BINDING TO PKC-3, BUT NOT TO PAR-3 OR TO A CONVENTIONAL PDZ DOMAIN LIGAND, IS REQUIRED FOR PAR-6 FUNCTION IN C. ELEGANS

Abstract

PAR-6 is a conserved protein important for establishment and maintenance of cell polarity in a variety of metazoans. PAR-6 proteins function together with PAR-3, aPKC and CDC-42. Mechanistic details of their interactions, however, are not fully understood. We studied the biochemical interactions between C. elegans PAR-6 and its binding partners and tested the requirements of these interactions in living worms. We show that PB1 domain-mediated binding of PAR-6 to PKC-3 is necessary for polarity establishment and PAR-6 cortical localization in C. elegans embryos. We also show that binding of PAR-6 and PAR-3 is mediated in vitro by a novel type of PDZ-PDZ interaction; the βC strand of PAR-6 PDZ binds the βD strand of PAR-3 PDZ1. However, this interaction is dispensable in vivo for PAR-6 function throughout the life of C. elegans. Mutations that specifically abolish conventional ligand binding to the PAR-6 PDZ domain also failed to affect PAR-6 function in vivo. We conclude that PAR-6 binding to PKC-3, but not to PAR-3 nor to a conventional PDZ ligand, is required for PAR-6 cortical localization and function in C. elegans.

INTRODUCTION

Proteins encoded by the anterior par (partitioning defective) genes, par-3, par-6 and atypical protein kinase C (pkc-3), are required in C. elegans for establishing embryonic polarity, for apicobasal polarization of non-epithelial early embryonic blastomeres and for proper function of polarized epithelial cells (Aono et al., 2004; Nance et al., 2003; Totong et al., 2007). PAR-6, PAR-3 and PKC-3 function together in early embryonic polarity; they have indistinguishable loss-of-function phenotypes, co-localize to the anterior cortex of the one-cell embryo, and are co-dependent for this distribution (Etemad-Moghadam et al., 1995; Hung and Kemphues, 1999; Tabuse et al., 1998; Watts et al., 1996).

Proteins homologous to the anterior PAR proteins are conserved among metazoans, and play similar roles in a variety of polarized cells (Goldstein and Macara, 2007). For example, in Drosophila, Bazooka (PAR-3), Par-6 and atypical protein kinase C (aPKC) are essential for establishing oocyte polarity and maintaining oocyte cell fate (Cox et al., 2001; Huynh et al., 2001). During Drosophila embryogenesis, they are also important for apical-basal polarity of embryonic epithelial cells and asymmetric cell division of neuroblasts (Kuchinke et al., 1998; Muller and Wieschaus, 1996; Petronczki and Knoblich, 2001; Wodarz et al., 2000). In mammalian epithelial cells, Par6, Par3/ASIP and aPKC play important roles in tight junction (TJ) formation and apical-basal polarity establishment (Shin et al., 2006; Suzuki and Ohno, 2006). PAR-6, PAR-3 and aPKC show extensive co-localization and are interdependent for their asymmetric distribution not only in C. elegans embryos but also in Drosophila epithelial cells and neuroblasts and in mammalian cells (Izumi et al., 1998; Lin et al., 2000; Petronczki and Knoblich, 2001; Suzuki et al., 2001).

PAR-6 and its homologues play key roles by serving as scaffolds that organize several other proteins (Par3, aPKC, Cdc42, Pals1/Stardust, Crumbs/CRB3, Lgl) or regulate their functions or both. Drosophila and mammalian homologues of PAR-6 and PKC-3(aPKC) bind directly, forming PB1 domain heterodimers (Hirano et al., 2005; Lin et al., 2000; Noda et al., 2003; Suzuki et al., 2001; Wilson et al., 2003; Yamanaka et al., 2001) and in Drosophila neuroblasts this interaction is regulated in a cell-cycle dependent manner (Wirtz-Peitz et al., 2008). The PDZ domain of mammalian and Drosophila Par-6 interacts with Par-3 PDZ1 (Joberty et al., 2000; Lin et al., 2000; Peterson et al., 2004). The well-studied polarity regulator CDC-42 also binds directly to PAR-6 and is required for PAR-6 function in C. elegans (Aceto et al., 2006; Gotta et al., 2001; Kay and Hunter, 2001; Schonegg and Hyman, 2006) and other animals (Etienne-Manneville and Hall, 2001; Hutterer et al., 2004; Joberty et al., 2000; Johansson et al., 2000; Lin et al., 2000; Qiu et al., 2000). A partial CRIB motif combines with the PDZ domain of Par-6 to bind Cdc42 (Garrard et al., 2003; Joberty et al., 2000; Johansson et al., 2000; Lin et al., 2000; Qiu et al., 2000). The Par-6 PDZ domain can also bind ligands through its hydrophobic binding pocket. The transmembrane ligand Crumbs/Crb3 binds to the Par-6 PDZ hydrophobic pocket through the Crumbs C-terminus in a Cdc42-dependent fashion (Kempkens et al., 2006; Lemmers et al., 2004; Peterson et al., 2004) whereas the Pals1/Stardust protein binds the Par-6 PDZ pocket through an internal portion of the protein in a Cdc42-independent fashion (Penkert et al., 2004; Peterson et al., 2004; Wang et al., 2004); although see (Hurd et al., 2003) for evidence for dependence on Cdc42. Lgl/Mlgl is another binding partner whose precise mechanism of interaction is unclear but could also involve binding to the PDZ domain of PAR-6 (Betschinger et al., 2003; Plant et al., 2003).

In C. elegans, the localization of PAR-6 is dependent upon PAR-3, PKC-3 and CDC-42. In par-3 mutants and pkc-3(RNAi) embryos, PAR-6 is absent from the cell cortex of early blastomeres (Hung and Kemphues, 1999; Tabuse et al., 1998; Watts et al., 1996). In cdc-42(RNAi) embryos and in embryos of par-6 mutants that block interaction with CDC-42, PAR-6 appears punctate or is undetectable at the cell cortex (Aceto et al., 2006; Gotta et al., 2001; Kay and Hunter, 2001; Schonegg and Hyman, 2006). Genetic analysis and co-localization results indicate that there are at least two modes by which PAR-6 can localize at the cortex, one that is CDC-42 dependent and one that is independent of CDC-42 (Beers and Kemphues, 2006; Hung and Kemphues, 1999); although see (Schonegg and Hyman, 2006) for evidence suggesting complete dependence on CDC-42.

