The pax-3 gene is involved in vulva formation in Pristionchus pacificus and is a target of the Hox gene lin-39

In nematodes, basic principles in developmental biology and in evolutionary developmental biology can be studied at the single-cell level(Hong and Sommer, 2006). In Caenorhabditis elegans, detailed genetic and molecular studies of vulva formation present a framework for studying pattern formation, cell fate specification, developmental competence and induction(Sternberg, 2005). At the same time, comparative studies in other nematodes, in particular the satellite organism Pristionchus pacificus, provide insight into the evolutionary changes of developmental processes and mechanisms(Sommer, 2005). Nearly all aspects of vulva formation are subject to evolutionary change. P. pacificus is amenable to forward and reverse genetic analysis, and provides a platform to identify genetic and molecular alterations that control evolutionary changes of developmental processes. P. pacificuspropagates as a self-fertilizing hermaphrodite and has an integrated genome map that contains a genetic linkage map of more than 600 molecular markers and a physical map of nearly 10,000 fingerprinted BAC clones(Srinivasan et al., 2002; Srinivasan et al., 2003). A whole-genome sequencing project has recently been finished(Dieterich et al., 2007).

The nematode vulva is a derivative of the ventral epidermis, which consists of 12 precursor cells, called P(1-12).p, in all nematodes studied to date(with `Pn.p' denoting the complete group of ventral epidermal precursor cells)(Fig. 1A). These 12 Pn.p cells adopt different cell fates according to positional information provided by homeotic control genes. In the C. elegans hermaphrodite, the first cell fate decision in the ventral epidermis is between potential vulval precursor cells (VPCs) and `non-vulval' cells. Cells in the anterior and posterior body region, P(1,2,9-11).p, fuse with the hypodermal syncytium hyp7 early in development. By contrast, P(3-8).p in the central body region remain non-fused, become VPCs and form a so-called vulva equivalence group (VEG)(Fig. 1A). Genetically, the Hox gene lin-39 defines the developmental competence of P(3-8).p and establishes this group of cells as VPCs. In C. elegans lin-39(Cel-lin-39) mutants, P(3-8).p fuse with the hypodermis, resulting in a phenotype that has been designated as `generation vulvaless' (Gev; Fig. 1B)(Clark et al., 1993; Wang et al., 1993). It has been suggested that the early role of Cel-LIN-39 is the indirect regulation of the cell fusion effector eff-1. Cel-LIN-39 regulates the GATA transcription factors egl-18 and elt-6, which in turn might regulate eff-1 (Fig. 2A) (Shemer and Podbilewicz,2002; Koh et al.,2002; Cassata et al.,2005).

In C. elegans, P(3-8).p adopt one of three alternative fates later in development (Sternberg,2005). P6.p generates eight progeny and forms the central part of the vulva, a fate that is designated as 1° fate. P(5,7).p have a 2°fate, generate seven progeny each and form the anterior and posterior part of the vulva, respectively. P(3,4,8).p do not participate in vulva formation in wild-type animals, remain epidermal and have the so-called 3° fate. After cell ablation of P(5-7).p, P(3,4,8).p can substitute for the other VPCs,indicating that all VPCs have the competence to form part of the vulva. Detailed genetic and molecular studies revealed a complex regulatory network involved in C. elegans vulva formation. An epidermal growth factor(EGF)-like molecule encoded by the gene lin-3 is secreted from the gonadal anchor cell (AC) and induces P(5-7).p to adopt 1° and 2°vulval fates. The LIN-3 signal is transmitted within the VPCs by an EGFR/RAS pathway. Wnt signaling and DSL/Notch signaling have been shown to participate in vulva development, and various specification events involve redundant functions of different signaling cascades. The Hox gene lin-39 is a downstream target of EGF and Wnt signaling during vulva induction(Maloof and Kenyon, 1998; Wagmaister et al., 2006). Thus, Cel-LIN-39 is used twice during vulva formation: first, to establish the VEG and, second, in response to vulva induction.

