C. elegans SUR-6/PR55 cooperates with LET-92/protein phosphatase 2A and promotes Raf activity independently of inhibitory Akt phosphorylation sites

The kinase Raf is a key target of the Ras GTPase during both normal development and oncogenesis, and acts upstream of MEK and ERK in the mitogen activated protein kinase (MAPK) cascade(Morrison and Cutler, 1997). Raf activation appears to involve multiple steps, including release of autoinhibition by the Raf N terminus, recruitment to the plasma membrane,Ras-GTP binding, multimerization and changes in phosphorylation status(Chong et al., 2003). Genetic screens in C. elegans and Drosophila have identified several gene products that function at a step between Ras and Raf and therefore appear to regulate Raf activation (Sternberg and Alberola-Ila, 1998; Moghal and Sternberg, 2003). These include KSR (kinase suppressor of Ras), a Raf-related protein that binds to Raf, MEK and ERK, and may function non-catalytically as a scaffold (Morrison,2001), SUR-8, a leucine-rich repeat protein that binds to Ras(Sieburth et al., 1998), and SUR-6, a PR55/B-regulatory subunit of protein phosphatase 2A (PP2A)(Sieburth et al., 1999). The mechanisms by which these proteins promote Raf activation are not well understood. However, the molecular nature of SUR-6 suggests that it may regulate the phosphorylation status of Raf or of another Raf regulatory protein.

PP2A is a heterotrimeric serine/threonine phosphatase composed of invariant catalytic (`C') and structural (`A') subunits and a variable regulatory subunit (`B') that directs the AC core complex to different substrates(Janssens and Goris, 2001). PP2A both positively and negatively influences the Ras/MAP kinase pathway in Drosophila and in mammalian cells, suggesting it may act on multiple Ras pathway substrates (Sontag et al.,1993; Alessi et al.,1995; Wassarman et al.,1996; Maixner et al.,1998; Ugi et al.,2002; Strack,2002; Silverstein et al.,2002). Unfortunately, the pleiotropic defects caused by interfering with PP2A activity in vivo and the very broad substrate specificity of PP2A in vitro have hampered attempts to identify its most functionally relevant substrates. The finding that partial loss-of-function alleles of sur-6/B specifically reduce Ras signaling in C. elegans (Sieburth et al.,1999) provides a potentially simpler genetic model system for studying the effects of PP2A on the Ras pathway.

C. elegans vulval development is a well characterized model system for studying the Ras signaling pathway(Moghal and Sternberg, 2003). The vulva is generated by a specialized subset of ventral ectodermal blast cells called vulval precursor cells (VPCs)(Fig. 1A). During larval development, Ras signaling induces three of six equipotent VPCs to execute a vulval lineage. The remaining three uninduced VPCs execute a non-vulval hypodermal lineage. The EGF receptor/Ras/MAP kinase pathway(Fig. 1B) is required for vulval induction, as complete loss of pathway activity causes a Vulvaless(Vul) phenotype in which no VPCs adopt vulval fates. Increased Ras activity causes a Multivulva (Muv) phenotype in which greater than three VPCs adopt vulval fates. Thus, the extent of vulval differentiation provides a sensitive readout of Ras signaling levels. Other signaling pathways, including a Wnt/β-catenin pathway, independently influence vulval fate induction and can also mutate to cause partial Vul or Muv phenotypes(Eisenmann et al., 1998; Gleason et al., 2002)(Fig. 1B).

The sur-6 gene encodes the only PR55/B regulatory subunit of PP2A in C. elegans (Sieburth et al.,1999). Two partial loss-of-function missense alleles of sur-6,ku123 and cs24, do not significantly perturb vulval development,but do strongly enhance the Vul phenotype caused by reducing Ras pathway activity (Sieburth et al.,1999). SUR-6 appears to function between (or in parallel to) Ras and Raf as sur-6 alleles suppress the Muv phenotype caused by activated Ras but not that caused by an activated form of Raf,Torso⁴⁰²¹-D-Raf (Sieburth et al., 1999). By contrast, mutations in the downstream genes mek-2 and mpk-1 efficiently suppress both the activated Ras and Torso-D⁴⁰²¹-Raf Muv phenotypes(Sieburth et al., 1998). Torso-D⁴⁰²¹-Raf contains the extracellular and transmembrane domains of a constitutively dimerizing Torso receptor tyrosine kinase fused to the kinase domain of D-Raf (Dickson et al., 1992; Baek et al.,1996). This fusion protein potentially bypasses multiple steps normally needed for Raf activation, including release of autoinhibition by the Raf N terminus, transport to and stable association with the plasma membrane,and multimerization (Morrison and Cutler,1997). The genetic data are therefore consistent with SUR-6 functioning at any of these steps in Raf activation.

