In zebrafish embryos, each myotome is typically innervated by three primary motoneurons (PMNs): the caudal primary (CaP), middle primary (MiP) and rostral primary (RoP). PMN axons first exit the spinal cord through a single exit point located at the midpoint of the overlying somite, which is formed beneath the CaP cell body and is pioneered by the CaP axon. However, the placement of CaP cell bodies with respect to corresponding somites is poorly understood. Here, we determined the early events in CaP cell positioning using neuropilin 1a (nrp1a):gfp transgenic embryos in which CaPs were specifically labeled with GFP. CaP cell bodies first exhibit an irregular pattern in presence of newly formed corresponding somites and then migrate to achieve their proper positions by axonogenesis stages. CaPs are generated in excess compared with the number of somites, and two CaPs often overlap at the same position through this process. Next, we showed that CaP cell bodies remain in the initial irregular positions after knockdown of Neuropilin1a, a component of the class III semaphorin receptor. Irregular CaP position frequently results in aberrant double exit points of motor axons, and secondary motor axons form aberrant exit points following CaP axons. Its expression pattern suggests that sema3ab regulates the CaP position. Indeed, irregular CaP positions and exit points are induced by Sema3ab knockdown, whose ectopic expression can alter the position of CaP cell bodies. Results suggest that Semaphorin-Neuropilin signaling plays an important role in position fine-tuning of CaP cell bodies to ensure proper exit points of motor axons.

INTRODUCTION

In a developing nervous system, the final neuronal pattern is often accomplished by fine-tuning the initial `rough sketch' pattern. For example,initial connective patterns of some neurons are refined into an appropriate subset of targets in the final step of determining the specificity of axonal connections (Purves and Lichtman,1980; Nakamura and O'Leary,1989; Lichtman and Colman,2000; Kantor and Kolodkin,2003). Such a process can provide a mechanism that ensures appropriate and complete formation of a neuronal pattern. However, because the process of fine-tuning in the nervous system has been studied mainly with regard to axonal pattern formation, the question of whether such a mechanism is applicable to other neuronal patterns, such as iterative positioning of neuronal elements, remains unsolved. We have investigated how a regularly spaced pattern of zebrafish primary motoneuron cell bodies is accomplished in the spinal cord and examined whether a fine-tuning process is also involved in the formation of the iterative pattern.

In zebrafish embryos, each myotome is typically innervated by three identifiable primary motoneurons (PMNs): the caudal primary (CaP), middle primary (MiP) and rostral primary (RoP)(Eisen et al., 1986; Myers et al., 1986; Westerfield et al., 1986). Their axons first exit the spinal cord from a single exit point adjacent to the medial surface of the somite; the exit point lies in close proximity to the CaP cell body and is pioneered by the CaP growth cone. All PMN axons migrate ventrally on the medial surface of the dorsal somite until they reach the horizontal myoseptal region. At the end of the dorsal somite pathway, the growth cones encounter a group of specialized cells called muscle pioneers and then follow divergent pathways that extend to the ventral, dorsal and horizontal myoseptal muscles within the myotomes(Eisen et al., 1986; Beattie, 2000). CaP axons start migrating in the numerical order of the segments, and CaP cell bodies show an iterant regularly spaced pattern (Fig. 1F). Some spinal hemisegments have two CaP cells; one of which is referred to as the variably primary (VaP). The two cells are equivalent to each other morphologically and in gene expression, but one dies about 36 hours after fertilization (Eisen et al.,1990; Eisen and Melancon,2001). The presence of VaP is different from other PMNs that show a segmentally arranged pattern; the distribution of VaPs is different for each embryo and does not always show a bilaterally symmetrical pattern(Eisen et al., 1990). CaP and VaP cells are positioned in close proximity and their axons exit from the spinal cord at the same exit point. The axonal exit points from the spinal cord are also used by axons of secondary motoneurons (SMNs) that develop later and follow PMN axons (Westerfield et al.,1986).

The mechanism of zebrafish PMN axonal migration has been studied extensively (Eisen et al.,1989; Melancon et al.,1997; Beattie and Eisen,1997; Zeller and Granato,1999; Beattie et al.,2000; Zhang and Granato,2000; Rodino-Klapac and Beattie, 2004) and several molecules that function in axon guidance have also been identified. These include Netrin 1a(Lauderdale et al., 1997),Semaphorins (Roos et al.,1999; Halloran et al.,2000; Sato-Maeda et al.,2006), Tenascin C (Schweitzer et al., 2005), and LH3(Schneider and Granato, 2006). Some of these molecules are distributed differently within the somite, and it may be necessary for the PMN axon to decide axonal exit points at the appropriate positions. However, little is known about the mechanism responsible for determining the positions of exit points. Although the position of an exit point may depend on the CaP cell body, the mechanism responsible for proper positioning of CaP cell bodies corresponding to the somites is unclear.

Here, we used the sensitivity of GFP expression in neuropilin 1a(nrp1a):gfp transgenic fish(Sato-Maeda et al., 2006) to trace CaP cell bodies, and examined early events establishing the iterative pattern. The regularly spaced pattern of CaP cell bodies was not present initially, but was achieved gradually by the time of axonogenesis. VaP was formed by spatial fine-tuning of CaP cell bodies, present in excess relative to the somites. This position adjustment was disrupted by Nrp1a knockdown,which often resulted in aberrant axonal exit points. Correspondingly,irregular patterns of CaPs were observed in Sema3ab knockdown embryos, and CaP cell bodies showed repulsive reaction to ectopic Sema3ab. These results suggest that the Semaphorin-Neuropilin signal plays an important role in fine-tuning the CaP position, which ensures proper exit points of axons and harmonizes development of spinal motoneurons and segmented somites.

Fig. 1.

The pattern of CaP cell bodies before and after axonogenesis.(A) Optical sections at different focal levels of an nrp1a:gfptransgenic embryo at the 28-somite stage (23 hpf). CaPs are labeled with anti-GFP antibody. Position of CaP cell bodies was not related to the corresponding somites. Also, two distinct and discrete cell bodies were observed at the 25th somite level. Arrows (lower panel) indicate the somite boundaries. (B) Irregular patterns of CaP cell bodies were also observed as the pattern of isl2-expressing cells (stars) during the pre-axonogenesis period in wild-type embryos. Two separate CaP cells were observed at the 11th somite level. (C) Appearance ratio of spinal hemisegments with two discrete CaPs. The two discrete CaPs were only seen in caudal segments where CaPs did not begin axonogenesis. Position 0 is defined as the most caudal segment in which a CaP axon was formed. Thirty-three pairs of separate CaPs from 26 embryos at the 26- to 29-somite stages were sorted by relative somite levels and scored for spinal hemisegments with two separated CaPs. (D) A transgenic embryo at the 29-somite stage (23.5 hpf). Although CaP cell bodies with axons were ellipsoid (20th, 21st), more caudal CaP cells appeared triangular or trapezoid (23rd, 24th). (E) Side view of a transgenic embryo at the 27-somite stage (22.5 hpf). (F) CaP cell bodies with axons were regularly spaced in the middle of overlying somites. Stars and triangles show mAb Sv2-labeled CaP/VaP cell bodies. Blue lines denote somite borders. Unless noted otherwise, embryos are oriented as rostral to the left and dorsal up. Sc, spinal cord; Nc, notochord. The numbers in the panels indicate the segmental order. Scale bars: 50 mm.

Fig. 1.

The pattern of CaP cell bodies before and after axonogenesis.(A) Optical sections at different focal levels of an nrp1a:gfptransgenic embryo at the 28-somite stage (23 hpf). CaPs are labeled with anti-GFP antibody. Position of CaP cell bodies was not related to the corresponding somites. Also, two distinct and discrete cell bodies were observed at the 25th somite level. Arrows (lower panel) indicate the somite boundaries. (B) Irregular patterns of CaP cell bodies were also observed as the pattern of isl2-expressing cells (stars) during the pre-axonogenesis period in wild-type embryos. Two separate CaP cells were observed at the 11th somite level. (C) Appearance ratio of spinal hemisegments with two discrete CaPs. The two discrete CaPs were only seen in caudal segments where CaPs did not begin axonogenesis. Position 0 is defined as the most caudal segment in which a CaP axon was formed. Thirty-three pairs of separate CaPs from 26 embryos at the 26- to 29-somite stages were sorted by relative somite levels and scored for spinal hemisegments with two separated CaPs. (D) A transgenic embryo at the 29-somite stage (23.5 hpf). Although CaP cell bodies with axons were ellipsoid (20th, 21st), more caudal CaP cells appeared triangular or trapezoid (23rd, 24th). (E) Side view of a transgenic embryo at the 27-somite stage (22.5 hpf). (F) CaP cell bodies with axons were regularly spaced in the middle of overlying somites. Stars and triangles show mAb Sv2-labeled CaP/VaP cell bodies. Blue lines denote somite borders. Unless noted otherwise, embryos are oriented as rostral to the left and dorsal up. Sc, spinal cord; Nc, notochord. The numbers in the panels indicate the segmental order. Scale bars: 50 mm.

