BF170 hydrochloride enhances the emergence of hematopoietic stem and progenitor cells

Developing an efficient differentiation method for producing abundant and safe therapeutic and transplantable hematopoietic stem and progenitor cells (HSPCs) ex vivo is still challenging. Suitable development conditions or cultured media can transform some of the yolk sac progenitors to definitive hematopoietic stem cells from their primary hematopoietic cell fate, if they are grafted into neonates (Xu et al., 2001; Yoder et al., 1997), or when they are cultured on stroma taken from the para-aortic region of the embryos (Matsuoka et al., 2001; Vodyanik et al., 2005). Recently, studies have unveiled a defined albumin-free culture system that has been developed to support the long-term ex vivo expansion of functional mouse HSPCs, and polyvinyl alcohol has been found to be a functionally superior replacement for serum albumin to sustain HSPC self-renewal ex vivo and to afford between 236- and 899-fold expansions of functional HSPCs over 1 month (Wilkinson et al., 2019).

The hematopoiesis process is conserved from zebrafish to mammals (Orkin and Zon, 2008). The theory that cells transform from mesoderm to hemogenic endothelium (HE) then to HSPCs guides the crucial process of inducing HSPC-like cells from a variety of cell types ex vivo and in vivo (Blaser and Zon, 2018; Vogeli et al., 2006). Some compounds, such as nitric oxide (NO) donors, enhance HSPC induction in vivo after the heartbeat begins in zebrafish, which is specific to the HSPC induction, with almost no effects on the vasculature, with the exception of SNAP (S-nitroso-N-acetylpenicillamine) (North et al., 2009). NO is a well-established direct regulator of blood flow, several NO donors or inhibitors are known modulators of heartbeat and blood flow (North et al., 2009), and blood flow-dependent klf2a-NO signaling is required for stabilization of HSPC programming in zebrafish embryos (Wang et al., 2011). Meanwhile, primary cilia on endothelial cells transduce Notch signaling to the earliest HE for proper HSPC specification during embryogenesis (Liu et al., 2019). NO donors act downstream of Notch signaling in regulating arterial identity and HSPC induction (North et al., 2009), suggesting a time-dependent signaling in HE for HSPC induction by the interaction complex of ex vivo chemical compounds with in vivo genes.

Using ex vivo zebrafish blastomere screen system, studies have identified chemical compounds that can induce blood vessel cells (Huang et al., 2012a), muscle progenitor cells (Xu et al., 2013) and neural crest cells (Ciarlo et al., 2017), and suppress MYB in adenoid cystic carcinoma (ACC) (Mandelbaum et al., 2018). The induction or suppression effects of the aforementioned chemical compounds are conserved from fish to mammals (Ciarlo et al., 2017; Xu et al., 2013). In 2011, we used a similar ex vivo zebrafish blastomere screen system from cmyb-GFP and mpx (myeloid-specific peroxidase)-GFP transgenic embryos, respectively, to screen nearly 2400 chemicals from a bioactive drug library (Xu et al., 2013). We found that 25 compounds could significantly enhance cmyb-derived GFP in primary cultured cells, and nine compounds could enhance mpx-derived GFP. Meanwhile, eight of the 25 compounds could also significantly increase the expressions of the hematopoietic genes tested in mouse embryoid bodies (EBs). Among the overlapping compounds, BF170 hydrochloride (C₁₅H₁₂N₂·HCl, abbreviated as BF170 hereafter) was chosen for the following functional studies.

BF170 has been reported to affect retinol (vitamin A) levels in cells (Chen et al., 2016) and to increase NFAT (nuclear factor of activated T cells) promoter-driven luciferase (Akiba et al., 2016), although its mechanism of action is still unknown. It is well known that retinol signaling and calcium/NFAT signaling play important roles in hematopoietic development (Ghatpande et al., 2002; Müller et al., 2009; Rönn et al., 2015), suggesting that BF170 might function potentially in hematopoiesis. In this study, we have found that the compound BF170 works effectively on apelin receptor-positive lateral plate mesoderm (LPM) cells and endothelial progenitor cells (EPCs) in human embryonic stem cells (H1 hESCs), and also on HE and HSPC specification before the heartbeat is established during zebrafish embryogenesis. Besides its roles in definitive hematopoiesis, we demonstrate that BF170 hydrochloride also works effectively on primary hematopoiesis. Furthermore, we also reveal previously unreported mechanisms underlying the effects of BF170, i.e. the increase in blood flow, the subsequent activation of primary cilia-Ca²⁺-Notch/NO signaling and, finally, the expansion of HSPC emergence during zebrafish embryogenesis.

Zebrafish embryonic pluripotent blastomere culture system has recently been successfully established and used to examine various biological processes (Ciarlo et al., 2017; Huang et al., 2012a; Xu et al., 2013). Meanwhile, this culture screening system for finding modulators of cmyb, a marker of HSPCs, has successfully found that retinoic acid agonists are potent suppressors of cmyb expression and adenoid cystic carcinoma (ACC) (Mandelbaum et al., 2018). Here, using the same cmyb-GFP embryonic pluripotent blastomere culture system (Mandelbaum et al., 2018) together with mpx-GFP embryonic pluripotent blastomere culture system (Fig. S1A,B), we identified the compounds enhancing cmyb expression, which might be the potential inducers of HSPCs ex vivo.

In this study, the zebrafish Tg (cmyb: GFP) and Tg (mpx: GFP) embryos were dissociated into single cells at the sphere stage [4 hours post-fertilization (hpf)], and plated in 384-well plates, with chemicals added 10 h later; on the third day, the fluorescence and gene expression were evaluated (Fig. S1A,B and Fig. 1A), similar to that recently reported (Mandelbaum et al., 2018). The sorted cmyb-GFP⁺ cells derived in culture contained a mixture of blood, neuroepithelial and retinal cell types (Mandelbaum et al., 2018), and exhibited abundant expression of both cmyb and runx1 (Fig. 1A), which are well-known HSPC markers in vertebrates, suggesting that the ex vivo cmyb-GFP⁺ cells successfully recapitulated the endogenous gene expression in zebrafish HSPCs. Meanwhile, the sorted mpx-GFP-expressed cells derived in culture exhibited abundant expression of both mpx and runx1, which successfully recapitulated the endogenous gene expression of in vivo embryonic mpx⁺ cells (Fig. S1B). The observations here not only demonstrated that cells from the zebrafish embryonic pluripotent blastomere culture system still possessed their corresponding cell properties, but also suggested that the ex vivo embryonic pluripotent blastomere culture system could be used in chemical compound screening for inducers of HSPCs in this study.

