During Caenorhabditis elegans development, the HSN neurons and the right Q neuroblast and its descendants undergo long-range anteriorly directed migrations. Both of these migrations require EGL-20, a C. elegans Wnt homolog. Through a canonical Wnt signaling pathway, EGL-20/Wnt transcriptionally activates the Hox gene mab-5 in the left Q neuroblast and its descendants, causing the cells to migrate posteriorly. In this report, we show that CAM-1, a Ror receptor tyrosine kinase (RTK) family member, inhibits EGL-20 signaling. Excess EGL-20, like loss of cam-1, caused the HSNs to migrate too far anteriorly. Excess CAM-1, like loss of egl-20, shifted the final positions of the HSNs posteriorly and caused the left Q neuroblast descendants to migrate anteriorly. The reversal in the migration of the left Q neuroblast and its descendants resulted from a failure to express mab-5, an egl-20 mutant phenotype. Our data suggest that CAM-1 negatively regulates EGL-20.
MANY cell types migrate during vertebrate development. For example, germ cells, cardiac precursors, melanocytes, neurons, and neuronal growth cones migrate extensively, often traversing long distances, to reach their targets. Several cells migrate long distances during Caenorhabditis elegans development (Figure 1A; Sulston and Horvitz 1977; Sulston et al. 1983; Hedgecock et al. 1987). For example, the canal-associated neurons (CANs) and anterior lateral microtubule (ALM) neurons migrate posteriorly to positions near the middle of the animal during embryogenesis. Also in embryos, the hermaphrodite-specific neurons (HSNs) migrate anteriorly from the tail to the middle of the animal, and the BDU neurons migrate a short distance anteriorly. The left (L) and right (R) Q neuroblasts and their descendants (hereafter referred to as QL and QR descendants, respectively) migrate during the first larval stage (Figure 1B; Sulston and Horvitz 1977). The QL descendants migrate posteriorly, whereas the QR descendants migrate anteriorly.
To understand how these long-range cell migrations are regulated, we have turned to genetic screens to identify genes that are required for normal cell migration to occur. An example of such a gene is cam-1 (also called kin-8; Forrester et al. 1999; Koga et al. 1999). Mutations in cam-1 cause anterior displacement of the final positions of cells that migrate during embryonic development. The CAN and ALM neurons, which migrate posteriorly during embryogenesis, stop prematurely along their migratory routes in cam-1 mutants, whereas the HSN and BDU neurons migrate anteriorly beyond their normal destinations (Forrester and Garriga 1997; Forrester et al. 1999). Mutations in cam-1 also disrupt postembryonic migrations. For example, the postembryonic migrations of the QR descendants are disrupted so that the cells sometimes fail to migrate to their normal positions (Forrester and Garriga 1997).
In addition to their cell migration defects, cam-1 mutants fail to properly orient the polarity of specific cells (Forrester et al. 1999). Six V cells located on each side of the animal divide asymmetrically during the first larval stage to produce an anterior daughter that fuses with the syncitial hypodermis and a posterior blast cell (Sulston and Horvitz 1977; Podbilewicz and White 1994). Similarly, the six Pn.aap neuroblasts P3.aap through P8.aap divide asymmetrically in males during the first larval stage to generate an anterior CA neuron, which accumulates low levels of the neurotransmitter serotonin, and a posterior CP neuron, which accumulates high levels of serotonin (Sulston et al. 1980; Loer and Kenyon 1993). The fates of the daughter cells of the most anterior V cell (V1) and male Pn.aap neuroblast (P3.aap) often are reversed in cam-1 mutants (Forrester et al. 1999). Finally, the CP neurons of wild-type males extend posteriorly directed axons, whereas the most anterior CP neuron of cam-1 mutant males often extends its axon anteriorly (Forrester et al. 1999).
