Olfactomedin-2 mediates development of the anterior central nervous system and head structures in zebrafish

Abstract

Olfactomedins comprise a diverse family of secreted glycoproteins, which includes noelin, tiarin, pancortin and gliomedin, implicated in development of the nervous system, and the glaucoma-associated protein myocilin. Here we show in zebrafish that olfactomedin-2 (OM2) is a developmentally regulated gene, and that knockdown of protein expression by morpholino antisense oligonucleotides leads to perturbations of nervous system development. Interference with OM2 expression results in impaired development of branchiomotor neurons, specific disruption of the late phase branchiomotor axon guidance, and affects development of the caudal pharyngeal arches, olfactory pits, eyes and optic tectum. Effects of OM2 knockdown on eye development are likely associated with Pax6 signaling in developing eyes, as Pax6.1 and Pax6.2 mRNA expression patterns are altered in the eyes of OM2 morphants. The specific absence of most cartilaginous structures in the pharyngeal arches indicates that the observed craniofacial phenotypes may be due to perturbed differentiation of cranial neural crest cells. Our studies show that this member of the olfactomedin protein family is an important regulator of development of the anterior nervous system.

Introduction

The mouse and human genomes encode at least 12 olfactomedin-related gene products These proteins contain a variable N-terminal domain and a conserved olfactomedin C-terminal domain (Karavanich and Anholt, 1998). The first member of the olfactomedin family was identified as a secreted glycoprotein in the bullfrog olfactory neuroepithelium (Snyder et al., 1991). Olfactomedin family members, however, are widely expressed and exhibit tissue-specific expression patterns (Kulkarni et al., 2000). Several olfactomedin proteins are expressed in the nervous system,. They include olfactomedin-1 (noelin-1) in brain, retina as well as kidney and lung (Kondo et al., 2000; Moreno and Bronner-Fraser, 2002), Olfactomedin-Noelin-Tiarin Protein 1 (ONT1) in midbrain and hindbrain roof plate, and axial and paraxial mesoderm (Sakuragi et al., 2006), tiarin in non-neural ectoderm (Tsuda et al., 2002), olfactomedin-2 (OM2) in brain and retina (this study), optimedin (olfactomedin-3) in brain and retina (Torrado et al., 2002), myocilin in the eye (Adam et al., 1997; Torrado et al., 2002), and gliomedin in Schwann cells (Eshed et al., 2005).

Recent studies have begun to shed light on the functions of olfactomedin proteins, and indicate that they play prominent roles in development of the nervous system. Noelins are necessary for development of the neural crest and regulation of neuronal fate (Barembaum et al., 2000; Moreno and Bronner-Fraser, 2002; Moreno and Bronner-Fraser, 2005). Tiarin modulates dorsalization in the developing spinal cord (Sakuragi et al., 2006; Tsuda et al., 2002). ONT1, a newly discovered member of the olfactomedin family, is also involved in development of neural crest cells, but opposes the functions of noelin and tiarin (Sakuragi et al., 2006). Myocilin mutations have been linked to congenital glaucoma and some cases of primary open angle glaucoma (Adam et al., 1997; Stone et al., 1997), and optimedin interacts with myocilin (Torrado et al., 2002). Optimedin is also a downstream target of Pax6 (Grinchuk et al., 2005), a transcription factor that is critical for CNS development (Simpson and Price, 2002). Gliomedin is a member of the olfactomedin protein family that is expressed by Schwann cells and required for molecular assembly of developing nodes of Ranvier in the peripheral nervous system (Eshed et al., 2005; Maertens et al., 2007). Collectively, these studies implicate members of the olfactomedin protein family as important regulators of nervous system development.

The function of OM2 has remained elusive, although OM2 mRNA expression pattern in the early developing CNS has been documented (Thisse and Thisse, 2004). OM2 shares a high level of similarity with myocilin, and a recent bioinformatics study predicted that myocilin may have been derived from OM2 via gene duplication and exon fusion events (Mukhopadhyay et al., 2004). We have capitalized on zebrafish as a powerful model for studies on early CNS development to begin to elucidate the function of OM2. In this study we have characterized the zebrafish OM2 orthologue and used morpholino antisense oligonucleotides directed against the translation start site and an intron-exon junction to interfere with OM2 protein expression during zebrafish development. OM2 protein knockdown produces highly penetrant phenotypes, including perturbation of the formation of axonal projections from branchiomotor neurons, disruption of anterior head and CNS development that includes severe defects in development of the olfactory pits, eyes and optic tectum, and abnormal development of the gill arches.

To explore possible molecular mechanisms underlying OM2 function, we analyzed Pax6 gene expression patterns in OM2 morphants, since Pax6 expression is prominent in the developing eyes and anterior CNS, and Pax6 disruption is associated with perturbed eye development (Hill et al., 1991). Our results show that the stereotypical expression patterns of Pax6.1 and Pax6.2 are perturbed in the eyes of OM2 morphants, which suggests that the normal function of Pax6.1 and Pax6.2 is dependent on expression of OM2. We also analyzed the expression of the neural crest marker crestin in OM2 morphants and the formation of the cartilaginous cells derived from neural crest cells. Such analysis provided strong evidence that OM2 plays a role in the differentiation of cranial neural crest cells. These studies demonstrate that OM2, along with other members of the olfactomedin protein family, is critical for development of the nervous system.

Results

Identification of zebrafish OM2

To identify the orthologue of human olfactomedin 2, we searched the zebrafish EST database for transcripts with olfactomedin motifs. We found OM2-encoding cDNAs from zebrafish, including BC044164 and its corresponding IMAGE clone (IMAGE ID 2639120). Alignment of OM2 cDNA to the UCSC genome browser showed a similar domain structure to that of human OM2 for exons 2-6. (Fig. 1A). The second exon of zebrafish OM2 contains 150bp (human OM2 150bp), the third exon contains 156bp (human 147bp), the fourth exon contains 220bp (human 220bp), the fifth exon contains 107bp (human 107bp), and the sixth exon up to the stop codon contains 675bp (human 675bp). However, unlike the first exon of human OM2, which contains 63bp starting from the ATG translation site, the sequence of the first exon of zebrafish OM2 calculated from BC044164 consists of only 27bp. Furthermore, this sequence could not be aligned in the same scaffold, but was aligned in another scaffold in the UCSC browser. In addition, the upstream ∼160kb region of zebrafish OM2 was well covered by paired BAC ends without any gaps, and this length was longer than any introns found in human olfactomedin genes (Mukhopadhyay et al., 2004), which indicates that the assembly of this region was reliable. Thus, the 27bp sequence not found in this region may have been incorrectly inserted in the EST clone (BC044164) during the generation of the EST library.

Figure. 1.