Based on the observations that cultured cmyb-GFP⁺ cells recapitulated the endogenous gene expression of HSPCs (Fig. 1A), we focused on screening compounds that could increase cmyb expression ex vivo. By screening nearly 2400 chemicals from a bioactive drug library (Xu et al., 2013), we found that 37 hits increased cmyb-GFP fluorescence of the cultured blastomere in the primary screen process, and 25 of the 37 hits (nearly 1% of the total compounds screened) exhibited a dose-dependent increase in the fluorescence of the cultured cmyb-GFP cells (Fig. S1D and Table S5). Nine chemicals, including BF170, also increased mpx-derived GFP signals (Table S5, highlighted in red). Next, we tested the in vivo induction effects of the 25 hits in zebrafish embryos. The chemicals were added to the embryos at 10 hpf and cmyb expression in the embryos analyzed at around 18 hpf (8 h after the chemicals were added). We found that 13 of the hits increased cmyb expression in zebrafish embryos (Table S6), including BF170 (Fig. 1B and Fig. S1D), which is a natural product with multiple biological activities and targets (Nurani et al., 2020; Okamura et al., 2005).

Next, we tested the effects of the aforementioned 25 hits on hematopoiesis in ex vivo mouse embryonic cells. Visual inspection revealed that the 25 hits had normal embryoid body (EB) formation, and eight out of the 25 hits increased the expression of the tested hematopoietic genes, such as Myb, Mpo, Gata1, Lmo2, Runx1 and Tal1 (Fig. S1C). Among these, BF170 significantly increased the expression of all the genes evaluated (Fig. S1D).

Furthermore, the effects of BF170 on the ex vivo hematopoietic differentiation of H1 hESCs were assessed using a chemically defined hematopoietic differentiation system described recently (Pang et al., 2013; Wang et al., 2018). The results indicated that BF170 had a minor enhancing effect on the generation of APLNR⁺ (Apelin receptor) lateral plate mesoderm (LPM) cells on day 3 (Fig. 1D) while significantly increasing the population of CD31⁺CD34⁺ endothelial progenitor cells (EPCs) on day 6 of hematopoietic differentiation (Fig. 1E). Meanwhile, it had almost no effect on the production of CD34⁺CD43⁺ and CD34⁺CD45⁺ hematopoietic progenitors on day 9 (Fig. S2A) and day 14 (Fig. S2B), respectively. In contrast, CD43⁺ cobblestone-like HPCs (hematopoietic progenitor cells) were observed in BF170-treated cells at day 14 (Fig. S2B, top panel).

In the ex vivo system, treating zebrafish cmyb-GFP embryonic pluripotent blastomere cells with 10 µM of BF170 in culture significantly increased the percentage of cmyb-GFP positive cells (from 13.59±0.47% to 24.57±0.51%) (Fig. S3A), and increased the expressions of HSPC genes runx1 and cmyb in both ex vivo zebrafish blastomere cultured clusters (Fig. 1C) and mouse embryonic stem cells (Fig. S1C). We, therefore, further analyzed the effects of BF170 on hematopoiesis in vivo in developing zebrafish embryos.

Embryos were treated with 10 µM BF170 at 10 hpf and fixed for analysis at 14, 18 and 20 hpf. We found that BF170 induced significant increases in the expression of cmyb, lmo2, scl (tal1) and flk1 (kdrl) during primitive hematopoiesis (Fig. S3B,C). These positive effects of BF170 on lmo2 and scl started at 14 hpf (Fig. S3B), earlier than its effects on cmyb and flk1 (Fig. S3C). By western blot analysis, BF170 treatment improved the protein levels of Lmo2, Gata1 and Scl at both 18 and 20 hpf (Fig. S3D,E). These observations here suggested that BF170 was an effective inducer in primitive hematopoiesis.

To evaluate whether BF170 could also induce definitive hematopoiesis in embryos, we treated embryos with BF170 at 10 hpf and fixed the embryos at 28 hpf to assess definitive hematopoiesis. In the treated embryos, we observed significantly increased expressions of a naïve HSPC (nHSPCs) gene, cmyb, and of a progenitor-like hemogenic endothelial cell (hemECs) gene, runx1 (Fig. 2A). We also found enhanced immunofluorescence of anti-Runx1 antibody staining in the aorta-gonad-mesonephros (AGM) region (Fig. S4B), together with increased flk1⁺fli1a⁺ double-positive cells in the ventral part of dorsal aorta (DA) (where HE cells were located) in Tg (flk1: mCherry/fli1a: nls-GFP) (Fig. S4A). Meanwhile, cmyb and runx1 expression was also increased in flk1⁺cmyb⁺ cells sorted from the BF170-treated embryos (Fig. 2B), with significantly increased nHSPCs (flk1⁺cmyb⁺ cells) and hemECs (flk1⁺runx1⁺ cells) emerging from the HE in BF170-treated embryos at 28 hpf (Fig. 2C and Fig. S4C). Furthermore, BF170 also induced an increase in the expression of the vascular endothelium markers flk1 and fli1a (fli1), the arterial endothelium marker ephrin B2a (efnb2a) and the venous marker flt4 in the AGM or in whole embryos (Fig. 2D and Fig. S4D), and in flk1⁺cmyb⁺ cells (Fig. S4E) at 28 hpf, suggesting that BF170 acted positively to promote hematovascular fate and HE fate determination.

To test the effects of BF170 on HSPC and endothelial cell proliferation, we treated Tg (cmyb: GFP), Tg(runx1: GFP) and Tg (fli1a: nls-GFP) embryos with BF170 at 10 hpf, and collected the treated embryos and their control at 28 hpf for BrdU incorporation and immunostaining analysis. BF170 induced increased numbers of cmyb⁺BrdU⁺ (Fig. 2E) and runx1⁺BrdU⁺ double-positive cells (Fig. 2F) in the AGM regions, and an increase of fli1a⁺BrdU⁺ cells and fli1a⁻BrdU⁺ cell numbers (Fig. S4F), suggesting that BF170 had positive effects on the proliferation of HE, hemECs and nHSPC cells during the HSPC emergence process, although this effect might not be highly cell specific.

In addition, BF170 induced the upregulation of cmyb, flk1 and fli1a at later developmental stages (33, 48 and 72 hpf) (Fig. S5A), accompanied by the downregulations of the erythroid markers βe1, βe2 and βe3 (hbbe1, hbbe2 and hbbe3) and gata1 at 28 hpf, 33 hpf (Fig. S5B) and at 96 hpf (Fig. S5C), and of the lymphoid markers rag1 and rag2 at 96 hpf (Fig. S5D,E). Expression of the myeloid markers L-plastin (lcp1) and mpx declined at 28 hpf (Fig. S5B) and then increased at 96 hpf (Fig. S5C). These analyses suggest that BF170 might impair the blood lineage differentiation in primitive hematopoiesis and play biased roles in definitive hematopoietic lineage differentiation, while enhancing the emergence of HSPCs.

We then treated embryos with BF170 between 24 and 28 hpf after the beginning of blood circulation (Fig. S6A). The results showed that the expression levels of cmyb and flk1 (Fig. S6B), and the number of flk1⁺cmyb⁺ cells (Fig. S6C) were not altered by the BF170 treatment at 10 µM, whereas 15 µM BF170 effectively increased flk1 expression, but did not change cmyb expression (Fig. S6D).

