunofluorescence in embryos incubated at 25°C for 11 hours. In the mutant, SEP-1 was still present on the anaphase I spindle (Fig. 8H,I), although it was reduced when compared with wild-type embryos. However, we did not detect the mutant protein on cortical granules during anaphase I (Fig. 8G-I). Therefore, the sep-1(e2406) allele appears to disrupt separase function partly through disrupting its localization to cortical granules. The observations that separase localizes to cortical granules and is required for degranulation suggests that separase could directly regulate exocytosis.

Cortical granule exocytosis is a phenomenon observed in the oocytes of many organisms (Wessel et al.,2001). We have identified cortical granules in C. elegansand have observed that degranulation occurs during anaphase I. We observed several steps of cortical granule trafficking and eggshell formation and identified a number of OID genes that perturb this process when inactivated(Fig. 9). Several cell cycle genes that cause OID are required for the exocytosis of cortical granules. In particular, separase localized to cortical granules during anaphase I and was required for degranulation (Fig. 9). It is possible that OID genes such as securin, spindle assembly checkpoint components and APC/C affect cortical granule exocytosis partly through regulation of separase.

It will be important to decipher the signals that trigger egg activation in C. elegans. Oocytes induced to mature by mutant sperm that are incapable of fertilization fail to complete anaphase I properly, do not form a proper eggshell and do not attempt meiosis II(McNally and McNally, 2005). Interestingly, cortical granules were still capable of fusing to the plasma membrane in unfertilized oocytes (Sato et al., 2006). As there is no discrete cell cycle arrest after maturation in C. elegans (McNally and McNally, 2005), unfertilized oocytes enter anaphase I. Entry into anaphase requires separase activation, which triggers degranulation. Further studies are required to understand the nature of the eggshell defect associated with unfertilized oocytes, and to know whether CGE is impaired. One possibility is that the wave of exocytosis may not occur with normal kinetics,disrupting eggshell formation. We hypothesize that signals possibly from both ovulation and fertilization stimulate an egg activation pathway acting upstream of, or in parallel with, separase to coordinate secretion of eggshell components with cell cycle progression in C. elegans.

Our data indicate that exocytosis of cortical granules is required for deposition of a mature three-layered eggshell in C. elegans. The outer vitelline layer, which covers oocytes, may function in sperm binding and restricting the free diffusion of cortical granule cargo, similar to its role in other species (Wessel et al.,2001). Interestingly, the vitelline envelope appears to separate from the plasma membrane at the time of fertilization, which could be part of a mechanism to prevent polyspermy. Our results also indicate that degranulation deposits the constituents of the inner two layers. Complex fibers made up of chitin and protein arranged in sheets have been observed in other nematodes (Wharton,1980). We suspect that the chitin synthase, chs-1,glycosylation enzymes such as sqv-4 and the cpg-1/-2proteoglycans are involved in eggshell layer formation(Herman et al., 1999; Hwang and Horvitz, 2002; Johnston et al., 2006; Olson et al., 2006; Zhang et al., 2005). As the C. elegans eggshell is permeable until late meiosis II, an elaborate process is probably involved in formation of mature eggshell. Further studies will be required to define how the secretions from the oocyte are assembled into the layers of the eggshell.

Our observations reveal that cortical granule exocytosis occurs during anaphase I and is regulated by separase. There are precedents for functions of separase outside cohesin cleavage. For example, in budding yeast, separase plays a non-proteolytic role in the regulation of anaphase spindle dynamics(Higuchi and Uhlmann, 2005; Sullivan and Uhlmann, 2003),controls asymmetric spindle pulling forces(Ross and Cohen-Fix, 2004) and activates a signaling pathway that promotes exit from mitosis and cytokinesis(de Gramont and Cohen-Fix,2005; Stegmeier et al.,2002). In frogs, separase plays a role in the centrosome duplication cycle (Tsou and Stearns,2006). In the mouse, proteolytic-inactive separase can rescue polar body extrusion but not chromosome segregation in separase-knockout oocytes (Kudo et al., 2006). Separase is required for progression through meiosis in both mouse and frog oocytes in part because it binds and inhibits CDK(Gorr et al., 2005; Gorr et al., 2006).

