Jagged 2b induces intercellular signaling within somites to establish hematopoietic stem cell fate in zebrafish

The ontogeny of the hematopoietic system is composed of two waves: the primitive and definitive waves. The primitive wave of hematopoiesis originates from erythroid- and myeloid-restricted progenitors, whereas the definitive wave initiates from multipotent hematopoietic progenitors, including hematopoietic stem cells (HSCs). During embryonic development, HSCs arise from a specific subset of vascular endothelial cells, termed hemogenic endothelial cells (HECs), in the ventral floor of the dorsal aorta (DA) through endothelial-to-hematopoietic transition (EHT) (Boisset et al., 2010; Bertrand et al., 2010; Kissa and Herbomel, 2010). Despite evolutionary divergence, the genetic programs governing hematopoiesis are highly conserved among vertebrates. Some key signaling molecules that regulate HSC formation have been identified using zebrafish and mouse embryos, such as Notch (Kumano et al., 2003; Hadland et al., 2004; Burns et al., 2005; Robert-Moreno et al., 2005; Kim et al., 2014), BMP4 (Wilkinson et al., 2009), FGF (Lee et al., 2014; Pouget et al., 2014) and Wnt (Clements et al., 2011; Grainger et al., 2016; Richter et al., 2018). These signaling molecules directly or indirectly confer HSC programs in a shared vascular precursor, the angioblast, to generate HECs in the DA.

Notch signaling plays a crucial role in HSC specification as well as patterning of the trunk vasculature. Canonical Notch signaling is induced when a membrane-bound Notch ligand (Delta or Jagged) on a signal-sending cell directly interacts with the Notch receptor on the cell surface of the signal-receiving cell. Upon binding to a Notch ligand, the receptor is cleaved by ADAM TACE metalloproteases at the S2 site, followed by further cleavage by γ-secretase at the S3 site. These cleavages of the Notch receptor result in translocation of the Notch intracellular domain (NICD) to the nucleus, activating transcription of Notch target genes (Brou et al., 2000; Bozkulak and Weinmaster, 2009; Mumm et al., 2000; Kopan and Ilagan, 2009). Studies on Notch1-deficient mice and chimeric mice generated from both wild-type and Notch1-deficient cells demonstrated that Notch1 signaling is required for HSC specification in a cell-autonomous manner (Kumano et al., 2003; Hadland et al., 2004).

In zebrafish, HSC specification requires three Notch genes, notch1a, 1b and 3, which are expressed in the somites and the posterior lateral plate mesoderm (PLPM) (Kim et al., 2014). Kim et al. showed that loss of notch1a impairs both HSC and DA formation, whereas loss of notch1b or notch3 results in loss of HSCs but does not affect DA formation. These data suggest that Notch1a is required for both HSC specification and aortic formation, whereas Notch1b and Notch3 are essential for HSC specification only. Unlike Notch1, however, Notch3 non-cell-autonomously regulates HSC specification. Enforced expression of NICD in somitic cells has been shown to rescue HSC formation in notch3-deficient zebrafish (Kim et al., 2014), providing evidence that Notch signaling within the somites indirectly regulates HSC specification.