Although binding partners of mammalian and Drosophila Par6 have been identified, much remains unknown about the mechanisms through which Par6 interacts with these partners and the consequences of these interactions in vivo. Furthermore, little has been reported about these interactions in C. elegans. To understand better how these interactions relate to the localization and function of PAR-6, we examined the biochemical interactions of C. elegans PAR-6 with PAR-3, PKC-3 and a heterologous PDZ domain ligand, Pals1, and investigated the function of these interactions in vivo with mutated par-6 transgenes that specifically block these interactions in vitro.

Here we report that, consistent with results from mammals and flies, C. elegans PAR-6 and PKC-3 associate through their PB1 domains; PAR-6 PDZ can bind to PAR-3 PDZ1; and the three proteins can exist as a protein complex in vivo. We also found that the PDZ-PDZ interaction between PAR-3 and PAR-6 occurs via a novel type of binding. By studying transgenic lines expressing mutated GFP–tagged PAR-6, we learned that the interaction between PAR-6 and PKC-3 is necessary for polarity establishment and PAR-6 cortical localization in the early embryo. Surprisingly, however, disrupting either the PAR-6 PDZ interaction with PAR-3 PDZ1 or disrupting the ability of the PAR-6 PDZ to bind a PALS-1-like ligand had little or no effect on PAR-6 function in vivo.

MATERIALS AND METHODS

Yeast two-hybrid screen and assays

Full-length par-6 cDNA was cloned into the pAS1-CYH2 vector (Bai and Elledge, 1996). Yeast transformants carrying pASPAR-6 were tested for PAR-6 expression by Western blot. A mixed-staged C. elegans library constructed in the pACT2 vector (a gift from Dr. Bob Barsted) was transformed into pASPAR-6 yeast according to the protocol of (Bai and Elledge, 1996). Approximately 600,000 yeast transformants were plated on selective medium [SC-Trp, Leu, His+ 50mM 3-amino 1,2,4 triazole (3-AT)]. The His+ clones were plated on selective medium and scored for β-galactosidase (β-gal) activity. The positive clones were identified by their dependence on PAR-6 to activate HIS3 and β-gal expression and were sequenced. To assay specific protein/protein interactions, full-length and fragments of par-6, pkc-3 and par-3 were cloned into pAS1-CYH2 and pACTII vectors (Bai and Elledge, 1996; Durfee et al., 1993; Harper et al., 1993) for yeast two-hybrid assays.

Binding assays with GST fusion proteins

For in vivo pull-down experiments, embryos were harvested from gravid worms (Etemad-Moghadam et al., 1995). Embryos were resuspended in C buffer (40 mM HEPES pH7.4, 150 mM NaCl, 10% glycerol, 1 mM EDTA, protease inhibitor cocktail from Roche) and proteins extracted by sonication. The embryo protein homogenate was centrifuged at 100,000 X g for 45 minutes. GST-PAR-3^245-932 or GST was bound to glutathione beads. Embryo extracts were mixed with 20μl of GST-PAR-3^245-932 or GST-bound beads, incubated at 4°C for 2 hrs, and subsequently washed six times with C buffer. Bound proteins were extracted using SDS sample buffer, and then subjected to SDS-PAGE and Western blotting.

For in vitro binding assays, full-length par-6 cDNA and fragments of the gene were cloned into the pQE32 (Qiagen) or pGEX-4T-1 vectors (Pharmacia Biotech). PKC-3 fragments were cloned into the pGEX-5X-1 vector (Pharmacia Biotech). PAR-3 fragments were cloned into the pMAL-c2 vector (NEB). Recombinant proteins were produced in E. coli BL21. 6His or MBP fusion protein-containing bacterial pellets were sonicated in binding buffer (1xPBS, 0.1% Triton X-100). After ultracentrifugation, the supernatant was incubated with GST fusion proteins immobilized on Glutathione beads for 2 hours at 4°C and washed with binding buffer six times. The beads were boiled with 2X SDS sample buffer. Eluted proteins were separated by SDS-PAGE and detected by Gelcode Blue staining (Pierce) or transferred to nitrocellulose membranes and detected by immunoblotting (Burnette, 1981) using rabbit antibodies to PAR-6 (1:1000) (Hung and Kemphues, 1999), rabbit antibodies to MBP tag (1:10,000) (NEB), HRP-conjugated goat anti-rabbit antibodies (1:10,000) (Jackson ImmunoResearch) and chemiluminescent reagents (Amersham Biosciences).

To test the interaction between PAR-6 and Pals1 or CDC-42, in vitro binding assays were carried out according to the methods of (Peterson et al., 2004).

Co-immunoprecipitation

Embryo protein extracts prepared as described above were pre-cleared by treatment with Protein A beads lacking antibody and then subjected to immunoprecipitation using beads bound with affinity-purified anti-PAR-6, anti-PAR-3, or anti-ZYG-9 (control) antibodies (Etemad-Moghadam et al., 1995; Hung and Kemphues, 1999; Matthews et al., 1998). After immunoprecipitation, Protein A beads were washed (40 mM HEPES pH7.4, 150 mM NaCl, 10% glycerol, 1 mM EDTA, 1% Triton-X 100), and eluted with 0.1M glycine, pH 2.8. The elutions were precipitated with Trichloroacetic acid (TCA) at a final concentration of 20%. The precipitates were washed once with ice-cold acetone, dried, boiled in SDS sample buffer and subjected to SDS-PAGE and Western blotting.

Production of transgenic lines

For analysis of the PAR-6-PKC-3 interaction in the early embryo, GFP::PAR-6 and GFP::PAR-6^Δ15-28 lines were generated according to the complex array method (Kelly et al., 1997; Mello et al., 1991). par-6 and par-6 ^Δ15-28 cDNAs cloned into pJAM (Aceto et al., 2006) were linearized and co-injected with PvuII-cut genomic DNA into the gonads of young adult KK818 par-6(zu222) unc-101(m1)/hIn1[unc-54(h1040)]I worms. Lines producing rolling progeny were tested for GFP expression and ability to rescue par-6(zu222) homozygous segregants in the F2 and F3 generations. Most lines generated using this transgenesis method underwent germline silencing within a few generations.

To analyze the effect of PAR-6 PDZ mutations on the maternal requirement for PAR-6 function in vivo, we generated transgenic lines carrying WT or mutated gfp::par-6 cDNA transgenes driven by the pie-1 promoter in the vector pAJS100 (pJunc) (Beers and Kemphues, 2006) using biolistic bombardment (Praitis et al., 2001) of unc-119(ed3); par-6(zu222)unc-101(m1)/hIn1[unc-54(h1040)]. To test the maternal effect of a particular mutation on embryonic viability, eggs laid by par-6(zu222) unc-101 homozygotes were monitored for hatch rate.

