Analysis of Dll4 Regulation Reveals a Combinatorial Role for Sox and Notch in Arterial Development

Analysis of Dll4 regulation reveals a combinatorial role for Sox and Notch in arterial development

Natalia Sacilottoa, Rui Monteirob, Martin Fritzschea, Philipp W. Beckera, Luis Sanchez-del-Campoa, Ke Liuc, Philip Pinheirob, Indrika Ratnayakaa, Benjamin Daviesd, Colin R. Godinga, Roger Patientb, George Bou-Ghariosc, and Sarah De Vala,1

aLudwig Institute for Cancer Research Ltd., Nufﬁeld Department of Clinical Medicine, University of Oxford, Oxford OX3 7DQ, United Kingdom; bMolecular Haematology Unit and British Heart Foundation Centre of Research Excellence, Weatherall Institute of Molecular Medicine, John Radcliffe Hospital, University of Oxford, Oxford OX3 9DS, United Kingdom; cKennedy Institute of Rheumatology, University of Oxford, Oxford OX3 7DQ, United Kingdom; and dWellcome Trust Centre for Human Genetics, University of Oxford, Oxford OX3 7BN, United Kingdom

Edited by Eric N. Olson, University of Texas Southwestern Medical Center, Dallas, TX, and approved June 4, 2013 (received for review January 14, 2013) The mechanisms by which arterial fate is established and main- DNA-binding factor RBPJ (CSL) and coactivator Mastermind tained are not clearly understood. Although a number of signaling to form an activation complex (10). Loss of Notch signaling in pathways and transcriptional regulators have been implicated in zebrafish results in defective arterial-venous differentiation al- arterio-venous differentiation, none are essential for arterial though a distinct dorsal aorta still forms, and some arterial- formation, and the manner in which widely expressed factors specific genes are expressed (10). Mice null for the Notch1 and may achieve arterial-specific gene regulation is unclear. Using both -4 receptors also display severe vascular remodeling defects, as mouse and zebrafish models, we demonstrate here that arterial do those lacking Rbpj and Delta-like ligand 4 (Dll4), the arte- specification is regulated combinatorially by Notch signaling and rial-specific Notch ligand (12–14). Loss of VEGF signaling in SoxF transcription factors, via direct transcriptional gene activa- fish can be rescued by activation of Notch signaling, suggesting tion. Through the identification and characterization of two arte- that Notch is downstream of VEGF in a common genetic rial endothelial cell-specific gene enhancers for the Notch ligand pathway (15). However, VEGF inhibition results in more severe

Delta-like ligand 4 (Dll4), we show that arterial Dll4 expression vascular defects, and arterial genes display differential sensi- BIOLOGY requires the direct binding of both the RBPJ/Notch intracellular tivity to Notch and VEGF signaling (10, 15). It is therefore likely DEVELOPMENTAL domain and SOXF transcription factors. Specific combinatorial, that additional, currently unidentified factors contribute to ar- but not individual, loss of SOXF and RBPJ DNA binding ablates terial differentiation downstream of VEGF. all Dll4 enhancer-transgene expression despite the presence of Spatiotemporal regulation of gene transcription during de- multiple functional ETS binding sites, as does knockdown of sox7; velopment is primarily accomplished by enhancers, although sox18 in combination with loss of Notch signaling. Furthermore, those located close to core promoters are also often termed triple knockdown of sox7, sox18 and rbpj also results in ablation proximal promoters (16). Investigation of endothelial enhancers, of endogenous dll4 expression. Fascinatingly, this combinatorial coupled with the analysis of mutant animal models, has dem- ablation leads to a loss of arterial markers and the absence of a de- onstrated the essential role of the ETS transcription factor family tectable dorsal aorta, demonstrating the essential roles of SoxF and during early endothelial cell specification and development (17, Notch, together, in the acquisition of arterial identity. 18). However, the transcriptional cascades regulating the differentiation of endothelial cells are less well understood. Few arterial-specific enhancer | CSL direct transcriptional targets of Notch have been identified in arterial endothelial cells, and the mechanisms regulating ndothelial cell commitment to an arterial or venous fate the expression of Notch components and other arterial-specific Eduring development, once thought to be primarily de- genes are poorly understood. termined by physiological cues, is now known to be mediated by Results the specific expression of genes, including those encoding fi fi ligands, receptors, and transcription factors (1). The endothelial- Identi cation of an Arterial Endothelial Cell-Speci c Enhancer Within Dll4 expressed SOXF subfamily of transcription factors has been the Third Intron of the Gene. An in silico search of the Dll4 implicated in this process. Similar to other SOX proteins, SOXF locus for regions enriched in human umbilical vein endothelial cells (HUVEC)-specific H3K27Ac histone modification, associ- factors directly bind the core DNA sequence WWCAAW (2, 3). fi In zebrafish, sox7 and sox18 are expressed in both the dorsal aorta ated with active enhancer regions (19), identi ed the highly conserved third intron of Dll4 (Dll4in3) as a putative enhancer and posterior cardinal vein during early development whereas (Fig. S1A). A 937-bp fragment containing the mouse Dll4in3 zebrafish sox17 is not vascular (4, 5). Combined knockdown of region was cloned upstream of the silent hsp68 minimal pro- sox7 and sox18 results in fusions between axial vessels and in moter-β-galactosidase (LacZ) reporter gene (Dll4in3hsp) or the down-regulation, but not loss, of arterial markers, including dll4 – endogenous Dll4 promoter-LacZ reporter gene (Dll4in3end) (4 6). In mammals, all three SOXF proteins are expressed in the (Fig. 1A and Fig. S1B). Mice transgenic for both Dll4in3end and developing aorta (7). However, expression is also detected in some Dll4in3hsp expressed the LacZ reporter gene in arterial but not veins, and Sox18 plays an additional role in early lymphatic vessel differentiation (7–9). Functional redundancy, overlapping expres-

