Abstract
Over half of the neurons in Caenorhabditis elegans send axons to the nerve ring, a large neuropil in the head of the animal. Genetic screens in animals that express the green fluorescent protein in a subset of sensory neurons identified eight new sax genes that affect the morphology of nerve ring axons. sax-3/robo mutations disrupt axon guidance in the nerve ring, while sax-5, sax-9 and unc-44 disrupt both axon guidance and axon extension. Axon extension and guidance proceed normally in sax-1, sax-2, sax-6, sax-7 and sax-8 mutants, but these animals exhibit later defects in the maintenance of nerve ring structure. The functions of existing guidance genes in nerve ring development were also examined, revealing that SAX-3/Robo acts in parallel to the VAB-1/Eph receptor and the UNC-6/netrin, UNC-40/DCC guidance systems for ventral guidance of axons in the amphid commissure, a major route of axon entry into the nerve ring. In addition, SAX-3/Robo and the VAB-1/Eph receptor both function to prevent aberrant axon crossing at the ventral midline. Together, these genes define pathways required for axon growth, guidance and maintenance during nervous system development.
INTRODUCTION
During development, neurons project axons that often travel long distances through diverse environments to contact their targets (Tessier-Lavigne and Goodman, 1996; Mueller, 1999). The nematode C. elegans provides a simple and well-characterized organism in which to study axon development. There are 302 neurons in the adult hermaphrodite, and the stereotyped position of each neuron, its processes and its putative synaptic connections are known from serial section electron micrographs (White et al., 1986). 180 of these neurons project axons into the nerve ring, the largest axon bundle in the animal. These axons must navigate to the nerve ring, recognize and enter the nerve ring, make specific contacts with each other and maintain these contacts during growth. Little is known about the genes that control nerve ring development.
Several conserved classes of guidance molecules direct axon guidance in C. elegans and other organisms. The UNC-6/netrin diffusible guidance cue is required for axon guidance along the dorsal-ventral axis in C. elegans, Drosophila and vertebrates (reviewed in Tessier-Lavigne and Goodman, 1996 and Culotti and Merz, 1998). Netrins attract or repel different subsets of axons in vivo and in vitro. Attraction to netrins is mediated by UNC-40/DCC, an immunoglobulin superfamily member that is thought to act as a netrin receptor on growing axons, while repulsion from netrins is mediated by UNC-5, a transmembrane receptor that acts in combination with UNC-40/DCC. UNC-6 is present in the region of the nerve ring (Wadsworth et al., 1996), and both the UNC-40 and UNC-5 receptors are expressed in nerve ring neurons (Leung-Hagesteijn et al., 1992; Chan et al., 1996), but it is not known whether these molecules influence nerve ring axon guidance. Nerve ring neurons also express SAX-3/Robo, a transmembrane protein in the immunoglobulin superfamily that is related to Drosophila and vertebrate Roundabout (Robo) proteins. Disruption of Robo function in Drosophila and C. elegans leads to inappropriate axon crossing of the ventral midline (Kidd et al., 1998; Zallen et al., 1998). sax-3 mutants are also defective in the ventral guidance of motor and interneuron axons to the midline, suggesting a role for SAX-3/Robo in multiple guidance events (Zallen et al., 1998).
A third guidance molecule expressed in the C. elegans nerve ring is the VAB-1 Eph receptor tyrosine kinase (George et al., 1998). Eph receptors recognize transmembrane or GPI-linked ephrin ligands, which have been implicated primarily in repulsion of growing axons, though they may have attractive properties as well (reviewed in Flanagan and Vanderhaeghen, 1998 and Holder and Klein, 1999). The VAB-1 Eph receptor is required for normal epithelial cell migrations in C. elegans (George et al., 1998), but its role in axon guidance has not been described.
We initiated a genetic analysis of axon guidance in the C. elegans nerve ring by focusing on a subset of nerve ring sensory neurons. The primary nematode sensory organ, the amphid, consists of twelve bilaterally symmetric amphid neuron pairs that function to detect chemicals, touch and temperature. Each amphid neuron has a single axon that extends in the nerve ring and a ciliated sensory dendrite that connects to an opening in the anterior cuticle (White et al., 1986). Most amphid axons reach the nerve ring after first growing ventrally in the amphid commissure, while one pair projects directly to the ring in a lateral position. The navigation of sensory axons in the nerve ring is largely completed by the end of embryogenesis. However, nerve ring axons continue to grow as the animal grows during the larval stages. During larval development, nerve ring axons increase 2.5-fold in length, while retaining the spatial organization established during embryogenesis. In the nerve ring, amphid sensory axons contact sensory and interneuron targets, establishing neural circuits essential to sensation and behavior (White et al., 1986). While little is known about the genes that direct amphid axon guidance, three cytoplasmic proteins are required for amphid axon extension: UNC-33/CRMP, UNC-44/ankyrin and UNC-76 (Hedgecock et al., 1985; Siddiqui and Culotti, 1991; McIntire et al., 1992; Otsuka et al., 1995; Bloom and Horvitz, 1997). In addition, the INA-1 α-integrin is required for amphid axon fasciculation (Baum and Garriga, 1997). Axon growth and maintenance in the larval stages may have requirements that are distinct from initial axon development. For some amphid sensory neurons, maintenance of axon morphology is regulated by neuronal electrical activity (Coburn and Bargmann, 1996; Coburn et al., 1998; Peckol et al., 1999).
