Genes Necessary for C. Elegans Cell and Growth Cone Migrations

Development 124, 1831-1843 (1997) 1831 Printed in Great Britain © The Company of Biologists Limited 1997 DEV9531

Genes necessary for C. elegans cell and growth cone migrations

Wayne C. Forrester* and Gian Garriga Department of Molecular and Cell Biology, University of California, Berkeley, CA 94720, USA *Author for correspondence (e-mail: [email protected])

SUMMARY

The migrations of cells and growth cones contribute to cell interactions between the CAN and HSN, a neuron that form and pattern during metazoan development. To study migrates anteriorly to a position adjacent to the CAN, the mechanisms that regulate cell motility, we have control the final destination of the HSN cell body. Third, C. screened for C. elegans mutants defective in the posteriorly elegans larval development requires the CANs. In the directed migrations of the canal-associated neurons absence of CAN function, larvae arrest development, with (CANs). Here we describe 14 genes necessary for CAN cell excess fluid accumulating in their pseudocoeloms. This migration. Our characterization of the mutants has led to phenotype may reflect a role of the CANs in osmoregula- three conclusions. First, the mutations define three gene tion. classes: genes necessary for cell fate specification, genes necessary for multiple cell migrations and a single gene Key words: cell migration, axonal outgrowth, C. elegans, canal necessary for final positioning of migrating cells. Second, associated neurons

INTRODUCTION by regulating actin dynamics, perhaps by regulating the pro- trusion of actin-rich structures at the leading edge of the cell Cell migration and axonal outgrowth play important roles in or growth cone to drive forward movement (reviewed by Hall, shaping form and pattern during development. The migrations 1994; Mitchison and Cramer, 1996; Lauffenburger and of growth cones, for example, contribute to the final pattern of Horwitz, 1996). connectivity that is essential for nervous system function. To identify genes necessary for cell and growth cone migra- Although much is known about the patterns of cell and growth tions in C. elegans, we isolated mutants defective in the migra- cone migrations, both genetic and molecular approaches are tions of the canal-associated neurons (CANs). The CANs are only beginning to define the molecules that regulate these cell a pair of bilaterally symmetric neurons that are born in the head movements. and migrate posteriorly to the middle of the embryo (Fig. 1A, Extracellular matrix molecules and cell-surface ligands can Sulston et al., 1983). After migrating, each CAN extends two function to guide cell and growth cone migrations (reviewed axons along the lateral body wall, one anteriorly to the head by Hynes, 1992; Goodman and Shatz, 1993; Luo and Raper, and one posteriorly to the tail (Fig. 1B). During their develop- 1994; Tessier-Lavigne, 1994). The product of the C. elegans ment, the CANs express two differentiation markers, the gene unc-6, for example, encodes a secreted protein that guides homeodomain proteins CEH-10 and CEH-23 (Svendsen and both cell and growth cone migrations along the dorsal-ventral McGhee, 1995; Wang et al., 1993). axis, acting through the UNC-5 and UNC-40 cell surface In this paper, we present a phenotypic characterization of the receptors (Hedgecock et al., 1990; Leung-Hagesteijn et al., CAN abnormal migration (Cam) mutants. This analysis has led 1992; McIntire et al., 1992; Chan et al., 1996). Vertebrate to three conclusions. First, the mutations define three gene homologs of UNC-6, the netrins, function similarly to guide classes. In the first class, ceh-10 and possibly cam-2 regulate growth cone migrations along the dorsal-ventral axis of the ver- CAN cell differentiation. For example, ceh-10 mutations tebrate spinal cord (Serafini et al., 1994; Kennedy et al., 1994). disrupt CAN migration, axonal outgrowth and expression of Furthermore, a netrin receptor, DCC, is an UNC-40 homolog CEH-23. In the second class, eight genes may function directly (Keino-Masu et al., 1996). in cell migration and axonal outgrowth. Mutations in these Signals from extracellular guidance cues are integrated by genes disrupt multiple cell and growth cone migrations. In the intracellular signaling molecules. Members of the Rho family third class, cam-1 may function to define the precise destina- of Ras-like small GTPases, Cdc42, Rac and Rho, are intracel- tions of certain migrating cells. Mutations in cam-1 disrupt lular signaling molecules that regulate cell motility and attach- final positioning of certain migrating cells, resulting in both ment (Ridley and Hall, 1992; Ridley et al., 1992; Nobes and incomplete and excessive migrations. In addition to these Hall, 1995), and recent genetic analysis of the unc-73 gene has genes, mutations in three syc (synthetic Cam) genes result in suggested that these molecules may function in C. elegans cell Cam defects only in the presence of a ceh-23-gfp reporter that and growth cone migrations (Run et al., 1996). Signaling expresses GFP in the CANs. molecules ultimately influence cell and growth cone migrations Second, cell interactions between the CANs and HSNs 1832 W. C. Forrester and G. Garriga

A. epi-1(gm139), epi-1(gm146), fam-2(gm94), kyIs5 [ceh-23-unc-76- gfp::lin-15] CAN LGV: unc-34(gm104), unc-34(gm114), unc-34(gm115), unc- Head 34(gm134), vab-8(gm99), vab-8(gm138) LGX: mig-2(gm38), mig-2(gm103sd), kyIs4 [ceh-23-unc-76- Tail gfp::lin-15] The isolation and genetic characterization of Cam mutants epi-1 HSN (epithelialization defective), ina-1 (integrin, α-subunit), unc-34 (uncoordinated), mig-2 (cell migration defective), unc-73, vab-8 Gonad (variable abnormal), ceh-10 (C. elegans homeobox), cam-1, cam-2 embryo (CAN abnormal migration), syc-1, syc-2, syc-3, (synthetic Cam), fam- 1 and fam-2 (fasciculation and cell migration defective) will be described in a subsequent paper by Forrester et al. (unpublished data). The ceh-10 alleles gm71, gm100, gm131, and gm133 all appear B. identical to gm58 for all phenotypes examined, and therefore only data for gm58 are presented in this report. dpy-5(e61) and unc-30(e191) CAN were described by Brenner (1974) and dpy-20(e1282ts) was described by Hosono et al. (1982). The chromosome III balancer qC1 was described by Austin and Kimble (1989). kyIs5 is a ceh-23-gfp reporter larva transgene that is integrated on LGIV (J. Zallen and C. Bargmann, personal communication) and kyIs4 is the same reporter integrated on Fig. 1. CANs and HSNs. (A) CAN and HSN migrations. Schematic LGX (J. Zallen and C. Bargmann, personal communication). lateral view of an embryo 400 minutes after first cleavage, the Because most of the mutants were isolated in a strain containing approximate time of the CAN and HSN migrations. Only one of each the kyIs5 reporter, we removed this transgene from the mutants by of the two bilaterally symmetric CANs and HSNs is shown. Arrows crossing to wild type. The epi-1 alleles gm121, gm139 and gm146, indicate routes of CAN and HSN migration. By the time the CANs and the fam-2(gm94) allele are tightly linked to kyIs5 and have not and HSNs reach their final destinations flanking the gonad been separated from the reporter. For quantitative determinations of primordium, however, the embryo is much longer than shown here. CAN axonal morphologies in mutants using green fluorescent protein After Sulston et al. (1983). (B) CAN morphology. Schematic lateral (GFP) (Chalfie et al., 1994), the kyIs5 reporter was crossed into the view of a newly hatched first larval stage animal. Anterior is to the mutant backgrounds. For epi-1(gm57), which was isolated in a back- left. The CAN cell body is located near the middle of the animal ground lacking kyIs5, the gm57 mutation was crossed into a kyIs4 along the anterior-posterior axis. Only one of each of the two CANs background. is shown. Each CAN extends one axon anteriorly to the head and one Other than in ceh-10(gm58) mutants, all cell migrations were axon posteriorly to the tail. scored in homozygous mutant animals derived from homozygous parents. Because ceh-10(gm58) is lethal, we examined cell positions appear to control the extent of the anteriorly directed HSN cell in homozygous animals derived from ceh-10(gm58)/qC1 parents. body migrations. The HSNs are born in the tail and migrate Because unc-73(gm67) and unc-73(gm123) progeny of homozygous anteriorly to positions adjacent to the CANs (Fig. 1A). In parents rarely survive to adulthood, we examined axonal morpholo- animals with missing or anteriorly displaced CANs, the HSNs gies of homozygous mutant animals derived from unc-73/dpy-5(e61) often migrate beyond their normal destinations, suggesting that hermaphrodites. Similarly, because epi-1(gm139) animals are sterile, the CANs signal the HSNs to stop migrating. we examined axonal morphologies in mutant animals derived from epi-1(gm139)/dpy-20(e1282ts) unc-30(e191) hermaphrodites. Third, C. elegans postembryonic development requires CAN function. In mutants that lack functional CANs, embryogen- Scoring of axons and cells esis proceeds at a normal rate, but larval development arrests. Axons were scored as defective in Table 2 if they deviated from wild- Furthermore, withering of the posterior body region displayed type morphology. This morphology was defined by electron micro- by many Cam mutants can be explained by a defect in scopic reconstructions of the C. elegans nervous system (White et al., outgrowth of posteriorly directed CAN axons that results in a 1986), immunocytochemical staining of wild-type animals (Desai et lack of CAN function in the tail. We speculate that the larval al., 1988; McIntire et al., 1992) and analysis of GFP expression in arrest and tail withering phenotypes of the mutants reflect a transgenic animals (J. Zallen and C. Bargmann, personal communi- role of the CANs in osmoregulation. cation and this report). The extent of cell migration in wild-type and mutant animals was determined by comparing the positions of nuclei relative to coordinate non-migratory hypodermal nuclei using Nomarski optics (Figs 1 MATERIALS AND METHODS and 2). For the CANs, HSNs, ALMs, coelomocyte mother cells, Z1, Z4 and M mesoblast, cells that migrate embryonically, we scored in Strains and genetics newly hatched hermaphrodite larvae the positions of the nuclei of Strains were grown at 20°C and maintained as described by Brenner these cells (or their progeny, the coelomocytes, in the case of coelo- (1974). In addition to the wild-type strain N2, strains with the mocyte mother cells) relative to non-migratory hypodermal V and P following mutations were used in this work. nuclei. In wild-type animals, cells migrate to occupy a range of LGI: dpy-5(e61), unc-73(gm67), unc-73(gm123), cam-2(gm124), positions. We considered a cell to have migrated 100% if it was syc-3(gm135) located anywhere within that range. To quantitate the cell positions, LGII: cam-1(gm105), cam-1(gm122) the distance between the starting point for a given migration (0%) and LGIII: ceh-10(gm58), ceh-10(gm71), ceh-10(gm100), ceh- the position after migration (the low end of the normal range was 10(gm120), ceh-10(gm127), ceh-10(gm131), ceh-10(gm133), syc- 100%) was divided into 100 increments. For cells that had migrated 2(gm132), fam-1(gm85), ina-1(gm144), syc-1(gm126) beyond their normal range of positions, we extended the metric LGIV: dpy-20(e1282ts), unc-30(e191), epi-1(gm57), epi-1(gm121), beyond 100% using the same scale that was used for that cell’s C. elegans cell migration mutants 1833

