Depletion of the co-chaperone CDC-37 reveals two modes of PAR-6 cortical association in C. elegans embryos

Asymmetric cell division is a fundamental mechanism for generating cell diversity. The first division of the C. elegans zygote provides a unique opportunity to study the effects of polarity on the ability of a cell to divide asymmetrically into daughters with different fates. The par genes(partitioning defective) are key regulators of polarity in the early embryo(Kemphues et al., 1988). The PAR proteins are conserved widely in animals and play roles in such diverse cellular functions as asymmetric cell division, apicobasal polarity in epithelial cells, oriented cell migration, neuronal axis specification and tight junction formation (Nishimura et al., 2004; Ohno,2001; Wiggin et al.,2005).

Considerable progress has been made in understanding events leading up to the first asymmetric division of C. elegans. After fertilization, the egg completes the meiotic divisions and produces egg coverings. Upon completion of meiosis, the zygotic surface begins to ruffle, the result of local cortical contractions. At the same time, the anterior PAR proteins(PAR-3, PAR-6 and PKC-3) appear uniformly at the cell periphery. Initially,the contractions are uniformly distributed over the surface, but shortly after the sperm pronucleus decondenses, the sperm centrosome signals to the cortex,leading to a local cessation of cortical ruffling and initiation of a cortical flow away from the posterior (Cheeks et al., 2004; Cowan and Hyman,2004; Munro et al.,2004). The anterior PAR proteins move with the flow toward the anterior, clearing them from the posterior, and PAR-1 and PAR-2 proteins are recruited from the cytoplasm to the smooth cortex in the posterior. Flow depends upon the activities of the PAR proteins(Cheeks et al., 2004; Munro et al., 2004). Mutations in the anterior PAR proteins allow an initial smoothing of the posterior cortex, but either block or greatly attenuate the flow into the anterior(Cheeks et al., 2004; Kirby et al., 1990; Munro et al., 2004). After the establishment phase, PAR-2 protein mediates the maintenance of asymmetry(Cuenca et al., 2003), in part by inhibiting a posterior-directed flow of cortical cytoplasm(Munro et al., 2004), and in part by antagonizing its own removal from the cortex by PKC-3 phosphorylation(Hao et al., 2006).

The establishment and maintenance of stable polar PAR domains is based on the principle of mutual exclusion. In the absence of the anterior PAR proteins, PAR-1 and PAR-2 are not restricted to the posterior, and, in the absence of PAR-2, the anterior proteins, as already described, are not stably restricted to the anterior (Boyd et al.,1996; Cuenca et al.,2003; Etemad-Moghadam et al.,1995; Tabuse et al.,1998; Watts et al.,1996). The mutual exclusion of anterior and posterior PAR proteins depends upon the 14-3-3 family member PAR-5(Morton et al., 2002).

Cortical localization of the anterior PAR proteins is dependent upon the activities of all three proteins: PAR-3, PAR-6 and PKC-3(Hung and Kemphues, 1999; Tabuse et al., 1998; Watts et al., 1996). In par-3 mutants, neither PAR-6 nor PKC-3 protein is detectable at the cell cortex. Similarly, in par-6 mutants, PKC-3 is not detectable at the cortex, and in pkc-3(RNAi) embryos PAR-6 is absent. However, in par-6 mutant and pkc-3(RNAi) embryos, PAR-3 accumulates weakly and somewhat asymmetrically at the cortex early in the cell cycle, but is undetectable after metaphase of mitosis. Cortical localization of the anterior PAR proteins also depends on a conserved interaction of PAR-6 with CDC-42, a Rho-family GTPase (Gotta et al.,2001; Kay and Hunter,2001) (D. Aceto, M.B. and K.K., unpublished).

To find other proteins with possible roles in establishing polarity in C. elegans, we identified homologs of a set of human proteins that co-purified with PAR proteins (Brajenovic et al., 2004) and tested them by RNA interference (RNAi). RNAi knockdown of one of them, W08F4.8, a homolog of human CDC37, caused a PAR-like phenotype. Concurrent with our analysis, a genome-wide RNAi screen identified this same gene as one of a small number of genes whose knockdown results in PAR-like early cleavages (Sonnichsen et al., 2005). In addition, Gunsalus and colleagues reported the distribution of GFP::CDC-37 in the early embryo(Gunsalus et al., 2005).

CDC37 was initially identified by a temperature-sensitive mutation defective in the cell cycle at Start in G1 in Saccharomyces cerevisiae (Reed, 1980). Subsequent analysis revealed that Cdc37p acts by maintaining the cyclin-dependent kinase Cdc28p in a state that is competent to associate with G1 and mitotic cyclins (Gerber et al.,1995). Vertebrate Cdc37 homologs have been identified in a number of studies as ∼50 kDa proteins that associate with Hsp90 and a variety of other proteins, particularly kinases(Hunter and Poon, 1997). The prevailing view is that Cdc37 functions as a cochaperone of Hsp90, interacting with protein kinases to regulate their activity, either by enabling them to fold correctly for proper activation or by preventing them from being degraded(Hunter and Poon, 1997). Among the many kinases regulated by Cdc37 and Hsp90 are Raf1(Grammatikakis et al., 1999; Perdew et al., 1997; Silverstein et al., 1998; Stancato et al., 1993), Cdk4(Dai et al., 1996; Stepanova et al., 1996), Akt(Basso et al., 2002) and Lkb1,a homolog of C. elegans PAR-4(Boudeau et al., 2003; Nony et al., 2003). However,evidence is accumulating that Cdc37 can act independently of Hsp90 and has a wider range of client proteins than has previously been appreciated(MacLean and Picard,2003).

