Identification of Caenorhabditis elegans Genes Required for Neuronal Differentiation and Migration

Corresponding author: Wayne C. Forrester, Department of Molecular and Cell Biology, 401 Barker Hall, University of California, Berkeley, CA 94720-3204, forrestr{at}mendel.berkeley.edu (E-mail).

Communicating editor: I. GREENWALD

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

To understand the mechanisms that guide migrating cells, we have been studying the embryonic migrations of the C. elegans canal-associated neurons (CANs). Here, we describe two screens used to identify genes involved in CAN migration. First, we screened for mutants that died as clear larvae (Clr) or had withered tails (Wit), phenotypes displayed by animals lacking normal CAN function. Second, we screened directly for mutants with missing or misplaced CANs. We isolated and characterized 30 mutants that defined 14 genes necessary for CAN migration. We found that one of the genes, ceh-10, specifies CAN fate. ceh-10 had been defined molecularly as encoding a homeodomain protein expressed in the CANs. Mutations that reduce ceh-10 function result in Wit animals with CANs that are partially defective in their migrations. Mutations that eliminate ceh-10 function result in Clr animals with CANs that fail to migrate or express CEH-23, a CAN differentiation marker. Null mutants also fail to express CEH-10, suggesting that CEH-10 regulates its own expression. Finally, we found that ceh-10 is necessary for the differentiation of AIY and RMED, two additional cells that express CEH-10.

DURING nervous system development, neurons migrate, extend growth cones to their targets, synapse with other cells, and express specific neurotransmitters. To achieve its final pattern of connectivity, a neuron must express proteins that regulate cell migration, axonal outgrowth, synapse formation and neurotransmitter expression. We have been particularly interested in understanding the mechanisms that regulate cell migration and axonal outgrowth.

Cell migration and axonal outgrowth are complicated processes that are likely to require many proteins. Extracellular matrix molecules such as laminin are thought to provide a permissive environment for cell motility (LANDER 1989 ). Additional extracellular matrix molecules such as the UNC-6/netrins play instructive roles, guiding migrations along the dorsoventral axis of nematodes, flies and vertebrates (SERAFINI et al. 1994 , SERAFINI et al. 1996 ; HARRIS et al. 1996 ; MITCHELL et al. 1996 ; WADSWORTH et al. 1996 ). Signals from extracellular guidance cues are mediated by receptors. For example, the UNC-5 and UNC-40/DCC netrin receptors appear to repel and attract growth cones, respectively (LEUNG-HAGESTEIJN et al. 1992 ; CHAN et al. 1996 ; KEINO et al. 1996 ; KOLODZIEJ et al. 1996 ; ACKERMAN et al. 1997 ; FAZELI et al. 1997 ; LEONARDO et al. 1997 ). Signals are integrated by intracellular signaling molecules. Members of the Rho family of Ras-like small GTPases, CDC-42, Rac and Rho, are intracellular signaling molecules that apparently act on discrete downstream targets to control cell migrations (LUO and RAPER 1994 ; NOBES and HALL 1995 ). Ultimately, signaling regulates actin dynamics that drive cell motility (LAUFFENBURGER and HORWITZ 1996 ; MITCHISON and CRAMER 1996 ).

Migratory cells must regulate molecules involved in cell motility, such as cell-surface receptors and signal transduction molecules, to ensure that cells or growth cones migrate at the proper time, in the appropriate direction, and along the correct path. Genetic analysis indicates that some of this regulation occurs transcriptionally. One of the best studied examples of a tran-scriptional regulator that controls the migration of a cell is the Caenorhabditis elegans gene mab-5, which encodes a homeoprotein that acts cell autonomously to direct the migrations of the QL neuroblast and its descendants (CHALFIE et al. 1983 ; KENYON 1986 ; COSTA et al. 1988 ; SALSER and KENYON 1992 ). Normally, QL descendants migrate posteriorly, whereas analogous QR descendants located on the contralateral side of the animal migrate anteriorly (SULSTON and HORVITZ 1977 ). Loss of mab-5 function causes QL to migrate anteriorly like QR, whereas expressing mab-5 ectopically in the QR descendants causes them to migrate posteriorly (KENYON 1986 ; SALSER and KENYON 1992 ). These results indicate that mab-5 controls the expression of proteins that direct QL and its descendants posteriorly.

In addition to molecules like MAB-5 that act specifically to regulate a cell's migratory behavior, other transcriptional regulators act more globally to control additional aspects of a cell's differentiation. The C. elegans gene egl-5, for example, encodes a homeoprotein that acts cell autonomously to specify HSN fate (DESAI et al. 1988 ; CHISHOLM 1991 ; WANG et al. 1993 ). Loss of egl-5 function disrupts not only HSN migration, but later aspects of HSN differentiation, such as neurotransmitter synthesis.

To identify genes required for cell migration and differentiation, we screened for mutants with defects in the migrations of the C. elegans canal-associated neurons (CANs). The CANs are a pair of bilaterally symmetric neurons that are born in the head and migrate posteriorly to the middle of the animal during embryogenesis. The CANs extend axons anteriorly and posteriorly along the lateral body wall and express two homeodomain proteins, CEH-10 and CEH-23 (WHITE et al. 1986 ; WANG et al. 1993 ; SVENDSEN and MCGHEE 1995 ). Our screens identified mutations in 14 genes required for CAN cell migration, including the first ceh-10 mutations. Analysis of ceh-10 mutants revealed that CEH-10 specifies CAN fate. Reduction of ceh-10 function results in CANs that partially fail to migrate. Complete loss of ceh-10 function results in CANs that fail to migrate and do not express CEH-23 or CEH-10.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Strains and genetics:
Strains were grown at 20° and maintained as described (BRENNER 1974 ). In addition to the wild-type strain N2, strains with the following mutations were used in this work.

LGI: cam-2(gm124), dpy-5(e61), let-204(e1719), syc-3(gm135), unc-11(e47), unc-54(e1092), unc-73(gm33), unc-73(gm67), unc-73(gm123), unc-95(su33sd), eDf3.

LGII: cam-1(gm105), cam-1(gm122), rol-6(e187), unc-4(e120), mnDf16.

LGIII: ceh-10(gm58), ceh-10(gm71), ceh-10(gm100), ceh-10(gm120), ceh-10(gm127), ceh-10(gm131), ceh-10(gm133), daf-2(e1370), dpy-17 (e164), dpy-18(e364), fam-1(gm85), ina-1(gm144), lin-12(n676sd n909), sDf121, syc-1(gm126), syc-2(gm132), unc-32(e189), unc-69 (e587), yDf10.

LGIV: bli-6(sc16), dpy-20(e1282ts), epi-1(gm57), epi-1(gm121), epi-1(gm139), epi-1(gm146), epi-1(ky152), fam-2(gm94), kyIs5 [ceh-23-unc-76-gfp::lin-15], lin-1(e1275), sDf23, unc-5(e53), unc-24(e128), unc-30(e191).

LGV: unc-34(gm104), unc-34(gm114), unc-34(gm115), unc-34(gm134), unc-34(e566), vab-8(gm99), vab-8(gm138), vab-8(e1017).