In P. pacificus, vulva formation differs substantially from that in C. elegans. Although, in both organisms, P(5-7).p form vulval tissue with a 2°-1°-2° fate pattern, novel cell-cell interactions that are unknown from C. elegans are crucial for vulva induction in P. pacificus (Sommer,2005). Four important differences in cell fate specification are apparent between P. pacificus and C. elegans(Fig. 1C). First, in P. pacificus, non-vulval Pn.p cells do not fuse with hyp7 as they do in C. elegans; instead, they die from programmed cell death (PCD)(Sommer and Sternberg, 1996). Second, this PCD involves P(1-4).p in the anterior body region and P(9-11).p in the posterior body region, and has an influence on the size of the VEG;therefore, P(3,4).p are members of the VEG in C. elegans but die in P. pacificus. Thus, PCD limits the size of the VEG in P. pacificus (Eizinger and Sommer,1997). Third, vulva induction in P. pacificus involves multiple cells of the somatic gonad and takes more than 10 hours, whereas the AC induces the vulva in a one-step interaction in C. elegans(Sigrist and Sommer, 1999). Fourth, the epidermal cell P8.p does not divide in P. pacificus and is not competent to adopt a vulva fate, whereas it is a VPC in C. elegans. However, P8.p is involved in cell fate specification of the VPCs by inhibiting P(5,7).p from taking a 1° fate, an interaction that has been designated as lateral inhibition (Jungblut and Sommer, 2000).

Genetic and molecular studies revealed that the VEG in P. pacificus is also specified by the Hox gene lin-39(Eizinger and Sommer, 1997). In P. pacificus lin-39 (Ppa-lin-39) mutants, P(5-8).p die from PCD like their anterior and posterior lineage counterparts, indicating that the ground state of ventral epidermal cells in P. pacificus is PCD(Fig. 1D, Table 1B). Thus, Ppa-LIN-39 prevents PCD, whereas Cel-LIN-39 prevents cell fusion, indicating that Ppa-LIN-39 must be part of a different regulatory network than Cel-LIN-39 during VEG formation.

Proteins encoded by the Pax gene family play a crucial role in animal development (for a review, see Chi and Epstein, 2002). Pax proteins contain up to three highly conserved protein domains: an N-terminal paired domain (PD) and a C-terminal homeodomain(HD), both of which are involved in DNA-binding, as well as an octapeptide. Pax proteins were shown to function as both transcriptional activators and repressors. For example, in mammals, Pax5 (also known as BSAP in humans) is converted from a transcriptional activator to a repressor via interactions with the co-repressors of the Groucho family in an octapeptide-dependent manner (Eberhard et al., 2000). At the sequence level, the presence or absence of the HD and/or the octapeptide, and the sequence similarities within the PD, allow the subdivision of Pax proteins into specific subfamilies(Chi and Epstein, 2002). Although there is a general conservation of Pax genes throughout the animal kingdom, many differences can be found in individual taxa. For example, the genome of the nematode C. elegans contains five bona-fide Pax genes(Hobert and Ruvkun, 1999). vab-3 belongs to the Pax-6 subfamily and is involved in head and tail development (Chisholm and Horvitz,1995; Zhang and Emmons,1995). egl-38 is a member of the Pax-2/5subfamily and regulates uterine and tail development(Chamberlin et al., 1997). The highly related gene K06B9.5 results from a recent gene duplication. Interestingly, the gene F27E5.2 is the only gene present in the C. elegans genome that contains all three protein domains, the PD, HD and octapeptide.

Here, we describe, by mutant analysis, the first specific postembryonic developmental function of a nematode Pax-3-type gene. We show that Ppa-pax-3 is involved in the formation of the VEG and that Ppa-pax-3 has distinct functions in cell fate specification of epidermal cells. Whereas Ppa-pax-3 regulates cell survival of the central Pn.p cells, it regulates cell death of posterior epidermal cells. Additional data indicate that Ppa-pax-3 is a direct target of Ppa-LIN-39. Together, these data show a function for a nematode Pax-3-type gene during vulva development and indicate a different regulatory network for the formation of the VEG.