In this paper, we report that reduction of CDC-37 in the C. elegans early embryo alters the dynamic interactions among the PAR proteins, leading to Par-like embryonic phenotypes. Our analysis of the CDC-37 phenotype has provided insight into the mechanisms of anterior PAR protein accumulation at the cortex.

Nematodes were cultured under standard conditions(Brenner, 1974). The Bristol N2 strain was used as wild type. Strains used for this analysis were KK0818, par-6(zu222) unc-101(m1)/hIn1 [unc-54(h1040)]I; KK0653, par-3(it71) unc-32(e0189)/qC1; JH1448, axEx1125 [pKR2.04,Ppie-1::GFP::MEX-5, pRF4, N2 genomic](Cuenca et al., 2003); SS629, [pie-1::gfp::pgl-1, unc-119(+)](Cheeks et al., 2004); KK0881, itIs160 [Ppie-1::GFP::PAR-6 unc-119(+)]; unc-119(ed3); JJ1473, zuIs45 [Pnmy-2::NMY-2::GFP unc-119(+)]; unc-119(ed3)(Nance et al., 2003); KK0911, itIs161 [Ppar-3::PAR-3 S863A::GFP unc-119(+)]; unc-119(ed3)] (B. Li and K.K., unpublished). The PAR-3::GFP protein in KK0911 carries an alanine substitution in the PKC-3 phosphorylation site of PAR-3 (Serine 863). PAR-3 S863A::GFP rescues par-3(it71) and gives a stronger cortical signal than does wild-type PAR-3::GFP.

For some experiments dsRNA was injected into the gonad arms of the worms(Fire et al., 1998). In this case, dsRNA was made in vitro from L4440 plasmids(Timmons and Fire, 1998)containing the gene of interest, using the RiboMAX kit (Promega). In other experiments, the L4440 plasmids containing the gene of interest were induced with IPTG in HT115(DE3) bacteria to produce dsRNA, and the bacteria were fed directly to worms to produce the RNAi effect(Timmons et al., 2001). For most RNAi studies, treated worms were incubated at 25°C for at least 18 hours before either antibody staining or time-lapse microscopy. For cdc-42(RNAi) studies, injected worms were incubated at 20°C for 8 hours, then 25°C for at least 16 hours before analysis. The efficiency of each RNAi experiment was determined by assessing the number of dead embryos laid by a subset of treated worms. Because of variable effectiveness of RNAi in C. elegans, comparisons between the results of different experiments are only reliable where the penetrance of each experiment is high.

Observations of live embryos with either DIC or fluorescence microscopy were made on a Leica DM RA2 microscope with a 63× or 100× Leica HCX PL APO oil emersion lens and Hamamatsu ORCA-ER digital camera. Digital images were captured using Openlab software (Improvision). All embryos were imaged at 22°C immersed in egg buffer (118 mM NaCl, 40 mM KCl, 3.4 mM CaCl₂, 3.4 mM MgCl₂, 5 mM Hepes, pH 7.4) on agar pads. The center of the embryo was chosen as the focal plane for the live observations. For cortical images, we focused at the cortex of partially flattened fixed embryos. We adjusted each image in PhotoShop, to remove background signal, by bringing the posterior cortical signal to zero. To count puncta, we sampled two similar-sized, non-overlapping windows from the center of the anterior cortex by laying an acetate sheet sequentially over the grayscale images of each fluorochrome and marking the positive puncta. We took the average of two samples for each embryo. Egg length was measured using Openlab software. Two-cell blastomere sizes were compared by measuring the length along the long axis of the embryo of AB and P1 in control and cdc-37(RNAi) embryos. Confocal images were collected on the Leica TCS SP2 system with the Leica DMRE-7 microscope and an HCX PL APO 63× oil immersion lens. Images were processed using the Leica Confocal SP2 software program and Adobe PhotoShop.

Immunostaining for PAR-2, PAR-3, PAR-6, PKC-3 and GFP was performed using a standard methanol fixation (Guo and Kemphues, 1995). Antibodies used were: anti-PAR-2 rabbit polyclonal (Boyd et al., 1996),1:3; anti-PAR-3 mouse monoclonal (Nance et al., 2003), 1:75; anti-PAR-6 rabbit polyclonal(Hung and Kemphues, 1999),1:15; anti-PKC-3 rat polyclonal (Hung and Kemphues, 1999), 1:10; and anti-GFP goat polyclonal (Rockland Immunochemicals), 1:3000. Slides were incubated with primary antibodies for 12-18 hours at 4°C. Slides were then washed three times in PBS, 0.5% Tween 20. All secondary antibodies were from Jackson Immunochemicals: goat anti-rabbit FITC, 1:200; goat anti-mouse Cy3, 1:250; donkey anti-rat Cy3,1:400; donkey anti-goat FITC, 1:100. Embryos were treated with secondary antibodies for 3 hours at 37°C. Slides were washed as before and then covered with VectaShield (Vector Laboratories).