CAM-1 belongs to the Ror class of receptor tyrosine kinases (RTKs). Ror kinases from humans, rat, mouse, Xenopus, and Drosophila have been reported (Masiakowski and Carroll 1992; Wilson et al. 1993; Oishi et al. 1997, 1999; Al-Shawi et al. 2001; Hikasa et al. 2002). Rors are characterized by the presence of conserved domains including extracellular immunoglobulin domains (Ig), cysteine-rich domains (CRD), and kringle (Kri) domains and intracellular tyrosine kinase (Kin) and serine/threonine-rich (S/T) domains (reviewed in Forrester 2002). Vertebrates and Drosophila contain two Ror genes. The C. elegans genome, by contrast, contains only a single Ror gene, cam-1 (Forrester et al. 1999; Koga et al. 1999). Vertebrate Rors are required for normal bone and heart development (Afzal et al. 2000; DeChiara et al. 2000; Oldridge et al. 2000; Schwabe et al. 2000; Takeuchi et al. 2000; van Bokhoven et al. 2000; Nomi et al. 2001).
Mutations in cam-1 and egl-20 affect many of the same cells. For example, migrating HSNs and QR descendants are misplaced posteriorly along their migratory routes in egl-20 mutants (Figure 2; Table 1; Harris et al. 1996; Whangbo and Kenyon 1999). Thus, the HSN migration defect in egl-20 mutants is opposite to that seen in cam-1 mutants, but the QR descendant defect is similar. QL descendants reverse direction to migrate anteriorly in egl-20 mutants, apparently because they fail to express the mab-5 homeobox gene in QL (Harris et al. 1996; Maloof et al. 1999; Whangbo and Kenyon 1999). QL descendant migration, by contrast, appears to be normal in cam-1 mutants (Kim and Forrester 2003). In egl-20 mutants, the polarity of the fifth V cell, V5, is often reversed (Whangbo et al. 2000); in cam-1 mutants, the polarity of V1 is sometimes reversed (Forrester et al. 1999). egl-20 encodes a Wnt protein that appears to function in a canonical Wnt signaling pathway that includes MIG-14, LIN-17 (Frizzled), MIG-1 (Frizzled), BAR-1 (β-catenin), the negative regulator PRY-1 (Axin), and POP-1 (TCF) to regulate QL migration (Harris et al. 1996; Sawa et al. 1996; Maloof et al. 1999; S. Clark and C. Bargmann, personal communication; reviewed in Herman 2002; Korswagen 2002). EGL-20 is produced by several cells in the tail (Whangbo and Kenyon 1999).
Analysis of cam-1 derivatives from which specific domains were deleted showed that the CRD was required for proper cell migration, but the intracellular region was not (Kim and Forrester 2003). Notably, a version of CAM-1 from which all domains besides the CRD and transmembrane domains had been deleted was able to prevent excessive HSN migration in a cam-1 mutant. High levels of CAM-1 CRD expression resulted in posterior displacement of the HSNs, defective V cell polarity, and defective Q cell migration, phenotypes that were strikingly similar to those that resulted from mutations in egl-20/Wnt (Harris et al. 1996; Maloof et al. 1999; Whangbo et al. 2000).
Here we show that mutations in the Wnt gene egl-20 and the Frizzled gene mig-1 suppressed the HSN overmigration defect of cam-1 mutants, whereas mutations in the Frizzled gene lin-17 did not. Excess egl-20, like cam-1 mutations, caused anterior displacement of the HSNs. Expression of the CAM-1 CRD reduced the expression of mab-5 in QL, as do mutations in egl-20. Furthermore, the QL migration defects that result from CRD expression can be suppressed by a gain-of-function mutation in mab-5. Our results support a model in which CAM-1 directs HSN and QL descendant cell migration by negatively regulating EGL-20 signaling. Previous reports that murine and Xenopus Rors are able to bind Wnts (Hikasa et al. 2002; Oishi et al. 2003) raise the possibility that the interaction is via direct binding.
MATERIALS AND METHODS
General procedure and strains:
Strains were grown at 20° and maintained as described (Brenner 1974). In addition to the wild-type strain N2, strains with the following mutations and transgenes were used in this work:
LG III: mab-5(e1751gf) (Hedgecock et al. 1987).
Extrachromosomal arrays: cwEx266[pCAM-1, rol-6(d)], cwEx152[pCAM-1ΔIgKriIntra, rol-6(d)], and cwEx34[pCAM-1ΔCRD, rol-6(d)] (Kim and Forrester 2003).
