The sinusoidal hematopoietic niche is formed by Jam1a via Notch signaling in the zebrafish kidney

Summary The zebrafish is a unique model to understand hematopoietic niches as hematopoietic stem/progenitor cells are maintained in the kidney. However, little is known about which cell types in the kidney play a role in hematopoietic niches. Here, we demonstrate that the sinusoidal endothelium is an essential and conserved niche component in the zebrafish kidney. Histological analysis revealed that runx1:mCherry+ hematopoietic cells were predominantly detected in the dorsolateral region of the kidney where sinusoids are highly developed. Loss of Junctional adhesion molecule 1a (Jam1a), which is expressed in both sinusoidal endothelial cells and hematopoietic cells, resulted in a remarkable reduction in sinusoids and a defect in hematopoietic niches. We found that Jam1a regulates jagged-1a expression in vascular endothelial cells to form a sinusoidal structure in the kidney. Collectively, these data suggest that sinusoids are formed by Jam1a via endothelial Notch signaling to provide hematopoietic niches in the zebrafish kidney.


INTRODUCTION
Hematopoietic stem cells (HSCs) are rare cells with the remarkable ability to both self-renew and generate all mature blood cell types over the life span of an individual. HSCs present within the adult bone marrow or newborn cord blood are by far the most widely utilized human stem cells in the clinic. To date, however, ex vivo expansion and maintenance of HSCs has not been possible for regenerative medicine approaches. A major obstacle has been the lack of understanding of the key cell types and signaling pathways that comprise the HSC niche, which is defined as the cellular and molecular components that regulate HSC quiescence, self-renewal, and differentiation. [1][2][3] Studies in the murine bone marrow have identified some essential cellular components of the HSC niche, such as osteoblasts, 4-6 sinusoidal endothelial cells, 7-10 perivascular cells, 9,[11][12][13] mesenchymal stem cells, 14 macrophages, 15 and megakaryocytes. [16][17][18] These niche cells highly express supporting factors for HSCs, also termed ''niche factors'', including angiopoietin 1, 6 stem cell factor (Scf), 9,19 C-X-C motif ligand 12 (Cxcl12; also known as stromal cell-derived factor 1 [SDF1]), 11,20 Jagged-1, 10 and thrombopoietin. 21,22 Nevertheless, there still remains controversy about the precise HSC niche, and a better understanding of the elements that regulate HSCs is required.
The zebrafish is an excellent model for a parallel view of the HSC niche. Many valuable tools and experimental methods have been established for the study of HSCs in zebrafish (e.g., transgenic/mutant animals, genome editing, transplantation assays, cell culture assays, etc.). Despite the high conservation of hematopoiesis at the cellular and molecular levels, the nature of the niche appears at face value to be different as the adult hematopoietic organ in teleost fish is the kidney. Various stages of hematopoietic cells and mature blood cells are observed in interstitial tissue of the kidney, termed the ''kidney marrow''. [23][24][25] Although, the zebrafish kidney is thus of great importance for understanding hematopoietic niches, little is known regarding which cell types play a role in hematopoietic niches in the kidney. This is due, at least in part, to the lack of specific HSC markers in zebrafish. Recently, we established a method to purify zebrafish HSCs from the adult kidney using a double-transgenic line, gata2a:GFP; runx1:mCherry. Based on these transgenes, three distinct hematopoietic cell populations can be resolved in kidney marrow cells (KMCs), gata2a:GFP + runx1:mCherry + (hereafter denoted as gata2a + runx1 + ), gata2a À runx1 + , and gata2a + runx1 À . Transcriptome analysis revealed that gata2a + runx1 + cells displayed typical molecular hallmarks of HSCs, whereas gata2a À runx1 + cells and gata2a + runx1 À cells showed expression signatures of erythroid/ myeloid cells and lymphoid cells, respectively. Competitive transplantation assays demonstrated that 100 gata2a + runx1 + cells fully reconstituted hematopoiesis over 16 weeks in irradiated zebrafish, indicating that zebrafish HSCs are highly enriched in the gata2a + runx1 + fraction in the kidney. 26,27 Here, we demonstrate that the sinusoidal endothelium is essential for hematopoietic niches in the zebrafish kidney. Histological analysis revealed that runx1:mCherry + hematopoietic cells were predominantly observed in the dorsolateral (DL) region of the kidney, where sinusoidal capillaries were abundant. Loss of junctional adhesion molecule 1a (Jam1a, also known as F11r), which is expressed by both sinusoidal endothelial cells and hematopoietic cells, resulted in a remarkable reduction in sinusoids and a defect in hematopoietic niches. We uncovered that Jam1a regulates jagged-1a expression in vascular endothelial cells to form a sinusoidal structure that provides hematopoietic niches in the kidney. Our data provide evidence that the sinusoidal endothelium is an evolutionarily conserved component of hematopoietic niches in vertebrates.

