Neurons throughout the vertebrate nervous system selectively activate the gene for a growth cone component, GAP-43, during embryonic development, and then decrease its expression abruptly as they form synapses. Distal interruption of mature axons in the central nervous system (CNS) of fish and amphibians, but not in the mammalian CNS reverses the developmental down-regulation of GAP-43 expression. To explore functional conservation and divergence of cis-acting elements that regulate expression of the GAP-43 gene, we studied activation, in transgenic zebrafish embryos, of mammalian GAP-43 genomic sequences fused to a marker gene. The DNA fragments containing the GAP-43 promoter, including a short fragment of 386 base pairs, were preferentially activated in the embryonic fish nervous system at times when extensive neuronal differentiation and neurite outgrowth take place.

After 2 days of development, expression of the mammalian transgenes was specifically downregulated in the fish spinal cord but increased in more rostral regions of the CNS. This expression pattern was well correlated with the regulation of the endogenous fish GAP-43 gene revealed by in situ hybridization. Elements of the mammalian gene located a substantial distance upstream of the minimal promoter directed additional expression of the marker gene in a specific set of non-neural cells in zebrafish embryos. Our results indicate that cis-acting elements of the GAP-43 gene, and signaling pathways controlling these elements during embryonic development, have been functionally conserved in vertebrate evolution.

Various stages of neuronal differentiation are characterized by expression of discrete sets of genes, presumably in response to distinct environmental or intracellular signals. One example is the gene for a major component of neuronal growth cones, GAP-43, which is expressed in most neurons shortly after their final mitosis, coincident with the onset of axon elongation (Skene, 1989; Biffo et al., 1990). As axons complete their initial outgrowth and establish synaptic terminals during development, most neurons selectively repress expression of the single-copy GAP-43 gene (Basi et al., 1987; Bendotti et al., 1991; Meberg and Routtenberg, 1991). Disruption of axons, often can reverse the chronic repression of this gene in adult neurons (Skene and Willard 1981a,b; Benowitz and Lewis, 1983; Perry et al., 1987; Hoffmann, 1989). These characteristic aspects of GAP-43 regulation suggest that its gene contains regulatory elements which allow it to be selectively induced at an appropriate stage of neuronal differentiation, and then to respond to signals that reversibly modulate its expression within differentiated neurons.

Comparison of the regulatory regions of various neuronal genes has revealed that several such genes contain relatively promiscuous core promoters, and that neuron-specific expression is achieved by the selective derepression of a common silencer element located upstream of the promoter (Kraner et al., 1992; Mori et al., 1992; Mandel and McKinnon, 1993). In contrast, a short fragment of DNA containing the core promoter of the mammalian GAP-43 gene, but lacking the previously identified silencer sequence, appears to confer neural-selective expression of this gene (Nedivi et al., 1992). This suggests that the cell type specificity of the GAP-43 gene is controlled by a signaling pathway distinct from those characterized so far. Here we examine whether activation of this novel signaling pathway is an evolutionally conserved feature of vertebrate neuronal differentiation.

In addition to its selective induction during neural differentiation, the GAP-43 gene responds to signals that chronically repress expression in most mature neurons. This reversible repression exhibits significant phylogenetic divergence between the central nervous systems (CNS) of mammals and fish. Interruption of axons far from their cell bodies reinduces GAP-43 synthesis in the CNS of fish and amphibians, which show a high capacity for regeneration (Benowitz et al., 1981; Skene and Willard, 1981b; Grafstein, 1986; LaBate and Skene, 1989), but not in the mammalian CNS (Redshaw and Bisby, 1984; Doster et al., 1991; Tetzlaff et al., 1991; Skene, 1992). This indicates an evolutionary divergence of some step in the signaling pathways controlling expression of the GAP-43 gene in mature neurons.

To explore the evolutionary conservation or divergence of signaling pathways controlling the developmental regulation of GAP-43 expression, we have examined the ability of mammalian (rat) GAP-43 promoter/enhancer elements to control expression of a β-galactosidase marker enzyme in transgenic zebrafish.

Animals

Zebrafish were obtained from the Oregon AB line, or in some cases from the golbl line which develops less pigmentation. Embryos were maintained as described previously (Westerfield, 1993) and staged by hours postfertilization at 28.5°C (h).

Preparation of transgenes

Fragments of the 5 ’ untranslated region of the rat GAP-43 gene, containing the GAP-43 core promoter and/or adjacent regulatory elements, were cloned into the unique HindIII site of the pCH 126 expression vector (generously provided by F. Lee, DNAX Corporation, Palo Alto, CA). The plasmid pCH 126 contains a functional gene for β-galactosidase linked to SV 40 polyadenylation signal sequences. Four fragments were used to generate transgenic zebrafish (Fig. 2): (A) A 386 base pair (bp) fragment (position –273 to 113 of the rat GAP-43 5 ’ untranslated region; Nedivi et al., 1992) included a proposed transcription start site and consensus sequences. (B) A 1 kb fragment (–522 to 470) encompassed the 386 bp fragment, an unusual homopurine stretch 3 ’ to the 386 bp fragment, and alternating purine and pyrimidine residues upstream of the 386 bp fragment. This 1 kb fragment was cloned into pCH 126 leaving only the HindIII site 5 ’ of the 1 kb insert intact. This allowed addition of upstream sequences. (C) 4 kb of unsequenced upstream GAP-43 genomic DNA (–520 to ∼–4500) was inserted 5 ’ of the 1 kb fragment using the unique HindIII site. In addition, the 4 kb fragment alone (D), without the GAP-43 minimal promoter, was inserted into the pCH 126 plasmid. As a control, a plasmid was used that contained Rous sarcoma virus (LTR-RSV) promoter sequences regulating the expression of the β-galactosidase gene (gift of S. Meakin, Stanford University, Stanford, CA).

Microinjection and analysis of transgene expression

Supercoiled plasmid DNA or isolated fragment was injected into fertilized zebrafish embryos prior to first cleavage essentially as described previously (Stuart et al., 1988). Approximately 1.5-3 nl, corresponding to ∼20 fmol of each promoter construct, was injected per embryo. This produced a reasonable number of transgene-expressing cells in each embryo. To control for the injected volume, 32P-dCTP was added to the injection solution so that each embryo contained approximately 200 cts/minute. Embryos with injected amounts of radioactivity that were more than 30% different from the average, or embryos that showed obvious malformations, were discarded. To determine the patterns of β-galactosidase expression, embryos were fixed and β-galactosidase activity visualized as described previously (Westerfield et al., 1992). Individual expressing cells were counted and assigned to specific tissues, according to their locations and morphologies. Cells that could not be identified or were found in tissues that only occasionally expressed the transgenes (e.g. notochord, notochord ensheathing cells, pronephric duct, etc.) were counted as ‘others’. Dead cells were identified by their refractile appearance in embryos under Nomarski differential interference contrast optics (Grunwald et al., 1988). To obtain a map of the spatial and early temporal expression pattern of a specific transgene, mean values of positive cells per tissue for a large number of embryos were calculated at 30 h and 55 h (386 bp: 30 h, 60 embryos; 55 h, 48 embryos; 1 kb: 30 h, 63 embryos; 55 h, 45 embryos; 1+4, 30 h, 67 embryos; 55 h, 58 embryos). Means were compared to each other with an independent t-test at the P = 0.05 level. For comparison of tissue specificities among promoter fragments, we calculated the percentages of positive cells located in various tissues of the embryo. In this analysis, we did not include expression of the transgenes in either the yolk cell or in cells of the enveloping layer (EVL), because both of these cell types arise as lineages distinct from the rest of the embryo very early in development (Kimmel and Law, 1985) and because they variably expressed all transgene constructs tested, including constructs with promoters in reverse orientation.

