Dishevelled attenuates the repelling activity of Wnt signaling during neurite outgrowth in Caenorhabditis elegans

Significance

The growth of neuronal processes is guided not only by external molecules but also by the modulation of the intracellular responses to those extracellular guidance cues. The mechanism for such modulation is unclear. Using a pair of sensory neurons in the nematode Caenorhabditis elegans, we found that guidance molecules activate two downstream molecules, which redundantly mediate the navigation of the main neurite away from the source of the repulsive guidance cues. However, one of the two molecules also provides feedback inhibition to downregulate the signaling activity, which allows the neurons to grow a second neurite against the gradient of the repulsive cues. This dual functionality of downstream signaling molecules thus fine-tunes the intracellular response to guidance molecules.

Abstract

Wnt proteins regulate axonal outgrowth along the anterior–posterior axis, but the intracellular mechanisms that modulate the strength of Wnt signaling in axon guidance are largely unknown. Using the Caenorhabditis elegans mechanosensory PLM neurons, we found that posteriorly enriched LIN-44/Wnt acts as a repellent to promote anteriorly directed neurite outgrowth through the LIN-17/Frizzled receptor, instead of controlling neuronal polarity as previously thought. Dishevelled (Dsh) proteins DSH-1 and MIG-5 redundantly mediate the repulsive activity of the Wnt signals to induce anterior outgrowth, whereas DSH-1 also provides feedback inhibition to attenuate the signaling to allow posterior outgrowth against the Wnt gradient. This inhibitory function of DSH-1, which requires its dishevelled, Egl-10, and pleckstrin (DEP) domain, acts by promoting LIN-17 phosphorylation and is antagonized by planar cell polarity signaling components Van Gogh (VANG-1) and Prickle (PRKL-1). Our results suggest that Dsh proteins both respond to Wnt signals to shape neuronal projections and moderate its activity to fine-tune neuronal morphology.

Axonal navigation along the anterior–posterior (A-P) axis is often guided by the gradients of morphogens, including those of the Wnt family of secreted glycoproteins (1). Wnt proteins repel axons expressing the receptor tyrosine kinase Derailed/Ryk (2, 3) or the Frizzled (Fzd, Fz) receptor (4–9), but also act as attractive cues for neurite outgrowth and dendritic formation through the Fzd receptor (10–12). Moreover, Wnt proteins can promote axon outgrowth and at the same time induce repulsive turning in cultured cortical neurons (6).

Wnt signals evoke downstream activity mainly through three different intracellular pathways: the β-catenin-dependent canonical pathway, the planar cell polarity (PCP) pathway, and the Wnt/calcium pathway (13, 14). Although the canonical pathway (7, 15) and the Wnt/calcium pathway (16) regulate axon guidance in a few instances, Wnt-mediated axon pathfinding mainly uses the PCP pathway, which activates the small GTPases Rho, Rac, and Cdc42, which regulate cytoskeletal dynamics (17, 18). Studies from mice, Drosophila, and Caenorhabditis elegans collectively suggest that the core components of PCP signaling involved in axon guidance are the Wnt receptor Fzd, the phosphoprotein Dishevelled (Dsh, Dvl), the transmembrane protein Van Gogh/Strabismus (Vang/Stbm, Vangl) and its adaptor Prickle (Prkl, Pk), and the atypical cadherin Flamingo/starry night (Fmi/Stan, Celsrs), although the requirement for each component may depend on the specific neuron type (17, 19).

An important question is how these components interact with each other to spatially and temporally control the neuronal response to the Wnt signal. One particularly intriguing aspect of this control is the relationship between Fzd and Dsh. Normally Dsh is thought to act as a downstream effector of Fzd (20–22), but Dsh can also promote the phosphorylation of Fzd, which attenuates Fzd activity and inhibits downstream PCP signaling (23–25). The physiological significance of this inhibition is unclear, but it suggests that Dsh proteins both promote and inhibit Wnt signaling in the same cellular context.

Here, using the morphologically well-defined PLM neurons in C. elegans, we find that two Dsh proteins, DSH-1 and MIG-5, act redundantly downstream of Fzd receptor to mediate the repelling activity of Wnt signal, which guides the outgrowth of a long, anteriorly directed neurite away from the cue. At the same time, DSH-1 also provides feedback inhibition to attenuate Fzd signaling. The net effect of these two actions is that the PLM neurons can grow a posteriorly directed neurite against the Wnt gradients. The dual functions of DSH-1 help establish the bipolar shape of the PLM neurons and other C. elegans posterior neurons.

Results

LIN-44/Wnt Proteins Repel PLM Neurites Through LIN-17/Fzd.

C. elegans has two pairs of embryonically derived touch receptor neurons (TRNs) (26). The anterior ALM neurons are unipolar, having a single anteriorly directed neurite, whereas the posterior PLM neurons are bipolar, with a long anterior neurite that extends to the center of the animal and a short posteriorly directed neurite that extends to the tail (Fig. 1A). Previous studies showed that mutations in lin-44/Wnt, which is normally expressed from the tail epidermis, and lin-17/Fzd cause PLM neurons to have short anterior neurites and a long posteriorly oriented neurite that reaches the tip of the tail, turns, and then extends anteriorly for a significant length (8) (Fig. 1B; this phenotype is hitherto referred to as the “Wnt phenotype”). Similarly, the double mutation of two other Wnt genes, cwn-1 and egl-20, reversed the orientation of ALM cells, which then had long posterior neurites and little or no anterior neurites (8, 27). The authors proposed that Wnt signaling establishes the A-P polarity of the TRNs. Although LIN-44 and LIN-17 regulate cellular polarity during asymmetric cell divisions (28, 29), the TRN morphological defect is the only instance in which the function of Wnt and Fzd receptors are thought to control the polarity of terminally differentiated neurons. Our results, however, challenge this interpretation and suggest that Wnt signaling plays its more expected role of providing guidance cues.