Fig. 2. Embryonic and larval migrations examined A. in the Cam mutants. Anterior is to the left. ALM CAN (A) Schematic lateral view of the left side of a newly hatched first larval stage hermaphrodite. Both the final positions of the ALM, CAN, ccsL and HSN cell bodies (ovals and circles) and their ccsL HSN migration routes (arrows) are indicated. (B) Schematic lateral view of the right side of a late first larval stage animal. The final positions of the B. SDQ cell bodies of the QR descendants, SDQ, AVM and AQR (circles), and their migration routes (arrows) are indicated. Soon after it begins migrating, QR divides. The anterior QR daughter subsequently divides to produce an AVM/SDQ precursor and a AQR AVM cell that dies, and the posterior QR daughter divides to produce AQR and a cell that dies. Cell deaths are not shown. After the SDQ/AVM precursor division, SDQ migrates dorsally and AVM ventrally. After its birth, AQR migrates anteriorly to a position near the posterior bulb of the pharynx. migration. The positions of individual nuclei were tabulated and these operated animals displayed the Clr phenotype, where the pseudo- numbers were used to calculate average distance migrated and coeloms had filled with fluid, causing the animals to appear clear or standard deviation. For the Q neuroblasts and their descendants, which translucent. We also killed both CAN nuclei in six kyIs5 L1 her- migrate during the first larval stage, we scored in late first larval stage maphrodites and checked whether the CANs could be detected by flu- hermaphrodites the final positions of the Q-descendant nuclei relative orescence microscopy. In none of the animals were we able to detect to the two daughter hypodermal nuclei derived from V1-6. the CANs immediately after the operations. Two of the animals never recovered from laser microsurgery, two arrested as L1s and two Immunocytochemical staining arrested as L4s. All arrested larvae were clear. As a control, we killed GABA- and serotonin-expressing neurons were examined by indirect the HSNs in early 16 wild-type L1s. All of the laser-operated animals immunofluorescence. Animals were collected, permeabilized, fixed survived to become adult egg-laying defective hermaphrodites (the and stained as described by McIntire et al. (1992). Anti-GABA and HSNs are necessary for normal egg laying). Thus, killing the CANs anti-serotonin antibodies were obtained from J. Steinbusch, Free Uni- may affect the ability of larvae to survive anesthesia in 10-20 mM versity, Amsterdam. Fluorescein isothiocyanate (FITC)-conjugated NaN3. secondary antibodies were obtained from Cappel, Inc (catalog number To test whether the CANs were required for later stages of larval 55646). Stained animals were examined using a Zeiss Axioskop and development, we killed both CANs in eight L2s, in six L3s and in Zeiss FITC filter number 487910. Photographs were taken on Kodak three L4s. One of the L2 laser-operated animals failed to recover from Ektachrome T160, Ektachrome Elite 400 or Technical Pan film. anesthesia, two arrested as L3s, one arrested as an L4, and three Images were digitized with a Nikon LS-1000 scanner and manipu- reached adulthood. All of the L2 laser-operated animals were Clr. One lated on an Apple Macintosh computer in Adobe Photoshop to of the L3 laser-operated animals failed to survive the anesthesia, three improve contrast. Images were printed on a Tektronix Phaser 440 arrested as Clr L4s, and two reached adulthood. One of the adults was printer. Clr. All three of the L4 laser-operated animals became normal adults. Laser cell killing Cell nuclei were destroyed using a laser microbeam as described by Avery and Horvitz (1987). To determine whether the CANs are RESULTS involved in HSN positioning, we killed CAN precursor cells, which divide to produce the CANs and the OLL socket cells, at approxi- 14 genes function in CAN cell migration mately 260 minutes after cleavage of the zygote (Sulston et al., 1983). To identify genes involved in cell migration, we conducted two In six embryos, we attempted to kill a single CAN precursor, and in screens for mutants with misplaced or abnormal CANs (W. C. the remaining thirteen embryos, we attempted to kill both CAN pre- Forrester et al., unpublished data). The screens identified 30 cursors. In all attempts, we successfully killed only one of the two mutations in 14 genes that function in CAN cell migration, CAN precursors. After hatching, we scored the laser-operated animals for missing CANs and for the positions of the HSNs. On three of 16 including four alleles of ceh-10 not presented here because sides where we killed the CANs, the HSNs had migrated beyond their they appear similar to ceh-10(gm58) (Table 1). The screens normal destinations. Because of variability in positioning of the CAN identified alleles of the six previously defined genes epi-1, ina- precursor, we often killed a cell near the CAN precursor; these served 1, mig-2, unc-34, unc-73 and vab-8, and of the seven new genes as controls. In all 37 of these control ablations, the HSN on the same cam-1, cam-2 (CAN abnormal migration), syc-1, syc-2, syc-3 side as the ablated cell migrated to its normal destination. Similar (synthetic Cam), fam-1 and fam-2 (fasciculation and migration numbers of embryos died during the manipulations when we killed defective). The screens also identified the first mutations in the only one CAN precursor. To determine whether a single CAN is suf- gene ceh-10, which encodes a homeodomain protein expressed ficient for viability, we followed 12 animals that lacked a single CAN in the CANs (Hawkins and McGhee, 1990; Svendsen and precursor. Eleven of the 12 animals grew to be normal adult her- McGhee, 1995). With the exception of mig-2(gm103sd), all of maphrodites, and one arrested as a first-stage larva. Both CAN nuclei of 21 first-larval-stage hermaphrodites (L1s) were the mutations are recessive and therefore likely to reduce or destroyed by laser microsurgery. Three of the laser-operated animals eliminate gene function (W. C. Forrester et al., unpublished never recovered from anesthesia, six arrested as L1s or second-stage data). larvae (L2), five arrested as third-stage larvae (L3), five arrested as Although six genes had been described, our alleles of mig- fourth-stage larvae (L4), and two reached adulthood. All laser- 2 and unc-73 behave differently from previously identified 1834 W. C. Forrester and G. Garriga

