ABSTRACT
The assembly of the nervous system in the nematode C. elegans requires the directed migrations of cells and growth cones along the anteroposterior and dorsoventral body axis. We show here that the gene vab-8 is essential for most posteriorly directed migrations of cells and growth cones. Mutations in vab-8 disrupt fourteen of seventeen posteriorly directed migrations, but only two of seventeen anteriorly directed and dorsoventral migrations. For two types of neurons that extend axons both anteriorly and posteriorly, vab-8 mutations disrupt only the growth of the posteriorly directed axon. vab-8 encodes two genetic activities that function in the guidance of different migrations. Our results suggest that most posteriorly directed cell and growth cone migrations are guided by a common mechanism involving the vab-8 gene.
INTRODUCTION
The migrations of neurons and their growth cones shape the structure and connectivity of developing nervous systems. Notable examples include the migrations of neural crest cells to generate the peripheral nervous system of vertebrates (Le Douarin, 1980) and the radial migrations of neurons that form the layers of the mammalian cortex (O’Rourke et al., 1992; Walsh and Cepko, 1993). In addition, extensive migrations of growth cones generate the axonal patterns of connectivity required for nervous system function (Goodman and Shatz, 1993).
Growth cones and migrating cells navigate through complex environments to reach their destinations by responding to guidance cues provided by other cells and the extracellular matrix. These cues include cell-associated molecules, immobilized matrix components and diffusible molecules. In vivo and in vitro studies have defined activities that stimulate or inhibit axon outgrowth nondirectionally, and directional activities that attract or repel migrating cells and growth cones (Reichardt and Tomaselli, 1991; Keynes and Cook, 1995). A major goal of developmental neurobiology is to understand how these cues are provided and interpreted as well as how individual cells integrate information from multiple cues.
Genetic analysis of the nematode Caenorhabditis elegans has identified a guidance mechanism that provides directional information along the dorsoventral axis to migrating cells and growth cones. The product of the C. elegans gene unc-6 acts as a directional cue for both dorsally and ventrally directed migrations (Hedgecock et al., 1990; McIntire, et al., 1992). unc-6 encodes a secreted, laminin-related protein that has been proposed to be distributed in a gradient along the dorsoventral axis (Hedgecock et al., 1990; Ishii et al., 1992). The unc-5 gene is specifically required for dorsally directed migrations and appears to encode an UNC-6 receptor (Leung-Hagesteijn et al., 1992; Hamelin et al., 1993).
UNC-6 is closely related to the netrins, vertebrate proteins that act in vitro as chemotropic molecules for commissural growth cones in the developing spinal cord (Tessier-Lavigne et al., 1988; Kennedy et al., 1994; Serafini et al., 1994). netrin1 is secreted by the ventral floor plate cells, suggesting that a gradient of netrin activity might direct commissural growth cones ventrally to the floor plate. Therefore, one cue that functions in guiding circumferential migrations appears to be functionally conserved between species (Goodman, 1994).
Although components involved in directing circumferential (dorsoventral) migrations have been identified, relatively little is known about the molecules involved in directing longitudinal (anteroposterior) migrations. Longitudinal axon outgrowth and cell migration is a prominent feature of the development of the C. elegans nervous system. Most neuronal cell bodies in C. elegans are located in ganglia of the head and tail or ventrally along the length of the animal (Fig. 1; Albertson and Thomson, 1976; White et al., 1976, 1986). Many of these neurons extend axons along the main longitudinal nerve bundles, the ventral and dorsal nerve cords, while others extend longitudinal axons along several smaller lateral nerve bundles. In addition, several cell bodies migrate from the head or tail to specific positions along the length of the body.
Mutations in the vab-8 (variable abnormal morphology) gene were shown previously to impair the longitudinal migrations of three cell types (Manser and Wood, 1990). Here, we describe in detail the role of vab-8 in guiding the migrations of cells and growth cones along the anteroposterior body axis. Our results indicate that mutations in vab-8 disrupt most posteriorly directed migrations, yet they fail to perturb most circumferential and anteriorly directed migrations. Based on vab-8 mutant phenotypes, we propose that vab-8 acts in the production or interpretation of a global longitudinal guidance cue.
MATERIALS AND METHODS
Strains and genetics
Standard methods of culturing and handling animals were used (Brenner, 1974). The vab-8 alleles e1017 and ct33, and the uncoordinated (Unc), withered-tail (Wit), multivulva (Muv), and protruding vulva (Pvl) phenotypes have been previously described (Manser and Wood, 1990). We also noted that many vab-8 mutants are constipated (Con; see Thomas, 1990). The vab-8 alleles gm84 and gm99 are EMS-induced mutations isolated in screens for mutants with displaced CANs (W.C.F., F. Wolf and G.G., unpublished results). ev411 was identified in a screen for Unc mutants (J. Culotti, personal communication) and formerly defined the gene unc-107. Our results indicate that ev411 is a vab-8 allele: ev411 causes defects in axon outgrowth like other vab-8 mutations (Table 1), maps to the same 0.3 mu interval as vab-8 mutations (B.W., A.M.T., and G.G., unpublished results), and fails to complement vab-8(ct33) for the Unc phenotype (Table 2).
