Volume 228, Issue 3 p. 358-369
Article
Free Access

Two divergent slit1 genes in zebrafish

Lara D. Hutson

Lara D. Hutson

Department of Neurobiology and Anatomy, University of Utah Medical Center, Salt Lake City, Utah

Search for more papers by this author
Michael J. Jurynec

Michael J. Jurynec

Program in Neuroscience, University of Utah, Salt Lake City, Utah

Search for more papers by this author
Sang-Yeob Yeo

Sang-Yeob Yeo

Laboratory for Developmental Gene Regulation, RIKEN Brain Science Institute, Wako, Saitama, Japan

Search for more papers by this author
Hitoshi Okamoto

Hitoshi Okamoto

Laboratory for Developmental Gene Regulation, RIKEN Brain Science Institute, Wako, Saitama, Japan

Core Research for Evolutional Science and Technology (CREST), Japan Science and Technology Corporation (JST), Chuo-ku, Tokyo, Japan

Search for more papers by this author
Chi-Bin Chien

Corresponding Author

Chi-Bin Chien

Department of Neurobiology and Anatomy, University of Utah Medical Center, Salt Lake City, Utah

Department of Neurobiology and Anatomy, 401 MREB, 20 North 1900 East, University of Utah Medical Center, Salt Lake City, UT 84132Search for more papers by this author
First published: 23 September 2003
Citations: 42

Abstract

Members of the Slit family regulate axon guidance and cell migration. To date, three vertebrate slit1 genes have been identified in mammals and orthologs of two, slit2 and slit3, have been identified in zebrafish. Here, we describe the cloning of full-length cDNAs for two zebrafish slit orthologs, slit1a and slit1b. Both predicted proteins contain the conserved motifs that characterize other vertebrate Slits. slit1a and slit1b are both expressed in the midline, hypochord, telencephalon, and hindbrain. Apart from these shared expression domains, however, their expression patterns largely differ. Whereas slit1a is expressed broadly in the central nervous system (CNS) and in the somites, pectoral fin buds, tail bud, and caudal fin folds, slit1b is expressed in the olfactory system throughout embryonic and larval development, and in the retina during larval stages. Their expression patterns, particularly that of slit1a, suggest that Slit proteins may have roles in tissue morphogenesis in addition to their established roles in axon guidance and cell migration. Developmental Dynamics, 2003. © 2003 Wiley-Liss, Inc.

INTRODUCTION

Slit proteins are large, secreted glycoproteins that regulate nervous system development and immune function (reviewed by Zinn and Sun, 1999; Brose and Tessier-Lavigne, 2000). The slit gene was first identified in Drosophila (Rothberg et al., 1988, 1990), and slit homologs have since been identified in vertebrates (Itoh et al., 1998; Nakayama et al., 1998) and C. elegans (Hao et al., 2001). It is now evident that mammals have three slit genes compared with one each in Drosophila and Caenorhabditis elegans. Some years after their initial discovery, Slits were shown by work in both Drosophila and vertebrate systems to be ligands for the Roundabout (Robo) axon guidance receptors (Battye et al., 1999; Brose et al., 1999; Kidd et al., 1999; Li et al., 1999).

The phenotypes of the slit and robo mutants in Drosophila indicate that midline-derived Slit is a repulsive cue that acts as a “gatekeeper” to prevent Robo-expressing axons from crossing the midline of the ventral nerve cord and allow Robo-nonexpressing axons to cross, but only once, because they up-regulate Robo after crossing (Kidd et al., 1999). In contrast, analyses of Slit1 and Slit2 knockout mice (Bagri et al., 2002; Plump et al., 2002) and the zebrafish mutant astray/robo2 (Karlstrom et al., 1996; Fricke et al., 2001; Hutson and Chien, 2002a) indicate that, in the vertebrate forebrain, Slits guide contralaterally projecting axons across the midline, dictating where the commissures form. Paradoxically, in addition to its repulsive activity, Slit also appears to have axon-branching and outgrowth-promoting activity in some contexts (Wang et al., 1999; Sang et al., 2002). Slit can also act on non-neuronal cells. For example, the Drosophila mutant phenotype indicates that Slit regulates migration of muscle cells (Battye et al., 1999; Kidd et al., 1999) and the expression patterns of Slits in mouse suggest that they may also play important roles in limbs, thymus (Yuan et al., 1999), kidney (Piper et al., 2000), and lung (Anselmo et al., 2003).