MATERIALS AND METHODS

Fish colony

Zebrafish (Danio rerio) were maintained in a laboratory breeding colony at 28.5°C on a 14 hours:10 hours light:dark cycle. Embryos collected from breeding fish were allowed to develop at 28.5°C and were staged as hours post fertilization (hpf) and by the number of somites (somite stage) (Kimmel et al.,1995).

RNA in situ hybridization and immunohistochemistry

Digoxigenin-labeled riboprobes were synthesized by in vitro transcription and hydrolyzed by limited alkaline hydrolysis(Cox et al., 1984). The procedure for whole-mount in situ hybridization was described by Schulte-Merker et al. (Schulte-Merker et al., 1992). Double labeling in situ hybridization was performed as described by Whitlock and Westerfield(Whitlock and Westerfield,2000). Whole-mount immunostaining was performed according to the procedures described previously (Shoji et al., 1998). The primary antibodies were used at the following dilutions: 1/50 for Sv2 (DSHB, University of Iowa), 1/50 for Znp1 (DSHB,University of Iowa), 1/10 for Zn5/Zn8 (DSHB, University of Iowa), 1/10 for anti-Myc mAb (Evan et al.,1985), and 1/400 for polyclonal anti-GFP antibody (Invitrogen). To label PMN axons at 48 hpf, we used mAb Znp1 instead of Sv2 because of its greater sensitivity. For immunostaining with in situ hybridization, fixed embryos were first processed for immunostaining followed by re-fixation and processing for in situ hybridization. Semi-thin sections were cut with a microslicer (Dosaka EM, Kyoto, Japan) after whole-mount hybridization or immunostaining according to the procedures describe previously (Sato-Meda et al., 2006).

Detection of CaP cell bodies

In addition to the specific molecular marker isl2 for CaPs(Appel et al., 1995; Tokumoto et al., 1995), we used a monoclonal antibody Sv2 and the nrp1a:gfp transgenic strain to identify CaP cell bodies. Sv2 was originally established as a monoclonal antibody that recognizes transmembrane glycoprotein in synaptic vesicles(Buckley and Kelly, 1985) and was reported to label all primary motor axons in zebrafish embryos(Panzer et al., 2005; Schneider and Granato, 2006). Instead, we found that Sv2 labeled the cell body and axon of CaPs during early axonogenesis in embryos earlier than 24 hpf (30-somite stage). In embryos at this stage, isl1-positive neurons in the ventral spinal cord that correspond to MiP and RoP showed no immunostaining with Sv2(Fig. 3F). Therefore, we determined the Sv2-labeled neurons to be CaPs, and not MiPs or RoPs, at this early stage of axonogenesis. For fluorescent live images, GFP-labeled cell bodies were examined with a Zeiss Axioskop upright microscope or with an Axiovert LSM5 Pascal confocal microscope.

CaP position irregularities were defined as follows. For pre-axonogenesis stages, CaPs situated on or by the borders of overlying somites at caudal five levels in 25- to 30-somite stages embryos. For stages after axonogenesis, CaPs located at the anterior or posterior marginal quarters(Fig. 3E) on the focus of the dorsal edge of the notochord.

Prediction of axonal extension according to segment and developmental stage

We determined the average status of CaP axonal development in segments at each stage (Sato-Maeda et al.,2006). This information enabled us to determine approximate segments in which the CaP axons begins axonogenesis at a given developmental stages (Table 1). Axons were detected using mAb Znp1.

Table 1.

Segment range where CaP axons begin axonogenesis at each somite stage

Somite stage 22 23 24 25 26 27 28 29 
Embryos examined 15 11 10 11 16 11 12 
Segment position* 14.4±0.88 15.2±0.56 15.8±0.98 17.1±0.88 18.6±1.03 19.4±0.81 19.7±1.10 22.3±0.65 
Segment range 13-16 14-16 14-17 16-18 17-20 18-21 18-21 21-23 
Somite stage 22 23 24 25 26 27 28 29 
Embryos examined 15 11 10 11 16 11 12 
Segment position* 14.4±0.88 15.2±0.56 15.8±0.98 17.1±0.88 18.6±1.03 19.4±0.81 19.7±1.10 22.3±0.65 
Segment range 13-16 14-16 14-17 16-18 17-20 18-21 18-21 21-23 
*

Average position (segment number) of the most caudal segment with a CaP axon, ± s.d.

Morpholino oligonucleotide injection

The antisense morpholinos (25-mer) were designed against DNA sequences for 5′-UTR or splicing donor sites (indicated as `a splice blocker'). The effectiveness of sema3aa MO, sema3ab MO1 and nrp1aMO1 was reported previously (Shoji et al.,1998; Lee et al.,2002; Torres-Vazquez et al.,2004). The control morpholino sequence had a five-base mismatch. The sequences were as follows:

sema3aa MO: 5′-CTTGTAGCCCACAGTGCCCAGAGCA-3′;

sema3aa MO control:5′-CTTCTAGCCGACAGAGCCCAGTGCA-3′;

sema3ab MO1 (a splice blocker):5′-AAATGTGTCTTACCGTTGAGCCATC-3′;

sema3ab MO1 control:5′-AAATCTGTGTTACGGTTCAGCGATC-3′;

sema3ab MO2: 5′-GTTCCGTATGCAGTCCCGTGGCCTC-3′;

sema3ab MO2 control:5′-GTTCGGTATCCACTCCCCTGGACTC-3′;

nrp1a MO1: 5′-GAATCCTGGAGTTCGGAGTGCGGAA-3′;

nrp1a MO1 control:5′-GAATGCTCGACTTCGGAGTCCGCAA-3′;

nrp1a MO2 (a splice blocker):5′-GCTCAACACTCACTTGCACTCTCGG-3′; and

nrp1a MO2 control:5′-GCTGAAGACTCAGTTGCAGTCTGGG-3′.

MOs were solubilized in 1× Danieau Solution(Nasevicius and Ekker, 2000)and were injected into recently fertilized eggs (approximately 6-9 ng for each embryo). To obtain 48 hpf Nrp1a knockdown embryos, diluted nrp1a MO1(2-3 ng/embryo) was injected.

DNA injection

Approximately 1 nl of a 50 ng/ml solution of hsp70:sema3ab-myc or hsp70:myc in water containing 0.1% Phenol Red was pressure injected from a micropipette into recently fertilized eggs as described previously(Sato-Maeda et al., 2006). To induce expression, embryos were incubated at 38°C for 30 minutes at the 22-somite stage. After heat induction, embryos were cultured until the 30-somite stage (24 hpf) and immunostained with Sv2 and anti-Myc or anti-GFP. Segments in which the construct was integrated into the floor plate cells and posterior to the 18th segment were examined.

RESULTS

CaP position is adjusted gradually before axon formation

The distribution of CaP cell bodies changes during development. The expression pattern of zebrafish isl2, a molecular marker for CaPs,did not correspond to the somite arrangement in the early spinal cord at the 8-somite stage when looking at the sixth to seventh somite level(Appel et al., 1995). By contrast, the CaP cell bodies at the axonogenesis stage are regularly spaced at the midportion of overlying somites(Eisen et al., 1986)(Fig. 1F, 17-20th segments at 29-somite stage; Fig. 3E,scheme). The difference suggests that the position of CaP cell bodies is regulated with the segmented somite. To examine when and how the position is established, we used nrp1a:gfp transgenic embryos as previously described (Sato-Maeda et al.,2006). In this transgenic strain, GFP is expressed by CaPs and VaPs according to endogenous nrp1a expression(Feldner et al., 2005; Sato-Maeda et al., 2006). Our series of live observations determined that CaPs and VaPs solely express GFP during the pre-axonogenesis stage in transgenic embryos (n=40). However, other motoneurons and interneurons in the ventral spinal cord begin expressing GFP after 24 hpf, which made longer tracing difficult(Sato-Maeda et al., 2006).

