Importance of endothelial Hey1 expression for thoracic great vessel development and its distal enhancer for Notch-dependent endothelial transcription

Thoracic great vessels such as the aorta and subclavian arteries are formed through dynamic remodeling of embryonic pharyngeal arch arteries (PAAs). Previous work has shown that loss of a basic helix-loop-helix transcription factor Hey1 in mice causes abnormal fourth PAA development and lethal great vessel anomalies resembling congenital malformations in humans. However, how Hey1 mediates vascular formation remains unclear. In this study, we revealed that Hey1 in vascular endothelial cells, but not in smooth muscle cells, played essential roles for PAA development and great vessel morphogenesis in mouse embryos. Tek-Cre–mediated Hey1 deletion in endothelial cells affected endothelial tube formation and smooth muscle differentiation in embryonic fourth PAAs and resulted in interruption of the aortic arch and other great vessel malformations. Cell specificity and signal responsiveness of Hey1 expression were controlled through multiple cis-regulatory regions. We found two distal genomic regions that had enhancer activity in endothelial cells and in the pharyngeal epithelium and somites, respectively. The novel endothelial enhancer was conserved across species and was specific to large-caliber arteries. Its transcriptional activity was regulated by Notch signaling in vitro and in vivo, but not by ALK1 signaling and other transcription factors implicated in endothelial cell specificity. The distal endothelial enhancer was not essential for basal Hey1 expression in mouse embryos but may likely serve for Notch-dependent transcriptional control in endothelial cells together with the proximal regulatory region. These findings help in understanding the significance and regulation of endothelial Hey1 as a mediator of multiple signaling pathways in embryonic vascular formation.

Thoracic great vessels such as the aorta and subclavian arteries are formed through dynamic remodeling of embryonic pharyngeal arch arteries (PAAs). Previous work has shown that loss of a basic helix-loop-helix transcription factor Hey1 in mice causes abnormal fourth PAA development and lethal great vessel anomalies resembling congenital malformations in humans. However, how Hey1 mediates vascular formation remains unclear. In this study, we revealed that Hey1 in vascular endothelial cells, but not in smooth muscle cells, played essential roles for PAA development and great vessel morphogenesis in mouse embryos. Tek-Cre-mediated Hey1 deletion in endothelial cells affected endothelial tube formation and smooth muscle differentiation in embryonic fourth PAAs and resulted in interruption of the aortic arch and other great vessel malformations. Cell specificity and signal responsiveness of Hey1 expression were controlled through multiple cis-regulatory regions. We found two distal genomic regions that had enhancer activity in endothelial cells and in the pharyngeal epithelium and somites, respectively. The novel endothelial enhancer was conserved across species and was specific to large-caliber arteries. Its transcriptional activity was regulated by Notch signaling in vitro and in vivo, but not by ALK1 signaling and other transcription factors implicated in endothelial cell specificity. The distal endothelial enhancer was not essential for basal Hey1 expression in mouse embryos but may likely serve for Notch-dependent transcriptional control in endothelial cells together with the proximal regulatory region. These findings help in understanding the significance and regulation of endothelial Hey1 as a mediator of multiple signaling pathways in embryonic vascular formation.
Transcription factors play essential roles in complex arrays of developmental events and are implicated in the etiologies of various human diseases (1,2). Multiple upstream signals regulate their expression and function in a cell type-and stage-specific manner, which in turn deeply influences cellular differentiation, proliferation, and movement through transcriptional control of downstream target genes. We and others previously identified the Hey family of basic helix-loop-helix transcriptional repressors that were enriched in the embryonic cardiovascular system (3)(4)(5)(6)(7)(8). Among three family members, the mice null for Hey2 die soon after birth, showing cardiac malformations and abnormal chamber gene expression (9)(10)(11)(12). Combined loss of Hey1 and Hey2 resulted in embryonic lethality due to impaired vascular network formation (11,13,14). In addition, we recently reported that the Hey1 deficiency caused lethal anomalies of the thoracic great vessels (15), which were similar to human congenital defects observed as isolated cardiovascular anomalies or as a manifestation of multiorgan syndromes such as 22q11.2 deletion syndrome (16). Hey1 null mice should serve as a new experimental model for human great vessel malformations, which possesses unique as well as overlapping features compared with existing mouse models (17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27). However, the relative importance of Hey1 actions in vascular cell types is not fully elucidated.
