SEARCH

SEARCH BY CITATION

Keywords:

  • axon guidance;
  • angiogenesis;
  • semaphorin;
  • VEGF;
  • primary motor neurons

Abstract

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

Neuropilin-1, a receptor for axon-repellent semaphorins and vascular endothelial growth factor (VEGF), functions both in angiogenesis and axon growth. Here, we show strong expression of neuropilin-1a in primary motor neurons in the trunk of embryonic zebrafish. Reducing the expression of neuropilin-1a using antisense morpholino oligonucleotides induced aberrant branching of motor nerves, additional exit points of motor nerves from the spinal cord, and migration of neurons out of the spinal cord along the motor axon pathway in a dose-dependent manner. These phenotypes could be partially rescued by coinjecting neuropilin-1a mRNA. Other axons in the spinal cord and head appeared unaffected by the morpholino treatment. In addition, neuropilin-1a morpholino treatment disturbed normal formation of blood vessels in the trunk of 24 hours postfertilization embryos, as shown by microangiography. Morpholinos to VEGF also disturbed formation of blood vessels but did not affect motor axons, indicating that correct formation of blood vessels is not needed for the growth of primary motor axons. Morpholinos to the semaphorin 3A homologs semaphorin 3A1 and semaphorin 3A2 also had no effect on motor axon growth. However, combined injections of neuropilin-1a morpholino, at a concentration that did not elicit axonal aberrations when injected alone, with VEGF, semaphorin 3A1, or semaphorin 3A2 morpholinos synergistically increased the proportion of embryos showing aberrant motor axon growth. Thus, neuropilin-1a in primary motor neurons may integrate signals from several ligands and is needed for proper segmental growth of primary motor nerves in zebrafish. Developmental Dynamics 234:535–549, 2005. © 2005 Wiley-Liss, Inc.


INTRODUCTION

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

Neuropilin-1 (NRP1) is a cell surface receptor for soluble class 3 semaphorins, of which semaphorin 3A (sema3A, semaIII, semaD, collapsin-1) has been studied most extensively in the developing nervous system (Raper, 2000). Interactions of sema3A with axons expressing NRP1 result in collapse of growth cones and repulsion of axons (reviewed in Bagri and Tessier-Lavigne, 2002; He et al., 2002), but there is also evidence for attractive functions of sema3A (Bagnard et al., 1998; Castellani et al., 2000). Mice deficient for either NRP1 or sema3A show defasciculation and target overshooting of cranial and spinal nerves (Kitsukawa et al., 1997; Taniguchi et al., 1997). NRP1 also functions as a receptor for the vascular endothelial growth factor (VEGF) in the formation of blood vessels (reviewed in Klagsbrun et al., 2002; Neufeld et al., 2002).

Recent observations suggest that semaphorin and VEGF signaling are probably not restricted to the nervous and vascular system, respectively. For example, sema3A acts as a repellent molecule for blood vessels in the quail forelimb (Bates et al., 2003). Both, sema3A-NRP1 and VEGF-NRP1 signaling is needed for heart morphogenesis in mice (Gu et al., 2003). In vitro, VEGF promotes neurite outgrowth and neuronal survival (Sondell et al., 2000; Bocker-Meffert et al., 2002; Rosenstein et al., 2003). Thus, several signaling molecules may converge on one receptor, NRP1, in the vascular and nervous system. In fact, there is evidence for functional competition of VEGF and sema3A for overlapping binding sites on NRP1 (Miao et al., 1999).

To gain further insight into the complex molecular interactions that govern axon guidance, we analyze the relative importance of NRP1a, one of the two zebrafish homologs of NRP1 (Lee et al., 2002; Bovenkamp et al., 2004; Martyn and Schulte-Merker, 2004; Yu et al., 2004), and its ligands in the outgrowth of primary motor axons in the trunk of embryonic zebrafish, a widely used model system to analyze the signals that pattern early motor axon growth (for review see Beattie, 2000). Several mutants that show deficits in the development of primary motor neurons have been described (Hutson and Chien, 2002). This system is relatively simple, with three primary motor neurons per spinal hemisegment, which grow axons out of the spinal cord along a common pathway in the middle of each trunk segment up to the horizontal myoseptum. The axon of the caudal primary motor neuron (CaP) is the first to grow, followed by the axons of the middle (MiP) and rostral (RoP) primary motor neurons. At the horizontal myoseptum, the CaP axon continues its growth toward the ventral somite forming the ventral motor nerve, whereas the MiP axon retracts and grows toward the dorsal somite. The RoP axon takes a lateral path from the horizontal myoseptum. In half of the hemisegments, a fourth primary motor neuron, called the variable primary motor neuron (VaP) is present. (Eisen et al., 1986, 1990; Myers et al., 1986; Westerfield et al., 1986).

During the outgrowth of primary motor axons, NRP1a is expressed in the ventral spinal cord, probably in primary motor neurons in embryonic zebrafish (Lee et al., 2002; Bovenkamp et al., 2004). At least three potential ligands for NRP1a are expressed in the zebrafish trunk. Sema3A2 (semaZ1b) transcripts are found in the posterior somite, bordering the midsegmental pathway of the ventral motor nerve (Roos et al., 1999), whereas sema3A1 (semaZ1a) mRNA is expressed in the dorsal and ventral somite leaving a corridor at the horizontal myoseptum negative of expression (Shoji et al., 1998). VEGF is expressed in the ventromedial somite (Liang et al., 1998). Overexpression of the two sema3A homologs, sema3A1 (semaZ1a, Halloran et al., 2000) and sema3A2 (semaZ1b, Roos et al., 1999), in the trunk of embryonic zebrafish induces truncations of primary motor nerves. This finding suggests axon-repellent functions of the molecules. Of interest, sema3A1 is also involved in the development of the dorsal aorta in the trunk (Shoji et al., 2003). VEGF-NRP1a signaling is necessary for blood vessel formation in the trunk (Nasevicius et al., 2000; Lee et al., 2002).

Here, we show that knockdown of NRP1a alone or in combination with sema3A1, sema3A2, or VEGF perturbs outgrowth of ventral motor nerves. This finding suggests that interactions of several ligands with NRP1a in primary motor neurons play a role in the patterning of peripheral nerves in zebrafish.

RESULTS

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

We analyzed axon outgrowth of the primary motor neurons in the trunk of embryonic zebrafish (Fig. 1A). Axon outgrowth starts with the CaP axon at 18 hours postfertilization (hpf). The axon grows on the common midsegmental path followed by the MiP and RoP axons toward the horizontal myoseptum, where axon paths diverge. By 24 hpf, when axon growth was analyzed by anti-tubulin immunohistochemistry, the CaP axon had reached the ventral myotome and the MiP axon had reached the dorsal myotome in most trunk segments, with the exception of the youngest, most caudal segments, in which axon growth occurred later. The RoP axon has advanced as far as the horizontal myoseptum by 24 hpf (Eisen et al., 1986; Myers et al., 1986; Westerfield et al., 1986).

