Cell and growth cone migrations along the dorsoventral axis of Caenorhabditis elegans are mediated by the UNC-5 and UNC-40 receptor subtypes for the secreted UNC-6 guidance cue. To characterize UNC-6 receptor function in vivo, we have examined genetic interactions between unc-5 and unc-40 in the migrations of the hermaphrodite distal tip cells. We report that cell migration defects as severe as those associated with a null mutation in unc-6 are produced only by null mutations in both unc-5 and unc-40, indicating that either receptor retains some partial function in the absence of the other. We show that hypomorphic unc-5 alleles exhibit two distinct types of interallelic genetic interactions. In an unc-40 wild-type genetic background, some pairs of hypomorphic unc-5 alleles exhibit a partial allelic complementation. In an unc-40 null background, however, we observed that unc-5 hypomorphs exhibit dominant negative effects. We propose that the UNC-5 and UNC-40 netrin receptors can function to mediate chemorepulsion in DTC migrations either independently or together, and the observed genetic interactions suggest that this flexibility in modes of signaling results from the formation of a variety of oligomeric receptor complexes.
SEVERAL genes, including unc-6, unc-5, and unc-40, are known to interact in guiding circumferential cell and growth cone migrations in Caenorhabditis elegans (Hedgecocket al. 1990; McIntireet al. 1992). unc-6 encodes a member of the secreted, laminin-related protein family called netrins (Ishiiet al. 1992; Kennedyet al. 1994; Serafiniet al. 1994). unc-6 is expressed ventrally in C. elegans, and mutations cause defects in both dorsally and ventrally directed cell and growth cone migrations (Hedgecocket al. 1990; Wadsworthet al. 1996). Similar functions have been observed for insect and vertebrate netrins (Colamarino and Tessier-Lavigne 1995; Harriset al. 1996; Mitchellet al. 1996; Serafiniet al. 1996). Although the mechanisms of function of the UNC-6/netrins are poorly understood, migrating cells and growth cones are thought to transduce relative differences in extracellular UNC-6/netrin concentration into local, intracellular changes in the actin cytoskeleton.
unc-5 and unc-40 encode transmembrane receptors of the immunoglobulin (Ig) superfamily (Leung-Hagesteijnet al. 1992; Chanet al. 1996). Expression of the C. elegans UNC-5 protein is, in most cases, sufficient to cause repulsion of migrating cells or growth cones away from ventral concentrations of UNC-6 (Hamelinet al. 1993, Wadsworthet al. 1996; Suet al. 2000). Vertebrate homologues of UNC-5 include the murine rostrocerebellar malformation (rcm) gene product (Ackermanet al. 1997), now renamed UNC5H3 (Przyborskiet al. 1998), and two rat homologues, UNC5H1 and UNC5H2 (Leonardoet al. 1997). unc5h3 mutants have defects in cell migrations in the developing cerebellum (Ackermanet al. 1997; Przyborskiet al. 1998), and all three homologues have been shown to bind directly to netrins (Leonardoet al. 1997).
C. elegans unc-40 is required primarily for ventrally oriented migrations, but also contributes to dorsally oriented and longitudinal migrations (Hedgecock et al. 1987, 1990). UNC-40 is related to the product of the mammalian deleted-in-colorectal-cancer (Dcc) gene (Chanet al. 1996), which is involved in axon guidance in the mouse spinal cord (Fazeliet al. 1997), and to the product of the frazzled gene of Drosophila, also involved in axon guidance (Kolodziejet al. 1996). DCC binds directly to netrins (Keino-Masuet al. 1996) and mediates responses to Netrin-1 of Xenopus retinal ganglion cells in culture (de la Torreet al. 1997). Thus, the UNC-6/netrin ligands together with the UNC-5 and UNC-40/DCC receptors compose a highly conserved system for the guidance of migrating cells and neuronal growth cones.
The functional relationships between the UNC-5 and UNC-40/DCC receptors are not entirely clear. Presumed null unc-40 mutations in C. elegans disrupt the same dorsally oriented cell and growth cone migrations as do unc-5 mutations, but with a lower penetrance (Hedgecocket al. 1990). These weak effects of unc-40 mutations suggest that UNC-40/DCC is not strictly required for UNC-5 function. However, ectopic expression studies in both C. elegans (Colavita and Culotti 1998) and Xenopus (Honget al. 1999) indicate that, in these unusual situations, all UNC-5-mediated repulsion requires UNC-40/DCC. We are using the hermaphrodite distal tip cells (DTCs) of C. elegans as a model in vivo system for the study of the mechanism, guidance, and regulation of cell migrations. We have sought to clarify the roles of and interactions between unc-40 and unc-5 in the migrations of the DTCs, which are repelled by UNC-6 in the ventral-to-dorsal phase of their migration. We propose a model in which the UNC-5 and UNC-40 receptor subtypes are capable, to a limited extent, of mediating repulsion from UNC-6 independently of one another in the DTCs. However, the two netrin receptors function best in combination and the types of genetic interactions that we observe between hypomorphic unc-5 alleles suggest that multimerization, possibly between UNC-5 receptors, is a key component of netrin receptor signal transduction.
