C. elegans is a canonical model for axon guidance given its conservation of axon guidance genes and stereotyped neuronal identities and morphologies (Chisholm et al., 2016). While the majority of C. elegans neurons are born embryonically, a subset is born during larval development which enables direct observations of axon guidance by time-lapse imaging experiments. Of these postembryonic neurons, five (AVM, HSN, PVM, PDE, and PQR) initiate pioneering axons toward the ventral region of the animal where they join the VNC (Figure 1A). (Another neuron, PVD, is born post-embryonically but its axon follows along the axon of PDE which initiates first.) We sought to leverage this common guidance trajectory and accessible visualization to provide new insights into mechanisms of axon guidance. First, we developed a paradigm that permits complete visualization of PDE axon’s ventral growth in wild-type animals. This strategy used a Plin-32::mCherry-PH integrated transgene to label the PDE cell membrane and imaged PDE axon outgrowth every 3 min for 4 hr (Figure 1B, Video 1). The mature axonal morphology of PDE is unbranched along the dorsoventral axis before bifurcating at the VNC (Figure 1B). The continuous, unbranched morphology of the mature PDE axon suggests continuous ventral growth after axon initiation. Surprisingly, however, we observed that the PDE axon consistently halts its ventral growth upon reaching the region near the ventral sublateral nerve cord (vSLNC), a small axon tract that runs along the anteroposterior body axis of the animal. During its pause, the axon establishes transient collateral branch from which additional dorsal and ventral branches emerge. Eventually, one ventral branch stabilizes at the VNC, bifurcates, and initiates fasciculated growth along VNC axons. Following VNC stabilization, the transient collateral branch at the vSLNC is pruned, resulting in the unbranched PDE axonal morphology observed in adulthood. The stereotyped PDE axonal pause and consistent transient branching at vSLNC indicate that this is an intermediate guidance target.

The vSLNC sits atop the ventral body wall muscles of the worm and adjacent to the epidermis, thereby demarcating an anatomical region that separates two distinct environments through which the axon grows (Figure 1C, D). In adult animals, the vSLNC is ensheathed by the epidermis, but the ensheathment process begins during larval stages and is not yet completed by PDE axonogenesis (Chalfie and Sulston, 1981). Prior to reaching the vSLNC, the PDE axon grows along the epidermal basal lamina, but additional ventral growth past the sublateral region requires that axons wedge themselves between fibrous adhesive junctions between the body wall muscles and the epidermis (Figure 1C, D; Cox and Hardin, 2004). Given these distinct growth environments, the sublateral region (either the vSLNC itself or the body wall muscles) serves as an intermediate target during ventral axon navigation. This stark contrast in growth environments correlates with alternate forms of morphological axon growth. Prior to reaching the vSLNC, the growth cone of PDE is wide and reminiscent of a lamellipodium. However, ventral axon branching from the transient collateral branch are consistent with filopodia-based motility.

Given that the body wall muscles tile the anteroposterior axis of the entire animal, other ventrally directed axons also must undergo a migration through these distinct regions. We next hypothesized that the other ventrally growing axons would similarly utilize the ventral sublateral region as an intermediate target. To test this, we imaged axon emergence in AVM, HSN, PVM, and PQR neurons. While these neurons position their somas more ventrally, we observe similar axonal trajectories as the PDE axon: axons initially emerge ventrally as lamellipodia, then adhere upon reaching the vSLNC. Multiple ventrally directed filopodia form and a ventral branch eventually stabilizes at the axons’ ultimate ventral target, the VNC (Figure 1E). To gauge the location of axon pausing relative to the vSLNC, we imaged HSN and PVM during axon emergence with a reporter construct expressing GFP under the control of the mec-7 promoter region (Figure 1F). This reporter construct labels mechanosensory neurons, including the PLM neuron whose axon comprises the vSLNC. Taken together, these data reveal that ventral axon guidance in C. elegans is a stepwise process facilitated by an intermediate target near the vSLNC. Notably, dorsally migrating axons in C. elegans also stall their growth cones at the borders of longitudinal nerve tracks and muscle borders, forming anvil-shaped structures (Knobel et al., 1999). These results argue that the stepwise migration and pausing of growth cone at specific tissue structures is a general phenomenon in C. elegans, prompting us to understand the molecular bases of this cellular behavior.

