Vertebrate MAX-1 is required for vascular patterning in zebrafish

Abstract

During embryogenesis, stereotypic vascular patterning requires guidance cues from neighboring tissues. However, key molecules involved in this process still remain largely elusive. Here, we report molecular cloning, expression, and functional studies of zebrafish max-1, a homolog of Caenorhabditis elegans max-1 that has been implicated in motor neuron axon guidance. During early embryonic development, zebrafish max-1 is specifically expressed in subsets of neuronal tissues, epithelial cells, and developing somites through which vascular endothelial cells migrate from large ventral axial vessels to form stereotypic intersegmental blood vessels (ISV). Blocking zebrafish max-1 mRNA splicing by morpholino injection led to aberrant ISV patterning, which could be rescued by injection of either C. elegans or zebrafish max-1 mRNA. Analysis of motor neurons in the same region showed normal neuronal axon pathfinding. Further studies suggested that the ISV defect caused by max-1 knockdown could be partially rescued by overexpression of ephrinb3 and that max-1 was involved in mediating membrane localization of ephrin proteins, which have been shown to provide guidance cues for endothelial cell migration. Our findings therefore suggest that max-1, acting upstream of the ephrin pathway, is critically required in vascular patterning in vertebrate species.

In vertebrates, the first blood vessels that develop are the large midline artery and vein, formed through vasculogenesis involving migration, differentiation and assembly of angioblasts (1). After the establishment of the major axial artery and vein structures, smaller blood vessels such as the trunk intersegmental vessels (ISV) are formed through angiogenesis, which involves the sprouting of endothelial cells from the preformed large vessels. The patterning of smaller angiogenic vessels is highly stereotypical and appears conserved among vertebrate species. Our understanding of angiogenic patterning during development has largely come from studies of model organisms, including chick, mouse, Xenopus, and zebrafish. In zebrafish, the first angioblasts arise in the posterior lateral mesoderm (2) and then migrate medially to form the primordia of the large axial vessels, the dorsal aorta and vein. Subsequently, endothelial cells migrate from the dorsal aorta to generate the secondary angiogenic ISV in the trunk region (3).

Both vascular and angiogenic development appear to involve signals from adjacent tissues. For instance, the notochord plays a key role in blood vessel development (4–6). In the zebrafish mutants floating head and sonic-you, notochord defects result in the absence of a dorsal aorta and ISV fail to sprout. Recently, a number of axon guidance molecules have been shown to be involved in vascular patterning. These molecules include the semaphorins, sema3a1and sema3a2 (7), netrin (8, 9), robo (10), neuropilin (11), and ephrin (12). These studies demonstrate a remarkable similarity between the molecular mechanisms that regulate neuronal and endothelial cell migration and patterning (13).

Max-1 was first identified and characterized as a neuronal guidance gene in Caenorhabditis elegans; its loss of function caused variable axon pathfinding defects (14). The C. elegans MAX-1 protein contains PH, MyTH4, and FERM domains arranged from N terminus to C terminus. This combination of domains suggests that MAX-1 represents a previously uncharacterized class of proteins. Huang et al. (14) also identified predicted homologs of max-1 gene in other species, including Drosophila, mouse, and human. It was further shown that human max-1 could partially rescue the C. elegans max-1 mutant phenotype, and RNA whole-mount in situ hybridization using a partial cDNA probe from mouse max-1 revealed primary expression of max-1 in the nervous system and muscle tissue of mouse embryos. These observations suggest that max-1 is conserved from C. elegans to human and that it might also function in the neural development of vertebrates.