Among a variety of cellular signaling pathways involved in embryonic development, it was first demonstrated that Notch signal activation stimulated the expression of Hey family genes through the Rbpj-dependent transcriptional control (28,29). Hey1 appears sensitive to Notch signaling in vivo, and the deletion of Rbpj or the Notch ligand Dll1 results in down-regulation of Hey1 expression in mouse embryos (5,(30)(31)(32). In addition, bone morphogenetic protein (BMP)-ALK receptor signaling activates the Hey1 and Hey2 transcription, which acts synergistically with Notch signaling in endothelial cells (33)(34)(35). Members of the Hey genes are under the control of other signaling pathways, such as those mediated by transforming growth factor b, hepatocyte growth factor, fibroblast growth factor, and Wnt (36)(37)(38)(39). Despite these lines of evidence, it is not clear how these upstream signaling pathways control cell type specificity of the Hey family expression in various embryonic tissues. For example, Hey1 and Hey2 are mutually exclusive in the atrial and ventricular myocardium, whereas they look co-regulated in the atrioventricular canal and endocardium (3,4). Expression in the pharyngeal epithelium is observed only for Hey1, whereas that in the dorsal ganglion is specific to Hey2. Detailed analyses of transcriptional regulation are necessary to further dissect distinct and complementary roles of Hey family genes in different embryonic tissues.
In this study, we generated Hey1 conditional knockout (cKO) mice and demonstrated that the Hey1 expression in endothelial cells was indispensable for proper pharyngeal arch artery (PAA) development and great vessel morphogenesis. We further performed a bacterial artificial chromosome (BAC)based enhancer screen and identified distal enhancers that were specific to different cell types with Hey1 expression. The novel endothelial enhancer possesses transcriptional activity recapitulating endogenous endothelial expression in large-caliber arteries downstream of Notch signaling. These data will provide important information to further understand Hey1 function and its regulatory mechanisms in cardiovascular development and disease.

Results
The Hey1 gene in endothelial cells is essential for thoracic great vessel development To clarify a cell type that requires Hey1 function for the great vessel formation, we generated a novel Hey1 cKO mouse line (Fig. S1). Hey1 is expressed in both endothelial and smooth muscle cells in developing vasculature (14), so we ablated the Hey1 gene in endothelial and smooth muscle cells using Tek-Cre and Tagln-Cre mice (40,41), respectively. The Tek-Cremediated Hey1 deletion in endothelial cells caused abnormalities of great vessel structure, namely interruption of the aortic arch type B (IAA-B), right-sided aortic arch (RAA), and aberrant origin of the right subclavian artery (ARSA) at embryonic day 18.5 (E18.5) and postnatal day 0.5 ( Fig. 1, A and B), which reproduced congenital anomalies observed in Hey1 null mice (15). We did not detect morphological defects of the heart and signs of circulation failure in Tek-Cre-mediated cKO embryos. Micro-CT analysis clearly showed the three-dimensional image of unordinary running and branching of the aorta in cKO embryos with RAA at E18.5 ( Fig. 1C and Movies S1 and S2). On the other hand, the Hey1 deletion in vascular smooth muscle cells did not affect great vessel morphogenesis (Fig. 1, A and B). These results indicated that endothelial Hey1 expression was indispensable for the thoracic great vessel formation.
