Mutations in the stumpy gene reveal intermediate targets for zebrafish motor axons

Transcription

1 Development 127, (2000) Printed in Great Britain The Company of Biologists Limited 2000 DEV Mutations in the stumpy gene reveal intermediate targets for zebrafish motor axons Christine E. Beattie 1,2, *, Ellie Melancon 1 and Judith S. Eisen 1 1 Institute of Neuroscience, University of Oregon, Eugene, OR 97403, USA 2 Neurobiotechnology Center and Department of Pharmacology, The Ohio State University, Columbus, OH 43210, USA *Author for correspondence ( beattie.24@osu.edu) Accepted 3 April; published on WWW 23 May 2000 SUMMARY Primary motoneurons, the earliest developing spinal motoneurons in zebrafish, have highly stereotyped axon projections. Although much is known about the development of these neurons, the molecular cues guiding their axons have not been identified. In a screen designed to reveal mutations affecting motor axons, we isolated two mutations in the stumpy gene that dramatically affect pathfinding by the primary motoneuron, CaP. In stumpy mutants, CaP axons extend along the common pathway, a region shared by other primary motor axons, but stall at an intermediate target, the horizontal myoseptum, and fail to extend along their axon-specific pathway during the first day of development. Later, most CaP axons progress a short distance beyond the horizontal myoseptum, but tend to stall at another intermediate target. Mosaic analysis revealed that stumpy function is needed both autonomously in CaP and non-autonomously in other cells. stumpy function is also required for axons of other primary and secondary motoneurons to progress properly past intermediate targets and to branch. These results reveal a series of intermediate targets involved in motor axon guidance and suggest that stumpy function is required for motor axons to progress from proximally located intermediate targets to distally located ones. Key words: Primary motoneuron, Muscle pioneer, Single-cell transplants, Mosaic analysis, Laser ablation, Genetic mapping INTRODUCTION Formation of appropriate connections between motoneurons and their target muscles is essential for animals to carry out their normal behaviors. En route to their targets, motor axons execute a series of intricate steps, including extension out of the spinal cord, navigation on the proper pathway and innervation of correct muscle targets. In vivo observations have revealed that axons extend towards their targets in a highly stereotyped manner, making few navigational errors (Eisen, 1994). The precision of these events is astonishing considering the complex, dynamic environment within a developing embryo and the distances many axons travel to reach their targets. There is considerable evidence that, as they navigate toward their targets, axons integrate information from various cues that determine when and where to extend (Goodman and Tessier- Lavigne, 1997). Mutagenesis screens in Drosophila melanogaster, in which antibody markers were used to visualize pathfinding defects, have uncovered many molecules that function in axon guidance (Goodman, 1996; Chiba and Keshishian, 1996). One of the themes that has emerged from these studies is that motor axons navigate to their muscle targets by extending along pathways punctuated by intermediate targets, where they pause before branching, turning or extending distally. Thus, both intermediate targets and distant pathway regions are likely to provide guidance information. Formation of stereotyped pathways may also involve inhibition of axons from invading regions adjacent to their normal pathways. Intermediate targets have also been described in vertebrates. For example, commissural axons navigate to the floor plate, pause and then turn to extend along other specific pathways (Bovolenta and Dodd, 1990; Kuwada et al., 1990). In the forebrain, a population of retinal axons crosses a zone along the midline of the optic chiasm whereas another population fails to cross, suggesting that this region is an intermediate target for retinal axons (Godement et al., 1990; Sretavan et al., 1995). Vertebrate motoneurons also extend axons to intermediate targets, the crural and sciatic plexus in avians (Lance-Jones and Landmesser, 1981; Tosney and Landmesser, 1985) and the muscle pioneer cells that define the horizontal myoseptum in zebrafish (Eisen et al., 1986; Melancon et al., 1997). Unlike studies in insects, however, studies in vertebrates have not yet revealed a series of closely spaced intermediate targets for navigating motor axons. The zebrafish is an excellent system for elucidating mechanisms of vertebrate motor axon guidance. Each myotome is innervated in a highly stereotyped manner by three or four primary motoneurons, each of which can be individually identified based on morphology and gene expression (Eisen, 1999). Axons from the primary

2 2654 C. E. Beattie, E. Melancon and J. S. Eisen motoneurons of each spinal hemisegment exit the spinal cord in the same ventral root and extend ventrally along a common pathway on the medial aspect of the myotome until they reach the muscle pioneers, an intermediate target at the nascent horizontal myoseptum, where they typically pause for up to 2 hours (Eisen et al., 1986; Fig. 1). After pausing, these axons diverge along cell-specific pathways: the Caudal Primary motoneuron, CaP, extends an axon along the ventromedial myotome; the Middle Primary motoneuron, MiP, sprouts a collateral that extends along the dorsomedial myotome and later retracts the original ventral axon; the Rostral Primary motoneuron, RoP, extends an axon laterally, through the horizontal myoseptum; the Variable Primary motoneuron, VaP, does not extend an axon beyond the horizontal myoseptum and typically dies after a few hours (Eisen et al., 1990). During the first day of development, primary motoneurons are the only cells with axons along these pathways; later they are joined by axons of secondary motoneurons (Pike et al., 1992). Although previous work has characterized the outgrowth, morphology, synaptic interactions and pathway regions of primary motor axons (Eisen et al., 1986; Myers et al., 1986; Melancon et al., 1997; Beattie and Eisen, 1997; Bernhardt et al., 1998), very little is known about the molecular mechanisms underlying their stereotyped pathfinding. Many of these mechanisms should be revealed by characterization of mutations affecting zebrafish motor axons (Granato et al., 1996; Zeller and Granato, 1999; Beattie et al., 1999). To identify genes involved in motoneuron pathfinding, we screened zebrafish mutants using antibodies to uncover defects in axon pathfinding (Beattie et al., 1999). Here we characterize the phenotype of zebrafish embryos homozygous for mutations in the stumpy (sty) gene. Analysis of these mutants reveals a series of intermediate targets for motor axons. Our data also suggest that stumpy may encode part of an attractive signaling system necessary for motor axons to extend beyond these intermediate targets onto distal pathway regions and for them to branch. We suggest that, like insects, vertebrate motor axons navigate to their targets by following a closely spaced series of intermediate targets. Uncovering these aspects of pathfinding paves the way for future molecular dissection of the interactions involved in this stepwise process. MATERIALS AND METHODS Fish strains Mutant strains were kept as heterozygous (stumpy b393, stumpy b398 ) and homozygous lines (stumpy b393 ) in the *AB