Fig. 1.

LIN-44/Wnt repels TRN neurites toward the anterior through Fzd receptor LIN-17 and Dsh proteins DSH-1 and MIG-5. (A) TRN morphologies visualized by RFP expression from the mec-17 promoter. (B) PLM in wild-type, lin-44(n1792), and lin-17(n671) animals and those expressing lin-44(+) and egl-20p::lin-44(+). (Scale bar, 20 μm.) (Right) Scheme of normal Wnt expression patterns adopted from ref. 30 and the diagrams of PLM shapes in three different phenotypes: Wnt phenotype, posterior underextension, and posterior overextension. (C) The length of PLM posterior neurites in various animals. (D) TRN morphology in dsh-1(u915) mig-5(u1030) double mutants. Arrows point to ALM and PLM neurons that showed the Wnt phenotype; triangles point to the posteriorly directed neurites of the TRNs with morphological defects. (E) The penetrance of Wnt phenotypes in various mutants.

First, overexpression of LIN-44 changed the length of PLM posterior neurite, which is not expected if LIN-44 only provides positional information to establish “cellular polarity.” Wnt ligands are expressed in a specific pattern along the A-P axis. lin-44 is expressed mainly posterior to the PLM cell body, whereas egl-20 is expressed exclusively anterior to the PLM cell body (Fig. 1B) (30). We found that overexpression of lin-44 from its own promoter led to shortened PLM posterior neurites (Fig. 1B), suggesting that Wnt signals act as repellents that suppress posterior outgrowth, rather than as a polarity signal. Misexpression of lin-44 from the egl-20 promoter elongated the posterior neurite but did not cause it to extend to the tail, turn, and then extend anteriorly (Fig. 1 B and C). This posterior “overextension” phenotype also differed from the Wnt phenotype, in that the anterior neurite was not shortened. As a control, we misexpressed lin-44 from the mom-2 promoter, which is expressed at a low level around the PLM cell body, and found that PLM morphology did not change significantly (Fig. 1C). These results suggest that the repelling activities of the misexpressed LIN-44 countered the activities of the endogenous LIN-44 to allow the growth of a longer PLM posterior neurite.

One previous argument for polarity reversal in lin-44 and lin-17 mutants is the mistargeting of the synaptic vesicle protein RAB-3, which is transported toward the anterior in the wild-type animals but localized in the posterior neurite in lin-44 and lin-17 mutants (8). We found, however, that GFP::RAB-3 localized primarily to the positions where PLM neurites physically contacted the ventral cord. In the wild-type animals, this occurred when the anterior neurite sent a distal branch to synapse onto the ventral cord, whereas in the mutants, the GFP was distributed along the ventral cord, where the anteriorly extending segment of the PLM posterior neurite, which made a U-turn at the tail, contacted the ventral cord (83%; n = 36; Fig. S1 A and B). GFP did accumulate in the tail region in the remaining 17% of PLM cells, as previously reported (8), but this distribution may be because these posterior neurites never made contact with the ventral cord. These results suggest that RAB-3 localization is more an indication of proximity to the ventral cord than a neurite-specific marker.

Fig. S1.

Localization of synaptic proteins in lin-44 and lin-17 mutants. (A) Presynaptic maker GFP::RAB-3 (arrows) is expressed from the TRN-specific mec-7 promoter and is normally localized in positions close to the vulva, where the PLM anterior neurite sends a branch to contact the ventral cord. (B) In lin-44(n1792) mutants, GFP::RAB-3 (arrows) is localized in the anteriorly running portion of the PLM posterior neurite at positions where the neurite contacts the ventral cord. (C and D) The gap junction marker UNC-9::GFP (arrows) is localized in the PLM anterior neurite close to the cell body at positions where the neurite contacts the PVC interneuron in both wild-type and lin-44 mutant animals. (E) A diagram showing the position of the synaptic markers. Similar phenotypes were also observed in lin -17 mutants [78% (n = 40) for mec-7p::GFP::rab-3 and 72% (n = 32) for mec-17p::unc-9::GFP].

We have, however, found a marker, the UNC-9 innexin gap junction protein (31), that localized to the anterior PLM neurite in both wild-type animals and lin-44 and lin-17 mutants. In wild-type animals, UNC-9::GFP was found in patches in the anterior neurite near the cell body. UNC-9::GFP was in the same position in lin-44 and lin-17 animals, even though the anterior neurite was much shorter (Fig. S1 C and D). The position of the label is where the PLM neurons form gap junctions with the downstream interneurons. Thus, synaptic proteins cannot be used to indicate polarity, undermining the main supporting evidence for the previous model.

Moreover, the growth of a posterior neurite in PLM neurons requires the cell-intrinsic, Wnt-independent activity of the Hox protein EGL-5; mutations in egl-5 specifically suppressed posterior, but not anterior, outgrowth in otherwise wild-type animals (32). If PLM polarity is simply reversed in lin-44 or lin-17 animals, the anterior neurite should act like the wild-type posterior neurite and be shortened by mutation of egl-5 in those animals. However, the loss of EGL-5 did not affect the growth of the short anterior neurite in lin-44 or lin-17 animals (Fig. S2), which further challenges the polarity theory.

