Zebrafish Semaphorin Z1a collapses specific growth cones and alters their pathway in vivo

During development, neuronal growth cones extend along specific pathways to their targets by utilizing both repulsive and attractive cues (Tessier-Lavigne and Goodman, 1996). The repulsive activity, upon growth cones, of a variety of factors has been extensively analyzed with in vitro assays (Walter et al., 1987; Raper and Kapfhammer, 1990; Fitzgerald et al., 1993; Drescher et al., 1995; Fan and Raper, 1995). The semaphorin/collapsin family of molecules has recently been of interest in this regard, because several of the members are thought to be repulsive for specific growth cones (Kolodkin, 1996). These proteins are either secreted or transmembrane, may have an Ig domain or thrombospondin type 1 domains, and all share a large, conserved sema domain. The first member of this family that was identified to be repulsive for growth cones was chick Collapsin 1, a secreted protein that causes the collapse of specific subsets of growth cones in vitro including those of the dorsal root ganglion (DRG) cells (Luo et al., 1993). Explant experiments showed that Sema III/D, which is the mammalian homolog to chick Collapsin 1, selectively repelled the axons of the cutaneous DRG cells but not the muscle afferent DRG cells (Messersmith et al., 1995; Puschel et al., 1995). This made sense since Sema III/D is normally expressed by the ventral spinal cord, and cutaneous but not muscle afferent DRG axons are restricted to the dorsal cord. This suggested that Sema III/D secreted by the ventral spinal cord cells normally acts to restrict the cutaneous DRG axons to the dorsal cord. Recent generation and examination of sema III/D knockout mice suggest that the requirement of sema III/D for this restriction depends upon the genetic background of the mice. In one strain some cutaneous DRG axons appear to extend aberrantly into the ventral cord (Behar et al., 1996), while in another strain cutaneous DRG axons remained within the dorsal cord (Taniguchi et al., 1997). In the latter strain, a subset of peripheral nerves were abnormally defasciculated but apparently extended to their correct target regions. The reason for the difference in phenotype between the two different knockout mice is presently unclear.

To better understand the in vivo role of sema III/D/collapsin 1, we focused on the role of a zebrafish Semaphorin, Sema Z1a, that is homologous to Semaphorin III/D/Collapsin 1, for pathfinding by a specific set of sensory growth cones, those of the posterior lateral line ganglion. Zebrafish embryos are useful for examining the behavior of growth cones because the embryos are transparent and features are accessible at all stages of development (Kimmel et al., 1995). Furthermore, the zebrafish nervous system is relatively simple and contains numerous, well characterized neurons, including the neurons of the posterior lateral line (Metcalfe, 1985; Metcalfe et al., 1985; Collazo et al., 1994; O’Malley et al., 1996).

The lateral line is a sensory structure that detects water-born vibrations, found in fishes and amphibians. It consists of 10–11 sensory organs that are located periodically along the anterior/posterior axis midway between the dorsal and ventral margins of the trunk and tail in larval zebrafish. The posterior lateral line ganglion contains the 15–20 sensory neurons that innervate the sensory organs. This is located posterior to the otocyst and anterior to the axial muscles. The sensory organs develop from cells derived from a mass of cells called the migrating primordium of the lateral line (Metcalfe, 1985; Metcalfe et al., 1985). The primordium migrates posteriorly, starting from the posterior lateral line ganglion at approximately 20 hours and reaching the posterior end of the tail by 42 hours. As the primordium migrates it periodically deposits small clusters of cells that differentiate into the sensory organs. Interestingly the growth cones of the neurons of the posterior lateral line ganglion that innervate the sensory organs extend together with the migrating primordium. The primordium and growth cones migrate posteriorly to the first myotome. Once they reach the axial muscles they migrate between the skin and the axial muscles of the horizontal myoseptal region.

Our analysis demonstrates that Sema Z1a is expressed by the axial muscles in a pattern that suggests it may guide the growth cones of the posterior lateral line by delimiting the pathway these growth cones can follow on the muscles. We further show that application of Sema Z1a can induce collapse of lateral line growth cones in vivo, and that the growth cones will deviate from their normal pathway when confronted with cells that are misexpressing Sema Z1a. These results strongly suggest that Sema Z1a normally guides posterior lateral line growth cones via a repulsive mechanism.