background. Homozygous mutant embryos (stumpy ) were obtained by natural crosses between identified carriers. Embryos raised between 25.5 and 28.5 C were staged by counting the number of raised somites and converting to hours post-fertilization at 28.5 C (h; Kimmel et al., 1995). Mutagenesis screen The mutagenesis screen was performed as described in Beattie et al., (1999). Briefly, adult male zebrafish exposed to 3 mm ethylnitrosourea (ENU) (Solnica-Krezel et al., 1994) were outcrossed to wild-type females to create an F 1 stock; mutations were screened in F 1 females by examining parthenogenetic diploid F 2 embryos created by the early pressure method (Streisinger et al., 1981; Beattie et al., 1999). Grossly abnormal embryos were removed at 2 and 24 h. The remaining embryos were fixed, labeled with antibodies recognizing primary motoneurons and analyzed under a stereomicroscope. Females identified as carrying a mutation were outcrossed to wild-type males to generate F 1 lines. Complementation was determined by scoring four separate matings between stumpy b398 and stumpy b398 heterozygotes. Genotypes were assigned based on CaP axon length at h, visualized by whole-mount antibody labeling with zn1 and znp1 (see below). Genetic mapping To map stumpy, mutants (*AB background) were outcrossed to WIK and SJD strains to create polymorphic mapping lines. Parthenogenetic F2 diploid embryos were produced by fertilizing eggs from identified carriers with UV-irradiated sperm (Streisinger et al., 1981). Genomic DNA was isolated from individual embryos designated either wild type or mutant, as determined by antibody labeling. PCR with Simple Sequence Length Polymorphic markers (SSLP markers; Knapik et al., 1996, 1998; Shimoda et al., 1999) was used to determine centromeric linkage (Johnson et al., 1995, 1996). stumpy b398 linkage was confirmed using normal diploid embryos from a *AB WIK cross. Whole-mount antibody labeling and in situ RNA hybridization Whole mount antibody labeling was performed as described in Eisen et al., (1989) with the following modifications. After fixation, embryos were incubated in water for 30 minutes (24-26 h embryos) to 4 hours (36, 60, 72 h embryos) to increase permeability, placed into PBDT blocking solution (phosphate buffered saline (PBS), 1% dimethylsulfoxide (DMSO), 1% bovine serum albumin (BSA), 0.5% Triton X-100) containing 2.5% goat serum, for 1 hour, incubated overnight at 4 C in antibodies diluted in PBDT containing 2.5% goat serum, and washed in PBS/0.5% Triton X-100 for 1 hour. Antibodies were detected using Clonal-PAP system (Sternberger) with diaminobenzidine (DAB) as a substrate (see Beattie and Eisen, 1997) or FITC-conjugated goat anti-mouse IgG. After antibody labeling, some embryos were embedded in 1.5% agar/5% sucrose and cut into 16 µm transverse cryostat sections. The zn1 and znp1 monoclonal antibodies (mabs) recognize primary and secondary motor axons (Trevarrow et al., 1990; Melancon et al., 1997) and the 4D9 mab recognizes zebrafish Engrailed proteins (Patel et al., 1989; Hatta et al., 1991). Whole-mount in situ RNA hybridization using digoxigenin-labeled anti-sense islet2 probes (Appel et al., 1995) was performed as described in Beattie et al. (1997). Single cell labels Individual motoneurons were labeled with rhodamine dextran ( MW; Molecular Probes) as described in Eisen et al., (1989). Embryos were mounted in 1.2% agar on a microslide for labeling, then removed from the agar and placed in embryo medium (Westerfield, 1995) containing 50 i.u. penicillin and 5 µg streptomycin at 28.5 C. Labeled cells were visualized using a Zeiss Axioscope. Images were captured with a Photometrics SPOT camera and were colorized using Photoshop (Adobe). Single cell and myotome transplants Myotome transplants were performed as described in Beattie and Eisen (1997). Single cell transplants were performed as described in Eisen (1991). Briefly, 16 h donor embryos labeled with rhodamine dextran ( MW; Molecular Probes) and unlabeled host embryos were mounted side by side in 1.2% agar on a microslide. Individual CaPs were transplanted from labeled donors to unlabeled host spinal hemisegments from which the native CaP and VaP (Eisen et al., 1990) motoneurons had been removed. Transplanted cells were visualized using a Zeiss Universal Compound Microscope equipped with a Dark Invader low light level camera. Images were captured using AxoVideo (Axon Instruments) and colorized using Photoshop (Adobe).

3 Motor axon pathfinding in stumpy mutants 2655 Muscle pioneer ablations Muscle pioneer ablations were performed by laser-irradiation as previously described (Melancon et al., 1997). Muscle pioneers in stumpy mutants were visualized under Nomarski DIC optics and ablations were performed in one or two somites at axial levels 7-10 prior to growth cone contact (approximately 17 h). Embryos were allowed to recover until h and then processed for anti-engrailed (4D9; Hatta et al., 1991) and znp1 immunoreactivity. As in previous studies, occasionally laser-irradiation created muscle scarring that precluded scoring extension of CaP axons (Melancon et al., 1997). RESULTS Mutations in the stumpy gene affect motor axons To identify mutations disrupting motoneurons, we screened h progeny from ENU-mutagenized fish with antibodies that recognize motoneurons (Beattie et al., 1999). We found three mutations in which embryos lacked ventral motor axons. Complementation tests suggested that two of these mutations affect the same gene, which we named stumpy (Fig. 2; Table 1). The third mutation, topped b458, complements stumpy and will be discussed elsewhere (C. M. Herpolsheimer and C. E. Beattie, unpublished). Both stumpy b393 and stumpy b398 show Mendelian inheritance (Table 1); however, stumpy b398 is a recessive, lethal mutation whereas stumpy b393 is a viable mutation that displays partial dominance; approximately 50% of the embryos from a heterozygous stumpy b393 mating show a phenotype intermediate between the phenotypes of homozygous mutants and wild types (Fig. 3; Table 1). stumpy embryos and larvae have no obvious motility defect; however, we cannot rule out subtle defects that might only be visible using high-speed video microscopy. As a first step toward molecular identification of stumpy, we Fig. 1. Schematic diagram of primary motor axon pathways. (A) Spinal cord hemisegments and overlying myotomes. CaP (red), MiP (blue) and RoP (green) cell bodies reside in stereotyped spinal cord locations. All primary motor axons leave the spinal cord at the same ventral root (1) and traverse a common pathway (a). At the horizontal myoseptum (dashed line), they pause and then diverge along axon-specific pathways (2). CaP axons extend ventrally along the medial myotome (b). MiP axons branch and extend along the dorsomedial myotome (c). By approximately 26 h (3), MiP has retracted its ventral axon and both the CaP and MiP axons have turned laterally. RoP axons extend mediolaterally through the horizontal myoseptum (d). VaP is not pictured here. Its cell body is directly adjacent to CaP and its axon only extends to the horizontal myoseptum. (B) Cross sectional view showing CaP and MiP (on separate sides for clarity), the common pathway (a), the CaP-specific pathway (b) and the MiP-specific pathway (c). sc, spinal cord; nc, notochord. Table 1. The stumpy mutant phenotype segregates in a Mendelian fashion Phenotypes from heterozygous matings (%) Wild type Heterozygous Homozygous b393 (n=155) 23.2± ± ±2.0 b398 (n=187) 74.4±1.2 na 