MATERIALS AND METHODS
Culture conditions: Culturing, handling, and genetic manipulations were as previously described (Brenner 1974). The following genes and alleles were used:
Linkage group (LG) I: unc-40(e1430), unc-40(ev457), unc-40 (ev643) (D. C. Merz, unpublished results).
LG IV: unc-5(e53), unc-5(e152), unc-5(ev432), unc-5(ev435), unc-5 (ev512), unc-5(ev585), unc-5(ev634), unc-5(ev642), unc-5 (ev644) (this study), gon-1(ev635) (D. C. Merz, unpublished results), dpy-20(e1282), unc-22(e66).
LGX: unc-6(ev400), unc-6(rh202).
Some strains not derived in our lab were obtained from the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health (NIH) National Center for Research Resources (NCRR). The isolation and phenotypic characterization of the unc-5 alleles e53, e152, e553, ev432, and ev435 have been previously described (Brenner 1974; Hedgecocket al. 1990; Leung-Hagesteijnet al. 1992). ev512, ev585, ev634, ev642, and ev644 were isolated by EMS mutagenesis in several genetic screens (D. C. Merz, A. Colavita and J. G. Culotti, unpublished data). The allele e53 is phenotypically identical to a group of severe unc-5 alleles (Hedgecocket al. 1990). Genetic and molecular analyses suggest that these alleles represent the complete elimination of UNC-5 function. For example, unc-5(e53) results from a nonsense mutation predicted to terminate the protein (W283STOP) in the region encoding the extracellular domain (see results below). Using largescale noncomplementation screens (>50,000 haploid genomes), we have been unable to generate unc-5 alleles more severe than existing strong alleles such as unc-5(e53).
unc-40(e1430) (Hedgecocket al. 1990) has been sequenced and is predicted to be a molecular null (Chanet al. 1996). e1430 has the most 5′ nonsense mutation of characterized unc-40 alleles. Several alleles with more 3′ truncations, including unc-40(ev457), have slightly more severe phenotypes (Hedgecocket al. 1990), possibly due to the production of truncated, trans-interfering proteins (Chanet al. 1996). The genetic interactions described here for unc-40(e1430) are identical with unc-40(ev457). On the basis of sequencing and phenotypic analysis, the unc-6(ev400) allele is considered a genetic and molecular null allele (Hedgecocket al. 1990; Wadsworthet al. 1996).
Scoring of DTC and Unc defects: Mutations in unc-5, unc-6, and unc-40 disrupt specifically the ventral-to-dorsal second phase of DTC migrations (Figure 1). The misshapen gonad arms thus produced are readily visible and quantifiable with a dissecting microscope. Most strains could be scored in this manner at the L4 or young adult stages of development. Strains that are egg-laying defective (Egl), especially those including unc-6 or unc-40 mutations, were scored entirely at higher magnification under differential interference contrast (DIC) optics. Gonad arms were scored as defective if the turn from the centrifugal first to the centripetal third migration phase occurred on the ventral side, indicating that the ventral-to-dorsal second phase did not occur. Some differences with previously published frequencies of DTC defects were found (Hedgecocket al 1990). Scoring of DTC migrations under DIC facilitates distinctions between qualitatively different classes of migration errors. For example, there is a distinct type of DTC migration defect often observed in unc-40 mutants in which the DTCs descend ventrally again immediately upon reaching the dorsal muscle band. Under low power microscopy, this defect is difficult to distinguish from a failure to migrate to the dorsal muscle band. The Unc-ness of worms was assayed using the standards of Hedgecock et al. (1990).
Standard errors for the proportions of defective DTC migrations in a population were calculated using the observed frequency and the actual sample size, assuming a binomial distribution, as previously described (Hedgecocket al. 1990).
Comparisons were done using a standard test (one-tailed) for comparing two proportions (Miller and Freund 1965). All P values represent the probability that the penetrance of DTC migration defects is lower in one strain than in another. A P value of 0.05 is considered significant. Where measures of statistical significance are reported in the text, the P value of the anterior or posterior DTC that is closer to 0.05 is given.
