In order to efficiently derive hematopoietic stem cells (HSCs) from pluripotent precursors, it is crucial to understand how mesodermal cells acquire hematopoietic and endothelial identities: two divergent, but closely related, cell fates. Although Npas4 has been recently identified as a conserved master regulator of hemato-vascular development, the molecular mechanisms underlying cell fate divergence between hematopoietic and vascular endothelial cells are still unclear. Here, we show in zebrafish that mesodermal cell differentiation into hematopoietic and vascular endothelial cells is regulated by Junctional adhesion molecule 3b (Jam3b) via two independent signaling pathways. Mutation of jam3b led to a reduction in npas4l expression in the posterior lateral plate mesoderm and defects in both hematopoietic and vascular development. Mechanistically, we show that Jam3b promotes endothelial specification by regulating npas4l expression through repression of the Rap1a-Erk signaling cascade. Jam3b subsequently promotes hematopoietic development, including HSCs, by regulating lrrc15 expression in endothelial precursors through the activation of an integrin-dependent signaling cascade. Our data provide insight into the divergent mechanisms for instructing hematopoietic or vascular fates from mesodermal cells.
Hematopoietic stem cells (HSCs) can produce all types of mature blood cells throughout the lifespan of an individual and are the therapeutic component of bone marrow transplantation. One of the major goals of regenerative medicine is to derive HSCs from induced pluripotent stem cells (iPSCs). The developmental mechanisms of HSCs in the embryo are of great interest, because elucidation of this process can provide insight into the cellular and molecular cues needed to instruct and expand HSCs from iPSCs (Murry and Keller, 2008). In vertebrates, HSCs arise from a subset of endothelial cells, termed hemogenic endothelial cells, in the ventral floor of the dorsal aorta (DA) (Boisset et al., 2010; Bertrand et al., 2010; Kissa and Herbomel, 2010). Studies in mouse and zebrafish embryos have greatly contributed to our understanding of the cellular and molecular programs controlling HSC development. Some transcription factors, such as Runx1, c-Myb (also known as Myb), Scl (Tal1) and Gata2, have been directly implicated in the regulation of HSC formation and maintenance during embryogenesis (Clements and Traver, 2013). It is believed that hematopoietic identity is conferred through the expression of these transcription factors within endothelial precursors. However, the molecular mechanisms underlying hematopoietic specification are still largely unknown, leading to difficulties in efficient derivation of hematopoietic lineages from iPSCs.
Hematopoietic and vascular endothelial cells develop in close proximity to each other within the mesoderm. It has been proposed that these two cell populations share a common precursor, termed the ‘hemangioblast’ (Choi et al., 1998; Huber et al., 2004; Vogeli et al., 2006). The zebrafish cloche mutant, in which both hematopoietic and vascular development are defective, has been identified as a unique model for the study of hemangioblast development (Stainier et al., 1995). Owing to the reduction of various endothelial and hematopoietic marker genes in this mutant embryo, the cloche gene is thought to act at the top of the hierarchy of hemangioblast specification (Sumanas et al., 2005). Recently, it has been determined that the cloche gene encodes a PAS-domain-containing bHLH transcription factor, neuronal PAS domain protein 4 like (Npas4l). Importantly, enforced expression of its human ortholog, NPAS4, can rescue the vascular defect of cloche mutant embryos, suggesting that NPAS4 plays a central role in hemato-vascular development in an evolutionarily conserved manner (Reischauer et al., 2016).
Junctional adhesion molecule (Jam) proteins belong to the immunoglobulin (Ig) superfamily of cell adhesion molecules and consist of two Ig-like domains, a transmembrane domain and a type II PDZ-binding motif. Jam3 (also known as Jam-C) is widely expressed on the epithelium, endothelium and some leukocytes, and has been implicated in tight junction formation and barrier function as well as leukocyte migration (Ebnet, 2017). Heterophilic interactions between Jam2 (also known as Jam-B) and Jam3 play an important role in leukocyte transmigration (Weber et al., 2007), spermatozoal differentiation (Gliki et al., 2004) and myocyte fusion (Powell and Wright, 2011). In mice, Jam3 is expressed on adult HSCs and promotes the adhesion of HSCs to Jam2-expressing stromal cells in the bone marrow (Arcangeli et al., 2011). Inhibition of Jam3 by a blocking antibody affects the homing and reconstitution ability of hematopoietic stem/progenitor cells (HSPCs) (Arcangeli et al., 2014). Jam3 also mediates an increase in endothelial permeability through the repression of Rap1, a member of the Ras family of GTPases that can enhance the adhesive contact between adjacent endothelial cells (Orlova et al., 2006). Although Jam3 functions in both hematopoietic and vascular endothelial cells, to date there has been no reported role for Jam3 in the development of hematopoietic or vascular endothelial cells.
Here, we show that a zebrafish ortholog of Jam3, Jam3b, plays dual roles in hemato-vascular development during early embryogenesis: it promotes endothelial specification in mesodermal cells via the regulation of npas4l, and it confers the hematopoietic fate in endothelial precursors through the regulation of Leucine rich repeat containing 15 (Lrrc15). In vertebrates, mesodermal induction requires fibroblast growth factor (FGF)-dependent extracellular signal-regulated kinase (Erk) signaling (Umbhauer et al., 1995; Kimelman, 2006; Xiong et al., 2015), a component of the mitogen-activated protein kinase (MAPK) cascade. We found that Jam3b suppresses the activation of Erk to enhance npas4l expression in the lateral plate mesoderm (LPM). Jam3b subsequently induces lrrc15 expression in endothelial precursors through an integrin-dependent signaling cascade, driving them to acquire the HSC fate.
Jam3b is required for both hematopoietic and vascular development
To investigate whether jam3b is expressed in hematopoietic and vascular endothelial cells in zebrafish embryos, we performed whole-mount in situ hybridization (WISH) using a jam3b-specific probe. The expression of jam3b was undetectable at 6 h post-fertilization (hpf), but was seen at ∼8 hpf, mainly in the dorsal part of the embryo (Fig. S1A). At 14 hpf, we could detect jam3b expression in the posterior LPM (PLPM), from which hematopoietic and endothelial precursors arise (Fig. S1B). Fluorescent WISH and immunofluorescence analysis of fli1a:GFP embryos, in which endothelial cells are labeled with GFP, showed that the expression of jam3b partially overlapped with fli1a:GFP at 14 hpf (Fig. S1C), indicating that jam3b is expressed in endothelial precursors. The expression of jam3b was also detected in the DA, particularly in the ventral floor of the DA (Fig. S1D), the site of HSC emergence. These data indicate that jam3b is expressed in endothelial/hematopoietic tissues during the time frame of hemato-vascular development.
