The C. elegans Chp/Wrch Ortholog CHW-1 Contributes to LIN-18/Ryk and LIN-17/Frizzled Signaling in Cell Polarity

Wnt signaling controls various aspects of developmental and cell biology, as well as contributing to certain cancers. Expression of the human Rho family small GTPase Wrch/RhoU is regulated by Wnt signaling, and Wrch and its paralog Chp/RhoV are both implicated in oncogenic transformation and regulation of cytoskeletal dynamics. We performed developmental genetic analysis of the single Caenorhabditis elegans ortholog of Chp and Wrch, CHW-1. Using a transgenic assay of the distal tip cell migration, we found that wild-type CHW-1 is likely to be partially constitutively active and that we can alter ectopic CHW-1-dependent migration phenotypes with mutations predicted to increase or decrease intrinsic GTP hydrolysis rate. The vulval P7.p polarity decision balances multiple antagonistic Wnt signals, and also uses different types of Wnt signaling. Previously described cooperative Wnt receptors LIN-17/Frizzled and LIN-18/Ryk orient P7.p posteriorly, with LIN-17/Fz contributing approximately two-thirds of polarizing activity. CHW-1 deletion appears to equalize the contributions of these two receptors. We hypothesize that CHW-1 increases LIN-17/Fz activity at the expense of LIN-18/Ryk, thus making the contribution of these signals unequal. For P7.p to polarize correctly and form a proper vulva, LIN-17/Fz and LIN-18/Ryk antagonize other Wnt transmembrane systems VANG-1/VanGogh and CAM-1/Ror. Our genetic data suggest that LIN-17/Fz represses both VANG-1/VanGogh and CAM-1/Ror, while LIN-18/Ryk represses only VANG-1. These data expand our knowledge of a sophisticated signaling network to control P7.p polarity, and suggests that CHW-1 can alter ligand gradients or receptor priorities in the system.


Introduction
Rho family small GTPases are intracellular signaling molecules that regulate cytoskeletal rearrangements and transcription, and which affect diverse cellular processes including cell adhesion, polarity, and migration [1]. Historically, the large majority of studies concerning Rho family GTPases have focused on the canonical members RhoA, Rac1, and Cdc42. However, other family members have been identified by strong sequence conservation, and many of them have unique functions [2]. Two of the less studied Rho family GTPases are the closely related Cdc42 subfamily proteins Wnt-regulated Cdc42 homolog-1 (Wrch-1/RhoU) and Cdc42 homologous protein (Chp/Wrch-2/RhoV). Wrch-1 was first identified as a gene whose expression increased when the Wnt signaling pathway was activated [3]. Wrch-1 is thought to be a key Wnt target in oncogenesis, as over-expression of Wnt-1 increases Wrch-1 expression and expression of activated Wrch-1 caused transformation similar to Wnt transformation [3,4]. Wnt-dependent expression of Wrch-1 does not require βcatenin, but does require c-Jun N-terminal kinase (JNK), implicating the non-canonical Wnt/Planar Cell Polarity (PCP) pathway [3,5].
The Wnt/PCP pathway guides developing epithelial tissues in orienting their cell divisions in the plane of the epithelium, and thus Wnt/PCP is critical for normal animal development [6,7]. Misregulation of the Wnt/PCP pathway is linked to cancer development and progression [8,9]. While the Wnt/PCP pathway is known to utilize Rho family GTPases, including Cdc42, as effectors [10][11][12][13], the role of Wrch-1 is unknown. However, a possible role of Wrch-1 in this pathway is consistent with its previously described role in epithelial apical-basal polarity, where Wrch-1 is asymmetrically distributed and binds as an effector the antero-posterior (A-P) and Planar polarity regulating protein Par6 [14,15].