Heteroallelic heterozygotes were constructed by mating vab-8/+ males to dpy-11(e224) vab-8(ct33) hermaphrodites (3.7 mu separate dpy-11 and vab-8). Non-Dpy F1 progeny were scored for Vab-8 phenotypes. The presence of gm84 or ev411 in the putative heteroallelic F1 progeny was confirmed by placing individual F1 progeny onto separate Petri plates and examining the F2 for the presence of 3/4 Vab non-Dpy and 1/4 Vab Dpy progeny. Animals hemizygous for vab-8 and the genetic deficiency ctDf1 were constructed by mating ev411/+, gm84/+, or rh205/+ males to ctDf1/nT1 unc(n754dm) hermaphrodites. F1 animals that did not display the dominant Unc phenotype caused by nT1 (n754dm) were scored for Vab-8 phenotypes. The presence of the vab-8 mutation was verified by placing individual F1 progeny onto separate Petri plates and examining the F2 progeny for the absence of wild-type animals.
Two additional vab-8 alleles were identified by their failure to complement the vab-8(ct33) mutation. lon-3(e2175) males were mutagenized with EMS (Sulston and Hodgkin, 1988) and mated to dpy-11(e224) vab-8(ct33) V; xol-1(y9) X hermaphrodites. The xol-1 mutation kills males and was included to prevent mating among the F1 progeny. The cross progeny (non-Dpy) were screened for hermaphrodites that displayed the Vab-8 Unc or Wit phenotypes. Animals homozygous for new vab-8 mutations were isolated by picking Lon F2 animals. From 5320 F1s screened, two new vab-8 alleles (gm107 and gm108) were identified.
Several results suggest that mutations causing the strongest phenotypes represent null alleles. First, homozygous (vab-8(strong)/vab-8(strong)) and hemizygous (vab-8(strong)/Df) animals exhibited Unc, Wit, Muv, Con, and ALA and AVKR axon outgrowth defects at similar penetrances and severities (see Results for description of types of vab-8 alleles; Table 2; B.W. and G.G., unpublished data). Second, a strong vab-8 allele behaved similarly to a deficiency when placed in trans to a weak vab-8 allele. That is, heterozygous vab-8(gm84)/vab-8(ct33) and hemizygous vab-8(gm84)/ctDf1 animals exhibited Unc, Wit, Pvl, Muv and Con defects at similar penetrances and severities (Table 2). Third, three of the vab-8 alleles, one (ct33) isolated in a previous study and two (gm107 and gm108) isolated in this study, were isolated in screens for mutations that failed to complement vab-8(e1017) or vab-8(ct33). Because hemizygous vab-8(e1017)/ctDf1 and vab-8(ct33)/ctDf1 animals are viable, null alleles could have been isolated in these screens, even if null alleles of vab-8 are lethal or sterile. In these screens no alleles were identified with more severe phenotypes than existing strong alleles. Finally, vab-8 alleles were isolated at frequencies typical for mutations that reduce or eliminate gene function (Brenner, 1974; Greenwald and Horvitz, 1980). In two separate screens, one for alleles that fail to complement vab-8(ct33) (see above) and one for CAN-migration mutants (W. C. F., F. Wolf and G. G., unpublished results), the forward mutation frequency was 3.8×10−4 and 1.9×10−4/mutagenized chromosome, respectively.
Because vab-8 mutant males cannot mate, we were unable to establish homozygous vab-8 mutant strains that produce the large number of males required for analysis of the CPn axons. vab-8 mutant males were obtained by crossing the vab-8 alleles into a him-6(e1423) mutant background, which causes a high incidence of males. Homozygous him-6(e1423) males were crossed to homozygous vab-8 mutant hermaphrodites. F1 wild-type progeny were obtained and numerous F2 Vab-8 mutant progeny were picked to individual plates. F2 animals that were also homozygous for him-6(e1423) were identified by the appearance of males in the F3 generation.
Indirect immunofluorescence histochemistry and microscopy
Animals were raised at 25°C prior to fixation in all staining protocols. Animals were fixed, permeabilized and incubated with rabbit antiserotonin or anti-GABA antisera (provided by J. Steinbusch, Free University, Amsterdam) as described by McIntire et al. (1992). Fixation, permeabilization and incubation of animals with a rabbit anti-FMRF antiserum (provided by Chris Li, Boston University, Boston, MA) and rabbit anti-β-galactosidase antiserum (Promega) was performed by the same methods used for anti-serotonin staining.