To test the effect of PAR-6 mutants on the zygotic requirement for PAR-6, we used biolistic bombardment to generate transgenic lines expressing GFP-tagged wild-type or mutated genomic par-6 under the control of its own promoter. Mutations were generated by fusion PCR using pJN284 as template (Nance et al., 2003). We bombarded par-6::gfp transgenes directly into KK1002, par-6(tm1425)/hIn1[unc-54(h1040)]; unc-119(ed3). In a typical bombardment, of about 30 non Unc-119 lines recovered, about 10 behaved as homozygous viable chromosomal integrations and three to five expressed GFP. Progeny from par-6(tm1425)/hIn1(unc-54) heterozygotes segregate three genotypes: par-6(tm1425)/par-6(tm1425), which arrest as L2 larvae, par-6(tm1425)/hIn1(unc-54), which are phenotypically wild-type, and hIn1(unc-54)/hIn1(unc-54), which are paralyzed Unc. Rescue was scored as the ability to recover individual wild-type-looking fertile worms that failed to segregate Unc-54 progeny. To verify the presumed genotype of the rescued lines, we collected fertile wild-type appearing adults from one line per each transgene, and genotyped them by single-worm PCR using primer sets that could detect the tm1425 deletion, the rescuing par-6::gfp transgene, and the par-6(+) gene in the balancer.

Microscopy and Immunofluorescence

To visualize GFP signals in living embryos, embryos were dissected out of gravid adult worms in distilled H₂O, mounted on agar pads and imaged at approximately 23°C. Openlab software was used to make time-lapse movies on a Leica DM RA2 microscope equipped with a 63X Leica HCX PL APO oil immersion lens and a Hamamatsu ORCAER digital camera.

For confocal analysis, embryos were fixed in methanol following previously published procedures (Guo and Kemphues, 1995). The following primary antibodies and dilutions were used: anti-PAR-3 mouse monoclonal (Nance et al., 2003) at 1:20; anti-PAR-6 rabbit polyclonal (Hung and Kemphues, 1998) at 1:20. Incubation times and temperatures were as described by Nance et al. (2003). Primary antibodies were detected by Cyc3 labeled goat anti-mouse (Jackson ImmunoResearch Laboratories, Inc.) at 1:200 and Alexa Fluor 488 labeled goat anti-Rabbit (Invitrogen) at 1:200. Confocal images were collected on a Leica TCS SP2 system with a Leica DMRE-7 microscope and an HCX PL APO 63x oil immersion lens. Images were processed using the Leica Confocal SP2 software program and Adobe PhotoShop. To quantify degree of co-localization of PAR-3 and PAR-6, three to four embryos were analyzed for each genotype. Six cortical sections 0.25 microns apart were projected to obtain each image for analysis. For each embryo, two independent regions were analyzed. For each image, background, as defined by the cytoplasmic signal at the posterior cortex, was removed; remaining cortical puncta were analyzed for overlap.

RESULTS

PAR-6, PAR-3 and PKC-3 can form a protein complex in embryo extracts

Par-3, Par-6 and atypical protein kinase C (aPKC) form complexes in mammalian cells and Drosophila embryos. To determine whether this is also true in C. elegans embryos, we tested whether the proteins could be co-purified from embryo extracts. Extracts were prepared from embryos collected from young gravid adults; such embryos vary in developmental age but are enriched for stages prior to gastrulation.

We first examined whether endogenous PKC-3 and PAR-6 proteins in embryo extracts can bind to GST-PAR-3^245-932, which contains all three PDZ domains and the putative PKC-3 binding/phosphorylation domain (Etemad-Moghadam et al., 1995). Indeed, as shown in figure 1A, PKC-3 and PAR-6 can be co-purified with GST-PAR-3^245-932 but not with GST alone, that PAR-6 and PKC-3 can associate with PAR-3 in embryo extracts.

Figure 1.

PAR-3, PAR-6 and PKC-3 form complexes in embryo extracts. A) Western blot showing co-purification of PKC-3 (upper band) and PAR-6 (lower band) with GST-PAR-3^245-932 from embryo extracts. B) Western blot showing co-purification of PKC-3 and PAR-6 from embryo extracts after immunoprecipitation (IP) with anti-PAR-3. Anti-ZYG-9 antibody was used for the control IP in panels B, C and D. C) Western blot showing co-purification of PAR-3 and PKC-3 after IP with anti-PAR-6. D) Gelcode Blue-stained poly-acrylamide gel showing results of IP with anti-PAR-6 antibodies. Note the absence of detectable PAR-3 protein (asterisk). ZYG-9, the control IP, migrates at a position similar to PAR-3.

To further explore the potential interaction between PAR-3, PAR-6 and PKC-3 in vivo, we examined whether PAR-6 and PKC-3 could be co-immunoprecipitated from embro extracts with PAR-3 (See Materials and Methods). As indicated in figure 1B, PAR-6 and PKC-3 co-precipitated with PAR-3 using anti-PAR-3 antibody but not control anti-ZYG-9 antibody, demonstrating that PAR-3 can associate with PAR-6 and PKC-3 in vivo.

In a reciprocal co-immunoprecipitation experiment using anti-PAR-6 antibodies, PAR-3 and PKC-3 were also detected in the IP complex by immunoblotting (Figure 1C). However, by protein staining, we could detect both PAR-6 and PKC-3 but could not detect PAR-3 (Figure 1D). This result suggests that there is a strong interaction between PAR-6 and PKC-3 in vivo; however, the interaction of these proteins with PAR-3 may be weaker or more dynamic. Alternatively, only a small fraction of PAR-3 may strongly associate with PAR-6 in vivo.

PAR-6 and PKC-3 bind to each other through their PB1 domains

To identify binding partners of PAR-6, we undertook a two-hybrid screen of a C. elegans mixed-stage cDNA library using full-length PAR-6 as bait. We recovered six positive clones out of 6 ×10⁵ transformants, five of which encode PKC-3.

To determine the precise nature of the interaction between PAR-6 and PKC-3 we divided the PKC-3 coding sequences in two: an N-terminal portion encoding the regulatory domain (amino acids 1-257) and the reciprocal C-terminal portion encoding the catalytic domain (amino acids 258-597). Like other PKCs, the N-terminal portion of PKC-3 contains a pseudosubstrate sequence that is proposed to act as a regulator of its kinase activity (Wu et al., 1998). In yeast two-hybrid assays, we found that full-length PAR-6 was able to interact with PKC-3^1-257 but not PKC-3^258-597 (Figure 2A) and by both yeast two-hybrid (Figure 2A) and by direct binding (Figure 2B) that this interaction was mediated by the N-terminal half of PAR-6 (amino acids 1-138).

Figure 2.