Fig. S2.

Mutations in egl-5 do not affect anteriorly directed neurite outgrowth in lin-44 and lin-17 mutants. (A) PLM morphologies in various mutant strains. Arrows point to the end of either PLM posterior neurites or the anterior ones. (Scale bar, 20 μm.) (B) The length of PLM posterior neurites in the wild-type and egl-5 animals; anterior neurite length of PLM neurons that displayed the Wnt phenotype in the indicated animals.

For the above reasons and because all other neurons affected by the loss of Wnt signals only show outgrowth, and not polarity defects (7, 11, 12), we feel that the defects in the PLM neurons are best described as alterations in outgrowth. Although we could not completely rule out the polarity model, the primary wild-type function of Wnt and Fzd appears to be the stimulation of anterior neurite outgrowth and the inhibition of posterior neurite outgrowth.

Dsh Proteins, DSH-1 and MIG-5, Act Redundantly Downstream of Wnt Ligand and Fzd Receptor to Repel TRN Neurites.

Dsh proteins link Fzd receptors to all three Wnt signaling pathways (14). C. elegans has three Dsh paralogs: DSH-1, DSH-2, and MIG-5. Mutations in a single Dsh gene did not cause the Wnt phenotype, but double mutations in dsh-1and mig-5 did (Fig. 1 D and E). Because of their close linkage, we used CRISPR (clustered regularly interspaced short palindromic repeats)/Cas9-mediated genome editing (33) to introduce mutations in dsh-2 and mig-5 in animals carrying the dsh-1(ok1445) allele. Loss-of-function dsh-2 and mig-5 alleles with small deletions that caused frameshifts were identified by genotyping (Fig. S3B). PLM neurons in dsh-1 mig-5 animals were morphologically similar to those in lin-44 or lin-17 mutants, and ALM neurons in dsh-1 mig-5 double mutants resembled the cells seen in cwn-1 egl-20 double mutants (Fig. 1D). ALM and PLM neurons in dsh-1 dsh-2 double mutants did not have the Wnt phenotype (Fig. 1E); in fact, they were similar to the cells in dsh-1 animals described in the next paragraph. These results suggest that Dsh proteins DSH-1 and MIG-5 act redundantly downstream of the Wnt ligand and Fzd receptor to propel TRN neurites toward the anterior.

Fig. S3.

The expression of dsh-1 in TRNs and the molecular lesions in the alleles of the Dsh genes. (A) The expression of dsh-1 reporter sIs12076 [C34F11.9a::gfp] in both ALM and PLM neurons (arrows), which are labeled by mec-17p::RFP. (Scale bar, 20 μm.) (B) Gene structures of the three Dsh paralogs, dsh-1, dsh-2, and mig-5, and the mutant alleles used in this study. Alleles u915, u952, and u953 were isolated from our forward genetic screen; u1029 and u1030 were generated using CRISPR/Cas9-mediated genome editing (32). (C) The length of PLM posterior neurites in dsh-1 mutants. Double asterisks indicate P < 0.01.

Mutations in Dsh Gene dsh-1 Caused the Shortening of the PLM Posterior Neurite.

Given its role downstream of Wnt and Fzd, we were surprised that dsh-1 single mutants had significantly shorter PLM posterior neurites compared with the wild-type (Fig. 2 A and B). We isolated three dsh-1 alleles (u915, u952, and u953) in a genetic screen for mutants with TRN morphological defects (Fig. S3B). All three (presumably loss-of-function) alleles and the deletion allele ok1445 resulted in a similar shortening of the PLM posterior neurites (Fig. 2B). DSH-1 is expressed and functions cell autonomously in the TRNs, as a dsh-1 promoter reporter sIs12076 [C34F11.9a::gfp] was expressed in both ALM and PLM neurons (Fig. S3A), and expression of a wild-type copy of the a isoform of dsh-1 (www.wormbase.org) from the TRN-specific mec-17 promoter rescued the PLM morphological defects (Fig. 2B). These results suggest DSH-1 normally promotes posteriorly directed neurite outgrowth. This function of DSH-1 depended on Wnt signals because dsh-1 lin-44 and dsh-1 lin-17 double mutants had the same defects as the lin-44 and lin-17 animals. In both the single and double mutants, about 80% of PLM neurons showed the Wnt phenotype (Fig. 1E), and the other 20% had a slightly elongated posterior neurite (Fig. S4), suggesting lin-44 and lin-17 are epistatic to dsh-1. These results support the conclusion that Wnt proteins act as guidance molecules for the PLM neurons.

Fig. 2.

DSH-1 promotes posteriorly directed neurite outgrowth by attenuating Wnt signals. (A) PLM posterior neurites in wild-type and dsh-1 animals. (B) The length of PLM posterior neurites in various mutants and transgenic animals. (C) The percentage of PLM neurons showing the Wnt phenotype in animals overexpressing Dsh proteins from mec-17 promoter. Constructs were injected at the concentration of 15 ng/μL. (D) The localization of DSH-1::GFP in the anterior (AN) and posterior (PN) neurites of PLM neurons. The dashed boxes, which are enlarged in the lower panels, represent 35 μm in neurite length; arrows point to the fluorescent puncta. (E) The average number of fluorescent DSH-1::GFP puncta/35 μm in ALM and PLM neurites.