The molecular identity of SUR-6 suggests a role for PP2A in modulating Raf phosphorylation. Indeed, in mammalian cells, PP2A has been proposed to activate Raf by removing inhibitory phosphates from the Raf N terminus(Abraham et al., 2000; Jaumot and Hancock, 2001; Dhillon et al., 2002; Kubicek et al., 2002). Phosphorylation of Raf-1 Ser259 by Akt and/or related serine/threonine kinases inhibits Raf activity (Zimmermann and Moelling, 1999; Rommel et al.,1999; Guan et al.,2000; Zhang et al.,2001). PP2A may dephosphorylate Raf-1 on Ser259 as one step in Raf activation. In support of this model, the PP2A catalytic subunit physically associates with Raf-1 (Abraham et al.,2000), Raf-1 Serine 259 phosphorylation increases upon treatment with the PP2A inhibitor okadaic acid(Abraham et al., 2000), and mutation of Ser259 to Ala in Raf-1 increases Raf kinase activity above basal levels (Michaud et al., 1995; Rommel et al., 1996; Clark et al., 1997). B-Raf and C. elegans LIN-45 RAF may be regulated in a similar manner to Raf-1 because each has multiple consensus Akt phosphorylation sites and mutation of these sites elevates their activities(Chong et al., 2001). Although the above experiments did not address which B regulatory subunit complexes with PP2A to target Raf, the data are consistent with a model in which SUR-6/PR55 and PP2A remove inhibitory phosphates from LIN-45 RAF. This model predicts that the PP2A catalytic subunit should also promote Ras signaling in C. elegans, and that mutating the candidate target sites on LIN-45 RAF should eliminate the requirement for SUR-6.

We analyze the effects of null mutations in sur-6 PR55/B and let-92 PP2A-C and provide support for the model that SUR-6 and PP2A cooperate to promote Raf activity. However, we find that mutating both consensus Akt phosphorylation sites in LIN-45 RAF does not eliminate the requirement for SUR-6. Therefore, SUR-6/PP2A does not act solely by dephosphorylating those inhibitory sites. We also provide genetic evidence that KSR activity is intact in sur-6 mutants, and that the kinase PAR-1 functions antagonistically to SUR-6 and KSR-1 during Ras-mediated vulval induction.

General methods for growing and handling of worms were performed as described (Brenner, 1974). Worms were grown at 20°C unless stated otherwise. The wild-type strain was Bristol N2. Genes, alleles and balancers used are described(Riddle et al., 1997) unless otherwise indicated.

LGI: unc-13(e51), sur-6(ku123) and sur-6(cs24)(Sieburth et al., 1999); mek-2(h294), mek-2(ku114), pry-1(mu38), ksr-2(dx27)(Ohmachi et al., 2002) and hT2[qIs48] (Wang and Kimble,2001).

LGIII: mpk-1(ku1) and unc-119(ed3). LGIV: let-60(n1046gf), let-60(n2031dn), let-60 (sy100dn), lin-45(ku112)(Rocheleau et al., 2002), lin-45(oz166) and lin-45(dx19)(Hsu et al., 2002); sur-8(ku167) (Sieburth et al.,1998), let-92(s504), let-92(s677), unc-22(s7), unc-22(e66),dpy-20(e1282), dpy-20(e1362), him-8(e1489), lip-1(zh15)(Berset et al., 2001), nT1[qIs5].

LGV: him-5 (e1490), let-341(cs41)(Rocheleau et al., 2002), par-1(b274), par-1(zu310ts), rol-4(sc8), unc-76(e911), gaIs36(Lackner and Kim, 1998).

LGX: lon-2(e678), ksr-1(n2526) and sem-5(n1779).

A deletion library of N2 worms mutagenized with ethyl methane sulfonate(Jansen et al., 1997) was screened for deletions in the sur-6 gene. Pooled genomic samples representing a total of 400,000 haploid genomes were used as templates in PCR reactions with two primers with the sequences 5′-CGG AGG ACA GCT GAT AAG TAA GAG GTT C-3′ and 5′-GAT GTA GAG ATT GTT AGT GGC AGC AAG AG-3′. A small amount of each reaction was used as template for a second round of PCR with the primers 5′-GAA GTT CTT CTC TGC GTG ATC GCA TAC-3′ and 5′-GAA GTT GAT CAG ATG AAA GAT CCT CTT CG-3′. The pool of worms containing the sur-6 deletion was thawed and used to establish cultures from individual worms, from which individual heterozygous animals were identified. Sequence analysis showed that the sv30deletion removes 1.8 kb of the sur-6-coding region; it extends from exon 2 to intron 8, eliminating at least five out of the seven WD40 repeats in the SUR-6 protein. PCR experiments confirmed that a wild-type copy of the sur-6 gene was not present elsewhere in the genome.