RESULTS
Hematopoietic cells are localized along the sinusoidal endothelium in the dorsolateral region of the kidney We first examined the structure of the zebrafish kidney by histological analysis. Zebrafish kidneys flank the dorsal aorta and cardinal vein and contain multiple types of renal epithelia including glomeruli, proximal and distal tubules, and collecting ducts. Although, there was no clear separation in renal structures as seen in mammalian kidneys (e.g., cortex and medulla), glomeruli tended to be distributed along the dorsal to lateral surface. We found that KMCs were predominantly observed in the DL region of the kidney ( Figures 1A-1C). It has recently been reported that melanocytes dorsally cover hematopoietic organs in aquatic animals, termed the ''melanocyte umbrella'', which is suggested to protect hematopoietic stem/ progenitor cells (HSPCs) from UV light. 28 Consistent with this observation, zebrafish kidneys were dorsally covered by melanocytes ( Figure 1B), suggesting that the kidney marrow is formed mainly under the melanocyte umbrella. To investigate the distribution of HSPCs, we performed whole-mount immunohistochemistry in kidneys dissected from gata2a:GFP; runx1:mCherry animals. Expression of runx1:mCherry was restricted in hematopoietic cells, whereas gata2a:GFP was detected not only in hematopoietic cells but also in vascular endothelial cells. Although the majority of runx1:mCherry + hematopoietic cells did not express gata2a:GFP, a few gata2a + runx1 + cells were sporadically detected in the DL region of the kidney ( Figure 1D). We found that gata2a + runx1 + cells were closely associated with gata2a:GFP + vascular endothelial cells. A transverse section of the kidney also showed that gata2a + runx1 + cells were in contact with a gata2a:GFP + vascular endothelial cell that typically shows spindle to long thin shapes with weak GFP expression and is morphologically distinguishable from gata2a:GFP + hematopoietic cells ( Figure 1E). By measuring the distance of individual gata2a + runx1 + cells or randomly selected cells to the gata2a:GFP + endothelium, we found that approximately 60% of gata2a + runx1 + cells were within 2 mm of the endothelium, which was in significantly higher frequency than randomly selected cells ( Figure 1F). While not significant, gata2a + runx1 + cells also tended to be detected at far distances from the dorsal aorta ( Figure 1G). However, there was no difference in the distance to renal tubules ( Figure 1H). Taken together, these data suggest that HSPCs closely interact with vascular endothelial cells in the kidney.
To further investigate vasculature in the kidney, we utilized kdrl:Cerulean animals, which express Cerulean under the control of vascular specific kdrl enhancers. 29 Whole-mount and section (B and C) Hematoxylin and eosin staining of zebrafish kidney. Red areas and black arrows denote kidney marrow and melanocytes observed in the dorsal surface of the kidney. A high magnification view of the yellow dotted area is shown in C. (D) Dorsal view of a gata2a:GFP; runx1:mCherry kidney. Inset shows the high magnification view of the white dotted area. Arrows and dotted lines indicate gata2a + runx1 + cells and endothelial cells, respectively. (E) Transverse section of a gata2a:GFP; runx1:mCherry kidney. Insets show the green (left), red (middle), and merged channel (right) of the white boxed area. Arrowheads indicate a gata2a + runx1 + cell, and white dotted lines outline a gata2a:GFP + vascular endothelial cell. The section is oriented dorsal side up. Erythrocytes within blood vessels are observed in white due to auto-fluorescence.  iScience Article immunohistochemistry revealed that there were at least three types of vascular endothelium in the kidney, the endothelium lining along the major blood vessel (e.g., dorsal aorta, cardinal vein, and their branches); the endothelium surrounding the renal tubule (''renal endothelium''); and the sinusoidal endothelium ( Figure S1A). The renal endothelium, which comprises small blood vessels to carry away the recovered solute, was entirely observed in the kidney. In contrast, the sinusoidal endothelium, which forms fenestrated capillaries, was predominantly observed in the DL region (Figures 2A-2D and S1A). Kidney tissue was subdivided into 4 