Circumferential axon migration in C. elegans is primarily driven by signaling through UNC-6/Netrin and its receptors (Hedgecock et al., 1990; Wadsworth et al., 1996). Dorsal guidance requires UNC-6 to signal through both its receptors, UNC-5 and UNC-40/DCC. Ventral guidance does not require UNC-5 signaling but is dependent on UNC-40 (Hedgecock et al., 1990). Most evidence supporting this notion is based on analyses of post-developmental axonal morphologies in mutants instead of visualizing the developmental process. We used time-lapse imaging of PDE axon outgrowth in null mutants of unc-6 (ev400) and unc-40 (e1430) to understand whether UNC-6/UNC-40 signaling is required for one or all of the events: (1) initial axon outgrowth direction, (2) stabilization at the intermediate target, (3) directional branching from the sublateral region, or (4) ventral growth from intermediate target to VNC. Importantly, the guidance of the vSLNC axons does not rely on UNC-6 or UNC-40, so the putative intermediate target is intact in these mutants (Li et al., 2008). The resulting movies demonstrate that each step of ventral growth is disrupted in unc-6 and unc-40 mutants (Figure 2A, Video 1 and Video 2, Figure 2—figure supplement 1). First, in most wild-type animals, the axon emerges from the ventral side of PDE, while axons emerge from non-ventral directions of PDE in unc-6 and unc-40 mutants (Figure 2—figure supplement 1A, B). To compare axon emergence between genotypes, we calculated the sine of angle of axon emergence which indicates where along the dorsal–ventral axis the axon is emerging. A sine value of 1 denotes dorsal emergence at 90°, 0 indicates emergence along the anterior–posterior axis, and −1 denotes 90° ventral emergence. These comparisons further implicate UNC-6 and UNC-40 in determining the direction of PDE axon emergence.

Second, over 80% of wild-type axons form a stable collateral branch immediately upon reaching the vSLNC. In contrast, nearly half of the unc-6 and unc-40 mutant axons form collateral branches at vSLNC but subsequently retract (Figure 2A, B). Typically, these axons ‘probe’ the vSLNC again after retraction and establish contacts with vSLNC again during the imaging window. However, retraction upon initial sublateral contact is rarely observed in wild-type animals, indicating that UNC-6 signaling stabilizes the growth cone at the sublateral intermediate target.

To further understand how UNC-6 and UNC-40 regulate axon development, we inserted a GFP tag at the C-terminus of the endogenous unc-40 locus. The resulting animals show UNC-40::GFP fluorescence in the nerve ring and major nerve cords of the worm (VNC and dorsal nerve cord), consistent with a published endogenous UNC-40::mNeonGreen strain (Figure 2—figure supplement 1C–E; Jayadev et al., 2022). We next analyzed movies of UNC-40::GFP during PDE axon outgrowth. These movies demonstrate that the PDE growth cone contains low levels of UNC-40::GFP prior to reaching the sublateral region. However almost immediately upon contacting the sublateral, UNC-40::GFP fills the growth cone along the transient collateral branch (Figure 2C, D). However, unc-6(ev400) mutants fail to increase growth cone UNC-40::GFP upon reaching the sublateral region and instead demonstrate seemingly random fluctuations in growth cone UNC-40::GFP intensity (Figure 2C, D). These results suggest that UNC-6 regulates UNC-40 levels in the growth cone.