sion, and a lack of direct gene targets has meant that the role of Author contributions: S.D.V. designed research; N.S., R.M., M.F., P.W.B., L.S.-d.-C., K.L., I.R., SOXF proteins in mammals is unclear, and their lack of specificity and B.D. performed research; R.M., P.P., and C.R.G. contributed new reagents/analytic within the vasculature suggests that they act in synergy with other, tools; N.S., R.M., R.P., and G.B.-G. analyzed data; and N.S. and S.D.V. wrote the paper. yet to be identified factors (6, 8). The authors declare no conflict of interest. The Notch signaling pathway also plays a key role in arterio- This article is a PNAS Direct Submission. venous differentiation, where most components of the Notch Freely available online through the PNAS open access option. pathway are specific to arteries (10, 11). Activation of Notch re- 1To whom correspondence should be addressed. E-mail: [email protected]. ceptor after ligand binding results in the translocation of Notch This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10. intracellular domain (NICD) to the nucleus, where it engages the 1073/pnas.1300805110/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1300805110 PNAS Early Edition | 1of6 Downloaded by guest on October 2, 2021 zebrafish. The 937-bp mouse Dll4in3 sequence, in conjunction with the E1b minimal promoter (21) and GFP reporter gene, was used to generate the stable transgenic zebrafish line tg(Dll4in3: GFP) (Fig. 1M). As in transgenic mice, reporter gene expression in tg(Dll4in3:GFP) embryos was detected in arteries (including the dorsal aorta), but not in veins (Fig. 1 N and O and Fig. S1P). These results clearly demonstrate that the Dll4in3 enhancer contains sufficient information to direct arterial endothelial cell- specific expression during embryonic development, independent of the endogenous promoter. The Dll4in3 enhancer is deeply conserved throughout verte- brate evolution, and the orthologous zebrafish region could be clearly identified, although only a small part of the intronic sequences shared notable sequence conservation (Fig. S1A). To establish whether the information directing arterial-specific expression of Dll4in3 was functionally conserved between mouse and zebrafish, we generated mice transgenic for a 699-bp zebrafish sequence including Dll4 intron 3 (zfishDll4in3) upstream of hsp68-LacZ (Fig. 1P). The zebrafish enhancer was able to direct an arterial endothelial cell-specific expression pattern in a manner similar to mouse Dll4in3hsp (Fig. 1 Q–U), suggesting that sequence motifs conserved within both enhancers play key roles in arterial specification.