Transgenic strains: Transgenic lines were generated using standard germline transformation techniques (Mello and Fire 1995). pZH85 contains 2.0 kb of genomic sequence from immediately upstream of unc-5 exon 2 (Leung-Hagesteijnet al. 1992) fused to a green fluorescent protein (GFP)-tagged unc-40 cDNA (pZH22; Chanet al. 1996), to express functional UNC-40 (Chanet al. 1996) in cells that normally express unc-5. This 2.0-kb unc-5 promoter fragment (unc-5B2), like the 4.6-kb promoter previously described (Suet al. 2000), can drive expression of reporter constructs in the DTCs and several classes of sensory neurons in the head. However, the 2.0-kb fragment does not drive expression in ventral cord motorneurons. pZH85 (20 or 80 ng/μl) was injected along with pMH86 (20 ng/μl), which contains the wild-type dpy-20 gene. Four independent non-Dpy transgenic lines (two at 20 ng/μl pZH85 and two at 80 ng/μl pZH85; they are called evEx117a–d, respectively) were generated in a dpy-20; him-5 background and passed into other mutant backgrounds, which were also usually in the dpy-20 background in order to follow the transgenic array. GFP expression in the DTCs was confirmed for each line. For each mutant background (e.g., unc-5) into which the extrachromosomal array containing pZH85 was passed, at least three of the four transgenic lines were used. No significant differences were observed between transgenic lines, and results are presented for evEx117a and evEx117b.
SSCP and sequencing of unc-5 alleles: cDNA was prepared from RNA isolated from unc-5 alleles. For unc-5(e53) and unc-5 (e152), SSCP analysis was carried out essentially as previously described (Chanet al. 1996), and mutations were confirmed by sequencing. The complete coding sequence of unc-5(e152) was sequenced, while unc-5(e53) was sequenced 5′ to the mutation site. For unc-5(ev585), single-stranded PCR products generated from cDNA were directly (and completely) sequenced for comparison to N2. The mutation was confirmed by sequencing of the opposing strand. A complete report of sequences of unc-5 alleles will be published elsewhere (M. T. Killeen, A. Krizus, I. Scott, R. Wilk and J. G. Culotti, unpublished results).
Construction of heteroallelic strains: For scoring of DTC migration defects, heterozygous strains of unc-5 were generated by crossing wild-type males with unc-5 dpy-20 double mutant strains. Non-Dpy progeny of these crosses were scored for DTC migration defects.
unc-5 alleles (e152 and ev585, for example) were placed in trans with one another as follows. dpy-20(e1282); him-5(e1490) males were crossed to unc-5(e152) hermaphrodites and male + dpy-20(e1282)/unc-5(e152) +; him-5(e1490)/+ cross-progeny were picked. These were crossed to doubly homozygous unc-5 (ev585) dpy-20(e1282) hermaphrodites. The non-Dpy hermaphrodite progeny from this cross, which were unc-5(e152) +/unc-5 (ev585) dpy-20(e1282), were scored for defects in the dorsal migrations of the anterior and posterior DTCs, as previously described (Hedgecocket al. 1990). The heterozygous dpy-20 mutation (dpy-20(e1282)/+) did not affect the frequencies of DTC defects caused by unc-5 mutations. For example, there were no significant differences in the defects of unc-5(e152) homozygotes compared with unc-5(e152) +/unc-5(e152) dpy-20 (e1282). Due to the temperature sensitivity of dpy-20(e1282) and of unc-5(ev585), all crosses were done at 25°. From each cross, the DTC migration defects of all of the non-Dpy hermaphrodite progeny were scored.
To examine the defects caused by heterozygous unc-5 mutants in an unc-40(e1430) background, the balanced strains unc-40(e1430); unc-5 + dpy-20(e1282) unc-22(e66)/+ gon-1 (ev635) ++ were constructed. Homozygous unc-22(e66) worms exhibit a Twitching (Twi) phenotype, while homozygous gon-1(ev635) worms exhibit a visible gonad (Gon) defect caused by failure of gonad arm extension. The gon-1 phenotype is easily distinguishable from the Mig phenotype caused by unc-5. Non-Twi non-Gon worms were scored for defects in the ventral-to-dorsal phase of DTC migration. Heterozygous gon-1 or unc-22 mutants did not affect the penetrance of DTC defects in the unc-40(e1430) background.
Background on DTC migration defects: The extension of each of the two arms of the bilobed hermaphrodite gonad during larval development is led by the migration of a DTC (Hedgecocket al. 1987). Mutations in unc-5, unc-6,or unc-40 specifically disrupt the ventral-to-dorsal second phase of the DTC migration pattern (Figure 1). The centrifugal first and centripetal third phases are normal in timing and extent. In these mutants, however, failures to execute the ventral-to-dorsal phase of migration result in both the first and third longitudinal phases of DTC migration occurring on the ventral side. unc-6 null mutants are not fully penetrant for these defects (Hedgecocket al. 1990), indicating that UNC-6-independent mechanisms must also exist to guide this DTC migration phase. The UNC-5 and UNC-40 receptors are expressed by the DTCs during their migrations (Chanet al. 1996; Suet al. 2000). Consistent with the idea that these receptors transduce UNC-6-mediated directional information, additional mutations in unc-5 and unc-40 do not enhance the second migration phase defects of an unc-6 null mutation (Hedgecocket al. 1990). Thus, the penetrance of DTC defects observed in an unc-6 null background represents a complete loss of function in this pathway, and a comparison of the penetrance of this specific defect in other unc-6, unc-5, or unc-40 mutants with that of the unc-6 null provides a measure of the efficacy of the repulsive UNC-6 signaling pathway.