To obtain information about the mechanisms of axon extension, guidance and maintenance in the C. elegans nerve ring, we conducted screens for mutants with altered trajectories of nerve ring sensory axons. We isolated mutations in 12 genes, including 8 new genes. Some mutations disrupted the initial outgrowth and guidance of axons in the nerve ring, as well as in other locations, while other mutations disrupted later aspects of nerve ring maintenance. In addition, we found that the SAX-3/Robo, VAB-1/Eph receptor and UNC-6/netrin, UNC-40/DCC guidance systems function together to guide sensory axons ventrally in the amphid commissure.
MATERIALS AND METHODS
Strain maintenance
Animals were maintained using standard methods (Brenner, 1974). Unless noted, sax-3(ky123, ky198, ky203), sax-5, sax-7 and sax-8 were grown at 20°C, and sax-1, sax-2, sax-3(ky200ts), sax-6, sax-9 and sax(ky213) were grown at 25°C, where their defects were more severe. All sax-3 alleles showed decreased viability at 25°C.
Strain construction
Details of plasmid construction are available upon request. Briefly, the ceh-23::gfp transgene contains a promoter and coding region fragment of the ceh-23 gene (Wang et al., 1993) fused to a peptide from unc-76 (Bloom and Horvitz, 1997) followed by the gfp coding region (Chalfie et al., 1994). The ceh-23::gfp construct was introduced into a lin-15(n765ts) mutant strain with the lin-15 plasmid pJM23 (Huang et al., 1994) by germline transformation, and stable transgenic strains were generated by gamma irradiation as described (Mello and Fire, 1995) to generate kyIs4. kyIs4 was outcrossed six times and mapped to LGXR. In kyIs4, GFP expression was consistently observed in the following neuron pairs: ADF, ADL, AFD, ASE, ASG, ASH, AWC and occasionally ASI in the amphid, BAG and AIY in the head, CAN in the central body, and PHA and PHB in the tail. kyIs4 animals exhibited wild-type amphid neuron morphology (White et al., 1986) and normal chemotaxis responses to the odorants benzaldehyde, 2-butanone and isoamyl alcohol, indicating that the AWC neurons were functional (Bargmann et al., 1993).
The mec-7::sax-3 transgene contains a partial sax-3 cDNA (nucleotides 1-1343) followed by the sax-3 genomic region containing the rest of the sax-3 open reading frame and part of the sax-3 3′ UTR cloned into the ppD96.41 mec-7 promoter vector (A. Fire and S. Xu, personal communication). The mec-7::sax-3 construct was injected at 10 ng/μl into sax-3(ky123) with 50 ng/μl of str-1::gfp as a coinjection marker.
Visualization of neuronal morphology
For DiI staining of amphid neurons, adult animals were exposed to a 15 μg/ml solution of DiI in M9 buffer for 1.5 hours. DiO staining was conducted as previously described (Coburn and Bargmann, 1996). These techniques label the ADL, ASH, ASI, ASJ, ASK and AWB amphid neuron pairs. To visualize other neuron types, we used the str-3::gfp transgene kyIs128 X (ASI sensory neurons in the head; E. Troemel and C. I. B., unpublished results) the glr-1::gfp transgene kyIs29 X (interneurons and motor neurons in the nerve ring and ventral nerve cord; Maricq et al., 1995) and the mec-4::gfp transgene bzIs7 IV (touch sensory neurons in the body; E. Wu and M. Driscoll, personal communication). The HSN neurons in the central body were visualized with antibodies to serotonin (Desai et al., 1988) in mutant strains that did not contain a gfp transgene.
Fluorescent animals were examined using a Zeiss Axioplan microscope and images assembled using Adobe Photoshop. Confocal images in Fig. 2 were generated from optical sections collected on a Zeiss LSM410 Invert confocal microscope and three-dimensional reconstructions were created using NIH Image 1.61/ppc.
Chemotaxis enrichment screen
kyIs4 animals were mutagenized with EMS (Brenner, 1974). Mutagenized animals were grown at 25°C for two generations and their F2 progeny assayed for chemotaxis to a point source of 5 nl benzaldehyde (Bargmann et al., 1993). After 1 hour, animals at the benzaldehyde source were removed and the remaining animals scored by fluorescence. sax-3, sax-5, sax-7, sax-8 and unc-44(ky110, ky115, ky116, ky186, ky256) alleles were obtained in the chemotaxis enrichment screen. Mutants were backcrossed three times by kyIs4 and once by N2 to separate the sax mutation from the ceh-23::gfp transgene. DiO staining detected similar amphid phenotypes in mutants with or without ceh-23::gfp.
daf-11 suppression screen
daf-11(m47) V; kyIs4 X and daf-11(m84) V; kyIs4 X strains were mutagenized with EMS and grown for one generation at 15°C. F1 progeny were shifted to 25°C as L4 larvae and F2 nondauers were cloned and their progeny scored for axon defects by fluorescence. sax-1, sax-2, sax-9, unc-33, unc-44(ky257) and unc-76 alleles were isolated as suppressors of daf-11(m47). sax-6 and sax(ky213) were isolated as suppressors of daf-11(m84). Mutations were not retested for suppression of daf-11. Mutants were outcrossed by a him-5(e1490) V; kyIs4 X strain to replace daf-11 with the closely linked him-5 mutation. Mutant him-5 strains were outcrossed twice by kyIs4 to separate the sax mutation from him-5 and once by N2 to separate the sax mutation from the ceh-23::gfp transgene. unc-76(ky258) was not separated from the daf-11 mutation.
