ABSTRACT
Axons must correctly reach their targets for proper nervous system function, although we do not fully understand the underlying mechanism, particularly for the first ‘pioneer’ axons. In C. elegans, AVG is the first neuron to extend an axon along the ventral midline, and this pioneer axon facilitates the proper extension and guidance of follower axons that comprise the ventral nerve cord. Here, we show that the ubiquitin ligase RPM-1 prevents the overgrowth of the AVG axon by repressing the activity of the DLK-1/p38 MAPK pathway. Unlike in damaged neurons, where this pathway activates CEBP-1, we find that RPM-1 and the DLK-1 pathway instead regulate the response to extracellular Wnt cues in developing AVG axons. The Wnt LIN-44 promotes the posterior growth of the AVG axon. In the absence of RPM-1 activity, AVG becomes responsive to a different Wnt, EGL-20, through a mechanism that appears to be independent of canonical Fz-type receptors. Our results suggest that RPM-1 and the DLK-1 pathway regulate axon guidance and growth by preventing Wnt signaling crosstalk.
INTRODUCTION
Proper nervous system development requires that neurons extend axons to precise targets. Groups of neurons exhibit similar patterns of axon outgrowth depending on their spatial position and sequential timing during development. Growing ‘pioneer’ axons often blaze pathways through the nervous system, with ‘follower’ axons extending along these pathways using the pioneers as a guide (Chitnis and Kuwada, 1991; Hidalgo and Brand, 1997; Hutter, 2003; Klose and Bentley, 1989; McConnell et al., 1989). Extracellular signals act as guidance cues for growing axons, yet how the nervous system employs a small number of guidance cues over a vast array of neuron types and developmental time points remains poorly understood (Kaplan et al., 2014; Morales and Kania, 2017; Petrovic and Schmucker, 2015). Modulation of the signal transduction pathways downstream of these guidance cues might provide one explanation for how the same cues are used in multiple contexts.
In the nematode Caenorhabditis elegans, one well-studied pioneer neuron is AVG. The AVG soma is in the retrovesicular ganglion, is the first neuron to extend an axon along the ventral midline, and is required for the guidance of the follower axons that comprise the ventral nerve cord (Durbin, 1987; Hutter, 2003; White et al., 1986). Conserved extracellular guidance cues, including Netrin/UNC-6, Slit/SLT-1 and TGFβ/UNC-129, guide the polarized migration of axons (Colavita et al., 1998; Garriga et al., 1993; Hao et al., 2001; Hedgecock et al., 1990; Ishii et al., 1992; Wadsworth, 2002). However, few such cues have been identified for pioneers like AVG (Hutter, 2003).
Whereas AVG axon guidance does not depend on established cues, the polarity of the AVG soma and the site of axon sprouting is regulated by the ubiquitin ligase PLR-1, which inhibits the surface expression of the Wnt receptors Ror/CAM-1 and Ryk/LIN-18 (Bhat et al., 2015; Moffat et al., 2014). Wnts are secreted glycoproteins that act as long-range signals in development (Gammons and Bienz, 2017; Loh et al., 2016). In the absence of PLR-1, the AVG neuron inappropriately responds to the Wnts CWN-1 and CWN-2, reorienting its polarity towards these anteriorly expressed cues during embryogenesis (Harterink et al., 2011; Kennerdell et al., 2009; Song et al., 2010). Unlike for AVG cell polarity, the factors that regulate AVG anterior-posterior axon guidance remain unknown.
One known regulator of anterior-posterior axon growth is RPM-1, a founding member of the Pam/Highwire/RPM-1 protein family (Grill et al., 2016; Schaefer et al., 2000). Loss of RPM-1 function results in abnormal neuronal morphology, axon overgrowth, impaired presynaptic differentiation, and internalization of postsynaptic AMPA receptors (Opperman and Grill, 2014; Park et al., 2009; Schaefer et al., 2000; Zhen et al., 2000). RPM-1 is a signaling hub with multiple protein-protein interaction domains. RPM-1 binds the F-box protein FSN-1, and together they act as a ubiquitin ligase to target the DLK-1 MAPKKK for degradation (Liao et al., 2004; Nakata et al., 2005). In the absence of RPM-1, stabilized DLK-1 acts through the MAPKK MKK-4 and the p38 MAPK PMK-3 to activate the translation of the transcription factor CEBP-1 (Yan et al., 2009). This MAP kinase pathway, which is activated in response to damage, promotes axon regeneration; however, its role in normal axon outgrowth is unclear. RPM-1 has a role in normal axon growth, as rpm-1 mutants have PLM mechanosensory neurons with overgrown axons and DA/DB motor neurons with dorsal-ventral axon guidance defects (Li et al., 2008; Schaefer et al., 2000). RPM-1 regulates these axons by reducing the surface levels of Robo/SAX-3 and UNC-5 guidance receptors (Li et al., 2008). This function is not mediated through FSN-1 or the DLK-1 MAP kinase cascade; rather, RPM-1 separately binds to the Rab GEF GLO-4, which activates the Rab GLO-1 to regulate axon guidance and outgrowth (Grill et al., 2007; Li et al., 2008).
Here, we reveal a new function for RPM-1: guidance and growth of the AVG pioneer axon through regulation of Wnt signaling crosstalk. Normally, AVG makes specific turns in the posterior and arrests growth at the lumbar ganglion. In the absence of RPM-1, AVG overgrows past the lumbar ganglion and into the tip of the tail. AVG overgrowth in rpm-1 mutants is not mediated through GLO-4, but through the overactivation of the DLK-1/p38 MAPK pathway. Surprisingly, the downstream target of this pathway, CEBP-1, is not required. Instead, the Wnt LIN-44 and its Fz-type receptor LIN-17 promote posterior growth of the AVG axon, and in loss-of-function rpm-1 mutants, AVG responds to an additional Wnt ligand, EGL-20, resulting in overgrowth. Our results indicate that RPM-1 inhibits AVG axon growth by modulating Wnt/Fz signaling, controlling the timing and specificity of ligand/receptor sensitivity.
RESULTS
RPM-1 and DLK-1 signaling regulates AVG axon outgrowth
To analyze AVG axon growth, we examined kyIs51[Podr-2b::GFP] and otIs182[Pinx-18::GFP] transgenic animals, both of which express GFP in AVG and a handful of other neurons (Chou et al., 2001; Sarin et al., 2009). In wild type, AVG extends its axon during embryogenesis (Durbin, 1987). By the L1 larval stage, its axon has extended beyond the end of the intestine and preanal ganglion, exited the ventral midline, and migrated dorsally towards the dorsorectal ganglion (Fig. 1A). When AVG reaches the DVA soma in the dorsorectal ganglion, it then turns and migrates posteriorly to the end of the lumbar ganglion before stopping (Fig. 1B-G). The cues for this guidance are unknown.