Table 1. Cell migration defects in Cam mutants Cell positions* Strain CAN cell† ALM cell‡ HSN cell¤ ccsL cell¦ SDQ+AVM** Wild type 100.0±0.0 (50) 99.5±3.3 (52) 100.0±0.0 (50) 99.9±0.4 (50) 99.2±2.7 (26) kyIs5 99.3±2.5 (26) 92.8±9.2 (24) 103.3±8.2 (6)†† 100.0±0.0 (10) 99.9±0.4 (24) unc-73(gm67) 56.0±42.6 (15) 57.3±25.2 (14) 46.7±28.8 (12) 31.3±9.5 (12) nd‡‡ unc-73(gm123) nd‡‡ nd‡‡ nd‡‡ nd‡‡ nd‡‡ cam-2(gm124) 57.8±36.4 (26) 94.1±13.1 (24) 104.8±8.7 (25) 95.0±7.8 (6) 95.2±14.7 (30) syc-3(gm135) 98.7±5.7 (27) 92.8±12.3 (28) 100.6±4.0 (26) 95.7±7.4 (30) 98.7±2.6 (25) cam-1(gm105) 90.7±25.3 (32) 99.6±1.6 (31) 102.6±7.1 (29) 96.1±11.4 (23) 93.7±17.1 (23) cam-1(gm122) 19.5±33.6 (28) 91.6±18.8 (32) 114.5±9.1 (29) 98.1±4.9 (13) 97.1±6.1 (23) ceh-10(gm58) none¤¤ 96.8±6.2 (29) 105.5±9.1 (29) 98.7±3.0 (26) 92.1±11.3 (32) ceh-10(gm120) 62.1±37.3 (32) 96.7±7.4 (32) 100.7±3.7 (30) 100.0±0.0 (26) 98.9±2.8 (26) ceh-10(gm127) 67.9±23.0 (29) 87.9±15.5 (35) 100.0±0.0 (37) 95.8±7.5 (28) 99.6±1.4 (25) syc-2(gm132) 99.5±3.0 (31) 100.0±0.0 (30) 99.3±4.1 (31) 99.7±1.1 (30) 99.8±0.8 (26) fam-1(gm85) 94.6±9.5 (29) 83.9±21.6 (28) 95.1±11.0 (31) 93.2±4.6 (29) nd‡‡ ina-1(gm144) 36.7±30.1 (31) 40.4±21.4 (25) 72.6±34.6 (35) 94.4±9.4 (12) 84.7±13.0 (19) syc-1(gm126) 98.9±3.8 (20) 92.0±22.8 (20) 98.7±4.9 (19) 100.0±0.0 (12) 99.4±2.2 (26) fam-2(gm94)¦¦ 9.2±22.7 (34) 28.2±19.8 (33) 52.5±36.1 (2) 98.9±3.5 (28) 98.9±2.8 (26) epi-1(gm57) 88.4±10.9 (21) 87.2±22.7 (19) 87.2±23.8 (18) 84.2±14.9 (10) 38.7±7.7 (6) epi-1(gm146)¦¦ 35.3±40.9 (10) 66.7±38.9 (6) 19.5±34.9 (8) nd‡‡ nd‡‡ epi-1(gm121)¦¦ 51.1±37.2 (30) 58.0±28.7 (29) 100±0 (24) 95±6.0 (29) 96.9±4.2 (25) epi-1(gm139)¦¦ nd‡‡ nd‡‡ nd‡‡ nd‡‡ nd‡‡ unc-34(gm104) 77.2±26.2 (30) 75.7±23.3 (29) 93.7±11.1 (30) 79.3±14.3 (24) 98.6±3.3 (26) unc-34(gm114) 71.2±32.5 (41) 77.0±22.7 (37) 92.5±13.1 (34) 84.9±7.6 (11) 97.0±4.4 (26) unc-34(gm115) 61.7±35.8 (33) 75.9±19.8 (19) 63.7±39.7 (30) 72.5±19.7 (19) 98.9±3.2 (26) unc-34(gm134) 59.2±36.2 (30) 79.1±25.2 (30) 76.4±36.3 (30) 81.5±9.1 (26) 97.7±5.9 (26) vab-8(gm99) 3.1±14.4 (38) 88.1±18.7 (36) 102.9±8.4 (35) 68.6±9.2 (28) 96.6±7.3 (22) vab-8(gm138) 2.5±10.8 (36) 85.4±18.7 (35) 103.2±7.5 (31) 82.5±9.3 (24) 98.4±3.6 (18) mig-2(gm38) 76.4±19.0 (31) 60.2±22.3 (38) 78.6±10.6 (37) 80.9±14.7 (10) 87.1±15.4 (30) mig-2(gm103) 29.1±32.6 (26) 53.9±30.3 (27) 82.0±13.8 (25) 62.9±16.6 (8) 80.9±16.0 (26)