regions, dorsomedial (DM), DL, ventromedial (VM), and ventrolateral (VL), and sinusoidal area was quantified in each region. The percentage of sinusoidal area in the DL region was approximately 54.0%, which was markedly higher than the DM (31.7%), VM (23.0%), or VL region (30.3%) ( Figures 2E and 2F). The distribution of sinusoids appeared to be correlated with the distribution of melanocytes as both were mainly localized in the dorsal side of the kidney (Figures S1B and S1C). To investigate the co-localization of hematopoietic cells and sinusoidal endothelium in the kidney, the kdrl:Cerulean line was combined with the runx1:mCherry line and kidney tissue was histologically analyzed. There were large numbers of runx1 + hematopoietic cells distributed around the kdrl + sinusoids in the kidney ( Figures 2G and 2H). The percentage of runx1 + cells distributed within the DL region was approximately 36.5%, significantly higher than that in the other regions ( Figure 2I). Moreover, the percentage of runx1 + cells distributed within 2 mm of the sinusoidal endothelium was approximately 58.2%, whereas the percentage of those distributed over 10 mm was only 6.1% ( Figure 2J). These data suggest that sinusoids are predominantly formed in the DL region of the kidney, and most hematopoietic cells are distributed adjacent to the sinusoidal endothelium. Both glomeruli and blood vessels were found to be more abundant dorsally in the kidney, especially in the DL region where they tended to be distributed within the region of the sinusoids; however, no localization of runx1 + cells around glomeruli or blood vessels was observed ( Figures S2A-S2D).

Loss of Jam1a results in a reduction of HSPCs in the kidney
In mammals, Jam1 (also known as Jam-A) is expressed in hematopoietic cells, vascular endothelial cells, and renal epithelium, and is involved in cell-cell interactions based on homophilic and/or heterophilic binding with other Jam molecules. [30][31][32] We examined Jam1a expression in the adult kidney under a kdrl:Cerulean background by immunohistochemistry and found that Jam1a was detected in both hematopoietic cells and sinusoidal endothelium in the kidney marrow ( Figures S3A and S3B).
To examine the role of Jam1a in adult hematopoiesis, we generated a genetic mutant line, jam1a sd43 , which is predicted to contain a premature stop codon due to a 10 base-pair deletion in exon 3 ( Figures S3C and S3D). Similar to Jam1 knockout mice, 33,34 homozygous jam1a sd43 zebrafish are partially viable into adulthood and fertile, and there were no obvious morphological abnormalities, whereas ja-m1a sd43 adult males showed slightly lower body weight than age-matched wild-type males ( Figures S3E-S3G). Western blotting analysis revealed that the wild-type form of the Jam1a protein was completely lost in jam1a sd43 animals ( Figures S3H and S3I). Jam1a was strongly expressed in the epithelium of collecting ducts and tubules in the kidney, as has been shown in the human kidney, 31 whereas this expression was never detected in jam1a sd43 animals ( Figures S3J and S3K). Despite the strong expression of Jam1a in the collecting ducts and tubules, there were no obvious morphological abnormalities in the renal tissue, suggesting that Jam1a may be dispensable for renal functions. It has previously been shown in the zebrafish embryo that loss of Jam1a results in a reduction of developing HSPCs in the dorsal aorta due to low levels of Notch signaling. 35 Consistent with this observation, approximately 87% of jam1a sd43 embryos showed reduced to no expression of the HSPC marker gene, runx1, in the dorsal aorta ( Figures S3L-S3N). To test if jam1a sd43 animals display a hematopoietic defect in the adult kidney, the jam1a sd43 line was combined with a double-transgenic line, kdrl:Cre; bactin2:loxP-STOP-loxP-DsRed (hereafter referred to as kdrlsw), which labels nearly all adult blood cells with dsRed except for mature erythrocytes. 26,36 Flow cytometric (FCM) analysis revealed that the absolute numbers of kdrl-swerythrocytes, kdrl-sw + non-erythrocyte blood cells, and granulocytes were approximately 41%, 35%, and 55% lower in jam1a sd43 animals than those in age-matched wild-type animals, respectively, whereas the numbers of lymphoid and precursor cells were unchanged ( Figures S4A-S4D).