The Wnt/PCP signaling pathway in C. elegans is implicated largely in anterior-posterior axon guidance and neuronal polarity [16]. Wnt/PCP components VANG-1/Van Gogh, PRKL-1/prickle, FMI-1/flamingo and DSH-1/dishevelled have been shown to function in neuronal polarity and neurite guidance [17,18], but have also been implicated in other polarity-based events [19,20]. Additionally, the polarization of cells during development of the vulval structure may involve an analogous mechanism [21]. The vulva develops from the invariant divisions of three vulval precursor cells (VPCs), P5.p, P6.p, and P7.p (Fig 1A; reviewed in [22]). The combination of Wnt, EGF, and Notch signaling pathways instruct P6.p to adopt the 1°cell fate and P5.p and P7.p to adopt the 2°cell fate. The presumptive 1°cell (P6.p) undergoes three cell divisions to generate an A-P symmetrical lineal group of eight cells that forms the central third of the vulva. In contrast, the flanking P5.p and P7.p undergo three divisions to form A-Pasymmetrical lineal groups of seven cells each that form the anterior and posterior thirds of the vulva, respectively. Importantly, P5.p and P7.p must be polarized in opposite directions, with mirror symmetry centered on the non-polarized central P6.p, to generate a functional vulva.
Redundant actions of diverse Wnt ligands (LIN-44, MOM-2 and CWN-2) also contribute to this process [28]. The phenotype was named P-Rvl (posterior-reversed vulval lineage). Interestingly, loss of two other Wnt homologs, EGL-20 and CWN-1, and orthologs of the PCP components Ror (CAM-1) and Van Gogh (VANG-1) suppress P-Rvl defects caused by removing LIN-17/Fz. Sophisticated control of multiple Wnt ligand gradients has a strong impact on P7.p polarization as well as contribution to vulval induction and competence [29,30]. In the absence of all Wnt activity, all P7.p cells exhibit the P-Rvl phenotype, revealing that removal of a redundant Wnt signal that confers vulval-oriented polarity reveals an underlying signal that confers posterior-oriented polarity. Further analyses suggest that the posterior-oriented polarity system is governed by mechanistically dissimilar Wnt signaling. [21].
These data suggest a model in which distinct Wnt signals control P5.p and P7.p polarization ( Fig 1B) [21]. The first pathway, termed the "ground polarity" signal, polarizes both P5.p and P7.p towards the posterior, and utilizes Wnt/EGL-20, Wnt/CWN-1, Ror/CAM-1, and Van Gogh/VANG-1. A second pathway, termed the "refined polarity" signal, polarizes both P5.p and P7.p towards the centrally located P6.p along a (P-Rvl)-distal (P-D) axis, and utilizes a Wnt signaling pathway involving ligands Wnt/LIN-44 and Wnt/MOM-2 and receptors Fz/ LIN-17 and Ryk/LIN-18 [25,28]. For the sake of clarity, in this study we will refer to "ground (posterior) polarity" and "refined (central) polarity." Thus, in the normally posteriorly oriented P5.p, ground (posterior) polarity and refined (central) polarity Wnt signals collaborate redundantly to promote the same polarity outcome, and A-P and P-D axes are aligned with the invaginated portion of the 2°lineage oriented towards the 1°lineage, posteriorly. Consequently when single components of either ground (posterior) polarity or refined (central) polarity are lost there are no P5.p A-Rvl phenotypes, because each polarity system maintains P5.p posterior orientation in the absence of the other. In contrast, in P7.p the ground (posterior) polarity and refined (central) polarity Wnt pathways act in opposition; the refined (central) polarity pathway specifying P-D polarity completely overrides the ground (posterior) polarity pathway with high fidelity to orient P7.p anteriorly, towards the 1°lineage and the center of the developing vulva ( Fig 1C). Consequently, loss of refined polarity components results in P7.p polarity reversal; loss of either LIN-17/Fz or LIN-18/Ryk displays a partially penetrant P-Rvl phenotype, while loss of both results in 100% P-Rvl [25], suggesting that it is collaboration of LIN-17/Fz and LIN-18/Ryk that interprets external Wnt signals and correctly orients P7.p [21].