RPM-1 prevents AVG axon posterior overgrowth. (A) Diagram of the C. elegans posterior, including the position of neuron cell bodies (circles) for the preanal ganglion (dark blue), dorsorectal ganglion (light blue) and lumbar ganglion (gray). The secretion of Wnt/LIN-44 from the tail hypodermis is indicated by red shading. The green line indicates the wild-type AVG after the completion of growth and guidance during embryogenesis. (B-G) L4-stage nematodes expressing GFP in the AVG axon (arrow) (B,E) and DsRed2 in the dorsorectal ganglion (C,F) from the indicated transgenes. Images in B-D and E-G are from slightly different focal planes to allow the separate visualization of the DVA, DVB and DVC neuron cell bodies (arrows). (D,G) DIC image. (H,I) GLR-1::GFP in wild type (H) and rpm-1(od14) mutants (I) in the PVC neuron cell body and the AVG axon (arrows). (J) Percentage of animals with the indicated AVG axon guidance phenotype. Red bars indicate animals in which the AVG axon is overgrown to the tail tip. Gray bars indicate animals in which the AVG axon stops growth in the lumbar ganglion (the wild-type phenotype). Light blue bars indicate animals in which the AVG axon stops at the dorsorectal ganglion. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants. n=50 animals/genotype. Scale bars: 20 μm.
RPM-1 prevents AVG axon posterior overgrowth. (A) Diagram of the C. elegans posterior, including the position of neuron cell bodies (circles) for the preanal ganglion (dark blue), dorsorectal ganglion (light blue) and lumbar ganglion (gray). The secretion of Wnt/LIN-44 from the tail hypodermis is indicated by red shading. The green line indicates the wild-type AVG after the completion of growth and guidance during embryogenesis. (B-G) L4-stage nematodes expressing GFP in the AVG axon (arrow) (B,E) and DsRed2 in the dorsorectal ganglion (C,F) from the indicated transgenes. Images in B-D and E-G are from slightly different focal planes to allow the separate visualization of the DVA, DVB and DVC neuron cell bodies (arrows). (D,G) DIC image. (H,I) GLR-1::GFP in wild type (H) and rpm-1(od14) mutants (I) in the PVC neuron cell body and the AVG axon (arrows). (J) Percentage of animals with the indicated AVG axon guidance phenotype. Red bars indicate animals in which the AVG axon is overgrown to the tail tip. Gray bars indicate animals in which the AVG axon stops growth in the lumbar ganglion (the wild-type phenotype). Light blue bars indicate animals in which the AVG axon stops at the dorsorectal ganglion. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants. n=50 animals/genotype. Scale bars: 20 μm.
We previously identified alleles of rpm-1 in a screen for mutants with defective GLR-1 synaptic localization (Park et al., 2009). GLR-1 is expressed in AVG, and we noticed that rpm-1 mutants expressing GLR-1::GFP had an elongated axon that extended to the posterior tip of the tail (Fig. 1H,I). We classified animals based on whether AVG terminated growth at the preanal ganglion or dorsal rectal ganglion (indicative of undergrowth), the lumbar ganglion (the normal termination point in wild type), or the most posterior tip of the tail (indicative of overgrowth) (Fig. 1J). All rpm-1 mutants have AVG axons that overgrow to the tail tip. Expression of a wild-type rpm-1 cDNA from the glr-1 promoter, which is expressed in AVG and a handful of other interneurons (Hart et al., 1995; Maricq et al., 1995), restored normal AVG migration in these mutants (Fig. 1J).
To characterize AVG axon guidance more thoroughly, we introduced kyIs51 into rpm-1(ok364) deletion (molecular null) mutants (Chou et al., 2001). We used DAPI staining to visualize neuronal and hypodermal nuclei in the tail, using conditions that preserved GFP fluorescence. Unlike in wild type, the AVG axon in rpm-1(ok364) null mutants extends posteriorly to the tail tip (Fig. 2A-C), similar to what we observed in rpm-1(od14) mutants expressing GLR-1::GFP. Interestingly, we often observed the AVG axon making a U-turn and migrating anteriorly back along the same ventral midline. Taken together, these results indicate that RPM-1 acts cell autonomously to stop AVG axon migration.
RPM-1 regulates AVG axon growth through DLK-1/p38 MAPK signaling. (A,B) Wild-type (A) and rpm-1 mutant (B) L4-stage nematodes expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue). The arrow points to a second fiber: the AVG axon making its U-turn. (C,D) Percentage of animals with the indicated phenotype. Red bars indicate animals in which the AVG axon is overgrowth to the tail tip. Gray bars indicate animals in which the AVG axon stops growth in the lumbar ganglion (the wild-type phenotype). Light blue bars indicate animals in which the AVG axon stops at the dorsorectal ganglion. Dark blue bars indicate animals in which the AVG axon stops at the preanal ganglion. ***P<0.001, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants or for the indicated comparison. n=50 animals/genotype. Scale bars: 20 μm.
RPM-1 regulates AVG axon growth through DLK-1/p38 MAPK signaling. (A,B) Wild-type (A) and rpm-1 mutant (B) L4-stage nematodes expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue). The arrow points to a second fiber: the AVG axon making its U-turn. (C,D) Percentage of animals with the indicated phenotype. Red bars indicate animals in which the AVG axon is overgrowth to the tail tip. Gray bars indicate animals in which the AVG axon stops growth in the lumbar ganglion (the wild-type phenotype). Light blue bars indicate animals in which the AVG axon stops at the dorsorectal ganglion. Dark blue bars indicate animals in which the AVG axon stops at the preanal ganglion. ***P<0.001, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants or for the indicated comparison. n=50 animals/genotype. Scale bars: 20 μm.
RPM-1 regulates mechanosensory axon guidance by binding to the Rab GLO-1 and its GEF GLO-4 (Grill et al., 2007; Li et al., 2008). We did not observe axon overgrowth in either glo-1 or glo-4 mutants (Fig. 2C). RPM-1 also acts as an ubiquitin ligase through its binding to the F-box protein FSN-1 (Liao et al., 2004; Nakata et al., 2005). We noted that fsn-1 mutants have a similar overgrowth phenotype to that of rpm-1 mutants (Fig. 2C), suggesting that the ubiquitin ligase function of RPM-1 regulates AVG growth.
RPM-1 and FSN-1 ubiquitylate the MAPKKK DLK-1, thereby inhibiting DLK/p38 MAPK signaling. Although loss-of-function mutations in dlk-1, mkk-4 and pmk-3 alone had no effect on AVG axon outgrowth, mutations in any of these genes suppressed the rpm-1 phenotype (Fig. 2C). Moreover, overexpression of a transgenic, wild-type dlk-1 cDNA from the glr-1 promoter in an otherwise wild-type animal was sufficient to cause AVG axon overgrowth (Fig. 2C). RPM-1 and the DLK/p38 MAPK cascade regulate presynaptic differentiation by activating the kinase MAK-2 and promoting the translation and activity of the transcription factor CEBP-1 (Yan et al., 2009). Surprisingly, mutations in either mak-2 or cebp-1 did not suppress the rpm-1 overgrowth phenotype (Fig. 2C). The DLK/p38 MAPK pathway can promote axon regeneration by activating PARG-1 and PARG-2 (Byrne et al., 2016). We found that mutations in parg-1 and parg-2 did not result in AVG axon growth defects and did not suppress axon overgrowth in rpm-1 mutants (Fig. 2D). Finally, DLK MAP kinase signaling can promote endocytosis through activation of the small GTPase RAB-5 (Park et al., 2009; van der Vaart et al., 2015). Although expression of a constitutively active RAB-5 mutant protein mimics the effects of rpm-1 mutations on GLR-1 recycling, it does not mimic the AVG axon overgrowth phenotype (Fig. 1J). Similarly, although expression of a dominant-negative, GDP-locked mutant form of RAB-5 can suppress the effects of rpm-1 on GLR-1 recycling, the same transgene does not suppress the AVG axon overgrowth phenotype of rpm-1 (Fig. 1J). Thus, RPM-1 prevents AVG axons from overgrowing by repressing DLK/p38 MAPK signaling that otherwise would promote axon growth by a novel mechanism.
