Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Nature Cardiovascular Research (2024 )Cite this article aav packaging
Signal-responsive gene expression is essential for vascular development, yet the mechanisms integrating signaling inputs with transcriptional activities are largely unknown. Here we show that RNF20, the primary E3 ubiquitin ligase for histone H2B, plays a multifaceted role in sprouting angiogenesis. RNF20 mediates RNA polymerase (Pol II) promoter-proximal pausing at genes highly paused in endothelial cells, involved in VEGFA signaling, stress response, cell cycle control and mRNA splicing. It also orchestrates large-scale mRNA processing events that alter the bioavailability and function of critical pro-angiogenic factors, such as VEGFA. Mechanistically, RNF20 restricts ERG-dependent Pol II pause release at highly paused genes while binding to Notch1 to promote H2B monoubiquitination at Notch target genes and Notch-dependent gene expression. This balance is crucial, as loss of Rnf20 leads to uncontrolled tip cell specification. Our findings highlight the pivotal role of RNF20 in regulating VEGF–Notch signaling circuits during vessel growth, underscoring its potential for therapeutic modulation of angiogenesis.
Formation of blood vessels requires precise coordination of cellular and molecular events guided by signaling cues. Highly mitogenic and plastic venous endothelial cells (ECs) serve as the primary source for angiogenic expansion, a process orchestrated by vascular endothelial growth factor (VEGF) and Notch signaling gradients1,2,3,4,5,6,7,8,9,10,11. Venous ECs migrate against the flow in a process known as reverse migration, and the ECs at the leading edge of the newly formed sprout, exposed to the highest VEGFA levels, give rise to tip cells3,6,7,10,11,12. VEGF binding to the pro-angiogenic receptor VEGFR2 in tip cells results in activation of tip cell-enriched genes, such as VEGFR3 and the Notch ligand delta-like 4 (DLL4), among others. Increased DLL4 in tip cells activates Notch signaling in the adjacent stalk ECs, thereby limiting VEGF signaling and sprouting behavior in these stalk cells1,2,3,4,5,6,7,8,9,10,11. Tip cells with high Notch activity become pre-specified for an arterial fate and will subsequently contribute to arteries through the same reverse migration process7. Signal-responsive gene expression is essential for the tight control of cell state transitions during vessel development, yet the mechanisms allowing the integration of signaling inputs and transcriptional activities during vascular growth and patterning are largely unknown.
In the past decade, it has become clear that signal-dependent transcription is primarily regulated after transcription initiation, through the pausing and release of promoter-proximal RNA polymerase (Pol II)13,14,15. After recruitment to the promoter by transcription factors (TFs) and transcription initiation, Pol II often pauses after the synthesis of a short nascent RNA (~20–60 nucleotides (nt)). Pol lI stalls after the binding of two pause-inducing factors, NELF and DSIF, and remains in a paused state until additional signals promote productive elongation. Upon activation by different cues, positive elongation factors P-TEFb and TFIIS induce pause release16. Paused Pol II is instrumental for rapid and synchronous gene induction by different signaling cues and for ensuring stoichiometric transcripts for genes encoding different subunits of multicomponent protein transcriptional complexes. Environmental (for example, heat shock, hypoxia and inflammation), developmental and differentiation signals have been shown to regulate Pol II pausing and release14,15. For example, before HIF1A activation upon hypoxia, its target genes harbor transcriptionally engaged paused Pol II. HIF1A binding and recruitment of the P-TEFb complex results in Pol II pause release for active elongation of HIF1A target genes17. Similarly, VEGFA stimulation rapidly induces overall Pol II pause release, primarily at early upregulated genes18. Sensing and responding to signals is central to the function of the endothelium; however, the molecular players that regulate these processes are poorly understood.
RNF20, the major E3 ubiquitin ligase responsible for monoubiquitination of histone H2B at lysine 120 (H2BK120ub, H2Bub1)19,20,21, plays a key role in transcriptional pausing and elongation by Pol II. On the one hand, RNF20-mediated H2Bub1 facilitates FACT function in promoting transcriptional elongation by destabilizing nucleosomes21. Conversely, RNF20 impedes the recruitment of TFIIS, necessary for releasing Pol II into active elongation at genes that promote tumorigenesis, thus repressing pro-oncogenic transcriptional programs22. Interestingly, mutations in the RNF20 gene have been associated with cardiovascular malformations in human patients23, and a recent study revealed that RNF20-regulated EC-born signals orchestrate neural precursor cell fate during embryogenesis24. Despite its important molecular functions and the intriguing findings mentioned above, the role of RNF20 in ECs remains largely unexplored.
Analysis of single-cell RNA sequencing (RNA-seq) datasets of ECs from postnatal day 6 (P6) and P10 retinas25 revealed temporal and spatial expression dynamics of Rnf20 in ECs. Rnf20 levels were lowest in venous and significantly higher in tip ECs, with highest levels in arterial ECs at early postnatal stages (P6) (Fig. 1a and Extended Data Fig. 1a). At P10, Rnf20 levels were generally lower when compared to P6 (Extended Data Fig. 1a). This differential expression pattern suggested a potential role for RNF20 in regulating tip and/or arterial identity of ECs.
a, Violin plot showing minimum, 25th percentile (Q1), median, 75th percentile (Q3) and maximum Rnf20 expression levels in venous (n = 504), tip (n = 189) and arterial (n = 192) ECs (left) and dot plot of the expression level and frequency of markers for these different cell populations in P6 and P10 retinas (right)25. The P values were calculated using a likelihood-ratio test from the FindMarkers function in Seurat (version 4.1.0). b, Schematic representation of the experimental setup for the data shown in c–s. c,d, Representative confocal images of control (n = 10) and Rnf20iEC-KO (n = 10) retina at P7 stained with the EC marker PECAM1 (CD31) (c) and quantification of the radial length (d). e,f, Overview of retinal vascular plexus of P7 control and Rnf20iEC-KO mice stained with IB4 (e) and quantification of the length of the vascular front (f); n = 10 retinas for each group (f). A, artery; V, vein. g,i, Co-immunostaining of mouse retina at P7 with CD31 and ICAM2 (labeling the vascular lumen) (g) or CD31 alone (i). h, Quantification of filopodia length of control and Rnf20iEC-KO retina ECs at P7; n = 5 control and n = 8 Rnf20iEC-KO retinas. j, Quantification of filopodia number per 100-µm vessel length; n = 5 control and n = 8 Rnf20iEC-KO retinas. k,l, Immunostaining for ESM1 (tip cell marker) and CD31 (k) and quantification of tip cell numbers (l); n = 5 control and n = 8 Rnf20iEC-KO retinas. m,n, Immunostaining for ERG (labeling EC nuclei) and CD31 (m) and quantification of EC number; n = 4 control and n = 3 Rnf20iEC-KO retinas (n). o,p, Labeling for EdU and ERG (o) and quantification of double-EdU/ERG-positive ECs; n = 5 control and n = 6 Rnf20iEC-KO retinas (p). q,r, Immunostaining for SOX17 and the Golgi marker Golph4/GPP130 of control and Rnf20iEC-KO retinas (q) and schematic drawing representing the Golgi orientation presented in q (r). s, Quantification of EC polarization; n = 4 retinas for each group. Figure represents mean ± s.e.m.; statistics were quantified with two-way ANOVA with Sidakʼs correction. Each dot in d, f, h, j, l, n and p represents quantification from different leaflets of the indicated number (n) of retinas, each isolated from an individual mouse. Data are represented as mean ± s.e.m.; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Numeric P values are shown within the figure panels. Ctr, control.