Given similar morphological polarization PDE and HSN growth cones, we sought to generalize these findings in HSN which develops more slowly than PDE and is therefore more amenable to imaging axon engagement with the ventral sublateral intermediate target. Indeed, endogenous UNC-40::GFP is enriched at the ventral surface of HSN cell body where its axon emerges. However, this enrichment is absent in unc-6 mutants, consistent with previous studies using overexpressed UNC-40::GFP constructs (Figure 2E; Adler et al., 2006). To quantify UNC-40 polarity, we measured the total GFP intensity along the periphery of the dorsal soma and sublateral contact site of HSN. The overall UNC-40::GFP polarity was calculated by dividing the difference between the two subcellular locations by the total GFP intensity along the cell periphery (Figure 2F, Figure 2—figure supplement 1F, G). The resulting polarity index ranges from –1 to 1 with negative numbers indicating a ventral UNC-40::GFP bias and positive numbers indicating a dorsal bias. A polarity value of 0 denotes an equal dorsoventral distribution of UNC-40::GFP. We also calculated the polarity of UNC-40 in AVM, PVM, and PQR in wild-type and unc-6(ev400) animals. Wild-type animals display negative polarity calculations, but in unc-6 mutant animals, the polarity index is close to 0 indicating that UNC-40 is evenly distributed between the dorsal and ventral neuronal domains (Figure 2E, G; Figure 2—figure supplement 1H–M). Together, these data indicate that upon vSLNC contact, UNC-6 signaling enriches UNC-40 at the contact site and facilitates adhesion with the surrounding environment.

Once the axon establishes a lateral branch at the vSLNC, it generates dorsally and ventrally oriented filopodia (Figure 2A). These filopodia are dynamic and undergo constant extension and retraction. Using time-lapse movies to capture the dynamics, we quantified the directionality, length, growth speed and stability of these branches (Figure 2H–K). Wild-type animals generate slightly more ventral branches than dorsal ones (55.6% ventrally directed branches vs. 44.4% dorsal). However, unc-6 and unc-40 mutants have a strong reversal of this bias where 61.4% and 71.9% of branching, respectively, is directed dorsally (Figure 2H). These data suggest that UNC-6/UNC-40 signaling promotes ventral branching. To further test this, we analyzed the growth trajectories of individual ventral branches. This analysis demonstrates that ventral growth stalls in unc-6 and unc-40 mutants, manifesting as a decreased average growth speed (Figure 2I, J). While unc-6 and unc-40 mutant PDE axonal filopodia reach the VNC, very few stabilize within the imaging window. Instead, most branches retract, indicating that UNC-6 and UNC-40 are also required for stabilizing filopodia once it reaches the VNC (Figure 2K).

Taken together, these data indicate that UNC-6 and UNC-40 are required for several steps of axon guidance including the direction of axonal initiation, axonal adhesion at the intermediate target, directional growth toward the VNC and axonal stabilization at the VNC. While the developmental process is dynamic with numerous growth and retraction events, UNC-6 and UNC-40’s impact on the stepwise events compound to ultimately disrupt the adult morphology of ventrally directed axons. In unc-6(ev400) mutants, 67.9% ± 5.1% of PDE axons and 66.9% ± 4.3% of HSN axons fail to reach the VNC by adulthood, compared with only 0% of PDE axons and 0.91% ± 0.9% of HSN axons in wild-type animals (Tables 1 and 2; Figure 2—figure supplement 1N).

In addition to UNC-6, SLT-1/Slit also directs ventral trajectories of C. elegans axons through its receptor SAX-3/Robo (Hao et al., 2001). SLT-1 is produced in dorsal regions of the worm and functions as a repellant. Given this, we next sought to leverage our dynamic imaging platform to analyze the precise growth cone behaviors that SLT-1 is responsible for. The resulting movies of slt-1(eh15) animals reveal that these PDE growth cones emerge from more dorsal regions of the cell body (Figure 3A–C, Video 3). Often, these misrouted axons corrected their trajectory by reorienting growth to the ventral region of the worm, likely in response to UNC-6. Further, we did not observe any defects in axon stabilization at the vSLNC or directional branching in slt-1 mutants (Figure 3D, E). Animals lacking SLT-1 were less likely than wild-type to stabilize at the VNC within the 4 hr imaging window (Figure 3F). We posit that this stabilization defect is transient because adult neuronal morphologies in these mutants present with only mild phenotypes in comparison to unc-6 mutants (Tables 1 and 2).