25.6±1.2 b393 b398 (n=158) 48.0± ± ±1.5 Embryos were collected from four separate matings of heterozygous stumpy b393, stumpy b398 and stumpy b393 stumpy b398 fish. Phenotypes were designated based on axon length, as visualized using whole-mount antibody labeling with zn1 and znp1 at h (see Fig. 3). Numbers did not significantly differ from expected Mendelian ratios (χ 2 test, data not shown). na, not applicable. placed both putative alleles on the zebrafish genome linkage map (Postlethwait et al., 1994, 1998). To map stumpy b393 to a linkage group (Lg), we used Simple Sequence Length Polymorphisms (SSLP; Knapik et al., 1996, 1998) on parthenogenetic diploid embryos generated by suppressing the second meiotic division (Streisinger et al., 1981; Johnson et al., 1995, 1996). Mapping lines were generated by crossing the mutant line, on the *AB background, to two other parental lines (SJD, WIK) commonly used for mapping because of the high degree of polymorphism with *AB (Nechiporuk et al., 1999). DNA was amplified by PCR from individual parthenogenetic diploid embryos scored as phenotypically mutant or wild type. Two markers near the centromere of Lg 13, z9995 and z3424, segregated with stumpy b393 in all mutants tested (n=50), indicating that stumpy maps to this Lg (data not shown). stumpy b398 was shown to segregate with z9995 using normal diploid embryos (data not shown), further supporting the notion that these two mutations are allelic. stumpy function is required for CaP pathfinding The striking absence of ventral motor axons at 24 h suggested that stumpy specifically affects CaP motoneurons. Antibody labeling showed that CaPs were present in stumpy mutants (see white arrowhead in Fig. 3C) and vital dye labeling revealing that stumpy CaP axons extended along the common pathway but stopped at the horizontal myoseptum (Fig. 4A). This defect could result from a requirement for stumpy function during either cell-fate specification or pathfinding. To address whether CaPs were properly specified, we analyzed expression of islet2, a LIM homeobox gene expressed in CaP and VaP (Appel et al., 1995; Tokumoto et al., 1995). The distribution of CaPs and VaPs in stumpy mutants was indistinguishable from that in wild types (Fig. 4B,C), suggesting that stumpy CaPs were properly specified. Thus, we hypothesized that the mutation affected CaP pathfinding. To analyze pathfinding, we labeled CaPs in living wild-type and mutant embryos and compared their axon projections over several days (Fig. 5; Table 2). At h, wild-type axons had extended past the horizontal myoseptum along a CaP-specific pathway on the ventromedial myotome (Table 2; Fig. 5A). At this stage, many axons had a prominent varicosity adjacent to the ventral aspect of the notochord (* in Fig. 5A). It is also common for axons to branch in this region (Eisen et al., 1986; Myers et al., 1986), thus it appears to be a second intermediate target for CaP axons. By 48 h, axons had reached the ventral aspect of the myotome (Fig. 5B; Table 2) and turned laterally.

4 2656 C. E. Beattie, E. Melancon and J. S. Eisen Fig. 2. stumpy mutants were isolated in a screen for aberrant motor axons. Lateral views (A,C,E,) and cross sections (B,D,F) of whole-mount antibody labeling at 26 h. (A,B) wild-type embryo, (C,D) stumpy b393 mutant embryo and (E,F) stumpy b398 mutant embryo. Arrowheads denote distal extent of ventral stumpy b398 axons. In this and all subsequent lateral views, anterior is to the left and dorsal is to the top. Bar, 50 µm. By 72 h, axons had extended dorsally along the rostral myotome boundary (arrow in Fig. 5C; Table 2;) and there were numerous branches throughout the ventral myotome. The most prominent of these branches originated at what appears to be a third intermediate target for CaP axons at the ventral aspect of the myotome (* in Fig. 5C; see Eisen et al., 1986; Myers et al., 1986; Liu and Westerfield, 1990). In contrast to wild types, at h CaP axons in stumpy b393 mutants were stalled at the horizontal myoseptum (Table 2; Fig. 5D). Even at 48 h, some axons (27%) remained stalled there (Table 2), while the remainder extended further, to the intermediate target at the ventral aspect of the notochord (Table 2; Fig. 5E). By 72 h, most CaPs that had not extended axons beyond the horizontal myoseptum were dead or dying, as revealed by the cell body and axon breaking apart, unless they made ectopic branches into the ventral muscle (Table 2). Even at this stage the vast majority (88%) of the axons remained stalled at the ventral notochord intermediate target (Table 2; Fig. 5F). We also labeled CaPs in homozygous stumpy b398 mutants. At 24 h, all CaP axons were either stalled at the horizontal myoseptum (67%) or at the ventral notochord intermediate target (33%; Table 2). However, in contrast to stumpy b393, by 48 h almost half the axons extended to the ventral myotome intermediate target, suggesting that this allele has a less severe phenotype (Table 2). As in stumpy b393 mutants, by 72 h most stumpy b398 CaPs that had not extended axons into the ventral myotome were dead or dying (Table 2). These results suggest that without wild-type stumpy function, CaP axons stall at intermediate targets along their pathway. Muscle pioneers may contribute to the stumpy CaP phenotype To learn what aspect of pathfinding was affected by stumpy gene function, we considered the myotome regions encountered during outgrowth. CaP axons traverse a common pathway, pause at the horizontal myoseptum and extend along the CaP-specific pathway (Fig. 1). Extension along the common pathway appeared normal in stumpy mutants (Fig. 4A). Although most stumpy CaP axons eventually extended beyond the horizontal myoseptum, they paused there for an excessive amount of time, up to 20 hours, compared with about 2 hours in wild types (Eisen et al., 1986). Thus, we reasoned that stumpy function types might regulate the ability of CaP axons to leave the horizontal myoseptum. Previous studies suggested that the muscle pioneers that define the horizontal myoseptum normally provide a signal that specifically affects MiPs, because ablation of muscle pioneers allowed MiPs to Table 2. CaP axons in stumpy mutants show developmental defects Number of CaPs with axons extending to a particular location (%) 24 h 48 h 72 h HS VNC VM HS VNC VM HS VNC VM Wild type 0/10 (0) 0/10 (0) 10/10 (100) 0/10 (0) 0/10 (0) 10/10 (100) 0/10 (0) 0/10 (0) 10/10 (100) b393 15/15 (100) 0/15 (0) 0/15 (0) 4/15 (27) a 11/15 (73) b 0/15 (0) 0/8 (0) 7/8 (88) 1/8 (12) c b398 8/12 (67) 4/12 (33) 0/12 (0) 2/12 (17) d 4/12 (33) e 6/12 (50) 2/8 (25) f 0/6 (0) 6/8 (75) CaP axon length was determined in living embryos after injecting dye into single CaP somata. Axons were scored at three time points (approximately 24, 48, 72 h) and the position of the most distal part of the axon was noted. HS, horizontal myoseptum; VNC, myotome region adjacent to the ventral edge of the notochord; VM, distal region of the ventral myotome. The number of CaPs with axons at a particular point are represented as a fraction with the percentage in parentheses. a One embryo had a CaP that extended slightly past the HMS and was alive at 72 h while the other three cells were dead at 72 h. b Four of these cells were dead or dying at 72 h. c Axon branched within the proximal ventral myotome and did not reach the ventral edge of the myotome. d Both of these cells were dead at 72 h. e Two of these cells were dead or dying at 72 h. f These cells had ectopic branches at the HS.