Drosophila and vertebrate Slits are proteolytically processed, and the different fragments appear to have different signaling properties (Brose et al., 1999; Wang et al., 1999). The leucine-rich repeats, contained in the amino-terminal fragment, are required for rescue of the Drosophila Slit mutant (Battye et al., 1999). Similarly, the amino-terminal fragment of vertebrate Slit2, which contains the leucine-rich repeats, is active in repulsion, collapse, and branching and outgrowth assays (Niclou et al., 2000; Nguyen Ba-Charvet et al., 2001a). An uncleavable variant of Slit2 does have repellent activity; however, it has no activity in collapse or branching/outgrowth assays, and, furthermore, blocks the branching/outgrowth activity of the amino-terminal fragment (Wang et al., 1999; Nguyen Ba-Charvet et al., 2001a). Intriguingly, mammalian Slit1 proteins do not contain the amino acid residues required for cleavage of human Slit2 (Brose et al., 1999), although they do appear to be cleaved at a different site (Whitford et al., 2002).

In addition to directly mediating cell behavior, Slits can also mediate responses to other factors. For example, Slit2 signaling through Robo1 can silence Netrin-1–mediated attractive signaling through DCC in vitro (Stein and Tessier-Lavigne, 2001), and Drosophila Slit signaling through Robo prevents N-cadherin–mediated adhesion (Rhee et al., 2002). Slits also interact with several signaling molecules in addition to the Robos. Slit2 binds with high affinity to laminin (Brose et al., 1999), which may modulate Slit activity (Nguyen-Ba-Charvet et al., 2001). Slit was also identified as a high affinity ligand for glypican, a heparan sulfate proteoglycan (Liang et al., 1999). As glypican interacts with the presumably inactive C-terminal fragment of Slit, it likely acts to tether Slit, thus regulating its distribution. However, the possibility that glypican actually transduces a signal must be considered (Ronca et al., 2001). Finally, Drosophila Slit interacts genetically with several integrins (Stevens and Jacobs, 2002).

In recent years, the zebrafish has become increasingly important for in vivo studies of axon guidance, due to its transparency during embryonic stages and its amenability to gain- and loss-of-function studies (reviewed by Hutson and Chien, 2002b). Orthologs of slit2 and slit3 have been cloned from zebrafish. One or both of these Slits may play roles in convergent extension (Yeo et al., 2001) and retinotectal axon pathfinding (Hutson and Chien, 2002a). To gain a complete understanding of Slit function in the zebrafish, we set out to characterize the remaining Slit family members. Here, we describe the cloning and expression patterns of two novel slit1 orthologs in zebrafish, slit1a and slit1b. Their expression patterns partially overlap in both neural and non-neuronal tissues. However, while slit1a is expressed broadly in the CNS, slit1b is expressed in a more specific pattern, primarily in the olfactory system. Whereas their expression patterns are largely consistent with roles for these Slits in regulating axon guidance and cell migration during development, their expression in non-neuronal tissues suggests that they may play additional roles there.

RESULTS

Slit1a and Slit1b Protein Structures and Map Positions

We have cloned cDNAs for two novel slit genes in zebrafish. The predicted proteins, translated in the longest open reading frames, are 66.6% similar to each other (Figs. 1, 2A; and Supplementary Figure S1, which is available online at http://www.interscience.wiley.com/jpages/1058-8388/suppmat/2003/index.html). Based on protein sequence alignments, along with their clearly identifiable signal peptides, we believe that both cDNAs contain full coding regions. However, as the existing slit1b cDNA lacks a stop codon upstream of the predicted start codon, we cannot rule out the possibility that another start codon lies upstream. Slit1a and Slit1b predicted signal peptides are predicted to be cleaved after amino acids 26 and 30, respectively (von Heijne, 1986). From Pfam analysis, neither protein is predicted to have a transmembrane domain, and both have four leucine-rich repeats, one laminin G domain, and a C-terminal cysteine knot, consistent with other known vertebrate Slit family members. Additionally, while Slit1b contains the expected nine EGF repeats, the eighth repeat in Slit1a is not detected at the highest search stringency. However, this region in Slit1a is highly similar to the other Slit family members (Fig. 1, Supplemental Figure 1), and the difference is unlikely to affect Slit1a function significantly, especially because Slit can signal independently of its EGF repeats (Battye et al., 2001; Nguyen Ba-Charvet et al., 2001a). As for the other known vertebrate Slit1 proteins, Slit1a and Slit1b lack a series of residues demonstrated to be required for cleavage of human Slit2 (Nguyen Ba-Charvet et al., 2001a) (Figs. 1, 2C). Finally, Slit1a contains a region of similarity to the signature motif for the Squash family of serine protease inhibitors (Otlewski, 1990). Of all Slits, vertebrate and invertebrate, Slit1a shows the highest degree of similarity with the Squash consensus motif. However, the first two residues of the signature, cysteine and proline, are lacking in Slit1a, making its biological relevance uncertain (Fig. 2D).