Fig. S4.

lin-44 and lin-17 are epistatic to dsh-1. (A) About 20% of the PLM neurons in lin-44 and lin-17 mutants do not show the Wnt phenotype and have generally normal cellular shape. The tail region (dashed box) is enlarged to visualize the PLM posterior neurites; arrows indicate the cell bodies of the PLM neurons that do not show the Wnt phenotype, and triangles point to the end of their posterior neurites. (B) In the dsh-1 lin-44 and dsh-1 lin-17 double mutants, the PLM neurons that do not display the Wnt phenotype have long posterior neurites instead of the shortened ones. (C) The posterior neurite length of PLM neurons showing the Non-Wnt phenotype in various strains. (Scale bar, 20 μm.)

Among the three Dsh genes, the effect on the growth of the PLM posterior neurite is specific to dsh-1; mutations in dsh-2 and mig-5 did not result in the shortening of PLM posterior neurite (Fig. 2B). Therefore, in addition to promoting the outgrowth of anterior neurites away from the Wnt signal DSH-1, but not the other two Dsh paralogs, also promoted posterior neurite outgrowth against the Wnt gradients. This second, inhibitory function of DSH-1 was further confirmed by the observation that dsh-1 overexpression from a mec-17 promoter produced the Wnt phenotype in wild-type PLM neurons (Fig. 2C), which suggests that an excess amount of DSH-1 can completely shut down Wnt signaling. Consistent with the absence of this second function in DSH-2 and MIG-5, overexpression of dsh-2 and mig-5 had little effect on PLM neurites (Fig. 2C).

The observation that overexpression of lin-44 from its own promoter caused a similar shortening of the PLM posterior neurite, as seen in dsh-1 mutants (Fig. 1B), is consistent with DSH-1 having a negative effect on Wnt signaling. Misexpression of lin-44 from the egl-20 promoter (i.e., from a more anterior site) partially rescued dsh-1 mutants (Fig. 2B). These results suggest that the loss of dsh-1 increased the Wnt repelling activity, presumably through mig-5, which led to defects in posteriorly directed outgrowth. The length of PLM anterior neurites was not affected by either lin-44 overexpression or dsh-1 mutations, suggesting that outgrowth toward the anterior is not sensitive to the hyperactivity of Wnt signaling. Because DSH-1::GFP was relatively enriched in the posterior neurite compared with the anterior neurite (Fig. 2 D and E), we propose that high Wnt concentration in the tail region leads to the accumulation of DSH-1 in the posterior neurites, which triggers the down-regulation of the Wnt signaling. This DSH-1-mediated negative feedback is not evoked in the anterior neurite, presumably because of the low LIN-44 concentration surrounding it.

The Dishevelled, Egl-10, and Pleckstrin Domain of DSH-1 Is Required for the Attenuation of Wnt Signaling and the Growth of PLM Posterior Neurites.

DSH-1-mediated down-regulation of the repulsive Wnt activities resembles the inhibition of Wnt5a-stimulated growth of commissural axons in mouse embryos by Dvl1, which induces phosphorylation of Fzd3 (24). Biochemical studies in Xenopus found that the dishevelled, Egl-10, and pleckstrin (DEP) domain of Dsh proteins was needed for Dsh-dependent Fzd3 phosphorylation (25). The DEP domain of DSH-1 is essential for attenuating the repelling activity of Wnt signaling that allows the posterior neurite to grow to a normal length. The expression of DEP-less DSH-1 or DSH-1 N terminus in dsh-1 animals failed to restore the PLM posterior neurite in dsh-1 mutants, whereas the expression of the DEP domain-containing C terminus did (Fig. 3 A and B). Moreover, the missense mutation in dsh-1(u952), which alters G512 in the DEP domain (Fig. S3B), gave the same shortened PLM posterior neurites as the null alleles (Fig. S3C). Overexpression of the full-length DSH-1 protein and the DEP domain-containing C terminus caused the Wnt phenotype, although at a lower penetrance than in lin-17 or lin-44 mutants (Fig. 3C). In contrast, overexpression of DSH-1 N terminus and DSH-1ΔDEP had little effect (Fig. 3C). These results suggest that DSH-1 provides inhibitory feedback to the Wnt signaling pathway through its DEP domain.

Fig. 3.

The DEP domain of DSH-1 mediates its inhibitory function. (A) The schemes representing the domain organization of DSH-1 protein and its variants. (B) The length of PLM posterior neurites in dsh-1 animals expressing various transgenes. Constructs were injected at the concentration of 3 ng/μL. (C) Penetrance of the Wnt phenotype in PLM neurons from animals overexpressing DSH-1 variants from mec-17 promoter. Constructs were injected at 15 ng/μL.

The DEP domains of DSH-1 and MIG-5 have 85% sequence similarity, and swapping the most divergent region of the domains (Lys-Leu-Lys-Tyr-Ile-Ala in DSH-1 and Ala-Ala-Gly-Leu-Ile-Arg in the MIG-5) did not affect the inhibitory function of DSH-1 (Fig. S5). Therefore, the functional differences between DSH-1 and MIG-5 cannot be attributed to the DEP domain alone.

Fig. S5.

The DEP domains of DSH-1 and MIG-5 are similar. (A) The alignment of the DEP domain of DSH-1 with that of MIG-5, using ClustalW2. Asterisks indicate identical amino acids, and dots indicate similar amino acids. Lys-Leu-Lys-Tyr-Ile-Ala (aa541-546) in red in DSH-1 is the most divergent region from MIG-5. (B) This region is replaced by the amino acids Ala-Ala-Gly-Leu-Ile-Arg from MIG-5 in the DSH-1_Ala-Ala-Gly-Leu-Ile-Arg mutant. (C) The length of PLM posterior neurites in dsh-1(u915) mutants expressing DSH-1_Ala-Ala-Gly-Leu-Ile-Arg mutant proteins from the mec-17 promoter.