The somitic Notch ligands Delta-c (Dlc) and Dld have been shown to be required for HSC specification in zebrafish. A non-canonical Wnt ligand, Wnt16, regulates the expression of both dlc and dld in somites (Clements et al., 2011). During somitogenesis, angioblasts emerging in the PLPM migrate axially along the ventral domain of the somites, where Dlc and Dld are expressed. A portion of migrating angioblasts adhere tightly to the somites based on the interaction of junctional adhesion molecules or integrins, and this intimate contact promotes efficient Dlc/Dld signal transduction from the somites to the angioblast (Kobayashi et al., 2014; Rho et al., 2019). Shortly after reaching the midline, angioblasts that received high levels of Notch signaling begin to express gata2b, which is essential for EHT and is directly regulated by Notch signaling (Butko et al., 2015; Robert-Moreno et al., 2005). Notch signaling from the somites thus plays an important role in the establishment of HSC fate in angioblasts. Here, we show in zebrafish that an additional somitic Notch ligand, Jagged 2b (Jag2b), promotes HSC specification by regulating wnt16 expression. Jag2 signaling has been shown to promote the survival and proliferation of hematopoietic progenitors in the murine bone marrow, and it is also involved in T-cell differentiation in the thymus (Tsai et al., 2000; Van de Walle et al., 2011). However, the role of Jag2 in HSC development has not been established in mammals or zebrafish. In the zebrafish embryo, we detected expression of jag2b in the somites as early as 11 h post-fertilization (hpf), when wnt16 was not yet expressed. Loss of jag2b resulted in a loss of wnt16 expression in the somites and a loss of HSCs in the DA, whereas wnt16-expressing cells were adjacent to Notch-activated cells in the somites. We found that Ephrin A1b mediates this intercellular signaling within the somites. In summary, we have identified signal transduction mechanisms of the Notch pathway that regulate HSC specification.

To investigate the expression pattern of jag2b in the zebrafish embryo, we first performed whole-mount in situ hybridization using jag2b-specific probes. Expression of jag2b was predominantly detected in the nervous system, somites, and intermediate mesoderm, which gives rise to the kidney. The expression of jag2b in the somites shifted anteroposteriorly during somitogenesis: it was first detected in segmented somites at 11 hpf [3 somite-stage (ss)], posteriorly elongated by 16 hpf (14 ss), and downregulated in anterior somites after 18 hpf (18 ss) (Fig. 1A-H). In addition, jag2b expression was also altered laterally within the somites, particularly at 13-14 hpf (8-10 ss) (Fig. 1D,E). Histological analysis revealed that jag2b expression at 13 hpf was detected throughout the somite, including both the lateral and adaxial domains of the somite (Fig. 1I), which later form fast and slow muscle fibers, respectively (Stickney et al., 2000). At 15 hpf (12 ss), however, expression was restricted to the adaxial domain (Fig. 1J), suggesting that jag2b expression is axially downregulated within the somites during somitogenesis. At 24 hpf, jag2b expression was detected in the ventral domain of the neural tube, lateral line, pronephros, and somites in the tail region (Fig. 1H,K).

To investigate the role of Jag2b in HSC formation in the zebrafish embryo, we utilized a gene knockdown method based on the CRISPR/Cas9 system, in which injection of four single guide RNAs (sgRNAs) redundantly targeting a single gene results in the recapitulation of null phenotypes (Wu et al., 2018). Four different sgRNAs targeting the jag2b gene were co-injected with Cas9 mRNAs in one-cell-stage embryos. The efficiency of jag2b knockdown was examined by quantitative reverse transcription PCR (qRT-PCR) using primer sets that recognize two different sgRNA target sites. We observed >96% reduction in jag2b expression in jag2b sgRNA-injected embryos (jag2b^sgRNA embryos) (Fig. S1A,B), indicating that jag2b function is largely blocked in jag2b^sgRNA embryos. Expression of an HSC marker gene, runx1, was detected in the DA of wild-type embryos, whereas this was largely reduced in jag2b^sgRNA embryos (Fig. 2A). Morpholino knockdown of jag2b also showed decreased expression of runx1 (Fig. S1C,D) and another HSC marker, cmyb, in the DA (Fig. 2B). Unexpectedly, however, genetic mutants of jag2b did not recapitulate the expression pattern of runx1 in jag2b^sgRNA embryos or jag2b morphants (Fig. S2A-E), suggesting that genetic compensation may occur in jag2b mutant embryos (Rossi et al., 2015). Indeed, we observed upregulation of jag1a and jag1b in jag2b mutant embryos, whereas the expression of these two genes was unchanged in jag2b^sgRNA embryos or jag2b morphants (Fig. S2F,G). To visualize developing HSCs, we utilized a gene trap line, gSAIzGFFM1770A, which contains Gal4FF in exon 3 of gata2b (hereafter denoted as gata2b-Gal4FF) (Kawakami et al., 2010). When combining gata2b-Gal4FF with UAS:GFP and fli1:lifeact-mCherry lines, GFP-expressing cells could be continuously detected in the ventral floor of the DA at 26 hpf, suggesting that the gata2b-Gal4FF; UAS:GFP line precisely recapitulates endogenous gata2b expression. Compared with wild-type embryos, the number of gata2b⁺ cells was significantly lower in jag2b^sgRNA embryos (Fig. 2C).