To understand the function of RNF20 in vessel growth, we generated tamoxifen-inducible EC-specific Rnf20 knockout mice by crossing Rnf20 fl/fl to Pdgfb-CreERT2 mice (hereafter referred to as Rnf20iEC-KO mice) (Extended Data Fig. 1b) and studied the impact of EC Rnf20 deletion on postnatal retinal angiogenesis. 4-hydroxy-tamoxifen (4-OHT) was administered to Rnf20iEC-KO and Cre-negative littermates (controls) from P1 to P4, and mice were analyzed at P7 (Fig. 1b). Gene deletion in the ECs was confirmed by real-time quantitative polymerase chain reaction (RT–PCR) and western blot (WB) analysis of ECs purified from P7 lungs (Extended Data Fig. 1c,d). Compared to controls, Rnf20iEC-KO mutant mice showed no differences in gross appearance or body weight (Extended Data Fig. 1e); however, endothelial Rnf20 loss led to a significant decrease in the retinal vessel outgrowth (Fig. 1c,d) accompanied by shortening of arteries and veins and an extended vascular front (Fig. 1e,f and Extended Data Fig. 1f). Endothelial sprouts at the extended vascular front displayed thinner morphology and a smaller lumen, indicated by CD31 and ICAM2 co-immunostaining (Fig. 1g). Mutant ECs also had more and longer filopodia at the sprouting front (Fig. 1h–j). As these cellular characteristics resemble tip-like ECs, we next immunostained P7 retinas for the VEGF-regulated tip cell marker ESM1 (ref. 26) (Fig. 1k,l and Extended Data Fig. 1g). This analysis revealed a striking increase in ESM1-positive ECs, suggesting that Rnf20 deficiency promotes the acquisition of a tip cell-like phenotype.
To study if the increased tip cell number is due to an increased EC proliferation, we labeled endothelial nuclei with antibodies against the TF ERG (marking endothelial nuclei) and CD31 (marking endothelial junctions). Quantification of the ERG-positive nuclei revealed a decreased number of ECs per vascular area (Fig. 1m,n). In addition, endothelial 5-ethynyl-2′-deoxyuridine (EdU) incorporation was also decreased in Rnf20iEC-KO ECs (Fig. 1o,p), suggesting that RNF20 is important for EC proliferation and that the increased tip cell number is independent of EC proliferation. Previous studies showed that CreERT toxicity might affect retinal angiogenesis and, thus, can confound interpretation of the effect of specific gene deletion27. Analysis of Pdgfb-CreERT2-positive versus Pdgfb-CreERT2-negative P7 retinas after the same 4-OHT administration regimen, however, did not find significant differences in radial length, vascular front, the length of arteries and veins or ESM1 staining (Extended Data Fig. 1h–k).
Timecourse imaging studies in zebrafish and mice showed that venous ECs polarize and migrate against the blood flow (reverse migration) to give rise to tip cells that are later recruited into arteries3,7,11,12. To assess if the increased tip cell number and the shortened arteries are due to defective polarization, we labeled control and Rnf20iEC-KO retinas for the Golgi marker Golph4/GPP130, SOX17 and isolectin B4 (IB4) to visualize the EC orientation (Fig. 1q–s). The relative position of the Golgi and nuclei at the vascular front close to arteries were then quantified (Fig. 1s). We found an increase of cells non-oriented or oriented with flow in Rnf20iEC-KO retinas, suggesting that EC polarization against the expected flow direction was impaired (Fig. 1s). These results suggest that compromised migration against the flow might lead to accumulation of the tip cells upon Rnf20 loss of function (LOF).
Notably, the vascular defects of Rnf20iEC-KO mice did not normalize during later phases of retinal development, and, by P14, we observed a substantially altered superficial network and reduced vascular network in the middle and deep vascular plexus (Fig. 2a,b). In contrast, EC-specific Rnf20 ablation at later timepoints—for example, after the superficial layer has formed—resulted in no major morphogenetic defects (Fig. 2c–f). These results are consistent with the lower expression of Rnf20 at these stages (Extended Data Fig. 1a). Taken together, our results suggest that RNF20 is an important regulator of vascular morphogenesis in the early postnatal window, acting as a selective suppressor of tip cell state and function.
a, Schematic representation of the experimental setup in b. b, Immunostaining for IB4 and confocal images of the superficial (left), middle and deep vascular plexus (right panels) at P14. c, Schematic representation of the experimental setup in d. d, Immunostaining for IB4 and confocal images of the superficial, middle and inner (deep) vascular plexus at P21. e, Schematic representation of the experimental setup in f. f, Immunostaining for IB4 and confocal images of the superficial, middle and inner (deep) vascular plexus at P28 and quantification of the vascular density (right panels); each dot represents quantification from individual leaflets from n = 6 retinas for each group. Data in f are represented as mean ± s.e.m. Differences between groups were assessed using an unpaired two-tailed Student’s t-test. NS, not significant.