The sv30 strain was outcrossed by crossing six times with wild type and by selecting for recombinants on LGI. During this procedure, a second mutation that increased the penetrance of the Vul phenotype of the strain was identified and genetically removed.

Vulval development was scored in early to mid fourth larval stage (L4)animals using differential interference contrast (DIC) microscopy. Animals with fewer than 22 vulval descendants (with losses in increments of three or four cells) and greater than six non-vulval descendants were scored as vulvaless (Vul). Animals with more than 22 vulval descendants (with gains in increments of three or four cells) and fewer than six non-vulval descendants were scored as multivulva (Muv). To calculate the number of induced vulval precursor cells (VPCs) each normal lineage was given a value of 1.0 and each partial lineage was given a value of 0.5, so that wild-type animals have a value of 3.0, Vul animals have a value less than 3.0 and Muv animals had a value greater than 3.0.

Embryonic lethality was assessed by allowing hermaphrodites to lay eggs for 12-18 hours and then counting unhatched eggs 24 hours later.

Embryos were allowed to develop on plates until most of them were at a point just before vulval development starts. The larvae were washed and∼200 larvae were mixed in an Eppendorf tube with 1 mg/ml of the appropriate dsRNA and OP50 bacteria at OD₅₉₅ of 1.0 in a 40 μl volume followed by incubation at 20°C with gentle rotation for 24 hours. The larvae were pipetted onto seeded plates and examined for vulval development by differential interference contrast microscopy.

Embryo immunostaining was performed by the freeze/crack method followed by methanol/acetone fixation (Miller and Shakes, 1995). Fixed embryos were incubated at 37°C for 1 hour with 1:100 dilution of YL1/2 rat anti-α-tubulin (Accurate Chemical &Scientific). Samples were washed twice with PBS+2% Tween-20 and incubated at 37°C for 30 minutes with 1:100 dilution of Cy3 conjugated donkey-anti rat IgG (Jackson Immuno Research) and washed three times as before. DAPI(4′,6-diamidino-2-phenylindole) was added to the penultimate wash at 0.5 mg/ml. Excess liquid was wiped off and a coverslip containing 5 μl of mowiol mounting medium was placed over the slide.

Worm lysates from 25-100 L4 animals were separated on 7.5% SDS-PAGE gels and transferred onto Hybond nitrocellulose (Amersham). Blots were probed with antibodies against di-phosphorylated MAPK (MAPK-YT, Sigma, 1:2500 dilution) or total MAPK (K23, Santa Cruz, 1:200 dilution) overnight at 4°C before incubation with horseradish peroxidase-conjugated secondary antibodies(Jackson Immuno Research) for 1 hour at room temperature and exposure to West Pico chemiluminescent substrate (Pierce). Membranes were stripped before reprobing with the second primary antibody.

A 5.2 kb genomic fragment from the F38H4.9 locus containing 3 kb of promoter sequence and 1 kb of predicted 3′UTR was amplified from N2 genomic DNA and cloned into pBluescript II (SK+) as a NotI/KpnI fragment to generate pGK182. Transgenic lines were generated in wild-type animals by co-injecting pGK182 at 10 ng/μl along with pTG96_2 (sur-5::GFP) (Yochem et al., 1998) at 30 ng/μl. Transgenic males were crossed with let-92(s504) unc-22 (s7)/DnT1 hermaphrodites. All GFP (+) Unc-22 progeny grew up to adulthood indicating rescue of the let-92 (s504)larval arrest phenotype. Rescued animals were sterile indicating a role for let-92 in the germline that could not be rescued as most C. elegans transgenes are transcriptionally silent in the germline(Kelly et al., 1997). The sterile phenotype of the rescued animals was similar to the sterile phenotype of wild-type larvae soaked in let-92 dsRNA. Vulva development was normal in the rescued animals.

Point mutations were introduced by PCR into the lin-45 cDNA and the mutated versions were cloned into pPD49.83 (kindly provided by A. Fire) as NheI/NcoI fragments to put them under control of the hsp16-41 promoter (Stringham et al., 1992). Mutagenic primers oMS103 (5′-GAT CGG AGC TCT GCT GCT CCG AAT ATC-3′) and oMS104 (5′-GAT ATT CGG AGC AGC AGA GCT CCG ATC-3′) were used on the lin-45 cDNA clone pRaf107a to change the codon for Ser312 to Ala in pGK167. Mutagenic primers oGK90 (5′-CGT AGT CGA GCG CCA GGC GAA CG-3′) and oGK91 (5′-CGT TCG CCT GGC GCT CGA CTA CG-3′) were then used to change Ser453 to Ala in pGK209. The lin-45 inserts were completely sequenced to verify that the only changes were the desired ones. lin-45(+) cDNA was cloned into pPD49.83 to generate pGK170.