Mutants for the cytoskeleton regulators NAV2/UNC-53 and Trio/UNC-73 have AVG axons that show some midline-crossing defects (Bhat et al., 2015). We found that a small number of unc-53(e2432) and unc-73(ev801) mutants terminated AVG axon growth prematurely, either at the dorsal rectal ganglion or the end of the preanal ganglion (Fig. 2D). Mutations in unc-53 strongly suppressed axon overgrowth in rpm-1, whereas unc-73 mutations showed only mild suppression (Fig. 2D), suggesting that the overgrowth observed in rpm-1 mutants requires UNC-53.
Wnt/LIN-44 signaling promotes AVG axon outgrowth
RPM-1 regulates the growth of some axons by regulating the signaling of guidance cues such as Netrin/UNC-6 and its receptor UNC-5 (Li et al., 2008; Tulgren et al., 2014). However, AVG axon growth is normal in 98% of unc-5(e53) and 96% of unc-6(ev400) mutants (n=50 animals for each genotype). A better candidate guidance cue is the Wnt LIN-44, which is expressed and secreted (Fig. 1A) from hypodermal cells in the tail of comma-stage embryos and larva (Harterink et al., 2011; Herman et al., 1995; Klassen and Shen, 2007). We examined lin-44(n1792) loss-of-function (null) mutants and observed that 96% of mutant animals had AVG axons that terminated prematurely at either the preanal ganglion (Fig. 3A) or the dorsal rectal ganglion (Fig. 3B). A wild-type Plin-44::gfp::lin-44 transgene robustly rescued the defects of lin-44 mutants (Fig. 3F). The Wnt EGL-20 is expressed near the rectum in comma-stage embryos and larva, where it also could act as a guidance cue (Harterink et al., 2011; Herman et al., 1995; Klassen and Shen, 2007). We examined egl-20(n585) mutants but observed wild-type AVG axon growth (Fig. 3F). Taken together, our results suggest that Wnt LIN-44 acts as a distal attractive cue for the AVG axons to migrate beyond the preanal ganglion.
Wnt/LIN-44 signaling promotes AVG axon outgrowth. (A,B) L4-stage mutants for lin-44 expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue), arrest growth at either the preanal ganglion (A) or the dorsorectal ganglion (B). (C) L4-stage mutant for lin-17 expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue). (D,E) LIN-17::GFP is localized to the growing axon tip of AVG (arrow). (F) Percentage of animals with the indicated phenotype. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with lin-44 or lin-17 mutants. n=50 animals/genotype. Scale bars: 20 μm.
Wnt/LIN-44 signaling promotes AVG axon outgrowth. (A,B) L4-stage mutants for lin-44 expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue), arrest growth at either the preanal ganglion (A) or the dorsorectal ganglion (B). (C) L4-stage mutant for lin-17 expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue). (D,E) LIN-17::GFP is localized to the growing axon tip of AVG (arrow). (F) Percentage of animals with the indicated phenotype. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with lin-44 or lin-17 mutants. n=50 animals/genotype. Scale bars: 20 μm.
The Fz receptor LIN-17 acts cell-autonomously to promote AVG axon outgrowth
Wnts regulate axon guidance through either Frizzled (Fz) receptors or Ryk/Derailed receptors (Ackley, 2014; Onishi et al., 2014; Wang et al., 2016). The Wnt LIN-44 acts as a repulsive cue through Fz LIN-17 in D-type motor neurons (Maro et al., 2009; Sawa et al., 1996). We examined lin-17(n671) loss-of-function (null) mutants and found that AVG terminated growth at the preanal ganglion (Fig. 3C,F). We generated transgenes, Pglr-1::LIN-17 and Podr-2b::LIN-17, that express wild-type LIN-17 cDNA in AVG, introduced these transgenes into lin-17 mutants, and found that they robustly restored AVG axon growth, indicating that LIN-17 acts cell-autonomously in AVG to promote posterior axon growth (Fig. 3F). We also generated Pglr-1::LIN-17::GFP, which expressed a tagged receptor that localized to the posterior axonal growth cone of AVG (Fig. 3D,E), consistent with LIN-17 functioning as a sensor of axon guidance cues.
There are three additional Fz-type receptors (MIG-1, MOM-5 and CFZ-2) in C. elegans (Gleason et al., 2006; Pan et al., 2006; Zinovyeva and Forrester, 2005). We observed premature AVG axon termination at the dorsal rectal ganglion in about one-third of mig-1(e1787) mutants, whereas we observed no statistically significant defects in mom-5(gk812) or cfz-2(ok1201) mutants (Fig. 3F). We conclude that initial AVG axon outgrowth from its anterior soma along the ventral midline does not require Wnt/Fz signaling, but the Fz-type receptor LIN-17 and (to some degree) the Fz-type receptor MIG-1 promote the AVG axon to grow past the preanal ganglion and into the tail.
Atypical Wnt signaling promotes AVG axon outgrowth via TCF/POP-1
The Wnt LIN-44 is a repulsive cue for the D-type motor neurons through the downstream effectors Dsh/MIG-5, GSK-3, Axin/PRY-1, β-catenin/BAR-1 and TCF/POP-1 (Maro et al., 2009). C. elegans has three Dsh proteins: DSH-1, DSH-2 and MIG-5. Unlike in D-type neurons, AVG axon guidance is, based on statistical significance, normal in mig-5(rh94) mutants (Fig. 4A). By contrast, the AVG axon terminates at the preanal ganglion in most dsh-1(ok1445) and dsh-2(ok2162) loss-of-function (null) mutants (Fig. 4A). In canonical Wnt signaling, Dsh sequesters a destruction complex, including the kinase GSK3 and the scaffolding molecule Axin, thereby preventing it from targeting β-catenin for proteolysis. Surprisingly, AVG axon growth looked statistically normal in gsk-3(nr2047), pry-1(mu38), axl-1(tm1095) and apr-1(ok2970) loss-of-function (null) mutants (Fig. 4A). Thus, whereas Dsh is required for AVG axon guidance, the destruction complex – the key inhibitor of β-catenin – is not.
Atypical Wnt signaling promotes AVG axon outgrowth via TCF/POP-1. (A-D) Percentage of L4-stage animals with the indicated phenotype. *P<0.05, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ##P<0.01, ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with hmp-2 mutants (C) or pop-1 mutants (D). n=50 animals/genotype.
Atypical Wnt signaling promotes AVG axon outgrowth via TCF/POP-1. (A-D) Percentage of L4-stage animals with the indicated phenotype. *P<0.05, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ##P<0.01, ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with hmp-2 mutants (C) or pop-1 mutants (D). n=50 animals/genotype.