We isolated the homolog of max-1 from zebrafish and analyzed its embryonic expression and function. Similar to its expression in mouse, zebrafish max-1 is mainly expressed in neuronal tissues and somites during development. An anti-sense morpholino oligo (MO) targeting a splice-site of zebrafish max-1 mRNA was designed, and embryos injected with this morpholino (morphant) were shown to lack a functional transcript of max-1. First, neuronal pathfinding in the morphant was examined by using transgenic zebrafish that have the somitic motor neurons labeled with specific GFP expression, but no apparent defects were observed. Because a number of neuronal guidance molecules have been implicated in vascular endothelial patterning, we examined ISV in the morphants of another transgenic zebrafish line that expresses GFP in blood vessels. Interestingly, abnormal ISV patterning was indeed observed, and the defect was specific to max-1 knockdown, because it could be rescued by mRNA injection of either C. elegans or zebrafish max-1. The ISV defect could also be partially rescued by overexpression of ephrinb3. In addition, max-1 appears to regulate membrane localization of ephrin proteins; the ephringb3-GFP fusion protein becomes abnormally aggregated in cytoplasm of max-1 knockdown cells. Our findings therefore suggest that max-1, acting epistatically upstream of ephrin pathway, is critically required in vascular patterning in vertebrate species.

Results

Molecular Cloning and Sequence Analysis of Zebrafish max-1.

A Blast search (15) of the Ensembl zebrafish genomic sequence database showed that zebrafish has one copy of the C. elegans max-1 homolog that is comprised of at least 28 exons on chromosome 17. Two nearby gene markers, XP_689775.1 and Q5BL95_BRARE, were colocalized with max-1 on one chromosome in human, mouse, and zebrafish. Therefore, this max-1 homolog is very likely an ortholog. To clone zebrafish max-1, we performed RT-PCR and 5′ RACE from total RNA isolated from 18-somites-stage zebrafish embryos and identified a 4.6-kb cDNA clone that contained the putative full-length coding sequence of max-1. In C. elegans, max-1 has two transcript variants, whereas in zebrafish, 5′-RACE and RT-PCR (six pairs of PCR primers distributed from 5′ end to 3′ end were used.) results revealed that there is only one isoform. Blast analysis of zebrafish max-1 cDNA in EST database didn't show any splicing alternatives either.

Zebrafish max-1 mRNA encodes a putative protein of 1,433 aa and has characteristic PH, MyTH4, and FERM domains arranged from the N terminus to the C terminus. Whereas MAX-1 proteins from other species have two PH domains, zebrafish MAX-1 contains only one (Fig. 1). PH domain corresponds to exon 10 to exon 12; MyTH4 domain corresponds to exon 17 to exon 20; FERM domain corresponds to exons 21–25. Sequence comparison revealed that zebrafish MAX-1 is highly homologous to the human MAX-1, with 60% identity and 73% similarity at the protein level. In the phylogenic analysis, zebrafish MAX-1 is most closely related to the human and mouse putative proteins (Fig. 1 and data not shown).

Fig. 1.

Diagrammatic comparison of putative MAX-1 proteins from zebrafish, C. elegans, Drosophila, mouse, and human. Percent identity and similarity are shown for each protein compared with the putative zebrafish protein. The coiled-coil domain (green) is present only in C. elegans, mouse, and human. C. elegans max-1 has two transcript forms: the long [1,099 aa (MAX-1A)] and the short [1,045 aa (MAX-1B)] form. MAX-1A and MAX-1B differ slightly at their N-terminal sequences. MAX-1A was chosen for comparison.

Expression Pattern of Zebrafish max-1 During Zebrafish Embryogenesis.

RNA whole-mount in situ hybridization showed that zebrafish max-1 was expressed initially at the bud stage [10 hours postfertilization (hpf)], and expression persisted for 5 days. RT-PCR results confirmed this observation (data not shown). During this period, the zebrafish max-1 was widely expressed in neural, mesoderm, and epithelial tissues.

In mesoderm and somites.

Zebrafish max-1 expression first appeared at the bud stage as two domains lateral to the tail bud (Fig. 2A, white arrowhead). A cross-section through this region showed that these domains are in the mesoderm on both sides of the tail bud, adjacent to prospective central nervous tissue (data not shown). As the embryo developed, the two domains elongated anteriorly along both sides of the prospective neural tube. At the 10-somites stage, expression of zebrafish max-1 had expanded into the newly formed somites (Fig. 2 C and E, white arrowheads), and maximum expression was seen at the late segmentation stage (18 somites to 24 hpf). At 28 hpf, anterior somites have lost zebrafish max-1 expression (Fig. 2I), whereas posterior somites and the borders between somites continue to express zebrafish max-1. The expression of zebrafish max-1 mRNA in somites disappear at ≈33 hpf.