Mapping
All sax mutations were recessive. Their genomic locations were determined by mapping with respect to recessive and dominant genetic markers. Details of mapping are available upon request. Mutations were followed using the amphid axon phenotype detected by ceh-23::gfp or DiO staining.
sax-1 mapped to X under the mnDp57 duplication. sax-2 mapped to III to the left or close to dpy-17 (0/4 dpy-17 non unc-32 recombinants and 15/15 unc-32 non dpy-17 recombinants were mutant). sax-3(ky123) mapping is described in Zallen et al. (1998). sax-5 mapped to IV between unc-26 and dpy-4 (13/19 dpy-4 non unc-26 recombinants were mutant). sax-6 mapped to I (5/7 non bli-3 unc-54 isolates were mutant and 2/7 were heterozygous). sax-7 mapped to IV to the left or close to unc-24 (0/7 unc-24 non dpy-20 recombinants and 1/1 dpy-20 non unc-24 recombinants were mutant). sax-8(ky201) mapped to III between dpy-17 and unc-32 (1/4 dpy-17 non unc-32 recombinants and 4/6 unc-32 non dpy-17 recombinants were mutant). sax-9(ky218) mapped to IV to the left or close to unc-26 (0/11 unc-26 non dpy-4 recombinants were mutant). kyIs4 mapped to XR to the left of unc-3 (1/12 unc-3 non lon-2 recombinants had kyIs4).
RESULTS
Identification of mutants with sensory axon defects
To visualize nerve ring axons in living animals, we used the ceh-23 cell-specific promoter (Wang et al., 1993) to express the green fluorescent protein (GFP). The ceh-23::gfp fusion was expressed in nine pairs of neurons in the head, including seven amphid neuron pairs (Fig. 1A,B), the BAG sensory neurons and the AIY interneurons, as well as the CAN neuron pair in the central body (Fig. 4A). The ceh-23::gfp transgene did not appear to disrupt the position, morphology or function of these neurons (Materials and Methods).
Behavioral and developmental assays that require amphid sensory function were used to enrich for mutations that disrupt amphid neuron development (Materials and Methods). In one screen, mutagenized animals were tested for chemotaxis to an odorant sensed by the AWC amphid neurons to identify chemotaxis-defective and movement-defective mutants. These animals were then screened visually for altered morphology of ceh-23::gfp-labeled neurons. A second screen made use of daf-11 mutants, which are growth arrested in the dauer larval stage (Riddle et al., 1981). Dauer formation occurs in response to sensory stimuli, and disruption of amphid neuron function suppresses the constitutive dauer arrest of daf-11 mutants (Vowels and Thomas, 1992; Schackwitz et al., 1996; Coburn et al., 1998). Mutagenized daf-11 animals were screened for suppressed mutants that relieved the daf-11 arrest and reached adulthood. These mutants were then examined visually for defects in axon morphology.
We screened 27,000 genomes using the chemotaxis enrichment strategy and isolated 15 mutations in six genes, determined by map position and complementation testing. We screened 6,000 genomes in the daf-11 suppression screen and isolated 9 mutations in at least seven genes. These screens identified different sets of genes, except for one gene isolated in both screens (Table 1). The new genes were designated sax genes (for sensory axon defects). These genes defined three main categories: (1) one gene required for nerve ring axon guidance (sax-3), (2) three genes required for both nerve ring axon extension and guidance (sax-5, sax-9 and unc-44), and (3) five genes where initial development proceeded normally, but defects occurred in the subsequent maintenance of nerve ring morphology (sax-1, sax-2, sax-6, sax-7 and sax-8).
sax-3/robo mutations disrupt multiple aspects of amphid axon guidance
Mutations in the sax-3/robo gene, a transmembrane protein that may function as a guidance receptor, caused amphid axons to extend inappropriately into the region anterior of the nerve ring (Fig. 1E,H; Table 2). As few as one or as many as all ceh-23::gfp-labeled axons extended anteriorly in an individual mutant animal. When all seven labeled axon pairs were anteriorly misrouted, they were often able to complete the dorsal component of their trajectories, forming an anteriorly displaced structure that resembled the nerve ring. All amphid neurons labeled by the ceh-23::gfp transgene or the fluorescent dyes DiO and DiI exhibited defects in sax-3 mutants, indicating that mutations in sax-3 disrupted the pathfinding of at least 11 of the 12 amphid neuron pairs. In addition, nerve ring axons from eight classes of interneurons and three classes of motor neurons labeled by the glr-1::gfp transgene (Maricq et al., 1995) also exhibited anterior axons (97% defective in sax-3(k123), n=31), indicating that sax-3 is likely to affect the position of many or all nerve ring axons.