Embryos that were assayed for β-galactosidase activity using onitrophenyl-β-D galactopyranoside (ONPG) were pooled after determination of the injected radioactivity. A minimum of 20 embryos were homogenized by sonication in 150 μl of FT buffer (0.25 M sucrose, 10 mM Tris-HCl pH 7.4, 10 mM EDTA). Homogenates were centrifuged in a microcentrifuge and the clear supernatant diluted to a total volume of 300 μl with Z buffer (60 mM Na2HPO4, 40 mM NaH2PO4, 10 mM KCl, 1 mM MgCl2, 50 mM β-mercaptoethanol). Eighty μl of ONPG (4 mg/ml in 60 mM Na2HPO4, 40 mM NaH2PO4) were added and the reaction carried out at 37°C, until a yellow shade was obvious. Color development was stopped by adding 100 μl of 1 M NaCO3 and the O.D. determined at 420 nm.

In situ hybridization

We examined whole-mount embryos by in situ hybridization between the ages of 12 and 60 h at intervals of 4-8 hours. Approximately 1020 embryos were examined at each developmental stage. Probes for in situ hybridization were generated from either a zebrafish cDNA generously provided by Dr Mark Fishman (Y. Lin and M. C. Fishman, GenBank accession number L27645), or a 280 bp EcoRI-BipEI fragment from the highly conserved 5 ’ end of a goldfish GAP-43 cDNA (LaBate and Skene, 1989). The zebrafish cDNA sequence was 65% identical to the previously published goldfish sequence at the nucleic acid level, and the predicted proteins were 61% identical over the full coding regions. Over the more highly conserved region used for the goldfish probe, the zebrafish and goldfish sequences were 83% identical at the nucleic acid level and 89% identical at the protein level. Antisense digoxigenin-labeled RNA probes were synthesized from the GAP-43 fish cDNAs after linearizing the plasmids (pBlue-script KS –) with XbaI using T3 RNA polymerase and an RNA labeling kit (Boehringer-Mannheim) according to the manufacturer’s directions. The full-length zebrafish probe was hydrolyzed for 50 minutes at 60°C in 400 mM N2CO3 and 600 mM NaHCO3. Approximately 100 ng of probe was used in 200 μl of hybridization solution.

In situ hybridization with the zebrafish probe was carried out as described by Oxtoby and Jowett (1993). In situ hybridization with the goldfish probe on 55 h embryos was performed as described by Püschel et al. (1992) with the exception that proteinase K treatment was extended up to 30 minutes. To doubly label embryos for GAP-43 RNA and β-galactosidase activity, injected embryos were fixed overnight in 4% paraformaldehyde in PBS at 4°C. The embryos were rinsed with PBS, incubated in X-gal solution (as described in Westerfield et al., 1992) for 2 hours at 37°C, and rinsed again. Embryos were postfixed for 30 minutes at room temperature, rinsed and placed directly into prehybridization solution (without methanol and proteinase K treatments). Embryos were then hybridized and the probes detected as described above.

Expression of the endogenous GAP-43 gene

During mammalian development, GAP-43 expression is first apparent in postmitotic neurons, coinciding with the onset of neurite outgrowth, and is confined almost exclusively to the nervous system (Skene, 1989; Biffo et al., 1990). To examine the expression of this gene during zebrafish embryonic development, we analyzed the distribution of the endogenous GAP-43 mRNA by in situ hybridization. Whole-mount embryos were hybridized with probes complementary to zebrafish (Yi Lin and Mark Fishman, unpublished; GenBank accession number L27645) and goldfish GAP-43 mRNA (LaBate and Skene, 1989). We first detected GAP-43 mRNA labeling by 17-18 h, the time that the earliest primary neurons are starting to grow axons (Kimmel and Westerfield, 1990). By 24 h, groups of cells in the forebrain expressed GAP-43 in regions that probably correspond to the previously identified telencephalic and diencephalic bands of acetylcholinesterase activity (Wilson et al., 1990) and in the presumptive epiphysis and nucleus of the posterior commissure (Fig. 1A). Segmentally reinterated groups of cells in the hindbrain expressed GAP-43 at this time, and in the caudal hindbrain, GAP-43 expression appeared in more or less continuous longitudinal columns of ventrolaterally located cells. These cells probably correspond to the early differentiating neurons in this region as shown previously by antibody labeling (Trevarro et al., 1990). Adjacent to the hindbrain, cells in the developing trigeminal (Figs 1C, 3D) and lateral line ganglia (Fig. 1C) also expressed GAP-43. At this developmental stage, GAP-43 mRNA was apparent in the spinal cord, where individual Rohon-Beard neurons showed distinct hybridization (Figs 1F, 5A), as well as more ventrally located cells that based on their sizes and locations (Kimmel and Westerfield, 1990) probably included primary motoneurons and interneurons (Figs 1G, 5A).

Fig. 1.

The regions of the brain containing the earliest neurons specifically express the GAP-43 gene. In situ hybridization of whole-mount zebrafish embryos with a probe complementary to the zebrafish GAP-43 cDNA. (A) Clusters of cells express the gene in the telencephalic (t) and diencephalic (d) bands, the epiphysis (e), the nuclei of the posterior commissure (pc) and the medial longitudinal fascicles (m), and in the hindbrain (h) and spinal cord, at 24 h. (B) More cells express the GAP-43 gene at 48 h. The hindbrain clusters have expanded into columns (arrowheads) and expression is now apparent at the midbrain/hindbrain junction (arrow). (C) Cells of the trigeminal (tr), anterior (a), and posterior (p) lateral line ganglia express the GAP-43 gene at 24 h. Expression in the hindbrain appears in reiterated sets of cells (C) located in the center of each rhombomere as shown in (D) for rhombomeres 5-7. (E) By 48 h, the hindbrain cells expressing GAP-43 appear in medial and lateral groups and expressing cells are present at the midbrain/hindbrain junction (arrow). The arrowhead (p) indicates the posterior lateral line ganglion. (F) Rohon Beard neurons near the dorsal surface of the spinal cord (F) and more ventrally located neurons (G) express GAP-43 at 24 h. (A,B) Sagittal views, anterior to the left; (C-G) dorsal views, anterior to the left; scale bar, 190 μm in A and B, 160 μm in C, 60 μm in D,F and G, and 256 μm in E.