*Positions were determined in mutant and wild-type strains using Nomarski optics. CAN, ALM, HSN, and ccsL (left coelomocytes) positions were determined in newly hatched larvae, and SDQ and AVM (QR descendants) were determined in mid L1 animals. Numbers represent average cell positions ± s.d. (see Materials and Methods). Numbers in parentheses represent number of cells scored. Genes are ordered based on their positions on linkage groups, with genes on the left arm of LGI first and genes on the right arm of LGX last. †CAN migrated 100% if its nucleus was located posterior to the V3 hypodermal nucleus and 0% if located anterior to H2 hypodermal nucleus. ‡ALM migrated 100% if its nucleus was located dorsal or posterior to the P3/4 hypodermal nucleus and 0% if located adjacent to the BDU neuron, its sister cell. ¤HSN migrated 100% if its nucleus was located anterior to the V4 hypodermal nucleus and 0% if located posterior to the P11/12 hypodermal nucleus. ¦The left coelomocyte mother cell migrated 100% if the nuclei of its descendants, the ccsL, were posterior to the V3 hypodermal nucleus and 0% if located anterior to H2 hypodermal nucleus. We were unable to identify left coelomocytes in a number of kyIs5, cam-2, cam-1(gm122), syc-1, epi-1, unc-34 and mig-2 mutants. **Because they occupy positions near each other, the SDQ and AVM data for each strain were combined, and used to calculate an average distance migrated. SDQ and AVM migrated 100% if their nuclei were anterior to the two daughter nuclei of the V2 cell and 0% if posterior to the two daughter nuclei of the V4 cell. The position of AQR, a third QR descendant, is not reported because AQR normally migrates to positions near other neurons with similar morphology. ††The HSNs were rarely detected in kyIs5 hermaphrodites. When present, the HSNs were in their normal positions. Surprisingly, the HSNs were almost always detectable in fam-2 and epi-1 alleles. ‡‡Not determined. We were unable to score the positions of these cells by Nomarski optics. ¤¤CANs were not observed in ceh-10(gm58) by Nomarski optics. The CANs also failed to express GFP in ceh-10(gm58) mutants containing a ceh-23-gfp reporter. ¦¦These strains contained the kyIs5 transgene. alleles of these genes. The mig-2 allele gm38 is recessive, reporter (W. C. Forrester et al., unpublished data). The ceh-23- whereas both the new allele gm103 and the previously reported gfp transgene confers a subtle Cam phenotype, and although mig-2 allele rh17 are semidominant (W. C. Forrester et al., the transgene increases Cam severity and penetrance in most unpublished data; I. Zipkin and C. Kenyon, personal commu- of the mutants, syc-1, syc-2 and syc-3 mutants require the nication). Both unc-73 alleles reported here lead to extensive transgene to express a Cam defect. Because each of the syc lethality and sterility, phenotypes not produced by weaker unc- genes is represented by a single mutant allele, these genes 73 alleles (Run et al., 1996). could play critical roles in CAN cell migration that stronger The severity and penetrance of the Cam phenotype varied mutations would reveal. widely among the mutants (Table 1). Mutations in 10 of the 14 Cam genes resulted in severely misplaced CANs; the CANs Cell migration defects migrated on average less than 60% of their normal distances. To determine whether cell migrations other than those of the Mutations in fam-1 resulted in a weak Cam phenotype. The CANs were perturbed in the mutants, we examined the CANs of animals bearing the ceh-23-gfp reporter transgene positions of several cells that migrate during C. elegans devel- express GFP, and syc-1, syc-2 and syc-3 mutants were isolated opment. We examined the positions of the ALM mechanosen- as animals with displaced CANs in a strain bearing this sory neurons, the HSN motor neurons, the mesodermal coelo- C. elegans cell migration mutants 1835 mocytes (ccs), the M mesoblast and the Z1/Z4 somatic gonad as well as both pioneer growth cone migrations and growth precursor cells, all of which migrate during embryogenesis cone migrations along axonal bundles, contribute to these (Fig. 2A; Table 1; Sulston et al., 1983). We also examined the axonal morphologies. positions of the Q neuroblast descendants and the P cell After the CAN cell bodies have migrated, they extend axons descendants. The left and right Q neuroblasts and their descen- both anteriorly and posteriorly. In the mutants, anterior dis- dants migrate during the first larval stage (Fig. 2B; Sulston and placement of the CAN cell bodies complicated our analysis of Horvitz, 1977). The six bilaterally symmetric pairs of P cells CAN axonal outgrowth. To reach its normal destination in the cover the ventral third of newly hatched first larval stage (L1) tail, the posteriorly directed CAN axon from an anteriorly animals. Midway through the L1, the P cells migrate ventrally displaced CAN must extend further than the axon from a and intercalate to form a row of 12 cells at the ventral midline. normally positioned CAN. Because posteriorly directed CAN The cells then divide longitudinally to produce the Pn.a and axons in ina-1(gm144) and mig-2(gm103sd) mutants often Pn.p cells. The Pn.a neuroblasts divide during the L1 to extend to their normal destinations even when the cell bodies produce ventral nerve cord motor neurons, and some of the are in the head (Tables 1 and 2), CAN axonal defects observed Pn.p cells divide during the third larval stage to produce the in other Cam mutants probably reflect a direct requirement for hermaphrodite vulva. gene function in CAN axonal outgrowth. Our analyses revealed that none of the mutations disrupted Although all Cam mutations disrupted posteriorly directed the migrations of Z1 and Z4, and only the cam-2(gm124) CAN axonal outgrowth, the severity of such defects varied sub- mutation appeared to disrupt M migration or development. We stantially (Table 2). Mutations in ceh-10 and vab-8 severely were unable to detect M in cam-2(gm124) mutants by disrupted CAN axonal outgrowth, mutations in cam-1, cam-2, Nomarski optics, suggesting that it was either severely epi-1, unc-34, and unc-73 caused modest defects, and misplaced or failed to differentiate normally. Therefore, data mutations in the remaining genes had little effect. for the M, Z1 and Z4 migrations are not presented in Table 1. Each laterally positioned HSN cell body extends an axon The left coelomocyte mother cell (the left and right coelomo- ventrally along the epithelium to the ventral nerve cord, a cyte mother cells each divide to produce two coelomocytes) paired bundle of axons that runs the length of the animal. Once migrates further than the right coelomocyte mother cell in the cord, each axon turns anteriorly and extends along the (Sulston et al., 1983). Similarly, the right Q neuroblast and its ipsilateral axon bundle to the nerve ring, the major neuropil descendants migrate further than the left Q and its descendants that encircles the pharynx (Fig. 3A-C). At the vulva, each HSN (Sulston and Horvitz, 1977). Because cells that migrate further axon defasciculates from the ventral nerve cord and branches are more severely affected in the mutants, we report here only (Fig. 3A-C; White et al., 1986; Garriga et al., 1993). our analyses of the left coelomocyte mother cell and the right The mutants exhibited a range of HSN axonal outgrowth Q and its descendants (Table 1). defects. In many mutants, the HSN cell bodies were displaced Mutations in the nine genes cam-1, epi-1, fam-1, fam-2, ina- posteriorly, consistent with the results obtained by Nomarski 1, mig-2, unc-34, unc-73 and vab-8 affect the migrations of optics (Fig. 3D; Table 1). Because severe posterior displace- many cells in addition to the CANs (Table 1). Mutations in the ment of the HSN cell bodies produces secondary defects in genes epi-1, mig-2 and unc-73 produce the most widespread ventrally directed outgrowth, fasciculation in the ventral nerve defects, affecting all cell migrations scored except those of M, cord and branching (Garriga et al., 1993), only observations for Z1 and Z4. In addition, epi-1, mig-2 and unc-73 mutations also HSNs that had migrated at least 75% of the wild-type distance disrupt the ventral migrations of the 12 hypodermal P cells are reported in Table 2. during the first larval stage. Defects in the ventral migrations The HSN axons of fam-2, mig-2, unc-34 and unc-73 mutants of the P cells produced by epi-1, mig-2 and unc-73 result in terminated prematurely in the ventral nerve cord before mutant animals with ventral nerve cord motorneurons along reaching the nerve ring, a phenotype previously described for their sides and hermaphrodites with laterally positioned vulvae mig-2, unc-73 and unc-34 mutants (Desai et al., 1988; McIntire (data not shown). As expected, anti-GABA staining revealed et al., 1992; Table 2; Fig. 3E). The axons of normally posi- that the VD motor neurons, descendants of the Pn.a neuro- tioned HSNs often failed to branch at the vulva in epi-1, fam- blasts, are found at ectopic lateral positions in these mutants 2, ina-1 and mig-2 mutants (data not shown). (Table 2; Fig. 4D). cam-1 mutations perturb the positioning of We also examined the axonal morphology of several the CANs, ALMs, HSNs and ccsL, which normally migrate to GABAergic neurons. The DD motor neurons are generated positions near the middle of the animal (Table 1; see below). during embryogenesis, and the VD motor neurons are generated during the L1. Both D-type motor neurons extend Axonal outgrowth is disrupted in CAN migration anteriorly directed processes along the ventral nerve cord that mutants branch near their anterior ends. Each branch produces a com- Cell migration and axonal outgrowth are similar processes that missure that extends dorsally along the lateral epithelium. require many of the same genes (for example, see Hedgecock Upon reaching the dorsal midline, the commissures branch and et al., 1987). To determine whether our mutations disrupt extend posteriorly and anteriorly directed processes along the growth cone migrations, we examined CAN axonal morphol- dorsal nerve cord. Most of the commissures extend along the ogy using the ceh-23-gfp reporter, HSN axonal morphology by right side (McIntire et al., 1992; Fig. 4A-C). anti-serotonin staining, and axonal morphology of the AVL, Dorsally directed commissural axons of epi-1, fam-1, mig- DVB, DD and VD motor neurons by anti-GABA staining. 2, unc-34 and unc-73 mutants often extended abnormally. The Anteriorly directed (CAN, HSN and DVB), posteriorly commissural axons of these mutants often branched, stopped directed (AVL and CAN), ventrally directed (DVB and HSN) prematurely, or extended longitudinally along the lateral body and dorsally directed (DD and VD) growth cone migrations, wall, failing to reach the dorsal midline (Table 2; Fig. 4G). 1836 W. C. Forrester and G. Garriga

Table 2. Axon defects in Cam mutants Cell and axon defects HSN‡ GABA-expressing neurons¤ CAN axon Stop Mis- Extend Cells Dorsal Comm. Strain migration† short¦ directed together lateral defective on left Defasc. n Wild type ÐÐÐ Ð (27) Ð (27) Ð (27) Ð Ð Ð Ð 14 kyIs5 ** 99.7±1.2 (67) Ð (1) +++ (3) na†† Ð Ð + +/Ð 31 unc-73(gm67)‡‡ 65.0±18.5 (31) na†† ++++ (2) na†† ++++ ++++ ++++ ++++ 6 unc-73(gm123)‡‡ 72.7±14.2 (29) ++++ (1) ++++ (8) na†† ++++ ++++ ++++ ++++ 8 cam-2(gm124) 63.1±30.5 (34) Ð (28) Ð (28) +++ (14) Ð Ð ++ Ð 18 syc-3(gm135) 89.3±9.9 (39) Ð (26) + (28) ++++ (12) Ð Ð + Ð 16 cam-1(gm105) 85.9±19.4 (57) Ð (27) Ð (27) + (11) Ð Ð + +/Ð 20 cam-1(gm122) 74.5±23.6 (26) +/Ð (40) +/Ð (42) + (18) Ð Ð + +/Ð 20 ceh-10(gm58)¤¤ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ÐÐÐ ceh-10(gm120) 45.9±30.8 (57) Ð (26) Ð (26) + (12) Ð Ð + Ð 19 ceh-10(gm127) 54.2±28.3 (52) Ð (24) Ð (24) Ð (12) Ð Ð +/Ð Ð 20 syc-2(gm132) 94.8±14.1 (55) + (27) +/Ð (28) + (13) Ð Ð + Ð 16 fam-1(gm85) 95.4±6.1 (29) Ð (30) +/Ð (31) ++++ (13) ++ ++ +++ ++ 19 ina-1(gm144) 88.9±17.9 (31) Ð (11) Ð (11) +++ (3) + + +++ + 20 syc-1(gm126) 96.9±8.3 (46) Ð (32) +/Ð (33) + (18) Ð Ð +++ Ð 19 fam-2(gm94) 86.7±9.7 (47) ++++ (5) + (6) +++ (4) ++ + +++ ++ 17 epi-1(gm57) 78.4±21.7 (30) + (15) + (16) ++++ (5) +++ + ++++ ++ 16 epi-1(gm146) 61.2±24.6 (36) + (17) + (20) +++ (5) ++++ + ++++ +++ 13 epi-1(gm121) 78.9±25.9 (27) Ð (28) Ð (28) +++ (13) + +++ ++ + 16 epi-1(gm139)‡‡ 69.7±17.9 (25) Ð (3) Ð (3) ++++ (1) ++++ ++++ ++++ +++ 4 unc-34(gm104) 75.7±15.8 (29) ++ (19) Ð (19) ++++ (9) + ++ +++ +++ 21 unc-34(gm114) 70.3±18.7 (53) ++ (17) Ð (17) ++++ (5) + ++++ + ++ 17 unc-34(gm115) 77.4±18.8 (60) ++ (35) Ð (35) ++++ (15) + + ++ +++ 20 unc-34(gm134) 85.1±12.3 (30) +++ (33) Ð (34) ++++ (16) + ++ ++ +++ 21 vab-8(gm99) 19.0±24.6 (42) Ð (28) Ð (28) +++ (13) + Ð + Ð 21 vab-8(gm138) 21.2±19.8 (47) Ð (25) Ð (25) ++ (12) Ð Ð + +/Ð 20 mig-2(gm38) 81.4±15.5 (34) +++ (11) +++ (18) ++++ (3) ++ ++ ++++ + 16 mig-2(gm103) 85.5±13.6 (54) ++++ (5) ++++ (5) Ð (2) +++ ++ +++ ++ 27