Worms were grown on 5 cm NGM agar plates seeded with OP50, a uracil-requiring mutant of Escherichia coli(Sommer and Sternberg, 1996). The following strains were used in this study: P. pacificus PS312(the wild-type strain) is a derivative of an isolate from Pasadena, California and represents the wild-type strain used for genetic analysis; P. pacificus PS1843, isolated from Port Angeles, Washington represents the polymorphic reference strain for mapping(Srinivasan et al., 2002). Other mutant strains used in this study were Ppa-lin-39(tu29)(Eizinger and Sommer, 1997) and Ppa-mab-5(tu357) (Jungblut and Sommer, 1998).

TMP/UV mutagenesis was carried out as described elsewhere(Jungblut and Sommer, 2001). In short, mixed-stage animals were washed off the plates in M9 buffer and were incubated for 20 minutes with 33 μg/ml TMP and then UV irradiated for 50 seconds with an intensity of 500 μW/cm². In the F2 generation,egg-laying defective mutants were isolated and their progeny were reanalyzed for vulva defects using Nomarski microscopy. Mutant hermaphrodites were backcrossed multiple times using wild-type males. Complementation tests were carried out using dpy-marked Gev mutants(Kenning et al., 2004).

For mapping, mutant hermaphrodites in the California background were crossed with males of the Washington strain. To extract genomic DNA, F2 mutant animals were picked to single tubes containing 2.5 μl of lysis buffer (50 mM KCl; 10 mM Tris-HCl pH 8.3; 2.5 mM MgCl₂; 0.45% NP-40; 0.45%Tween; 0.01% gelatin; 5 μg/ml Proteinase K) and incubated for 1 hour at 65°C, followed by inactivation of the Proteinase K at 95°C for 10 minutes. To assign linkage of a mutation to a certain chromosome, two representative single-strand conformational polymorphism (SSCP) markers per chromosome were tested against 42 Washington-backcrossed mutant animals. For SSCP detection, PCR samples were diluted 1:1 in denaturing solution (95%formamide, 0.1% xylene cyanol, 0.1% bromophenol blue), denatured at 95°C for 5 minutes and loaded onto a GeneGel Excel prepoured 6% acrylamide gel(PharmaciaBiotech, Piscataway, NJ). Gels were fixed and silver stained to detect the DNA.

Oligonucleotides (Gene Tools) were dissolved in water and subsequently diluted to a concentration of 100 μM. Primer sequences are available on request.

A total of 120 J1 animals were picked into 15 μl of 1:10 diluted single worm lysis buffer. RNA was extracted with 100 μl of TRIZOL using a repeated freeze-thaw protocol in liquid nitrogen. RNA was reverse transcribed in 20μl total volume reaction (Invitrogen) with negative controls without reverse transcriptase included for each sample. Quantitative PCR was performed on a Roche LC480 LightCycler using the manufacturer's SYBR green PCR mix. Primer concentrations were 0.5 mM.

Ppa-LIN-39 and Ppa-CEH-20 proteins were expressed by coupled in vitro transcription/translation (Promega) using plasmids pBY01 and pBY02. Complementary oligonucleotides were annealed and end-labeled with[γ32P]-ATP using T4 polynucleotide kinase (NEB, Beverly, MA). Labeled oligonucleotides (1.5 pmol) were used in bandshift experiments. Electrophoresis was carried out at 4°C on a 6% non-denaturing polyacrylamide gel.

We used the SITEBLAST software (Michael et al., 2005) and consensus sequences from the JASPAR database to identify conserved putative Hox target enhancers in the P. pacificusgenome.

In various TMP/UV mutagenesis screens, we isolated two new mutants(tu214 and tu358) with a Gev phenotype. Complementation tests revealed that tu214 and tu358 were not allelic to Ppa-lin-39, but were allelic to one another (data not shown). Therefore, we have named this novel gene gev-2 for generation vulvaless-2.