Given that loss of slt-1 enhances axon guidance defects in unc-6 mutants, we next sought to characterize PDE axon guidance in this double mutant background. We found that the angle of axon emergence in the unc-6(ev400); slt-1(eh15) animals are completely random with no ventral preference (Figure 3C, Video 4). The double mutant also showed strong defects in vsLNC stabilization, axonal branching and stabilization at VNC (Figure 3D–F). This is consistent with the strong axon trajectory phenotypes found in the adult animals of this genotype (Tables 1 and 2).

Our data thus far demonstrate that PDE axon outgrowth is an excellent in vivo model of Netrin-mediated guidance where precise axonal behaviors can be probed across distinct change-of-function genotypes. Therefore, we next asked whether regulators of UNC-6/Netrin signaling might only affect growth or adhesion. For this analysis, we chose the E3 ubiquitin ligase MADD-2/Trim9/Trim67 because of its conserved role in Netrin-dependent guidance and morphological polarization of the HSN emerging axon (Akin and Zipursky, 2016; Hao et al., 2010; Plooster et al., 2017). We captured movies of PDE axon guidance in madd-2 null (ok2226) animals (Figure 4A, Video 5). These analyses revealed that madd-2 mutants showed normal growth cone orientation at emergence (Figure 4—figure supplement 1A, B). However, the majority of the madd-2 mutant axons retracted upon vSLNC contact similar to unc-6 mutants, suggesting that MADD-2 contributes to UNC-6-mediated adhesion to the intermediate target (Figure 4B).

Given UNC-40’s requirement for adhesion and ventral polarization, we next asked whether MADD-2 is required for the polarized subcellular localization of UNC-40. To test this, we observed endogenous UNC-40::GFP polarization within HSN, AVM, PVM, and PQR in madd-2 mutants. These images reveal that UNC-40 loses its ventral polarity in madd-2 mutants (Figure 4C, D; Figure 4—figure supplement 1E–N). We confirmed these results in PDE by analyzing growth cone UNC-40 levels during axon guidance. Consistent with the results in HSN, each of AVM, PVM, and PQR, the PDE growth cone showed reduced UNC-40 polarization at the vSLNC (Figure 4—figure supplement 1C–H). Together, these data reveal a requirement for MADD-2 to polarize UNC-40 at endogenous levels. Further, we analyzed madd-2; unc-6 double mutants for UNC-40::GFP polarity in HSN. These animals demonstrate enhanced mispolarization of UNC-40 (Figure 4C, D), suggesting that UNC-6 and MADD-2 might promote the polarity of the UNC-40 receptor in parallel pathways.

Existing literature suggests that Trim9 ubiquitinates DCC and that Netrin antagonizes this activity (Plooster et al., 2017). To test whether UNC-40 ubiquitination is required for its polarization at the vSLNC, we generated unc-40(4KR) animals with lysine to arginine mutations at the four intracellular lysine residues adjacent to the UNC-40 transmembrane domain to hamper potential ubiquitination. These mutations were made in the UNC-40::GFP background to allow visualization of endogenous levels of mutant UNC-40. These four amino acid substitutions are sufficient to mispolarize UNC-40 in HSN, comparable to polarity measurements in madd-2(ok2226) animals (Figure 4C, E, F). Critically, UNC-40 mispolarization is not enhanced in unc-40(4KR); madd-2(ok2226) animals, suggesting that UNC-40 ubiquitination functions in the same pathway as MADD-2. Given that MADD-2 directly regulates UNC-40 localization through ubiquitination, these data suggest that MADD-2 ensures that UNC-40 is polarized and enriched at the growth cone at the sublateral region to promote adhesion.