Is β-catenin even required for AVG axon outgrowth? We expected that loss-of-function β-catenin mutants would show the same early termination of AVG growth as loss-of-function lin-44 and lin-17 mutants. C. elegans has four different β-catenin orthologs (BAR-1, WRM-1, HMP-2 and SYS-1) with separable functions (Korswagen et al., 2000). Unlike in D-type neurons, AVG axon guidance is statistically normal in bar-1(ga80) loss-of-function (null) mutants (Fig. 4B). Whereas transgenic animals that overexpress wild-type BAR-1 show synaptic defects in AVG and other command interneurons (Dreier et al., 2005), they have normal AVG axon guidance (Fig. 4B). Null mutants for sys-1 are embryonic lethal, precluding analysis of AVG; however, the partial loss-of-function mutation sys-1(q544) results in viable homozygous animals with reduced Wnt signaling (Kidd et al., 2005; Miskowski et al., 2001). We observed weakly penetrant defects in AVG guidance in sys-1(q544) single mutants and sys-1(q544) bar-1(ga80) double mutants (Fig. 4B). Null mutants for wrm-1 are also embryonic lethal, although wrm-1(ne1982) mutants are viable and show a temperature-sensitive embryonic-lethal phenotype (Nakamura et al., 2005; Wu and Herman, 2006). We examined wrm-1(ne1982) mutants at (20°C), an intermediate temperature that results in embryonic lethality for some but not all animals, but we did not observe AVG axon guidance defects that were statistically significant (Fig. 4B). WRM-1 is an atypical β-catenin that functions as a regulatory subunit of the LIT-1 MAPK (Rocheleau et al., 1999; Yang et al., 2015). We examined lit-1(or131) partial loss-of-function mutants at 20°C but did not observe AVG axon guidance defects (Fig. 4B). As sys-1 and wrm-1 mutants are embryonic lethal, and given that AVG axon guidance occurs during embryogenesis, we have not been able to examine the sys-1 and wrm-1 temperature-sensitive mutations at the most restrictive temperature. Cell-specific RNAi (Esposito et al., 2007) against sys-1 or wrm-1 also only resulted in weakly penetrant AVG axon defects (Fig. S1A).
In C. elegans, HMP-2 is the sole β-catenin that interacts with the α-catenin HMP-1 and the cadherin HMR-1 at adherens junctions (Costa et al., 1998; Korswagen et al., 2000; Natarajan et al., 2001). Surprisingly, hmp-2(qm39) partial loss-of-function mutants showed overgrowth of the AVG axon past the lumbar ganglion and into the tail tip (Fig. 4C), the opposite phenotype that would be expected if HMP-2 were a canonical β-catenin effector of this pathway. If Wnt LIN-44 and Fz LIN-17 signaling were transduced through HMP-2 in AVG, then hmp-2 mutations should be epistatic to lin-44, lin-17 and dsh-1/2 mutations. We examined double mutants between hmp-2(qm39) and either lin-44(n1792), lin-17(n671) or dsh-1(ok1445). All three of these double mutant combinations resulted in early termination of AVG axon growth at the preanal ganglion, suggesting that lin-44, lin-17 and dsh-1 are epistatic to hmp-2 (Fig. 4C). One explanation for why hmp-2 mutants have axon overgrowth defects could be that AVG axon guidance is sensitive to cadherin/catenin adhesion; however, AVG axon growth appears normal in mutants for hmp-1, which encodes α-catenin (Lockwood et al., 2008; Fig. 4C). Our results suggest that HMP-2 inhibits AVG axon overgrowth, but does so not as a β-catenin effector acting downstream of Wnt/LIN-44 and Fz/LIN-17, or as part of a cadherin/catenin adhesion complex.
Canonical Wnt signaling also acts through TCF, which forms a complex with β-catenin, binding specific sites in DNA to regulate transcription (Gammons and Bienz, 2017; Loh et al., 2016). C. elegans has a single TCF gene called pop-1 (Lin et al., 1998, 1995; Rocheleau et al., 1997). We examined AVG in two different mutants: pop-1(hu9) and pop-1(q645) (Korswagen et al., 2002; Siegfried and Kimble, 2002). AVG stops early at either the preanal ganglion or the dorsal rectal ganglion in both mutants (Fig. 4D). We generated a transgene, Podr-2b::LIN-17, that expresses wild-type POP-1 cDNA in AVG, introduced it into pop-1 mutants, and found that it robustly restored AVG axon growth, indicating that POP-1 acts cell-autonomously in AVG to promote posterior axon growth. Overexpression of either mammalian TCF or its C. elegans ortholog POP-1 lacking the amino-terminal β-catenin interaction domain results in a potent dominant-negative phenotype (Clevers and van de Wetering, 1997; Korswagen et al., 2000; Molenaar et al., 1996). Thus, we also generated a transgene containing dominant-negative POP-1 (ΔN-POP-1) lacking the first 44 amino acids and expressed from the glr-1 promoter. Transgenic animals show an AVG axon growth arrest (Fig. 4D). We introduced the pop-1(q645) mutation into hmp-2(qm39) mutants and found that AVG axons stop early in their migration in pop-1 hmp-2 double mutants (Fig. 4C). Taken together, POP-1 acts in AVG to promote axon outgrowth, although whether it requires a β-catenin co-activator remains unclear.
RPM-1 regulates AVG sensitivity to specific Wnts
If Wnt/LIN-44 promotes posterior AVG axon growth past the preanal ganglion, then RPM-1 might prevent overgrowth by inhibiting Wnt signaling. To test this hypothesis, we generated lin-44(n1792) rpm-1(ok364) double mutants. Like in wild type (Fig. 5A,B), the AVG axon in rpm-1 single mutants grows past the preanal ganglion, turns up through the dorsal rectal ganglia, and migrates through the lumbar ganglion (Fig. 5C). Unlike in wild type, the axon in rpm-1 mutants then continues to grow past the lumbar ganglion to the tail tip at L3 stage and beyond (Fig. 5D). By contrast, the axon in lin-44 mutants stops at the preanal ganglion in L1 (Fig. 5E) and does not turn dorsal at any subsequent stage (Fig. 5F). The AVG axon stops prematurely at the preanal ganglion in lin-44 rpm-1 double mutants in L1 larvae (Fig. 5G), like the phenotype observed in lin-44 single mutants. This arrest appears temporary, however, as AVG resumes growth and extends into the tail tip starting at the L2 stage and onward into adulthood, without turning to grow up through the dorsal rectal ganglion (Fig. 5H). Thus, the Wnt LIN-44 is required for AVG to grow dorsally after reaching the preanal ganglion, but AVG responds to other cues to grow posteriorly after a period of time if LIN-44 and RPM-1 are impaired.
RPM-1 regulates AVG sensitivity to specific Wnts. (A-H) L1-stage (A,C,E,G) or L4-stage (B,D,F,H) nematodes expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue, with bracket indicating the lumbar ganglion). Genotypes include wild type (A,B), rpm-1 mutants (C,D), lin-44 mutants (E,F) and rpm-1 lin-44 double mutants (G,H). (I) Percentage of animals with the indicated phenotype for each indicated genotype and developmental stage. *P<0.001, **P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type, L1 stage. #P<0.05, ##P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants at L1 stage, or for the indicated comparison. n=50-100 animals/genotype. Scale bars: 20 μm.