Fig. 2.

Expression pattern of zebrafish max-1 and eprhinb3 detected by in situ hybridization. Embryos are anterior to the left and dorsal to the top, except in A, D, E, and G, which are dorsal views with anterior to the left. O and P are transverse sections through the trunk shown in N and H, respectively. A–K and P show max-1 in situ hybridization; L–O show ephrinb3 in situ hybridization. (A) Bud stage. The white arrowhead points to the max-1 expression region near the tail bud. (B) The 3-somites stage. The white arrowhead indicates the prospective hindbrain. (C) The 10-somites stage. The white arrowhead points to the somites. (D) The 10-somites stage (dorsal view). The white arrowhead points to the first rhombomere, and the red arrowhead indicates the third rhombomere. The white arrow points to the otic placode. (E) The 10-somites stage (dorsal view). The white arrowhead points to the somites. (F) The 14-somites stage. The white arrowhead indicates the epithelial tissue at the edge of the tail bud. (G) The 21-somites stage. The white arrowhead points to the dorsal midline. (H) The 26-somites stage. The white arrowhead points to the mouth. (I) The 28-hpf stage. (J) The 2-dpf stage. The white arrowhead points to the pectoral fin bud. (K) The 5-dpf stage. The white arrowhead points to the ear. (L) The 10-somites stage (ephrinb3 expression pattern). The white arrowhead points to the somites. (M) The 15-somites stage (ephrinb3 expression pattern). The white arrowhead points to the somites. (N) The 1-dpf stage (ephrinb3 expression pattern). The white arrowhead points to the ventral somites. (O) The 1-dpf stage (ephrinb3 expression pattern). Shown is a transverse section through the plate in N. The white arrowhead points to the ventral somites. (P) The 26-somites stage (max-1 expression pattern). Shown is a transverse section through the plate in H.

In neural tissues.

At the 3-somites stage, the prospective hindbrain and spinal cord began to express zebrafish max-1 (Fig. 2B). At the 10-somites stage, a number of sensory placodes, including the otic placode expressed zebrafish max-1, and the first and third rhombomeres had a higher level of zebrafish max-1 expression than other rhombomeres (Fig. 2D). At the 26-somites stage, staining in the brain began to disappear (Fig. 2H), and, at 28 hpf, expression in the brain was remarkably reduced (Fig. 2 I and J). Max-1 expression in the ear persists up to 5 days postfertilization (dpf) (Fig. 2K).

In epithelial tissues.

At the 14-somites stage, the distal edge of the tail bud began to show weak expression of zebrafish max-1 (Fig. 2F); this staining was stronger at the 18-somites stage. As embryogenesis continued, this staining extended anteriorly along the dorsal edge of embryo. At the 21-somites stage, staining of the dorsal midline was observed (Fig. 2G); in transverse sections, this staining corresponds to epithelial tissue (data not shown). Expression in the dorsal midline disappears at 33 hpf. Other epithelial tissues that expressed zebrafish max-1 were the lining of the mouth (from the 26-somites stage to 3 dpf) (Fig. 2H), the edge of pectoral fin bud (36 hpf to 3 dpf) (Fig. 2J), and the edge of the operculum at 3 dpf (data not shown).

Knockdown of Zebrafish max-1 Resulted in an ISV Patterning Defect.

