The C. elegans genome has two gna homologues, gna-1 and gna-2. RNAi with gna-1 results in a wild-type phenotype, while RNAi with gna-2 is embryonic lethal (http://WS140.wormbase.org). To explore the role of GNA-2 in C. elegans development, we isolated (qa705), an allele lacking the entire coding sequence (Figure 2A).

gna-2(qa705) males are fertile, and hermaphrodites are viable but 100% penetrant for the maternal effect embryonic lethal (Mel) phenotype. An extrachromosomal array containing the coding sequence for gna-2 (nucleotides 20340–23478 in reference cosmid T23G11) partially rescued the gna-2(qa705) embryonic lethal phenotype and resulted in a high incidence of males (Him) phenotype. Hatchling numbers/hermaphrodite, % Him and number (n) of hermaphrodites assayed were as follows. (i) Wild type with no extrachromosomal array: 261 hatchlings, 0.2% Him, n = 10; (ii) wild type carrying an extrachromosomal array of the visible Rol6 marker, pRF4: 219 hatchlings, 0.1% Him, n = 10; and (iii) two independent lines of gna-2(qa705) carrying the visible Rol6 marker, pRF4, and also carrying an extrachromosomal array bearing a translational fusion of gna-2::GFP (line 1: 48 hatchlings, 2.7% Him, n = 10; line 2: 39 hatchlings, 4.6% Him, n = 10). In adult gna-2(qa705) hermaphrodites, UDP-GlcNAc was undetectable (Figure 2B, red arrow), and its epimer, UDP-GalNAc, was decreased to 19% of wild-type levels (Figure 2B, blue arrow). Three alleles (qa707, qa708, qa709) of a single suppressor (sup-46) of gna-2(qa705) were identified, and sup-46 required gna-1 for suppression of lethality, suggesting that UDP-GlcNAc synthesis is essential for embryonic development (Figure 2C,D). Furthermore, the gna-2(qa705) Mel phenotype was partially rescued by injection of 50–250 mM UDP-GlcNAc into the distal germline (Figure 2E). These results demonstrate that the Mel phenotype of gna-2(qa705) results from loss of GNA-2 catalytic activity, which causes UDP-GlcNAc deficiency in the germline or early embryo. Thus, de novo synthesis of UDP-GlcNAc is essential for early development in C. elegans. The UDP-GlcNAc end product, chitin, was markedly reduced in embryos dissected from gna-2(qa705) hermaphrodites (henceforth referred to as gna-2(qa705) embryos) (Figure 2F), as was reactivity with the lectins WGA, LEA, STA and SBA (Figure 2G). In wild-type embryos, WGA staining revealed transient cortical-membrane punctae that appeared by the first meiotic division and disappeared by the two-cell stage (Figure 2G). LEA, STA, SBA and CBD-F staining were not punctate, indicating that WGA reactivity in wild-type embryos reveals a non-chitin glycoconjugate(s) (WGA-G). WGA binds with high affinity to α2–3-linked sialic acids in mammalian cells. However, these structures are absent in C. elegans, and Natsuka et al 51 have shown that WGA binding in C. elegans most likely reflects α-GalNAc modification of mucin-type glycoproteins.

Live in utero recordings of gna-2(qa705) embryos expressing tubulin::GFP showed meiotic spindles that wax and wane with normal timing (Figure 3A,C). Meiosis I+II was almost identical for wild type (29 min) and gna-2(qa705) (31 min). Furthermore, the time during which a meiosis II spindle could be visualized in wild type (8 min) and gna-2(qa705) (12 min) (Figure 3A,B; bottom rows) is consistent with the timing for wild type reported by Liu et al53, (13 min), but very different from the timing for cul-2(RNAi) (meiosis II spindle still present at 36 min), which has a perdurant meiosis II spindle.

As an independent measure of meiotic timing, we examined embryos expressing histone H2B::GFP (Figure 3D,F). Results for wild-type meiotic timing (13 min for meiosis I, 25 min for meiosis II) were similar to those previously reported for wild type (15–17 min for meiosis I, 30–31 min for meiosis II) 5253. Furthermore, gna-2(qa705) had meiosis I and meiosis II timings similar to wild type. This is very different from results reported for cul-2(RNAi) or zyg-11(RNAi), in which meiosis I timing was normal, but meiosis II time was extended (60–75 min) 5253. Combined with the tubulin::GFP results, the histone H2B::GFP results support the conclusion that gna-2(qa705) does not have a perdurant meiosis II spindle. Using the histone H2B::GFP, we detected stray chromatin (distinct from the two masses segregated at anaphase) in 8% of gna-2(qa705) embryos at anaphase I, and 15% of gna-2(qa705) embryos at anaphase II, suggesting chromosome nondisjunction (Figure 3E, red arrow, Table 1). Additionally, a polar body was not extruded following anaphase I, and all maternal DNA entered meiosis II, resulting in 12 sister chromatid pairs at prophase II (100% of embryos), rather than the normal number of 6 (Figure 3D, 17 min time point, Table 1). A second polar body formed after anaphase II, but was not extruded (100% of embryos; Table 1). Whereas polar-body extrusion was defective, maternal and paternal pronuclei decondensed and met with normal timing (Figure 3D, yellow asterisk, Figure 3F). The unextruded second polar body (Figure 3D, red asterisk) often migrated to meet the pronuclei, although its migration usually lagged behind that of the maternal pronucleus.