Fig. 1.

The regions of the brain containing the earliest neurons specifically express the GAP-43 gene. In situ hybridization of whole-mount zebrafish embryos with a probe complementary to the zebrafish GAP-43 cDNA. (A) Clusters of cells express the gene in the telencephalic (t) and diencephalic (d) bands, the epiphysis (e), the nuclei of the posterior commissure (pc) and the medial longitudinal fascicles (m), and in the hindbrain (h) and spinal cord, at 24 h. (B) More cells express the GAP-43 gene at 48 h. The hindbrain clusters have expanded into columns (arrowheads) and expression is now apparent at the midbrain/hindbrain junction (arrow). (C) Cells of the trigeminal (tr), anterior (a), and posterior (p) lateral line ganglia express the GAP-43 gene at 24 h. Expression in the hindbrain appears in reiterated sets of cells (C) located in the center of each rhombomere as shown in (D) for rhombomeres 5-7. (E) By 48 h, the hindbrain cells expressing GAP-43 appear in medial and lateral groups and expressing cells are present at the midbrain/hindbrain junction (arrow). The arrowhead (p) indicates the posterior lateral line ganglion. (F) Rohon Beard neurons near the dorsal surface of the spinal cord (F) and more ventrally located neurons (G) express GAP-43 at 24 h. (A,B) Sagittal views, anterior to the left; (C-G) dorsal views, anterior to the left; scale bar, 190 μm in A and B, 160 μm in C, 60 μm in D,F and G, and 256 μm in E.

During the second day of development, the pattern changed only slightly with more cells throughout the nervous system expressing GAP-43. The initial pattern, however, was still apparent, with the bands of expressing cells in the forebrain, segmentally arranged clusters of expressing cells in the hindbrain, and individual cells in the spinal cord. More cells began expressing GAP-43 after 24 h in anterior regions than in more posterior regions. By 48 h, bands of cells along the anterior wall of the midbrain/hindbrain junction and the anterior border of the midbrain expressed GAP-43 (Fig. 1B). In the rostral hindbrain, the segmental clusters of cells extended dorsally to form vertical columns of cells expressing GAP-43 (Fig. 1B) and medial and lateral cell groups of the clusters were apparent (Fig. 1E) similar to previous descriptions of neurons expressing the zn-5 antibody in this region (Trevarrow et al., 1990). Beginning around 48 h, the number of GAP-43-expressing cells in the spinal cord began to drop and we could detect little or no expression by 60 h (Figs 3B, 5B). At the same time, however, the number of expressing cells in more anterior regions of the nervous system continued to increase. At each developmental stage, expression was variegated, there were both expressing and non-expressing cells present in each region of the CNS. In the spinal cord, for example, Rohon Beard neurons were positive for the GAP-43 message at all stages in which positive cells were present, even though other cells were negative. Thus, in zebrafish as in mammals, the GAP-43 gene is expressed selectively in the nervous system, coinciding spatially and temporally with the differentiation of neurons and the outgrowth of axons. With continued maturation, as in the spinal cord after 48 h, expression of the gene is repressed in many neurons. The general features of neural selective expression and temporal regulation of the GAP-43 gene, therefore, are functionally conserved between fish and mammals.

DNA constructs

To determine whether trans-acting factors in embryonic fish neurons could recognize regulatory elements from the mammalian GAP-43 gene, we used partially characterized fragments, spanning the GAP-43 promoter/enhancer region (Nedivi et al., 1992) to control the expression of β-galactosidase in zebrafish embryos (Fig. 2). A 386 bp fragment (Fig. 2A) contained the basic GAP-43 core promoter. Transcription of the marker gene controlled by this core promoter was strictly orientation-dependent in vitro, as well as in transient expression assays in transgenic zebrafish. A second construct (1 kb fragment; Fig. 2B) encompassed the 386 bp core promoter and contained additional regions previously shown to influence expression of the marker gene in in vitro assays in mammalian cell lines (Nedivi et al., 1992). The largest construct (1+4 kb) included the 1 kb fragment plus an additional 4 kb of 5 ’ DNA (Fig. 2C). In order to verify that no additional promoter(s) were located within the extra 4 kb of genomic GAP-43 DNA, we also analyzed the regulation of the 4 kb construct (Fig. 2D) that lacked the basic GAP-43 promoter. The promoter of the Rous sarcoma virus (RSV), linked to β-galactosidase served as a control.

Fig. 2.

Fragments of the rat GAP-43 promoter/enhancer region analyzed in transient expression assays in transgenic zebrafish. Diagrams of the transgenes are aligned with a map of the 5 untranslated region of the GAP-43 gene. The positions of repeating sequences (GA, light hatched box and GT, dark hatched box) are indicated for reference, and the length of each promoter fragment (in base pairs; A-D) is indicated to the left. Downward arrows indicate restriction sites, and the tall arrow indicates the position of the proposed transcription start site. The 386 bp fragment (A) contains the GAP-43 core promoter. The 1 kb (B) and 1+4 kb (C) fragments include the core promoter. The fourth fragment (D) contains 4 kb of 5 ’ DNA and lacks the 386 bp core promoter. The broken lines show a region that is not drawn to scale.

Fig. 2.

Fragments of the rat GAP-43 promoter/enhancer region analyzed in transient expression assays in transgenic zebrafish. Diagrams of the transgenes are aligned with a map of the 5 untranslated region of the GAP-43 gene. The positions of repeating sequences (GA, light hatched box and GT, dark hatched box) are indicated for reference, and the length of each promoter fragment (in base pairs; A-D) is indicated to the left. Downward arrows indicate restriction sites, and the tall arrow indicates the position of the proposed transcription start site. The 386 bp fragment (A) contains the GAP-43 core promoter. The 1 kb (B) and 1+4 kb (C) fragments include the core promoter. The fourth fragment (D) contains 4 kb of 5 ’ DNA and lacks the 386 bp core promoter. The broken lines show a region that is not drawn to scale.

Transient expression of transgenes

We visualized the activity of the various mammalian promoter fragments in fish by injecting circular or linear plasmid DNA into zebrafish embryos at the 1- to 2-cell stage and then assaying β-galactosidase expression in the embryos after fixation at 30 h and 55 h. We observed a variegated expression pattern of the transgenes, probably due to an uneven, mosaic distribution of the injected DNA (Westerfield et al., 1992). The degree of variegation was independent of whether the injected DNA was circular or linear.