gev-2 mutant animals were strongly egg-laying defective. More-detailed cell lineage analysis in the ventral epidermis of gev-2mutants revealed two independent defects in Pn.p cell fate specification(Table 1). First, the central cells, P(5-8).p, died from PCD in many gev-2 mutant animals,resulting in the absence of a vulva (Fig. 3). Although both alleles had a similar PCD phenotype, they showed a slight posterior bias in that P8.p survived with a higher frequency than P5.p (Table 1C,D). Second,mutations in gev-2 resulted in a distinct phenotype in the posterior body region. In wild-type and Ppa-lin-39 mutant hermaphrodites P(9-11).p die from PCD (Table 1A,B) (Sommer and Sternberg,1996). However, P(9-11).p survive in gev-2 mutant animals with an frequency of 10-20% (Table 1C,D). Thus, gev-2 has different functions in Pn.p cell fate specification in the central and the posterior body region: it positively regulates cell survival of P(5-8).p and induces the PCD of P(9-11).p(Fig. 1E).

Mutations in lin-39 result in a Gev phenotype in both P. pacificus and C. elegans(Clark et al., 1993; Wang et al., 1993; Eizinger and Sommer, 1997). However, Ppa-lin-39 and Cel-lin-39 differ with regard to their involvement in vulva induction. Whereas Cel-lin-39 is crucial for vulva induction and was shown to represent a downstream target of EGF/RAS and WNT signaling (Maloof and Kenyon,1998; Wagmaister et al.,2006), Ppa-lin-39 is dispensable for vulva induction: in Ppa-lin-39; Ppa-ced-3 double mutants, the PCD of Pn.p cells is rescued and the surviving VPCs undergo normal vulva differentiation(Table 1E)(Sommer et al., 1998).

To determine whether gev-2 has a role in P. pacificusvulva induction, we generated similar double mutants with a cell-death-defective mutant to overcome the PCD of P(5-8).p. Because genetic mapping of gev-2 revealed that it is located on chromosome II in close proximity to the Ppa-ced-3 gene (see below for details), we generated double mutants containing gev-2 with a second cell-death-defective mutant previously described as ipa-2(Sommer et al., 1998). ipa-2 shows complete cell survival of all Pn.p cells, similar to Ppa-ced-3. ipa-2 maps to chromosome III and has been shown to be the Ppa-ced-4 gene, which encodes an ortholog of the cell-death adaptor CED-4/APAF-1 (Dinkelacker, Lee and Sommer, unpublished information). Interestingly, the phenotype of the Ppa-gev-2; Ppa-ced-4 double mutant was allele-specific: whereas Ppa-gev-2(tu358);Ppa-ced-4(tu324) mutant animals mostly formed a normal vulva(Table 1F), Ppa-gev-2(tu214); Ppa-ced-4(tu324) double mutants were slightly induction vulvaless. Specifically, 20-30% of the VPCs remained un-induced in Ppa-gev-2(tu214); Ppa-ced-4(tu324) double-mutant animals(Table 1G). This is consistent with findings in the gev-2(tu214) single mutant, in which the majority of the surviving VPCs remained epidermal. Only 12% of P5.p cells, 31%of P6.p cells and 17% of P7.p cells, respectively, formed vulva tissue in tu214 mutant animals, although a higher percentage of cells did survive (Table 1C). By contrast, nearly all of the surviving VPCs underwent vulva differentiation in the tu358 allele (Table 1D). When we scored vulva induction in transheterozygous tu214/tu358 animals, we found vulva differentiation in most of the surviving VPCs (Table 1H). Together, these results suggest that gev-2(tu214) has an allele-specific effect indicating a function of gev-2 during vulva induction.

Pn.p cell fate specification differs between P. pacificushermaphrodites and males. Whereas the anterior cells P(1-4).p died in both sexes, P(5-8).p survived only in hermaphrodites and P(9-11).p survived only in males (Fig. 1; Tables 1, 2). In P. pacificusmales, P10.p and P11.p formed the hook and various neural structures,respectively, and thus have cell fates similar to their counterparts in C. elegans (Fig. 1F,H). P9.p did not divide in P.pacificus and remained epidermal, a fate pattern that is also identical to P9.p in C. elegans. Given that mutations in gev-2 result in distinct phenotypes in P(5-8).p and P(9-11).p in hermaphrodites, we wanted to know whether this phenotype is sex-specific. We generated gev-2; Ppa-him-2 double mutants, which result in a high incidence of males, and found that P(9-11).p survived and differentiated normally, as in wild-type animals(Table 2B). Similarly, in the occasionally observed spontaneous gev-2 males in an otherwise wild-type background, P(9-11).p differentiated normally(Fig. 1J).