Identification and expression of zebrafish olfactomedin 2. (A) Schematic representation of the zebrafish OM2 gene. The first exon includes the sequence obtained by 5′ RACE. The gray box at the 5′ end indicates a signal sequence-coding sequence predicted by the SignalP3.0 algorithm (http://www.cbs.dtu.dk/services/SignalP/). Numbers in exons (boxes) indicate the numbers of nucleotides. The positions targeted by the two MOs used in this study are indicated by black lines. Arrows indicate the primers used for RT-PCR to verify the efficiency of 3i4e MO (panel D). Intron lengths are not to scale. (B) Coding sequence of OM2 first exon and corresponding amino acid sequence. The underlines on the amino acid sequence indicate signal peptides predicted by SignalP 3.0 server using neural networks (top line) and hidden Markov model (bottom line). (C) Analysis of OM2 mRNA expression during zebrafish development using RT-PCR. (D), RT-PCR with cDNA from 37hpf fish shows that the intron-exon boundary targeting morpholino (3i4e MO) specifically reduces mRNA with the correct splicing at the 3^rd intron and 4^th exon junction. OM2 primers used for this experiment are indicated in Fig. 1A (arrows). As a control, zebrafish actin mRNA level was not altered by OM2 3i4e MO, which indicates that disruption of mRNA splicing was specific for the target sequence.

This notion was further strengthened by absence of a predicted signal peptide sequence, critical for the secretion of OM proteins, from the 27 bp segment. For these reasons, we conducted 5′ RACE to identify the initial coding exon of zebrafish OM2. When the sequence obtained from 5′RACE was compared to the UCSC browser, the newly identified exon1 was found within the same scaffold. Furthermore, this new sequence could be aligned with exon1 of human OM3 and, most importantly, has a predicted signal peptide (Fig. 1B). Thus, we are confident that we identified the complete coding region of the zebrafish OM2 gene.

OM2 mRNA expression is developmentally regulated

To begin to assess OM2 mRNA expression during zebrafish development, we conducted RT-PCR analysis of OM2 mRNA. OM2 mRNA expression was detected at the earliest embryonic age examined, 3 hours post-fertilization (hpf), indicating maternal expression of OM2 mRNA (Fig. 1C). OM2 mRNA maternal expression was also detected at 6 hpf, but subsequently OM2 mRNA was markedly decreased in 10 hpf embryos (Fig. 1C). Weak expression of OM2 mRNA was again detected at 14 hpf, and increased continuously through 30 hpf, the latest embryonic age analyzed by RT-PCR (Fig. 1C). Later expression of OM2 mRNA was analyzed by in situ hybridization (see below).

OM2 mRNA is expressed in the developing CNS

We used the 5′RACE PCR product to clone a full-length coding sequence of zebrafish OM2 for preparation of riboprobes to analyze the distribution of OM2 mRNA in zebrafish embryos. OM2 mRNA expression was analyzed by in situ hybridization with digoxigenin-labeled riboprobes. OM2 mRNA was found exclusively in the anterior part of the body.

At 10 hpf there was no detectable OM2 expression, consistent with RT-PCR data (Fig. 2A). However, beginning at 19 hpf discrete and strong OM2 mRNA expression was observed in primary motoneurons in the branchiomotor nuclei (III, V, VII, and X in Fig. 2B,C). This is particularly evident by analysis of GFP expression in branchiomotor nuclei of islet1-GFP transgenic zebrafish (Figure 2E). At 19 hpf, OM2 mRNA was also detected in the proximal region of the optic primordium (Fig. 2C inset). As development progresses OM2 continues to be expressed in branchial motor nuclei (Fig. 2D-G). However, the well-defined OM2 mRNA expression pattern in branchial motor nuclei with fine cellular resolution was gradually lost due to its increasingly broader and diffuse distribution (compare Fig. 2C,E with Fig. 2M,N). By 54 hpf, widespread expression of OM2 mRNA in the midbrain and hindbrain obscured its expression in the branchiomotor nuclei, although we could still discern the OM2 mRNA expression pattern in some branchiomotor nuclei with close examination (Fig. 2N).

Figure. 2.

Localization of OM2 mRNA in developing zebrafish using in situ hybridization. A, OM2 mRNA is not detected in 10hpf embryos. B and C, 19 hpf embryo shows OM2 mRNA in the developing eye and branchiomotor nuclei (III, V, VII, and X). Expression in the proximal region of the optic primordium is shown in the inset (yellow dotted line shows the boundary of optic primordium). D and E, in 22.5 hpf embryos broader domains of OM2 mRNA expression were detected as well as the sharply defined branchiomotor nuclei-specific expression. GFP expression in branchiomotor nuclei of an islet-GFP embryo is also shown in E, demonstrating the branchiomotor identity of these nuclei. Arrowhead in D indicates a strong expression of OM2 in the tectum, which is also observed in a cryostat section (arrowhead in the inset). F and G, OM2 expression pattern in a 27 hpf embryo. Strong expression is detected in the forebrain and the tectum as shown in G. H and I, 36 hpf embryo shows strong expression of OM2 mRNA in the anterior CNS, including ventral regions that will develop into the lower jaw and pharyngeal arches (arrows in H). In addition, OM2 is clearly detected in the eyes. J and K, 48 hpf embryos, showing OM2 expression in eye and brain, the epiphysis (arrow in J), with branchiomotor nuclei expression still being observed (J). L and M, in 54 hpf embryos OM2 expression in forebrain, tectum, epiphysis and eyes is detected. More stratified patterns of OM2 mRNA are observed in the inner nuclear cell layer (in) and retinal ganglion cell layer (rg) in the eye (M). Close observation with cryosections shows that OM2 mRNA is still expressed in vagal (CN X) branchiomotor nuclei (arrows in N: red dotted box in M indicates the region shown in cryosection N). OM2 in developing pharyngeal arches is more restricted and this pattern is continued to 70 hpf embryos. O and P, 70 hpf embryo shows a similar pattern to 54 hpf embryos. Notice a strong, yet highly restricted pattern of OM2 mRNA expression is found along the pharyngeal arches (red inset in O was obtained from an oblique angle). Also in cryosections of 70 hpf embryonic eye OM2 mRNA is expressed in defined, laminated layers of the INL and RGC, as in 54 hpf embryos (inset in P). Small arrowheads in P indicates OM2 mRNA expression in pharyngeal arches.

By 36 hpf, OM2 mRNA was also present in the developing eyes and anterior CNS (Fig. 2H,I). As mentioned above, OM2 mRNA was first detected in the proximal region of the optic primordium in a restricted pattern (Fig. 2C inset). The expression domain of OM2 mRNA gradually expands so that by 36hpf, the entire eyes were stained by in situ hybridization (Fig. 2I). However, as the eyes become more stratified, OM2 mRNA becomes restricted to the inner nuclear layer and the retinal ganglion cell layer, as shown by analysis of cryostat sections of the embryonic eye (Fig. 2M, P inset). In addition, OM2 mRNA was consistently detected in the developing optic tectum from 22.5 hpf to 70 hpf (Fig. 2D-O). The epiphysis is another prominent CNS region with a high level of OM2 mRNA expression (arrows in Fig. 2J,L).