UNC-5 and UNC-40 function at the same time and place: The functional and biochemical relationships between the UNC-5 and UNC-40 receptors are unclear. We sought to examine in detail the role of each receptor in the second, ventral-to-dorsal, DTC migration phase. As described above, the identical DTC migration defect is caused by mutations in unc-5,-6,or-40. Previous mosaic and transgenic studies have demonstrated that UNC-5 acts cell autonomously within the DTCs to initiate the ventral-to-dorsal migration phase (Leung-Hagesteijnet al. 1992; Suet al. 2000). Reporter construct analyses suggest that unc-40 has a broad pattern of expression, including the DTCs and surrounding tissues throughout the time of the DTC migrations (Chanet al. 1996). Although UNC-40 can act cell autonomously to guide neuronal axons (Chanet al. 1996), its site and time of action in DTC migrations have not been determined. A functional GFP-tagged unc-40 transgene was placed downstream of a fragment of the unc-5 promoter to express UNC-40 in the DTCs, but not in surrounding tissues, at the time of the initiation of the ventral-to-dorsal migration phase (Figure 2; Suet al. 2000). This GFP-tagged UNC-40 transgene was previously shown to be capable of rescuing an unc-40 mutant when expressed under the control of its endogenous promoter (Chanet al. 1996). When expressed in the DTCs downstream of the unc-5B2 promoter, this unc-40 transgene (in the evEx117 lines) was able to rescue the DTC migration defects of an unc-40 null mutant (Figure 3), as the frequency of defects in evEx117; unc-40(e1430) lines was significantly lower than in unc-40(e1430) alone (P < 0.01; Figure 3A). The dumpy (Dpy) and uncoordinated (Unc) phenotypes of unc-40 were not affected. Expression of UNC-40 from this transgene was not able to rescue the DTC migration defects of the unc-6(ev400) null mutant (P > 0.05; Figure 3B), indicating that the function of the transgenic UNC-40GFP in the DTCs, like endogenous UNC-40, requires the UNC-6 guidance cue. From these results, we conclude that UNC-40, like UNC-5, acts cell autonomously within the DTCs at the time of the ventral-to-dorsal migration to mediate a response to UNC-6.
UNC-5 and UNC-40 can function independently of one another: As previously described, the relatively weak defects in ventral-to-dorsal migrations caused by unc-40 mutations suggest that UNC-5 does not absolutely require UNC-40 to signal repulsion (Hedgecocket al. 1990). In ectopic expression assays in vitro or in vivo, however, UNC-5 does absolutely require UNC-40/DCC for growth cone repulsion (Colavita and Culotti 1998; Honget al. 1999). The penetrance of DTC defects caused by a null mutation in unc-6 is significantly greater than that caused by null mutations in either unc-5 or unc-40 (P < 0.001; Figure 4A). However, the null allele unc-40(e1430) in combination with the null allele unc-5 (e53) resulted in a frequency of DTC migration defects identical to that caused by a complete loss of unc-6 function (P > 0.09; Figure 4A). This demonstrates that, in DTC migrations, UNC-5 can transduce in part the UNC-6 signal independently of UNC-40 (Figure 4B). In addition, UNC-40 can signal repulsion independently of UNC-5, indicating that UNC-5 is not absolutely required for UNC-40 to assume a repulsive role (Figure 4B). This guidance system is, however, fully functional only if both receptors are present.
We asked whether the residual function of UNC-40 in the absence of UNC-5 could be mimicked by expression of the unc-40GFP transgene in the DTCs. In addition we asked whether a high level of UNC-40 expression in the DTCs could partially rescue the DTC migration defects of an unc-5 mutation. As shown in Figure 3C, the unc-5B2::unc-40GFP transgene of evEx117 was capable of rescuing the DTC migration defects of an unc-40(e1430); unc-5(e53) double mutant strain to a level approximately equal to that of an unc-5(e53) mutant alone, but could not fully rescue the DTC migration defects, presumably due to its inability to substitute fully for UNC-5. Consistent with this interpretation, transgenic expression of UNC-40 together with wild-type endogenous UNC-40 was also unable to compensate for a null mutation in unc-5 (Figure 3D). Therefore, in addition to acting together to mediate repulsion from UNC-6, UNC-5 and UNC-40 can act independently to carry out this same function and do so in non-interchangeable ways.