To identify mechanisms underlying RNF20 function in angiogenic growth, we performed RNA-seq of retinal ECs sorted from control and Rnf20iEC-KO mice (Fig. 3a and Extended Data Fig. 2a,b). These analyses identified profound changes in gene expression in RNF20-deficient ECs. Among the upregulated transcripts were genes involved in blood vessel morphogenesis, cell activation and cell motility and MAPK cascade as well as genes involved in negative regulation of cell proliferation, consistent with the observed phenotypic changes (Fig. 3b,c and Source Data Fig. 3). For example, tip cell marker genes such as Flt4 (Vegfr3), Dll4, Esm1 and Cxcr4; venous EC marker genes Nrp2 and Ephb4; and the negative regulator of cell proliferation Cdkn1a were significantly upregulated in Rnf20iEC-KO ECs (Fig. 3b). In contrast, genes such as Cc2d2a, Ift88, Bbs7, Arl6 and Tmem67, which are involved in the formation and organization of cilia, integral to flow response, were downregulated. Gene set variation analysis (GSVA) revealed that genes upregulated upon Rnf20 loss are characteristic for tip and venous ECs25 (Fig. 3d). In contrast, gene signatures of arterial and proliferative ECs were suppressed upon Rnf20 depletion. Consistent with the RNA-seq analysis, we observed greatly increased FLT4 and DLL4 protein levels in Rnf20iEC-KO retinas (Fig. 3e).
a, Schematic representation of the experimental setup. FACS, fluorescence-activated cell sorting. b, Volcano plot showing the distribution of differentially expressed genes in Rnf20iEC-KO versus control retina ECs. n = 3; log2 fold change ≤ −1, ≥1; P < 0.05. Differential expression analysis was performed using DESeq2 (version 1.40.0). Different colored dots represent genes involved in pathways presented in c. c, Top GO terms of genes upregulated and downregulated upon Rnf20 loss in retina ECs. d, GSVA to calculate gene set scores for EC signature summaries for genes upregulated in Rnf20iEC-KO compared to control retina ECs. e, Immunostaining for VEGFR3 (FLT4) (e, left) and DLL4 (e, right), together with IB4. f, Top GO terms of genes upregulated and downregulated upon RNF20 silencing in HUVECs (n = 3; log2 fold change ≤ −0.58, ≥0.58; P < 0.05). g, Heatmap representation of overlapping transcriptional changes in mouse retinas and HUVECs upon RNF20 loss. h, Schematic representation of Pol II average profiles and the method used for defining the Pol II PI. PI was calculated by the ratio of −50 bp to +300 bp Pol II signal divided by the Pol II signal within +300 bp to the TTS + 3 kb. Highly paused genes were identified as those falling within the third quartile (Q3) of the PI in control HUVECs. i, Boxplot showing the minimum, 25th percentile (Q1), median, 75th percentile (Q3) and maximum of Pol II PI after RNF20 depletion in HUVECs for highly paused genes (n = 3,816) and for all genes (n = 15,455); n = 3 Pol II ChIP-seq for each group. j, Genome tracks of merged Pol II ChIP-seq reads in HUVECs transfected with control siRNA or siRNA against RNF20 (n = 3, for each group). k, GO analysis of highly paused genes in HUVECs. GO term enrichment analysis in c, f and k was performed using Metascape (version 3.5). The bars represent the −log10(P value) for enriched GO terms.
To understand whether changes in gene expression are simply due to increased tip cell numbers in the developing Rnf20iEC-KO retinas, we performed RNA-seq in human umbilical vein endothelial cells (HUVECs) transfected with control small interferring RNA (siRNA) or siRNA against RNF20 (Extended Data Fig. 2c–e and Source Data Fig. 3). Similarly to Rnf20iEC-KO ECs, siRNF20 HUVECs showed overrepresentation of genes involved in cell adhesion and motility and, interestingly, an enrichment of genes involved in the VEGFA–VEGFR2 signaling pathway (Fig. 3f,g). Consistent with the decreased EC proliferation in the developing Rnf20 LOF retinas, siRNF20 HUVECs also showed downregulation of cell cycle regulators (Fig. 3f and Extended Data Fig. 2f). RT–PCR analysis further confirmed these transcriptional alterations (Extended Data Fig. 2g). Together, these findings suggest that the morphogenetic defects in the Rnf20 mutants, particularly the tip cell phenotype, might be linked to aberrant VEGFA signaling.
Because RNF20 has been shown to regulate Pol II activity, we next performed chromatin immunoprecipitation followed by sequencing (ChIP-seq) for total Pol II to calculate the pausing index (PI) in control and siRNF20 HUVECs (Fig. 3h–j, Extended Data Fig. 2h,i and Source Data Fig. 3). PI, employed as a surrogate for assessing the extent of promoter-proximal Pol II pausing at a gene, was calculated by the ratio of Pol II density at the promoter (−50 bp to +300 bp) over the elongating Pol II at gene bodies (+300 bp to the transcription termination site (TTS) + 3,000 bp) derived from total Pol II ChIP-seq datasets (Fig. 3h). Genes were classified based on their PI in control HUVECs, with highly paused genes categorized in the third quartile (Q3) of the PI distribution. We found a significantly decreased PI at highly paused genes upon RNF20 loss (Fig. 3i), whereas the PI at all genes was slightly but significantly increased, suggesting a role of RNF20 in mediating promoter-proximal pausing at highly paused genes (Fig. 3i). Gene Ontology (GO) analysis of highly paused genes revealed overrepresentation of genes involved in cellular response to stress, including fluid shear stress, VEGFA–VEGFR2 signaling pathway, RNA metabolism and splicing as well as regulation of cell cycle and cell death (Fig. 3j,k and Extended Data Fig. 2i).
Transcriptional elongation by Pol II is coupled to mRNA processing28, and, therefore, RNF20 LOF might result in alternative splicing. In addition, we also observed genes involved in RNA metabolism and splicing to be paused by RNF20 (Fig. 3k). Thus, we next performed differential splicing analysis using the endothelial in vivo and in vitro RNA-seq datasets. In both, we found a large number of differentially spliced genes (Source Data Fig. 4). Among the differentially spliced and upregulated genes in Rnf20iEC-KO retina ECs were genes involved in extracellular matrix (ECM) organization, VEGFA–VEGFR2 signaling, signaling through tyrosine kinases and genes involved in cell adhesion and cytokine stimulation (Figs. 4a,b and Extended Data Fig. 3a). Among the downregulated and differentially spliced genes were candidates involved in cilium organization. We also found a large number of genes that were differentially spliced but did not show altered expression (Extended Data Fig. 3b). Nevertheless, these alternative splicing events could interfere with protein functions. These genes belonged to the GO terms related to DNA repair, cell cycle, mRNA metabolic process, chromatin organization and mRNA processing (Extended Data Fig. 3b). In this context, RNF20 is required for recruitment of double-strand break (DSB) repair proteins and timely DNA damage repair, and its loss results in accumulation of DNA damage29,30. Indeed, RNF20-depleted HUVECs showed increased levels of γH2AX, a marker of DSBs, without any DNA damage-inducing agents (Extended Data Fig. 3c), in concordance with another study showing accumulation of DNA damage in ECs upon Rnf20 LOF24.