To investigate the role of Jam3b in the hemato-vascular development, we utilized a jam3b mutant zebrafish line, jam3bsa37, which contains a missense point mutation in a cysteine residue (C136Y) that is predicted to form a structurally critical disulphide bond (Fig. S1E) (Powell and Wright, 2011; Weber et al., 2007). We first examined the expression of HSC marker genes, runx1 and cmyb, in homozygous jam3bsa37 embryos. The expression of runx1 at 30 hpf and cmyb at 36 hpf was detected in the DA of wild-type (WT) embryos, but expression of both genes was largely reduced in jam3bsa37 embryos (Fig. 1A). The expression of runx1 in the DA of jam3bsa37 embryos could be partially rescued using an exogenous WT jam3b mRNA (Fig. S1F). Developing HSPCs can be visualized as cd41:GFP; kdrl:mCherry double-positive cells in the ventral floor of the DA at 52 hpf (Clements et al., 2011). As shown in Fig. 1B, the number of double-positive cells was approximately four times lower in jam3bsa37 embryos compared with WT embryos. After budding from the DA, nascent HSPCs migrate to and colonize the caudal hematopoietic tissue (CHT), an equivalent tissue to the mammalian fetal liver in which HSPCs expand (Tamplin et al., 2015), and the thymus in which T lymphocytes are produced (Trede and Zon, 1998). The expression of cmyb in the CHT and rag1 (a T cell marker) in the thymus was nearly undetectable in jam3bsa37 embryos at 4 days post-fertilization (dpf) (Fig. 1C). These results indicate that Jam3b is required for HSC formation in the developing embryo.
In zebrafish, the N-terminal-truncated isoform of scl, scl-β, is expressed in hemogenic endothelial cells and shown to act upstream of runx1 (Qian et al., 2007; Zhen et al., 2013). A paralog of zebrafish gata2 genes, gata2b, is also known to regulate runx1 expression in hemogenic endothelial cells (Butko et al., 2015). We found that the expression of scl-β, gata2b and gfi1aa, an additional early HSC marker gene (Thambyrajah et al., 2016), was also remarkably reduced in the DA of jam3bsa37 embryos (Fig. S1G), suggesting that Jam3b mutation impairs the early specification program of HSCs.
We next investigated the formation of vascular endothelium and primitive hematopoietic cells in jam3bsa37 embryos. In WT embryos, the expression of a pan-endothelial marker gene, fli1a, was detected throughout the vasculature including the DA, posterior cardinal vein (PCV) and intersegmental vessels (ISVs) at 24 hpf. In contrast, fli1a expression was weak and discontinuous in the trunk of jam3bsa37 embryos at the same stage. We also examined the expression of efnb2a (a DA marker) and flt4 (a PCV marker) in jam3bsa37 embryos and observed an expression pattern similar to fli1a in the DA and PCV, respectively (Fig. 1D). The expression of the primitive erythrocyte marker genes gata1a and hbae3 in the intermediate cell mass (ICM) and posterior blood island (PBI), which are the major sites of primitive erythropoiesis (Galloway and Zon, 2003; Warga et al., 2009), was largely reduced in jam3bsa37 embryos compared with WT embryos (Fig. 1E). Embryos injected with a jam3b morpholino oligo (MO), which blocks the translation of jam3b mRNA (Fig. S1H) (Powell and Wright, 2011), also showed a reduction in runx1 expression in the DA, fli1a in the trunk vasculature, and gata1a in the ICM and PBI (Fig. S1I), confirming the results observed in jam3bsa37 embryos. In contrast, the head vasculature and primitive macrophages developed normally in jam3bsa37 embryos, as evidenced by the expression of fli1a in the head and l-plastin (lcp1; a macrophage marker) in the rostral blood island (Fig. 1F), the major site of primitive myelopoiesis. Collectively, these data indicate that mutation of Jam3b resembles the cloche phenotype, with defects in both hematopoietic and vascular development, although the phenotypes are restricted to the posterior part of the embryo.
Jam3b maintains the expression of npas4l in the PLPM
Given the defect of both hematopoietic and vascular development in jam3bsa37 embryos, we speculated that Jam3b is involved in the early hemato-vascular development. We next examined, therefore, the expression of early hematopoietic/endothelial marker genes in jam3bsa37 embryos. As shown in Fig. 2A, the expression of gata1a, scl, etv2 (also known as etsrp), fli1a and npas4l was reduced in the PLPM in jam3bsa37 embryos at 14 hpf compared with WT embryos. Quantitative polymerase chain reaction (qPCR) analysis in isolated fli1a:GFP (+) cells at 15 hpf confirmed the reduction of npas4l as well as fli1a expression in LPM cells of jam3bsa37 embryos (Fig. 2B). We also observed the downregulation of other LPM marker genes, alas2, klf17, hhex and ets1, in jam3bsa37 embryos, whereas expression of cdx4 and its downstream target gene lmo2 (Paik et al., 2013) was near normal or slightly upregulated (Fig. S2A,B). In contrast to the PLPM, the expression of spi1b (a myeloid marker, also known as pu.1), scl and etv2 in the anterior LPM (ALPM), the site from which primitive myeloid cells and endothelial precursors emerge, was not greatly affected in jam3bsa37 embryos (Fig. 2C), which is consistent with our finding that Jam3b is dispensable for the formation of primitive myelocytes and head vasculature (Fig. 1F). The expression of pax2a (a pronephros marker) and dlc (a somite marker) was also unaffected in jam3bsa37 embryos (Fig. S2B), suggesting that mutation of jam3b does not affect mesodermal induction.
As mutation of jam3b led to the reduction of various hematopoietic and endothelial marker genes including npas4l in the PLPM, we next investigated whether Jam3b is involved in the initial formation of the PLPM. The expression of npas4l in the PLPM is first detected at ∼11 hpf as small bilateral stripes, which will extend anteriorly and posteriorly by 14 hpf (Reischauer et al., 2016). We found that npas4l expression was unaffected at 11 hpf in jam3bsa37 embryos. Consistent with this, the expression of fli1a was also normal at 12 hpf (Fig. 2D). Taken together, these data suggest that Jam3b is required for the maintenance of npas4l expression in the PLPM rather than the PLPM formation.