To study function of Max-1, two MOs (MO1 and MO2) were designed to block max-1 splicing (Fig. 3). MO1 is complementary to the putative exon 9/intron 9 boundary, and MO2 is complementary to exon 8/intron 8 boundary. Respectively, 5.2 ng of MO1 and MO2 or equal amount of each in combination was injected into the 1- to 2-cell-stage, flk-1:GFP-transgenic zebrafish embryos (morphant), and development of GFP-labeled blood vessels in the live transgenic zebrafish embryos was directly observed under a fluorescent microscope for 5 days. At this dosage, although morphants appeared slightly thinner than the uninjected controls little other general morphological defects were observed. At higher dose, 10.4 ng each of MO1 and MO2 per embryo, morphants showed general developmental defects, including a small head and curled trunk (data no shown).

Fig. 3.

Location of splice-blocking MO1 and MO2 and the resulting zebrafish max-1 splicing variants. Above the gel is the putative protein structure of zebrafish max-1. The PH domain corresponds to exons 10, 11, and 12. The gene structure of zebrafish max-1 from exon 5 to exon 14 is shown. MO1 binds to the exon 9/intron 9 boundary, and MO2 targets the exon 8/intron 8 boundary. In morphant, no wild-type mRNA (band a) was detected by RT-PCR. Band a is wild-type mRNA. Bands b and c lose part of coding sequence (black shadows) and result in a premature stop codon before the PH domain.

Specificity and efficiency of splice-blocking were determined by RT-PCR with primers corresponding to exons flanking the MO target sites. Although injection of each MO reduced the max-1 transcript (data not shown), coinjection of MO1/MO2 resulted in complete loss of the expected wild-type fragment (1.8 kb; Fig. 3, band a) and generation of two additional smaller bands (Fig. 3, bands b and c). Sequencing showed that abnormal transcripts had lost exon 9 for the band b and exon 8 and exon 9 for band c (Fig. 3). The resulting aberrant mRNA transcripts contained translational frame shifts and premature stop codons ahead of the PH domain, suggesting that no dominant-negative proteins were formed. Zebrafish ef-1α was used as an RT-PCR control, showing that equal amount of PCR product was amplified, and no abnormal splicing was observed (Fig. 3).

At 1 dpf, injection of MO1 and MO2 resulted in missing or misdirected ISVs (Fig. 4B). Injected embryos develop slightly slower than the control group, but overall embryonic structures appeared normal. As development continued, those ISV missed at 1 dpf eventually sprouted from the dorsal aorta, but most did not follow the normal stereotypical intersomitic pathway, following instead an irregular network (Fig. 4D). At 3 dpf, ≈83.7% of the morphants still showed an aberrant ISV pattern that persisted until at least 5 dpf, the last time point observed (Fig. 4L).

Fig. 4.

Zebrafish max-1 knockdown causes defects in ISV patterning but not in neuronal axon guidance. Embryos are anterior to the left and dorsal to the top. A–H and K are flk-1:GFP embryos. I and J are gata2:GFP embryos. (A) Control-injected embryo at 1 dpf. (B) max-1 knockdown morphant at 1 dpf. The white star indicates a missing ISV, and the white arrowhead points to an aberrant ISV. (C) Control embryo at 3 dpf. (D) max-1 knockdown morphant at 3 dpf. Most ISVs are aberrantly patterned. (E) Zebrafish max-1 rescue. The MOs were coinjected, with 24 pg of zebrafish max-1 mRNA per embryo. ISV patterning appears normal. (F) C. elegans max-1 rescue. The MOs were coinjected, with 24 pg of C. elegans max-1 mRNA per embryo. (G) max-1 morphant at 2 dpf. White arrowheads point to the ISVs that do not connect to the DLAV. (H) Angiography with TRITC in the same embryo in G. White arrowheads point to the same ISVs as in G. No circulation was observed at those two positions. (An embryo with a less severe phenotype was used to compare connected and nonconnected ISVs.) (I) gata2:GFP control embryo at 30 hpf. The motor neuron axon pattern is normal. (J) max-1 MO knockdown in a gata2:GFP embryo at 30 hpf. The motor neuron axon pattern is normal. (K) Partial rescue of morphant ISV phenotype with zebrafish ephrinb3 mRNA:MO coinjected, with 32 pg of zebrafish ephrinb3 mRNA per embryo. Most ISVs connect to the DLAV. However, some still show an aberrant pattern. (L) Quantitative comparison of zebrafish max-1 knockdown by MO1 and MO2 coinjection and phenotypic rescue at 3 dpf. Of the morphant embryos (n = 106), 83.7% showed a defective phenotype, and 16.3% showed a normal phenotype. Of the zebrafish max-1 (z.max-1) rescue embryos (n = 109), 12.3% showed a defective phenotype, 60.8% were rescued (normal), and 26.9% were partially rescued. Of the max-1 rescue embryos (n = 122), 16.3% showed a defective phenotype, 33.6% were rescued (normal), and 50.1% were partially rescued. Of the ephrinb3 rescue embryos (n = 135), 23.1% showed a defective phenotype, 18.4% were rescued (normal), and 58.5% were partially rescued. Error bars show the standard errors (P < 0.001).