Several hypomorphic unc-6 alleles, including unc-6 (rh202), selectively disrupt ventral-to-dorsal migrations (Hedgecocket al. 1990). These alleles result from deletions of the V-2 module of the UNC-6 EGF-like repeats and are predicted to disrupt UNC-6-UNC-5 interactions (Wadsworthet al. 1996), although this has not been tested biochemically. unc-6(rh202) mutants are, like unc-5 (e53), less severe than an unc-6 null allele (P < 0.001; Figure 4A). However, double mutants of unc-40(e1430); unc-6(rh202) were not significantly different from the unc-6 null allele ev400 in the penetrance of DTC defects (P > 0.25; Figure 4A), nor were they significantly different from the unc-40; unc-5 double null mutant (P > 0.19). This is consistent with the idea that unc-6(rh202) selectively eliminates UNC-5-dependent functions of UNC-6, but leaves intact a repulsion induced by UNC-6-UNC-40 interactions.
unc-5 interactions in an unc-40(+) background: Genetic interactions were examined between unc-5 alleles predicted to partially reduce UNC-5 function. Most of the unc-5 alleles examined were fully recessive and exhibited more frequent DTC defects when placed in trans to a null unc-5 allele [unc-5(e53); Figure 5]. Rare (<1%) posterior DTC defects were observed only in unc-5(e53)/+ or unc-5(e152)/+ strains. unc-5(ev634) has the same frequency of DTC migration defects as unc-5(e53) and this frequency is not increased when ev634 is placed in trans to unc-5(e53) (Figure 5). This strain is, however, less uncoordinated than unc-5(e53), suggesting some residual UNC-5 function, at least in the nervous system. Thus, we consider this and the other unc-5 alleles examined [excepting unc-5(e53)] to be hypomorphic alleles.
When placed in trans to one another, some but not all pairs of two different hypomorphic alleles exhibited a partial complementation. For example, the unc-5(ev585) allele in trans to the unc-5(e152) allele resulted in a significantly (P < 0.0001) lower frequency of DTC defects than that observed in either homozygous ev585 or homozygous e152 mutants (Figure 5). Whereas homozygous ev585 or e152 hermaphrodites both exhibited defects in ∼20% of anterior DTCs and 40% of posterior DTCs, the trans-heterozygotes (ev585/e152) had defects in only 5/226 (2%) of anterior and 54/226 (24%) of posterior DTCs.
A similar partial complementation was observed for other pairs of unc-5 alleles (Figure 5). This interallelic complementation was never complete, but was in all cases a partial amelioration of DTC migration defects. Six of eight of these unc-5 hypomorphs could be placed into one of two groups in which partial complementation was observed between but not within each group. Thus, the group ev435, ev512, ev585, and ev642 partially complemented e152 and ev644. In cases for which the site of the molecular lesion is known, the former group comprises alleles with UNC-5 ectodomain mutations, while the latter group comprises cytodomain domain mutations (M. T. Killeen, A. Krizus, I. Scott, R. Wilk, M. Nygiem and J. G. Culotti, unpublished results). For example, unc-5(ev585) results from a missense mutation in the second Ig domain (C181Y), while unc-5(e152) results from a nonsense mutation (Q507STOP) predicted to truncate the cytodomain. One of the exceptions to this grouping of alleles was unc-5(ev634), which partially complemented ev432, ev512, and ev644. Another exception was unc-5(ev432), which complemented only ev634.
unc-5 interactions in an unc-40(–) background: Although all unc-5 alleles examined were recessive and some unc-5 hypomorphic alleles exhibited interallelic complementation, different genetic properties were observed when these unc-5 alleles were examined in an unc-40 null background. For example, an unc-40(e1430); unc-5(e53)/+ strain was more severely defective in DTC migrations than unc-40(e1430) alone (P < 0.0001; Figure 6A). As unc-5(e53) is a null allele, this indicates the haploinsufficiency of unc-5 in the absence of UNC-40. The unc-5 hypomorphic alleles e152 and ev585 exhibited stronger dominant enhancement of the DTC defects of unc-40(e1430) than did the unc-5 null allele e53 (Figure 6A). Both unc-40(e1430); unc-5(e152)/+ and unc-40 (e1430); unc-5(ev585)/+ strains had significantly higher frequencies of DTC migration defects than unc-40 (e1430) alone (P < 0.0001). In addition, both had significantly more DTC defects than unc-40(e1430); e53/+ [P < 0.0001 for unc-40(e1430); unc-5(e152)/+ and P < 0.012 for unc-40(e1430); unc-5(ev585)/+]. The most severe effects were observed with unc-40(e1430); unc-5 (e152)/+, in which 43/136 (32%) of anterior and 110/136 (81%) of DTC migrations were defective (Figure 6A). These worms were also more uncoordinated than unc-40(e1430) alone. Thus, in an unc-40(–) background, unc-5 null alleles can exhibit haploinsufficiency while hypomorphic alleles also exhibit dominant negative effects.