Live in utero recording also showed that the SPCC failed to associate closely with the cortex in gna-2(qa705) (69% of embryos) (Table 1, Figure 4). In wild-type embryos, SPCC-cortical association correlates with the initiation of polarization, which results in the anterior spread of a PAR-1,2-binding smooth posterior cortex that culminates in the pseudocleavage furrow. In addition to showing failed SPCC-cortical association, a pseudocleavage furrow was never observed in gna-2(qa705) (100% of embryos; Table 1, Figure 4). Furthermore, cortical polarity was defective in gna-2(qa705) embryos, as indicated by a failure to segregate PAR-3 to the anterior cortex (Figure 5A), and cortical PAR-2 was reduced and mislocalized following GNA-2 depletion by gna-2 RNAi (Figure 5A, Table 2; 150 mM KCl treatment). Cytoplasmic polarity was also defective, as posterior localization of P granules was not observed and PIE-1 was undetectable (Figure 5A).

Differential interference contrast (DIC) imaging (Leica Leitz DMRB) revealed that gna-2(qa705) embryos had some form of eggshell (Figure 5B, arrows); however, the shell was deficient in chitin and other lectin-reactive glycoconjugates (Figure 2F,G) and did not maintain a rigid oval shape in the uterus. gna-2(qa705) embryos were also osmotically sensitive, swelling in H₂O and shrinking in 300 mM KCl, and were permeable to the lipophilic dye, FM 4–64, consistent with an osmotic/permeability barrier defect (Figure 5B,C). Together with the polarity defects, this identifies gna-2(qa705) as a novel Pod mutant. It is possible that polarization defects in Pod mutants might be secondary to osmotic defects, as has been shown for the cytokinesis defect of cyk-3 mutants 54. As such, the chitinous eggshell may be essential to provide an osmotically appropriate environment for A-P polarization. To determine if GNA-2-deficient embryos could polarize if provided with synthetic osmotic support, worms expressing PAR-2::GFP were subjected to gna-2 RNAi, and embryos were dissected and incubated in minimal embryonic growth medium (EGM) 55. EGM provided little rescue of polarization (Table 2), suggesting that polarization defects in GNA-2-deficient embryos may be dissociable from osmotic sensitivity. However, it is also possible that EGM did not provide adequate and/or timely osmotic support.

The C. elegans genome encodes two chitin synthase (chs) homologues. chs-2 is expressed in the pharynx and head neurons of larvae and adults, and chs-1 is expressed in the hermaphrodite germline, in eggs and in dauer larvae 5657. Consistent with its expression during early development, chs-1 RNAi blocked eggshell chitin production and resulted in a highly penetrant embryonic lethality that resembled gna-2(qa705) (Figure 6A,B). Embryos were fragile, osmotically sensitive and FM 4–64-permeable (Figure 6C,D). Meiotic timing was normal (Figure 3F), but embryos failed to extrude the polar bodies and had 12 sister chromatid pairs at meiosis II (100% of embryos; Table 1). Furthermore, stray chromatin detected at anaphase II (10% of embryos) suggested meiotic chromosome segregation defects (Table 1). Additionally, the SPCC failed to associate closely with the cortex (30% of embryos) and a pseudocleavage furrow was not detected (100% of embryos) (Table 1, Figure 4). PAR-3 and P granules were mislocalized and PIE-1 was undetectable (Figure 6E), showing that polarization was defective. Moreover, PAR-2::GFP was undetectable or mislocalized, and incubation in minimal EGM did not rescue localization (Figure 6E, Table 2). The nearly identical phenotypes of gna-2(qa705) and chs-1(RNAi) support the conclusion that chitin deficiency is responsible for the gna-2(qa705) Mel phenotype. The Pod phenotype of hypomorphic alleles of genes encoding APC subunits might suggest that APC activity is required for chitin synthesis. However, while mat-2(ax76) embryos at the non-permissive temperature (25°C) were arrested at meiotic metaphase I, they still produced a chitinous eggshell, showing that chitin synthesis does not depend on APC activity (Figure 6F). This result also indicates that the Pod phenotype of hypomorphic APC alleles does not result from chitin deficiency.

