Identification of Domains of Netrin UNC-6 that Mediate Attractive and Repulsive Guidance and Responses from Cells and Growth Cones

Mapping domains required for UNC-6 function

Derivatives of unc-6 were constructed to encode UNC-6 with specific deletions or substitutions (Fig.1). For each construct, the microinjected DNA was integrated into a chromosome, and three independently derived transgenic lines were established. The proteins were designed to include an N-terminal HA epitope tag, and the expression of each protein was analyzed by Western blot. The derivatives were tested in the unc-6 null background for their ability to rescueunc-6 null guidance defects. For each migration assay, 100 animals from each line were scored.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Schematic representation of the netrin UNC-6 protein and UNC-6 derivatives used in this study. UNC-6 comprises an N-terminal domain VI (residues 1–268) similar to the N-terminal domain VI of laminin subunits, three cysteine-rich repeats (residues 269–437) similar to those of domain V of laminin subunits, and a C-terminal domain named C (residues 438–591) that is not found in laminins but is phylogenetically conserved among other extracellular proteins. An epitope tag comprising three tandem copies of the HA epitope was engineered into a site immediately after the predicted signal peptide (SP).

The UNC-6ΔVI construct deletes domain VI. UNC-6/netrin domain VI was named because of its similarity with the N-terminal domain VI of laminin subunits (Ishii et al., 1992). Laminin is a large heterotrimeric extracellular matrix protein formed from three polypeptide chains (α, β, and γ), and in the prototypical laminin each subunit chain contains a domain VI (Timpl and Brown, 1996). Overall, UNC-6 domain VI is most similar to the domain VI of the laminin γ subunit; however, a short motif (SADFGKTW) within the UNC-6 domain VI is a hallmark of laminin β subunits. Because the phylogenetic conservation of this motif in netrins and laminin β subunits suggests a critical function, we specifically deleted these amino acids to create a UNC-6ΔVI8β derivative.

The UNC-6/netrin V domain was named because of the similarity to the tandem array of cysteine-rich repeats of the V-1, V-2, and V-3 motifs of the laminin subunits (Ishii et al., 1992). Overall the UNC-6 V domains are most similar to the repeats of the laminin γ subunit, with the exception of V-2, which is most similar to the V-2 domain of the laminin β subunit. The V-2 repeat is the most highly conserved repeat among the netrins. Within vertebrate netrins, the other laminin-like domains more closely resemble either the laminin γ or β subunit (Koch et al., 2000). To study the requirement of these domains for UNC-6 guidance, we constructed a series of derivatives that specifically remove each domain, UNC-6ΔV-1, UNC-6ΔV-2, and UNC-6ΔV-3. In addition, we constructed a derivative, UNC-6 V-1–3-3, to test whether the V-3 repeat can functionally substitute for the V-2 repeat.

The third domain of UNC-6 is designated domain C (Ishii et al., 1992). The domain was found at the C terminus and showed similarity to the C termini of the complement C345 protein family. This module was later identified in frizzled related proteins, type I PCOLCEs, and TIMPs (Leyns et al., 1997; Banyai and Patthy, 1999). Interestingly, functionally divergent members of the UNC-6/netrin family have been described that differ mainly in the similarity between C domains (Nakashiba et al., 2000, 2002; Yin et al., 2002). To examine the role of UNC-6 C we constructed the derivative UNC-6ΔC. In addition, we investigated the effect that the expression of the C module alone would have by constructing a UNC-6ΔVI-V derivative using a glr-1 promoter to drive expression in ventral nerve cord interneurons.