5 Motor axon pathfinding in stumpy mutants 2657 extend axons slightly beyond the horizontal myoseptum (Melancon et al., 1997). The stumpy phenotype could result if wild-type stumpy function was required to prevent CaPs from responding to this signal. To test this idea, we ablated muscle pioneers in stumpy b393 mutants and analyzed CaP axons. In non-ablated, control segments, 17% (2/12) of stumpy CaP axons had extended beyond the horizontal myoseptum by 26 h. In contrast, after removal of muscle pioneers, 55% (5/9) of stumpy CaP axons had extended beyond the horizontal myoseptum by this stage; four of these axons (4/5, 80%) were stopped at the ventral notochord intermediate target (Fig. 6). These results support the idea that stumpy CaP axons may respond to a muscle pioneer-derived stop signal that they normally ignore in wild types. Even in experimental segments, however, CaP axons were not normal, suggesting that stumpy also functions to promote CaP axon extension along its cell-specific pathway independent of the muscle pioneers. Genetic mosaics reveal that stumpy function is needed in both CaP and other cells To address whether stumpy function was required cellautonomously in CaP, we created genetic mosaics by transplanting stumpy b393 CaPs into wild-type hosts and wildtype CaPs into stumpy b393 hosts prior to axogenesis. Wild-type CaPs transplanted into wild-type hosts developed normal axon projections (8/8; Fig. 7A). In contrast, in stumpy hosts, most wild-type CaP axons stalled at the horizontal myoseptum (86%; 6/7; Fig. 7B), although one CaP appeared wild type. Most stumpy CaPs also stalled at the horizontal myoseptum in wild-type hosts (67%; 6/9; Fig. 7C), although three CaPs appeared wild type. We also created genetic mosaics in which stumpy b393 myotomes were replaced with wild-type myotomes; these also failed to rescue stumpy CaP axons (n=6, data not shown). Together these results suggest that wild-type stumpy function is needed both in CaP and in other cells for CaP to extend normally along its cell-specific pathway. MiP and RoP axons in stumpy mutants display defects in branching Because stumpy function appears to be needed in surrounding cells for proper extension of CaP axons, we wondered whether it was also needed for proper extension of MiP and RoP axons. To test this possibility, we labeled MiPs and RoPs with vital dye and followed their axon projections in living stumpy b393 mutants. Both motoneurons extended axons normally along the common pathway (data not shown). stumpy RoP axons looked normal during the first day of development but at 48 h they were less branched than wildtype RoP axons (n=6; Fig. 8A,B). Although antibody labeling revealed that MiP dorsal axons were present in stumpy mutants (see Fig. 3), examination of vital dye-labeled cells Fig. 3. stumpy mutants have short ventral axons. Whole-mount antibody labeling of 26 h stumpy b393 wild-type sibling (A), heterozygous (B) and homozygous (C) mutants. Arrows denote the choice point at the horizontal myoseptum; black arrowheads indicate dorsally projecting MiP axons; the asterisk in B indicates neurite extending beyond this choice point; the white arrowhead in C denotes zn1-labeled CaP cell body. Bar, 25 µm. Fig. 4. stumpy function is unnecessary for CaP specification. (A) CaP motoneuron (red) labeled with vital dye in a living stumpy mutant, which was then sectioned and processed for immunohistochemistry (zn1 and znp1; green). Arrowhead denotes the CaP cell body and arrows denote the horizontal myoseptum. (B,C) Whole-mount RNA in situ hybridization using a digoxigeninlabeled probe for anti-sense islet2 (blue) in wild-type (B) and stumpy embryos (C) at approximately 24 h. Arrowheads point to CaPs and asterisks denotes CaP/VaP pairs. Bar, 10 µm.