GFP-expressing CaPs were first detected when the corresponding somite was newly segmented. At younger somite levels, where the CaPs had not yet begun axonogenesis, the cell bodies were irregularly distributed(Fig. 1A and Table 1; 24-27th segments at 28-somite stage). Some cell bodies were situated in the anterior or posterior portion; others were situated on the boundary of the overlying somite. In addition, two CaP cell bodies were often observed within the region of a single somitic segment (cells in the 25th segment in Fig. 1A), which were also observed by isl2 expression (Fig. 1B: 11th segment at 17-somite stage). Severe displacement of CaP cell bodies on or by the border of the overlying somite was seen in 52% of the spinal segments (34 out of 65), in which 35% and 17% had one and two CaPs,respectively. The two CaPs discretely located under a somitic segment were seen only during the pre-axonogenesis period(Fig. 1C) and merged at the midportion after that period. This time course indicates that the CaP position is established by the time of axonogenesis.

CaP cell bodies altered their appearance dynamically(Fig. 1A,D,E). At first, they appeared flat, clinging to the floor plate (26th and 27th segments in Fig. 1A). In more rostral (i.e. older) levels, they were triangular or trapezoidal (23rd and 24th segments in Fig. 1D). When CaP started to form axons, they were all ellipsoidal (20th and 21st segments in Fig. 1D). Transition of the CaP cell shape was more clearly demonstrated in live observations using nrp1a:gfp transgenic embryos (Fig. 2). Here, we were able to identify GFP-labeled CaP cells that altered their appearance from flat to triangular(Fig. 2A star, Fig. 2B).

The irregular initial pattern of CaPs was adjusted to the regularly spaced pattern by active cellular movement. When we followed GFP-positive cells at the newly forming somite level, a flat-shaped CaP originating under the somite boundary migrated posteriorly to the midsection (arrowhead in Fig. 2B) and it eventually overlapped with another CaP. Then, the two equivalent neurons were distributed under the midportion of a corresponding somite, on being the CaP and the other the VaP. Similar movement was frequently seen in which two discrete cells overlapped 80 minutes later (arrowhead in Fig. 2A; the 21st segment at the 27-somite stage in the top panel). As mentioned above, these two separate CaP cells were only seen at younger (caudal) somite levels when CaP axons were not formed (Fig. 1C), which indicates that migration is completed before axonogenesis.

Our observations showed that the regularly spaced pattern of CaP cell bodies was achieved by cellular migration. They were distributed initially in an irregular pattern and then adjusted their position to correspond with the somite by the time of axonogenesis. This process appears to be responsible for the heterogeneity in the spinal segments; some have only a CaP, whereas the others have a CaP and a VaP (Fig. 1F).

Antisense knockdown of nrp1a results in abnormal CaP position even after axon formation

Our observations indicated that the position fine-tuning of the CaP cell body occurs prior to axonogenesis. To identify the molecules involved, we studied the effects of Nrp1a knockdown on the CaP position. nrp1a is expressed by CaPs when fine-tuning their position, and knockdown of Nrp1a has reportedly brought about various degrees of aberrant axonal phenotype(Feldner et al., 2005). To determine whether Nrp1a is required for CaP cell positioning, we injected antisense MOs against nrp1a into recently fertilized embryos and assayed the CaP cell position with isl2 in situ hybridization and with Sv2 immunostaining.

Fig. 2.

Live observations of CaP cell bodies in nrp1a:gfp transgenic embryos. (A) Trunk region of the same embryo shown at different stages. Two discrete cells underlying a single somite seen in the upper panel(the 21st somite level at the 27-somite stage: arrowhead) overlapped after 80 minutes, shown in the lower panel. A transition in cell shape between different stages was also observed (stars). The numbers in the panels indicate the segmental order. Scale bar: 50 μm. (B) Confocal micrographs of CaP cell bodies in another embryo. Sequential images were captured of the 16-18th segment, starting from the 17-somite stage (17.5 hpf). A CaP cell body originally located beneath the somite boundary (arrowhead) migrated posteriorly and overlapped with another CaP in the middle of the somite. The numbers denote the time elapsed. The lines indicate segment borders of the somites. Scale bar: 20 μm.

Fig. 2.

Live observations of CaP cell bodies in nrp1a:gfp transgenic embryos. (A) Trunk region of the same embryo shown at different stages. Two discrete cells underlying a single somite seen in the upper panel(the 21st somite level at the 27-somite stage: arrowhead) overlapped after 80 minutes, shown in the lower panel. A transition in cell shape between different stages was also observed (stars). The numbers in the panels indicate the segmental order. Scale bar: 50 μm. (B) Confocal micrographs of CaP cell bodies in another embryo. Sequential images were captured of the 16-18th segment, starting from the 17-somite stage (17.5 hpf). A CaP cell body originally located beneath the somite boundary (arrowhead) migrated posteriorly and overlapped with another CaP in the middle of the somite. The numbers denote the time elapsed. The lines indicate segment borders of the somites. Scale bar: 20 μm.

In Nrp1a knockdown embryos, CaP cell bodies initially arose in an irregular pattern similar to that in wild-type embryos (see Fig. S1 in the supplementary material). However, CaPs were out of normal alignment when the position became regularly spaced in intact and control MO-injected embryos(Fig. 3A-D). Significant numbers of CaPs were dislocated severely at the marginal quarters to the corresponding somite (Fig. 3E)compared with the control. Moreover, two separate CaP cell bodies did not merge (Fig. 3B,D), and their axons left the spinal cord independently. These two discrete exit points were the most notable phenotype in Nrp1a knockdown embryos(Fig. 3F,G; see Table 3), which was associated with the position fine-tuning defect.

Table 3.

Knockdown of Nrp1a and Sema3ab causes abnormal axonal exit points

MO injected (ng/embryo)Embryos examinedEmbryos with double exit points (%)
nrp1a MO1 (7) 49 15* (30.6) 
nrp1a MO1 control (7) 42 (0) 
nrp1a MO1(3) 30 (16.7) 
nrp1a MO2 (5) 45 10* (22.2) 
nrp1a MO2 control (5) 42 (2.4) 
sema3aa MO (9) 39 2 (5.1) 
sema3aa MO control (9) 39 (0) 
sema3ab MO1 (9) 41 6* (14.6) 
sema3ab MO1 control (9) 51 (2.0) 
sema3ab MO2 (3) 46 4* (8.7) 
sema3ab MO2 control (3) 53 (0) 
sema3aa MO (7) + sema3ab MO1 (7) 31 (16.1) 
nrp1a MO1 (7) + sema3ab MO1 (7) 33 12 (36.4) 
MO injected (ng/embryo)Embryos examinedEmbryos with double exit points (%)
nrp1a MO1 (7) 49 15* (30.6) 
nrp1a MO1 control (7) 42 (0) 
nrp1a MO1(3) 30 (16.7) 
nrp1a MO2 (5) 45 10* (22.2) 
nrp1a MO2 control (5) 42 (2.4) 
sema3aa MO (9) 39 2 (5.1) 
sema3aa MO control (9) 39 (0) 
sema3ab MO1 (9) 41 6* (14.6) 
sema3ab MO1 control (9) 51 (2.0) 
sema3ab MO2 (3) 46 4* (8.7) 
sema3ab MO2 control (3) 53 (0) 
sema3aa MO (7) + sema3ab MO1 (7) 31 (16.1) 
nrp1a MO1 (7) + sema3ab MO1 (7) 33 12 (36.4) 

Injected embryos were fixed at the 25-29 somite stage.

*

By Fisher's exact test of independence, morpholino oligonucleotide (MO)injection was considered effective at producing the double exit phenotype(P<0.05).

Effect of the MO injection was not significant. The possibility that the double exit phenotype and MO injection might have been independent events cannot be rejected at a 0.05 significance level.

Less MO was applied to the embryos shown in Fig. 6 to circumvent embryonic death by circulation defects so as to enable examination of secondary motoneuron axons.