Given that hematopoietic and vascular endothelial cells develop in close proximity to each other within the PLPM, we next investigated whether loss of jag2b affects vascular formation or primitive hematopoiesis. Expression of fli1 (a pan-endothelial marker), efnb2a (a DA marker) and gata1a (an erythrocyte marker) was intact in jag2b^sgRNA embryos (Fig. 2D-F). Furthermore, expression of the PLPM markers fli1 and npas4l was also unchanged in jag2b^sgRNA embryos at 14 hpf (Fig. S3A,B), suggesting that loss of jag2b specifically affects HSC formation. To investigate further the role of Jag2b in HSC formation, jag2b was overexpressed in wild-type embryos by injection of jag2b mRNA. Expression of runx1 in the DA was upregulated by injection of 50 pg jag2b mRNA, whereas it was downregulated by injection of 150 pg mRNA (Fig. S3C). These data suggest that Jag2b promotes HSC formation in the zebrafish embryo, although extremely high doses disrupt this process.

Because Jag2b is a Notch ligand expressed predominantly in the somites, we next examined the spatiotemporal distribution of Notch-activating cells in the somites using a Notch reporter line, Tp1:GFP, which expresses GFP under the control of Notch-responsive elements (Parsons et al., 2009). Expression of Tp1:GFP was weakly detected in the somites at 11 hpf and upregulated by 13 hpf (Fig. 3A-F), a time window when jag2b was entirely expressed within the somites (Fig. 1B-D,I). Histological analysis revealed that the majority of Tp1:GFP-expressing cells were adjacent to jag2b-expressing cells at 13 hpf (Fig. 3G). Moreover, Tp1:GFP expression was reduced in the somites of jag2b^sgRNA embryos compared with wild-type embryos (Fig. 3H), suggesting that Notch signaling in the somites is activated by Jag2b. To test whether ectopic activation of Notch signaling in the somites is sufficient to rescue HSCs in jag2b^sgRNA embryos, expression of a dominant activator of the Notch pathway, NICD, was forced in the somites using a double-transgenic line, phldb1:Gal4; UAS:NICD, which expresses NICD under the control of somite-specific phldb1 enhancers (Burns et al., 2005; Kim et al., 2014). The expression of runx1 in the DA was partially restored in NICD (+) jag2b^sgRNA embryos (Fig. 3I), indicating that Jag2b-driven Notch signaling in the somites regulates HSC formation in the zebrafish embryo.

Given the requirement of Jag2b-driven Notch signaling in the somites for HSC formation, the expression of some somite markers was investigated in Jag2b-deficient embryos. Expression of both desma (a somite marker) and nkx3-1 (a sclerotome marker) was unaffected in jag2b^sgRNA embryos at 15 hpf (Fig. 4A,B), indicating that loss of jag2b does not affect somite and sclerotome formation. Although the expression of notch1a and 1b was slightly upregulated, the expression of notch3 in the somites was unchanged in jag2b morphants (Fig. S4A-C). However, we observed a reduction in wnt16 expression in the somites of both jag2b^sgRNA embryos and jag2b morphants (Fig. 4C, Fig. S4D). Consistent with this, loss of jag2b also reduced both dlc and dld expression in the somites (Fig. 4D,E, Fig. S4E,F), which are the downstream target genes of Wnt16 (Clements et al., 2011). Conversely, injection of 50 pg jag2b mRNAs in wild-type embryos increased the expression of wnt16, dlc and dld in the somites (Fig. S4G-I). To test whether forced expression of dlc and dld can rescue HSC formation in Jag2b-deficient embryos, as has been shown in wnt16 morphants (Clements et al., 2011), dlc and dld mRNAs were co-injected with jag2b MO into embryos. The expression of runx1 in the DA was restored by co-injection of dlc and dld mRNA in jag2b morphants (Fig. S4J). These data suggest that Jag2b-driven Notch signaling regulates HSC formation by regulating the Wnt16–Dlc/Dld signaling axis in the somites.

