Histone acetyltransferase 7 (KAT7)-dependent intragenic histone acetylation regulates endothelial cell gene regulation

Although the functional role of chromatin marks at promoters in mediating cell-restricted gene expression has been well characterized, the role of intragenic chromatin marks is not well understood, especially in endothelial cell (EC) gene expression. Here, we characterized the histone H3 and H4 acetylation profiles of 19 genes with EC-enriched expression via locus-wide chromatin immunoprecipitation followed by ultra-high–resolution (5 bp) tiling array analysis in ECs versus non-ECs throughout their genomic loci. Importantly, these genes exhibit differential EC enrichment of H3 and H4 acetylation in their promoter in ECs versus non-ECs. Interestingly, VEGFR-2 and VEGFR-1 show EC-enriched acetylation across broad intragenic regions and are up-regulated in non-ECs by histone deacetylase inhibition. It is unclear which histone acetyltransferases (KATs) are key to EC physiology. Depletion of KAT7 reduced VEGFR-2 expression and disrupted angiogenic potential. Microarray analysis of KAT7-depleted ECs identified 263 differentially regulated genes, many of which are key for growth and angiogenic potential. KAT7 inhibition in zebrafish embryos disrupted vessel formation and caused loss of circulatory integrity, especially hemorrhage, all of which were rescued with human KAT7. Notably, perturbed EC-enriched gene expression, especially the VEGFR-2 homologs, contributed to these vascular defects. Mechanistically, KAT7 participates in VEGFR-2 transcription by mediating RNA polymerase II binding, H3 lysine 14, and H4 acetylation in its intragenic region. Collectively, our findings support the importance of differential histone acetylation at both promoter and intragenic regions of EC genes and reveal a previously underappreciated role of KAT7 and intragenic histone acetylation in regulating VEGFR-2 and endothelial function.

Histone acetylation is a chromatin modification involved in transcriptional activation, and it is dynamically regulated by histone acetyltransferases (KATs) 6 and histone deacetylases (HDACs) (1). Although promoter histone acetylation is commonly associated with transcriptional activation, intragenic histone acetylation also plays a pivotal regulatory role (1). In yeast cells lacking the KATs gcn5 and elp3, histone H3 hypoacetylation in the intragenic region is associated with transcriptional inhibition (2). Gcn5 deficiency causes transcriptional inhibition by preventing nucleosome eviction and transcription elongation (3). Also, intragenic histone acetylation levels are critical for cotranscriptional spliceosome assembly (4). However, the importance of intragenic histone acetylation is not well characterized in mammals.
To further our understanding on the epigenetic regulation of EC-enriched genes (EC genes), especially locus-wide chromatin marks, we characterized the pan-AcH3 and -AcH4 profiles of HUVECs and a non-EC, HuAoVSMCs, at 19 EC genes via locus-wide chromatin immunoprecipitation followed by ultrahigh-resolution (5 bp) tiling array analysis (ChIP-chip). Almost all of the EC genes were differentially H3-and H4-acetylated, and these profiles were highly correlated with their expression and RNA polymerase II (pol II) ChIP-seq profiles. Similar findings were evident with low resolution ChIP-seq AcH3K9 and AcH3K27 profiles of HUVECs and normal human epithelial keratinocytes (NHEK) from the Encyclopedia Of DNA Elements (ENCODE) database (9). Interestingly, some notable genes, including vascular endothelial growth factor receptor-1 and -2 (VEGFR-1 and VEGFR-2), exhibited broad differential intragenic histone acetylation. Consistent with this observation, HDAC inhibition with trichostatin A (TSA) in HuAoVSMCs increased VEGFR-1 and VEGFR-2 expression.
We found that KAT7 localized to the VEGFR-2 locus, and in vitro KAT7 depletion in human ECs perturbed EC gene expression, including VEGFR-2. The reduced VEGFR-2 expression level in KAT7-depleted EC was mediated transcriptionally by reducing AcH4, AcH3K14, and pol II occupancy in the VEGFR-2 intragenic region. Perturbed EC gene expression in KAT7-depleted ECs was associated with deficient EC function. Parallel findings were made in vivo. Importantly, KAT7 morpholino (MO) inhibition in zebrafish embryos (3 dpf) resulted in aberrant vessel formation with compromised circulatory integrity, especially hemorrhage, that could be rescued by human KAT7 RNA. Furthermore, these zebrafish embryos showed reduced expression of EC-enriched genes that were similarly perturbed in human KAT7-depleted ECs. Expression of a KAT7-regulated gene, the VEGFR-2 homolog kdrl, in KAT7-depleted zebrafish salvaged aberrant vessel formation and partly restored the disrupted circulatory integrity in these zebrafish. Overall, our findings suggest that histone acetylation is important for EC-enriched gene expression and that KAT7 plays a significant role in endothelial function, in part, by regulating intragenic histone acetylation at VEGFR-2.

Differential histone acetylation occurs at genes with endothelial cell-enriched expression
In our previous studies, we observed that differential histone acetylation at eNOS is functionally important for its EC-enriched expression (6). To determine whether this finding extends to other genes with preferential EC expression, we designed an ultra-high-resolution tiling DNA microarray at a 5-bp resolution for 34 genes comprising the following: (i) 19 EC genes; (ii) six broadly expressed genes; and (iii) nine EC-excluded genes (Table S1). These genes were tiled by an average of 24 distinct DNA probes per bp. Both DNA strands were examined. Although prior studies have focused on promoters, intragenic modifications are gaining attention. Therefore, we were also interested in intragenic modifications across cell types for EC genes. Thus, our tiling arrays probed the 34 genes at the genomic regions that are upstream of transcription initiation (50 kb), intragenic regions, and downstream of the last transcript exon (50 kb). Using the tiling arrays, we assessed for differential pan-AcH3 (AcH3K9 and AcH3K14) and pan-AcH4 (AcH4K5, AcH4K8, AcH4K12, and AcH4K16) enrichment at these genes in HUVEC and HuAoVSMC via ChIP-chip analysis. Of the 19 EC genes, 15 were preferentially acetylated at H3, H4, or both in ECs with 12 showing EC-enriched differential histone acetylation at their promoters (Figs. 1A, 2A, 3A, and 4A, B, D, and E and Table S1). Importantly, we recapitulated our previous findings at eNOS and found differential H3 and H4 acetylation in its proximal promoter region in ECs versus non-ECs ( Fig. 3A) (6). For the 15 preferentially histone-acetylated EC genes (e.g. eNOS and VEGFR-2), their mRNA expression in ECs versus HuAoVSMCs was Ն3-fold, and their HuAoVSMC basal expression was low (19). The four non-preferentially histone-acetylated EC genes showed only modestly higher mRNA expression in ECs (e.g. TFPI and EphB4), because they also evidenced modest HuAoVSMC basal expression. Thus, genes exhibited differential histone acetylation when their mRNAs were robustly and selectively expressed in ECs versus non-ECs. Focal differential histone acetylation occurred in the intragenic regions of many EC genes (e.g. eNOS). Interestingly, some genes, notably KDR (VEGFR-2) and FLT1 (VEGFR-1), showed broad differential histone acetylation across their transcriptional units (Figs. 1A, 2A, and 4 and Table S1). This was not expected. We were able to validate differential histone acetylation at VEGFR-2 by ChIP-qPCR (Fig. 4, A and B). Importantly, normalization for histone H3 density, which addresses nucleosomal density, did not affect this conclusion (Fig. 4, C-E). Differential histone acetylation at EC genes is not due to greater global histone acetylation levels in EC, as noted previously (6). Furthermore, the broadly expressed PPIA (cyclophilin A) showed similar high histone acetylation levels in both cell types (Fig. 5A). Also, the non-expressed CDH1 (E-cadherin) showed sparse histone acetylation levels in the two cell types (Fig. 5B). Taken together, genes with enriched expression are preferentially histone acetylated in ECs versus non-expressing cells.

