SEARCH

SEARCH BY CITATION

Keywords:

  • laminin;
  • zebrafish;
  • axon guidance;
  • retinotectal;
  • bashful;
  • Mauthner;
  • spinal cord;
  • motor axons;
  • reticulospinal;
  • branchiomotor

Abstract

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Laminin is known to provide a highly permissive substratum and in some cases directional information for axon outgrowth in vitro. However, there is still little known about laminin function in guiding axons in vivo. We investigated the axon guidance role of laminin-α1 in the developing zebrafish nervous system. Analysis of zebrafish bashful (bal)/laminin-a1 mutants revealed multiple functions for laminin-α1 in the outgrowth and guidance of central nervous system (CNS) axons. Most CNS axon pathways are defective in bal embryos. Some axon types, including retinal ganglion cell axons, early forebrain axons, and hindbrain reticulospinal axons, make specific pathfinding errors, suggesting laminin-α1 is required for directional decisions. Other axon tracts are defasciculated or not fully extended in bal embryos, suggesting a function for laminin-α1 in regulating adhesion or providing a permissive substratum for growth. In addition, some neurons have excessively branched axons in bal, indicating a potential role for laminin-α1 in branching. In contrast to CNS axons, most peripheral axons appear normal in bal mutants. Our results, thus, reveal important and diverse functions for laminin-α1 in guiding developing axons in vivo. Developmental Dynamics 235:213–224, 2006. © 2005 Wiley-Liss, Inc.


INTRODUCTION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

During development of the nervous system, neurons extend axons that navigate through a complex environment to innervate specific targets. Growing axons are guided along their correct pathways by environmental cues that can be permissive, attractive, or repulsive. Many of these cues are found in the basal lamina, a complex of extracellular matrix proteins involved in many developmental and morphogenetic processes (Serafini et al., 1994; Streuli et al., 1995; Miyagoe et al., 1997; Cote et al., 1999; Hopker et al., 1999; Deng and Ruohola-Baker, 2000; Patton, 2000). Laminins are important components of the basal lamina that serve both as structural scaffolding for morphogenesis as well as signaling molecules, exerting many of their effects by binding integrin receptors on cell surfaces (Powell and Kleinman, 1997). The laminins are heterotrimers composed of distinct combinations of α, β, and γ subunits. Invertebrates have one or two different laminins, consisting of one or two possible α subunits and one β and one γ subunit. Mammals have 15 known laminin chains: 5 α, 4 β, and 3 γ, as well as different splice variants of these genes, producing 15 known laminin heterotrimers (Miner and Yurchenco, 2004).

Laminin-1, composed of α1β1γ1 subunits, was the first laminin to be identified and has been shown to be a permissive substrate for neurite outgrowth for a variety of neural cell types (Powell and Kleinman, 1997). Several in vitro studies have demonstrated that laminin-1 acts not only as a simple permissive growth substrate but can also provide directional information to growing axons (Kuhn et al., 1995, 1998; Halfter, 1996; Patton et al., 1997; Adams et al., 2005). More recently, laminin-1 also has been shown to be an important modifier of other specific guidance cues. For example, laminin-1 causes Xenopus retinal axons to change their response to the guidance signal Netrin-1 from attractive to repulsive (Hopker et al., 1999). In addition, laminin-1 modifies the response of these axons to ephrin-A5, changing the response from repulsive to attractive (Weinl et al., 2003). These in vitro studies show that laminin-1 can have complex effects on axon growth and guidance.

The role of laminin-1 in axon guidance in vivo is still not well understood, although several studies have shown that laminins are important. A Caenorhabditis elegans laminin mutant displays axon guidance defects such as commissural axons stopping at the midline or longitudinal axons joining the contralateral tract (Forrester and Garriga, 1997; Huang et al., 2003). Drosophila laminin-α1,2 subunit mutants display defects in ocellar axon pathfinding (Garcia-Alonso et al., 1996). Similarly, antibody perturbation of laminin in grasshopper causes defects in sensory axon turning (Bonner and O'Connor, 2001). Much less is known about axon guidance roles for vertebrate laminins in vivo. In mice, laminin-1 is one of the earliest expressed laminins, and mutants in any of the laminin-1 subunits die before gastrulation (Smyth et al., 1999; Miner et al., 2004), preventing analysis of potential axon guidance functions. In zebrafish, three mutants identified in a screen for retinal axon guidance defects, bashful (bal), grumpy (gup), and sleepy (sly; Karlstrom et al., 1996), were subsequently shown to be mutants in laminin subunits α1, β1, and γ1, respectively (Parsons et al., 2002; Steven Pollard and Derek Stemple, personal communication).

We isolated an allele of the zebrafish bal/laminin-α1 mutant in a screen for axon guidance mutants, and here, we investigate axon guidance defects in this mutant. Zebrafish are an ideal model system to look at laminin's role in axon pathfinding in the vertebrate brain. The nervous system initially forms as a simple and well-characterized scaffolding of axon tracts (Kimmel et al., 1982; Mendelson, 1986; Metcalfe et al., 1986; Bernhardt et al., 1990; Chitnis and Kuwada, 1990; Wilson et al., 1990). We find that bal embryos exhibit axon outgrowth and pathfinding defects in most central nervous system (CNS) pathways. These defects include specific pathfinding errors by retinal ganglion cell (RGC) axons and many earlier developing axons in the brain, defasciculation of some axon tracts, failure of several axon types to advance along their pathway, and excessive branching of spinal motor neuron axons. Additionally, the cell bodies of some hindbrain branchiomotor neurons fail to migrate to their mature position. In contrast to CNS pathways, most peripheral axons, including branchiomotor axons and sensory ganglion axons, show normal pathfinding. Our results show that laminin-α1 has several important roles in axon outgrowth and guidance in vivo.