During our studies on the effects of BF170 on HSPC induction, an interesting phenomenon we noticed was that BF170 increased blood flow in Tg (flk1: mCherry/runx1: GFP) embryos at 28 hpf (Fig. 3A, Movies 1 and 2). Studies have reported that higher flow velocity resulted in an increase in blood vessel size via outward remodeling (Baeyens and Schwartz, 2016; Langille and O'Donnell, 1986; Sugden et al., 2017; Tuttle et al., 2001). Consistently, we found that BF170 led to larger DA and PCV (posterior cardinal vein) diameters in Tg (flk1: mCherry/runx1: GFP) embryos than those in the control group (Fig. 3B). As blood flow velocity is associated with heart rate (Banjo et al., 2013; Sankari et al., 2017; Weijts et al., 2018), we measured the heartbeats of the embryos at 28 hpf and observed that BF170 slightly increased heart rate (Fig. S6E and Movies 3 and 4).

Blood flow has previously been reported to play an important role in HSPC development (Adamo et al., 2009; North et al., 2009; Wang et al., 2011), and primary cilia may lie downstream of blood flow to regulate HSPC specification (Liu et al., 2019). Therefore, we next examined the effects of BF170 on primary cilia at 28 hpf. We found that BF170 upregulated the expressions of cilia genes kif3a, ift88, and fsd1 in the early cmyb⁺flk1⁺ HSPC cells (Fig. 3C). Knockdown of fsd1, which is essential for ciliogenesis (Tu et al., 2018), almost diminished the effects of BF170 on HSPC induction (Fig. 3D-G).

Meanwhile, BF170 slightly increased the number of cilia and decreased the ciliary angle in the AGM region at 28 hpf (Fig. 4A). Treatment with CBD (ciliobrevin D), a small molecule that inhibits cilia function (Firestone et al., 2012; Liu et al., 2019), inhibited the promoting effects of BF170 on blood flow (Fig. S6F, Movies 5-8). Meanwhile, CBD treatment almost blocked the effects of BF170 on HSPC induction in zebrafish embryos (Fig. 4B-D), suggesting that cilia could potentially mediate the effects of BF170 on HSPC induction by sensing the increased blood flow.

In addition, RNA expression analysis of fli1a⁺ and flk1⁺ cells sorted from the control and the BF170-treated embryos at 20 hpf and 28 hpf not only showed the enrichment of hematopoiesis-associated KEGG terms, such as ‘T cell receptor signaling pathway’, ‘signaling pathways regulating pluripotency of stem cells’, ‘PI3K-Akt signaling pathway’, ‘MAPK signaling pathway’, ‘FoxO signaling pathway’ and ‘Notch signaling pathway’, etc. (Fig. S7A-D and Table S7), for the differentially expressed genes (DEGs), but also the enrichment of cilia, ciliary transition fiber and others in the sorted vascular and hemogenic endothelium (HE) cells from BF170-treated embryos (Fig. 5A, Fig. S7E and Table S8).

Notch signaling is notable regulator that acts downstream of primary cilia (Liu et al., 2019) to regulate HSPC emergence (Butko et al., 2016; Kim et al., 2013). BF170-induced DEGs in fli1a⁺ and flk1⁺ cells were significantly enriched in Notch-related GO terms, as mentioned above (Fig. 5A, Fig. S7E and Table S8). We therefore further tested the effects of BF170 on Notch signaling and the subsequent HSPC induction in embryos. The protein levels of NICD (Notch1 intracellular domain, an activated form of Notch1) and Notch1 were also increased (Fig. 5B, Fig. S8A), with an increase in Notch1 reporter fluorescence in Tg (tp1: EGFP) embryos treated with BF170 (Fig. 5C). BF170 treatment significantly increased expression of dll4, notch1a and notch3, and of Notch signaling target genes hey1, hey2, her1, her2 and her5 in early HSPCs, the cmyb⁺flk1⁺ cells (Fig. 5D). Additionally, the Notch signaling inhibitor DAPT significantly blocked the induction effects of BF170 on HSPC emergence (Fig. 5E,H) and HE induction (Fig. S8C). Similar tendencies were also observed in notch3^−/− and jag1a^−/− mutants, and in notch1a, notch1b and notch3 morphants (Fig. 5F,G,I,J, Fig. S8B).

The aforementioned results indicated that BF170 also positively affected primitive hematopoiesis and expression of the vessel genes flk1 and fli1a in zebrafish embryos before heartbeat. Primitive hematopoiesis and vessel development are tightly regulated by multiple signaling pathways, such as BMP, WNT, activin, TGFβ and VEGF (Flamme et al., 1995; Gupta et al., 2006; Kanatsu and Nishikawa, 1996; Park et al., 2004). Consistently, in this study, we also observed that BF170-induced DEGs in fli1a⁺ and flk1⁺ cells at 20 hpf were significantly enriched in VEGF signaling, TGFβ signaling, Wnt signaling and BMP signaling pathways (Fig. S7F and Table S9). The expression of bmp2b and bmp4 was also obviously increased in BF170-treated embryos at 14 hpf (Fig. S7G).

NO signaling is a well-established direct regulator of vascular tone and reactivity, thereby influencing blood flow (Baeyens et al., 2016; Lucitti et al., 2007; North et al., 2009). Meanwhile, NO acts downstream of Notch signaling in the induction of HSPCs (North et al., 2009), and downstream of blood flow to regulate the emergence of HSPCs (Wang et al., 2011). In this study, we observed increased blood flow (Fig. 3A, Movies 1 and 2) and increased klf2a expression (Fig. 6A) by BF170 treatment. Meanwhile, increased expressions of the NOS (nitric oxide synthase) genes nos1 and nos2b, which are downstream of klf2a (Wang et al., 2011), were observed in the cmyb⁺flk1⁺ cells sorted from BF170-treated embryos (Fig. 6B). L-NAME (nitro-L-arginine methyl ester), a NOS inhibitor (Thengchaisri et al., 2022), significantly inhibited the effects of BF170 on early HSPC and vascular induction in embryos (Fig. 6C,D and Fig. S9A), and knockdown of the heartbeat gene sih (tnnt2a) (Lancrin et al., 2009; Wang et al., 2011) also blocked the induction effects of BF170 on HSPC emergence (Fig. S9B), suggesting that BF170 might activate the blood flow and the subsequent NO signaling in embryos to induce HSPC production, consistent with the report that NO signaling is the direct link between blood flow and HSPC development (North et al., 2009; Wang et al., 2011).

Additionally, we tested the HSPC emergence in BF170-treated embryos co-treated with both Notch inhibitor DAPT and NO inhibitor L-NAME, and observed that simultaneous inhibition of Notch and NO signaling caused a greater reduction in HSPCs compared with DAPT or L-NAME inhibition alone. However, simultaneous inhibition of Notch and NO signaling did not completely suppress the effect of BF170 on HSPCs (Fig. S9C).

