The study was conducted conforming to the local institutional laws and the Chinese law for the protection of animals. All adult zebrafish (Dario rerio) were maintained under standard conditions in accordance with our previous protocols (Huang et al., 2013; Xu et al., 2014). The AB/WT, Tg(kdrl:ras-mCherry) and Tg(mnx1:EGFP) zebrafish used in this article have been described previously (Nicoli et al., 2012; Jao et al., 2013). Zebrafish embryos after 24 hpf were treated with 0.2 mM 1-phenyl-2-thio-urea to prevent pigment formation.

The zebrafish sema6D protein sequences (NP_998164.3) and zebrafish sema6dl sequences (XP_005173705.1) together with other nine species, including Bos taurus (XP_024853302.1), Gallus gallus (XP_040562440.1), Homo sapiens (XP_024305841.1), Mus musculus (XP_030105194.1), Oryctolaguscuniculus (XP_017203544.1), Oryziaslatipes (XP_011486693.1), Rattus norvegicus (XP_038960910.1), Sus scrofa (XP_020951762.1), and Xenopus tropicalis (XP_031753477.1) were got from NCBI and used for phylogenetic analysis. The alignments of these sequences were constructed by MEGA7and the phylogenetic tree was constructed by ML.

Total RNA of zebrafish embryos at various stages was extracted with TRizol according to the manufacturer’s instruction (Invitrogen, Waltham, MA, USA) and genomic contaminations were removed by DNaseI. Quantity of isolated RNA was verified using gel electrophoresis and Nanodrop, followed by cDNA synthesis using Transcriptor First Strand cDNA Synthesis Kit (Roche), and then was stored at −20°C.

Primers for PCR were designed by the Primer Premier six software and listed as following:

sema6D-QF: 5’-CCTCCTCCTATTCCTCTTCTGTT-3’;

sema6D-QR: 5’-ACTACGGTGCGGTTCTTATGA-3’;

ef1a–QF: 5’-GAGTTGTGCCGTACATCAG-3’;

ef1a–QR: 5’-CGTGAGAGTACATGGTCATG-3’.

Quantitative RT-PCR was conducted in a total 20 μl reaction volume with 10 μl SYBR premix (TIANGEN). The relative RNA amounts were calculated with the comparative CT (2-DDCT) method and normalized with elongation factor 1-alpha (ef1a) as the reference. Whole-mount in situ hybridization (WISH) with antisense RNA probes was synthesized as described previously (Wang et al., 2016). The cDNA fragments used for sema6D RNA probe transcription as templates were amplified using the forward primer 5’-CGACGGCTATCACTTCACTCT-3’and reverse primer 5’-TGGAACATTCTGACGGCTCTT-3’. Then a 548bp sequence of sema6D was inserted into pGEM-T-easy vector. Digxigenin (DIG)-labeled sense and antisense probes were performed from the linearized pGEM-T-easy plasmids using the DIG RNA Labeling Kit (Roche).

Whole-mount in situ hybridization (WISH) was performed according to our previous procedures (Huang et al., 2013). Digoxigenin-labeled antisense probes were constructed as described above. Zebrafish embryos without pigment at different developmental stages were collected and fixed with 4% PFA overnight at 4°C. After incubated with the probe overnight, an alkaline phosphatase-conjugated antibody against digoxigenin and AP-substrate NBT/BCIP solution (Roche, Switzerland) was used to detect the digoxigenin-labeled RNA probe.

Splicing-blocking Morpholino (5’- TGTGAGCTGAGTGAATGCAGACCT -3’) that was specific for sema6D gene was synthesized by Gene Tools. The Morpholino was diluted to 0.3 mM with RNase-free water. The single cell stage embryos of Tg(kdrl:ras-mCherry) and Tg(mnx1:EGFP) zebrafish were obtained for microinjections as described previously (Wang et al., 2016). Then, the embryos were raised in E3 medium at 28.5°C for following imaging.