Cell migration and polarity:
The extent of cell migration in wild-type, mutant, and transgenic animals was determined by comparing the positions of nuclei relative to nonmigratory hypodermal nuclei using Nomarski optics with a Nikon E600 microscope. In transgenic animals carrying extrachromosomal arrays, the CAM-1 proteins were tagged with GFP, and the transgenic animals were identified by expression of GFP. For ALM, BDU, CAN, and HSN cells that migrate embryonically, we scored the positions of the nuclei of these cells relative to nonmigratory hypodermal V and P nuclei in newly hatched hermaphrodite larvae (L1). For the Q neuroblasts and their descendants, which migrate during the L1 stage, we scored the final positions of the Q descendant nuclei relative to the two daughter hypodermal nuclei Vn.a and Vn.p derived from V1-6 in mid-L1 stage hermaphrodites.
QL and QR descendants were identified using Nomarski optics. To assess GFP levels in the Q cell descendants of animals carrying a mab-5::gfp transgene, we examined the animals visually by epifluorescence. mab-5::gfp is expressed in QL descendants but not in QR descendants, showing that its expression is properly regulated in Q cell descendants despite causing QL descendants to migrate anteriorly (not shown). Fluorescence in QL descendants was categorized as bright if the level was indistinguishable from that seen in wild-type animals bearing mab-5::gfp, as faint if levels were still detectable visually, as very faint if it was detectable after a 2-sec exposure using a Cool Snap CCD camera (Photometrics, Tucson, AZ) but not by eye, or as undetectable if it was not detectable after a 2-sec exposure using a Cool Snap CCD camera. Images were captured with a Cool Snap CCD camera controlled with Image-Pro Express software (Media Cybernetics) and manipulated using Adobe Photoshop on an Apple Macintosh computer.
HSN migration and CAM-1:
Several observations suggest that CAM-1 and EGL-20 function together in cell migration. In wild-type embryos, the HSNs migrate anteriorly from their birthplace in the tail to their destinations between the P5/6 and V4 hypodermal cells, a position near the middle of the animal (Figure 1A; Sulston et al. 1983). Loss of cam-1 often causes the HSNs to migrate beyond their normal destinations, while excess cam-1 often causes the HSNs to terminate their migrations prematurely (Figure 2; Table 1; Forrester and Garriga 1997; Forrester et al. 1999; Kim and Forrester 2003). The role of CAM-1 in HSN migration appears to reside in the extracellular CRD since it is both necessary and sufficient for proper HSN migration (Kim and Forrester 2003). CRDs are also found in Frizzled molecules, receptors for Wnts (Bhanot et al. 1996; Masiakowski and Yancopoulos 1998; Saldanha et al. 1998; Xu and Nusse 1998), and Wnts binds Frizzled receptors via their CRD (Bhanot et al. 1996; Lin et al. 1997; Hsieh et al. 1999; Dann et al. 2001). Finally, murine and Xenopus Rors are able to bind Wnts (Hikasa et al. 2002; Oishi et al. 2003). In this context, it is interesting that mutations in the Wnt gene egl-20 disrupt HSN migration, often causing the HSNs to occupy positions posterior to those found in wild-type hermaphrodites (Figure 2; Table 1; Desai et al. 1988; Harris et al. 1996). These observations raise the possibility that CAM-1's role in cell migration is to inhibit EGL-20. In this report, we test this hypothesis.