thumbnail image

Figure 1. Expression of neuropilin-1a (NRP1a) and semaphorin 3A2 (sema3A2) in the trunk of embryonic zebrafish. A: A schematic side view of trunk segments at 18 and 24 hours postfertilization (hpf) is given. At 18 hpf, the caudal primary motor neuron (CaP) grows an axon out of the spinal cord. At 24 hpf, the axons of the middle (MiP) and rostral (RoP) primary motor neurons have followed on the common pathway to the horizontal myoseptum. The CaP axon is the only one growing ventrally beyond the horizontal myoseptum. B: Expression patterns of NRP1a (blue) and its potential ligands sema3A1 (orange), sema3A2 (red), and VEGF (green) are summarized for CaP in one trunk hemisegment. C: In a lateral view of a whole-mounted 16 hpf embryo at mid-trunk level (rostral is left), NRP1a mRNA is expressed in the dorsal spinal cord (asterisks), in motor neurons (mn), and in the hypochord (arrow). Arrowhead indicates expression in putative angioblasts. D: In a cross-section through the trunk at 16 hpf, expression is obvious in the dorsal spinal cord (asterisks) and the motor neurons (mn). E: In a lateral view of the caudal trunk of a 24 hpf embryo, expression of NRP1a is reduced in the dorsal spinal cord but is still strong in motor neurons (mn) and in a forming blood vessel (arrow). F: Sema3A2 is expressed in the caudal half of trunk myotomes (arrows; age and orientation as in E). G: In situ hybridization with an NRP1a sense RNA probe did not yield a signal (age and orientation as in E). H: A lateral view of a whole-mounted 24 hpf embryo is shown at caudal trunk level. NRP1a mRNA is labeled in red and tubulin protein in brown. Anti-tubulin–immunopositive motor axons (ax) can be seen to emerge from NRP1a mRNA-labeled motor neurons (mn). Rohon-Beard neurons (RhB) are labeled by the anti-tubulin antibody but not by the NRP1a in situ hybridization probe. I–L: Double labeling of NRP1a (red) and islet-1/-2 (brown) mRNAs at 24 hpf is shown (lateral trunk views, rostral is left). In I, the islet probes were omitted. In J, the NRP1a probe was combined with the islet-1 probe and in K with the islet-2 probe. In L, the NRP1a probe was combined with both probes. Arrows indicate cells labeled in red only (NRP1a). Arrowheads indicate double-labeled cells, and open arrowheads indicate cells labeled only in brown (islets). Scale bars = 25 μm in C,D,H, 50 μm in G (applies to E–G), 12.5 μm in L (applies to I–L).

Download figure to PowerPoint

NRP1a Is Expressed in Trunk Motor Neurons

To confirm earlier reports of NRP1a expression in motor neurons (Lee et al., 2002; Bovenkamp et al., 2004; Yu et al., 2004), we analyzed expression of NRP1a mRNA in the trunk during the outgrowth of motor axons by in situ hybridization. At 16 hpf, embryos show conspicuous expression of NRP1a mRNA in small clusters of cells at the ventral edge of the spinal cord consistent with the position of primary motor neurons in the trunk. There was also expression in the dorsal aspect of the spinal cord, in the hypochord, and in putative angioblasts in the ventral somite (Shoji et al., 2003; Fig. 1C,D).

At 24 hpf, expression of NRP1a mRNA in the position of motor neurons was strongest in the youngest, most caudal trunk segments. This finding indicates a developmental down-regulation of NRP1a mRNA in more mature rostral trunk segments. Expression in the dorsal spinal cord and in the ventral somite was strongly reduced at 24 hpf (Fig. 1E). Hybridization with sense probes did not yield any signal (Fig. 1G). Similarly, mRNA for one of the putative ligands of NRP1a, sema3A2, was also most strongly expressed in caudal trunk segments, where it was found in the posterior half of each somite (Fig. 1F and Roos et al., 1999). Other putative NRP1a ligands, such as sema3A1 and VEGF are also expressed during the outgrowth of primary motor axons (Fig. 1B). Sema3A1 mRNA is located in the dorsal and ventral somite, with a gap of expression at the horizontal myoseptum (Shoji et al., 1998), and VEGF mRNA is found in the ventromedial somite (Liang et al., 1998).

To directly demonstrate that expression of NRP1a mRNA in the ventral spinal cord was in primary motor neurons, we double labeled NRP1a mRNA together with probes for islet-1, a marker for RoP and MiP (Inoue et al., 1994), and islet-2, a marker for CaP and VaP (Appel et al., 1995; Tokumoto et al., 1995) by in situ hybridization. This strategy goes beyond previous studies in which motor neurons were identified only by position (Lee et al., 2002; Bovenkamp et al., 2004; Yu et al., 2004). Islet-1 and islet-2 probes were applied simultaneously and developed in brown to label all primary motor neurons. Developing the NRP1a signal in the same embryos (n = 19) in red yielded double-labeled cells, with no cells detectable that were only labeled in red (Fig. 1L). Omitting the islet probes as a negative control yielded only cells labeled in red (Fig. 1I, n = 17). Thus, we conclude that NRP1a is expressed only in primary motor neurons in the ventral spinal cord. The NRP1a signal (red) appeared to be weaker in the most rostral islet-1/-2–labeled cell, suggesting lower expression of NRP1a in RoP. This expression was also found when NRP1a was labeled together with islet-1 alone (n = 15). As expected, there were cells caudal to the double-labeled cells that were only labeled for NRP1a (Fig. 1J). These were probably CaP and VaP, the most caudal primary motor neurons, which are not labeled by the islet-1 probe. Conversely, double labeling of NRP1a with islet-2 alone (n = 17) yielded double-labeled cells and cells that were only labeled for NRP1a rostral to the double-labeled cells (Fig. 1K). This is because islet-2 only labels CaP and VaP and not the two most rostral primary motor neurons RoP and MiP. Double labeling of NRP1a mRNA and tubulin protein showed a close apposition of ventral motor nerves to the NRP1a-positive somata, further demonstrating that NRP1a-positive cells are motor neurons (Fig. 1H). Taken together, these results suggest expression of NRP1a in VaP, CaP, MiP, and to lower levels in RoP.

Morpholinos to NRP1a Induce Errors in the Growth of Primary Motor Axons

To analyze the function of NRP1a in motor axon outgrowth, we reduced the expression of NRP1a by specific morpholinos. Labeling of ventral motor nerves with an antibody to tubulin at 24 hpf revealed three major types of aberrations in the growth of these nerves after application of 1mM NRP1a morpholino1 (Fig. 2). Embryos (n = 53) were affected by aberrant branching of ventral motor nerves (79.4% affected embryos, P < 0.0001 against all controls), multiple exit points of ventral motor nerves (77.6% affected embryos, P < 0.0001 against all controls), or ventrally displaced neuronal somata (63.4% affected embryos, P < 0.0001 against all controls), compared with 7.9%, 7.9%, and 13.9% for the respective phenotypes in controls injected with a morpholino in which four bases were mismatched (1 mM NRP1a 4mm morpholino, n = 64, Table 1). Percentages of embryos with aberrations differed also significantly from those in uninjected embryos, in embryos injected with a standard control morpholino at 2 mM and in embryos injected with buffer for all phenotypes (Table 1). Truncations were rare and not significantly increased vs. all controls (data not shown).

thumbnail image

Figure 2. Neuropilin-1a (NRP1a) morpholinos induce aberrations of primary motor nerves. A–F: Lateral views of anti-tubulin–labeled whole-mounted 24 hpf embryos at mid-trunk levels are shown (rostral is left). A,B: In uninjected embryos (A) or those injected with 1 mM NRP1a 4mm morpholino (4mm, B), single unbranched motor nerves (arrows in A,B) grow ventrally out of the spinal cord. C–E: Injection of 1 mM NRP1a morpholino1 induced branching (arrows in C), a second exit point for motor axons per hemisegment (arrows in D), or anti-tubulin–labeled cells that appear to have migrated out of the spinal cord along the pathway of the ventral motor nerve (arrows in E). F: Injection of 1 mM NRP1a morpholino2 also induced aberrant branching (arrow) of the ventral motor nerve. Scale bar = 25 μm in F (applies to A–F).