Details are in the caption following the image

ClustalW alignment of predicted zebrafish Slit proteins. Residues identical in all four proteins are indicated with white lettering on a black background. Residues that are similar between at least three of four proteins are boxed, with similar residues in bold. Conserved domains detected by Pfam are underscored and abbreviated as follows: CKn, cysteine knot; EGFR1–9, EGF repeats; LRR1–4, leucine-rich repeats; LamG, Laminin G domain.

Details are in the caption following the image

Phylogenetic analysis of predicted zebrafish Slit proteins based on ClustalW alignment. A: Slit1a and Slit1b are more closely related to human Slit1 than to Slit2, Slit3, or Drosophila or Caenorhabditis elegans Slit. B: Phylogenetic tree of predicted Slit proteins. C: Alignment of peptide sequence required for cleavage of the human Slit2 protein (boxed region). Slit1 proteins lack residues required for cleavage of Slit2. D: Alignment of putative serine protease inhibitor domain in Slit1a. “Signature” denotes residues conserved between Squash family serine protease inhibitors. Active site is indicated with an asterisk. “Consensus” denotes consensus sequence derived from alignment of all family members from the Pfam 4.0 database. Similar residues are boxed.

ClustalW analysis indicates that both cDNAs represent orthologs of slit1 from other vertebrates (Fig. 2A,B). We believe that slit1a, slit1b, slit2, and slit3 are the full complement of zebrafish Slit genes, based on extensive searches of the EST and genome databases. Furthermore, searches of the Fugu genome database (http://www.ensembl.org/Fugu_rubripes/) revealed that this pufferfish also has two slit1, one slit2, and one slit3, with the two slit1 genes being clear orthologs of the zebrafish slit1s (C.-B.C, unpublished data; Fig. 2B). Thus, it is likely that this gene was duplicated early in the teleost radiation. By using radiation hybrid mapping, we mapped zebrafish slit1a to approximately 4.7 cM from the top of linkage group (LG) 13 and slit1b to approximately 43.3 cM from the top of LG22. To further test whether zebrafish slit1a and slit1b are true Slit1 orthologs, we analyzed conservation of synteny near Slit1 genes by using the existing genetic and genomic databases. Zebrafish slit1a, Fugu slit1a, and Fugu slit1b all have immediate neighbors that are clear orthologs of human genes mapping close to SLIT1 at 10q24.1. Together, the combination of sequence similarity and conserved synteny shows unequivocally that we have indeed identified two Slit1 orthologs in zebrafish.

Developmental Expression Patterns

16-cell to dome stage (1.75–3 hours postfertilization [hpf]).

The expression patterns of slit1a and slit1b were analyzed by using whole-mount in situ hybridization between the 16-cell stage and 48 hpf. From the 16-cell stage to dome stage, both mRNAs are expressed ubiquitously (data not shown), suggesting that they are maternally derived.

6–14 somites (12–16 hpf).

At the six-somite stage, both slit1a and slit1b are expressed at low levels in the ventral midline (data not shown). At the 10-somite stage, slit1a continues to be expressed in the ventral midline, specifically the floorplate of the spinal cord and is now expressed in the midbrain, dorsal hindbrain, somites, and tail bud (Fig. 3A). At the 14-somite stage, slit1a continues to be expressed in these same structures (Fig. 3B,C). In the hindbrain, it is expressed most strongly caudal to the mid-hindbrain boundary, in the approximate region of rhombomere 2 (Fig. 3B). At this stage, slit1a is also expressed in somites (Fig. 3C).

Details are in the caption following the image

slit1a and slit1b in situ hybridization at 10- to 18-somite stages. Lateral (A,D,G,I), flat-mount dorsal (B,E,H,J), and transverse views (C,F) of slit1a (A–C,G,H) and slit1b (D–F,I,J) at the 10-somite (A,D), 14-somite (B,C,E,F), and 18-somite (G–J) stages are shown. Rostral is to the left (A,D,G,I) or up (B,E,H,J). Dorsal is up (C,F). At 10 somites, slit1a is expressed in the midbrain, dorsal hindbrain, somites, and tail bud (A) and slit1b in the ventral midline (D). At 14 somites, slit1a is expressed in midbrain and hindbrain (B), and also in the floorplate and somites (C). It is also expressed at low levels in the rest of the spinal cord and in the notochord. E: slit1b extends rostrally to approximately the mid-hindbrain boundary. F: slit1b in floorplate and notochord. G: At 18 somites, slit1a expression in floorplate, hypochord, somites, hindbrain, and midbrain tegmentum, also in tail bud mesenchyme and fin folds. H: The rostral-most extent of slit1a expression in the CNS corresponds approximately to the forebrain-midbrain boundary. I: slit1b in the floorplate and hypochord. J: slit1b expression in hindbrain floorplate is scalloped, presumably due to morphologic segmentation of the hindbrain at this stage. fp, floorplate; hc, hypochord; hb, hindbrain; mhb, mid-hindbrain boundary; not, notochord; sc, spinal cord; som, somites; tb, tail bud; teg, tegmentum. Scale bars = 100 μm in A,B,D,E,G–K, 50 μm in C,F.