cej-1 and B0280.5 encode proteins predicted to be secreted mucins with multiple chitin-binding Peritrophin-A domains (Figure 7A). Furthermore, cej-1 and B0280.5 are targets of the germline translational repressor GLD-1, and combined cej-1 and B0280.5 RNAi (cej-1+B0280.5(coRNAi)) has an embryonic lethal phenotype that resembles gna-2(RNAi) 22 (Figure 7B). These results suggested that CEJ-1 and B0280.5 might be required for eggshell assembly or function. cej-1+B0280.5(coRNAi) did not prevent eggshell chitin deposition (Figure 7C), and meiotic timing was normal (Figure 3F). However, embryos had stray chromatin at anaphase I (6% of embryos) and anaphase II (12% of embryos), failed to extrude the polar bodies (76% of embryos for polar-body 1, 82% of embryos for polar-body 2), and had 12 sister chromatid pairs at meiosis II (76% of embryos) (Table 1). Moreover, SPCC-cortical association failed (47% of embryos), and a pseudocleavage furrow was not detected (94% of embryos) (Table 1, Figure 4). Additionally, PAR-3 and P granules were mislocalized, PIE-1 was undetectable, PAR-2::GFP was undetectable or mislocalized, and PAR-2::GFP localization defects were not rescued by incubation in minimal EGM (Figure 7D; Table 2). Furthermore, like gna-2(qa705) or chs-1(RNAi), cej-1+B0280.5(coRNAi) resulted in embryos that were osmotically sensitive and FM 4–64-permeable (Figure 7E,F). These findings support a model in which GNA-2 is required for CHS-1-dependent synthesis of eggshell chitin, which binds CEJ-1 and B0280.5, proteins with essential but overlapping roles in meiotic chromosome segregation, polar-body extrusion, osmotic barrier function and polarization.

The DYRK-related kinase, MBK-2, is required for polarization and asymmetric division. Furthermore, like WGA-G (Figure 2G), it is cortically localized in oocytes in the proximal germline, and in the embryo its cortical distribution changes from uniform to dramatically punctate during meiosis 41. To determine if MBK-2 and WGA-G colocalize, hermaphrodites expressing MBK-2::GFP were dissected, and germline and embryos were costained with a GFP antibody and WGA conjugated to tetramethylrhodamine-5-(and-6)-isothiocyanate (WGA-TRITC). WGA-G and MBK-2 colocalized in the cortex of oocytes in the proximal germline and in the cortex of newly fertilized embryos (Figure 8A). Furthermore, cortical WGA-G and MBK-2 punctae were completely coincident at anaphase of meiosis I (Figure 8A). MBK-2 localizes to centrosomes and chromosomes 41 by first mitosis, at which point, WGA-G punctae are no longer detected.

WGA-G and MBK-2 colocalize in the germline, and WGA-G was intact in mbk-2(pk1427) (Figure 8B), a putative null allele, suggesting that WGA-G could be required upstream of MBK-2 localization. In that case, disruption of embryonic WGA-G by chs-1 RNAi might also prevent MBK-2 localization. Hermaphrodites expressing MBK-2::GFP were subjected to chs-1 RNAi, and germline and embryos were costained with a GFP antibody and WGA-TRITC. Cortical MBK-2 was undetectable in oocytes and in embryos following chs-1 RNAi (Figure 8C). In contrast, neither MBK-2 nor WGA-G was mislocalized following cej-1+B0280.5(coRNAi) (Figure 8D). Chitin was not detected in oocytes in the germline, using the highly specific chitin-binding probe, CBD-F (Figure 8E), but chs-1 RNAi disrupted oocyte localization of MBK-2. The simplest explanation for these findings is that chs-1 RNA or CHS-1, rather than chitin itself, is required for MBK-2 localization. In this regard, chitin synthases 58 and the homologous vertebrate hyaluronan synthases 59, are unusual in their localization to the plasma membrane, rather than the endoplamic reticulum/Golgi apparatus. Saccharomyces cerevesiae chitin synthase-III is found in membranous organelles called chitosomes, and chitin synthase activation probably requires both enzyme phosphorylation and transport of chitosomes to the plasma membrane 586061. Therefore, one possibility is that MBK-2 may be localized to the oocyte cortex by virtue of its association with a CHS-1-containing compartment. Alternatively, chitin may be present in oocytes in the germline at a level that was too low to detect, or in a form that was masked by binding to a germline-expressed chitin-binding protein.

chs-1 RNAi disrupts MBK-2 localization, including the cortical patches that presage MBK-2-dependent APC degradation of MEI-1 and MEI-2. Furthermore, chs-1(RNAi) and mbk-2(pk1427) both fail to localize P granules 41, suggesting that MBK-2 deficiency might contribute to the chs-1(RNAi) phenotype. However, failed SPCC-cortical association in chs-1(RNAi) may be sufficient to explain P-granule mislocalization. Furthermore, cej-1+B0280.5(coRNAi) embryos mislocalize P granules, but MBK-2 localization is preserved. Taken together, these results suggest that MBK-2 mislocalization is not responsible for P-granule mislocalization in chs-1(RNAi). Similarly, while chs-1(RNAi) and mbk-2(pk1427) both exhibit a multinucleated phenotype, cej-1+B0280.5(coRNAi) shares this phenotype; therefore, it is probably not attributable to MBK-2 mislocalization. The apparent insignificance of disruption of MBK-2 in chs-1(RNAi) may reflect the earlier and/or more severe developmental defects in chs-1(RNAi).

Treatment of C. elegans embryos with latrunculin A, a drug that inhibits actin polymerization by sequestering G-actin, prevents polar-body extrusion and results in a phenotype of 12 sister chromatids at meiotic metaphase II 62. Embryos deficient in PFN-1, an actin-binding protein, have a similar phenotype 62. Here we demonstrate that in spite of normal meiotic timing, gna-2(qa705), chs-1(RNAi) or cej-1+B0280.5(coRNAi) embryos have a latrunculin A-like meiotic phenotype, with failed polar-body extrusion, and 12 sister chromatid pairs at meiosis II. Therefore, a critical role of the eggshell in meiosis appears to be support of actin-dependent polar-body extrusion.