Excess egl-20 mimicked the effects of mutating cam-1:
One prediction of the model that CAM-1 antagonizes EGL-20 is that excess EGL-20 might exceed the capacity of CAM-1 to restrict its function, thereby causing excessive HSN migration. A complication of testing this prediction is that the CANs, which migrate from the head to positions just anterior to the HSNs, prevent the HSNs from migrating beyond them (Forrester and Garriga 1997). In CAN migration mutants or animals from which the CANs were removed by laser microsurgery, the HSNs often migrate beyond the P5/6 cell, which does not occur in wild-type hermaphrodites (Forrester and Garriga 1997). To circumvent this control, we used a mutation in vab-8 to remove the CANs from their normal positions. Mutations in the gene vab-8 specifically disrupt posteriorly directed cell migrations, including those of CANs (Manser and Wood 1990; Wightman et al. 1996; Wolf et al. 1998). In vab-8 mutant hermaphrodites, the CANs usually are located in the head, and ∼20% of the HSNs migrate beyond their normal destinations (Figure 3; Table 1; Manser and Wood 1990; Wightman et al. 1996; Forrester and Garriga 1997). To test whether overexpression of EGL-20 would cause the HSNs to migrate too far anteriorly, we introduced into vab-8 mutant animals the egl-20 gene fused to gfp, a transgene that fully rescues egl-20 mutant phenotypes (Whangbo and Kenyon 1999). The addition of functional egl-20 genes to a genetic background containing the two endogenous egl-20 genes should result in excess EGL-20; however, we lacked the reagents to determine the extent to which EGL-20 was overexpressed in these transgenic animals. In egl-20::gfp; vab-8 animals, the HSNs migrated excessively 59.3% of the time, a frequency much higher than that seen in vab-8 mutants alone (Figure 3; Table 1). Therefore, increased egl-20 caused the HSNs to migrate too far, a cam-1 mutant phenotype.
If CAM-1 antagonizes EGL-20, then excess egl-20 in hermaphrodites that lack cam-1 function should drive the HSNs even further anterior. To test this prediction, we examined HSN position in cam-1(gm122) mutants that carried the egl-20::gfp transgene. We found that 95.5% of HSNs overmigrated (Figure 3; Table 1).
Mutations in egl-20 suppressed the HSN cell migration defects of cam-1 mutants:
The model that CAM-1 antagonizes EGL-20 function also predicts that mutations in egl-20 should suppress the HSN defects of cam-1 mutants. We scored HSN position in hermaphrodites doubly mutant for both cam-1 and egl-20 and found that the egl-20 mutations completely suppressed the HSN overmigration phenotype caused by the cam-1 mutations. Mutations in cam-1 suppressed the effects of the egl-20 mutations, but to a lesser extent: more HSNs migrated to wild-type positions in the double mutants (Figure 2; Table 1). These effects were observed with two alleles each of cam-1 and egl-20, demonstrating that the suppression was not allele specific.
Mutations in mig-1 suppressed the HSN migration phenotype of cam-1 mutants:
We examined the effects of mutating other genes that have been reported to act with egl-20. lin-17 and mig-1, for example, encode C. elegans Frizzled receptors (Sawa et al. 1996; S. Clark and C. Bargmann, personal communication). EGL-20 and both Frizzled receptors are required for QL cell migration, raising the possibility that both LIN-17 and MIG-1 function as EGL-20 receptors (Harris et al. 1996; reviewed in Herman 2002; Korswagen 2002). Mutations in mig-1 and lin-17 disrupt HSN cell migration, although the effects of lin-17 mutations are weak (Figure 4; Desai et al. 1988; Hedgecock et al. 1987; Harris et al. 1996). Because of the genetic interactions between cam-1 and egl-20, we wondered whether similar interactions might occur between cam-1 and genes that encode putative EGL-20 receptors. To examine possible interactions, we scored HSN position in hermaphrodites doubly mutant for cam-1 and either lin-17 or mig-1.
The positions of the HSNs in mig-1; cam-1 double mutants were similar to those in mig-1 single mutants (Figure 4; Table 1). HSNs no longer migrated beyond their normal positions in the double mutants and often were misplaced posteriorly as in mig-1 single mutants. This observation demonstrates that like egl-20, mutations in mig-1 can suppress the HSN phenotypes of cam-1 mutants. The final positions of HSNs in lin-17; cam-1 double mutants were similar to those seen in cam-1 single mutants (Figure 4; Table 1).
We also analyzed the effects on HSN migration of two other genes involved in Wnt signaling. Both pry-1, which encodes a C. elegans axin homolog, and bar-1, which encodes C. elegans β-catenin, function in QL cell migration (Eisenmann et al. 1998; Maloof et al. 1999). Mutations in neither pry-1 nor bar-1 significantly disrupted HSN cell migration (Figure 4; Table 1; Harris et al. 1996). We found that neither pry-1 nor bar-1 interacted genetically with cam-1 in HSN migration (Figure 4; Table 1). HSN cell positions in pry-1; cam-1 and cam-1; bar-1 double mutants resembled HSN positions in cam-1 single mutants.