*Data for the HSN, DD and VD axons are presented as symbols: −, no defects; +/−, 5% or less defective; +, 5-24.9% defective; ++, 25-49.9% defective; +++, 50-74.9% defective, and ++++, >75% defective (see Materials and Methods). Few unc-73(gm67), unc-73(gm123) and epi-1(gm139) mutant animals were identified after staining, and therefore little data was obtained for these mutants. †Data for the CAN axon are presented as percentage migrated ± s.d. CAN axonal morphology was examined in mixed stage animals bearing the ceh-23-gfp reporter. The position of the terminus of the posterior CAN axon was determined. This position was expressed as the percentage of the distance from the point at which the CAN cell would be located if it failed to migrate (0%) to the point at which it normally terminates in wild-type animals (100%). None of the mutations substantially affected anterior CAN axonal outgrowth (data not shown). Numbers in parentheses are number of CAN axons scored. ‡HSN axonal morphology was determined in adult hermaphrodites by anti-serotonin staining. HSN axons were examined for defects in their ability to extend to the nerve ring (Stop short), for defects in the ability of the HSN axons to extend along their normal pathway (Misdirected), and for defects in the ability of both HSN axons to extend separately along the ventral nerve cord (Extend together). Axons were scored as stopping short if they did not extend to the nerve ring. Axons were scored as misdirected if they did not project to the ventral nerve cord and turn anteriorly. Axons were scored as extending together if both axons ran together as a single fascicle in the ventral nerve cord. Because severe displacement of HSN cell bodies can cause secondary defects in HSN axonal outgrowth, we only included results for HSNs that had migrated more than 75% of the distance migrated by wild-type HSNs. Numbers in parentheses are number of sides (Stop short or Misdirected) or animals (Extend together) scored. ¤AVL, DVB, DD and VD axonal morphologies were examined in late-larval stage and adult hermaphrodites by anti-GABA staining. Hermaphrodites were examined for defects in VD cell body positioning (Cells lateral), for defects in dorsal extension of commissures (Dorsal defective), for defects in commissures sidedness (Comm. on left), and for defects in axon fasciculation (Defasc.). VDs were scored as mispositioned if they were not at the ventral midline. Commissures were scored as defective if they extended dorsally along the left side. The GABAergic neurons were scored as defasiculated if the axons split into more than one bundle. Commissures were scored as dorsal defective if they failed to extend to the dorsal nerve cord prior to branching, or if they extended longitudinally along the lateral body wall. Numbers of animals scored are shown in the last column (n). ¦The number of HSN axons scored as stopping short can be an underestimate of the true value. Because HSN axons of mutants that exhibited the stopping short phenotype often extended together as a single fascicle, if only one of the axons extended to the nerve ring, both axons would be scored as reaching the nerve ring. **HSNs were rarely detected in the kyIs5 strain. Although kyIs5 was present in fam-2 and epi-1(gm146, gm121, gm139) animals, these mutants did not display a significant HSN-misdirected phenotype. ††Not applicable. No HSNs scored had migrated more than 75% of the distance migrated by wild-type HSNs (stop short and misdirected), or both HSNs of an animal failed to migrate more than 75% of the distance migrated by wild-type HSNs (extend together). ‡‡We scored unc-73 and epi-1(gm139) homozygotes from heterozygous mothers. unc-73 homozygotes from homozygous mothers rarely reach adulthood, and epi-1(gm139) homozygotes are sterile. ¤¤CAN axons were not detected in ceh-10(gm58). In addition, HSN and GABA-expressing neurons were not examined because the mutants die at the first or second larval stage.

Similarly, the HSN axons of unc-73 and mig-2 mutants (Table priately along the left side in at least 25% of animals (Table 2; 2; Fig. 3G) and the CAN axons of mig-2, epi-1 and unc-73 Fig. 4E). mutants (data not shown) often extended along aberrant The ventral nerve cord often was disorganized in epi-1, fam- pathways. In 16 mutants, the commissures extended inappro- 1, fam-2, mig-2, unc-34 and unc-73 mutants. In wild-type C. elegans cell migration mutants 1837

HSN axon branch

HSN cell body nerve ring HSN axons vulva

Fig. 3. HSN axonal morphology in wild- type and mutant hermaphrodites. Anterior is to the left and dorsal is up. (A) Schematic left oblique view of an adult hermaphrodite. The left HSN cell body and both HSN axons in the ventral nerve cord are shown. (B-H) Immunoﬂuorescence photomicrographs of wild-type and mutant adult hermaphrodites stained with an anti- serotonin antiserum. In all panels, positions of the left HSN cell body (small arrowhead), HSN axons (arrows), and vulva (large arrowhead) are indicated. (B,C) Oblique views of the same wild-type adult hermaphrodite. Midbody (B) and posterior body (C) are shown. In B, the right HSN axon is also visible as it extends parallel to the left axon. (D) Left lateral view of an epi-1(gm57) hermaphrodite tail. The HSN cell body is located posterior to its normal position. The vulva is off the panel to the left. (E) Left lateral view of the midbody of an unc-34(gm134) adult hermaphrodite. The arrow indicates the position in the ventral nerve cord where the HSN axon terminates prematurely (Stop short phenotype in Table 2). (F) Slightly oblique view of an unc-34(gm134) adult hermaphrodite. The arrow indicates the position where the left and right HSN axons join in the ventral nerve cord (Extend together phenotype in Table 2). (G) Left lateral view of the midbody of a mig- 2(gm38) adult hermaphrodite. The HSN cell body is located in its normal position, but the axon extends anteriorly along the lateral body wall (one class of Misdirected phenotype in Table 2). (H) Left lateral view of a cam-1(gm122) adult hermaphrodite. The HSN cell body has migrated beyond its normal destination and is located anterior to the vulva. As shown here, the axons of anteriorly displaced HSNs often extend anteriorly along the lateral body wall. Scale bar indicates 50 µm. animals, AVL and DVB also extend axons along the right anteriorly as a single fascicle (Table 2, Fig. 3F). This HSN ventral nerve cord (McIntire et al., 1992). Anti-GABA defect and the GABAergic defasciculation defects seen in staining reveals that all four GABAergic axons extend near these mutants may result from general defects in ventral nerve each other along the right bundle (Fig. 4B). The ventral nerve cord fasciculation. cords of 12 mutants representing six genes defasciculate, splitting into additional bundles that separate widely (Table 2; Anterior displacement of the HSNs Fig. 4F). This defasciculation was common in epi-1, mig-2, In Cam mutants, the HSNs either migrated incompletely and unc-34 and unc-73 mutants and was occasionally seen in fam- on average were displaced posteriorly, or they migrated exces- 1 and fam-2 mutants. Instead of extending as two separate sively and on average were displaced anteriorly. Speciﬁcally, axons, the HSNs of these mutants often extended both axons mutations in cam-1, cam-2, ceh-10 and vab-8 cause HSNs to 1838 W. C. Forrester and G. Garriga