C. elegans males arise at a frequency of about 0.5%, by non-disjunction of the X chromosome, and a phenotype showing a high incidence of males (Him) is a common finding in mutants with meiotic chromosome-segregation defects 63. gna-2(qa705) rescued by extrachromosomal array was Him, and stray chromatin was detected at meiosis I and meiosis II, in some gna-2(qa705) embryos. Furthermore, chs-1(RNAi) or cej-1+B0280.5(coRNAi) embryos also had stray chromatin at meiosis. These findings indicate that the eggshell is required for faithful meiotic chromosome segregation. Meiotic spindles localized to the embryonic cortex in gna-2(qa705), chs-1(RNAi) or cej-1+B0280.5(coRNAi) embryos and meiosis occurred with normal timing, suggesting that gross meiotic spindle defects did not underlie the chromosome-segregation defects. However, it is possible that more subtle defects in spindle microtubule structure could explain the lagging chromosome phenotype. In starfish, and perhaps also in Xenopus and mouse 646566, actin is essential for chromosome capture by the spindle during chromosome segregation. Therefore, the chromosome segregation defects identified in this study may also reflect a requirement for the eggshell in supporting actin-dependent chromosome capture.

FM 4–64 dye permeability is a reliable indicator of osmotic sensitivity in Pod mutants, and in wild-type embryos treated with a laser pulse to perforate the eggshell 5. In wild-type embryos, the first polar body stained with FM 4–64, demonstrating that it is external to the barrier. Therefore, meiosis I and extrusion of the first polar body precede development of the osmotic/permeability barrier. Accordingly, the defects in meiosis I segregation and first polar-body extrusion in gna-2(qa705), chs-1(RNAi) and cej-1+B0280.5(coRNAi) are not likely to be secondary to osmotic defects. CEJ-1 and B0280.5 bind chitin 11; furthermore, they are substituted with chondroitin chains and are predicted to be mucins, which are highly hydrophilic and hydrating glycoproteins 11. Chondroitin proteoglycans are required in the C. elegans hermaphrodite to prevent collapse of the developing vulva 12. Furthermore, they are essential for polar-body extrusion and for the separation of the embryonic plasma membrane from the eggshell 12. Therefore, analogous to the role of chondroitin-proteoglycans in the developing vulva, CEJ-1/B0280.5 may support actomyosin-mediated cytokinesis of polar bodies by generating a hydrated matrix.

SPCC-cortical association failed in gna-2(qa705), chs-1(RNAi) and cej-1+B0280.5(coRNAi), and pseudocleavage furrow formation was not detected. Furthermore, in fixed embryos, cortical PAR-3 was not asymmetric, cortical PAR-2 was reduced and mislocalized, cytoplasmic P granules were not confined to the posterior, and cytoplasmic PIE-1 was undetectable. A perdurant meiotic spindle has been proposed as a possible explanation for reversal of polarization in cul-2(RNAi) 53. However, in this study, we determined that meiotic spindle duration and meiotic timing were normal; therefore, the polarization defects associated with deficiency of eggshell chitin or CEJ-1/B0280.5 cannot be a consequence of an extended meiosis II and/or a perdurant meiosis II spindle.

The importance of actomyosin contraction in polarization is underscored by the previous observation that loss of function of actin-associated proteins, such as POD-1, NMY-2, MLC-4 or PFN-1, prevents both cytoplasmic flow and polarization 5366768. Our results demonstrate that actomyosin-dependent polarization also requires eggshell chitin and CEJ-1/B0280.5. While most gna-2(qa705), chs-1(RNAi) and cej-1+B0280.5(coRNAi) embryos showed failed SPCC-cortical association, in a few embryos the SPCC appeared to associate closely with the cortex, but pseudocleavage was not seen. These results suggest that in addition to being required for SPCC movement to or association with the cortex, the eggshell may also be required subsequently for the anterior-directed cortical-domain movement that follows SPCC-cortical association.