We created a vector for expressing GFP::CDC-37 by inserting cdc-37cDNA (W08F4.8a, 1113 nucleotides, amplified from the Gibco Proquest Library)into the pJunc vector, a derivative of pMW1.03, constructed by Aaron Schetter(Gunsalus et al., 2005), which contains a 5.7 kb genomic fragment of unc-119 (pPDMM016)(Maduro and Pilgrim, 1995) and a par-2 coding sequence fused to GFP and driven by the pie-1promoter (Wallenfang and Seydoux,2000). The par-2 gene was removed from the vector by Spe1 cleavage and cdc-37 cDNA was inserted into the same site. Low-copy integrated lines expressing GFP::CDC-37 in the germline were identified among unc-119(ed3) worms following microparticle bombardment (Praitis et al.,2001).

To test for rescue, worms carrying the GFP::CDC-37 transgene were injected with dsRNA specific to the first 160 nucleotides of the endogenous cdc-37 3′UTR. dsRNA specific to the 3′UTR of a gene is an effective RNAi trigger (Parrish et al.,2000). Progeny of the injected worms were assayed for lethality 48 hours after injection. Eighty-five to 100% of embryos from 20 wild-type worms were dead compared with 32-50% of embryos from 28 GFP::CDC-37-expressing worms.

Populations of control or RNAi worms were collected from feeding plates 20 hours after feeding began; adults were bleached to obtain early embryos; and embryos were lysed in SDS sample buffer. Samples were analyzed by Bradford Assay (Bradford, 1976) to determine total protein present in each sample. Equal amounts of total protein were loaded onto an 8% SDS PAGE gel. The proteins were then transferred to nitrocellulose membrane for western blotting(Burnette, 1981) and stained with the MemCode Reversible Membrane Staining Kit (Pierce) to verify equal loading of samples. All primary antibodies were diluted 1:1000 in 1×TBST with 0.5% milk and incubated overnight at 4°C. Appropriate HRP-conjugated secondary antibodies (Jackson ImmunoResearch) were diluted 1:5000 in 1×TBST with milk, incubated for one hour at 22°C, then washed with TBST three times for a total of 30 minutes. Bands were visualized using a chemiluminescence detection reagent (Amersham Biosciences).

The wild-type S. cerevisiae GAL1/10 promoter was cloned into the pRS416 (CEN) URA3 plasmid (New England BioLabs). Wild-type yeast CDC37 or C. elegans cdc-37 (W08F4.8) cDNA was cloned in frame behind the promoter. These constructs were transformed into the yeast strains DP007 (cdc37-34ts) or RMY326 (control; kindly provided by D. Picard, Geneva University, Switzerland). Yeast were assayed for growth on galactose media at 25, 30, 35 and 37°C to test for rescue. In the absence of an antibody specific to CDC-37, we cannot rule out the possibility that our negative result was due to low expression of the worm protein.

In a study to identify proteins that interact with the mammalian Par proteins, Cdc37 co-purified in complexes with both aPKC and LKB1(Brajenovic et al., 2004). On the basis of this result, we sought to test Cdc37 homologs for roles in polarity establishment in C. elegans. Of the three potential Cdc37 homologs identified by sequence similarity, only one, W08F4.8, was predicted to encode a protein similar in overall size and domain structure to the H. sapiens CDC37 protein (see Fig. S1 in the supplementary material). W08F4.8 protein was found to interact with the Hsp90 family member DAF-21 in a yeast two-hybrid assay (Li et al.,2004).

In spite of sequence similarity to CDC37 and conservation of HSP-90 binding, we found that W08F4.8 cDNA expressed under the control of the yeast GAL1/10 promoter was unable to complement a temperature-sensitive cdc37 yeast mutation. Even though C. elegans W08F4.8 failed this test for orthology with yeast CDC37, it is the C. elegans gene most similar in sequence to yeast, fruit fly and mammalian Cdc37 proteins. Therefore we refer to the gene as cdc-37.

To test whether CDC-37 has a role in embryonic polarity in C. elegans, we depleted the message from early embryos by using RNAi. Wild-type worms injected with dsRNA specific to cdc-37 show increasing levels of embryo lethality: 30% at 0-17 hours versus >80% 18-24 hours post-injection. Embryos collected at this time show an approximately 80%reduction in steady-state GFP::CDC-37 levels (see Fig. S2 in the supplementary material). All embryos used for phenotypical analysis were dissected during this latter interval. Injected worms become sterile approximately 24 hours after injection. Sterile animals have sperm in the spermatheca as determined by DAPI staining, and developing oocytes are present; however, oocytes accumulate behind the spermatheca, suggesting a failure in egg maturation or ovulation (McCarter et al.,1999). Approximately 18% of the worms that escape embryonic lethality show a protruding vulva phenotype. Other less frequently observed post-embryonic phenotypes include larval arrest and tail rupture.