We have shown that separase localizes to cortical granules and loss of separase activity reduces exocytosis during anaphase I. Cortical granule exocytosis during anaphase I might create the necessary extracellular space,as well as provide additional membrane, for polar body extrusion. Consistent with this notion, inactivation of separase dramatically reduced the extracellular space near the polar body and restricted polar body extrusion(Fig. 6F-K). In sep-1(e2406) mutant embryos the initial separation of homologous chromosomes was essentially normal, but cortical granule exocytosis was still reduced. This is consistent with our observation that sep-1(e2406)mutant protein does not localize properly to cortical granules but retains some localization to chromosomes. These observations suggest that the sep-1(e2406) mutation might preferentially affect the role of separase in cortical granule exocytosis while retaining proteolytic activity toward cohesin. This suggestion is strengthened by the position of the sep-1(e2406) mutation (Siomos et al., 2001), which does not reside in the catalytic domain, but rather within a large N-terminal domain predicted to contain ARM repeats(Viadiu et al., 2005).

We observed rare cortical granule exocytic events during anaphase II in separase mutant and RNAi-depleted embryos (data not shown), further suggesting that cortical granule exocytosis is promoted specifically during anaphase. The observation that cell cycle regulatory genes control cortical granule exocytosis is surprising given observations made in other organisms. In Xenopus, inhibiting exit from mitosis by a non-degradable cyclin B did not prevent cortical granule exocytosis during egg activation(Murray et al., 1989). Although mouse oocytes can be parthenogenetically activated to undergo degranulation without cell cycle resumption(Ducibella et al., 2002),another study indicated that chromosome separation initiates before cortical granule exocytosis (Tombes et al.,1992). Studies in hamster oocytes indicated that limited exocytosis of cortical granules over the meiosis I spindle occurs during anaphase, contributing membrane for polar body extrusion(Okada et al., 1986). These data suggest that cortical granule exocytosis could also be regulated by separase during anaphase in oocytes of other organisms.

The finding that separase regulates cortical granule exocytosis during anaphase I leads us to speculate that separase may also regulate secretion during mitotic cell divisions. This control mechanism provides an elegant solution for synchronizing membrane trafficking during cell cleavage with chromosome segregation by linking these events through the activity of separase. We have observed multinucleate cells in sep-1(e2406)mutants shifted to the non-permissive temperature after eggshell formation(data not shown), suggesting that loss of separase activity can cause cytokinesis defects in mitotic divisions. It is well known that secretion is required during late anaphase and cytokinesis(Albertson et al., 2005), and separase was identified in a screen for genes required for secretion(Bard et al., 2006). Further analysis of how separase promotes exocytosis during anaphase I, and whether it regulates cytokinesis by a similar mechanism should be a fertile focus of future studies.

We thank Kraig Kumfer for cloning bombardment vectors, Haining Zhang for staining protocols and other lab members for feedback; Peter Crump for statistical analysis and Leanne Olds for drawing illustrations; Bill Vogt,Willy Hausner and Mike Szulczewski of Prairie Technologies (Middleton, WI) for technical assistance with the swept-field Confocal; Bill Bement, Anna Skop,Bob Goldstein, Ji Liu and Francis Pelegri for sharing equipment, reagents and expertise. The Barbara Meyer lab graciously provided the HIM-10 antibody. cDNA clones were obtained from Yuji Kohara. This work is supported by the University of Basel to A.S., by the Intramural Research Program at the National Institutes of Health (NIH) National Institute of Diabetes and Digestive and Kidney Diseases to A.G. and C.R., a NIH grant (R01 GM7583) to J.W., a NIH SBIR grant (5R44MH065724) with Prairie Technologies to K.E. and a NIH training grant (5F32 GM076867-02) to J.B.