The potential for hematopoietic reconstitution can be evaluated through an in vivo competitive repopulation assay, in which the contributions of donor-and competitor-derived cells are compared in irradiated recipients. 26,37 DsRed-labeled wild-type or jam1a sd43 kMCs (donors; kdrl-sw background) were co-transplanted with equal numbers of blue fluorescent protein (BFP)-labeled wild-type KMCs (competitors; bactin2:BFP background) into sub lethally irradiated recipients (wild-type or jam1a sd43 group). At 4 weeks post-transplantation (wpt), KMCs from each recipient group were further transplanted into secondary recipients, followed by analysis of KMCs at 6 wpt ( Figure 3A). In the wild-type group, the percentage of donor-derived dsRed + cells within the total dsRed + and BFP + cells was closed to 50% in both primary and secondary transplantations, confirming the reliability and validity of this transplantation assay. In contrast, contribution was approximately 24.1% in primary recipients and 13.2% in secondary recipients in the jam1a sd43 group ( Figures 3B-3D), indicating that mutation of jam1a results in low levels of hematopoietic reconstitution activity. We also evaluate the homing capacity of jam1a sd43 kMCs. DsRed-labeled wildtype or jam1a sd43 kMCs were co-injected with equal numbers of BFP-labeled wild-type KMCs into wildtype recipients (wild-type or jam1a sd43 group), followed by analysis of KMCs at 2 days post-transplantation iScience Article (dpt). The percentage of donor-derived dsRed + cells within the total dsRed + and BFP + cells in the wild-type group was approximately 42.3%, whereas that in the jam1a sd43 group was 11.7% ( Figures S5A-S5C). These data suggest that mutation of jam1a also affects the homing capacity of KMCs.
To compare the number of HSPCs between wild-type and jam1a sd43 animals, we performed FCM analysis in KMCs under gata2a:GFP; runx1:mCherry background ( Figures S6A and S6B). The absolute numbers of gata2a + runx1 + cells were approximately 38% lower in jam1a sd43 animals than wild-type animals ( Figures 4A-4C), while there was no significant difference in the percentage of gata2a + runx1 + cells in KMCs ( Figure S7A). We also observed that both the percentage and absolute number of gata2a À runx1 + cells (erythroid and myeloid cells) were approximately 30% lower in jam1a sd43 animals than those in wild-type animals, whereas those of gata2a + runx1 À cells (lymphoid cells) were approximately 50% higher ( Figures S7B and S7C).
In order to evaluate the hematopoietic reconstitution activity of gata2a + runx1 + cells in jam1a sd43 animals, wildtype, and jam1a sd43 animals under the background of gata2a:GFP; runx1:mCherry; bactin2:BFP were used for competitive repopulation assay. One hundred gata2a + runx1 + cells from wild-type or jam1a sd43 animals were co-transplanted with DsRed-labeled wild-type KMCs (competitors) into sub lethally irradiated recipients (wildtype or jam1a sd43 group) ( Figure 4D). The percentage of donor-derived BFP + cells in the wild-type group was approximately 30.8% at 16 wpt, whereas that in the jam1a sd43 group was only 0.02% ( Figure 4E). These data suggest that mutation of jam1a affects the long-term hematopoietic reconstitution of HSCs.
The differentiation potential of gata2a + runx1 + cells was evaluated by an in vitro colony assay in the presence of Epoa or Csf3b (Gcsfb). We did not observe significant differences between wild-type and jam1a sd43  Figures S7D and S7E). Because there was a reduction in the number of erythrocytes and granulocytes in the jam1a sd43 kidney despite the unchanged differentiation potential of gata2a + runx1 + cells, the expression levels of epoa and csf3b were compared between the wild-type and jam1a sd43 kidney by quantitative polymerase chain reaction (qPCR). We observed a large reduction in the expression of both genes in the jam1a sd34 kidney ( Figure S7F), suggesting that decreased erythrocytes and neutrophils are attributed to decreased stimulating factors for erythropoiesis and granulopoiesis in the jam1a sd34 kidney.