a, Venn diagram showing the overlap between differentially expressed genes (upregulated (Up) and downregulated (Down)) and differentially spliced genes (DSGs) (P < 0.05, log2 fold change ≤ −1, ≥1) in Rnf20iEC-KO compared to control retinal ECs. b, Top GO terms of DSGs upregulated or downregulated in Rnf20-depleted ECs. Representative DSGs in upregulated genes are presented to the right. c, Genome tracks of merged RNA-seq reads at the Vegfa gene locus in control and Rnf20iEC-KO retinal ECs, showing alternative splicing upon Rnf20 LOF. d, PCR products with primers for amplification of full-length Vegfa using cDNA of isolated retinal EC, followed by Sanger sequencing confirming that Rnf20iEC-KO retinal ECs express Vegfa111 and Vegfa121, whereas control ECs express mainly a non-coding transcript. M, marker (DNA). e, Immunostaining for VEGFA (white) together with IB4 (red) showing substantially higher VEGFA levels in Rnf20iEC-KO retina. f, Top GO terms of DSGs (P < 0.05, log2 fold change ≤ −1, ≥1, n = 3) in HUVECs transfected for 60 h with siRNA against RNF20 versus control siRNA. g, qPCR for different VEGFA isoforms in HUVECs transfected with control and RNF20 siRNAs (n = 4 biological replicates for each group). h,i, Sprouting assay with HUVECs transfected with control and RNF20 siRNA without and with VEGFA supplementation, showing that RNF20-deficient HUVECs sprout without VEGFA addition (h) and quantification of the cumulative sprout length (i); n = 5 siControl basal, n = 6 siControl + VEGFA, n = 9 siRNF20 basal and n = 7 siRNF20 + VEGFA (p). Data in g and i are mean ± s.e.m.; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Numeric P values are shown within the figure panels. GO term enrichment analyses in b and f were performed on DSGs using Metascape (version 3.5). The bars represent the −log10(P value) for each enriched GO term.
Notably, Vegfa was differentially spliced in control versus Rnf20iEC-KO retina ECs. In control ECs, Vegfa coding exons 2–7 were largely excluded, leading to a non-functional transcript, and Rnf20 deficiency resulted in the inclusion of exons 2–4/5 (Fig. 4c,d). Sanger sequencing of the PCR products resulting from amplification of retinal ECs cDNAs with primers located at the transcription start and end site confirmed the expression of Vegfa111 and Vegfa121 isoforms in Rnf20iEC-KO (Fig. 4c,d). Consistent with the increased Vegfa111 and Vegfa121 expression, we observed increased total VEGFA protein levels in the retinal vasculature (Fig. 4e). Likewise, silencing of RNF20 in HUVECs resulted in major changes in VEGFA splicing and other genes involved in the VEGFA–VEGFR2 signaling pathway, tube morphogenesis (including endothelial tube formation), protein catabolic processes and ECM organization (Extended Data Fig. 4f). RT–PCR analysis revealed that VEGFA111 was highly upregulated, whereas the anti-angiogenic VEGFA121b isoform was decreased, upon RNF20 depletion (Fig. 4g). Consistent with the increased expression levels of pro-angiogenic and the decreased expression of anti-angiogenic VEGFA isoforms, RNF20-deficient HUVECs were able to sprout even without the addition of VEGFA, and this effect was further enhanced by the presence of exogenous VEGFA (Fig. 4h,i). Taken together, these data emphasize that endothelial RNF20 limits cell-autonomous VEGF signaling. This occurs through transcriptional and post-transcriptional control mechanisms and ensures proper sprouting angiogenesis.
Because our results indicated that RNF20 mediates transcriptional pausing at highly paused genes, we next studied whether it integrates signals and TF activities. To this end, we performed assay for transposase-accessible chromatin with sequencing (ATAC-seq) to analyze the accessibility of chromatin within the genome, followed by TOBIAS footprinting analysis31 (Fig. 5a–c, Extended Data Fig. 4a–c and Source Data Fig. 5). We identified a significant increase in chromatin accessibility at genes involved in cell activation, migration and the VEGFA–VEGFR signaling pathway (Fig. 5a,b), whereas chromatin permissiveness was decreased at genes involved in cytoskeleton-dependent intracellular transport and cilium organization (Extended Data Fig. 4b,c), in concordance with the observed transcriptomic alterations. Footprinting analysis identified an enrichment of motifs for Ets, Stat and Fos-Jun in Rnf20iEC-KO retinal ECs (Fig. 5c and Extended Data Fig. 4d). In contrast, TF motifs associated with components of the Notch signaling pathway (for example, Hes and Hey) were more frequently identified in open chromatin in control ECs. These data suggest that RNF20 might control the VEGF–Notch signaling circuit during sprouting angiogenesis.
a,b, Venn diagram showing the overlap between genes characterized by increased chromatin accessibility (n = 2, log2 fold change ≤ −0.58, ≥0.58; P < 0.05) and increased expression in Rnf20iEC-KO retina (a) and GO analysis of genes within the overlap (b). P values were calculated using Metascape (version 3.5). c, TOBIAS footprinting analysis31 of control and Rnf20iEC-KO retinal ECs. d, Cellular fractioning followed by WB analysis for N1ICD, RBPJ, H2Bub and H2B of control and RNF20 siRNA-transfected HUVECs. H2B and α-tubulin (TUB) served as loading control. e, Co-immunoprecipitation with N1ICD and RNF20. f, Relative fold enrichment of N1ICD binding at the HES1 locus in siRNF20 versus control HUVECs, determined by N1ICD ChIP–qPCR analysis (n = 3). g, HES1-luciferase Notch reporter activity in control and siRNF20 cells transfected with N1ICD (n = 8). h, Genome tracks of merged H2Bub1 ChIP-seq reads in HUVECs transfected with control and siRNF20 (n = 3 ChIP-seq per group). i, Bar plot showing the statistical significance (−log10(P value)) of H2Bub1 decrease in siRNF20 versus control HUVECs on the gene body of hg38 genome-wide transcriptome, at downregulated upon RNF20 LOF genes and at Notch target genes (GO:0007219). j, qPCR for Notch target genes (HES1, HES2, HEY1 and NOTCH3) as well as other downregulated upon RNF20 LOF genes, such as CCND1 and SOX17, in HUVECs transfected with control and siRNF20 for 24 h (n = 3). k, Statistical significance (−log10(P value) > 50) of differential TF motifs at ATAC-seq peaks in Rnf20iEC-KO versus control retinal ECs (n = 2) and in siRNF20 versus control HUVECs (n = 4). Statistical significance was assessed using TOBIAS (version 0.14.0). l, Line plots representing the average ERG ChIP-seq read density (RPM) at TSS ± 2 kb of genes upregulated, downregulated or non-changed (NC) in siRNF20 HUVECs (n = 4). Shaded areas represent the ±s.e.m. m, Genome tracks of ERG ChIP-seq reads at RNF20-dependent genes. Primers used for the ChIP–qPCR analysis in n are indicated with a black bar. n, Relative fold enrichment of ERG binding in siRNF20 versus control HUVECs, determined by ChIP–qPCR analysis (n = 4). Data in f, g, j and n represent mean ± s.e.m.; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Differences in i and l were calculated using a two-sided Wilcoxon test. Numeric P values are shown within the figure panels. CTRL, control; KO, knockout; RPM, reads per million.