Comparison of histone acetylation ChIP-chip data with ChIPseq data from the ENCODE consortium
Whole-genome mapping of AcH3K9, AcH3K27, H3K36me3, and pol II has previously been conducted on HUVECs and NHEKs by the ENCODE consortium and represents a comprehensive, yet relatively low resolution ChIP-seq dataset (9). These data allowed us to compare our high-resolution pan-AcH3 to AcH3K9 and AcH3K27 profiles to gain further insight into the histone acetylation status of the 19 EC genes. Regions of differential AcH3K9 and

Role of KAT7 in EC gene expression
AcH3K27 enrichment were noted for all 19 EC genes in ECs versus NHEKs (Fig. 1B, 2B, and 3B and Table S1). Importantly, our pan-AcH3 data mirrored the AcH3K9 and AcH3K27 datasets (Figs. 1-3 and Table S2). However, regions of differential pan-AcH3, but not AcH3K9 and AcH3K27, enrichment were detected, which might reflect AcH3K14 enrichment (Figs. 1-3 and Table S2). Regions that were exclusively differentially AcH3K9-and/or AcH3K27enriched were also observed (Figs. 1-3 and Tables S1 and S2). Aside from SELP and VCAM1, which exhibited low basal expression, EC genes were differentially H3K36me3 enriched in their intragenic region in ECs versus NHEKs with 15 exhibiting widespread differential enrichment (Fig. 1B, 2B, and 3B and Table S1). This observation is in contrast to select EC genes that are differentially acetylated throughout their loci.
pol II ChIP-seq profiles in HUVECs and NHEK showed that pol II selectively engaged EC genes in ECs versus NHEKs (Figs.  1B, 2B, and 3B and Table S1). We confirmed the pol II enrichment across the VEGFR-2 locus by ChIP-qPCR (Fig. 4F). Importantly, pol II enrichment overlaps the widespread regions of differential pan-AcH3 and pan-AcH4 enrichment in VEGFR-2 ( Fig. 1 and Table S1). Aside from SELP, which exhibits low basal EC mRNA expression, EC genes with differential pan-AcH3 and -AcH4 profiles were also enriched for pol II in ECs (19). This observation suggests that differential pan-AcH3 and -AcH4 enrichment is associated with transcription at these genes (Figs. 1-3 and Table S1).

Effects of TSA treatment on VEGFR-1 and VEGFR-2
In our previous studies, we noted that differential promoter DNA methylation was not observed at VEGFR-1 and VEGFR-2 between ECs and non-ECs suggesting that it did not regulate their EC-enriched expression (19). Indeed, DNA methyltransferase (DNMT) inhibitors did not affect VEGFR-1 and VEGFR-2 expression. This is in contrast to EC genes, such as eNOS, CD31, vWF, and CD144 (VE-cadherin), that exhibited differential proximal promoter DNA methylation across cell types and showed increased mRNA expression upon DNMT inhibition in non-expressing cell types where their promoters are normally methylated (19). Because broad differential histone acetylation occurred at VEGFR-1 and VEGFR-2, we assessed the role of histone acetylation on their expression by treating HUVECs and HuAoVSMCs acutely (4 h) with trichostatin A (TSA), a class I and II HDAC inhibitor. Both genes were up-regulated by TSA in HuAoVSMCs, a non-EC that does not basally express either gene (Fig. 6A). In contrast, high basal VEGFR-1 and VEGFR-2 expression did not change significantly in HUVECs (Fig. 6B). Taken together, histone acetylation plays a functional role in the EC-enriched expression of VEGFR-1 and VEGFR-2. A, histone H3 (pan-AcH3; AcH3K9 and AcH3K14) and H4 (pan-AcH4; AcH4K5, AcH4K8, AcH4K12, and AcH4K16) acetylation profiles of KDR/VEGFR-2 in HUVEC and HuAoVSMC. The histone acetylation level at each genomic location is defined as the fold difference between the pan-AcH3/pan-AcH4 ChIP (Cy5) and the control rIgG ChIP (Cy3) hybridization signals. Only fold differences between pan-AcH3/pan-AcH4 ChIP and isotype control ChIP that are Ն2-fold are shown. Biological replicates of pan-AcH3 (n ϭ 2) and pan-AcH4 (n ϭ 3) ChIP-chip analysis showed similar results. B, AcH3K9, AcH3K27, H3K36me3, and RNA pol II ChIP-seq profiles of KDR/VEGFR-2 in HUVECs and NHEK. The HUVEC and NHEK-associated ChIP-seq profiles were obtained from the ENCODE database (9). The enrichment levels of AcH3K9, AcH3K27, H3K36me3, and pol II are represented by normalized intensities.

Role of KAT7 in EC gene expression KAT7 depletion disrupts the expression of VEGFR-2 and other genes involved in EC physiology
In view of the fact that differential histone acetylation occurred broadly across VEGFR-2 and HDAC inhibition in HuAoVSMC up-regulated VEGFR-2, we pursued the hypothesis that broad histone acetylation in the VEGFR-2 locus is important for its EC expression. Because KAT7-containing MYST complexes can be recruited broadly across target genes, KAT7 is a candidate for regulating broad histone acetylation at VEGFR-2 (13). Thus, we assessed the expression and role of KAT7 in ECs. We found KAT7 was expressed ubiquitously in various primary cell types and was unaffected by environmental stimuli, such as hypoxia (Ͻ1% O 2 ; Fig. 7A and Fig. S1). To assess whether KAT7 affects VEGFR-2 EC expression, we depleted KAT7 by RNAi and found it reduced VEGFR-2 mRNA and protein expression by 36% (Ϯ11% S.E., n ϭ 4, p Յ 0.05) and 65% (Ϯ9% S.E., n ϭ 4, p Յ 0.05), respectively (Fig. 7, B, C, and E). In contrast, TIE1, TIE2, and eNOS expressions were not significantly altered (Fig. 7, B and E). To determine whether KAT7 regulated other genes involved in EC physiology, gene expression analysis was conducted. Of the 26,109 analyzed protein-coding transcripts, 263 were differentially regulated with 117 down-regulated and 146 up-regulated by KAT7 depletion (p Յ 0.05, Ն2-fold change, n ϭ 4) ( Fig. 7C; Table  S3). However, we were unable to find an effect of KAT7 depletion on EC expression of VEGF-A itself. Interestingly, gene ontology (GO) analysis of mRNA species that are down-regulated by KAT7 depletion are enriched for biological processes associated with EC physiology, including inflammatory response, vasodilation, response to wounding, regulation of blood vessel size, regulation of tube size, and positive up-regulation of cholesterol efflux (Fig. 7D). In contrast, GO analysis of mRNA species up-regulated by KAT7 is enriched for genes associated with cell cycle and mitosis ( Fig.  S2A and Table S3). Using independent HUVEC isolates and quantitative RT-qPCR, we confirmed the regulation of 11 select genes with roles in vessel tone regulation (ACE1 and ASS1), vessel formation (CASZ1, CEACAM-1, CXCL16, RGS4, and VEGFR-1-3), EC activation (E-selectin), and cholesterol efflux (ABCG1) (Fig. 7E) (20 -28). Thus, these in vitro studies suggest that KAT7 is important in EC gene expression. A, histone H3 (pan-AcH3; AcH3K9 and AcH3K14) and H4 (pan-AcH4; AcH4K5, AcH4K8, AcH4K12, and AcH4K16) acetylation profiles of FLT1/VEGFR-1 in HUVEC and HuAoVSMC. The histone acetylation level at each genomic location is defined as the fold difference between the pan-AcH3/pan-AcH4 ChIP (Cy5) and the control rIgG ChIP (Cy3) hybridization signals. Only fold differences between pan-AcH3/pan-AcH4 ChIP and isotype control ChIP that are Ն2-fold are shown. Biological replicates of pan-AcH3 (n ϭ 2) and pan-AcH4 (n ϭ 3) ChIP-chip analysis showed similar results. B, AcH3K9, AcH3K27, H3K36me3, and RNA pol II ChIP-seq profiles of FLT1/VEGFR-1 in HUVECs and NHEK. The HUVEC and NHEK-associated ChIP-seq profiles were obtained from the ENCODE database (9). The enrichment levels of AcH3K9, AcH3K27, H3K36me3, and pol II are represented by normalized intensities.