PKC-3 and PAR-6 interact via their N-terminal domains. A) Interaction between PAR-6 NT (amino acids 1-138) and PKC-3 NT (amino acids 1-257) by yeast two-hybrid assay (A) and by direct binding of bacterial fusion proteins (B). Panel B shows a western blot probed with anti-PAR-6 antibody.

Analysis of mammalian Par-6 and aPKC revealed that both proteins contain a conserved PB1 domain at their N-termini and that these domains bind to each other (Hirano et al., 2005; Noda et al., 2003; Wilson et al., 2003). We confirmed this interaction for the C. elegans proteins using yeast two-hybrid and in vitro binding assays; a PAR-6 fragment containing the PB1 domain (amino acids: 1-126) interacts with a fragment of PKC-3 containing its PB1 domain (amino acids:18-95) whereas fragments of PAR-6 or PKC-3 deleted for all or part of either of the PB1 domains did not interact (Figure 3A and data not shown). Thus, similar to their mammalian homologues, PAR-6 and PKC-3 appear to interact with each other in vitro through their PB1 domains.

Figure 3.

The PB1 domain of PAR-6 is essential for binding to PKC in vitro and for function in vivo. A) Diagram showing deleted region of PAR-6 (amino acids 15-28) within the PB1 domain. B) Yeast two-hybrid assays showing interaction of wild-type PAR-6 or PAR-6^Δ15-28 with PKC-3, PAR-3 or constitutively active CDC-42. PAR-6^Δ15-28 does not interact with PKC-3. C) Distribution in living embryos of GFP::PAR-6 (top) and GFP::PAR-6^Δ15-28 (bottom). The mutant protein fails to localize to the cell cortex. The transient localization of GFP::PAR-6^Δ15-28 to the nucleoplasm just after nuclear envelop breakdown, as seen in the second panel, also occurs with GFP::PAR-6.

Interaction of PAR-6 with PKC-3 is necessary for polarity establishment and PAR-6 cortical localization

To determine the importance of the interaction between PKC-3 and PAR-6 in living animals, we constructed a mutant form of PAR-6 unable to bind to PKC-3. We found that deleting the highly conserved amino acids 15-28 within the PAR-6 PB1 domain disrupted the interaction of PAR-6 with PKC-3 in yeast two-hybrid assays without preventing its interaction with PAR-3 or constitutively active CDC-42 (Figure 3A). In pkc-3(RNAi) embryos, PAR-6 is no longer localized to the cortex (Tabuse et al., 1998), raising the possibility that binding to PKC-3 mediates PAR-6 cortical localization. To test this possibility, we used DNA microinjection to generate lines of worms that expressed the gfp::par-6^Δ15-28 transgene under the control of the pie-1 promoter and determined the distribution of the mutant protein and its ability to rescue viability of embryos from par-6(zu222) mothers, which lack maternally provided PAR-6 (see Materials and Methods). Control par-6(zu222) mothers expressing a wild-type gfp::par-6 transgene produced 672/2053 viable embryos, whereas none of seven independent lines expressing the gfp::par-6^Δ15-28 transgene rescued the lethality or polarity defects of embryos from homozygous par-6(zu222) mothers (0/5225 embryos hatched).

PAR-6 normally accumulates at the anterior cortex of the C. elegans zygote. In both par-6(+)/par-6(zu222) and par-6(zu222) embryos we observed cytoplasmic accumulation of GFP::PAR-6^Δ15-28, but no cortical localization (figure 3B). This result indicates that PAR-6 interaction with PKC-3 is important for polarity establishment and PAR-6 cortical localization. However, because the pie-1 promoter only expresses maternally and the zu222 mutation is maternal-specific, this experiment does not address a possible role for PAR-6-PKC-3 interaction in late embryogenesis or larval development.

To test whether interaction of PAR-6 with PKC-3 is important during later development of C. elegans, we tested whether expression of PAR-6^Δ15-28::GFP from the par-6 endogenous promoter could rescue the putative par-6 null allele, tm1425 (Totong et al., 2007). The par-6(tm1425) allele is a 853bp deletion that spans the first exon to the second intron of T26E3.3a, which is the longer of the two isoforms of PAR-6. tm1425 homozygotes can proceed through embryogenesis due to the maternal load of PAR-6, but arrest as young larvae. Out of 29 lines transformed with par-6^WT::gfp, we recovered nine integrated lines, three of which expressed PAR-6^WT::GFP driven by the endogenous promoter. The par-6^WT::gfp transgene fully rescued homozygous tm1425 worms and exhibited tissue and subcellular distributions consistent with previous analyses (Totong et al., 2007). For par-6^Δ15-28::gfp we recovered 34 lines in two independent biolistic transformation experiments; only two lines integrated the transgene into the genome and neither expressed GFP. We examined ten non-integrated lines, and recovered seven lines that express some GFP. When we examined expression in these lines, we noted that GFP accumulated in the cytoplasm of various tissues including hypodermis, vulva, and cells near the pharynx. Surprisingly, very few worms showed accumulation of PAR-6^Δ15-28::GFP in either pharynx or intestine, where PAR-6^WT::GFP is readily detected. In those few worms, expression was mosaic with only one or two of the intestinal cells or the pharyngeal cells expressing GFP. In those cells the mutant protein accumulated in the cytoplasm, but not the apical cortex (Figure 4), consistent with our results from expression in the early embryo. We tested two lines with rare mosaic expression for rescue of the larval lethality in par-6(tm1425) and saw no rescue; however, this could be due to the mosaic expression of the par-6^Δ15-28::gfp transgene. Because we did not recover any integrated lines with PAR-6^Δ15-28::GFP expression and lines with extrachromosomal arrays show only rare and mosaic GFP expression in the tissues where PAR-6 is normally expressed, it is possible that expressing PAR-6^Δ15-28::GFP in late stage embryos or larvae is toxic and we were only able to recover lines with weak or mosaic expression. Alternatively, coding sequences deleted in creating the transgene may be essential for proper expression.

Figure 4.

Distribution of GFP::PAR-6 wild-type (WT) and GFP::PAR-6 mutant fusion proteins, as indicated, in adult pharynx and intestine, larval vulvae, and embryonic pharynx and intestine. Note the mosaic expression of GFP::PAR-6^Δ15-28 and its failure to localize apically.