At 36 hpf, OM2 mRNA was detected in the developing pharyngeal arches (Fig. 2H,I). Interestingly, as development progresses OM2 mRNA expression in pharyngeal arches became increasingly more discrete from its earlier, diffuse pattern. Thus, at 70hpf, OM2 mRNA was clearly detected in developing pharyngeal arches (Fig. 2O inset and arrowheads in 2P). This pattern is intriguing as we found possible functions of OM2 in cartilage differentiation of neural crest cells in the pharyngeal arches (see below). Overall, these OM2 mRNA expression patterns resemble archived images deposited in the zFIN database (Thisse and Thisse, 2004), but provide more detailed information regarding the expression of OM2 mRNA during zebrafish development.

OM2 knockdown disrupts anterior CNS and head development, including neural crest cell-derived cartilaginous structures of the pharyngeal arches

To address the function of OM2 in zebrafish development, we injected 1-cell stage embryos with MOs that targeted either the translation start site (ATG MO) or the splice junction of intron 3 and exon 4 (3i4e MO) (Fig. 1A). To confirm that OM2 treatment results in OM2 knockdown, OM2 mRNA splicing was assessed by RT-PCR following treatment with 3i4e MO (Fig. 1D). These data show that normal splicing of OM2 mRNA is disrupted in these OM2 morphants.

Microinjection of 1-cell stage zebrafish embryos with either ATG MO or 3i4e MO resulted in similar, highly penetrant phenotypes (Table 1). Thus, for the majority of our experiments the 3i4e MO was used for OM2 knockdown. With increased concentration or volume of morpholino oligos, non-specific developmental defects were observed. Therefore, careful dosage analysis was conducted to find the optimal dose for each morpholino (Table 1). We used two criteria to determine the optimal dose. As a positive criterion, we used a stereotypical defect found in the midbrain hindbrain boundary (MHB) at ∼24hpf: absence or grossly hypomorphic tectal wall with spared cerebellar wall, which results in a single-walled MHB when viewed from the lateral angle (Fig. 3A-D). As an indicator of non-specific morpholino toxicity, we used a decrease in spontaneous tail coiling largely to a single side: morphants with higher MO dose showed more severe rigidity in their ability to coil to both sides. Since this mode of spontaneous coiling is entirely controlled by the spinal cord (Drapeau et al., 2002), where OM2 is not found, we concluded the coiling defects are non-specific and avoided MO doses displaying such non-specificity (Table 1). We found that 1mM, ∼0.4 nl for 3i4eMO, and 1mM, ∼0.7 nl for ATG MO, were optimal conditions that did not give rise to non-specific defects at 24 hpf. When control morpholinos were injected under identical conditions, no developmental defects were found, which strongly indicates that OM2 protein expression is specifically knocked down by treatment with either OM2 MO and the observed phenotypes were not caused by non-specific morpholino toxicity. Thus, unless otherwise noted, uninjected embryos were used for controls in MO injection experiments. Moreover, to address the possibility of a MO-mediated general delay of development, we used sonic hedgehog (SHH) in situ hybridization as a negative control. The level and pattern of SHH mRNA expression did not change in OM2 morphants (23/24: Fig. 3 O,P). Thus, we conclude that OM2 morphants show specific defects within the range of morpholinos we opted to use.

Table 1.

Summary of OM2 morphant phenotypes at 25 hpf

	Fused MHB fish / Total injected fish	Fused MHB (specific phenotype)	Spontaneous tail coiling**
			Normal	Defective
Control MO (3∼3.5nl)	0/48	0 %	48 (100 %)	0
3i4e MO (1∼1.5nl)	129/160	81 %	160 (100 %)	0
3i4e MO (2∼2.5nl)*	156/156	100 %	147 (94 %)	9 (6 %)
3i4e MO (3∼3.5nl)	147/147	100 %	103 (70 %)	44 (30 %)
ATG MO (3∼3.5nl)*	146/152	96 %	138 (91 %)	14 (9 %)

^*optimal dose determined

^**defects indicate non-specific MO toxicity

Figure 3.

OM2 morphants show specific defects in anterior CNS development (n=78). A and B, 25 hpf embryos from the lateral view show specific defects of OM2 morphants in the midbrain-hindbrain boundary (MHB), which appears to have been formed by the fusion of tectal (arrow) and cerebellar walls (arrowheads). Slightly smaller size of the morphant eyes is also shown. C and D, at 37 hpf, the MHB defect is still present. E-H, Pax2a in situ hybridization patterns are not affected by OM2 knockdown. Asterisks indicate otic vesicles. Pax2a mRNA expression at the boundary MHB is intact in OM2 morphants (arrowheads), despite the morphological defects (B,D). I and J, MHB-specific Fgf8 mRNA expression pattern is not altered by OM2 knockdown. Arrowheads indicate Fgf8 mRNA at the MHB. K and L, at 60 hpf, gross morphology of OM2 morphants is similar to that of control MO injected fish, except for mild edema in the 4^th ventricle (arrow in L). M and N, dorsal side of the head shows markedly decreased size of the optic tectum (asterisks). Notice also that the MHB is significantly thinner and less developed than its control counterpart. O and P, in situ hybridization with SHH probes shows no alteration in SHH mRNA expression patterns in OM2 morphants (arrows; n=24).

Consistent with the localization of OM2 mRNA expression in anterior regions of the zebrafish embryo, our observed phenotypes were primarily associated with alterations in anterior nervous system structures. OM2 morphants do not show dramatic alterations in overall body plan (Fig. 3K,L). However, striking morphological defects in the anterior CNS are evident upon careful examination. For example, as early as at 25 hpf, marked defects in optic tectal development are observed, which results in the stereotypic single-walled midbrain-hindbrain boundary (MHB) viewed from the lateral side (Fig. 3A,B). This MHB phenotype is likely due to the role of OM2 in tectal development, as hindbrain morphology appeared normal and the expression of MHB patterning markers, such as Pax2a and Fgf8, were unaffected in OM2 morphants at 20-28 hpf (Fig. 3C-J). Such tectum defects became progressively more pronounced as shown in 37 hpf (Fig. 3C,D), and 60 hpf morphants (average 64% decrease in tectum area, n=36; Fig. 3 M,N). 60 hpf embryos also show mild edema in the fourth ventricle and smaller eyes (average 24% decrease in eye area, n=36; Fig. 3 K,L).

We also observed developmental defects at the ventral side of the head. For example, development of the olfactory pits (arrows in Fig. 4A,B) and formation of the oral cavity (asterisks in Fig. 4A,B) was severely disrupted in OM2 morphants, with a less well-formed oral cavity found in a more caudal location. In addition, staining of 4 dpf OM2 morphants with Alcian Blue revealed dramatic perturbation of development of gill cartilage corresponding to the region of OM2 mRNA expression (Fig. 4C-4F), when no other significant morphological defects were detected in the overall body plan. These later developmental phenotypes related to cartilage development likely manifest at earlier developmental ages, as cell types associated with cartilage formation are perturbed in 33 hpf OM2 morphants (see Fig. 8). These observations are reminiscent of recent findings on the role of other olfactomedin family members in neural crest cell differentiation and migration (Barembaum et al., 2000; Moreno and Bronner-Fraser, 2002; Moreno and Bronner-Fraser, 2005; Sakuragi et al., 2006), and could be attributed to disruption of normal development of neural crest cells, as elaborated below.