Jam3b regulates the expression of drl and lrrc15
In order to explore the molecular mechanisms governing Jam3b-dependent hemato-vascular development, homozygous jam3bsa37 embryos obtained by an in-cross of jam3bsa37 animals were used for RNA-seq analysis at 12 or 16 hpf (Fig. S3A). Principal component analysis (PCA) and correlation based on expression read counts revealed close associations between each expression profile (Fig. S3B). Previous studies in zebrafish showed that draculin (drl), which encodes a zinc-finger transcription factor, is expressed in the early LPM (Herbomel et al., 1999). Microarray analysis of isolated drl:GFP (+) and (−) cells at 11 hpf identified a number of uncharacterized genes expressed in the LPM (Mosimann et al., 2015). We therefore combined our RNA-seq datasets with published microarray datasets of drl:GFP (+) and (−) cells to identify genes that were differentially expressed between LPM cells from WT and jam3bsa37 embryos. Within genes that were more than twofold enriched in drl:GFP (+) cells (329 genes), 15 genes were selected as up- (>2-fold) or downregulated (<0.5-fold) genes in either 12 or 16 hpf jam3bsa37 embryos based on two independent RNA-seq analyses (Fig. S3C and Table S1). Among these 15 genes, eight genes involved in signal transduction or encoding membrane proteins (four downregulated genes, drl, gfi1aa, gstt1b and lrrc15, and four upregulated genes, cldnh, nos1apa, tcf21, and xkr9) were further examined by qPCR in fli1a:GFP (+) cells at 15 hpf. The expression of six out of the eight selected genes was successfully detected by qPCR. The expression of drl, lrrc15 and tcf21 decreased in fli1a:GFP (+) cells of jam3bsa37 embryos (<0.3-fold), whereas gfi1aa, gstt1b and nos1apa were nearly unchanged (Fig. 3A, Fig. S3D). Because tcf21 was detectable in neither the PLPM nor the DA by WISH (Fig. S3E), we focused on drl and lrrc15 as final candidates (Fig. S3F). drl was previously identified in a WISH screen of zebrafish cDNA libraries (Herbomel et al., 1999). There are four paralogs of drl genes in the zebrafish genome: drl, drl-like 1 (drl-l1; drll.1), drl-l2 (drll.2) and drl-l3 (drll.3). Each paralog contains three exons with very high homology in both coding and non-coding sequences. Although loss of drl-l3, which has the longest coding sequence of the group, affected primitive erythropoiesis, the role of drl as well as drl-l1 and drl-l2 in hemato-vascular development has not been clearly demonstrated (Pimtong et al., 2014). The expression of drl was specifically detected in the PLPM, but it was markedly decreased in jam3bsa37 embryos at 15 hpf (Fig. 3B). Lrrc15 belongs to the leucine rich repeat superfamily and is involved in cell-cell or cell-extracellular matrix interactions. Lrrc15 is frequently increased in tumor cells, particularly in breast and prostate cancers (Schuetz et al., 2006; Stanbrough et al., 2006). However, the role of Lrrc15 in hematopoietic and vascular cells has not been determined. In WT embryos, the expression of lrrc15 was detectable by WISH after 17 hpf specifically in the vascular cord, a linear mass of endothelial precursors at the midline. The expression of lrrc15 was also detected in the ventral floor of the DA and a portion of primitive erythrocytes at 24 hpf. We found that lrrc15 expression was also largely decreased in the vascular cord at 18 hpf and the DA and primitive erythrocytes at 24 hpf in jam3bsa37 embryos (Fig. 3C,D). Because the expression of drl and lrrc15 was specific in the hematopoietic and/or endothelial tissue and was very strongly affected by mutation of jam3b, we hypothesized that Drl and Lrrc15 are core components of Jam3b-dependent hemato-vascular development.
To examine the functional roles of drl and lrrc15 in hemato-vascular development, validated drl MO or lrrc15 MO was injected into WT embryos (Fig. S4A,B) (Pimtong et al., 2014). In lrrc15 morphants, the expression of runx1 in the DA and gata1a in the ICM and PBI was reduced, whereas the expression of fli1a in the trunk vasculature and npas4l in the PLPM was unaffected. In contrast, the expression of all four of these genes was severely reduced in drl morphants (Fig. 3E). We generated lrrc15 (lrrc15kz1) and drl (drlkz2) genetic mutant lines, which contain premature stop codons at early exons (Fig. S4C,D). Homozygous lrrc15kz1 embryos exhibited reduced runx1 and gata1a expression, but no change in fli1a and npas4l expression. In contrast, the expression of all four genes was reduced in homozygous drlkz2 embryos (Fig. S4E), recapitulating both lrrc15 and drl MO phenotypes. These results suggest that Lrrc15 is involved in hematopoietic development, including the formation of HSCs, whereas Drl is involved in both hematopoietic and vascular development.
To test whether the defects in hemato-vascular development in jam3bsa37 embryos could be rescued by enforced expression of drl and/or lrrc15, we injected mRNAs for these two genes into jam3bsa37 embryos. The expression of npas4l in the PLPM was increased in jam3bsa37 embryos injected with drl mRNA compared with uninjected jam3bsa37 embryos (Fig. S4F). Injection of drl mRNA also rescued the expression of fli1a in the trunk vasculature. However, the expression of runx1 in the DA and gata1a in the ICM and PBI was not restored by injection of drl mRNA into jam3bsa37 embryos (Fig. 4A), although gata1a expression increased in the PLPM (Fig. S4G). These data suggest that enforced expression of drl is sufficient to rescue vascular development, but insufficient to restore hematopoietic development in jam3bsa37 embryos. In contrast to drl mRNA, injection of lrrc15 mRNA did not rescue runx1, fli1a or gata1a in jam3bsa37 embryos. Interestingly, however, co-injection of drl and lrrc15 mRNA fully rescued the expression of runx1 and gata1a as well as fli1a in jam3bsa37 embryos (Fig. 4A). We examined the number of cd41:GFP; kdrl:mCherry double-positive HSPCs in the ventral floor of the DA and found that the average number of HSPCs in jam3bsa37 embryos was significantly increased by co-injection of drl and lrrc15 mRNA (Fig. 4B). In addition, the expression of rag1 in the thymus was restored by co-injection of these two mRNAs (Fig. 4C). These data clearly indicate that enforced expression of both drl and lrrc15 is sufficient to rescue hemato-vascular development in jam3bsa37 embryos. Our results also suggest that Drl acts upstream of Npas4l to promote vascular development, whereas Lrrc15 is essential for hematopoietic development.