Because of these defects in the positions of CaPs, we assumed that the aberrantly positioned cells would be subjected to different environments that may affect the subtype identity of the CaP. However, molecular marker analysis indicated that this is unlikely because both of the separated cells expressed isl2, not isl1, markers of MiP and RoP cells(Fig. 3F,G). In addition, isl2+ cell number in the ventral spinal cord throughout a given segmental level did not change between Nrp1a knockdown and the control embryos(Table 2). These results suggest that differentiation and subtype specification of CaP were unaffected in the Nrp1a-knockdown embryos. Furthermore, axons of the two separated CaPs,caused by Nrp1a knockdown, had features similar to those of a CaP and a VaP;only one CaP can extend its axon to the ventral myotome, and the other cell(i.e. VaP) ceases axonal extension at the horizontal myoseptal region(Eisen et al., 1990; Eisen and Melancon, 2001). As shown in Fig. 3G and Fig. 7A, only one axon migrated further onto the ventral myotome, and the other axon did not migrate beyond the horizontal myoseptal region in most cases (26 of 29 pairs of axons).

Table 2.

Number of islet 2(+) primary motoneurons in nrp1aembryos injected with morpholino oligonucleotides (MOs)

Somite stageSegments*MOEmbryos examinedMean ± s.d.
16-18 10 segments nrp1a 16 12.56±1.26 
 (4th-13th) Control 16 13.31±0.95 
20-23 10 segments nrp1a 17 13.24±1.03 
 (8th-17th) Control 19 13.84±0.83 
Somite stageSegments*MOEmbryos examinedMean ± s.d.
16-18 10 segments nrp1a 16 12.56±1.26 
 (4th-13th) Control 16 13.31±0.95 
20-23 10 segments nrp1a 17 13.24±1.03 
 (8th-17th) Control 19 13.84±0.83 
*

The number of islet 2(+) cells throughout 10 somitic segments was scored. Either side of the 4th-13th or 8th-17th segments were examined at the 16-18 or 20-23 somite stages, respectively.

No significant difference at a significance level of 0.05 (Student's t-test).

These results suggest that Nrp1a is required for position fine-tuning of CaP cell bodies, and that the process occurs after differentiation and subtype specification of the CaP. Defects in the fine-tuning result in aberrant exit points due to the separation of CaP and VaP cells.

Knockdown of Sema3ab also results in abnormal CaP position

We next investigated class III semaphorins, ligands of Nrp1a, which would function in the position fine-tuning of CaP cell bodies. Two copies of the zebrafish sema3a genes, sema3aa and sema3ab(previously called sema3a1 and sema3a2, respectively)(Yee et al., 1999; Roos et al., 1999), were potential candidates because they were expressed in myotomes. However, the level of sema3aa expression was much weaker compared with sema3ab during the CaP pre-axonogenesis period(Shoji et al., 1998; Yee et al., 1999; Bernhardt et al., 1998). First,we examined the effects of the knockdown of these genes by screening the double exit phenotype of CaP axons, which indicates displacement of CaP cell bodies. Injection of sema3aa MO caused few aberrant exit phenotypes(Table 3). These phenotypes were mild, with two cell bodies overlapped slightly (data not shown). By contrast, knockdown of Sema3ab frequently caused double exit phenotype of axons with irregular distribution of the CaP cell bodies(Fig. 4A,B; Table 3). Interestingly, the extent of the effect was less than that with Nrp1a knockdown, which may indicate that there would be more ligands involved with Sema3ab. However,Sema3aa is not likely to play this role with Sema3ab, because the defect with Sema3aa/Sema3ab double knockdown was not greater than that with the single Sema3ab knockdown. By contrast, another set of double knockdowns for Nrp1a and Sema3ab resulted in an added effect on the double exit phenotype compared with each single knockdown (Table 3), which is consistent with the idea that the products of these genes work together as a ligand-receptor complex. Alternatively, this additive effect would suggest Nrp1b, which is weakly expressed by CaPs (M.S.-M.,unpublished observation), may partially compensate for the CaP positioning. Thus, Semaphorin-Neuropilin signaling is considered necessary for the proper positioning of CaP cell bodies, and Sema3ab functions as a ligand of Nrp1a in this process.

The expression pattern of sema3ab in zebrafish was reported previously (Bernhardt et al.,1998; Roos et al.,1999; Shoji et al.,2003); its expression is detected from 12 hpf in homogeneous unsegmented mesoderm. As the somite develops to the segmented form, the striped expression pattern emerges as it localizes to the posterior half of each somite (Bernhardt et al.,1998). Here, we further examined the expression in relation to the positions of CaP cell bodies during the CaP pre-axonogenesis period. Expression of sema3ab was most noticeable in somites in which the corresponding CaP cell bodies were being adjusted (19th-21st; Fig. 4C). In these somites,CaPs were situated under the margin of the expressing region, which suggests that CaP cell bodies may determine their position when they detect a particular level of Sema3ab in their environment.

Fig. 3.

Knockdown of Nrp1a results in abnormal CaP cell positioning even after axon formation. Position of CaP cell bodies after axonogenesis was examined using isl2 expression (purple) and mAb Sv2 (brown).(A) Side view of a control embryo at the 28-somite stage. (B) A Nrp1a-knockdown embryo at the 29-somite stage with an irregular CaP pattern.(C,D) Horizontal sections of control and Nrp1a-knockdown embryos. CaPs are indicated by stars. (E; left) Severe dislocation was defined as the center of the CaP cell bodies being located in either the anterior or the posterior quarter of the overlying somite (indicated by `Mar'in the schema). (Right) Dislocation of CaP was significantly increased in Nrp1a-knockdown embryos, as determined by Fisher's exact test of independence(star, P<0.05).Nc, notochord. (F) Abnormal exit points of CaP axons by two separate CaPs in Nrp1a-knockdown embryos. CaP axons were labeled with mAb Sv2. (G) The positions of CaPs (stars) relative to MiP and RoP (diamonds, labeled by isl1) were unaffected by Nrp1a knockdown. (H) Different lengths of two axons from separate CaPs correspond to those of a CaP and a VaP. One axon extended onto the ventral myotome (*), whereas the other axon did not extend beyond the horizontal myoseptum (**). Arrowhead shows the level of the horizontal myoseptum. MiP axons extended normally along the dorsal pathway(arrows). The numbers in A, B and F-H denote the somitic segment order. The blue lines indicate segment borders. Scale bar: 50 μm.

Fig. 3.

Knockdown of Nrp1a results in abnormal CaP cell positioning even after axon formation. Position of CaP cell bodies after axonogenesis was examined using isl2 expression (purple) and mAb Sv2 (brown).(A) Side view of a control embryo at the 28-somite stage. (B) A Nrp1a-knockdown embryo at the 29-somite stage with an irregular CaP pattern.(C,D) Horizontal sections of control and Nrp1a-knockdown embryos. CaPs are indicated by stars. (E; left) Severe dislocation was defined as the center of the CaP cell bodies being located in either the anterior or the posterior quarter of the overlying somite (indicated by `Mar'in the schema). (Right) Dislocation of CaP was significantly increased in Nrp1a-knockdown embryos, as determined by Fisher's exact test of independence(star, P<0.05).Nc, notochord. (F) Abnormal exit points of CaP axons by two separate CaPs in Nrp1a-knockdown embryos. CaP axons were labeled with mAb Sv2. (G) The positions of CaPs (stars) relative to MiP and RoP (diamonds, labeled by isl1) were unaffected by Nrp1a knockdown. (H) Different lengths of two axons from separate CaPs correspond to those of a CaP and a VaP. One axon extended onto the ventral myotome (*), whereas the other axon did not extend beyond the horizontal myoseptum (**). Arrowhead shows the level of the horizontal myoseptum. MiP axons extended normally along the dorsal pathway(arrows). The numbers in A, B and F-H denote the somitic segment order. The blue lines indicate segment borders. Scale bar: 50 μm.

CaP cell bodies are repulsed by cells misexpressing Sema3ab

The nrp1a and sema3ab expression patterns and results of knockdown studies suggest that Sema3ab may repulse CaP cell bodies. Roos et al. (Roos et al., 1999)reported that the CaP axons stalled in embryos injected with sema3abmRNA, and the authors suggested that Sema3ab can repulse CaP axons. To determine whether Sema3ab can repel CaP cell bodies as well as their axons, we injected embryos with DNA constructs of hsp70:sema3ab-myc or hsp70:myc. In these embryos, a random mosaic of cells expressed exogenous Sema3ab after heat induction.