Here, we characterize the role of the gene chw-1, which encodes the sole C. elegans ortholog of Chp and Wrch-1. Using an in vivo cell migration assay, we present evidence that wild-type CHW-1 signaling is partially constitutively active, unlike most Rho family members, whose activities are tightly regulated. Therefore, transcriptional regulation of CHW-1 may suffice to activate CHW-1-dependent pathways. Using genetic epistasis analysis we show that CHW-1 affects LIN-17/Fz and LIN-18/Ryk differently; loss of chw-1 suppresses the lin-17 P-Rvl phenotype, but enhances the lin-18 P-Rvl phenotype, a pattern not yet described for any other proteins functioning in this system. The contributions of LIN-17/Fz and LIN-18/Ryk to P7.p polarity are unequal, with LIN-17/Fz responsible for roughly two-thirds of the polarizing activity and LIN-18/Ryk for one-third. But in the absence of CHW-1, LIN-17/Fz and LIN-18/Ryk contributions are presumed to be equal. The ground (posterior) polarity receptors CAM-1/Ror and VANG-1/Van Gogh also have genetically distinguishable pathway interactions: loss of CAM-1 suppresses lin-17 but not lin-18 mutations, while VANG-1 loss suppresses both lin-17 and lin-18 P-Rvl defects. A chw-1 promoter::GFP fusion construct is excluded from the three VPCs assuming vulval fates, including the polarized P5.p and P7.p, but is expressed in uninduced VPCs. We hypothesize that CHW-1 non-autonomously promotes the contribution of LIN-17/Fz to P7.p polarity at the expense of LIN-18/Ryk contribution.

CHW-1 is the C. elegans ortholog of human Chp/Wrch
The C. elegans genome contains a single gene, F22E12.2, that encodes a protein similar to human Rho family small GTPases Chp/RhoV and Wrch-1/RhoU (Fig 2A-2C). We have named this gene chw-1 (Chp and Wrch). By amino acid identity, the GTPase domain of CHW-1 is most similar to that of Wrch-1 ( Fig 2B). Furthermore, the core effector-binding motifs of CHW-1 and Wrch-1 are nearly identical, while that of Chp has notable differences (Fig 2A). These differences may be reflected by CHW-1 effector selectivity resembling that of Wrch-1 as opposed to Chp. While the CHW-1 GTPase domain is highly conserved, the N-and C-termini are divergent (Fig 2A). First, the N-terminal extension found in Wrch-1 and Chp, which functions to auto-inhibit the proteins (Fig 2A) [31,32], is mostly absent in CHW-1. Second, whereas Rho family GTPase C-termini generally terminate in CAAX signals for prenyl lipid modifications essential for membrane targeting and protein function [33,34], Wrch-1 and Chp lack any CAAX motif but terminate instead in CXX motifs that specify palmitoylation and support their membrane targeting and biological activity [4,31]. However, the CHW-1 protein lacks CAAX, CXX or other lipid modification motifs, suggesting that CHW-1 does not require membrane targeting for its function.
Native CHW-1 protein may be partially constitutive active Given these differences, we first assayed CHW-1 activity by testing whether ectopic CHW-1 expression imposed any biological defects. Migration of gonadal distal tip cells (DTCs) is regulated by Rho family GTPases [35,36], and we have shown that ectopic mutationally activated CED-10/Rac or CDC-42/Cdc42 were sufficient to perturb migration [37], while control expression of wild-type CED-10/Rac and CDC-42 were phenotypically inconsequential. For these assays we used the lag-2 promoter [38] to drive expression of CHW-1 or GFP in DTCs and evaluated disrupted DTC migration as a readout of CHW-1 activity. We found that expression of wild-type CHW-1 significantly increased the frequency of errant DTC migration, as compared to expression of GFP ( Fig 2D). Wild-type CHW-1 proteins from C. elegans, C. briggsae and C. remanei contain an alanine at position 18, analogous to position 12 in most Ras and Rho small GTPase superfamily members (Fig 2A; S1 Fig). In Wrch-1 and most Rho family GTPases, the wild type residue at this position is a glycine. Mutation of this glycine in small GTPases generally disrupts intrinsic GTPase activity/GAP interaction, thus locking the protein in the GTP-bound state, derepressing the protein and activating downstream signaling. Expression of CHW-1(A18G) abolished the ability of ectopic CHW-1 to cause DTC migration defects, suggesting that the wild-type CHW-1 protein is at least partially active. In contrast, introduction of theoretically stronger activating mutations in residues required for GTPase function [39], A18V and Q69L, disrupted DTC migration more than the wild-type protein ( Fig 2D).