Dll4in3 Enhancer Is Bound and Regulated by a Combination of SOXF and RBPJ Transcription Factors. The sequences conserved between the mouse and zebrafish Dll4in3 enhancers were investigated using ClustalW (22) to identify putative transcription factor binding motifs, which were then verified by electrophoretic mo- bility shift assays (EMSA) and chromatin immunoprecipitation (ChIP) (Fig. 2A, Fig. S2). Binding sites for SOXF (sites a and b), RBPJ (site a), and MEF2C, TBX, and ETS (sites a, b, g, and h) were identified and confirmed (Fig. S2 A–C). Additional sites for ETS (sites c, d, e, and f), FOXC, and RBPJ (site b) showed no detectable binding in EMSA (Fig. S2 C and D). To determine which motifs contributed to arterial expression, we generated a series of modified enhancers in which individual binding motifs were destroyed through sequence mutation or deletion (exact sequences of mutations and deletions are provided in SI Mate- rials and Methods). These constructs were then used to generate Tol2-mediated mosaic transient transgenic zebrafish, and GFP expression in the fish vasculature was assessed at 24 hours post- fertilization (hpf) (Fig. 2B and Fig. S3). Unsurprisingly, in light of the known role of ETS factors in early endothelial development (20), mutation of the functional ETS binding motifs Fig. 1. The Dll4in3 enhancer directs arterial-specific expression in both trans- resulted in near-total ablation of Dll4in3 expression (Fig. 2B and genic mice and zebrafish. (A) Schematic representation of mouse Dll4 gene Fig. S3C). ETS proteins are expressed in both veins and arteries, (Upper) and Dll4in3hsp transgene (Lower). (B–L) The Dll4in3hsp transgene and ETS binding motifs are required elements in all known directs arterial-specific expression. Representative transgenic whole-mount vascular enhancers (18). Consequently, we hypothesized that embryos (B and D) and transverse sections (C and E)showreportergeneex- alternative transcription factors may contribute to arterial-spe- pression (X-Gal staining, blue) in dorsal aorta (da) but not the cardinal vein (cv). cific Dll4in3 expression. Surprisingly, however, alterations af- μ (Scale bar: 50 m.) At E15, X-Gal staining is detected in arterial (a) but not ve- fecting binding to any other single cis-motif within the Dll4in3 nous (v) cells in whole-mount embryo (F and G), yolk sac (H), and transverse fi fi enhancer did not result in a signi cant loss of expression in the section (I). E15 transverse paraf n sections showed overlapping expression fi – of X-gal staining (J), or β-gal (K and L) with the arterial markers Neuropilin1 transgenic sh assay (Fig. 2B and Fig. S3 D G). (Nrp1) (J′) and Ephrin B2 (Efnb2) (K′′). Expression of the venous marker Endo- To ensure that these observations were not specific to the fish mucin (Emcn) did not overlap that of the reporter gene (L′′). (Scale bar: 10 μm.) model, and to more clearly examine the expression pattern of (M–O) The mouse Dll4in3 transgene (M) directs arterial-specific expression of mutated transgenes within the vasculature, the same modified the GFP reporter gene in transgenic zebrafish line tg(Dll4in3:GFP) to dorsal enhancers were tested in transgenic mice (Fig. 2 C–F, Fig. S4, aorta (da) and intersegmental arteries (ISA) but not the posterior cardinal and Table S1). Because FOXC transcription factors have been vein (pcv) or intersegmental veins (ISVe)(N and O). (P–U) The zebrafish Dll4in3 Dll4 fi reported to regulate expression (23), we also mutated the transgene (P) directs arterial-speci c expression in transgenic mice in arteries putative FOXC binding motif although we found no evidence of but not veins in whole-mount embryos (Q and S) and transverse sections (R and fi T). No overlap was seen with the venous marker Emcn (U). See also Fig. S1. FOXC biding in EMSA (Fig. S2C). As in zebra sh, ablation of protein binding to any individual motif, including FOXC, had no detectable effect on arterial-specific expression of the reporter in venous endothelial cells throughout embryonic development and mice (Fig. 2 C–F, Fig. S4). Mutating the RBPJ-a site in combi- postnatally, mimicking the expression pattern of endogenous nation with the RBPJ-b site (which did not bind in EMSA Dll4 (20) (Fig. 1 B–L, and Fig. S1 C–O). The Dll4in3 enhancer analysis) also had no detectable effect on arterial-specific expres- was also able to drive arterial-specific expression in transgenic sion (Fig. 2C, Fig. S4). However, mutation of SOX-a combined

2of6 | www.pnas.org/cgi/doi/10.1073/pnas.1300805110 Sacilotto et al. Downloaded by guest on October 2, 2021 BIOLOGY DEVELOPMENTAL

Fig. 2. The Dll4in3 enhancer contains cis-motifs that bind RBPJ and SOXF and are required for arterial-specific expression. (A) Multispecies alignment of the conserved region of the Dll4in3 enhancer using ClustalW. Colored sequences are confirmed by EMSA; gray sequences are motifs identified in silico that did not bind in EMSA. (B) Summary of reporter gene expression detected at 24 hpf in Tol2-mediated mosaic transient transgenic zebrafish embryos. GFP expression was scored as positive in the vasculature when at least one intersegmental vessel, or equivalent expression in axial or head vessels, was observed. See also Fig. S3.(C) Summary of reporter gene expression detected in transient transgenic mice. Expressor/transgenic denotes the number of embryos with detectable X-gal staining anywhere compared with the total number of transient transgenic embryos. Arterial specificity was determined by whole-mount, dissection, and histological analysis. Intensity of arterial staining is the average for each transgene, from a qualitative assessment of individual transgenic embryos (Table S1 and Fig. S4). (D–J) Transient transgenic mouse embryos generated using Dll4in3 mutant transgenes. Whole-mount (Upper) and transverse sections (Lower) of representative X-Gal– stained embryos. X-Gal staining can be detected in the dorsal aorta (da) but not cardinal vein (cv) of ΔRBPJ-a, ΔSOX-a, ΔSOX-b, ΔSOX-a/b, and ΔRBPJ-a/SOX- a embryos (D–H). No vascular staining is detected in E11 (I) and E9 (J) embryos transgenic for the ΔRBPJ-a/SOX-a/b construct. (Scale bar: 100 μm.)