The number of cells expressing the transgene varied from 12 cells to as many as 50 cells in individual embryos. For a given amount (copy number) of injected DNA, the average number of positive cells per embryo depended on the DNA construct tested. Injection of approximately 20 fmol of the 1+4 GAP-43/ β-galactosidase fusion gene produced a mean of 11.7±1.3 cells per embryo expressing the transgene (n=64) at 30 h. Both, the 386 bp and 1 kb GAP-43 promoter constructs were significantly less activated, with a mean of 4.4±0.59 (n=59) and 3.8±0.49 (n=73) cells per embryo after injection of the same amount of DNA. This result, that zebrafish cells expressed the longest GAP-43 transgene more strongly than the shorter constructs, was reinforced by quantitative assays of total β-galactosidase activity in whole-embryo homogenates. Embryos injected with the 1+4 transgene consistently yielded homogenates with 1.3 to 1.8 times higher enzyme activity than embryos injected with the two smaller constructs (not shown). However, the 4 kb fragment alone, fused to the gene for β-galactosidase, was unable to direct expression of the marker gene, indicating that an additional GAP-43 promoter is absent from sequences upstream of the 1 kb fragment.

Neuronal expression

We first examined whether activation of the rat GAP-43 genomic fragments was restricted to the zebrafish nervous system, as previously described for endogenous GAP-43 expression in mammals (Skene, 1989). In embryos containing a large number of cells expressing the transgene, it was possible to visualize the tissue distribution of transgene expression within a single embryo (Fig. 3A,C). The majority of animals, however, contained only a small number of positive cells per embryo. We therefore plotted the distribution of all positive cells in a large number of embryos injected with each DNA construct. We found that zebrafish embryos preferentially activated rat GAP-43 promoter/enhancer elements in the nervous system at stages when large numbers of neurons become postmitotic, extend neurites, and express the endogenous GAP-43 gene (Fig. 3). At 55 h, for example, more than 70% of cells expressing the transgenes containing the 386 bp or 1 kb fragment were in the CNS (brain and spinal cord; Fig. 4). Cells expressing the transgenes often appeared in restricted regions of the nervous system (Fig. 3A,C,D) corresponding to sites of expression of the endogenous GAP-43 gene revealed by in situ hybridization (Fig. 3B,D).

Fig. 3.

The patterns of transgene and endogenous GAP-43 sense RNA expression are very similar. (A) In 60 h embryos, the 386 bp fragment is activated almost exclusively in the developing brain, visualized by the blue reaction product of β-galactosidase. Only minor expression appears in the spinal cord. (B) Distribution of the endogenous GAP-43 mRNA is almost identical to the pattern of GAP-43 transgene expression at the same developmental stage. Hybridization to the goldfish antisense probe appears predominantly in the brain (dark blue reaction product), whereas posterior regions of the embryo, including the spinal cord, show no obvious hybridization signals. The brown cells in the head and tail regions are pigment cells. (C) High magnification of a zebrafish brain activating the 1 kb fragment at 30 h. Arrows indicate axons of individual neurons that contain the β-galactosidase reaction product. (D) Double labeling for expression of the 386 bp β-galactosidase construct (blue reaction product) and the endogenous GAP-43 gene (purple/gray reaction product) in a 24 h embryo. Several trigeminal ganglion neurons, one with a process (arrow), at the posterior end of the ganglion express the transgene. Scale bars: 200 μm in A-C; 60 μm in D.

Fig. 3.

The patterns of transgene and endogenous GAP-43 sense RNA expression are very similar. (A) In 60 h embryos, the 386 bp fragment is activated almost exclusively in the developing brain, visualized by the blue reaction product of β-galactosidase. Only minor expression appears in the spinal cord. (B) Distribution of the endogenous GAP-43 mRNA is almost identical to the pattern of GAP-43 transgene expression at the same developmental stage. Hybridization to the goldfish antisense probe appears predominantly in the brain (dark blue reaction product), whereas posterior regions of the embryo, including the spinal cord, show no obvious hybridization signals. The brown cells in the head and tail regions are pigment cells. (C) High magnification of a zebrafish brain activating the 1 kb fragment at 30 h. Arrows indicate axons of individual neurons that contain the β-galactosidase reaction product. (D) Double labeling for expression of the 386 bp β-galactosidase construct (blue reaction product) and the endogenous GAP-43 gene (purple/gray reaction product) in a 24 h embryo. Several trigeminal ganglion neurons, one with a process (arrow), at the posterior end of the ganglion express the transgene. Scale bars: 200 μm in A-C; 60 μm in D.

Fig. 4.

Cells in the nervous system preferentially express GAP-43 promoter transgenes. The tissue distribution of cells expressing each transgene was plotted for 45-63 embryos as described under Experimental Procedures. Both the 386 bp (light bars) and 1 kb GAP-43 (dark bars) promoter transgenes were preferentially expressed in the nervous system at 55 h. At the same stage of development, a Rous sarcoma virus (RSV) promoter (hatched bars) directed expression broadly in the embryo, but not in the nervous system. At earlier times (30 h), a broader spectrum of cells (neuronal as well as non-neuronal) expressed the GAP-43 promoter transgenes. At 55 h, the mean number of cells expressing the transgene from the 386 bp GAP-43 promoter fragment was 1.3±0.3 cells per embryo; from the 1 kb fragment 3.8±0.9 cells per embryo; and from the RSV promoter 2.7±0.5 cell per embryo. At 30h, the 386 bp fragment was activated in 4.4±0.6 cells per embryo and the 1 kb fragment in 3.8±0.5 cells per embryo.

Fig. 4.

Cells in the nervous system preferentially express GAP-43 promoter transgenes. The tissue distribution of cells expressing each transgene was plotted for 45-63 embryos as described under Experimental Procedures. Both the 386 bp (light bars) and 1 kb GAP-43 (dark bars) promoter transgenes were preferentially expressed in the nervous system at 55 h. At the same stage of development, a Rous sarcoma virus (RSV) promoter (hatched bars) directed expression broadly in the embryo, but not in the nervous system. At earlier times (30 h), a broader spectrum of cells (neuronal as well as non-neuronal) expressed the GAP-43 promoter transgenes. At 55 h, the mean number of cells expressing the transgene from the 386 bp GAP-43 promoter fragment was 1.3±0.3 cells per embryo; from the 1 kb fragment 3.8±0.9 cells per embryo; and from the RSV promoter 2.7±0.5 cell per embryo. At 30h, the 386 bp fragment was activated in 4.4±0.6 cells per embryo and the 1 kb fragment in 3.8±0.5 cells per embryo.