Observed gonadal tube-shaped morphology was normal, but the DTC migration was aberrant, as we previously described for defects in CED-10/Rac, and PAK-1 or MAX-2 Pak mutant combinations or ectopically expressed mutationally activated CED-10/Rac and CDC-42 [37]. In conjunction with this prior study, we hypothesize thatectopically expressing mutationally activated CED-10/Rac, CDC-42 or CHW-1 disrupts spatial recognition of guidance cues, but expression of a non-activated GTPase is neutral, due to the tight regulation of the GTP-GDP cycle typical of Rho family small GTPases [39]. Deletion of chw-1 (see below) caused no DTC migration defects, nor did we observe P chw-1 ::gfp expression in the DTCs or other parts of the gonad. Though not an in vitro assay of GTP loading of CHW-1, this assay has the benefits of being an in vivo assessment of intrinsic CHW-1 activity, particularly since the experiment is controlled with expression of wild-type and A18G mutant proteins, the latter of which compares with similar ectopic expression of wild-type CED-10/Rac and CDC-42 [37]. We hypothesize that the presence of alanine 18 in wild-type CHW-1 confers partially compromised GTP hydrolysis, and hence CHW-1 may be partially activated in the absence of stimulation by GEFs. Thus, expression of CHW-1 may be sufficient to activate CHW-1 effectors. But the ability to mutate the protein to stronger activation argues that CHW-1 retains some GTP hydrolysis capability and negative regulatory input, presumably from a GTPase activating protein (GAP).

Loss of CHW-1 suppresses the lin-17 P-Rvl mutant phenotype
Wnt/PCP signaling induces expression of Wrch-1 in vertebrates [5]. Therefore, we hypothesized that CHW-1 expression is likely to function in the Wnt-dependent polarization of vulval precursor cells that bears some resemblance to PCP. However, using the ok697 deletion mutation that removes most of the chw-1 coding sequences or RNA interference (RNAi) targeting the chw-1 genomic region, we detected no vulval polarity defects (Table 1). In double mutant strains, chw-1(ok967) suppressed by one-third the P-Rvl defect conferred by two different lin-17 alleles (Table 1; S2 Fig). The suppression of the lin-17(0) P-Rvl phenotype by chw-1(ok697) was similar in strength to the suppression of lin-17(0) by vang-1(0) (VanGogh) and cam-1(0) (Ryk), but weaker than suppression of lin-17(0) by egl-20(n585) (Wnt) [21]. chw-1(RNAi) corroborated the chw-1(ok697) suppression of the lin-17(n671) P-Rvl defect, while control gfpdirected RNAi was neutral. CHW-1 is the C. elegans ortholog of human Wrch-1/RhoU and Chp/Wrch-2/RhoV. A. Sequence alignment of CHW-1, human Wrch-1, and human Chp. Identical residues have a black background, conservative changes have a gray background, and non-conservative changes have a white background. CHW-1 lacks the N-terminal extension found in Wrch-1 and Chp (gray bracket). CHW-1 also contains an atypical residue (alanine; arrowhead) at position 18 (analogous to position 12 in Cdc42 and most Rho and Ras family members). The core effector domain (black bracket) is most similar to that of Wrch-1, with one conservative variant (V!I) between CHW-1 and Wrch-1. Indicated with a gray caret are the atypical A residue at position 18 and the typical Q residue at position 69, both of which we mutated for our DTC locomotion studies in panel D. B. An identity and similarity comparison of the GTPase-domain sequences of CHW-1 with Wrch-1, Chp and Cdc42. C. Gene structure of chw-1 and an overlapping gene prediction (F22E12.3). Below, sequences deleted in chw-1 (ok697) and sequences used for feeding RNAi are indicated. D. An assessment of CHW-1 activity by ectopic expression in DTCs. The lag-2 promoter was used to drive ectopic CHW-1 expression in DTCs and migration defects were analyzed by DIC microscopy. A total of 200 DTC migrations were analyzed for each construct. Expression of wild-type CHW-1 caused significantly more frequent migration defects than did expression of GFP or CHW-1(A18G). Expression of CHW-1(A18V) or CHW-1(Q61L) caused significantly more DTC migration defects than did expression of WT CHW-1. Tests of statistical significance were performed using Fisher's exact test.