Expression of wnt16 was first detected at 14 hpf in anterior somites and was posteriorly elongated during somitogenesis (Fig. 4F-J). To determine whether Notch-activated cells do indeed express wnt16, the expression pattern of wnt16 was examined in the somites of Tp1:GFP embryos. Interestingly, histological analysis revealed that the majority of wnt16-expressing cells were adjacent to, but not merged with, Tp1:GFP-expressing cells in the medial region of the somites at 15 hpf (Fig. 4K,L). The distribution of wnt16- and Tp1:GFP-expressing cells at 15 hpf was similar to that of jag2b- and Tp1:GFP-expressing cells at 13 hpf; the two cell types were intermittently distributed to neighbor each other in the somites (Fig. 3G). These data suggest that Jag2b-driven Notch signaling indirectly regulates wnt16 expression, and raise the possibility that additional paracrine/juxtacrine signaling may mediate this intercellular signal transduction.

Ephrin–Eph signaling is transduced between neighboring cells and is known to be involved in the formation of somite boundaries (Durbin et al., 1998; Davy and Soriano, 2007; Watanabe et al., 2009; Watanabe and Takahashi, 2010). In zebrafish, two Ephrin genes, efna1b and efnb2a (encoding Ephrin A1b and B2a, respectively), have been shown to be expressed in somites (Durbin et al., 1998). As Ephrin signaling fits our model of intercellular signaling in the somites, we next investigated whether efna1b or efnb2a are regulated by Jag2b. Expression of efna1b was reduced in the somites of jag2b^sgRNA embryos and jag2b morphants, although the expression of efnb2a was unchanged (Fig. 5A,B, Fig. S5A,B). Injection of jag2b mRNAs conversely increased the expression of efna1b, but not efnb2a, in the somites (Fig. S5C,D), suggesting that efna1b expression in somites is regulated by Jag2b. Expression of efna1b in the somites was first detected at 13 hpf and increased by 14 hpf (Fig. 5C-G), a time window when Notch signaling is highly active in somites (Fig. 3C,D). Moreover, efna1b and Tp1:GFP expression mostly overlapped in the somites at 14 hpf (Fig. 5H-J). Double-fluorescence whole-mount in situ hybridization using probes for efna1b and wnt16 revealed that efna1b-expressing cells were laterally and anteroposteriorly adjacent to wnt16-expressing cells at 15 hpf (Fig. 5K). Similar to jag2b expression, efna1b expression shifted predominantly to the adaxial region of the somite at 16 hpf, whereas wnt16 expression was localized to the dorsomedial region of the somite (Fig. S5E,F). To determine whether Ephrin A1b is involved in HSC formation, four different sgRNAs targeting the efna1b gene were co-injected with Cas9 mRNA into embryos (Fig. S6A,B). Embryos injected with efna1b sgRNAs (efna1b^sgRNA) showed a reduction in runx1 expression in the DA and wnt16, dlc and dld expression in the somites. In contrast, jag2b, desma and nkx3-1 expression in the somite remained intact (Fig. 5L,M, Fig. S6C-G). Consistent with these observations, a genetic mutant line of efna1b, efna1b^kz5, which contains a premature stop codon in exon 2 of efna1b, also showed reduced runx1 expression in the DA and wnt16, dlc and dld expression in the somites (Fig. S7A-F). These data suggest that loss of efna1b phenocopies loss of jag2b.