RESULTS

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

baluw1 Is a Strong Allele of bashful/laminin-α1

To identify genes important for axon guidance, we screened zebrafish mutants by labeling axons with an anti-acetylated α-tubulin antibody at 24–27 hours postfertilization (hpf). We isolated a mutant with a gross morphological phenotype similar to previously described bal/laminin-α1 alleles, including a shortened body axis, notochord defects, and irregular brain morphology (Odenthal et al., 1996; Schier et al., 1996; Stemple et al., 1996). We made complementation crosses between carriers of this mutation and carriers of bala69, an allele identified in a screen for lens defects (Link et al., 2001), and found that 25% of the offspring displayed the bal phenotype, including a shortened body axis, notochord, and brain defects. This noncomplementation suggests that this mutation is an allele of bal, which we named baluw1. To further confirm that our mutation was an allele of bal, we mapped the mutation. We performed a mapcross between the baluw1 carriers in the AB line and WIK fish. The resulting embryos were raised to maturity and incrossed to produce embryos used for mapping. bal has been mapped to linkage group (LG) 24, near 76.9 cM (Steven Pollard and Derek Stemple, personal communication). We used the z11862 primers to amplify a simple sequence-length polymorphism (SSLP) marker in the same location and compared the polymorphisms present in the AB and WIK founders with those from homozygous mutant embryos (mut) or their wild-type siblings (heterozygous for the mutation mixed with noncarriers, WT). The AB polymorphism, represented by the band at 177 bp, is the only one present in the homozygous mutants, whereas the mixed population of WT siblings showed both the AB band at 177 bp and the WIK one at 210 bp (Fig. 1A). Thus, our mutation is strongly linked to the same chromosomal region as bal/laminin-α1. baluw1 homozygous embryos show notochord defects characteristic of the strongest bal alleles (Stemple et al., 1996; Fig. 1B,C). The type of mutation causing the strong phenotype in these alleles has not been reported. At 24 hpf, the notochord in baluw1 embryos is twisted and the notochord cells display a rounded appearance, indicative of an impairment in differentiation. At 5 days postfertilization (dpf), baluw1 larvae have a curved tail and shortened body axis (Fig. 1D,E). This allele is completely recessive, as heterozygotes appear normal.

thumbnail image

Figure 1. Mapping and morphology of baluw1 embryos. A: Gel of polymerase chain reaction products showing a band at 177 bp from AB founder (AB), baluw1 mutant embryos (mut), and wild-type sibling embryos (WT). The band at 210 bp is only present in WIK founder (WIK) and WT lanes. B,C: Lateral views (anterior left) of 24 hpf WT (B) and baluw1 (C) trunks. N denotes notochord. D,E: 5 days postfertilization (dpf) WT (D) and baluw1 (E) embryos. Scale bars = 20 μm in A,B, 200 μm in C,D.

Download figure to PowerPoint

RGC Axons Show Specific Pathfinding Errors in baluw1

In zebrafish, RGC axons project to the contralateral optic tectum. Beginning approximately 32 hpf, RGC axons extend out of the eye and then cross the midline in the ventral diencephalon. Axons in the contralateral optic tract turn dorsally and posteriorly toward the contralateral tectum. bal was identified previously in a screen for retinotectal axon guidance mutants and was shown to have several defects, including aberrant ipsilateral and anterior projections by RGC axons (Karlstrom et al., 1996). We have analyzed the RGC axon pathfinding errors in more detail to determine at which developmental stages and specific locations along the retinotectal pathway they occur.

At 2 dpf, many RGC axons have extended across the midline to the contralateral diencephalon. We first examined RGC pathways at this stage by labeling all RGC axons with an antibody, ZN-5. All wild-type embryos (n = 28) had normal contralateral projections (Fig. 2A). In a small number of bal embryos (10%; n = 29), no RGC axons crossed the midline, but instead all projected along the ipsilateral tracts (Fig. 2B, arrows). The remaining bal embryos had RGC axon tracts that were indistinguishable from those of wild-type embryos. Because it is not possible to detect whether a small number of axons make ipsilateral errors when axons from both eyes are labeled with antibodies, we next used 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) injection to label RGC axon trajectories from only one eye at 2.5–5 dpf. At 2.5 dpf, wild-type RGC axons have extended across the midline and have reached the dorsal turn on the contralateral side (n = 14; Fig. 2C). In bal embryos, RGC axons made pathfinding errors in two locations. Some bal axons projected into the ipsilateral optic tract from the midline region (40%; n = 15; Fig. 2D, arrow). In addition, on the contralateral side just after the dorsal turn, the optic tract split with some axons misdirected anteriorly instead of posteriorly (27%; Fig. 2E, arrowhead). In the remaining embryos at this age (33%), the RGC axons extended normally. At 3.5–4 dpf, wild-type RGC axons have grown into the contralateral tectum (n = 15; Fig. 2F). In some bal embryos, axons projected to the ipsilateral tectum (26%; n = 19; not shown), and some had axons that projected anteriorly from the contralateral optic tract (68%). At this stage, the aberrant anterior axons have extended into the telencephalon (Fig. 2G, arrowhead). In addition, we also saw projections extending anteriorly to the ipsilateral telencephalon at this age in 26% of the embryos (not shown). At 5 dpf, wild-type RGC axons have arborized in their correct topographic locations within the tectum (n = 3; Fig. 2H). In bal, similar pathfinding defects are seen as in earlier ages. In all embryos (n = 8), there were some normal projections to the contralateral tectum, whereas 75% also projected axons to the ipsilateral tectum (Fig. 2J, arrow). In addition, 63% had axons that projected to the contralateral anterior telencephalon, and 38% projected axons to the ipsilateral anterior telencephalon (Fig. 2I J, arrowheads). In the 4 and 5 dpf embryos, we injected DiI into the nasal retina and DiA into the temporal retina to determine whether axons from a particular region of the retina were more likely to make errors. We found that axons from either retinal location could make anterior errors (Fig. 2G,I,J, arrowheads) or ipsilateral tectal errors (Fig. 2J, arrow). Of interest, the axons that projected either to the normal tectum or the ipsilateral tectum roughly innervated the correct topographic location (Fig. 2J).