Jam3b suppresses Rap1a-Erk activity to promote vascular development
In mammals, JAM3 has been shown to negatively regulate Rap1 activity (Orlova et al., 2006). Rap1 exists in an inactive guanine nucleotide diphosphate (GDP)-bound state and is activated when GDP is exchanged for guanine nucleotide triphosphate (GTP). Upon activation, Rap1 can bind to a variety of effectors, such as Raf family proteins, which triggers Erk activation and other signaling pathways (Stork and Dillon, 2005). We compared Rap1 activity between WT and jam3bsa37 embryos using Raichu-Rap1, a fluorescence resonance energy transfer (FRET)-based Rap1 activation monitoring-probe. Binding of GTP-Rap1 to Raf induces FRET from CFP to YFP within cells expressing Raichu-Rap1 (Fig. S5A) (Sakurai et al., 2006). Raichu-Rap1 mRNA was injected into WT or jam3bsa37 embryos under the kdrl:mCherry background, which labels endothelial cells with mCherry, and live-imaging analysis was performed at 18 hpf. In WT embryos, FRET signals were nearly undetectable in kdrl:mCherry (+) endothelial precursors and other cells. In contrast, strong FRET signals were detected in jam3bsa37 embryos, especially in kdrl:mCherry (+) endothelial precursors within the posterior part of the embryo (Fig. 5A). This suggests that Jam3b suppresses Rap1 activation in the zebrafish embryo, as has been shown for human JAM3 (Orlova et al., 2006).
To determine whether Jam3b regulates hemato-vascular development by suppressing Rap1 activity, Rap1 activation was induced by treating WT embryos with the Epac-selective cAMP analog 8-pCPT-2'-O-Me-cAMP-AM (known as 007-AM). Hyperactivation of Rap1 through 007-AM treatment led to the cloche-like phenotype observed in jam3bsa37 embryos, marked by reduced expression of drl, npas4l in the PLPM, runx1 in the DA and fli1a in the trunk vasculature compared with dimethyl sulfoxide (DMSO)-treated control embryos (Fig. S5B).
The Ras-Raf-Mek-Erk kinase cascade triggered by FGF signaling is one of the most important pathways for mesodermal induction in vertebrates (Umbhauer et al., 1995; Kimelman, 2006; Xiong et al., 2015). It has been shown that the duration and strength of Erk activation can influence cell fate determination, including proliferation, differentiation, survival or apoptosis (Shaul and Seger, 2007). Because Rap1 is also known as an essential mediator of Erk activation in hematopoietic and vascular endothelial cells (Stork and Dillon, 2005; Yan et al., 2008), it is likely that the high level of Rap1 activation in jam3bsa37 embryos causes excessive Erk activation, possibly leading to impaired hemato-vascular development. We therefore measured the level of phosphorylated Erk1 and 2 (pErk1/2), the active form of Erk1/2, in jam3bsa37 embryos. We found that, although the amount of total Erk1/2 was unchanged, pErk1/2 was higher in jam3bsa37 embryos than in WT embryos at 23 hpf. The higher ratio of pErk1/2 to total Erk1/2 in jam3bsa37 embryos indicates that jam3b mutation results in increased Erk activation (Fig. 5B). Whole-mount immunostaining analysis further confirmed that pErk1/2 levels at 23 hpf were higher throughout the body of jam3bsa37 embryos compared with WT embryos (Fig. 5C). At 15 hpf, pErk1/2 signals were strongly detected in the developing somite, but nearly undetectable in fli1a:GFP (+) PLPM cells in WT embryos, as has been shown previously (Shin et al., 2016). In contrast, pErk1/2 signals were detectable not only in the somite but also in a portion of fli1a:GFP (+) PLPM cells in jam3bsa37 embryos (Fig. 5D). Importantly, MO knockdown of both rap1aa and rap1ab (rap1aa/b) in jam3bsa37 embryos reduced the level of Erk activation in PLPM cells (Fig. 5D, Fig. S5C), suggesting that Jam3b represses Erk activation in PLPM cells through the regulation of Rap1a. MO knockdown of rap1aa/b in WT embryos also led to a reduction in drl expression in the early mesoderm (Fig. S5D). These data suggest that Rap1a activates Erk to induce early mesodermal development, but this Rap1a-Erk signaling cascade is repressed later by Jam3b in PLPM cells.
To establish the link between Rap1a-Erk activation and the cloche-like phenotype in jam3bsa37 embryos, we attempted to rescue the jam3bsa37 phenotype by suppression of Erk activity. We treated embryos with a chemical Mek inhibitor, either SL-327 or PD98059, to block Erk phosphorylation. Treatment of jam3bsa37 embryos with 15μM of SL-327 or 10μM of PD98059 from 6-23 hpf reduced the phosphorylation of Erk1/2 (Fig. 5B). We observed a marked increase in drl and npas4l expression in the PLPM of PD98059- or SL-327-treated jam3bsa37 embryos compared with DMSO-treated jam3bsa37 embryos. Interestingly, however, the expression of lrrc15 in the vascular cord was not restored by PD98059 or SL-327 treatment (Fig. 5E, Fig. S5E), indicating that the regulation of lrrc15 by Jam3b is independent from the Erk signaling pathway. A previous study showed that the formation of ISVs, but not the DA or PCV, requires Erk activation in endothelial precursors (Shin et al., 2016). We found that the expression of fli1a was greatly increased in the DA and PCV in PD98059- or SL-327-treated jam3bsa37 embryos, whereas it was unchanged in ISVs (Fig. 5E, Fig. S5E). Taken together, these data suggest that Jam3b represses the Rap1a-Erk signaling cascade to promote endothelial specification in mesodermal cells.
We next aimed to rescue HSC formation in jam3bsa37 embryos by enforcing expression of lrrc15 while blocking Erk signaling. It should be noted, however, that treatment with SL-327 severely affected the formation of the somite (Fig. S5F), which expresses several important signaling molecules that are required for HSC specification, such as Wnt16, Dlc, Dld and Vegfa (Clements et al., 2011; Genthe and Clements, 2017). We found that the effect on somitogenesis was much less pronounced when embryos were treated with PD98059 (Fig. S5F). We therefore injected lrrc15 mRNA into jam3bsa37 embryos and treated with PD98059 from 6-14 hpf. The expression of runx1 in the DA of jam3bsa37 embryos was slightly increased by PD98059 treatment, and it was greatly increased by injection of lrrc15 mRNA together with PD98059 treatment (Fig. 5F). These data confirmed that Jam3b plays a second role in the acquisition of HSC fate through the regulation of lrrc15 independently from drl regulation.
A previous study in mammals suggests that Rap1a-mediated Erk activation in vascular endothelial cells is induced at least in part by FGF signaling (Yan et al., 2008), raising the possibility that the high level of Erk activation observed in jam3bsa37 embryos originates from FGF signaling. To test this hypothesis, we treated embryos with a chemical inhibitor of FGF receptors, SU5402. In WT embryos, treatment with 5μM of SU5402 from 8-14 hpf resulted in enhanced expression of drl in the PLPM without causing morphological defects, whereas treatment from 2-14 hpf led to the deformation of the PLPM (Fig. S5G). Similar to Mek inhibitors, treatment of jam3bsa37 embryos with SU5402 from 8-14 hpf rescued the expression of drl and npas4l in the PLPM and fli1a in the DA and PCV (Fig. 5G). Taken together, these results suggest that suppression of the FGF-Rap1a-Erk signaling pathway by Jam3b is crucial for endothelial specification in mesodermal cells.