In C. elegans, the Hox gene mab-5 is crucial for P(9-11).p development in males and has a function similar to the Hox gene Cel-lin-39 in hermaphrodites (Fig. 1G) (Kenyon,1986). To study whether Ppa-mab-5 is involved in P(9-11).p determination in males, we investigated Ppa-mab-5(tu357)mutant males. Indeed, P(9-11).p underwent PCD in Ppa-mab-5(tu357)males, indicating that the role of MAB-5 in P(9-11).p male-specification is conserved between P. pacificus and C. elegans(Fig. 1G,I; Table 2C). Similarly, P(9-11).p also underwent PCD in gev-2(tu214); Ppa-mab-5(tu357) double mutants(Table 2D). Taken together, the analysis of P(9-11).p determination in P. pacificus males suggests strong similarities to C. elegans, but does not provide any evidence for a role of gev-2. Thus, gev-2 represents a hermaphrodite-specific regulator of Pn.p cell fate specification with distinct functions in the central and the posterior body region.

To identify the molecular nature of gev-2, we mapped the locus using the polymorphic reference strain P. pacificus var. Washington(Srinivasan et al., 2002). gev-2 maps to chromosome II between the molecular markers S264 and S152, and is most closely linked to the marker S98, which is associated with the BAC clone BACPP17D-2 (Fig. 4A). Fine mapping determined recombination breakpoints to the left and right of S98, and established an interval of three BAC clones. Light shotgun sequencing of all three BAC clones and BLAST searches of these sequences with the draft of the P. pacificus genome sequence identified the pax-3 gene as a potential candidate gene for gev-2 (Fig. 4B). When we sequenced the pax-3 gene as a candidate for gev-2, we found that it is mutated in both alleles of gev-2(Fig. 4C). We identified a deletion of 830 bp from intron 7 to exon 9 in tu214. Sequence analysis of mutant cDNA clones revealed the use of a novel splice acceptor site, which caused skipping of the complete exon 9, resulting in a deletion of the HD and the 3′ UTR (Fig. 4C, Fig. 5A). tu358 results in an amino acid replacement of His to Arg, which is one of the highly conserved residues in the octapeptide(Fig. 4C,D). We conclude that gev-2 is Ppa-pax-3.

Although Ppa-pax-3(tu214) represents a strong reduction-of-function allele, the available alleles do not necessarily indicate the null phenotype of Ppa-pax-3. To identify a potential null allele of Ppa-pax-3, we used reverse genetic deletion library screens, but were unable to identify a third allele after screening more than 4 million gametes (data not shown). To study the function of Ppa-pax-3 further, we used morpholino (MO) knockdown experiments,which were previously shown to function in sensitized genetic backgrounds(Zheng et al., 2005). Consistent with these previous findings, we saw no effect of a Ppa-pax-3 MO when applied in wild-type animals(Table 1I). However, a Ppa-pax-3 MO in a Ppa-pax-3(tu358) mutant background showed a strong enhancement of the cell-death and the vulva-differentiation phenotype(Table 1J), further indicating that the observed effects are due to a reduction of Ppa-pax-3function.

Ppa-PAX-3 has bona-fide PD and HD domains and an octapeptide, as known from other PAX-3-type proteins (Fig. 4B) (Chi and Epstein,2002). The overall amino acid sequence identity between Ppa-PAX-3 and Cel-PAX-3 is 49%, and is highest in the three conserved domains (Fig. 4D). Interestingly, Ppa-pax-3 and Cel-pax-3 differ in gene structure and isoform formation. Whereas Cel-pax-3 has six exons and forms a single isoform, Ppa-pax-3 contains nine exons and two isoforms were isolated in reverse transcriptase (RT)-PCR experiments. Using the spliced leader SL1 as the forward primer in RT-PCR experiments, two alternatively spliced forms were obtained(Fig. 5A). The second Ppa-pax-3 isoform misses the N-terminal part of the PD, but contains the octapeptide and the HD (Fig. 5A).