thumbnail image

Figure 2. Retinal ganglion cell (RGC) axon defects in bal embryos. K–N are in situ hybridizations. All views are anterior left. A,B: Ventral views of ZN-5–labeled RGC axons in 2 days postfertilization (dpf) wild-type (WT; A) and bal (B) embryos showing ipsilateral projections in bal (arrows in B). C–E: Ventral views of 2.5 dpf WT (C) or bal (D,E) embryos injected with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) into eye on bottom. The arrow in D indicates an ipsilateral projection, and the arrowhead in E shows the split optic tract. The dashed line indicates the midline. F,G: Lateral views, contralateral to injected eye, of 4 dpf WT (F) and bal (G) embryos injected with DiI (red) in nasal and DiA (green) in temporal retina. RGC axons extend anteriorly in bal (arrowhead). Injection sites of DiI and 4-4-dihexadecylaminostyryl-N-methylpyridinium iodide (DiA) overlap in G. H–J: Dorsal views of 5 dpf WT (H) and bal (I,J) embryos injected with DiI nasal and DiA temporal. Arrowheads indicate anterior projections, and the arrow denotes ipsilateral tectal projections. K,L: Lateral views of 48 hpf WT (K) and bal (L) embryos showing ephrinA5a in the retina (arrow). M,N: Ventral views of 48 hpf WT (M) and bal (N) embryos showing foxa2 along optic pathway (arrowheads) and in midbrain (arrows). E denotes eye. Scale bars = 60 μm.

Download figure to PowerPoint

To address whether the RGC axon errors are an indirect result of abnormal tissue patterning or altered expression of other guidance cues, we examined the expression patterns of several patterning and axon guidance molecules expressed in the RGCs or along their axon pathway. We first looked at expression of several genes implicated in RGC axon guidance: netrin1a, sema3D, ephrinA5a, ephrinA5b, ephA4, slit1a, slit1B, slit2, slit3, robo1, robo2, robo3, and robo4 by performing in situ hybridization in wild-type and bal embryos. The general expression patterns of these genes appeared normal in RGC cells or in regions surrounding the RGC axon pathway. For example, the expression of ephrinA5a in the retina (arrows) is shown in Figure 2K,L. In addition, we performed in situ hybridizations to examine the expression patterns of several genes implicated in tissue patterning: foxa2, wnt1, pax2, pax6, shh, zic1, and zic2. The expression patterns of these genes were generally normal, although some were slightly disrupted, in a manner suggesting an expansion of midline tissue. For example, the transcription factor foxa2 (arrowheads) is shown in Figure 2M,N. This expanded midline pattern in the midbrain (Fig. 2M,N, arrows) is similar to the expansion of shh expression shown previously in bal (Schier et al., 1996). Thus, although there may be some disruption of tissue that could indirectly affect axons, the basic expression patterns of many axon guidance molecules are not altered in bal.

Anterior Commissure and Postoptic Commissure Fail to Cross the Midline in bal

We also investigated the early forming axon pathways throughout the nervous system. Two of these early tracts cross the midline in the forebrain: the anterior commissure (AC) that connects telencephalic neurons and the more ventral postoptic commissure (POC) in the diencephalon. The POC is positioned just ventral to the RGC axons. In bal, these commissures are not properly formed (Fig. 3A,B). At 24 hpf, 68% (n = 25) of wild-type embryos had AC axons crossing the midline and 100% had POC axons crossing the midline. In contrast, only 20% (n = 25) of bal embryos had AC axons crossing the midline and 28% had POC axons crossing the midline. Axons that failed to cross often strayed dorsally and ventrally from the presumptive tract or appeared to stop on either side of the midline. By 27 hpf, all wild-type embryos had well-formed fascicles with many axons crossing both the AC and POC (n = 18; Fig. 3A). In bal, there were still few embryos with axons crossing the AC (18%; n = 17), but slightly more embryos had axons crossing the midline in the POC (41%; Fig. 3B). These defects are not the result of an age delay, as they persist in embryos as late as 40 hpf (33% of bal embryos had no AC or POC crossing midline; n = 15; data not shown). The midline crossing errors of the AC and POC axons are similar to those of the RGC axons, suggesting the mechanisms underlying these defects may be related.

thumbnail image

Figure 3. Axon tracts of the forebrain show pathfinding defects in bal. A,B: Ventral views (anterior left) of 27 hours postfertilization (hpf) wild-type (WT; A) and bal (B) embryos labeled with anti-α-tubulin showing anterior commissure (AC) and postoptic commissure (POC) axons do not cross the midline in bal (asterisk). C–F: Ventral views (C,D) and lateral views (E,F) of 24 hpf WT (C,E) and bal (D,F) embryos labeled with anti–α-tubulin showing the tract of the postoptic commissure (TPOC) axon errors in bal (arrows in D,F). Anterior left in all. G,H: Lateral views of 24 hpf WT (G) and bal (H) embryos injected with 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) into the epiphysis. In bal embryos, epiphysial axons are misguided (arrow in H). Dashed line indicates the anterior edge of embryo. I,J: Acridine Orange labeling of 24 hpf WT (I) and bal (J) embryos showing increased levels of apoptosis in bal (arrows). Scale bars = 60 μm.

Download figure to PowerPoint

Neurons of the Bilateral Nuclei of the Tract of the Postoptic Commissure and Epiphysial Axons Make Pathfinding Errors in bal

Two other diencephalic axon tracts are also misrouted in bal. Neurons of the bilateral nuclei of the tract of the postoptic commissure (nucTPOC) reside in the ventral diencephalon and, in addition to sending axons through the POC, extend axons posteriorly at 24 hpf to form the TPOC (Fig. 3C,E). In 100% of bal embryos (n = 24), the TPOC axons spread out broadly, extend in multiple directions, and do not form an organized tract (Fig. 3D,F, arrows).