Loss of Jam1a causes a defect in hematopoietic niches in the kidney
Since Jam1a was expressed not only in hematopoietic cells but also in sinusoidal endothelial cells in the kidney, we next evaluated the capacity of hematopoietic niches in jam1a sd43 animals by transplantation assays. Side scatter low (SSC low ) cells (containing HSPCs) were collected from wild-type kdrl-sw animals and transplanted into irradiated wild-type or jam1a sd43 recipients, followed by FCM analysis at 4 wpt ( Figure 5A). The absolute numbers of donor-derived dsRed + cells were approximately 3-fold lower in jam1a sd43 recipients than in wild-type recipients ( Figure 5B), suggesting that mutation of jam1a causes a defect in hematopoietic niches. We also investigated if wild-type hematopoietic cells can be home to the kidney of iScience Article jam1a sd43 animals. CFSE-labeled wild-type KMCs were transferred to non-irradiated wild-type or jam1a sd43 recipients (280,000 cells/recipient), and KMCs in each recipient at 2 dpt were analyzed by FCM ( Figure 5C). The absolute number of CFSE-labeled KMCs in wild-type recipients was 810 G 222 (n = 11, G s.e.m.), whereas that in jam1a sd43 recipients was only 24 G 11 (n = 11, G s.e.m.) (Figures 5D and 5E).
Given the defect in hematopoietic niches in jam1a sd43 animals, we next investigated the vasculature in the jam1a sd43 kidney in a kdrl:Cerulean background. Although the renal endothelium was normally observed, the sinusoidal endothelium was remarkably reduced in jam1a sd43 animals compared to wild-type animals ( Figure 6A). The percentage of sinusoidal area in the wild-type kidney was approximately 46.2%, whereas the jam1a sd43 kidney contained only 21.7% ( Figures 6B and 6C).
In mammals, Cxcl12 is known to be expressed in stromal cells and endothelial cells in the bone marrow to regulate HSC homing. 11,20 Given the defect in hematopoietic cell homing to the jam1a sd43 kidney, we next compared cxcl12 expression between wild-type and jam1a sd43 kidneys. Whole-mount fluorescent in situ hybridization analysis revealed that the expression of cxcl12a was predominantly detected in the renal endothelium, and there were no obvious differences between wild-type and jam1a sd43 kidneys in terms of expression patterns and levels ( Figures S8A and S8B). qPCR analysis also revealed that the expression of cxcl12a and cxcl12b was unchanged between the wild-type and jam1a sd43 kidney ( Figure S8C).
To investigate the molecular mechanisms underlying the defect in hematopoietic niches in the jam1a sd43 kidney, we performed expression analysis of some niche factor genes in kdrl:Cerulean + endothelial cells from wild-type or jam1a sd43 kidneys. Although expression of cxcl12b and kitlga (kit ligand a) was upregulated, angpt1 (angiopoietin 1), jagged-1a (jagged canonical Notch ligand 1a), vcam1b (vascular cell adhesion molecule 1b), sele (selectin E), and cx43 (connexin 43) were downregulated in jam1a sd43 endothelial cells compared to wild-type endothelial cells, notably, jagged-1a was nearly undetected in jam1a sd43 endothelial cells (Figures 6D and 6E). While sinusoids and other blood vessels share most vascular markers, sinusoidal endothelial cells endocytose acetylated low-density lipoprotein (Ac-LDL). 38,39 In order to compare jagged-1a expression between sinusoids and other endothelial cells, Alexa Fluor 488-conjugated Ac-LDL was intravenously injected into kdrl:Cerulean animals, and kidney cells were analyzed by FCM. In the wild-type kidney, kdrl:Cerulean + endothelial cells were subdivided into two populations based on uptake of Ac-LDL, Ac-LDL high , and Ac-LDL low . Expression of jagged-1a was only detected in Ac-LDL high kdrl:Cerulean + endothelial cells ( Figures 6F and 6G), suggesting that jagged-1a is predominantly expressed in sinusoidal endothelial cells in the kidney. Although the percentage of Ac-LDL high within the kdrl:Cerulean + population was unchanged in the jam1a sd43 kidney ( Figures S8D-S8F), expression of jagged-1a was undetected in Ac-LDL high kdrl:Cerulean + endothelial cells in jam1a sd43 animals ( Figure 6H).