For UNC-86 and 611B1 staining, animals were fixed, permeabilized and incubated with 1% UNC-86 antiserum (provided by M. Finney and G. Ruvkun, Massachusetts General Hospital, Boston, MA) or 0.5% monoclonal antibody 611B1 (provided by G. Piperno, Rock-efeller University, New York, NY), as described by Finney and Ruvkun (1990). Occasionally animals were not permeabilized using this procedure. In these cases the animals were placed in a Dounce homogenizer and permeabilized with three strokes of the pestle. After the final wash, 5 μl of stained worms were mixed with 5 μl of a solution containing 20 mg/ml n-propyl gallate, 70% glycerol, 30 Mm Tris-HCl, pH 9.5.
ADE and PDE are dopaminergic neurons. To visualize the axons of these neurons, we indirectly detected the activity of aromatic amino acid decarboxylase (Loer and Kenyon, 1993), the enzyme that synthesizes dopamine and serotonin from dopa and 5-hydroxytryptophan, respectively. 250 μl of a solution containing 20 mg/ml 5-hydroxytryptophan (Sigma) in M9 was added to a single 60 mm plate containing a growing population of nematodes, and the animals were incubated for 4 hours to overnight at 25°C. After the 5-hydroxy-tryptophan treatment, the animals were fixed, permeabilized and stained with anti-serotonin antiserum as described by McIntire et al. (1992). Dopaminergic neurons take up the exogenously supplied 5-hydroxytryptophan, and because they contain aromatic amino acid decarboxylase, these cells synthesize serotonin from 5-hydroxytryptophan.
Stained worms were viewed by immunofluorescence microscopy using a Zeiss microscope and Zeiss filters no. 487910 and no. 487915. FITC- or Texas Red-conjugated secondary antibodies were obtained from Cappel, Inc. Images were photographed using Ektachrome T160 film, scanned with a Nikon scanner to a computer graphics file, and annotated and enhanced to improve contrast using the Adobe Photoshop graphics program.
glr-1::GFP, glr-1::mec-4(dm), ceh-23::lacZ and UL64A1 transgenic strains
To visualize the AVA, AVB, AVD and PVC neurons, we expressed an UNC-76-GFP fusion protein under the control of the glr-1 promoter. The glr-1 gene encodes a C. elegans glutamate receptor subunit, which is expressed in AVA, AVB, AVD, PVC and additional neurons. A complete description of the glr-1::unc-76::GFP strain is given by Maricq et al. (1995). Briefly, a hybrid gene containing the coding region for the first 197 amino acids of the UNC-76 protein (L. Bloom and H. R. Horvitz, personal communication) and the green fluorescent protein (GFP) of the jellyfish Aequorea victoria (Chalfie et al., 1994) was placed under the control of the glr-1 promoter. Animals carrying a glr-1::unc-76::GFP extrachromosomal array were produced by germline transformation of lin-15 animals with lin-15 and glr-1::unc-76::GFP DNA (Mello et al., 1991). We treated these animals with γ-irradiation to yield a strain that possessed the glr-1::unc-76::GFP gene integrated into the genome. The fusion of the N-terminal region of unc-76 to β-galactosidase had been shown to enhance the expression of β-galactosidase within axons and to exclude it from the nucleus (L. Bloom and H.R. Horvitz, personal communication). We observed a similar result using the glr-1::unc-76::GFP fusion protein (hereafter referred to as glr-1::GFP).
We used a glr-1::mec-4(dm) fusion gene to kill the AVA, AVB, and AVD neurons (see Results). This construct utilizes the same glr-1 promoter described above, but fuses it to the cytotoxic mec-4(dm) gene (Driscoll and Chalfie, 1991). The details of the glr-1::mec-4(dm) fusion gene are described in Maricq et al. (1995).
We used a ceh-23::lacZ transgene to visualize the CAN axons. This construct is a translation fusion of the promoter and N-terminal region of ceh-23, which encodes a homeobox gene (Wang et al., 1993), to the 197 amino acid fragment of unc-76 described above and the lacZ gene of E. coli (J. Zallen and C. I. B., unpublished data).
The UL64A1 transgene allowed the visualization of the excretory canals by β-galactosidase staining. This transgene is the UL6 line described by Young and Hope (1993).
In all cases the transgenes were crossed into vab-8 mutant back-grounds by mating wild-type males to hermaphrodites that bore the transgene, and crossing F1 males to homozygous vab-8 hermaphrodites. Wild-type cross progeny were isolated and picked individually to single plates. Multiple F2 vab-8 progeny were picked individually to single plates and screened for the presence of the transgene by direct examination for fluorescence or β-galactosidase activity.