with deletion of SOX-b resulted in reduced vascular expres- Second Dll4 Arterial Enhancer Is also Regulated by SOX and RBPJ. It is sion although arterial specificity was still observed (Fig. 2G). now well established that dynamically expressed genes often Similar results were seen after combined mutation of SOX-a and contain multiple enhancers, some of which direct overlapping, or RBPJ-a motifs (Fig. 2H). More interestingly, ablation of both at least very similar, patterns of expression. This phenomenon SOXF-a and -b in combination with the RBPJ-a mutation has been reported for a number of endothelial-expressed genes, resulted in a total absence of transgene expression in arteries, including Mef2c, Tal1, and Kdr (18, 24). Some arterial β-gal ex- + in both embryonic day (E) 11 and E9 transgenic embryos (Fig. pression is detectable in Dll4LacZ/ mice, where a region in- 2 C, I,andJ and Fig. S4), and a near-absolute silencing of the cluding the Dll4in3 enhancer was replaced by LacZ (14), enhancer transgene in transgenic fish (Fig. 2B and Fig. S3J). suggesting that the Dll4 gene was under the regulation of addi- Because SOXF factors are known to interact with MEF2, care tional endothelial enhancers present somewhere else. Because was taken to confirm that the RBPJ-a mutation did not alter the the 51-kb Dll4 locus contained four peaks of HUVEC-specific binding of MEF2C (Fig. S2C, lane 9). These results suggest that H3K27Ac (Fig. S1A), we tested the remaining three peaks RBPJ and SOXF transcription factors combinatorially regu- (−16, −12, and +14) upstream of the hsp68 minimal promoter late Dll4 arterial expression through direct DNA binding at and LacZ reporter gene in transgenic mice. Although Dll4-16hsp the Dll4in3 enhancer. and Dll4+14hsp were unable to direct expression in endothelial