RPM-1 regulates AVG sensitivity to specific Wnts. (A-H) L1-stage (A,C,E,G) or L4-stage (B,D,F,H) nematodes expressing GFP in the AVG axon (yellow), with nuclei stained for DAPI (blue, with bracket indicating the lumbar ganglion). Genotypes include wild type (A,B), rpm-1 mutants (C,D), lin-44 mutants (E,F) and rpm-1 lin-44 double mutants (G,H). (I) Percentage of animals with the indicated phenotype for each indicated genotype and developmental stage. *P<0.001, **P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type, L1 stage. #P<0.05, ##P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants at L1 stage, or for the indicated comparison. n=50-100 animals/genotype. Scale bars: 20 μm.
The Wnt EGL-20 is also expressed in the posterior of the nematode, although we did not observe defects in AVG axon growth in egl-20(n585) (Fig. 3F). We introduced egl-20(n585) mutations into animals with rpm-1(ok364) and found that these double mutants had the same axon overgrowth phenotype observed in rpm-1 single mutants (Fig. 5I). Reasoning that redundancy with LIN-44 Wnt signaling might obscure any contribution from EGL-20 Wnt signaling, we examined AVG axon growth in lin-44 egl-20 double mutants and lin-44 egl-20 rpm-1 triple mutants. Double mutants for lin-44 and egl-20 show a similar axon arrest phenotype to that observed in lin-44 single mutants even in older animals (Fig. 5I). Surprisingly, more than two-thirds of lin-44 egl-20 rpm-1 triple mutants show a similar axon arrest phenotype as in lin-44 single mutants even at later stages of development (Fig. 5I). Our data indicate that the AVG axon overgrowth observed in rpm-1 mutants is due to a combination of LIN-44 and EGL-20 Wnt signaling, as well as possibly other cues given that some overgrowth remains in the triple mutant.
AVG axon growth in egl-20 mutants is like that in wild type, and axon growth arrest is similar in lin-44 single mutants compared with lin-44 egl-20 double mutants, suggesting that normally AVG is not sensitive to the EGL-20 Wnt. However, in the absence of RPM-1 and LIN-44, the AVG neuron becomes sensitive to the EGL-20 Wnt signal, as EGL-20 is required for the resumption of AVG axon growth observed in older larvae. One possible explanation is that both Wnts are able to bind and activate the Fz-type receptor LIN-17, with LIN-44 being the preferred ligand. We examined axon growth in lin-17 rpm-1 double mutants and found that AVG arrests growth at the preanal ganglion in L1 larvae, but then resumes growth in L2 larvae, moving to the tail tip without turning dorsally first (Fig. 5I). We also examined lin-17 egl-20 rpm-1 triple mutants and found that their AVG axons arrested growth in the preanal ganglion throughout development, like in lin-44 egl-20 rpm-1 triple mutants (Fig. 5I). Thus, our results suggest that the reactivation of growth triggered by EGL-20 in rpm-1 mutants is unlikely to be through LIN-17.
We also observed the AVG axon not only stopping at the preanal ganglion in rpm-1 mutants that are also impaired for Wnt signaling (e.g. rpm-1 lin-44 egl-20 triple mutants), but making a U-turn at this position and continuing growth in the anterior direction (Fig. 6A-C), similar to the phenotype observed at the tail tip of rpm-1 single mutants. Neither the LIN-44/EGL-20 Wnt combination, nor the LIN-17 receptor, is required for this U-turn (Fig. 6D). However, DLK/p38 MAPK signaling is required, as pmk-3 lin-17 rpm-1 triple mutants and pmk-3 rpm-1 double mutants do not make the U-turn and grow anteriorly. Triple mutants for cebp-1, lin-17 and rpm-1 still make the U-turn regardless of where they arrest posterior growth, suggesting a CEBP-1-indendent role for DLK/p38 MAPK signaling in promoting this growth. It appears that absolute axon growth and the direction of axon growth become uncoupled in rpm-1 mutants.
RPM-1 regulates both guidance and growth of axons through disparate mechanisms. (A,B) The AVG axon grows posteriorly to the tail tip (arrow), makes a U-turn, and then grows anteriorly in rpm-1 single mutants (A) and rpm-1 lin-44 double mutants (B) (L4 stage shown). (C) L4-stage quintuple mutant for rpm-1, mig-1, lin-17, mom-5 and cfz-2. The AVG axon arrests posterior growth at the pre-anal ganglion (arrow), but then makes a U-turn and migrates anteriorly along the lateral body wall. (D) Percentage of L4-stage animals with the indicated phenotype. Black bars indicate that growth terminates without any subsequent change in direction regardless of the position of termination. Orange bars indicate that growth continues in the anterior direction after making a U-turn at either the tail tip or the preanal ganglion. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants. n=50 animals/genotype. Scale bars: 20 μm.
RPM-1 regulates both guidance and growth of axons through disparate mechanisms. (A,B) The AVG axon grows posteriorly to the tail tip (arrow), makes a U-turn, and then grows anteriorly in rpm-1 single mutants (A) and rpm-1 lin-44 double mutants (B) (L4 stage shown). (C) L4-stage quintuple mutant for rpm-1, mig-1, lin-17, mom-5 and cfz-2. The AVG axon arrests posterior growth at the pre-anal ganglion (arrow), but then makes a U-turn and migrates anteriorly along the lateral body wall. (D) Percentage of L4-stage animals with the indicated phenotype. Black bars indicate that growth terminates without any subsequent change in direction regardless of the position of termination. Orange bars indicate that growth continues in the anterior direction after making a U-turn at either the tail tip or the preanal ganglion. ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants. n=50 animals/genotype. Scale bars: 20 μm.
AVG axon overgrowth in rpm-1 mutants appears to be independent of Fz-type receptors
Is AVG using a different receptor to respond to EGL-20? There are four Fz-type receptors in C. elegans: LIN-17, MIG-1, MOM-5 and CFZ-2 (Gleason et al., 2006; Pan et al., 2006; Thorpe et al., 1997). AVG grows past the dorsal rectal ganglion in about one-third of mig-1(e1787) mutants (Fig. 3F). Mutants for mom-5(gk812) show normal AVG axon growth, as do mutants for cfz-2(ok1201) (Fig. 3F). Unlike for lin-17 mutations, which can temporarily arrest AVG axon growth at the L1 stage when introduced into rpm-1 mutants, mutations in any of the other three Fz-type receptors do not dramatically prevent axon overgrowth when placed in combination with rpm-1 mutations (Fig. 7A). A quadruple mutant for all four Fz-type receptors has arrested AVG axon growth like that of lin-17 single mutants (Fig. 7A). Surprisingly, AVG axon outgrowth still resumes in 75% of older rpm-1 mutants that lack all four Fz-type receptors (Fig. 7A), suggesting that EGL-20 might be activating a non-canonical Wnt receptor when RPM-1 is removed. The axon often makes a U-turn and continues growth in the anterior direction, suggesting that axon growth, regardless of direction, does not require the Fz-type receptors (Fig. 6C,D).
AVG axon overgrowth in rpm-1 mutants appears to be independent of Fz-type receptors. (A-D) Percentage of L4-stage animals with the indicated phenotype. ***P<0.001, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. #P<0.05, ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants or for the genotypes compared by brackets. Expression of wild-type pop-1 in AVG (via the odr-2b promoter) in rpm-1 pop-1 double mutants suppresses axon growth arrest and results in overgrowth similar to that in rpm-1 single mutants. n=50 animals/genotype.