Loss of endothelial Hey1 expression leads to disrupted tubular structure of fourth PAAs IAA-B, RAA, and ARSA are all attributable to the fourth PAA defects at early developmental stages (42). We then analyzed the structures of pharyngeal arches and PAAs of endothelial cKO embryos. In control embryos, third, fourth, and sixth PAAs were well-formed in corresponding pharyngeal arches by E10.5 ( Fig. 2A). In contrast, the disruption of endothelial tube structure was observed in fourth PAAs of endothelial cKO embryos at E10.5 and E11.5 ( Fig. 2A), whereas the size and structure of pharyngeal arches were maintained. Pan-endothelial (Pecam1) and arterial endothelial (Nrp1 and Gja5) markers were clearly expressed in the disorganized vasculature (Fig. 2, A  and B). Quantitative RT-PCR analysis of Pecam1 1 endothelial cells indicated that other endothelial marker genes were also expressed at normal levels in Hey1 cKO as well as null embryos (Fig. S2, A and B), suggesting that general endothelial cell differentiation was not compromised. Migration and distribution of neural crest-derived cells, which were marked with Tfap2a (AP2a) and Crabp1, were not altered by the Hey1 deficiency (Fig. S3, A and B); on the other hand, the expression of a smooth muscle marker Acta2 (a smooth muscle actin) was almost undetectable in the affected fourth PAA (Fig. 2C). These characteristics were identical to that in Hey1 null embryos (15), substantiating the importance of endothelial Hey1 for fourth PAA-derived great vessel formation.
Great vessel anomalies occurred in some endothelial cKO mice but not in others (Fig. 1B), which was unrelated to the extent of Hey1 down-regulation because its mRNA level in Pecam1 1 cells did not markedly vary among cKO embryos (Fig. S2A). As is generally accepted for other genetic models showing great vessel malformations (24,(43)(44)(45), it is likely that endothelial Hey1 deficiency predisposes cKO mice to fourth PAA defects. cKO embryos retained ;30% of Hey1 transcripts in the Pecam1 1 endothelial cell population, whereas null embryos showed only negligible Hey1 mRNA expression (Fig.  S2A). As described in its original paper (40), the Tek-Cre-mediated recombination of the Rosa allele was observed in all identifiable endothelial cells by E10.5 (Fig. S4); however, the Hey1 floxed allele should have relatively low efficiency of Cre-mediated recombination. There may be mosaicism of the Hey1 allele recombination, although it is difficult to examine it with singlecell resolution. Nevertheless, it is of note that even a partial reduction of endothelial Hey1 expression was enough to affect fourth PAA development.

Hey1 expression in embryonic tissues is regulated through multiple cis-regulatory regions
Considering the importance of endothelial Hey1 expression for proper vascular development, we attempted identification of tissue-specific enhancers for Hey1 transcription to understand how Hey1 expression was regulated during embryonic development. The BAC construct that encompassed ;173 kb surrounding mouse Hey1 gene (2140 to 133kb) was used for F0 transgenic mouse LacZ reporter analysis (Fig. 3A). As shown in Fig. 3 (A and B) and Fig. S5 (A and B), the full-length BAC-LacZ reporter (designated as "a") showed transcriptional activity that reproduced endogenous Hey1 expression in the cardiovascular system, pharyngeal epithelium, and somites at E9.5. A shorter reporter that lacked the 2140 to 243 kb region (b) lost the activity in the atria, pharyngeal epithelium, and somites but was positive specifically in large-caliber arteries, including PAA, the dorsal aorta, and intersomitic vessels. On the other hand, the deletion without the 243 to 20.3 kb region (c) did not drive the reporter expression in these Endothelial Hey1: Great vessel development and enhancer arteries, whereas it retained the activity in the pharyngeal epithelium and somites. These results suggested presence of multiple cis-regulatory regions for Hey1 transcription in different embryonic tissues.