The affected embryos from cdc-37(RNAi) worms exhibit defects consistent with a loss of embryonic polarity(Fig. 1). The extent of cortical ruffling in cdc-37(RNAi) embryos varies from almost normal to barely detectable (Fig. 1F,G). The position of the pseudocleavage furrow in cdc-37(RNAi) embryos does not differ from that of controls(n=9 wild type and 11 cdc-37(RNAi); P=0.24). However, the pseudocleavage invaginations in cdc-37(RNAi) embryos are shallow (37.9±13.2 μm versus 47.4±5.1 μm in wild type, P=0.012), tend to be unilateral(Fig. 2G; 6/11 embryos), and are short lived (113±22 seconds, n=11, versus 191±56 seconds in wild type, n=11; P=0.0003). In cdc-37(RNAi) embryos, the two pronuclei meet more centrally(Fig. 1C,H; 56±8% egg length versus 70±3% egg length in wild type; n=14; P<0.0001) and the first cleavage is more equal(Fig. 1I; AB, the anterior two-cell stage blastomore, occupies 52±4% of total egg length, n=39, versus 58±2% in wild type, n=17; P<0.0001). The second cleavage is more synchronous after cdc-37(RNAi) (Fig. 1J)and is slightly slower than wild type. In cdc-37(RNAi) embryos, the period between nuclear envelope breakdown (NEB) in the zygote, P0, to NEB in its daughter cells AB and P1 was 1070±82 seconds, whereas the time between NEB in P0 and AB in wild type was 859±49 seconds(P<0.0001), and the time between NEB in P0 and P1 was 966±58.5 (P=0.0018; n=9, wild type; n=11, cdc-37(RNAi)). In contrast to the reproducible orthogonal spindle orientations of wild-type two cell embryos (n=11), cdc-37(RNAi) embryos display a variety of spindle orientations, with 85% exhibiting phenotypes different from wild type(Fig. 1J; n=39): 56%of the two-cell embryos divide with both spindles transverse to the long axis.

We also examined molecular markers of embryonic polarity, GFP::MEX-5(Cuenca et al., 2003; Schubert et al., 2000) and P granules using GFP::PGL-1 (Cheeks et al.,2004; Strome and Wood,1982; Strome and Wood,1983). We found that asymmetry of both markers is dependent upon CDC-37 (Fig. 2; for GFP::MEX-5,asymmetry was normal in 8/8 wild-type versus 1/10 cdc-37(RNAi)embryos; for GFP::PGL-1, asymmetry was normal in 12/12 wildtype versus 6/16 cdc-37(RNAi) embryos).

Many of the phenotypes described above are also exhibited by mutations in the par genes (Kemphues et al.,1988; Kirby et al.,1990; Morton et al.,2002; Tabuse et al.,1998; Watts et al.,1996). We tested whether depletion of CDC-37 alters the distributions of the PAR proteins. Five PAR proteins have a highly reproducible asymmetry in the developing embryo. PAR-3, PAR-6 and the atypical protein kinase C PKC-3 colocalize to the anterior cortex of the one-cell embryo (Etemad-Moghadam et al.,1995; Hung and Kemphues,1999; Tabuse et al.,1998), and PAR-1 and PAR-2 colocalize to the posterior cortex of the embryo (Boyd et al., 1996; Guo and Kemphues, 1995). We found that the distribution of both sets of proteins is dramatically altered by depletion of CDC-37.

In contrast to control embryos (Fig. 3A-C, n=32; Fig. 4, column 3, n=13) PAR-1, PAR-2, PAR-6 and PKC-3 are evenly distributed around the entire cortex of one-cell cdc-37(RNAi)embryos throughout the cell cycle beginning just after the completion of meiosis (Fig. 3D-I; Fig. 4, column 4; n=12; data not shown for PAR-1). The abundance of PAR-1 and PAR-2 at the cortex appears to be reduced relative to control embryos. The relative abundance of PAR-6 and PKC-3 does not appear to be affected, although PKC-3 levels are lower by western blot assays (see below). GFP::PAR-6 in control embryos enters the nucleus at nuclear envelope breakdown in 10 out of 10 embryos (not shown) (Cuenca et al.,2003). This fails to occur in eight out of nine cdc-37(RNAi) embryos, and these embryos consistently exhibit a mitosis-specific peri-centrosomal accumulation of GFP::PAR-6 that is never seen in control embryos (Fig. 4). Finally, whereas PAR-6 accumulates in cortical puncta in wild-type embryos (Hung and Kemphues,1999), in cdc-37(RNAi) embryos it has a much smoother cortical distribution (data not shown).