Figure 4.

Craniofacial abnormalities in OM2 morphants. A, B Ventral view of the head shows several defects (n=170). Arrows point at underdeveloped olfactory pits. Brackets indicate smaller eyes (A and B). Asterisk shows a malformed mouth. C-F Alcian blue staining of 4dpf morphants shows the severely impaired pharyngeal arch cartilages (n=27). Arrows in C indicate cartilages in developing pharyngeal arches, which appear absent in OM2 morphants. H, hyoid arch; M, mandibular arch. In E and F the dramatically reduced cartilage in the pharyngeal arch is again apparent (asterisks in F).

Figure 8.

Analysis of cranial neural crest cells in OM2 morphants. A and B, confocal micrographs of 33 hpf fliI-GFP fish. Head is left. The first and second (yellow) asterisks indicate the developing mandibular and hyoid arches, which are only mildly affected by OM2 knockdown. More caudal pharyngeal arches (red asterisks) are severely affected in OM2 morphants. C and D, dlx2 in situ hybridization of 23 hpf fish. Asterisks in C and D indicates the migrated cranial neural crest cells within the developing mandibular and hyoid arches. Cranial neural crest cells in more caudal pharyngeal arches are indicated by arrowheads. Although hypomorphic, the pattern and position of the dlx2-positive cells suggest that initial specification and migration of cranial neural crest cells are normal in OM2 morphants. E-H, crestin in situ hybridization of 25 hpf zebrafish. Rostral streams of cranial neural crest cell migration are indicated by r2, r4, and r6, which originate from rhombomeres 2, 4, and 6, respectively. In morphants (F and H), r2 and r4 streams were not well segregated (68/70: dotted lines) and r6 stream was hypomorphic (69/70).

Axon guidance of branchiomotor neurons is perturbed in OM2 morphants

Our initial observation of OM2 mRNA in branchiomotor nuclei (Fig. 2C-G) and the abnormal development of anterior CNS structures in OM2 morphants prompted us to analyze the development of cranial nerves in greater detail. We first used islet-1-GFP transgenic zebrafish to examine axonal pathways of developing cranial motor neurons. Expression of GFP under the promoter of the islet 1 transcription factor visualizes developing motor neurons (Higashijima et al., 2000). When cranial motor neuron axons were examined from a lateral angle in 28 hpf to 54 hpf zebrafish, several axon pathways appear affected by OM2 MO (Fig. 5A-H). At 28 hpf, pioneering primary branchiomotor neurons are positioned along the dorsal midline of the head with a few newly formed axon bundles (arrows in Fig. 5A). Note that these early events take place when OM2 mRNA levels are still low. Consistently, OM2 morphants do not show significantly different phenotypes at this early stage, when viewed both laterally and dorsally (Fig. 5A, B). An apparent fusion of the trigeminal nuclei was observed, however, in 28 hpf OM2 morphants (Fig. 5 A,B, asterisk). Pronounced phenotypic differences in islet-1-GFP zebrafish were observed between controls and OM2 morphants at later stages when OM2 mRNA levels normally increase. At 36 hpf, OM2 morphants have shorter axons of neurons in the trigeminal (V) and facial (VII) nuclei (V, VII in Fig. 5C, D). In addition, axon branches of newly formed gill motor neurons were missing in morphants compared to controls (arrowheads in Fig. 5C, D). Moreover, precise axon guidance appeared abnormal in OM2 morphants (Fig. 5D). Thus, in 36 hpf controls, a growing axon terminus from the glossopharyngeal nucleus traverses between the first and second pharyngeal endodermal pouches (Fig 5C inset). Likewise the axon terminus of the first gill motor nerve traverses between the second and third pharyngeal pouches as shown in Fig. 5C inset. These precise axonal trajectories were frequently disrupted in OM2 morphants as shown in Fig. 5D inset, where neither of these axon growth cones successfully navigated their normal paths.

Figure 5.

Axon guidance of branchiomotor neurons is perturbed in OM2 morphants. Head is to the left in A-F. All images were captured by a laser confocal microscope and z-stacked images were collapsed. A, B Lateral and dorsal view of 28 hpf embryos showing branchiomotor nuclei and initial axon outgrowth (n=23 for A, n=40 for B). Arrows show newly formed branchiomotor axon bundles and primary branchiomotor neurons. No apparent defects were detected at 28 hpf, except for fusion of the trigeminal nuclei in OM2 morphants (asterisks). C, D, 36 hpf embryos, showing normal development of branchiomotor nuclei in morphants (D), but mildly impaired outgrowth of branchiomotor axons (n=20 for C, n=44 for D). Inset in D shows failure of axons to grow into the pharyngeal arches in OM2 morphants (red: zn5 staining). Arrowheads in C indicate newly formed gill motor axons. Numbers in insets of C, D show pharyngeal endodermal pouches, absence/fusion of pouches in morphants. E-H, 54hpf embryos analyzed for branchiomotor development (n=25 for E and G, n=73 for F and H). In control islet1-GFP fish (E), cranial nerve IV, V, and VII are labeled with Roman numerals. SoL indicates superior orbital lateral line axons. F, Lateral view of branchiomotor axons of a 54hpf ATG OM2 morphant fish. Islet-1 derived GFP expression shows the stereotypically perturbed projection of cranial nerve V and VII w ith abrupt endings and numerous fine filopodia (inset in F). Gill motor neurons from CN X do not form nerve bundles that innervate gill muscles (asterisks in F). Despite the relatively normal position of branchiomotor nuclei V and VII, close observation reveals slightly hypomorphic nuclei in morphants (bracket with CN V & VII in F). Aberrant axons with anteriorily wandering processes were detected from the vagus (X) nerve, as indicated by arrows in F. Absence of SoL is indicated by red asterisks. G, Ventral view of 54 hpf control fish. The characteristic crossed axon pathway formed by cranial nerve V and VII is apparent (double arrow). Single arrows indicate intermediate branches of cranial nerve V. H, In OM2 ATG morphants cranial nerve V and VII prematurely terminate as shown in F, which results in the absence of the crossed axon pathway (double arrow). Notice also the absence of peripherally extended cranial nerve VII in morphants (double asterisk). Single asterisk indicates hypomorphic telencephalic nuclei. I, Lateral view of 72 hpf control fish. Anteriorily projected CN V and VII are indicated by white arrowheads (V) and red arrows (VII). J, Lateral view of 72 hpf OM2 morphant fish. One of the well-recovered fish is shown here. CN V axon (white arrowheads) grew further compared with F and made a stereotypic medio-caudal turn, whereas CN VII axons did not grow much further (red arrows). K, Ventral view of 72 hpf control fish. As in G, a cross-shaped axon pathway made by CN V and CN VII axons is readily visible (white arrowheads and red arrows). L, Ventral view of 72 hpf OM2 morphant. White arrowheads indicate the axon pathway formed by further-projected CN V axons, which could reach the midline in this case. Notice much more caudal location of the axon pathway compared with that in control fish (K). Red arrows indicate CN VII axons that did not grow further from the earlier observation (54 hpf ventral view, H).