The BAC-LacZ reporter analysis located the pharyngeal/ somitic enhancer to the region between 2140 and 243 kb. Consistently, a 0.6-kb fragment in the 250 kb region showed robust enhancer activity in the pharyngeal epithelium and somites (Fig. S6A). The activity of full-length BAC-LacZ reporter was reduced by the lack of the 0.6-kb fragment selectively in these tissues, whereas that in the vasculature was maintained ( Fig. S6A). In addition, CRISPR/Cas9-mediated excision of the corresponding area significantly decreased endogenous Hey1 expression in a tissue-specific manner (Fig. S6B), clearly indicating its sufficiency and necessity as a new pharyngeal/ somitic enhancer. Hey1-proximal region responds to vascular signals but is insufficient for endothelial specific transcription in vivo BAC-LacZ studies indicated that the endothelial enhancer was present between 243 and 20.3 kb (Fig. 3, A and B), and we further examined how endothelial Hey1 transcription was controlled in mouse embryos. It was previously reported that the HEY1 expression was synergistically activated by Notch and ALK1 signaling in endothelial cells (Fig. S7A) (34). Transcriptional activity driven by the proximal region (20.5 to 10.2 kb) was up-regulated by the Notch 1 intracellular domain (N1ICD) expression and BMP9 treatment (Fig. 4A) in luciferase reporter assays, which was consistent with the previous finding using BMP6 (33). The proximal region contained three potential Rbpj-binding sites, one of which was required for transcriptional activation ( Fig. 4A and Fig. S7B).
Because Notch and ALK1 signaling pathways are heavily involved in endothelial differentiation and vascular formation (34,35), we first tested the importance of the proximal region for endothelial specific transcription in mouse embryos. Unexpectedly, the proximal region did not show reproducible enhancer activity in the vasculature of LacZ reporter embryos ( Fig. 4B and Fig. S7C). More importantly, the ablation of proximal region from the full-length BAC-LacZ reporter did not reduce the activity in vascular endothelial cells ( Fig. 4B and Fig.  S7C), strongly suggesting that the proximal region could serve for signal responsiveness but was insufficient to achieve endothelial transcription in vivo.
A distal endothelial enhancer is located 18 kb upstream of the mouse Hey1 gene With this result, it was the best conceivable that an additional enhancer for endothelial Hey1 expression was present in the region between 243 and 20.3 kb. We then analyzed publicly available ChIP sequencing (ChIP-Seq) (Gene Expression Omnibus, under accession number GSE88789) data sets and found the transcriptional coactivator p300 binding 18 kb upstream of the mouse Hey1 gene in embryonic endothelial cells (Fig. 5A) (46). Indeed, the ablation of a 1.6-kb sequence at the 218 kb region clearly deprived the full-length BAC-LacZ reporter of vascular transcriptional activity at E9.5 (Fig. 5 (A and B) and Fig. S8A). In addition, the LacZ reporter analysis indicated that the 1.6-kb fragment had highly specific enhancer activity in embryonic vasculature (Fig. 5 (A and B) and Fig. S8A). A detailed analysis using the reporter mouse lines revealed that the transcriptional activity was restricted to endothelial cells of large- Tek-Cre 1 ) embryos at E10.5 and E11.5. B, Gja5 immunohistochemistry at E11.5 also showed the impaired endothelial tube formation of fourth PAAs. Expression levels of these endothelial markers remained unchanged. C, expression of a smooth muscle marker Acta2 was almost undetectable in the affected fourth PAA. PAAs are numbered. Note that the staining of different markers is shown using serial sections for E11.5. Scale bars, 100 mm.
Endothelial Hey1: Great vessel development and enhancer caliber arteries, such as PAAs and the dorsal aorta at E9.5 and E10.5 ( Fig. 5C and Fig. S8B). The 1.6-kb sequence was highly conserved among species (Fig. 5A). A comparable 1.9-kb fragment in the human HEY1 216 kb region also showed LacZ reporter expression in vascular endothelial cells, suggesting relevance of the new, distal endothelial enhancer in human gene regulation (Fig. 5 (A and B) and Fig. S8C). To characterize the distal enhancer specificity, we further established an EGFP reporter line using the mouse 218 kb fragment. Similar to the LacZ reporter lines (Fig. 5C), the EGFP mice showed intense fluorescent signals in endothelial cells of large-caliber arteries, but not in the cardinal veins (Fig. 5D). Endogenous Hey1 mRNA level was significantly higher in EGFP high cells of the Pecam1 high endothelial population, and the expression of arterial endothelial genes was also enriched in EGFP high endothelial cells (Fig. 5, E and F).