CAM-1 influences QL descendant position by regulating the expression of mab-5:
mab-5 encodes an Antennapedia homolog that functions as a key regulator of Q neuroblast migrations (Costa et al. 1998). In wild-type animals, the QL descendants migrate posteriorly, whereas the QR descendants migrate anteriorly (Sulston and Horvitz 1977). This difference in the direction of migration between the two neuroblasts and their descendants is controlled by mab-5 expression. Expression of mab-5 in QL results in posteriorly directed migrations, whereas the lack of mab-5 expression in QR results in anteriorly directed migrations (Salser and Kenyon 1992). EGL-20 and components of a canonical Wnt signaling pathway regulate mab-5 expression in QL descendants (Harris et al. 1996; Maloof et al. 1999; Korswagen et al. 2000). Mutations that disrupt EGL-20 signaling eliminate mab-5 expression, resulting in QL descendants that migrate anteriorly.
While loss of cam-1 does not significantly affect QL descendant migrations (Forrester and Garriga 1997), high levels of CAM-1 driven from its own promoter can cause QL descendants to migrate anteriorly, an egl-20 mutant phenotype (Kim and Forrester 2003). A transgene that retains little more than the extracellular CRD and transmembrane domains of CAM-1 (pCAM-1ΔIgKriIntra) can produce this effect (Table 1; Kim and Forrester 2003).
If CAM-1 inhibits EGL-20 signaling, then expression of the CRD should reduce mab-5 expression in QL descendants, the same phenotype caused by loss of egl-20. To assess mab-5 expression, we used a chromosomally integrated mab-5::gfp transgene (muIs16) that contains 10 kb of DNA upstream of mab-5 fused to gfp (Cowing and Kenyon 1996). mab-5::gfp was introduced into cam-1 mutants in the absence or presence of cam-1 transgenes that express full-length CAM-1 (pCAM-1), a CAM-1 derivative that contains the CRD and transmembrane domains (pCAM-1ΔIgKriIntra), or a CAM-1 derivative that lacks its CRD (pCAM-1ΔCRD) (Kim and Forrester 2003). pCAM-1, pCAM-1ΔIgKriIntra, and pCAM-1ΔCRD are expressed at similar levels from the transgenes used in these experiments (Kim and Forrester 2003). Wild-type animals or cam-1 mutants that carry mab-5::gfp expressed high levels of GFP in QL descendants but not in QR descendants (Figure 5; Table 2). Both the pCAM-1 and pCAM-1ΔIgKriIntra transgenes caused a dramatic reduction in mab-5::gfp expression in QL, while the pCAM-1ΔCRD transgene had no effect on the expression of mab-5::gfp in QL (Figure 5; Table 2). These observations support the hypothesis that the CAM-1 CRD domain inhibits EGL-20 function.
A mab-5 gain-of-function mutation rescued the anterior QL migration defects that result from CAM-1 CRD expression:
If CRD expression causes QL descendants to migrate anteriorly because they fail to activate mab-5, then providing excess mab-5 activity might suppress their anterior migration. To provide mab-5 activity in QL, we utilized the gain-of-function mutation mab-5(e1751gf), which causes ectopic production of MAB-5 (Salser and Kenyon 1992; Salser et al. 1993). mab-5(e1751gf) causes both QL and QR to produce MAB-5 and consequently to migrate posteriorly (Hedgecock et al. 1987; Salser and Kenyon 1992). In egl-20 and mig-1 mutants, mab-5(e1751gf) causes QL descendants to remain in the posterior (Harris et al. 1996). In mab-5(e1751gf) animals that overexpressed the CAM-1 CRD, QL descendants remained in posterior positions (Table 1). This result along with the observation that mab-5 expression is reduced in animals that overexpress the CAM-1 CRD indicates that the anterior migration of QL descendants caused by expression of the CRD results from their failure to express MAB-5 and supports the model that CAM-1 negatively regulates EGL-20.