To test whether preferential expression in the nervous system reflects specific activation of GAP-43 promoter elements, we injected zebrafish embryos with a construct containing the β-galactosidase cDNA fused to the Rous sarcoma virus (RSV) promoter. Although the RSV promoter has some tendency for preferential expression in cells of mesodermal origin (e.g. muscle, bone and skin), it can direct expression in a broad range of cell types, including neural cells (Brinster et al., 1984; Overbeek et al., 1986; Swain et al., 1987; Stuart et al., 1990; Nedivi et al., 1992). When injected into zebrafish embryos, the RSV-promoter was activated predominantly in muscle and skin, but never in neurons (Fig. 4). A different viral promoter, from cytomegalovirus (CMV), and promoter fragments from the mammalian Hox 1.1 and Hox 3.3 genes are also activated in zebrafish embryos in patterns clearly distinct from that observed here for the GAP-43 promoter fragments (Westerfield et al., 1992 and unpublished observations). These data indicate that the pattern of transgene expression obtained with the GAP-43 fragments results from specific activation of the GAP-43 promoter in the nervous system.

It is not yet clear whether expression of the GAP-43 constructs in the CNS reflects expression in neurons only or in both neurons and glial cells. Although neurons predominantly express the endogenous GAP-43 gene, glial cells can also express it under some circumstances (Da Cunha and Vitkovic, 1990; Deloulme et al., 1990; Curtis et al., 1992). At least some of the cells expressing the transgenes in zebrafish embryos could be identified as neurons by histological detection of β-galactosidase activity that filled their axonal or dendritic processes (Fig. 3C,D). Because β-galactosidase is not readily transported into established axons and dendrites (Liu et al., 1991), these may represent cells that expressed the enzyme just prior to neurite extension, subsequently carrying it into their extending processes.

Temporal changes

Both the 386 bp and 1 kb promoter constructs displayed substantial neural selectivity at 55 h (Fig. 4; 55 h). At an earlier stage of development (Fig. 4; 30 h), transgenes were also expressed preferentially in the nervous system, but at this stage expression of both constructs was common outside the nervous system (Fig. 4; 30 h). At this time the 1 kb construct appeared to show somewhat greater neural selectivity than the 386 bp construct. In situ hybridization revealed no clear evidence of endogenous GAP-43 expression in non-neural cells at this stage. Within the nervous system, however, transgene expression was very well correlated with expression of the endogenous GAP-43 gene (Fig. 3D). In 25 embryos doubly labeled, 90% (45 out of 50) of the cells in the nervous system expressing β-galactosidase also expressed the zebrafish GAP-43 gene. In situ hybridization showed that expression of the endogenous GAP-43 gene is downregulated throughout the spinal cord by 55 h (Figs 3B, 5A,B), while expression remained elevated in the forebrain. Both the 386 bp and 1 kb rat promoter fragments exhibited a corresponding reduction in expression between 30 and 55 h in spinal cord, but not in forebrain (Fig. 3A,C,D).

Fig. 5.

Large spinal cord cells, including Rohon Beard sensory neurons, express both the transgene and the endogenous GAP-43 gene at 24 h (A), but not at 60 h (B). Double labeling for expression of the 386 bp β-galactosidase construct (blue reaction product) and the endogenous GAP-43 gene (purple/gray reaction product). The Rohon Beard neurons, which are double labeled in A, were identified on the basis of their cell body positions and size. Scale bar, 30 μm. C and D show the mean number of cells in the brain and spinal cord that express the 386 bp (light box) or 1 kb (dark box) transgene at 30 h and 55 h, respectively (386 bp transgene: 30 h, 60 embryos, 55 h, 48 embryos; 1 kb transgene: 30 h, 63 embryos, 55 h, 45 embryos). Error bars show standard errors. Both constructs were selectively downregulated at 55 h, reflecting the regulation of the endogenous GAP-43 gene.

Fig. 5.

Large spinal cord cells, including Rohon Beard sensory neurons, express both the transgene and the endogenous GAP-43 gene at 24 h (A), but not at 60 h (B). Double labeling for expression of the 386 bp β-galactosidase construct (blue reaction product) and the endogenous GAP-43 gene (purple/gray reaction product). The Rohon Beard neurons, which are double labeled in A, were identified on the basis of their cell body positions and size. Scale bar, 30 μm. C and D show the mean number of cells in the brain and spinal cord that express the 386 bp (light box) or 1 kb (dark box) transgene at 30 h and 55 h, respectively (386 bp transgene: 30 h, 60 embryos, 55 h, 48 embryos; 1 kb transgene: 30 h, 63 embryos, 55 h, 45 embryos). Error bars show standard errors. Both constructs were selectively downregulated at 55 h, reflecting the regulation of the endogenous GAP-43 gene.

Additional upstream elements enhance ectodermal expression

At 55 h, cells in the brain preferentially expressed the two smaller GAP-43 transgenes, containing up to 1 kb of 5 ’untranslated genomic DNA. Addition of another 4 kb of genomic GAP-43 DNA to the 1 kb fragment, however, extended expression to cells in the skin (Fig. 6). Targeted expression in these hexagonally shaped cells was statistically significant at the 99% confidence level and differed from the transient expression in non-neural cells earlier in development (Fig. 4; 30 h), because it depended on an element located at a considerable distance from the core promoter (within a 4 kb piece of genomic DNA). In contrast, the early non-neural expression was independent of any elements outside the 386 bp core promoter. Activation of the 386 bp and 1 kb fragments in skin at 55 h was statistically insignificant and therefore the same as expression of the marker gene in all other non-neural tissues. Expression of the 1+4 transgene in these non-neural cells seemed to persist longer than in the brain (not shown). The 4 kb fragment alone, fused to the gene for β-galactosidase, was unable to direct expression in any tissue, indicating that expression in epidermal cells is directed by an enhancer element and not by an additional promoter.

Fig. 6.

(A) Tissue distribution of cells expressing the 1 kb (light box) and 1+4 kb (dark box) GAP-43 construct at 55 h. 64 and 58 embryos, respectively were analyzed and the total number of transgene expressing cells per tissue was divided by the number of embryos examined. Error bars show the standard errors. Both transgenes are preferentially activated in brain. In addition, statistically significant expression of the 1+4 transgene is directed to cells in the skin. (B) Hexagonal-shaped cells in the zebrafish epidermis express the 1+4 transgene (blue reaction product) at 55 h. Black cells are pigment cells. Scale bar, 50 μm.

Fig. 6.

(A) Tissue distribution of cells expressing the 1 kb (light box) and 1+4 kb (dark box) GAP-43 construct at 55 h. 64 and 58 embryos, respectively were analyzed and the total number of transgene expressing cells per tissue was divided by the number of embryos examined. Error bars show the standard errors. Both transgenes are preferentially activated in brain. In addition, statistically significant expression of the 1+4 transgene is directed to cells in the skin. (B) Hexagonal-shaped cells in the zebrafish epidermis express the 1+4 transgene (blue reaction product) at 55 h. Black cells are pigment cells. Scale bar, 50 μm.

A variety of studies have shown that the neural specificity of GAP-43 expression is a conserved feature among vertebrates, at least for some developmental stages. Additionally, the specific downregulation of GAP-43 expression, when synapses are formed, occurs in most vertebrate neurons. By introducing mammalian GAP-43 promoter fragments into fish, we have now obtained direct evidence that neural induction and repression of the GAP-43 gene are mediated by functionally conserved mechanisms.