The newly identified distal enhancer drove endothelial transcription that recapitulated endogenous Hey1 expression in large caliber arteries. Contrary to our expectation, however, CRISPR/Cas9-mediated excision of this region did not cause significant decrement of endogenous Hey1 mRNA level in mouse embryos (Fig. S9, A and B). These results suggest that the proximal region or an unidentified enhancer may compensate the lack of the distal endothelial activity, whereas it was technically difficult to test such a hypothesis in mouse embryos by deleting multiple candidate regions around the Hey1 gene.

Hey1 distal endothelial enhancer is controlled by Notch signaling
Last, we examined regulatory mechanisms of the distal enhancer for its endothelial activity. In a screen for potential transcriptional regulators based on the consensus binding motif analysis, Foxc1, Foxc2, and N1ICD showed highest induction of the luciferase reporter expression driven by the distal enhancer (Fig. 6, A and B). The responsiveness to Foxc proteins was assigned to a 430-bp fragment (Fig. S10A) and three binding sites (Fig. 6, A and C). Nevertheless, their mutations in the 1.6-kb enhancer did not affect LacZ reporter expression in mouse embryos ( Fig. 6D and Fig. S10B). In marked contrast, mutations in Rbpj-binding sites abolished the response to N1ICD in luciferase analysis ( Fig. 6C and Fig. S11  (A and B)) and resulted in the complete loss of LacZ reporter activity in the embryonic vasculature ( Fig. 6D and Fig. S11C). These Rbpj-binding elements are present in the corresponding genomic region of various species, suggesting that they are functional in humans and other creatures (Fig. S11A). There are no consensus motifs for SMAD binding in the distal enhancer, and ALK1 signaling did not show synergy with Notch signaling for the distal enhancer activation (Fig. 6E), unlike its effect for the proximal region (Fig. 4A).
As previously reported (31), Hey1 mRNA expression in multiple embryonic tissues, including PAAs and the aorta, was markedly decreased in Rbpj null embryos, whereas severe developmental defects made its interpretation difficult (data not shown). We then analyzed whether the distal endothelial enhancer activity was dysregulated in Rbpj null embryos by intercrossing them with EGFP reporter mice. As shown in Fig. 6F, vascular EGFP expression was suppressed to a virtually neglectable level with the Rbpj null background even in E8.5 embryos showing relatively mild developmental defects. These results verified complete Notch dependence of the distal endothelial enhancer and further indicated its usefulness as a surrogate to monitor the Notch signal activity in the embryonic vasculature.

Discussion
In this study, we demonstrate that the conditional deletion of Hey1 in endothelial cells causes abnormalities of thoracic great . Hey1 proximal region responds to vascular signals but is insufficient for endothelial transcription in vivo. A, luciferase reporter assays in human umbilical vein endothelial cells revealed that the N1ICD expression and the BMP9 treatment synergistically induced the transcriptional activity driven by the proximal region, which required a Rbpj-binding site. The sequence of mouse Hey1 proximal region is shown in Fig. S7. **, p , 0.01; ns, not significant. B, in F0 transgenic mouse reporter analyses, the proximal region was insufficient for endothelial transcription (d) and dispensable for the full-length BAC-LacZ reporter activity in vascular endothelial cells (e). Scale bars, 500 mm. Results of other F0 embryos are displayed in Fig. S7. vessel morphogenesis. Embryonic arterial defects occur only in fourth PAAs, which is identical to those observed in Hey1 null mice (15). Fourth PAAs are prone to be disorganized during remodeling due to unique characteristics, including a nonmuscular region where the expression of a smooth muscle actin is reduced or absent (22,47). Left and right fourth PAAs normally form a part of the aortic arch and right subclavian artery, and their defects cause IAA-B and ARSA, respectively. RAA results from the connection of right fourth PAA to the aorta, which compensates for the abnormal involution of left fourth PAA. It is still unclear how the Hey1 deficiency in endothelial cells acts as a predisposing factor for such abnormalities. Among numerous signaling factors implicated in fourth PAA development (17, 19-21, 23, 24, 27, 48-51), Hey1 may be functionally interconnected with endothelial regulatory genes, such as Plxnd1  (Cxcr4, Efnb2, Gja4, Gja5, Dll4, Jagged1, and Hey2), although the expression levels of pan-endothelial (Cdh5 and Pecam1) and venous (Ephb4) genes were equivalent between EGFP