In order to confirm the role of Ephrin A1b in wnt16 regulation, efna1b mRNAs were injected into embryos. Overexpression of efna1b in wild-type embryos increased wnt16 expression in the somites and runx1 expression in the DA (Fig. 6A,B). Co-injection of efna1b mRNAs together with jag2b sgRNAs and Cas9 mRNAs resulted in recovery of runx1 expression in the DA (Fig. 6C). Furthermore, co-injection of wnt16 mRNAs together with efna1b sgRNAs and Cas9 mRNAs rescued runx1 expression in the DA (Fig. 6D). Taken together, these data strongly suggest that Jag2b-driven Notch signaling regulates efna1b expression in the somite to modulate the Wnt16–Dlc/Dld signaling axis, which is required for HSC specification.

In the present study, we have shown in the zebrafish embryo that Jag2b regulates HSC specification by triggering intercellular signal transduction in the somites. Expression of jag2b in the somites is initiated as early as 11 hpf (3 ss), activating Notch signaling in the segmented somite at 11-13 hpf (3-8 ss). Jag2b-driven Notch signaling then regulates efna1b expression at 13-14 hpf (8-10 ss). Ephrin A1b transduces signals to neighboring cells to drive wnt16 expression at 14-15 hpf (10-12 ss), which modulates the expression of dlc and dld to establish HSC fate in angioblasts during axial migration at 15-17 hpf (12-16 ss) (Clements et al., 2011; Kobayashi et al., 2014). Thus, our data clearly demonstrate spatiotemporal regulatory mechanisms of intercellular signal transduction in the somites that are required for HSC specification (Fig. 7).

In zebrafish, R-spondin 1 (Rspo1) has been implicated in the regulation of wnt16 expression. R-spondins have been shown to regulate various developmental and physiological processes through enhancement of both canonical and non-canonical Wnt signaling (Glinka et al., 2011; Jin and Yoon, 2012). Loss of rspo1 in the zebrafish embryo has been shown to result in reduced expression of wnt16 in the somites and impaired HSC formation in the DA. The HSC formation defect in rspo1-deficient embryos could be rescued by injection of dlc and/or dld mRNAs, suggesting that Rspo1 regulates HSC formation via the Wnt16–Dlc/Dld signaling axis (Genthe and Clements, 2017). However, as rspo1 is broadly expressed in the zebrafish embryo, it is unclear how Rspo1 spatiotemporally regulates wnt16 expression in the somites. Our data demonstrate that wnt16 expression was initiated shortly after Notch-activated cells were detected in the somites (14-15 hpf). During this time window, wnt16 expression was induced by neighboring Notch-activated cells via Ephrin A1b signaling. Loss of either jag2b or efna1b decreased wnt16 expression in the somites, but overexpression of either one increased wnt16 expression. These observations suggest that spatiotemporal expression of wnt16 is predominantly regulated by Jag2b-driven Notch signaling, and Rspo1 may boost wnt16 expression in the somite.

The expression domain of jag2b in the somite is dynamically altered at early somitogenesis, and this spatiotemporal pattern of jag2b reflects the distribution of efna1b- and wnt16-epxressing cells. At 11-13 hpf, jag2b was expressed entirely in the anterior somites, which in turn activated Notch signaling in neighboring cells at 12-13 hpf. At 14 hpf, however, jag2b expression was restricted to the adaxial domain of somites. Similarly, efna1b and wnt16 expression was also detected in the entire domain of the somite at 13-14 hpf and 14-15 hpf, respectively, whereas they were later detected mainly in the adaxial to dorsomedial domain. Given that HSC fate is established by receiving Dlc–Dld signaling during angioblast migration along the ventral surface of the sclerotome (15-17 hpf) (Kobayashi et al., 2014), dlc and dld expression in the sclerotome could be induced by Wnt16 at 14-15 hpf, a time window when wnt16 is entirely expressed in the somites. Collectively, these observations suggest that, although the expression of jag2b, efna1b and wnt16 is shifted during somitogenesis, each requisite signal for HSC specification is transmitted to neighboring somitic cells within a very narrow timeframe (∼1 h).