At 54 hpf, more pronounced abnormalities in axonal projections were observed. Superior orbital lateral line axons were missing or highly hypomorphic in OM2 morphants (SoL, Fig 5E, F). Furthermore, trigeminal (V) and facial (VII) nerves showed a premature termination of axon growth with fine branches (Fig. 5F inset). In normal developing embryos long axons from cranial nuclei V and VII give rise to curved pathways towards the midline, which form a stereotypical crossed pathway when observed from a ventral angle (double arrow in Fig. 5G). In contrast, premature termination of axonal projections from cranial nerves V and VII in OM2 morphants (Fig. 5F) results in the absence of this crossover structure (double arrow in Fig. 5H). OM2 morphants consistently showed premature termination of cranial nerves V and VII at 54hpf (73/73). In addition, peripherally projecting axons from cranial nerve VII form a distinct midline structure in normal embryos (double asterisk in Fig. 5G), but in OM2 morphants this axonal pattern is entirely missing due to premature termination of axon extension (single arrows in Fig. 5H). Strikingly, gill motor nerves that arise from the vagal nucleus (X) did not form branches that innervate gill muscles (asterisks in Fig. 5F). Instead, aberrant axon bundles were often observed (∼20%). Figure 5F shows an example of such aberrant axon outgrowth that grew posteriorly from vagal sensory ganglia and developed into long wandering, defasciculated axons (arrows in Fig. 5F). Expression of islet1-GFP also consistently showed mild defects in the number of cells and their clustering patterns in cranial nuclei of OM2 morphants, despite the generally conserved positions of these nuclei (CN V&VII in Fig. 5A-F and asterisks in Fig. 5G, H), whereas the initial differentiation/localization of primary branchiomotor neurons was largely unaffected due to the absence of OM2 in early embryonic stages (Fig. 5A-D). Interestingly, when OM2 morphants were allowed to develop further until 72hpf, most fish (60/73) showed partial recovery of CN5 axons, with some almost reaching to the midline (5/73: arrowheads in Fig. 5J,L). In stark contrast, only few morphants at 72hpf show further growth of CN7 axons (10/73) and even this growth was very limited (arrows in Fig. 5J,L). CN7 axons in morphants could not grow to the midline (0/73). Comparison between 54hpf (Fig. 5H) and 72hpf (Fig. 5L) indicates that axon outgrowth from CN5 and CN7 is differentially affected by OM2 knockdown.

As OM2 morphants display most specific phenotypes after 24 hpf when OM2 expression reaches high levels, we hypothesized that late-onset events which begins within the temporal and spatial boundary of the OM2 expression pattern would be more severely affected by OM2 morpholino treatment. Thus, we extended our analysis of neuronal development by assessing the effects of OM2 knockdown on a population of neurons that are born around the time when OM2 expression is high. We visualized such late-onset, dorso-laterally positioned hindbrain commissural neurons by immunohistochemistry with the Zn-5 monoclonal antibody (Trevarrow et al., 1990). Laser confocal micrographs of 28, 36 and 54hpf fish show pronounced differences in Zn-5 immunostaining in OM2 morphants. When we identified the more laterally located commissural nuclei (Higashijima et al., 2000) adjacent to the primary cranial nuclei, we found that several such commissural nuclei were entirely missing or significantly underdeveloped in OM2 morphants. The abnormal development of these commissural neurons is not the result of defects in rhombomere boundaries, as radical fringe mRNA expression appeared normal in OM2 morphants (data not shown). Affected commissural nuclei were labeled by their nearest cranial nuclei, which include IV, V, VI, VII, IX, and anterior-most X (labels in Fig. 6E and arrowheads in Fig 6F). Interestingly, the innervation domains of primary motor neurons (islet-1 GFP) and commissural neurons (Zn-5) were mutually exclusive as evident from the non-overlapping green and red fluorescence patterns (Fig. 6E, F). In 28 hpf morphants, Zn-5 immunostaining indicates that OM2 knockdown results in impaired development of non-neuronal structures that include less-well separated pharyngeal endodermal pouches (PEP) (Figure 6A,B). In 36 and 54 hpf morphants the formation of commissural neuronal nuclei was severely affected by OM2 MOs (brackets in Fig. 6C,D and arrowheads in Fig. 6F). Around the PEPs, through which the axons of primary gill motor neurons grow, control fish show sharply demarcated boundaries formed by Zn-5-positive cells (arrows in Fig. 6E and inset) and primary gill motor axons course through regions where Zn-5-positive cells are absent. In contrast, OM2 morphants do not show such parallel innervation patterns of growing gill motor axons, and Zn-5 positive cells form less sharply defined areas (arrows in Fig. 6F and inset). The same results were obtained with the OM2 ATG MO (Fig. 6F) and the 3i4e MO (Fig. 6B, D). The nearly identical phenotypes obtained with the two different MOs suggest that both MOs reduce OM2 protein levels rather than generating truncated OM2 proteins with abnormal functions. It is of note that Zn-5-positive endodermal pouches are formed, albeit with defects, in the correct region. This is significant as one could argue that late phenotypes such as cranial nerve axon guidance defects (Fig. 5F,H) and absence of cartilaginous structures (Fig. 4E,F) in the ventral head are due simply to the absence of tissue growth in the region. The presence of PEP as shown by immunohistochemistry provides direct evidence that ventral tissues along the jaw and gill arches are still formed in OM2 morphants. We conclude that OM2 plays important roles in axon guidance of cranial motor neurons and the development of late-onset, laterally positioned hindbrain commissural neurons.

Figure 6.

Formation of dorso-laterally located commissural nuclei is disrupted in OM2 morphants (number of samples analyzed are the same as in Fig. 5). Later onset, ascending commissural nuclei within the hindbrain were visualized using Zn-5 immunostaining. islet-1 positive primary motor neurons are detected by GFP (green). A, C, E, control embryos. B, D, F, OM2 morphants. A, B, 28 hpf embryos, showing that pharyngeal arch endodermal pouch (PEP) formation is moderately defective in OM2 morphants (arrows). Dotted line in B denotes partially divided endodermal pouches in OM2 morphants. M, mandibular arch; H, hyoid arch. C, D, 36 hpf embryos, showing intricate structures of Zn-5 positive cells. Hindbrain commissural neurons (brackets) are almost completely lacking or severely underdeveloped in OM2 morphants, although commissural neurons adjacent to the CN VI do form partially in morphants (dotted circle). PEP formation in OM2 morphants also appears abnormal, with pharyngeal endodermal pouches 2 and 3 possibly fused. g1, first gill endodermal pouch. E, F, In 54 hpf embryos continued absence of development of hindbrain commissural neurons is observed in OM2 morphants. Gill motor axons from cranial nerve X send their axons through Zn-5 negative areas (inset in E), which is sharply contrasted by morphant embryos where gill motor axons fail to grow through the PEPs (inset in F). All photographs were taken with 200x magnification and z-stacked images were flattened to show structures located at different depths. The scale bar in F indicates 100 micrometer.