To confirm the specificity of MO1 and MO2, zebrafish or C. elegans max-1 mRNA was coinjected with the morpholino to determine whether they could rescue the morphant phenotype. As shown in Fig. 4, both zebrafish and C. elegans max-1 mRNA rescued the morphant ISV phenotype (Fig. 4 E and F). Of the embryos injected with zebrafish max-1 mRNA, 60.8% showed complete rescue and 26.9% showed partial rescue. Of the embryos injected with C. elegans max-1, 33.6% showed complete rescue and 50.1% showed partial rescue (Fig. 4L).

We then performed microangiography with TRITC to determine whether the aberrant ISV could support circulation. As shown in Fig. 4 G and H, ISVs that connected to both the dorsal aorta and the dorsal longitudinal anastomotic vessels (DLAV) contributed to blood circulation, whereas truncated ISV blocked circulation.

The effect of MO1 and MO2 was specific to vascular structures because the somites and motor neuron patterning in the trunk region appeared normal. The motor neurons were visualized in live gata2:GFP transgenic zebrafish embryos (Fig. 4 I and J). In zebrafish injected with the higher amount of morpholino (10.2 ng each of MO1 and MO2), motor neuron axons also started to show abnormal patterns (data not shown).

Max-1 Genetically Interacts with ephrinb3 in Zebrafish.

Several neural guidance molecules have been shown to be involved in patterning of angiogenesis. Previous studies suggested that sema3Ab (16), ephrinb1, ephprinb2a, ephrinb2b, and ephrinb3 (17) were expressed in the zebrafish somites during ISV formation. We therefore selected these genes as candidates for potential rescue of the max-1 morphant phenotype by coinjection of their mRNA transcripts with the zebrafish max-1 MOs. Ephrinb3 was the only gene tested that could partially rescue a max-1 morphant (Fig. 4 K and L).

From 14 hpf to 24 hpf, ephrinb3 is expressed in the ventral somites and ventral mesenchymal cells, a similar expression pattern to max-1 in somites. Biochemical studies have established that ephrin family genes encode transmembrane proteins. To test whether max-1 plays a role in mediating the subcellular localization of the ephrinb3 protein, we produced an ephrinb3-GFP fusion protein construct and confirmed that this fusion protein indeed was predominantly localized on the plasma membrane of cells from dissociated embryos. Coinjection of mRNA encoding the ephrinb3-GFP fusion protein and MO1 plus MO2 resulted in abnormal distribution of the ephrinb3-GFP fusion protein in cells isolated from injected embryos at 24 hpf. Compared with control-injected embryos, the fusion protein formed more aggregates in cytoplasm of cells from morphant embryos (Fig. 5). This effect was specific to ephrinb3-GFP fusion protein because subcellular localization of RFP membrane protein, an internal control, remained normal in the same cells where ephrinB3-GFP was affected.