Download figure to PowerPoint

Table 1. Morpholinos to NRP1a Induce Aberrant Ventral Motor Nerve Growtha
Injection typenEmbryos with aberrant ventral motor nerve branching (%)Embryos with multiple exits of ventral motor nerves (%)Embryos with displaced neurons (%)
  • a

    Morpholino doses are indicated in brackets. n, numbers of embryos analyzed. NRP1a, neuropilin-1a; MO, morpholino; NRP1a MO1/MO2, morpholino1/2 against NRP1a; NRP1a 4mm MO, morpholino with four-mismatched bases based on NRP1a morpholino1.

  • *

    P < 0.05 (Fisher's exact test) tested against all controls.

  • **

    P < 0.01 (Fisher's exact test) tested against all controls.

  • ***

    P < 0.001 (Fisher's exact test) tested against all controls.

Uninjected966.5 ± 3.42.3 ± 1.11.4 ± 1.4
Vehicle414.8 ± 4.82.4 ± 2.40.0 ± 0.0
Standard control MO (2 mM)544.8 ± 4.81.6 ± 1.63.2 ± 1.6
NRP1a 4mm MO (1 mM)657.9 ± 3.27.9 ± 3.213.9 ± 6.2
NRP1a MO1 (0.1 mM)7018.0 ± 3.711.0 ± 4.79.8 ± 7.6
NRP1a MO1 (0.25 mM)3417.9 ± 2.127.0 ± 20.4**17.9 ± 2.1
NRP1a MO1 (0.5 mM)17850.5 ± 7.2***60.5 ± 6.4***56.8 ± 5.5***
NRP1a MO1 (1 mM)5379.4 ± 6.2***77.6 ± 3.9***63.4 ± 8.9***
NRP1a MO2 (0.25 mM)3812.8 ± 7.213.1 ± 1.95.0 ± 5.0
NRP1a MO2 (1 mM)5350.6 ± 4.9***12.6 ± 4.05.6 ± 5.6
NRP1a MO2 (2 mM)6935.6 ± 14.4***21.7 ± 0.5*2.9 ± 0.1

In the following, we describe the phenotypes after injection of 1 mM NRP1a morpholino1 in more detail. Aberrant branching: instead of growing ventrally from the spinal cord as a single nerve toward the ventral myotome in an unbranched way (Fig. 2A,B), 40.7% of the nerves in affected hemisegments were aberrantly branched (Fig. 2C). Most of these branches (69.6%) were directed caudally. This nerve branching could be due to axonal branching with one axonal branch remaining on the midsegmental pathway, or due to one of the primary motor axons taking an aberrant course. Rostrally (20.2%) and bilaterally (10.2%) branched nerves were observed less frequently. On average, 4.2 ± 0.4 (SEM) hemisegments/embryo showed aberrant branching in affected embryos.

Multiple exit points of ventral motor nerves.

Instead of exiting the spinal cord only in one position in the middle of each hemisegment, nerves in 35.6% of the affected hemisegments showed mostly one additional exit point, with a nerve of variable length growing ventrally (Fig. 2D). This second nerve either ran parallel to the main nerve or joined it at variable positions on the somite. In those segments in which the segment borders could be visualized in differential interference contrast, it was possible to determine where additional nerve exit points were located. Most of the additional exit points were found in the posterior half of the somites (74.7%), 22.4% were located in the anterior half of the somite and 2.9% of the hemisegments had additional exit points rostral and caudal to the midsegmental pathway. On average, 3.7 ± 0.4 hemisegments/embryo showed multiple exits in affected embryos.

Ventrally displaced neurons.

In several of the affected hemisegments (18.7%), anti-tubulin–positive neuronal somata were found in the midsegmental pathway outside the spinal cord in contact with the nerve (Fig. 2E). This finding was almost never observed in controls (one displaced cell in 84 uninjected embryos). Most of these cells were dorsoventrally elongated along the nerve, giving the impression of having migrated out of the spinal cord along the nerve. On average, 2.5 ± 0.3 hemisegments/embryo showed ventrally displaced neurons in affected embryos. Taking all three phenotypes together, we found that of all hemisegments analyzed, 35.1% (438 of 1,247 analyzed hemisegments, n = 53 embryos) were abnormal.

The effect of NRP1a morpholino1 was dose-dependent (Table 1). The morpholino had no significant effect on ventral motor nerve branching at 0.1 mM and 0.25 mM (18.0% and 17.9% affected embryos, n = 70 and n = 34, respectively), but significantly increased the number of embryos with aberrant ventral motor nerve branching at 0.5 mM (50.5%, n = 178, P < 0.0001 against all controls) and 1 mM (see above), compared with uninjected controls (6.5% affected embryos, n = 96), buffer-injected embryos (4.8% affected embryos, n = 41), embryos injected with standard control morpholino at 2 mM (4.8% affected embryos, n = 54), and embryos injected with NRP1a 4mm morpholino at 1 mM (7.9% affected embryos, n = 65). Multiple exits of ventral motor nerves were significantly increased at 0.25 mM (27.0% affected embryos, P < 0.01 against all controls), at 0.5 mM (60.5% affected embryos, P < 0.0001 against all controls), and 1 mM (see above) but not at 0.1 mM (11.0% affected embryos) compared with all controls (2.3% of uninjected embryos, 2.4% of embryos injected with vehicle, 1.6% of embryos injected with 2 mM standard control morpholino and 7.9% of embryos injected with 1 mM NRP1a 4mm morpholino). At 0.1 mM and 0.25 mM, displaced neurons were found in 9.8% and 17.9% of the embryos, respectively, which was not significantly more than in all controls (uninjected, 1.4%; vehicle injected, 0%; standard control morpholino injected, 3.2%; and NRP1a 4mm morpholino injected, 13.9%), whereas at concentrations of 0.5 mM and 1 mM, the frequency of this phenotype was significantly increased to 56.8% (P < 0.0001 against all controls) and 63.4% of the embryos, respectively.

A second morpholino to NRP1a (NRP1a morpholino2) of nonoverlapping sequence with morpholino1 also induced aberrant ventral motor nerve branching (Fig. 2F) and additional exits of ventral motor nerves but not displaced neurons (Table 1). At 0.25 mM, NRP1a morpholino2 was ineffective, with 12.8% of embryos (n = 38) showing aberrant ventral motor nerve branching and 13.1% of embryos showing multiple exits of ventral motor nerves. At 1 mM, 50.6% of the embryos were affected by aberrant ventral motor nerve branching (n = 53), which was significantly more than in all controls (P < 0.0001 against all controls). This percentage was not increased by a higher concentration of 2 mM (35.6%, n = 69, P < 0.001 against all controls). The frequency of multiple exits of ventral motor nerves was not significantly different from all controls at 1 mM (12.6%) but was significant at 2 mM (21.7%, P < 0.05 against all controls). Thus, NRP1a morpholino2 confirmed most of the effects of NRP1a morpholino1 on motor axon growth but was less effective than NRP1a morpholino1.