6 2658 C. E. Beattie, E. Melancon and J. S. Eisen Fig. 5. CaP axons are truncated in stumpy mutants. Individual CaP motoneurons were labeled in live wild-type (A-C) and stumpy mutants (D-F) with a vital fluorescent dye and analyzed at approximately 22 h (A,D), 48 h (B,E), and 72 h (C,F). Arrowheads denote the horizontal myoseptum; asterisk in A indicates a branch along the myotome adjacent to the ventral aspect of the notochord. The arrow in C indicates the axon turning laterally (the distal part of the axon is in a different focal plane and not shown here) and the asterisk denotes the major distal branch. White lines indicate the dorsal and ventral aspects of the notochord. Bar, 20 µm. Fig. 6. Muscle pioneer ablations do not rescue stumpy CaP axons. Lateral view of a 26 h stumpy mutant after muscle pioneer ablations. The asterisk denotes segment in which muscle pioneers were ablated; arrows indicate labeled muscle pioneers in control segments; arrowheads denote distal tip of CaP axons. Axons were labeled with zn1 and znp1 and muscle pioneers were labeled with 4D9 mabs. Bar, 25 µm. revealed that stumpy MiP axon projections were variable. Half the labeled MiPs retracted the ventral axon on a normal, wild-type schedule but had a dorsal axon that was less branched and failed to make the normal lateral turn during the second day of development (6/12; Fig. 8C,D). One third (4/12) had a similarly affected dorsal axon but also retained a ventral axon that extended aberrantly through the horizontal myoseptum to the caudal segment boundary and retrogradely on the lateral myotome surface. One of the remaining MiPs had a normal dorsal axon but an aberrant ventral axon and the other MiP looked wild type. In wild types, the MiP dorsal axon often pauses at a specific distal location and forms a caudal branch as it turns laterally (Fig. 8C; Eisen et al., 1986; Myers et al., 1986). In 83% of the cases (10/12), stumpy MiPs stalled in this region suggesting that it is an intermediate target for MiP axons. These results illustrate that most stumpy MiP and RoP axons looked similar to wild-type axons during the first day of development, but later showed a variety of defects. stumpy function is unnecessary for pathfinding by other types of neurons To learn whether axons of other types of neurons were affected by stumpy mutations, we examined Rohon-Beard, trigeminal and lateral line sensory neurons (Metcalfe et al., 1990) and VeLD (Bernhardt et al., 1990) and Mauthner (Metcalfe et al., 1986) interneurons. Pathfinding by all of these cell types appeared unaffected in stumpy mutants (data not shown), suggesting that stumpy does not function globally in axon pathfinding, but is specific for motor axons. stumpy function is required for proper formation of motor nerves To learn whether stumpy mutations affected other motor axons, we examined wild-type and stumpy embryos at later developmental stages. The znp1 monoclonal antibody recognizes the axons of early developing primary motoneurons and later developing secondary motoneurons (Melancon, 1994), both of which contribute to the dorsal and ventral motor nerves. We followed these nerves between h in the caudal trunk (segments 9-13) and rostral tail (segments 14-18). As with many other aspects of development along the body axis, at a single developmental stage, motor nerves in the caudal trunk were developmentally more advanced than those in the rostral tail. We observed several phases of nerve extension similar to those described previously (Pike et al., 1992). First, nerves extended along characteristic medial pathways to the distal aspects of the ventral and dorsal myotome (see Fig. 9K). Next, nerves extended back toward the horizontal myoseptum along characteristic pathways on the lateral myotome (see Fig. 9K). Finally, nerves branched within the myotome. Wild-type nerves had extended distally (Fig. 9A) and begun to turn laterally (data not shown, see also Pike et al., 1992) by 36 h and projected many branches by 72 h (Fig. 9I).

7 Motor axon pathfinding in stumpy mutants 2659 Fig. 7. Wild-type stumpy function is needed in CaP and other cells. (A) Lateral view of CaP motoneuron transplanted from wildtype donor into wild-type host. (B) Wildtype CaP motoneuron transplanted into stumpy host. (C) stumpy CaP motoneuron transplanted into wild-type host. White lines demarcate dorsal and ventral aspects of the underlying notochord and arrows mark the horizontal myoseptum. Bar, 10 µm. Development of dorsal and ventral motor nerves was delayed in stumpy mutants and these nerves showed aberrant branching patterns. We first examined motor nerves in stumpy b393 mutants. In contrast to wild type, most stumpy nerves had not reached the distal aspect of the medial pathway by 36 h (Fig. 9A, compare with B). Approximately half the nerves (45±11.2%) were stalled at the ventral notochord intermediate target (Fig. 9B), the same region where CaP stalled (see Fig. 5E). Unlike CaP, however, most nerves had extended to the distal aspect of the medial pathway by 48 h (85±11.2%; Fig. 9C, compare with D), although some axons in the rostral tail remained short (arrow in Fig. 9C). By 60 h, nerves extended to the ventral aspect of the myotome (Fig. 9E, compare with F), but stalled there as well, so turning onto the lateral myotome was dramatically delayed, occurring some time between h as compared to 36 h in wild types. Moreover, some nerves also extended along the same aberrant pathway as ventral MiP axons, through the horizontal myoseptum and retrogradely along the lateral myotome (Fig. 9G, compare with H). Not only did these nerves extend through the myotome in an abnormal position, they also extended in abnormal locations along the lateral myotome (Fig. 9G, compare with H), and sometimes crossed segment boundaries, unlike nerves in wild types (data not shown). However, although the proximal portions of the nerves along the lateral myotome were in abnormal locations, the positions were more normal distally. Branching of both dorsal and ventral nerves was also dramatically reduced at 72 h (Fig. 9I, compare with J). We also examined motor nerves in homozygous stumpy b398 mutants. We found similar defects; however, the phenotype was markedly less severe than stumpy b393 and by 48 h it was difficult to distinguish stumpy b398 mutants from wild types (data not shown). mutations were isolated using antibodies to screen for axon defects. It seems unlikely that either allele is a protein null, because one allele is recessive and embryonic lethal and the other allele is partially dominant and viable. Motoneurons of embryos homozygous for the partially dominant allele are affected more dramatically than are those of embryos homozygous for the recessive allele. These mutations reveal previously unrecognized aspects of motoneuron pathfinding and will lead to elucidation of molecules that function during this process. Pathfinding begins normally in stumpy mutants Normal motoneuron projections require proper cell-fate specification (Eisen, 1998, 1999). islet2 expression and the normal correspondence of motoneuron cell body positions and axon trajectories reveals that specification of primary motoneurons is normal in stumpy mutants. Extension along the DISCUSSION We describe two mutations in the stumpy gene that disrupt pathfinding by zebrafish primary and secondary motoneurons. These Fig. 8. MiP and RoP axons show decreased branching in stumpy mutants. Individuallylabeled RoP (A,B) and MiP (C,D) motoneurons in living wild-type (A,C) and stumpy embryos (B,D) at 48 h. Arrows denote axon branches; asterisk in C signifies MiP axon turning laterally. Bar, 10 µm.