In the next stage of growth, the movies of madd-2 mutant demonstrate that PDE axons primarily initiate ventrally directed branches similar to wild-type animals, indicating that the directional growth of PDE axon is independent of MADD-2 (Figure 4G). This is notable given that vertebrate mutants for the MADD-2 ortholog Trim9 display aberrant branching (Winkle et al., 2014). However, these ventral branches also display decreased stabilization at the VNC similar to unc-6 mutants (Figure 4H). Together, these experiments reveal two important insights: axonal adhesion and stabilization are dependent on MADD-2 but directional growth is not.

UNC-6/Netrin has been proposed to function via either chemotaxis or haptotaxis (Comer et al., 2019; Varadarajan et al., 2017; Wu et al., 2019). Chemotactic growth occurs in response to a gradient where axons sense the Netrin concentration in their immediate environment to determine their direction of travel toward or away from the Netrin source. In contrast, haptotactic growth does not rely on a long-distance gradient and instead uses local, minimally diffusive Netrin to anchor growth along a laminar surface. In the vertebrate spinal cord, commissural axons utilize both mechanisms of Netrin-mediated guidance to properly establish ventral tracts that avoid the developing motor column (Varadarajan et al., 2017; Wu et al., 2019).

Therefore, we sought to distinguish between these two UNC-6 growth mechanisms during stepwise PDE axon guidance by expressing a non-diffusible form of UNC-6 in the intermediate target region in the unc-6 mutants. Specifically, we fused the unc-6 coding sequence to the transmembrane domain sequence of nlg-1/neuroligin so that the molecule is tethered to the plasma membrane (Teichmann and Shen, 2011). We generated two membrane-tethered UNC-6 expression constructs: one under the control of mec-7 promoter elements for expression along the vSLNC and one under the control of the unc-4 promoter for expression at the VNC (Figure 5A). Importantly, UNC-4 expressing cholinergic motor neurons produce endogenous UNC-6 in larval stages and MEC-7 expressing mechanosensory neurons do not (Basu et al., 2021). However, membrane-tethered UNC-6 would be present on the axons in VNC but unable to diffuse toward the PDE cell body.

Using these constructs, we generated three different transgenic strains in the unc-6 mutant background, a strain expressing Pmec-7::unc-6, a strain expressing Punc-4::unc-6 and a strain carrying both. We validated that these rescue constructs were expressed in the proper cells by including a SNAP ligand-binding domain after the transmembrane domain. SNAP labeling confirms that these constructs are expressed along the vSLNC and VNC, consistent with the promoter fusions (Figure 5—figure supplement 1A). We also did not detect SNAP signals outside of the nerve tracks, indicating that the membrane-tethered UNC-6 is not cleaved and shed at significant quantity. Importantly, these transgenes do not disrupt wild-type axon guidance (Figure 5—figure supplement 1D, Tables 1 and 2). With these transgenes, we aimed to determine whether the distinct axon behaviors at the vSLNC intermediate target and at VNC are mediated by UNC-6’s chemotactic or haptotactic activity.

These transgenes are unable to restore ventral directionality of axon emergence (Figure 5—figure supplement 1B, C). Interestingly, stabilization at vSLNC is restored when non-diffusible UNC-6 is present at vSLNC (Pmec-7::unc-6) but not when expressed only at the VNC (Punc-4::unc-6), consistent with stabilization being contact-mediated (Figure 5B, C, Videos 6–8). Given the rapid and spatially specific accumulation of growth cone UNC-40 observed in the movies of PDE axon outgrowth, we next tested whether membrane-tethered UNC-6 could rescue UNC-40::GFP polarization in HSN. Indeed, non-diffusible UNC-6 along the vSLNC restored UNC-40::GFP polarity in HSN (Figure 5D, E). In contrast, non-diffusible UNC-6 expressed along the VNC could not restore ventral UNC-40 polarization (Figure 5D, E, Figure 5—figure supplement 1E, F), presumably due to the inability of HSN to contact this UNC-6 population. These data indicate that axonal stabilization at the vSLNC intermediate target requires local UNC-6, and membrane-tethered UNC-6 is sufficient for this function. Therefore, haptotaxis is sufficient to account for adhesion and UNC-40 polarization.