Sacilotto et al. PNAS Early Edition | 3of6 Downloaded by guest on October 2, 2021 (Fig. 5 A and B and Fig. S7 A and B), suggesting that the loss of dll4 expression observed in Fig.4 was not due to a total absence of vasculature, and indicating that fli1a was not a downstream target of the soxF/rbpj code. Indeed, analysis of early zebrafish embryos revealed that triple sox7;sox18;rbpj MO did not affect the assembly of angioblasts at the posterior lateral mesoderm, nor the coalescence of these angioblasts at the midline to form a single vascular cord (Fig. 5A). However, by 26 hpf, tg(Fli1a: GFP) embryos injected with control MO displayed a detectable dorsal aorta and cardinal veins whereas triple sox7;sox18;rbpj MO embryos had only one discernible axial vessel, suggesting a defect in arterio-venous segregation (Fig. 5, Fig. S7B) (27). In addition, no evidence of increased apoptosis was detected (Fig. S7 C and D). These morphological differences persisted at 48 hpf Fig. 3. The Dll4-12 enhancer contains essential RBPJ and SOXF motifs. (A) (Fig. 5B and Movies S1, S2, S3, and S4). In agreement with Schematic representation of the mouse Dll4 locus (Upper, exons are vertical previous reports, rbpj or sox7;sox18 MOs injected separately lines, putative enhancers depicted by colored horizontal lines) and Dll4- resulted in an intermediate phenotype in which a dorsal aorta 12hsp transgene (Lower). (B–E) A representative E11 X-Gal–stained trans- and cardinal vein could still be seen (Fig. 5B and Fig. S7B)(4–6, genic whole-mount embryo (B) and transverse section (C) show reporter 10). Examination of tg(kdrl:GFP);tg(gata1:dsRed) embryos 48 hpf gene expression in the dorsal aorta (da) but not the cardinal vein (cv). (Scale after triple sox7;sox18;rbpj MO further confirmed the phenotype bar: 100 μm.) (D and E) Ablation of RBPJ, SOX-b, and SOX-d sites in the Dll4- 12 enhancer results in loss of enhancer expression. Representative E11 whereas gata1:dsRed expression patterns demonstrated the loss X-Gal–stained whole-mount embryo (D) and transverse section (E). (Scale of blood flow in triple MO embryos (Fig. S7E). Interestingly, bar: 100 μm.) The values on the bottom right represent the number of sprouting of intersegmental vessels, although retarded, was still vascular expressing embryos per number of transgenic embryos. See also detected at 48 hpf in both tg(fli1a:GFP) and tg(kdrl:GFP) em- Figs. S4 and S5 and Table S1. bryos after triple sox7;sox18;rbpj MO-injection. Although this sprouting may be a consequence of MO dilution during de- – − velopment, efnb2a MO injected embryos, unable to form a dis- cells, the 857-bp Dll4 12 enhancer directed arterial endothelial- tinct dorsal aorta, also form intersegmental vessels (27). specific expression in a pattern similar to Dll4in3 (Fig. 3 A–C, fi These observations, combined with the loss of dll4 expression Fig. S5 A and B) although it was not conserved in zebra sh, nor seen in Fig.4, suggest that, whereas angioblast formation and able to drive expression in transgenic zebrafish. Analysis of Dll4- fi migration are unaffected by triple sox7;sox18;rbpj MO, endo- 12 identi ed two functional SOX sites and one RBPJ site (Fig. thelial cells do not take on an arterial identity and the dorsal S5 C–E). A mutant Dll4-12 enhancer lacking functional versions aorta fails to form. Therefore, we looked at the expression of of these sites (SI Materials and Methods) was no longer able to markers of venous and arterial differentiation in zebrafish after direct transgene expression to the arterial compartment (Fig. 3 D control, single rbpj,doublesox7;sox18,andtriplesox7, sox18, and E), further demonstrating the key roles of SOXF and RBPJ rbpj MO knockdown (Fig. 5 and Fig. S8). The expression of flt4 in Dll4 gene regulation during embryonic development. remained robust, indicating that venous identity was not com- fi Both SOXF and RBPJ Factors Are Required for Arterial Development. promised. However, we could nd no detectable expression of any known arterial marker in triple sox7, sox18, rbpj embryos, We next examined the consequences of morpholino (MO) fl knockdown of the orthologous genes sox7, sox18, and rbpj in including efnb2a, dlC, notch1b, t1, notch1b, notch3 and hey2 – – WT and tg(Dll4in3:GFP) zebrafish embryos. All MOs used were (grl) (Fig. 5 C E, Fig. S8 A D). notch3 expression was also ab- validated in previous studies (SI Materials and Methods). MO- sent after rbpj MO alone and is likely a direct Notch signaling induced knockdown of rbpj had little effect on the expression of either Dll4in3:GFP or endogenous dll4 although the known hy- persprouting phenotype was clearly detectable (Fig. 4, Fig. S6A) (25). The double sox7;sox18 MO led to diminished, but still detectable, expression of GFP in tg(Dll4in3:GFP) embryos, with similar reductions in endogenous dll4 levels, agreeing with previous studies (Fig. 4B, Fig. S6A) (6). Remarkably, triple sox7; sox18;rbpj MO resulted in complete silencing of GFP expression in tg(Dll4in3:GFP) fish, with cognate near-total ablation of endogenous dll4 expression in the vasculature (Fig. 4 A and B, Fig. S6). Inhibition of NICD translocation to the nucleus by treating tg(Dll4in3:GFP) embryos with the γ-secretase inhibitor N-[N- 3,5-Difluorophenacetyl]-L-alanyl-S-phenylglycine Methyl Ester (DAPM) exactly recapitulated the results obtained with the rbpj MO (Fig. S6 B–E). Furthermore, chemical inhibition of the VEGF pathway resulted in total loss of Dll4in3:GFP expres- Fig. 4. Combinatorial knockdown of sox7, sox18, and rbpj in zebrafish sion, alone and in combination with rbpj and sox7;sox18 MOs results in ablation of both Dll4in3:GFP and endogenous dll4 expression. (A) (Fig. S6 F–I). These results indicate that Notch and SoxF fac- Analysis of GFP expression in scrambled, rbpj, double sox7;sox18, and triple tors combinatorially regulate Dll4 expression in arteries down- sox7;sox18;rbpj MO-injected tg(Dll4in3:GFP) embryos at 34 hpf. (B) Analysis stream of VEGF. of endogenous dll4 gene expression in scrambled, rbpj, double sox7;sox18, and triple sox7;sox18;rbpj MO-injected WT zebrafish embryos at 26 hpf. dll4 To more accurately understand the phenotypic effects of triple expression was detected by in situ hybridization. Values on the bottom right sox7;sox18;rpbj knockdown, we investigated the expression pat- fl indicate number of embryos with the predominant and displayed phenotype tern of the pan-vascular marker i1a after MO injections (26). per total number of embryos analyzed. In all cases, 0.125 pmol of sox7 and Triple sox7;sox18;rbpj MO did not result in notable reductions in sox18 and 0.15 pmol of rbpj MO were used. *, highlights ectopic neural GFP levels in tg(Fli1a:GFP) embryos, nor of endogenous fli1a expression; arrowhead shows dorsal aorta. See also Fig. S6.

4of6 | www.pnas.org/cgi/doi/10.1073/pnas.1300805110 Sacilotto et al. Downloaded by guest on October 2, 2021 the inhibition of the entire early arterial gene expression pro- gram. These results are notably more severe than those seen after dll4 knockdown (Fig. S8 E–K). Overall, the data presented here suggest that SoxF and Notch signaling controls arterial gene expression beyond the regulation of the Dll4 gene, re- vealing a combinatorial SoxF/Notch transcriptional code for arteriogenesis. Discussion Collectively, our data provide evidence supporting essential roles for SoxF and Notch, together, in the acquisition of arterial identity. It has been hypothesized that Notch signaling acquires specificity through the formation of transcriptional complexes with other, more specific factors. However, while the re- spective expression patterns of SoxF and Notch factors mean that they overlap in only the arterial compartment of the vasculature, potentially providing the required specificity, our data do not necessarily support the idea of a SoxF/Notch transcriptional complex. Dll4 expression occurs to some extent in the absence of direct binding of either SOXF or RBPJ, albeit at reduced levels. Although it is possible that SOXF and RBPJ binding was not completely ablated in the Dll4 mutant transgenes, due to cryptic binding sites elsewhere in the enhancers, the silencing of the transgene after combined RBPJ/SOX site mutations and the similarity of these results to the morphant fish suggest otherwise. Although RBPJ, NICD, and SOXF may form