Fig. 4. GABAergic neuron process A. morphology in wild-type and mutant hermaphrodites. Anterior is to the left. (A) Schematic right lateral view of an adult hermaphrodite showing cell bodies and processes of the DD, VD, AVL and DVB motor neurons. GABAergic RME and RIS AVL neurons are not shown. (B-H) Immunofluorescence photomicrographs of wild-type and mutant hermaphrodites stained with an anti-GABA antiserum. (B) Right lateral view of a fourth larval stage wild-type hermaphrodite. The ventral nerve cord processes (arrowheads) and two commissures (arrows) are indicated. The dorsal nerve cord is out of the plane of focus. (C) Ventral view of a wild-type fourth larval stage hermaphrodite. The ventral nerve cord processes (arrowheads) and two commissures as they exit the ventral cord (arrows) are indicated. (D) Left lateral view of a unc-73(gm123) adult hermaphrodite. A VD cell body is misplaced laterally (large arrow) from its normal position in the ventral nerve cord, which is out of focus (arrowheads indicate the ventral side of the animal; cells Lateral phenotype in Table 2). The commissures of unc-73 mutants often extended aberrantly along the lateral body wall (small arrow). (E) Ventral view of a mig-2(gm103sd) adult hermaphrodite. Commissure exits the ventral nerve cord (arrowheads) on left side (arrow; Comm. on left phenotype in Table 2). (F) Ventral view of an unc-34(gm115) adult hermaphrodite. The GABAergic neurons of the ventral nerve chord split into two widely separated bundles (arrowheads; Defasc. phenotype in Table 2). (G) Right lateral view of third larval stage unc-34(gm104) hermaphrodite. Commissural axon growing along lateral body wall (arrow; Dorsal defective phenotype in Table 2). Scale bar indicates 50 µm. be displaced anteriorly. In the remaining Cam mutants the nations. For example, in cam-1(gm122) mutants, the HSNs HSNs were displaced posteriorly (Table 1). In cam-2, ceh-10 migrated beyond their normal destinations 72% of the time and vab-8 mutants, the HSNs migrate to positions anterior to (Figs 3H, 5E). Taken together, these data suggest that cam-1 their normal destinations approximately 10-28% of the time. functions to define the final destinations of ALMs, CANs, ccsL The anterior displacements of the HSNs in ceh-10(gm58) and HSNs. mutants shown in Fig. 5C are typical for mutants with missing or severely displaced CANs. Postembryonic development requires CAN function To test the hypothesis that misplaced or missing CANs cause Killing the CANs in newly hatched L1 hermaphrodites by the anterior displacement of the HSNs, we killed the CAN laser microsurgery caused animals to die at various stages precursor cells in wild-type embryos. In three of 16 laser- during larval development (J. Sulston, personal communica- operated embryos (18.8%) where we killed a single CAN tion; Materials and Methods). Observing the laser-operated precursor cell, the HSN on the side lacking CAN migrated animals by Nomarski optics revealed that their pseudo- beyond its normal destination (Fig. 5D). In 37 laser-operated coeloms had filled with fluid, causing the animals to appear controls where we killed a cell adjacent to the CAN precursor, clear (Clr) or translucent. Both ceh-10 and cam-2(gm124) the HSN migrated to its normal destination. These results mutants also display this Clr phenotype, and animals bearing confirm that the CANs play a role in final positioning of the the strongest ceh-10 mutations lack CANs and arrest as first- migrating HSNs. or second-stage Clr larvae (Fig. 6B). These observations In cam-1 mutants, the ALMs, CANs, ccsL and HSNs occa- indicate that the CAN cells are necessary for postembryonic sionally migrated incompletely. The HSNs, and to a lesser development in C. elegans and may function in osmoregula- extent the ALMs, can also migrate beyond their normal desti- tion. C. elegans cell migration mutants 1839

A Manser and Wood (1990) proposed that the withered tail (Wit) phenotype displayed by Cam mutants resulted from a lack of CAN function in the posterior body region (Fig. 6D). Similar phenotypes displayed by Wit and Clr mutants, such as B the ability to detect the boundaries of cells by Nomarski optics, support this hypothesis. Moreover, weak ceh-10 mutants display a Wit phenotype, and strong ceh-10 mutants display a Clr lethal phenotype, supporting the hypothesis that the two phenotypes are related. Manser and Wood (1990) also proposed that the anterior dis- C placement of the CAN cell body in Cam mutants caused a lack of CAN function in the posterior body and the resulting Wit phenotype. The results of our study, however, reveal that several mutants with severely displaced CAN cell bodies do not display the Wit phenotype (Fig. 7A). To test whether the D Wit phenotype might result from defects in posteriorly directed CAN axon outgrowth, we compared the penetrance of the axon defect to the penetrance of the Wit defect. As shown in Fig. 7B, we note a correlation between the penetrance of the posterior CAN axon outgrowth defect and Wit defects of E different mutants. Therefore, the Wit phenotype is likely to result from failure of the posterior CAN axon to grow to its normal destination.

Fig. 5. CAN and HSN cell body positions in wild-type and mutant DISCUSSION hermaphrodites. (A) Diagram of a newly hatched larva. Long vertical lines indicate positions of landmark V cell nuclei, and short vertical We have identified 14 Cam genes that are required for C. lines indicate positions of P cell nuclei. (B-E) CAN (solid) and HSN elegans cell migrations. We find that these genes represent (cross-hatched) positions. y-axes on the left represent the number of three classes: genes required for cell fate specification, genes CANs and HSNs. (B) Cell positions in wild-type hermaphrodites. required for multiple cell and growth cone migrations, and a (C) HSN positions in ceh-10(gm58) mutants. This mutant lacks single gene required for cell positioning (Fig. 8). In addition, detectable CANs. (D) HSN positions in wild-type animals that lack CANs because the CAN precursors were killed by laser analysis of the Cam mutants demonstrates that the CANs are microsurgery. (E) Cell positions in cam-1(gm122) mutants. Cell required for proper positioning of the HSNs. Finally, we show positions were determined by Nomarski optics as described in the that the CAN cells are required for larval viability in C. legend to Table 1. elegans.

Fig. 6. Clr and Wit phenotypes. In all panels, anterior is to the left and dorsal is up. (A) Left lateral view of a wild-type ﬁrst stage larva viewed by Nomarski optics. Note cell bodies are not visible. (B) Left lateral view of a ceh-10(gm58) larva viewed by Nomarski optics. The distinct outlines of cell bodies are a manifestation of the Clr phenotype. (C) Lateral view of an adult wild-type hermaphrodite. The girth of the region posterior to the vulva (arrowhead) is the same as that anterior to the vulva. (D) Lateral view of a vab- 8(gm99) hermaphrodite. The girth of the region posterior to the vulva (large arrowhead) is reduced compared to that anterior to the vulva (the Wit phenotype). Wit animals often display abnormal vulval development; they produce protruding vulvae (large arrowhead) and can produce ectopic ventral protrusions consisting of vulval cells (small arrowhead) in the withered region (Manser and Wood, 1990; Wightman et al., 1996). Scale bars, 20 µm (A and B) and 50 µm (C and D). 1840 W. C. Forrester and G. Garriga

A. 90 B B 80 Differentiation ceh-10 70 Migration Head epi-1 60 fam-1 fam-2 50 ina-1 Tail mig-2 40 B unc-34

percent Wit unc-73 30 vab-8 B B 20 B B Positioning cam-1 10 B B B B 0 BB B B B Fig. 8. Cam genes define three steps in cell migration. The diagram 0 102030405060708090100 is similar to Fig. 1A, where only one of the two bilaterally symmetric percent migrated (CAN cell) CANs is shown, and the arrow indicates the CAN migration route. B. 100 We propose that the ceh-10 gene specifies CAN cell fate B (Differentiation), that eight genes function directly in cell migration 90 (Migration) and that cam-1 functions in final positioning of 80 migrating cells (Positioning). We have not assigned four genes, cam- 70 2, syc-1, syc-2 and syc-3, to any of these categories. 60 B 50 the neurotransmitter serotonin (Desai et al., 1988; Wang et al., 40 B