Domains VI, V-2, and V-3 primarily function in dorsal cell and axon guidance

To determine whether any of the UNC-6 derivatives are capable of directing dorsal axon migrations, we examined the dorsal migrations of the DA and DB motor neuron axons and of the SDQR axon when each UNC-6 derivative was expressed in the unc-6 null background (Fig. 2). To score the DA and DB migrations, we genetically crossed in anunc-129:: gfp transgene that allows these neurons to be visualized by epifluorescence microscopy (Colavita et al., 1998). In the wild-type embryo, axons of the DA and DB neurons migrate circumferentially from the ventral midline cell bodies to form the dorsal nerve cord (Fig. 3). Inunc-6 null larvae, the axons wander and longitudinally migrate, resulting in 98% of the axons failing to reach the dorsal midline (Table 1). We find that all the axons reach the dorsal cord in unc-6ΔV-1animals, and only 9% failed to reach the dorsal midline inunc-6ΔC animals (Table 1) (Lim et al., 1999). Similar results were obtained when the SDQR axon migrations were scored. The SDQR cell body is located along the lateral body wall, and the axon migrates dorsally to the dorsal sublateral nerve (Fig. 2). The axon was visualized by pan-neural expression of green fluorescent protein (GFP) (Kim et al., 1999). Although 91% of the axons fail to reach the dorsal sublateral nerve in unc-6 null animals, 3% fail inunc-6ΔV-1 animals and 56% fail in unc-6ΔCanimals (Table 1). In contrast, expression of UNC-6ΔVI, UNC-6ΔV-2, UNC-6ΔV-3, or UNC-6V-1–3-3 fails to rescue dorsal axon migrations (Tables 1, 2). These results indicate that the UNC-6 VI, V-2, and V-3 domains are required primarily for the dorsal axon guidance activities of UNC-6.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Summary of cell and axon positions.A, Cell bodies and axons of representative sensory and motor neurons. The migration of the axons of DA and DB motor neurons and of SDQR were assayed to measure the ability of unc-6mutations and transgenes to guide dorsal axon migrations. Only DA motor neurons are represented for simplicity. PVCL and AVM axons were assayed to measure ventral axon migrations. B, Migrating mesodermal cells, the distal tip cells, and the anchor cell were assayed to measure dorsal and ventral cell migrations, respectively. Details of the phenotypes of these cells in unc-5, unc-6, and unc-40 mutants were reported byHedgecock et al. (1990). Anterior is to the left; dorsal is at top.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Dorsal axon migrations of DA and DB motor neurons.A, In wild-type animals, the DA and DB cell bodies (arrowheads) are positioned along the ventral nerve cord, and each has an axon that migrates longitudinally along the ventral nerve cord (vc) and an axon that migrates circumferentially to the dorsal nerve cord (dc). B, In some mutants, the dorsal circumferential migration defects are relatively mild. In thisunc-40 mutant, a single axon (arrows) has abnormally migrated at the dorsal sublateral position and fails to reach the dorsal nerve cord. C, In unc-6mutants, the axons rarely reach the dorsal nerve cord. In thisunc-6 mutant, most axons turn and migrate at the ventral sublateral and lateral positions (arrows). Anterior is to the left; dorsal is at top. Scale bars, 25 μm.

To measure the ability of the UNC-6 derivatives to guide dorsal cell migrations, we scored the dorsal migration of the two hermaphrodite distal tip cells (Hedgecock et al., 1990). These cells migrate across muscle and epidermis basement membranes to form the hermaphrodite gonad. At one point, both the anterior and posterior distal tip cells migrate dorsally from their positions along the ventral muscle quadrant (Fig. 2). In unc-6 null animals, 38% of the anterior and 62% of the posterior cells fail to migrate dorsally (Table 1). With the exception of UNC-6ΔV-1 and UNC-6ΔC, all of the other UNC-6 derivatives fail to rescue these dorsal cell migrations (Tables 1, 2). All cells migrated normally in unc-6ΔV-1 animals, and only 3% of the anterior and 13% of the posterior cells failed to migrate in unc-6ΔC animals (Table 1) (Lim et al., 1999). These results indicate that the UNC-6 VI, V-2, and V-3 domains are required primarily for the dorsal cell guidance activities of UNC-6.

Domain VI functions primarily in ventral cell and axon guidance, and domain V-3 functions in ventral cell guidance but not ventral axon guidance