To study the expression of Ppa-pax-3, we used quantitative RT-PCR experiments with synchronized embryos, J2, J3, J4 and adult animals. We found Ppa-pax-3 expression in only embryos and young larvae up to the J2 stage (Fig. 5B). Ppa-pax-3 was not expressed in P. pacificus adults, which,unlike C. elegans, do not retain many eggs. This expression coincides with the generation of the VPCs and the PCD of their anterior and posterior sister cells. PCR primers in the N-terminal part of the cDNA resulted in a similar expression profile, suggesting that the two isoforms are regulated in a similar way (data not shown). These results suggest that Ppa-pax-3is primarily expressed in the embryo and early larval stages during the generation of the Pn.p cells, but prior to vulva induction.

Given that mutations in Ppa-pax-3 and Ppa-lin-39 have a similar PCD phenotype in P(5-8).p, we wanted to know whether these two loci interact with one another. First, we generated a Ppa-pax-3(tu214);Ppa-lin-39(tu29) double mutant and analyzed the cell survival frequencies of P(5-8).p. Ppa-pax-3(tu214); Ppa-lin-39(tu29) double mutants had a cell survival index of 0.9 cells per animal, which is similar to the survival index of the Ppa-lin-39(tu29) single mutant(Table 1B,K). This result indicates that the frequency of cell survival in Ppa-pax-3;Ppa-lin-39 double mutants is not additive, providing a first indication that these genes might act in a linear pathway.

To obtain further evidence, we analyzed the expression of Ppa-pax-3 in Ppa-lin-39 mutant and wild-type animals, and the expression of Ppa-lin-39 in Ppa-pax-3 mutant and wild-type animals, by quantitative RT-PCR. Whereas the expression of Ppa-lin-39 was unchanged between wild-type and Ppa-pax-3(tu214) mutant animals, Ppa-pax-3 expression was drastically reduced in Ppa-lin-39(tu29) mutant animals when compared with wild-type (Fig. 5C,D). Specifically, the relative pax-3 transcript level was reduced eightfold in Ppa-lin-39 mutants. These results suggest that Ppa-lin-39 might act upstream of Ppa-pax-3.

To test whether Ppa-pax-3 is directly regulated by Ppa-LIN-39, we searched for Hox-binding sites in the Ppa-pax-3 promoter. Hox-binding sites have been studied intensively by in vivo and in vitro studies, and Hox proteins are known to interact with cofactors, such as Extradenticle-like proteins of the PBC class in Drosophila (Pearson et al.,2005). A common property of Hox target enhancers is the requirement for multiple Hox-monomer-binding sites, most of which possess an ATTA (or TAAT) core sequence (Pearson et al., 2005). In addition, many Hox-response elements require Hox-PBC heterodimer binding sites. We searched for potential binding sites in the Ppa-pax-3 promoter using SITEBLAST software(Michael et al., 2005) and found one putative Hox-PBC-binding site, which conformed to the sequence TGATNNAT (Fig. 6A).

Next, we tested whether Ppa-LIN-39 could bind this putative binding site by using electrophoretic mobility shift assays (EMSA)(Fig. 6B). Ppa-LIN-39 caused a shift in mobility of an oligonucleotide containing a high-affinity binding site for Drosophila Antennapedia, but not for an oligonucleotide containing the binding site of the Ppa-pax-3 promoter(Fig. 6B). However, when we added the PBC-type cofactor Ppa-CEH-20 to Ppa-LIN-39, the oligonucleotide with the binding site of the Ppa-pax-3 promoter was shifted in a similar way as the Drosophila Antennapedia control oligonucleotide (Fig. 6B). At the same time, Ppa-CEH-20 alone did not cause a shift of the oligonucleotide (Fig. 6B). When the binding site of the Ppa-pax-3 promoter had been mutated, the shift of the oligonucleotide by Ppa-CEH-20 and Ppa-LIN-39 was abolished. We conclude that a heterodimer of Ppa-LIN-39 and Ppa-CEH-20 can bind to the Ppa-pax-3 promoter in vitro.