percent Wit B 1993; Chisholm, 1991; Baumeister et al., 1996). Similarly, ceh- 30 10 functions in the CANs to regulate their migrations and their B ability to express ceh-10 and ceh-23 (W. C. Forrester et al., 20 BB B B B BB unpublished data). Thus, the two homeodomain proteins CEH- 10 10 and EGL-5 may serve analogous functions during CAN and BBB 0 BB HSN development, respectively. 0 102030405060708090100 cam-2 may also play a regulatory role in CAN cell percent migrated (posterior CAN axon) migration, but unlike ceh-10, ceh-23 expression does not appear to require cam-2 (data not shown). Consistent with the Fig. 7. Withered tail phenotype correlates with defects in posterior hypothesis that cam-2 plays a regulatory role during develop- CAN axon outgrowth. Each data point represents Wit phenotype and CAN cell body or axon position for a specific Cam mutant. (A) Plot ment, cam-2 mutants display a low penetrance multivulval of average CAN cell body position in Cam mutant strains (x-axis) phenotype and may lack differentiated M mesoblasts, pheno- versus percent Wit (y-axis). (B) Plot of average location of posterior types that could result from cell fate transformations. terminus of CAN axon in Cam mutant strains (x-axis) versus percentage Wit (y-axis). Average cell body or CAN axon positions Eight genes function in multiple cell and growth were determined as described in the legends to Tables 1 and 2. cone migrations Percentage Wit was determined by counting Wit and non-Wit adults In contrast to mutations in ceh-10 and cam-2, which specifi- bearing (B) or lacking (A) the ceh-23-gfp transgene. cally affect CAN migrations, we found that mutations in epi- 1, ina-1, fam-1, fam-2, mig-2, unc-34, unc-73 and vab-8 disrupt the migrations of multiple cells and growth cones. Cell The ceh-10 homeobox gene specifies CAN fate migration and axonal outgrowth defects have been described The gene ceh-10, which encodes a homeodomain protein that previously for mig-2, unc-34, unc-73 and vab-8 (Hedgecock et is expressed in the CANs (Svendsen and McGhee, 1995), al., 1985, 1987; Desai et al., 1988; Manser and Wood, 1990; regulates CAN cell fate. In weak ceh-10 mutants, the CAN cell McIntire et al., 1992; Wightman et al., 1996). The first epi-1 bodies and growth cones fail to migrate normally. In strong and ina-1 mutants were isolated on the basis of defects in ceh-10 mutants, the CANs do not express the ceh-23-gfp epithelialization (E. Hedgecock, personal communication) and transgene (data not shown). We propose that CEH-10 tran- HSN migration (P. Baum, D. Parry and G. G., unpublished scriptionally regulates other genes that function directly in data), respectively. CAN differentiation. CEH-10 could directly regulate genes Molecules that direct cell and growth cone migrations can involved in CAN cell migration or act through other regulatory guide the migrations along a specific body axis or in a specific genes such as ceh-23, which encodes a homeodomain protein direction. For example, C. elegans unc-6 guides migrations expressed in the CANs (Wang et al., 1993). both dorsally and ventrally, whereas unc-5 guides only dorsally Several genes that regulate HSN differentiation are directed migrations (Hedgecock et al., 1990; McIntire et al., necessary for HSN migration (Desai et al., 1988). The homeo- 1992). Similarly, vab-8 guides posteriorly directed migrations domain gene egl-5, for example, functions in the HSNs to along the longitudinal body axis (Wightman et al., 1996), and regulate their migrations, their ability to express the HSN consistent with these results, we find that vab-8 functions in differentiation protein UNC-86, and their ability to synthesize posteriorly directed migrations. We find that the seven other C. elegans cell migration mutants 1841 genes of this class are neither body axis- nor directional- is not expressed in the HSNs, or in any cells near the HSNs specific. Mutations in these genes disrupt both circumferential other than CANs (Svendsen and McGhee, 1995). Finally, and longitudinal migrations (Tables 1 and 2). killing the CAN precursors also caused the HSNs to migrate Pleiotropic defects in cell migration and axonal outgrowth beyond their normal destinations. The simplest interpretation caused by mutations in these eight genes suggest that they of these results is that altered CAN cell body positions result encode products that function directly in cell and growth cone in anteriorly displaced HSN cell bodies. migrations. Recent molecular analysis of three of these genes In our model, the CANs express a secreted or membrane- supports this view. The gene epi-1 encodes a laminin α chain bound cue that inhibits HSN migration. Since most HSNs stop homolog (K. Joh, D. Hall, J. Yochem, I. Greenwald and E. at their normal destinations in cam-1, cam-2, ceh-10, mig-11, Hedgecock, personal communication). Laminin is a het- unc-39 and vab-8 mutants, additional mechanisms must con- erotrimeric extracellular matrix molecule that functions in cell tribute to the termination of HSN migration. and growth cone migrations (for review see Lander, 1989). The gene ina-1 encodes an α integrin subunit (P. Baum and G. The cam-1 gene functions in final positioning of Garriga, in preparation). Integrins are α/β heterodimeric cell migrating cells surface receptors that function in cell adhesion events The cam-1 mutants differ from other mutants with incomplete necessary for cell and growth cone migrations (for review see CAN and excessive HSN migrations. In cam-1(gm122) Hynes, 1992). In several organisms, specific integrin receptors mutants, for example, the HSNs migrate too far anteriorly 72% act as receptors for laminin, raising the possibility that the of the time, compared to 10-28% for other mutants with products of the epi-1 and ina-1 genes interact (reviewed in excessive HSN migrations. In cam-1 mutants, the higher Hynes, 1992). Finally, unc-73 encodes a protein that is related frequency of excessive HSN migrations does not result from to a guanine nucleotide exchange factor for the Rho subfamily more severe CAN migration defects, since ceh-10 and vab-8 of Ras related G-proteins (Run et al., 1996). Rho family mutations completely disrupt CAN migrations. In addition to members include the yeast protein Cdc42 and mammalian being excessive, HSN migrations in cam-1 mutants can also be proteins Cdc42, Rac and Rho. Yeast Cdc42 functions in actin incomplete (Fig. 5E). Moreover, ALM migrations can also be assembly during budding (Johnson and Pringle, 1990), and the incomplete or excessive in cam-1 mutants, although the ALM mammalian proteins function in filopodial and lamellipodial defects are less severe than the HSN defects (data not shown). formation during fibroblast migration, and in stress fiber In summary, cells appear to migrate to their approximate, but formation during fibroblast attachment (Ridley and Hall, 1992; not precise, destinations in cam-1 mutants. These results Ridley et al., 1992; Nobes and Hall, 1995). In fibroblasts, suggest that cam-1 acts to define the final positions of several activated Rac causes integrin receptor aggregation (Hotchin migrating cells. and Hall, 1995), raising the possibility that unc-73 and ina-1 Cell interactions between the migrating sex myoblasts function in the same pathway. (SMs) and cells of the C. elegans hermaphrodite somatic gonad The fam-1, fam-2, mig-2, unc-34 and vab-8 gene products define the precise destinations of the SMs (Thomas et al., have not yet been reported. Given the molecular identities of 1990). The SMs migrate anteriorly from positions in the other members of this group, it seems likely that these genes posterior body to the precise center of the gonad during the will encode products that function directly in cell migration third larval stage. Killing hermaphrodite somatic gonad pre- and axonal outgrowth. cursors by laser microsurgery disrupts precise SM positioning, but not the ability of SMs to migrate anteriorly to their approx- CANs are required to properly position the HSNs imate destinations. Thomas et al. (1990) proposed that a signal As noted by Manser and Wood (1990) for the mig-11, unc-39 emanating from the somatic gonad attracts the SMs to their and vab-8 Cam mutants, the HSNs occasionally migrated precise destinations. By analogy, the ALMs, CANs, ccsL and beyond their normal destinations. We have extended these HSNs could initially follow general directional cues, but then observations to cam-1, cam-2 and ceh-10 mutants. migrate to precise positions by following more specific cues Based on incomplete CAN migrations and excessive HSN emanating from cells near the center of the embryo. The cam- migrations, Manser and Wood (1990) argued that positional 1 gene could encode or regulate such a specific cue, or it could information used by migrating cells was shifted anteriorly in function in the interpretation of such a cue by the migrating these mutants. In this model, altered guidance information cells. causes excessive HSN migrations. We favor an alternate hypothesis, in which the CANs stop the HSNs from migrating CAN may function in osmoregulation beyond their normal destinations. First, EM reconstructions of The CANs are essential for larval development. Killing the embryos from serial sections has shown that after migrating, CANs in larvae by laser microsurgery caused the animals to the CAN and HSN cell bodies occupy positions adjacent to one arrest at various stages of development. This variability pre- another (Durbin, 1987). This proximity could facilitate cell sumably reflects differences in the effectiveness of CAN interactions between the CANs and HSNs. Second, no Cam killing, since five different strong ceh-10 mutations, which mutant displays normal HSN migrations. For each mutant, on produce animals that lack detectable CANs, all result in early average the HSNs either fail to migrate to or migrate beyond larval arrest. In contrast to larval development, embryonic their normal destinations. It seems unlikely that mutations in development proceeds at the normal rate in ceh-10 mutants. cam-1, cam-2, ceh-10, mig-11, unc-39 and vab-8 all disrupt Both the laser-operated animals and ceh-10 mutants express positional information to cause incomplete CAN and excessive a Clr phenotype: the pseudocoelom of these animals fills with HSN migrations. In particular, mutations in ceh-10 disrupt fluid, causing the animals to appear clear or translucent. This CAN migration indirectly by altering CAN cell fate. CEH-10 phenotype suggests that the CANs function in an osmoregula- 1842 W. C. Forrester and G. Garriga tion process that operates during larval development. The CAN Garriga, G., Desai, C. and Horvitz, H. R. (1993). Cell interactions control the axons are associated with the excretory canal, a structure direction of outgrowth, branching and fasciculation of the HSN axons of involved in osmoregulation (White et al., 1986; Nelson and Caenorhabditis elegans. Development 117, 1071-87. Goodman, C. S. and Shatz, C. J. (1993). Developmental mechanisms that Riddle, 1984). Perhaps the CANs regulate the activity of the generate precise patterns of neuronal connectivity. Cell 72, 77-98. excretory canal. Hall, A. (1994). Small GTP-binding proteins and the regulation of the actin cytoskeleton. Annu. Rev. Cell Biol. 10, 31-54. Cam genes define three steps in cell migration Hawkins, N. C. and McGhee, J. D. (1990). Homeobox containing genes in the Analysis of the Cam mutants indicates that three genetically nematode Caenorhabditis elegans. Nucleic Acids Res. 18, 6101-6. Hedgecock, E. M., Culotti, J. G. and Hall, D. H. (1990). The unc-5, unc-6 and defined steps are necessary for cell migrations (Fig. 8). First, unc-40 genes guide circumferential migrations of pioneer axons and cells decide to migrate. For the CANs, the gene ceh-10, either mesodermal cells on the epidermis in C. elegans. Neuron 4, 61-85. directly or through other genes, controls this decision to Hedgecock, E. M., Culotti, J. G., Hall, D. H. and Stern, B. D. (1987). migrate. Second, cues guide the cell along its route. The gene Genetics of cell and axon migrations in Caenorhabditis elegans. Development 100, 365-82. vab-8 guides cells and growth cones posteriorly (Manser and Hedgecock, E. M., Culotti, J. G., Thomson, J. N. and Perkins, L. A. (1985). Wood, 1990; Wightman et al., 1996). Seven additional genes, Axonal guidance mutants of Caenorhabditis elegans identified by filling epi-1, fam-1, fam-2, ina-1, mig-2, unc-34 and unc-73, function sensory neurons with fluorescein dyes. Dev. Biol. 111, 158-70. in multiple cell and growth cone migrations, and products of Hosono, R. Hirahara, K., Kuno, S. and Kurihra, T. (1982). Mutants of three of these genes function directly in cell and growth cone Caenorhabditis elegans with dumpy and rounded head phenotype. J. Exp. Zool. 224, 135-44. migrations as extracellular matrix molecules, cell surface Hotchin, N. A. and Hall, A. (1995). The assembly of integrin adhesion receptors and signal transduction molecules. Third, extracellu- complexes requires both extracellular matrix and intracellular rho/rac lar cues define the final destinations of migrating cells. A cue GTPases. J. Cell Biol. 131, 1857-65. from the CAN appears to stop the migrating HSN. In addition, Hynes, R. O. (1992). Integrins: versatility, modulation, and signaling in cell adhesion. Cell 69, 11-25. the gene cam-1 may function to define the final destinations of Johnson, D. I. and Pringle, J. R. (1990). Molecular characterization of several migrating cells. Further molecular analysis of the genes CDC42, a Saccharomyces cerevisiae gene involved in the development of that function in multiple cell migrations should lead to cell polarity. J. Cell Biol. 111, 143-52. important insights into how the pathways and destinations of Keino-Masu, K., Masu, M., Hinck, L., Leonardo, E. D., Chan, S. S.-Y., migrating cells are defined. Culotti, J. G. and Tessier-Lavigne, M. (1996). Deleted in Colorectal Cancer (DCC) encodes a netrin receptor. Cell 87, 175-85. Kennedy, T. E., Serafini, T., de la Torre, J. R. and Tessier-Lavigne, M. We thank Jen Zallen and Cori Bargmann for providing the kyIs4 (1994). Netrins are diffusible chemotropic factors for commissural axons in and kyIs5 reporter strains. We thank Dianne Parry and Paul Baum for the embryonic spinal cord. Cell 78, 425-35. providing fam-1(gm85) and mig-2(gm38) strains, and Kate Guenther Lander, A. D. (1989). Understanding the molecules of neural cell contacts: for help with the identification of embryonic cells. We are grateful to emerging patterns of structure and function. Trends Neurosci. 12, 189-95. Paul Baum, Nancy Hawkins and Jen Zallen for critical reading of this Lauffenburger, D. A. and Horwitz, A. F. (1996). Cell migration: a physically manuscript and to Claire Walczak for artistic assistance. This work integrated molecular process. Cell 84, 359-69. was supported by NIH grant NS32057 to G. G. W. C. F. was supported Leung-Hagesteijn, C., Spence, A. M., Stern, B. D., Zhou, Y., Su, M.-W., by postdoctoral fellowships from the National Institutes of Health and Hedgecock, E. M. and Culotti, J. G. (1992). UNC-5, a transmembrane from the Breast Cancer Research Program of the US Army Medical protein with immunoglobulin and thrombospondin type 1 domains, guides cell and pioneer axon migrations in C. elegans. Cell 71, 289-99. Research and Material Command. Luo, Y. and Raper, J. A. (1994). Inhibitory factors controlling growth cone motility and guidance. Curr. Opin. Neurobiol. 4, 648-54. Manser, J. and Wood, W. B. (1990). Mutations affecting embryonic cell REFERENCES migrations in Caenorhabditis elegans. Dev. Genet. 11, 49-64. McIntire, S. L., Garriga, G., White, J., Jacobson, D. and Horvitz, H. R. Austin, J. and Kimble, J. (1989). Transcript analysis of glp-1 and lin-12, (1992). Genes necessary for directed axonal elongation or fasciculation in C. homologous genes required for cell interactions during development of C. elegans. Neuron 8, 307-22. elegans. Cell 58, 565-71. Mitchison, T. J. and Cramer, L. P. (1996). Actin-based cell motility and cell Avery, L. and Horvitz, H. R. (1987). A cell that dies during wild-type C. locomotion. Cell 84, 371-9. elegans development can function as a neuron in a ced-3 mutant. Cell 51, Nelson, F. K. and Riddle, D. L. (1984).Functional study of the Caenorhabditis 1071-78. elegans secretory-excretory system using laser microsurgery. J. Exp. Zool. Baumeister, R., Liu, Y. and Ruvkun, G. (1996). Lineage-specific regulators 231, 45-56. couple cell lineage asymmetry to the transcription of the Caenorhabditis Nobes, C. D. and Hall, A. (1995). Rho, rac, and cdc42 GTPases regulate the elegans POU gene unc-86 during neurogenesis. Genes Dev. 10, 1395-410. assembly of multimolecular focal complexes associated with actin stress Brenner, S. (1974). The genetics of Caenorhabditis elegans. Genetics 77, 71- fibers, lamellipodia, and filopodia. Cell 81, 53-62. 94. Ridley, A. J. and Hall, A. (1992). The small GTP-binding protein rho regulates Chalfie, M., Yuan, T., Euskirchen, G., Ward, W. W. and Prasher, D. C. the assembly of focal adhesions and actin stress fibers in response to growth (1994). Green fluorescent protein as a marker for gene expression. Science factors. Cell 70, 389-99. 263, 802-5. Ridley, A. J., Paterson, H. F., Johnston, C. L., Diekmann, D. and Hall, A. Chan, S. S.-Y., Zheng, H., Su, M.-W., Wilk, R., Killeen, M. T., Hedgecock, (1992). The small GTP-binding protein rac regulates growth factor-induced E. M. and Culotti, J. G. (1996). UNC-40, a C. elegans homolog of DCC membrane ruffling. Cell 70, 401-10. (Deleted in Colorectal Cancer), is required in motile cells responding to Run, J.-Q., Steven, R., Hung, M.-S., Van Weeghel, R., Culotti, J. G. and UNC-6 netrin cues. Cell 87, 187-95. Way, J. C. (1996). Suppressors of the unc-73 gene of Caenorhabditis Chisholm, A. D. (1991). Control of cell fate in the tail regions of C. elegans by elegans. Genetics 143, 225-36. the gene egl-5. Development 111, 921-32. Serafini, T., Kennedy, T. E., Galko, M. J., Mirzayan, C., Jessell, T. M. Desai, C., Garriga, G., McIntire, S. L. and Horvitz, H. R. (1988). A genetic and Tessier-Lavigne, M. (1994). The netrins define a family of axon pathway for the development of the Caenorhabditis elegans HSN motor outgrowth-promoting proteins homologous to C. elegans UNC-6. Cell 78, neurons. Nature 336, 638-46. 409-24. Durbin, R. (1987). Studies on the Development and Organisation of the Sulston, J. E. and Horvitz, H. R. (1977). Post-embryonic cell lineages of the Nervous System of Caenorhabditis elegans. PhD thesis, Cambridge nematode, Caenorhabditis elegans. Dev. Biol. 56, 110-56. University, UK. Sulston, J. E., Schierenberg, E., White, J. G. and Thomson, J. N. (1983). C. elegans cell migration mutants 1843