To investigate whether any of the UNC-6 derivatives are capable of directing ventral axon migrations, we examined the ventral axon migrations of the PVCL and AVM axons in the unc-6null background (Fig. 2). The PVC neurons are in the lateral lumbar ganglia near the tail of the animal; PVCL is in the left lumbar ganglion and PVCR is in the right (Fig.4). These neurons send axons ventrally along the lumbar commissures in a process that requires the UNC-6-expressing commmissure pioneer PVQ and the UNC-6-expressing midline PVT neuron (Durbin, 1987; Ren et al., 1999). After entering the ventral cord, the PVC axons eventually go the entire length of the cord. To examine the PVC neuron, we used aglr-1:: GFP transgene that labels 11 interneurons in the ventral nerve cord, the PVC axons being the only ones from the lumbar ganglion (Maricq et al., 1995). We also examined the AVM neuron, which is born in the first larval stage and sends an axon ventrally from the lateral cell body during subsequent developmental stages (Sulston and Horvitz, 1977). The AVM neuron is positioned alone along the right lateral body wall, and its axon migration was scored using pan-neural GFP expression.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Ventral axon migrations of PVC in the tail region.A, In wild-type animals, the PVC neurons extend an axon ventrally along the lumbar commissure to the ventral nerve cord (vc). The left PVC neuron (PVCL) is shown. B, In unc-6 mutants, the PVC axons often migrate at the lateral position instead of directly entering the ventral nerve cord. In this animal, the axon from the left PVC neuron (PVCL) has migrated laterally, whereas the axon from the right PVC neuron (PVCR) has correctly migrated to the ventral nerve cord. Anterior is to the left; dorsal is at top. Scale bars, 10 μm.

In unc-6 null mutants, 39% of the PVCL and 27% of the AVM axons fail to migrate ventrally (Table 1). Expression of UNC-6ΔVI or UNC-6ΔVI8β does not rescue this phenotype, whereas expression of UNC-6ΔV-1, UNC-6ΔV-2, UNC-6ΔV-3, UNC-6V-1–3-3, or UNC-6ΔC at least partly rescues these ventral axon migration defects (Tables 1,3). These results are in contrast to the dorsal axon migration results, where only UNC-6ΔV-1 and UNC-6ΔC could rescue the unc-6 null phenotype. Thus, domain VI is required for both dorsal and ventral axon migrations, whereas V-2 and V-3 are required primarily for dorsal axon migrations but not for ventral axon migrations. Domains V-1 and C are not essential for either dorsal or ventral axon migrations.

To determine whether expression of a UNC-6 derivative affects ventral cell migration, we examined the ability of each transgene to rescue the egg-laying defects of unc-6 null mutants. In wild-type hermaphrodites, the anchor cell moves to the center of the ventral uterus and induces the ventral epidermal cells to become vulval precursor cells (Fig. 2) (Sternberg and Horvitz, 1986). In theunc-6 null hermaphrodites, the anchor cell is often displaced laterally or dorsally, and the vulva does not develop properly, resulting in animals that are unable to lay eggs (Hedgecock et al., 1990). This occurs in 55% of the null mutants (Table 1). The expression of UNC-6ΔVI, UNC-6ΔVI8β, or UNC-6ΔV-3 does not rescue the egg-laying defect; however, the expression of UNC-6ΔV-1 does rescue and expression of UNC-6ΔV-2, UNC-6 V-1–3-3, and UNC-6ΔC partly rescues (34, 37, and 30% defective, respectively). These results show that the ventral cell migration function requires VI and V-3 and that V-2 and C influence the migration but are not essential. Domain V-1 is not required.

Domain C is not required for guidance activity but inhibits an axon-branching activity

UNC-6ΔC has guidance activity (Lim et al., 1999) (Table 1). In addition, we reported previously that the expression of UNC-6ΔC causes ventral nerve cord motor neurons to extend additional circumferential branches (Lim et al., 1999). Further studies have indicated that the UNC-6 C domain silences ectopic branching, which is dependent on the CaMKII and DAG signaling pathways (Wang and Wadsworth, 2002). We report here that the expression in the ventral nerve cord interneurons of UNC-6ΔVI-V, which comprises only UNC-6 C, does not rescue any of the motor neuron guidance defects in unc-6null animals, nor are any other abnormalities observed (data not shown). This is consistent with the proposal that the UNC-6 C domain acts in cis to the N-terminal domains to silence the axon (Wang and Wadsworth, 2002).

unc-6 transgenes encoding domain deletions cause phenotypes similar to selective loss-of-function alleles

A caveat concerning the interpretation of the transgene results stems from the difficulty of proving that the phenotypes are not the result of inappropriate expression. Although Western blot analysis indicates that a protein from each transgene is produced, whether the precise wild-type levels and distribution of each protein are achieved, is difficult to ascertain. Because mutations within the coding sequence of the endogenous gene are more likely to cause phenotypes that result from altered protein structure, we used our assays to examine mutants with selective loss-of-function alleles. If individual domains primarily determine a specific guidance activity, then mutations disrupting the coding sequence of that domain might cause some phenotypes that are similar to those caused by the expression of a transgene encoding the protein without that domain. Thus, complementary phenotypes would suggest that a transgene phenotype is the result of altered protein structure rather than inappropriate expression.