Pax6.1, Pax6.2 mRNA expression patterns are altered in OM2 morphants

Pax6.1 is prominently expressed in the early zebrafish CNS, in particular the anterior forebrain and eyes (Scholpp et al., 2003). Mutations in Pax6 lead to the small eye phenotype (Hill et al., 1991). With regard to olfactomedin protein function, overexpression of tiarin mRNA in Xenopus embryos expanded the expression domain of Pax6 (Tsuda et al., 2002), suggesting that olfactomedin proteins may modulate Pax6 expression. Based on these observations we decided to assess Pax6.1 gene expression in OM2 morphants as this might provide insights into possible molecular mechanisms that underlie OM2 function in eye development.

We analyzed Pax6.1 and Pax6.2 mRNA expression in normal 26 hpf-54hpf zebrafish. We detected a decreased level of Pax6.1 mRNA in the eyes of 28 hpf, 36 hpf and 54 hpf OM2 morphants (Fig. 7A-H). More importantly, a well-delineated band of Pax6.1 mRNA expression, corresponding to retinal ganglion cells and inner nuclear layer neurons, which appears in 54hpf zebrafish eyes (arrows in Fig. 7 E,G), was substantially altered in OM2 morphants (Fig 7 F,H). We also examined Pax6.2 mRNA expression in OM2 morphants. Like Pax6.1, the expression of Pax6.2 was diminished and its characteristic well defined expression pattern was abolished in OM2 morphants (Fig. 7I-P). These observations suggest that OM2 is needed for proper expression of Pax6 genes during development of the zebrafish eye.

Figure 7.

Pax 6 mRNA expression is disrupted in 54 hpf OM2 morphants. A,C,E,G, Pax6.1 in situ hybridization of 28-54 hpf control fish eyes. B,D,F,H, Pax6.1 in situ hybridization of 3i4e MO-injected fish eyes. Notice the reduction of Pax6.1 mRNA level in the eye at all ages, but Pax6.1 mRNA can still be observed in the brain. Arrows in E and G indicates a sharply delineated Pax6.1 mRNA expression pattern at 54 hpf, which was lost in OM2 morphant eyes (F and H). I,K,M,O, Pax6.2 in situ hybridization of 25-54 hpf control fish. High level of Pax6.2 mRNA expression is particularly evident in the 54 hpf eye with sharp, circular boundaries (arrows in M and O). J,L,N,P, Pax6.2 in situ hybridization of 25-54 hpf 3i4e MO-injected fish. Much weaker Pax6.2 mRNA expression with less well-defined layers is observed in the eye at all developmental stages.

Cranial neural crest cell development is disrupted in OM2 morphants

Since functional analysis of other olfactomedin family members, such as noelins or tiarin, have suggested roles in development of neural crest cells (Barembaum et al, 2000; Sakuragi et al, 2006), and our OM2 morphant phenotypes also suggest a role for OM2 in neural crest cell function, we further investigated the role of OM2 in cranial neural crest cell development. To determine if initial specification or migration of cranial neural crest cells, which give rise to pharyngeal cartilage, is perturbed in OM2 morphants, we first analyzed fliI-GPF transgenic zebrafish (Lawson and Weinstein, 2002). FliI is expressed in the cranial neural crest in Xenopus (Meyer et al, 1995), especially in mandibular, hyoid and branchial arches. In fliI-GFP transgenic zebrafish we observed that perturbed expression of fliI-GFP was perturbed in the region of the branchial arches (Fig. 8A, B). Frontal arches (mandibular and hyoid) were less affected (yellow asterisks in Fig. 8A,B) compared to the severe defects found in the caudal branchial arches (red asterisks in Fig. 8A,B; n=72). This observation is consistent with the more severe defects observed in the formation of cartilaginous tissue in caudal pharyngeal arches, in contrast to only mild cartilaginous defects in mandibular and hyoid arches (Fig. 4C-F). That fliI-positive pharyngeal arches are formed at all, albeit with defects, demonstrates that initial specification and migration of cranical neural crest cells are largely intact. This conclusion was corroborated by our in situ hybridization analysis of the dlx2 gene, which is a marker for pre-migratory and migrating cranial neural crest cells (Luo et al., 2007; Yan et al., 2002). As shown in figure 8C and D, dlx2 mRNA expression patterns in OM2 morphants were not markedly altered compared with control MO treated fish, with only a mild decrease in expression levels. These results suggest that the formation and migration of cranial neural crest cells are not significantly altered by OM2 knockdown.

We next employed the neural crest cell marker crestin (Rubinstein et al., 2000) to examine by in situ hybridization the development and migration of cranial neural crest cells in OM2 morphants. In 25 hpf embryos three distinct migratory streams of cranial neural crest cells are observed in hindbrain rhombomeres r2, r4 and r6 (arrowheads for r2 and r4, and arrow for r6 in Fig. 8E,G). At r6 the migrating neural crest cells comprise a rostral population that initiates migration through the third branchial arch (arrows in Fig. 8E,G; Halloran and Berndt, 2003). In OM2 morphants the most anterior populations of cranial neural crest cells in r2 and r4 are not well segregated (dotted lines in Fig. 8F), and the r6 neural crest cells exhibit a hypomorphic migratory stream (Fig. 8H). The in situ hybridization pattern of crestin in OM2 morphants strongly indicates that initial specification of cranial neural crest cells was not affected by OM2 knockdown, as is shown by the relatively well-preserved patterns of crestin-positive cells. On the other hand, defective migratory streams, especially in r6, suggest that OM2 is necessary for normal neural crest cell migration along this route. However, OM2 knockdown did not completely eliminate the migration of cranial neural crest cells, as demonstrated in OM2 morphant fliI-GFP fish and by dlx2 in situ hybridization (Fig. 8A-D), where neural crest cell derived tissues in pharyngeal arches are still formed, albeit with defects,. We conclude that our data should be interpreted that OM2 is likely to play a minor role in cranial neural crest cell migration. Therefore, severely hypomorphic cartilaginous tissues in caudal pharyngeal arches (Fig. 4D,F) is likely the consequence of perturbed neural crest cell differentiation, rather than defective specification or migration. On this account, it is particularly salient that OM2 mRNA is found around the developing pharyngeal arches, with a progressively more defined pattern within gill arches shortly before neural crest cells undergo differentiation into cartilaginous cells (Fig. 2L,N). These observations reinforce our conclusion that OM2 is critical for normal neural crest cell development in pharyngeal arches.

Discussion

Zebrafish OM2 morphants are characterized by anterior CNS abnormalities

In this study we capitalized on the zebrafish as an advantageous developmental model system to investigate the function of OM2, using microinjection of antisense morpholino oligonucleotides to knockdown expression of OM2 protein during early development. We used two different MO targets of the gene together with control MOs to insure specificity of phenotypic effects. We also attempted to demonstrate the specificity of OM2 MOs by co-injecting in vitro-transcribed capped OM2 RNA in order to rescue the morphant phenotypes. However, we have not successfully achieved rescue of OM2 MO phenotypes. We believe this is likely due to two factors: the late onset of OM2 mRNA expression, which is first detected at 14hpf by us, and at 19 hpf by others (Thisse and Thisse, 2004), and degradation of injected RNAs. In a recent study, similar rescue attempts with late onset gene plexinD1 capped RNA were also unsuccessful at rescuing severely affected vasculature in plexin D1 morphants (supplement in (Torres-Vazquez et al., 2004)). Thus, late onset gene expression may not be readily amenable to mRNA rescue in zebrafish morphants.