The severity of the amphid axon defects prompted us to examine the trajectories of individual axons in sax-3 mutants. The str-3::gfp transgene specifically labels the two ASI amphid neurons (E. Troemel and C. I. B., unpublished data), each of which projects a single axon ventrally to the ventral midline in the amphid commissure and then dorsally within the nerve ring (Fig. 2A). In sax-3 mutants, the ASI axons often deviated anteriorly from their normal trajectory in the nerve ring (Fig. 2B; Table 3) or failed to grow ventrally in the amphid commissure and instead projected directly to the nerve ring in a lateral position. These laterally misrouted axons either continued dorsally in the nerve ring or terminated in a lateral position, perhaps because of a failure to encounter their normal substrata for nerve ring entry. A similar defect in the ventral guidance of six amphid neuron pairs was detected by DiI labeling (Fig. 5). These results indicate that sax-3 is required for two distinct guidance decisions of amphid sensory axons: first to guide amphid commissure axons ventrally to the nerve ring entry point and then to prevent axons in the nerve ring from wandering anteriorly.
sax-3 mutations also disrupted the cell migrations of some neuron types. Amphid neurons are born at the anterior tip of the animal and undergo short-range posterior migrations (Sulston et al., 1983). Amphid neurons were occasionally displaced anteriorly in sax-3 mutants, consistent with a failure to complete their normal migrations (Fig. 1I, 22% of amphid neurons defective in sax-3(ky123) n=55). In addition, sax-3 mutants exhibited defects in two long-range cell migrations, the posterior migration of the CAN neurons (Fig. 4B; Table 5) and the anterior migration of the HSN motor neurons (7% defective in sax-3(ky123), n=27). These phenotypes indicate that sax-3 is required for cell migrations in both directions along the longitudinal axis.
sax-3 mutants also exhibited a number of other phenotypes. Defects in head morphology produced a notched head phenotype (Fig. 1I, 38% defective in sax-3(ky123) n=71). sax-3 mutants exhibited a high incidence of embryonic lethality, with 82% of laid eggs (n=150) failing to hatch in the strong sax-3(ky123) allele (Zallen et al., 1998). The notched head phenotype and lethality are characteristic of mutants with defects in epithelial cell migration and adhesion (George et al., 1998). Consistent with the observation that sax-3 mutations cause widespread neuronal defects, mutant animals were defective in a number of behaviors, including chemotaxis, locomotion and egg-laying.
sax-5, sax-9 and unc-44 mutations disrupt both extension and guidance of amphid axons
Eleven mutations representing six genes caused amphid sensory axons to terminate prematurely before completing their trajectories in the nerve ring (Figs 1F, 2C; Tables 2, 3). In these mutants, most axons arrived at the nerve ring as in wild type, but terminated within the ring before reaching the dorsal midline. These mutants include three previously identified genes, unc-33, unc-44 and unc-76, which are required for axon outgrowth in several neuron classes, including amphid neurons (Hedgecock et al., 1985; Desai et al., 1988; Siddiqui, 1990; Siddiqui and Culotti, 1991; McIntire et al., 1992). In addition, a gene originally designated sax-4(ky112) was found to be allelic to the vab-3 gene (J. Hao, E. Lundquist, J. A. Z. and C. I. B., unpublished data). vab-3 encodes a Pax family transcription factor (Chisholm and Horvitz, 1995) and characterization of vab-3 alleles revealed severe defects in amphid axon extension (Fig. 1F, 63% defective in vab-3(e648), n=99 by DiI labeling).
Mutations in sax-5, sax-9 and unc-44 caused defects in amphid axon guidance as well as extension (Table 2). In sax-5 and unc-44 mutants, nearly all axons labeled by ceh-23::gfp, DiO and str-3::gfp were defective, suggesting that these mutations affect at least 11 of the 12 amphid neuron pairs. Both sax-5 and unc-44 mutants retained a ring of motor neuron and interneuron axons that express glr-1::gfp around the circumference of the nerve ring, excluding the possibility that these mutations cause a complete loss of the nerve ring. Mutations in sax-9 caused lower penetrance defects in amphid axon extension and guidance (Table 2).
Mutations in sax-5, sax-9 and unc-44 also caused misrouting of the HSN motor axon (Table 4), consistent with a role for these genes in axon pathfinding. In wild-type animals, the two HSN neurons in the central body region each project a single axon ventrally to the ventral midline and then anteriorly to the head (Fig. 3A). In sax-5, sax-9 and unc-44 mutants, the HSN axon often grew in a lateral position or wandered and branched extensively in the vicinity of the cell body (Fig. 3B). In some animals, the HSN axon grew posteriorly instead of anteriorly in the ventral cord, and when the axon did grow anteriorly as in wild type, it often terminated prematurely. Consistent with previous reports, unc-44 mutants also exhibited defects in HSN axon outgrowth, leading to premature termination (Table 4; McIntire et al., 1992). Axon termination was also observed in the CAN neurons in the central body (Table 5). In addition, unc-44 mutants exhibited ectopic HSN axons, with up to four axons from a single neuron, suggesting a role for unc-44 in axon initiation (Fig. 3C; Table 4). Ectopic axons were also observed in unc-33(e204) mutants (data not shown).