Effect of NRP1a Morpholinos Can Be Partially Rescued by NRP1a mRNA Overexpression

To show the specificity of the observed effects of morpholinos to NRP1a, we coinjected NRP1a mRNA and NRP1a morpholino1 to rescue the morpholino-induced phenotype. Expression of myc-tagged NRP1a protein translated from the injected mRNA was demonstrated by immunohistochemistry (Fig. 3A,B). The outlines of individual cells were most strongly labeled, suggesting that the exogenous protein was associated with the plasma membrane. With progressing development, less protein was detectable. The protein was detected at 5 hpf in 16 of 17 embryos, at 16 hpf in 7 of 10 embryos, at 18 hpf in 3 of 10 embryos, and at 24 hpf in 1 of 19 embryos. The mRNA was expressed throughout the embryos in a mosaic pattern, i.e., the mRNA was not expressed in all cells (Fig. 3A). This finding is commonly observed with this type of mRNA overexpression study (McWhorter et al., 2003).

thumbnail image

Figure 3. Overexpression of neuropilin-1a (NRP1a) mRNA. A,B: In lateral views of 16 hours postfertilization (hpf) whole-mount embryos (rostral is left, dorsal is up; yolk sac has been removed), myc-tagged NRP1a mRNA is expressed in a mosaic pattern (A). B: No signal is observed in uninjected animals. The insert in A shows that outlines of cells are most prominently labeled, suggesting membrane-associated expression of the exogenous protein. Scale bars = 250 μm in B (applies to A,B), 10 μm in inset.

Download figure to PowerPoint

Analyzing motor axon outgrowth with anti-tubulin immunolabeling in 24 hpf embryos indicated that injection of the NRP1a mRNA alone had no significant effect on the growth of ventral motor nerves in the trunk (10.4% of NRP1a mRNA-injected embryos displayed branched nerves vs. 6.5% in uninjected controls; 7.7% NRP1a mRNA-injected embryos had multiple exits of motor nerves vs. 2.3% in uninjected controls; 0% of NRP1a mRNA-injected embryos had displaced neurons vs. 1.4% in uninjected controls; n = 36 NRP1a mRNA-injected embryos). We then performed paired experiments in which 0.5 mM NRP1a morpholino1 was injected either alone or in combination with the NRP1a mRNA. NRP1a morpholino1 is not complementary to the sequence of the overexpression construct and, thus, can only bind to endogenous NRP1a mRNA. In these experiments, the proportion of affected animals was significantly reduced for severe aberrant branching (> two branched nerves per animal: 42.3% affected animals with NRP1a morpholino1-injected vs. 34.1% in coinjected animals, P = 0.023), for multiple exits (65.9% in NRP1a morpholino1-injected vs. 46.5% in coinjected animals, P < 0.0001) and for displaced neurons (62.9% in NRP1a morpholino1-injected vs. 40.3% in coinjected animals, P = 0.0002; Table 2). Thus, all three observed motor axon phenotypes in NRP1a morpholino1-injected animals could be partially rescued by overexpression of NRP1a mRNA.

Table 2. Overexpression of NRP1a mRNA Partially Rescues the Motor Axon Phenotypea
Injection typenEmbryos with severe aberrant ventral motor nerve branching (%)Embryos with multiple exits of ventral motor nerves (%)Embryos with displaced neurons (%)
  • a

    Abbreviations as in Table 1.

  • *

    P < 0.05 (Fisher's exact test).

  • ***

    P < 0.001 (Fisher's exact test).

NRP1a MO1 (0.5 mM)13342.3 ± 6.865.9 ± 6.062.9 ± 5.1
NRP1a MO1 (0.5 mM) + NRP1a mRNA14034.1 ± 6.2*46.5 ± 4.0***40.3 ± 5.1***

Trunk Structures and Other Axon Trajectories Appeared Unaffected in NRP1a Morpholino-Treated Embryos

To exclude that the effect of the morpholino treatment was secondary to possible alterations of trunk morphogenesis, we labeled several structures after application of 1 mM NRP1a morpholino1, which had the strongest effect on ventral motor nerves, and compared patterns with those in uninjected embryos. The notochord, which underlies the pathway of ventral motor axons, and the spinal floor plate were labeled with antibodies to chondroitin sulfates (Bernhardt and Schachner, 2000) and appeared normal (Fig. 4A,B; n = 22 NRP1a morpholino1-injected embryos). In double-labeling experiments, vertical myosepta, labeled with antibodies to tenascin-C (Bernhardt et al., 1998), appeared normal in segments in which motor axons, labeled with anti–HNK-1 antibodies, grew aberrantly (Fig. 4C,D; n = 6 NRP1a morpholino1-injected embryos). At the horizontal myoseptum, an important choice point for growing motor axons in the trunk, double labeling of muscle pioneer cells (Melancon et al., 1997) with antibodies to the engrailed protein and motor axons with anti-tubulin antibodies (Fig. 4E,F; n = 16 NRP1a morpholino1-injected embryos) indicated normal differentiation of muscle pioneer cells in segments with aberrant motor axons. For each labeling pattern analyzed, 10 to 24 uninjected control embryos were used as a reference. Thus, aberrations in motor axons were probably not caused by aberrations in trunk structures, such as vertical myosepta, notochord, and muscle pioneer cells at the horizontal myoseptum.

thumbnail image

Figure 4. Trunk structures appear normal after injection of 1 mM neuropilin-1a (NRP1a) morpholino1. Lateral views of whole-mounted 24 hours postfertilization (hpf) embryos at mid-trunk levels are shown (rostral is left). A–F: Notochord (nc) and spinal floor plate (fp) labeled with an anti-chondroitin sulfate antibody (A,B), vertical myosepta (arrowheads) labeled with an anti–tenascin-C antibody (red), and ventral motor axons labeled with an anti–HNK-1 antibody (green, C,D), as well as muscle pioneer cells (mp) at the horizontal myoseptum labeled with an antibody to engrailed, and ventral motor axons labeled with an anti-tubulin antibody (E,F) did not show systematic differences between uninjected embryos (A,C,E) and those injected with 1 mM NRP1a morpholino1 (B,D,F). Arrows in D and F indicate aberrant branches of ventral motor nerves. Scale bars = 25 μm in B (applies to A,B), 25 μm in D (applies to C,D), 25 μm in F (applies to E,F).

Download figure to PowerPoint

The presence and normal positioning of motor neurons was controlled with an antibody to islet-1/-2 proteins (Becker et al., 2002). In embryos injected with 1 mM NRP1a morpholino1, immunopositive cells in the ventral spinal cord were labeled at a density that was not significantly different from that in uninjected control embryos (60.3 ± 3.4 cells in segments 5–7 of six NRP1a morpholino1-injected embryos vs. 58.7 ± 2.1 cells in segments 5–7 of six uninjected embryos, Fig. 5A,B). The only exception was occasional labeling of single cell nuclei ventral to the spinal cord in some hemisegments (Fig. 5B). These ectopic cells reflect the presence of the anti-tubulin antibody-labeled cells in the somitic pathway of motor nerves. Therefore, these cells may be motor neurons that had migrated out of the spinal cord. Intensely labeled large nuclei of putative Rohon-Beard cells in the dorsal spinal cord were located in their normal position and density (26.8 ± 2.0 Rohon-Beard cells in segments 5–7 of six NRP1a morpholino1-injected embryos vs. 25.7 ± 1.6 Rohon-Beard cells in segments 5–7 of six uninjected embryos).

thumbnail image

Figure 5. Other neurons and axons in the head and spinal cord are not affected by NRP1a morpholino1. Lateral views of 24 hpf whole-mounted embryos at mid-trunk (A,B,E,F) or head (C,D) levels are shown (rostral is left). A,B: Labeling Rohon-Beard (RhB) and motor neurons (mn) with an antibody to islet-1 indicates comparable numbers of these cell types in uninjected embryos (A) and those injected with 1mM NRP1a morpholino1 (B). Whereas Rohon-Beard neurons and most motor neurons are found in their correct locations, one immunopositive cell (arrow) is displaced ventral to the spinal cord in B. C,D: Anti-tubulin labeling of the dorsoventral diencephalic tract (dvdt) and the posterior commissure (pc) reveals no significant differences between uninjected embryos (C) and those injected with 1 mM NRP1a morpholino1 (D). E,F: Labeling of commissural primary ascending interneurons in the spinal cord with the 3A10 antibody indicated normal positioning of somata (CoPA) and contralateral axons (arrowheads), which eventually join the dorsal longitudinal fascicle (DLF) in uninjected embryos (E) and those injected with 1mM NRP1a morpholino1 (F). Scale bars = 25 μm in B (applies to A,B), 25 μm in D (applies to C,D), 25 μm in F (applies to E,F).