Next, we analyzed the directional filopodia growth in animals expressing membrane-tethered UNC-6. Interestingly, none of the transgenes were sufficient to restore biased ventral branching (Figure 5F). This is consistent with a model where the biased direction of axon growth after adhesion at the vSLNC intermediate target requires diffusible UNC-6. Lastly, the frequency of axon stabilization at the VNC is partially rescued only in animals with membrane-tethered UNC-6 at both the vSLNC and VNC (Figure 5G). This result suggests that VNC stabilization is contact-mediated, but its efficiency requires UNC-6 at the intermediate target along earlier portions of the axon’s path of growth. Taken together, these data suggest that contact-mediated haptotaxis drives axon stabilization at the vSLNC intermediate target and VNC, and that polarized UNC-40 toward the ventral cell body and growth cone is required for adhesion.

Given the ability of non-diffusible UNC-6 constructs to restore contact-mediated axon guidance behaviors, we considered their impact on the ultimate morphology of HSN and PDE neurons. By assaying adult neuronal morphology, we found that only the double transgene of membrane-tethered UNC-6 was sufficient to partially restore adult axonal morphology with 68.6% ± 5.0% of PDE axons and 47.4% ± 4.6% of HSN axons reaching the VNC by adulthood (Tables 1 and 2). This is in comparison to 100% of PDE axons and 99.1% ± 0.9% of HSN axons doing so in wild-type animals and 32.1% ± 5.1% of PDE axons and 33.1% ± 4.3% of HSN axons in unc-6 mutants. These data indicate that membrane-tethered UNC-6 can partially rescue axon guidance when present at both vSLNC and VNC. This implies that diffusible UNC-6, possibly in the form of a gradient, is also required for axon guidance of PDE and HSN.

To further understand UNC-6’s distribution and activity, we used an endogenous C-terminus mNeonGreen (mNG) tag of UNC-6 (UNC-6::mNG) (Naegeli et al., 2017) to visualize its localization in vivo. Our genetic evidence suggests that UNC-6 functions not only locally at vSLNC and VNC but also likely over the entire distance between VNC and vSLNC. We closely examined the localization of UNC-6::mNG in this region. We first imaged the emerging PDE growth cone interacting with UNC-6::mNG. In these images, bright mNG puncta surround the growth cone and its ventral path (Figure 6A). Notably, however, we are unable to detect an obvious gradient of UNC-6::mNG between the VNC and dorsal region of the animal.

We next imaged the zone between VNC and vSLNC where numerous faint UNC-6 puncta are present. Interestingly, bright UNC-6 puncta are also localized along the axonal filopodia of HSN. A similar localization along the axonal filopodia is also observed for PDE and AVM (Figure 6B). Such signal is completely absent in strains without the endogenous UNC-6::mNG tag, validating that the signal represents endogenous UNC-6 (Figure 6—figure supplement 1). We measured the intensity of UNC-6 through this region along the ventrodorsal axis. The average intensity over this region is the highest at the VNC. Moving dorsally, the average intensity steadily decreases but increases again near the vSLNC (Figure 6C). These data are consistent with a model where UNC-6 is enriched in both the vSLNC and VNC regions separated by two shallow yet opposing gradients. Axon growth past the vSLNC intermediate target is likely due to permissive growth in response to high local concentrations of UNC-6, as recently proposed by Hooper and Lundquist, 2024.

Given the striking enrichment of UNC-6 along axonal filopodia, we hypothesized that UNC-40 is required for UNC-6 clustering along the filopodia. Animals carrying unc-40 null mutations display a dorsoventral gradient similar to wild-type animals (Figure 6D, E). However, axonal filopodia that extend toward the VNC are not associated with UNC-6 puncta (Figure 6D). Taken together, these data suggest that UNC-6 is present in a gradient and decorates the periphery of the growth cone during growth.