8 2660 C. E. Beattie, E. Melancon and J. S. Eisen common pathway is also normal, and thus appears to be independent of stumpy function; it has recently been shown to depend on function of another gene, diwanka (Zeller and Granato, 1999). The stumpy phenotype is first manifest when axons pause for an abnormally long time at the horizontal myoseptum. This data supports the idea that the horizontal myoseptum is the first intermediate target encountered by motor axons. Although axonal projections beyond the horizontal myoseptum are abnormal, when axons do project they typically select the appropriate pathway, suggesting that any positive and negative cues that distinguish the cell-specific pathways are normally distributed in stumpy mutants. stumpy is required for motor axons to extend past intermediate targets and for proper axon branching Although stumpy motor axons projected normally along the common pathway, projection distally was marked by stalling at specific locations: the horizontal myoseptum, the ventral aspect of the notochord, and the ventral and dorsal aspects of the myotome. These are regions where wild-type primary motor axons normally pause, make conspicuous varicosities and often branch (Eisen et al., 1986; Myers et al., 1986), suggesting that they may be intermediate targets (Melancon et al., 1997; Goodman and Tessier-Lavigne, 1997). The stalling of stumpy primary, and in some cases secondary, motor axons in these regions provides strong evidence that these regions are intermediate targets and that stumpy function is normally required for axons to leave intermediate targets and extend into more distal territory. One interpretation consistent with these data is that attractive cues that normally prompt motor axons to extend beyond intermediate targets are not present in stumpy mutants, or that motor axons are unable to respond to them. Alternatively, stumpy may function to destablize the interaction between growth cones and intermediate targets, thus enabling growth cones to progress beyond these regions; without stumpy function, growth cones stall or stop at these regions. There is evidence for this type of anti-adhesive function in Drosophila where the gene beaten path functions to regulate defasciculation from choice points (Fambrough and Goodman, 1996). stumpy gene function also regulates branching of motor axons at intermediate targets. For example, wild-type MiPs retract their ventral axons following prolonged contact with muscle pioneer cells at the horizontal myoseptum, but in the absence of these cells the ventral axon persists and may extend abnormal branches (Melancon et al., 1997). Most stumpy MiPs failed to retract the ventral axon, which then branched Fig. 9. stumpy motor nerves are aberrant. Wild-type (A,C,E,G,I) and stumpy mutant (B,D,F,H,J) embryos labeled with znp1 at 36 h (A,B; medial focal plane), 48 h (C,D; medial focal plane), 60 h (E,F; medial focal plane), 48 h (G,H; lateral focal plane) and 72 h (I,J; lateral focal plane). Arrows point to axons along the medial myotome, arrowheads point to axons along the lateral myotome and asterisks indicate axons extending laterally through the horizontal myoseptum. (K,L) Schematic drawings of axons in cross section (K, wild type) and (L, mutant) at 48 h. Arrows denote direction of axon growth. For each time point, 40 axons from segments 8-17 were analyzed in 4 embryos. Bars, 100 µm (A-D); 75 µm (E-F,I-J); 50 µm (G-H). Fig. 10. Model for stumpy function. In wild-type embryos, primary motor axons extend along the medial myotome and pause or branch at intermediate targets (blue dots). In stumpy mutants, primary motor axons stall at proximal intermediate targets and fail to reach distal ones. Thus, stumpy may be a molecule important for attracting growth cones to distal pathway regions (green).

9 Motor axon pathfinding in stumpy mutants 2661 aberrantly, suggesting that stumpy function is required to sculpt the MiP arbor by preventing the ventral axon from branching and ensuring that it is retracted. stumpy function also regulates branching of secondary motor axons, some of which project aberrantly through the horizontal myoseptum in stumpy mutants. Because this is similar to the aberrant path taken by some MiP ventral axons, it is unclear whether stumpy affects secondary motoneurons directly, or whether this effect is through interactions between primary and secondary motor axons (see Pike et al., 1992). CaP survival may depend on axons contacting appropriate muscle Survival of vertebrate motoneurons has been linked to receiving proper trophic support from target muscles (Oppenheim, 1996). Interestingly, stumpy CaP motoneurons that fail to extend an axon to their normal ventral muscle target typically die by 72 h. In contrast, most stumpy CaP motoneurons that extend a short distance along the ventral muscle, either by forming ectopic branches at the horizontal myoseptum or by extending to the ventral notochord intermediate target, survive longer. Previous studies have shown that CaP motoneurons can activate contractions in muscle pioneers (Melancon et al., 1997), suggesting intimate interactions between these two cell types. However, our results suggest that ventral muscle provides trophic support required for CaP motoneuron survival, and that this support cannot be derived from muscle pioneers. stumpy mutations do not simply convert CaPs to VaPs A superficial interpretation of the stumpy phenotype is that CaPs are converted into VaPs. All known molecular markers, as well as transplantation studies, show that CaP and VaP are both specified to be CaP and that interactions between the cells force one of them to adopt the VaP fate (Eisen, 1992). VaPs fail to extend axons beyond the horizontal myoseptum and typically die by 36 h (Eisen et al., 1990). However, although stumpy CaPs that fail to extend axons beyond the horizontal myoseptum also die, they survive more than a day longer than wild-type VaPs, a very significant amount of time in zebrafish development. Further, most stumpy CaPs do extend axons beyond the horizontal myoseptum, something never observed for VaPs in wild-type embryos. Although we cannot rule out a role for stumpy in specifying VaP, the clear differences between VaPs and stumpy CaPs suggest that the interactions determining the VaP fate are more complicated than a simple alteration in stumpy function. stumpy function is required both in CaP and in surrounding cells Our genetic mosaic analysis suggests that stumpy function is required both in motoneurons and in other cells for normal pathfinding. This idea is supported by previous mosaic analyses which distinguished cell-autonomous and cellnonautonomous functions