Download figure to PowerPoint

Trajectories of other axons were also analyzed in anti-tubulin–labeled embryos. In the head of embryos injected with 1 mM NRP1a morpholino1, the prominently visible dorsoventral diencephalic tract and the posterior commissure appeared normal (n = 12; Fig. 5C,D). In the spinal cord, Rohon-Beard cells, the dorsal longitudinal fascicle and the medial longitudinal fascicle, peripheral processes of Rohon-Beard neurons, as well as trigeminal neurons and axons appeared normal compared with uninjected embryos (n = 21; not shown). MiP and RoP primary motor axons could not be evaluated, because their trajectories are not discernible in anti-tubulin immunohistochemistry.

The 3A10 antibody to a neurofilament-associated protein specifically labels somata and axons of the Mauthner neurons in the brainstem and the commissural primary ascending interneurons in the spinal cord at 24 hpf, similar to the CON1 antibody (Bernhardt et al., 1990). In embryos injected with 1 mM NRP1a morpholino1 (n = 11), the Mauthner neurons were normally positioned and sent their crossed axons into the spinal cord in a manner that was indistinguishable from uninjected controls (n = 10; not shown). Large commissural primary ascending interneurons were also normally located in the dorsal spinal cord, projected ventrally, crossed the midline, and projected in the contralateral dorsal longitudinal fascicle, as in uninjected controls (Fig. 5E,F). Thus, several other axon trajectories were not affected by the injection of NRP1a morpholino1.

Formation of Primary Motor Nerves Does Not Depend on the Presence of Blood Vessels

Injection of NRP1a morpholino1 also disturbs the formation of blood vessels in the trunk so that aberrations in motor axon growth could be secondary to the loss of blood vessels (Lee et al., 2002). To analyze if motor axons would grow in their normal pathway in the absence of blood vessels in the trunk, we injected morpholinos to VEGF, known to inhibit blood vessel development (Nasevicius et al., 2000; Lee et al., 2002). Injection of 1 mM VEGF morpholino did not induce ventral motor axon aberrations (2.5% embryos with ventral motor nerve branching, 1.7% embryos with multiple exits of ventral motor nerves or displaced neurons, n = 49; Fig. 6B,D).

thumbnail image

Figure 6. Differential effects of different morpholinos on blood vessel and on motor axon development in the trunk of 24 hours postfertilization (hpf) embryos. A,C,E: Lateral views of whole embryos subjected to microangiography are shown. Fluorescence of yolk sacs is near the injection site and does not indicate the presence of blood vessels. B,D,F: Mid-trunk levels of embryos labeled with anti-tubulin antibodies subsequent to micro-angiography are shown (rostral is left). A,B: In uninjected embryos, the trunk vasculature (A, arrow) and ventral motor nerves (B) develop normally. C,D: In embryos injected with 1 mM vascular endothelial growth factor (VEGF), morpholino trunk vessels fail to develop (C) but motor nerves grow normally (D). E,F: In embryos injected with 1 mM NRP1a morpholino1, no trunk vessels are labeled (E) and ventral motor nerves grow abnormally (F). Arrow in F indicates a branched ventral motor nerve. Scale bars = 250 μm in E (applies to A,C,E), 25 μm in F (applies to B,D,F).

Download figure to PowerPoint

To analyze the effect on blood vessels and motor axons in the same set of embryos, microangiography was performed on uninjected embryos (n = 7), on embryos injected with 1mM NRP1a morpholino1 (n = 8), and on embryos injected with 1 mM VEGF morpholino (n = 11) before labeling of motor axons with an anti-tubulin antibody. In all uninjected embryos, blood flow through axial trunk blood vessels at 24 hpf could be detected by microangiography (Fig. 6A). At this developmental stage, intersegmental vessels are just beginning to form (Childs et al., 2002) and are not filled by microangiography. However, there was no blood flow through axial vessels in VEGF morpholino- and NRP1a morpholino1-injected embryos (Fig. 6C,E). Subsequent analysis of ventral motor axon growth revealed that aberrations of ventral motor nerves were only present in those embryos injected with NRP1a morpholino1 (Fig. 6B,D,F). Thus, primary motor axons can grow correctly in the absence of normal blood vessel differentiation in the trunk, as shown in VEGF morpholino-injected embryos.

VEGF, sema3A1, and sema3A2 Morpholinos Act Synergistically With NRP1a Morpholinos

To determine the contribution of potential NRP1a ligands to the motor axon phenotype observed after NRP1a morpholino treatment, we injected morpholinos to potential ligands of NRP1a alone or in combination with NRP1a morpholinos. Results are summarized in Table 3. Similar to injecting VEGF morpholinos, injection of morpholinos to the sema3A homologs sema3A1 and sema3A2 had no significant effects on the growth of ventral motor axons when injected at a concentration of 2 mM (sema3A1 morpholino: 14.0% embryos with aberrantly branched ventral motor nerves, 10.0% embryos with multiple exits of ventral motor nerves, and 1.8% embryos with displaced neurons, n = 47; sema3A2 morpholino: 14.9% embryos with aberrantly branched ventral motor nerves, 3.3% embryos with multiple exits of ventral motor nerves, and 6.7% embryos with displaced neurons, n = 43). There are several potential NRP1a ligands expressed in the trunk, which may act redundantly and could compensate for the reduction in the expression of single ligands, such as VEGF, sema3A1, and sema3A2. We hypothesized that compensation might not be possible if the availability of NRP1a was reduced in experimental embryos. Thus, we decided to slightly reduce NRP1a expression by injecting 0.1 mM NRP1a morpholino1, a concentration that is ineffective on its own to elicit a motor axon phenotype, in combination with 1 mM VEGF, 2 mM sema3A1 or 2 mM sema3A2 morpholinos, which also did not affect motor axon outgrowth when injected alone. Coinjections of NRP1a morpholino1 with either VEGF or sema3A2 morpholinos significantly induced motor axon branching (+ VEGF morpholino: 44.2%, n = 61, P = 0.0032; + sema3A2 morpholino: 32.2% affected embryos, n = 80, P = 0.0008; Fig. 7A,D) and displaced neurons (+ VEGF morpholino: 31.8%, P < 0.0001; + sema3A2 morpholino: 9.7%, P = 0.0082; Fig. 7B,E) compared with combinations with 0.1 mM NRP1a 4 mismatch morpholino (+ VEGF morpholino: 23.0% embryos with aberrant branching and 0% embryos with displaced neurons, n = 71; + sema3A2 morpholino: 10.8% embryos with aberrant branching and 0% embryos with displaced neurons, n = 66) but not multiple exits of motor nerves (+ VEGF morpholino: 15.1% vs. 12.8% control embryos; + sema3A2 morpholino: 18.4% vs. 13.0% control embryos). The combination with Sema3A1 morpholino showed only a significant increase of the percentage of embryos with displaced neurons (30.4%, n = 62, P < 0.0001, Fig. 7C) compared with the control combination (2.8%, n = 76), whereas no significant differences were found regarding motor nerve branching (34.3% vs. 19.7%) and multiple exits (17.5% vs. 9.8%).