sax-5 and sax-9 mutations also disrupt axon guidance in the ventral nerve cord, the second largest axon bundle in C. elegans. The ventral nerve cord is normally asymmetric; over 30 axons travel on the right side of the ventral midline, while only 4-6 axons travel on the left side (White et al., 1976). The decision to join one side of the ventral cord is made upon arrival at the ventral midline, and axons in the body region do not cross the midline subsequent to this choice. When the HSN axon arrives at the ventral midline, it joins the ipsilateral cord and travels anteriorly to the head. While wild-type HSN axons travel on opposite sides of the ventral cord (Fig. 3D), axons often traveled together on the same side in sax-5 and sax-9 mutants (Fig. 3E; Table 4). Similarly, eleven glr-1-expressing interneuron axons that normally extend in the right ventral cord occasionally grow on the left side in sax-5 mutants (Table 4). Mutations in both sax-5 and sax-9 also disrupted the posteriorly directed CAN cell migrations (Fig. 4C; Table 5), while only mutations in sax-9 disrupted the anteriorly directed HSN migration (7% defective, n=110). Cell migration defects were not observed in unc-33, unc-44 or unc-76 mutants.
sax-1, sax-2 and sax-6 mutations disrupt the maintenance of neuronal morphology
Mutations in the sax-1, sax-2 and sax-6 genes caused amphid neurons to send aberrant processes into the region posterior to the nerve ring (Fig. 1D). The ceh-23::gfp marker detected one to several aberrant posteriorly directed processes per amphid, in addition to a set of apparently normal axons in the nerve ring. Defects were observed both with the ceh-23::gfp transgene and with DiO and DiI filling, where some aberrant processes were traceable to the ASJ neuron. sax-1, sax-2 and sax-6 were identified as suppressors of daf-11 dauer constitutivity, an assay that can detect subtle sensory defects. Laser ablation of the ASJ neuron pair suppresses the dauer arrest of daf-11 mutant animals (Schackwitz et al., 1996).
Therefore, it is possible that the ASJ axon defects in these mutants altered ASJ function. To characterize the behavior of individual axons in sax-1 and sax-2 mutants, we examined the ASI neuron pair. In all cases, the posteriorly directed processes were found to be ectopic neurites, since each ASI neuron also extended an apparently normal axon in the nerve ring (Fig. 2D; Table 3).
To determine whether the phenotypes in sax-1 and sax-2 mutants reflect defects in the establishment or maintenance of neuronal morphology, we characterized the morphology of mutant neurons throughout development. While the outgrowth of amphid axons is completed by the end of embryogenesis, few ectopic neurites were detected by the ceh-23::gfp transgene in first-stage (L1) larval sax-1 and sax-2 mutant animals, and aberrant neurites increased in penetrance during the later larval stages (data not shown). These results are consistent with a role for sax-1 and sax-2 in the maintenance of neuronal morphology during larval growth.
The sax-1 and sax-2 mutant phenotypes were not specific to amphid sensory neurons, since aberrant ectopic neurites were also observed in the CAN neurons (29% in sax-1, n=121; 53% in sax-2, n=36) and glr-1::gfp-labeled inter-and motor neurons (46% in sax-1, n=54; 39% in sax-2, n=31). However, the cell migrations and guidance of the primary axons of amphid, CAN, HSN and glr-1::gfp-labeled neurons were normal. Consistent with the observation that these aspects of neuronal morphology are preserved in these mutants, mutant animals performed normally in a range of behavioral assays, including locomotion, egg-laying and osmotic avoidance (data not shown).
sax-7 and sax-8 mutations disrupt the maintenance of nerve ring placement
In sax-7 and sax-8 mutants, the nerve ring contained an apparently normal complement of axons, and initial nerve ring morphology was wild type in postembryonic first-stage (L1) larvae. However, as mutant animals progressed through the four larval stages to adulthood, nerve ring axons became displaced posteriorly relative to the correctly positioned cell bodies (Fig. 1G; Table 2). glr-1::gfp-expressing axons of sax-7 mutants were also posteriorly displaced (data not shown), indicating a general displacement of many or all nerve ring axons. However, no defects were observed in the ventral guidance of amphid axons in the amphid commissure (Fig. 1G; Table 2), HSN motor axons or CAN axons. These phenotypes suggest a specific role for the sax-7 and sax-8 genes in the maintenance of nerve ring placement.