LGX: kyIs4 [ceh-23-unc-76-gfp::lin-15], mig-2(gm38), mig-2-(gm103sd), mnDf1.

dpy-5(e61), dpy-17(e164), dpy-18(e364), unc-4(e120), unc-5(e53), unc-11(e47), unc-24(e128), unc-30(e191), unc-32(e189), unc-34(e566), and unc-69(e587) were described by BRENNER 1974 . dpy-20(e1282ts) was described by HOSONO et al. 1982 , bli-6(sc16) by PARK and HORVITZ 1986 , epi-1(ky152) by S. CLARK and C. I. BARGMANN (personal communication), let-204(e1719) by ANDERSON and BRENNER 1984 , lin-12(n676sd n909) by GREENWALD et al. 1983 , unc-54(e1092) by WATERSTON et al. 1980 , unc-73(gm33) by D. PARRY, P. BAUM and G. GARRIGA (personal communication), unc-95(su33sd) by ZENGEL and EPSTEIN 1980 and vab-8(e1017) by MANSER and WOOD 1990 . The chromosomal rearrangement eDf3 was described by ANDERSON and BRENNER 1984 , mnDf1 by MENEELY and HERMAN 1979 , mnDf16 by SIGURDSON et al. 1984 , yDf10 by DELONG et al. 1993 . sDf23 was isolated by D. BAILLIE (personal communication), and sDf121 by H. STEWART, D. COLLINS and D. BAILLIE (personal communication). The LGIII balancer qC1 was described by AUSTIN and KIMBLE 1989 . The ceh-23-gfp transgene will be described elsewhere (J. A. ZALLEN and C. I. BARGMANN, unpublished results). Briefly, the GFP coding region was fused to the ceh-23 5' regulatory region and coding sequence up to amino acid 25 of the homeodomain (WANG et al. 1993 ). Amino acids 1–197 of the Unc-76 protein (BLOOM and HORVITZ 1997 ) were included to enhance staining of neuronal processes. The ceh-23-gfp transgene was injected with the lin-15 plasmid pJM23 (HUANG et al. 1994 ) as a coinjection marker. kyIs5 is a ceh-23-gfp reporter transgene that is integrated on LGIV, and kyIs4 is the same reporter integrated on LGX (J. A. ZALLEN and C. I. BARGMANN, personal communication). pInt1[ceh-10-lacZ::rol-6] and pEx9[ceh-10::rol-6] were described in SVENDSEN and MCGHEE 1995 .

Because most of the mutants were isolated in a strain containing the kyIs5 reporter, we removed this transgene from the mutants by crossing to wild type. The epi-1 alleles gm121, gm139 and gm146, and the fam-2(gm94) allele are tightly linked to kyIs5 and have not been separated from the reporter.

Most phenotypes were tabulated from homozygous mutant animals derived from homozygous parents. However, we could not score ceh-10(gm58), unc-73(gm67), unc-73(gm123) and epi-1(gm139) mutant progeny of homozygous parents. Because ceh-10(gm58) homozygotes die as larvae, we examined the phenotypes of homozygous ceh-10 animals from ceh-10(gm58)/qC1 hermaphrodites. Because self-progeny of parents homozygous for unc-73(gm67) and unc-73(gm123) rarely survive to adulthood, we examined phenotypes of homozygous unc-73 animals from unc-73 +/+ dpy-5(e61) hermaphrodites. Similarly, because epi-1(gm139) homozygotes are sterile, we examined phenotypes of homozygous epi-1(gm139) animals from + epi-1(gm139) + /dpy-20(e1282ts) + unc-30(e191) hermaphrodites.

Screens:
Wild-type animals were mutagenized with EMS as described (BRENNER 1974 ). Mutagenized hermaphrodites were cultured individually and allowed to self-fertilize. Three F₁ progeny of each parent were cultured individually, allowed to self-fertilize, and screened for the presence of withered tails (Wit) or clear larvae (Clr) offspring. Lethal or sterile mutants were maintained by selecting heterozygous siblings. Genes were mapped as described in Table 1. Once the genes were mapped, lethal or sterile mutations were balanced by generating animals heterozygous for the mutation and for a chromosome bearing mutations that result in visible phenotypes. CAN cells in candidate Cam mutants were examined by crossing kyIs4 or kyIs5 GFP reporter transgenes into the mutant strains and determining CAN cell position by fluorescence microscopy. We examined the progeny of approximately 1000 mutagenized F₁ parents and identified two Wit mutants with CAN migration defects [epi-1(gm57) and unc-73(gm67)] and two Clr mutants that lacked differentiated CANs [ceh-10(gm58 and gm71)].

In the second screen, animals bearing the kyIs5 [ceh-23-unc-76-gfp::lin-15] transgene were mutagenized and cultivated as described for the first screen (BRENNER 1974 ). From each F₁, approximately 20 first larval stage F₂ progeny were transferred to 5% agar pads on microscope slides. To determine CAN cell position by GFP expression, animals were examined on a Nikon Labophot microscope equipped with epifluorescence. Animals were illuminated using a Nikon BV-2 filter set and a 40x dry objective. F₁s that segregated 25% or more progeny with misplaced or missing CANs were characterized further. Lethal or sterile mutants were maintained as described for the first screen.

All mutants were outcrossed to wild type at least three times to remove unlinked mutations. The kyIs5 GFP reporter was also removed from all strains identified in the GFP screen by mating to wild type and reisolating the mutation in the absence of kyIs5 (except gm121, gm139, gm146 and gm94, which are all tightly linked to kyIs5). To determine CAN position, the kyIs5 reporter was crossed into all mutants except gm57, which maps near the position of kyIs5 integration, using standard genetic methods. CAN cell position was determined in gm57 mutants by crossing in kyIs4.

Complementation tests:
Mutations that mapped near one another, or near other mutations known to disrupt cell migration, were tested for complementation. This general strategy is illustrated here with gm121, a mutation identified in the GFP screen that maps near epi-1, a gene required for the migrations of several cells (FORRESTER and GARRIGA 1997 ). Heterozygous gm121 males were crossed into epi-1(ky152) unc-24(e128) (kindly provided by S. CLARK and C. BARGMANN). Cross-progeny were identified because they no longer displayed the Unc-24 phenotype. Half of these cross-progeny resembled epi-1 mutants. In addition, these animals had misplaced CAN cells. Thus, based on mapping and complementation, gm121 is an epi-1 allele.

We have not been able to eliminate the possibility that gm135 is an unusual allele of unc-73. We have mapped gm135 and unc-73 by placing the mutations in trans to unc-11(e47) dpy-5(e61) and assessing whether recombinants in this interval carry the syc-3 or unc-73 mutation. We find that zero of eight Unc non-Dpy and nine of 10 Dpy non-Unc recombinants in this interval carry the gm135 mutation and one of nine Unc non-Dpy and 19 of 24 Dpy non-Unc carry the unc-73 mutation. gm135 animals exhibit a "loopy" Unc phenotype in which animals move with exaggerated body curvature, whereas unc-73(gm33) animals rarely move. Although gm135/unc-73(gm33) animals are generally not Unc (rare animals are loopy Unc), their CANs are misplaced as often as in gm135 homozygotes, indicating that gm135 complements unc-73 for its movement defects, but not for its CAN migration defects (not shown). gm135/qDf3 hemizygotes are loopy Unc, similar to gm135 mutants alone, and have misplaced CANs. unc-73(gm33)/qDf3 hemizygotes are occasionally loopy Unc and have misplaced CANs. The simplest interpretation of these data is that gm135 defines a new gene, syc-3, which is uncovered by qDf3. However, we cannot rule out the alternative model in which syc-3 is an unusual allele of unc-73 and that qDf3 deletes sequences that would complement the CAN cell migration defect of unc-73 mutants.