When CaP cell bodies encountered floor plate cells expressing focal ectopic Sema3ab, they shifted anteriorly or slightly dorsally away from the Sema3ab-expressing cells (nine out of nine CaPs were mislocated; Fig. 5A,B). In controls, CaP cell bodies were all at the normal position and overlapped with cells expressing the Myc epitope (four out of four CaPs; Fig. 5C,D). These results suggest that Sema3ab regulates the position of CaPs in a repulsive manner, to achieve position fine-tuning.

Fig. 4.

Knockdown of Sema3ab causes the double exit phenotype of CaP axons similar to Nrp1a knockdown. (A) The position of CaP cell bodies after axonogenesis was examined using isl2 expression. Separate isl2+ CaP cell bodies were detected in Sema3ab-knockdown embryos(11th segment of a 25-somite stage embryo). (B) The double exit phenotype of CaP axons was observed in a Sema3ab-knockdown embryo. CaP axons and cell bodies were labeled with mAb Sv2. (C) sema3ab was expressed in the posterior halves of somites, and CaP cell bodies (detected by isl2 expression) were observed around the margin of the sema3ab expressing region. The inset shows the 21st segment of the same embryo at the focal plane of the CaP cell bodies. In older segments in which CaP axonogenesis had begun, expression of sema3ab was downregulated (17th segment of a 25-somite stage embryo). The numbers indicate the segmental order, and the stars indicate the positions of the CaP cell bodies. Scale bar: 50 μm.

Fig. 4.

Knockdown of Sema3ab causes the double exit phenotype of CaP axons similar to Nrp1a knockdown. (A) The position of CaP cell bodies after axonogenesis was examined using isl2 expression. Separate isl2+ CaP cell bodies were detected in Sema3ab-knockdown embryos(11th segment of a 25-somite stage embryo). (B) The double exit phenotype of CaP axons was observed in a Sema3ab-knockdown embryo. CaP axons and cell bodies were labeled with mAb Sv2. (C) sema3ab was expressed in the posterior halves of somites, and CaP cell bodies (detected by isl2 expression) were observed around the margin of the sema3ab expressing region. The inset shows the 21st segment of the same embryo at the focal plane of the CaP cell bodies. In older segments in which CaP axonogenesis had begun, expression of sema3ab was downregulated (17th segment of a 25-somite stage embryo). The numbers indicate the segmental order, and the stars indicate the positions of the CaP cell bodies. Scale bar: 50 μm.

Fig. 5.

Abnormal CaP cell positioning induced by focal ectopic Sema3ab.(A,B) Optical sections at different focal levels of an embryo in which Sema3ab-Myc was transiently expressed. Ectopic Sema3ab was labeled with anti-Myc (black) and CaPs with mAb Sv2 (brown). In a case in which ectopic Sema3ab was expressed in floor plate cells (white double arrows in A), the CaP cell body was shifted anteriorly (arrowhead in B) away from the cells expressing Sema3ab. (C,D) Optical sections of a control embryo. Myc epitope expression (white double arrows in C) did not affect the location of the CaP cell body (arrowhead in D) and overlapped with it. Lines in B and D indicate somitic borders. Sc, spinal cord; FP, floor plate; Nc, notochord. Scale bar: 50 μm.

Fig. 5.

Abnormal CaP cell positioning induced by focal ectopic Sema3ab.(A,B) Optical sections at different focal levels of an embryo in which Sema3ab-Myc was transiently expressed. Ectopic Sema3ab was labeled with anti-Myc (black) and CaPs with mAb Sv2 (brown). In a case in which ectopic Sema3ab was expressed in floor plate cells (white double arrows in A), the CaP cell body was shifted anteriorly (arrowhead in B) away from the cells expressing Sema3ab. (C,D) Optical sections of a control embryo. Myc epitope expression (white double arrows in C) did not affect the location of the CaP cell body (arrowhead in D) and overlapped with it. Lines in B and D indicate somitic borders. Sc, spinal cord; FP, floor plate; Nc, notochord. Scale bar: 50 μm.

Irregular CaP position causes aberrant exit points of SMN axons

We examined whether abnormal CaP exit points would affect following axons. In the zebrafish spinal cord, axons of SMNs recognized by monoclonal antibody Zn5 (black axons in Fig. 6A)follow the pathway of PMN axons (brown axons in Fig. 6A) and use a common and narrow exit region in the midportion. Pike et al.(Pike et al., 1992) reported that SMN ventral axons extended slowly and form aberrant branches after ablation of CaP, but were still able to reach ventral myotomes. This study suggested that CaP axons facilitated SMN axonal development, but it is still unclear how exit points of SMN axons are determined. Here, we examined SMN axons in Nrp1a-knockdown embryos in which double exit points of PMN were induced.

In Nrp1a-knockdown embryos, two separate SMN exit points adjacent to a single somite were observed (four segments in two out of 21 embryos; Fig. 6B,C). By contrast, we did not observe aberrant SMN exit points in relation to the PMN axon in the control embryos (19 embryos). The defects in SMN axons were relatively moderate and less frequent, as correlated with the lesser amount of MO applied in the experiment, but were consistent with the frequency of CaP axon defects induced by the same dose (Table 3). Although nrp1a MO induced strong double exit phenotype and resulted in embryonic death by 48 hpf, probably because of circulation defects (Lee et al.,2002), it was necessary to reduce the amount of MO injected. Even at this less effective dose, these double exit points induced the formation of two bundles of SMN axons; one bundle was thicker and clearly lay alongside the PMN axons (Fig. 6B,C,arrowheads). The other bundle, some located anteriorly and others posteriorly,was thinner and sometimes loose (Fig. 6B,C, arrows). One of the two bundles did not associate with the PMN axon, which probably mirrors VaP axons that would degenerate at the stage examined. We further examined the correlation between CaP and SMN defects by tracking the double exit points of CaP using nrp:gfp transgenic embryos with Nrp1a knockdown. When we found the double exit points of CaP axons at 24-25 hpf, the double exit phenotype in SMN axons appeared at 48 hpf in most cases (eight out of 11 cases: Fig. 7). In several remnant cases, separated CaPs formed double exit points once; however, two axons merged near the exit points after sometime,resulting in a SMN abnormality, which was difficult to observe.

These results indicate that SMN exit points are dependent on the PMN exit points. Thus, the position fine-tuning of CaP is also important for SMN axon development. If the fine-tuning is not correct, subsequent SMN axons behave abnormally, thereby disrupting the coupled segmental architecture between the spinal cord and segmented somites.

DISCUSSION

Spatial and numerical fine-tuning of CaP cell bodies

In the present study, we investigated how CaP cell bodies form a regularly spaced iterative pattern in the zebrafish spinal cord. Our results showed that the cell bodies are distributed irregularly at first and do not correspond precisely to newly forming somitic segments(Fig. 8A,B). After the neighboring paraxial mesoderm is segmented, the CaP cell body migrates anteriorly or posteriorly to adjust its position corresponding to the middle area of the overlying somite (Fig. 8C, left). This fine-tuning of the CaP position ensures regularly spaced axonal exit points beneath the cell bodies, which are later followed by SMNs. This `spatial fine-tuning' is required to establish segmental architecture of the spinal motor nerve and somite, in which each segmental nerve bundle passes through iterative vertebrae and innervates the same level of myotomes. In cases where the CaP position was not properly adjusted(Fig. 8C, right), the common exit points were abnormally localized (Fig. 8D, right). Although the reason why CaP axons, but not secondary motor axons, exit the spinal cord beneath their cell bodies remains unclear,our results indicate that the CaP position determines the common exit points. The spatial fine-tuning of CaP cell bodies is thus important for harmonizing the spinal cord and somite development.

Fig. 6.

Aberrant exit points of secondary motoneuron (SMN) axons in Nrp1a-knockdown embryos. (A) Axons of primary motoneurons (PMNs)and SMNs at 48 hpf in a control MO-injected embryo. PMN axons were labeled with a brown stain, recognized by mAb Znp1, but not by Zn5(Zeller et al., 2002). SMN axons were stained in blue-black, recognized by Zn5(Fashena and Westerfield,1999), regardless of Znp1 reactivity. SMN axons extended ventrally along with PMN axons using an exit point common to PMN and SMN. (B)Double exit phenotype of SMN axons was observed in Nrp1a-knockdown embryos at 48 hpf. Generally, one axonal bundle of SMNs was thicker (arrowhead) than the others (arrow), regardless of its anterior or posterior position. (C)Another example of a double exit phenotype of SMN. Scale bar: 50 μm.