Mutations in multiple guidance pathways disrupt ventral guidance in the amphid commissure
To supplement genetic screens for new genes involved in C. elegans nerve ring development, we investigated whether mutations in known genes disrupted the trajectories of amphid sensory axons. The UNC-6/netrin secreted axon guidance cue and its receptor, UNC-40/DCC, are required for the guidance of many C. elegans axons along the dorsoventral axis (Hedgecock et al., 1990; Ishii et al., 1992; Chan et al., 1996). We found that, in unc-6 and unc-40 mutants, the ASI amphid axon sometimes failed to grow ventrally in the amphid commissure and instead traveled directly to the nerve ring in a lateral position (Fig. 2E; Table 3). In addition, other amphid axons labeled by DiI filling also exhibited ventral guidance defects (Fig. 5), indicating that, like sax-3/robo, unc-6 and unc-40 mutations disrupt the ventral guidance of multiple amphid neuron types. In mutant animals where ventral guidance occurred normally, amphid axons sometimes terminated prematurely while traveling dorsally in the nerve ring (Table 3; 17% defective in unc-6(ev400) n=240, 11% defective in unc-40(e271) n=332 by DiI filling). UNC-5, a transmembrane receptor that mediates repulsion from UNC-6 in combination with UNC-40 (Hedgecock et al., 1990; Leung-Hagesteijn et al., 1992; Hamelin et al., 1993), is not required for amphid axon guidance (Table 3; 1% defective in ventral guidance in unc-5(e53) by DiI filling, n=168).
Ventral guidance in the amphid commissure was also disrupted in animals mutant for VAB-1, a C. elegans Eph receptor that is expressed in neurons that project axons to the nerve ring (George et al., 1998). DiI filling revealed a partially penetrant defect in which many amphid axons grew directly to the nerve ring in a lateral position, a defect qualitatively similar to the defects of unc-6 and unc-40 mutants (Fig. 5). vab-1 mutants had a strong lateral axon phenotype in the ASI neuron, but no other ASI defects (Table 3). Amphid phenotypes were most severe in animals with the vab-1(dx31) deletion that removes the first four exons, but were also present in animals with the vab-1(e2) missense mutation that abolishes kinase activity (George et al., 1998). vab-1 mutations caused minimal defects in ventral guidance of the HSN axons (Table 4).
Like SAX-3/Robo, the VAB-1/Eph receptor also affects axon guidance at the C. elegans ventral midline. In half of vab-1(e2) and vab-1(dx31) mutant animals, the HSN axon aberrantly crossed and recrossed the ventral midline (Table 4). Likewise, glr-1-expressing interneuron axons sometimes failed to remain in the right ventral cord and crossed over to the left side (Fig. 3G; Table 4).
SAX-3/Robo functions in parallel to VAB-1 and UNC-6/UNC-40 in the ventral guidance of amphid axons
SAX-3, UNC-6, UNC-40 and VAB-1 are all required for normal amphid axon trajectories; how do these molecules function together to guide axons ventrally in the amphid commissure? If these genes act in a single pathway, disrupting two genes should not cause a more severe defect than a complete loss of function in either one. Alternatively, if these genes act in parallel, then disrupting two genes together may cause a defect that is more severe than the defect in either single mutant.
The vab-1(e2); sax-3(ky200) double mutant was significantly more defective than animals lacking either vab-1 or sax-3 function (Fig. 5A). Similarly, double mutant combinations between sax-3 and either unc-6 or unc-40 were significantly more defective than animals with strong mutations in either sax-3, unc-6 or unc-40 (Fig. 5B). By contrast, unc-40; unc-6 double mutants were not enhanced compared to unc-6 single mutants (Fig. 5C), consistent with evidence that UNC-6 and UNC-40 participate in a single guidance pathway (Hedgecock et al., 1990). These results establish that the guidance functions of SAX-3/Robo do not absolutely require the VAB-1 Eph receptor or the UNC-6/netrin, UNC-40/DCC pathway.
The ventral guidance of amphid axons was significantly more defective in vab-1(e2); unc-6(ev400) or unc-40(e1430); vab-1(e2) double mutants than in unc-6 or unc-40 single mutants (Fig. 5C), indicating that the VAB-1 Eph receptor does not require UNC-6 or UNC-40 to execute guidance functions. However, the defects in double mutants with the weak vab-1 allele did not exceed those caused by the strong vab-1(dx31) allele alone, so UNC-6 could participate in either a subset of VAB-1 activities or in a separate pathway. Unfortunately, the lethality of double mutants precluded characterization of genetic combinations with strong loss-of-function sax-3(ky123) or vab-1(dx31) mutations.
sax-3/robo functions cell autonomously in ventral guidance of the AVM sensory axon
The requirement for SAX-3/Robo in ventral guidance is consistent with two models: SAX-3 could act cell autonomously as a receptor on growing axons or non-autonomously, either as a receptor in another pioneer axon or as a ligand on the substratum. Since amphid axons grow out in the amphid commissure bundle, their ventral guidance may be achieved through the combined action of axon-axon and axon-substratum interactions. Like amphid neurons, the AVM mechanosensory neuron has a cell body that is located on the lateral hypodermis and an axon that grows ventrally to the ventral midline; however, the AVM axon travels independently using only axon-substratum interactions. For this reason, we chose to examine the question of sax-3 autonomy in the AVM neuron. AVM, like amphid neurons, relies on sax-3 for its ventral axon guidance (Fig. 6B,D) as well as unc-6 and unc-40 (Siddiqui, 1990; Chan et al., 1996).
To determine the site of sax-3 action for AVM ventral guidance, we expressed the sax-3 cDNA from the mec-7 promoter, which drives expression in six mechanosensory neurons, including AVM (Hamelin et al., 1993). The mec-7::sax-3 fusion rescued the AVM defects of sax-3 mutants in two independent transgenic lines (Fig. 6C,D), indicating that SAX-3 can function cell autonomously in AVM ventral guidance. This result does not rule out the possibility that SAX-3 could behave non-autonomously in other cell contexts, such as the nerve ring.