Phenotypic characterization:
To quantitate phenotypic defects of mutant strains, homozygous viable lines were cultivated at 20°. Individual adult animals were examined for Wit, uncoordinated movement (Unc), egg-laying defective (Egl) and multiple vulvae (Muv) phenotypes, and individual early-stage larvae were examined for Clr or lumpy (Lmp) phenotypes (Table 2). Because the Egl phenotype is transient, we scored Egl by culturing 20 late larval stage animals and monitoring them daily for retention of greater than 15–20 embryos or for hatched larvae within the parent. Wit, Unc and Muv phenotypes were scored by counting adults in random-staged populations that did and did not exhibit the phenotype. Clr and Lmp phenotypes were scored by counting young larvae that did and did not exhibit the phenotype. Phenotypes of mutants not viable as homozygotes were tabulated by monitoring homozygous mutant offspring of heterozygous parents.

Genetic analysis of mutations:
To determine whether mutations in newly identified genes were enhanced by deficiencies, we compared the phenotypes of m /Df to that of m /m, where m is the mutation and Df is a deficiency for the locus. To examine cam-1/Df animals, we generated a rol-6(e187) cam-1/mnDf16 strain and examined rol-6 cam-1/mnDf16 self-progeny for the phenotypes shown in Table 3. mnDf16 deletes cam-1, but not rol-6, so non-Rol self-progeny will be hemizygous (m/Df ) for the cam-1 mutation. Strains were scored in the presence of homozygous kyIs5 (CAN position column of Table 3) or in its absence (all other columns of Table 3) for behavioral or morphological defects. To examine fam-2/sDf23 animals, we generated fam-2 kyIs5/sDf23 animals by crossing fam-2 kyIs5 heterozygous males to sDf23/+ hermaphrodites. From fam-2 kyIs5/sDf23 animals, fam-2 kyIs5/sDf23 self-progeny were identified by their reduced GFP fluorescence, and their phenotypes were scored. To confirm that the animals scored contained sDf23, each animal was transferred to a plate and shown to segregate 1/4 dead embryos. To examine syc-3/Df animals, we generated syc-3 dpy-5/qDf3 animals. qDf3 deletes syc-3, but not dpy-5, so non-Dpy self-progeny are hemizygous for syc-3. To examine syc-2/Df, we generated syc-2/mnDf1 and scored the phenotypes. We verified that animals harbored Df by culturing each animal individually and determining that it segregated 1/4 dead embryos. The phenotype of syc-1/Df was not determined.

cam-2 was not deleted by eDf3.fam-1 was not deleted by yDf10 or by sDf121. Chromosomal deficiencies that delete ceh-10 are not currently available. Therefore, we were unable to assess whether the phenotypes of these mutants were enhanced by chromosomal deficiencies.

ceh-10 rescue:
To determine whether the wild-type ceh-10 gene could rescue ceh-10 mutant phenotypes, we generated strains of the genotype ceh-10(gm58), pEx9[ceh-10::rol-6], and ceh-10(gm127); pEx9[ceh-10::rol-6]. pEx9 is described in SVENDSEN and MCGHEE 1995 . Almost all viable offspring from ceh-10(gm58); pEx9 parents were Rol, indicating that the wild-type ceh-10 gene present on the extrachromosomal array rescued the lethality of ceh-10(gm58). The CANs were positioned normally in both ceh-10(gm58); pEx9 and ceh-10(gm127); pEx9 Rol animals, demonstrating that wild-type ceh-10 also rescued the CAN cell migration defect.

Sequencing ceh-10 alleles:
To determine the DNA sequence of the coding region and the intron-exon boundaries of mutant ceh-10 alleles, DNA was cloned by polymerase chain reactions (PCR) from single mutant worms (WILLIAMS et al. 1992 ). To amplify the ceh-10 gene, we used two sets of primers: 5'-dGATGAATTCTCGACTACTGCACACCGG-3' [127–109; numbers indicate nucleotide position in the genomic sequence beginning at the A of the initiator methionine (SVENDSEN and MCGHEE 1995 )] and 5'-dGCCATCGATCCCCAG GTTTTCTCGG-3' (1055–1039) to amplify the 5' half of the gene, and 5'-dGATGAATTCGGCTTACTTACTGAAACCTGA GACC-3' (913–933) and 5'-dGCCATCGATTGTTTCCTGACC GCTC-3' (2064–2049) to amplify the 3' half. The first primer for each reaction contained an EcoRI site, and the second primer contained a ClaI site. We cleaved the PCR products with EcoRI and ClaI and purified them by electrophoresis in 0.8% low melting temperature agarose (SAMBROOK et al. 1989 ). We cloned the products into ClaI and EcoRI cleaved pBSKS+ (Stratagene). DNA was extracted from the transformed cells and sequenced using the following primers: 5'-dGATGAATTCTCGACTACTGCACACCGG-3' (1054–1072), 5'-dCACGGATATTGTCCTCAG-3' (168–185), 5'-dCCAATCT CAACATCAGCGG-3' (359–377), 5'-dGATGAATTCGGCTTA CTTACTGAAACCTGAGACC-3' (913–933), 5'-dAATTACCCT GAACCACC-3' (1602–1618). We determined the DNA sequence of the mutant ceh-10 genes from three independent PCR reactions.

Identification of AIY in ceh-10 mutants:
To determine whether AIY cells were produced but failed to express ceh-10-lacZ or ceh-23-gfp transgenes in strong ceh-10 mutants, or were absent, we examined ceh-10(gm58) mutants for the presence of AIY cells by Nomarski microscopy. AIY was identified by its characteristic location in these animals. AIY cells were detected in 15 of 16 sides of animals examined.

Histochemical and immunocytochemical staining:
We detected ß-galactosidase activity histochemically in wild-type and ceh-10(gm58) first stage larvae (L1) that carried the pInt1[ceh-10-lacZ::rol-6] transgene. We transferred several worms in 200 µl water to a multiwell glass dish. The glass dish was placed under vacuum until the water had evaporated. Three hundred microliters of -20° acetone was added to worms and then allowed to evaporate. One hundred to two hundred microliters of staining solution (300 µl 2x phosphate buffer [360 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 µl 1 M MgCl₂], 100 µl 100 mM Redox buffer [50 mM K ferricyanide, 50 mM K ferrocyanide], 4 µl 1% SDS, 400 µl dH₂O, 12 µl 2% X-gal in dimethyl formamide) was added to worms and allowed to incubate at room temperature for 1 hr to overnight in a sealed, humidified container (FIRE et al. 1990 ). Animals were stained with antibodies to the neurotransmitter GABA as described previously (WIGHTMAN et al. 1996 ).

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Screens for cell migration mutants:
To identify mutants defective in CAN development and migration, we performed two mutant screens. In the first screen, we took advantage of the Clr and Wit visible phenotypes that are displayed by animals lacking normal CAN function. Killing both CANs in newly hatched larvae causes animals to die with a distinctive transparent appearance, the clear (Clr) phenotype (Figure 1B; FORRESTER and GARRIGA 1997 ). In addition, mutants with misplaced CAN cell bodies often develop a dramatic withering of the posterior half of the animal, the withered-tail (Wit) phenotype (Figure 1E; MANSER and WOOD 1990 ; WIGHTMAN et al. 1996 ; FORRESTER and GARRIGA 1997 ). We screened the progeny of individual F₁s from mutagenized parents for the Clr or Wit phenotypes and then determined whether these mutants displayed CAN defects. By screening the progeny of approximately 1000 F₁ animals, we identified two Wit mutants with CAN migration defects, epi-1(gm57) and unc-73(gm67), and two Clr mutants that lacked differentiated CANs, ceh-10(gm58 and gm71) (Table 2). We also identified eight Clr mutants that had morphologically normal CANs. These mutants were not analyzed further.