A mixture of pGK167 (100 ng/μl) and pPD#MM016 (unc-119+)(Maduro and Pilgrim, 1995) (30 ng/μl) was injected into unc-119(ed3) animals. Stable extrachromosomal array lines that gave a robust Muv phenotype after heat-shock were kept. One such line was irradiated with 1800 rads of X-rays to integrate the array into the genome. One line carrying the insertion csIs34 on the X chromosome was outcrossed four times with wild type before use in strain construction. pGK167 was injected at 20 ng/μl along with punc-119:GFP at 100 ng/μl to yield csEx2.

pGK209 was injected into N2 animals at 20 ng/μl along with pTG96_2(sur-5::GFP) at 30 ng/μl and pBluescript at 50 ng/μl to yield transgenic lines csEx52 and csEx53. pMS88 containing hsp16-41::torso⁴⁰²¹-Draf(Sieburth et al., 1998) was similarly injected to yield csEx64. Stable transgenes were crossed into the desired genetic backgrounds using standard genetic methods.

Although both previously described sur-6 missense alleles are viable, RNA-mediated interference (RNAi) of sur-6 indicated that sur-6 is an essential gene(Sieburth et al., 1999; Piano et al., 2001). Introduction of sur-6 double-stranded (ds) RNA into wild-type hermaphrodites (such that both maternal and zygotic sur-6 functions were inhibited) resulted in highly penetrant embryonic lethality in the F1 generation. However, introduction of sur-6 dsRNA into RNAi-resistant rde-1 hermaphrodites (Tabara et al., 1999) (such that maternal sur-6 function remained intact) had no effect on rde-1/+ F1 progeny (R. Howard and M.V.S.,unpublished). These RNAi experiments suggested that sur-6 is required maternally but not zygotically for viability, and that sur-6 null alleles might therefore cause maternal-effect embryonic lethality.

To further characterize sur-6, we generated a deletion mutation, sv30 (Materials and methods). As predicted for a null allele, sv30 causes maternal effect embryonic lethality(Table 1). This embryonic lethality could not be rescued by mating sv30 hermaphrodites with wild-type males (data not shown). The deficiency qDf8, which removes sur-6, fails to complement sv30 for the maternal effect lethal phenotype (Table 1). In addition, the sv30/qDf8 phenotype was not more severe than that of sv30/sv30 homozygous animals(Table 1) supporting the notion that sv30 is a genetic null. The sur-6(cs24) allele complements sv30 for the maternal effect lethal phenotype(Table 1), indicating that cs24 does not significantly perturb the essential function of sur-6 (see Discussion).

We examined sur-6(sv30) and sur-6(RNAi) embryos to determine the cause of lethality. Both types of embryos appeared similar and exhibited spindle defects and abnormal timing of the earliest cell divisions(Fig. 2). In wild-type embryos,the zygote divides asymmetrically to generate a larger anterior blastomere(AB) and a smaller posterior blastomere (P1). AB then divides symmetrically to generate ABa and ABp (Fig. 2A),and shortly thereafter P1 divides asymmetrically to generate EMS and P2(Fig. 2C). In sur-6(sv30) embryos, P1 often initiated division prior to AB(Fig. 2D, 4/11 embryos),spindles appeared morphologically abnormal or collapsed during anaphase(Fig. 2B, 4/11 embryos), and chromatin bridges (Fig. 2H) and supernumerary centrosomes (Fig. 2J) were observed by DAPI and anti-tubulin staining, respectively. These defects suggest a general requirement for sur-6 in mitotic progression during embryogenesis.

Maternally rescued sur-6(sv30) homozygotes do not, however, show obvious cell cycle defects. They have normal gonadal and germline morphology and are fertile, but are mildly uncoordinated (Unc) and have a weakly penetrant Vul defect that resembles the Vul defect of sur-6(cs24)missense mutants (Table 1, Fig. 3B). The relatively weak Vul phenotype could be explained in part by perdurance of maternally provided gene product. However, sur-6(sv30) homozygotes from sur-6(sv30)/sur-6(cs24) mothers have only marginally more severe Vul defects than sur-6(sv30) homozygotes from sur-6(sv30)/+mothers (Table 1). Therefore,it appears that sur-6 promotes but is not absolutely essential for vulval induction.