Zebrafish (Danio rerio) embryos were collected and allowed to develop at 28.5°C, and staged as described by Westerfield (1995). In some cases pigmentation was inhibited by exposure to 0.2 mM phenylthiourea (Sigma) at about 20 hpf (hours postfertilization; Burrill and Easter S. S, 1994). floating head (flh) carrier fish were obtained from M. Halpern. you-too (yot) embryos were supplied by P. Haffter.

Antisense sema Z1a digoxigenin (DIG)-labeled riboprobes were synthesized and then adjusted to approximately 300 bps fragment length by limited alkaline hydrolysis (Cox et al., 1984). Hybridization on wholemounted embryos was performed as described by Westerfield (1995). For sectioning, fixed embryos were immersed in 30% sucrose, embedded in OCT compound (Miles, Elkhart, IN), and cut using a Reichert- Jung 2800 cryostat. Sections were air dried and mounted in 70% glycerol under a coverslip.

To label cells expressing Sema Z1a^myc and growth cones expressing the HNK1 antigen differentially, embryos were first stained using anti-myc then anti-HNK1 antibodies. Briefly, embryos were fixed in 4% paraformaldehyde in PBS, incubated with a 1:10 dilution of anti- myc mAb (Evan et al., 1985), incubated with a 1:100 dilution of anti- mouse IgG-HRP, and treated with 0.3 mg/ml diaminobenzidine (DAB), 0.08% NiCl², and 0.08% CoCl₂ which gives a black peroxidase reaction product . Embryos were then re-fixed, incubated with 1:4000 dilution of mAb zn12 (anti-HNK-1) (Metcalfe et al., 1990; Trevarrow et al., 1990), incubated with anti-mouse IgG-HRP, and then with 0.5 mg/ml DAB, which results in a brown peroxidase reaction product.

To mark DiI-labeled growth cones with a brown DAB reaction product, embryos were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer, embedded in 0.7% agar in 0.1 M phosphate buffer on glass slides, and under visual control through DIC optics of a compound microscope 2 mg/ml DiI in n,n-dimethylformamide was passed from a micropipette into the posterior lateral line ganglion iontophoretically. Embryos were then held overnight at 4°C in a dark humid chamber to allow for spread of the DiI into axons and growth cones and the fluorescently labeled cells were marked with a brown reaction product by following the photo-oxidation procedure with 0.5 mg/ml DAB and epifluorescence illumination (Lubke, 1993).

Growth cones were labeled within 30 minutes by ejecting DiI iontophoretically from a micropipette inserted into the posterior ganglion of the lateral line in living embryos that were anesthetized with 0.01% tricane (3-aminobenzonic acid ethylester). The embryos were then embedded in 0.7% agar. Following labeling, the growth cones were exposed to Sema Z1a-containing medium, Sema Z1a^myc- containing medium, or control conditioned medium as described below. The embryos were mounted onto a modified plastic dish with a coverslip bottom (Kapfhammer and Raper, 1987), and growth cone behavior was monitored on a epifluorescence microscope (Zeiss) equipped with a SIT camera (DAGE-MTI) and a temperature controlled stage (Medical Systems Corp.). Images were processed using Image-1 software (Universal Imaging Corp.) and transferred onto optical disks. Sixteen frames were averaged per image (exposure=1.03 second), and images were captured every 2 minutes for up to 2 hours 18 minutes. Time lapse records started approximately 15 minutes after the ejection of the conditioned medium due to the time required for transferring embryos from the injection microscope to the modified recording chamber on the microscope dedicated to time lapse recordings.