Epiphysial neurons in the dorsal diencephalon project axons ventrally in 1 or 2 fascicles to join the TPOC and then anteriorly to join the POC (Wilson and Easter, 1991; Fig. 3G). These axons run parallel to those of the nearby posterior commissure (PC) but remain separate from it. We examined epiphysial axon pathways with either antibody labeling or DiI injections into the epiphysis. The bal epiphysial axons initially extended in the correct ventral direction but then were misguided. At 27 hpf, 21% of embryos (n = 24) showed some epiphysial axons joining the PC but still projecting ventrally to the TPOC (not shown). In 13%, epiphysial axons did not reach the TPOC but instead grew anteriorly toward the telencephalon (Fig. 3H, arrow). In all the embryos examined, the axons of the PC appeared normal (not shown).

Because loss of adhesion to the extracellular matrix has been shown to cause a variety of cell types to undergo apoptosis (Grossmann, 2002), we labeled embryos with Acridine Orange (5 μg/ml in E3; 3,6-bis[dimethylamino]acridine hemi-zinc chloride) to determine whether there were any areas of increased apoptosis around forming axon tracts at 24 hpf. bal embryos showed increased numbers of apoptotic cells in the forebrain, which were localized near the paths of the epiphysial axons and the TPOC (Fig. 3I,J, arrows). In wild-type embryos, there was an average of 75.8 apoptotic cells in the forebrain (n = 5), whereas bal embryos had on average of 238 apoptotic cells (n = 4). This finding suggests the axon guidance defects seen here could be caused by alterations of this underlying tissue.

Nuclei of the Medial Longitudinal Fasciculus Neurons Are Disorganized and Fail to Extend Their Axons Fully

The bilateral nuclei of the medial longitudinal fasciculus (nucMLF) reside in the ventral midbrain and extend axons posteriorly into the spinal cord, pioneering the MLF (Fig. 4A). In 100% of bal embryos (n = 38), the cell bodies of the nucMLF were disorganized and more laterally positioned in the midbrain (Fig. 4B, arrow). This disorganization was evident at the earliest stages of nucMLF differentiation and axonogenesis, approximately 17 hpf. The nucMLF axons in 100% of bal embryos (n = 38) initially projected their axons in multiple directions before eventually converging along their posteriorly directed pathway (Fig. 4B,D, arrowheads). Of interest, we often saw multiple processes projecting from nucMLF cell bodies in bal as compared with the normal unipolar morphology, suggesting an effect of laminin-α1 on initial cell polarity (Fig. 4C,D, arrow and inset). Although most of the nucMLF axons eventually projected posteriorly, they were defasciculated and made little forward advance in bal embryos (Fig. 4B, asterisk). At 24 hpf, the leading MLF growth cone in wild-type embryos had reached, on average, the level of trunk somite 12 (n = 26), whereas bal MLF growth cones had not exited the hindbrain (100%; n = 21). Later, at 27 hpf, MLF axons in wild-type embryos had extended to somite 20, on average (n = 18). However, the MLF axons in bal embryos have not yet made any further posterior progress (n = 17; not shown). Thus, although nucMLF axons eventually find their correct pathway in the absence of laminin-α1, they do not advance properly through the hindbrain, suggesting that laminin-α1 may be serving as a permissive substrate to promote growth of these axons along their pathway. However, it remains possible that the effects on MLF axons are caused indirectly by other changes in the surrounding cells.

thumbnail image

Figure 4. bal embryos have defects in midbrain and hindbrain neurons. All are ventral views with anterior left. A–D: Nuclei of the medial longitudinal fasciculus (nucMLF) neurons labeled with ZN-12 in wild-type (WT; A,C) and bal (B,D) embryos at 24 hours postfertilization (hpf). C,D: Higher magnifications images of A,B. B: Cell bodies are laterally misplaced and disorganized in bal (arrow). B,D: Axons initially project in multiple directions (arrowheads in B,D) and then form a loose fascicle (asterisk in B). D: Arrow and inset shows cell body with multiple processes. E,F: Hindbrain neurons labeled with anti–α-tubulin in 24 hpf WT (E) and bal (F) embryos. Axons are defasciculated (asterisk), are found at the midline (arrows), or extend longitudinally along the midline (arrowhead). The open arrowhead shows the posterior most extent of axons in MLF. G–I: Mauthner neurons labeled with 3A10 in 27 hpf WT (G) and bal (H,I) embryos showing extra projection from soma (asterisks in I), anteriorly extending axons (arrow in I), branched axons (arrow in I), straight contralateral axon projection (arrow in H), posterior extent of axons (arrowheads in H,I). Scale bars = 40 μm for all full panels, 20 μm for inset in D.