Calcium (Ca²⁺) signaling has been reported to be an essential regulator in hematopoietic progenitor maintenance and differentiation (Ho et al., 2021; Luchsinger et al., 2019; Shim et al., 2013). In this study, we observed the enrichment of Ca²⁺ signaling-related GO terms in the DEGs of the vascular and HE cells sorted from BF170-treated embryos (Fig. S10 and Table S10). Using Calcium Orange fluorescence live imaging, we found that BF170 induced a marked increase in intracellular Ca²⁺ levels (Fig. S11A,B and Movies 9 and 10) and Ca²⁺ accumulation around cilia in the AGM region in Tg (actb2: Arl13b-GFP) embryos at 28 hpf (Fig. S11B, marked by white arrowheads), accompanied with increased Ca²⁺ signaling in the ventral part of DA in Tg (flk1: mCherry/Cca.actb: GCaMP6 s) embryos at 28 hpf (Fig. 7A,B, marked by white arrowheads; Movies 11 and 12). Expression of pkd2, a calcium channel gene in cilia, was significantly increased in cmyb⁺flk1⁺ cells from BF170-treated embryos (Fig. 7C). Meanwhile, knockdown of pkd2 caused a significant decrease in Ca²⁺ signaling in the ventral endothelium of the DA (Fig. S11C, Movies 13 and 14), and obviously weakened the effects of BF170 on HSPC induction (Fig. 7D-G) and on HE induction (Fig. S11E, right panel). These results demonstrate that BF170 induced HSPC production by enhancing cilia Ca²⁺ signaling.

Next, we tested whether the increased Ca²⁺ signaling in BF170-treated embryos was related to Notch signaling and NO signaling. Knockdown of pkd2 resulted in the reductions of NICD protein levels and dll4 expression, and partially repressed the positive effects of BF170 on the induction of NICD and dll4 (Fig. S11D,E, left panel). In addition, pkd2 knockdown significantly decreased klf2a expression and blocked the induction of BF170 on klf2a (Fig. S11E, middle panel). These results indicate that BF170 might modify the effects of Notch signaling and NO signaling on HSPCs by enhancing cilia Ca²⁺ signaling.

Zebrafish chemical genetics screens have been used successfully to find that the drug PGE2 and NO donors induce HSPCs in vivo in vertebrates (North et al., 2009, 2007). Recently, a zebrafish chemical screen system using ex vivo cultured pluripotent blastomere cells has been successfully used to find compounds involved in the induction of specific cells (Ciarlo et al., 2017; Huang et al., 2012a; Xu et al., 2013) and in the suppression of ACC (Mandelbaum et al., 2018). Meanwhile, in a mammalian ex vivo compound screening system, polyvinyl alcohol has been found to sustain mouse HSPC self-renewal and to afford between a 236- and 899-fold increase in functional HSPCs over 1 month (Wilkinson et al., 2019). In this study, using the ex vivo screening system and cultured zebrafish cmyb-GFP and mpx-GFP pluripotent blastomere cells, we have successfully elucidated that the compound BF170 is a potential and previously unreported inducer of hematopoiesis ex vivo in both zebrafish pluripotent blastomere culture cells and mouse embryonic cells. Both ex vivo cmyb-GFP and mpx-GFP zebrafish pluripotent blastomere cells possess their respective cell properties, and mpx-GFP, which marks the myeloid lineage, serves as a control to show the drug-induced effects on the hematopoietic cells. During hematopoietic differentiation induced from H1 hESCs, BF170 promotes the generation of lateral plate mesoderm (LPM) and endothelial progenitor cells (EPCs), suggesting that BF170 is also a pro-hematopoiesis molecule in ex vivo cells, even though the ex vivo BF170 treatments need to be optimized further, especially when treating ex vivo mammalian cells for the HSPC induction analysis. Next, this study further demonstrates that BF170 positively affects both primitive and definitive hematopoiesis (HSPC induction) in vivo in zebrafish embryos. Mechanistically, BF170 induces HSPC emergence via the blood flow-cilia-Ca²⁺ signaling-Notch/NO signaling axis.

In this study, we find that BF170 effectively induces primitive hematopoiesis marked by an increase in primitive blood marker genes, such as lmo2 and scl, in both ex vivo cultured zebrafish blastomere cells and in vivo zebrafish embryos. BF170 significantly increases flk1 and fli1a expression, and the number of endothelial cells (fli1a⁺ cells). Furthermore, BF170 causes a significant increase in cmyb⁺flk1⁺ and runx1⁺flk1⁺ cells in the ventral wall of DA, and in the expression of cmyb and runx1 in the AGM regions at 28 hpf. These observations demonstrate that BF170 acts positively not only in primitive hematopoiesis and vascular cell specification, but also in the induction of HE cells and early HSPCs. The findings in this study are different from those for some chemical compounds that induce HSPCs with almost no effects on hematopoiesis and vascular cell specification in zebrafish embryos (North et al., 2009), suggesting a new model for studying HSPC induction. The hematopoietic ontogeny from embryonic mesoderm to HE and then to HSPCs is conserved from zebrafish to mammals (Orkin and Zon, 2008), and the observations in this study, that BF170 acts in the induction of HE cells, will help in the generation of hematopoietic progenitor-like cells ex vivo with multilineage potential (Blaser and Zon, 2018).

We observed an increased cmyb and runx1 expression by BF170 treatment of zebrafish embryos, which is partially attributed to the increase of cmyb and runx1 transcripts in the specific cells (flk1⁺cmyb⁺ or flk1⁺runx1⁺ cells) and to the cells specifically expressing the genes. The increase of cmyb and runx1 transcripts in the flk1⁺cmyb⁺ or flk1⁺runx1⁺ cells partially explains why the increased proliferation of HSPCs and endothelial cells (ECs) occurs in BF170-treated embryos, as runx1 and cmyb have been shown to be required for the emergence and proliferation of HSPCs and ECs (Chen et al., 2009; Kalev-Zylinska et al., 2002; Zhang et al., 2011). Additionally, we reveal that the positive effects of BF170 on proliferation are not restricted to runx1⁺ or cmyb⁺ HSPCs or fli1a⁺ cells (endothelial cells), since the proliferation of fli1a⁻ cells is also significantly increased by BF170 treatment, suggesting that BF170 functions effectively in the proliferation of other cell types in addition to naïve HSPCs (cmyb⁺), the progenitor-like hemogenic endothelial cells (runx1⁺) and endothelial cells (fli1a⁺).

BF170 was initially used as a probe for in vivo imaging of tau pathology in the AD (Alzheimer's disease) brain (Okamura et al., 2005), but little is known about its effects on hematopoiesis. In this study, we observe, for the first time, the positive impact of BF170 on both primitive and definitive waves of hematopoiesis. BF170 has been reported to increase levels of retinol and NFAT promoter-derived luciferase in cell lines (Akiba et al., 2016; Chen et al., 2016), and retinol signaling is linked with blood flow (Al Haj Zen et al., 2016; Papiernik et al., 2020) and Ca²⁺ signaling (Garattini et al., 2004; Karlsson et al., 2010). These reports are consistent with our findings, to a certain extent, that BF170, as a non-NO donor, is a previously unreported positive regulator of definitive hematopoiesis, in addition to its role in enhancing blood flow.