Regulation of GAP-43 within the nervous system

Using transient expression assays, we have demonstrated that promoter and enhancer elements in the rat GAP-43 gene initiate transcription of a marker gene in transgenic zebrafish, and are activated in neurons at stages when extensive neurite outgrowth takes place (Kimmel and Westerfied, 1990; Wilson et al., 1990). At 55 h, for example, a 386 bp fragment containing the GAP-43 core promoter and a larger fragment of 1 kb encompassing the smaller piece were preferentially expressed in the brain (Fig. 4). At this stage, expression of the mammalian transgenes correlates well with the distribution of endogenous GAP-43 mRNA revealed by in situ hybridization. At an earlier stage of development (30 h), the GAP-43 promoter fragments were activated in a broader range of tissues, but even at this stage the transgenes were expressed preferentially in the nervous system (Fig. 4), where transgene expression occurred almost exclusively in cells expressing the endogenous GAP-43 gene. Expression of the endogenous GAP-43 gene, in turn, is well correlated at both stages (Fig. 1) with the onset of neurite outgrowth (Chitnis and Kuwada, 1990; Kimmel and Westerfield, 1990; Wilson et al., 1990; Wilson and Easter, 1991). The pattern of transgene expression is specific for the GAP-43 promoter fragments and different from those of other viral and mammalian gene promoters (Fig. 4 and Westerfied et al., 1992). Moreover, addition or deletion of upstream elements changed the expression pattern of the GAP-43 promoter fragments (Fig. 6). Thus, it is unlikely that the predominant activation of GAP-43 promoter fragments in the nervous system is due to the distribution of the injected transgenes, the relatively large number of CNS cells, or some other bias of the transient expression technique. Rather, selective expression of the transgene in neural cells indicates that factors in zebrafish CNS cells specifically transactivate the mammalian GAP-43 promoter.

Previous analysis, using transient expression assays in dissociated cell cultures, showed that the 386 bp fragment directed expression preferentially in mammalian neuronal cells compared to several non-neuronal cell lines (Nedivi et al., 1992). Because this fragment lacks the neural-specific silencer element common to the type II sodium channel, SCG10, and synapsin I genes (Kraner et al., 1992; Mori et al., 1992; Mandel and McKinnon, 1993), the pathway(s) to which this region of the GAP-43 gene responds appear to represent a separate pathway controlling the selective expression of genes in developing vertebrate neurons (Nedivi et al., 1992). Our present results show that the same fragment also directs appropriate neural-selective expression in zebrafish embryos in vivo. This indicates that some transacting factors have retained the ability to recognize regulatory elements within this region of the GAP-43 gene over the 400 million years since fish and mammals diverged from a common ancestor, and that mechanisms coupling the expression or activities of these trans-acting factors to specific stages in neural differentiation have been functionally conserved over the same period.

At least one feature of GAP-43 gene regulation has undergone an apparent phylogenetic divergence between fish and mammals. In fish as in mammals, the majority of neurons sharply reduce expression of this gene as developing neurons complete their initial axon outgrowth and this repression ordinarily persists throughout adult life. However, lesion studies indicate that the mechanisms that maintain this repression in the CNS are regulated differently in fish and mammals. Interruption of axons far from their cell bodies reverses the repression of GAP-43 synthesis in the CNS of fish and amphibians, which show a high capacity for regeneration (Benowitz et al., 1981; Skene and Willard, 1981b; Grafstein, 1986; LaBate and Skene, 1989), but not in the mammalian CNS (Doster et al., 1991; Skene, 1992).

Transient expression assays do not allow us to address in detail the chronic repression of mammalian GAP-43 transgenes in mature animals or after lesions of specific axonal pathways. However by 2 days of development, in situ hybridization indicates that specific downregulation of the GAP-43 gene has occurred in zebrafish spinal cord (Figs 3, 5), allowing us to investigate the initial repression of the GAP-43 gene in these maturing neurons. At that time, expression of the mammalian GAP-43 promoter fusion genes also was specifically downregulated in the spinal cords of transgenic fish (Figs 3, 5). This indicates that, despite substantial divergence in the pathways controlling the repression of the GAP-43 gene in mature CNS neurons, this aspect of GAP-43 gene regulation is also subject to strong evolutionary constraints. At least some of the cis and trans-acting factors responsible for this late regulation have been sufficiently conserved that maturing CNS neurons in fish are able to regulate appropriately the mammalian transgenes.

Expression outside the nervous system

Expression of the mammalian GAP-43 transgenes in fish is neural selective at 55 h (Figs 3, 4A). At earlier times, the transgenes were also expressed preferentially in the nervous system and this expression was well correlated with the distribution of the endogenous gene (Figs 3D, 5A). However, at these early times, we also observed considerable expression of the transgenes outside the nervous system (Fig. 4) which was not obviously correlated with expression of the endogenous gene. It is possible that this extraneural regulation reflects a biologically significant induction of pathways that can activate GAP-43 regulatory elements. Recently GAP-43 expression has been seen in various glial cells (Da Cunha and Vitkovic 1990; Deloulme et al., 1990; Curtis et al., 1992) and in at least two populations of mesenchymal cells in the developing chick limb (Stocker et al. 1992a,b). In some regions of the limb, GAP-43 immunoreactivity co-localized with cells that were also positive for meromyosin, a muscle-specific marker, and in the interdigital mesenchyme that undergoes programmed cell death. Alternatively, activation of the mammalian transgenes outside the fish nervous system might represent illegitimate transcription.

The gradual restriction of both the transgene and endogenous GAP-43 expression suggest that expression may be actively repressed at a specific time in development in particular cell types. Progressive restriction of gene expression from an early distribution in multiple cell types, to a limited subset of neurons in adults, is a developmental pattern exhibited not only by the GAP-43 gene, but by several other ‘neural-specific’ genes (Reinhard and Skene, 1992). Neuronal glutamate receptors, for example, can be expressed not only in glial cells in the CNS, but also transiently in non-neural tissues of mouse embryos (Hermanns-Borgmeyer, unpublished observation). Similarly, virtually all tissues of the rat embryo express SCG 4, a gene cloned from an adult rat superior cervical ganglion library. From this initial widespread distribution, expression of SCG 4 mRNA declines in all tissues during late gestation but then returns to high levels postnatally, specifically in the nervous system (Anderson and Axel, 1985).