Surprisingly, PAR-3 behaves differently from PAR-6 and PKC-3 in cdc-37(RNAi) embryos. In control embryos, PAR-3 is evenly distributed along the entire cortex early in the cell cycle and becomes restricted to the anterior of the cell along with PAR-6 and PKC-3(Etemad-Moghadam et al., 1995)(Fig. 3B; Fig. 4; n=82). In cdc-37(RNAi) embryos, PAR-3, like PKC-3 and PAR-6, has a significant overlap with PAR-2 (Fig. 3D-F; n=57). However, in these embryos PAR-3 forms large cortical aggregates, and is more patchily distributed than are PAR-6 and PKC-3(Fig. 3E,F; Fig. 4). Furthermore, unlike PAR-6 and PKC-3, which remain uniformly distributed around the cortex in cdc-37(RNAi) embryos, PAR-3 clears from the posterior cortex during pronuclear migration. In control embryos, PAR-3 clears from the posterior cortical region to 49%±5% of the total egg length (n=10); in cdc-37(RNAi) embryos, the maximal clearing reaches only 40%±11% (n=19) of the egg length. Finally, whereas PAR-6 and PKC-3 remain cortical throughout the cell cycle in cdc-37(RNAi)embryos, PAR-3 consistently disappears from the cortex at metaphase(Fig. 4, column 2, fourth image). PAR-3 returns to the cortex late in the cell cycle and is distributed to both daughter cells in a patchy and variable distribution, disappearing again at the next metaphase and coming back to the cortex late in the second cell cycle. This pattern of PAR-3 dynamics is similar to that observed in par-6 mutant and pkc-3(RNAi) embryos(Tabuse et al., 1998; Watts et al., 1996).

These results were unexpected because in the cdc-37(+) background,cortical localization of PAR-6 and PKC-3 is dependent upon PAR-3(Hung and Kemphues, 1999; Tabuse et al., 1998). Therefore, we hypothesized that CDC-37 activity contributes to the normal dynamics of these proteins at the cortex, but when it is reduced or absent PAR-6 and PKC-3 can accumulate at the cortex independently of PAR-3.

To test this hypothesis, we compared the distribution of PAR-6 and PKC-3 in par-3(it71) embryos with or without RNAi-mediated depletion of CDC-37(Fig. 5). As previously reported (Hung and Kemphues,1999; Tabuse et al.,1998) in cdc-37(+); par-3(it71) embryos, PAR-6 and PKC-3 are not detectable at the cortex at any stage in the cell cycle(Fig. 5A,C; n=50). However, in cdc-37(RNAi); par-3(it71) embryos, PAR-6 and PKC-3 accumulate at the cortex but are not asymmetrically localized(Fig. 5B,D; n=42). Thus, cortical accumulation of PAR-6 and PKC-3 is dependent upon PAR-3 only in the presence of normal levels of CDC-37.

To determine whether CDC-37 acts preferentially through PAR-6 or PKC-3, we depleted CDC-37 in par-6(zu222) worms(Fig. 5G), and doubly depleted PKC-3 and CDC-37 in wild-type worms (Fig. 5H), and assayed the remaining anterior complex proteins by immunofluorescence microscopy. As previously shown by Hung and Kemphues, in par-6(zu222) worms, PKC-3 is not detectable at the cortex(Hung and Kemphues, 1999)(Fig. 5G; n=21). PKC-3 is also absent from the cortex in cdc-37(RNAi); par-6(zu222) embryos at all stages of the first cell cycle (Fig. 5H; n=25). As shown by Tabuse and colleagues, in pkc-3(RNAi) embryos, PAR-6 is not detectable at the cortex(Tabuse et al., 1998)(Fig. 5E; n=102). However, after double RNAi of PKC-3 and CDC-37, PAR-6 is present symmetrically at the cortex of one-cell embryos at all stages of the first cell cycle(Fig. 5F; n=170). Thus, in a cdc-37(RNAi) background, PAR-6 is required for PKC-3 localization to the cortex, but PKC-3 is not required for PAR-6 cortical localization. We interpret these results to mean that CDC-37 is required to prevent PAR-6 from associating with the cortex, and that PAR-3 normally acts to block the action of CDC-37.

We also examined the distribution of PAR-3 in these doubly depleted embryos. We found, as expected, that PAR-3 becomes partially restricted to the anterior cortex in par-6 mutant embryos, pkc-3(RNAi)embryos, cdc-37(RNAi) embryos, pkc-3(RNAi); cdc-37(RNAi) and par-6(zu222); cdc-37(RNAi) embryos (data not shown). However, we noted that removal of PAR-6 or depletion of PKC-3 in cdc-37(RNAi)embryos led to a slight increase in the levels of cortical PAR-3 (data not shown).

As described above, cdc-37(RNAi) allows constitutive cortical accumulation of PAR-6. If this reflects a role for CDC-37 in the normal mechanism for PAR-6 cortical accumulation and asymmetry, we would expect this cortical accumulation to be dependent upon the Rho-GTPase CDC-42. Binding to CDC-42 is required for the proper accumulation and maintenance of PAR-6 at the cortex of the one-cell embryo (Gotta et al., 2001; Kay and Hunter,2001) (D. Aceto, M.B. and K.K., unpublished). In cdc-37(RNAi) embryos, GFP::PAR-6 is evenly distributed around the entire cortex (Fig. 4, column 4). By contrast, in cdc-42(RNAi) embryos, a reduced amount of GFP::PAR-6 accumulates at the cortex early in the first cell cycle and becomes partially restricted to the anterior, but is undetectable after metaphase(Fig. 6A,B; D. Aceto, M.B. and K.K., unpublished). We found that depleting both CDC-42 and CDC-37 by RNAi results in a phenotype that differs from either single RNAi. In cdc-42(RNAi); cdc-37(RNAi) embryos, GFP::PAR-6 does not accumulate at the cortex (7/11 embryos; Fig. 6C,D). Thus, CDC-42 is required for the constitutive cortical accumulation of PAR-6 in cdc-37(RNAi) embryos, and CDC-37 is required for the residual PAR-6 in cdc-42(RNAi) embryos.