In the context of Ras pathway activity, sur-6(sv30) behaves similarly to the previously described sur-6 missense alleles. The sur-6(sv30) Vul phenotype is strongly enhanced by hypomorphic alleles of lin-45 raf or the Raf regulator sur-8(Table 2A). sur-6(sv30) also dominantly suppresses the Muv phenotype caused by an activated allele of let-60 ras, but does not suppress the Muv phenotypes caused by activated torso⁴⁰²¹-D-raf or by activated forms of MEK and ERK (Table 2B, Fig. 3C,D). We also found that sur-6(sv30) fails to suppress the Muv phenotype caused by loss of pry-1 Axin, a negative regulator of bar-1β-catenin (Gleason et al.,2002; Korswagen et al.,2002), confirming that sur-6 specifically affects Ras signaling and not Wnt signaling (Table 2B). Finally, western blot analyses using a monoclonal antibody specific for the dually phosphorylated active form of ERK revealed a fivefold decrease in MPK-1 ERK phosphorylation in sur-6(sv30) mutants(Fig. 4). These data support the model that sur-6 promotes Ras signaling and acts upstream of Raf.

We next wanted to test the relationship between SUR-6 and the PP2A catalytic core. F38H4.9 encodes the single catalytic subunit of PP2A (PP2A-C)in C. elegans. Introduction of F38H4.9 dsRNA into wild-type hermaphrodites resulted in embryonic lethality in the F1 generation(Sieburth et al., 1999; Kamath et al., 2003), whereas introduction of F38H4.9 dsRNA into rde-1 hermaphrodites resulted in early larval lethality in the rde-1/+ F1 progeny (R. Howard and M.V.S., unpublished). These RNAi experiments suggested that PP2A-C is required both maternally and zygotically for viability, and that PP2A-C mutations might cause larval lethality. The cosmid B0033 and a plasmid containing the F38H4.9 gene can rescue the larval lethality of let-92 mutants (see Materials and methods), suggesting that PP2A-C corresponds to the genetic locus let-92 (Rogalski and Baillie,1985). This assignment has been independently established(Ogura et al., 2003). We henceforth refer to F38H4.9/PP2A-C as let-92.

let-92(RNAi) embryos showed catastrophic failures in cytokinesis during the earliest cell divisions (data not shown). These defects are more severe than those seen in sur-6 mutant embryos, consistent with studies in yeast and mammals that suggest the catalytic subunit acts in concert with multiple regulatory subunits to participate in distinct cellular and developmental events (Janssens and Goris, 2001). Although the sur-6 and let-92embryonic arrest phenotypes are distinct, a let-92 loss-of-function mutation shows strong dominant synthetic lethal interactions with both viable sur-6 missense alleles (Table 3), suggesting that sur-6 and let-92 do function together during embryogenesis.

The different regulatory subunits of PP2A can either promote or inhibit the catalytic activity of PP2A towards different substrates, and both cooperative and antagonistic relationships between different PP2A subunits have been documented in S. cerevisiae and Drosophila (van Wyl et al.,1992; Maixner et al., 1998; Janssens and Goris, 2001). We sought to determine if sur-6 promotes or inhibits PP2A activity towards a Ras pathway substrate by examining the effect of reducing let-92 activity. If sur-6 promotes PP2A activity then reducing let-92 should cause a Vul phenotype similar to that caused by sur-6(sv30), whereas if sur-6 inhibits PP2A activity then reducing let-92 should have the opposite effect.

Our previous experiments suggested a positive role for let-92because, like sur-6(RNAi), let-92(RNAi) could partially suppress the let-60(gf) Muv phenotype(Sieburth et al., 1999). We also found that let-92(RNAi) and sur-6(RNAi) caused similar weak synthetic Vul phenotypes in sur-8 mutant larvae(Table 4), further supporting a positive role for let-92.

Homozygous let-92 mutants die as very young larvae(Rogalski and Baillie, 1985),before vulval development starts, preventing us from directly examining vulval development in these strains. Instead, we tested let-92 alleles for dominant genetic interactions with other Ras pathway components. let-92/+ did not modify the let-60(gf)/+ Muv phenotype(Table 4). However, let-92/+ strongly enhanced the penetrance and expressivity of the Vul defects caused by two different dominant-negative alleles of let-60 ras (Table 4). This effect of let-92 was similar to that of a null allele of lin-45 raf(Table 4), consistent with let-92 mutations causing a reduction in Raf activity. let-92/+ did not affect the vulval defects of partial loss-of-function alleles of other Ras pathway genes such as sem-5/Grb2, let-341/Sos, mek-2/MEK, mpk-1/ERK or lip-1/MKP (data not shown, see Materials and methods for alleles used). Therefore, sensitivity to reduced let-92dosage may be specific for the Ras/Raf level of the signal transduction pathway.

Based on these experiments, we conclude that let-92, like sur-6, plays a positive role in vulval fate induction. We find no evidence for a negative role. Thus, sur-6 most probably promotes PP2A activity toward a Ras pathway substrate. If so, SUR-6/PP2A either removes inhibitory phosphates from a positively acting substrate or removes stimulatory phosphates from a negatively acting substrate.