Loss of CHW-1 enhances the lin-18 P-Rvl mutant phenotype
Our observed lin-17 genetic interactions with chw-1 were consistent with CHW-1 functioning with EGL-20/Wnt, VANG-1/VanGogh, and CAM-1/Ror to promote refined (central) polarity in opposition to ground (posterior) polarity. Previous studies suggested that the partially redundant refined (central) polarity receptors LIN-17/Fz and LIN-18/Ryk function in preempting EGL-20/Wnt-mediated ground (posterior) polarity [21,25]. Therefore we tested the hypothesis that CHW-1 functions in ground (posterior) polarity by examining the genetic interaction between chw-1 and lin-18 mutations. Approximately 35% of lin-18(e620) animals are P-Rvl [25], and we corroborated this observation with lin-18(n1051) ( Table 2;   Counter-intuitively, we found that loss of chw-1 enhanced rather than suppressed P-Rvl defects conferred by loss of lin-18 ( Table 2), suggesting that CHW-1 does not function as part of the ground (posterior) polarity program in opposition to refined (central) polarity. Interpretation of chw-1 genetic interactions with lin-17 and lin-18 was complicated by the structure of the overlapping gene, F22E12.3, whose first two exons are predicted to lie within intron 1 of chw-1. In addition to disrupting chw-1, the ok697 deletion removes these two F22E12.3 exons (wormbase.org release WS246; Fig 2C). Similarly, the original library clone for bacterially mediated feeding RNAi ("chw-1 + F22E12.3 RNAi") [40] targets both chw-1 and the first two exons of F22E2.3. Therefore, it is theoretically possible that the observed suppression of lin-17 P-Rvl and enhancement of lin-18 P-Rvl phenotypes was due to loss of chw-1, F22E12.3, or both. We resolved this issue by using F22E12.3-specific feeding RNAi, which failed to suppress lin-17 or enhance lin-18 mutants (Table 3). Conversely, we constructed chw-1-directed feeding RNAi targeting only the full-length chw-1 cDNA, and this RNAi suppressed lin-17 and enhanced lin-18 mutants. We conclude that it is loss of chw-1 that caused lin-17 suppression and lin-18 enhancement.
Taken together, our results 1) corroborate that LIN-17/Fz and LIN-18/Ryk are both quantitatively non-equivalent in directing refined (central) vulval polarity and 2) suggest that the function of CHW-1 is to equalize the signaling contributions of these two refined (central) polarity Wnt receptor signals.
As  Table 3. Effects on P7.p polarity are chw-1-specific. Animals were grown at 23°C and scored by DIC at late L4 stage. n is number of animals scored. Data for strains marked with an asterisk (*) are from Table 1 (lin-17) or  [21]. To further test whether CHW-1 functions as part of the EGL-20/Wnt (and CWN-1/Wnt) ground (posterior) polarity pathway, we constructed the lin-17; chw-1; lin-18 triple mutant strain. We found no suppression of the 100% P-Rvl phenotype by chw-1(ok697) ( Table 4), suggesting that CHW-1 does not play a substantial role in ground (posterior) polarity signaling.

CHW-1 function may be non-autonomous
To assess the endogenous CHW-1 expression pattern, we constructed a plasmid with chw-1 promoter sequences driving green fluorescent protein (GFP), and generated the reIs3 transgene containing this plasmid. Prior to vulval induction, the reIs3 P chw-1 ::gfp-containing transgene failed to express GFP in the six VPCs, but is visible in surrounding cells (Fig 3A and 3B). Upon EGF induction GFP was expressed in non-vulval (3°) VPC lineages and surrounding cells, but was excluded from VPC lineages that had been induced to form vulva (P5-7.p and subsequent 2°-1°-2°lineages; Fig 3C and 3D). The exclusion extended throughout larval development, and all expression faded after the L4 stage.
We hypothesized that Wnt pathway activity might actively exclude GFP and presumably CHW-1 expression from vulval lineages, and tested this hypothesis with single and double  CHW-1 has differential effects on lin-17 and lin-18 mutants. Alternatively, our promoter fusion construct may not accurately represent endogenous CHW-1 expression. We were limited in our inclusion of upstream promoter sequences by transgene toxicity; we were unable to isolate live transgenic animals harboring constructs containing F22E12.3 exon 3 or further upstream sequences, or chw-1 intron 1, implying the existence of regulatory elements that titrate essential transcription factors. This genomic region also harbors non-coding RNA genes F22E12.7 and F22E12.13, whose over-expression could also cause toxicity.