Consistent with the decreased footprints of TFs involved in Notch signaling in the ATAC-seq datasets of Rnf20iEC-KO retinal ECs, we observed a reduced abundance of the Notch1 intracellular domain (N1ICD) on chromatin using cellular fractionation of RNF20-depleted HUVECs (Fig. 5d and Extended Data Fig. 4e). Furthermore, co-immunoprecipitation experiments revealed that RNF20 and N1ICD interact (Fig. 5e), and RNF20 deletion resulted in decreased binding of N1ICD at the promoter of the Notch target gene HES1 (Fig. 5f) and decreased HES1 reporter gene activity (Fig. 5g), suggesting that RNF20 might be required for Notch-dependent gene activation. Interestingly, dBre1 (the Drosophila homolog of RNF20) was shown to be required for the expression of Notch target genes in Drosophila by coupling H2Bub1 to H3K4me3 (ref. 32). Thus, we next performed ChIP-seq for H2Bub1 in control and siRNF20-transfected HUVECs (Fig. 5h). We observed significant decrease of H2Bub1 genome wide upon RNF20 depletion, consistent with its function as the main E3 ligase of H2Bub1 (Fig. 5i and Extended Data Fig. 4f). Notably, H2Bub1 levels were more significantly depleted at downregulated genes upon RNF20 LOF, such as SOX17 and CCND1, with particularly significant depletion observed at Notch target genes (Fig. 5h,i and Extended Data Fig. 4f). Consistent with the major decrease in H2Bub1, the expression of these genes was significantly downregulated already 24 h after RNF20 siRNA transfection (Fig. 5j).
On the other hand, we found enrichment of motifs for ERG, STAT3 and FOSL2, which play important roles in the regulation of VEGFA expression and its downstream effects, at open chromatin regions in Rnf20iEC-KO retinal ECs. TF-gene regulatory network analysis of ERG, STAT3 and FOSL2 footprints suggested that ERG functions upstream of STAT3 and FOSL2 in activating pro-angiogenic genes, such as Vegfa, Vegfc, Flt4 and Nrarp (Extended Data Fig. 4g). To better discern primary from secondary effects, we conducted ATAC-seq in HUVECs transfected with either control siRNA or siRNA targeting RNF20. Comparative analysis of TF motifs enriched in accessible chromatin in both retinal ECs and HUVECs upon RNF20 LOF revealed enrichment of ETS family members in both datasets (Fig. 5k and Extended Data Fig. 4h). In contrast, STAT3 and FOSL2 motifs were exclusively enriched in retinal ECs, indicating an essential role for ETS family members in RNF20 LOF ECs.
Among the ETS family of TFs, ERG was found majorly upregulated in Rnf20iEC-KO retinal ECs (Fig. 3b). ERG is instrumental for EC transcriptional programming33,34 and regulates Notch signaling and vascular stability35 but also VEGF-inducible transcription36. Analyses of the genome-wide ERG binding profiles37 revealed significantly higher ERG occupancy at genes upregulated in RNF20-deficient HUVECs, including STAT3, FOSL2, DLL4, VEGFA, VEGFC and NRARP (Fig. 5l,m), supporting our footprinting and network analysis. Furthermore, ChIP–qPCR analysis revealed significantly increased binding of ERG at the DLL4, VEGFC and STAT3 regulatory regions upon RNF20 depletion (Fig. 5m,n). ETS1, on the other hand, has been demonstrated to enhance VEGFA-dependent gene expression by promoting the release of paused Pol II (ref. 38).
To further investigate the potential involvement of ETS family members in Pol II pause release after RNF20 LOF, we intersected genes exhibiting ERG (ref. 34) or ETS1 (ref. 38) enrichment at the transcription start site (TSS) (±1 kb) with genes displaying reduced transcriptional pausing in RNF20-depleted HUVECs. Notably, 59% of genes demonstrating decreased Pol II pausing upon RNF20 loss (Fig. 6a) were bound by ERG, ETS1 or both (Fig. 6a,c and Extended Data Fig. 5a). GO analysis of genes bound by both ERG and ETS1, exhibiting decreased pausing in RNF20-depleted HUVECs (825 genes), revealed enrichment in pathways related to VEGFA–VEGFR2 and Rho GTPase signaling, DNA damage response and cell division (Fig. 6b). Genes exclusively bound by ERG (74 genes) were associated with cell cycle checkpoints and division, whereas those bound solely by ETS1 (269 genes) were linked to cell proliferation, tube morphogenesis and VEGFA–VEGFR2 signaling (Extended Data Fig. 5b).
a, Overlap of genes bound by ETS1 (ref. 38) or ERG (ref. 34) at the TSS (±1 kb) with genes showing decreased PI in HUVECs transfected with siRNF20 versus control siRNA. b, GO analysis of genes bound by both ERG and ETS1 and showing decreased PI upon RNF20 loss. c, Genome tracks of ERG and ETS1 ChIP-seq reads and merged total Pol II ChIP-seq reads in control (n = 3) and siRNF20 (n = 3) HUVECs. d, Images of ERG and CD31 immunostaining of control and Rnf20iEC-KO retinas. e, Box plot showing the minimum, 25th percentile (Q1), median, 75th percentile (Q3) and maximum of Pol II PI after ERG overexpression in HUVECs for highly paused genes (n = 3,554) and for all hg38 genes. f, Overlap of genes bound by ETS1 or ERG at the TSS (±1 kb) with genes showing decreased PI in siRNF20 versus control HUVECs (n = 3) or ERG-overexpressing (OE) versus control HUVECs (n = 2 ChIP-seq per group). g, GO analysis of genes bound by ERG and showing decreased PI upon both ERG overexpression and RNF20 depletion. h, Cellular fractioning of control and ERG-OE HUVECs, followed by WB analysis for RNF20 and H3 (loading control) in the chromatin-bound fraction (CBF) and quantification (n = 4) to the right. i, qPCR analysis of RNF20-dependent genes in control (n = 3) and ERG-OE (n = 4) HUVECs. j, Sprouting assay with HUVECs (left) and quantification of the cumulative sprout length (right). Basal: siControl (n = 15) and ERG OE (n = 10); +VEGFA: siControl (n = 15) and ERG OE (n = 9). k, qPCR analysis of HUVECs transfected with control (n = 3), with RNF20 siRNAs alone (n = 3) or with 1-nm ERG siRNA to reduce ERG levels in RNF20-depleted HUVECs to control levels (n = 4 siRNF20 + siERG). l, Sprouting assay (left) and quantification of the cumulative sprout length (right). Basal: siControl (n = 13), siRNF20 (n = 26), siERG (n = 9) and siRNF20 + siERG (n = 8); +VEGFA: siControl (n = 11), siRNF20 (n = 14), siERG (n = 13) and siRNF20 + siERG (n = 12). Data in h, i and j represent mean ± s.e.m.; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Data in k and l represent mean ± s.e.m.; differences between groups were assessed using one-way ANOVA multiple comparisons test. P values in b and g were calculated using Metascape (version 3.5).