The filopodia between vSLNC and VNC during axon outgrowth are dynamic, with constant growth and retraction. This implies that the UNC-6 puncta associated with the filopodia membrane are also likely not static. To directly test this, we simultaneously imaged five HSN growth cones and UNC-6::mNG puncta during its outgrowth in the region between the VSLNC and VNC. For this analysis, we only imaged growth cones that had extended into the space between the vSLNC and the VNC. Indeed, we observed dynamic assembly and disassembly of mNG puncta near the growth cone and filopodia on ~1 min time scales (Figure 7A, Video 9). In movies of unc-40 null animals, these foci assemble but quickly disperse and do not associate with the growth cone (Figure 7A, Video 10). To quantify this, we calculated the intensity of each UNC-6 cluster adjacent to the cell membrane. These measurements reveal decreased UNC-6 labeling around 5 HSN growth cones in unc-40 mutants compared to wild-type animals (Figure 7B). Axonal growth was rarely observed in unc-40 mutants within the 120 s imaging window (Figure 7A), consistent with the reduced growth speed (Figure 2I). Together, these data suggest that there is a diffusive population of UNC-6 in the axonal path and that the extending growth cone remodels the UNC-6 landscape by assembling transient, highly concentrated foci around the cell to stimulate growth via UNC-40.

Netrin has conserved functions in nervous system development across phylogeny. Initially, Netrin was proposed to function as a chemotactic cue that stimulates directional axon growth (Kennedy et al., 1994; Serafini et al., 1994). Indeed, secreted Netrin from the vertebrate floor plate functions to attract axons toward the ventral midline, avoiding the developing motor column (Wu et al., 2019). However, additional investigations in the vertebrate spinal cord as well as in the Drosophila visual system and midline have identified haptotactic roles for Netrin in axon guidance (Akin and Zipursky, 2016; Brankatschk and Dickson, 2006; Varadarajan et al., 2017). Our data clearly demonstrate roles for both chemotaxis and haptotaxis in Netrin-mediated attraction. Specifically, haptotaxis (mediated by membrane-tethered UNC-6) is sufficient to explain growth cone stabilization by providing adhesive force to maintain axonal contact with an intermediate target. Conversely, chemotaxis likely provides the basis of directional growth toward the VNC.

Importantly, the endogenous distribution of UNC-6/Netrin supports dual migration mechanisms of haptotaxis and chemotaxis. In C. elegans, the ventral sublateral intermediate target is separated from the VNC by a gradient of UNC-6 with the highest concentration along the VNC. Yet, we also observe that this gradient is modified by an accumulation of UNC-6 in the sublateral region, consistent with the location of the axon’s intermediate target. However, this poses an additional conundrum where axons must extend past the vSLNC into a region with lower relative UNC-6 concentrations. One possibility is that UNC-6 acts as a permissive cue to stimulate growth past the vSLNC. Such permissive functions for UNC-6 have also been reported in repulsive axon guidance in C. elegans (Hooper and Lundquist, 2024).

The precise mechanisms which produce elevated levels of UNC-6 in the sublateral region remains unclear. As a laminin-related protein, Netrin is likely interacting with extracellular matrix (ECM) proteins, and the enrichment of an ECM-binding partner in the sublateral could restrain UNC-6 in such a way as to promote haptotaxis. One candidate class of molecules is heparin sulfate proteoglycans which have high affinity to UNC-6 (Serafini et al., 1994). In fact, structural analysis of the Netrin/DCC interaction has identified a putative binding site dependent on extracellular sulfate (Finci et al., 2014).