Protein Kinase C Phosphorylation Sites in LIN-17/Fzd Cytoplasmic Tail Mediate the DSH-1 Functions in Attenuating Wnt Signaling.

We next confirmed that phosphorylation of the Fzd receptor LIN-17 was required for DSH-1-mediated attenuation of Wnt signaling activity. Although the lin-17 gene produces several isoforms varying in their C terminus, we found that the expression of the cDNA of the lin-17a isoform (www.wormbase.org) from a TRN-specific mec-17 promoter fully rescued the morphological defects of lin-17 mutants (Fig. 4D). Thus, LIN-17a responds to Wnt ligands cell-autonomously in PLM neurons.

Fig. 4.

Phosphorylation sites in the cytoplasmic tail of LIN-17 are needed for DSH-1-mediated inhibition of Wnt signaling. (A) The sequence of the C-terminal cytoplasmic region of LIN-17a proteins. Predicted PKC phosphorylation sites are labeled in red. Five serine/threonine sites were mutated to alanine in LIN-17a(5A) and to glutamic acid in LIN-17a(5E) mutants. (B) Percentage of the Wnt phenotype in PLM neurons of lin-17 animals expressing wild-type or mutant LIN-17a proteins in TRNs. (C–E) PLM posterior neurites in wild-type, lin-17, and dsh-1 animals carrying transgenes that express LIN-17a mutants.

The cytoplasmic tail of Fz1 can be phosphorylated and then inhibited by protein kinase C ζ (PKCζ) in Drosophila eyes (23), and multiple phosphorylation sites have been identified in mouse Fzd3 and Xenopus Fz3 (24, 25). Using the online tools of the Group-based Prediction System (34), we identified four potential PKC phosphorylation sites (S523, T546, S551, and S555) in the C-terminal cytoplasmic region of LIN-17a and created LIN-17a mutants with nonphosphorylatable (5A) and phosphomimetic (5E) amino acids at those sites plus the adjacent T524 (Fig. 4A).

Expression of either LIN-17a(5A) or LIN-17a(5E) in the TRNs rescued the PLM neurite Wnt phenotype in lin-17 null mutants (Fig. 4B). However, the morphologically restored PLM neurons that express the nonphosphorylatable LIN-17a(5A) had significantly shorter posterior neurites, whereas cells that express LIN-17a(5E) carrying the phosphomimetic mutations had longer posterior neurites compared with cells rescued by the wild-type LIN-17a (Fig. 4 C and D). These results suggest that the potential PKC phosphorylation sites in the cytoplasmic tail of LIN-17/Fzd are important for the regulation of neurite outgrowth by Wnt signaling, although we could not rule out the phosphorylation of these sites by other kinases.

Unlike the overexpression of lin-44/Wnt, overexpression of lin-17(+) did not cause the premature termination of the PLM posterior neurite (Fig. 4D). This result could occur if Wnt, and not its receptor, is limiting, as seems likely. In contrast, overexpression of the LIN-17a(5A) in wild-type animals did result in a short PLM posterior neurite (Fig. 4D), possibly through competition with endogenous LIN-17a. We also found that the expression of LIN-17a(5E), but not LIN-17a(+) or LIN-17a(5A), was sufficient to rescue the defects in posteriorly directed outgrowth in dsh-1 mutants (Fig. 4 D and E). These results suggest that DSH-1 negatively regulates Wnt signaling activity by affecting the phosphorylation state of LIN-17.

The Canonical Pathway Contributes Weakly to the Wnt Activities in Neurite Guidance.

We tested whether both the canonical and noncanonical Wnt pathways are needed for Wnt repellent activity in PLM neurons. In the canonical pathway, the binding of Wnt ligands to Fzd receptor leads to the disassembly of the GSK-3/Axin/APC destruction complex, which leads to the stabilization and nuclear translocation of β-catenin and activation of downstream genes (35). This signaling cascade requires the C-terminal cytoplasmic Lys-Thr-X-X-X-Trp motif in Fzd receptors (36). We found that the expression of LIN-17a lacking amino acids 497–502 (Lys-Thr-Val-His-Ala-Trp) rescued the Wnt phenotype in lin-17 mutants (Fig. 5A), suggesting the canonical pathway is not required for Wnt activities. In addition, mutations in either pry-1/Axin, a component of the destruction complex, or any of the four β-catenin genes (bar-1, hmp-2, wrm-1, and sys-1) did not significantly change the length of PLM posterior neurites in wild-type animals (Fig. 5B). Loss of bar-1 and overexpression of LIN-17a (ΔLys-Thr-Val-His-Ala-Trp) could, however, partially restore the PLM posterior neurite defect in dsh-1 mutants where the strength of Wnt signaling was elevated (Fig. 5B). These results suggest that the β-catenin-dependent canonical pathway indeed contributes to the guidance function of Wnt proteins, as had been previously found for the anteroposterior guidance of d-type axons in C. elegans (7).

Fig. 5.

The canonical pathway contributes weakly to the activity of Wnt signaling. (A) The penetrance of the PLM Wnt phenotype in lin-17 animals expressing LIN-17a that lacks the transmembrane motif Lys-Thr-Val-His-Ala-Trp (aa 497–502). (B) The length of PLM posterior neurites in various strains.

PCP Signaling Components Vang, Prkl, and Fmi Antagonize the Function of DSH-1 in Promoting Outgrowth.