DISCUSSION
Identification of genes required for normal sensory axon trajectories
Mutations in eight sax genes disrupt the trajectories of sensory axons in the C. elegans nerve ring. sax-3/robo is required for two distinct guidance decisions in the amphid commissure and the nerve ring, while sax-5, sax-9 and unc-44 are required for nerve ring axon extension and guidance. Once amphid axons have completed their initial trajectories in the nerve ring, sax-1, sax-2, sax-6, sax-7 and sax-8 are required for the maintenance of nerve ring structure. Mutations in the sax-3, sax-5, sax-9 and unc-44 genes disrupt axon guidance or cell migration in other neuron types, indicating that their function is not limited to the nerve ring. Most of the sax mutants were not defective for locomotion, and therefore would not have been isolated in previous screens for axon guidance mutants in a collection of behaviorally uncoordinated mutants.
sax-3, vab-1 and unc-6/unc-40 function in parallel for ventral axon guidance
The SAX-3/Robo, VAB-1/Eph receptor and UNC-6/netrin, UNC-40/DCC guidance systems are required to guide amphid sensory axons ventrally in the amphid commissure, but mutations in these genes cause relatively mild ventral guidance defects individually. Double mutants are more severe, suggesting that the SAX-3/Robo, VAB-1/Eph receptor and UNC-6/netrin, UNC-40/DCC pathways can function independently in ventral guidance, although these molecules may also share some functions. While the sequence and expression of SAX-3, UNC-40 and VAB-1 suggest that they could act as guidance receptors in nerve ring axons, cell autonomy studies will be required to determine the roles of these guidance molecules in the complex environment of the nerve ring.
The SAX-3/Robo immunoglobulin superfamily member functions cell autonomously in the AVM sensory neuron, consistent with a role as a receptor for ventral guidance. sax-3 is expressed in amphid neurons as well as epidermal substratum cells during the time of amphid axon outgrowth in the embryo (Zallen et al., 1998). SAX-3 may act as a receptor for ventral guidance of amphid sensory neurons; alternatively, SAX-3 could function non-autonomously as a receptor in a nerve ring pioneer axon, as a ligand in an axon-substratum interaction, or in an earlier developmental process (see below). The UNC-40 guidance receptor is widely expressed in neurons (Chan et al., 1996). In the region of the nerve ring, its ligand UNC-6 is present in four neurons and two ventral CEP sheath cells that surround the amphid commissure, but is absent from the two dorsal CEP sheath cells (Wadsworth et al., 1996). This expression pattern suggests a model in which UNC-6 secreted by the ventral CEP sheath cells may attract axons from the amphid commissure that express the UNC-40 receptor. In this model, UNC-6 could act either as a diffusible chemoattractant or as a local substratum to direct axon growth.
The VAB-1 Eph receptor is expressed in many head neurons but not in their epidermal substratum cells. In these neurons, its first developmental function is to direct the epidermal cell migrations that take place during ventral enclosure of the embryo (George et al., 1998). vab-1 and sax-3 mutants share some epidermal abnormalities, including a partially penetrant notched head phenotype. These defects are associated with lethality that is enhanced in sax-3; vab-1, sax-3;unc-6/40 and vab-1;unc-6/40 double mutants, suggesting that SAX-3, UNC-6 and UNC-40 may also affect epidermal morphogenesis.
We show here that vab-1 mutations disrupt the ventral guidance of amphid axons. There are three possible explanations for this phenotype: VAB-1 and SAX-3 could mediate amphid axon guidance directly, amphid axon defects could arise indirectly as a consequence of abnormal epidermal migration, or altered interactions between neuroblasts in vab-1 mutants could change neuronal substrata for axon migration (George et al., 1998). We favor the possibility that some effects of SAX-3 and VAB-1 are due to their actions in neurons. First, individual axons often exhibited guidance defects in sax-3 and vab-1 mutants even when the remainder of the nerve ring appeared normal. Second, in both sax-3 and vab-1 mutants, animals were observed with axon defects but no visible head morphogenesis phenotypes. Third, ventral guidance defects and anterior axon misrouting were not present in other notched-head mutants, such as dpy-23 and ina-1 (J. A. Z. and C. I. B., unpublished results). These results do not support a model where head morphogenesis defects account for all axon defects in sax-3 and vab-1 mutants.
sax-3, sax-5, sax-9 and vab-1 affect axon crossover at the ventral midline
Both sax-3/robo and the vab-1 Eph receptor are required to prevent aberrant axon crossing in the C. elegans ventral midline (Zallen et al., 1998 and the present study), along with the new genes sax-5 and sax-9. Axons from the left ventral cord can travel aberrantly on the right side in the absence of left cord pioneers (Durbin, 1987). However, axons from both the left cord (HSNL) and the right cord (HSNR and glr-1::gfp-labeled interneurons) cross the midline inappropriately in these mutants, indicating that these defects do not merely reflect an absence of the left cord. Drosophila Robo has been shown to function as a receptor on axons that prevents inappropriate crossing at the midline (Kidd et al., 1998). In vertebrates and Drosophila, Robo proteins bind to Slit, a candidate ligand molecule that is secreted by midline cells (Brose et al., 1999; Kidd et al., 1999; Li et al., 1999), suggesting that Robo may detect a localized midline repellent.