Given that ERG was significantly upregulated in Rnf20iEC-KO retinal ECs (Figs. 3b and 6d) and over 46% of genes exhibiting reduced transcriptional pausing in RNF20-depleted HUVECs were also bound by ERG, we investigated whether elevated levels of ERG might similarly facilitate the release of paused Pol II. To minimize the potential for non-specific effects associated with the shared core DNA binding motif of ETS family TFs, we overexpressed ERG approximately two-fold compared to the control, matching the level of ERG upregulation observed in RNF20 LOF retinal ECs (Fig. 3b). ChIP-seq analysis of total Pol II in both control and ERG-overexpressing HUVECs revealed a significant reduction in Pol II pausing at highly paused genes upon ERG overexpression, with no impact on Pol II distribution across the genome (Fig. 6e). Genes bound by ERG, showing decreased pausing in ERG-overexpressing HUVECs, were associated with chromatin organization, RNA metabolism, DNA damage response, regulation of cellular response to stress as well as nervous system and vasculature development (Extended Data Fig. 5c). Notably, genes bound by ERG exhibiting decreased pausing in both RNF20-depleted and ERG-overexpressing HUVECs were linked to receptor tyrosine kinase signaling, signaling by Rho GTPases and VEGFA, cell activation and sprouting angiogenesis (Fig. 6f,g). Consequently, we further investigated whether and how the increased levels of ERG contribute to the transcriptional alterations and enhanced sprouting observed in RNF20-depleted HUVECs. Interestingly, overexpression of ERG in HUVECs, at levels similar to Rnf20-deficient ECs, led to significantly decreased binding of RNF20 to chromatin, despite the increased RNF20 mRNA expression (Fig. 6h and Extended Data Fig. 5d). This suggests that increased ERG levels restrict RNF20 binding to chromatin. Notably, ERG overexpression increased expression of RNF20-repressed genes (Fig. 6i) and increased sprouting in basal medium, similarly to RNF20-deficient HUVECs (Fig. 6j). Furthermore, reducing ERG levels in RNF20-depleted HUVECs to those seen in controls normalized the transcriptional changes dependent on ERG and reduced the excessive sprouting in response to RNF20 LOF (Fig. 6k,l). Further reduction of ERG levels resulted in significantly lower expression of DLL4 and NRARP, whereas ETS1 silencing had a less pronounced effect (Extended Data Fig. 5e).
Taken together, these results suggest that ERG plays a major role in mediating the transcriptional and functional alterations of RNF20-deficient ECs.
Because the VEGFA–VEGFR signaling pathway was highly enriched in all datasets, we next assessed the downstream signaling events. Interestingly, siRNF20 HUVECs grown in basal medium showed significantly higher levels of VEGFR2 and pTyr1175-VEGFRTyr1175 as well as increased levels of activated Thr202/Tyr2 phosphorylated (p-) ERK1/2 and higher expression of the RNF20-dependent gene DLL4 (Fig. 7a,b), whereas activated p-p38 levels were not significantly affected. Either silencing of VEGFR2 or pharmacological inhibition of p-ERK1/2 with the MEK inhibitor (MEKi) PD0325901 resulted in decreased protein levels of the RNF20-dependent genes, DLL4 and VEGFR3 (Fig. 7c,d and Extended Data Fig. 6a,b). Furthermore, treatment of control and siRNF20 HUVECs with the VEGFR2 signalling inhibitor vandetanib or the MEKi PD0325901 decreased sprouting of RNF20-depleted HUVECs (Fig. 7e and Extended Data Fig. 6c,d). Because ERG activity was shown to be regulated by VEGF/MAPK/ERK signaling36, activation of the VEGFA–ERK1/2–ERG axis might be responsible for the pro-angiogenic activity of RNF20-depleted ECs. Next, we sought to corroborate our findings in vivo either by blocking VEGFR2 signaling using the anti-VEGFR2 antibody DC101 or by pharmacological inhibition of ERK phosphorylation using the MEKi SL327. Both treatments normalized the number of tip cells and number of filopodia in Rnf20iEC-KO similar to control retinas (Fig. 7f–i and Extended Data Fig. 6e–h), emphasizing that increased VEGFA signaling through ERK1/2 activation plays an instrumental role in the cellular phenotypes caused by endothelial Rnf20 loss.
a,b, WB analysis for RNF20, DLL4, activated Thr1175 phosphorylated VEGFR2, total VEGFR2, activated Thr202/Tyr2 phosphorylated ERK1/2, total ERK1/2, activated Thr180/Tyr182 phosphorylated p38, total p38 as well as the loading control α-tubulin of HUVECs transfected with control and RNF20 siRNA (a) and quantification of protein levels (b); n = 7 siControl, n = 7 siRnf20 for VEGFR2; n = 7 siControl, n = 7 siRnf20 for pVEGFR2; n = 8 siControl, n = 8 siRnf20 for pERK1/2. c, WB analysis for RNF20, VEGFR2, VEGFR3, pERKThr202/Tyr204, DLL4 and α-tubulin (TUB) of total protein extracts of HUVECs transfected with control, VEGFR2 and RNF20 siRNA. d, WB analysis for RNF20, VEGFR2, VEGFR3, pERKThr202/Tyr204, DLL4 and α-tubulin of total protein extracts of HUVECs transfected with control or RNF20 siRNA and treated with either DMSO or the MEKi PD0325901 for 24 h. e, Sprouting assay with HUVECs transfected with control or RNF20 siRNA and treated with DMSO, the VEGFR inhibitor vandetanib (ZD6474) or the MEKi PD0325901 cultured in basal medium without VEGFA supplementation (left) and quantification of the cumulative sprout length (right) (n = 10 for siControl, siControl + vandetanib, siControl + MEKi, n = 18 for siRNF20, n = 10 for siRNF20 + vandetanib, siRNF20 + MEKi). f,g, Immunostaining for CD31 and ESM1 in control and Rnf20iEC-KO retinas treated with either PBS or DC101 (f) or with oil and the MEKi SL327 (g). h,i, Quantification of tip cells per area of vascular coverage in control and Rnf20iEC-KO retinas treated with either PBS or DC101; each data point represents different leaflets from n = 4 Control + PBS, n = 6 Rnf20iEC-KO + PBS, n = 6 Control + DC101, n = 6 Rnf20iEC-KO + DC101 retinas (h) or with oil and the MEKi SL327; n = 4 Control + oil, n = 4 Rnf20iEC-KO + oil, n = 2 Control + SL327, n = 4 Rnf20iEC-KO + SL327 retinas (i). j, Schematic representation of the molecular and cellular changes resulting from RNF20 loss. Data in b and e represent mean ± s.e.m.; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Data in h and i represent mean ± s.e.m.; differences between groups were assessed using two-way ANOVA with Tukey’s correction. Numeric P values are shown within the figure panels.