View larger version (86K): [in this window] [in a new window]   Figure 1. —Cam mutant phenotypes. Lateral views of late first stage larval (A–C) or adult (D–G) hermaphrodites viewed by bright field microscopy. In all panels, anterior is to the left and dorsal is up. (A) Wild-type first stage larva. (B) A ceh-10(gm58) first-stage larva that displays the Clr phenotype. Accumulation of fluid between intestine and body wall makes the animals appear translucent or clear (arrowheads). (C) A fam-1 first stage larva that displays the Lmp phenotype. Note the large ventral protrusion on the head of this animal. (D) Wild-type adult hermaphrodite. Arrowhead indicates the position of the vulva. (E) An adult vab-8(gm99) hermaphrodite that displays the Wit phenotype. Note the reduced girth in the region posterior to the vulva (large arrowhead) relative to the anterior half. Wit animals often have protruding vulvae (large arrowhead) and produce ectopic ventral protrusions consisting of vulval cells (small arrowhead). (F) An adult cam-1(gm105) hermaphrodite that displays the Egl phenotype. The mutant hermaphrodite has retained additional late-stage embryos (small light ovals visible within the animal). Arrowhead indicates the position of the vulva. (G) An adult cam-2 hermaphrodite that displays ectopic vulval protrusions (small arrowheads) anterior to the vulva (large arrowhead). Also note that the magnification of this panel is twice that of D–F, reflecting the small phenotype of cam-2 mutants. Scale bar for A–C shown in A represents 20 µm, and for D=nF, shown in D represents 50 µm. The scale bar in G represents 50 µm.

In the second screen, we took advantage of a reporter that fused regulatory sequences of ceh-23, a C. elegans homeobox gene, to Aequorea victorea green fluorescent protein (GFP; CHALFIE et al. 1994 ). Transgenic animals that contain the ceh-23-gfp reporter express GFP in the CANs, as well as several neurons of the head and tail (Figure 2A). We screened the progeny of individual F₁s from mutagenized parents for animals with missing or misplaced CANs by fluorescence microscopy (for examples, see Figure 2, B and C). By screening the progeny of 2567 F₁ animals, we identified 30 mutants. Of these mutants, six were weak or genetically complex and were not pursued. Of the remaining mutants, three lacked differentiated CANs, and 21 had misplaced CANs. Two mutants, fam-1(gm85) and mig-2(gm38), were identified in a similar screen for cell migration mutants using Nomarski optics microscopy (D. PARRY, P. BAUM and G. GARRIGA, personal communication). Their characterization is included here.

View larger version (71K): [in this window] [in a new window]   Figure 2. —ceh-10 mutants are defective in CAN development. Immunofluorescence photomicrographs of wild-type and ceh-10 mutant larvae bearing the kyIs5 transgene. All panels show a left lateral view with anterior to the left and dorsal up. The arrowhead indicates the position of the CAN cell body. Left and right arrows indicate positions of the ceh-23-gfp-expressing cells in the head and of the gonad primordium in the midbody, respectively. (A) Wild-type early larval stage hermaphrodite. Left CAN is in the middle of animal. (B) ceh-10(gm120) early larval stage hermaphrodite. Left CAN is misplaced anteriorly. (C) ceh-10(gm100) early larval stage hermaphrodite. The CANs are not detected by ceh-23-gfp expression. Small spots seen in each animal are due to intestinal autofluorescence. Scale bar, 20 µm.

To define the genes identified in the screens, the chromosomal positions of the mutations were mapped using standard two- and three-factor mapping and deficiency mapping (MATERIALS AND METHODS, Table 1). Mutations that mapped near one another, or near previously identified genes known to affect cell migrations, were tested for complementation. In this way, the 30 mutations identified in the screens were assigned to 14 genes (Figure 3; Table 1). We identified mutations in the six previously defined genes: epi-1 (epithelialization defective), ina-1 (integrin -subunit), mig-2 (migration defective), unc-34, unc-73 (uncoordinated movement), and vab-8 (variable abnormal), and in the seven new genes: cam-1, cam-2 (CAN abnormal migration); fam-1, fam-2 (fasciculation and cell migration defective); and syc-1, syc-2, and syc-3 (synthetic Cam). In addition, the first mutations in ceh-10 were identified. ceh-10 previously had been identified molecularly as a C elegans homeobox-containing gene (HAWKINS and MCGHEE 1990 ; SVENDSEN and MCGHEE 1995 ).

View larger version (14K): [in this window] [in a new window]   Figure 3. —Map positions of Cam genes. Lines represent the portions of the six C. elegans linkage groups where the Cam genes map. Cam genes identified in our screen are in bold print. The other genes represented on the map were used in Cam gene mapping. Cam genes set off from the line have not been mapped precisely relative to indicated markers. The lines below the LGs represent the extents of deficiencies used for mapping. The map positions of the Cam genes is based on previously published results and data presented in Table 1 and Table 3.

All three Syc mutants require the presence of kyIs5, which contains the ceh-23-gfp transgene, to express a penetrant Cam phenotype (Figure 4). kyIs5 confers a subtle Cam phenotype and enhances the severity and penetrance of the Cam defect in the other mutants. The Syc mutants differ from other Cam mutants, however, in that they require kyIs5 to express a Cam defect. The kyIs5 transgene contains a ceh-10-unc-76-gfp fusion gene, as well as the gene lin-15. The UNC-76 sequences were inserted to help localize GFP to neuronal axons, and the gene lin-15 was a marker used originally to follow the presence of the transgene. We have not rigorously tested whether ceh-23, unc-76, gfp or lin-15 sequences, or the site of kyIs5 integration, confer the Syc phenotype to these mutants.

View larger version (16K): [in this window] [in a new window]   Figure 4. —Cam phenotypes in the Syc mutants. At the top is a schematic representation of a first larval stage (L1) animal. Anterior is to the left and ventral is down. The arrow indicates the direction and extent of CAN migration. Each vertical line along the ventral side of the worm indicates the positions of hypodermal cell nuclei. The CAN nuclei were scored in newly hatched L1 animals relative to the indicated hypodermal nuclei. The CAN positions for each strain are represented as a percentage of all the CANs scored in the bar graphs below.

Phenotypic characterization of Cam mutants:
We investigated whether the mutants were Clr or Wit, phenotypes associated with defects in CAN function (Figure 1, Figure B and E; Table 2). We also examined the mutants for uncoordinated movement (Unc), multiple vulvae (Muv), egg-laying defective (Egl), or lumpy larvae (Lmp) phenotypes (Figure 1, Figure C, Figure F and G; Table 2). Finally, we determined whether mutant phenotypes were enhanced when the mutations were placed over a deficiency for the locus (Table 3).

cam-1(gm105, gm122) II: cam-1 mutant animals are Wit, Unc and Egl. The cam-1 Unc phenotype is distinctive; mutant animals move forward with an exaggerated waveform, often coiling into a tight spiral. In response to high temperatures, high population densities, and limited food, wild-type animals enter a morphologically distinct alternate third larval stage called dauer (CASSADA and RUSSELL 1975 ; GOLDEN and RIDDLE 1984 ). In the presence of low population densities and sufficient food, both cam-1 mutations cause a dauer constitutive phenotype, with approximately 15% of the larvae becoming dauers at 20°. Under these conditions, wild-type animals never become dauers (CASSADA and RUSSELL 1975 ; GOLDEN and RIDDLE 1984 ).