The finding that mutations in the VAB-1/Eph receptor lead to midline crossover defects suggests the possibility that Eph receptors may participate in axon guidance at the midline. Interestingly, a transmembrane ephrin is expressed at the ventral midline of the developing neural tube in vertebrates (Bergemann et al., 1998). It is possible that the VAB-1/Eph receptor may interact with a conserved ephrin at the C. elegans ventral midline to prevent axons from crossing inappropriately.
sax-1, sax-2, sax-6, sax-7 and sax-8 function in nerve ring maintenance
The genes required for nerve ring maintenance are distinct from those involved in early axon outgrowth and guidance. Five sax genes are specifically required for the maintenance of axon morphology during larval stages. In sax-1 and sax-2 mutants, neurons initiate ectopic neurites that extend late in development. Although the pathfinding of embryonic axons proceeds normally, late-growing ectopic processes wander aberrantly into the region posterior to the nerve ring. These ectopic neurites may fail to grow into the nerve ring because molecules that direct early axon guidance, such as SAX-3, are downregulated postembryonically (Zallen et al., 1998); alternatively, ectopic neurites could be indifferent to conventional guidance systems.
Posterior sensory axons are also present in mutants with abnormal sensory activity. These include mutations in a cyclic nucleotide-gated sensory channel, mutations that affect the structure of the amphid sensory cilia, and mutations that alter ion channel function (Coburn and Bargmann, 1996; Coburn et al., 1998; Peckol et al., 1999). sax-1 and sax-2 might participate in an activity-dependent pathway for maintaining axon morphology, or in an alternative pathway.
The sax-7 and sax-8 genes identify a distinct mechanism of nerve ring maintenance. In these mutants, the nerve ring is initially correctly positioned relative to the cell bodies in first larval stage animals. However, by the second larval stage, these mutants begin to display an altered morphology, with the nerve ring increasingly posteriorly displaced relative to neuronal cell bodies. These mutants may provide information about the cell types and molecular pathways that maintain the position of the nerve ring bundle in the growing animal.
sax genes have distinct and overlapping roles in axon guidance, extension, initiation and cell migration
The diverse phenotypes of sax mutant animals may reflect a versatility of proteins involved in morphogenesis. Mutations in sax-5, sax-9 and unc-44 disrupt both axon guidance and extension. Similarly, netrins promote both outgrowth and guidance of vertebrate commissural axons (Kennedy et al., 1994; Serafini et al., 1994). This overlap between guidance and extension responses could occur if components of the basic cytoskeletal machinery for axon extension are targets for regulation by axon guidance pathways.
unc-44 and unc-33 mutants exhibit ectopic neurite defects in the HSN and PDE neurons, suggesting that these genes may regulate axon initiation in multiple neuron types (this work and Hedgecock et al., 1985). UNC-44 is a C. elegans ankyrin, a protein that links transmembrane receptors to the actin cytoskeleton (Otsuka et al., 1995) and UNC-33-related proteins may act downstream of receptors for the semaphorin guidance cue (Goshima et al., 1995). UNC-44 is required for the proper localization of UNC-33 (W. Li and J. Shaw, personal communication); perhaps the localization of UNC-33 or other factors directs axon initiation to a single site in wild-type animals.
Mutations in the sax-3, sax-5 and sax-9 genes affect cell migration as well as axon pathfinding. All three mutations selectively disrupt the cell migration of the CAN neurons without affecting CAN axon outgrowth. A reciprocal requirement for sax-5 and sax-9 was observed in sensory neurons: mutations in these genes disrupted amphid axon trajectories while leaving amphid cell migrations intact. These genes may participate in axon guidance in one cellular context and cell migration in another. A parallel can be found in vertebrates, where ephrin guidance molecules direct the diverse processes of retinal axon guidance, neural crest cell migration and development of the vasculature (reviewed in Flanagan and Vanderhaeghen, 1998 and Holder and Klein, 1999).
The mechanisms by which guidance cues influence axon behavior in diverse ways are not understood, but may reflect differences in ligand presentation, receptor expression or signal transduction. The characterization of multifunctional genes required for axon growth, guidance and maintenance may provide insight into the ways guidance systems operate in distinct cellular and molecular contexts.
ACKNOWLEDGEMENTS
We thank Monica Driscoll, Ed Wu, Andy Fire and Emily Troemel for gfp transgenes and promoters, and Joe Hao and Erik Lundquist for identifying ky112 as a vab-3 allele and generating the vab-1; unc-40 double mutant. We are also grateful to Tim Yu, Katja Brose, Zemer Gitai, Erin Peckol and Joe Hao for comments on the manuscript. Some strains were provided by the Caenorhabditis Genetics Center. This work was supported by the Howard Hughes Medical Institute and an NIH Postdoctoral Fellowship (to S. A. K.). J. A. Z. was a predoctoral fellow of the National Science Foundation and C. I. B. is an Assistant Investigator of the Howard Hughes Medical Institute.