Fig. 5.

ephrinb3-GFP fusion protein localization in control and max-1 morphant zebrafish cells. Control cells contained 47 pg of ephrinb3-egfp mRNA, memb-mCherry mRNA, and 10.2 ng of MAX1MO4mis per embryo. Both GFP and red fluorescent protein (RFP) signals are evenly localized on the cell membrane. MAX1MO4mis is a control MO that has no phenotype at this concentration. Morphant cells contained 47 pg of ephrinb3-egfp mRNA, memb-mCherry mRNA, and 5.2 ng of MO1/MO2 per embryo. GFP expression aggregates are shown in the cytoplasm, whereas RFP is evenly localized on membrane. (A and B) Detection under GFP filter. (C and D) Detection under RFP filter. (E and F) Superimposition of GFP and RFP images. White arrowheads in B and F point at GFP aggregation. (G) Quantitative comparison of control and morphant. of the control cells (n = 550), 68.7% have a GFP signal only on the membrane and 31.3% have a GFP signal both on the membrane and in the cytoplasm. Of the morphant cells (n = 551), 49.5% have a GFP signal only on the membrane, and 50.5% have a GFP signal both on the membrane and in the cytoplasm. In the key for G, membrane means that the GFP signal is only on membrane, and cytoplasm means that the GFP signal is both on the membrane and in the cytoplasm. Error bars show standard errors (P < 0.001).

Discussion

In C. elegans, max-1 is involved in motor neuron axon guidance and acts genetically downstream of the netrin pathway (14). In this report, we identified and analyzed expression and function of the max-1 gene in zebrafish. Embryos that had been injected with two splice-blocking MOs (MO1 and MO2; 5.2 ng each per embryo) lacked functional max-1 mRNA transcripts yet developed without any apparent general defects. Approximately 83.7% of these morphants exhibited abnormal ISV patterning (Fig. 4L). Fish injected under the same conditions showed no obvious defects in motor neuron axon guidance (Fig. 4 I and J). In zebrafish injected with the higher amount of morpholino (10.2 ng each of MO1 and MO2), both ISVs and motor neuron axons appeared abnormally patterned. It is therefore possible that MAX-1 also plays a role in neuronal axon guidance in vertebrates.

We focused on analysis of the more specific function of max-1 on ISV patterning seen in zebrafish injected with the lesser amount of morpholino. In control embryos, individual ISVs are restricted ventrally to the border between adjacent somites. At the dorsal–ventral midpoint, the ISV leaves the somite border and runs directly dorsal to connect with the DLAV (3). However, in morphants, ISVs are often truncated or branch out aberrantly, without following a specific pattern, into adjacent somites. We therefore speculated that certain guidance signals might be disturbed in max-1 morphants.

Several signaling molecules that function in both axon guidance and angiogenesis have been reported in zebrafish, including netrin, semaphorin, robo, neuropilin, and ephrin proteins. To examine whether any of these proteins interact with max-1, we selected a few that were reported to be expressed in somites during the same developmental time frame that max-1 is expressed as candidates that might rescue the max-1 morphant phenotype. Among these candidates, zebrafish ephrinb3 was found capable of partially rescuing the ISV phenotype of max-1 morphants. We confirmed that the expression of ephrinb3 and max-1 indeed overlaps in the ventral somites of zebrafish embryos from the 10-somites stage to 26 hpf (Fig. 2 C, F, H, and L–P). These studies suggest that ephrinb3 functions downstream of max-1 to regulate ISV patterning. Given that ephrin proteins have been implicated in tissue patterning and boundary formation through repulsive guidance cues (18, 19), we propose a hypothesis that ISV patterning in the trunk of zebrafish involves repulsion mechanisms mediated by MAX-1 and the ephrin protein pathway. We have attempted to knock down ephrinb3 by morpholino injection, but so far the data are inconclusive because of our inability to confirm the effectiveness of the ephrinb3 morpholino and perhaps because of the functional redundancy of additional ephrin proteins expressed in the somites.