Analysis of PHA and PHB axon outgrowth requirements
PHA and PHB are sensory neurons in the tail that have shortened axons in vab-8 mutants. To examine whether interneurons that express glr-1 are required for normal PHA and PHB axon outgrowth, we used a strain that contained a glr-1::mec-4(dm) extrachromosomal array (kyEx53), which killed the glr-1-expressing cells. This strain also carried the integrated glr-1::GFP transgene (kyIn29), which allowed for the identification of the glr-1-expressing cells that were not killed by the array (Maricq et al., 1995). By epifluorescence microscopy, we identified animals that expressed the glr-1::GFP transgene in particular neurons and scored the PHA and PHB axons by filling with the fluorescent dye 3,3′-dioctadecyloxacarbocyanine (DiO; Molecular Probes) using a modification of the procedure developed by Hedgecock et al. (1985).
Scoring of axons and cells
Axons were scored as defective in Table 1 if they failed to reach their normal destination as defined by electron microscopic reconstructions of the C. elegans nervous system (White et al., 1986). Occasionally, in wild-type animals (or, in the case of the CPn axons, in him-6(e1423) animals) axons failed to reach their defined normal destination. This background of variability is shown in the wild-type column of Table 1. The extent of cell migration in wild-type and mutant animals was determined by comparing the positions of migratory cell bodies relative to the positions of non-migratory hypodermal cell bodies (see Fig. 8). In Table 1, for each migratory cell the region between hypodermal cells in which most of the cells were located in wild-type animals was defined as the normal destination. Cell bodies that were located in other regions along their normal route were scored as defective for migration. Thus, the wild-type column in Table 1 represents the natural variability of the position of a given cell. The data shown in Table 1 and Fig. 8 are for three representative alleles. The other vab-8 alleles were scored for the following phenotypes and found to be similar to the corresponding representative allele: gm107 and gm108 were scored for AVA, AVB, AVD, AVKR, ALA, RID axons and all cell migrations described in Table 1; ct33 and rh205 were scored for ADE, AVA, AVB, AVD, AVKR, ALA, CPn, HSN, PDE, PVC, RID, RMG axons and all cell migrations described in Table 1; and gm99 was scored for AVKR, ALA, RID axons and all cell migrations described in Table 1.
RESULTS
vab-8 mutations cause the premature termination of posteriorly directed axons
In vab-8 mutants, we observed defects in four classes of neurons that extend axons posteriorly along the ventral nerve cord from cell bodies located in the head. The AVAs, AVBs, AVDs and AVKs are pairs of bilaterally symmetrical interneurons located near the nerve ring (Figs 2, 3). Each neuron extends a single axon partially around the nerve ring and then posteriorly along the ventral nerve cord to the tail (White et al., 1976, 1986).
The AVA, AVB and AVD axons can be visualized in a glr-1::GFP transgenic strain (Maricq et al., 1995; see Materials and Methods) and one AVK axon, AVKR, can be detected using an anti-FMRFamide antiserum (Schinkmann and Li, 1992). In vab-8 mutants, the cell bodies of these neurons occupied their normal positions, but their axons terminated prematurely, ending in the ventral nerve cord at positions roughly midway between the nerve ring and the anus (Figs 2C, 3C; Table 1). Therefore, vab-8 is required for these axons to extend to their normal length.
The six male-specific CP neurons are located in the ventral cord along the midbody and each extends an axon posteriorly along the ventral nerve cord to the diagonal muscles located in the posterior half of the body (Sulston et al., 1980; Fig. 4). The anatomy of the CPn axons was examined using anti-serotonin antibodies (Loer and Kenyon, 1993). In wild-type males, the CP axons extended posteriorly in a bundle that thickened near the tail as more CP axons were added. By contrast, in vab-8 males, the diameter of the CP axon bundle near the tail was thinner than in wild-type males, suggesting that some CP axons failed to reach their normal destinations (data not shown). In addition, CPn axons often projected anteriorly beyond the CP1 cell body in vab-8 mutants, implying that some CPn growth cones extended anteriorly rather than posteriorly from a CPn cell body (Fig. 4C; Table 1).
ALA is a single interneuron located in the head that extends two bilaterally symmetrical processes along lateral nerve bundles to the tail (White et al., 1986 and Fig. 5A). Immunofluorescent staining of animals with an anti-FMRFamide antiserum revealed that the ALA axons failed to reach the tail in vab-8 mutants (Fig. 5C; Table 1). Thus, vab-8 is required for the posteriorly directed outgrowth of axons in the ventral nerve cord and in a lateral nerve cord.
We identified one posteriorly directed axon unaffected by vab-8 mutations. RID is a single motorneuron that extends several axons, the longest of which extends along the dorsal cord to the tail (White et al., 1986). Using an anti-FMRFamide antiserum, we observed no defects in RID axon morphology in vab-8 mutants (Fig. 5C; Table 1). Thus, the posteriorly directed outgrowth of eight of nine classes of axons examined depend upon vab-8 function.