At six somites, slit1b is expressed at low levels in the ventral midline (data not shown), and this pattern continues through 10 somites (Fig. 3D). At 14 somites, slit1b continues to be expressed in the ventral midline in both notochord and floorplate (Fig. 3E,F). Unlike slit1a, slit1b is not expressed elsewhere at this stage.

18 somites (18 hpf).

At the 18-somite stage slit1a expression has expanded to include most of the hindbrain and the ventral midbrain (Fig. 3G,H). It continues to be expressed in somites and is now detected in the floorplate, hypochord, tail bud, and caudal fin folds (Fig. 3G).

At 18 somites, slit1b continues to be expressed in the floorplate and, like slit1a, is now expressed in the hypochord but is no longer expressed in the notochord (Fig. 3I,J).

24 hpf.

At 24 hpf, slit1a is expressed in the floorplate (Fig. 4A,C), and is expressed broadly in the forebrain, midbrain, and hindbrain (Fig. 4B,D). This broad expression, however, is not uniform, and is strongest in bilateral clusters of cells in the dorsal telencephalon and in a rostrocaudal stripe in the rostral diencephalon. It is also expressed throughout the tegmentum and in bilateral clusters in the hindbrain. This expression appears to correspond to the earliest-differentiating neurons in the CNS (Chitnis and Kuwada, 1990; Wilson et al., 1990; Ross et al., 1992). slit1a is also expressed in some non-neural tissues, specifically the hypochord, somites of the tail, and tail bud mesenchyme and is expressed at low levels in the tail fin folds (Fig. 4A,C).

Details are in the caption following the image

slit1a and slit1b in situ hybridization at 24 hpf. Lateral (A–C,E–G) and dorsal (D,H) views of slit1a (A–D) and slit1b (E–H). The optic recess is indicated in brain views and is considered the rostral-most point in the central nervous system. A: slit1a is expressed throughout the nervous system, at particularly high levels in the floorplate, and in tail somites. B: slit1a in dorsal telencephalon (arrow 1), rostral diencephalon (arrow 2), tegmentum, and hindbrain. C: slit1a in the floorplate, hypochord, tail somites (som), tail bud mesenchyme, and fin folds (arrowhead). D: slit1a in bilateral cell clusters in dorsal telencephalon and hindbrain. E: slit1b in the hindbrain, floorplate, and hypochord. F: slit1b in the dorsal telencephalon (arrow 3), rostral diencephalon (arrow 4), trigeminal ganglion, and hindbrain. G: slit1b in floorplate, hypochord, and caudal notochord. H: slit1b is primarily expressed in the telencephalon, although it is beginning to be expressed at low levels in the olfactory epithelium at this stage. In the hindbrain, slit1b is expressed in the floorplate and bilateral cell clusters. fb, forebrain; hb, hindbrain; hc, hypochord; not, notochord; op, olfactory placode; OR, optic recess; t, telencephalon; teg, tegmentum; TG, trigeminal ganglion. Scale bars = X in A (applies to A,E), X in B (applies to B,C,F,G), X in D (applies to D,H).

Like slit1a, slit1b continues to be expressed in the floorplate throughout the nervous system (Fig. 4E). slit1b is also expressed in a very small number of cells in the olfactory placodes, in small bilateral clusters in the dorsal telencephalon, and in several cells in the rostral diencephalon (Fig. 4F,H). The cells in the dorsal telencephalon are likely to be the olfactory bulb primordium (Wilson et al., 1990). The location of slit1b-expressing cells in the diencephalon corresponds to the rostral-most slit1a-expressing cells in the diencephalon at this stage. slit1b is also detected in the trigeminal ganglion and in bilateral clusters in the hindbrain (Fig. 4F,H). slit1b is also expressed in some non-neuronal structures, including the hypochord and caudal-most notochord (Fig. 4G).

36 hpf.