The gm122 allele appears stronger than the gm105 allele. For example, 9% of gm105 animals are Wit and 87% are Unc, while 20% of gm122 animals are Wit and 100% are Unc (Table 2). gm122 animals are also smaller and grow more slowly than gm105 animals. Gene dosage experiments suggest that gm105 reduces cam-1 function, whereas gm122 may eliminate it. cam-1(gm122)/mnDf16 hemizygous animals display most phenotypes with similar frequency and severity to cam-1(gm122) homozygotes, whereas cam-1(gm105)/mnDf16 hemizygous animals are more often Wit and exhibit more severe CAN migration defects than cam-1(gm105) homozygotes (Table 3).

cam-2 (gm124) I: cam-2 mutants often die as Clr larvae. Animals that survive to adulthood are often Wit, Unc and Egl. In addition, nearly 10% of the adult animals are multivulval (Muv; Figure 1G). Mutants displaying a severe Wit phenotype are Muv, with ectopic vulvae always located posterior to the vulva (WIGHTMAN et al. 1996 ). By contrast, cam-2 mutants also develop ectopic vulvae at positions anterior to the vulva, suggesting that the cam-2 Muv phenotype is not a secondary consequence of tail withering. cam-2 mutants are often substantially smaller than wild type and exhibit greatly reduced brood sizes (Figure 1G).

fam-1 (gm85) III: fam-1 mutant larvae are often lumpy in appearance and frequently develop notched heads resembling those seen in vab-3 or ina-1 mutants (Figure 1F; CHISHOLM and HORVITZ 1995 ; BAUM and GARRIGA 1997 ). fam-1 mutants are often Unc, and occasionally Wit and Egl. In addition, fam-1 mutants are slightly shorter and fatter than wild type, a phenotype referred to as dumpy (Dpy).

fam-2 (gm94) IV: fam-2 mutants are rarely Wit or Unc (Table 2). Because fam-2 was identified in the GFP screen and is tightly linked to the GFP reporter, we have been unable to separate it from the reporter. Because hermaphrodites carrying the ceh-23-gfp transgene are Egl, we have not established whether fam-2 hermaphrodites are Egl in the absence of the GFP reporter. fam-2/sDf23 hemizygous animals are more often Wit and Unc than fam-2 homozygotes (Table 3). The penetrance of the Egl defect decreases in hemizygous animals. The lower Egl penetrance may be caused by the decreased dose of the ceh-23-gfp transgene since this transgene causes the Egl phenotype when homozygous, but not when heterozygous.

syc-1(gm126) III: syc-1 mutants are Dpy, Unc and Egl. The Dpy phenotype is maternally rescued; homozygous syc-1 self-progeny from a heterozygous hermaphrodite are less Dpy than syc-1 self-progeny from a homozygous hermaphrodite (data not shown). Similarly, the Cam phenotype of syc-1 mutants is maternally rescued.

syc-2(gm132) X: syc-2 mutants are Unc, generally exhibiting a strong "kinker" locomotion phenotype. Instead of moving in a smooth sinusoidal wave, syc-2 mutants kink with moving. syc-2/mnDf1 animals display the Unc phenotype with similar severity and penetrance to syc-2 mutants alone (Table 3).

syc-3(gm135) I: syc-3 mutants are weakly Unc, generally exhibiting exaggerated curvature of the body when moving, particularly when moving backward. syc-3/qDf3 hemizygous animals are more severely Unc, suggesting that syc-3(gm135) may reduce, but not eliminate gene function (Table 3).

epi-1 (gm57, gm121, gm139, gm146) IV: epi-1 mutants are Wit, Unc, Muv and Lmp (Table 2). Based on the penetrance of these phenotypes, as well as brood size and viability of the mutant strains, the epi-1 alleles can be ordered into an allelic series where gm121 is the weakest, gm57 and gm146 are intermediate, and gm139 is strongest. In addition to the phenotypes described above, epi-1 mutants are somewhat Dpy and exhibit defects in gonadal morphology (not shown and K. JOH, D. HALL, J. YOCHUM, I. GREENWALD and E. HEDGECOCK, personal communication). gm57 and gm146 have greatly reduced brood sizes relative to wild type, and gm139 mutants are sterile. Unlike the other epi-1 alleles, epi-1(gm121) mutants often die as Clr larvae. Because other epi-1 mutants are not Clr, including the three stronger alleles isolated in our screens, the Clr phenotype of epi-1(gm121) mutants may be caused by a second, linked mutation.

ina-1 (gm144) III: ina-1 mutants are Unc, Egl and Lmp, and occasionally Wit. The Unc animals are generally able to move well, but occasionally fail to move smoothly and instead "coil" transiently. ina-1(gm144) is likely to represent a partial loss-of-function mutation because stronger mutations in ina-1 result in larval lethality (BAUM and GARRIGA 1997 ).

mig-2 (gm38, gm103sd) X: mig-2 mutants are Unc and Egl (Table 2). mig-2(gm38) animals are often able to move, but do so poorly, and often exhibit a kinked appearance. Homozygous mig-2(gm103sd) mutants are severely Unc, as well as Wit and Egl. The mutant animals are generally paralyzed, with a strongly kinked appearance. In addition, mig-2(gm103sd), but not mig-2(gm38), mutants often have ectopic laterally positioned vulvae as seen in unc-40 mutants (HEDGECOCK et al. 1990 ). mig-2(gm103sd) is the only obvious semidominant mutation identified in our screens. Although heterozygous animals are rarely Unc, their CANs migrate only an average of 73 ± 22% of their normal distance. By contrast, the CANs of mig-2(gm38)/+ heterozygous animals appear normal.

unc-34 (gm104, gm114, gm115, gm134) V: unc-34 mutants exhibit a severe and 100% penetrant Unc phenotype. The animals either fail to move, or do not move with the smooth motion of wild-type animals.

unc-73 (gm67, gm123) I: unc-73 mutants are often Unc, Egl and Wit. In addition, they have ectopic laterally positioned vulvae like those seen in mig-2(gm103sd) mutants. Some Unc-73 phenotypes are partially maternally rescued. Homozygous self-progeny from heterozygous hermaphrodites are less Wit, do not exhibit lateral vulvae as often, and have larger numbers of progeny than homozygous self-progeny from homozygous hermaphrodites.

vab-8 (gm99, gm138) V: vab-8 mutants are often Wit, Unc, Muv and occasionally Egl. Ectopic vulvae are always located posterior to the vulva (Figure 1E; WIGHTMAN et al. 1996 ).

ceh-10 (gm58, gm71, gm100, gm120, gm127, gm131, gm133) III: We isolated seven ceh-10 mutants of two phenotypic classes. One mutant class, represented by the two alleles gm120 and gm127, are viable. These mutants are occasionally Wit, Unc, Egl and Muv (Table 2). As with vab-8 mutants, we propose that the Unc and Muv phenotypes of these ceh-10 mutants result from tail withering (WIGHTMAN et al. 1996 ). The CANs of these mutants are partially disrupted in their migrations (Figure 2; FORRESTER and GARRIGA 1997 ). The level of CEH-23-GFP expression appears similar to wild type (Figure 2B). A second mutant class, represented by the five alleles gm58, gm71, gm100, gm131 and gm133, die as Clr larvae. The CANs of these mutants appear completely disrupted in their migrations (FORRESTER and GARRIGA 1997 ), and they fail to express the ceh-23-gfp transgene (Figure 3). The visible and CAN phenotypes of the mutants indicate that gm120 and gm127 are weaker than the other ceh-10 alleles, and thus are partial loss-of-function mutations.