Most anteriorly directed and circumferential axons are normal in vab-8 mutants
We also examined anteriorly directed axon outgrowth in vab-8 animals. The PVC interneurons, which can be visualized in the glr-1::GFP strain, are located in the tail and extend axons along the ventral nerve cord to the nerve ring (Fig. 2). The HSN motorneurons, which can be detected using an anti-serotonin antiserum, are located in the midbody and extend axons ventrally to the ventral nerve cord and then anteriorly to the nerve ring. The ALMs and PLMs are each symmetric pairs of sensory neurons that can be detected using an anti-tubulin monoclonal antibody. The ALMs are located in the midbody and extend lateral axons anteriorly to the nerve ring, while the PLMs are positioned in the tail and extend lateral axons anteriorly to the midbody (White et al., 1986, and Fig. 1). The anteriorly directed PVC, HSN, ALM and PLM axons were normal in vab-8 mutants (Table 1).
Circumferential growth cone migrations were also unperturbed in vab-8 mutants. The ventral migrations of ADE, HSN, RMG, and PDE growth cones and the dorsal migrations of the DDn and VDn motor neuron growth cones were all normal in vab-8 animals (Table 1).
vab-8 mutations preferentially disrupt outgrowth of posteriorly directed axons of two bipolar neurons
In principle, vab-8 could function in both anteriorly and posteriorly directed growth cone migrations in a subset of neurons. A general function of vab-8 in migration may not be reflected in our data if the neurons we sampled were fortuitously biased towards neurons with posteriorly directed axons that require vab-8 and anteriorly directed and circumferential axons that do not require vab-8. Evidence that vab-8 is specifically required for posteriorly directed migrations was provided by examining the PDE and CAN neurons, each of which extend both an anteriorly and a posteriorly directed axon. The PDEs are a pair of bilaterally symmetrical dopaminergic mechanosensory neurons located laterally in the posterior body (Fig. 6A). Each PDE neuron extends an axon to the right ventral nerve cord, where it bifurcates and extends one axon anteriorly to a position behind the pharynx and a second axon posteriorly to a position near the anus (White et al., 1986 and Fig. 6).
We examined PDE axon morphology using an immuno-cytochemical technique that detects aromatic amino acid decarboxylase activity (see Materials and Methods). In wild-type and vab-8 animals, the PDEs extended axons to the ventral nerve cord, and the anteriorly directed axons of the PDEs reached their normal positions behind the pharynx. However, the posteriorly directed PDE axons were shortened or missing in vab-8 mutants (Fig. 6C; Table 1).
The CANs are a pair of bilaterally symmetrical neurons located near the center of the animal (Fig. 7). Each CAN extends one axon anteriorly to the nerve ring and a second axon posteriorly to the tail, both along a lateral nerve bundle (White et al., 1986; Durbin, 1987). We examined CAN axon morphology in animals bearing a ceh-23::lacZ transgene (J. Zallen and C. I. B., unpublished results). In vab-8(ev411) and vab-8(gm107) animals (see below for discussion of different vab-8 alleles), the CAN cell bodies were found in their normal positions, and the anterior CAN axons extended normally to the nerve ring. However, the posteriorly directed CAN axons terminated prematurely (Fig. 7C; Table 1). In other vab-8 mutants (ct33, gm84, gm99, gm108, e1017, and rh205), the posteriorly directed CAN axons also terminated before reaching their normal destination in the tail. However, these vab-8 mutations also disrupted CAN cell migration (Fig. 7D, and see below). Therefore, anterior displacement of the CAN cell body may contribute to the severity of the CAN axon defect in these mutants.
These data show that for two cell types, vab-8 mutations disrupt posteriorly directed growth cone migrations, but have no effect on migrations in other directions.
vab-8 mutations and ventral cord lesions cause PHA and PHB axon outgrowth defects
The PHAs and PHBs are bilaterally symmetrical bipolar sensory neurons located in the tail. Each PHA and PHB neuron extends an axon ventrally and then anteriorly along the ventral nerve cord. PHA and PHB can be visualized by staining with the vital dye DiO (Hedgecock et al., 1985; see Materials and Methods). In vab-8 mutants, the anteriorly directed axons of PHA and PHB were often shorter than in wild type (Table 1; J. Culotti, personal communication).
A similar PHA and PHB outgrowth defect can be caused by killing other ventral cord neurons. In a glr-1::mec-4 (dm) transgenic strain (Maricq et al., 1995), many neurons, including AVA, AVB, AVD, PVC and PVQ, degenerate because of the expression of the toxic MEC-4 degenerin protein (Driscoll and Chalfie, 1991). We found that the PHA and PHB axons stopped prematurely in glr-1::mec-4(dm) animals, a phenotype like that seen in vab-8 mutants. Because the glr-1::mec-4(dm) transgene is not expressed in PHA and PHB (Maricq et al., 1995), these results indicate that defects in PHA and PHB axon outgrowth can be caused by defects in other neurons.