PAR mutants are polarity-defective, but osmotically insensitive. Therefore, polarization is not required for development of the osmotic barrier. In contrast, polarity is usually defective in osmotically sensitive mutants, and the Pod phenotype results from deficiency of a diverse group of proteins, including POD-1 (an actin-binding protein), APC subunits (MAT-1–3, EMB-27,30), enzymes required for fatty-acid synthesis (POD-2, an acetylCoA carboxylase, EMB-8, an NADPH-cytochrome P450 reductase, F32H2.5, a fatty acid synthase) 54649, a ubiquitin C-terminal hydrolase required for actin-dependent processes (CYK-3) 54 and proteins required for eggshell synthesis (GNA-2, CHS-1, CEJ-1/B0280.5) (this study). These results indicate that polarization may require an osmotically appropriate environment, and that Pod-class polarity defects could be secondary to osmotic sensitivity. The results of our study, and those of Rappleye et al49 for emb-8(hc69), emb-8(RNAi), pod-2(RNAi) or fatty-acid synthase (F32H2.5)(RNAi), showing that polarization was not rescued by incubation in isotonic medium, suggest that in some Pod mutants polarity defects may not simply be a consequence of osmotic defects. Intriguingly, Rappleye et al49 also determined that pod-2(ye60) supplemented with long-chain fatty acids (C16-C22) were normally polarized when dissected in isotonic medium. Unfortunately, however, in that study it was not reported whether polarity was defective in embryos dissected in non-isotonic medium, probably reflecting the fact that osmotically sensitive embryos often burst (or crenate) without osmotic support. Therefore, while the results showed that a combination of lipid supplementation and osmotic support rescued polarization in pod-2(ye60), it is not clear whether the rescue was provided by the lipids or by the isotonic medium. Additionally, the osmotically sensitive mutants emb-8(t1462) and 3-hydroxy-3-methylglutaryl coenzyme A (HMG-CoA) reductase (F08F8.2)(RNAi) were normally polarized when dissected in isotonic medium but no data were presented for embryos dissected in non-isotonic medium. As such, the question of whether or not polarization defects of Pod mutants are secondary to osmotic sensitivity remains unresolved.

As has been suggested for the ascaroside layer in Ascaris species 6, a role of the osmotic barrier in the C. elegans embryo may be to prevent desiccation after the egg is laid. The molecule(s) responsible for this barrier has not been identified. It may include lipid molecules of the inner eggshell layer playing a role analogous to that of ascarosides. However, EM images of pod-1(ye11) showed that all three eggshell layers appeared intact, but embryos were osmotically sensitive and dye-permeable 5. Therefore, it seems likely that the C. elegans permeability barrier resides inside the inner eggshell layer, as suggested by our results showing that the first polar body (putatively in the inner eggshell layer) stains with FM 4–64, while the second does not. A deficiency of HMG-CoA reductase causes osmotic sensitivity, suggesting that the C. elegans osmotic barrier requires cholesterol 49. Furthermore, as long-chain fatty acids [palmitate, stearate, oleate, linoleate, arachidonate, docosohenaenoate or phosphatidylinositol (linoleic + palmitoleic)] were insufficient to rescue the osmotic sensitivity of pod-2(ye60) 49, the barrier may require shorter-chain fatty acids or other malonyl CoA-derived molecules. In this study, we determined that chs-1(RNAi) or cej-1+B0280.5(coRNAi) embryos have defects in the osmotic/permeability barrier. These results suggest that the chitinous middle layer of the eggshell is required for synthesis/extrusion of the inner eggshell layer, which could be the osmotic barrier itself, or is required for subsequent synthesis/extrusion of molecule(s) comprising the osmotic barrier.

The eggshell is required for multiple early developmental events, including meiosis, polar-body extrusion, osmotic barrier function and polarization. However, while the eggshell is physically associated with the embryonic plasma membrane at meiosis I, by polarization, additional layers separate the chitinous eggshell from the embryonic plasma membrane. This indicates that the embryonic plasma membrane is in contact with different extracellular molecules at different stages in the development of the one-cell embryo, which may suggest that the eggshell serves distinct functions at different stages. For example, during meiosis I, when it is closely associated with the plasma membrane, the eggshell might act as a physical scaffold for cortical molecules overlying the meiotic spindle. Subsequently, during polar-body extrusion, the eggshell might provide a fluid space into which the polar bodies can be extruded. Additionally, from second polar-body extrusion onwards, the eggshell could be required primarily to provide an osmotic barrier. Moreover, as long-chain fatty-acid supplementation suggests that membrane properties are crucial for polarization 49, the eggshell might be required during polarization for integrity of the plasma membrane. While these or other distinct, separate roles could underlie the requirement for the eggshell in multiple early developmental processes, a common feature of these events may explain, at least in part, the requirement for the eggshell. Specifically, as meiosis, polar-body extrusion and polarization all appear to be actomyosin-dependent, it seems likely that the eggshell supports actomyosin-dependent contractile events.

At both the cellular and the whole-animal level, some form of skeletal support usually facilitates contractile events. For example, in motile mammalian cells, dynamic adhesion to the extracellular matrix supports the pulling forces of the actomyosin network that move the cell forward 70. At the whole-animal level, skeletal-muscle contraction is facilitated by a skeleton, which can be internal or external, and rigid or hydrostatic. For example, in vertebrates and "hard-shelled" arthropods, contraction requires muscle attachment to the bony skeleton or to a rigid exoskeleton, respectively. In animals that lack a rigid skeleton, a hydrostatic skeleton can provide a zone of incompressible fluid to transmit the force of muscle contraction (reviewed by Chapman 71). By way of example, coelomic fluid in annelids is propelled forward by a wave of contraction of circular and longitudinal muscles, which results in a net movement of the worm in the direction of coelomic fluid flow 7172. Furthermore, while crustaceans normally depend on a rigid exoskeleton for body muscle contraction, a hydrostatic skeleton is used during the post-molt, soft-shell stage 73. Hydrostatic skeletons can also be utilized within the context of a rigid 'container'. For example, fluid in the gut of the nematode Aphelenchoides acts as a hydrostatic skeleton that facilitates movement either of gut contents or of the whole animal, depending on the coordination of body-wall muscle contraction. Specifically, when localized regions of dorsal and ventral body muscles contract in phase, gut contents are moved; conversely, a wave of alternating dorsal and ventral contraction results in sinusoidal locomotion of the worm.