Zebrafish OM2 mRNA is most prominently expressed in the developing brain, in particular the midbrain and hindbrain (Thisse and Thisse, 2004), as well as in the developing eye and epiphysis. The majority of morphological phenotypes we observed in OM2 morphants correlate well with OM2 gene expression patterns and were restricted to the developing anterior CNS, eye, and gill arches; alterations in GFP expression in islet-1-GFP transgenic zebrafish were primarily associated with hindbrain branchiomotor axons. Developmental functions of OM2 in neural crest cell differentiation into cartilaginous cells were suggested by our observation of OM2 morphant phenotypes in the caudal pharyngeal arches, with little changes in early neural crest cell markers such as dlx2 and fliI.

Prominent morphological phenotypes that were observed in OM2 morphants included smaller eyes, disruption of olfactory pits, and severe deformation of the optic tectum. These phenotypes were observed by 25hpf during early development, and persisted at 60 hpf. Each of these phenotypes corresponds to expression patterns of OM2 mRNA prior to and/or during the developmental stage when the phenotype was observed.

OM2 morphants exhibit disruption of branchiomotor axon pathways

A prominent phenotype observed in islet-1-GFP transgenic zebrafish treated with OM2 MOs was a disruption of axon growth from branchiomotor neurons. The general GFP expression pattern in primary branchiomotor nuclei in the brainstem was unaffected by OM2 knockdown, but axon pathway formation by branchiomotor nuclei was disrupted. While the majority, if not all, primary branchiomotor axons display growth defects, the most pronounced effects of OM2 knockdown on growing branchiomotor axons were associated with cranial nerves IV, V, VII and X. The trigeminal and facial axons were truncated and failed to innervate target tissues in the head. The gill motor axons were also truncated and failed to grow into the gill pharyngeal arches.

Recent studies have provided some insight into potential regulatory mechanisms for branchiomotor axon growth. A genetic screen for cranial nerve mutations in mouse indicated that Pax3 is required for growth of the facial nerve, as well as formation of the lower jaw (Mar et al., 2005). While these Pax3 phenotypes resemble some of the OM2 phenotypes in our current studies, it is clear that the OM2 phenotypes suggest a more complex regulation of branchiomotor axon growth than has been described for Pax3 in mouse. Studies in zebrafish have implicated semaphorins, specifically Sema4E, in the establishment of branchiomotor axon pathways (Xiao et al., 2003). Misexpression of Sema4E, or knockdown with Sema4E MOs, resulted in pronounced effects on facial and gill motor axon growth that resemble the axonal growth patterns observed in OM2 morphants. However, other branchiomotor axons, such as trigeminal axons, were unaffected by Sema4E misexpression or knockdown (Xiao et al., 2003). Interestingly, some cranial nerves in OM2 morphants appear to be more sensitive to OM2 perturbation following recovery from MO treatment. For example, CN7 axons exhibited more retarded growth when compared to CN5 axons at 54 hpf (Fig. 5I-L). These data are similar to findings reported in Sema4E studies (Xiao et al., 2003). It is therefore tempting to speculate that OM2 is an extracellular regulator of semaphorin function during branchiomotor axon growth, possibly via its regulation of semaphorin binding to its receptors.

The relationship between OM2 expression and the early neurodevelopmental transcriptional regulator Pax 6

Pax 6 is critical for eye development, with the small-eye mutant phenotype being due to loss of Pax6 function (Hill et al., 1991; Matsuo et al., 1993). Overexpression of tiarin mRNA in Xenopus embryos produced enhanced expression of Pax6 mRNA (Tsuda et al., 2002), and optimedin (olfactomedin-3) is a downstream target of Pax6.1 in mouse (Grinchuk et al., 2005). Based on these previous observations, we asked whether Pax6 gene expression could be functionally related to OM2. Two Pax6 genes are expressed in zebrafish, Pax6.1 and Pax6.2. We observed a decrease in Pax6.1 and Pax6.2 mRNA expression in OM2 morphants. In normal 54 hpf embryos Pax6.1 and Pax6.2 are primarily expressed in the retinal ganglion cell layer and inner nuclear layer of the eye (Cheng et al, 2006), and this well-defined expression was not detected in OM2 morphants, where much less defined, weaker expression was observed. Although we have not established a direct causal relationship between OM2 expression and Pax 6 function, it is tempting to hypothesize that OM2 knockdown leads to a reduction in eye size due to disruption of Pax6.1 mRNA expression pattern. The reduction in eye size in OM2 morphants is not as pronounced as in Pax6 small-eye mutants, probably because OM2 expression does not occur until 14 hpf.

Mutations in Pax6 have also been associated with disruption of the formation of nasal cavities in mice (Grindley et al., 1995). Again, this observation parallels our observed effects of OM2 MOs on development of the olfactory pits in zebrafish, which were nearly absent in OM2 morphants. Thus, our data and reports from other studies of olfactomedin protein functions raise the possibility that OM2 either directly or indirectly regulates Pax6.1 expression.

Role of olfactomedins in cranial neural crest cell development

Initial studies on neurodevelopmental functions of olfactomedin proteins have focused on their effects on neural crest cells. Noelin-1 regulates expression of the neurogenic gene NeuroD in Xenopus (Moreno and Bronner-Fraser, 2001). Noelin-1 overexpression results in excessive and prolonged neural crest cell migration with concurrently elevated expression of the neural crest cell marker, Slug, again implicating olfactomedins in regulation of neural crest cell differentiation and migration (Barembaum et al., 2000). Another olfactomedin family member, ONT1, has antagonistic effects on the expression of neural crest cell markers when compared to noelin-1 (Sakuragi et al., 2006), but like noelin-1 prolongs the generation and migration of neural crest cells. Tiarin induces the expression of dorsal neural tube markers, such as Pax3 (Tsuda et al., 2002), whereas noelin-1 and ONT1 do not induce dorsal neural tube markers (Sakuragi et al., 2006). Tiarin also has variable effects on the expression of neural crest cell markers during chick early development, suppressing expression of FoxD3 in a manner similar to noelin-1 (Sakuragi et al., 2006); however, the later neural crest markers Twist and Ets are suppressed by tiarin while these markers are up-regulated by ONT1 and noelin-1 (Sakuragi et al., 2006).