To determine which glr-1::mec-4(dm)-expressing neurons were needed for normal PHA and PHB axon extension, we identified animals that retained a subset of glr-1-expressing neurons (see Materials and Methods). We identified nine animals that lacked AVA, AVB and AVD and retained one PVC (two animals), one PVQ (four animals), or one PVC and one PVQ (three animals). All nine animals had shortened PHA and PHB axons. These results suggest that normal PHA and PHB axon outgrowth depends on axons that extend from the head to the tail. Therefore, defects in the outgrowth of AVA, AVB, and/or AVD axons, which normally overlap the PHA/PHB axons in the preanal ganglion, might lead to the premature termination of the PHA and PHB axons in vab-8 mutants.
vab-8 mutations disrupt posteriorly directed cell migrations
Manser and Wood (1990) reported that two vab-8 mutations, e1017 and ct33, disrupt the migrations of the ALM and CAN neurons and the coelomocyte mother cell, ccmL. We examined the positions of eight cell types that migrate posteriorly: the neurons CAN and ALM, the progeny of the left and right coelo-mocyte mother cells (ccmL and ccmR), the mesoblast M, and the somatic gonad precursors Z1 and Z4, which all migrate during embryonic development, and the PQR neurons, which migrate during larval development (Sulston and Horvitz, 1977; Kimble and Hirsh, 1979; Sulston et al., 1983; Hedgecock et al., 1987). To determine the positions of these cells in wild-type and in vab-8 mutants, we observed the cells directly by Nomarski differential interference contrast microscopy. In strong vab-8 mutants, we often found the cell bodies of ALM, CAN, ccmL, ccmR and PQR in positions anterior to their normal location, suggesting that vab-8 activity is needed for these posteriorly directed cell migrations (Fig. 8; Table 1). By contrast, none of the vab-8 mutations caused defects in the positions of M, Z1 and Z4, indicating that these cell migrations do not require vab-8 function (Table 1; data not shown). The dorsal migrations of distal tip cells (dtc) of the gonad were also normal.
We also examined the morphology of the processes extended by the excretory cell. The excretory cell is located ventral to the posterior bulb of the pharynx and extends four tubular processes, called canals (Nelson and Riddle, 1984). In wild-type animals, two canals extend anteriorly and two canals extend posteriorly along the left and right lateral body wall past the anus (Nelson et al., 1983; Sulston et al., 1983; White et al., 1986). To examine the morphology of the excretory cell canals, we used a promoter-trap transgene (UL64A1) that expresses β-galactosidase in the canals (Young and Hope, 1993). In vab-8 mutants, the posteriorly directed canals of the excretory cell terminated prematurely (Table 1).
Some anteriorly directed cell migrations occasionally overshoot their normal destinations in vab-8 mutants
Most anteriorly directed cell migrations were normal or enhanced in vab-8 mutants. The HSN cell bodies migrate anteriorly during embryonic development (Sulston et al., 1983). As reported previously (Manser and Wood, 1990), in some animals one or both HSNs migrated slightly anterior to their normal destinations (11/51 HSNs migrate beyond their normal destination in vab-8(rh205) mutants; Fig. 8; Table 1).
The neuroblast QL migrates posteriorly during postembryonic development, while its homolog QR and its descendants migrate a longer distance anteriorly (Sulston and Horvitz, 1977). As noted above, the posterior migration of the QL daughter PQR is shortened in vab-8 mutants. By contrast, the QR daughters, AVM and SDQR, migrated to positions that were slightly anterior to their normal locations (5/14 for vab-8(e1017) mutants). These results suggest that like the HSNs, QR and its descendants occasionally migrate beyond their normal destinations in vab-8 mutants.
vab-8 mutations do not alter cell fates along the longitudinal body axis
As in other animals, anteroposterior positional information in C. elegans is conferred by a cluster of Hox (or homeotic) genes (Bürglin and Ruvkun, 1993; Wang et al., 1993). Cell fate determination of many stationary cell types as well as the position of some migrating cells is altered when the antero-posterior axis is shifted by mutations in the Hox genes. We examined the fates of several cell types along the longitudinal axis and found no alterations in anteroposterior patterning in vab-8 mutants.
During postembryonic development, twelve Pn.a neuro-blasts (P1.a-P12.a), which are located along the ventral cord, each divide to generate several types of neurons. The Pn.a precursors at different longitudinal positions produce different cell types; e.g. the centrally positioned neuroblasts (P3.a-P8.a) each produce a VC motor neuron, while the other Pn.a neuroblasts do not. In vab-8 mutants, the VC neurons were centrally positioned as in wild type, suggesting that they were derived from P3.a-P8.a. Homologous neuroblasts in the male give rise to the CP neurons described above; the cell bodies of the CP neurons were also found in their correct positions in vab-8 mutants.