We tested whether SUR-6/PP2A acts by removing inhibitory phosphates from LIN-45 RAF. Mammalian Raf-1 has a single consensus Akt phosphorylation site(Serine 259) that is subject to inhibitory phosphorylation and that PP2A has been proposed to dephosphorylate (Chong et al., 2003). C. elegans LIN-45 RAF has two consensus Akt phosphorylation sites (Ser312 and Ser453) that appear to function analogously to the single Raf-1 Ser259 site (Chong et al., 2001). We mutated both serines to alanine and expressed the mutant LIN-45 proteins under the control of the heat shock promoter hsp16-41 (Stringham et al.,1992). When overexpressed in this manner, LIN-45⁺ had no apparent effect (data not shown). However, LIN-45^S312A caused a moderate Muv phenotype and LIN-45^{S312A S453A} caused a somewhat stronger Muv phenotype (Table 2C, Fig. 3E). These results are similar to those reported previously(Chong et al., 2001), except that those authors saw a Muv phenotype with LIN-45^{S312A S453A} but not LIN-45^S312A, possibly because of lower expression levels. Taken together, our results are consistent with the model that both Ser312 and Ser453 in LIN-45 RAF are sites of inhibitory phosphorylation.

If Ser312 and Ser453 in LIN-45 RAF are the relevant targets of SUR-6/PP2A during vulval development, then mutation of those serines to alanine should eliminate the need for sur-6. However, the Muv phenotypes of LIN-45^S312A and LIN-45^{S312A S453A} still required sur-6 (Table 2C, Fig. 4E,F). This requirement for sur-6 cannot be explained by an effect on endogenous LIN-45⁺ as LIN-45^{S312A S453A} produced a potent Muv phenotype even in a strong loss-of-function lin-45(dx19lf) mutant background (Table 2C). Because removing both inhibitory sites did not eliminate the requirement for sur-6, SUR-6 must promote LIN-45 RAF activity via a mechanism distinct from dephosphorylating those sites. SUR-6/PP2A may regulate LIN-45 RAF through as yet unidentified phosphorylation sites, or it may regulate LIN-45 RAF indirectly by targeting other Raf regulatory proteins.

The putative scaffold protein KSR is a positive Raf regulator whose function can be inhibited by phosphorylation(Muller et al., 2001), making it another candidate PP2A substrate. Furthermore, KSR and SUR-6 appear to act at a similar step of Raf activation, as ksr-1 also suppresses the Muv phenotype caused by LIN-45^{S312A S453A} but not Torso⁴⁰²¹-Draf (Table 2B,C) (Sieburth et al.,1999). C. elegans has two ksr genes, ksr-1 and ksr-2, that are redundantly required for viability and vulval development (Ohmachi et al.,2002). Although ksr-1 and ksr-2 single mutants are viable and have normal vulvae, the mutants are very sensitive to further reductions in KSR activity (Ohmachi et al., 2002). Therefore, if sur-6 mutations reduce KSR activity, we would expect to see a strong genetic interaction between a sur-6 null mutation and ksr-1 or ksr-2. Contrary to this prediction, ksr-1 and ksr-2 mutations each failed to enhance sur-6(sv30) Vul defects(Table 2A). These results suggests that KSR activity is relatively intact in sur-6 mutants. Therefore, we do not favor the model that KSR is a key substrate of SUR-6/PP2A.

The serine/threonine kinase C-TAK1/PAR-1 has been identified biochemically as an inhibitor of mammalian KSR (Muller et al., 2001). Most C. elegans par-1 mutants have wild-type vulval fate specification (Hurd and Kemphues, 2003). However, we find that par-1(b274lf)/par-1(zu310ts) transheterozygotes (in which both maternal and zygotic par-1 activities are reduced) are weakly Muv(Table 5). Additionally we find that reducing par-1 function strongly reverts the suppressed(non-Muv) phenotype of sur-6(ku123);let-60(n1046gf) and let-60(n1046gf);ksr-1(n2526) double mutants(Table 5). By contrast,reducing par-1 function does not revert the suppressed (non-Muv)phenotype of lin-45(ku112) let-60(n1046gf). Therefore par-1has an inhibitory role in vulval development, probably acts upstream or parallel to lin-45 raf, and functions antagonistically to sur-6/PP2A and ksr-1.

Our results suggest that SUR-6, the PR55/B regulatory subunit of PP2A, and LET-92, the catalytic subunit of PP2A, cooperate to positively regulate mitotic progression and Ras signaling in C. elegans. Although we have been unable to use epistasis analysis to place LET-92/PP2A at a particular step of the Ras pathway, the simplest model is that SUR-6/PR55 and LET-92/PP2A act at the same step. Epistasis analysis using a sur-6 null allele supports the model that SUR-6 acts upstream of LIN-45 RAF, but is inconsistent with the model that SUR-6 and PP2A act solely by dephosphorylating inhibitory Akt sites on LIN-45 RAF. Instead, SUR-6 and PP2A may regulate Raf indirectly by influencing another Raf regulatory protein.