All seven ceh-10 mutations mapped genetically to a region of LG III that contained ceh-10, a gene encoding a homeoprotein that is expressed in the CANs (SVENDSEN and MCGHEE 1995 ). To determine whether our mutations were ceh-10 alleles, we introduced a wild-type ceh-10 transgene into gm58 and gm127 mutant strains and found that the gm58 and gm127 mutant phenotypes were rescued by the ceh-10 transgene. When carrying the transgene, gm58 mutants survived to adulthood, and gm127 adult hermaphrodites were no longer Wit, Unc, Egl or Muv. Furthermore, the CANs of gm58 mutants expressed the ceh-23-gfp transgene and the CANs of both gm58 and gm127 mutants were positioned normally.

ceh-10 mutations:
To determine the molecular nature of the ceh-10 mutations, we cloned and sequenced ceh-10 from the mutant strains. The two independent partial loss-of-function mutations, gm120 and gm127, were identical; the conserved GT of the first intron's 5' splice site was mutated to AT (Figure 5). Presumably splicing still occurs at this site, or from a cryptic site, to produce functional product since these mutations do not eliminate ceh-10 function. Two of the stronger alleles, gm58 and gm71, were missense mutations, and two, gm100 and gm133, were nonsense mutations (Figure 5). The gm133 mutation should terminate translation of the protein 10 amino acids into the homeodomain to produce a truncated product that is predicted to lack DNA binding activity. Thus, the strong ceh-10 alleles are likely to eliminate gene function.

View larger version (19K): [in this window] [in a new window]   Figure 5. —ceh-10 mutations. A portion of the ceh-10 gene is shown beginning at the start codon and ending at the stop codon. Boxes represent coding sequences, and lines represent introns. Positions of the mutations, and the molecular lesions, are indicated.

ceh-10 regulates ceh-23 and ceh-10 expression:
Although the partial loss-of-function ceh-10 mutations disrupt only CAN migration, the more severe mutations also appear to perturb CAN differentiation. By Nomarski optics, CANs are not detected in their proper positions or along their migratory routes (FORRESTER and GARRIGA 1997 ). A complete failure of CAN migration would place the cells among other neurons of the head ganglia, making it impossible to distinguish ectopic CANs from neighboring cells by Nomarski optics. In addition, the CANs fail to express the ceh-23-gfp transgene (Figure 2C). The absence of detectable CANs raises several possibilities. The ceh-10 mutations could disrupt divisions in the CAN lineage in such a way that the CANs are never produced. Alternatively, the CANs could be generated, but then fail to differentiate or die.

In ceh-10 mutants, the RMED and AIY neurons, which normally express ceh-10 (SVENDSEN and MCGHEE 1995 ), fail to express differentiation markers. The four RME neurons express the neurotransmitter GABA. RMED, the only RME that expresses ceh-10, failed to express detectable levels of GABA in ceh-10(gm127) mutants; RMED was not detected by GABA staining in 15/16 mutant animals, whereas it was detected in 14/14 wild-type animals. Like misplaced CANs, the RMED neuron occupies a position that made its identification by Nomarski optics difficult. AIY, an interneuron that expresses both ceh-10 and ceh-23 also fails to express ceh-23-gfp in strong ceh-10 mutants (not shown). Unlike the CANs and RMED, the AIY neuron is relatively isolated with three other neurons at the posterior end of the ventral ganglia, making its identification by Nomarski optics straightforward. We have detected AIY in strong ceh-10 mutants by Nomarski optics, indicating that AIY is produced, but fails to differentiate (MATERIALS AND METHODS). We propose that the CANs and RMED are also produced in ceh-10 mutants, but fail to differentiate.

To determine whether ceh-10 expression requires ceh-10 function, we introduced a ceh-10-lacZ transgene into a ceh-10(gm58) mutant background. In wild type, the ceh-10-lacZ transgene is expressed in AIYL/R, CEPDL/R, RID, ALA, RMED, AINL/R, AVJL/R and CANL/R (Figure 6; SVENDSEN and MCGHEE 1995 ). In ceh-10(gm58) mutants, we detect no ceh-10-lacZ transgene expression (Figure 6). This result suggests that ceh-10 regulates its own expression.

View larger version (84K): [in this window] [in a new window]   Figure 6. —ceh-10 expression requires CEH-10 function. Histochemical staining was used to detect ß-galactosidase expression from ceh-10-lacZ transgene. Anterior is to the left, and dorsal is up. (A) Wild-type transgenic animal expressing ß-galactosidase in the CANs and at least one cell in the head (arrowheads). The number of ß-galactosidase expressing cells in transgenic animals varies. (B) ceh-10(gm58) transgenic animals never express ß-galactosidase. Scale bar, 20 µm.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Genetic screens for Cam genes:
We used two genetic screens for CAN migration mutants. First, we isolated mutants that were Wit or Clr, phenotypes associated with defects in CAN function. Second, we screened directly for mutants with misplaced or missing CANs in a strain that expressed GFP in the CANs. Although more labor intensive, the second screen proved more productive; screening approximately two and a half times the number of mutagenized chromosomes identified six times as many mutants.

There are two likely explanations for the greater success of the GFP screen. First, nearly half of the mutants isolated in the direct screen were less than 10% Wit or Clr and would have been missed in the Wit and Clr screen. Second, several of the mutants developed slowly. The Wit phenotype is most easily scored in adult animals, while defects visualized by GFP could be scored in younger animals. In the Wit and Clr screen, mutants that are slow to reach adulthood might be lost among the next generation of nonmutant animals.

Three Syc mutants require the presence of kyIs5 to express the Cam phenotype: mutations in the syc genes result in defects in CAN cell migration only when kyIs5 is present (Figure 4). Mutations in each of the Syc genes results in weak (syc-1 and syc-3) or strong (syc-2) Unc phenotypes in the absence of the transgene, suggesting that the mutations disrupt the development or function of additional neurons. The effects on CAN migration are enhanced when the syc-3 mutation is placed in trans to a deficiency, raising the possibility that stronger alleles of this gene would result in CAN migration defects in the absence of kyIs5.

Most mutations reduce or eliminate gene function:
Our interpretation of the roles played by these genes in CAN cell migration and differentiation assumes that our mutations reduce or eliminate rather than alter gene function. Several considerations support this hypothesis. First, all mutations except mig-2(gm103sd) result in recessive phenotypes, which are likely to result from reducing or eliminating rather than altering gene function (PARK and HORVITZ 1986 ). Second, several of the genes were identified by multiple alleles. We identified seven alleles of ceh-10, four alleles each of epi-1 and unc-34, and two alleles each of cam-1, unc-73, and vab-8. Mutations that reduce or eliminate gene function are more common than mutations that alter function (BRENNER 1974 ). Third, we analyzed mutations of cam-1, fam-2, syc-2 and syc-3 in trans to chromosomal deficiencies of these loci. For the cam-1(gm122) and syc-2(gm132) mutations, the phenotypes of hemizygous and homozygous mutants were similar, suggesting that these mutations severely reduce, and possibly eliminate, gene function. For the cam-1(gm105), fam-2(gm94) and syc-3(gm135) mutations, the phenotypes of hemizygous animals were more severe than those of homozygotes, suggesting that these mutations reduce gene function. Finally, two of the strong ceh-10 alleles are nonsense mutations in the homeodomain, which should result in truncated proteins that lack DNA binding activity.