Fig. 6.

Aberrant exit points of secondary motoneuron (SMN) axons in Nrp1a-knockdown embryos. (A) Axons of primary motoneurons (PMNs)and SMNs at 48 hpf in a control MO-injected embryo. PMN axons were labeled with a brown stain, recognized by mAb Znp1, but not by Zn5(Zeller et al., 2002). SMN axons were stained in blue-black, recognized by Zn5(Fashena and Westerfield,1999), regardless of Znp1 reactivity. SMN axons extended ventrally along with PMN axons using an exit point common to PMN and SMN. (B)Double exit phenotype of SMN axons was observed in Nrp1a-knockdown embryos at 48 hpf. Generally, one axonal bundle of SMNs was thicker (arrowhead) than the others (arrow), regardless of its anterior or posterior position. (C)Another example of a double exit phenotype of SMN. Scale bar: 50 μm.

Fig. 7.

Abnormal CaP cell positioning results in double exit points in both primary motoneurons (PMNs) and secondary motoneurons (SMNs). (A)Sequential images at the 18th somite level in an nrp1a:gfp transgenic embryo with Nrp1a-knockdown. Two separated CaP cell bodies (21 hpf) extended their axons to form double exit points (24 hpf). At 27 hpf, one axon extended beyond the horizontal myoseptal level (*), but the other axon did not (**). Arrowheads indicate the level of the horizontal myoseptum. (B) The double exit phenotype of SMN axons was observed at the same position as in A at 48 hpf. SMN axons were immunostained with mAb Zn5. Scale bars: 20 μm.

Fig. 7.

Abnormal CaP cell positioning results in double exit points in both primary motoneurons (PMNs) and secondary motoneurons (SMNs). (A)Sequential images at the 18th somite level in an nrp1a:gfp transgenic embryo with Nrp1a-knockdown. Two separated CaP cell bodies (21 hpf) extended their axons to form double exit points (24 hpf). At 27 hpf, one axon extended beyond the horizontal myoseptal level (*), but the other axon did not (**). Arrowheads indicate the level of the horizontal myoseptum. (B) The double exit phenotype of SMN axons was observed at the same position as in A at 48 hpf. SMN axons were immunostained with mAb Zn5. Scale bars: 20 μm.

Along with the irregular position, the initial cell number of CaPs exceeds the number of somites (Fig. 8B,C, left; Table 2), which is necessary to provide at least one cell under each somitic segment during the tuning process. The excessive CaPs also result in two CaPs along corresponding somitic segments, although they do not disrupt the one exit point per somitic segment rule because the two CaPs overlap at their proper position through spatial fine-tuning(Fig. 8D, left). Subsequently,only one CaP survives and the other is eliminated as a VaP(Beattie et al., 2000; Eisen and Melancon, 2001), and this `numerical fine-tuning' of CaPs can be achieved without affecting either the exit points or other segmental structure as spatial fine-tuning is accomplished successfully (Fig. 8E, left). Neural systems often adopt an `excess innervations and subsequent elimination' strategy, i.e. an excessive number of axons are allowed to connect to a limited number of targets, but eliminated later by activity and/or trophic support-dependent processes(Oppenheim, 1991; Balice-Gordon and Lichtman,1994; Katz and Shatz,1996; Pettmann and Henderson,1998). Excessive CaPs and elimination as VaPs in zebrafish spinal cord can be regarded as a similar type of numerical fine-tuning. Here, spatial fine-tuning is an important event for maintaining coupled development between spinal motoneurons and their target somitic myotomes within the numerical fine-tuning process.

Fine-tuning and subtype specification of CaP

Although the results of this study indicate the position fine-tuning of CaPs is in accordance with the segmented somites, several lines of evidence suggest that the paraxial mesoderm also provides cues to specify the differentiation of primary motoneurons to CaP, MiP or RoP in each spinal segment (Eisen, 1991; Lewis and Eisen, 2004). This raises the possibility that the two processes, i.e. subtype specification and position fine-tuning, may be interdependent. The extreme version of this hypothesis has the CaP identity being established by the cell's position within the somitic segment. For example, only cells that migrate and are located at the midportion of the somite would become CaPs, whereas those that did not reach this position would differentiate into other types of neurons. However, this seems unlikely because our real-time observations showed that nrp1a:gfp-positive neurons that will become CaPs are already present at the newest somite level in an irregular pattern(Fig. 1A, Fig. 2B) and that their migration defect does not change molecular marker expression or ventral projecting feature of the CaP feature (Fig. 3F-H). The above mentioned evidence indicates that the subtype identity of the CaP is determined before the somite establishes morphological boundaries, and the position fine-tuning occurs subsequent to cell-type specification. The results of other studies support this sequence. Eisen(Eisen, 1991) reported that CaP cell bodies transplanted to the MiP positions moved back to the original positions. In addition, Lewis and Eisen(Lewis and Eisen, 2004)reported that several zebrafish mutants lack morphological somitic segments but retain early cryptic segmentation as revealed by her1 and cs131 expression in the presomitic mesoderm. In these mutants,primary motoneurons are specified as CaPs or MiPs, but the precise spacing is disturbed. Further studies are needed to understand the subtype specification of PMNs; however, here we concluded that the specification as a CaP and its fine-tuning occur as sequential and separate processes.

Fig. 8.

Schematic summaries for fine-tuning of CaP cell positioning to correspond to overlying somitic segments. The differentiation and subtype specification of CaPs (A), somite segmentation (B), spatial fine-tuning (C), axonogenesis (D), and numerical fine-tuning(E). CaP cell bodies initially do not correspond precisely in position to newly segmented somites (A,B). In wild type, by the time of axonogenesis,CaP cells are adjusted to the proper positions in relation to somitic segments(C, left). When two CaPs are overlaid by a single somite, they overlap their position with each other, thereby maintaining one axonal exit point per adjacent somite (D, left). Later, one of the two CaPs is eliminated as VaP and each ventral myotome comes to be innervated by only one CaP (E, left upper diagram). If the spatial fine-tuning is successful, secondary motor axons born later (purple, lower diagram) follow the single exit point in each segment. However, after knockdown of Nrp1a or Sema3ab, CaP cell bodies remain in the initial irregular positions (C, right). Abnormal CaP cell positioning brings about the double exit phenotype of CaP axons (D, right) and these are followed by secondary motor axons (purple; E, right lower).

Fig. 8.

Schematic summaries for fine-tuning of CaP cell positioning to correspond to overlying somitic segments. The differentiation and subtype specification of CaPs (A), somite segmentation (B), spatial fine-tuning (C), axonogenesis (D), and numerical fine-tuning(E). CaP cell bodies initially do not correspond precisely in position to newly segmented somites (A,B). In wild type, by the time of axonogenesis,CaP cells are adjusted to the proper positions in relation to somitic segments(C, left). When two CaPs are overlaid by a single somite, they overlap their position with each other, thereby maintaining one axonal exit point per adjacent somite (D, left). Later, one of the two CaPs is eliminated as VaP and each ventral myotome comes to be innervated by only one CaP (E, left upper diagram). If the spatial fine-tuning is successful, secondary motor axons born later (purple, lower diagram) follow the single exit point in each segment. However, after knockdown of Nrp1a or Sema3ab, CaP cell bodies remain in the initial irregular positions (C, right). Abnormal CaP cell positioning brings about the double exit phenotype of CaP axons (D, right) and these are followed by secondary motor axons (purple; E, right lower).