Role of KAT7 in EC gene expression KAT7 mediates chromatin-based transcriptional regulation of VEGFR-2
To determine whether KAT7 regulated VEGFR-2 expression transcriptionally in human ECs, we assessed pol II occupancy throughout the VEGFR-2 locus (Fig. 8A). Aside from a genomic region at the 3Ј end of the VEGFR-2 genomic locus (ϩ28760/ ϩ28706), pol II occupancy at the assessed genomic regions throughout the locus was reduced in KAT7-depleted ECs (Fig.  8B). We then examined the AcH3 and AcH4 residues that are EC-enriched at the VEGFR-2 locus to determine whether they correspond to KAT7 catalytic activity. Indeed, AcH3K9, AcH3K14, and AcH4K8 were found to be differentially ECenriched across the VEGFR-2 locus in HUVEC versus HuAoVSMC (Fig. S3). To determine whether KAT7 directly regulated the chromatin structure of the VEGFR-2 locus, we assessed the histone acetylation status at VEGFR-2 upon KAT7 depletion. Consistent with KAT7 catalytic activity, pan-AcH4 and AcH3K14 levels were reduced at many, particularly central, VEGFR-2 intragenic regions that were assessed in KAT7-depleted ECs (Fig. 8, C and D). To determine whether KAT7 can localize to the VEGFR-2 locus, ECs were transduced with KAT7-expressing lentiviruses, and KAT7 enrichment at the VEGFR-2 genomic locus was assessed by ChIP-qPCR analysis ( Fig. 9). KAT7 showed enriched localization at the VEGFR-2 genomic locus in ECs transduced with KAT7-expressing lentiviruses versus ECs transduced with His-eGFP-expressing lentiviruses (Fig. 9B). In contrast, KAT7 was not enriched at the eNOS promoter in ECs that are transduced with KAT7 and His-eGFP-expressing lentiviruses (Fig. 9B). Taken together, KAT7 affects VEGFR-2 transcriptional activity, in part, by localizing to the VEGFR-2 genomic locus and regulating intragenic histone acetylation.

KAT7 regulates EC phenotype and survival
To determine whether KAT7 regulates the EC phenotype, we conducted broadly accepted assays of EC phenotypes, namely matrigel and wound-healing assays. We noted fewer branch points, a decreased number of branches in EC networks, and a greater portion of branches outside of a network in matrigel assays of KAT7-depleted ECs (Fig. 10A). Furthermore, KAT7 depletion reduced the surface area recovered in wound-healing assays (Fig. 10B). . Histone modification and RNA polymerase II profiles of NOS3/eNOS. A, representative pan-AcH3 (pan-AcH3; AcH3K9 and AcH3K14) and -H4 (pan-AcH4; AcH4K5, AcH4K8, AcH4K12, and AcH4K16) acetylation profiles throughout (A) NOS3 in HUVEC and HuAoVSMC. The histone acetylation level at each genomic location is defined as the fold difference between the pan-AcH3/pan-AcH4 ChIP (Cy5) and the isotype control ChIP (Cy3) hybridization signals. Only fold differences between pan-AcH3/pan-AcH4 ChIP and isotype control ChIP Ն2-fold are shown. Biological replicates of pan-AcH3 (n ϭ 2) and pan-AcH4 (n ϭ 3) ChIP-chip analysis showed similar results. B, ChIP-seq data profiles of AcH3K9, AcH3K27, H3K36me3, and pol II throughout NOS3 in HUVEC and NHEK. The HUVEC and NHEK-associated ChIP-seq profiles were obtained from the ENCODE database (9). The enrichment levels of AcH3K9, AcH3K27, H3K36me3, and pol II are represented by normalized intensities.

Role of KAT7 in EC gene expression
In view of the fact that a major target of KAT7 in EC is the VEGF axis, we determined whether KAT7 depletion perturbs the VEGF-A response of ECs. Therefore, we assayed cell migra-tion and signal transduction activity of ECs in response to VEGF-A. We found that VEGF-A significantly induced cell migration activity of control siRNA-treated ECs in a chemoat-  A and B, representative pan-AcH3 (pan-AcH3; AcH3K9 and AcH3K14) and -H4 (pan-AcH4; AcH4K5, AcH4K8, AcH4K12, and AcH4K16) acetylation profiles throughout PPIA (A) and CDH1 (B) in HUVEC and HuAoVSMC. The histone acetylation level at each genomic location is defined as the fold difference between the pan-AcH3/pan-AcH4 ChIP (Cy5) and the isotype control ChIP (Cy3) hybridization signals. Only fold differences between pan-AcH3/pan-AcH4 ChIP and isotype control ChIP Ն2-fold are shown. Biological replicates of pan-AcH3 (n ϭ 2) and pan-AcH4 (n ϭ 3) ChIP-chip analysis showed similar results.

Role of KAT7 in EC gene expression
tractant assay (Fig. 10C). In contrast, KAT7 depletion blunted EC migration activity in response to VEGF-A (Fig. 10C). We also observed an appreciable increase in cell migration activity in KAT7-depleted ECs relative to control siRNA-treated ECs in the absence of VEGF-A (Fig. 10C). This increase in the cell migration activity of KAT7-depleted ECs might be due to the effects of KAT7 on other regulators of cell migration basally. However, this difference was not statistically significant. Furthermore, as noted earlier, KAT7 depletion did not affect basal expression of VEGF-A in ECs. With respect to signal transduction activity of ECs in response to VEGF-A, we found that VEGFR-2-dependent signal transduction activity is disrupted in KAT7-depleted ECs as shown by reduced VEGFR-2 tyrosine 1175 (Tyr-1175) phosphorylation at 5 min in response to VEGF-A induction, even after taking account of reduced VEGFR-2 protein expression (Fig. 10D). The physiology of VEGFR-2 Y1175 phosphorylation in response to VEGF-A administration has been implicated in VEGF-A induced increase in cell migration and proliferation (29,30). The blunted VEGF-induced migratory response is consistent with the change in biochemical signaling in KAT7-depleted ECs.
KAT7 can also affect cell proliferation and survival in other cell types, perhaps due to effects on DNA replication (11,31). We also found that KAT7 regulated genes involved in cell cycle, cell division, and DNA replication ( Fig. S2A and Table S3). Thus, we conducted cell proliferation and apoptosis assays on KAT7-depleted ECs. We found reduced cell proliferation but no changes in apoptosis (Fig. S2, B and C). Thus, these in vitro studies suggest that KAT7 is important in EC cellular phenotype and highlights its specific role in VEGFR-2 regulation and VEGF-A response.