The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev. Kenyon, C. (1993). A homeotic gene cluster patterns the anteroposterior Biol. 100, 64-119. body axis of C. elegans. Cell 74, 29-42. Svendsen, P. C. and McGhee, J. D. (1995). The C. elegans neuronally White, J. G., Southgate, E., Thomson, J. N. and Brenner, S. (1986). The expressed homeobox gene ceh-10 is closely related to genes expressed in the structure of the nervous system of Caenorhabditis elegans. Phil. Trans. R. vertebrate eye. Development 121, 1253-62. Soc. Lond. B 314, 1-340. Tessier-Lavigne, M. (1994). Axon guidance by diffusible repellants and Wightman, B., Clark, S. G., Taskar, A. M., Forrester, W. C., Maricq, A. V., attractants. Curr. Opin. Genet. Dev. 4, 596-601. Bargmann, C. I. and Garriga, G. (1996). The C. elegans gene vab-8 guides Thomas, J. H., Stern, M. J. and Horvitz, H. R. (1990). Cell interactions posteriorly directed axon outgrowth and cell migration. Development 122, coordinate the development of the C. elegans egg-laying system. Cell 62, 671-82. 1041-52. Wang, B. B., Muller, I. M., Austin, J., Robinson, N. T., Chisholm, A. and (Accepted 19 February 1997)