Finally, we wanted to know whether the HOX-PBC-binding site in the Ppa-pax-3 promoter is evolutionarily conserved. P. pacificusand C. elegans are members of different nematode families, and promoter elements cannot be compared over such evolutionary distances(Grandien and Sommer, 2001). Therefore, we cloned the pax-3 gene, including the 5′regulatory regions, from Pristionchus sp. 11 and Pristionchus maupasi, two additional members of the genus Pristionchus(Herrmann et al., 2006; Herrmann et al., 2007). In these species, cell fate specification of the VPCs is identical to P. pacificus (data not shown). Although the pax-3 coding regions were found to be highly conserved between all three Pristionchusspecies, the 5′ regulatory regions were much less conserved. We found that the HOX-PBC-binding site that caused the shift in the EMSA assay is nearly completely conserved between all three Pristionchus species,further supporting the significance of the Ppa-pax-3 promoter element for regulation by Ppa-LIN-39 (Fig. 6C).

In this study, we provide the genetic and molecular characterization of a genetic locus involved in cell fate specification of the ventral epidermis of P. pacificus hermaphrodites. Ppa-pax-3, together with Ppa-lin-39, is involved in the formation of the P. pacificusVEG and in the regulation of the survival of P(5-8).p. In contrast to Ppa-lin-39, Ppa-pax-3 has an additional function in the posterior body region and regulates the PCD of P(9-11).p. We show that Ppa-pax-3 is expressed in the embryo and early larval stages, and we provide molecular evidence that Ppa-LIN-39 regulates Ppa-pax-3 and that Hox-PBC-binding sites in the Ppa-pax-3promoter are evolutionarily conserved.

This study provides the first detailed analysis of the function of a Pax-3-type gene in nematode postembryogenesis and indicates a role for Ppa-pax-3 in a Hox gene-regulated developmental process, namely the establishment of the VEG. Ppa-pax-3 mutants are Gev and represent the second complementation group besides Ppa-lin-39 with such a phenotype. Also, in C. elegans, two complementation groups with a Gev phenotype have been identified - Cel-lin-39(Wang et al., 1993; Clark et al., 1993) and Cel-ceh-20, the latter of which encodes a PBC-like gene(Liu and Fire, 2000; Takacs-Vellai et al., 2007). Although Ppa-ceh-20 was originally a candidate for the genetic locus described in this study, further mapping analysis indicated that Ppa-ceh-20 and gev-2 are located on different chromosomes.

pax-3 is unique among the nematode Pax genes with regard to the presence of all three functional domains: the PD, the HD and the octapeptide. Our mutant analysis revealed that Ppa-pax-3(tu358) represents a point mutation in the octapeptide, resulting in an amino acid substitution of the first amino acid from histidine to arginine. This histidine is highly conserved in PAX-3-type proteins throughout the animal kingdom. The fact that the cell-survival phenotype of tu358 is nearly identical to the one of the deletion mutant tu214, which eliminates part of the HD,supports the idea that the octapeptide is of functional importance for the repression of PCD. Interestingly, studies of the mammalian Pax5 protein BSAP indicate that the octapeptide is only required for the repressive function of BSAP, whereas it is dispensable for the activator function(Eberhard et al., 2000). Given that surviving VPCs have no vulva phenotype in the tu358 allele,whereas mutant animals of the deletion allele tu214 result in vulva differentiation defects, we speculate that the octapeptide is dispensable for the role of Ppa-pax-3 in vulva induction, which is most probably involved in transcriptional activation.