high and EGFP low cells. *, p , 0.05; **, p , 0.01; ns, not significant. and Edn1. We performed RNA-Seq analysis of embryonic endothelial cells from Hey1 endothelial cKO as well as null mice, but the results did not show the disturbance of a single gene or a few genes that explained vascular abnormalities by the Hey1 deficiency. There was no significant enrichment that could be directly linked to the pathogenesis in gene ontology or KEGG pathway analysis (data not shown). ChIPgrade antibodies against Hey1 are not available, but ChIP-Seq was previously performed using FLAG-tagged HEY1 overexpression in cultured cells (52,53). These studies suggested possible Hey1 target genes, including Dll4, Kdr, and Foxc1, but the expression of those candidate genes did not change in endothelial cells of Hey1 cKO and null embryos (data not shown). Mosaicism of the Hey1 allele recombination may have made such expression analyses difficult, and a specific population of endothelial cells in a particular region of the embryos might be more sensitive to the Hey1 dosage. Importance in endothelial cells is also evident for complementary functions by Hey1 and Hey2 at earlier vascular formation because their loss in endothelial cells reproduces vascular defects and embryonic lethality observed in doublenull mice (14). Advanced technologies such as multiomics Endothelial Hey1: Great vessel development and enhancer analyses at the single-cell level may give a clue to downstream signaling pathways that are central to Hey-dependent endothelial functions.
Whereas the Hey1 deletion in smooth muscle cells did not cause great vessel anomalies, that does not necessarily negate its supportive functions in smooth muscle cells. Consistently, the incidence of great vessel anomalies appeared lower in endothelial Hey1 cKO mice compared with null mice (15). Although significant retention of Hey1 transcripts in endothelial cells, probably due to low recombination efficiency of the Hey1 floxed allele, can explain low phenotypic penetrance in cKO mice, Hey1 functions in smooth muscle cells or other cell types may also be required for proper fourth PAA development. However, it is noteworthy that Hey1-deficient phenotypes are clearly distinguishable from neural crest-related abnormalities. IAA-B and other great vessel malformations are often observed in patients with 22q11.2 deletion syndrome, and TBX1 is one of the most important genes in the minideletion region (19,(54)(55)(56). Tbx1 transcription factor is required for migration, proliferation, and/or survival of neural crest-derived cells, and its mutant mice are defective also in the outflow tract, thymus, and craniofacial structures (19,44,(55)(56)(57). In sharp contrast, Hey1 null (15) as well as endothelial cKO mice (Fig. S3) did not show abnormalities in neural crest-derived cell behavior as well as malformation of the pharyngeal arches, cardiac outflow tract, and other neural crest-related structures. Mouse embryos with inactivation of Notch signaling in neural crestderived cells showed down-regulation of Hey family expression in the smooth muscle layer of PAAs, but the fourth PAA defects only rarely occurred in these mutant mice (58). Furthermore, the heterozygous Tbx1 deletion in the pharyngeal epithelium leads to great vessel abnormalities (59), indicating that Tbx1 and Hey1 control distinct signaling pathways for PAA development in different cell types.
The present study further elucidates that Hey1 transcription in embryonic tissues is controlled through multiple cis-regulatory regions. In particular, endothelial Hey1 expression essential for vascular development appears maintained through at least two complementary regions. The proximal region adjacent to exon 1 responds to both Notch and ALK1 signaling but is not sufficient to implement endothelial transcription in mouse embryos. The novel enhancer at the distal 218 kb region is solely regulated by Notch signaling and can reproduce the endogenous Hey1 expression pattern in endothelial cells. Nonetheless, the distal enhancer does not single-handedly control endothelial Hey1 transcription because its deletion does not alter Hey1 mRNA level in embryonic endothelial cells. In silico analysis of ChIP-Seq data (46) indicates a couple of additional p300-binding sites around the Hey1 locus (data not shown), although they are located outside the minimal 243 to 20.4 kb region determined in BAC-LacZ analysis. As is often seen with essential regulatory genes for embryonic development (60,61), it is likely that the proximal, distal, and possibly additional regulatory regions supply the place of each other to ensure robust Hey1 expression in endothelial cells.