CEJ-1 and B0280.5 bind chitin, and it is possible that they also bind, directly or indirectly, to plasma-membrane molecules that link to the actomyosin cytoskeleton. In this way, the eggshell and/or molecules in the PVF could act as a dynamic adhesive matrix to anchor focal actomyosin contraction during meiosis, polar-body extrusion and polarization, analogous to the role played by the extracellular matrix in supporting focal adhesion in migrating cells or extending axonal growth cones. Conversely, or additionally, as CEJ-1 and B0280.5 are chondroitin-modified and have multiple mucin domains, their role may reflect an ability to generate a fluid-filled space. In this case, transmission of actomyosin contractile forces in the one-cell C. elegans embryo may depend on the hydrated eggshell/PVF acting as a hydrostatic skeleton. This would be analogous to the situation in jumping spiders, where it has been proposed that a hydrostatic skeleton within a rigid exoskeleton allows application of a large amount of force to a small deformation, resulting in a forceful and quick movement 71. In the case of polar-body extrusion, localized asymmetric force application may facilitate cytokinesis. Similarly, during the actomyosin-dependent ruffling stage that is coincident with movement of the large SPCC to the cortex, fluid translocated into the ingressing ruffles (and away from the non-ruffling adjacent membrane) might be required to generate sufficient force to propel the SPCC to the cortex.

gna-2(qa705) was isolated by screening a chemical deletion library (1.1% ethyl methanesulphonate (EMS)-treated N2, generated in the laboratory of J. Culotti) by nested PCR using the primer pairs 5'-ACTCCGCCCCCAATCTTGTTGC-3' and 5'-AACCAGCGTACCGCACAGTTGC-3' and 5'-TTGTCGAAGGAAGGGGAGGAGC3' and 5'-AATCATTGTGGCGGCGAGGAGG-3'. Deletion breakpoints were established by sequencing the truncated PCR product. The gna-2 allele, qa705, was outcrossed 10 times to yield strain XA780 (gna-2(qa705)/unc-109(n499)I). The gna-2(qa705)-bearing chromosome was marked by recombination with unc-55(e1170) and balanced with gld-1(q485) to produce XA774 (gld-1(q485)/gna-2(qa705) unc-55(e1170)I). Stable transformation rescue of gna-2(qa705) was demonstrated with a fragment of the cosmid T23G11(nucleotides 20340–23478, Genbank accession number Z81130) sub-cloned into pBSKS^- to yield the construct pWJ8. pAK42-1, a rescuing gna-2::GFP translational fusion, was generated by inserting the GFP coding sequence of pPD95.79 (a gift of A. Fire) into pWJ8 just prior to the stop codon of gna-2 (nucleotide 23146 of Genbank accession Z81130). Transgenic animals were generated by gonad injection 75. Rol-6 (pRF4) was used as the visible co-injection marker, and empty pBSKS- was used to bulk out the arrays. Injections were conducted using a total DNA concentration of 150 ng/μl and a plasmid ratio of 1:9:5 rescuing construct: pBSKS^-:pRF4. For pWJ8, wild-type hermaphrodites were injected, and arrays were passed into a gna-2(qa705) unc-55(e1170)I background. For pAK42-1, gld-1(q485)/gna-2(qa705) unc-55(e1170) hermaphrodites were injected, and rolling Uncs were selected. To confirm rescue, individual fertile, rolling Unc worms were genotyped for the deletion by PCR.

To identify suppressors of gna-2(qa705) an F2 screen was performed. gld-1(q485)/gna-2(qa705) unc-55(e1170) L4 hermaphrodites were mutagenized using EMS (50 mM). F2 progeny were fed gna-2 RNAi to enrich for suppressors, and Unc-55 F3s were picked and pooled (5/plate). Plates were examined for fertility and lines were established from hermaphrodites throwing only Unc progeny. A minimum of 6000 haploid genomes was screened. The recovered alleles were outcrossed 10 times, subjected to complementation analysis, and mapped to linkage group I. qa707 and qa708 were isolated from gna-2(qa705) and then reconstructed with gna-2(qa705) by recombination, with the aid of unc-101(m1).