We also tested the role of Van Gogh (VANG-1)/Vang, Prickle (PRKL-1)/Prkl, and FMI-1/Fmi, three core components of PCP signaling in axonal guidance (17) in PLM outgrowth. Although their loss did not produce the Wnt phenotype seen in lin-44 and lin-17 mutants, mutations in vang-1 and prkl-1 resulted in the abnormal elongation of the PLM posterior neurite (Fig. 6A). This posterior overextension phenotype, although less severe than the Wnt phenotype, suggests a weakening of the repulsive Wnt signal and a role, albeit a weak one, for VANG-1 and PRKL-1 in Wnt signaling activity in the PLM neurons. This conclusion is supported by the observation that mutations in vang-1 and prkl-1 partially rescued the dsh-1 phenotype by increasing the length of PLM posterior neurite. Mutations in fmi-1 also rescued the PLM posterior outgrowth defects in dsh-1 mutants, but elongated the posterior neurite only slightly in the wild-type background (Fig. 6A). Thus, these three positive regulators of PCP signaling antagonize the DSH-1-mediated feedback inhibition. Mutations in unc-44/Diego-like and lin-18/Derailed/Ryk did not affect PLM outgrowth (SI Results).

Fig. 6.

The PCP signaling pathway is downstream of the Wnt activities in neurite guidance. (A–C) The length of PLM posterior neurites in various strains. In A, lin-17a(5A) was expressed from the mec-17 promoter and repressed the phenotype of vang-1 and prkl-1.

Expression of nonphosphorylatable LIN-17a(5A) suppressed the elongation of the PLM posterior neurite in vang-1 and prkl-1 mutants (Fig. 6A). These results suggest that VANG-1 and PRKL-1 normally act to reduce LIN-17 phosphorylation and, thus, counteract the increase in LIN-17 phosphorylation caused by DSH-1. These data are consistent with a previous report that mouse Vangl2 promotes Wnt5a-stimulated outgrowth of commissural axons by suppressing Fzd3 hyperphosphorylation (24).

DAAM-1 and JNK-1 Act Downstream of PCP Signaling.

The PCP signaling pathway activates Rac and its downstream kinase JNK, as well as the small GTPase Rho and its effector ROCK, through the Formin family protein Daam1 (37). We found that mutations in jnk-1, the sole homolog of JNK, caused the elongation of PLM posterior neurites, but not the Wnt phenotype, and mutations in daam-1 had much weaker effects (Fig. 6C). Mutations in both genes significantly increased the length of PLM posterior neurites in dsh-1 background (Fig. 6C). jnk-1 daam-1 double mutants had slightly longer posteriorly directed neurites than jnk-1 animals (Fig. 6 B and C). These results suggest that both Rac/JNK and Daam1/Rho/ROCK pathways contribute to the repulsive Wnt effect in the PLM neurons.

bar-1 jnk-1 daam-1 triple mutants did not show the Wnt phenotype either, although the PLM posterior neurites were the most extended of any of the mutant combinations (Fig. 6C). Thus, both the canonical and the PCP pathways contribute to Wnt signaling in the PLM neurons. In contrast, the Wnt/calcium pathway did not appear to be important (Fig. S6 and SI Results).

Fig. S6.

The Wnt/Calcium pathway has little effect on PLM posterior outgrowth. The length of PLM posterior neurites were normalized to the diameter of cell bodies in various mutant strains. n.s. indicates no significant difference in an ANOVA analysis.

DSH-1-Mediated Attenuation of Wnt Signaling Generally Enables Posteriorly Directed Neurite Outgrowth in Tail Neurons.

The tail region of C. elegans contains the cell bodies of more than a dozen neurons in addition to the PLM neurons. Two pairs of these neurons, ALNL/R and PLNL/R, are bipolar, similar to the PLM neurons (Fig. S7). We found that the posterior ALN and PLN neurites were also repelled by LIN-44/Wnt signals, and DSH-1 promoted posteriorly directed outgrowth against the Wnt gradients. In lin-44, lin-17, and dsh-1 mig-2 mutants, the extent of the two neurites of the ALN and PLN neurons were reversed: the anterior neurite was shorter and the posterior neurites grew to the end of the tail, turned, and extended anteriorly for a considerable length (Fig. S7 A–D). In the dsh-1 single mutants, the posterior neurites of the ALN and PLN neurons were significantly shorter than in the wild-type (Fig. S7E). These results suggest that DSH-1-mediated feedback inhibition may generally balance the repulsive activities of Wnt signals and enable the growth of posterior neurites in tail bipolar neurons.

Fig. S7.

DSH-1 promotes posteriorly directed neurite outgrowth in ALN and PLN neurons. (A and B) The morphologies of ALN and PLN neurons, which are labeled by GFP expressed from the lad-2 promoter, in the wild-type, lin-44, and dsh-1 mutant animals. In lin-44 animals, the posterior neurite is overextended toward the tail, turns, and runs anteriorly for a significant length. This phenotype was termed as the “Wnt phenotype” for ALN and PLN neurons. The posterior neurites of these neurons were shortened in dsh-1 animals. (C) The penetrance of Wnt phenotype in various mutants. (D) The expression of lad-2p::GFP was not observed in ALN neurons of the lin-17 mutants; instead, extra cells expressing the TRN marker mec-17p::RFP were observed at the tail region. Therefore, ALN neurons may adopt the fate of their sister cells, PLM neurons. Similar lineage transformation was observed in lin-17 mutants before (28). (E) The length of ALN and PLN posterior neurites in dsh-1 mutants.