Developmental regulation of GAP-43 expression

From our analysis of endogenous GAP-43 gene expression, we have found that expression is very dynamic during the first few days of development. Throughout the CNS, GAP-43 is first expressed in individual cells that we can, in most cases, identify as primary neurons (Figs 1, 5A). Later, additional cells express the gene (Figs 1, 3). This increase in the number of expressing cells is most apparent in the forebrain and midbrain, less so in the hindbrain, and much less obvious in the spinal cord. Beginning after the second day, expression drops off rapidly in the spinal cord, while still increasing in more anterior regions (Figs 1, 3). At no single time did we see all cells in any CNS region expressing the gene, and even when expression is dropping off in the spinal cord, the first cells, for example Rohon Beard neurons, still express the gene. This suggests that some neurons may never express GAP-43, although we cannot rule out the possibility that some cells express very low levels of GAP-43 very transiently and hence were missed by our analysis.

Evidence for an additional upstream regulatory element

By 55 h, the initially broad expression of the 386 bp and 1 kb constructs had become restricted predominantly to the nervous system. At this time, however, the 1+4 construct, bearing an additional 4 kb of upstream DNA from the mammalian GAP-43 gene, directed strong expression in epidermis as well as neural tissue. This indicates that an additional regulatory element(s) located within the upstream 4 kb fragment can direct gene expression in a specific type of non-neural cells. This highly selective expression in epidermal cells is clearly distinct from the transient expression in multiple cell types observed during earlier embryogenesis, which occurred independently of the element(s) in the upstream 4 kb fragment. Because of the technical impediments to in situ hybridization in epidermal cells, it is not yet clear whether the endogenous GAP-43 gene is actually expressed in these cells or whether other factors can override this potential for expression in skin, keeping the endogenous gene silent. These results indicate that the cell-type specificity of GAP-43 expression may be regulated by cis-acting elements distinct from the neural-specific elements in or near the core promoter.

We are particularly grateful Dr. K. Hatta, Dr. T. Schilling and R. Warga for their help with zebrafish anatomy and Yi Lin and Dr. M. Fishman for providing the zebrafish GAP-43 cDNA. We thank Dr. A. Püschel for help with the in situ hybridization technique and critical discussions. This work was supported by NIH grants EY07397, HD22486, and NS21132. E. Reinhard is supported by the ‘Schweizerische Stiftung für medizinisch-biologische Stipendien.’ The sequence of the zebrafish GAP-43 cDNA is recorded in GenBank under the accession number L27645.