The following additional strains were also constructed: XA781 (gld-1(q485)/gna-2(qa705) unc55(e1170)I; unc-119(ed3) ruIs32 [unc-119(+)pie-1::GFP::H2B]III), XA782 (gld-1(q485)/gna-2(qa705) unc55(e1170)I; unc-119(ed3) III; ruIs57 [unc-119(+) pie-1::GFP::tubulin]), XA785 (gna-2(qa705) unc-55(e1170)I; qaEx747 [gna-2::GFP, pRF 4 pBS KS^-]), XA787 (gna-2(qa705) unc-55(e1170)I; qaEx749 [gna-2::GFP pRF 4 pBS KS^-]), XA792 (sup-46(qa707)I), XA794 (gna-2(qa705) unc55 (e1170) sup-46(qa707)I), XA796 (gna-2(qa705) unc55 (e1170) sup-46(qa708)I), XA797 (sup-46(qa708)I) and DE2 (dnEx2 [pRF 4 pBS KS^-]). We also obtained the following CGC strains: N2, CB4856, AZ212 (unc-119(ed3); ruIs32 [unc-119(+) pie-1::GFP::H2B]III), AZ244 (unc-119(ed3) III; ruIs57 [unc-119(+) pie-1::GFP::tubulin], DR1 (unc-101(m1)I), DR101 (dpy-5(e61)unc-55(e1170) I), DS97 (mat-2(ax76)II), JK1466(gld-1(q485)/dpy-5(e61)unc-13(e51)I), JH1576(unc-119(ed3) III; axIs1140 [pie-1::gfp-mbk-2]), JH1580(unc-24(e1172)mbk-2(pk1427)IV/nT1 [let-?(m435)](IV;V) and KK866 (itIs153 [pie-1::gfp::par-2, pRF4 N2 genomic DNA]).

RNAi was performed by feeding 76. The following RNAi feeding vectors were produced by cloning PCR-generated genomic sequence into the Sma1 site of pPD129.36 (a gift of A. Fire): pAK57-2 (B0280.5 primers 5'-TTCCTTCAAGGTGCGAGTTT-3' and 5'-AGTGCAAGCAGTGAATGACG-3'), pAK56-2 (cej-1 primers 5'-GAGACCACCACAGAAGCATATGCTCCAG-3' and 5'-GTGGTTTCAACGACTTCTGACTCGATC-3'), pAK55-1 (chs-1 primers 5'-TTCGTCCATCGGAAGTTTTC-3' and 5'CCAAGTACATGTCGATTGCG-3'). The gna-2 RNAi feeding vector, pAK38-2, was a PCR-generated cDNA fragment (primers 5'-CGGGATCCGATGCTATGAAAAAGGC-3' and 5'-GCTCTAGACTACAAGAGTGGCAGCACC-3') cloned into the BamH1 and XbaI sites of pPD129.36. Double-stranded RNA production was induced in liquid culture, and bacteria were collected by centrifugation and spread on NGM plates supplemented with ampicillin, tetracycline and isopropyl-beta-D-thiogalactopyranoside (IPTG). Escherichia coli strain OP50 on unsupplemented NGM was used as a control, unless otherwise stated. For combined RNAi (co-RNAi), approximately equal amounts of each bacterial preparation were mixed before spreading on the plates.

To determine hatchling number, single L1/L2 larvae (P_o) were transferred to individual plates and allowed to lay eggs. Total hatchling number/hermaphrodite was determined by aspirating F1 larvae prior to adulthood. For experiments quantifying % Him, the P_o was transferred to a new plate each day until no new hatchlings were observed, and the sex of the hatchlings on each progeny plate was determined at the L4 or adult stage. Results were analysed by one-way analysis of variance (ANOVA) followed by Tukey's multiple comparison test. Hatchling number was determined, rather than brood size or % of brood that hatched, because brood-size measurements would have been inaccurate for the following reasons: (i) embryos with a defective eggshell are fragile and sometimes break as they are laid; these embryos would have be excluded from the brood count; (ii) embryos with a defective eggshell are often laid as part of a large cohesive mass, which precludes accurate quantification of the number of embryos in the mass; and (iii) gna-2(qa705) and chs-1(RNAi) embryos are pale brown and non-refractile, making them easily confused with (unfertilized) oocytes.

To test biochemical rescue of the gna-2(qa705) Mel phenotype, young adult Unc-55 hermaphrodites were picked from gld-1(q485)/gna-2(qa705) unc55 (e1170)I plates. UDP-GlcNAc was dissolved in water (10, 250, 100 or 250 mM) and injected into the rachis of a single gonad arm as described previously 75. The number of hatchlings from each injected hermaphrodite was quantified by transferring individual hatchlings to single plates. Hermaphrodites throwing any wild-type or Gld-1 progeny were excluded from further analysis.

For HPLC analysis 1000 adult hermaphrodite wild type or gna-2(qa705) were collected in M9 buffer, washed three times in M9, incubated for 30 min at room temperature (RT), washed three times in H₂O, resuspended in 50 μl H₂O, frozen in dry ice/ethanol and stored at -80°C for no longer than 1 week. Samples were thawed rapidly, sonicated for 20 seconds, centrifuged, and the supernatant filtered and injected onto an anion exchange column (Zorbax SAX; Agilent Technologies, Chromatographic Specialities Inc., Brockville, ON). Detection was by ultraviolet absorption at 254 nm. UDP-GlcNAc, UDP GalNAc, UDP-Glc and UDP-Gal were quantified and UDP-GlcNAc and UDP-GalNAc were expressed as a percentage of UDP-Glc + UDP-Gal