Primary cilia have specialized functions in shear-stress sensation, chemosensation, etc., and in the maintenance of stem cells in a variety of tissues (Mohieldin et al., 2016; Tong et al., 2014; Tummala et al., 2010; Winkelbauer et al., 2005), which function in the endothelium to induce Notch signaling for HSPC induction (Liu et al., 2019). Meanwhile, studies report that NO is a well-established direct regulator of vascular tone and reactivity that influences blood flow and acts downstream of Notch signaling in HSPC induction (North et al., 2009), and klf2a in endothelium senses blood flow to activate NO signaling, thereby inducing HSPCs (North et al., 2009; Wang et al., 2011). Here, we observe increased expressions of cilia genes in both BF170-treated embryos and sorted cmyb⁺flk1⁺ cells, with the increased blood flow, which is in line with previous studies demonstrating that blood flow mediates primary cilia to sense and transmit signals in zebrafish (Goetz et al., 2014; Li et al., 2020). Meanwhile, elevated Notch signaling and increased levels of Notch1 and NICD proteins are revealed in BF170-treated embryos during the HSPC emergence process, e.g. the increased expression of Notch-related genes in cmyb⁺flk1⁺ HSPCs. Furthermore, increased NO signaling and increased klf2a expression are also observed in BF170-treated embryos, suggesting that primary cilia may sense BF170 stimulation for enhancing the emergence of the HE and HSPCs via stimulation of Notch and NO signaling.

In this study, we find that blocking blood flow by sih knockdown, by using the Notch signaling inhibitor DAPT, by knockdown or knockout of Notch signaling, or by using the NO inhibitor L-NAME all attenuate the beneficial effects of BF170 on HSPC induction. This further suggests that BF170 may induce blood flow and Notch signaling for the initiation of HE specification and HSPC programming, and that the expanded dorsal aorta, and consequently active vascular tone, stimulate NO signaling for the maintenance of HSPC programming. In contrast to NO donors, which expand HSPCs after the heartbeat initiation and have almost no effect on the vascular specification in embryos (North et al., 2009), we find in this study that BF170 is a potential inducer of primitive hematopoiesis and hematovasculature, suggesting that BF170 induces HSPCs via a more-complex program, with NO signaling as one of the downstream mediators.

Ca²⁺ signaling plays a key role in hematopoietic development (Paredes-Gamero et al., 2012; Sipka et al., 2021). In primary cilia mechanosensors, chemosensors and electrosensors often cooperate with Ca²⁺ to regulate a series of cellular processes (Djenoune et al., 2023; Norris and Jackson, 2016; Saternos et al., 2020). In this study, we observe that BF170 treatment significantly improves intracellular Ca²⁺ levels around primary cilia in HE in the AGM region, and functional loss of Pkd2 suppresses the induction effects of BF170 on HSPCs, suggesting that Ca²⁺ released by primary cilia mediates the positive effects of BF170 on HSPC induction. We also observe that blockage of primary cilia-released Ca²⁺ signaling impairs the positive effects of BF170 on Notch signaling, suggesting that BF170 promotes HSPC development through primary cilia-Ca²⁺-Notch signaling pathways. When endothelial cilia are subjected to fluid shear stress generated by blood flow, eNOS (endothelial nitric oxide synthase) is activated by intracellular Ca²⁺ to produce NO to further expand endothelial cells (Boo and Jo, 2003; Nauli et al., 2008). Consistently, we find that BF170 increases Ca²⁺ and NO levels, and that pkd2 deficiency blocks the positive effects of BF170 on NO signaling. We therefore hypothesized that the induction of BF170 on HSPCs may also be regulated via primary cilia-Ca²⁺-NO signaling, cilia-Ca²⁺-Notch signaling or primary cilia-Ca²⁺-Notch-NO signaling.

We notice that inhibition of blood flow, ciliary function, calcium signaling, NO signaling or Notch signaling alone only partially blocks the beneficial effects of BF170 on HSPCs, and simultaneous inhibition of Notch and NO signaling cannot completely inhibit the induction effects of BF170 on HSPCs, implying that BF170 may activate additional pathways, such as MAPK, FoxO, PI3K-Akt, mTOR and TNF signaling pathways. These pathways are known to play important roles in regulating HSPC emergence (Espín-Palazón et al., 2014), maintenance (Huang et al., 2012b), lifespan (Ito et al., 2006), expansion (Perry et al., 2011) and long-term regeneration (Tothova et al., 2007), and are also significantly enriched in BF170-treated flk1⁺ cells or fli1a⁺ cells at 28 hpf by KEGG pathway analysis.

In addition to elaborating the mechanisms underlying the effects of BF170 on HSPC development, we also observe that BF170 induces significant enrichment of several signaling pathways involved in primitive hematopoiesis, such as BMP, WNT, TGF-β and VEGF (Flamme et al., 1995; Gupta et al., 2006; Kanatsu and Nishikawa, 1996; Lengerke et al., 2008; Park et al., 2004; Trompouki et al., 2011) in fli1a⁺ and flk1⁺ cells at 20 hpf, and increases the expression of bmp2b and bmp4 at 14 hpf, which might contribute to the enhanced primitive hematopoiesis and vascular development in BF170-treated zebrafish embryos. BMPs have been reported to coordinate with Wnt to regulate hematopoiesis (Lengerke et al., 2008; Trompouki et al., 2011). However, these need further verification in future studies.

In the ex vivo systems without blood flow, we reveal that BF170 induces significantly increased expression of hematopoietic genes and promotes the generation of LPM and EPCs in early hematopoietic differentiation in hESCs, suggesting that BF170 may enhance the development of mesoderm and endothelial progenitor cells via the aforementioned signaling pathways both in vivo and ex vivo. Meanwhile, the ex vivo system for BF170 in subsequent HSPC induction needs to be optimized in the future, and the underlying mechanisms other than blood flow will be studied in the ex vivo mammalian cell culture systems.

Currently, there are still a few unresolved issues in this study. For example, what are the primary cells responding to BF170 to increase the blood flow? Although we have found effects of BF170 on endothelial cells, the direct target cells of BF170 remain uncertain. Furthermore, although we observe increased heart rates and enlarged vessel diameters in BF170-treated embryos, how BF170 increases blood flow and subsequently promotes primary cilia function remain to be further elucidated. Answering these questions will provide further insights into the molecular mechanisms of BF170 in HSPC induction and will help develop therapies for hematological disorders.

In summary, this study identifies BF170 as a new inducer of HSPCs by combining an ex vivo chemical screening system and the in vivo zebrafish embryo system, and that BF170 promotes hematopoietic development, including primitive and definitive hematopoiesis, in zebrafish. Mechanistically, in BF170-treated embryos, endothelial cilia sense the change in fluid shear stress caused by BF170-induced increased blood flow and then induce downstream Ca²⁺ and subsequent Notch signaling and NO production, thereby facilitating HSPC emergence and the maintaining HSPC programming stabilization (Fig. 7H).