PAR-6 interacts with PDZ1 of PAR-3 in vitro

Previous work in mammals and flies revealed a weak interaction between PDZ1 of Par3 and the PDZ domain of Par6 (Joberty et al., 2000; Lin et al., 2000; Peterson et al., 2004). C. elegans PAR-6 and PAR-3 have protein structures similar to their mammalian and Drosophila homologues. We first confirmed that C. elegans PAR-3 and PAR-6 could interact in yeast two-hybrid and in vitro binding assays. Full-length PAR-6 could interact in both assays with a fragment of PAR-3 containing its three PDZ domains (amino acids 383-754) and with a fragment containing PDZ1 (amino acids 383-516), but not with a PAR-3 N-terminal fragment (amino acids 1-383), fragments containing PDZ2 (amino acids 501-651 for yeast two-hybrid and 517-651 for in vitro assays) or PDZ3 (amino acids 627-724 for yeast two hybrid and 652-775 for in vitro assays) or a C-terminal fragment including PDZ3 and the remainder of the carboxy-terminus (amino acids 637-1397) (Figure 5A). As evidence for specificity of this interaction, we found that neither PAR-3 nor PAR-6 interact with LIN-7, another C. elegans PDZ containing protein (Kaech et al., 1998) (Figure 5A). We further confirmed an interaction between PAR-6 PDZ (amino acids 139-252) and PAR-3 PDZ1 (amino acids 383-516) in both yeast two-hybrid and in vitro binding assays (Figure 5B). This appears to be a relatively weak interaction. Under the same in vitro binding conditions both MBP::PAR-3 PDZ1 and MBP::mPals1 peptide (see below) can be purified from bacterial cell extracts using GST::PAR-6 PDZ. Although both can be detected by western blot, only MBP::mPals1 peptide is detectable by protein staining (Figure 6A, first lane).

Figure 5.

PAR-6 PDZ and PAR-3 PDZ1 mediate binding between the two proteins A) PAR-6 binds to PDZ1 of PAR-3. Two-hybrid assays are on the left. The western blot on the right shows detection of MBP-tagged PAR-3 fragments with anti-MBP antibody following binding to GST-PAR-6 on glutathione agarose beads. Coomassie Brilliant Blue staining of GST-PAR-6 in the reaction is shown below the western blot. B) PAR-3 PDZ1 binds to PAR-6 PDZ. Two hybrid assays are on the left. The western blot on the right shows detection of MBP-PAR-3 PDZ1 with anti-MBP antibody following binding to GST-tagged PAR-6 fragments, which are shown below the western blot.

Figure 6.

Binding of PAR-3 PDZ1 to PAR-6 PDZ is not likely to depend on binding through the hydrophobic pocket. A) Commassie Brilliant Blue stained gel showing co-purification of MBP-Pals1 (aa 30-42) expressed in bacteria with wild type bacterially expressed GST-PAR-6 PDZ domain but not with GST-PAR-6 PDZ domain containing a four amino acid substitution in the hydrophobic pocket. B) Western blot showing co-purification of MBP-PAR-3 PDZ1 with wild type and mutant forms of GST-PAR-6 PDZ. The lower panel shows Coomassie staining of GST-PAR-6 PDZ eluted from glutathione beads; the PAR-3 PDZ1 protein was not detectable by protein stain. C) GST pull-down assay for MBP fusions of wild type and mutant PAR-3 PDZ1. Middle panel is Coomassie stain, top and bottom are western blots with anti-MBP antibody. MBP-PAR-3 PDZ2 serves as negative control. D) Co-purification of GST-PAR-6 PDZ and GST-PAR-6 PDZ^{F192Y D198P} with MBP-PAR-3 PDZ1 and MBP-Pals1 peptide showing reduced binding to the mutated fusion protein as seen with anti-MBP (upper panel) but no effect on binding to the MBP-Pals1 peptide as seen with Commassie stain (lower panel).

PAR-6/PAR-3 interaction is not essential for PAR-6 function

PDZ domains are 80-100 amino acid regions that form a globular domain composed of five or six β strands (βA-F) and two α helices arranged into what has been described as an ‘up-and-down β barrel’ (Doyle et al., 1996; Hillier et al., 1999; Morais Cabral et al., 1996). The βB strand, the αB helix and the loop connecting the βA and βB strand generate a hydrophobic binding pocket, through which PDZ domains interact with their binding partners (Doyle et al., 1996; Morais Cabral et al., 1996; Penkert et al., 2004). The most common PDZ-binding partners are transmembrane proteins that insert their intracellular carboxyl terminus into the hydrophobic binding pocket of PDZ domains (Doyle et al., 1996; Hung and Sheng, 2002). In a few cases, internal protein segments can insert into the binding pocket. For example, the nNOS PDZ domain interacts with the syntrophin PDZ domain through a β finger which is a C-terminal extension of the nNOS PDZ (Hillier et al., 1999). The PDZ domain of Par6 can also bind an internal segment of the protein Pals1 (Penkert et al., 2004), but Pals1 does not have a β finger. To accommodate the internal peptide of Pals1, the Par6 PDZ domain is deformed (Penkert et al., 2004).

We investigated the mechanism by which PAR-6 PDZ (amino acids: 139-252) interacts with PAR-3 PDZ1 (amino acids: 383-516). To test whether PAR-3 PDZ could be binding to the peptide-binding pocket of PAR-6 PDZ, we first verified that C. elegans PAR-6 PDZ could bind to a known ligand. Because there is no known ligand for C. elegans PAR-6, we tested the ability of mammalian Pals1 (Hurd et al., 2003) to bind to the C. elegans PAR-6 PDZ domain in a GST pull-down assay. As shown in Figure 6A, MBP::mPals1(amino acids 30-42) binds strongly enough to the PAR-6 PDZ domain that we can detect co-purified protein by staining. Mutations in the carboxylate binding loop of mPar6 PDZ disrupted the interaction between mPar6 and Pals1 (Hurd et al., 2003). We mutated the corresponding residues (RPLG166-169AAAA) of C. elegans PAR-6 PDZ and found that this quadruple mutation abolished C. elegans PAR-6 interaction with mammalian Pals1 (Figure 6A). However, the mutation did not affect PAR-6 binding to PAR-3 in vitro (Figure 6B). Another mutation, M234W), altering the C. elegans PAR-6 PDZ domain deep within the hydrophobic binding pocket, weakens PAR-6 PDZ interaction with mammalian Pals1, and, like the quadruple mutant, does not affect PAR-6 to PAR-3 binding (data not shown). This suggests that in C. elegans, PAR-3 does not bind to the canonical hydrophobic binding pocket of PAR-6 PDZ. We also created mutations in the carboxylate binding loop of PAR-3 PDZ1 (KQLG395-398AAAA) and found no effect on interaction with PAR-6 (Figure 6B), suggesting that C. elegans PAR-6 does not bind to the canonical hydrophobic binding pocket of PAR-3 PDZ1.