Ephrin–Eph signaling plays pivotal roles in neural tube patterning (Gale and Yancopoulos, 1997; Stolfi et al., 2011), neural crest cell migration (Smith et al., 1997; Wang and Anderson, 1997; Santiago and Erickson, 2002; Davy and Soriano, 2007) and somite boundary formation (Durbin et al., 1998; Davy and Soriano, 2007; Watanabe et al., 2009; Watanabe and Takahashi, 2010). In mammals, Eph B4-expressing HSCs interact with Ephrin B2-expressing mesenchymal stromal cells, and this interaction triggers detachment of HSCs from stromal cells, leading to erythroid differentiation of HSCs (Foo et al., 2006; Suenobu et al., 2002). Within the presomitic mesoderm, when an Eph A4-expressing cell is adjacent to an Ephrin B2-expressing cell, these cells repel each other, resulting in boundary formation followed by epithelialization at this somite boundary (Watanabe et al., 2009; Watanabe and Takahashi, 2010). In zebrafish, blocking Eph signaling by injecting mRNAs encoding a dominant-negative form of EphA4 resulted in abnormal boundary formation in somites. Similar phenotypes were also observed with overexpression of soluble Ephrin A1b or B2a, both of which can bind to EphA4 to block its signaling (Durbin et al., 1998). However, we did not observe somite defects in efna1b^sgRNA or efna1b^kz5 embryos, as evidenced by the expression of desma and nkx3-1. These observations suggest that, although Ephrin A1b can potentially bind to EphA4, Ephrin A1b is dispensable for boundary formation in somites. Expression of efna1b and epha4 was detected not only in the somite boundary but also in the middle region of the somite where no boundary formed (Durbin et al., 1998). Clements et al. reported that loss of wnt16 led to reduced expression of sclerotome maker genes at 22 hpf, a time point after sclerotome migration adjacent to the neural tube and notochord (Clements et al., 2011). In contrast, we observed intact expression of the sclerotome marker nkx3-1 in jag2b^sgRNA and efna1b^sgRNA embryos at 15 hpf despite a reduction in wnt16 expression. Because wnt16 expression begins at 14 hpf, it is likely that early sclerotome formation at the ventromedial surface of the somite is independent of wnt16 expression. Previous cell-lineage analysis in zebrafish demonstrated that somitic cells within the newly segmented somite undergo rearrangement relative to their neighbors prior to whole-somite rotation (Hollway et al., 2007), indicating that somitic cells rearrange their positions dynamically during somitogenesis. In the intestinal epithelium, Ephrin–Eph signaling in epithelial cells drives Wnt signaling to modulate the spatial distribution of cells in the crypt-villus axis (Batlle et al., 2002, 2005). Although further studies are needed to determine whether Ephrin A1b controls somitic cell rearrangements, Ephrin A1b is a crucial mediator of Jag2b-triggered intercellular communication, representing a previously unappreciated cellular mechanism in HSC development.

There are at least five Notch ligands that are involved in establishing HSC programs in the zebrafish embryo. Dlc and Dld are involved in early HSC specification during angioblast migration (Clements et al., 2011; Kobayashi et al., 2014), and we have shown here that the expression of these Notch ligand genes is regulated in part by Jag2b-driven Notch signaling in the somites. It has been shown that Delta-like 4 (Dll4) is required for specification of the arterial endothelium; however, knockdown of dll4 has also been shown to lead to loss of HSCs, suggesting that arterial programs are required before formation of HECs (Bonkhofer et al., 2019). Within the aortic floor, Jag1a expressed by arterial endothelial cells activates Notch in neighboring endothelial cells to establish HSC programs (Espín-Palazón et al., 2014; Monteiro et al., 2016). Thus, multiple waves of Notch signaling provided by various cell types and ligands are involved in HSC development. Further studies of Notch-related genes will elucidate the spatiotemporal regulatory mechanisms of HSC development.