Our hypothesis that ephrin signaling is involved in ISV patterning in zebrafish is consistent with previous studies showing that Eph receptors and ephrins can trigger an adhesive response of endothelial cells and are required for the remodeling of blood vessels (20). Biochemical studies suggest that the extent of multimerization of Eph receptors and ephrins on the cell membrane modulates the cellular response. MAX-1 was shown to be localized beneath cytoplasm membrane mediated by its FERM domain (14). In addition, PH domains are often involved in targeting proteins to the plasma membrane (21). Our studies showed that an ephrinb3-GFP fusion protein is normally localized on the membrane and that coinjection of mRNA encoding this fusion protein and MO1 and MO2 caused excess aggregation of GFP signal in cytoplasm. This finding suggests that appropriate membrane association of ephrin proteins may require proper function of MAX-1 in zebrafish. Overall, our studies have identified MAX-1 as a critical component of the guidance pathway underlying endothelial cell migration and blood vessel patterning and further strengthen the similarity between vascular and neuronal developmental regulatory mechanisms.

Materials and Methods

Zebrafish Stocks.

The wild-type AB strain zebrafish, gata2:GFP (22, 23) and flk-1:GFP transgenic (24), was used in this study. Zebrafish embryos were staged as described by Kimmel et al. (25).

Molecular Cloning of Zebrafish max-1 and Phylogeny Analysis.

The Ensembl zebrafish database was searched with the translated protein sequence from C. elegans max-1, and a predicted protein (ctg10711.1.70473.135987) sharing sequence homology with max-1 was identified. Primers were designed according to the sequence of ctg10711.1.70473.135987 and used in RT-PCR to isolate the zebrafish max-1 cDNA from total RNA isolated from 18-somites-stage embryos. In addition, 5′RACE (First Choice RLM-RACE kit, Ambion, Austin, TX) was used to determine the 5′ end of the zebrafish max-1 mRNA, and a full-length putative cDNA clone was constructed. The nucleotide sequence of zebrafish max-1 was submitted to GenBank, (accession no. DQ501275). Clustalx (version 1.83) was used for the phylogenic analysis (26).

Whole-Mount in Situ Hybridization.

Digoxigenin-labeled antisense RNA probes were generated in vitro by using the zebrafish max-1 and ephrinb3 cDNA as templates with T3 or T7 RNA polymerase (Promega, Madison, WI). Whole-mount RNA in situ hybridizations were performed essentially as described by Westerfield (27) on embryos at the following developmental stages: bud, 3 somites, 10 somites, 14 somites, 15 somites, 21 somites, 26 somites, 1 dpf, 28 hpf, 2 dpf, 3 dpf, 4 dpf, and 5 dpf.

Morpholino Design, Phenotype Analysis, and Detection of Splice-Blocking Variants of Zebrafish max-1 mRNA with RT-PCR.

Two splice-blocking MOs of zebrafish max-1, MO1 and MO2, were synthesized by Gene Tools (Philomathe, OR). The sequence of MO1 is 5′ TAATTCTGTCATCTTACCCAAGAGG3′, which corresponds to the putative exon 9/intron 9 boundary of zebrafish max-1; the sequence of MO2 is 5′GGAGATGCTGATGTACCTGCAGCAGAGC3′, which targets the putative exon 8/intron 8 junction of zebrafish max-1.

MO1 and MO2 were dissolved in nuclease-free water at 0.2 mM and injected into 1- to 2-cell-stage embryos (5.2 ng for each oligo per embryo or 2.6 ng of both oligos). The amount of injection was determined by measuring the volume of liquid injected (20 times) into a 1-μl capillary glass (34 mm long) using a ruler, and volume per microinjection was thereby calculated. Injected embryos were cultured in fish water at 28.5°C.

The defective phenotype was defined as one having five or more ISVs absent or misdirected in the region above the yolk extension and tail. The normal or rescued phenotype was defined as having no more than two ISVs absent or misdirected in the same region. The partially rescued phenotype was defined as having three to four ISVs absent or misdirected in the region from yolk extension to tail.