Kat7 inhibition results in defective vascular structure and circulation in zebrafish
Previous studies showed that kat7-deficient mouse embryos were developmentally arrested at the 10-somite stage due to defects in post-gastrulation mammalian development, which disrupted somites, mesenchyme, and possibly blood vessel formation (12). However, functional defects in the vasculature were not addressed, and the abnormal vascular structure might be confounded by embryonic defects that affect developmental progression (12). Thus, we characterized kat7 function in the zebrafish vasculature, in which structure and function can be monitored in real time and a functioning cardiovascular system is not required for viability at relatively advanced stages of embryogenesis. We deduced that zebrafish have two predicted kat7 orthologs, namely kat7a and kat7b (Zebrafish Genome Build GRCz10), which encode orthologs with 13% and 86% amino acid sequence identity to human KAT7, respectively. We focused on kat7b, which is predicted to have the conserved zinc finger and MYST domains (13). To determine the expression profile of kat7b in zebrafish embryos, tg(kdrl:eGFP) zebrafish embryos were sorted for EC-enriched GFP ϩ and EC-depleted GFP Ϫ cells. Consistent with the ubiquitous expression of human KAT7 in multiple cell types, kat7b expression is expressed in both GFP ϩ and GFP Ϫ cells (Fig. 11A). This observation is consistent with the non-tissue-specific expression of kat7b that was noted previously by others in zebrafish embryos between 0 and 60 h (32). To characterize zebrafish kat7 function, we depleted kat7b in tg(fli1:nGFP) zebrafish using morpholinos (MO) and assessed initial intersegmental vessel (ISV) formation (24 -34 h post-fertilization), an early stage of vascular development (Fig. 11, B-D, and Fig. S4, A and B). Kat7 depletion reduced EC numbers in zebrafish ISV throughout early ISV formation, suggesting the process is impeded (Fig. 11, C and D). We further characterized kat7 function by depleting kat7 using independent morpholinos (E1ATG MO and E3I3 MO) in tg(kdrl:eGFP, gata1:dsRED) zebrafish embryos to simultaneously visualize EC vascular patterning and circulating erythrocytes at 3 days post-fertilization (dpf) (Fig. 12). Kat7depleted zebrafish exhibited abnormal vessel formation as shown by the disrupted vascular plexus of the subintestinal vein (SIV) (Fig. 12, A and B). Although overall vascular architecture appeared grossly intact, the circulatory integrity of kat7-depleted zebrafish was compromised. Specifically, we observed reduced blood circulation in the ISVs, dorsal aorta, caudal vein, and SIV; and blood accumulated in the yolk sac and trunk (Fig. 12, A and C). Hemorrhage was also prominent in the brain (Fig. 12, A and D). Importantly, these phenotypes were rescued with human KAT7 RNA, which is insensitive to the E1ATG MO (Fig. 12). Perturbations in the closed cardiovascular system might reflect blood vessel underdevelopment as suggested by the decreased ISV EC numbers in kat7-depleted tg(fli1:nGFP) zebrafish (3 dpf) (Fig. 13). Our assessments of time-based imaging indicated that the decreased ISV EC numbers of kat7-depleted embryos did not reflect augmented migration in and out of the vessels (Fig. 11, C and D), as described for other EC genes (33). Taken together, kat7 is important for forming a functional closed cardiovascular network.

Role of KAT7 in EC gene expression Kat7 regulation of EC gene expression is critical for forming a functional cardiovascular network in zebrafish
To determine whether kat7 affects EC gene expression in zebrafish development, MO-treated tg(kdrl:eGFP) zebrafish (3 dpf) were subjected to fluorescence-activated cell-sorting (FACS) to isolate GFP ϩ and GFP Ϫ cells that are EC-enriched and EC-depleted, respectively. We noted robust expression of the VE-cadherin (cdh5), VEGFR1 (flt1), and VEGFR-2 (kdr)

Role of KAT7 in EC gene expression
orthologs and the VEGFR-2 paralog (kdrl) in GFP ϩ versus GFP Ϫ cells (Fig. 14, A and C-E) (34). Importantly, kdr, kdrl, and flt1 mRNA expression was reduced in KAT7-depleted GFP ϩ versus control GFP ϩ cells (Fig. 14, C-E). Consistent with these observations, kdrl and flt1 protein expression was reduced in kat7-depleted embryos relative to control MO-treated embryos (Fig. 14F). In contrast, cdh5, which is EC-enriched, and nos1, which in zebrafish is not EC-enriched, were not affected by kat7 depletion (Fig. 14, A and B). Thus, as with human ECs, kat7 is important for zebrafish EC-enriched gene expression.
Because kat7 regulates EC gene expression in zebrafish embryos, we assessed whether the abnormal phenotypes of kat7-depleted zebrafish can be rescued by the exogenous expression of a target gene, namely kdrl. kdrl RNA was able to restore the disrupted SIV of kat7-depleted zebrafish (Fig. 15, A  and B). Moreover, the reduced blood circulation was restored, in part, with exogenous kdrl mRNA in kat7-depleted zebrafish as shown by similar levels of blood circulation in the ISVs, dorsal aorta, caudal vein, and SIV as control MO-treated zebrafish (Fig. 15, A and C). However, hemorrhages were prominently observed in the brain of kat7-depleted zebrafish even with the exogenous expression of kdrl suggesting that the disrupted expression or activity of other kat7-regulated genes might be responsible for this phenotype (Fig. 15, A and D). Taken together, kat7 regulation of EC gene expression is important for a functional cardiovascular network in zebrafish.

Discussion
The importance of intragenic epigenetic modifications is beginning to be appreciated. Intragenic histone modifications and DNA methylation are thought to mediate gene expression by affecting transcriptional processivity and alternative mRNA transcript expression (1). However, their role in cell-enriched gene expression has received modest attention, in part, because epigenomic studies need to be performed in a cell-by-cell-type fashion. Here, we used an ultra-high-resolution and highthroughput approach to analyze the pan-AcH3 and pan-AcH4 profiles in the genomic loci of 19 EC genes in HUVEC versus HuAoVSMC and compared these findings to widely expressed genes and EC-excluded genes. We also integrated these data with ENCODE cell-specific ChIP-seq data for AcH3K9, AcH3K27, H3K36me3, and pol II (9). Many transcriptional regulation studies have focused on chromatin modifications at promoters and enhancers (1,9). Consistent with these studies and our eNOS studies, we report that preferential histone acetylation enrichment at EC gene promoters in expressing ECs versus non-ECs is a general feature (6). Importantly, we argue that differential intragenic histone acetylation is critical