Our work uncovered a crucial role of RNF20 in the regulation of the VEGF–Notch signaling circuits orchestrating EC state transitions, which are the driving force for angiogenic growth and physiological tissue repair but also for pathological angiogenesis (Fig. 7j and Extended Data Fig. 7).
Our data show that RNF20 mediates Pol II promoter-proximal pausing specifically at genes that are highly paused in ECs. These genes play crucial roles in various cellular processes, including the cellular response to fluid shear stress, the VEGFA–VEGFR2 signaling pathway, cell cycle regulation, cell death and RNA metabolism/splicing. Notably, consistent with the rapid transcriptional changes observed in genes encoding splicing factors, we observed substantial alterations in RNA splicing. Additionally, Pol II elongation rate, also termed transcription speed or velocity, influences splicing-related processes, such as constitutive splicing, alternative splicing and back-splicing39. H2Bub1 is closely linked to Pol II elongation rate40; therefore, it is plausible that RNF20 regulates RNA diversity by tightly controlling Pol II pause release at genes involved in the regulation of splicing and/or by the regulation of H2Bub1 levels. Specifically, RNF20 LOF resulted in significant changes in the splicing of the critical regulator of angiogenesis and EC behavior, VEGFA. Alternative mRNA splicing of VEGFA gives rise to many isoforms that show different affinities to their receptors, diffusion properties, stabilities and cellular functions41,42,43. Notably, we found increased levels of VEGFA111 in RNF20-depleted HUVECs and in Rnf20iEC-KO retinal ECs, accompanied by elevated VEGFA protein levels. VEGFA111 encodes a freely diffusible and stable VEGF isoform lacking the heparin-binding ability encoded by exons 6 and 7 as well as the major proteolytic cleavage sites encoded by exon 5, thereby displaying resistance to proteolysis41. VEGFA111 is highly increased in ECs from Dll4 LOF retinas, which exhibit increased tip cell formation and enhanced sprouting angiogenesis7. This isoform is also specifically induced by genotoxic stress41, which significantly alters alternative splicing44. Notably, RNF20 is required for recruitment of DSB repair proteins and timely DSB repair29,30, and RNF20 depletion in ECs results in DNA damage24; thus, it would be interesting to explore to what extent DNA damage induced by RNF20 loss, or by other stressors such as hypoxia, contributes to VEGFA splicing specifically as well as to RNA diversity in general. Recombinant VEGFA111 binding to VEGFR2 was shown to signal specifically through ERK1/2 activation to regulate EC proliferation, migration, sprouting and EC fate choices41,45. Our data indicate that elevated VEGFA signaling, followed by activation of the ERK1/2 signaling cascade and ETS family members, plays a central role in activating pro-angiogenic genes and promoting sprouting upon RNF20 loss.
ETS1 has been demonstrated to boost VEGFA-dependent gene expression by facilitating the release of paused Pol II through the recruitment of the positive elongation factor P-TEFb38. Our study uncovered that increasing ERG levels can also effectively trigger the release of paused Pol II by displacing RNF20 from highly paused genes. Although our analysis focused on TSSs, it is important to note that ERG also regulates endothelial gene expression by binding to enhancers34. Enhancer-bound TFs can also stimulate Pol II pause release through enhancer–promoter loops46. Thus, ERG might promote Pol II pause release by engaging with enhancers and facilitating enhancer–promoter interactions through homodimerization or heterodimerization with other proteins, including other members of the ETS family. In our study, we limited ERG overexpression to approximately two-fold to minimize potential non-specific effects due to the shared core DNA binding motif of ETS family TFs; however, this limitation should be considered when interpreting the results. Notably, ERG appears to play a dominant role in mediating the phenotype in the context of RNF20 LOF. Genes showing increased Pol II release upon ERG overexpression and after RNF20 depletion were associated with receptor tyrosine kinase signaling, signaling by Rho GTPases and VEGFA, cell activation and sprouting angiogenesis. This finding aligns with the role of ERG in dynamically regulating VEGF-inducible genes36. Given the substantial transcriptional changes observed with RNF20 loss, the combination of RNF20 depletion and increased ERG activity may synergistically enhance VEGFA-dependent gene expression.
Together with elevated VEGF signaling, we observed decreased Notch signaling. The Drosophila homolog of RNF20, dBre1, has been shown to be required for the expression of Notch target genes by coupling H2Bub1 to H3K4me3 (ref. 32). Similarly, we observed that the mammalian RNF20 plays a crucial role in regulating Notch target gene expression. We found that N1ICD interacts with RNF20, and depletion of RNF20 led to a global reduction in N1ICD association to chromatin and major decrease of H2Bub1 levels at Notch target genes in ECs, suggesting an evolutionary conserved role of RNF20 in the regulation of Notch-dependent transcriptional program through modulating the histone code. On the other hand, Dll4 and Nrarp, targets of both ERG and Notch34,47, were among the most upregulated genes upon Rnf20 loss. ERG interacts with N1ICD35, suggesting that elevated ERG levels may be responsible for the increased expression of a subset of Notch target genes upon RNF20 LOF. Nrarp is known to inhibit the transcription of Notch target genes48,49, either by forming an inhibitory complex with CSL–NICD or by destabilizing NICD. Thus, its increased expression may also contribute to the attenuated Notch signaling observed in Rnf20iEC-KO ECs.