a transcriptional complex that requires only one factor to directly BIOLOGY

bind DNA, neither knockdown of sox7;sox18 and rbpj separately, DEVELOPMENTAL nor chemical inhibition of the Notch pathway, mimics the severe arteriogenesis phenotype seen in the triple morphant fish, agreeing with previous reports that SOXF regulation occurs through a Notch-independent pathway (6). These results suggest that some degree of arterial identity can be acquired downstream of either Notch or SoxF, such that only perturbation of both pathways results in the loss of arterial identity. Despite the genetic data linking the VEGF and Notch signaling pathways, it is still not clear how these two pathways interact to drive arterial endothelial cell differentiation. Because the Dll4 ligand plays such an integral, dose-dependent role in Notch signaling in the vasculature, it is enticing to think that VEGF may be upstream of Notch in part via the regulation of SOXF factors. VEGF-induced SOXF expression in endothelial cells would consequently achieve low-level Dll4 activation, in- creasing Notch signaling in neighboring cells, and thus reinforcing Fig. 5. Triple sox7;sox18;rbpj knockdown results in the loss of a detectable Dll4 expression. It will therefore be fascinating to uncover the dorsal aorta. (A) Analysis of GFP expression in tg(fli1a:GFP) zebrafish embryos mechanisms by which VEGF may regulate SOXF transcriptional after injection of control (Upper) and triple sox7;sox18;rbpj MO (Lower) activation. during the migration of angioblasts from the lateral plate mesoderm toward The abnormal vascular remodeling and arterio-venous differ- the midline and coalescence into a vascular cord. At 26 hpf, the dorsal aorta entiation seen in β-catenin mutant embryos resemble that seen (red bracket) and posterior cardinal vein (blue bracket) can be detected in the after modulation of Notch signaling and have been attributed to control MO-injected embryo but not in the triple MO (white bracket high- direct targeting of β-catenin to the Dll4 promoter through a TCF lights single axial vessel). See also Fig. S7 A and B.(B) Lateral view (Upper) and binding site (28). β-catenin can also form an activation complex optical transversal sections (Lower, location indicated by white bar) of con- with NICD and RBPJ at a binding motif in the Dll4 third intron trol, rbpj, sox7;sox18, and sox7;sox18;rbpj MO-injected fli1a:EGFP embryos at 2 dpf. See also Movies S1, S2, S3, and S4 for 3D animations of each embryo. (here known as Dll4in3 RBPJ-a) (29). Although our data dem- (C–E) Whole-mount in situ hybridization of venous (C) and arterial (D and E) onstrate that both this motif and the promoter are dispensable markers on control, rbpj, double sox7;sox1,8 and triple sox7;sox18;rbpj MO at for arterial Dll4 expression, β-catenin is also known to interact 26 hpf. See also Fig. S8 A–D. Black arrowhead, dorsal aorta; white arrowhead, with SOXF proteins to regulate gene expression in a TCF- where dorsal aorta should be; black arrow, posterior cardinal vein; white independent manner (30). It is consequently likely that β-catenin bracket, single axial vessel; black bracket, notochord. Probes used are depic- may regulate Dll4 expression patterns in combination with, or ted in the bottom left of each picture. Values on the bottom right indicate upstream of, both RBPJ and SOXF factors. Previous studies number of embryos with the predominant and displayed phenotype per total have also proposed FOXC transcription factors as key regulators number of embryos analyzed. In all cases, 0.125 pmol of sox7 and sox18 and of arterial identity, at least in part through direct transcriptional 0.15 pmol of rbpj MO were used. activation of the Dll4 gene promoter (23). However, the two Dll4 enhancers identified here work independently of the FOXC- target as previously reported (10) whereas other markers were binding endogenous promoter and contain no functional Fork- absent only in the triple sox7, sox18, rbpj MO embryos. This total head binding motifs, suggesting that FOXC factors influence loss of arterial markers was unprecedented and suggests that arterial development independently of Dll4 transactivation dur- combinatorial loss of sox7, sox18,andrbpj expression results in ing early development.