SI Results

Diego and Derailed/Ryk Are Not Required for PLM Outgrowth.

Loss of some components of the Wnt signaling pathways did not affect PLM outgrowth in wild-type, lin-44, or dsh-1 animals. Specifically, unc-44 encodes an ankyrin-repeat protein that has high sequence identity in the functional domains to Diego (19), the only PCP core component not involved in axon guidance (17), and lin-18 encodes the C. elegans homolog of Derailed/Ryk that is needed for Wnt-mediated axon repulsion in flies and mice (2, 3). Mutations in neither unc-44(e362) nor lin-18(e620) changed the length of PLM neurites.

Wnt/Calcium Pathway Is Not Important for the Repelling Activity of LIN-44.

We tested the contribution of the Wnt/Calcium pathway, which is mainly mediated by PKC and calcium/calmodulin-dependent protein kinase II (CaMKII/UNC-43) (42). Mutations in pkc-1 and unc-43 alone or together did not significantly elongate the PLM posterior neurites in the wild-type or dsh-1 animals, and at least mutations in pkc-1 did not enhance the phenotype of bar-1 or bar-1 jnk-1 mutants (Fig. S6). Although there are multiple paralogs of PKC and many calcium-activated enzymes and signaling molecules, our results suggest that the calcium pathway may have little effect on PLM guidance.

Discussion

In this study, we found that LIN-44/Wnt and LIN-17/Fzd act to repel PLM neurites toward the anterior instead of controlling neuronal polarity, as previously thought. In this reinterpretation, Wnt proteins function in their conventional role as guidance molecules during PLM morphogenesis. We propose that the default outgrowth pattern in PLM neurons is determined by the intrinsic propensity to grow toward the posterior in the absence of the guidance cues. The Wnt signal released from the posterior side of the cell body likely determines the length of the two neurites by regulating neurite extension. The anterior neurite grows down the repulsive Wnt gradient, whereas the posterior neurite grows against it.

Importantly, we found that the outgrowth of both neurites requires DSH-1, which has two seemingly paradoxical functions in both promoting and inhibiting Wnt activity. Two Dsh paralogs, DSH-1 and MIG-5, redundantly mediate the intracellular response to the repulsive activities of Wnt proteins. This repulsive activity promotes the production of a long, anteriorly directed neurite in PLM and other tail neurons. At the same time, DSH-1 acts specifically in the posterior neurite to attenuate the strength of Wnt signaling, allowing it to grow against the repelling gradients. The dsh-1 mig-5 redundancy meant that dsh-1 mutants only displayed a loss of the inhibitory Dsh function, so this activity could be studied in isolation.

A dual role for Dsh proteins provides explanations to several previously puzzling observations. For example, the axons of posterior d-type motor neurons in C. elegans are repelled by LIN-44/Wnt; however, mutations in Dsh genes mig-5 and dsh-1 resulted in the underextension phenotype, opposite to the overextension phenotype observed in lin-44 mutants (7). Results from our study suggest that in this scenario, MIG-5 and DSH-1 (or the three Dsh proteins) may act redundantly to mediate the repulsive activities of LIN-44, but both also carry the inhibitory functions. A dual function could explain the opposing phenotypes between Dsh and Wnt mutants. When Wnt proteins act as attractive cues, Dsh proteins also play both positive and negative roles, depending on the cellular contexts. The mushroom body axons and embryonic sensory axons in Drosophila require Dsh for Wnt-stimulated outgrowth (20, 21), whereas mammalian Dvl1 inhibits Wnt5a-stimulated growth of commissural axons (24). Although our model is consistent with the regulatory function of Dvl1 in providing feedback inhibition of Fzd3, we do not know whether Dvl1 also positively contributes to Wnt signaling. A more recent study found that Dvl2 positively promotes Wnt/PCP signaling in the commissural axons and antagonizes Dvl1-mediated inhibition of Fzd3 (38), which suggests that the dual function of Dsh proteins could also be executed separately by distinct Dsh paralogs.

Because the shortened PLM posterior neurite in dsh-1 mutants represents a state of hyperactive Wnt signaling, these animals provided a sensitized background to identify genes that positively contribute to the actions of Wnt proteins. We found that the PCP core components VANG-1, PRKL-1, and FMI-1 promoted the repulsion of neurites, but the relatively weak effects of their loss compared with Wnt and Fzd suggest they play modulatory, instead of essential, roles in Wnt signaling. This modulation involves antagonizing the inhibitory function of DSH-1 and reducing LIN-17/Fzd phosphorylation. Consistent with our results, a previous study reported that Vangl2 in mouse commissural axons blocked the Dvl1-mediated inhibition of Fzd3 (24). Because Vang physically interacts with Prkl and recruits it to the membrane (39), our studies further establish that the Vang/Prkl complex positively promotes Wnt/PCP signaling and axon guidance by Wnt proteins. Furthermore, Fmi, a cadherin-like membrane protein that serves as coreceptors for Fzd and Vang (40), promotes the function of both in neurite guidance.

Our results suggest that the effectors downstream of Wnt signaling are most probably highly redundant. First, in the saturated screen that searched for mutants with TRN morphological defects and identified dsh-1 alleles, we isolated four lin-44 alleles and four lin-17 alleles but did not find mutants of any other genes that showed a similar PLM Wnt phenotype. Second, the bar-1 jnk-1 daam-1 triple mutants, in which both the canonical pathway and PCP signaling pathway are blocked, still failed to reproduce the lin-44 phenotype, although an intermediate phenotype with a markedly elongated PLM posterior neurite was observed. These results suggest that a strong genetic redundancy and/or compensating mechanism exist downstream of the Wnt/Fzd/Dvl complex, which ensures a robust output of Wnt signaling.