AVG axon overgrowth in rpm-1 mutants appears to be independent of Fz-type receptors. (A-D) Percentage of L4-stage animals with the indicated phenotype. ***P<0.001, ****P<0.0001, Kruskal–Wallis/Dunn's correction compared with wild type. #P<0.05, ####P<0.0001, Kruskal–Wallis/Dunn's correction compared with rpm-1 mutants or for the genotypes compared by brackets. Expression of wild-type pop-1 in AVG (via the odr-2b promoter) in rpm-1 pop-1 double mutants suppresses axon growth arrest and results in overgrowth similar to that in rpm-1 single mutants. n=50 animals/genotype.
Wnts can signal through non-canonical receptors, including Ryk, ROR, the Flamingo/Vangl2/PCP pathway, and LRP5/6. We did not observe AVG axon guidance defects in mutants for C. elegans Ryk (lin-18), ROR (cam-1) or LRP5/6 (lrp-1 and lrp-2), and none of these mutations dramatically blocked the overgrowth defects caused by rpm-1 mutations (Fig. 7B,C). We examined Flamingo (fmi-1) and Vangl (vang-1) mutants, but did not observe AVG axon guidance defects. We were not able to examine fmi-1 rpm-1 double mutants due to their close linkage. However, vang-1 mutations did not block the overgrowth of AVG observed in rpm-1 mutants (Fig. 7C). We also examined combinatorial mutants for multiple Fz-type receptors, the non-canonical Ryk receptor lin-18, and rpm-1. Like in lin-17 rpm-1 double mutants, the AVG axon initially stops during L1 but then resumes growth during later larval stages (Fig. 7B). These results suggest that AVG axon overgrowth might occur in rpm-1 mutants even in the absence of all known Wnt receptors.
Finally, we also examined whether overgrowth in rpm-1 mutants is dependent on other Wnt signaling components even if it is not dependent on the receptors. Wntless/MIG-14 promotes the release of Wnts through retrograde recycling, and the secretion of most (perhaps all) Wnts is depressed in mig-14(ga62) mutants (Pan et al., 2008; Yang et al., 2008). AVG axons arrest prematurely in both mig-14 single mutants and mig-14 rpm-1 double mutants throughout development (Fig. 7D). Mutations in dsh-1 or pop-1 also prevent much of the overgrowth observed in rpm-1 mutants at all developmental stages (Fig. 7D). Thus, AVG axon posterior overgrowth in rpm-1 mutants requires downstream Wnt signaling components even though Wnt receptors are not required. Interestingly, mutations in pop-1 do not prevent the AVG axon from making a U-turn and growing in the anterior direction when RPM-1 is absent (Fig. 6D), supporting an additional, Wnt-independent role for RPM-1 in AVG axon growth.
DISCUSSION
The mechanism by which pioneer axons such as AVG navigate the nervous system during development is not understood. We found that the ubiquitin ligase comprising RPM-1 and FSN-1 keeps the AVG pioneer axon from overgrowing past the lumbar ganglion and into the tail tip. RPM-1 and FSN-1 repress the activity of a p38 MAPK pathway comprising DLK-1, MKK-4 and PMK-3, but do not act through CEBP-1. Instead, RPM-1 and the DLK-1 pathway regulate how AVG responds to the Wnts LIN-44 and EGL-20. LIN-44 promotes the posterior growth of the AVG axon. In the absence of RPM-1 activity, AVG responds to EGL-20, which appears to promote AVG axon growth even in the absence of all four Fz-type receptors. Taken together, our results suggest that RPM-1 and the DLK-1/p38 MAPK pathway regulate axon guidance and growth by modulating signal transduction downstream of distinct Wnt signals.
The Wnt LIN-44 is required for proper AVG axon guidance, yet its role is more complicated than that of a simple attractive guidance cue. In the absence of LIN-44 (and indeed all the Wnts), the AVG axon migrates along the ventral cord to the preanal ganglion, suggesting that signals other than Wnts guide the initial journey of the growth cone. After reaching the preanal ganglion, the AVG growth cone responds to extracellular LIN-44 to turn dorsally and migrate up to the dorsal rectal ganglion, then to turn and migrate posteriorly through the lumbar ganglion. Dorsal migration is orthogonal to the source of LIN-44 Wnt release, so it seems unlikely that the growth cone is simply moving towards the Wnt as an attractive cue.
Wnts can act as either attractive or repellant guidance cues depending on the specific circuit. In Drosophila, Wnt5 repels axons in the CNS through its interaction with the RYK receptor Derailed (Yoshikawa et al., 2003). In mammals, Wnts either attract or repel corticospinal axons along the anteroposterior axis through Fz-type or RYK receptors, respectively (Liu et al., 2005; Lyuksyutova et al., 2003; Schmitt et al., 2006). Wnts also repel axons in the corpus callosum via RYK receptors (Hutchins et al., 2011). Wnts can act as attractants for the dorsal root ganglion and monoaminergic neurons in the CNS in mice and in retinal photoreceptor cells in Drosophila (Fenstermaker et al., 2010; Lu et al., 2004; Sato et al., 2006). The Wnt EGL-20 acts as a repellant to guide the touch receptor neurites via the Fz-type receptors MIG-1 and MOM-5 in C. elegans (Pan et al., 2006; Prasad and Clark, 2006). Indeed, LIN-44 itself acts through the Fz-receptor LIN-17 to repel the D-type motor neurons from extending into the posterior tail (Maro et al., 2009).
Wnts typically act through the planar cell polarity (PCP) pathway to guide axon growth cones (Goodrich and Strutt, 2011; Onishi et al., 2014; Zallen, 2007). This seems an unlikely mechanism for how Wnt LIN-44 and Fz LIN-17 promote AVG axon outgrowth, as mutations in various conserved PCP components did not regulate AVG axon migration, whereas canonical Wnt-Fz downstream signaling components, such as TCF POP-1, are required for AVG axon migration.
The role of Wnt-Fz signaling in AVG axon growth likely requires changes in transcription. We found that Fz/LIN-17 and TCF POP-1 act cell-autonomously in AVG to promote axonal growth. TCF/POP-1 binds to β-catenin, and this resulting complex acts as a transcriptional activator on multiple target genes. Surprisingly, none of the mutants for the C. elegans β-catenin genes has penetrant defects in AVG migration out of the preanal ganglion. One possibility is that there is functional redundancy between these genes, although this has not been observed previously (Korswagen et al., 2000). Alternatively, we had to analyze sys-1 hypomorphs and cell-specific RNAi knockdowns rather than sys-1 null mutants to circumvent the problem of embryonic lethality observed in the nulls (Kidd et al., 2005; Miskowski et al., 2001). The sys-1 hypomorphs might possess enough β-catenin activity to accomplish AVG axon guidance for most animals, as a small percentage of these mutants did have AVG axon guidance defects. Finally, there are some reports of POP-1 acting as a transcriptional repressor independently of β-catenin (Calvo et al., 2001; Shetty et al., 2005; Yang et al., 2011). This explanation also seems unlikely, as the expression of a POP-1 protein lacking the β-catenin interaction domain created a dominant-negative effect on AVG axon growth, similar to the phenotype observed in pop-1 loss-of-function mutants. Regardless of whether POP-1 is a lone repressor or an activator working with SYS-1, it is not intuitive how a signal transduction pathway with a final output of a transcription factor could specify direction or polarity in a mononuclear cell.