The abbreviations of genes tested in this study are listed in Table S1. All animals and experiments were conducted in accordance with the Guidelines for Experimental Animals approved by the Institutional Animal Care and Use Ethics Committee of Huazhong Agricultural University (permit number HZAUFI2016-007).

Wild-type zebrafish (AB line) and transgenic zebrafish Tg (runx1: GFP) (Zhang et al., 2015a), Tg (cmyb: GFP) (North et al., 2007), Tg (mpx: GFP), Tg (rag2: DsRed) (Ma et al., 2012), Tg (flk1: mCherry/runx1: GFP) (Zhang et al., 2015a), Tg (flk1: mCherry/cmyb: GFP), Tg (EPV. Tp1-Mmu. Hbb: EGFP) [CZ355, purchased from CZRC (China Zebrafish Resource Center)] (Parsons et al., 2009), Tg (actb2:Mmu.Arl13b-GFP) (CZ1619, purchased from CZRC) (Borovina et al., 2010), Tg (fli1a: nls-GFP) (a generous gift from Prof. Weijun Pan, Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, Shanghai, China) and Tg (Cca.actb: GCaMP6 s) (a generous gift from Associate Prof. Yong Long, Institute of Hydrobiology, Chinese Academy of Sciences, Wuhan, China) (Li et al., 2021), as well as mutants of notch3^+/− (Alunni et al., 2013; Quillien et al., 2014) and jag1a^−/− (CZ1048, purchased from CZRC), were cultured and bred according to the technical service data provided by the CZRC (http://www.zfish.cn/inforscan/1365.html). Adult zebrafish were raised in recirculating aquaculture systems (28±0.5°C, 14:10 h light:dark) and were fed with Artemia salina hatched from commercial Artemia salina eggs (Tianjin Fengnian Aquaculture, China) three times per day. The ages of the embryos and larvae are expressed in hours post-fertilization (hpf) or days post-fertilization (dpf). Embryos or larvae aged 4 hpf to 96 hpf were used in this study. The classification of these specimens as males or females was not feasible due to sex being influenced by various internal and external factors, with definitive determination typically occurring near adulthood (Liew and Orbán, 2014).

The H1 hESC line (WiCell Research Institute, Madison, WI, USA) was maintained in mTeSR1 medium (StemCell Technologies) supplemented with 1% penicillin-streptomycin. Regular microscopic inspections were conducted to monitor and confirm the absence of contamination within the cell line.

This experiment followed the procedures of Ciarlo et al. (2017). Disassociated blastomere cells were grown in zebrafish embryonic stem cells (zESC) medium, composed of 70% LDF medium and 30% RTS34st-conditioned medium. LDF medium contained 50% Leibowitz's L-15, 35% DMEM and 15% Ham's F-12, supplemented with 15 mM HEPES (15630, Invitrogen), 1% L-glutamine (25030, Invitrogen), 10 μg/ml ciprofloxacin (17850, Sigma), 100 μg/ml piperacillin (P8396, Sigma), 10 μg/ml amphotericin B (A9528, Sigma), 10 nM sodium selenite (S9133, Sigma), 1% N2 (17502, Invitrogen) and 2% B27 (17504, Invitrogen). Fusion cultures of RTS34st cells (a stromal cell line from rainbow trout spleen) were incubated in RTS34st conditional medium for 3 days (Leibowitz's L-15 plus 15% FBS). Tg (cmyb: GFP) and Tg (mpx: GFP) de-chorionated embryos were collected at 4 hpf and gently shaken about 20 times in a 50 ml centrifuge tube with 5 ml zESC medium (the proportion of 600 embryos and 25 ml medium) to obtain disassociated blastomere cells. The disassociated cells were diluted to 10 ml of zESC medium, filtered with 40 mm nylon mesh filter and aliquoted 30 μl per well for 384-well uncoated plates. After incubating at 28.5°C for 10-16 h, compounds were added to each well.

Bioactive compounds from the following libraries were used for Tg (cmyb: GFP) and Tg (mpx: GFP) primary cultured blastomere cells: Sigma LOPAC (1280), BIOMOL ICCB Known Bioactives (480), BIOMOL Fatty Acids (68), BIOMOL Protease Inhibitor (53), BIOMOL Phosphatase Inhibitor (33) and BIOMOL Orphan Ligands (84) (Xu et al., 2013). After the culturing and treatment described above, primary blastomere cells were imaged on the third day using a Celigo cytometer (Cyntellect) under channels of GFP and white light (Ciarlo et al., 2017; Xu et al., 2013). The compounds that increased the GFP fluorescence were selected as candidate compounds for the next ex vivo and in vivo tests.

Experiments on mouse embryonic stem (ES) cell culture and formation were performed with reference to previous guidelines (Guan et al., 2012). ES cells cultured ex vivo were collected, and were suspended in KO-DMEM (Invitrogen), 20% fetal bovine serum (Stem Cell Technologies), 100 μM non-essential amino acids (Invitrogen), 200 μM L-glutamine (Invitrogen), 50 μg/ml ascorbic acid (Sigma), 200 μg/ml H-transferrin (Sigma) and 100 μM dimercaptoethanol (Sigma). The suspended three-dimensional cell aggregates, which contained embryoid bodies (EBs), were formed by shaking the ES cells in 10 cm ultra-low-attachment dishes for 1 day. The next day, ∼30 h or more later, EBs were seeded into 48-well plates, and the candidate compounds, such as BF170, were added to each well at an optimal dose, with DMSO as a control. The expression of different blood genes was measured by one-step cell and direct quantitative RT-PCR on day 4.

Hematopoietic differentiation of H1 hESCs was performed using a chemically defined hematopoietic differentiation system as described previously (Pang et al., 2013; Wang et al., 2018). First, H1 hESCs were cultured in mTeSR1 medium (StemCell Technologies) supplemented with 1% penicillin-streptomycin. At 70-80% confluency, single cells were dissociated using Accutase (Invitrogen), seeded on growth factor-reduced (GFR) Matrigel-coated plates and cultivated in mTeSR1 with 10 μM Y27632 (Selleck). After 24 h (day 0), the cells were cultured in mTeSR1 medium without select factors (05892, StemCell Technologies). On day 2, the medium was replaced with mTeSR1 medium without selected factors. On day 5, cells were cultured in mTeSR1 medium without selected factors with 50 ng/ml FGF2, 40 ng/ml VEGF and 20 μM SB 431542 (Stemgent) added for 3 days. From day 8, to induce the generation of CD45⁺ cells, differentiated cells were cultured in fresh mTeSR1 medium without selected factors [with 20 ng/ml SCF (Peprotech), 50 ng/ml TPO (Peprotech), 20 ng/ml IL3 (Peprotech), 1 mM glutamax (Gibco), 2% B27 (Gibco), 0.1 mM MTG (Sigma-Aldrich), 1% ITS (Gibco), 1% NAA (Gibco) and 1% penicillin/streptomycin], and fresh culture medium was replaced at 2-day intervals. The candidate compounds, including BF170, quazinone, flupirtine maleate, were added from day 0 with DMSO as control, and this continued throughout the experimental stages.