At least six types of neurons [PLM, ALN, PLN (this study), DD6, VD12, and VD13 (7)] require Dsh proteins to grow toward the posterior and against the gradient of repulsive LIN-44 concentration. Given the specific expression of LIN-44 from the tail hypodermis (8, 29), we propose that Dsh-mediated attenuation of the Wnt signaling is a common mechanism used by the posteriorly directed neurite to navigate against the Wnt gradients. This region-specific activity suggests that fine-tuning the strength of Wnt activity by intracellular mechanisms may be essential for the local shaping of the posterior neuroanatomy.

Materials and Methods

C. elegans wild-type (N2) and mutant strains were maintained at 20 °C, as previously described (41). Most strains were provided by the Caenorhabditis Genetics Center, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). To create dsh-1 mig-5 and dsh-1 dsh-2 double mutants, constructs that express CRISPR/Cas9 with sgRNAs targeting 5′-GAGAAGGAGTAGCGACGCTTGG-3′ in exon 2 of mig-5 and 5′-GATGTTTCTAACATTTATGTGG-3′ in exon 1 of dsh-2, respectively, were injected into dsh-1(u915) mutants, using a previously described method (33). Most of the constructs were made using the Gateway cloning system (Life Technologies). Q5 Site-Directed Mutagenesis Kit (New England Biolabs) was used to create constructs that express dsh-1a and lin-17a mutants. Transgenes uIs31[mec-17p::GFP] III, uIs115[mec-17p::RFP] IV, and uIs134[mec-17p::RFP] V were used to visualize TRN morphology. Transgenes jsIs821[Pmec-7::GFP::rab-3] and uIs178[mec-17p::unc-9::GFP; mec-17p::RFP] were used to label presynaptic vesicles and gap junctions, respectively. The relative length of PLM neurites was calculated by dividing the length of the neurite by the diameter of PLM cell body along the A-P axis. To test transgenic animals, DNA constructs were injected at a concentration of 5 ng/μL (unless otherwise indicated) into the animals to establish stable lines carrying the extrachromosomal array. For statistical analysis, ANOVA and a post hoc Tukey–Kramer method were used to identify significant difference between the sample means in multiple comparisons. Single and double asterisks indicated P < 0.05 and P < 0.01, respectively. More details are provided in SI Materials and Methods.

SI Materials and Methods

Strains.

C. elegans wild-type (N2) and mutant strains were maintained at 20 °C, as previously described (40). Most strains were provided by the Caenorhabditis Genetics Center, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440). dsh-1 alleles u915, u952, and u953 were isolated in a screen by visually searching for mutants with TRN morphological defects using a TU4069 strain that carries uIs134[mec-17p::RFP] transgene. daam-1(gk960790) is a deletion allele isolated from the Million Mutation Project (43) and was outcrossed at least three times before characterization.

To create dsh-1 mig-5 and dsh-1 dsh-2 double mutants, constructs that express CRISPR/Cas9 with sgRNAs targeting 5′-GAGAAGGAGTAGCGACGCTTGG-3′ in exon 2 of mig-5 and 5′-GATGTTTCTAACATTTATGTGG-3′ in exon 1 of dsh-2, respectively, were injected into dsh-1(u915) mutants, using a previously described method (32); the resulting mutations were identified by genotyping (Fig. S2).

Constructs and Transgenes.

Most of the constructs were made using the Gateway cloning system (Life Technologies). A 1.9-kb mec-17 promoter was used to drive TRN-specific expression, and a 2.2-kb egl-20 promoter and a 2.1-kb mom-2 promoter were used to misexpress lin-44. These promoters were cloned into pDONR P4-P1r vectors. lin-44 genomic DNA and cDNAs of dsh-1a and lin-17a were cloned into Gateway pDONR221. un-54 3′UTR and the destination vector pDEST-R4-R3 were used to generate the final expression vector. The Q5 Site-Directed Mutagenesis Kit (New England Biolabs) was used to create constructs that express dsh-1a and lin-17a mutants.

Transgenes uIs31[mec-17p::GFP] III, uIs115[mec-17p::RFP] IV, and uIs134[mec-17p::RFP] V were used to visualize TRN morphology. Transgene uIs129[lad-2p::GFP], which is generated by integrating otEx331 into the genome, was used to visualize the morphology of ALN and PLN neurons. Transgene sIs12076 [C34F11.9a::gfp] was used to monitor dsh-1 expression (44). Transgenes jsIs821[Pmec-7::GFP::rab-3] and uIs178[mec-17p::unc-9::GFP; mec-17p::RFP] were used to label presynaptic vesicles and gap junctions, respectively.

Phenotype Scoring and Statistical Analysis.

The relative length of the PLM posteriorly directed neurite was calculated by measuring the length of the process and dividing the length by the diameter of the PLM cell body along the A-P axis. At least 30 gravid adults were measured. To test transgenic animals, DNA constructs were injected at a concentration of 5 ng/μL (unless otherwise indicated) into the animals to establish stable lines carrying the extrachromosomal array. At least three independent lines were tested. For statistical analysis, ANOVA and a post hoc Tukey–Kramer method were used to identify significant difference between the sample means in multiple comparisons. Single and double asterisks indicated P < 0.05 and P < 0.01, respectively.