The similarity between the behavior of PAR-3 in cdc-37(RNAi) and the residual PAR-6 in cdc-42(RNAi) led us to wonder whether most or all of the residual cortical PAR-6 in cdc-42(RNAi) embryos might be associated with PAR-3. To test this, we examined the extent of overlap of PAR-3 and PAR-6 in cortical puncta in cdc-42(RNAi) embryos (see Materials and methods). Cortical sampling of uninjected control GFP::PAR-6 embryos stained with antibodies specific to GFP and PAR-3 showed that 65% of the PAR-6 puncta overlap with PAR-3 in the wild type(Fig. 7A-C; see also Table S1 in the supplementary material). By contrast, cortical sampling in GFP::PAR-6 embryos treated with cdc-42(RNAi) showed that 90% of PAR-6 puncta overlap with PAR-3 (Fig. 7D-F;see Table S1 in the supplementary material). This extent of overlap is similar to that seen previously in control experiments using two different anti-PAR-3 antibodies (Hung and Kemphues,1999), and thus is consistent with the hypothesis that most or even all cortical PAR-6 in cdc-42(RNAi) embryos is associated with PAR-3. The proportion of puncta that contained only PAR-3 did not show a statistically significant difference between control and cdc-42(RNAi)embryos (see Table S1 in the supplementary material). These results suggest that there are two sites for PAR-6 cortical association, one dependent on CDC-42, and the other independent of CDC-42 and colocalizing with PAR-3.

CDC-37-depleted embryos exhibit altered cortical behaviors in the zygote. To understand the basis for this, we examined the distribution of NMY-2, the heavy chain of embryonic cytoplasmic myosin(Guo and Kemphues, 1996). Just after the completion of meiosis in wild-type embryos, NMY-2::GFP is found in distinct foci and filaments throughout the entire cortex. In response to the polarization signal, the cortical cytoskeleton, including the NMY-2::GFP foci,flows into the anterior (Munro et al.,2004). To determine the effect of knockdown of CDC-37 on NMY-2 localization, we examined cdc-37(RNAi) and control embryos expressing NMY-2::GFP. We found that, in cdc-37(RNAi) embryos, NMY-2::GFP forms foci and filaments like wild-type controls do. The foci flow into the anterior, clearing approximately 40% of the posterior cortex in contrast to the 50% clearing observed in control embryos. Based on descriptions by Munro and colleagues (Munro et al.,2004), this behaviour is similar to that seen when anterior PAR proteins are reduced or eliminated. As the two pronuclei migrate to the center of the embryos, the large NMY-2 foci in the anterior are replaced by fine fibers, as occurs in controls; however, in contrast to controls, asymmetry is lost. Cortical myosin drops to a level normally seen in the posterior (see Fig. S3 in the supplementary material).

Previous work had shown that GFP::CDC-37 exhibited a distribution typical of free cytosolic proteins (Gunsalus et al., 2005). We independently constructed a GFP::CDC-37 translational fusion expressed under the control of the pie-1promoter and generated transgenic lines by biolistic bombardment(Praitis et al., 2001). We recovered nine independent lines showing similar GFP expression in the embryo. The Ppie-1::gfp::cdc-37 transgene suppresses the lethality of cdc-37(RNAi), indicating that the transgene is functional (see Materials and methods). As previously reported, from meiosis until NEB,GFP::CDC-37 is evenly distributed throughout the cytoplasm in a pattern consistent with free cytosolic proteins(Fig. 8A-C). In our strains, at NEB, GFP::CDC-37 remains predominantly cytoplasmic but becomes transiently enriched around the chromosomes (Fig. 8D). As mitosis progresses, CDC-37 becomes slightly enriched on the spindle (Fig. 8E). Upon nuclear envelope reformation, CDC-37 is excluded from the nucleus(Fig. 8F).

Because of the known role of CDC37 proteins as co-chaperones, we investigated whether CDC-37 depletion might influence the level of any of the PAR proteins in the early embryo. In two independent RNAi experiments, the steady state levels of PAR-1, PAR-3, PAR-5 and PAR-6 proteins were indistinguishable between control and cdc-37(RNAi) embryo extracts by western analysis.