In this study, BF170 powder (C₁₅H₁₂N₂·HCl, B4311, Sigma), DAPT powder (D5942, Sigma), and CBD (Ciliobrevin D, HY-122632, MedChemExpress, China), quazinone powder (BML-PD170-0010, Enzo Life Sciences), flupirtine maleate powder (F8927, Sigma), were dissolved in dimethyl sulphoxide (DMSO) (D2650, Biosharp, China) to prepare 10 mM BF170, 10 mM quazinone, 10 mM flupirtine maleate, 100 mM DAPT and 10 mM CBD stock solutions, respectively, which were stored at −80°C away from light. L-NAME (N5751, Sigma) powder was made into 10 mM stock solution with water, and was diluted to 10 μM for treatment. In ex vivo hESCs, 10 μM BF170, 10 μM quazinone and 10 μM flupirtine maleate were used. In embryos, 100 μM DAPT was added to embryos at 6 hpf, and 10 μM BF170, 10 μM L-NAME or 10 μM CBD was added at 11 hpf. For BF170 treatment, DMSO was used as control. For DAPT treatment, embryos were divided into a DAPT group, a BF170 group, a BF170 plus DAPT treatment group and a DMSO group. For L-NAME treatment, embryos were exposed to L-NAME, BF170, L-NAME plus BF170 or DMSO. For CBD treatment, embryos were treated with CBD, BF170, CBD plus BF170 or DMSO. Additionally, BF170 at 15 μM and at 10 μM were used to treat embryos after the presence of a heartbeat was detected, and the concentration of 15 μM for BF170 is indicated in figure panels whereas BF170 with no concentration indication means BF170 at 10 μM in this study. Three replicates were set in each group, and the treated embryos were cultured at 28.5°C. The embryos were collected during the required developmental period according to different experimental needs.

Zebrafish primary blastomere cells and mouse EBs, which were treated with the candidate compounds, with their respective controls, were collected at the indicated stage and used for one-step cell direct qRT-PCR. Meanwhile, Tg (flk1: mCherry/cmyb: GFP) and Tg (flk1: mCherry/runx1: GFP) blastomere cells were cultured with GFP fluorescence, collected and washed, and then were homogenized into single cells in PBS containing 5% FBS. After filtration, the desired positive cells were sorted by flow cytometry (FACS) (BD FacsAria SORP, 650110M3, BioDot). One-step cell direct qRT-PCR was performed as reported recently (Chen et al., 2019; Li et al., 2023). Different groups of cells were sorted into the lysate of the CellsDirect One-Step qRT-PCR Kit (11753-100, Invitrogen). The lysed solution was then used as template for one-step cell direct qRT-PCR. Primer sequences for the genes tested in this study are shown in Table S2, including mouse genes Myb, Runx1, Mpo, Gata1, Lmo2 and Scl, and zebrafish genes mpx, βe3, shha, ifabp (fabp2), mylpfa, cmyb, runx1, fli1a, flk1, flt4, efnb2a (ephrin B2a), dll4, notch1a, notch3, hey1, hey2, her1, her2, her5, nos1, nos2b, pkd2, kif3a, ift88 and fsd1. The expression of GAPDH, 18s or β-actin genes was used as a reference control, and the expression levels of target genes relative to reference genes were calculated using 2^−ΔΔCt.

Whole-mount in situ hybridization detection was performed as described in our previous studies (Jin et al., 2021; Li et al., 2022, 2023; Liu et al., 2017). The gene probes for cmyb, runx1, fli1a, flk1, rag1, flt4, mpx, lcp1 (l-plastin), gata1, hbbe1, hbbe2, hbbe3, scl (tal1), lmo2, bmp2b, etc. have been reported previously (Galloway et al., 2005; Li et al., 2022; Zhang et al., 2018; Zhou et al., 2016). Other probes (bmp4, efnb2a , dll4 and klf2a) were amplified from cDNA using the primers shown in Table S3. Digoxigenin-labeled antisense RNA probes were used to detect the gene expression in the whole-mount embryos at designated periods. The specific positive signals of probes were purple, and the relative strength and distribution of the signal indicate the transcription level of the marker gene. Samples were observed and photographed under a stereoscopic microscope (M205FA, Leica). The signal area and strength of each image was calculated using ImageJ. The percentage of embryos with altered gene expression was counted, as reported previously (Zhang et al., 2015b). Meanwhile, embryos from incrossing notch3^+/− were individually genotyped after whole-mount in situ hybridization.

Flow cytometry analysis was performed as reported previously (Wang et al., 2020). Briefly, the differentiated cells at different periods were dissociated with Accutase (Invitrogen), and washed with pre-cooled PBS. After 300 g centrifugation, the precipitated cells were resuspended in PBS and subsequently filtered through a 70 μm cell strainer to obtain a single-cell suspension. Next, the cells were incubated with fluorochrome-conjugated antibodies against surface markers for 30 min at 4°C, and stained with DAPI for 5 min before testing to exclude dead cells, followed by flow cytometry analysis using a FACS CantoII flow cytometer (BD Biosciences). The primary antibodies in this study were as follows: anti-APLNR (FAB856A, R&D Systems; 1:100), anti-CD31 (555446, BD Biosciences; 1:100), anti-CD34 (555824, BD Biosciences; 1:100), anti-CD43 (560198, BD Biosciences; 1:100) and anti-CD45 (555485, BD Biosciences; 1:100). The control IgG antibodies included PE cyanine 7 mouse IgG1 (25-4714-42, eBioscience; 1:100) and APC mouse IgG3 (IC007A, R&D Systems; 1:100).

BrdU labeling and immunofluorescence were performed based on our previously reported methods (Li et al., 2023; Tai et al., 2022). BrdU (10 mM; ST1056, Beyotime, China) was injected into BF170 and DMSO-treated Tg (cmyb: GFP), Tg (runx1: GFP) and Tg (fli1a: nls-GFP) embryos at 26 hpf, respectively, and incubated for 2 h. The 28 hpf embryos were fixed in 4% paraformaldehyde (PFA) and preserved overnight at 4°C. Next, the fixed embryos were dehydrated, and then rehydrated with methanol and 1×PBS, followed by hydrochloric acid (HCl) incubation at room temperature. Finally, embryos were stained using mouse anti-BrdU mAb (1:200, A1482, ABclonal, China) and rabbit anti-GFP-Tag pAb (1:200, AE011, ABclonal, China), and then double-stained with Alexa Fluor 555-conjugated goat anti-mouse IgG (H+L) (1:200, AS057, ABclonal, China) and goat anti-rabbit IgG FITC (H+L) (1:200, BL033A, Biosharp, China). Immunofluorescence for anti-Runx1 antibody (1:200, ab92336, Abcam) was performed on control and BF170-treated embryos, as reported recently (Li et al., 2023; Tai et al., 2022). The images of immunofluorescence were obtained with a confocal microscope (Leica M205FA).