In other systems, PR55/B and PP2A have been found to have both positive and negative effects on Ras signaling. For example, in Drosophila a positive role for PR55/PP2A is supported by findings that mutations in tws/PR55 suppress the lethality caused by activated sevenless receptor and activated Ras (Maixner et al., 1998), and mutations in the PP2A catalytic subunit enhance photoreceptor defects caused by a hypomorphic Draf allele(Wassarman et al., 1996). However, a negative role for PR55/PP2A is supported by findings that RNAi against tws/PR55 elevates the level of phospho-ERK in cultured S2 cells (Silverstein et al.,2002), and mutations in the PP2A catalytic subunit enhance photoreceptor defects caused by activated Ras(Wassarman et al., 1996). Thus, in Drosophila the role of PR55/PP2A appears complex, and PP2A may act on multiple substrates within the Ras pathway. Similarly, in mammalian cells PP2A has been suggested to positively regulate Ras signaling by removing inhibitory phosphates from Raf (Abraham et al., 2000; Jaumot and Hancock,2001; Dhillon et al.,2002; Kubicek et al.,2002) and to negatively regulate Ras signaling by removing activating phosphates from MEK or ERK(Alessi et al., 1995). By contrast, we find no evidence for a negative role of SUR-6/PR55 or LET-92/PP2A in C. elegans, despite having tested sur-6 and let-92 mutations in numerous genetic backgrounds. Therefore either PP2A lacks a negative role in C. elegans, or its negative role is masked by its stronger positive role.

Prior studies of SUR-6 and its genetic placement between (or in parallel to) Ras and Raf relied on partial loss-of-function sur-6 alleles(Sieburth et al., 1999). Our characterization of a sur-6 null mutation, sv30, is consistent with the prior studies but clarifies several previously unresolved points. First, our data suggest that SUR-6 promotes Ras signaling but is not absolutely essential for Ras signaling under normal circumstances. Therefore,SUR-6/PP2A may dephosphorylate a site that has modest effects on substrate activity, or SUR-6/PP2A function may be partly redundant with that of another phosphatase complex. Second, our finding that sur-6(sv30) reduces the levels of activated MPK-1 ERK in vivo argues that SUR-6 acts upstream of rather than in parallel to MEK and ERK. SUR-6 could still act either upstream or in parallel to LIN-45 RAF. Finally, our data dispute two of the prevailing models for SUR-6 function.

Like mammalian Raf proteins, LIN-45 RAF appears to be inhibited by phosphorylation on sites that match the consensus sequence for the Akt kinase(Chong et al., 2001). To date we have not seen effects of akt-1 or akt-2 RNAi on Ras signaling (G.K. and M.V.S., unpublished), so it is unclear whether Akt itself or some other kinase(s) normally phosphorylates these LIN-45 RAF inhibitory sites. Although a simple and attractive model was that SUR-6 and PP2A dephosphorylate these LIN-45 RAF inhibitory sites, our data are inconsistent with that being their sole mechanism of action. We found that sur-6(sv30) suppresses the Muv phenotype caused by LIN-45^{S312A S453A}, which lacks both presumptive inhibitory Akt phosphorylation sites, indicating that SUR-6 must promote LIN-45 RAF activity independently of those sites. One possibility is that SUR-6/PR55 and PP2A dephosphorylate LIN-45 RAF on other inhibitory sites; however, no such sites have been identified as yet. An alternative possibility is that SUR-6 and PP2A indirectly influence LIN-45 RAF by dephosphorylating some other Raf regulatory protein(s).

A second proposed model was that SUR-6 and PP2A regulate the scaffold protein KSR (Sieburth et al.,1999; Ory et al.,2003). KSR may regulate Raf at a similar step as SUR-6, as ksr-1 mutations also suppress the Muv phenotype caused by LIN-45^{S312A S453A} but not Torso⁴⁰²¹-Draf. However, as discussed above, the failure of sur-6 mutations to genetically interact with ksr-1 or ksr-2 suggests that KSR activity is relatively intact in sur-6 mutants. Therefore, KSR is unlikely to be the sole SUR-6/PP2A substrate. It remains possible that SUR-6/PP2A has multiple substrates, and that LIN-45 RAF and/or KSR are among these, but our data argue that another important substrate(s) remains to be identified.