Table 3. Synergistic Effects of Coinjection of Different Morpholinosa
Injection typenEmbryos with aberrant ventral motor nerve branching (%)Embryos with multiple exits of ventral motor nerves (%)Embryos with displaced neurons (%)
  • a

    VEGF, vascular endothelial growth factor. Other abbreviations as in Table 1. VEGF 4mm MO, morpholino with four mismatched bases derived from the VEGF morpholino.

  • **

    P < 0.01 (Fisher's exact test).

  • ***

    P < 0.001 (Fisher's exact test).

NRP1a MO1 + VEGF 4mm MO810.0 ± 0.07.2 ± 0.85.2 ± 3.7
     
NRP1a 4mm MO + VEGF MO7123.0 ± 7.212.8 ± 4.80.0 ± 0.0
NRP1a MO1 + VEGF MO6144.2 ± 15.1**15.1 ± 3.131.8 ± 3.1***
     
NRP1a 4mm MO + Sema3A1 MO7619.7 ± 5.19.8 ± 4.62.8 ± 1.7
NRP1a MO1 + Sema3A1 MO6234.3 ± 8.217.5 ± 3.530.4 ± 18.2***
     
NRP1a 4mm MO + Sema3A2 MO6610.8 ± 5.813.0 ± 4.60.0 ± 0.0
NRP1a MO1 + Sema3A2 MO8032.2 ± 8.9***18.4 ± 7.19.7 ± 3.7***
     
NRP1a 4mm MO + L1.1 MO494.2 ± 2.40.0 ± 0.02.3 ± 2.3
NRP1a MO1 + L1.1 MO4711.5 ± 7.88.7 ± 2.17.0 ± 7.0
thumbnail image

Figure 7. Subthreshold concentrations of neuropilin-1a (NRP1a) morpholino1 in combination with vascular endothelial growth factor (VEGF), semaphorin (sema) 3A1, or sema3A2 morpholinos induce aberrant growth of motor nerves. Lateral views of anti-tubulin–labeled whole-mounted 24 hours postfertilization (hpf) embryos at mid-trunk levels are shown (rostral is left). A–E: Combinations of NRP1a morpholino1 with VEGF (A) or sema3A2 (D) morpholinos induce aberrant motor nerve branching (arrows in A,D), whereas displaced neurons in the path of the ventral motor nerve (arrows in B,C,E) occur after combination of NRP1a morpholino1 with VEGF (B), sema3A1 (C), or sema3A2 (E) morpholinos. Scale bar = 25 μm in E (applies to A–E).

Download figure to PowerPoint

As an additional control, subthreshold concentrations of NRP1a morpholino1 (0.1 mM) were injected in combination with a four-base mismatch control morpholino for VEGF (1 mM), which did not elicit aberrant growth of motor axons (0% embryos with aberrant ventral motor nerve branching, 7.2% embryos with multiple exits of ventral motor nerves, and 5.2% embryos with displaced neurons, n = 81). The synergistic effects observed in the experiments with subthreshold amounts of morpholinos indicate a contribution of VEGF as well as of sema3A1 and sema3A2 signaling to the correct outgrowth of the ventral motor nerve in the trunk.

We also tested a potential coreceptor for NRP1a, the cell recognition molecule L1.1 (Castellani et al., 2000), which is expressed in primary motor neurons (Tongiorgi et al., 1995a). Injection of L1.1 morpholino alone had no significant effect on the growth of ventral motor axons when injected at a concentration of 2 mM (4.9% embryos with aberrant ventral motor nerve branching, 9.8% embryos with multiple exits of ventral motor nerves, and 0% embryos with displaced neurons, n =51). Injecting a concentration of 2 mM L1.1 morpholino in combination with the subthreshold concentration of NRP1a morpholino1 (0.1 mM) also did not induce a significant increase of either aberrant branching of ventral motors (11.5% vs. 4.2% embryos coinjected with 0.1 mM NRP1a 4mm morpholino, n = 49 and 47, respectively), of additional exits of ventral motor nerves (8.7% vs. 0% embryos coinjected with 0.1 mM NRP1a 4mm morpholino), or of displaced neurons (7.0% vs. 2.3% embryos coinjected with 0.1 mM NRP1a 4mm morpholino), even though the L1.1 morpholino effectively abolishes L1.1 immunoreactivity on motor axons for at least 24 hpf (data not shown). This finding indicates that, in the context of primary motor axon outgrowth, L1.1 may not play a major role.

DISCUSSION

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

We show here that NRP1a expression in motor neurons is necessary for the correct outgrowth of ventral motor nerves in embryonic zebrafish. Synergistic effects of injecting morpholinos to NRP1a together with morpholinos to the potential ligands VEGF, sema3A1, and sema3A2 suggest a role of NRP1a in integrating signals of several ligands in the trunk environment.

Morpholino Injections Directly Affect Motor Neurons

Several observations suggest that the observed effects were directly due to reduced NRP1a expression in the motor neurons, rather than being secondary to alterations of the trunk or spinal cord environment. Two morpholinos of nonoverlapping sequence show very similar effects. All effects of morpholino knockdown also could be partially rescued by overexpression of NRP1a mRNA. This finding argues that the effect of the morpholino is due to reduced NRP1a mRNA levels. That the rescue was not complete may best be explained by the mosaic expression of NRP1a protein after mRNA injection. Moreover, that during late phases of axon growth (24 hpf) protein expression from the exogenous mRNA could hardly be detected by immunohistochemistry for the myc epitope indicates diminished abundance of the exogenous protein as embryos develop. A similar incomplete rescue of motor axon phenotypes by mRNA overexpression has also been observed by others (McWhorter et al., 2003).

In addition to affecting motor axon growth, morpholinos to NRP1a inhibit blood vessel formation (Lee et al., 2002), which we confirm here. However, the motor axon phenotypes observed after NRP1a morpholino injection are probably independent of altered blood vessel formation. This is suggested by the observation that treatment with a morpholino to VEGF inhibits the formation of the vasculature in the trunk but does not detectably influence the outgrowth of the ventral motor nerve. Sema3A1 morpholinos also affect blood vessel formation in the trunk (Shoji et al., 2003), but the morpholino used here had no detectable effect on motor axons when injected alone. Thus, early motor axon development is probably independent of blood vessel formation.

Furthermore, vertical and horizontal myosepta formed and differentiated correctly in NRP1a morpholino-injected embryos as indicated by tenascin-C labeling of vertical myosepta and labeling of muscle pioneer cells at the horizontal myoseptum, a critical choice point for growing motor axons, with an antibody to engrailed (Melancon et al., 1997). This finding suggests that the somitic pathway of motor axons was not grossly changed in morpholino-injected embryos.

Even though NRP1a mRNA is expressed in the dorsal spinal cord in uninjected embryos at 16 hpf, the organization of the spinal cord appeared unaltered in morpholino-injected embryos at 24 hpf. Labeling with the 3A10 antibody indicated correct positioning of commissural primary ascending neurons and their axons in the dorsal spinal cord. Mauthner axons in the ventral spinal cord were also unaffected by morpholino injections. Immunolabeling with an antibody to islet-1/-2 proteins indicated normal densities and positioning of Rohon-Beard cells and motor neurons, with the notable exception of those putative motor neurons that had migrated out of the spinal cord. Thus, motor axon aberrations were probably not secondary to effects on the spinal cord organization. Several other axon trajectories, such as the dorsoventral diencephalic tract, the posterior commissure in the head, or peripheral axons of Rohon-Beard neurons all appeared normal, indicating that there was no generalized alteration of axon growth in morpholino-injected embryos. However, we cannot exclude defects of MiP and RoP axons, because these could not be selectively labeled.