UNC-5 can act through an UNC-40-dependent pathway: As described above, null mutant phenotypes suggest that the UNC-5 and UNC-40 receptor subtypes can function independently of one another. However, complete function requires both receptors, and recent biochemical and in vitro assays on vertebrate homologues suggest direct interactions between the two receptor subtypes (Honget al. 1999). We asked whether, for any unc-5 hypomorphs, the residual function in DTC migrations was dependent upon UNC-40. Double mutant strains comprising a null allele of unc-40 and an unc-5 hypomorph were all more severe than the unc-5 hypomorph alone (Figure 6B). Therefore, eliminating unc-40 function enhances the defects of unc-5 mutants. Some, but not all, alleles were enhanced to the severity of unc-6 null mutants. In particular, a double mutant comprising unc-40(e1430) and unc-5(e152) was not significantly different from an unc-6 null (P > 0.39) or from unc-40(e1430); unc-5(e53) (P > 0.26), either in the frequency of DTC migration errors (Figure 6B) or in the severity of uncoordination (data not shown). Similar results were obtained with other unc-40 alleles together with unc-5(e152). Two other strong and putative null unc-40 alleles, ev457 and ev643, isolated in separate genetic screens and backgrounds, also eliminated UNC-6-dependent DTC migrations as double mutants with unc-5(e152), as the frequency of DTC migration defects was not significantly different from that of unc-6 (ev400). An unc-40(ev457); unc-5(e152) strain (N = 129) exhibited 49% anterior (P = 0.46) and 88% posterior (P = 0.13) DTC defects. unc-40(ev643); unc-5(e152) (N = 122) had 47% anterior (P = 0.42) and 84% posterior (P = 0.5) DTC defects.
Most other unc-5 hypomorphic alleles did not exhibit this interaction (Figure 6B). For example, although the e152 and the temperature-sensitive ev585 unc-5 alleles are quantitatively similar in their DTC and axon guidance defects at 25° (Figure 5 and D. C. Merz, unpublished data), a double mutant of unc-40(1430) together with unc-5(ev585) exhibited a less severe phenotype than an unc-6 null (Figure 6B). Thus, in an unc-40 null background, the function retained by some hypomorphic unc-5 alleles, such as e152, is dependent on UNC-40, while the function retained by other unc-5 alleles is not fully dependent on UNC-40. These two groups of alleles, therefore, distinguish UNC-40-dependent and UNC-40-independent signaling functions of UNC-5. Expression of UNC-40 GFP in the DTCs from a multicopy array (evEx117) partially rescued the DTC defects of unc-5 (e152), but not unc-5(ev585) or unc-5(e53) (Figure 3, D–F), providing additional evidence that the cytoplasmically truncated UNC-5 receptor is dependent upon UNC-40 for its residual function.
Extracellular guidance cues such as UNC-6/netrins and semaphorins are thought to act on migrating cells and growth cones by increasing or decreasing the extension or stabilization of membrane structures such as lamellipodia and filopodia. The mechanisms by which local concentration differences in a ligand are interpreted at the leading edge of a migrating cell or growth cone are not understood. To learn how the UNC-6 directional signal is transduced by the UNC-5 and UNC-40 receptors at the cell surface, we have examined genetic interactions between unc-5 and unc-40 in the ventral-to-dorsal migration phase of the DTCs. Transgenic and mosaic experiments indicate that each of the UNC-6 receptor subtypes can act cell autonomously within the DTCs at the time of this migration phase. This allows us to propose from the genetic interactions a model for how the UNC-5 and UNC-40 receptor proteins function.
The UNC-6/netrins are capable of acting as either attractant or repellant guidance cues. Of the two known UNC-6 receptors, UNC-5 is associated only with repulsive migrations, i.e., away from sources of UNC-6/netrin. UNC-40, on the other hand, is involved in both attraction toward and repulsion away from UNC-6 in C. elegans (Hedgecocket al. 1990). This suggests that the two receptor subtypes may act together to mediate repulsion, and evidence for this interaction has been observed in in vitro studies of axon guidance functions of vertebrate homologues (Honget al. 1999). Indeed, DTC migration phenotypes indicate that a complete response to UNC-6 requires both receptors. We observe, however, that UNC-5 can partially function in the absence of UNC-40, consistent with previous in vivo observations (Hedgecocket al. 1990). In addition, UNC-40 can partially function in the absence of UNC-5 to mediate repulsion of the DTCs away from UNC-6. This indicates that there are multiple mechanisms by which the UNC-6 signal can be transduced to produce DTC repulsion (Figure 4B).