Role of KAT7 in EC gene expression
to EC gene regulation. We noted unexpected broad EC-enriched histone acetylation in the intragenic regions, especially at VEGFR-1 and VEGFR-2, that was not noted at other ECenriched genes (e.g. eNOS and CD31). Notably, VEGFR-1 and VEGFR-2 expression increased in non-expressing HuAoVSMC following acute HDAC inhibition. We further implicated KAT7 as being mechanistically and functionally important in maintaining VEGFR-1 and VEGFR-2 expression in human ECs and in vivo in zebrafish ECs.
Broad enrichment of acetylated histone residues in the intragenic regions has been noted before, but its functional contribution to cell-specific gene expression has not been studied (12,35). Intragenic histone acetylation can regulate gene expression via the co-transcriptional assembly of the spliceosome and transcription elongation (2,3,36,37). In particular, intragenic histone acetylation can facilitate transcription elongation by promoting histone eviction and recruiting bromodomain-containing proteins (3,37). We found that KAT7 can localize to the VEGFR-2 genomic locus to mediate broad differential AcH3K14 and pan-AcH4 enrichment, especially in the central intragenic genomic region. These KAT7-mediated modifications possibly promote transcription elongation based on the overall reduced broad pol II occupancy in KAT7-depleted ECs (3,37). To our knowledge, we are the first to demonstrate that broad differential intragenic histone acetylation occurs at unique EC genes and that this intragenic mark is functionally important for EC-enriched expression. Our findings can be compared with broad intragenic H3K36me3, which reflects active transcription across a transcriptional unit. Here, the H3K36me3 modification is a generalized component of gene transcription, instead of a feature of specific genes (1). The one nuance would be the intragenic signature of both H3K27me3 and H3K36me3, which is thought to reflect chromatin marks at inactive and active alleles, respectively, of monoallelically expressed autosomal genes (38).
VEGFR-2 and VEGFR-1 might specifically be regulated by KAT7-mediated broad intragenic histone acetylation as they are important in establishing the endothelial and hematopoietic lineages during early embryonic development. Previous studies have suggested that a subset of developmental genes is not completely silent in embryonic stem cells (ESCs), and their modest expression is thought to give ESCs the potential to differentiate into various lineages (39). These genes show the bivalent H3K4me3 and H3K27me3 signature at their promoter in murine ESCs (39,40). This suggests that they are transcriptionally poised in murine ESCs. This raises the possibility that their robust expression may be regulated at the transcription elongation stage upon lineage commitment (39). Because both VEGFR-1 and VEGFR-2 show a bivalent H3K4me3 and H3K27me3 signature at their promoters in human ESCs, their expression might be regulated in a similar manner (9). Thus, KAT7 might catalyze broad intragenic histone acetylation of VEGFR-2 in ECs to facilitate efficient RNA polymerase II transcription elongation to establish its robust expression (Fig. S5). Consistent with the role of KAT7 in contributing to the robust expression of target genes, we observed reduced, but appreciable, VEGFR-1 and VEGFR-2 expression even after KAT7 depletion. The loss of KAT7 therefore results in RNA polymerase II pausing and, ultimately, transcription termination. Indeed, we

Role of KAT7 in EC gene expression
noted overall reduced RNA polymerase II occupancy across the VEGFR-2 genomic locus after KAT7 knockdown (Fig. S5) (41).
Similar to VEGFR-2, the expression of other KAT7 target genes in ECs might also be regulated at the transcription elongation level. Additionally, KAT7-regulated genes in ECs could be the common targets of transcription factors. Indeed, it is suggested that KAT7 can be recruited by trans factors, including NF-B and the androgen receptor, to regulate transcription (42,43). Alternatively, KAT7 may indirectly regulate the expression of some genes in ECs by acting on other upstream regulators.
Regarding our key finding of broad differential intragenic histone acetylation at select EC genes, it is likely not due to higher nucleosome density. In VEGFR-2, differential AcH3 and AcH4 enrichment is still observed after normalizing for histone H3 density. Instead, it is likely dependent on the competing

Role of KAT7 in EC gene expression
activity of KAT and HDAC that can localize to intragenic regions (3,10). Indeed, we showed that KAT7 could localize to VEGFR-2, but not eNOS, to mediate broad differential AcH3K14 and pan-AcH4 enrichment in its intragenic region. Furthermore, our TSA studies suggest that class I and/or II HDACs might be regulating differential broad histone acetylation at VEGFR-1 and VEGFR-2. However, H3K36me3-dependent HDAC recruitment is unlikely to regulate broad differential histone acetylation at select EC genes as H3K36me3 occurred at all expressed genes (44).
We previously analyzed the DNA methylation profiles of key EC-enriched genes and found differential promoter DNA methylation at a subset of these genes (e.g. eNOS). This finding was associated with their EC-enriched mRNA expression and early S phase replication in non-ECs. The latter observation contradicts the common paradigm of genes replicating early and late in S phase in expressing and non-expressing cells, respectively. In contrast, VEGFR-1 and VEGFR-2 were not differentially DNA-methylated and followed the common paradigm of replication timing (19). Here, we demonstrate that broad differential intragenic histone acetylation also distinguishes VEGFR-1 and VEGFR-2 from other EC-enriched genes.
Although we argue that KAT7 acts on EC physiology through VEGFR-2 regulation, KAT7's role in DNA replication may contribute to its effects on EC physiology via specific effects on VEGFR-1 and VEGFR-2 replication. KAT7 was noted to have a role in replication licensing during the G 1 -S phase transition (31). However, no studies have found replication timing defects Specifically, a total of 147 control MO-treated zebrafish were used for assessing the frequency of all phenotypes. A total of 229, 125, and 140 E1ATG MO-treated zebrafish were used for assessing the frequency of the SIV, circulatory integrity (blood circulation), and hemorrhage phenotypes, respectively. A total of 251, 142, and 132 E1ATG MO ϩ KAT7 RNA-treated zebrafish were used for assessing the frequency of the SIV, circulatory integrity (blood circulation), and hemorrhage phenotypes, respectively. A total of 101, 156, and 149 E3I3 MO-treated zebrafish were used in determining the frequency of the SIV, circulatory integrity (blood circulation), and the hemorrhage phenotypes, respectively. * denotes a statistical significance of p Յ 0.05 compared with control MO, and ¶ represents a statistically significant difference between E1ATG MO versus E1ATG MO ϩ KAT7 RNA. Representative fluorescence images of tg(kdrl:eGFP, gata1:dsRED) zebrafish embryos (3 dpf) treated with MOs and human KAT7 RNA showing effects on SIV formation (magnification: ϫ111, numerical aperture: 0.10) (B), circulatory integrity (magnification: ϫ38.3, numerical aperture: 0.12) (C), and hemorrhage (see arrow; magnification: ϫ119, numerical aperture: 0.25) (D) are shown. The dorsal aorta, intersegmental vessels, and caudal vein are labeled as DA, ISV, and CV, respectively. All fluorescence images were acquired using a Leica M205 FA dissecting microscope equipped with a Leica DFC 365 FX digital camera using the Leica Application Suite Advanced Fluorescence software.