At 36 hpf, slit1a is expressed at reduced levels in the floorplate (Fig. 5A). Although it continues to be expressed broadly in the telencephalon, diencephalon, midbrain, and hindbrain, it appears to be somewhat reduced in the dorsal telencephalon relative to 24 hpf. Two predominant patches of expression are observed in the forebrain: one in the ventral telencephalon dorsal to the optic recess, and another in the ventral diencephalon in a patch just rostral to and parallel to the post-optic commissure (Fig. 5C). Also at 36 hpf, slit1a begins to be expressed in the epiphysis, the dorsal-most structure of the diencephalon. From this structure, slit1a expression extends downward and outward, rostrally into the telencephalon, ventrally toward (but not into) the hypothalamus, and ventrocaudally into the tegmentum (Fig. 5C). slit1a is also expressed at low levels in the optic tectum and cerebellar primordium, while it is absent from the hypothalamus (Fig. 5C). slit1a continues to be expressed in some non-neuronal structures, in particular the caudal fin folds, and begins to be expressed in the pectoral fins (Fig. 5B). By this time, slit1a expression has diminished to very low levels in the hypochord and is no longer expressed in the somites (Fig. 5A).

Details are in the caption following the image

slit1a and slit1b in situ hybridization at 36 hours postfertilization (hpf). Lateral (A,C,D,F) and dorsal (B,E,G) views of slit1a (A–C) and slit1b (D–G). Isolated brains (C,F). Rostral is to the left. A: slit1a in floorplate and ventral fin folds (arrowhead). B: slit1a in the telencephalon, hindbrain, and pectoral fins. C: slit1a in the dorsal telencephalon (arrow 1), bilateral patches between the optic recess and anterior commissure (arrow 2), dorsal diencephalon, ventral diencephalon (arrow 3), tegmentum, and cerebellar primordium. D: slit1b in the floorplate and caudal notochord. E: slit1b in two pairs of bilateral cell clusters in the forebrain, trigeminal ganglion, and hindbrain. F: slit1b in dorsal telencephalon (arrow 4). G: slit1b expression in the floorplate is confined to the medial floorplate. cb, cerebellum; di, diencephalon; ep, epiphysis; fb, forebrain; fp, floorplate; hb, hindbrain; hc, hypochord; hy, hypothalamus; not, notochord; op, olfactory placode; OR, optic recess; pf, pectoral fin bud; som, somites; t, telencephalon; TG, trigeminal ganglion; tec, tectum; teg, tegmentum. Scale bars = 100 μm in A (applies to A,D), in B (applies to B,E), in C (applies to C,F), 50 μm in G.

At 36 hpf, slit1b continues to be expressed in the floorplate, telencephalon, diencephalon, and trigeminal ganglion. In the latter three structures, it is expressed at higher levels and in more cells than at 24 hpf, likely due in part to growth of these structures (Fig. 5D–F). In addition, expression is now evident in the tegmentum. However, in contrast to slit1a, which is expressed throughout the tegmentum, slit1b is expressed primarily in the caudal tegmentum (Fig. 5F). Also in contrast to slit1a, slit1b expression is not detected in the dorsal diencephalon, optic tectum, or cerebellar primordium. At 36 hpf, slit1b persists in the caudal-most notochord, but is no longer detected in the hypochord (Fig. 5D). Unlike slit1a, slit1b is not expressed in the pectoral fins (Fig. 5E). slit1b in the floorplate is confined to a single row of cells which corresponds to the medial floorplate (Fig. 5G).

48 hpf.

At 48 hpf, slit1a continues to be expressed at reduced levels in the floorplate (Fig. 6A–C). It also continues to be expressed broadly in the telencephalon, diencephalon, midbrain, cerebellum, and hindbrain. It is also now expressed more strongly in the optic tectum (Fig. 6A), and very low levels are now observed in the retinal ganglion cell layer of the eye (Fig. 6K). slit1a also continues to be expressed in the pectoral fin (Fig. 6B) and caudal fin folds (Fig. 6C) but is now absent from the tail bud mesenchyme (Fig. 6C).

Details are in the caption following the image

slit1a and slit1b in situ hybridization at 48 hours postfertilization (hpf). Lateral (A,C,D,F–L) and dorsal (B,E) views of slit1a (A–C,K) and slit1b (D–J,L) expression. (A,D) Eyes have been removed. A: slit1a in the dorsal telencephalon, dorsal diencephalon, ventral diencephalon, optic tectum, cerebellar primordium, hindbrain, and heart. B: slit1a in the cerebellar primordium, pectoral fin bud, and floorplate. C: slit1a in the floorplate and ventral fin folds (arrowhead). D: slit1b in the telencephalon, tegmentum, hindbrain, trigeminal ganglion, and heart. Boxed region is shown at higher magnification in G. E: slit1b in the trigeminal and posterior lateral line ganglia. F: slit1b in the floorplate. G: The trigeminal (box H) and posterior lateral line (box I) ganglia. Boxed regions are shown at higher magnification in H and I. J: slit1b mRNA is in or adjacent to axons of the anterior commissure. K: slit1a at very low levels in retinal ganglion cell layer of the eye. L: slit1b at high levels in the inner nuclear layer of the eye. AC, anterior commissure; cb, cerebellum; ep, epiphysis; fp, floorplate; h, heart; hb, hindbrain; hy, hypothalamus; OR, optic recess; PC, posterior commissure; pf, pectoral fin; PLLG, posterior lateral line ganglion; t, telencephalon; TG, trigeminal ganglion; tec, tectum. Scale bars = 100 μm in A (applies to A–F), 50 μm in G,J–L, 25 μm in H (applies to H,I).