Jagged-1a is required for sinusoid formation in the kidney
In mice, Jagged-1 is known to be a proangiogenic regulator and is required for promoting HSC selfrenewal in the bone marrow. 10,40 To investigate the role of Jagged-1a in hematopoiesis in the zebrafish kidney, we generated jagged-1a mutant zebrafish (jagged-1a kz4 ) by the CRISPR/Cas9 system. qPCR analysis confirmed that the wild-type form of jagged-1a mRNA was lost in the jagged-1a kz4 kidney ( Figures S9A  and S9B). We examined the vasculature of the jagged-1a kz4 kidney in a kdrl:Cerulean background. Similar to jam1a sd43 kidneys, jagged-1a kz4 kidneys also displayed a reduction in sinusoidal endothelium, whereas renal endothelium was unaffected ( Figures 7A and 7B). Section immunocytochemistry revealed that the sinusoidal area in the jagged-1a kz4 kidney was approximately 48% lower than the wild-type kidney ( Figures 7C and 7D). To examine the capacity of hematopoietic niches in the jagged-1a kz4 kidney, DsRed-labeled wild-type KMCs were transplanted into irradiated wild-type or jagged-1a kz4 recipients, followed by FCM analysis at 4 wpt. We found that the absolute numbers of donor-derived dsRed + cells were iScience Article approximately 50% lower in jagged-1a kz4 recipients than in wild-type recipients (Figures 7E and 7F). In addition, homing assays also revealed that the absolute number of BFP + cells that home to the kidney was reduced in jagged-1a kz4 recipients compared to wild-type recipients ( Figures 7G and 7H), suggesting that mutation of jagged-1a phenocopies a defect in hematopoietic niches in jam1a sd43 animals.
Since Jagged-1a is a Notch ligand that activates Notch signaling via interaction with Notch receptors, we next investigated whether enforced activation of Notch in vascular endothelial cells can recover sinusoids in the jam1a sd43 kidney. A dominant activator of Notch signaling, the Notch intracellular domain (NICD), was expressed in vascular endothelial cells in jam1a sd43 animals using the fli1:Gal4; UAS:NICD line. 35 We first examined the expression of Notch target genes, her2 (also known as hes5), her6 (also known as hes1), hey1, and hey2, in kdrl:Cerulean + Ac-LDL high sinusoidal endothelial cells in wild-type, jam1a sd43 , and jam1a sd43 with forced expression of NICD by qPCR. Compared to wild-type sinusoidal endothelial cells, her2, hey1, and hey2 were downregulated in jam1a sd43 sinusoidal endothelial cells, whereas her6 was almost unchanged. In contrast, forced expression of NICD in jam1a sd43 sinusoidal endothelial cells resulted in upregulation of her2, her6, and hey1 ( Figure 8A). Histological analysis revealed that the sinusoidal structure was remarkably increased in the kidney of NICD (+) jam1a sd43 animals ( Figures 8B-8F), suggesting that formation of the sinusoidal structure is required for Jagged-1a-derived Notch signaling in endothelial cells. However, transplantation of wild-type BFP + cells into NICD (À) or NICD (+) jam1a sd43 recipients showed that the absolute number of donor-derived BFP + cells was unchanged between these two types of jam1a sd43 recipients ( Figures S10A and S10B). This is likely because HSPCs cannot home to the kidney without Jam1a expression in vascular endothelial cells. Indeed, homing assays revealed that forced expression of NICD in vascular endothelial cells did not increase the number of wild-type BFP + cells homing to the jam1a sd4 kidney (Figures S10C and S10D). Taken together, these data suggest that Jam1a regulates sinusoidal niche formation in the kidney by regulating jagged-1a expression in vascular endothelial cells, and Jam1a also plays a role in recruiting hematopoietic cells to the kidney.

DISCUSSION
In the present study, we demonstrate that sinusoidal endothelium is a key component of the HSC niche in the zebrafish kidney, and that Jam1a plays a pivotal role in its formation and function in large part via regulation of jagged-1a expression in vascular endothelial cells.
Endothelial cells that comprise sinusoidal blood vessels have been implicated directly in regulating HSCs in murine bone marrow. Most CD150 + CD48 À CD41 À Lineage À HSCs are in contact with sinusoidal endothelial cells, 7 and endothelial-specific deletion of Jagged-1 results in depletion of the HSC pool in the bone marrow. 10 In addition, conditional deletion of Scf, which regulates HSC proliferation, in endothelial cells also leads to HSC depletion in the bone marrow. 9 In vivo imaging analysis in the bone marrow revealed that sinusoids are a major target for HSC homing and engraftment. 8 Consistent with these observations, we showed in zebrafish that runx1 + hematopoietic cells predominantly localized along the sinusoidal endothelium in the DL region of the kidney. The efficiency of HSPC engraftment was lower in jam1a sd43 or jagged-1a kz4 recipients, whereas the sinusoidal structure was recovered in jam1a sd43 animals by forced Notch activation in vascular endothelial cells. These data suggest that the major role of Jagged-1a in HSC niches is to generate sinusoidal niches and that direct interactions with sinusoids may be essential for HSC maintenance and self-renewal in the zebrafish kidney.