Role of KAT7 in EC gene expression
at specific genes following KAT7 disruption. Future studies should address this possibility, especially because KAT7 depletion in ECs and HeLa cells in this study and others, respectively, reduced cell proliferation and dysregulated genes with roles in cell proliferation, cell cycle, and DNA replication (18). In particular, we noted an increase in mRNA species associated with cell cycle and mitosis even though cell proliferation is reduced. The increased expression of these mRNA species may represent a frustrated cellular response to cell cycle impairment. Interestingly, only nine mRNAs were shared between KAT7depleted ECs and HeLa cells. 7 These data highlight the role of KAT7 function and cell-specific gene expression.
KAT7 has been noted to have cell-specific functions. Here, we argue that KAT7 regulates EC gene expression and physiology. Previously, others have suggested that Brpf2-containing KAT7 complexes play a role in murine fetal liver erythropoiesis and T lymphocyte development (15,45). KAT7 is also implicated to be involved in the early stages of adipocyte differentiation by being required for the process of mitotic clonal expansion (46). Surprisingly, we noted that KAT7 is ubiquitously expressed in multiple human cell types and zebrafish embryos (32). Thus, the cell-specific functions of KAT7 may instead be imparted by the unique composition of cell-specific KAT7 complexes (14,15,17,45). Indeed, JADE and BRPF proteins can act as interchangeable scaffolding subunits of KAT7 complexes that determine their residue preference for histone acetylation (14). Future studies should address the composition of KAT7 complexes in different cell types and how their composition can affect gene expression.
Previous studies suggest that KAT7 is required for early embryonic development. Kat7-deficient mice have gross abnormalities in their immature vascular architecture, but developmental arrest at the 10-somite stage and embryonic lethality (E10.5) precluded later assessments (12). Here, we found that kat7 depletion perturbed EC function based on cellbased assays. Moreover, kat7-depleted zebrafish showed disrupted SIV formation and compromised circulatory integrity, especially hemorrhage. Importantly, these defects could be rescued with human KAT7 RNA. Overall, our study complements and extends the findings in the Kat7-deficient mice by invaluably offering further insight into KAT7's role in mature ECs and EC of the vasculature at advanced stages of embryonic development (47).
Consistent with the vascular phenotype of kat7-depleted zebrafish, KAT7 disrupted human and zebrafish EC gene expression. Indeed, mRNA species that are down-regulated by KAT7 depletion are enriched for biological processes associated with EC physiology, including the regulation of blood vessel size. In particular, we found that KAT7 depletion has a major effect on VEGF biology in vitro and in vivo. Although basal VEGF-A expression was not affected by KAT7 knockdown in human ECs, we found that KAT7 depletion disrupted VEGFR-1 and VEGFR-2 expression in human ECs and that this regulation was conserved in zebrafish based on the reduced expression of their homologs, flt1, kdrl, and kdr. Furthermore, acute VEGF-A signaling and VEGF-A EC phenotypic responses are blunted in cell culture. Importantly, exogenous kdrl expression rescued the vascular plexus in the SIV, and reduced blood circulation was restored in kat7-depleted zebrafish. These findings further support KAT7 as a major regulator of EC physiology and that it plays a major role in VEGF-A biology in vitro and in vivo.
Surprisingly, kat7-depleted zebrafish showed a perturbed vascular phenotype that was distinct from gene-specific depletion of either kdrl, kdr, or flt1 alone. Kdrl and kdr regulate the formation and function of vessels generated by angiogenesis with kdrl playing a more prominent role (27,48). Although SIV were disrupted in both kat7-depleted and kdrl-deficient zebrafish, kat7-depleted zebrafish did not exhibit truncated ISVs (48). Also, they did not display aberrant ISV hyperbranching observed in flt1-depleted zebrafish (49). We infer that the phenotypic differences are due to the dysregulation of other genes in kat7-depleted zebrafish. Alternatively, the respective phenotypes observed by deficiency in the VEGFR-1 and VEGFR-2 orthologs might have been neutralized by their combined reduced expression. Regardless, the ISVs of kat7depleted zebrafish were perturbed based on reduced EC numbers in these vessels (3 dpf). Furthermore, initial ISV sprouting

Role of KAT7 in EC gene expression
appeared dysregulated suggesting a defect in cell migration and proliferation that is consistent with reduced wound healing, cell migration, VEGFR-2 tyrosine 1175 phosphorylation, and cell proliferation activity of KAT7-depleted ECs (50). The compromised circulatory integrity in KAT7-depleted zebrafish might be due to vessel underdevelopment in the ISVs. Alternatively, it might reflect disrupted hematopoiesis as suggested by Kat7's role in murine fetal liver erythropoiesis (15). However, this is unlikely as others had previously reported that primitive and definitive hematopoiesis was only mildly affected in kat7-depleted zebrafish (51). Furthermore, hemorrhage and blood accumulation in the trunk occurred in kat7-depleted zebrafish.
In summary, our findings support the importance of differential histone acetylation at both the promoter and intragenic regions of EC genes. Although previous studies have shown a gene regulatory function of intragenic histone modifications, their role in EC gene regulation has not been noted (1). We posit that differential intragenic histone acetylation plays a crit-ical role in EC gene regulation and physiology and reveal a previously underappreciated role for KAT7 in these processes, especially as it relates to VEGFR-2. Also, evidence suggests that known intragenic histone-modifying enzymes contribute to vascular development and function (52,53). These emerging findings may suggest that intragenic chromatin modifiers may be viable therapeutic targets for treating cardiovascular disease and warrant further study (1).

Role of KAT7 in EC gene expression
and normal human epidermal keratinocytes (NHEK) (Lonza) were cultured according to the supplier's instructions. Cells were maintained at 37°C in 5% CO 2 in a humidified Steri-Cycle incubator (ThermoForma, Model 370). Total cellular RNA was extracted from cultured cells as described previously (7) using the Solution D RNA extraction method (54). Total cellular protein was extracted from cultured cells using RIPA cell lysis buffer (Cell Signaling Technology) according to the manufacturer's instructions. Protease/phosphatase inhibitor mixture (Cell Signaling Technology) was added in the RIPA cell lysis buffer for total cellular protein extraction in VEGF-A signal transduction studies.

Cell treatments
Trichostatin A (1 M; Sigma) studies were performed on HUVECs and HuAoVSMCs as described previously for 4 h with an equivalent amount of ethanol added to control plates (6). siRNA knockdowns were conducted with 40 nM siGENOME human KAT7 siRNA-SMART pool (M-017668-00), which is composed of four unique KAT7 siRNAs (D-017668-01, D-017668-02, D-017668-03, and D-017668-04) and non-targeting siRNA (D-001210-03) from Dharmacon as described previously (7,55). The pool of KAT7 siRNAs was used to reduce potential off-target effects, without jeopardizing KAT7-specific knockdown (56). Potential off-target effects are reduced as a result of the siRNA pool being composed of individual siRNAs at low concentrations (56). Briefly, siRNA transfections were conducted on 90% confluent HUVECs grown on 60-mm tissue culture plates using 33 l of Oligofectamine transfection reagent (Thermo Fisher Scientific) in a total volume of 2000 l. After 4 h of transfection at 37°C in Opti-MEM medium (Thermo Fisher Scientific), M-199 medium (Thermo Fisher Scientific) containing fetal bovine serum (Hyclone), heparin, and endothelial cell growth supplement (Biomedical Technologies) was added. Cells were cultured for another 48 h prior to use in matrigel, wound healing, BrdU incorporation cell prolif- Specifically, a total of 239 and 289 control MO-treated zebrafish were used for assessing the frequency of SIV defects and other phenotypes, respectively. A total of 520 and 425 E1ATG MO-treated zebrafish were used for assessing the frequency of SIV defects and other phenotypes, respectively. A total of 461 and 412 E1ATG MO ϩ kdrl RNA-treated zebrafish were used for assessing the frequency of SIV and other phenotypes, respectively. * denotes a statistical significance of p Յ 0.05 compared with control MO, and ¶ represents a statistically significant difference between E1ATG MO versus E1ATG MO ϩ kdrl RNA. Representative fluorescence images of tg(kdrl:eGFP, gata1:dsRED) zebrafish embryos (3 dpf) treated with MOs and zebrafish kdrl RNA showing effects on SIV formation (magnification: ϫ74.7, numerical aperture: 0.10) (B), circulatory integrity (blood circulation) (magnification: ϫ23.9, numerical aperture: 0.05) (C), and hemorrhage (see arrow; magnification: ϫ74.7, numerical aperture: 0.10) (D) are shown. The dorsal aorta, intersegmental vessels, and caudal vein are labeled as DA, ISV, and CV, respectively. All fluorescence images were acquired using a Leica M205 FA dissecting microscope equipped with a Leica DFC 365 FX digital camera using the Leica Application Suite Advanced Fluorescence software.