Download figure to PowerPoint

Hindbrain Reticulospinal Axons Show Defects in bal

Shortly after the nucMLF axons extend, the reticulospinal neurons in each hindbrain rhombomere project axons posteriorly to the spinal cord, fasciculating along the MLF. Some reticulospinal neurons project along the ipsilateral MLF and others cross the midline and join the contralateral MLF (Kimmel et al., 1982; Mendelson, 1986; Metcalfe et al., 1986; Fig. 4E). In bal embryos, the reticulospinal axons had highly undulating trajectories and many of the contralateral projecting axons did not cross the midline but instead appeared to stop there (Fig. 4F, arrows) or extended longitudinally along the midline (Fig. 4F, arrowhead). Of interest, in bal embryos in which the nucMLF axons did not extend far into the hindbrain, most reticulospinal axons still extended posteriorly for a short distance in the general correct position of the MLF, although they also appeared to stop after a short distance. At 24 hpf, these axons, on average, had reached the level of somite 3 (n = 21; Fig. 4F, open arrowhead) and did not progress further by 27 hpf (n = 17 embryos). Furthermore, these axons were highly defasciculated (Fig. 4F, asterisk). To visualize the individual trajectories of the Mauthner neurons, the contralaterally projecting reticulospinal neurons of rhombomere 4, we labeled these cells with a specific antibody, 3A10. We saw several Mauthner axon pathfinding defects in bal embryos. At 27 hpf, 97% of bal embryos (n = 32) had Mauthner axons that did not extend beyond the hindbrain, whereas axons in wild-type embryos had extended well into the spinal cord at this stage (Fig. 4H,I, arrowheads). In addition, 53% of embryos had at least one axon that traveled anteriorly for a short distance before finding its correct path (Fig. 4I, arrow), 25% had axons that branched (Fig. 4I, arrow), 13% had axons that extended toward the midline but then turned back to join the ipsilateral MLF (not shown), and 38% had axons that extended straight across the midline to the contralateral cell body instead of taking the normal curved trajectory to the contralateral MLF (Fig. 4H, arrow). Of interest, 25% of the bal embryos had an extra short process emerging from the cell body (Fig. 4I, asterisks), suggesting an effect on initial cell polarity, as we saw in the nucMLF.

Spinal Cord Axons Exhibit Defects in bal

The early axons in the spinal cord form a well-characterized set of longitudinal and commissural tracts (Bernhardt et al., 1990; Kuwada et al., 1990a, b; Bernhardt et al., 1992; Fig. 5A). There are two main longitudinal tracts: the dorsal longitudinal fasciculus (DLF), and the ventral MLF, which is continuous with and pioneered by the axons of the midbrain nucMLF. In bal embryos at 24 hpf, there was no MLF in the spinal cord, consistent with our finding that the nucMLF axons did not extend beyond the hindbrain (Fig. 5B, asterisk). In addition, the DLF was defasciculated, and the axons showed undulating trajectories in bal (Fig. 5B, arrow), similar to the axon tracts in the hindbrain. The primary commissural ascending neurons (CoPAs) are the earliest neurons in the spinal cord to cross the midline. In bal embryos, these axons have trajectories that are somewhat wandering and irregular (Fig. 5A,B, arrowhead), although many appear to cross the midline normally (Fig. 5C,D, arrows).

thumbnail image

Figure 5. Spinal and branchial axon phenotypes in bal. All views are anterior left except E,F (dorsal up). Labeling is anti–α-tubulin (A–D), ZNP-1 (E–H), F59 (I,J), or Tg(isl1:GFP; K–P). A–D, K–P are confocal projections. A–H: Embryos are wild-type (WT; A,C,E,G) or bal (B,D,F,H). A–D: Lateral (A,B) or dorsal (C,D) views of spinal cord in 23 hours postfertilization (hpf) embryos. dorsal longitudinal fasciculus (DLF; arrow in B) and primary commissural ascending neuron (CoPA) axons (arrowhead in B) are disorganized and the medial longitudinal fasciculus (MLF) is absent (asterisk in B) in bal (B). CoPA axons (arrows in C,D) correctly cross the midline. E–H: Cross-sections (E,F) or lateral views (G,H) of 36 hpf embryos showing extra axon running parallel to caudal primary axon (CaP; arrowhead in F), absence of normal rostral primary axon (RoP; open arrowhead in F) and axon branching (arrows in H). The asterisk in E indicates out of focus CaP in the next segment. I,J: Dorsal views of 22 hpf embryos showing normal slow muscle differentiation. K,L: Ventral views of hindbrain in 3 days postfertilization (dpf) embryos showing that some nVII branchiomotor neurons fail to migrate. In L, more nVII cells are in r4, rather than r6. M–P: Ventral (M,N) and lateral views (O,P) of branchiomotor axons innervating pharyngeal arches (arrows) in 4 dpf embryos. Scale bars = 20 μm in A–D, 40 μm in E–P.

Download figure to PowerPoint

The spinal motor neurons lie in the ventral spinal cord and send their axons out of the CNS to innervate the trunk muscle somites. Three primary motor neurons in each segment project axons ventrally along a common pathway to the horizontal myoseptum, where they split into three separate pathways. The rostral primary axon (RoP) extends laterally to innervate somites in the horizontal myoseptum region, the middle primary axon (MiP) extends dorsally along the medial edge of the somite to innervate the dorsal somite, and the caudal primary axon (CaP) extends ventrally along the medial edge of the somite to innervate the ventral somite (Eisen et al., 1986) (Fig. 5E,G). We examined spinal motor axon pathways in trunk cross-sections of approximately 200 μm. In all wild-type sections (n = 81, 11 embryos), the spinal motor neurons showed correct pathfinding. In bal embryos, the spinal motor neurons showed multiple infrequent axon guidance errors (80 sections, 11 embryos). In 19% of the sections, no RoP axon could be detected throughout the thickness of a somite (Fig. 5F, open arrowhead). In 5% of the sections, an extra axon appeared to extend parallel to the normal CaP axon, suggesting the RoP axon could be misrouted (Fig. 5F, arrowhead). In 24% of the sections, no MiP axon could be detected throughout the thickness of a somite. Furthermore, all the motor axons branch more extensively in bal than motor axons in age-matched wild-type embryos at all time points examined (24–48 hpf) (Fig. 5H, arrows). Thus, although laminin-α1 plays a minor role in motor axon pathfinding, it appears to be important for regulation of axon branching. We further examined the patterning and differentiation of the somite tissue and notochord to determine whether disruptions in patterning could be causing the motor axon phenotypes. The expression of two notochord genes, shh and robo4, appeared completely normal at 24 hpf, indicating the bal notochords express some mature markers, despite their undifferentiated morphology. We also performed in situ hybridizations for three genes that appear to be expressed differently along the anterior–posterior axis of the somites: gli1, gli2, and myoD. Despite the narrower, U-shaped morphology of the somites, the expression patterns of all three of these genes were normal. Additionally, the antibody F59, which labels the myosin heavy chain in slow muscle cells shows that, in both the WT and bal embryos, the slow muscle cells have differentiated and migrated to their lateral positions correctly (Fig. 5I,J, arrows). Some bal muscle fibers appear to have a slight defect in orientation, but the overall morphology is not grossly disrupted.