Interactions between cam-1 and Wnt signaling mutants:
Another prediction of our model is that removing cam-1 function should not affect the QL migration defect of animals lacking egl-20 function. In cam-1; egl-20 double-mutant animals that contain the severe egl-20 mutation n585, the QL descendants were in similar positions in egl-20 single and cam-1; egl-20 double mutants (Figure 6; Table 1). The QL defects of the severe egl-20 mutant n1437, however, were suppressed by removing cam-1; more QL descendants were in their normal positions in the double mutants (Figure 6; Table 1). Mutations in cam-1 did not detectably increase mab-5 expression in QL descendants in either egl-20 mutant (Table 2). The ability of cam-1 mutations to partially suppress egl-20(n1437) but not egl-20(n585) is confusing since previous genetic analysis is consistent with both of these alleles being severely disrupted for egl-20 function (Harris et al. 1996). One interpretation of this suppression is that the egl-20(n1437) mutants retain some egl-20 activity and that removing cam-1 increases egl-20 activity to levels that drive mab-5 expression in QL descendants to sufficient levels to influence their migrations, although still below levels that we can detect.
Two Frizzled homologs, MIG-1 and LIN-17, are necessary for mab-5 expression in QL descendants and hence for their posteriorly directed migrations (Harris et al. 1996). The requirement for two Frizzleds in QL migration is curious and suggests several possible roles for these molecules in QL migration. One simple interpretation of these data is that both Frizzled molecules act as receptors for EGL-20. Consistent with this hypothesis, Harris et al. (1996) have shown that while egl-20 is absolutely required for mab-5 expression in QL descendants, both mig-1 and lin-17 are only partially required. These observations are consistent with both LIN-17 and MIG-1 acting as EGL-20 receptors in QL migration. Removal of either one leaves the function of the other to mediate, albeit less efficiently, the EGL-20 response. If this hypothesis is correct and if CAM-1 inhibits EGL-20 function by direct binding as our model predicts, then loss of cam-1 should partially suppress the QL defect of both lin-17 and mig-1 single mutants by increasing EGL-20 signaling through the remaining functional receptor.
Analysis of cam-1 double mutants did not fit this simple model. In mig-1; cam-1 double mutants, QL descendants often migrated anteriorly as they did in mig-1 single mutants, but they were often located in more posterior positions and mab-5 expression was detected more often in QL descendants than in mig-1 single mutants (Figure 7; Tables 1 and 2). While this result fits the model, mutations in cam-1 enhanced the QL defects of lin-17 mutants; QL descendants migrated anteriorly and failed to express mab-5 much more often in cam-1; lin-17 double mutants than in lin-17 single mutants (Figure 7; Tables 1 and 2).
EGL-20 appears to signal through a canonical Wnt signaling pathway to regulate mab-5 expression in QL and its descendants. To test whether cam-1 could interact with other components of this pathway we made double mutants between cam-1 and either pry-1, which encodes an axin homolog that negatively regulates the pathway (Maloof et al. 1999), or bar-1, which encodes a β-catenin homolog (Eisenmann et al. 1998; Maloof et al. 1999). As with cam-1 or pry-1 single mutants, QL migration was normal in the double mutant (Figure 7; Table 1). The QL migration defect of bar-1 mutants, however, was suppressed by loss of cam-1. The fact that cam-1 loss could suppress the QL defect of mig-1 and bar-1 mutants but not the QL defect of lin-17 mutants argues that mig-1 and bar-1 function in the same pathway to regulate mab-5 expression.
Interactions between cam-1 and egl-20 did not generally affect all cell migrations:
Mutations in cam-1 disrupt CAN, ALM, and BDU migrations as well as those of HSN and Q descendants discussed above (Forrester and Garriga 1997). In contrast, mutations in egl-20, mig-1, lin-17, bar-1, or pry-1 do not substantially disrupt CAN, ALM, or BDU migrations (Harris et al. 1996; Maloof et al. 1999; Table 1). We looked for genetic interactions between cam-1 mutations and mutations in egl-20, mig-1, lin-17, bar-1, and pry-1 for these migrations and found that mutations in these Wnt signaling molecules had no effect on the distribution of these cells in cam-1 mutants (Table 1).