Culture medium containing recombinant Sema Z1a was generated by transfecting COS-1 cells grown at 37°C, 5% CO₂ in Dulbecco’s Modified Eagle’s Medium supplemented with 10% fetal calf serum, 50 U/ml penicillin, and 50 mg/ml streptomycin. COS-1 cells were transfected with DNA expression constructs pBK-CMV-sema Z1a, pBK-CMV-sema Z1a^myc, or as a control pBK-CMV (Stratagene) using the calcium phosphate method (Ausubel et al., 1991). After 48 hours the conditioned media were collected and centrifuged at 5000 g for 10 minutes to remove debris. To apply the conditioned media onto the growth cones, embryos were anesthetized, embedded in 0.7% agar on a slide, and mounted onto a compound microscope with DIC optics. A sharp micropipette filled with the conditioned medium was inserted, under visual control, into the leading edge of the migrating primordium, and the medium was ejected with 10 pressure pulses (10 mseconds, 5-20 psi) over a period of 10 minutes.

Approximately 1 nl of 200 ng/μl of capped RNA or 50 ng/μl of DNA that was suspended in water containing 0.1% phenol red was pressure- injected from a sharp micropipette into a single blastomere of zebrafish embryos at the 1–8 cell stage as described previously (Ekker et al., 1995). In all cases, more than 80% of injected embryos developed normally. To induce expression of sema Z1a or sema Z1a^myc driven by the zebrafish hsp70 promoter, embryos in a Petri dish were transferred into a water bath at 40°C and incubated for 1 hour. Following induction, embryos were allowed to develop for 2–4 hours at 28.5°C and fixed for analysis.

To clarify the in vivo role of semaphorins in zebrafish embryos, we cloned sema Z1a, a homolog of semaphorin D/III/collapsin 1, and analyzed its expression pattern (Yee and Kuwada, unpublished data). The zebrafish possesses a second gene referred to as sema Z1b that is also highly homologous to semaphorin III/D/collapsin 1 (Roos, Bernhardt and Schachner, personal communication; see Discussion). Interestingly, sema Z1a is expressed in a striped pattern by the metamerically arranged axial muscles in the trunk and tail (Figs 1A,B, 2A). Each myotome consists of a dorsal and ventral region separated by the horizontal myoseptal region (Eisen et al., 1986). sema Z1a is expressed by the dorsal and ventral regions but not by the horizontal myoseptal region at times when axons are extending within and upon these myotomes. This sema Z1a mRNA free region is 5-6 muscle cells wide (20-25 μm) and centered upon the horizontal myoseptum that separates the myotome into dorsal and ventral halves. Furthermore, sema Z1a is not expressed by the epidermis overlying the horizontal myoseptal region (Fig. 2A). Thus, sema Z1a expression delineates a longitudinal stripe in the trunk and tail that is free of sema Z1a mRNA.

Interestingly, the growth cones of the posterior lateral line ganglion extend posteriorly upon the muscles of the horizontal myoseptal region that do not express sema Z1a. The growth cones are first projected at about 20 hpf, and after reaching the axial muscles extend caudally between the epidermis and axial muscles of the horizontal myoseptal region (Metcalfe, 1985) (Figs 1C,D, 2E). The coincidence between the pathway of the posterior lateral line growth cones and the Sema Z1a free stripe suggests that Sema Z1a guides these growth cones along their pathway via a repulsive action.

sema Z1a is also expressed by cells at the leading edge of the migrating primordium of the posterior lateral line in front of the growth cones (Fig. 1B). As previously mentioned, the migrating primordium moves caudally together with the lateral line growth cones. Although the nature of the relationship between the primordium and growth cones is presently unclear, the growth cones are never observed in advance of the leading edge of the migrating primordium (Metcalfe, 1985). Thus expression of sema Z1a by the leading edge of the primordium may act to regulate the motility of growth cones so that the growth cones do not overtake the primordium and, therefore, act to keep the two components of the lateral line system in register.

As a first step in examining the function of Sema Z1a, several different mutant embryos that were likely to exhibit alterations in the myotome expression of sema Z1a (see Discussion) were analyzed. sema Z1a expression was examined in the myotomes of floating head (flh) (Halpern et al., 1995), and you-too (yot) (van Eeden et al., 1996) mutant embryos. Both mutant embryos are characterized by a reduction or lack of the horizontal myoseptum and specialized muscle cells called muscle pioneer cells normally found in the horizontal myoseptal region. In both embryos, sema Z1a is uniformly expressed by the entire myotome, and, therefore, there is no longitudinal Sema Z1a free zone in the myotomes (Fig. 2B,C for flh; data not shown for yot).