Branchiomotor Neuron Migration but Not Axon Outgrowth Is Affected in bal

The branchiomotor neurons are located in a segmental pattern in the hindbrain and extend axons out of the CNS to innervate pharyngeal arch-derived muscles (Chandrasekhar, 2004). Most of the branchiomotor neurons are born in their final segmental position, with an exception being the cranial nerve VII (nVII) neurons, which are born in rhombomere 4 (r4) and then migrate to r6–7 during development (Bingham et al., 2002; Chandrasekhar, 2004). We examined branchiomotor neuron development by crossing baluw1 into Tg(Isl1:gfp) fish, in which branchiomotor neurons express green fluorescent protein (GFP; Higashijima et al., 2000). By 3 dpf, wild-type nVII neurons have completed migration and are spread along r6 and r7 (Fig. 5K). In bal embryos, some of the nVII neurons have failed to migrate and are still located in r4, with the others spread out in r5 to r7 (Fig. 5L). This is a less severe phenotype than other identified zebrafish mutants, such as trilobite/strabismus, and landlocked/scribble1, in which all of the nVII neurons fail to migrate (Jessen et al., 2002; Wada et al., 2005). Despite this aberrant cell body position, the branchiomotor axons in these mutants are still able to correctly innervate their targets. Similarly, the bal embryos all appear to have correct axon pathfinding of the branchiomotor neurons to the pharyngeal arches, despite their aberrant cell body positions (Fig. 5M–P, arrows). Thus, we have identified another gene important for migration but not axon guidance of branchiomotor neurons.

Other Peripheral Axon Pathways Are Normal in bal

Although most of the axon tracts in the CNS are abnormal in bal, most peripheral axon tracts appear normal, similar to the branchiomotor axons. We examined several peripheral axons, including those of the posterior lateral line ganglion (PLLg). The PLLg sensory ganglion lies just posterior to the otocyst and extends axons posteriorly along the horizontal myoseptum in association with the migrating lateral line primordium. The somite level to which the primordium has migrated is an accurate indicator of embryonic age (Kimmel et al., 1995). At 24 hpf, the PLLg axons have extended, on average, to somite 6 (n = 21) in wild-type embryos and also to somite 6 (n = 16) in bal embryos (Fig. 6A,B, arrows). Although the PLLg axons have extended to the same somite level, the absolute distance was shorter in bal (an average of 249 μm vs. 326 μm in wild-type). Similarly, at 27 hpf, the PLLg axons in wild-type embryos have extended to somite 10 (n = 36) and the bal axons have reached somite 9 (n = 29), representing an average distance of 267 μm for wild-type and 170 μm for bal. This absolute difference reflects the shortened body axis and narrower somites characteristic of bal embryos. Thus, although the PLLg axons extended more slowly in bal, they extended along the correct pathway to the appropriate somite level for their age.

thumbnail image

Figure 6. Other peripheral axons and migrating neural crest cells are normal in bal. A–F: Neurons were labeled with anti–α-tubulin (A,B), or ZN-12 (C–F). All are lateral views with anterior to the left. A,B: The 24 hours postfertilization (hpf) wild-type (WT; A) and bal (B) trunks showing posterior lateral line ganglion (PLLg) axons at same somite level (arrows). C,D: The 24 hpf WT (C) and bal (D) heads showing trigeminal axons extending from the trigeminal nucleus (asterisk) along the surface of the head (arrows). E,F: The 24 hpf WT (E) and bal (F) trunks showing Rohon–Beard axons extending along the surface of the trunk (arrows). G,H: In situ hybridization with crestin in 25 hpf WT (G) and bal (H) trunks showing ventral streams of neural crest cells (arrows). N denotes notochord. Scale bars = 40 μm.

Download figure to PowerPoint

In addition to the PLLg, other peripheral sensory neurons appeared normal in bal. The peripheral axons of the trigeminal ganglion (Fig. 6C,D, arrows) and the spinal Rohon–Beard sensory neurons (Fig. 6E,F, arrows) extended normally and arborized fully in the epidermis of the head and trunk. In addition, there were no detectable differences in axonal projections of the other cranial sensory ganglia.

Finally, we also examined neural crest cell migration in bal embryos. Neural crest cells originate in the neural tube and migrate through peripheral tissue to specific locations throughout the embryo. The directed migration of these cells has in many cases been shown to be mediated by the same cues that guide axons, such as laminins and ephrins, for example (Perris and Perissinotto, 2000; Halloran and Berndt, 2003). We examined the position of migrating neural crest cells by labeling with the neural crest marker crestin (Luo et al., 2001). At all time points examined (18–36 hpf), the overall number and distribution of migrating neural crest cells in bal embryos appeared normal (Fig. 6G,H, arrows), suggesting laminin-α1 is not required for their migration.

DISCUSSION

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

In this study, we have investigated the function of laminin-α1 in vivo in the developing vertebrate brain. We show that zebrafish bal/laminin-α1 mutants exhibit multiple defects in axon outgrowth and pathfinding. These defects include deviation of axons from their correct pathway, failure of axons to extend fully, defasciculation of axon tracts, and excessive branching in target regions. Many CNS axons in the brain and spinal cord are affected, and defects are seen in longitudinal tracts, dorsal–ventral tracts, and midline crossing pathways. However, not all axons require laminin-α1 for correct growth. Some CNS axons and most peripheral axons are normal in bal.