While we have not addressed the precise mechanism by which zebrafish OM2 regulates the differentiation of neural crest cells, our analysis of crestin and dlx2 gene expression as well as fliI-GFP expression patterns in OM2 morphants, suggests that OM2 has a limited function in the initial specification of the neural crest and more significant, but not critical, roles in migration of cranial neural crest cells. Conversely, Alcian Blue staining of OM2 morphants together with OM2 in situ hybridization in pharyngeal arches indicates that OM2 is a critical element in cranial neural crest cell differentiation into cartilaginous cells within developing pharyngeal arches. Our studies, however, indicate that OM2's regulatory function on neural crest cell differentiation does not extend along the entire anterior-posterior body axis, as OM2 mRNA expression is primarily localized to the anterior CNS. This suggests that subpopulations of neural crest cells, such as cranial neural crest cells that migrate in the head, are subject to regulation by OM2.

Elucidating the interplay between OM2 and other members of the olfactomedin gene family in neural development, the mechanisms that control their expression and the cellular targets they regulate are likely to reveal new aspects of neural development. Our studies show that zebrafish OM2 is an essential regulator of early development of the anterior central nervous system with diverse effects on the development of cranial motor nuclei, sensory organs, and tissues that derive from the neural crest.

Experimental Procedures

Fish maintenance

Wild-type adult zebrafish (Ekkwill background) and islet-1-GFP transgenic zebrafish (Higashijima et al., 2000) were maintained at 28.5°C under standard laboratory conditions (Westerfield, 2000). Embryos were allowed to develop and staged by hours or days after fertilization at 28.5°C using morphological criteria (Kimmel et al., 1995).

Molecular cloning

To obtain the 5′ UTR sequence of the OM2 gene, 5′ RACE was performed with a gene specific primer (5′ – CTTCCTCCCTCAGCCGCAGAATCAT – 3′) using the GeneRacer kit (Invitrogen). The 145 bp PCR product was cloned into the TOPO TA vector (Invitrogen) for sequencing. The clones from 5′ RACE were sequenced and analyzed by BLAST and UCSC genome browser. In order to construct a full-length open reading frame of OM2, an IMAGE clone (2639120) containing the published OM2 sequence (BC044164) was obtained from Open Biosystems. The OM2 encoding insert was subcloned into the Xho1 restriction site of the pCMV-Script expression vector (Stratagene). To replace the 5′ region of OM2 from the IMAGE clone with our 5′ RACE product, a forward PCR primer with a PstI restriction site and a reverse primer downstream of a unique PstI site in the published OM2 sequence were used to PCR-amplify the 5′ region of the 5′ RACE product. The corresponding segment of the 5′ region of OM2 IMAGE clone was removed by PstI digestion of XhoI-cloned OM2 in pCMV-Script, which was used as a vector to ligate to the PstI digested PCR insert. Final product of pCMV-OM2 full coding sequence was used to generate digoxigenin-labeled riboprobes and capped RNA.

For Pax6.1 and Pax6.2 riboprobe synthesis, the following RT-PCR primers were used for PCR amplification: Pax6.1 forward: 5′ – GGAACCGTGCGTCTCATAAC – 3′; Pax 6.1 reverse: 5′ – GAAGTGGCACTATCCCCGTA – 3′; Pax6.2 forward: 5′ – TTCACTGTTTTGCTCGGAAG – 3′; Pax6.2 reverse: 5′ – GTCCAGCTTCTCGCTCAGTC – 3′. Amplicons were subcloned in the TOPO TA cloning vector (Invitrogen).

Antisense morpholino injection

Antisense morpholino nucleotides (MOs) (Gene Tools, LLC) were designed against the 5′ sequence near the start site of translation (Nasevicius and Ekker, 2000) and an intron/exon splice site between the 3^rd intron and 4^th exon (Draper et al., 2001). The sequences of MOs used were: OM2-ATG-MO: 5′- AGCGGAACGCTCATGTCGTCCCC -3′ (underline indicates the complementary sequence of the initial ATG site), OM2 3i4e MO: 5′- CCTTTAACTCCTGCGGTAAACAGAA - 3′, and Control-MO : 5′-CCTCTTACCTCAGTTACAATTTATA-3′. MOs were solubilized in water at a concentration of 2 mM and diluted to 1mM in water and phenol red before injection into one cell stage embryos. A series of injections with different MO volumes were conducted to establish optimal amounts of OM2 MOs. Approximate estimates of optimal volume obtained with mineral oil tests are about 0.4∼0.5 nl. We determined the optimal dose to be a volume that results in specific phenotypes in over 90% of injected fish with less than 10% showing non-specific MO toxicity as measured by defective spontaneous tail coiling. Since such MO toxicity could be observed as early as at 24 hpf, and OM2 morphant phenotypes are evident primarily at later stages, we were able to discard morphants with non-specific MO toxicity and to focus on specific phenotypes that appear at later time points. This pre-screening for non-specific effects enabled us to observe consistently reproducible morphant phenotypes without non-specific MO toxicity. Both OM2 ATG MO and splice-modifying 3i4e MO resulted in similar phenotypes. The control MO at the same concentration and volume produced no distinguishable effects compared to uninjected fish.

Whole-mount in situ hybridization

We chose the N-terminal region of OM2 from ATG to the first PstI restriction site, which does not include the well-conserved olfactomedin domain as a target for sense and anti-sense digoxigenin-labeled riboprobes generated by in vitro transcription. Whole-mount in situ hybridization was performed as described previously (Thisse et al., 1993; Yuan et al., 1997), with hybridization at 65°C. Embryos were cleared in benzyl alcohol/benzyl benzoate prior to viewing. In situ hybridization for mRNA expression for Pax2a, Pax6.1, Pax6.2, crestin, Fgf8, and radical fringe was also at 65°C.

Immunohistochemistry and Alcian Blue staining

Immunohistochemical localization of zebrafish antigens was conducted, as described previously, with some modification (Kim et al., 2007), using the mouse MAb Zn-5 antibody (Developmental Hybridoma Bank). Zebrafish embryos were fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) and incubated overnight at −20° C in methanol. Embryos were washed three times in PBS-0.2% Triton X-100, treated 20 min with proteinase K (1 μg/ml) in PBS-0.2% Triton X-100, and again washed three times in PBS-0.2% Triton X-100. Following blocking for 2 h in PBS-0.2% Triton X-100-5% normal donkey serum-1% DMSO, embryos were incubated overnight at 4° C with antibodies diluted in the same buffer. The Zn-5 antibody was used at a dilution of 1:500. Embryos were washed three times in PBS-0.2% Triton X-100 and incubated 2 h in Cy3-conjugated donkey anti-mouse IgG (1:800 dilution; Jackson Immunobiologicals). Following additional washing, embryos were examined using a Zeiss LSM 510 confocal laser microscope.

For Alcian Blue staining, fish were fixed in 4% paraformaldehyde overnight and washed in water before overnight staining in 0.02% Alcian Blue in 70% ethanol/30% acetic acid. After destaining in 70% ethanol/30% acetic acid and dehydration in 100% ethanol, fish were rehydrated in a series of diluted ethanol to pure distilled water and cleared in 1.7% trypsin in sodium borate buffer (pH 8.5) for 5 min. Cleared fish were washed twice in 0.5% KOH for 5 min prior to observation with a Nikon SMZ1500 stereomicroscope.