At 26 hpf, total RNA was extracted from the morphant (MO1 and MO2 coinjected embryos) and control embryos with a Qiagen RNeasy kit. Reverse transcription was carried out with random decamers by using SSII reverse transcriptase (Invitrogen, Carlsbad, CA). Two rounds of PCR were performed to detect splice-blocking variants of zebrafish max-1 mRNA. The primers used were S6 (base pairs 600–622), 5′-ACATGCAGCCTAGCGAAACAGAA-3′; R11 (base pairs 2,403–2,382), 5′-ATCACGTCACTCGGCGACTTAT-3′; and R12 (base pairs 2,452–2,431), 5′-GGCAATATGACACGAGGAGTTG-3′. RT-PCR products were cloned and sequenced. Zebrafish ef-1α (GenBank accession no. NM_131263) was used as an internal control. The ef-1α primers used were ef-1α forward (base pairs 496–516), 5′-TCACCCTGGGAGTGAAACAGC-3′, and ef-1α reverse, (base pairs 1,188–1,168), 5′-ACTTGCAGGCGATGTGAGCAG-3′. The product size of ef-1α is 692 bp.

A translation-blocking MO, MAX1MO (5′-TCTCTGTTCCAGCTCCTGCATCTAC-3′), and a 4-bp mismatch control MO, MAX1MO4mis (5′-TCTCTcTTCCAcCTCCTcCATgTAC-3′), were generated by Gene Tools. These oligos were used as controls for injection because neither of them had any defective phenotype at 10 ng per embryo.

mRNA Preparation.

pXT7, a derivative of pGEM4Z (Promega, Madison, WI) and pSP64T (28), was used to make mRNA. A region of the zebrafish max-1 cDNA containing the coding sequence (base pairs 503-4660) was subcloned into pXT7. The C. elegans max-1 cDNA clone was provided by Y. Jin (University of California, Santa Cruz); memb-mCherry construct encodes a membrane red fluorescent protein and was a gift from S. Megason (California Institute of Technology, Pasadena); zebrafish ephrin cDNA clones (ephrinb1, ephrinb2a, ephrinb2b, and ephrinb3) were kindly provided by J. Chan (Children's Hospital, Boston, MA). Ephrinb3 was subcloned in frame to the 5′ end of EGFP (Clontech, Mountain View, CA) to generate a fusion protein. A zebrafish Sema3Ab clone was a gift from C. Becker (Zentrum für Molekulare Neurobiologie, Hamburg, Germany). Sense transcripts from the cDNA clones were generated by using the mMESSAGE mMACHINE kit (Ambion, Austin, TX) and were purified with the RNeasy Mini kit (Qiagen, Hilden, Germany).

Microangiography.

TRITC dextran at 20 mg/ml (molecular weight 70,000; Research Organics, Cleveland, OH) dissolved in double distilled water was microinjected into the sinus venosus/cardinal vein of zebrafish embryos at ≈48 hpf essentially as described by Weinstein and coworkers (29) and Chen et al. (30).

Cell Dissociation.

Dechorionated embryos were deyolked with a forceps and digested in 0.25% trypsin (Sigma, St. Louis, MO) and 1 mM EDTA in PBS (13.7 mM NaCl/0.27 mM KCl/0.43 mM Na₂HPO₄/0.14 mM KH₂PO₄, pH 7.3) at 37°C for 20 min with periodic agitations. Then the cells were centrifuged at 1,000 × g at 4°C for 2 min and washed with PBS two times.

Imaging.

Images of zebrafish embryos were acquired by using a Axioplan 2 microscope (Zeiss, Oberkochen, Germany) equipped with a C4742–95 digital camera (Hamamatsu, Tokyo, Japan) or an AxioCam digital camera and OpenLab (Improvision, Lexington, MA) software, then edited with PhotoShop 7.0 (Adobe Systems, San, Jose, CA).

Statistical Analysis.

Phenotypes of morphants, rescue results by mRNA injections, and cellular sublocalization of the ephrinb3 protein were analyzed by a two-way χ² test. The same experiments with similar numbers of embryos or cells were repeated three times.

Abbreviations

DLAV

dorsal longitudinal anastomotic vessels

dpf

days postfertilization

hpf

hours postfertilization

ISV

intersegmental blood vessels

MO

morpholino oligo