Hey1 and other Hey family genes have been recognized as typical Notch downstream genes enriched in the cardiovascular system (3,28), and both the Hey1 proximal region and distal en-dothelial enhancer have Rbpj-binding elements that mediate Notch responsiveness. Notch signaling is implicated in early steps of endothelial differentiation, such as arterial-venous specification and tip cell-stalk cell interaction (62,63). In addition, fluid shear stress activates Notch signaling and up-regulates the expression of Notch target genes, including Hey1 (64). It is known that an adequate range of blood flow into developing PAAs is necessary for proper morphogenesis of great vessel structure (65), and Notch-dependent Hey1 expression is probably an important factor that responds to the variable blood flow during drastic PAA remodeling. The present study suggests that a decrement of Hey1 expression in endothelial cells can predispose embryos to PAA defects. Strict regulation of Hey1 expression is likely crucial to prevent congenital malformations of thoracic great vessels.
The Hey1 distal enhancer activity in arterial endothelial cells is abolished by the mutations of Rbpj-binding sites and is clearly repressed in Rbpj null embryos. Other arterial endothelium enhancers show different modes of Rbpj dependence. The Dll4 enhancer is regulated by Notch signaling in combination with SoxF and Ets factors, but its activity is influenced neither by the Rbpj site mutations nor by Rbpj deficiency (66,67). The Notch1 and Ece1 endothelial enhancers do not contain Rbpj sites but display arterial specificity by the actions of SoxF, Ets, and Foxc factors (68,69). On the other hand, Rbpj acts as a transcriptional repressor in venous endothelial cells for the Flk1 arterial enhancer (70). Such a diversity of regulatory mechanisms is certainly important for unique characteristics of their expression.
During embryonic development, Hey1 is expressed in various other tissues, such as the pharyngeal epithelium, somites, and cardiac atrium (3,4). This study further identified a distal enhancer for the pharyngeal epithelium and somites. Marked reduction of endogenous Hey1 expression by its deletion clearly shows importance as a specific enhancer, but a low level of Hey1 mRNA is still detectable in these tissues. Because the BAC-LacZ reporter activity in these two areas is suppressed by the lack of proximal region (Fig. 4B), the proximal region may serve for Hey1 transcription also in the pharyngeal epithelium and somites. In addition, Hey1 expression restricted to the atrium of the heart is particularly interesting because its family gene Hey2 is differentially expressed in the ventricle (3,4). We recently reported that ventricular Hey2 expression was controlled by Tbx20 and Gata family proteins through its distal enhancer (71). Although the full-length BAC apparently contains transcriptional activity in the atrium, we have not specified where the Hey1 atrial enhancer is located. Studying the precise mechanisms of Hey1 transcriptional regulation will help understand how tissue-specific gene expression is achieved in the cardiovascular system and other organs during embryonic development.
Increasing evidence supports the importance of the HEY family in human physiology and disease. Their expression is often dysregulated downstream of Notch signaling in various diseases (72)(73)(74). In addition, a fusion of HEY1 and NCOA1 causes chondrosarcoma and aberrant HEY1 expression is correlated with the metastasis, therapeutic response, and patient survival in multiple cancer types (38,(75)(76)(77)(78). A variant near the HEY2 locus is associated with clinical characteristics of Brugada syndrome (79), and HEY2 also acts as a key factor for the differentiation of pluripotent stem cells into ventricular myocytes (80). It is tempting to hypothesize that variations in the protein-coding or enhancer regions of the HEY1 gene affect its molecular function or expression level in human patients with isolated great vessel anomalies. HEY1 may also be associated with clinical variability of 22q11.2 deletion syndrome as a modifier gene. It will be of clinical interest to examine possible involvement of HEY1 in such entities of human congenital diseases.