We propose a model where the CAM-1 protein inhibits the function of EGL-20 and perhaps other Wnts by direct binding. Several observations support this model. First, the extracellular CRD of CAM-1 is both necessary and sufficient for CAM-1 function in HSN and QL migrations (Kim and Forrester 2003; this study), and both migrations are mediated by EGL-20 (Harris et al. 1996). Second, the CRDs of Ror kinases are similar to the CRDs of Frizzled family members (Masiakowski and Yancopoulos 1998; Saldanha et al. 1998; Xu and Nusse 1998), and Frizzled proteins are receptors that bind Wnts directly via their CRD (Bhanot et al. 1996; Hsieh et al. 1999; Uren et al. 2000; Dann et al. 2001; Wu and Nusse 2002). Finally, the prediction that Ror kinases can bind Wnts has recently been confirmed for Ror2 from Xenopus and mouse (Hikasa et al. 2002; Oishi et al. 2003). The genetic results presented here support the hypothesis that CAM-1 interferes with EGL-20 function.
CAM-1 and EGL-20 antagonism during HSN and QL migration:
Supporting the hypothesis that CAM-1 interferes with EGL-20 function are results indicating that egl-20 and cam-1 antagonize one another. First, loss of cam-1 resembles the phenotype caused by excess egl-20, and the loss of egl-20 resembles the phenotype caused by excess cam-1. Second, while loss of cam-1 or excess egl-20 had no obvious QL defect, loss of egl-20 or excess cam-1 caused a loss of mab-5 expression in the QL lineage, resulting in the anterior migrations of QL descendants (this study; Harris et al. 1996). Third, loss of both cam-1 and egl-20 resulted in a mutual suppression of the HSN migration defects exhibited by both mutants. The egl-20 mutations clearly suppressed the excessive migration caused by mutations in cam-1; to a lesser extent, cam-1 mutations suppressed the undermigration phenotype caused by mutations in egl-20. For QL migration, loss of cam-1 appeared to suppress the egl-20 mutation n1437, but not n585. This difference may reflect the severity of the egl-20 mutation, with loss of cam-1 capable of suppressing weaker, but not stronger egl-20 mutants. Although the QL descendant migration defects of egl-20(n1437) mutants are at least as severe as those of egl-20(n585) mutants, other phenotypes of egl-20(n1437) mutants appear less severe. For example, egl-20(n585) mutants are smaller, more uncoordinated, and display more severe HSN migration defects than egl-20(n1437) mutants do (Figure 2; not shown). An alternative explanation for this discrepancy is that these strains contain differences in the genetic background that influence QL migration. Because we see no mab-5::gfp expression in either cam-1; egl-20 double mutant, it is possible that these putative strain differences affect the migration of the QL descendants independently of Wnt signaling. At this point we cannot distinguish between these two general explanations.
The differences in cam-1/egl-20 genetic interactions between HSN and QL migrations might result from the involvement of other Wnts in HSN, but not QL migration. mab-5 expression in Q descendants requires EGL-20 (Harris et al. 1996), and thus EGL-20 may be the only Wnt involved in regulating expression of this Hox gene. This would explain why cam-1 mutations did not suppress the strong egl-20 allele n585. The ability of cam-1 to partially suppress the HSN migration defect of both of the egl-20 alleles tested could mean that these egl-20 alleles retain some activity. While Harris et al. (1996) showed that n585 and n1437 behaved like strong loss-of-function alleles in genetic tests, the n585 allele is a missense mutation, and the sequence of n1437 has not been reported (Maloof et al. 1999). An alternative explanation for the ability of cam-1 mutations to suppress the HSN defect of both egl-20 alleles is that other Wnts antagonized by CAM-1 promote HSN migration. The C. elegans genome encodes five Wnts (Shackleford et al. 1993; Herman et al. 1995; Rocheleau et al. 1997; Thorpe et al. 1997; Maloof et al. 1999). In this scenario, cam-1 loss would enhance the activity of other Wnts involved in HSN migration, resulting in bypass suppression of strong egl-20 alleles.