Murine C-TAK/PAR-1 has been suggested to phosphorylate KSR to modulate KSR localization (Muller et al.,2001). The relevant phosphorylation site is not conserved in C. elegans KSR-1 or KSR-2, although related sites are present elsewhere. PAR-1 can also phosphorylate Raf as well as other substrates(Benton et al., 2002) and therefore could have broader roles in Ras signaling. C. elegans par-1has been previously found to play a role in vulval morphogenesis(Hurd and Kemphues, 2003). We found that par-1 also plays an inhibitory role in vulval fate specification. This inhibitory role of par-1 is partly masked by perdurance of maternally provided gene product, but could be seen in animals in which both maternal and zygotic par-1 contributions were diminished, as well as in some sensitized genetic backgrounds. We found that zygotic removal of par-1 reverts the suppressor of ras (gf)phenotypes of sur-6 and ksr-1 but not lin-45 raf,consistent with models where PAR-1 acts on KSR, LIN-45 RAF or both to inhibit vulval development. Our results also raise the possibility that SUR-6/PP2A inhibits PAR-1 to indirectly affect LIN-45 RAF activity.

The sur-6 maternal effect lethal phenotype reveals that in addition to Ras signaling, sur-6 is required for mitotic progression. sur-6(sv30) and sur-6(RNAi) embryos display a variety of mitotic defects such as ectopic and aberrant cytokinesis, the collapse and re-elaboration of well-extended anaphase spindles, abnormally shaped spindles and chromatin bridges during anaphase. Similar mitotic defects have been observed in Drosophila tws/PR55 mutants(Gomes et al., 1993; Mayer-Jaekel et al., 1993). Premature sister chromatid separation and cytokinesis defects have also been observed in S. cerevisiae cdc55/PR55 mutants(Minshull et al., 1996; Wang and Burke, 1997). Thus,the mitotic role of PR55 appears to be evolutionarily conserved. The early C. elegans embryo is a particularly tractable system for further study of this poorly understood mitotic role of PR55.

Interestingly, the original two missense alleles of sur-6, ku123and cs24, behave similarly to the sur-6 null allele sv30 with respect to Ras signaling but do not cause mitotic defects or embryonic lethality even when placed in trans to sur-6(sv30) or a deficiency of the sur-6 locus as shown here and previously(Sieburth et al., 1999). Thus,the functions of sur-6 in Ras signaling and mitotic progression are genetically separable. The cs24 allele causes an E to K change at an absolutely conserved site of the third WD repeat of SUR-6/PR55(Sieburth et al., 1999). Equivalent mutations in the human PR55 protein have been shown to severely compromise binding of PR55 to the A subunit of PP2A, preventing PR55 association with the catalytic core(Strack et al., 2002). Thus, cs24 would be predicted to similarly compromise the interaction of SUR-6 with the PP2A core. As cs24 is nearly wild type for mitotic function but nearly null for Ras signaling function, one possible model is that SUR-6 acts independently of the A and C subunits during mitosis but acts with the A and C subunits during Ras signaling. An alternative model is that SUR-6 acts with the A and C subunits during both processes, but that only very low levels of SUR-6/PP2A are required for mitotic function. The latter model is supported by the fact that maternally rescued sur-6(sv30) mutants show few or no mitotic defects during larval development (presumably owing to perdurance of low levels of maternal product), and by the fact that reducing the let-92/PP2A dose by half causes synthetic embryonic lethality in both sur-6 missense backgrounds. Thus, there appear to be different threshold requirements for the various roles of SUR-6-associated PP2A, with Raf activation being most sensitive to reductions in sur-6activity.

In conclusion, we have shown that SUR-6/PR55 promotes mitotic progression during embryogenesis and Raf activation during vulval induction and that it cooperates with LET-92/PP2A-C during both processes. Therefore, SUR-6/PP2A probably dephosphorylates mitotic and Ras pathway substrates. Thus far, the identification of PP2A targets has been hampered by the very broad specificity of PP2A in vitro and the widespread requirement for PP2A in vivo. The fact that sur-6 null mutants show specific phenotypes makes C. elegans a useful genetic model system in which to test candidate PP2A targets and identify those that are functionally relevant.

We are very grateful to A. Eriksson for help in isolating sur-6(sv30); C. Rocheleau for help with embryological techniques; S. Cibulsky for help with densitometry; D. Garbe, J. Doto and E. Gaskin for excellent technical assistance; and K. Ogura and Y. Kohara for sharing unpublished information. We also thank the Caenorhabditis Genetics Center, K. Siegfried and J. Kimble for providing strains; Y. Kohara for EST clones; A. Fire for vectors; and G. Deshpande, R. Howard, D. Levitan, R. Lints, D. Raizen, C. Rocheleau and J. Yochem for stimulating discussions and comments on the manuscript. This work was supported by a NSERC grant to D.B. and NIH grant CA87512 to M.V.S. Work in S.T.'s laboratory is supported by grants from Cancerfonden and the Wallenberg Foundation.