Sacilotto et al. PNAS Early Edition | 5of6 Downloaded by guest on October 2, 2021 Our data indicate that other components of the Notch path- licensed by the UK Home Office. Analysis was performed as described pre- way, in addition to other arterial-specific genes including the viously, with care taken to ensure all embryos were consistently treated (31). VEGF receptor flt1 and the ephrin ligand efnB2a, are also Mosaic transient transgenic zebrafish embryos were generated using the downstream of the combinatorial SoxF/Notch code. Conversely, Tol2 system, and scored for GFP reporter expression at 24 hpf (32). The tg triple sox7:sox18:rbpj MO-injection had no effect on angioblast (Dll4in3:GFP) stable line was created by initial outcross of adult F0 carriers migration and midline coalescence, or pan-vascular gene ex- generated using the Tol2 system. The tg(fli1a:GFP), tg(kdrl:GFP) and tg pression. This suggests that the SoxF/Notch code is not required (gata1:dsRed) lines were described previously (26, 33, 34). for early endothelial cell specification, and appears strikingly different from the vascular regulation directed by the Ets family Morpholinos and Chemical Treatments. Antisense morpholino oligonucleo- of transcription factors. Loss of ETS factors in both mouse and tides were as described previously (4, 25). MOs were injected into 1- to 2-cell fish disrupts early endothelial cell specification, and ETS binding wild-type or tg(Dll4in3:GFP) embryos, using 0.5 nL of the corresponding MO motifs are found in all endothelial-expressed gene enhancers, solution at a concentration depicted in figure legends. For pharmacological regardless of expression pattern (17, 18). These observations inhibition of VEGF and Notch signaling pathways, embryos were manually μ μ suggest that ETS factors regulate gene expression during early dechorionated and incubated with 1 M of SU5416 (Sigma) or 100 M DAPM endothelial cell specification, other transcription factors, in- (Calbiochem), respectively, starting at 10 hpf. cluding Sox and NotchICD/Rbpj, contribute more precise spatial and temporal control of gene expression within the differenti- In Situ Hybridization and Immunohistochemistry. In situ hybridization and ating vasculature. The study of additional arterial- and venous- immunohistochemistry were performed as previously described (31, 35). specific enhancers will therefore likely reveal further information ChIP. ChIP assays were performed on confluent HUVECS as described (36) about the transcriptional cascades regulating vascular patterning. using antibodies against Sox7, Sox18, Pol II (Santa Cruz), NICD1, RbpJ, and Materials and Methods IgG (Abcam). Cloning. Dll4in3, Dll4-12, and zfishDll4in3 enhancers were generated by PCR EMSA. EMSAs were performed as described previously (31). from BAC clone RP23-46P4 (CHORI) and genomic DNA respectively. Reporter See SI Materials and Methods for additional details. vectors were generated using Gateway technology (Invitrogen). For Dll4i- n3end, Dll4in3 was cloned into pAUG-βGal upstream of the Dll4 promoter using SacII-NotI and SpeI-BamHI sites respectively. All mutated versions of ACKNOWLEDGMENTS. We thank N. Ahituv for providing GW vectors, M. Shipman for help with imaging, and J. Koth for help with caspase-3 Dll4in3 were generated by PCR (sequences in SI Materials and Methods). IHC. This work was supported by the Ludwig Institute for Cancer Research, the British Heart Foundation (BHF) (N.S. and S.D.V.), the BHF Centre of Transgenic Animals. Transgenic mice were generated by oocyte microinjec- Research Excellence, Oxford (RE/08/004) (to R.M. and S.D.V.), and the tion. All animal procedures were approved by local ethical review and Medical Research Council (MR/J007765/1) (N.S., K.L., G.B.-G., and S.D.V.).