It has been reported that T-cell acute lymphoblastic leukaemia, a strongly Notch-dependent cancer, up-regulates Wrch-1/RhoU in a Notch activity-dependent manner to promote cell migration and chemotaxis [41]. A large set of putative C. elegans Notch-LAG-1/CSL transcriptional client genes were identified computationally [42]. We searched this dataset, but the chw-1 promoter was not identified as a putative Notch transcriptional target, consistent with our reIs3 reporter not evincing expression in 2°cells.
We also found that loss of CHW-1 had different effects in cam-1 or vang-1 mutant backgrounds (Table 5). chw-1-directed RNAi did not alter suppression of lin-17; by cam-1 in the triple mutant compared to the gfp-directed RNAi control (Table 5, line 5 vs. 4). In contrast, chw-1-directed RNAi was additive with vang-1 in suppressing the lin-17 P-Rvl phenotype compared to the gfp RNAi baseline: the chw-1 or vang-1 single mutant or RNAi suppressed the lin-17 mutant, while the chw-1; vang-1 double knockout suppressed the lin-17 mutant even further ( Table 5, lines 8 vs. 9 vs. 10 and 12 vs. 13 vs. 14). We note that the degree of the lin-17; chw-1; vang-1 P-Rvl phenotype, 35%, is the same as the lin-18 single mutant phenotype, also 35%. Whether this value represents a coincidence or reflects a fundamental property of the signaling system is unknown.
We additionally observed that P-Rvl phenotype of cam-1; lin-18 mutant animals we enhanced by chw-1 loss (Table 5, lines 18 vs. 19). Yet the P-Rvl phenotype of vang-1 lin-18 was unaltered by chw-1 loss (Table 5, lines 21 vs. 22 vs. 23 vs. 24). Based on these interactions we can make models interpreting CHW-1 interactions with LIN-17 and LIN-18, and also models interpreting LIN-17 and LIN-1 interactions with VANG-1 and CAM-1, but we cannot reconcile both models with the data we have. We propose that missing information is required to make sense of these inconsistencies. However, we still consider the model that CHW-1 promotes LIN-17/Fz signaling at the expense of LIN-18/Ryk signaling to best describe the data.

Discussion
In the absence of CHW-1, the relative contributions of LIN-17/Fz and LIN-18/Ryk to P7.p polarity appear to be roughly equivalent. We therefore hypothesize that CHW-1 regulates relative contribution of LIN-18/Ryk and LIN-17/Fz receptors to the refined (central) polarity decision, or to antagonism of the ground (posterior) polarity decision, which may be the same thing. We propose that CHW-1 boosts LIN-17/Fz signaling at the expense of LIN-18/Ryk, or shifts ligand signaling to favor LIN-17/Fz activation over LIN-18/Ryk activation. Thus, in a lin-17 mutant loss of CHW-1 increases LIN-18/Ryk contribution, thus partially compensating for LIN-17/Fz loss. Conversely, in a lin-18 mutant, loss of CHW-1 decreases LIN-17/Fz contribution, thus exacerbating the lin-18 mutant P-Rvl phenotype. It is unknown why P7.p would Table 5. Loss of VANG-1 but not CAM-1/Ror in the lin-17 (Fz) mutant background retains sensitivity to CHW-1 activity. Also, loss of CAM-1 but not VANG-1 in the lin-18 (Ryk) background retains sensitivity to CHW-1 activity. In other words, VANG-1 loss abolishes the ability of lin-18 mutant animals to respond to chw-1 mutation or RNAi. Animals were maintained at 23°C and scored by DIC at late L4 stage. n is number of animals scored. Analyses of statistical significance were performed using Fisher's exact test.  need non-equivalent signaling of these diverse Wnt receptor systems, or mechanistically how CHW-1 could act to make their contributions nonequivalent. Use of unexpected signaling input to regulate receptor activity or output has multiple precedents in growth cone guidance. The UNC-40/DCC and SAX-3/Robo transmembrane receptors, which canonically regulate dorsal-ventral axon guidance in response to UNC-5/Netrin and SLT-1/Slit ligands, respectively, can be modulated by VAB-8/kinesin-related, UNC-73/ RhoGEF and MIG-2/RhoG signaling input to regulate anterior-posterior migrations instead [43,44]. The EVA-1 co-receptor alters SAX-3/Robo interactions with UNC-40/DCC signaling [45], and the UNC-129/TGF-β-related protein regulates UNC-6/Netrin outcomes through UNC-40/DCC or UNC-5/Netrin receptors [46]. We