Laminin traditionally has been thought to act as a general growth permissive component of the basal lamina, rather than as an instructive guidance cue (Powell and Kleinman, 1997). Our results show that, for many developing axons in the zebrafish brain, laminin-α1 has an instructive function. For example, the RGC axons make very specific guidance errors, leaving their pathway at particular locations and projecting either to the ipsilateral tectum or the anterior forebrain. These axons grow to the same extent as wild-type axons, suggesting that laminin-α1 is not required as a permissive substrate for RGC axons. In addition, RGC axon outgrowth is not completely unregulated in bal. In comparison, RGC axons in the astray mutant, which lacks one guidance receptor, Robo2, misproject to many more locations in the brain (Karlstrom et al., 1996; Fricke et al., 2001). The more restricted growth in bal suggests laminin-α1 is required for axon pathway choice at a few specific decision points. One possibility is that laminin-α1 regulates the RGC growth cone response to another guidance cue at these locations. In vitro, laminin has been shown to reverse the response of RGC growth cones to the guidance signals Netrin-1 and ephrin-A5 (Hopker et al., 1999; Weinl et al., 2003). In addition, we show that many axon guidance cues are expressed normally in bal, supporting the idea that laminin-α1 has a direct effect and/or may act specifically to modify growth cone responses to guidance cues. It remains possible that previously defined defects in the eye and optic stalk could be altering the optic pathway in an indirect manner (Karlstrom et al., 1996). For example, although many of the RGC axons do leave the eye and correctly extend along the optic pathway, at least initially, it is possible that changes in the eye or the optic stalk may affect the ability of the RGC growth cone to correctly read and respond to later guidance cues.

The AC, POC, and the hindbrain reticulospinal axons also show particular pathfinding defects in bal. AC and POC axons fail to cross the midline. This failure is likely a guidance defect rather than a lack of outgrowth ability because the telencephalic neurons that give rise to the AC and the nucTPOC axons that project through the POC are both able to extend axons in other directions in bal. Labeling of individual Mauthner axons clearly showed that these hindbrain axons are misrouted in bal. In addition to being misrouted, Mauthner axons also appear to stop prematurely or fail to extend fully in bal, suggesting that laminin-α1 may act as both an instructive and permissive cue for these axons. It is also possible that this defect is an indirect effect of abnormal notochord differentiation. Although we have seen proper expression of shh and robo4 in the bal notochord, there may be other, untested cues that could affect the differentiation and patterning of surrounding tissues. Of interest, our results also show that hindbrain reticulospinal axons that normally fasciculate along nucMLF axons do not require these axons to recognize and grow along the correct general location of the MLF pathway, at least for a short distance.

The epiphysial axons and nucTPOC axons both stray from their pathways in the same region of the diencephalon in bal embryos. These axons project aberrantly in many directions, suggesting there may be a loss of all guidance information in this region. Of interest, bal embryos also show an increase in apoptosis in this region. In fact, a variety of cell types have been shown to undergo apoptosis after loss of adhesion to the extracellular matrix (Grossmann, 2002). It is possible that the effects on epiphysial and TPOC axon guidance in bal may be caused indirectly by loss of some cells in the surrounding tissue that normally produce important guidance signals for these axons.

For some developing axons, laminin-α1 may act simply as a permissive growth promoting substrate. The nucMLF axons, for example, initially extend in multiple directions in bal but find their correct pathway and extend for a short distance along it. These axons then appear to stop prematurely, and although they are defasciculated, they do not make further directional errors, suggesting laminin-α1 is not required for them to locate their correct pathway but is required for them to advance. It is interesting that laminin-α1, through some direct or indirect mechanism, does appear to regulate the organization of the nucMLF cell bodies and the initial direction the axons emerge from the cell body.

The spinal motor neuron axons in bal embryos exhibit excessive branching. Laminin-1 is expressed in the somites (Parsons et al., 2002) and could potentially play a role in inhibiting ectopic branching of the motor axons. However, it is also possible that this defect is an indirect effect of abnormal notochord differentiation. The notochord is important for differentiation of the somites (Halpern et al., 1995; Currie and Ingham, 1996; Weinberg et al., 1996), and signals from both the somites and the notochord affect spinal motor axon pathfinding (Eisen et al., 1986; Myers et al., 1986; Beattie and Eisen, 1997). We have showed that several markers of somite patterning, including gli1, gli2, myoD, and slow muscle myosin appear normal in bal, and slow muscle cells appear to have migrated normally, suggesting the effect on spinal motor axons may not be due to defective somite patterning. Finally, the neural crest cells appear to migrate along the somites correctly, indicating that the molecules necessary for their guidance are normal.

Some of the CNS axons and most of the peripheral axons examined appear normal in bal. This finding may reflect that neither these axons nor the surrounding tissue requires laminin-α1 for correct development. Recent work has shown that laminin-α1 is largely absent from peripheral tissues in the mouse, except for the lung and kidney (Pierce et al., 1998, 2000). More specifically, laminin-α1 has not been found in the axon pathways or targets of peripheral sensory axons in mice during development (Lentz et al., 1997), or in the sciatic nerve of rats or humans postnatally or during regeneration (Wallquist et al., 2002). Recent work has shown that notochords in bal embryos are not as severely affected as those in grumpy (laminin-β1) and sleepy (laminin-γ1) due to a redundancy in α subunits. It is only by disrupting laminin-α4 as well as laminin-α1 that the severe notochord phenotype is copied (Stemple, 2005). It is possible that other laminins, such as laminin-8 (α4β1γ1) may be influencing the guidance of some of the normally projecting neurons in bal embryos. The presence of other laminins may also be the reason why the bal embryos survive much later in development than similar mouse mutants.

The laminin-α1 subunit is also a component of laminin-3 (α1β2γ3). It is unknown when and where this laminin is expressed in zebrafish, but previous investigations into the function of laminin-3 in mice have not implicated it in axon guidance (Miner and Yurchenco, 2004). Therefore, laminin-1 is the most likely candidate to be mediating the effects we have seen in bal mutants.