PKC-3 and PAR-4 protein levels, however, were affected by cdc-37(RNAi). PKC-3 levels were reduced in two out of three experiments by more than 80% (Fig. 9A). The effect on PAR-4 was less pronounced. When wild-type embryonic proteins were separated by gel electrophoresis and probed with a PAR-4 antibody, two predominant bands were observed: a faster migrating band representing about 90% of the antibody-reactive protein and a slower band(Fig. 9B). The slower migrating band is a phosphorylated form of the protein, as this band is eliminated by treatment of the sample with phosphatase (data not shown). Several independent samples of cdc-37(RNAi) embryos have shown a consistent reduction of the slower migrating PAR-4 band.

We have found that a homolog of yeast and vertebrate CDC37 proteins is required for embryonic polarity in C. elegans. Based on RNAi phenotypes, C. elegans CDC-37 is required for two major activities in the establishment of polarity. CDC-37 is required for the mutual exclusion of PAR-2 and the anterior PAR proteins, and for normal interactions among the anterior PAR proteins. We propose that the RNAi phenotypes arise in part because CDC-37 activity is required to maintain PKC-3 levels, and in part because CDC-37 is required for the proper dynamic association of PAR-6 with two different cortical-binding sites.

cdc-37(RNAi) embryos exhibit polarity-defective phenotypes,including altered cortical dynamics, equal first cleavage and failure to localize MEX-5 and P granules. These phenotypes apparently arise through the effect of CDC-37 on the distribution of the PAR proteins. It is probable that the embryonic phenotypes we describe do not represent the null condition, as at the time we sample embryos some CDC-37 protein remains.

In wild-type embryos, stable cortical localization of the anterior PAR proteins, PAR-3, PAR-6 and PKC-3, requires the proper function of all three proteins (Hung and Kemphues,1999; Tabuse et al.,1998; Watts et al.,1996). PAR-6 and PKC-3 are absolutely dependent upon PAR-3(Hung and Kemphues, 1999; Tabuse et al., 1998), whereas PAR-3 exhibits only a partial dependence on the other two proteins. The reduced PAR-3 asymmetry, its patchy distribution and its disappearance at metaphase in cdc-37(RNAi) embryos is identical to what is observed in par-6(zu222) and pkc-3(RNAi) embryos(Tabuse et al., 1998; Watts et al., 1996). Indeed,PAR-3 distribution and dynamics are identical in cdc-37(RNAi)embryos, par-6(zu222); cdc-37(RNAi) embryos and pkc-3(RNAi);cdc-37(RNAi) embryos. A simple explanation for the effect on PAR-3 distribution and dynamics comes from the observation that PKC-3 levels are consistently reduced after cdc-37(RNAi) (see below). Reduced levels of PKC-3 can also account for the failure of PAR-2 to become restricted to the posterior, as exclusion of PAR-2 from the anterior requires the phosphorylation of PAR-2 by PKC-3 (Hao et al., 2006). Similar arguments may also hold for PAR-1(Hao et al., 2006). NMY-2::GFP behaviour in cdc-37(RNAi) embryos can also be explained by reduced levels of PKC-3.

A reduction of PKC-3 cannot explain the entire cdc-37(RNAi)phenotype, however, as pkc-3(RNAi) and cdc-37(RNAi)phenotypes differ. In contrast to the failure of PAR-6 to accumulate cortically in pkc-3(RNAi) embryos(Tabuse et al., 1998), PAR-6 and the residual PKC-3 remain at the cortex throughout the cell cycle in cdc-37(RNAi) embryos, and do so even in the absence of PAR-3. Our epistasis analysis reveals that PAR-3-independent cortical localization of PKC-3 requires PAR-6, but that PAR-6 does not require PKC-3. Thus, reduction of CDC-37 renders PAR-6 and PKC-3 independent of PAR-3 for cortical localization, and renders PAR-6 independent of PKC-3 for its cortical localization.

We interpret these results to mean that CDC-37 activity normally influences the dynamics of the anterior complex proteins. We propose that CDC-37 and one or more of its client protein(s) function in wild-type cells to block the binding of PAR-6 to an unknown cortically associated protein. We further propose that binding to this cortical protein is part of the normal mechanism by which PAR-6 and PKC-3 localize to the anterior cortex. In wild-type cells,PAR-3 acts locally to antagonize the action of the CDC-37 client and allows PAR-6 to bind only in the anterior. In wild-type cells, this binding also requires PKC-3, perhaps because PAR-3 exerts its blocking effect via PKC-3.

It is also possible that the PAR-3-independent, cortical-binding site for PAR-6 in cdc-37(RNAi) embryos is spurious and has no normal role in anterior PAR protein function. Although we cannot rule this out, two observations make this unlikely. First, in keeping with previous immunolocalization results (Hung and Kemphues, 1999), we found that in wild-type embryos a substantial fraction of cortical PAR-6 does not colocalize in puncta with PAR-3. Second,the accumulation of PAR-6 at the cortex in cdc-37(RNAi) embryos is dependent upon CDC-42, just as it is in cdc-37(+) embryos. We also found that, in cdc-42(RNAi) embryos, all or most of the residual PAR-6-containing puncta also contain PAR-3. A further complexity of our results is that double depletion of CDC-37 and CDC-42 results in a complete absence of GFP::PAR-6 at the cortex, a phenotype that differs from that resulting from depletion of CDC-42 alone. This indicates that CDC-37 is required for the residual GFP::PAR-6 puncta in cdc-42(RNAi) embryos,suggesting that these puncta are not the result of partial depletion of CDC-42, but rather represent a different class of PAR-6-binding site.