The mig-2(gm103sd) mutation, by contrast, is semidominant. Heterozygous mig-2(gm103sd)/+ animals are occasionally Unc and have misplaced CANs (not shown), and homozygous mig-2(gm103sd) animals are more severely Unc and have CANs that are more severely misplaced (Table 2; FORRESTER and GARRIGA 1997 ). The mig-2(gm38) mutation, by contrast, appears recessive. These results suggest that gm103 alters mig-2 function.

Neural function in the Cam mutants:
Two observations suggest that the Clr lethal and Wit phenotypes are related, resulting from defects in CAN function. First, Clr and Wit phenotypes both result from defects in CAN function. Animals lacking CANs, generated by laser microsurgery or by mutation, die as Clr larvae, and Cam mutants are often Wit (MANSER and WOOD 1990 ; WIGHTMAN et al. 1996 ; FORRESTER and GARRIGA 1997 ). Second, the tails of Wit mutants appear Clr, suggesting that tail withering results from a lack of CAN function in the posterior body. A correlation between defects in outgrowth of the posterior CAN axon and the Wit phenotype further support this hypothesis (FORRESTER and GARRIGA 1997 ).

Cell migration and axon outgrowth defects appear to result in other behavioral and morphological defects. During embryogenesis, the bilaterally symmetric HSNs migrate anteriorly (SULSTON et al. 1983 ). Later during larval development, each HSN extends an axon that innervates the egg-laying muscles to establish synapses that are essential for egg laying (WHITE et al. 1986 ; DESAI et al. 1988 ; GARRIGA et al. 1993 ). cam-1, cam-2, fam-2, ina-1, mig-2, syc-1 and unc-73 mutants display significant defects in egg laying. The HSN migration or axon pathfinding defects of cam-1, fam-2, ina-1, mig-2 and unc-73 may contribute to the egg-laying defective (Egl) phenotype of these mutants (FORRESTER and GARRIGA 1997 ). Because the HSNs of cam-2 and syc-1 mutants appear normal, the Egl phenotype of these mutants results from defects in HSN function or from defects in the development or function of other components of the egg-laying system. Consistent with this view, cam-2 mutants are abnormal for development of the vulva, the opening though which eggs are laid.

Locomotion in C. elegans requires the coordinated contractions of body wall muscles that the nervous system regulates. cam-1, cam-2, epi-1, fam-1, ina-1, mig-2, syc-2, unc-34, unc-73 and vab-8 mutants are usually uncoordinated (Unc). epi-1, fam-1, mig-2, unc-34 and unc-73 displayed defects in the morphology of the VD and DD motor neurons, which regulate locomotion (FORRESTER and GARRIGA 1997 ), and vab-8 mutants displayed defects in the morphology of the interneurons involved in movement (WIGHTMAN et al. 1996 ), making it likely that these axon outgrowth defects contribute to Unc phenotypes of these mutants. Possibly, defects in the outgrowth of other axons involved in movement contribute to the Unc phenotype of cam-1, cam-2, ina-1 and syc-2 mutants.

ceh-10 is a member of a gene family necessary for cell differentiation:
CEH-10 is homologous to Chx10 in mouse and Vsx-1 in goldfish, two proteins that are expressed during retinal development (LEVINE et al. 1994 ; LIU et al. 1994 ). The proteins are members of the Pax-like class of homeoproteins, containing a conserved octapeptide sequence and Pax-like homeodomain, but lacking the paired-box of Pax homeoproteins. In addition, all three proteins contain a highly conserved 60-amino acid region located immediately C-terminal to the homeodomain, referred to as the CVC domain [for ceh-10, Vsx-1 and Chx10, also referred to as the extended conservation region (LEVINE et al. 1994 ; SVENDSEN and MCGHEE 1995 )]. The function of the CVC domain is not known.

As discussed above, the strong ceh-10 alleles are likely to eliminate ceh-10 function. In particular, the ceh-10(gm133) allele is a nonsense mutation that is predicted to terminate translation at the tenth amino acid of the homeodomain to produce a truncated protein with no DNA binding activity (Figure 5). The shared phenotypes of the other four strong mutants argue that they also completely lack ceh-10 function. Two of these mutants, gm58 and gm71, contain changes in amino acids that are conserved among the CEH-10, mouse Chx10 and goldfish Vsx-1 CVC domains, suggesting that the CVC domain is essential for function.

The gene Chx10 specifies the fate of bipolar neurons of the retina. Chx10 is expressed in the neuroretina, the hindbrain and spinal cord, and its expression is maintained at high levels in bipolar cells of the retina (LIU et al. 1994 ). Ocular retardation (or) is a nonsense mutation in the Chx10 gene that causes reduced proliferation of retinal progenitors and a lack of differentiated bipolar cells (BURMEISTER et al. 1996 ).

Like Chx10, we propose that ceh-10 specifies the fate of neurons that express it. Both the CANs and AIY interneurons normally express the ceh-10 and ceh-23 genes (WANG et al. 1993 ; SVENDSEN and MCGHEE 1995 ). In strong ceh-10 mutants, the CANs fail to migrate, and neither the CANs nor AIYs express ceh-23-gfp or ceh-10-lacZ. RMED neurons in ceh-10 mutants no longer express detectable levels of their neurotransmitter GABA, suggesting that the fate of these cells has also been altered. Although we have not been able to unambiguously detect the RMED or displaced CAN neurons among the other neurons of the head ganglia, we have detected the AIY neuron in strong ceh-10 mutants, indicating that AIY is produced, but fails to differentiate normally (see MATERIALS AND METHODS). We also propose that some aspects of ceh-10-dependent differentiation of the CANs and AIY interneurons are regulated by CEH-23. Because ceh-23 mutants have not been reported, we do not know how CEH-23 regulates CAN or AIY differentiation. By analogy to CEH-10, we propose that Chx10 might act through other transcriptional regulators to control bipolar cell fate.

Unlike complete loss-of-function mutants, CAN cells in partial loss-of-function ceh-10 mutants express ceh-23-gfp, and migrate more than half their normal distance (FORRESTER and GARRIGA 1997 ). In these mutants, CAN cells must retain function, because a complete lack of CAN function is lethal.

Mutants that retain partial ceh-10 activity exhibit CAN cell migration defects, but still express ceh-23-gfp in CANs and AIY, raising the possibility that ceh-10 acts independently of ceh-23 to regulate genes involved in cell migration. Alternatively, ceh-10 could act through ceh-23 to regulate cell migration genes, and ceh-23 expression in weak ceh-10 mutants would be insufficient to properly regulate genes that control CAN cell migration. Identification of transcriptional targets of the ceh-10 and ceh-23 homeodomain genes should provide insight into the mechanisms by which neurons acquire their unique migratory and functional properties.