To find the minimal regions that mediate the interaction between PAR-6 and PAR-3, we made a series of deletion constructs within MBP::PAR-3 PDZ1 and GST::PAR-6 PDZ and tested their ability to interact in in vitro in a GST pull-down assay. As summarized in supplemental Figure S1, we found that the βC strand of PAR-6 PDZ is necessary to interact with PAR-3 PDZ1 and the βD strand of PAR-3 PDZ1 is necessary to interact with PAR-6 PDZ (for example, Figure 6C, lane 2; Supplemental Figure S1). We then tested the βC strand of PAR-6 PDZ and the βD strand of PAR-3 PDZ1 in the same assay and determined that these fragments were sufficient to promote binding to the other PDZ domain (Supplemental Figure S1).

To determine how disrupting the interaction with PAR-3 affects PAR-6 function in C. elegans in vivo, we first identified mutations that specifically disrupt this interaction in vitro. Because the LIN-7 PDZ domain does not bind to PAR-3 PDZ1 (Figure 5A), we mutated two residues in the βC strand of PAR-6 PDZ to match the corresponding residues in the βC strand of LIN-7 PDZ (F192Y, D198P;(Kaech et al., 1998)). These mutations significantly reduced the binding between PAR-6 and PAR-3 (Figure 6D; upper panel). PAR-6 PDZ^{F192Y, D198P} can still interact with mammalian Pals1 (Figure 6D; lower panel), suggesting that the overall PDZ structure is not affected by these mutations. Mutations of three residues in the βD strand of PAR-3 PDZ1 (C445R, A448E, D450N) to match the corresponding residues in the βD strand of PAR-3 PDZ2 also significantly weakened the interaction between PAR-6 and PAR-3 (Figure 6C; right lane).

We introduced the GFP::PAR-6 PDZ^{F192Y, D198P} construct into worms lacking maternally provided PAR-6 to determine the in vivo effect of these mutations. Expression of the GFP::PAR-6^F192Y,D198P mutant rescued the embryonic lethality of par-6(zu222) worms as well as control GFP::PAR-6^WT (85% of 303 embryos hatched compared to 86% of 313 control embryos). Consistent with embryonic rescue, the subcellular distribution of GFP::PAR-6 PDZ^{F192Y, D198P} matched that of the endogenous PAR-6 protein (Figure 7), with normal anterior cortical accumulation in the first cell cycle. Because par-6(zu222) is a maternal null allele, and has normal zygotic par-6 expression, we expected that the rescued embryos would grow to adulthood. However, we observed that only 11% of the hatched embryos reached adulthood. This larval lethality appeared to be due to a combined effect of dominant negative effects of transgene expression and the subvital nature of the unc101 marker, since animals carrying the transgene and also heterozygous for unc-101 par-6(zu222) exhibited 25% larval lethality and control unc101 homozgotes showed 40% larval lethality. Assuming that this mutant GFP::PAR-6 PDZ^{F192Y, D198P} severely compromises direct binding to PAR-3 in vivo, this result indicates that the direct interaction between PAR-6 and PAR-3 PDZ1 is not essential for viability or proper distribution in the C. elegans early embryo. Although this result does not address whether there is an essential role for this interaction in later stage embryos or during larval development, independent results with structure/function analysis of PAR-3 ruled this out (B. Li and K. K.; see discussion).

Figure 7.

Embryos expressing GFP::PAR-6 and GFP::PAR-6 with the indicated mutations in the PDZ domain. Embryos are shown at one-cell pseudocleavage, one-cell prophase, two-cell and four-cell stages.

The observation that direct interaction between PAR-3 and PAR-6 appears not to be required raised the question of whether the PAR-6^{F192Y, D198P}::GFP exhibited an altered co-localization with PAR-3 in vivo. To test this possibility, we examined the distribution of PAR-3 and PAR-6^{F192Y, D198P}::GFP in par-6(zu222) homozygotes. In wild type embryos, PAR-3 and PAR-6 co-localization is dynamic such that only about 40% of the cortical puncta recognized by the two proteins contain detectable levels of both proteins (Hung and Kemphues, 1999). If the direct interaction between the two proteins plays a significant role in complex formation, we expected that we might see fewer cortical puncta that contained both PAR-3 and PAR-6^{F192Y, D198P}::GFP relative to PAR-6::GFP. We found no significant difference between the extent of co-localization of the two proteins; 41± 4%of the puncta contained both PAR-3 and PAR-6::GFP and 44±1% of the puncta contained both PAR-3 and PAR-6 ^{F192Y, D198P}::GFP (Figure 8). As a control for possible effects of transgene expression, we also examined untransformed wild type embryos (N2) and noted that although the overall level of PAR-6 protein appeared lower, the extent of co-localization was similar; 35%±4.

Figure 8.

Confocal cortical images of one-cell embryos double labeled with anti-PAR-3 and anti-PAR-6 antibodies. Embryos expressing GFP::PAR-6 and GFP::PAR-6^{F192Y D198P} were homozygous for par-6(zu222). The three panels below each embryo show images of the boxed region after processing to remove background fluorescence; anti-PAR-6 is shown in green, anti-PAR-3 in red and the overlay of the two channels is the rightward panel. Two regions of each of three embryos were used for quantifying the degree of co-localization for GFP::PAR-6 and GFP::PAR-6^{F192Y D198P}; four embryos were used for N2 (see Materials and Methods).

Disrupting ligand binding to the PDZ domain does not affect PAR-6 cortical localization and function in C. elegans embryos

The PDZ domain of Par6 is highly conserved (Joberty et al., 2000; Qiu et al., 2000), cooperates with the semi-CRIB domain to bind Cdc42 (Garrard et al., 2003; Joberty et al., 2000; Qiu et al., 2000) and can bind to the C-terminus of Crumbs/CRB3 (Kempkens et al., 2006; Lemmers et al., 2004) or to an internal segment of Pals1 (Penkert et al., 2004; Wang et al., 2004). We wished to determine the in vivo consequences of blocking the binding of PAR-6 PDZ to its presumed and unknown C. elegans ligand(s). To control for disruption of the overall PDZ domain structure, we tested our mutations in the context of a fragment of PAR-6 that includes the semi-CRIB domain and the PDZ domain (amino acids 126-252). Because binding to CDC-42^Q61L requires proper folding of the PDZ domain (Garrard et al., 2003), we sought mutations that blocked binding to Pals1 but could still bind to CDC-42^Q61L. We tested twelve mutations shown in other studies to disrupt PDZ-ligand interactions (Hurd et al., 2003; Peterson et al., 2004; Wang et al., 2004) to identify three that blocked ligand binding to mPals1 in vitro without reducing binding to CDC-42 or PAR-3 PDZ1: P167G, K161A and the double mutant R166A P167A (Supplemental Figure S2). P167G and K161A seemed to increase binding to CDC-42 ^Q61L. Then we generated GFP::PAR-6 transgenic lines carrying these mutations to test their effect on PAR-6 distribution and function in vivo. Unexpectedly, all three mutant constructs at least partially rescue the maternal effect embryonic lethality and polarity defects of par-6(zu222) homozygotes [71% of embryo viability for GFP::PAR-6^K161A, 59% of embryo viability for GFP::PAR-6^R166A,P167A and 76% of embryo viability for GFP::PAR-6^P167G]. The GFP signal in the embryos of all these transgenic lines is asymmetric at the cortex, similar to wild-type PAR-6 (Figure 7). Assuming that the mutations block the ability to bind PDZ ligands in vivo, ligand binding to the PDZ domain is not essential for PAR-6 localization or function in the early embryo.