Similarly, the twelve V cell precursors (V1L/R-V6L/R), located along the mid-lateral surface of the animal, undergo several rounds of division to generate epithelial cells and neurons. V5L and V5R, but not other V cells, produce the posterior deirid sensory organs. Each posterior deirid sensory organ was properly positioned in vab-8 mutants, indicating that it was derived from V5 (data not shown). Thus, vab-8 mutations do not alter the fates of the Pn.a and V cells along the longitudinal axis.
vab-8 mutants did display defects in some non-migrating cells, such as those involved in vulval development (Manser and Wood, 1990; Table 2). The strongest vab-8 mutations caused two defects in vulval development: a protruding vulva (Pvl) and a multivulva phenotype (Muv) (Table 2). Unlike most Muv mutants (Ferguson et al., 1987), the ectopic vulval pro-trusions in vab-8 mutants were always posterior to the vulva.
vab-8 encodes two genetic activities
All eight vab-8 alleles characterized were recessive and behaved as loss-of-function mutations (Table 2). We grouped vab-8 mutations into three classes based on behavioral and anatomical phenotypes: strong (ct33, e1017, gm99, rh205), weak (gm84, gm108), and uncoordinated or Unc (ev411, gm107).
Strong and weak vab-8 mutants displayed Unc, Wit, Muv, Pvl and Con phenotypes (Table 1), as well as the axon outgrowth and cell migration defects described above (Table 2). The strong vab-8 alleles behaved as genetic null alleles (see Materials and Methods) and therefore likely eliminate most or all vab-8 activity. All phenotypes of the weak alleles were less severe than the strong alleles and were enhanced in trans to a deficiency (Table 1 and 2). Thus, the weak alleles appear to be partial loss-of-function alleles that retain some vab-8 activity.
The Unc vab-8 mutations caused defects in movement, axon outgrowth and some cell migrations comparable to strong mutations, but did not cause Wit, Pvl, Muv or Con phenotypes (Tables 1 and 2; data not shown). Moreover, Unc vab-8 mutations did not affect CAN cell migration (Fig. 8; Table 1). Unlike the weak allele gm84, the phenotypes caused by vab-8(ev411) in trans to a deficiency were similar to vab-8(ev411) homozygotes (Table 2). Furthermore, ev411/ctDf1 animals did not display Wit, Pvl or Muv phenotypes.
These results indicate that vab-8 is a complex locus that encodes two genetic activities. Unc mutations eliminate one vab-8 activity, which is required for a subset of migrations, while strong and weak mutations eliminate or reduce both vab-8 activities.
DISCUSSION
Three observations indicate that vab-8 acts in a global mechanism for directing posteriorly directed cell migration and axon outgrowth. First, vab-8 mutations disrupted fourteen of the seventeen posteriorly directed axon and cell migrations examined, but only two of the seventeen circumferential and anteriorly directed axon and cell migrations examined. Second, vab-8 mutations had selective effects on cells of similar type that differed only in their direction of migration. The only known difference between AQR and PQR, the descendants of QR and QL, respectively, is that AQR migrates anteriorly and PQR migrates posteriorly. vab-8 mutations disrupted the pos-teriorly directed migration of PQR but not the anteriorly directed migration of AQR. Third, vab-8 mutations disrupted the outgrowth of the posteriorly directed axons of the CAN and PDE neurons, while the anteriorly directed axons of these neurons extended to their normal destinations.
vab-8 mutations had widespread effects on development. The mutations disrupted both cell migrations and axon outgrowth, affected both neuronal and non-neuronal cell types, and disrupted migrations that occur during both embryonic and postembryonic stages of development. The trajectories of these migrations vary in route and environment, occurring in the ventral nerve cord and along the lateral body wall. Thus, the affected cell types have little in common other than migration along the longitudinal axis.
The migration defects in vab-8 animals are often incomplete and vary in penetrance. For example, although few of the CAN cell bodies reached their normal destination in strong vab-8 mutants, 17% of CANs in vab-8(rh205) animals did migrate a short distance (Fig. 8). Because the strong vab-8 alleles appear to eliminate all gene activity, we propose that other genes provide or interpret directional information in the absence of vab-8. As the elimination of unc-5, unc-6, or unc-40 activity also fails to block circumferential migrations completely (Hedgecock et al., 1990), partial redundancy in providing or interpreting directional information might be common to much of C. elegans development. Redundancy may also account for the observation that the migrations of RID, M and Z1/Z4 are not affected by vab-8 mutations.
Although we analyzed the morphology of mature axons, we believe that the premature termination phenotype observed in vab-8 animals represents a defect in axon outgrowth. We examined the ALA, AVA, AVB, AVD, AVKR and CAN axons, which normally complete outgrowth during embryogenesis, in first and second stage larvae (L1 and L2) and found that these axons extended to relative positions similar to those seen in adults (data not shown). Because adults are much longer than young larvae, these data suggest that these axons do not extend first to their normal positions and then fail to elongate as the animal grows.