We suggest that POP-1 regulates the expression of another receptor, which in turn makes the AVG axon competent to respond to other guidance cues present in the posterior of the animal (Fig. 8A). In this model, the AVG axon reaches the preanal ganglion, where it senses LIN-44. LIN-44 activates Fz LIN-17 on the growth cone, in turn activating DSH-1 and DSH-2 and sending POP-1 back to the AVG neuronal nucleus. POP-1 activates the expression of another receptor, which is then placed on the growth cone membrane to sense local guidance cues. Those cues and that hypothesized receptor then promote exit of the growth cone from the preanal ganglion and entry into the dorsal rectal ganglion and the lumbar ganglion. Given that the initial migration of the growth cone is dorsal, one candidate cue is Netrin/UNC-6; however, we found that mutations in either unc-5 or unc-6 had no effect on AVG migration into the dorsal rectal ganglion.
A model for the regulation of AVG axon guidance and growth by Wnts and RPM-1. (A) (1) As the AVG axon (red line) migrates posteriorly into the tail, it encounters Wnt/LIN-44 (red shading). (2) We hypothesize that Wnt/LIN-44 activity alters gene expression in AVG (indicated by the change in line color from red to green) such that AVG expresses a new receptor sensitive to dorsal guidance cues (green shading), (3) allowing the axon to grow in the dorsal direction towards the dorsorectal ganglion. (B) Our findings suggest that Wnt/LIN-44 regulates AVG axon guidance through Fz/LIN-17 signaling (solid arrows), resulting in POP-1-mediated transcriptional changes that promote dorsal axon guidance. RPM-1 prevents AVG overgrowth by preventing crosstalk from the Wnt/EGL-20 through an unknown receptor. RPM-1 has an additional role, independent of Wnt signaling, in restricting growth. Damage and stress can disinhibit these two mechanisms (dotted lines) to allow new axon growth and regeneration in older animals.
A model for the regulation of AVG axon guidance and growth by Wnts and RPM-1. (A) (1) As the AVG axon (red line) migrates posteriorly into the tail, it encounters Wnt/LIN-44 (red shading). (2) We hypothesize that Wnt/LIN-44 activity alters gene expression in AVG (indicated by the change in line color from red to green) such that AVG expresses a new receptor sensitive to dorsal guidance cues (green shading), (3) allowing the axon to grow in the dorsal direction towards the dorsorectal ganglion. (B) Our findings suggest that Wnt/LIN-44 regulates AVG axon guidance through Fz/LIN-17 signaling (solid arrows), resulting in POP-1-mediated transcriptional changes that promote dorsal axon guidance. RPM-1 prevents AVG overgrowth by preventing crosstalk from the Wnt/EGL-20 through an unknown receptor. RPM-1 has an additional role, independent of Wnt signaling, in restricting growth. Damage and stress can disinhibit these two mechanisms (dotted lines) to allow new axon growth and regeneration in older animals.
Whereas Wnt/LIN-44 promotes AVG axon guidance in the tail, we also uncovered a role for RPM-1 in preventing Wnt crosstalk and AVG axon overgrowth (Fig. 8B). The function of RPM-1 in axon growth is best understood with respect to axon damage, where it regulates the damaged-induced DLK/p38 MAPK pathway to allow axon regeneration (Hammarlund et al., 2009; Xiong et al., 2010; Yan et al., 2009). RPM-1 also prevents inappropriate anterior overgrowth and turning of the touch neuron axon PLM, as well as the anterior and posterior overgrowth of the D-type GABAergic motor neurons (Maro et al., 2009; Opperman and Grill, 2014; Schaefer et al., 2000). This regulation is through the combined actions on the DLK-1/p38 MAPK pathway and the GLO-4/GLO-1 small GTPase pathway, whereas we find that RPM-1 regulates AVG axon growth solely through its actions on the DLK-1/p38 MAPK pathway (Baker et al., 2014; Grill et al., 2007; Li et al., 2008; Opperman and Grill, 2014).
One common element of how RPM-1 regulates PLM, AVG and GABAergic motor neuron axon growth is Wnt signaling. A small percentage of lin-44 and bar-1 mutants show the same PLM axon termination defects observed in rpm-1 mutants (Tulgren et al., 2014). By contrast, lin-44 and rpm-1 mutants show the opposite phenotypes for AVG axon termination, and BAR-1 has no obvious role. Moreover, PLM overgrowth in rpm-1 mutants depends on the regulation of BAR-1 nuclear import by EMR-1 and ANC-1, and we did not observe a similar phenotype for AVG growth in these mutants (Fig. S1B). Similarly, lin-44 mutants show a BAR-1-dependent overgrowth of GABAergic motor neuron axons, suggesting that LIN-44 acts as a repellant for these neurons (Maro et al., 2009). By contrast, we find that lin-44 mutants have an undergrowth of the AVG axon, indicating that LIN-44 acts as an attractant or stimulator of AVG posterior growth. Given the opposite effects of LIN-44 on growth, as well as the disparate reliance on BAR-1 and downstream RPM-1 effectors, our results demonstrate a novel mechanistic role for Wnt signaling and RPM-1 in regulating AVG axon growth and guidance.
In addition to increased Wnt signaling, the overgrowth defects for PLM in rpm-1 mutants are also due to augmented signaling of the guidance cues Netrin/UNC-6 and Slit/SLT-1 through their respective receptors UNC-5 and Robo/SAX-3 (Li et al., 2008). RPM-1 might regulate how PLM responds to guidance cues by inhibiting the amount of UNC-5 and SAX-3 receptors that accumulate on the neuron surface. We have not observed a change in LIN-17::GFP subcellular localization in rpm-1 mutants (data not shown). Instead, we favor a model in which RPM-1 prevents one or more Wnt receptors from inappropriate activation by the EGL-20 Wnt. A more detailed mechanistic understanding of how RPM-1 prevents Wnt crosstalk will require the identification of the receptor(s) activated by EGL-20 in rpm-1 mutants, as none of the Fz-type receptor mutations or known non-canonical receptor mutations appears to block the overgrowth defects observed in rpm-1 mutants to the same extent as double Wnt lin-44 egl-20 mutations. However, it is important to note that we cannot examine AVG axon growth in a strain in which both maternal and zygotic contributions are completely null for all the Fz-type receptors; a small amount residual receptor activity could be sufficient.
Even in the absence of both LIN-44 and EGL-20 Wnts, the AVG axon still continues to overgrow in rpm-1 mutants, albeit in the wrong (anterior) direction. Although this overgrowth requires the DLK-1/p38 MAPK pathway, it does not require the lone TCF POP-1, suggesting that it is entirely independent of β-catenin signaling. It also does not require CEBP-1, suggesting that it works through a different mechanism than that used following axonal injury (Yan et al., 2009). This additional mechanism of axon growth regulation might be related to the ability of RPM-1 and its homologs to inhibit axon growth by promoting growth cone collapse (Borgen et al., 2017; Hendricks and Jesuthasan, 2009). This mechanism is through the inhibition of DLK-1/p38 MAPK signaling, which otherwise stabilizes microtubules. In the absence of RPM-1, axonal microtubules become overly stabilized, which not only provides additional structural support for axon growth but also facilitates transport of materials to the growing axon tip.