A recent study in C. elegans supports these conclusions using a transmembrane-tethered allele of unc-6 (Hooper and Lundquist, 2024). This allele displays a partial loss-of-function phenotype that can be restored through UNC-6 produced by an ectopic source. These data suggest that the diffusibility of UNC-6 is more critical than the location of its expression in C. elegans. Therefore, the gradient that we observe is likely established using local interactions between UNC-6 and the surrounding ECM. Additional studies combining cell biological and structural information may provide further insight on UNC-6 interactions with the ECM and the dichotomous signaling mechanisms that promote haptotactic and chemotactic responses.

Haptotactic growth cone adhesion serves a central function in growth cone migrations (Comer et al., 2019). Investigations of the molecular signaling mechanisms that promote Netrin-driven haptotaxis have implicated an ‘adhesion-clutch model’ where DCC generates force through connections to the actin cytoskeleton (Qiu et al., 2024). Our data identify another molecular regulatory feature of haptotaxis: guidance receptor localization. In ventrally migrating axons, endogenous levels of UNC-40 receptor are polarized toward the ventral side of the cell upon contact with the sublateral intermediate target. The E3 ubiquitin ligase MADD-2/Trim9/Trim67 cooperates with UNC-6 in this polarization. These results are consistent with cell invasion mechanisms in the non-neuronal anchor cell in C. elegans (Wang et al., 2014). In our neuronal system, genetic paradigms that disrupt UNC-40 polarization also exhibit decreased adherence to the sublateral region (Figures 4 and 5). Therefore, we posit that polarization of UNC-40 likely amplifies the cell’s capacity to adhere to its intermediate target.

Given our data suggesting that MADD-2 ubiquitinates UNC-40, we propose that ubiquitination of UNC-40 by MADD-2 triggers UNC-40 endocytosis. Similar data in vertebrate systems identify that Netrin antagonizes Trim9-induced ubiquitination of DCC (Plooster et al., 2017). Together, these data suggest that in these neurons MADD-2 likely functions in areas where UNC-6 concentration is low (i.e. the dorsal region of the cell/growth cone). In this model, UNC-40 would be preferentially endocytosed in dorsal regions of the cell, thereby contributing to the overall polarity of UNC-40. In contrast, UNC-6 likely stabilizes UNC-40 expression on the cell membrane and prevents ubiquitination at the ventral side of the cell, further establishing UNC-40 polarity. Consistently, experiments in vertebrate systems have long suggested that exposure to Netrin causes DCC to be added to the cell membrane through exocytosis (Matsumoto and Nagashima, 2010; Plooster et al., 2017).

Notably, Drosophila photoreceptors depend on Trim9 for axon stabilization but not elongation in Netrin guidance, consistent with our movies of madd-2 mutant axon guidance (Akin and Zipursky, 2016). Additionally, studies of vertebrate neurons demonstrate that Trim9 ubiquitinates DCC when the concentration of Netrin is low to regulate DCC exocytic balance (Plooster et al., 2017). Given the requirement of four juxtamembrane lysine residues in UNC-40 polarization, our data suggest that this regulatory function of MADD-2/Trim9 is conserved and crucial for UNC-40 cell polarity and growth cone adhesion. While the data here do not directly examine cytoskeletal regulation in relation to haptotaxis, Trim9 and Trim 67 are known to regulate members of the Ena/VASP family through ubiquitination to control filopodia density in response to Netrin-1 (Boyer et al., 2020; McCormick et al., 2024; Menon et al., 2015). Additionally, the lone Ena/VASP family member in C. elegans (UNC-34) is required for ventral axon guidance downstream of UNC-40 (Gitai et al., 2003; Norris et al., 2009). Systematic investigations of the interplay between ubiquitination, receptor localization, and cytoskeletal dynamics are required to fully elaborate haptotactic signaling mechanisms in growth cones.

Collectively, our data provide spatiotemporal access to Netrin-mediated attraction. During growth, the axon alternates between haptotactic and chemotactic migration, and we delineate the precise axonal behaviors controlled by each migration mechanism. Overall, this study predicates future investigations of the precise molecular cascades that govern discrete chemotactic and haptotactic responses to Netrin within the same growth cone.