We examined four different alleles of unc-6. The first wasunc-6(rh204), an allele found previously to contain a deletion that causes the production of a protein lacking the V-2 domain (Wadsworth et al., 1996). Earlier studies indicated that this allele specifically disrupts dorsal cell and axon migrations (Hedgecock et al., 1990). Here we assayed the effects of this mutation by directly scoring axons using neuronal GFP expression. This allows a direct comparison to our transgene expression results. We find that both theunc-6(204) mutation and the expression of the V-2 deletion transgene cause similar phenotypes; that is, dorsal migrations are primarily disrupted (Table 1).

The unc-6(e78) allele was noted previously as having partial gene function (Hedgecock et al., 1990). To quantitate dorsal axon migrations, we scored the DA and DB axons and found that 90% of the DA and DB circumferential axons failed to migrate dorsally. However, for dorsal cell migrations, 0% of the anterior and only 15% of the posterior distal tip cells failed to migrate dorsally (Table 1). We find that the e78 mutation introduces a tyrosine residue in place of the conserved cysteine at position 410 within domain V-3. These results suggest that a misfolding of the V-3 domain inunc-6(e78) mutants does not affect dorsal cell migration functions but does affect dorsal axon migration functions. Both theunc-6(e78) mutation and the expression of the V-3 deletion transgene cause similar phenotypes: disruption of dorsal axon migrations with less affect on ventral axon migrations. Furthermore, both show differences in the ability to guide cell and axon migrations, suggesting that the V-3 domain is responsible for different interactions between UNC-6 and either migrating cells or growth cones.

The unc-6(rh46) and unc-6(rh46 ev436) are alleles that disrupt the coding sequence of the VI domain. The rh46allele introduces an alanine to proline substitution near the middle of VI, whereas rh46 ev436, is a pseudorevertant ofrh46 that introduces another substitution, a threonine for an alanine, near the end of VI (Wadsworth et al., 1996). We assayed the effects of these alleles to compare the effects of these mutations with those caused by the expression of VI transgenes. Similar to the phenotypes caused by the expression of the unc-6ΔVI8βtransgene, the unc-6(rh46)and unc-6(ev436)mutations also cause a temperature-sensitive loss of guidance activity (Tables 2, 3). The results from previous assays suggested that theev436 pseudorevertant restores growth cone but not cell guidance activity (Hedgecock et al., 1990). The results presented here show that the restored activity is heat sensitive. The degree to which different migrations are rescued depends on the temperature at which the assays are performed.

Together these results indicate that point mutations within the coding sequences of the VI, V-2, and V-3 domains each produce unique phenotypes that are similar to those produced by the expression of the transgenes that delete or alter the same domain. This strongly supports the idea that the phenotypes produced by the transgenes are caused by expression of altered protein products.

UNC-6 domains that mediate different guidance functions

Previous studies indicated that the guidance activities of UNC-6 are mediated through distinct domains (Hedgecock et al., 1990; Wadsworth et al., 1996). Several alleles of unc-6that cause selective loss of dorsal guidance activity encode for a protein specifically lacking the V-2 domain. Furthermore, two selective loss-of-function alleles that disrupt tissue- and direction-specific guidance were found to be point mutations within domain VI. However, these studies did not show whether the V-2 domain was the only domain required for dorsal guidance, whether other domains are associated with other UNC-6 guidance functions, or how indicative the VI domain point mutations are of domain VI function. The systematic approach taken in this study has addressed these issues and provided evidence that both direction-specific (dorsal versus ventral) and tissue-specific (ectodermal growth cones versus mesodermal cells) guidance requires a combination of different UNC-6 domains.