Role of KAT7 in EC gene expression
eration, and caspase-3 apoptosis assays. For all other assays that use siRNA-transfected HUVECs, cells were cultured for another 72 h instead. Hypoxia treatments of HUVECs (Ͻ1% O 2 ) were conducted as described previously for 0, 4, and 24 h using a temperature-and humidity-controlled incubator with a sealed anaerobic system (ThermoForma, Model 1025) using a high-purity anaerobic gas mixture (5% CO 2 , 10% H 2 , 85% N 2 ; Linde). VEGF-A induction studies were performed with recombinant human VEGF 165 (293-VE) from R & D Systems. HUVECs were induced with VEGF-A for 3, 5, 10, and 30 min at a final concentration of 50 ng/ml after serum starvation for 1 h with Media 199 (Thermo Fisher Scientific) in the VEGF-A signal transduction studies.

Lentiviral transduction
His-eGFP and KAT7 (UHN Vector Core Facility) lentiviruses were used to transduce HUVECs. Specifically, 30,000 -60,000 HUVECs were transduced with the respective lentivirus at 10 multiplicities of infection in the presence of protamine at a final concentration of 8 g/ml for 24 h.

ChIP
ChIP was performed as described previously with the aforementioned antibodies using ϳ1 ϫ 10 6 cells per ChIP, except for the KAT7 and His-tag ChIPs (5-7). The KAT7 and His-tag ChIPs used ϳ3 ϫ 10 6 cells per ChIP instead. Samples were purified using the QIAquick PCR purification kit (Qiagen) and analyzed by qPCR as reported previously with Power SYBR Green Master Mix (Thermo Fisher Scientific) using primers in Table 1 (5-7).

ChIP-custom high-resolution tiling array and ChIP-seq data analysis
ChIP samples were processed for ChIP-chip analysis by the University Health Network microarray center (Toronto, Can-ada) using custom tiling microarrays. The microarrays were designed to tile across the non-repetitive DNA sequences of 34 genes and the 50-kb regions flanking upstream of their transcriptional start site and downstream of their polyadenylation and cleavage signal. The genes included the following: (i) 19 EC genes; (ii) six broadly expressed genes; and (iii) nine EC-excluded genes (Table S1). Both DNA strands of the genes were tiled by an average of 24 distinct DNA probes per 5 bp on a 1 M Agilent feature array with 60-mer oligonucleotides. The oligonucleotides were designed with the Agilent's eArray program using the UCSC HG18 March 2006 human genomic assembly. The data were processed according to a predefined pipeline. Briefly, quantified data files from Agilent Feature Extraction software (Agilent) were first collected into a single directory. Each file contained sense and corresponding antisense tiled probes to genomic regions surrounding genes of interest (Table  S1). Mean expression measurements (Cy5 and Cy3 channels) for each of the strand probes were extracted and sorted into an ascending chromosomal order as defined by the mid-point of the probe location in a format acceptable for input into the Tilemap command line software using custom scripts written in PERL (57). Preprocessing of the data consisted of quantile normalization followed by taking the log base 2. Bar files were then created for input to the Tilemap version 2 program to find pulled down fragment peaks using the moving average (MA) mode. The MA algorithm uses a modified t-statistic averaged over a sliding 400-bp window of adjacent probes (58). False discovery rate calculations are based on a left tail distribution estimate of the MA test statistics. We used a MA cutoff of 3.0 for peak selection. Based on the average probe spacing on the custom Agilent arrays, a window size of 3 was chosen for the tilemap configuration parameter file. Various choices for the configuration of tilemap were tested during the course of experiments to optimize peak findings using both real and simulated data (58). Resulting peaks in each analysis were visualized with the Integrated Genome Browser (59). Genomic regions with a log(2) fold change difference between experimental and control ChIPs in HUVEC that were Ն0.6 versus in HuAoVSMC were considered EC-enriched. The ChIP-chip data for pan-AcH3 (GSE93868) and pan-AcH4 (GSE93962) are deposited in the NCBI GEO database. ChIP-seq signal enrichment tracks of AcH3K9, AcH3K27, and pol II for HUVEC and NHEK were obtained from the ENCODE database and visualized using Integrated Genome Browser (9, 59).

Gene expression microarray analysis
Total RNA of HUVEC samples treated with control siRNA or KAT7 siRNAs (n ϭ 4) were processed by Arraystar Inc. for gene expression microarray analysis. The microarray used for sample hybridization is the human LncRNA microarray V3.0, a custom 8 ϫ 60K Agilent array containing 58,944 probes that detect 26,109 mRNA transcripts and 30,586 long non-coding RNA transcripts. Differentially regulated genes had Ն2-fold change in expression between conditions (p Յ 0.05). The microarray data (GSE93608) is deposited in the NCBI GEO database.

Role of KAT7 in EC gene expression qRT-PCR
First-strand cDNA synthesis was conducted on total cellular RNA using the SuperScript III First-Strand Synthesis Supermix for qRT-PCR kit (Thermo Fisher Scientific). cDNA was quantified by qPCR as described previously using primers in Table 1 (5-7). For the TSA experiments, VEGFR-1 and VEGFR-2 mRNA expression was determined by absolute quantification with plasmid standard curves and subsequent normalization to cyclophilin A mRNA levels. For other qRT-PCR experiments on human ECs, relative mRNA expression was determined by either absolute quantification with plasmid standard curves or the comparative C t method using 18S rRNA expression for normalization. For qRT-PCR experiments on zebrafish embryo samples, mRNA expression was determined by comparative C t using zebrafish ␤-actin for normalization.  -AGG AAT TGA CGG AAG GGC AC-3Ј  5Ј-GGA CAT CTA AGG GCA TCA CA-3Ј  ABCG1 transcript  5Ј-GCT GGA GCT GGT GAA CAA C-3Ј  5Ј-GGT GGA TGG TGC AAA TGA T-3Ј  ACE1 transcript  5Ј-AGG CCA ACT GGA ACT ACA A-3Ј  5Ј-TCT GCA ACT GGT TCA CAT C-3Ј  ASS1 transcript  5Ј-GCC CGC AAA CAA GTG GAA AT-3Ј  5Ј-CAT CCT CCA GGG AGC AAT GAC-3Ј  CASZ1 transcript  5Ј-CCT CCC TGT CCT TCA ACA CT-3Ј  5Ј-TGA CGG CTG GTT TAT CTG TG-3Ј  CD34 transcript  5Ј-CAG GCA TCA GAG AAG TGA AAT T-3Ј  5Ј-CCC TCT CCC CTG TCC TTC TTA AA-3Ј  CEACAM1 transcript  5Ј-CAG CCC CAC TTC ACA GAG TG-

Role of KAT7 in EC gene expression
Immunoblots Immunoblots were conducted as described previously (7,55). Briefly, total protein extracts were size-fractionated with NuPAGE Novex 4 -12% BisTris or 3-8% Tris acetate gels (Thermo Fisher Scientific) using the XCell SureLock Mini-Cell (Thermo Fisher Scientific) and transferred onto 0.45-m nitrocellulose membranes using the XCell II Blot Module according to the manufacturer's recommendations (Thermo Fisher Scientific). Immunoreactive bands were detected using the aforementioned primary and secondary antibodies and visualized with the Amersham Biosciences ECL Prime Western blotting Detection Reagent (GE Healthcare). Signal quantification was performed using ImageJ (National Institutes of Health) and normalized with ␣-tubulin.