EGL-20 signaling in HSN and QL descendant migration:
If CAM-1 acts by modulating EGL-20 signaling, then how might EGL-20 function? For QL migration, EGL-20 appears to act though a canonical signaling pathway that regulates the ability of the TCF transcription factor POP-1 to activate mab-5 (reviewed in Herman 2002; Korswagen 2002). While the role of EGL-20 in QL migration is well established, how EGL-20 regulates HSN migration is unclear. EGL-20 could use a canonical signaling pathway that transcriptionally regulates genes necessary for HSN migration. Several transcription factors are necessary for HSN migration, including the Hox protein EGL-5 (Desai et al. 1988; Garriga et al. 1993; Wang et al. 1993; Baum et al. 1999). While it is possible that EGL-20 contributes to the regulation of these transcription factors, EGL-20 cannot be the sole regulator since mutations in egl-20 result in less severe HSN defects than those caused by mutations in any of the genes encoding these transcription factors.
Another difference in EGL-20 signaling in these migrations is that QL migration requires bar-1, whereas HSN migration does not (Harris et al. 1996). The lack of an obvious role for BAR-1, a β-catenin, in HSN migration suggests that signaling downstream of EGL-20 is different in QL and the HSN. HSN migration might require one of the two additional β-catenins in C. elegans, WRM-1, and HMP-2 (Rocheleau et al. 1997; Costa et al. 1998), but because these proteins are essential for embryonic development, their roles in HSN migration remain uninvestigated. Another possibility is that HSN migration uses a Wnt signaling pathway that does not utilize a β-catenin. Wnt signaling pathways that do not involve β-catenins are referred to as “noncanonical” and regulate processes as diverse as planar cell polarity in Drosophila and gastrulation movements in vertebrates (for a recent review, see Veeman et al. 2003).
Wnts have recently been implicated in growth cone guidance. The Drosophila Wnt5 appears to be a repellant for axons that express the Derailed RTK, ensuring that the axons cross the midline at the anterior and not the posterior commissure (Yoshikawa et al. 2003). During vertebrate spinal cord development, commissural axons first extend ventrally to the floor plate, which they then cross. Once across the floor plate, the axons turn anteriorly. A decreasing anterior to posterior gradient of Wnt4 guides the axons and has been proposed to directly attract the growth cones anteriorly (Lyuksyutova et al. 2003). It is also possible that EGL-20 guides the HSNs out of the tail. Because EGL-20 is expressed in cells of the tail (Whangbo and Kenyon 1999), it could act as an HSN repellant.
Two Frizzled receptors in QL migration:
One of the enigmas of Wnt signaling in C. elegans is that QL migration requires two Frizzled receptors. The ability of cam-1 loss of function to suppress the QL defect of mig-1 mutants and enhance the defect of lin-17 surprised us and suggests that MIG-1 and LIN-17 function differently in QL migration. The ability of cam-1 loss to also suppress the QL defect of bar-1 and presumptive hypomorphic egl-20 mutants suggests that EGL-20, MIG-1, and BAR-1 function in a canonical pathway that regulates mab-5. In this regard it is noteworthy that mig-1 but not lin-17 plays an important role in HSN migration (Harris et al. 1996). If this interpretation of our results is correct, the role of LIN-17 in QL migration is unclear.
The ability of cam-1 loss to suppress the QL defects of mig-1 and bar-1 mutants suggests that there are additional Frizzled and β-catenin molecules or noncanonical pathways involved in EGL-20's activation of mab-5, that CAM-1 has other functions besides inhibiting EGL-20 function, or that these alleles retain some activity. Our observation that cam-1(gm122) mutation does not suppress a mab-5 loss-of-function mutation (data not shown) suggests that the mab-5 gene is a major target of cam-1 function in QL migration.
We thank the Caenorhabditis Genetics Center for providing strains used in this work. This work was supported by National Institutes of Health grant R01 HD37815 to W.F. and R01 NS32057 to G.G.
Communicating editor: K. Kemphues
- Received May 28, 2004.
- Accepted September 10, 2004.
- Genetics Society of America