Anderson
,
D. J.
and
Axel
,
R.
(
1985
).
Molecular probes for the development and plasticity of neural crest derivatives
.
Cell
42
,
649
662
.
Basi
,
G. S.
,
Jacobson
,
R. D.
,
Virag
,
I.
,
Schilling
,
J.
and
Skene
,
J. H. P.
(
1987
).
Primary structure and transcriptional regulation of GAP-43, a protein associated with nerve growth
.
Cell
49
,
785
791
.
Bendotti
,
C.
,
Servadio
,
A.
and
Samanin
,
R.
(
1991
).
Distribution of GAP-43 mRNA in the brain stem of adult rats as evidenced by in situ hybridization: Localization within monoaminergic neurons
.
J. Neurosci
.
11
,
600
607
.
Benowitz
,
L. I.
,
Shashoua
,
V. E.
and
Yoon
,
M. G.
(
1981
).
Specific changes inrapidly transported proteins during regeneration of the goldfish optic nerve
.
J. Neurosci
.
1
,
300
307
.
Benowitz
,
L. I.
and
Lewis
,
E. R.
(
1983
).
Increased transport of 44,000- to 49,000 dalton acidic proteins during regeneration of the goldfish optic nerve: a two-dimensional gel analysis
.
J. Neurosci
.
3
,
2153
2163
.
Biffo
,
S.
,
Verhaagen
,
J.
,
Schrama
,
J.
,
Schotman
,
P.
,
Danho
,
W.
and
Margolis
,
F. L.
(
1990
).
B-50/GAP-43 expression correlates with process outgrowth in the embryonic mouse nervous system
.
Eur. J. Neurosci. 2, 487-499
.
Brinster
,
R. L.
,
Chen
,
H. Y.
,
Messing
,
A.
,
Van Dyke
,
T.
,
Levine
,
A. J.
and
Palmiter
,
R. D.
(
1984
).
Transgenic mice harboring SV-40 T antigen genes develop characteristic brain tumors
.
Cell
37
,
367
379
.
Chitinis
,
A. B.
and
Kuwada
,
Y.
(
1990
).
Axonogenesis in the brain of zebrafish embryos
.
J. Neurosci
.
10
,
1892
1905
.
Curtis
,
R.
,
Stewart
,
H. J. S.
,
Hall
,
S. M.
,
Wilkin
,
G. P.
,
Mirsky
,
R.
and
Jessen
,
K. R.
(
1992
).
GAP-43 is expressed by nonmyelin-forming schwann cells ofthe peripheral nervous system
.
J. Cell Biol
.
116
,
1455
1464
.
Da Cunha
,
A.
and
Vitkovic
,
L.
(
1990
).
Regulation of immunoreactive GAP-43 expression in rat cortical macroglia is cell type specific
.
J. Cell Biol
.
111
,
209
215
.
Deloulme
,
J.-C.
,
Janet
,
T.
,
Au
,
D.
,
Storm
,
D. R.
,
Sensenbrenner
,
M.
and
Baudier
,
J.
(
1990
).
Neuromodulin (GAP-43): a neural protein kinase C substrate is also present in O-2A glial cell lineage. Characterization of neuromodulin in secondary cultures of oligodendrocytes and comparison with the neuronal antigen
.
J. Cell Biol
.
111
,
1559
1570
.
Doster
,
S. K.
,
Lozano
,
A. M.
,
Aguayo
,
A. J.
and
Willard
,
M. B.
(
1991
).
Expression of the growth-associated protein GAP-43 in adult rat retinal ganglion cells following axon injury
.
Neuron
6
,
1
13
.
Grafstein
,
B.
(
1986
).
The retina as a regenerating organ
. In
The Retina, part II. pp. 275-335. New York: Academic Press, Inc
.
Grunwald
,
D. J.
,
Kimmel
,
C. B.
,
Westerfield
,
M.
,
Walker
,
C.
and
Streisinger
,
G.
(
1988
).
A neural degeneration mutation that spares primary neurons in the zebrafish
.
Dev. Biol
.
126
,
115
128
.
Hoffman
,
P. N.
(
1989
).
Expression of GAP-43, a rapidly transported growth-associated protein, and class II beta tubulin, a slowly transported cytoskeletal protein, are coordinated in regenerating neurons
.
J. Neurosci
.
9
,
893
897
.
Kimmel
,
C. B
and
Law
,
R. D.
(
1985
).
Cell lineage of zebrafish blastomeres. III. Clonal analyses of the blastula and gastrula stages
.
Dev. Biol
.
108
:
94
101
.
Kimmel
,
C. B.
and
Westerfield
,
M.
(
1990
).
Primary neurons of the zebrafish
. In
Signal and Sense
. (eds.
G. M.
Edelman
,
W. E.
Gall
and
W. M.
Cowan
) pp.
561
588
.
New York
:
Wiley-Liss
.
Kraner
,
S. D.
,
Chong
,
J. A.
,
Tsay
,
H.-J.
and
Mandel
,
G.
(
1992
).
Silencing of the type II sodium channel gene: a model for neural-specific gene regulation
.
Neuron
9
,
37
44
.
LaBate
,
M. E.
and
Skene
,
J. H. P.
(
1989
).
Selective conservation of GAP-43 structure in vertebrate evolution
.
Neuron
3
,
299
310
.
Liu
,
Y.
,
Chapman
,
E. R.
and
Storm
,
D. R.
(
1991
).
Targeting of neuromodulin (GAP-43) fusion proteins to growth cones in cultured rat embryonic neurons
.
Neuron
6
,
411
420
.
Mandel
,
G.
and
McKinnon
,
D.
(
1993
).
Molecular basis of neural-specific gene expression
.
Ann. Rev. Neurosci
.
16
,
323
345
.
Meberg
,
P. J.
and
Routtenberg
,
A.
(
1991
).
Selective expression of protein F1/(GAP-43) mRNA in pyramidal but not granule cells of the hippocampus
.
Neurosci
45
,
721
733
.
Mori
,
N. C.
,
Schoenherr
,
D. J.
,
Vandenbergh
,
D.J.
and
Anderson
,
D. J.
(
1992
).
A common silencer element in the SCG10 and type II Na+ channel genes binds a factor present in nonneuronal but not neuronal cels
.
Neuron
9
,
45
54
.
Nedivi
,
E.
,
Basi
,
G. S.
,
Virag-Akey
,
I.
and
Skene
,
J. H. P.
(
1992
).
A neural-specific GAP-43 core promoter located between unusual DNA elements that interact to regulate its activity
.
J. Neurosci
.
12
,
691
704
.
Overbeek
,
P. A.
,
Lai
,
S.
,
Van Quil
,
K. R.
and
Westphal
,
H.
(
1986
).
Tissuespecific expression in transgenic mice of a fused gene containing RSV terminal sequences
.
Science
231
,
1574
1577
.
Oxtoby
,
E.
and
Jowett
,
T.
(
1993
).
Cloning of the zebrafish krox-20 gene (krx-20) and its expression during hindbrain developmen
.
Nucleic Acids Res
.
21
,
1087
1095
.
Perry
,
G. W.
,
Burmeister
,
D. W.
and
Grafstein
,
B.
(
1987
).
Fast axonally transported proteins in regenerating goldfish optic axons
.
J. Neurosci
.
7
,
792806
.
Püschel
,
A. W.
,
Gruss
,
P.
and
Westerfield
,
M.
(
1992
).
Sequence and expression pattern of pax-6 are highly conserved between zebrafish and mice
.
Development
114
,
643
651
.
Redshaw
,
J. D.
and
Bisby
,
M. A.
(
1984
).
Fast axonal transport in central nervous system and peripheral nervous system axons following axotomy
.
J. Neurobiol
.
15
,
109
118
.
Reinhard
,
E.
and
Skene
,
J. H. P.
(
1992
).
Multiple elements may be used for regulation of the GAP-43 gene in different cell-types
.
Perspectives on Developmental Neurobiology
1
,
29
37
.
Skene
,
J. H. P.
and
Willard
,
M.
(
1981a
).
Axonally transported proteins associated with axon growth in rabbit central and peripheral nervous systems
.
J. Cell Biol
.
89
,
96
103
.
Skene
,
J. H. P.
and
Willard
,
M.
(
1981b
).
Changes in axonally transported proteins during axon regeneration in toad retinal ganglion cells
.
J. Cell Biol
.
89
,
86
95
.
Skene
,
J. H. P.
(
1989
).
Axonal growth-associated proteins
.
Ann. Rev. Neurosci
.
12
,
127
156
.
Skene
,
J. H. P.
(
1992
).
Retrograde pathways controlling expression of a major growth cone component in the adult CNS
. In
The Nerve Growth Cone
. (eds.
P. C.
Letourneau
,
S. B.
Kater
,
E. R.
Macagno
), pp.
463
475
.
New York
:
Raven Press, Ltd
.
Stocker
,
K. M.
,
Baizer
,
L.
and
Ciment
,
G.
(
1992a
).
Transient expression of GAP-43 in non-neuronal cells of the embryonic chicken limb
.
Dev. Biol
.
149
,
406
414
.
Stocker
,
K. M.
,
Ciment
,
G.
and
Baizer
,
L.
(
1992b
).
GAP-43 in non-neuronal cells of the embryonic chick limb: clues to function
.
Perspectives on Developmental Neurobiology
1
,
53
62
.
Stuart
,
G. W.
,
Mc Murry
,
J. V.
and
Westerfield
,
M.
(
1988
).
Replication, integration and stable germ-line transmission of foreign sequences injected into early zebrafish embryos
.
Development
103
,
403
412
.
Stuart
,
G. W.
,
Vielkind
,
J. R.
,
McMurray
,
J. V.
and
Westerfield
,
M.
(
1990
).
Stable lines of transgenic zebrafish exhibit reproducible patterns of transgene expression
.
Development
109
,
577
584
.
Swain
,
J. L.
,
Stewart
,
T. A.
and
Leder
,
P.
(
1987
).
Parental legacy determines methylation and expression of an autosomal transgene: A molecular mechanism for parental imprinting
.
Cell
50
,
719
727
.
Tetzlaff
,
W.
,
Alexander
,
S. W.
,
Miller
,
F. D.
and
Bisby
,
M. A.
(
1991
).
Response of facial and rubrospinal neurons to axotomy: Changes in mRNA expression for cytoskeletal protein and GAP-43
.
J. Neurosci
.
11
,
2528
2544
.
Trevarrow
,
B.
,
Marks
,
D. L.
and
Kimmel
,
C. B.
(
1990
).
Organization of hindbrain segments in the zebrafish embryo
.
Neuron
4
,
669
679
.
Westerfield
,
M.
,
Wegner
,
J.
,
Jegalian
,
B. G.
,
De Robertis
,
E. M.
and
Püschel
,
A. W.
(
1992
).
Specific activation of mammalian Hox promoters in mosaic transgenic zebrafish
.
Genes Dev
.
6
,
591
598
.
Westerfield
,
M.
(
1993
).
The Zebrafish Book, 2nd edition
.
University of Oregon Press
.
Wilson
,
S. W.
,
Ross
,
L. S.
,
Parrett
,
T.
and
Easter
,
S. S.
, Jr.
(
1990
).
The development of a simple scaffold of axon tracts in the brain of the embryonic zebrafish, Brachydanio rerio
.
Development
108
,
121
144
.
Wilson
,
S.W.
and
Easter
,
S.S.
Jr.
(
1991
).
Stereotyped pathway selection by growth cones of early epiphysial neurons in the embryonic zebrafish
.
Development
112
,
723
746
.