At this stage, slit1b continues to be expressed in the telencephalon, hindbrain, floor plate, and trigeminal ganglion (Fig. 6D–H). It may also be expressed in the anterior lateral line ganglion, which is in close proximity to the trigeminal ganglion at this stage. slit1b is also expressed in a structure caudal to the otic vesicle (Fig. 6I), likely the posterior lateral line ganglion. slit1b is now also expressed in the heart (Fig. 6D), and at high levels in the inner nuclear layer (INL) of the eye (Fig. 6L). Interestingly, careful examination of slit1b in the forebrain revealed expression that is closely associated with the anterior commissure, either in the axons themselves or in neighboring cells (Fig. 6J). Many of these axons originate in the telencephalon (Wilson et al., 1990), where slit1b is expressed. In contrast, the posterior commissure does not express slit1b (Fig. 6J). These in situ hybridization results are summarized in Table 1.

Table 1. Summary of Zebrafish slit1 Developmental Expressiona
Structure slit1a slit1b
10/14/18 som 24 hpf 36 hpf 48 hpf 10/14/18 som 24 hpf 36 hpf 48 hpf
Olfactory epithelium ± +
Forebrain
 Dorsal telencephalic nuclei NA ++ ++ ++ NA + ++ ++
 Ventral telencephalon NA ++ + + NA
 Dorsal diencephalon NA ± ++ ++ NA
 Ventral diencephalon (excluding hypothalamus) NA ± + + NA
 Hypothalamus NA NA
Midbrain
 Optic tectum NA ± + NA
 Tegmentum −/−/+ ++ ++ ++ −/−/− + ++
Hindbrain
 Cerebellum NA + ++ NA
 Hindbrain floorplate −/−/+ ++ ++ ++ +/+/+ ++ ++ ++
 Bilateral clusters NA ++ ++ ++ NA ++ ++ ++
 Broad expression +/+/+ + ± ± −/−/−
Cranial ganglia
 Trigeminal ganglion NA NA + + +
 Post. lateral line ganglion NA NA + +
 Others NA NA
Spinal cord
 Floorplate +/+/+ + + + +/+/+ ++ ++ ++
 Roof plate −/−/− −/−/−
 Broad expression +/+/+ −/−/−
Notochord +/+/+ +/+/+
Hypochord −/+/+ + −/+/+ ++ ±
Somites +/+/+ + −/−/−
Heart NA ± NA +
Pectoral fin bud NA NA + + NA
Fin folds +/+/+ + + + −/−/−
Tail bud +/+/+ + −/−/−
  • a 10/14/18som, 10, 14, and 18 somites; hpf, hours postfertilization; −, no expression; ±, little or no expression; +, moderate expression; ++, high expression; NA, structure does not exist at this stage.

DISCUSSION

Two Slit1 Genes in Zebrafish

Many mammalian genes have two zebrafish orthologs due to a duplication of the teleost genome that is presumed to have occurred during evolution (Postlethwait et al., 1998). Therefore, the discovery of two slit1 genes is not surprising. slit1a and slit1b appear to be orthologs of human SLIT1, based on nucleotide and amino acid sequence comparisons, and on conserved synteny between the human, zebrafish, and Fugu genomes.

We can also ask whether slit1a and slit1b are functional orthologs of Slit1 genes in other vertebrates. Duplicated genes may be retained in evolution by partitioning the functions of the original gene (Force et al., 1999). One might expect, therefore, the combined phenotypes of mutants for fish slit1a and slit1b to equal the phenotype of a mammalian Slit1 mutant. However, no known zebrafish slit mutants exist to date, and Slit1 knockout mice are virtually normal, at least with respect to CNS axon guidance (Bagri et al., 2002; Plump et al., 2002). Alternatively, gene expression patterns can be used to infer function. That is, the combined expression of slit1a and slit1b might be expected to mirror the expression pattern of mammalian slit1. More specifically, the expression patterns of slit1 between different species should be more similar than between slit1 in one species and slit2 or slit3 in another. We consider several structures in zebrafish and rodents in which slit expression patterns are reasonably well-characterized.