Semaphorin-Neuropilin signaling plays an important role in position fine-tuning

The fine-tuning process is regulated by the Semaphorin-Neuropilin signal that functions in various types of cell migration and neural growth cone guidance (Kolodkin, 1998; Raper, 2000; Kruger et al., 2005). sema3ab, a secreted class III semaphorin gene, is expressed by newly forming somites (Roos et al.,1999; Shoji et al.,2003), whereas nrp1a, which encodes a receptor subunit for Sema3A (Kolodkin et al.,1997; He and Tessier-Lavigne,1997) is expressed specifically by CaPs(Sato-Maeda et al., 2006). Knockdown of each gene resulted in similar position defects in CaP cell bodies, consistent with the hypothesis that the products of these genes work together as a ligand-receptor complex to determine the CaP position. In analogy with the repulsive action of Sema3A on growth cone guidance(Luo et al., 1993), we hypothesized that Sema3ab affects CaP migration in a repulsive manner. In fact, CaP position was shifted by the presence of nearby ectopic Sema3ab-expressing cells (Fig. 5). An endogenous sema3ab expression pattern, at the posterior region of segmented somites(Roos et al., 1999; Shoji et al., 2003)(Fig. 4C), provides partial support for this idea. In cases where a CaP faces the posterior of the segmented somite, it would migrate anteriorly by detecting the Sema3ab level in the environment. In this model, however, the position fine-tuning cannot be fully explained in the case in which the CaP faces the anterior of the segment. One potential explanation for this is that an additional repulsive factor, such as another semaphorin, may be expressed in the anterior region of the somite and play a role. Interestingly, the frequency of abnormal CaP due to Sema3ab knockdown was about half that caused by Nrp1a knockdown. This result indicates that additional molecules may be involved as ligands for Nrp1a. Of the nine members of the class III semaphorins identified in zebrafish (Halloran et al.,1998; Roos et al.,1999; Yee et al.,1999; Yu et al.,2004; Stevens and Halloran,2005), sema3h is expressed at the anterior region of the segmented somite (Halloran et al.,1998; Stevens and Halloran,2005) and is thus a good candidate to support our model. However,the potential relationship between this novel semaphorin and CaP migration should be examined in future studies.

Semaphorin-Neuropilin signal functions in cell migration and axon guidance of CaP

The results presented herein along with those of our previous study(Sato-Maeda et al., 2006)suggest that two copies of the zebrafish sema3a gene function in position fine-tuning and in pathfinding of CaP axons. First, sema3abexpressed in newly forming somites regulates the position fine-tuning described above; second, sema3aa expressed, in turn, in a different pattern controls the navigation of axon pathfinding(Shoji et al., 1998; Shoji et al., 2003; Sato-Maeda et al., 2006). Along with this transition, somitic expression of the sema3a genes changes from the posterior region by sema3ab to the dorsal and ventral regions, but not in between by sema3aa(Shoji et al., 1998; Shoji et al., 2003; Sato-Maeda et al., 2006). This subfunctionalization by two homologous genes(Lynch and Force, 2000) is also supported by our knockdown studies. In Sema3ab-knockdown embryos, the CaP position is abnormal, whereas the axon pathfinding appears normal(Fig. 4B). By contrast, Sema3aa knockdown barely affects the CaP position(Table 3), whereas the axon behaves abnormally (Sato-Maeda et al.,2006). During position fine-tuning, Sema3ab regulates migration of the cell body. However, after axonogenesis, Sems3aa does not regulate the cell body, but regulates migration of growth cones and axons. Thus, two sema3a genes regulate the fine-tuning and axon pathfinding processes sequentially and separately. It remains unclear how these different events are finely regulated in individual cells. Because the cell morphology is quite different - flat during cell migration, changing to triangular, trapezoidal, and finally to ellipsoidal during axon formation - we speculate that subcellular domains that respond to semaphorins may be switched between these two processes. Alternatively, the cell body may be anchored by its adhesive nature after position tuning to maintain the cell arrangement inside the spinal cord. The regulatory mechanism controlling which portion of neuronal cells are motile and which immotile is a profound issue, and CaP development would be a good model to investigate such questions in future studies.

In conclusion, a stepwise fine-tuning process accomplishes the regularly spaced pattern of CaP cell bodies and its relationship to somitic segments. After subtype specification, the initial `rough sketch' of the CaP cell pattern is adjusted to a `fine pattern', which ensures the proper axonal exit point and harmonizes the spinal cord and somite development. The Semaphorin-Neuropilin signal plays an important role in this process, although at present it can only be partially explained. Further studies on other semaphorins as well as other molecules are required to understand the molecular mechanism underlying this process.

Acknowledgements

We thank Drs Halloran and Kuwada for helpful discussions and comments, and M. Ajiro, H. Tawarayama and L. Li for their technical expertise. This work was funded by Grant-in Aid for Scientific Research and Dynamics of Extracellular Environments to W.S. National BioResource Project, Japan, provided fish lines. Developmental Studies Hybridoma Bank, University of Iowa, provided antibodies.