In conclusion, our results provide insight into how laminin-α1 functions in vivo to guide growing axons. We show that laminin-α1 is required for the guidance of most CNS axons, although it appears to serve different functions for different axon populations. Further investigation will help reveal the mechanisms by which laminin-α1 mediates its effects on each of these different neuronal types.

EXPERIMENTAL PROCEDURES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

Animals

Zebrafish (Danio rerio) were maintained in a laboratory breeding colony on a 14-/10-hr light/dark cycle. Embryos were maintained at 28.5°C and staged as described previously (Kimmel et al., 1995). Embryo stage was defined as hours postfertilization or days postfertilization. baluw1 homozygotes were identified by morphology when 18 hpf or older. Controls designated as wild-type were either wild-type strain AB or heterozygous bal+/− siblings. We received bala69 fish from Brian Link. baluw1 heterozygotes were crossed with Tg(isl1:GFP) fish (Higashijima et al., 2000) to visualize the branchiomotor axons.

Mapping the Mutant Locus

An F1 mapcross was constructed by crossing baluw1 heterozygous AB fish with WIK fish. DNA was then extracted from the founders of this mapcross and from F2 embryos derived from incrossing F1 mapcross carriers (Johnson et al., 1994). Embryo DNA was pooled from 20 homozygous mutant fish (mut) or from 20 wild-type siblings (WT) at 24 hpf. Polymerase chain reaction (PCR) analysis with SSLP markers (Shimoda et al., 1999) was carried out in a manner similar to that previously described (Liao and Zon, 1999). Briefly, 20-μl PCR reactions were carried out with 100-ng genomic DNA template, 10 pg each of L and R end primers, 2 μl of 10× Taq buffer with KCl, and 1 μl (5 U) Taq DNA polymerase (Fermentas Life Sciences, Hanover, MD). PCR products were electrophoresed on a 2% MetaPhor agarose gel (Cambrex, East Rutherford, NJ) with 1× TAE and ethidium bromide at a total concentration of 1.2 μg/ml. SSLP markers were for z11862 (DNA Synthesis Laboratory at the University of Wisconsin Biotechnology Center, Madison, WI), and the primer sequences were as follows: upstream, 5′TCAGACATAAGCATGAGCGG3′; downstream, 5′GAGGTGTGTGTTCAGCTCGA3′.

Immunohistochemistry and In Situ Hybridization

Digoxigenin-UTP–labeled riboprobes for foxa2, gli1, gli2, myoD, crestin, wnt1, pax2, zic1, pax6, netrin1a, sema3D, shh, ephrinA5a, ephrinA5b, ephA4, zic2, slit1a, slit1B, slit2, slit3, robo1, robo2, robo3, and robo4 were synthesized by in vitro transcription and hydrolyzed to an average length of approximately 300 bases by limited alkaline hydrolysis (Cox et al., 1984). Whole-mount in situ hybridization was performed as described previously (Halloran et al., 1999).

Whole-mount immunohistochemistry was carried out as described previously (Wolman et al., 2004). Briefly, embryos were fixed in 4% paraformaldehyde overnight, blocked in 5% sheep serum and 2 mg/ml bovine serum albumin in phosphate buffered saline, and incubated in primary antibody. The primary antibodies used were ZN-12 (1:500) ZNP-1 (1:100) ZN-5 (1:500; all from Zebrafish International Resource Center, Eugene, OR), anti-acetylated α-tubulin (1:1,500; Sigma, St. Louis, MO), 3A10 (1:50), and F59 (1:100; both from Developmental Studies Hybridoma Bank, Iowa City, IA). Antibody labeling was completed with the Vectastain Mouse IgG ABC immunoperoxidase labeling kit (Vector Laboratories, Burlingame, CA) or with Alexa 488–conjugated secondary antibodies (4 μg/ml; Molecular Probes, Eugene, OR).

DiI–DiA Injections

DiI injections were carried out as described previously (Liu et al., 2004). Briefly, embryos were fixed in 4% paraformaldehyde and mounted in 1% low melting point agarose. A 0.2% solution of DiI (Molecular Probes) or DiA (4-4-dihexadecylaminostyryl-N-methylpyridinium iodide) in dimethyl formamide was pressure injected into the retina or the epiphysis. The dye was allowed to transport overnight.

Acridine Orange Staining

Live 24 hpf embryos were dechorionated and incubated in Acridine Orange (Sigma, St. Louis, MO) for 30 min. The embryos were then washed, mounted on coverslips, and visualized.

Imaging

All brightfield images were captured on a Nikon TE300 inverted microscope equipped with a Spot RT camera (Diagnostic Instruments, Sterling Heights, MI) and processed with MetaMorph software (Universal Imaging, West Chester, PA). Fluorescent images are confocal projections captured on a Zeiss Axiovert 100M microscope and the Bio-Rad 1024 Lasersharp Confocal microscope.

Acknowledgements

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES

We thank Derek Stemple, Steven Pollard, and Brian Link for sharing unpublished data; Hitoshi Okamoto and Anand Chandrasekhar for the Tg(Isl1:gfp) fish; Brian Link for the bala69 fish; and numerous other colleagues for probes. We thank Kari Obma for technical assistance and Aidan Tesch for fish care. M.C.H. was funded by NINDS. NSF supported acquisition of the confocal microscope to Jim Pawley (Department of Zoology, University of Wisconsin). The 3A10 antibody developed by Thomas M. Jessell and Jane Dodd and the F59 antibody developed by Frank E. Stockdale were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biological Sciences (Iowa City, IA).

REFERENCES

  1. Top of page
  2. Abstract
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. EXPERIMENTAL PROCEDURES
  7. Acknowledgements
  8. REFERENCES