Two distinct and opposing types of genetic interactions were observed between unc-5 hypomorphic alleles. First, in the presence of wild-type UNC-40, some unc-5 alleles exhibit a partial allelic complementation. In cases where the site of the molecular lesion is known (M. T. Killeen, A. Krizus, I. Scott, R. Wilk, M. Nygiem and J. G. Culotti, unpublished results), complementing pairs usually comprise one allele with an extracellular and one allele with an intracellular mutation. Noncomplementing pairs comprise two extracellular or two intracellular mutations. For example, unc-5(e152), which encodes a protein with a predicted cytodomain truncation, complements unc-5(ev585), which causes a missense mutation in the second Ig domain of the UNC-5 ectodomain. Interallelic complementation usually signifies the functional importance of close or direct proteinprotein interactions (Razet al. 1991; Sibleyet al. 1994). We propose that complementation results from one UNC-5 receptor compensating for a defect in input through the extracellular domains and the other for a defect in output to cytoplasmic signaling pathways through the intracellular domains.
In contrast to these genetic interactions in a wild-type unc-40 background, the dominant negative interactions observed in the absence of functional UNC-40 suggest that the products of hypomorphic unc-5 alleles can, under these conditions, interfere with the functions of wild-type UNC-5. Together, these observations strongly argue for an intimate and possibly direct association between UNC-5 proteins, in addition to that between UNC-5 and UNC-40.
The formation by the UNC-6 receptors of a variety of functional oligomeric complexes may be required to allow the sensitivity of cells and growth cones to a wide range of UNC-6 concentrations. Alternatively, different concentrations of UNC-6 may activate, through the formation of different receptor complexes, distinct cellular responses.
The idea that UNC-5 and UNC-40 cannot substitute for one another despite having the same general role in the second DTC migration phase suggests that they may have distinct downstream targets. Although signaling pathways linking transmembrane guidance receptors like UNC-5 and UNC-40 to the cytoskeleton are not well understood, there are in principle several ways in which cell adhesion and cytoskeletal dynamics may be regulated in cell migration (Lauffenburger and Horwitz 1996). As motility requires the comprehensive and coordinated function of the cytoskeleton, guidance systems may have to send several parallel signals to various targets to effectively direct the cytoskeleton. It is also possible that the UNC-5 and UNC-40 receptor subtypes regulate distinct subsets of the motility apparatus. For example, one receptor subtype may be more important for filopodial extension and the other for filopodial retraction. Although we have provided evidence that both receptor subtypes act within the DTCs at the same time, we have not examined their subcellular localization during signaling. A combination of precise localization and functional studies in combination with biochemical assays will be necessary to address these issues.
In ectopic expression studies in different systems, UNC-5 appears to absolutely require UNC-40/DCC to mediate repulsion (Colavita and Culotti 1998; Honget al. 1999). As we have shown, however, this is not the case in cells that normally express UNC-5 and that are repelled by UNC-6. One reason for such differences may be that ectopic expression studies reflect an artificial situation that results in enhanced reliance on some, but not all, components of the normal signaling pathway. For example, we have proposed that cells that normally express UNC-5, which include the DTCs, express some component, possibly a coreceptor or downstream signaling partner, that allows UNC-5 to function independently of UNC-40/DCC. This component may be absent or inactive in cells that normally exhibit only attractive responses to UNC-6/netrin cues through the UNC-40/ DCC receptor subtype. In such cells, therefore, UNC-40-independent repulsive functions of UNC-5 would be absent.
Further evidence for the sensitized nature of ectopic expression studies comes from a genetic screen for suppressors of the ventral-to-dorsal axon projections caused by ectopic expression of UNC-5. This screen identified mutations at eight loci (Colavita and Culotti 1998). Most of these loci, although required for UNC-5 function in this particular set of touch sensory neurons, are not absolutely required for UNC-5 function in commissural neurons or DTCs. We have proposed (Merz and Culotti 2000) that the touch sensory neurons do not possess the ability to generate a complete response downstream of UNC-5 and that this results in a dependence upon the products of unc-40 and other genes identified as suppressors in this screen.
UNC-5/UNC-40 signaling models: Although null mutations in unc-40 do not completely eliminate wild-type UNC-5 function in DTC migrations, they do completely eliminate the function of some unc-5 hypomorphic alleles. The protein products of these particular unc-5 alleles may be defective mainly in UNC-40-independent signaling, but retain UNC-40-dependent signaling. The unc-5(e152) cytoplasmic truncation allele, for example, appears to require UNC-40 for its residual function. In addition, excess UNC-40 in the DTCs expressed from a transgene partially rescues the DTC migration defects of this unc-5 allele. This suggests that the UNC-5 cytoplasmic domain encoded by unc-5(e152) is involved in signaling functions that largely require UNC-40, whereas the deleted portion of the cytoplasmic domain is involved in signaling functions that do not require UNC-40. Conversely, other hypomorphic alleles that do not require UNC-40 presumably retain some signaling through an UNC-40-independent pathway.