There are many similarities between expression of slit1 genes in zebrafish and rodent. slit1a and slit1b in zebrafish and Slit1 in rat are expressed in the presumptive olfactory bulb, where both species' slit2 and slit3 are expressed at substantially lower levels or not at all (Marillat et al., 2002). Only zebrafish slit1a and mouse and rat Slit1 are expressed in the optic tectum/superior colliculus (Yuan et al., 1999; Marillat et al., 2002). Only zebrafish slit1b and mouse Slit1 in are expressed in the trigeminal ganglion (Yuan et al., 1999). Both zebrafish slit1a and slit2 and rat Slit1 and Slit2 are expressed in the cerebellar primordium while neither zebrafish nor rat slit3 is (Marillat et al., 2002). Finally, zebrafish slit1a and slit2 (Yeo et al., 2001) and mouse Slit1 and Slit2 (Yuan et al., 1999) are expressed in the developing somites.

On the other hand, there are several inconsistencies between expression of slit genes in zebrafish and rodent. First, whereas zebrafish slit1a and slit3 are expressed in developing pectoral fin buds (this article and data not shown), only Slit2 and Slit3 are expressed in developing limb buds in mouse (Yuan et al., 1999). Therefore, during evolution of the paired appendages, the deployment of slit genes seems to have diverged. Second, in zebrafish only slit3 is expressed in the spinal cord roofplate (Yeo et al., 2001), and slit2 is expressed in the dorsal midline of the brain (data not shown), while Slit1 and Slit2 in mouse and rat are expressed in the roofplate and in neural folds of the cortex (Wang et al., 1999; Yuan et al., 1999). In most vertebrates, the neural tube forms as the neural folds fuse dorsally to become the roofplate, whereas in zebrafish, the neural tube forms from the neural keel by cavitation at comparatively late stages of development (Papan and Campos-Ortega, 1994). Thus, the differences in gene expression may reflect these differences in morphogenesis of the neural tube.

Overall, the slit1a and slit1b expression patterns are more similar to Slit1 than they are to either Slit2 or Slit3 in mouse and rat. This further supports the proposition that slit1a and slit1b are indeed Slit1 orthologs.

Slit1a and Slit1b Structure: Functional Implications

Slit1a and Slit1b predicted proteins contain all of the expected motifs conserved in vertebrate Slits. Like Slit1 proteins in other vertebrates, Slit1a and Slit1b also lack the amino acid residues required for cleavage of Slit2. However, they may be cleaved at a different location, as is mammalian Slit1 (Whitford et al., 2002). An unexpected finding was that Slit1a contains a domain similar to the signature motif for the Squash family of serine protease inhibitors. Although it is not known whether this motif is functional, its presence is intriguing in light of the importance of serine proteases for neurite extension (Monard, 1988) and cell migration (Tani et al., 2001). The significance of these structural features must be addressed by using functional assays.

Slit1a and -1b Expression Patterns Suggest Novel Functions

Slits are well known for their roles in axon guidance, and the expression patterns of slit1a and slit1b in the CNS and peripheral tissues are consistent with these roles. We were surprised to find that slit1b mRNA is localized to the anterior commissure, indicating that it is either expressed in a narrow band of cells closely associated with the commissure or that it is specifically targeted to commissural axons. Given that local protein translation in growth cones can be important for axon guidance in some contexts (Campbell and Holt, 2001; Brittis et al., 2002), the latter interpretation suggests a novel role for Slits as proteins secreted from growth cones. Growth-cone–derived Slit1b could act either in an autocrine or paracrine manner, although it is conceivable that it is provided anterogradely to target cells, as has been demonstrated for the neurotrophins (reviewed by von Bartheld et al., 2001).

slit1a and slit1b expression in non-neuronal tissues also suggests roles in developmental processes in addition to axon guidance. Ubiquitous misexpression of zebrafish Slit2 during early embryonic development can cause defects in convergent extension (Yeo et al., 2001). If, indeed, Slits normally play a role in gastrulation, Slit1a and Slit1b are good candidates, because their mRNAs are both supplied maternally. Slits can also affect vascular endothelial cell migration in vitro (Park et al., 2003). Our results are consistent with a role for slit1a and 1b derived from the notochord, hypochord, and/or somites in the regulation of vascular development. slit1b expression in heart might reflect a role in heart development. Finally, slit1a is expressed in early developing pectoral fin buds. While this Slit may act to guide axons, it could also regulate morphogenesis or patterning of the fin buds.

Summary

The similarities between the Slit1a and Slit1b predicted proteins indicate that they are likely to have similar molecular properties. However, differences in expression suggest that, although their coding regions are largely similar, their promoters and enhancers may have diverged more substantially, leading to differences in transcriptional regulation. Future functional tests of both genes will determine how their functions have diverged, and to what extent they regulate other developmental processes in addition to axon guidance and cell migration.

EXPERIMENTAL PROCEDURES

Fish Care

Fish were maintained and bred under standard conditions and in accordance with University of Utah IACUC Protocol A3031-01. Strains used were primarily Tübingen and Tup Longfin. Embryos were staged according to Kimmel et al. (1995) as equivalent hours postfertilization at 28.5°C.