Western blotting was performed as previously described (Liu et al., 2017). The protein samples were extracted from BF170- and DMSO-treated embryos at 16 hpf, 20 hpf, 24 hpf and 28 hpf. The following primary antibodies were used: anti-Scl (Tal1) (55317-1-AP, Proteintech; 1:1000), anti-Lmo2 (21966-1-AP, Proteintech; 1:1000), anti-GATA1 (10917-2-AP, Proteintech; 1:1000), anti-Notch1 (20687-1-AP, Proteintech; 1:1000), anti-NICD (AF5307, Affinity; 1:1000), anti-H3 (A2348, ABclonal; 1:2000), anti-β-Actin (AC026, ABclonal; 1:10,000) and anti-PCNA (A0264, ABclonal; 1:1000). HRP-labeled goat anti-rabbit IgG (H+L) (A0208, Beyotime; 1:5000) and HRP-labeled goat anti-rat IgG (H+L) (A0192, Beyotime; 1:5000) were used as secondary antibodies. The Fuji Film LAS4000 mini luminescent image analyzer was used for photographing the blots. ImageJ was used for quantifying the protein levels based on the band density obtained in the western blot analysis.

In this study, fsd1 morpholino (MO) (Liu et al., 2019) and sih (tnnt2a) MO (Wang et al., 2011) were kindly provided by Professor Liu Feng (Institute of Zoology, Chinese Academy of Sciences, Beijing, China), and the MOs of notch1b (Kim et al., 2014), notch1a, notch3 (Ma and Jiang, 2007) and pkd2 (Djenoune et al., 2023) were purchased from Gene Tools. The MO sequences are provided in Table S4. These MOs were injected into one-cell stage embryos at a final concentration of 1 mM, and the injection volume was controlled at about 1 nl/embryos. Embryo phenotype was observed after microinjection, and embryos of a specific age were collected for different tests.

Sample collection, the sequencing process and the analysis method followed the previously reported procedure (Chen et al., 2019). After DMSO and BF170 treatment, 20 hpf and 28 hpf Tg (flk1: mCherry) and Tg (fli1a: GFP) embryos were collected, and GFP- and mCherry-positive cells were sorted out into SMART-Seq V4 kit lysis buffer containing RNase inhibitor. The samples were then used for RNA-sequencing. Briefly, all samples were amplified to obtain cDNA product (1∼2 kb) by using the Smart Seq2 method, and the concentrations of amplified products were measured using a Qubit 3.0 Flurometer (Life Technologies). For library construction, cDNA samples were first fragmented to ∼350 bp by Bioruptor Sonication System (Diagenode). The fragments were subjected to end repair, addition of ‘A’ base and adaptor ligation, and then purified using Beckman AMPure XP beads. After quality inspection using an Agilent 2100/LabChip GX Touch, the libraries were run on Illumina HiSeq platform by double-end sequencing program (PE150). After quality control for sequencing data, bioinformatics analysis was performed based on GRCz10.88 zebrafish reference genome. FPKM (fragments per kilobase per million mapped fragments) was used to quantify the value of gene expression. Differential expression analysis was conducted using DEGSeq (Wang et al., 2010), and the genes with |log₂Ratio|>1 and P<0.05 were defined as differentially expressed genes (DEGs). Next, all DEGs were used for GO (Gene Ontology) and KEGG (Kyoto Encyclopedia of Genes and Genomes) functional analysis. The DEG, GO and KEGG pathway analysis data have been deposited in the National Genomics Data Center (NGDC) OMIX Database under accession number OMIX006675 and in the Science Data Bank (ScienceDB) (https://doi.org/10.57760/sciencedb.agriculture.00027).

In this study, fluorescent reporter embryos from Tg (flk1: mCherry/runx1: GFP), Tg (flk1: mCherry/cmyb: GFP), Tg (flk1: mCherry/fli1a: nls-GFP), Tg (rag2: DsRed), Tg (actb2: Arl13b-GFP/flk1: mCherry), Tg (EPV. Tp1-Mmu. Hbb: EGFP) and Tg (flk1: mCherry/Cca.actb: GCaMP6s) were treated according to experimental requirements. Live embryos were stripped of egg membranes and anesthetized before being photographed. Images of live embryo, video and immunofluorescence were obtained using confocal microscopy (Leica M205FA). The fluorescence intensity and the number of fluorescent cells were quantified by ImageJ, and the data were analyzed using GraphPad Prism 8.0.

The embryos treated with DMSO or BF170 at 28 hpf were microinjected with 2 µM Calcium Orange AM (Invitrogen, Molecular Probes, C3015). After incubation for 1 h at 28.5°C (protected from light), embryos were anesthetized and mounted in 1% low melt agarose. The Calcium Orange signals were imaged by super resolution microscopy (STORM-A1R) and analyzed using a NIS-Elements Viewer 4.50.

The sample size used for different experiments in each group was larger than 10 embryos (n>10), with at least three biological replicates for each test. The number of embryos used for FACS was 100 to 150, and three biological replicates were performed for each group. NIH ImageJ was used to quantify the whole-mount in situ hybridization signal and the number of labeled cells, as well as the fluorescence intensity and the number of fluorescent cells in zebrafish embryos, and the data were analyzed and visualized using GraphPad Prism 8.0. Pixel area was calculated using NIH ImageJ. First, the whole-mount in situ hybridization images were converted to type 8-bit, and pixel size was then determined by a known distance and unit of length (scale bar on the image) using the ‘Analyze-Set Scale’ option in the toolbar. Next, the area of positive signals was adjusted by Image Adjust Brightness/Contrast, and the final value was measured. Blood flow velocity for every sample was evaluated using the formula: blood flow velocity (μm/s)=distance (μm) traveled by every GFP-labeled blood cell per second (s), and five GFP-labeled cells per sample were assessed. Ca²⁺ fluorescence intensities at different time points were measured using NIS-Elements Viewer 4.50. The data measured at every time point for blood flow velocity and Ca²⁺ fluorescence was input into GraphPad Prism 8.0 for visualization and statistical analysis. The results of quantitative real-time PCR and other experiments were tested using a paired t-test and a post-hoc Turkey's test in SPSS 26.0. The statistical significance between groups was determined as *P<0.05, **P<0.01 or ***P<0.001.

We are grateful to Prof. Feng Liu (Institute of Zoology, Chinese Academy of Science) for providing morpholinos, we thank Prof. Chengtian Zhao (Ocean University of China, College of Marine Life Sciences) and Prof. Lihong Shi (State Key Laboratory of Experimental Hematology) for providing project conception, and we thank Associate Prof. Yong Long (Institute of Hydrobiology, Chinese Academy of Science) and Prof. Weijun Pan (Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences) for providing transgenic lines, Tg (Cca.actb: GCaMP6s) and Tg (fli1a: nls-GFP), respectively. We also thank China Zebrafish Resource Center (CZRC) for providing Tg (EPV. Tp1-Mmu. Hbb: EGFP) and Tg (actb2: Mmu.Arl13b-GFP) transgenic lines, and jag1a^+/− and notch3^+/− mutants. We thank the core facilities of the State Key Laboratory of Agricultural Microbiology of Huazhong Agricultural University (Wuhan, China) for technical support.