Since flh and yot embryos lacked the longitudinal sema Z1a free zone that the posterior lateral line growth cones normally follow, the pathways followed by these growth cones were assayed in these mutant embryos by labeling them with the HNK-1 antibody, which labels the lateral line axons and growth cones, or with DiI. In 22 hpf wild-type embryos, the growth cones form tight bundles and appear to be extending sandwiched between and in contact with the lateral border of myotomes and the medial border of the migrating primordium when viewed from a dorsal perspective (Fig. 2E). In contrast, in flh embryos most of the growth cones are defasciculated and appear not to be in contact with the myotome border, having shifted laterally into the migrating primordium (Fig. 2F). Later the lateral line growth cones extended both ventrally and posteriorly onto the yolk sack and tube, and then extended posteriorly ventral to the myotomes (Fig. 2G) rather than upon the myotomes as in wild-type embryos. In each case the aberrantly extending growth cones remained associated with the migrating primordium. The lateral line growth cones followed similar aberrant pathways in yot mutant embryos (not shown). Thus, in flh and yot mutants a change in sema Z1a expression within the myotomes is correlated with alteration of the pathway that lateral line growth cones follow.

Since both mutations likely affect signalling from the axial mesoderm required for the differentiation of the horizontal myoseptal region (Halpern et al., 1995; van Eeden et al., 1996), we next tried to rescue the normal sema Z1a expression pattern in mutant embryos. The myotome phenotype of flh embryos is attributed to the lack of an inductive signal from the notochord since the embryos are missing the notochord (Halpern et al., 1995). One candidate signal is sonic hedgehog (shh) which is a factor secreted by the notochord that is normally involved in patterning of nearby tissues such as the spinal cord (Tanabe and Jessell, 1996). Therefore, overexpression of sonic hedgehog (shh) in flh mutants might induce axial muscles to adopt a horizontal myoseptal fate. This in turn may lead to an inhibition of sema Z1a expression by axial muscles and, therefore, change the pathways followed by lateral line growth cones. To test this, shh RNA was injected into 1- to 8-cell stage flh embryos and they were assayed for sema Z1a expression at 27 hpf or the pathways followed by the lateral line axons at 36 hpf. 68% of injected embryos assayed for sema Z1a expression (n=31) exhibited various degrees of reduction of sema Z1a expression by the myotomes, predominantly by the muscle cells midway between the dorsal and ventral borders of the myotomes flh embryos (Fig. 2D). The putative effect of shh on sema Z1a expression indicated some specificity since expression in many other regions of the embryos was not changed (not shown). In 66% of injected flh embryos in which the lateral line axons were assayed (n=32), the axons extended upon the myotomes along a longitudinal pathway approximately midway between the dorsal and ventral borders of the myotomes (Fig. 2H). So the lateral line axons extended posteriorly in the same region of the myotome in which sema Z1a was inhibited. In contrast, neither the sema Z1a expression pattern (n=28) nor the pathways of the lateral line axons (n=30) were affected in flh embryos injected with green fluorescent protein (gfp) RNA. The examination of flh, yot, and shh overexpressed flh embryos demonstrates that the expression pattern of sema Z1a in the myotomes is tightly correlated with the pathways followed by the lateral line growth cones.

To directly examine the effect of Sema Z1a on the lateral line growth cones, recombinant Sema Z1a was applied directly onto the growth cones in vivo. Since the lateral line growth cones co-migrate with the migrating primordium of the lateral line, conditioned medium containing recombinant Sema Z1a protein or control medium was pressure ejected from a micropipette inserted into the leading edge of the migrating primordium under visual guidance in 26-27 hpf embryos. Embryos were fixed 30 minutes after application of the medium and the morphology of the lateral line growth cones were assayed by labeling them with DiI. Previously we had shown that the medium conditioned for 48 hours by COS cells transiently transfected with cmv-sema Z1a or cmv-sema Z1a^myc collapses chick DRG growth cones just as chick Collapsin 1, and confirmed by western blotting that the conditioned medium contained the Sema Z1a^myc protein (Yee and Kuwada, unpublished data).