Sinusoids are commonly observed in both adult and embryonic hematopoietic organs in mammals, including the bone marrow, 7-10 spleen, 41 and fetal liver. 42,43 In zebrafish embryos, HSCs arise from the dorsal aorta and move to the transient hematopoietic organ, the caudal hematopoietic tissue (CHT), an equivalent organ to the mammalian fetal liver. Previous studies have shown that sinusoidal structures are highly developed in the CHT in the zebrafish embryo. 44,45 Live-imaging analysis of HSPCs in CHT revealed that HSPCs are enwrapped in sinusoidal endothelial cells, which are described as ''endothelial cuddling''. This specific interaction with sinusoidal endothelial cells results in close contact between HSCs and a single mesenchymal stromal cell, leading to oriented cell division of the HSC. 46 These observations, in The expression of Jam1 in HSCs is highly conserved in human, mouse, and zebrafish 30,47 ; however, there have been no reports of its function in the hematopoietic niche. We demonstrated that the number of cells homing to the kidney was decreased in either case when jam1a sd43 animals were used as donor or recipient, suggesting that Jam1a regulates hematopoietic cell homing to the kidney in both a cell autonomous and non-cell autonomous manner. Since Jam1a expression in hematopoietic niches is required for iScience Article hematopoietic cell homing, forced expression of NICD in endothelial cells is not sufficient to rescue niche functions in the jam1a sd43 kidney, but NICD expression is sufficient to restore sinusoidal structures. Collectively, these data suggest that Jam1a not only regulates jagged-1a expression in vascular endothelial cells to form a sinusoidal structure, but also functions to recruit hematopoietic cells to the kidney. Given that Jam1 expression in hematopoietic cells and vascular endothelial cells is highly conserved, the function of Jam1 in hematopoiesis may be conserved among vertebrates.
In zebrafish, melanocytes that located in the dorsal region of the kidney have been shown to play a role in protecting HSPCs from UV light-induced DNA damage. The melanocyte umbrella of the hematopoietic organ is highly evolutionarily conserved in aquatic animals, including jawless fish, teleost fish, and amphibians. 28 We found that runx1 + hematopoietic cells were predominantly detected in the DL region of the kidney that is covered by melanocytes. Moreover, the distribution of sinusoids in the kidney was correlated with that of melanocytes, supporting the idea that hematopoietic niches are protected by melanocytes in aquatic animals.
Since teleost fish and mammals share a common ancestor, the zebrafish kidney provides a complementary system to understand the universal regulatory mechanisms of vertebrate HSCs. Further studies in zebrafish and other vertebrates will define core cellular components that have been conserved over evolutionary time, and lead to new discoveries regarding the molecular cues required for HSC maintenance and expansion ex vivo.

Limitations of the study
In this study, we found in zebrafish that sinusoids are predominantly formed in the DL region of the kidney, and most runx1 + hematopoietic cells are distributed adjacent to the sinusoidal endothelium; however, there are several limitations. First, we cannot show if gata2a + runx1 + HSPCs are closely associated with sinusoidal endothelial cells due to the technical difficulty of visualizing kdrl + sinusoidal endothelial cells under the background of gata2a:GFP in the kidney. Ac-LDL cannot also be used to visualize sinusoidal endothelial cells as Alexa 488, the fluorescent label used for Ac-LDL, fades upon removal of melanin from the kidney by hydrogen peroxide treatment. Since much of the sinusoidal endothelium overlaps with the distribution of the melanin pigment, the sinusoidal endothelium in the kidney cannot be visualized without the removal of melanin. Second, it is not possible to show if gata2a + runx1 + HSPCs home to the DL region of the kidney. Given that only a few gata2a + runx1 + HSPCs are expected to home to the kidney by transplantation of KMCs, no reliable data on HSPC homing can be presented in zebrafish. Third, although we showed that Jam1a regulates the formation of a sinusoidal structure via jagged-1a expression in vascular endothelial cells, the regulatory mechanisms of jagged-1a expression by Jam1a require further exploration.