NRP1a in Primary Motor Neurons May Be a Receptor for Repellent Signals in the Somite Environment

Different phenotypes were observed after application of NRP1a morpholinos: branching of ventral motor nerves, more than one nerve exiting the spinal cord, and the ventral migration of putative motor neurons out of the spinal cord. The most frequently observed phenotype, branching of the ventral motor nerve (40.7% of all aberrant hemisegments), was either due to one of the three primary motor axons taking an aberrant path or due to axonal branching of a primary motor axon. This differentiation cannot be distinguished in anti-tubulin–labeled preparations. However, this result clearly shows axon growth into territories normally not invaded. This is also true for multiple-exit phenotypes and is in agreement with the notion that NRP1a is a receptor for axon-repellent signals (Bagri and Tessier-Lavigne, 2002). Reducing NRP1a expression may release axons from these repulsive signals. Of interest, the majority of nerve branches were directed into the posterior part of the somite (69.6%). Additional exit points from the spinal cord were also most frequently observed in the posterior part of the somite (74.7%). Transcripts for the NRP1a ligand sema3A2 are concentrated in the posterior part of the somite (Roos et al., 1999). However, other potential ligands of NRP1a show a different distribution. Sema3A1 is expressed in the dorsal and ventral somite, leaving a corridor at the horizontal myoseptum that is free of sema3A1 transcript.

Overexpression of the putative NRP1a ligands sema3A1 and sema3A2 in transgenic animals or by mRNA injections both induced truncations of ventral motor nerves (Roos et al., 1999; Halloran et al., 2000). This truncation can be considered a complementary phenotype to the abnormal axon branching and multiple exits observed in NRP1a morpholino-treated animals. Similar complementary phenotypes are also observed when the function of axon-repellent chondroitin sulfates is analyzed in the pathway of the ventral motor nerve. Enzymatic removal of these glycostructures induces abnormal branching of ventral motor nerves, whereas injecting a chondroitin sulfate mixture leads to nerve truncations (Bernhardt and Schachner, 2000). Therefore, our findings further support the hypothesis that NRP1a is a receptor for repellent semaphorin ligands in the somite environment.

We observed dorsoventrally elongated cells along the ventral motor nerve pathway in morpholino-injected embryos. That these ectopic cells were labeled by antibodies to the neuronal marker tubulin and to islet-1, a marker of motor neurons in zebrafish (Tokumoto et al., 1995), suggests that ectopic cells were motor neurons. Their shape and position is suggestive of a scenario in which these cells migrated out of the spinal cord along the motor axon pathway. One possible explanation for this phenotype is that, normally, repulsive signals from the somite restrict motor neuron somata to the spinal cord and that this repulsion is released under conditions of reduced NRP1a expression. Of interest, this phenotype is reminiscent of motor neurons exiting the ventral spinal cord after the ablation of so-called boundary cap cells at the motor axon exit point in chicks. The molecular signals from boundary cap cells are unknown (Vermeren et al., 2003).

Overexpression of NRP1a mRNA alone did not induce aberrations in ventral motor nerves, whereas in transgenic mice, overexpression of NRP1 induces abnormal sprouting and defasciculation of nerves (Kitsukawa et al., 1995). It is possible that the dose of NRP1a protein that was reached in our experiments was too low to induce such effects, or that these would occur only later, when secondary motor axons join the nerves.

Multiple Guidance Cues Are Present in the Trunk Environment

Synergistic effects of morpholinos to sema3A1, sema3A2, and VEGF in combination with subthreshold concentrations of NRP1a morpholino suggest that these molecules are interdependent in vivo. That morpholino knockdown of individual ligands was ineffective suggests that some of these potential NRP1a ligands act redundantly and may substitute for each other in experiments in which morpholinos to only a single ligand were injected. The observation that, when injected alone, VEGF morpholino had a severe effect on blood vessel formation but no detectable effect on motor axons indicates that the VEGF signal is indispensable for angiogenesis but not for axon growth. Nevertheless, VEGF appears to contribute to the guidance of motor axons as revealed in coinjection experiments with NRP1a morpholino. The same is true for sema3A1 and sema3A2. However, not all combinations of morpholinos tested elicit all of the phenotypes observed when NRP1a expression alone is knocked down at above threshold concentrations (Table 3), and the multiple-exit phenotype is not elicited by any of the combinations, suggesting some nonoverlapping functions of additional ligands. For example, sema3D has been shown to signal through NRP1a in the developing zebrafish CNS (Wolman et al., 2004), and the class 3 semaphorin sema3G (semaZ8, Halloran et al., 1998), which may act through NRP1a, is expressed in the somite.

Of interest, coinjections of VEGF morpholinos with NRP1a morpholinos induced phenotypes similar to those of coinjecting semaphorin morpholinos with NRP1a morpholinos (branching and ventrally migrating cells), even though mRNAs for VEGF and individual sema3A homologs are all differentially expressed in the somite environment (Fig. 1B). Moreover, VEGF and semaphorins have been described as functional competitors in the vascular system (Miao et al., 1999) and VEGF has been reported to promote neurite outgrowth in vitro (Sondell et al., 2000; Bocker-Meffert et al., 2002; Rosenstein et al., 2003), whereas sema3A mostly repels axons (reviewed in Bagri and Tessier-Lavigne, 2002; He et al., 2002). Disturbing the interactions of different ligands with NRP1a on motor axons in vivo may destabilize a complex axon guidance system. This destabilization may then result in comparable phenotypes. Similarly, overexpression and morpholino knockdown of sema3A1 both inhibit the formation of blood vessels in the trunk of embryonic zebrafish (Shoji et al., 2003).

Thus, there are multiple guidance cues present in the somite for motor nerves growing out of the spinal cord. Some of these (sema3A1, sema3A2, and VEGF) may act through NRP1a as shown in this study, whereas others, such as chondroitin sulfates may act on the axons in a different way. The presence and parallel action of multiple guidance cues in motor axon outgrowth is not unique to zebrafish but has also been recognized in other vertebrates (Tannahill et al., 2000; Schneider and Granato, 2003). Here, we show that NRP1a is necessary for the correct outgrowth of primary motor axons in the trunk of embryonic zebrafish.

EXPERIMENTAL PROCEDURES

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

Zebrafish

Adult zebrafish were kept at a 14-hr light and 10-hr dark cycle and fed dried fish food and live brine shrimp several times a day. Embryos were collected from our breeding colony according to standard procedures and staged in hours postfertilization at the standard temperature of 28.5°C (Kimmel et al., 1995).

Injection of mRNA and Morpholinos

For NRP1a mRNA overexpression, a vector was generated that contained untranslated 5′ sequences and the entire open reading frame of NRP1a followed by a myc-tag. Capped mRNA was synthesized with the mMessage mMachine Kit (Ambion, Huntingdon, England) followed by extension of the polyA tail with the Poly(A) tailing kit (Ambion), each according to the manufacturer's instructions. Sema3A1 and sema3A2 overexpression constructs used to show the efficiency of the morpholinos in vivo were generated in a similar way, containing only the first ∼300 bp of the coding sequence, including the morpholino binding site. For the sema3A2 construct, polyA tailing was required, whereas the myc-tagged protein of the sema3A1 construct was sufficiently detectable in 16 hpf embryos without polyA tailing.