Jam3b promotes HSC specification through an integrin-dependent signaling pathway
Having shown that Jam3b regulates lrrc15 expression to promote HSC development independently from the Rap1a-Erk signaling cascade, we next investigated how Jam3b regulates lrrc15 expression in endothelial precursors. In human carcinoma cells, JAM3 controls integrin-dependent signal transduction via the modulation of β1-integrin activation (Mandicourt et al., 2007). The activation of integrin via binding to the extracellular matrix (ECM) mediates various intracellular signaling cascades, such as the phosphoinositide 3-kinase (PI3K), c-Jun N-terminal kinase (JNK) and Erk signaling cascades (Hynes, 2002). Recently, it has been shown that β1-integrin is required for HSC specification in zebrafish embryos (Rho et al., 2019). To investigate whether integrin regulates lrrc15 expression in endothelial precursors, we used a transgenic zebrafish that expresses a dominant-negative form of β1-integrin fused with mCherry (mCherry-itgb1aDN). Enforced expression of mCherry-itgb1aDN leads to stacking of intracellular cytoskeleton-associated proteins, such as talin and actinin, which inhibits endogenous integrin signaling (Iida et al., 2018). The UAS:mCherry-itgb1aDN line was crossed with an endothelial cell-specific Gal4 driver line, fli1a:Gal4 (Fig. 6A). We found that the expression of runx1 in the DA at 28 hpf and lrrc15 in the vascular cord at 20 hpf was reduced in mCherry-itgb1aDN (+) embryos compared with mCherry-itgb1aDN (−) embryos (Fig. 6B,C). Although forced expression of mCherry-itgb1aDN in endothelial cells induced blood vessel abnormalities (Iida et al., 2018), the expression of efnb2a in the DA was intact in mCherry-itgb1aDN (+) embryos (Fig. S6A), suggesting that the defect in HSC formation in mCherry-itgb1aDN (+) embryos was not due to the secondary effect of blood vessel abnormalities.
Focal adhesion kinase (FAK) is a nonreceptor tyrosine protein kinase that plays a central role in the transduction of integrin signaling. Upon binding of integrin to the ECM, FAK is phosphorylated, which triggers the recruitment of Src family kinases and signaling complex formation (Kleinschmidt and Schlaepfer, 2017). Mutation of lysine residues to glutamic acid in the activation loop of FAK (K578E/K581E) mimics phosphorylation of FAK and enhances integrin signaling (Gabarra-Niecko et al., 2002). We generated mRNA for zebrafish constitutively activated Faka (ca-faka), which contains mutations of K580E and K583E (Fig. S6B). Injection of ca-faka mRNA was sufficient to rescue runx1 expression in the DA in mCherry-itgb1aDN (+) embryos (Fig. 6B). In addition, injection of lrrc15 mRNA also rescued runx1 expression in mCherry-itgb1aDN (+) embryos (Fig. 6D). Together, these data suggest that integrin-dependent signaling regulates the expression of lrrc15 in endothelial precursors to form HSCs in developing embryos.
To examine whether mutation of jam3b affects the adhesion of endothelial precursors to the ECM, we isolated fli1a:GFP (+) cells from WT or jam3bsa37 embryos at 18 hpf and performed an adhesion assay on a fibronectin-coated dish. Although adherent fli1a:GFP (+) cells were detected in both WT and jam3bsa37 embryos (Fig. S6C), the percentage of adherent fli1a:GFP (+) cells was significantly lower in jam3bsa37 embryos compared with WT embryos (Fig. S6D). In addition, time-lapse imaging of fli1a:GFP revealed that some migrating fli1a:GFP (+) cells in jam3bsa37 embryos showed less spreading and a more rounded morphology compared with WT embryos, resulting in a delay in medial convergence and discontinuous vascular cord (Fig. 6E, Movies 1 and 2). These phenotypes are very similar to those in integrin-deficient embryos (Rho et al., 2019). These findings allowed us to test whether HSC formation in jam3bsa37 embryos can be rescued by injection of ca-faka mRNA. As shown in Fig. 6F, the expression of lrrc15 in the vascular cord was restored in jam3bsa37 embryos injected with both drl and ca-faka mRNA, but it was not restored in jam3bsa37 embryos injected with ca-faka mRNA alone. Consistent with this, the expression of runx1 in the DA was also restored in jam3bsa37 embryos injected with both mRNAs (Fig. 6G). These results strongly suggest that Jam3b regulates an integrin-dependent signaling cascade to promote HSC specification.
As FAK is known to mediate multiple signaling pathways, including PI3K, JNK and Erk, we then tried to determine which pathway is involved in HSC formation. We could exclude the Erk signaling pathway because impaired hemato-vascular development in jam3bsa37 embryos is caused in part by a high level of Erk activation in PLPM cells (Fig. 5B-F). Therefore, we treated WT embryos with a PI3K inhibitor (wortmannin) or a JNK inhibitor (SP600125) from 6-24 hpf and examined runx1 expression in the DA. We found that wortmannin treatment reduced runx1 expression, whereas SP600125 treatment did not (Fig. S6E). Treatment of wortmannin also reduced lrrc15 expression in the vascular cord, whereas enforced expression of lrrc15 in wortmannin-treated embryos rescued runx1 and gata2b expression in the DA (Fig. 6H,I, Fig. S6F). Collectively, these data suggest that Jam3b plays a second role in the regulation of lrrc15 to form HSCs via the activation of the integrin-FAK-PI3K signaling pathway.
As heterophilic interactions between Jam2 and Jam3 have been implicated in many biological processes (Gliki et al., 2004; Weber et al., 2007; Powell and Wright, 2011; Arcangeli et al., 2011), we investigated whether Jam3b requires Jam2 to regulate hemato-vascular development. Previous studies determined the physical binding properties of zebrafish Jam proteins by surface plasmon resonance: Jam3b can bind to Jam2a, Jam2b and Jam3b itself, but not to Jam1a, Jam1b and Jam3a (Powell and Wright, 2011, 2012). It has also been shown that the interaction of Jam1a and Jam2a mediates Notch signal transduction between endothelial precursors and the somite to establish HSC fate (Kobayashi et al., 2014), whereas the role of Jam2b in hemato-vascular development has not been determined. MO knockdown of jam2a, but not jam2b, reduced the expression of drl in the PLPM and lrrc15 in the vascular cord (Fig. S7A,B). Although the phenotype of jam2a morphants was similar to that of jam3bsa37 embryos, jam2a expression was not detected in the PLPM (Fig. S7C), suggesting that non-endothelial Jam2a is involved in the activation of Jam3b in the PLPM to promote hemato-vascular development. In contrast to jam3bsa37 embryos, however, runx1 expression in the DA was not restored in embryos injected with jam2a MO together with drl and lrrc15 mRNA (Fig. S7D). This is likely because of the additional role that Jam2a plays in regulating HSC development via Notch signal transduction (Kobayashi et al., 2014). Taken together, our data suggest that Jam3b regulates both endothelial and hematopoietic specification through two independent signaling pathways, and this dual role of Jam3b may be promoted by Jam2a (Fig. 7).