required for motoneuron development. For example, individual primary motoneurons transplanted between wild types and spadetail mutants always behaved in the fashion appropriate for the genotype of the host, showing that the mutant motoneuron phenotype is cellnonautonomous (Eisen and Pike, 1991). In contrast, wild-type cells transplanted into detour mutant hosts can develop into branchiomotor neurons, which are absent from mutants, suggesting that detour function is required cell-autonomously in branchiomotor neurons (Chandresakhar et al., 1999). Without knowing precisely which other cells require stumpy function, it is difficult to predict what kind of molecule might be required both in motoneurons and in other cells for proper pathfinding. One possibility is a homophilic guidance or adhesion molecule, although other possibilities, such as a molecule that disrupts cell-cell interactions, cannot be excluded. stumpy functions in motor axon guidance Our studies reveal a series of intermediate targets along the pathways traversed by zebrafish motor axons. Studies in insects have shown that axon pathways are broken up into relatively short segments by a series of intermediate targets that provide important guidance information during pathfinding (Goodman and Tessier-Lavigne, 1997). Although intermediate targets have also been described during pathfinding in vertebrate embryos (Goodman and Tessier-Lavigne, 1997), there have not been clear examples in which an individual pathway is divided into short segments. We have shown that the pathways followed by spinal motor axons in embryonic zebrafish are punctuated by intermediate targets at relatively close intervals. Genetic screens (Eisen, 1996) will be important to establish the identities of the cells that define each of these intermediate targets and the specific molecular interactions they mediate. Axons normally extend directly to intermediate targets and transiently pause there, suggesting that these regions are more attractive than surrounding areas. This idea is supported by experiments showing that removal of at least one of these intermediate targets enhances motor axon extension, in both wild types (Melancon et al., 1997) and in stumpy mutants; however, extension of axons beyond these intermediate targets also suggests the presence of other attractive cues located more distally. Together these observations suggest the following model (Fig. 10). During extension, axons are attracted to intermediate targets. They pause there, but later leave as the intermediate targets become less attractive, because there are more attractive targets located distally, or because attractive interactions are disrupted. Perhaps the simplest way of thinking about stumpy gene function is that it encodes part of an attractive system required for axons to extend beyond intermediate targets whose attractiveness is mediated by molecules other than Stumpy. When stumpy function is absent, axons do not receive signals that prompt them to extend beyond intermediate targets, and thus they tend to stall; much later the attractive signals at these intermediate targets may diminish, and this allows axons to extend to the next intermediate target. This may also explain the lack of branching; if axons are stalled at an intermediate target, they may be unable to respond to signals that instruct them to branch. Eventual cloning of stumpy will help elucidate the molecular cues involved in these steps of motor axon pathfinding. We thank Chuck Kimmel, Monte Westerfield, Paul Henion and Mark Seeger for comments on the manuscript, the University of Oregon Zebrafish Facility staff, Kirsten Stoesser, Katie Davenport and Deanna Grant for fish care, Cindy Herpolsheimer for technical assistance and fish care, Karen Larison and Rushu Luo for sectioning

10 2662 C. E. Beattie, E. Melancon and J. S. Eisen and Don Kane and Michael Granato for advice regarding mapping. Supported by an AHA (Oregon Chapter) postdoctoral grant and NSF IBN to C. E. B. and NIH NS23915 to J. S. E. REFERENCES Appel, B., Korzh, V., Glasgow, E., Thor, S., Edlund, T., Dawid, I. B. and Eisen, J. S. (1995). Motoneuron fate specification revealed by patterned LIM homeobox gene expression in embryonic zebrafish. Development 121, Beattie, C. E. and Eisen, J. S. (1997). Notochord alters the permissiveness of myotome for pathfinding by an identified motoneuron in embryonic zebrafish. Development 124, Beattie, C. E., Hatta, K., Halpern, M. E., Liu, H., Eisen, J. S. and Kimmel, C. B. (1997). Temporal separation in the specification of primary and secondary motoneurons in zebrafish. Dev. Biol. 187, Beattie, C. B., Raible, D. W., Henion, P. D. and Eisen, J. S. (1999). Early pressure screens in zebrafish. In Methods in Cell Biology (ed. H. W. Detrich, M. Westerfield and L. Zon), pp San Diego: Academic Press. Bernhardt, R. R., Chitnis, A. B., Lindamer, L. and Kuwada, J. Y. (1990). Identification of spinal neurons in the embryonic and larval zebrafish. J. Comp. Neurol. 302, Berhnardt, R. R., Goerlinger, S., Roose, M. and Schachner, M. (1998). Anteriorposterior subdivision of the somite in embryonic zebrafish: Implications for motor axon guidance. Dev. Dyn. 213, Bovolenta, P. and Dodd, J. (1990). Guidance of commissural growth cones at the floor plate in embryonic rate spinal cord. Development 109, Chandrasekhar, A., Schauerte, H. E., Haffter, P. and Kuwada, J. Y. (1999). The zebrafish detour gene is essential for cranial but not spinal motor neuron induction. Development 126, Chiba, A. and Keshishian, H. (1996). Neuronal pathfinding and recognition: Roles of cell adhesian molecules. Dev. Biol. 180, Eisen, J. S. (1991). Determination of primary motoneuron identity in developing zebrafish embryos. Science 252, Eisen, J. S. (1992). The role of interactions in determining cell fate of two identified motoneurons in the embryonic zebrafish. Neuron 8, Eisen, J. S. (1994). Development of motoneuronal phenotype. Annu. Rev. Neurosci. 17, Eisen, J. S. (1996). Zebrafish make a big splash. Cell 87, Eisen, J. S. (1998). Genetic and molecular analysis of motoneuron development. Curr. Opin. Neurobiol. 8, Eisen, J. S. (1999). Patterning motoneurons in the vertebrate nervous system. Trends Neurosci. 22, Eisen, J. S. and Pike, S. H. (1991). The spt-1 mutation alters segmental arrangement and axonal development of identified neurons in the spinal cord of the embryonic zebrafish. Neuron 6, Eisen, J. S., Myers, P. Z. and Westerfield, M. (1986). Pathway selection by growth cones of identified motoneurones in live zebra fish embryos. Nature 320, Eisen, J. S., Pike, S. H. and Debu, B. (1989). The growth cones of identified motoneurons in embryonic zebrafish select appropriate pathways in the absence of specific cellular interactions. Neuron 2, Eisen, J. S., Pike, S. H. and Romancier, B. (1990). An identified motoneuron with variable fates in embryonic zebrafish. J. Neurosci. 