	ACKNOWLEDGMENTS

We thank CORI BARGMANN, PAUL BAUM, SCOTT CLARK, DAVID BAILLIE, JIM MCGHEE, DIANNE PARRY, HELEN STEWART, PIA SVENDSEN and THERESA STIERNAGLE and the Caenorhabditis Genetics Center for providing strains. Some strains used in this work were provided by the C. elegans Genetic Toolkit Project, which is funded by a grant from the National Institutes of Health to ANN M. ROSE, DAVID L. BAILLIE and DONALD L. RIDDLE. We thank NANCY HAWKINS for critical reading of this manuscript and CLAIRE WALCZAK for assistance with producing figures. This work was supported by National Institutes of Health (NIH) grant NS-32057 to G.G., by postdoctoral fellowships from the NIH and from the Breast Cancer Research Program of the U.S. Army Medical Research and Materiel Command to W.C.F., and by a predoctoral fellowship from the National Science Foundation to J.A.Z.

Manuscript received June 26, 1997; Accepted for publication September 8, 1997.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ACKERMAN, S. L., L. P. KOZAK, S. A. PRZYBORSKI, L. A. RUND, and B. B. BOYER et al., 1997  The mouse rostral cerebellar malformation gene encodes an UNC-5-like protein. Nature 386:838-842[Medline].

ANDERSON, P. and S. BRENNER, 1984  A selection for myosin heavy chain mutants in the nematode Caenorhabditis elegans.. Proc. Natl. Acad. Sci. USA 81:4470-4474[Abstract/Free Full Text].

AUSTIN, J. and J. KIMBLE, 1989  Transcript analysis of glp-1 and lin-12, homologous genes required for cell interactions during development of C. elegans.. Cell 58:565-571[Medline].

BAUM, P. D. and G. GARRIGA, 1997  Neuronal migrations and axon fasciculation are disrupted in ina-1 integrin mutants. Neuron 19:51-62[Medline].

BLOOM, L. and H. R. HORVITZ, 1997  The Caenorhabditis elegans gene unc-76 and its human homologs define a new gene family involved in axonal outgrowth and fasciculation. Proc. Natl. Acad. Sci. USA 94:3414-3419[Abstract/Free Full Text].

BRENNER, S., 1974  The genetics of Caenorhabditis elegans.. Genetics 77:71-94[Abstract].

BURMEISTER, M., J. NOVAK, M. Y. LIANG, S. BASU, and L. PLODER et al., 1996  Ocular retardation mouse caused by Chx10 homeobox null allele: impaired retinal progenitor proliferation and bipolar cell differentiation. Nat. Genet. 12:376-384[Medline].

CASSADA, R. C. and R. L. RUSSELL, 1975  The dauerlarva, a post-embryonic developmental variant of the nematode Caenorhabditis elegans.. Dev. Biol. 46:326-342[Medline].

CHALFIE, M., J. N. THOMSON, and J. E. SULSTON, 1983  Induction of neuronal branching in Caenorhabditis elegans.. Science 221:61-63[Abstract/Free Full Text].

CHALFIE, M., Y. TU, G. EUSKIRCHEN, W. W. WARD, and D. C. PRASHER, 1994  Green fluorescent protein as a marker for gene expression. Science 263:802-805[Abstract/Free Full Text].

CHAN, S. S., H. ZHENG, M. W. SU, R. WILK, and M. T. KILLEEN et al., 1996  UNC-40, a C. elegans homolog of DCC (deleted in colorectal cancer), is required in motile cells responding to UNC-6 netrin cues. Cell 87:187-195[Medline].

CHISHOLM, A., 1991  Control of cell fate in the tail region of C. elegans by the gene egl-5.. Development 111:921-932[Abstract/Free Full Text].

CHISHOLM, A. D. and H. R. HORVITZ, 1995  Patterning of the Caenorhabditis elegans head region by the Pax-6 family member vab-3.. Nature 376:52-55.

COSTA, M., M. WEIR, A. COULSON, J. SULSTON, and C. KENYON, 1988  Posterior pattern formation in C. elegans involves position-specific expression of a gene containing a homeobox. Cell 55:747-756[Medline].

DELONG, L., J. D. PLENEFISCH, R. D. KLEIN, and B. J. MEYER, 1993  Feedback control of sex determination by dosage compensation revealed through Caenorhabditis elegans sdc-3 mutations. Genetics 133:875-896[Abstract].

DESAI, C., G. GARRIGA, S. L. MCINTIRE, and H. R. HORVITZ, 1988  A genetic pathway for the development of the Caenorhabditis elegans HSN motor neurons. Nature 336:638-646[Medline].

FAZELI, A., S. L. DICKINSON, M. L. HERMISTON, R. V. TIGHE, and R. G. STEEN et al., 1997  Phenotype of mice lacking functional Deleted in colorectal cancer (Dcc) gene. Nature 386:796-804[Medline].

FIRE, A., S. W. HARRISON, and D. DIXON, 1990  Modular set of lacZ fusion vectors for studying gene expression in Caenorhabditis elegans. Gene 93:189-198[Medline].

FORRESTER, W. and G. GARRIGA, 1997  Genes necessary for C. elegans cell and growth cone migrations. Development 124:1831-1843[Abstract].

GARRIGA, G., C. DESAI, and H. R. HORVITZ, 1993  Cell interactions control the direction of outgrowth, branching and fasciculation of the HSN axons of Caenorhabditis elegans. Development 117:1071-1087[Abstract].

GOLDEN, J. W. and D. L. RIDDLE, 1984  The Caenorhabditis elegans dauer larva: developmental effects of pheromone, food, and temperature. Dev. Biol. 102:368-378[Medline].

GREENWALD, I. S., P. W. STERNBERG, and H. R. HORVITZ, 1983  The lin-12 locus specifies cell fates in Caenorhabditis elegans.. Cell 34:435-444[Medline].

HARRIS, R., L. M. SABATELLI, and M. A. SEEGER, 1996  Guidance cues at the Drosophila CNS midline: identification and characterization of two Drosophila Netrin/UNC-6 homologs. Neuron 17:217-228[Medline].

HAWKINS, N. C. and J. D. MCGHEE, 1990  Homeobox containing genes in the nematode Caenorhabditis elegans.. Nucleic Acids Res. 18:6101-6106[Abstract/Free Full Text].

HEDGECOCK, E. M., J. G. CULOTTI, and D. H. HALL, 1990  The unc-5, unc-6 and unc-40 genes guide circumferential migrations of pioneer axons and mesodermal cells on the epidermis in C. elegans.. Neuron 4:61-85[Medline].

HOSONO, R., K. HIRAHARA, S. KUNO, and T. KURIHRA, 1982  Mutants of Caenorhabditis elegans with dumpy and rounded head phenotype. J. Exp. Zool. 224:135-144.

HUANG, L. S., P. TZOU, and P. W. STERNBERG, 1994  The lin-15 locus encodes two negative regulators of Caenorhabditis elegans vulval development. Mol. Biol. Cell 5:395-411[Abstract].

KEINO, M. K., M. MASU, L. HINCK, E. D. LEONARDO, and S. S. CHAN et al., 1996  Deleted in Colorectal Cancer (DCC) encodes a netrin receptor. Cell 87:175-185[Medline].

KENYON, C., 1986  A gene involved in the development of the posterior body region of C. elegans.. Cell 46:477-487[Medline].

KOLODZIEJ, P. A., L. C. TIMPE, K. J. MITCHELL, S. R. FRIED, and C. S. GOODMAN et al., 1996  frazzled encodes a Drosophila member of the DCC immunoglobulin subfamily and is required for CNS and motor axon guidance. Cell 87:197-204[Medline].

LANDER, A. D., 1989  Understanding the molecules of neural cell contacts: emerging patterns of structure and function. Trends Neurosci. 12:189-195[Medline].

<