In the present study, we define a novel molecular mechanism by which Jam3b plays dual roles in both hematopoietic and vascular development in mesodermal cells. During early embryogenesis, Erk is activated by FGF signaling to induce mesodermal development. This default state of Erk activation in mesodermal cells is suppressed by Jam3b through inhibition of the Rap1a activity, promoting the acquisition of early endothelial identity. Jam3b then regulates the expression of lrrc15 in endothelial precursors via an integrin-dependent signaling pathway to confer hematopoietic fate, leading to the formation of HSCs in the DA. Thus, our study delineates the fundamental mechanisms for the divergence of hematopoietic and vascular lineages from mesodermal cells.
Jam proteins play major roles in the modulation of cell-cell interaction, cell migration and cell polarization in many tissues. In some cases, however, Jam proteins are also involved in signal transduction to promote cell proliferation and differentiation. For example, Jam1 forms a complex with integrin αVβ3 in endothelial cells under physiological conditions. After activation of endothelial cells by FGF signaling, Jam1 dissociates from integrin αVβ3, which induces angiogenetic signals, such as the Erk signaling pathway (Naik et al., 2003). Our current data and previous reports in mammals (Orlova et al., 2006; Mandicourt et al., 2007) demonstrated that Jam3 negatively regulates the Rap1a signaling pathway, but positively regulates an integrin-dependent signaling pathway. Although further studies are required to address the detailed molecular mechanism behind the regulation of Rap1a and integrin by Jam3, these regulatory pathways may be evolutionarily conserved among vertebrates.
Identification of npas4l has greatly contributed to our understanding of how hematopoietic/endothelial precursors are specified from the mesodermal cell in vertebrates (Reischauer et al., 2016). In the present study, we have demonstrated that Drl acts upstream of npas4l in the LPM. Loss of drl results in the downregulation of npas4l, whereas enforced expression of drl increased npas4l expression in the PLPM of jam3bsa37 embryos. Interestingly, injection of drl mRNA did not lead to ectopic expression of npas4l or other hematopoietic/endothelial marker genes, possibly because of the presence of a cofactor that regulates npas4l expression in the LPM. On the other hand, drl expression is decreased in the PLPM by the loss of scl or lmo2 (Dooley et al., 2005; Patterson et al., 2007), two downstream targets of Npas4l (Reischauer et al., 2016). These findings, in combination with our data, suggest that there are cross-talk regulatory mechanisms between Drl, Scl, Lmo2 and Npas4l in the PLPM, and these factors may cooperatively regulate early hemato-vascular development.
The duration and strength of Erk activation is tightly controlled by the balance between positive and negative regulatory pathways. Imbalanced Erk signaling is frequently associated with malignant transformation. Megakaryocyte maturation is regulated by sustained Erk activation via Rap1 activation, but it is inhibited by transient Erk activation via Ras activation (Stork and Dillon, 2005). These observations highlight that sustained or transient activation of Erk signaling can direct different cellular outcomes. In the present study, we found that mutation of jam3b results in the hyperactivation of the Rap1a-Erk signaling cascade in PLPM cells. Although Erk activation induced by FGF signaling is required for early mesodermal induction, prolonged Erk activation in the PLPM impaired hemato-vascular development. Thus, our data suggest that Jam3b plays an important role in the repression of Rap1a activity, enabling transient Erk activation in the PLPM to promote endothelial specification.
In vertebrates, it is believed that hemangioblasts can give rise to hematopoietic cells through two types of intermediate endothelial precursors, hemogenic angioblasts and hemogenic endothelial cells. Hemogenic angioblasts can give rise to primitive hematopoietic cells in the yolk sac, and hemogenic endothelial cells can generate HSCs in the DA (Gritz and Hirschi, 2016; Lacaud and Kouskoff, 2017). This concept suggests that endothelial specification is an essential first step for mesodermal cells to differentiate into hematopoietic lineages. Consistent with this idea, our present data suggest that, after promoting endothelial specification through Npas4l regulation, Jam3b plays a second role in the acquisition of hematopoietic fate by regulating lrrc15. Although the role of Lrrc15 in HSC formation remains elusive, our work uncovers the importance of Jam3 in the acquisition of both endothelial and hematopoietic fates by mesodermal cells, which may aid in the generation of HSCs from iPSCs.
MATERIALS AND METHODS
All animal experiments were performed in accordance with a protocol approved by the Committee on Animal Experimentation of Kanazawa University. Zebrafish strains AB*, jam3bsa37, lrrc15kz1, drlkz2, Tg(fli1a:GFP)y1, Tg(-6.0itga2b:eGFP)la2, Tg(kdrl:memCherry)s896, Tg(fli1a:Gal4FF)ubs4 and Tg(UAS:mCherry-itgb1aDN)ko112 were raised in a circulating aquarium system (AQUA) at 28.5°C in a 14/10 h light/dark cycle and maintained in accordance with guidelines of the Committee on Animal Experimentation of Kanazawa University. Because homozygous jam3bsa37 animals are partially viable into adulthood and fertile, homozygous jam3bsa37 embryos were obtained by an in-cross of jam3bsa37 animals.
All chemical compounds used in this study were dissolved in DMSO (Wako) and were used at the following concentrations: 8-pCPT-2′-O-Me-cAMP-AM (007-AM; Biolog Life Science Institute), 20 μM; SU5402 (Wako), 5 μM; SL-327 (Santa Cruz Biotechnology), 15 μM; PD98059 (Tocris Bioscience), 10 μM; wortmannin (Tocris Bioscience), 1 μM; SP600125 (Tocris Bioscience), 5 μM. Embryos were incubated in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2, 0.33 mM MgSO4) containing DMSO or a chemical compound in the dark, followed by dechorionation with 20 mg/ml of pronase (Sigma-Aldrich) in E3 medium and fixation with 4% paraformaldehyde (PFA; Wako) in phosphate buffered saline (PBS).