A similar paradigm for interactions between receptor subtypes has been proposed for the p75NTR and trkC neurotrophin receptors. Truncated trkC receptors that lack cytodomain tyrosine kinase activity retain partial signaling functions, but this residual function is dependent upon the p75NTR coreceptor (Hapneret al. 1998). This indicates the function of distinct p75NTR-dependent and -independent signaling mechanisms for the trkC cytodomain. A different signaling model has been proposed for the two receptor subtypes for the semaphorin guidance cues. The neuropilin receptor subclass appears to increase the affinity of the semaphorin ligands for the plexin receptor subclass (Takahashiet al. 1999). However, cytoplasmic signaling occurs only through the plexin receptor cytodomain and does not require the cytodomain of the neuropilin receptor (Takahashiet al. 1999; Tamagnoneet al. 1999).
Hong et al. (1999) have reported that direct physical interactions between the cytoplasmic domains of UNC-5 and DCC are required for repulsion of growth cones from a netrin source in vitro. It was reported that a specific region of the cytodomain of UNC-5, the DB domain, is essential for both the functional and the biochemical UNC-5-DCC interactions. However, in C. elegans the unc-5(e152) mutation that retains full dependence upon UNC-40 contains a nonsense mutation (Q507STOP) predicted to truncate UNC-5 between the transmembrane region and the domain that has homology to ZO-1 (a.k.a. the ZU5 domain), thus eliminating the proposed DB domain.
It is unclear how an UNC-5 mutation that is predicted to eliminate the functional interaction between UNC-5 and UNC-40/DCC can retain full dependency on UNC-40 for repulsive guidance mechanisms. One possibility is that UNC-5 in C. elegans functions differently in this respect from vertebrate UNC-5s or that functional interactions between UNC-5 and UNC-40 in DTCs are different from that in neurons. It is also possible that in vitro and ectopic expression experiments like those carried out in Xenopus to define the DB domain of UNC-5 are artificially sensitized to perturbations in the UNC-5 protein that, in a normal situation, might not be as important to function.
Switching of UNC-40/DCC: Studies of Xenopus retinal ganglion cell growth cone responses to exogenously applied Netrin-1 in vitro have revealed that levels of cAMP within the growth cone can regulate the role of DCC (Minget al. 1997). Prior incubation of the neurons with an inhibitor of protein kinase A or a nonhydrolyzable analogue of cAMP appeared to switch DCC from a netrin receptor that mediates attraction to one that mediates repulsion. Other experiments suggest that the expression of the UNC-5 receptor may also act as a switching mechanism for UNC-40. For example, touch sensory axons in C. elegans that normally project ventrally are redirected dorsally by ectopic expression of UNC-5 in these neurons (Hamelinet al. 1993). UNC-40 is required for both the abnormal dorsalward and the normal ventralward axonal projections of the touch neurons (Hedgecocket al. 1990; Colavita and Culotti 1998). It is not known whether the cAMP and UNC-5-mediated switching mechanisms are related.
In the DTCs, UNC-40 can mediate repulsion from UNC-6 to some extent in the absence of UNC-5. It is interesting to note that UNC-40 is present in the DTCs during the longitudinal first migration phase (Chanet al. 1996), yet the DTCs do not exhibit a repulsive response to UNC-6 at this time. UNC-40 is presumably inactive or ineffective in mediating a repulsive response to UNC-6 until the appropriate time of initiation of the second phase of migration. At the time of the initiation of the ventral-to-dorsal migration phase, UNC-40 generates a repulsive response to UNC-6 due to the expression at this time of UNC-5 (Suet al. 2000) and also due to an unidentified UNC-5-independent switching mechanism.
The UNC-6 signaling model proposed here suggests genetic screening strategies capable of isolating particular signaling pathways in the DTCs downstream of UNC-5 and UNC-40. The isolation of mutations in genes encoding downstream components will permit further examination of UNC-6 signal transduction mechanisms through the UNC-5 and UNC-40 receptors.
We thank members of the Culotti lab for comments and discussions, especially Drs. R. Steven and N. Levy-Strumpf for comments on the manuscript. Some worm strains not derived in our lab were obtained from the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources. This work was supported by a National Cancer Institute of Canada Terry Fox Postdoctoral Fellowship (to D.C.M.), a Fellowship from the Medical Research Council (MRC) of Canada (to M.T.K.), and grants from the MRC of Canada and the Spinal Cord Research Foundation of Canada (to J.G.C.).
Communicating editor: R. K. Herman
- Received May 10, 2000.
- Accepted April 6, 2001.
- Copyright © 2001 by the Genetics Society of America