Morpholinos were purchased from Gene Tools (Philomath, OR) and solubilized in Danieau solution according to the manufacturer's instructions. We used two morpholinos of nonoverlapping sequence for NRP1a, designated NRP1a morpholino1, which has been described (Lee et al., 2002) and NRP1a morpholino2 (GATCAACACTAATCCACAATGCATC). A morpholino in which four bases were mismatched, based on morpholino1 (NRP1a 4mm morpholino: GATTCCAGGAGTTCGGACTGCCGAA) and a standard control morpholino with a random sequence (CCTCTTACCTCAGTTACAATTTATA) were injected as controls. The morpholino to VEGF has been described previously (Nasevicius et al., 2000). Morpholinos to sema3A1 (AAAAATCCCAACAAGGTAATCCATG), sema3A2 (GTACAA- TCCACCACAAGTAGTCCAT), and L1.1 (ATGAAAACAGCCCCGACTCCAGACA) were also used.

The efficiencies of NRP1a morpholino1 (Lee et al., 2002), the VEGF morpholino (Nasevicius et al., 2000), and the morpholino to L1.1 (Becker et al., 2004) have been demonstrated previously. The sequence of NRP1a morpholino2 overlaps with the NRP1a overexpression construct. Therefore, we could test activity of the morpholino by coinjecting mRNA and morpholino. Coinjection suppressed detectability of the myc epitope in 23 of 25 embryos, whereas injection of the overexpression construct alone led to detectability of the myc epitope in 20 of 20 embryos at 5 hpf. Thus, the morpholino binds to the injected mRNA and suppresses its translation. The activities of the sema3A1 and sema3A2 morpholinos were tested in the same way, using the myc-tagged reporter constructs described above. Injecting the sema3A1 construct alone induced protein expression in 19 of 21 embryos. Protein expression was not detectable in any of the 16 embryos coinjected with the mRNA and the sema3A1 morpholino. After injection of the sema3A2 construct, 9 of 10 embryos expressed the protein compared with 0 of 7 embryos coinjected with the mRNA and the sema3A2 morpholino. This finding indicate specific binding of morpholinos to sema3A1 and sema3A2.

For injections, rhodamine dextran (0.5%, Mr = 10 × 103, Molecular Probes) was added to mRNA or morpholino solutions. A glass micropipette was filled with the mRNA (1–2 μg/μl) or morpholino solutions (up to 2 mM) and a volume of 0.5 to 1 nl per egg (one- to four-cell stage) was injected using a Picospritzer (PLI-100, Medical Systems Corp., Greenvale, NY).

In Situ Hybridization

The probes for detecting NRP1a (Lee et al., 2002), sema3A2 (Roos et al., 1999), and islet-1 and islet-2 (Tokumoto et al., 1995) have been described. Probes were labeled with digoxigenin using the Megascript kit (Ambion) and used on 16 and 24 hpf whole-mounted embryos as described (Tongiorgi et al., 1995b). Briefly, whole embryos were incubated with the probe at 55°C overnight and washed extensively at 55°C. The hybridized probe was detected using an alkaline phosphatase–coupled anti-digoxigenin antibody (Roche, Mannheim, Germany). The signal was developed using SIGMAFAST 5-bromo-4-chloro-3-indolyl phosphate/nitro blue tetrazolium (BCIP/NBT) tablets (Sigma-Aldrich, Deisenhofen, Germany), resulting in a brownish reaction product. For double labeling with immunohistochemistry, we used SIGMAFAST Fast Red TR/Naphthol AS-MX tablets (Sigma-Aldrich) to yield a red precipitate. The sense probes served as negative controls and did not show a signal.

Double in situ hybridization was performed with the NRP1a probe labeled with fluorescein and the islet-1/-2 probes labeled with digoxigenin according to previously published protocols (Jowett, 2001). Embryos were simultaneously incubated with the probes and sequentially detected with alkaline phosphatase–coupled antibodies to fluorescein (Roche) and digoxigenin. NRP1a probes were developed with Fast Red and, after inactivation of the alkaline phosphatase with 0.1 M glycine-HCl, 0.1% Tween-20, pH 2.2, islet-1/-2 probes were developed with BCIP/NBT. Specificity of the labeling was tested by omitting the islet probes, which did not yield a brown precipitate (Fig. 1I).

Immunohistochemistry

The polyclonal antibody against tenascin-C of zebrafish has been described (Tongiorgi, 1999). Monoclonal antibody (mAb) CS-56 to chondroitin sulfates was purchased from Sigma-Aldrich. mAb 9E10, recognizing the myc-epitope, was purchased from Santa Cruz Biotechnology (Santa Cruz, CA). The 40.2D6 antibody to islet-1/-2 and the antibody 3A10 to a neurofilament-associated antigen were both developed by Dr. T.M. Jessell (Columbia University, NY). These antibodies and the 4D9 antibody to engrailed (Patel et al., 1989) were obtained as cell culture supernatants from the Developmental Studies Hybridoma Bank maintained by The University of Iowa, Department of Biological Sciences (Iowa City, IA). Ventral motor axons were visualized with a monoclonal antibody against acetylated tubulin (6-11B-1; Sigma-Aldrich) or the 412 monoclonal antibody to the HNK-1 epitope (Becker et al., 2001). Cy2-, Cy5-, and horseradish peroxidase (HRP) -conjugated secondary antibodies to rabbit and mouse were used (Dianova, Hamburg, Germany). Immunohistochemistry was performed as described previously (Becker et al., 2001). Briefly, embryos were deeply anesthetized in 0.1% aminobenzoic acid ethyl methyl ester in phosphate buffered saline, fixed in 4% paraformaldehyde, incubated with primary antibodies overnight, and with the appropriate secondary antibodies, again overnight. HRP labeling was detected with diaminobenzidine as a substrate. Omitting the primary antibody did not result in any labeling.

Microangiography

Microangiography was performed as previously described (Lee et al., 2002; Goishi et al., 2003). Briefly, fluorescein isothiocyanate (FITC) -dextran (Sigma) in 75 mM NaCl solution was injected into the sinus venosa, which results in labeling of the entire vascular system. For fluorescent microscopy, a FITC filter was used. Zebrafish embryos were visualized with an Olympus SZX12 stereomicroscope and photographed using an Olympus DP11 digital camera.

Analysis of Nerve Growth

We analyzed anti-tubulin–labeled peripheral nerves in whole-mounted 24 hpf embryos. Only the ventral motor nerve was clearly visible at this stage, because the dorsal motor nerve was obscured by the underlying spinal cord and no axons had grown into the specific lateral pathway of the ROP axon at that stage. Only the rostral 12 pairs of motor nerves were scored, because all of these had grown beyond the ventral edge of the notochord into the ventral somite in uninjected embryos at 24 hpf. Trunk hemisegments were scored as abnormal when nerves were branched at or above the ventral edge of the notochord. This strategy is to exclude naturally occurring branching that is sometimes observed ventral to the notochord. We also scored trunk hemisegments as abnormal when nerves were truncated (i.e., did not grow beyond the horizontal myoseptum), more than one anti-tubulin–immunolabeled axon fascicle exited the spinal cord (multiple exits) and when tubulin-immunopositive cells were present in the ventral motor pathway. Embryos were scored as affected by the respective phenotypes when more than 1 of 24 nerves were aberrantly branched and at least 1 of 24 hemisegments showed multiple exits of nerves or anti-tubulin–positive cells in the ventral somite. Embryos were scored as severely affected by branching when more than two of the motor nerves were aberrant. For each treatment at least two and for most treatments three or more experiments were performed. Values for affected embryos are given as mean ± standard error of the mean (S.R.M.). Statistical analyses were done using Fisher's exact test.

Acknowledgements

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

We thank Dr. Diane Bovenkamp for critical reading, Linda Schuldt for excellent technical assistance, and Angelika Nest for fish care. T.B., C.G.B., and M.S. are supported by the Deutsche Forschungsgemeinschaft; C.G.B. is supported by the CHS Stiftung; and M.K. is supported by NIH NCI.

REFERENCES

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