Generation of mutant lines
For CRISPR/Cas9-mediated generation of mutant lines, crRNA recognizing the genomic loci was designed and purchased from Integrated DNA Technologies (IDT) and hybridized with tracrRNA according to the manufacturer's instructions. The gRNA complex (50 pg/nl) was then co-injected with 400 ng/nl of Cas9 protein (IDT) into one-cell-stage embryos. Fish from the F1 generation were screened using primers that flanked the target region (Table S2) and mutant alleles were identified by sequencing. Embryos from the F3 generation were used for experiments. The crRNA sequences used in this study are: lrrc15, CTGGCAGCTTTCTGGACATG; drl, ACAGAGCATCACAATGCTGA.
WISH and immunostaining
cDNAs were cloned by reverse transcription (RT)-PCR using specific primers (listed in Table S2) and ligated into the pCRII-TOPO vector (Invitrogen). Digoxigenin (DIG)-labeled RNA probes were prepared by in vitro transcription with linearized constructs using the DIG RNA Labeling Kit (SP6/T7, Sigma-Aldrich). Embryos were fixed in 4% PFA in PBS. For permeabilization, embryos were treated with proteinase K (10 μg/ml, Sigma-Aldrich) for between 30 s and 30 min, refixed with 4% PFA and washed with 0.1% Tween-20 (Sigma-Aldrich) in PBS (PBST). Hybridization was then performed using DIG-labeled antisense RNA probes diluted in hybridization buffer [50% formamide, 5× standard saline citrate (SSC), 0.1% Tween-20, 500 μg/ml torula RNA, 50 μg/ml heparin] for 3 days at 65°C. For detection of DIG-labeled RNA probes, embryos were blocked in 0.2% bovine serum albumin (BSA; Sigma-Aldrich) in PBST and incubated overnight at 4°C with the alkaline phosphatase-conjugated anti-DIG antibody (Sigma-Aldrich, 1093274) at a dilution of 1:5000. After washing with PBST, embryos were developed using nitroblue tetrazolium chloride and 5-bromo-4-chloro-3-indolyl-phosphate (NBT/BCIP; Sigma-Aldrich) in staining buffer [100 mM Tris (pH 9.5), 100 mM NaCl, 50 mM MgCl2, 0.1% Tween-20]. For histological analysis, embryos developed with NBT/BCIP were embedded in paraffin and sectioned at 3 μm. Deparaffinized tissue sections were mounted with Mount-Quick Aqueous (Cosmo Bio). For fluorescent WISH, embryos were permeabilized with acetone at −20°C for 7 min and hybridized with DIG-labeled RNA probes as described above. Embryos were blocked with 2% blocking reagent (Sigma-Aldrich) in PBST and incubated with the peroxidase-conjugated anti-DIG antibody (Sigma-Aldrich, 11207733910) at a dilution of 1:500. Embryos were developed using TSA Plus Cyanine 3 System (PerkinElmer) for 30 min. For whole-mount immunostaining, fixed embryos were permeabilized with acetone, blocked with 2% blocking reagent, and incubated overnight at 4°C with the rabbit anti-phospho-p44/42 MAPK (for pErk1/2) (Cell Signaling Technology, 9101) at 1:250, and/or chicken anti-GFP (Aves Labs, GFP-1020) at 1:1000 dilution. After washing with PBST, embryos were incubated overnight at 4°C with the goat anti-chicken IgY Alexa Fluor 488-conjugated (Abcam, ab150173) at 1:1000 and/or donkey anti-rabbit IgG Alexa Fluor 647-conjugated (Abcam, ab150075) at 1:1000 dilution.
Cell preparation and flow cytometry
Embryos were dechorionized and digested with 50 μg/ml of Liberase TM (Roche) in PBS for 1 h at 37°C. Cells were then filtered through a 40 μm stainless steel mesh and washed with 2% fetal bovine serum (FBS; Cosmo Bio) in PBS by centrifugation (400 g). Immediately before flow cytometric analysis, Sytox Red solution (Thermo Fisher Scientific) was added to the samples at a final concentration of 5 nM to exclude nonviable cells. Flow cytometric acquisition and cell sorting was performed on a FACS Aria II (BD Biosciences). Cells were sorted into 10% FBS in PBS and were used for expression analysis.
Sorted fli1a:GFP (+) cells were resuspended in the medium containing 40% Leibovitz's L-15 medium (Wako), 32% Dulbecco's Modified Eagle's Medium (Wako), 12% Ham's F12 medium (Wako), 8% FBS, 2 mM L-glutamine (Wako), 15 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES, Sigma-Aldrich), 100 U penicillin (Wako) and 100 μg/ml streptomycin (Wako). Cells were then plated on a 96-well plate (Violamo) or a glass-bottom dish (Matsunami) coated with 50 μg/ml of fibronectin (Cosmo Bio) and incubated for 12 h at 30°C, 5% CO2. For the adhesion assay, cells in a 96-well plate were washed 3× with PBS and GFP (+) adherent cells were counted using an EVOS FL imaging system (Thermo Fisher Scientific). Percent adhesion was calculated by dividing the number of GFP (+) adherent cells by the total number of GFP (+) cells plated in the well.
qPCR and RT-PCR
Total RNA was extracted from embryos using RNeasy Mini Kit (Qiagen), and cDNA was synthesized using ReverTra Ace qPCR RT Master Mix (Toyobo). qPCR assays were performed using GoTaq qPCR Master Mix (Promega) on a ViiA 7 Real-Time PCR System according to the manufacturer's instructions (Thermo Fisher Scientific). The expression of ef1a (eef1a1l1) was used for normalization. RT-PCR was performed using GoTaq Green Master Mix (Promega), and gel analysis was performed in 1.5% agarose or 5-10% acrylamide gels. Primers used for qPCR or RT-PCR are listed in Table S2.
Cap analysis gene expression (CAGE) library preparation, sequencing, mapping and gene expression analysis were performed by Dnaform (Haberle et al., 2015). We used 2 μg of total RNA from embryos for synthesis of cDNA using a Library Preparation kit (Dnaform). The 5′ caps of RNAs were biotinylated after reverse transcription. RNA-cDNA hybrids were then captured by streptavidin-conjugated magnetic beads. The cDNA was released from RNAs and ligated with 5′ and 3′ linkers. Second-strand synthesis was performed to create the final double-stranded DNA product. Sequencing of CAGE tags was performed using the Illumina NextSeq500, and base-calling was performed using the Illumina RTA software (ver. 2.4.11). Sequenced CAGE tags were mapped to the full genome sequences for Danio rerio (danRer10) using the Burrows-Wheeler Aligner (BWA, ver. 0.7.12-r1039) and HiSAT2 software (ver. 2.0.5). Tags per million (TPM) were calculated using the CAGEr package of Bioconductor in R (ver. 3.3.3) with a cutoff of <0.5. To evaluate quality and reproducibility, Pearson correlation coefficients were calculated and PCA was performed using R. Heatmaps were generated in R with the Bioconductor gplot package. The data have been deposited in Gene Expression Omnibus (GEO; National Center for Biotechnology Information) under accession number GSE114416. Microarray data of drl:GFP (+) and drl:GFP (−) were obtained from the GEO database (accession number GSE70881).