Matrigel assay
Growth factor-reduced matrigel (BD Biosciences) was plated onto 96-well plates (50 l/well), incubated at 37°C for 30 min, and seeded with HUVECs at 16,250 cells/well 48 h post-siRNA transfection. Images were captured after 8 h using a Nikon Eclipse TS100 for quantification. Specifically, one central image was captured from each well for a total of three wells for each condition at ϫ4 magnification (numerical aperture: 0.13). Images were analyzed using a commercial imaging service (Wimasis) to obtain the number of branch points and branches per field. The number of isolated and network branches were determined for the images with the ImageJ angiogenesis analyzer using default parameters (60). These experiments were conducted on three biological replicates. Representative images were acquired using a Zeiss AxioObserver.Z1 microscope with environmental control (37°C, 5% CO 2 with humidity) at ϫ10 magnification (numerical aperture: 0.3). The microscope was programmed to image a 4 ϫ 4 tile region in a well that represents a sample in rapid succession with an approximate 10% overlap between adjacent tiles/images. The overlaps were used to stitch individual tiles/images to generate composite images for samples using the Zen Blue 2012 software.

Wound-healing assay
Wound-healing assays were performed in HUVECs grown on 6-well plates 48 h post-siRNA transfection as described previously (61). Images were acquired from the Zeiss Axio-Observer.Z1 microscope at 0 and 20 h at ϫ10 magnification (numerical aperture: 0.3). The microscope was programmed to image a 3 ϫ 15 tile region that represents most of the scratched area of a sample in rapid succession with an approximate 10% overlap between adjacent tiles/images. The overlaps were used to stitch individual tiles/images to generate composite images for samples using the Zen Blue 2012 software. Images were analyzed using a commercial imaging service (Wimasis).

Cell-migration assay
HUVECs treated with control siRNA or KAT7 siRNAs were serum-starved for 3 h with Media 199 (Thermo Fisher Scientific) after 72 h of siRNA transfection and subsequently seeded (50,000 cells/well) on fibronectin-coated (50 g/ml; Sigma-Aldrich) 8.0-m pore polycarbonate transwell membrane inserts (Corning Inc.). Cell migration was conducted for 4 h in the presence of 0.1% BSA or VEGF 165 (25 ng/ml) at the bottom of each well. Non-migrated cells were removed, and inserts were subsequently fixed with 10% formalin (Thermo Fisher Scientific) for 15 min. Inserts were then washed and stained with 0.5% crystal violet (Bioshop) and dried overnight. Membranes from the inserts were removed and mounted on glass slides with Permount (Thermo Fisher Scientific). Membranes were visualized using an Olympus upright BX50 microscope equipped with a DP72 camera, and five unique fields of view at ϫ10 magnification (numerical aperture: 0.2) per membrane were captured for each sample. The total number of cells in the five fields of view were quantified by manual counting.

BrdU incorporation cell-proliferation assay
HUVEC treated with control siRNA or KAT7 siRNAs were subjected to BrdU cell-proliferation assays (Millipore) at 48 h post-treatment following the manufacturer's instructions in 96-well plates (20 000 cells/well). Colorimetric measurements were conducted using the SpectraMax M5e microplate reader (Molecular Devices).

Caspase-3 apoptosis assay
Caspase-3 assays were conducted on HUVECs treated with control siRNA or KAT7 siRNAs using the EnzChek caspase-3 assay kit 1 (Thermo Fisher Scientific) at 48 h post-treatment according to the manufacturer's instructions. The volume of cell lysates used in the assays was adjusted according to the amount of protein present in each sample. Fluorescence measurements were conducted using the SpectraMax M5e microplate reader (Molecular Devices).

Zebrafish husbandry and ethics statement
Zebrafish were housed in the Li Ka Shing Knowledge Institute research vivarium (Toronto, Canada) and maintained and staged as described previously (62). All zebrafish (Danio rerio) experiments were conducted under the approved protocol ACC403 of St. Michael's Hospital Animal Care Committee (Toronto, Canada). Briefly, all fish strains were housed under a 14-h light/10-h dark cycle at 28°C. Pairs of appropriate zebrafish strains were mated to produce embryos that were raised in 1ϫ E3 embryo medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl 2 , 0.33 MgSO 4 ).

Role of KAT7 in EC gene expression
(dsRED) fluorescence signal. Images of living tg(fli1:nEGFP) y7 embryos were also taken using a confocal microscope (Zeiss LSM 700). Embryos were anesthetized with 0.16 mg/ml tricaine methanesulfonate (Finquel) and embedded in 2.5% methylcellulose (Sigma-Aldrich) or 1% low-melting agarose (Bioshop) in the desired orientation prior to imaging with the dissection or confocal microscope, respectively. Total cellular protein was extracted from zebrafish embryos (3 dpf) with RIPA cell lysis buffer (Cell Signaling Technology) containing dissolved mini-protease inhibitor mixture tablets (Roche Applied Science) upon homogenization with a microcentrifuge pestle. To sort GFP ϩ and GFP Ϫ cells from tg(kdrl:eGFP) zebrafish embryos (3 dpf), embryos were dechorionated, deyolked, and digested with TrypLE (Thermo Fisher Scientific) for 1 h at 28.5°C. The dissociated cells were resuspended in PBS, 1% BSA and sorted using a BD FACS Aria I cell sorter as described previously (63). RNA was purified from pre-sorted, GFP ϩ , and GFP Ϫ cells using the PicoPure RNA isolation kit (Thermo Fisher Scientific) according to the manufacturer's instructions. Zebrafish RNA was also used for standard RT-PCR to confirm kat7 inhibition by the E3I3 MO. It was performed with the following kat7 primers, kat7 ex3-ex4 amplicons: 5Ј-TTC GGA TGA CTC TGG AGA TC-3Ј and 5Ј-TCC TGG GAG GAG ACT CAT CT-3Ј; and kat7 ex3-in3 amplicons: 5Ј-AAG CCA ACC AGC GCA ATA AC-3Ј and 5Ј-CAG CCT GGC CTT TTT TCT GT-3Ј under the following PCR conditions: 94°C denaturation for 5 min, 35 cycles of 94°C for 30 s, 57.9 and 55.6°C for 30 s, respectively, and 72°C for 30 s, and 72°C final extension for 15 min.

Statistical analysis
Statistical analyses were performed using a Student's t test or analysis of variance and Newman-Keuls post hoc test, as appropriate. A p value Յ0.05 was considered statistically significant.