In summary, we suggest that RPM-1 acts as a dual-purpose switch. First, it prevents axon guidance cue crosstalk as the expression of different cues wax and wane over developmental time and as a growth cone enters new regions of the body containing new cues. Second, it provides a brake on growth cone extension by destabilizing microtubules. Axonal injury can deactivate this switch, allowing for the needed crosstalk required to get axons re-growing and regenerating in adult tissue, long after the original guidance cue landscape of the embryo has vanished.
MATERIALS AND METHODS
Growth conditions and strains
Standard methods were used to culture C. elegans (Brenner, 1974). Animals were grown at 20°C on standard NGM plates seeded with OP50 Escherichia coli. Strains were backcrossed to our laboratory N2 two to four times. Genes and mutations used in this study are listed in Table S1. Transgenic strains include: OH4887 otIs182[Pinx-18::GFP], OH3701 otIs173[Prgef-1::DsRed2], KP1476 nuIs25[Pglr-1::GLR-1::GFP], OR856 rpm-1(od14), OR1977 nuIs25; odEx[Pglr-1::RFP::RAB-5(GDP)], nuIs142[Pglr-1::bar-1::gfp], OR1129 nuIs25; odEx[Pglr-1::RFP::RAB-5(GTP)], OR1456 rpm-1(od14); nuIs25; odEx[Pglr-1::RPM-1(+)], OR1125 nuIs25; odEx[Pglr-1::RFP::DLK-1, rol-6dm] and kyIs51[Podr-2b::GFP].
Transgenes and germline transformation
Plin- 44::signal sequence::flag::gfp::lin-44 genomic coding::lin-44 3'UTR and Pegl-20:: lin-44(+) were kind gifts from Kang Shen (Stanford University, CA, USA). These constructs were injected into OR3385 lin-44(n1792); kyIs51 at 3 ng/μl and 15 ng/μl, respectively.
The Pglr-1::lin-17(pOR826) and Pglr-1::lin-17::gfp(pOR827) transgenic plasmids were generated by PCR-amplifying cDNA from an OpenBiosystems clone followed by Gateway recombination to introduce the products into the Gateway destination vectors (pOR298 and pOR478, respectively) containing the glr-1 promoter. These constructs were injected into N2 at 10 ng/μl.
To generate a dominant-negative version of POP-1, Pglr-1::pop-1(D/N), lacking the first 44 amino acids (the activation domain), cDNA was PCR amplified from the ORFeome RNAi collection using upstream primer AAAATGGCAGAATTAGACGGTGCCGGTCGAAATCCATC followed by the Gateway recombination explained above. This construct was injected into kyIs51 strain at 10 ng/μl.
Transgenes that synthesize a sense and an antisense mRNA under the control of the glr-1 promoter to knock down the lin-17, wrm-1 and sys-1 genes were generated (Esposito et al., 2007). The lin-17 cDNA was amplified by PCR using upstream primer ATGATGCATTCTTTGGGCATCATTCTACTAT and downstream primer GACGACCTTACTGGGTCTCCATGAATTCTG. The sys-1a and wrm-1a cDNA were synthesized by GENEWIZ (South Plainfield, New Jersey). The cDNAs were subcloned into pCR8 (Invitrogen) by the TOPO cloning reaction, resulting in sense and antisense donor vectors. The donor cDNAs were then moved into pOR298, a destination vector containing the glr-1 promoter and unc-54 3′UTR, resulting in sense transgenes and antisense transgenes. Plasmids for the sense and antisense transgenes were equally mixed (5 or 10 ng/µl each) and injected into the kyIs51 strain.
Podr-2b::lin-17 (pOR830) and Podr-2b::pop-1 (pOR831) transgenic plasmids were generated by the Gibson Assembly cloning technology with fusion of the three overlapping linearized DNA fragments: PCR-amplified odr-2b promoter, PCR-amplified pop-1 or lin-17 cDNA, and a linearized ∼2.8 kb DNA as a backbone, which is a digestion product of pOR829 with SacI and EcoRV. For the Podr-2b::lin-17 (pOR830) plasmid, the odr-2b promoter was PCR amplified with upstream primer TCACGGTACCCTTAATTAACGAGCTCTGTGAGTTAATTGAACTGATACTAG and downstream primer AGAATGCATCATTTTTTCTGTCTGAAATATAAATGTTCC, and lin-17 cDNA was amplified with upstream primer ATATTTCAGACAGAAAAAATGATGCATTCTTTGGG and downstream primer TAGACCCATATGCCACGCGTCCGATTTAGACGACCTTACTGGGTC. For the Podr-2b::pop-1 (pOR831) plasmid, the odr-2b promoter was PCR amplified with upstream primer TCACGGTACCCTTAATTAACGAGCTCTGTGAGTTAATTGAACTGATACTAG and downstream primer GTCGGCCATCATTTTTTCTGTCTGAAATATAAATGTTCC, and pop-1 cDNA was amplified with upstream primer ATATTTCAGACAGAAAAAATGATGGCCGACGAAGAG and downstream primer TAGACCCATATGCCACGCGTCCGATTTAAATAGTACACATCGATTCCTGCATAAG.
DAPI staining
To stain animals bearing kyIs51[Podr-2b::GFP] with DAPI, animals were washed twice with PBS and then fixed with 4% paraformaldehyde (PFA) at 4°C for 2 h. PFA-fixed animals were washed twice with cold PBS and then re-suspended in 100% methanol (pre-chilled at −20°C) for 5 min at −20°C. After twice washing with PBS, animals were stained with DAPI solution (200 ng/ml in PBS; Sigma-Aldrich) for 30 min.
Fluorescence microscopy
GFP fluorescence and DAPI staining were visualized in nematodes by mounting on 2% agarose pads with 20 mM levamisole. Fluorescent images were observed using an Axioplan II (Carl Zeiss) with 10× (N.A.=0.3), 40× (N.A.=1.3) and 63× (N.A.=1.4) PlanApo objectives. Imaging was performed with an ORCA charge-coupled device camera (Hamamatsu) using iVision software (Biovision Technologies). Maximum intensity projections of z-series stacks were obtained and out-of-focus light was removed with a constrained iterative deconvolution algorithm (iVision).
Acknowledgements
We thank Roshni Shah for technical assistance. We also thank the C. elegans Genetics Center, Kang Shen, Bill Wadsworth, Yishi Jin, Massimo Hilliard, Oliver Hobert, Peter Juo and Lars Drier for reagents and strains.
Footnotes
Author contributions
Conceptualization: E.C.P., C.R.; Methodology: E.C.P.; Validation: E.C.P.; Formal analysis: E.C.P., C.R.; Investigation: E.C.P.; Resources: E.C.P.; Writing - original draft: C.R.; Writing - review & editing: E.C.P., C.R.; Visualization: E.C.P., C.R.; Supervision: C.R.; Project administration: C.R.; Funding acquisition: C.R.
Funding
This work was supported by funding from the National Institutes of Health (R01NS42023, R01GM101972 and R21NS102780 to C.R.). Deposited in PMC for release after 12 months.
References
Competing interests
The authors declare no competing or financial interests.