hypothesize that CHW-1 could perform a similar function in regulating LIN-17/Fz versus LIN-18/Ryk receptor contributions to P7. p polarity. CAM-1/Ror and VANG-1/Van Gogh receptors also contribute differentially to However, these mechanisms are cell autonomous, and our data suggest that CHW-1 functions non-autonomously, though combined early non-vulval and later 3°-specific expression preclude the use of heterologous promoter expression of CHW-1 to conclusively resolve this question. If CHW-1 function is non-autonomous, we hypothesize that it controls the composition of Wnt gradients. There is a precedent for control Wnt gradients impacting P-Rvl phenotypes. CAM-1/Ror has been argued to sequester Wnts, thus impacting P-Rvl phenotype [21]. This phenomenon is so pronounced that even anterior displacement of the CAM-1/Rorexpressing canal-associated neurons (CANs) and their axons disrupts Wnt gradients and impacts the P-Rvl phenotype. This observation led to the intriguing model that alteration of nervous system architecture impacts subsequent epithelial cell development [29]. With this in mind we scrutinized chw-1(ok697) mutant animals using pan-neural and CAN/HSN GFP markers, but observed no alterations in nervous system structure, CAN axon extension or CAN cell migration. FGF was found to induce the SM cells to secrete CWN-1/Wnt, thus contributing to P7.p polarization [47]. We did not observe GFP expression in the SMs using the reIs3 transgene. Additional players in this system are the PAK-1 S/T kinase, NCK-1/Nck adaptor, and CED-10/Rac [48]. The PAK S/T kinase is an established effector of mammalian Chp/ Wrch [3,32,[49][50][51], [52], suggesting that C. elegans PAK-1 could function as a CHW-1 effector. Thus CHW-1 may signal through a different cascade than the implied Rac-Pak signal or interact somehow with PAK-1.
We also incorporate the CHW-1 modulating activity into this model, proposing that CHW-1 promotes activation of LIN-17/Fz at the expense of LIN-18/Ryk. Given the observed GFP expression pattern driven by transgenic chw-1 promoter with reIs3, we hypothesize that CHW-1 functions non-autonomously to impact multiple Wnt gradients or receptor activity. There exist two precedents for non-autonomous impacts on refined (central) polarity Wnt gradients. FGF secreted by the induced presumptive 1°VPC, P6.p, induced the SMs, neighboring the Anchor Cell, to secrete Wnts in support of AC polarizing signal [47]. Nervous system development and placing of potential Wnt sinks like the CANs and their axons can also sculpt Wnt gradients, thus impacting refined (central) polarity and perhaps ground (posterior) polarity [29]. It is currently unclear how CHW-1 fits into this paradigm, but future studies will fit CHW-1 amongst the many emerging players.

Materials and Methods
Nematode strains and analysis C. elegans strains were cultured using standard techniques [53]. All strains used were derivatives of wild-type Bristol strain N2 and were maintained at 23°C.
Bacterial-mediated RNAi was performed essentially as described previously [55]. Briefly, L4 animals were placed on a lawn of E. coli (strain HT115(DE3)) expressing dsRNA from plasmid L4440, and their progeny were scored at the L4 stage. RNAi constructs targeting both chw-1 and F22E12.3 (Wormbase RNAi ID WBRNAi000113757) were described previously [40]. Constructs to specifically target either chw-1 or F22E12.3 were generated for this work (details below). A construct targeting GFP was used as a negative control [56].

Phenotypic analysis
Defects in DTC migration were scored at the late L4 stage using DIC microscopy. Each gonad arm was analyzed separately and was considered abnormal if the gonad arm deviated from the typical "U" shape [57]. The significance of differences in the frequency of DTC migration defects was analyzed using Fisher's exact test.
Vulval phenotypes were scored at the mid-L4 larval stage using DIC microscopy. Animals were classified as P-Rvl if we observed an ectopic posterior invagination that was separated from the primary vulval invagination by adherent cells. Additionally, we examined the putative reversed P7.p lineage for number of cells, to account for the possibility of ectopic induction.