The Sema Z1a- or Sema Z1a^myc-conditioned medium but not medium conditioned by COS cells transfected with the expression vector alone had dramatic effects on the lateral line growth cones. Following application of control medium, lateral line growth cones were complex with numerous lamellopodia and filopodia (26-27 hpf embryos, n=11; Fig. 3A). However, after application of Sema Z1a (26-27 hpf embryos, n=8) or Sema Z1a^myc (26-27 hpf embryos, n=7) conditioned media, the growth cones exhibited simple, stick-like morphology (Fig. 3B,C). When recombinant Sema Z1a was injected into the tectum at stages when many retinal growth cones have reached the tectum (50 hpf embryos; n=7), these growth cones appeared unperturbed and similar in morphology to retinal growth cones following injection of control medium (n=6; data not shown). Thus, direct application of recombinant Sema Z1a induces the collapse of a subset of growth cones.

The above static analysis of the effect of recombinant Sema Z1a upon the morphology of the lateral line growth cones was corroborated and extended by a dynamic analysis of the response of the growth cones to applied Sema Z1a. For these experiments a single lateral line growth cone was labeled with DiI, Sema Z1a (n=3) or control medium (n=3) was injected onto the growth cone, and the behavior of the growth cone was recorded with time-lapse video microscopy in 26-27 hpf embryos. By 15 minutes following the ejection of recombinant Sema Z1a, growth cones were already collapsed and for the next 15 minutes the main body of the collapsed growth cone was immotile (Fig. 4). The collapsed growth cones had short filopodia extending from their tips. During this 15 minutes period these filopodia retracted. Thirty minutes following Sema Z1a exposure, the growth cones resumed their extension at a rate comparable to that of control growth cones (Fig. 4C), but the complex morphology of the growth cones recovered much more slowly. This demonstrates that the inhibitory effects of Sema Z1a are transient, and suggests that Sema Z1a may affect the mechanisms that control growth cone morphology and the mechanisms that control axon extension differentially.

The collapse activity of recombinant Sema Z1a upon the lateral line growth cones suggests that these growth cones will change their pathway to avoid cells in the horizontal myoseptal region that misexpress sema Z1a. To test this hypothesis an inducible DNA expression construct controlled by a zebrafish hsp70 promoter (Warren and Kuwada, unpublished data) was used to heat activate the expression of sema Z1a^myc. This construct is useful for examination of axon pathfinding since exposure to elevated temperatures used to induce expression does not by itself induce errors in axonal outgrowth (Warren and Kuwada, unpublished data). The hsp70-sema Z1a^myc DNA construct was injected into 1- to 8-cell stage embryos, embryos were transiently placed at elevated temperatures at 21-23 hpf, and the pattern of sema Z1a^myc expression assayed with a mAb against the c- myc epitope, and the lateral line axons with the HNK-1 antibody, 3 hours later. Control embryos were injected with a hsp70-myc epitope DNA expression construct.

When lateral line growth cones encountered ectopic Sema Z1a^myc-expressing muscle cells in the horizontal myoseptal region (n=27 embryos), they appeared to change their normal pathway to avoid these cells in all cases. Usually (19/27 cases) the growth cones avoided the misexpressing cells but remained within the horizontal myoseptal region (Fig. 5A,B). Occasionally (8/27 cases) the growth cones avoided the misexpressing cells but traversed more dorsal or ventral regions of the myotomes that express sema Z1a (Fig. 5C). Often the axons made sharp turns away from a cell expressing Sema Z1a^myc suggesting that the growth cones were repulsed by the Sema Z1a^myc. Interestingly, the axons often appeared to have extended quite close to the misexpressing cells before turning away. This might signify that effective amounts of Sema Z1a do not diffuse much from the secreting cell. Consistent with this possibility, labeling of Sema Z1a^myc with the mAb against the c-myc eptope was not detected beyond the expressing cells. Growth cones never took such aberrant routes when there was no exogenous Sema Z1a^myc expression by cells in the horizontal myoseptal region (n=35) or when they encountered horizontal myoseptal cells that expressed the myc epitope alone (n=15; Fig. 5D). These results strongly suggest that when posterior lateral line growth cones encounter cells ectopically making Sema Z1a, they will change their pathway to avoid the cells.