We infer from these results that there are two different ways PAR-6 can localize to the cortex, and that both are mediated by CDC-37(Fig. 10). The first mechanism is through cdc-42-independent binding to a PAR-3-containing complex,perhaps by direct binding to PAR-3. The second mechanism, mentioned above, is CDC-42 dependent, and is blocked by the action of CDC-37 and involves an unknown cortical-binding site. This site could be CDC-42 itself, or possibly a PDZ domain ligand whose binding to PAR-6 is regulated by CDC-42. This latter possibility is consistent with biochemical and structural studies of mammalian and fly PAR-6 proteins (Peterson et al.,2004). In these studies, the PAR-6 PDZ domain was found to be able to bind to a carboxy-terminal-type synthetic ligand in a CDC-42-regulated manner. PAR-6 was also found to bind weakly to the N-terminal PDZ domain of Bazooka (PAR-3) in a CDC-42-independent manner. Furthermore, the peptide ligand did not compete with PAR-3 for binding(Peterson et al., 2004). We also infer from our results and previously published results the existence of a class of PAR-3-containing puncta that do not contain PAR-6, and that in the early part of the cell cycle are independent of CDC-42, CDC-37, PKC-3 and PAR-6 for their cortical localization.

Our results contribute to a growing body of evidence that supports a dynamic regulation of anterior PAR protein distribution, as opposed to the simple concept of an obligate complex suggested by early biochemical and co-localization results. PAR-6 can bind to a variety of partners including Lethal Giant Larva (Betschinger et al.,2003; Plant et al.,2003; Yamanaka et al.,2003), Crumbs (Kempkens et al., 2006; Lemmers et al.,2004) and Pals1 (Hurd et al.,2003; Wang et al.,2004). Distinct distributions for Bazooka (PAR-3) versus PAR-6 and aPKC have been described in Drosophila epithelial cells and photoreceptors (Harris and Peifer,2005; Nam and Choi,2003), and Par3-independent cortical recruitment of Par6 and aPKC occurs in mammalian astrocytes(Etienne-Manneville and Hall,2001). It will be interesting to learn to what extent the mechanisms for regulating these complexes are evolutionarily conserved.

CDC-37 is known to function as a protein chaperone in many systems,regulating the stability and function of many proteins, especially kinases(MacLean and Picard, 2003). We found that the steady-state level of PKC-3 is affected by depletion of CDC-37. Cdc37 was identified among proteins complexed with aPKC(Brajenovic et al., 2004), and thus may directly regulate its stability and, perhaps, its activity.

Biochemical support for our genetic evidence that another kinase (or kinases) might also be a client of CDC-37 comes from the observation that cdc-37(RNAi) consistently reduced the level of a phosphorylated form of PAR-4. The effect on this minor PAR-4 isoform, however, cannot account for all of the polarity phenotypes of the cdc-37(RNAi) embryos, because these phenotypes differ substantially from those of par-4loss-of-function mutants (Watts et al.,2000). We considered that PKC-3 might be functioning to phosphorylate PAR-4, but the phosphorylated form of PAR-4 was unaffected by pkc-3(RNAi) (M.B. and K.K., unpublished). We also considered the possibility that the unique cdc-37(RNAi) phenotype might be revealing a redundant function among the three known polarity kinases PAR-1, PAR-4 and PKC-3. However, doubly or triply depleted embryos (e.g. par-4(it57);par-1(RNAi); pkc-3(RNAi)) did not replicate the cdc-37(RNAi)phenotype (M.B. and K.K., unpublished). We also ruled out the possibility that the phenotype was a gain-of-function defect due to an imbalance between PAR-4 isoforms, by showing that cdc-37(RNAi); par-4(it57) embryos had early polarity defects identical to those of cdc-37(RNAi) alone (M.B. and K.K., unpublished).

We have shown that C. elegans CDC-37 has a role in the establishment of polarity in the one-cell embryo. We also present evidence that we interpret as indicating the existence of two modes of binding for PAR-6 at the cortex: CDC-42-dependent binding at a site that is normally blocked by the action of a CDC-37 client; and CDC-42-independent binding at a site that co-localizes with PAR-3 and that is also dependent upon a CDC-37 client. We propose that, in wild-type embryos, CDC-37-mediated inhibition of the CDC-42-dependent binding site and PAR-3-mediated release of this inhibition provide a key mechanism for the anterior accumulation of PAR-6. Identifying the ligand or ligands for cortical binding, and the relevant client or clients of CDC-37 that regulate this binding will advance our understanding of the mechanism of action of the anterior PAR proteins.

We thank Wendy Hoose and Mona Hassab for technical support, and Diane Morton, Kelly Liu and Sylvia Lee for helpful comments on the manuscript. We thank Didier Picard for yeast strains. M.B. was supported in part by NIGMS Predoctoral Training Grant GM007617. This work was funded by NICHD grant HD27689 and NIGMS grant GM079112 to K.K.