For fixed embryos, young adult hermaphrodites were washed in 150 mM KCl and transferred to 0.1% poly-lysine-coated slides. Embryos were released by cutting at the gonad, or gonads were released by cutting just behind the pharynx, followed by freeze-fracture 63. Slides were fixed in 4°C MeOH for 10 min (PGL-1), in -20°C MeOH for 15 min (PAR-3) or in -20°C MeOH for 8 min (other antibodies, lectins and chitin-binding probe). After fixation, slides were washed for 2 min in phosphate-buffered slaine (PBS), and for 10 min in PBS with 0.5% Tween-20 (PBST), and were blocked in PBS containing 10% goat serum and 1% BSA (PBGS) for 30 min at RT. Primary antibodies were diluted in PBGS and applied overnight at 4°C at the following dilutions: anti-PGL-1 77, 1:10000; anti-PIE-1, 1:100 78; anti-PAR-3 1:30 79. PAR-2::GFP and MBK-2::GFP (in fixed embryos), were detected by an anti-GFP antibody (1:100; Molecular Probes, Inc., Eugene OR). The following secondary antibodies (diluted in PBGS, applied at RT for 2–4 h) were used: goat anti-mouse (GAM)-fluorescein isothiocyanate (FITC) (1:100; Pierce, Rockford IL); goat anti-rabbit (GAR)-rhodamine (1:100; Jackson ImmunoResearch Laboratories, Inc, Westgrove PA); GAR-FITC (1:100; Sigma-Aldrich, Oakville ON). Lectins (LEA-FITC, SBA-FITC, STA-FITC, WGA-FITC, WGA-rhodamine; all Sigma-Aldrich) were diluted in PBGS (40 μg/ml) and applied overnight at 4°C. The chitin-binding probe (chitinase A1 chitin-binding domain fused to maltose binding protein and labeled with TRITC (CBD-F); gift of Y. Zhang, New England Biolabs, Inc., Ipswich MA) was diluted 1:250 in PBGS and applied overnight at 4°C. DNA was stained with Hoechst (Molecular Probes), then SlowFade (Molecular Probes) was added prior to mounting, and fluorescence was detected on a Leica Leitz DMRB microscope. To test osmotic sensitivity, living embryos were dissected in H₂O, 150 or 300 mM KCl, transferred to the appropriate (H₂O, 150 or 300 mM KCl) 2% agarose pad on a glass slide, and viewed by DIC. Compression by the glass coverslip was prevented by depositing embryos into the fluid-filled space that was created by removing a central disc from the pad. To test whether osmotic support could rescue PAR-2 polarization wild-type worms expressing PAR-2::GFP were fed gna-2 RNAi, chs-1 RNAi or cej-1+B0280.5 coRNAi, and embryos were dissected quickly in 150 mM KCl followed by freeze-fracture (control), or were dissected in minimal EGM 55 and incubated at 22°C for 30 min before freeze fracture. Instead of M9, 150 mM KCl was used as a control because gna-2(RNAi), chs-1(RNAi) or cej-1+B0280.5(coRNAi) embryos dissected in M9 are very fragile and usually break, whereas the KCl provides some osmotic support. However, dissection and freezing was completed in <5 min, which should not allow for reversal of polarity defects.

To examine meiotic spindles, meiotic DNA segregation, polar-body extrusion, SPCC-cortical association, pseudocleavage furrow formation and pronuclear migration, wild-type or gna-2(qa705) hermaphrodites expressing tubulin::GFP or histone H2B::GFP were transferred to agarose pads, and embryos were examined in utero, using a Leica DMLSFA confocal microscope. The effect on meiotic DNA segregation, polar-body extrusion, SPCC-cortical association, pseudocleavage furrow formation and pronuclear migration of chs-1(RNAi) or cej-1+B0280.5(coRNAi) was also determined in worms expressing histone H2B::GFP. For meiosis duration (determined with tubulin::GFP), timing was defined as follows:

• Meiosis I: time from entry into the spermatheca until tubulin::GFP waned to a small spot (the tubulin::GFP signal does not wane to undetectable before the meiosis II spindle waxes). Time from entry into the spermatheca was used instead of nuclear envelope breakdown (NEBD) because it is easily and objectively assessed without DIC optics, and because it excludes any oocytes that undergo NEBD but fail to be ovulated (and fertilized) in a timely manner.

• Meiosis II: time from the end of meiosis I (defined above) until the meiosis II spindle waned to a very pale structure.

• Meiosis I + II: time from entry into the spermatheca until the meiosis II spindle waned to a very pale structure.

As an additional way to determine meiotic timing, we also examined chromosome segregation using histone H2B::GFP. Timing was defined as follows:

• Meiosis I: time from entry into the spermatheca until chromatin masses were separated (anaphase I).

• Meiosis II: time from entry into the spermatheca until chromatin masses were separated a second time (anaphase II).

• Pronuclear meeting: time from entry into the spermatheca until maternal and paternal pronuclei met.

Failure of the SPCC to associate closely with the cortex was scored as any embryo in which the sperm pronucleus (visualized with histone H2B::GFP) failed to come within half of the sperm pronuclear diameter of the cortex. Results were analysed by t test (tubulin::GFP experiment) or one-way ANOVA followed by Tukey's multiple comparison test (histone H2B::GFP experiment). Lipophilic dye permeability was tested by DIC/UV microscopy of living embryos dissected in 150 mM KCl containing 30 μM FM 4–64 (Molecular Probes).