In the past several years the role of molecules that act to inhibit the complexity and motility of growth cones have been intensively studied (Tessier-Lavigne and Goodman, 1996). For the most part these molecules have been reported to be involved in the formation of specific neuronal projection patterns and maps (Drescher et al., 1995; Matthes et al., 1995; Messersmith et al., 1995; Puschel et al., 1995; Behar et al., 1996; Nakamoto et al., 1996). This study presents in vivo evidence that an inhibitory growth cone signal may also play an important role in navigating growth cones over long distances in embryos. First, the pathway followed by the growth cones of the posterior lateral line ganglion correlates with the expression pattern of sema Z1a by the axial muscles and is consistent with a repulsive action of Sema Z1a. In wild- type embryos the growth cones extend upon a pathway that is delimited by cells expressing sema Z1a. In flh and yot homozygous embryos sema Z1a expression in the muscles is altered and the pathway followed by the growth cones is concomitantly altered. Furthermore, when a relatively normal expression pattern of sema Z1a is restored in the muscles of flh embryos by overexpression of shh, the growth cones follow a relatively normal pathway. Second, direct application of recombinant Sema Z1a onto the growth cones in living zebrafish embryos demonstrates that Sema Z1a can collapse these growth cones in vivo. Interestingly, following collapse, axons began to extend at relatively normal rates prior to the recovery of a complex morphology by the growth cones. This suggests the possibility that Sema Z1a may in some cases act primarily to control the morphology of the growth cones rather than regulate the rate of axon extension. Third, the growth cones appeared to avoid muscle cells within their normal pathway when these cells are induced to misexpress sema Z1a. In most cases the growth cones veered around the misexpressing cell but stayed within the otherwise sema Z1a free zone. However, in some cases growth cones actually extended beyond their normal pathway onto muscles that normally express sema Z1a. In these cases it is possible that the misexpressing horizontal myoseptal cells generated high enough levels of Sema Z1a to cause the growth cones to actually extend upon muscles that normally express sema Z1a. This suggests that growth cones can detect local differences in Sema Z1a concentration. Supporting this idea, it has been reported that temporal retinal growth cones can habituate to a repulsive signal upon continuous exposure and can respond to small gradients of the signal (Walter et al., 1990a,b).

In addition to the myotome, sema Z1a is expressed at relatively low levels by cells at the leading edge of the migrating primordium. The migrating primordium and the lateral line growth cones comigrate with the growth cones always found in the proximal portion of the primordium (Metcalfe, 1985). Since Sema Z1a is repulsive for these growth cones, it seemed odd that the leading edge of the primordium would secrete Sema Z1a in advance of the growth cones. However, the presumed relatively low amounts of Sema Z1a protein that is secreted may act to primarily regulate the motility of the growth cones so that the growth cones do not overtake the primordium. This would ensure that the primordium and the growth cones retain their spatial relationship as they advance. Although little is really known about the mechanisms that control the formation of the posterior lateral line sensory system, it may rely upon the close association between growth cones and the primordium cells that will give rise to sensory organs.

Our experiments took advantage of the finding that sema Z1a is aberrantly expressed by the entire myotome in flh and yot homozygous embryos, in order to correlate changes in expression of sema Z1a with pathfinding by the lateral line growth cones. These mutants possess abnormal myotomes that are missing the muscle pioneers and horizontal myoseptum (Halpern et al., 1995; van Eeden et al., 1996). Both the muscle pioneers and the horizontal myoseptum can be rescued by transplanting wild-type notochord cells into flh embryos signifying that notochord-derived signals normally induce these structures (Halpern et al., 1995; Odenthal et al., 1996). Thus, the myotome defect including the aberrant expression of sema Z1a is likely a consequence of the missing notochord function in these mutants. This suggests that downregulation of sema Z1a by the horizontal myoseptal region cells is induced by notochord derived signals.