Dechorionized embryos were collected and lysed in lysis buffer [25 mM Tris-HCl (pH 7.4), 1 mM EDTA, 0.1 mM EGTA, 150 mM NaCl, 5 mM MgCl2, 2 mM Na3VO4, 20% glycerol, 0.1% Triton X-100, 1 mM dithiothreitol, proteinase inhibitor (Wako, 165-26021)]. The lysate was then centrifuged to remove debris, resuspended in 2× sample buffer [4% SDS, 0.2 M dithiothreitol, 0.1 M Tris-HCl (pH 6.8), 10% glycerol, 20 mg/ml bromophenol blue], and boiled to prepare SDS-PAGE samples. The samples were separated by a 10% polyacrylamide gel and transferred to a PVDF membrane (Millipore). The membrane was blocked with the Can Get Signal/PVDF blocking reagent (Toyobo) for 60 min at 37°C, and then incubated with rabbit anti-phospho-p44/42 MAPK (for pErk1/2; Cell Signaling Technology, 9101, 1:500), rabbit anti-p44/42 MAPK (for total Erk1/2; Cell Signaling Technology, 9102; 1:1000), or rabbit anti-α/β-Tubulin (Cell Signaling Technology, 2148; 1:1000) antibodies overnight at 4°C. After washing with TBS-T [50 mM Tris-HCl (pH 7.4), 138 mM NaCl, 2.7 mM KCl, 0.1% Tween-20], the membrane was incubated with the anti-rabbit IgG HRP-conjugated secondary antibody (Cosmo Bio, SA00001-2; 1:10,000) for 2 h at room temperature. The primary and secondary antibodies were diluted with the Can Get Signal/Solution-1 and -2 (Toyobo), respectively. After washing with TBS-T, the chemiluminescence reaction was performed using the Western Lightning Plus-ECL (PerkinElmer) and the signals were developed on X-ray films. Expression levels were quantified using ImageJ software (ver. 1.51s).
For fluorescent and time-lapse imaging, embryos were mounted in a glass-bottom dish filled with 0.6% low-gelling agarose (Sigma-Aldrich) in E3 medium containing 200 μg/ml of tricaine (Sigma-Aldrich) and imaged using an FV10i confocal microscope and Fluoview FV10i-SW software (ver. 2.1.1) (Olympus). For time-lapse imaging, z-stacks were taken every 5 min and movies were created with a Fluoview FV10i-SW software. For fluorescent WISH samples, embryos were deyolked and washed with PBT to remove yolk granules. Embryos were then mounted with 70% glycerol/PBT and were imaged by confocal microscopy. Visible light imaging or fluorescent images of GFP mRNA-injected embryos were captured using an Axiozoom V16 microscope (Zeiss) with a TrueChrome II digital camera (BioTools) and TCapture software (ver. 188.8.131.522) (Tucsen Photonics) or an Axioskop 2 plus microscope (Zeiss) with a DS74 digital camera and CellSens Standard software (ver. 1.16) (Olympus).
Morpholino and mRNA injection
Embryos were injected at the one-cell stage with 1 nl of MO (GeneTools) or mRNA. The MO sequences and concentrations used in this study are as follows: jam3b MO, TTAACGCCATCTTGGAGTCGGTGAA (3.5 ng/nl) (Powell and Wright, 2011); lrrc15 MO, AACCACGCCAGGTCCATTGATTCCG (1.5 ng/nl); drl MO, TCAAGAACAGACACCAACCTCTTTG (3 ng/nl) (Pimtong et al., 2014); rap1aa MO, TGGTGGCCTGTAAGATTACATCACA (1 ng/nl); rap1ab MO, GTACAATGTACTTACCAAGGCTGAC (4 ng/nl). jam2a MO, AGGAACTACAGCAGAAACAGGTCAA (3.5 ng/nl) (Kobayashi et al., 2014); jam2b MO, GGTTAAAGCAATGCACCAACCAGAC (13 ng/nl). The Raichu-Rap1 was amplified using pCAGGS-Raichu-Rap1 vector, which consisted of EYFP-V68L/Q69K, zebrafish rap1b, Raf RBD and ECFP from the amino terminus, and was ligated into the pCS2+ vector. Capped mRNAs were synthesized from linearized pCS2+ constructs using the mMessage mMachine SP6 kit (Thermo Fisher Scientific). mRNAs were injected into embryos at the following concentrations; jam3b, 100 pg/nl; drl, 50 pg/nl; lrrc15, 280 pg/nl; Raichu-Rap1, 50 pg/nl; jam3b-GFP, 50 pg/nl; lrrc15-GFP, 50 pg/nl; ca-faka, 100 pg/nl.
Quantification and statistical analyses
Data were analyzed for statistical significance after at least two repeated experiments. Statistical differences between groups were determined by unpaired two-tailed Student's t-test. A value of P<0.05 was considered to be statistically significant.
The authors thank G. Wright for providing the jam3bsa37 line, D. Traver for providing the fli1a:GFP, cd41:GFP and kdrl:mCherry lines, A. Iida for providing the UAS:mCherry-itgb1aDN line, M. Kobayashi for providing in situ probes, D. Voon, M. Hazawa and K. Kitamura for supporting experiments, and D. Voon and K. Lewis for critical evaluation of the manuscript.
Conceptualization: I.K., M.Y.; Methodology: I.K., J.K.-S., H.K., S.F.; Validation: I.K., J.K.-S., Y.H., M.O., K.Y., H.K.; Formal analysis: I.K.; Investigation: I.K., J.K.-S., Y.H., M.O., K.Y., H.K., S.F.; Resources: J.K.-S., S.F.; Data curation: I.K., J.K.-S., Y.H., M.O., K.Y., H.K., S.F.; Writing - original draft: I.K.; Writing - review & editing: M.Y., H.K.; Visualization: I.K.; Supervision: I.K., M.Y.; Project administration: I.K.; Funding acquisition: I.K.
This work was supported in part by a Grant-in-Aid for Young Scientists (B) from the Japan Society for the Promotion of Science (17K15393), a Research Award for Young Investigators from the Hokuriku Bank, a Grant for Young Researchers from the JGC-S Scholarship Foundation, and a Grant-in-Aid from the Takeda Science Foundation, Mochida Memorial Foundation for Medical and Pharmaceutical Research, and the Naito Foundation.
RNA-seq data have been deposited in GEO under accession number GSE114416.
The authors declare no competing or financial interests.