To test whether a functional hydrophobic binding pocket is essential for PAR-6 PDZ domain function during later development of C. elegans, when zygotically expressed PAR-6 is required, and to verify the surprising results with the maternally expressed protein, we generated transgenic lines that express wild-type or mutant PAR-6::GFP driven from the par-6 promoter in a par-6(tm1425)/hIn1(unc-54) background. We recovered integrated and expressing lines for PAR-6^WT::GFP (3 lines) PAR-6^R166A,P167A::GFP (4 lines) and for PAR-6^P167G::GFP (5 lines). For two lines from each genotype, we determined the pattern of GFP accumulation in homozygous par-6(tm1425) worms by whole-mount florescence imaging and tested the ability of the lines to rescue the lethality. In transgenic lines expressing PAR-6^WT::GFP, GFP accumulated as previously reported for this construct (Totong et al., 2007), including on the apical surface of the pharynx and apical junctions of intestine cells in late embryos from comma stage to about 2-fold stage. In larvae and adults, GFP accumulated on the apical surface of the pharynx, intestine, anus, spermatheca, uterus, and vulva. Both mutant proteins showed a distribution indistinguishable from wild type. Consistent with this, all of these mutant transgenic lines rescued tm1425 homozygotes as well as wild- type PAR-6::GFP. Indeed, for both mutant transgenes, we were able to establish stable lines homozygous for tm1425 and carrying the transgene as the only source of full-length PAR-6 as verified by PCR genotyping.

DISCUSSION

C. elegans PAR-3, PAR-6, and PKC-3 co-localize at the anterior cortex of one-cell zygotes, and mutations in the three genes share a common maternal-effect lethal cell polarity phenotype (Cheng et al., 1995; Etemad-Moghadam et al., 1995; Hung and Kemphues, 1999; Kemphues et al., 1988; Tabuse et al., 1998; Watts et al., 1996). The overall protein structures of the three and their involvement in cell polarity are conserved (Goldstein and Macara, 2007; Suzuki and Ohno, 2006). Studies of mammalian and fruit fly homologues of PAR-3, PAR-6, and PKC-3 reveal conserved biochemical interactions among the proteins (Izumi et al., 1998; Joberty et al., 2000; Lin et al., 2000; Nagai-Tamai et al., 2002; Petronczki and Knoblich, 2001; Qiu et al., 2000; Suzuki et al., 2001; Wodarz et al., 2000; Yamanaka et al., 2001). We previously showed that interaction of PAR-6 with CDC-42 was essential for viability and protein localization in early C. elegans embryos (Aceto et al., 2006). Here we report the biochemical interactions of PAR-6 with its partners in C. elegans along with the results of tests of the role of these interactions in vivo.

The PB1 domain is required for PAR-6 localization and function

An internal deletion in PAR-6 that prevents or dramatically reduces its binding to PKC-3 renders the protein unable to localize cortically and unable to compensate for absence of maternally-provided PAR-6. This result is consistent with previous observations that PAR-6 is not localized at the cortex in pkc-3(RNAi) embryos (Beers and Kemphues, 2006). One possible interpretation of our result is that in wild-type embryos, direct binding to PKC-3 recruits PAR-6 to the cortex. However, previous work from our lab showed that upon reduction of the chaperone protein CDC-37 by RNAi, PAR-6 localizes to the cortex independently of PKC-3, while the cortical localization of PKC-3 still requires PAR-6 (Beers and Kemphues, 2006). Thus it is unlikely that PKC-3 directly brings PAR-6 to the cortex. Instead, we propose that PAR-6 targets PKC-3 to the cortex but that PKC-3 binding, and perhaps activity, is required for stable cortical association of PAR-6. One activity of PKC-3 might be to antagonize the activity of clients of the chaperone CDC-37 that normally block PAR-6 recruitment to the cortex (Beers and Kemphues, 2006). Reduction of PKC-3 activity at the cortex would also be expected to result in abnormal accumulation of PAR-2 and perhaps PAR-1 at the anterior cortex (Hao et al., 2006) and these two proteins would be expected to antagonize accumulation of PAR-6 (Hung and Kemphues, 1999).

We do not know whether this requirement for interaction between PAR-6 and PKC-3 extends to epithelial cells during late embryonic and larval development because our experiments to test this were inconclusive, possibly because blocking the interaction renders PAR-6 dominant negative in this developmental context.

C. elegans PAR-6 can bind directly to PAR-3 in vitro through a novel PDZ-PDZ interaction

As expected from analysis in other animals, we found that PAR-6 binds to PAR-3 in vitro and that this binding occurs via a PDZ/PDZ interaction. We determined that the binding between the two PDZ domains occurs via a novel interaction between the βC strand of PAR-6 PDZ and the βD strand of PAR-3 PDZ1. This PDZ-PDZ interaction differs from that of nNOS PDZ with the PDZ domains of Syntrophin or PSD95, in which a C-terminal extension of nNOS PDZ forms a β finger that mimicks a canonical C-terminal ligand (Hillier et al., 1999). Our results are consistent with another study showing that Pals1 and a synthetic ligand, both of which are known to bind at the hydrophobic pocket of the PDZ domain, do not compete with Par3 binding to Par6 (Peterson et al., 2004). Consistent with results from Drosophila (Peterson et al., 2004), the interaction between PAR-6 and PAR-3 is weak. This might explain why roughly equal amounts of PKC-3 co-precipitated with PAR-6 from embryonic extracts, but only a small amount of PAR-3 could be detected. This weak binding could explain the discrepancies that exist among data from different groups on whether mammalian Par3 and Par6 directly interact with each other, and if so, which regions are sufficient to mediate their interaction (Joberty et al., 2000; Lin et al., 2000; Suzuki et al., 2001). It could also explain why attempts to co-immunoprecipitate the proteins from Drosophila embryo extracts by either Par6 antibodies or Par3 antibodies were unsuccessful (Petronczki and Knoblich, 2001).