Cas9 mRNA was obtained by in vitro transcription with the linearized plasmid pXT7-Cas9 according to the procedure previously described (Nakayama et al., 2013). The sema6D guide RNA (gRNA; 5’-GGCGTGGCAGAAGTAATGAGTGG-3’) was designed and synthesized followed the previously reported (Chang et al., 2013). Transgenic zebrafish lines Tg(kdrl:ras-mCherry) and Tg(mnx1:EGFP) were natural mated to obtain embryos for microinjection. One to two-cell stage zebrafish embryos were injected with 2–3 nl of a solution containing 250 ng/μl Cas9 mRNA and 15 ng/μl sgRNA (Gong et al., 2017). At 72 hpf, 10 zebrafish embryos were randomly collected and mixed for genomic DNA extraction according to the previous methods (Gong et al., 2017). Then, the amplicons from the genomic DNA were cloned into pGEM-T-easy vector and after transformation, 50 clones were selected randomly for sequencing. The inserted fragments were sequenced by using the forward primer 5’-CCTGTGCATATAGATTGTTG-3’ and reverse primer 5’-AAGTCTACAGACAGTAACG-3’. The mutation efficiency of Crispr-Cas9 was then calculated based on the sequencing results (Wu et al., 2018).

Full-length and truncated coding sequences of sema6D were synthesized and inserted into PCS²⁺ vector as templates for in vitro transcription. The mRNA synthesis was carried out by using the mMESSAGEmMACHINESp6 Ultra Kit (Ambion) and purified with the MEGAclearTM Transcription Clean-Up Kit (Ambion) before the injection. Finally, 2 nl capped mRNA was co-injected with sema6D Mo into one-cell stage embryos. The fli1a:sema6D plasmid was constructed by LR recombination as described in the LifetechMultiste Gateway Manual (Life Technologies, Carls-bad, CA, USA). Then, the construct was injected into one cell stage embryos of Tg(mnx1:EGFP::kdrl:ras-mcherry) zebrafish for tissue specific rescue experiments (1 ng per embryo).

After being anesthetized with tricaine, the zebrafish embryos were mounted in 0.8% low melt agarose and then photographed by Leica TCS-SP5 LSM confocal microscope. For the in situ hybridization, Photographs were taken using an Olympus stereomicroscope MVX10. Statistical analyses were performed by one-way analysis of variance (ANOVA) and the Mann-Whitney test. Statistical differences were considered significant for P-values <0.05.

To analyze the homology of zebrafish sema6D with other homologous genes, the multiple alignments and phylogenetic analyses of sema6D/sema6Dl were performed with species including Danio rerio, Bostaurus, Gallus gallus, Homo sapiens, Musmusculus, Oryctolaguscuniculus, Oryziaslatipes, Rattusnorvegicus, Sus scrofa, and Xenopus tropicalis. As shown in Figures 1A,B, the sema6D proteins are significantly conserved during evolution, especially for the sema domain, suggesting their important functions. In addition, zebrafish sema6D was clustered in a separate clade with Oryziaslatipes and was close to Xenopus tropicalis in the phylogenetic tree constructed by complete amino acid sequences of all the sema6D proteins above (Figure 1C). Interestingly, the sema6Dl, a sema6D isoform gene in zebrafish, was clustered with sema6D from Oryziaslatipes, suggesting it may not be functionally related to sema6D in zebrafish.

To explore the roles of sema6D during zebrafish embryonic development, the expression level of sema6D was studied using QRT-PCR and WISH. According to QRT-PCR results, sema6D exhibited steady expression from 24 hpf to 96 hpf, with the highest level at 72 hpf, followed by 96 hpf (Figure 2A). Then, the spatial expression pattern of sema6D in early embryonic development was further studied by WISH. From 24 hpf to 48 hpf, sema6D was mainly expressed in the nervous system and also in the blood vessels (Figures 2B–D). Its expression in the nervous system is much restricted to the brain and head structures, being very weak in the spinal cord. Its expression in the vascular system was mainly observed in the ISVs. To further analyze the expression of sema6D in the zebrafish vessels, the endothelial cells from Tg(fli1a:EGFP) were sorted for RT-PCR (Figure 2E). The results showed that both fli1a and sema6D were detected in the selected endothelial cells, which was consistent with the previously reported single-cell RNA sequencing data of zebrafish endothelial cells (Shi et al., 2020; Figure 2F; Supplementary Table S1). Besides, mef2aa, which is specifically expressed in somite and heart, was chosen as a negative control to validate the purity of selected cells (Lv et al., 2017). The results showed that no mef2aa signals were detected in the sorted EGFP-positive cells (Figure 2F). Taken together, these results suggested that sema6D might participate in the development of zebrafish vessels and nerves.