Cloning and Sequence Analysis

A 991 nt, coding fragment near the 5′ end of the slit1a coding region was obtained by screening a 22–24 hpf random-primed lambda-gt11 library at low stringency using a probe derived from a ∼1.5-kb BglII fragment of zebrafish slit2 cDNA (see Yeo et al., 2001). Additional slit1a cDNA sequence was cloned by reverse transcriptase-polymerase chain reaction (RT-PCR) from 36 hpf total RNA and primers were designed to recognize putative slit1a exons predicted from sequence found in the Sanger Centre Trace Repository (http://trace.ensembl.org/). PCR with primers S1A3F (ATGCCTGCGAGAGGAGGATG), derived from known slit1a sequence, and S1A5R (GAACCTCTTTGCAGGCCG-TG), derived from a putative slit1a exon, yielded a 2,716-nt fragment. PCR with primers S1A5F (CAATCCTCTGCACTGTGACTG) and S1A2R (GTTGATGTACAGATTCTGGATGC), both derived from predicted slit1a exons, yielded a 1,357-nt fragment. The two fragments overlap and together include most of the coding region beginning with the predicted start codon. We also performed RT-PCR to clone 143 bases of 5′ untranslated region using primer S1A3R (CTCGGTGTTTCTGGGAATGTTC), derived from known slit1a sequence, and S1A1F (CGCTGGAGCAACTCTAGTAG), derived from genomic sequence upstream of the start codon. To clone the 3′ end, we used nested upstream primers S1A7F (GAAGCTCCGCTGT-ATGTAGGAGGCATG) and S1A8F (GGGTCTGGCAGATCCAGAACAG-CTC) and amplified using SMART rapid amplification of cDNA ends (Clontech). All PCR reactions used 20–25 cycles with Pfu Turbo (Stratagene). For all PCR reactions except for S1A3F/S1A5R, three independent clones were sequenced.

Two different overlapping slit1b fragments were initially cloned by degenerate PCR using CODEHOP-designed (http://www.blocks.fhcrc.org/codehop.htm) primer pairs (CACCTGCACCGGCACCACHGTBGAYTG or GCGCCTTCCACGACATGAARSARCTSGA, upstream; CTCGGACAGCCAGGCCAGRTGRCARTC, downstream). One of these clones was used to generate a probe to screen a 22–26 hpf random-primed lambda-gt11 library (kindly provided by Dr. Motoko Aoki), from which clone 17E1-3 containing the full coding region was obtained.

Sequencing was performed in the University of Utah Sequencing Core Facility, and both strands of the coding regions were sequenced. The alignment in Figure 1 was performed by using Multalin (Corpet, 1988). Protein motifs were determined by scanning the Pfam-A 4.0 database (Bucher and Bairoch, 1994; Hofmann et al., 1999) using PFSCAN (pftools version 2.2, June 1999, Philipp Bucher, [email protected]) through the San Diego Supercomputing Center Biology Workbench (http://workbench.sdsc.edu).

slit1a was mapped on the T51 radiation hybrid panel (Geisler et al., 1999) using primers S1A3F and S1A3R. Results were submitted to the Geisler lab for assignment of map position (http://www.map.tuebingen.mpg.de/). slit1b was mapped by the laboratory of Dr. Leonard Zon (http://134.174.23.167/zonrhmapper).

In Situ Hybridization

Embryos were fixed for 1 day at 4°C or between 1 hr and overnight at 22°C, then dehydrated through a methanol series and stored at −20°C until use. Whole-mount in situ hybridizations were performed by using standard methods (Lee et al., 2001). slit1a and slit1b probes were generated by digesting plasmids pS1A11.35 and p17E1-3 with NotI and SmaI (NEB) and transcribing with SP6 and T7 (Life Technologies), respectively. Probes were used at roughly 200 ng/ml. We occasionally performed in situ hybridizations with RNaseA (Sigma) treatment after hybridization. The hybridization pattern observed under these conditions was identical to that without RNaseA treatment.

All samples were mounted in 80% glycerol and viewed by using differential interference contrast optics on an Olympus BX50WI Compound microscope. Digital images were captured by using an Olympus Magnafire camera on a Macintosh G4 computer. In-focus regions from multiple focal planes were combined using Adobe Photoshop.

Acknowledgements

We thank Dr. Leonard Zon's laboratory for mapping slit1b, and the Geisler lab for assistance with mapping slit1a. We also thank Jeong-Soo Lee and Cornelia Fricke for assistance with the library screen, Amy Kugath for general technical assistance, and Douglas Campbell, Arminda Suli, and Andrew Pittman for comments on the manuscript.