Zebrafish strains AB*, jag2b^hu3425, jag2b^kz6, jag2b^kz7, efna1b^kz5, Tg(Tp1:GFP)^um14 (Parsons et al., 2009), gSAIzGFFM1770A (gata2b-Gal4FF), Tg(UAS:GFP) (Asakawa et al., 2008), Tg(fli1:lifeact-mCherry)^ncv7 (Wakayama et al., 2015), Et(phldb1:Gal4-mCherry) (Distel et al., 2009) and Tg(UAS:NICD)^kca3 (Burns et al., 2005) were raised in a circulating aquarium system (AQUA) at 28.5°C on a 14 h/10 h light/dark cycle and maintained in accordance with guidelines from the Committee on Animal Experimentation of Kanazawa University.

Previously designed guide RNA (gRNA) sequences were utilized (Wu et al., 2018) and are listed in Table S1. sgRNAs were synthesized as previously described (Kobayashi-Sun et al., 2020). Briefly, a single-strand DNA oligo containing the gRNA target and T7 promoter sequence was annealed with the gRNA scaffold primer, followed by synthesizing the double-strand DNA with MightyAmp DNA polymerase (Thermo Fisher Scientific). sgRNAs were then synthesized by in vitro transcription using the MEGAshortscript T7 Transcription Kit (Thermo Fisher Scientific) and purified with the mirVana miRNA Isolation Kit (Thermo Fisher Scientific) according to the manufacturer's protocol. For knockdown experiments, embryos were injected with a mixture of four different sgRNAs (200 ng/µl each) and Cas9 mRNA (100 ng/µl) at the one-cell stage. For generation of a mutant line, embryos injected with a single sgRNA and Cas9 mRNA were raised to adulthood to obtain the F1 generation. Fish from the F1 generation were screened using primers that flanked the target region, and mutant alleles were identified by sequencing. The gRNA scaffold primer and genotyping primers are listed in Table S1.

cDNAs were cloned by reverse transcription-polymerase chain reaction (RT-PCR) using the specific primers listed in Table S1 and ligated into the pCRII-TOPO vector (Invitrogen). Digoxigenin (DIG)- and fluorescein-labeled RNA probes were prepared by in vitro transcription with linearized constructs using the DIG RNA Labeling Kit (SP6/T7; Sigma-Aldrich) and Fluorescein RNA Labeling Mix with T7 polymerase (Sigma-Aldrich), respectively. For permeabilization, embryos fixed with 4% paraformaldehyde in PBS were treated with proteinase K (10 μg/ml) (Sigma-Aldrich) for 30 s to 15 min at room temperature or acetone for 7 min at −20°C, refixed with 4% paraformaldehyde, and washed twice 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 (Sigma-Aldrich) in PBST and incubated overnight at 4°C with alkaline phosphatase-conjugated anti-DIG antibody (Sigma-Aldrich, 1093274) at 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 double-fluorescence in situ hybridization, embryos were simultaneously hybridized with DIG-labeled efna1b and fluorescein-labeled wnt16 probes, followed by staining with horseradish peroxidase (HRP)-conjugated anti-DIG antibody (Sigma-Aldrich, 11207733910) at 1:500. Embryos were developed using the TSA Plus Cyanine 3 System (Perkin Elmer). After quenching with 0.2% H₂O₂ in methanol, embryos were stained with HRP-conjugated anti-fluorescein antibody (Sigma-Aldrich, 11426346910) at 1:100, followed by development with the TSA Plus Fluorescein System (Perkin Elmer). For immunohistochemistry, embryos were embedded in paraffin and sectioned at 3 μm thickness. Deparaffinized tissue sections were incubated with Tris-EDTA buffer (10 mM Tris, 1 mM EDTA, 0.05% Tween-20) at 90°C for 90 min for antigen retrieval. Sections were then blocked with 0.2% bovine serum albumin in PBS for 60 min at room temperature and stained with chicken anti-GFP (Aves Labs, GFP-1020) at 1:1000 at 4°C overnight, followed by staining with anti-chicken IgY HRP-conjugated secondary antibody (Abcam, ab6877) at 1:1000 at room temperature for 1 h. Sections were developed with 3, 3'-diaminobenzidine (DAB) substrate solution (Wako Chemicals) for 1-2 min. After washing, sections were mounted with mounting medium (Mount-Quick Aqueous, Daido Sangyo). For whole-mount immunohistochemistry, fixed embryos were blocked with 2% blocking reagent, and incubated overnight at 4°C with chicken anti-GFP and/or rabbit anti-RFP (Abcam, ab34771) at 1:1000. After washing with PBST, embryos were incubated overnight at 4°C with goat anti-chicken IgY Alexa Fluor 488-conjugated (Abcam, ab150173) and/or donkey anti-rabbit IgG Alexa Fluor 647-conjugated (Abcam, ab150115) at 1:1000.