Considering the significant expression of sema6D in the zebrafish brain, it is rational to speculate it might modulate the development of neural system. To validate the hypothesis, specific morpholino antisense oligonucleotide (sema6D-MO) was used to knockdown the expression of sema4D in Tg(mnx1:EGFP) transgenic zebrafish. The results of cDNA amplification provided evidence that the injection of sema6D-MO efficiently altered the transcription pattern of sema6D (Supplementary Figure S1). Themorphology of PMNs was examined by confocal microscopy at 48 and 72 hpf. Although the morphant zebrafish embryos grossly appeared normal, the deficiency of sema6D caused dramatic developmental defects of PMNs (Figure 3A). Compared with the controls, the development of caudal primary motor neurons (Caps) was significantly inhibited at 48 hpf, where the average length of CaPs in the sema6D morphants was particularly shorter than that of the control groups (Figure 3B). At 72 hpf, although the length of CaP in the morphants were similar to the controls, the axonal trajectories of PMNs were significantly misled and many of the truncated axons in the sema6D deficient zebrafish could not recover completely (Figure 3C). Moreover, the number of CaP branches in the morphants decreased obviously at 72 hpf (Figure 3D). Taken together, these results suggest that sema6D is required for the development and navigation of neural networks.

Since sema6D was found to express in zebrafish vessels, we speculated that it might participate in the development of blood vessel. To further explore the roles of sema6D in blood vessel formation, the morphology of ISVs in sema6D knockdown zebrafish was observed by confocal microscopy at different stages. The results showed that the deficiency of sema6D resulted in significant path finding defects of zebrafish ISVs, which was consistent with observation in the development of motor neuron. In control groups, ISVs grew from the dorsal aorta and reached the dorsal roof in an orderly manner to form dorsal anastomotic vessels (DLAV). In contrast, the ISVsin sema6D deficiency zebrafish grew upwards halfway, then turned to connect with adjacent ISVs disorderly, and could not form complete DLAV (Figures 4A,B). In addition, a small number of ISVs which grew only halfway or even less usually failed to cross the horizontal myoseptum in the sema6D knockdown zebrafish (Figure 4C). These results indicated that sema6D seems not necessary for the initial stages of ISVs sprouting, but rather regulates the vascular patterning.

In order to confirm that sema6D is required for the development of PMNs and ISVs, the CRISPR/Cas9 system was utilized to knockout sema6D in Tg(mnx1:EGFP::kdrl:ras-mCherry) transgenic zebrafish line. In order to ensure complete disruption of functional proteins, the target sites near and downstream of the translation start codon (ATG) of sema6D coding sequence were selected for gRNAs design (Figure 5A). The editing efficiency and the knockout patterns of the selected gRNA-Cas9 system were identified by sequencing (Figures 5B,C). It was found that the phenotypes of PMNs and ISVs in the F0 generation of sema6D knockouts were consistent with the deficient morphants by confocal imaging analysis (Figure 5D). The PMNs were shorter and the axonal trajectories were apparently misled in the F0 knockouts (Figure 5E). Besides, the disorganized vasculature was also observed (Figure 5F). Furthermore, injection of sema6D gRNA without cas9 caused no obvious developmental defects, confirming the phenotype was a specific consequence of sema6D knockdown.

In order to confirm the defects of vascular and neuronal development were specifically caused by sema6D deficiency, the in vitro synthesized sema6D mRNA and sema6D-Mo were co-injected into one cell stage zebrafish embryos. Confocal imaging analysis revealed that the overexpression of sema6D mRNA could greatly rescue the defective phenotypes of PMNs and ISVs (Figures 6A–C) in sema6D deficient embryos. In addition, single sema6D mRNA injection also caused phenotypes of PMNs and ISVs, which are similar to those in deficient morphant (Figures 6A–C). This result validated that sema6D could regulate growing guidance of vessels and neurons in zebrafish. To further investigate the consequences of tissue specific rescue, sema6D was over-expressed in morphants with the fli1a promoter. Comparing with the morphant, embryos co-injected with sema6D-MO and fli1a:sema6D plasmid exhibited rescue phenotypes (Figures 6D–F). Taken together, these results suggested that endothelial-derived sema6D were involved in both neural and vascular development.