None was found. For some genotypes, we did observe a small number of animals (<2%) with abnormal vulval morphology distinct from VPCpolarization. These animals were included in the normal category because VPCs were properly polarized. The significance of differences in P-Rvl frequency between strains was analyzed using Fisher's exact test.

Plasmids and transgenes
To generate a chw-1-specific RNAi construct, the entire chw-1 ORF was amplified from oligo-dT primed cDNA and cloned into pBluescript SK+. The chw-1 ORF was then shuttled into the bacterial-mediated RNAi plasmid, pPD129.36 (L4440; A. Fire, personal communication), using Not I and Sal I. We attempted to clone a full F22E12.3 ORF, but were unable to amplify any F22E12.3 fragments from cDNA. Therefore, an F22E12.3-specific feeding RNAi construct was generated by cloning a PCR fragment amplified from genomic DNA lacking chw-1 coding sequences into pPD129.36 using Bgl II. The F22E12.3 PCR product was amplified from cosmid DNA (F22E12) using primers TK35 (5'-ttttagatctttgcctttgtaggctggatt) and TK36(5'-ttttagatctttccgacattaattggaaattg). All inserts were confirmed by sequencing.
To express chw-1 cDNA ectopically in DTCs, we used the pJK590 vector, which contains lag-2 promoter sequences that drive expression in DTCs [38]. To clone chw-1 into pJK590, we first used site-directed mutagenesis to eliminate a Bsm I site in the promoter (pJK590-mut), thus allowing use of the polylinker Bsm I site for cloning the chw-1 cDNA. We then amplified chw-1 cDNA yk1263f9 with primers DJR543 (AAAAAAaccggtGGCCGGCCgaaaaaATGG CAAAAGAAGAAGATGAGAAATG) and DJR544(AAAAAAgaatgctGCGGCCGCttaattaa TCAAATGAGGTTGCGTGTACG) and cloned the resulting product into pJK590-mut using Age I and Bsm I to create plasmid pCM14.1. Mutagenesis of the resulting construct yielded plasmids pCM14.2 (A18G), pCM14.3(A18V) and pCM14.4(Q69L).
The chw-1 promoter plasmid was constructed by PCR amplification and subcloning of chw-1 promoter sequences. We attempted to use primers within the F22E12.3 gene upstream of chw-1, but resulting constructs caused lethality, preventing isolation of transgenes. Instead we used primers VM59 (forward) TCGAGTATTTCGAACCGTTACTGGTGGAGG (with Xma I site) and DJR412 (reverse) TCGAAGTTTTTCCAACTGCC (with Xma I site) to amplify promoter sequences not including F22E12.3 coding sequences, and subcloned the resulting product into plasmid pPD95.67, which contains GFP and an unc-54 3'UTR (A. Fire, personal communication). This plasmid was injected at 100 ng/μl with co-injection marker plasmid pRF4 (rol-6(su1006d)) (80 ng/μl) to make extrachromosomal array reEx6, which was then integrated by UV irradiation to form reIs3. reIs3 was outcrossed 5x to the wild type (N2) by following its Rol phenotype, and the resulting strain was used to construct all subsequent strains.
Since reIs3-bearing transgenic animals are Rol, it is difficult to see all ventrally located VPCs in one animal. We therefore scored reIs3 GFP expression in animals grown on sqt-1-directed RNAi, which mitigates the reIs3 Rol phenotype. We independently verified that sqt-1(RNAi) did not alter GFP expression patterns.

3'RACE
Because the predicted CHW-1 protein lacked typical C-terminal sequences, we performed 3' RACE experiments to detect all possible 3' exons. We used forward primers DJR557 in exon 3 (GGAGCTGAATTTTTTGAATGTTCAGC), DJR556 in exon 4 (ATGTTTGACGCAGCAATTTTGGCC) and DJR555 spanning the exon 4/putative 3'UTR boundary (GATTCATCACACGTACTCGCAACC) to amplify from reverse transcriptase-synthesized reverse strand cDNA generated by an Invitrogen 3'RACE kit using total C. elegans RNA. In each case only one significant 3'RACE band was identified, and sequence analysis indicated that the band corresponded to the Wormbase-predicted gene structure for exon 4, which was also independently validated by cDNA (Wormbase WS226). This predicted gene structure also corresponds to CHW-1 homologs from related nematode species.
Supporting Information S1 File. An accompanying file contains the list of C. elegans strains used in this study.