STAR+METHODS
Detailed methods are provided in the online version of this paper and include the following:

Materials availability
This study did not generate new unique reagents.

Data and code availability
All data reported in this paper will be shared by the lead contact upon request.
No original code was reported in this study.
Any additional information required to reanalyse the data reported in this paper is available from the lead contact upon request.

Generation of mutant zebrafish
For CRISPR/Cas9-mediated generation of a mutant line, the guide RNA (gRNA) was designed to target exon 3 of jam1a or exon2 of jagged-1a and was synthesized as previously described. 52,53 Two complementary oligonucleotides corresponding to the target sequence were annealed and ligated into the pT7-gRNA vector. The gRNA was then synthesized using a linearized vector with MEGAshortscript T7 Transcription Kit (Thermo Fisher Scientific) and purified by mirVana miRNA Isolation Kit (Thermo Fisher Scientific). Cas9 mRNA was synthesized using a linearized pCS2-Cas9 vector with mMESSAGE mMACHINE SP6 Transcription Kit (Thermo Fisher Scientific) and purified by RNeasy Mini Kit (Qiagen). gRNA (50 pg/nL) and Cas9 mRNA (150 pg/nL) were co-injected into one-cell stage embryos. Fish from the F1 generation were screened by genomic PCR. For jam1a sd43 , the mutant allele was identified by sequencing. For jagged-1a kz4 , due to a large deletion in the jagged-1a genomic loci, the mutant allele was determined by genomic PCR and was confirmed by RT-PCR. The sequences of the designed gRNA and primers used for genotyping were listed in Table S1.

Administration
Alexa Fluor 488-conjugated (AF488) acetylated low-density lipoprotein (Ac-LDL) (Thermo Fisher Scientific) was diluted at the final concentration of 66.7 ng/mL in phosphate buffered saline (PBS). Adult zebrafish were retro-orbitally injected with 7 mL of AF488-Ac-LDL solution. At 3 h post-injection, kidneys were dissected to prepare cells for flow cytometric (FCM) analysis.

Cell preparation and flow cytometry
Kidney marrow cells (KMCs) were prepared as previously described. 26 Cells were obtained by pipetting of a dissected kidney in 1 mL of ice-cold 2% fetal bovine serum (FBS) in PBS (2% FBS/PBS). After centrifugation, the pellet was gently mixed with 1 mL of distilled water by pipetting to lyse erythrocytes by osmotic shock. Subsequently, 1 mL of 2X PBS was added. For preparation of vascular endothelial cells, a dissected kidney was digested with Liberase TM (Roche) in PBS for 1 h at 37 C. Cells were then filtered through a 40-stainless mesh and washed with 2% FBS/PBS by centrifugation. Just before FCM analysis, the Sytox Red (Thermo Fisher Scientific) was added to exclude dead cells. FCM acquisition and cell sorting were performed on a FACS Canto 2 (BD Biosciences) or FACS Aria III (BD Biosciences). Data analysis was performed using the Kaluza software (ver. 1.3, Beckman Coulter). The absolute number of cells was calculated by flow cytometry based on the acquisition events, maximum acquisition times, and the percentage of each cell fraction.
qPCR For sorted cells, whole-transcript amplification and double-strand cDNA synthesis was performed as previously described. 26 Cells were directly sorted in a lysis buffer containing 1 mg/mL of polyinosinic-polycytidylic acid, and total RNA was extracted using RNeasy Mini Kit. Reverse transcription (RT) was performed using Super Script III (Thermo Fisher Scientific) and an RT primer, which contains oligo-dT, T7 promoter, and PCR target region sequences. After digestion of remaining RT primers by exonuclease I (Takara), a poly-A tail was added to the 3' ends of the first-strand cDNAs using terminal transferase (Sigma). The second-strand DNA was then synthesized using MightyAmp DNA polymerase (Thermo Fisher Scientific) and a tagging primer, which contains oligo-dT and PCR target region sequences. PCR amplification was performed using a suppression primer, which allow to amplify small-size DNA that contains complementary sequences at both ends of the template DNA. The amplified double-strand cDNA was purified using QIAquick PCR Purification Kit (Qiagen). For the kidney tissue, total RNA was extracted using RNeasy Mini Kit (Qiagen), and cDNA was synthesized using ReverTra Ace qPCR RT Master Mix (Toyobo