Direct and indirect effects of vab-8 mutations
The observed defects in cell migration and axon outgrowth provide a basis for understanding the behavioral and morphological phenotypes of vab-8 mutants. The uncoordinated movement phenotype of all vab-8 mutants is most severe in the posterior body. Because locomotion is controlled by the AVA, AVB, AVD and PVC interneurons (Chalfie et al., 1985), the premature termination of the AVA, AVB and AVD axons in vab-8 mutants likely contributes to the vab-8 locomotion defects.
During larval development, the posterior body of strong and weak vab-8 mutants fails to grow normally, leading to a Wit phenotype. Manser and Wood (1990) proposed that defects in the function of the CAN neurons in the posterior portion of the animal cause the Wit phenotype. vab-8 strains with more severe CAN migration defects display a more penetrant Wit phenotype, consistent with the hypothesis that the anterior displacement of the CAN cell bodies and axons results in a lack of CAN function in the posterior body that leads to the Wit phenotype.
The reasons for the multivulva and protruding vulva phenotypes of vab-8 mutants are unclear. A defect in vulval morphogenesis, rather than the generation of the cells that produce the vulva, is likely to cause the Pvl defect (D. Eisenmann, personal communication). vab-8 might function directly in vulval morphogenesis, or these vulval phenotypes might result indirectly from other vab-8 defects, such as the withering of the tail. Consistent with the latter possibility, in both strong and weak vab-8 mutants, all Muv animals are also Wit (B. W., F. Wolf and G. G., unpublished results), and the ectopic ventral protrusions of vab-8 Muv animals are located posterior to the vulva in the withered body region.
One observation at odds with the model that vab-8 is required for posteriorly directed migration is that the anteriorly directed PHA and PHB axons typically terminate prematurely in vab-8 animals. However, similar PHA and PHB axonal defects were observed in glr-1::mec-4 animals. This observation raises the possibility that in vab-8 mutants the PHA and PHB outgrowth defects arise as a secondary consequence of the shortening of posteriorly directed axons, such as AVA, AVB and/or AVD. In particular, the AVA and AVD axons are candidates for interacting with PHA and PHB growth cones, because these axons are closely associated, and the AVA and PHB axons synapse extensively (White et al., 1986). However, these experiments do not exclude a direct role of vab-8 in PHA and PHB outgrowth.
vab-8 functions in a global guidance system along the longitudinal axis
Taken together, our results suggest that vab-8 encodes or regulates a component of a global signaling system that provides directional information to migrating cells and growth cones. One possibility is that vab-8 produces or interprets a graded guidance cue distributed along the longitudinal axis. For example, an attractive chemotropic cue might be produced from a source in the tail so that growth cones and migrating cells would migrate posteriorly toward the source (for reviews see Tessier-Lavigne and Placzek, 1991; Tessier-Lavigne, 1994). Alternatively, specific concentrations of a guidance cue might define specific positions along the longitudinal axis. In this model the guidance cue would function similarly to a morphogen for organizing spatial information (Driever and Nüsslein-Volhard, 1988).
Both of these models are similar to those proposed for UNC-6 in C. elegans and the netrins in the vertebrate spinal cord, molecules that function in directing growth cones and migrating cells circumferentially. A molecular analysis of the vab-8 gene should reveal whether vab-8 encodes a signaling ligand, a receptor, or part of the machinery for producing or responding to the signal. In addition, it will be of interest to determine whether the mechanisms controlling longitudinal guidance are also conserved between nematodes and vertebrates.
ACKNOWLEDGEMENTS
We are grateful to Monica Driscoll, Mike Finney, Ed Hedgecock, Ian Hope, Chris Li, James Manser, Barbara Meyer, Gianni Piperno, Fred Wolf, Bill Wood, Jen Zallen, and the Caenorhabditis Genetics Center for kindly providing strains and materials used in this work and to David Eisenmann and Fred Wolf for sharing unpublished results. We also thank Jeff Way and the anonymous reviewers for comments on the manuscript, and members of the Garriga, Meyer and Bargmann laboratories for valuable discussion. This work was supported by a research grant from the Muscular Dystrophy Association, NIH grant NS32057 and a McKnight Scholars Award to G. G. and NIH grant DC01393 to C. I. B. C. I. B. is a Markey Scholar and a Searle Scholar, and some funds for this work were provided by these trusts. C. I. B. is an Assistant Investigator of the HHMI. B. W. was supported by a postdoctoral fellowship from the American Cancer Society, S. G. C. by a postdoctoral fellowship from the Helen Hay Whitney Foundation, W. C. F. by a postdoctoral fellowship from the National Institutes of Health, and A. V. M. by a ADAMHA Scientist Development Award.