References

Appel, B., Korzh, V., Glasgow, E., Thor, S., Edlund, T., Dawid,I. B. and Eisen, J. S. (
1995
). Motoneuron fate specification revealed by patterned LIM homeobox gene expression in embryonic zebrafish.
Development
121
,
4117
-4125.
Balice-Gordon, R. J. and Lichtman, J. W.(
1994
). Long-term synapse loss induced by focal blockade of postsynaptic receptors.
Nature
372
,
519
-524.
Beattie, C. E. (
2000
). Control of motor axon guidance in the zebrafish embryo.
Brain Res. Bull
.
53
,
489
-500.
Beattie, C. E. and Eisen, J. S. (
1997
). Notochord alters the permissiveness of myotome for pathfinding by an identified motoneuron in embryonic zebrafish.
Development
124
,
713
-720.
Beattie, C. E., Melancon, E. and Eisen, J. S.(
2000
). Mutations in the stumpy gene reveal intermediate targets for zebrafish motor axons.
Development
127
,
2653
-2662.
Bernhardt, R. R., Goerlinger, S., Roos, M. and Schachner, M.(
1998
). Anterior-posterior subdivision of the somite in embryonic zebrafish: implications for motor axon guidance.
Dev. Dyn
.
213
,
334
-347.
Buckley, K. and Kelly, R. B. (
1985
). Identification of a transmembrane glycoprotein specific for secretory vesicles of neural and endocrine cells.
J. Cell Biol.
100
,
1284
-1294.
Cox, K. H., Deleon, D. V., Angerer, L. M. and Angerer, R. C.(
1984
). Detection of mRNAs in sea urchin embryos by in situ hybridization using asymmetric RNA probes.
Dev. Biol
.
101
,
485
-502.
Eisen, J. S. (
1991
). Determination of primary motoneuron identity in developing zebrafish embryos.
Science
252
,
569
-572.
Eisen, J. S. and Melancon, E. (
2001
). Interactions with identified muscle cells break motoneuron equivalence in embryonic zebrafish.
Nat. Neurosci.
4
,
1065
-1070.
Eisen, J. S., Myers, P. Z. and Westerfield, M.(
1986
). Pathway selection by growth cones of identified motoneurones in live zebra fish embryos.
Nature
320
,
269
-271.
Eisen, J. S., Pike, S. H. and Debu, B. (
1989
). The growth cones of identified motoneurons in embryonic zebrafish select appropriate pathways in the absence of specific cellular interactions.
Neuron
2
,
1097
-1104.
Eisen, J. S., Pike, S. H. and Romancier, B.(
1990
). An identified motoneurons with variable fates in embryonic zebrafish.
J. Neurosci
.
10
,
34
-43.
Evan, G. I., Lewis, G. K., Ramsay, G. and Bishop, J. M.(
1985
). Isolation of monoclonal antibodies specific for human c-myc proto-oncogene product.
Mol. Cell. Biol
.
5
,
3610
-3616.
Fashena, D. and Westerfield, M. (
1999
). Secondary motoneuron axons localize DM-GRASP on their fasciculated segments.
J. Comp. Neurol
.
406
,
415
-424.
Feldner, J., Becker, T., Goishi, K., Schweitzer, J., Lee, P.,Schachner, M., Klagsbrun, M. and Becker, C. G. (
2005
). Neuropilin-1a is involved in trunk motor axon outgrowth in embryonic zebrafish.
Dev. Dyn
.
234
,
535
-549.
Halloran, M. C., Severance, S. M., Yee, C. S., Gemza, D. L. and Kuwada, J. Y. (
1998
). Molecular colning and expression of two novel zebrafish semaphorins.
Mech. Dev
.
76
,
165
-168.
Halloran, M. C., Sato-Maeda, M., Warren, J. T., Su, F., Lele,Z., Krone, P. H., Kuwada, J. Y. and Shoji, W. (
2000
). Laser-induced gene expression in specific cells of transgenic zebrafish.
Development
127
,
1953
-1960.
He, Z. and Tessier-Lavigne, M. (
1997
). Neuropilin is a receptor for the axonal chemorepellent Semaphorin III.
Cell
90
,
739
-751.
Kantor, D. B. and Kolodkin, A. L. (
2003
). Curbing the excesses of youth: molecular insights into axonal pruning.
Neuron
38
,
849
-852.
Katz, L. C. and Shatz, C. J. (
1996
). Synaptic activity and the construction of cortical circuits.
Science
274
,
1133
-1138.
Kimmel, C. B., Ballard, W. W., Kimmel, S. R., Ullmann, B. and Schilling, T. F. (
1995
). Stages of embryonic development of the zebrafish.
Dev. Dyn.
203
,
253
-310.
Kolodkin, A. L. (
1998
). Semaphorin-mediated neuronal growth cone guidance.
Prog. Brain Res
.
117
,
115
-132.
Kolodkin, A. L., Levengood, D. V., Rowe, E. G., Tai, Y. T.,Giger, R. J. and Ginty, D. D. (
1997
). Neuropilin is a semaphorin III receptor.
Cell
90
,
753
-762.
Kruger, R. P., Aurandt, J. and Guan, K.-L.(
2005
). Semaphorins command cells to move.
Nat. Rev. Mol. Cell Biol
.
6
,
789
-800.
Lauderdale, J. D., Davis, N. M. and Kuwada, J. Y.(
1997
). Axon tracts correlate with netrin-1a expression in the zebrafish embryo.
Mol. Cell. Neurosci
.
9
,
293
-313.
Lee, P., Goishi, K., Davidson, A. J., Mannix, R., Zon, L. I. and Klagsbrun, M. (
2002
). Neuropilin-1 is required for vascular development and is a mediator of VEGF-dependent angiogenesis in zebrafish.
Proc. Natl. Acad. Sci. USA
99
,
10470
-10475.
Lewis, K. E. and Eisen, J. S. (
2004
). Paraxial mesoderm specifies zebrafish primary motoneuron subtype identity.
Development
131
,
891
-902.
Lichtman, J. W. and Colman, H. (
2000
). Synapse elimination and indelible memory.
Neuron
25
,
269
-278.
Luo, Y., Raible, D. and Raper, J. A. (
1993
). Collapsin: a protein in brain that induces the collapse and paralysis of neuronal growth cones.
Cell
75
,
217
-227.
Lynch, M. and Force, A. (
2000
). The probability of preservation of a duplicate gene by subfunctionalization.
Genetics
154
,
459
-473.
Melancon, E., Liu, D. W., Westerfield, M. and Eisen, J. S.(
1997
). Pathfinding by identified zebrafish motoneurons in the absence of muscle pioneers.
J. Neurosci.
17
,
7796
-7804.
Myers, P. Z., Eisen, J. S. and Westerfield, M.(
1986
). Development and axonal outgrowth of identified motoneurons in the zebrafish.
J. Neurosci.
6
,
2278
-2289.
Nakamura, H. and O'Leary, D. D. M. (
1989
). Inaccuracies in initial growth and arborization of chick retinotectal axons followed by couorse corrections and axon remodeling to develop topographic order.
J. Neurosci
.
9
,
3776
-3795.
Nasevicius, A. and Ekker, S. C. (
2000
). Effective targeted gene `knockdown' in zebrafish.
Nat. Genet.
26
,
216
-220.
Oppenheim, R. W. (
1991
). Cell death during development of the nervous system.
Annu. Rev. Neurosci
.
14
,
453
-501.
Panzer, J. A., Gibbs, S. M., Dosch, R., Wagner, D., Mullins, M. C., Granato, M. and Balice-Gordon, R. J. (
2005
). Neuromuscular synaptogenesis in wild-type and mutant zebrafish.
Dev. Biol
.
285
,
340
-357.
Pettmann, B. and Henderson, C. E. (
1998
). Neuronal cell death.
Neuron
20
,
633
-647.
Pike, S. H., Melancon, E. F. and Eisen, J. S.(
1992
). Pathfinding by zebrafish motoneurons in the absence of normal pioneer axons.
Development
114
,
825
-831.
Purves, D. and Lichtman, J. W. (
1980
). Elimination of synapses in the developing nervous system.
Science
210
,
153
-157.
Raper, J. A. (
2000
). Semaphorins and their receptors in vertebrates and invertebrates.
Curr. Opin. Neurobiol
.
10
,
88
-94.
Rodino-Klapac, L. R. and Beattie, C. E. (
2004
). Zebrafish topped is required for ventral motor axon guidance.
Dev. Biol.
273
,
308
-320.
Roos, M., Schachner, M. and Bernhardt, R. R.(
1999
). Zebrafish semaphorin Z1b inhibits growing motor axons in vivo.
Mech. Dev.
87
,
103
-117.
Sato-Maeda, M., Tawarayama, H., Obinata, M., Kuwada, J. Y. and Shoji, W. (
2006
). Sema3a1 guides spinal motor axons in a cell- and stage-specific manner in zebrafish.
Development
133
,
937
-947.
Schneider, V. A. and Granato, M. (
2006
). The myotomal diwanka (lh3) glycosyltransferase and type XVIII collagen are critical for motor growth cone migration.
Neuron
50
,
683
-695.
Schulte-Merker, S., Ho, R. K., Herrmann, B. G. and Nusslein-Volhard, C. (
1992
). The protein product of the zebrafish homologue of the mouse T gene is expressed in nuclei of the germ ring and the notochord of the early embryo.
Development
116
,
1021
-1032.
Schweitzer, J., Becker, T., Lefebvre, J., Granato, M.,Schachner, M. and Becker, C. G. (
2005
). Tenascin-C is involved in motor axon outgrowth in the trunk of developing zebrafish.
Dev. Dyn
.
234
,
550
-566.
Shoji, W., Yee, C. S. and Kuwada, J. Y. (
1998
). Zebrafish semaphorin Z1a collapses specific growth cones and alters their pathway in vivo.
Development
125
,
1275
-1283.
Shoji, W., Isogai, S., Sato-Maeda, M., Obinata, M. and Kuwada,J. Y. (
2003
). Semaphorin3a1 regulates angioblast migration and vascular development in zebrafish embryos.
Development
130
,
3227
-3236.
Stevens, C. B. and Halloran, M. C. (
2005
). Developmental expression of sema3G, a novel zebrafish semaphorin.
Gene Expr. Patterns
5
,
647
-653.
Tokumoto, M., Gong, Z., Tsubokawa, T., Hew, C. L., Uyemura, K.,Hotta, Y. and Okamoto, H. (
1995
). Molecular heterogeneity among primary motoneurons and within myotomes revealed by the differential mRNA expression of novel islet-1 homologs in embryonic zebrafish.
Dev. Biol
.
171
,
578
-589.
Torres-Vazquez, J., Gitler, A. D., Fraser, S. D., Berk, J. D.,Pham, V. N., Fishman, M. C., Childs, S., Epstein, J. A. and Weinstein, B. M. (
2004
). Semaphorin-plexin signaling guides patterning of the developing vasculature.
Dev. Cell
7
,
117
-123.
Westerfield, M., McMurray, J. V. and Eisen, J. S.(
1986
). Identified motoneurons and their innervation of axial muscles in the zebrafish.
J. Neurosci.
6
,
2267
-2277.
Whitlock, K. E. and Westerfield, M. (
2000
). The olfactory placodes of the zebrafish form by convergence of cellular fields at the edge of the neural plate.
Development
127
,
3645
-3653.
Yee, C. S., Chandrasekhar, A., Halloran, M. C., Shoji, W.,Warren, J. T. and Kuwada, J. Y. (
1999
). Molecular cloning,expression, and activity of zebrafish semaphorin Z1a.
Brain Res. Bull
.
48
,
581
-593.
Yu, H. H., Houart, C. and Moens, C. B. (
2004
). Cloning and embryonic expression of zebrafish neuropilin genes.
Gene Expr. Patterns
4
,
371
-378.
Zeller, J. and Granato, M. (
1999
). The zebrafish diwanka gene controls an early step of motor growth cone migration.
Development
126
,
3461
-3472.
Zeller, J., Schneider, V., Malayaman, S., Higashijima, S.,Okamoto, H., Gui, J., Lin, S. and Granato, M. (
2002
). Migration of zebrafish spinal motor nerves into the periphery requires multiple myotome-derived cues.
Dev. Biol.
252
,
241
-256.
Zhang, J. and Granato, M. (
2000
). The zebrafish unplugged gene controls motor axon pathway selection.
Development
127
,
2099
-2111.