The hedgehog family of secreted molecules are attractive candidates for regulation of sema Z1a. These molecules have been demonstrated to be essential for normal patterning of the ventral neural tube (Chiang et al., 1996; Ericson et al., 1996; Tanabe and Jessell, 1996). In zebrafish shh and echidna hedgehog (ehh) are expressed by the notochord (Kraus et al., 1993; Ekker et al., 1995; Currie and Ingham, 1996). Furthermore, injection of ehh RNA can induce muscle pioneers (Currie and Ingham, 1996). Consistent with the purported role of hedgehogs for regulation of sema Z1a, we found that overexpression of shh leads to downregulation of sema Z1a in the myotomes of flh mutants.

Our experiments clearly demonstrate that Sema Z1a can affect the behavior of the lateral line growth cones. However, it is possible that Sema Z1a may exert its effects upon the growth cones indirectly by regulating migration of the migrating primordium. In fact, the primordium and growth cones remain associated and are affected in identical fashion in the flh and yot embryos and in flh embryos overexpressing shh. The most extreme version of this hypothesis would have Sema Z1a guiding the primordium with the growth cones towed along passively. This seems unlikely since our dynamic analysis demonstrates the active, motile nature of the growth cones. Furthermore, application of recombinant Sema Z1a dramatically reduces the motility of these growth cones. We cannot rule out the possibility that Sema Z1a may act upon the migrating primordium cells which in turn secrete another factor that actually collapses the growth cones. However, the fact that recombinant Sema Z1a can collapse growth cones from chick DRG explants grown on laminin substrates (Yee and Kuwada, unpublished data) suggests that the effect of Sema Z1a upon the lateral line growth cones is likely to be a direct one.

Our experiments demonstrate that Sema Z1a can collapse lateral line growth cones, and that the growth cones will avoid Sema Z1a in vivo. Whether Sema Z1a is necessary for normal pathfinding is still an open question. The recent finding that the phenotypes of sema III/D knockout mice varies with strain suggests that some strains are better able to compensate for the loss of Sema III/D (Behar et al., 1996; Taniguchi et al., 1997). One intriguing possibility is that a different semaphorin may be the basis for this compensation. Interestingly, the zebrafish appears to have two cDNAs, sema Z1a and sema Z1b, that share high sequence homology with semaphorin D/III/collapsin 1 (Yee and Kuwada, unpublished data; Roos, Bernhardt and Schachner, personal communication). sema Z1b is expressed by newly generated myotomes but is then down- regulated so that the anterior myotomes no longer express sema Z1b by 20 hpf (Roos, Bernhardt, and Schachner, personal communication) when the growth cones are first being extended from the posterior lateral line ganglion. Although the distribution of the Sema Z1b protein is not known, the down- regulation of sema Z1b by the anterior myotomes, determined by in situ hybridization, prior to extension of lateral line growth cones upon the myotomes suggests that sema Z1b unlike sema Z1a may not play a significant role in guiding posterior lateral line growth cones.

We thank members and former members of the Kuwada lab, Drs Mary Halloran, Anand Chandrasekhar, James Lauderdale, and James Warren, and Dr Hitoshi Okamoto for many helpful discussions during the course of our experiments. Thanks to Drs Marc Roos, Robert Bernhardt, and Melitta Schachner for sharing unpublished data. Thanks to Dr Marnie Halpern for providing us with flh carriers, and Dr Christiane Nüsslein-Volhard and Pascal Haffter for providing yot embryos. Special thanks to Dr Fengyun Su for supervision of our zebrafish breeding colony. Molecular analysis was in part supported by a NIH grant (M01RR00042) to the General Clinical Research Center at the University of Michigan. The research was supported by a long term postdoctoral fellowship to W. S. from the Human Frontier Science Program Organization and grants from the NINDS (NS24848) and APA (KAI-9603 and KAR1-9704) to J. Y. K.