Dorsal and ventral migrations are mediated by the UNC-6 V-2 and V-3 domains. Deletion of either the V-2 or V-3 domain disrupts dorsal migrations to the same degree as the unc-6(ev400)loss-of-function allele (Table 1). The dorsal guidance activities appear to require the combination of the two domains because substituting a V-3 for the V-2 domain does not restore dorsal guidance functions. The severity of the dorsal guidance defects is similar amongunc-5 null mutants and the mutants expressing UNC-6ΔV-2, UNC-6ΔV-3, or UNC-6V1–3-3 (Table 1), suggesting that disrupting the V-2 or V-3 domains interferes with the ability of UNC-6 to interact with the UNC-5 receptor. Among the UNC-6 derivatives, ΔV-3 most closely resembles the cell and axon ventral defects of theunc-40 mutant (Table 1), suggesting that V-3 is involved in the interaction with UNC-40. Loss of either flanking domain, as in ΔV-2, ΔC, and V-1–3-3 animals, causes a less severe loss of ventral guidance function, suggesting that the flanking domains may influence the interaction between domain V-3 and UNC-40.

The V-3 domain appears to mediate different interactions between UNC-6 and either migrating cells or growth cones. The unc-6(e78)allele, which disrupts domain V-3, specifically interferes with dorsal axon migrations but not dorsal cell migrations or any ventral migrations (Table 1). Furthermore, deletion of V-3 more strongly disrupts ventral cell migrations than ventral axon migrations. Interestingly, the phenotypes of UNC-5 and UNC-40 mutants do suggest that these receptors mediate responses differently for cells and growth cones. This raises the possibility that unknown cell type-specific UNC-6 receptors associate with the V-3 domain. Alternatively, domain V-3 may influence a difference in the manner in which UNC-6 interacts on cells and growth cones with the UNC-5 and UNC-40 receptors.

Domain VI is essential for all UNC-6 guidance functions. Interestingly, the short but conserved laminin β subunit motif within domain VI is critical, suggesting that it may be a site of molecular interaction. Two mutations within domain VI selectively disrupt the guidance activities. The temperature-sensitive unc-6(rh46 ev436)allele selectively restores growth cone migrations, whereas theunc-6(ev400 ev437) allele, which replaces the domain VI residues glutamine–serine–histidine at positions 78–80 with two serines, disrupts ventral migrations but not dorsal migrations of both axons and cells (Hedgecock et al., 1990; Wadsworth et al., 1996). The observation that these two alleles selectively disrupt the different guidance activities of UNC-6 provides further evidence that domain VI is required for all guidance activities. It is possible that this domain is involved directly in the association of UNC-6 with UNC-5 and UNC-40. Alternatively, this domain could indirectly affect UNC-5 and UNC-40 signaling through interactions with other receptors and the extracellular matrix. By influencing properties of cell adhesion, the domain could alter cytoskeletal dynamics that are tied to both UNC-5 and UNC-40 responses.

Model for parallel UNC-6 signaling

Guidance involves dynamic changes to the cytoskeleton and cell adhesion properties of the migrating cell or growth cone. Our results suggest that the combined activities of different domains mediate UNC-6 guidance (Fig. 5). We propose that domain VI plays a key role in coordinating the cytoskeletal machinery and allowing a response to UNC-6 and that domains V-2 and V-3 elicit signals that specifically direct attraction or repulsion. Finally, the C domain provides a signal that prevents axon branching in response to the signals elicited by the N-terminal domains (Lim et al., 1999; Wang and Wadsworth, 2002). In this model, the modular nature of the UNC-6 molecule is associated with the parallel signals, and the combination of UNC-6 modules regulates multiple aspects of the guidance machinery.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

A summary of the proposed function for each of the UNC-6 domains. The combination of domains elicits parallel signals that together mediate the different cytoskeletal changes necessary for guidance.

Our results form the basis for further biochemical studies to investigate molecular interactions between UNC-6/netrin and its receptors. Two UNC-6 receptors are known: UNC-5 mediates repulsion, i.e., migration away from UNC-6 sources, whereas UNC-40 is involved in both attraction and repulsion (Hedgecock et al., 1990). For the repulsive response, it is thought that UNC-5 and UNC-40 can act either independently or together through the formation of UNC-5 or UNC-5 and UNC-40 receptor complexes (Merz et al., 2001). Studies in vitro support such receptor interactions (Hong et al., 1999). We speculate that these receptor complexes form primarily through an association with the V-2 and V-3 domains. Although our results also support the possibility that domain VI is involved, we favor the possibility that this domain interacts with other receptors and the extracellular matrix to mediate cytoskeletal and adhesion changes necessary for the UNC-5 and UNC-40 responses. Finally, our results indicate that domain V-3 is responsible for a difference between the ability of cells and growth cones to respond to UNC-6. The molecular explanation for this difference also requires further research.