Mouse strains
The founder mice with the Hey1 floxed allele were generated as follows (Fig. S1). The targeting vector was designed to delete exons 2, 3, and 4 of the Hey1 gene by Cre-mediated recombination and was electroporated into HK3i ES cells (81) with two single guide RNAs (Table S1) and the human optimized Cas9 expression plasmid px330 (Addgene plasmid 42230) (82). Recombinant ES cells were injected into eight-cell stage embryos to produce chimera mice. The Pgk-Neo cassette flanked by FRTs was removed by crossing the chimera mice with Flp-expressing female mice (Jackson Laboratory, JAX 003800). PCR primers for the genotyping are shown in Fig. S1 and Table S1. The Hey1(loxP) line information is available at RRID:SCR_019138 (accession no. CDB1302K). Tek-Cre (JAX 008863) and Tagln (SM22a)-Cre (JAX 004746) mice were used for the tissue-specific deletion (40,41).
The mouse lines with the enhancer deletion at the Hey1 218 or 250 kb region were generated by CRISPR/Cas9 genome editing. Human optimized Cas9 mRNA or Alt-R S. p. Cas9 Nuclease 3NLS (Integrated DNA) was introduced into BDF1 fertilized eggs with two single guide RNAs (Table S1) (83). The targeted regions were PCR-amplified for the sequencing at the F0 and F1 generations. Mice at the F1 and later generations were used for expression and phenotype analyses.
All animal experiments were approved by the institutional animal care and use committees of the National Cerebral and Cardiovascular Center and RIKEN Kobe Branch.

Micro-CT imaging
A contrast reagent eXIA 160XL (Summit Pharmaceutical International) was introduced from the right ventricle of E18.5 embryos, and micro-CT images were taken with an 8-mm pixel size using SkyScan 1276 (Bruker).

LacZ reporter analysis
A BAC clone encompassing the mouse Hey1 gene, RP23-255P16, was purchased from Advanced GenoTechs Co. The BAC-LacZ reporter construct was generated by inserting the nuclear localization signal (nls)-LacZ-poly(A) fragment at the first ATG site of the Hey1 gene as we reported previously (84). Deletion BAC series were prepared with the kanamycin R insertion. Plasmid-based LacZ reporters of the proximal region and distal enhancers were made using the nls-LacZ-poly(A) and hsp68 promoter-nls-LacZ-poly(A) vectors, respectively. Rbpj-binding sites in the 218 kb enhancer were mutated as follows: site 1, ACGTGATGGGAATTGGA ! ACGTGActt-tAATTGGA; site 2, GGAGCGTGGGAACCCCG ! GGA-GCGctttAACCCCG. Foxc-binding sites were mutated as follows: site 1, AGCTTTATTGAGATACA ! AGCTTaAagGA-GATACA; site 2, TCACTTAAAATATGTGA ! TCACT-TAActTtTGTGA; site 3, TAGAATATTTTCATTTT ! TAGAAaAagTTCATTTT. Transgenic mice were generated using standard methods (71). In brief, the circular BAC reporter or linear enhancer reporter fragment (1-3 ng/ml) was injected into the pronuclei of BDF1 fertilized eggs. b-Gal reaction of mouse embryos was performed as we described previously (71).

Luciferase reporter analysis
A DNA fragment of the 218 kb enhancer was inserted into the pGL4.10[luc2] plasmid containing the MLP promoter (85). Human umbilical vein endothelial cells were seeded in the EGM2 medium (Lonza) 24 h before the transfection, and the plasmids for the luciferase, CMV-b-gal, and mouse N1ICD expression were introduced using Lipofectamine PLUS and Endothelial Hey1: Great vessel development and enhancer LTX reagents (Life Technologies, Inc.). After 1 h of transfection, the cells were preconditioned in the EBM2 medium (Lonza) containing 0.2% BSA and 0.2% fetal bovine serum for 4 h, followed by the treatment with BMP9 (R&D Systems) or vehicle. Transfection of 293T cells was performed according to our previous report (86). Luciferase and b-gal activities were measured using FLUOStar Omega (BMG LABTECH). Triplicated assays were independently performed three times, which gave reproducible results. Statistical analysis was performed using Tukey's test.

Data availability
All data are contained within the article and supporting information.