10, Fambrough, D. and Goodman, C. S. (1996). The Drosophila beaten path gene encodes a novel secreted protein that regulates defasciculation at motor axon choice points. Cell 87, Godement, P., Salaun, J. and Mason, C. A. (1990). Retinal axon pathfinding in the optic chiasm of crossed and uncrossed fibers. Neuron 5, Goodman, C. S. (1996). Mechanisms and molecules that control growth cone guidance. Annu. Rev. Neurosci. 19, Goodman, C. and Tessier-Lavigne, M. (1997). Molecular Mechanisms of axon guidance and target recognition. In Molecular and Cellular Approaches of Neural Development (ed. W. M. Cowan, T. M. Jessel and L. Zipursky), pp Oxford: Oxford University Press. Granato, M., van Eeden, F. J., Schach, U., Trowe, T., Brand, M., Furutani-Seiki, M., Haffter, P., Hammerschmidt, M., Heisenberg, C. P., Jiang, Y. J., Kane, D. A., Kelsh, R. N., Mullins, M. C., Odenthal, J. and Nüsslein-Volhard, C. (1996). Genes controlling and mediating locomotion behavior of the zebrafish embryo and larva. Development 123, Hatta, K., Bremiller, R., Westerfield, M. and Kimmel, C. B. (1991). Diversity of expression of engrailed-like antigens in zebrafish. Development 112, Johnson, S. L., Africa, D., Horne, S. and Postlethwait, J. H. (1995). Half-tetrad analysis in zebrafish: mapping the ros mutation and to the centromere of linkage group I. Genetics 139, Johnson, S. L., Gates, M. A., Johnson, M., Talbot, W. S., Horne, S., Baik, K., Rude, S., Wong, J. and Postlethwait, J. H. (1996). Centromere-linkage analysis and consolidation fo the zebrafish genetic map. Genetics 142, Kimmel, C. B., Ballard, W. W., Kimmel, S. R., Ullmann, B. and Schilling, T. F. (1995). Stages of embryonic development of the zebrafish. Dev. Dyn. 203, Knapik, E. W., Goodman, A., Atkinson, O. S., Roberts, C. T., Shiozawa, M., Sim, C. U., Weksler-Zangen, S., Trolliet, M. R., Futrell, C., Innes, B. A., Koike, G., McLaughlin, M. G., Pierre, L., Simon, J. S., Viallonga, E., Roy, M., Chiange, P.-W., Fishman, M. C., Driever, W. and Jacob, H. J. (1996). A reference cross DNA panel for zebrafish (Danio rerio) anchored with simple sequence lenght polymorphisms. Development 123, Knapik, E. W., Goodman, A., Ekker, M., Chevrette, M., Delgado, J., Neuhauss, S., Shimoda, N., Driever, W., Fishman, M. C. and Jacob, H. J. (1998). A microsatellite genetic linkage map for zebrafish (Danio rerio). Nature Genetics 18, Kuwada, J. Y., Bernhardt, R. R. and Chitnis, A. B. (1990). Pathfinding by identified growth cones in the spinal cord of zebrafish embryos. J. Neurosci. 10, Lance-Jones, C. and Landmesser, L. T. (1981). Pathway selection by chick lumbrosacral motoneurons during normal development. Proc. R. Soc. Lond. B 214, Liu, D. W. and Westerfield, M. (1990). The formation of terminal fields in the absence of competitive interactions among primary motoneurons in the zebrafish. J. Neurosci. 10, Melancon, E. (1994). The role of muscle pioneers in pathfinding of primary motoneurons in the embryonic zebrafish. MA thesis, University of Oregon. Eugene, Oregon. Melancon, E., Liu, D. W. C., Westerfield, M. and Eisen, J. S. (1997). Pathfinding by indentified zebrafish motoneurons in the absence of muscle pioneers. J. Neurosci. 17, Metcalfe, W. K., Mendelson, B. and Kimmel, C. B. (1986). Segmental homologies among reticulospinal neurons in the hindbrain of the zebrafish larva. J. Comp. Neurol. 251, Metcalfe, W. K., Myers, P. Z., Trevarrow, B., Bass, M. B. and Kimmel, C. B. (1990). Primary neurons that express the L2/HNK-1 carbohydrate during early development in the zebrafish. Development 110, Myers, P. Z., Eisen, J. S. and Westerfield, M. (1986). Development and axonal outgrowth of identified motoneurons in the zebrafish. J. Neurosci. 6, Nechiporuk, A., Finney, J. E., Keating, M. T. and Johnson, S. L. (1999). Assessment of polymorphism in zebrafish mapping strains. Genome Res. 9, Oppenheim, R. W. (1996). Neurotrophic survival molecules for motoneurons: an embarrassment of riches. Neuron 17, Patel, N. H., Martin Blanco, E., Coleman, K. G., Poole, S. J., Ellis, M. C., Kornberg, T. B. and Goodman, C. S. (1989). Expression of engrailed proteins in arthropods, annelids, and chordates. Cell 58, Pike, S. H., Melancon, E. F. and Eisen, J. S. (1992). Pathfinding by zebrafish motoneurons in the absence of normal pioneer axons. Development 114, Postlethwait, J. H., Johnson, S. L., Midson, C. N., Talbot, W. S., Gates, M., Ballinger, E. W., Africa, D., Andrews, R., Carl, T., Eisen, J. S., Horne, S., Kimmel, C. B., Hutchinson, M., Hohnson, M. and Rodriguez, A. (1994). A genetic linkage map for the zebrafish. Science 264, Postlethwait, J. H., Yan, Y. L., Gates, M. A., Horne, S., Amores, A., Brownlie, A., Donovan, A., Egan, E. S., Force, A., Gong, Z., Goutel, C., Fritz, A., Kelsh, R., Knapik, E., Liao, E., Paw, B., Ransom, D., Singer, A., Thomson, M., Abduljabbar, T. S., Yelick, P., Beier, D., Joly, J.-S., Larhammar, D., Rosa, F., Westerfield, M., Zon, L., Johnson, S. L. and Talbot, W. S. (1998). Vertebrate genome evolution and the zebrafish gene map. Nat. Genet. 18, Shimoda, K., Knapik, E. W., Ziniti, J., Sim, C., Yamada, E., Kaplan, S., Jackson, D., de Sauvage, F., Jacob, H. and Fishman, M. C. (1999). Zebrafish genetic map with 2000 microsatellite markers. Genomics 58, Solnica-Krezel, L., Scheir, A. F. and Driever, W. (1994). Efficient recovery of ENU-induced mutations from the zebrafish germline. Genetics 136, Sretavan, D. W., Pure, E., Siegel, M. W. and Reichardt, L. F. (1995). Disruption of retinal axon ingrowth by ablation of embryonic mouse optic chiasm neurons. Science 269, Streisinger, G., Walker, C., Dower, N., Knauber, D. and Singer, F. (1981). Production of clones of homozygous diploid zebra fish (Brachydanio rerio). Nature 291, Tosney, K. W. and Landmesser, L. (1985). Growth cone morphology and trajectory in the lumbrosacral region of the chick embryo. J. Neurosci. 5, Tokumoto, M., Gong, Z., Tsubokawa, T., Hew, C. L., Uyemura, K., Hotta, Y. and Okamoto, H. (1995). Molecular heterogeneity among primary motoneurons and within myotomes revealed by the differential mrna expression of novel islet-1 homologs in embryonic zebrafish. Dev. Biol. 171, Trevarrow, B., Marks, D. L. and Kimmel, C. B. (1990). Organization of hindbrain segments in the zebrafish embryo. Neuron 4, Westerfield, M. (1995). The Zebrafish Book. Eugene, Oregon: University of Oregon Press. Zeller, J. and Granato, M. (1999). The zebrafish diwanka gene controls an early step of motor growth cone migration. Development 126,