Total RNA was extracted from embryos using the RNeasy Mini Kit (Qiagen), and cDNA was synthesized using the ReverTra Ace qPCR RT Master Mix (Toyobo). qRT-PCR assays were performed using TB Green Premix Ex Taq II (Takara Bio) on a ViiA 7 Real-Time PCR System following the manufacturer's instructions (Thermo Fisher Scientific). Expression of ef1a was used for normalization. Primers used for qRT-PCR are listed in Table S1.

For fluorescence imaging, embryos were mounted in a glass-bottomed dish filled with 0.6% low-gelling agarose (Sigma-Aldrich) in E3 medium and imaged using a FV10i confocal microscope and Fluoview FV10i-SW software (v2.1.1) (Olympus). Visible light imaging was captured using an Axiozoom V16 microscope (Zeiss) with a TrueChrome II digital camera (BioTools) and TCapture software (ver. 4.3.0.602) (Tucsen Photonics) or an Axioplan2 microscope (Zeiss) with a Moticam Pro 285B digital camera and Motic Images Plus 2.3S software (v2.3.4) (Shimazu).

The morpholino oligo (MO; GeneTools) sequence against jag2b is as follows: TCCTGATACAATTCCACATGCCGCC (Lorent et al., 2004). Capped mRNAs were synthesized from linearized pCS2+ constructs using the mMessage mMachine SP6 kit (Thermo Fisher Scientific). Embryos were injected at the one-cell stage with 1 nl of MOs and/or mRNAs at the following concentrations: jag2b MO, 400 μM; Cas9 mRNA, 100 pg/nl; jag2b mRNA, 50-150 pg/nl; dlc mRNA, 50 pg/nl; dld mRNA, 50 pg/nl; efna1b mRNA, 100 pg/nl; wnt16 mRNA, 100 pg/nl.

Data were analyzed for statistical significance after at least two repeated experiments. To quantify the expression levels of each tested gene, individual embryos were classified into three (high, middle or low) or two (high or low) categories based on the signal intensity. Statistical differences of phenotypic distributions between groups were determined using Pearson's χ² test. The statistical difference in count data between groups was determined by unpaired two-tailed Student's t-test. Values of P<0.05 were considered statistically significant.

We thank K. Kawakami for providing the gSAIzGFFM1770A and UAS:GFP line; S. Fukuhara for providing the fli1:lifeact-mCherry line; R. Wong and M. Hazawa for supporting confocal imaging; and K. Lewis for critical review of the manuscript.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.200339.