Our analysis of Ppa-pax-3 indicates significant differences in the regulation of the VEG between P. pacificus and C. elegans. Whereas studies in C. elegans suggest an indirect role of LIN-39 in the regulation of the cell fusion effector eff-1 via the GATA transcription factors egl-18 and elt-6(Shemer and Podbilewicz, 2002; Koh et al., 2002; Cassata et al., 2005), our studies show that Ppa-LIN-39 is part of a different regulatory network. Our genetic and molecular results suggest that Ppa-LIN-39 acts upstream of the Ppa-pax-3 gene(Fig. 2B). We speculate that Ppa-PAX-3 is directly involved in the suppression of PCD of P(5-8).p. However, the exact molecular mechanism of this suppression remains to be identified and several distinct mechanisms might be at work. One potential target for Ppa-PAX-3 might be the ortholog of egl-1, a key regulator of PCD in C. elegans (for a review, see Yuan, 2006). So far, we have been unable to detect an ortholog of egl-1, which encodes a small peptide, in the P. pacificus genome assembly. However, alternative mechanisms could also be involved in the regulation of PCD. Park et al. have recently shown that the Pax2/5/8 member egl-38 and pax-2promote cell survival in C. elegans(Park et al., 2006). These studies suggest a role for egl-38 and pax-2 in the regulation of ced-9 (Park et al.,2006). Future work in P. pacificus will indicate whether,and how, Ppa-pax-3 is coupled to the PCD machinery.

Besides its role in vulva formation, Ppa-PAX-3 plays a role during the regulation of the PCD of P(9-11).p. This finding indicates for the first time that P(9-11).p are regulated in a different way than the ventral epidermal cells in the anterior body region. Interestingly, the function of Ppa-pax-3 in P(9-11).p specification is independent of Ppa-LIN-39, indicating that Ppa-pax-3 is regulated differently in these cells. Indications about the mechanism of this regulation might come from the analysis of a second gene, ped-9, which, when mutated, causes the survival of P(9-11).p, and which is currently being analyzed further (R. Molnar and R.J.S., unpublished observation). With regard to P(1-4).p, recent studies have shown that P(3,4).p are regulated differently from P(1,2).p; mutations in Ppa-hairy, a gene that has no counterpart in the C. elegans genome, or in Ppa-groucho, the ortholog of C. elegans unc-37, result in the survival of P(3,4).p, but not of P(1,2).p (Schlager et al.,2006). Genetic and biochemical studies revealed that Ppa-HAIRY and Ppa- GROUCHO downregulate the activity of Ppa-lin-39 and thereby restrict the size of the VEG to P(5-8).p. The finding that Ppa-pax-3 can rescue P(9-11).p, but not P(1,2).p, from PCD indicates that at least five different types of positional information subdivide the 12 ventral epidermal cells of P. pacificus(Fig. 2C): in the central body region, P(5-8).p survive based on positional information provided by Ppa-LIN-39 and Ppa-PAX-3(Eizinger and Sommer, 1997)(this study). In the posterior body region, P(9-11).p die from PCD and Ppa-PAX-3 has a role distinct from the one described for P(5-8).p. P12.pa is a special cell that might be regulated by the Hox gene Ppa-egl-5, as occurs in C. elegans. However, a Ppa-egl-5 mutant has not been identified yet. In the anterior region,P(3,4).p die because of the downregulation of Ppa-lin-39 by Ppa-HAIRY and Ppa-GROUCHO(Schlager et al., 2006). The regulation of PCD in P(1,2).p is distinct from the PCD of P(3,4).p and P(9-11).p, but no specific mutants have been isolated that rescue P(1,2).p in a specific manner. Because all Pn.p cells survive in Ppa-ced-3 and Ppa-ced-4 mutants, the PCD of P(1,2).p is dependent on the normal PCD machinery (Sommer et al.,1998). One might speculate that the fate of P(1,2).p represents the ground state of Pn.p cells and that no specific positional information is required for the determination of this fate. Taken together, the seven cell-death events in the ventral epidermis of the P. pacificushermaphrodite are highly regulated and require three distinct genetic mechanisms in the different regions of the animal(Fig. 2C). In C. elegans, the distinct cell fates of Pn.p cells in the hermaphrodite are also regulated in a region-specific manner. In particular, cell fusion of P(1,2).p and P(9-11).p are regulated by distinct genetic programs(Alper and Kenyon, 2001).

We thank M. Grbic for help with the isolation of the tu358 allele;C. Dieterich for support with the bioinformatics analyses; D. Rudel, B. Schlager and H. Zauner for critically reading the manuscript; and members of the Sommer laboratory for discussion.