Direct interaction between PAR-3 and PAR-6 is not essential for PAR-6 function

In contrast to our results with blocking PAR-6 binding to PKC-3, we found that a par-6 mutation that blocks or greatly reduces the ability of PAR-6 to bind to PAR-3 has little or no effect on function of the protein. This result adds to accumulating evidence that despite the frequent observation of biochemical interactions between the proteins, direct binding between PAR-3 and PAR-6 may not play a significant role in their function. In Drosophila epithelial cells and photoreceptors, the distribution pattern of Baz (Par-3) is different from that of Par-6 and aPKC (Harris and Peifer, 2005; Nam and Choi, 2003), also suggesting that direct binding of Par-6 and Par-3 may not be an important component of the polarity system in some contexts. Furthermore, PDZ1 was found to be nonessential for human Par3 to rescue the consequences of RNAi-induced depletion of Par-3 from MDCK II cells (Chen and Macara, 2005). We obtained similar results in C. elegans; a par-3 transgene lacking the PDZ1 domain, which binds to the PAR-6 PDZ domain in vitro rescues a par-3 loss-of-function mutation as well as a wild type par-3 transgene (B. Li and K. Kemphues, submitted).

We cannot rule out the possibility that interaction of PAR-3 with PAR-6 is an in vitro artifact. However, repeated reports of binding using different assays in other systems argue against this possibility (Joberty et al., 2000; Johansson et al., 2000; Lin et al., 2000; Peterson et al., 2004). Alternatively, this weak interaction may make a subtle contribution to the normal function of the system, providing a level of redundancy that ensures faithful execution of cell polarity. Perhaps other proteins in the PAR-6 complex link PAR-6 and PAR-3 together even when their direct interaction is disrupted. Our observation that PAR-6 ^{F192Y, D198P}:GFP co-localizes to cortical puncta containing PAR-3 to the same extent as PAR-6::GFP and the observation by Suzuki and colleagues that T7-wt Par6 and T7-Par6 (ΔCRIB/PDZ) could pull down the same amount of Par3 from COS cells (Suzuki et al., 2001) are both consistent with this possibility.

Ligand binding ability of PAR-6 PDZ may be not essential in C. elegans

The PDZ domain of Par6 can interact with many different partners in different contexts in other animals. In mammalian cells and Drosophila the PDZ domain can bind to Crumbs/Crb3 (Kempkens et al., 2006; Lemmers et al., 2004), Stardust/Pals1 (Hurd et al., 2003; Wang et al., 2004) and perhaps Lgl (Betschinger et al., 2003; Plant et al., 2003; Yamanaka et al., 2003). A properly folded PDZ domain also plays an important role in binding of PAR-6 to CDC-42 (Garrard et al., 2003; Joberty et al., 2000; Lin et al., 2000). Two of the ligands, Crumbs and Stardust/Pals1 bind through the hydrophobic pocket of the Par6 PDZ domain, although in slightly different ways. Crumbs is typical of C-terminal ligands (Kempkens et al., 2006; Lemmers et al., 2004) although its binding may be regulated by Cdc-42 (see below). Pals1 also binds PAR-6 via interaction with the PDZ binding pocket, but the binding region is internal (Hurd et al., 2003; Wang et al., 2004). Structural analysis demonstrated that this internal peptide inserts into the hydrophobic binding pocket of Par6 PDZ in a novel way that involves conformational change of Par6 PDZ imposed by the ligand (Penkert et al., 2004). Par6 PDZ can also interact with Lgl (Betschinger et al., 2003; Plant et al., 2003; Yamanaka et al., 2003), but the mechanism for this interaction is not clear.

Another interesting feature of Par6 PDZ is that it is structurally and functionally coupled to the adjacent semi-CRIB motif. Although CDC-42 mainly makes contacts with the semi-CRIB motif of Par6, this interaction also requires the presence of the PDZ domain (Joberty et al., 2000). Structural studies indicated that the PDZ domain makes minor contact with Cdc42 and stabilizes the interaction between the semi-CRIB motif and Cdc42 (Garrard et al., 2003). In turn, Cdc42 binding to the semi-CRIB motif can also modify the conformation of the PDZ domain, which may increase the binding affinity between Par6 PDZ and its ligand (Peterson et al., 2004).

Taken together, these results led us to the hypothesis that the PAR-6 PDZ domain binds in a conventional way to an unknown ligand and that the binding might be regulated by CDC-42. To test that hypothesis and to determine how PAR-6 behaves in the absence of ligand binding, we mutated the binding pocket to block binding to Pals1 but not CDC-42.

In our analysis, most mutations, including single point mutations in PAR-6 PDZ affected the interaction of PAR-6 with CDC-42. Surprisingly, PAR-6^{R166A P167A} and PAR-6 ^P167G, mutations that severely reduced ability to bind Pals1 without disrupting CDC-42 binding, were able to function as well as wild type throughout the life of the worm. A third mutant form of the protein, PAR-6^K161A rescued the maternal requirement for PAR-6 but was not tested for rescue of zygotic function. Although we cannot easily determine whether these mutant PAR-6 proteins are completely blocked for binding of the more typical C-terminal ligands, it seems unlikely that none of the mutations we tested would disrupt binding to a conventional PDZ ligand. Thus, our results raise the possibility that ligand binding to the PDZ domain is not essential for PAR-6 function in C. elegans. In contrast to our results with the C. elegans PAR-6 PDZ, the PDZ domain of mammalian Par6 may be important for its function. In MDCK cells, overexpression of WT Par6 inhibited tight junction formation, whereas overexpression of Par6^{KPLG167-170AAA} or Par6^P171G failed to inhibit (Joberty et al., 2000; Peterson et al., 2004). However, it may be important to revisit this question taking into account the possibility that the effect was due to blocking binding to Cdc-42 (Peterson et al., 2004) or reducing binding to Par-3 (Joberty et al., 2000) and using rescue of RNAi knockdown rather than suppression of dominant negative effects to assay for function.

Does the C. elegans PAR-6 PDZ domain function solely to facilitate binding to CDC-42?

Assuming that our mutations successfully block the interactions that we targeted, we are left with the surprising possibility that the PDZ domain of C. elegans PAR-6 functions primarily to facilitate binding to CDC-42. Our results indicating that direct binding of PAR-6 to PAR-3 is not essential are consistent with a number of previous observations. More unexpected is our finding that mutations known to block ligand binding do not detectably disrupt PAR-6 function. The high degree of conservation of amino acids in the binding pocket argues fairly compellingly for evolutionary pressure to maintain the precise configuration of the pocket. Although it is reasonable to assume that conservation in the binding pocket reflects constraints on ligand binding, it is possible that this conservation instead reflects constraints on structure that preserve CDC-42 binding. Our finding that 9 of 12 mutations we chose for their potential to disrupt ligand binding also affect CDC-42 binding supports this latter possibility.