1. Potente M, Gerhardt H, Carmeliet P (2011) Basic and therapeutic aspects of angio- 19. Heintzman ND, et al. (2009) Histone modifications at human enhancers reflect global genesis. Cell 146(6):873–887. cell-type-specific gene expression. Nature 459(7243):108–112. 2. Schepers GE, Teasdale RD, Koopman P (2002) Twenty pairs of sox: Extent, homology, 20. Benedito R, Duarte A (2005) Expression of Dll4 during mouse embryogenesis suggests and nomenclature of the mouse and human sox transcription factor gene families. multiple developmental roles. Gene Expr Patterns 5(6):750–755. Dev Cell 3(2):167–170. 21. Birnbaum RY, et al. (2012) Functional characterization of tissue-specific enhancers in 3. Mertin S, McDowall SG, Harley VR (1999) The DNA-binding specificity of SOX9 and the DLX5/6 locus. Hum Mol Genet 21(22):4930–4938. other SOX proteins. Nucleic Acids Res 27(5):1359–1364. 22. Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: Improving the sensitivity of 4. Cermenati S, et al. (2008) Sox18 and Sox7 play redundant roles in vascular de- progressive multiple sequence alignment through sequence weighting, position- fi – velopment. Blood 111(5):2657–2666. speci c gap penalties and weight matrix choice. Nucleic Acids Res 22(22):4673 4680. 5. Herpers R, van de Kamp E, Duckers HJ, Schulte-Merker S (2008) Redundant roles for 23. Seo S, et al. (2006) The forkhead transcription factors, Foxc1 and Foxc2, are required fi sox7 and sox18 in arteriovenous specification in zebrafish. Circ Res 102(1):12–15. for arterial speci cation and lymphatic sprouting during vascular development. Dev – 6. Pendeville H, et al. (2008) Zebrafish Sox7 and Sox18 function together to control Biol 294(2):458 470. arterial-venous identity. Dev Biol 317(2):405–416. 24. Frankel N (2012) Multiple layers of complexity in cis-regulatory regions of de- – 7. Sakamoto Y, et al. (2007) Redundant roles of Sox17 and Sox18 in early cardiovas- velopmental genes. Dev Dyn 241(12):1857 1866. cular development of mouse embryos. Biochem Biophys Res Commun 360(3): 25. Siekmann AF, Lawson ND (2007) Notch signalling limits angiogenic cell behaviour in developing zebrafish arteries. Nature 445(7129):781–784. 539–544. 26. Lawson ND, Weinstein BM (2002) In vivo imaging of embryonic vascular development 8. Francois M, Koopman P, Beltrame M (2010) SoxF genes: Key players in the de- using transgenic zebrafish. Dev Biol 248(2):307–318. velopment of the cardio-vascular system. Int J Biochem Cell Biol 42(3):445–448. 27. Herbert SP, et al. (2009) Arterial-venous segregation by selective cell sprouting: An 9. Engert S, Liao WP, Burtscher I, Lickert H (2009) Sox17-2A-iCre: A knock-in mouse line alternative mode of blood vessel formation. Science 326(5950):294–298. expressing Cre recombinase in endoderm and vascular endothelial cells. Genesis 47(9): 28. Corada M, et al. (2010) The Wnt/beta-catenin pathway modulates vascular re- 603–610. modeling and specification by upregulating Dll4/Notch signaling. Dev Cell 18(6): 10. Lawson ND, et al. (2001) Notch signaling is required for arterial-venous differentia- 938–949. tion during embryonic vascular development. Development 128(19):3675–3683. 29. Yamamizu K, et al. (2010) Convergence of Notch and beta-catenin signaling induces 11. Gridley T (2007) Notch signaling in vascular development and physiology. De- arterial fate in vascular progenitors. J Cell Biol 189(2):325–338. velopment 134(15):2709–2718. 30. Sinner D, Rankin S, Lee M, Zorn AM (2004) Sox17 and beta-catenin cooperate to 12. Krebs LT, et al. (2004) Haploinsufficient lethality and formation of arteriovenous regulate the transcription of endodermal genes. Development 131(13):3069–3080. – malformations in Notch pathway mutants. Genes Dev 18(20):2469 2473. 31. De Val S, et al. (2004) Mef2c is activated directly by Ets transcription factors through 13. Duarte A, et al. (2004) Dosage-sensitive requirement for mouse Dll4 in artery de- an evolutionarily conserved endothelial cell-specific enhancer. Dev Biol 275(2): – velopment. Genes Dev 18(20):2474 2478. 424–434. fi 14. Gale NW, et al. (2004) Haploinsuf ciency of delta-like 4 ligand results in embryonic 32. Kawakami K (2005) Transposon tools and methods in zebrafish. Dev Dyn 234(2): lethality due to major defects in arterial and vascular development. Proc Natl Acad Sci 244–254. – USA 101(45):15949 15954. 33. Jin SW, Beis D, Mitchell T, Chen JN, Stainier DY (2005) Cellular and molecular 15. Lawson ND, Vogel AM, Weinstein BM (2002) Sonic hedgehog and vascular endo- analyses of vascular tube and lumen formation in zebrafish. Development 132(23): thelial growth factor act upstream of the Notch pathway during arterial endothelial 5199–5209. differentiation. Dev Cell 3(1):127–136. 34. Traver D, et al. (2003) Transplantation and in vivo imaging of multilineage engraft- 16. Maston GA, Evans SK, Green MR (2006) Transcriptional regulatory elements in the ment in zebrafish bloodless mutants. Nat Immunol 4(12):1238–1246. human genome. Annu Rev Genomics Hum Genet 7:29–59. 35. Gering M, Patient R (2005) Hedgehog signaling is required for adult blood stem cell 17. Lee D, et al. (2008) ER71 acts downstream of BMP, Notch, and Wnt signaling in blood formation in zebrafish embryos. Dev Cell 8(3):389–400. and vessel progenitor specification. Cell Stem Cell 2(5):497–507. 36. Sacilotto N, Espert A, Castillo J, Franco L, López-Rodas G (2011) Epigenetic tran- 18. De Val S, Black BL (2009) Transcriptional control of endothelial cell development. Dev scriptional regulation of the growth arrest-specific gene 1 (Gas1) in hepatic cell Cell 16(2):180–195. proliferation at mononucleosomal resolution. PLoS ONE 6(8):e23318.

6of6 | www.pnas.org/cgi/doi/10.1073/pnas.1300805110 Sacilotto et al. Downloaded by guest on October 2, 2021