Finally, we found consistently higher total levels of VEGFR2, which were not affected by ERK1/2 inhibition. The Vegfr2 promoter is repressed by Hes and NICD50. Thus, the inability of these factors to bind to chromatin might be responsible for the high level of VEGFR2 expression and the further potentiation of VEGF signaling. Additionally, ephrin-B2, which enables VEGF receptor endocytosis and downstream signal transduction51, is a highly paused gene significantly upregulated in RNF20-depleted HUVECs and in Rnf20iEC-KO retinal ECs. Thus, because VEGFA and VEGFC levels, as well as their receptors and their signal transduction ability, are increased in RNF20 LOF ECs, our model suggests that exacerbation of VEGF–VEGFR levels results in uncontrolled tip cell specification. Notably, our data emphasize that the increased number of tip cell-like ECs upon Rnf20 loss occurs independently of EC proliferation. Indeed, Rnf20-defcient ECs showed significantly decreased proliferation, which is likely a consequence of excessive VEGFA/C levels above the threshold levels. High mitogenic VEGF stimulation has been proposed to induce cell cycle arrest in angiogenic vessels through p-ERK-dependent upregulation of p21 (ref. 8). CDKN1A, encoding p21, belongs to the highly paused genes in ECs, which are released for active elongation upon RNF20 loss, providing a possible explanation for the decreased proliferation but increased tip cell specification in Rnf20iEC-KO mutants. Additionally, we observed defective reverse migration and compromised artery formation upon Rnf20 LOF. High ephrin-B2 expression in tip cells promotes arterial specification, which involves Notch activation and Sox transcription factor activity10. Exploring to what extent decreased Sox17 levels, Notch signaling activity, altered shear stress response or cilia dysfunction upon RNF20 LOF compromises artery formation would be interesting for future studies.
Interestingly, mutations in the RNF20 gene have been associated with cardiovascular malformations in human patients23, and RNF20 levels are altered in many cancers52. Beyond ECs, RNF20 may, therefore, play a broader role in controlling the VEGF–NOTCH signaling circuit, leading to cardiovascular and cancerous diseases when the function of this E3 ubiquitin ligase is corrupted. Delineating the mechanisms that control RNF20 expression and activity might, thus, provide insights into diseases caused by vascular dysfunction.
The Rnf20tm1a(EUCOMM)Wtsi line was generated by microinjection of Rnf20 tm1a(EUCOMM)Wtsi embryonic stem cells, obtained from the European Conditional Mouse Mutagenesis Program (EUCOMM), into blastocysts (Extended Data Fig. 1a). For the generation of a conditional (floxed) allele Rnf20tm1c(EUCOMM)Wtsi line, the Rnf20tm1a(EUCOMM)Wtsi mouse line was crossed with germline FLP deleter mouse line. To induce endothelial-restricted deletion, the resulting Rnf20tm1c(EUCOMM)Wtsi mice carrying a conditional floxed allele were bred to a line carrying the tamoxifen-inducible recombinase CreERT2 driven by the platelet-derived growth factor subunit B (Pdgfb) promoter—that is, Pdgfb-iCreERT2 (ref. 53). To activate CreERT2, pups were administered 25 μl of 3 mg ml−1 4OHT (Sigma-Aldrich, H7904) intraperitoneally from P1 to P4. Retinas were harvested between P7 and P28. Mice analyzed in this study were on a C57BL/6J background.
All animal experiments were performed according to the regulations issued by the Committee for Animal Rights Protection of the State of Baden-Württemberg (Regierungspraesidium Karlsruhe; experimental protocol Az.: 35-9185.81/G-181/17).
Whole-mount immunofluorescence stainings were performed according to ref. 54. In brief, enucleated eyes were fixed in 4% paraformaldehyde (PFA) either for 2.5 h on ice (for stainings: PECAM1, ESM1, FLT4, ICAM2) or for 30 min at room temperature, followed by retina dissection. After dissection, retinas were incubated in blocking buffer (3% FBS, 1% BSA, 0.25% Tween 20, 0.25% Triton X-100, 0.01% sodium deoxycholate in PBS) for 1 h at room temperature. Primary antibodies were incubated in blocking buffer or in blocking buffer diluted 1:1 in PBS (for stainings: PECAM1, ESM1, FLT4, ICAM2) overnight at 4 °C. After three washes with PBST (0.1% Tween 20 in PBS), retinas were then incubated with Alexa Fluor–conjugated secondary antibodies either in modified PBLEC buffer (1 mM CaCl2, 1 mM MgCl2, 0.1 mM MnCl2, 0.5% Triton X-100) or in blocking buffer diluted 1:1 in PBS (for stainings: Pecam1, Esm1, Flt4, Icam2) for 2 h at room temperature. Next, retinas were washed and flat mounted with ProLong Gold antifade mountant (Life Technologies, P36930). For staining with IB4 alone, dissected retinas were blocked and stained in PBLEC buffer (1 mM CaCl2, 1 mM MgCl2, 0.1 mM MnCl2, 1% Triton X-100) for 2 h at room temperature or overnight at 4 °C. EdU incorporation was detected, after injecting pups with 225 µg of EdU in 50 µl of PBS 4 h before harvest, using a Click-iT EdU kit (Thermo Fisher Scientific, C10337) according to the manufaturer’s instruction. For details on antibody dilutions, see Extended Data Table 1.
Images were acquired with a Zeiss LSM 700 confocal microscope or Axio Scan.Z1. To compare signal intensities between different groups, settings (laser excitation and detection) were kept constant. FIJI/ImageJ (version 1.53) and Zeiss ZEN software (version 2.3) were used for image acquisition and processing.
Endothelial density analysis was performed with the FIJI/ImageJ plugin Vessel analysis. The number of tip cells was quantified by ESM1/ERG double-positive cells, normalized to the length of the angiogenic front and defined by a line at the base of the tip cells. Filopodia per vessel length were analyzed by the number of filopodia normalized to the length of the angiogenic front, defined by a line at the base of the tip cells per field. Vessel (artery and vein) length was normalized to radial length. EC proliferation was measured by the number of EdU/ERG double-positive cells to the total number of ERG-positive cells per field.
The ‘n’ number stands for the number of retinas characterized, which were derived from at least two litters. Individual dots represent measurement data from images taken from different leaflets of each retina.
HUVECs from pooled donors were purchased from Lonza (CC-2519) or PromoCell (C-12208) and cultured in complete EGM-MV2 medium (PromoCell, C-22022) supplemented with 1× penicillin–streptomycin. Human embryonic kidney (HEK293T) cells were purchased from Life Technologies (Thermo Fisher Scientific, R70007) and cultured in DMEM high-glucose GlutaMAX (Gibco, 61965059) supplemented with 10% FBS, 1× penicillin–streptomycin and 1× sodium pyruvate.
siRNA-mediated knockdown of RNF20 was performed by transfection of ON-TARGETplus Human RNF20 siRNA (Dharmacon/Horizon, L-007027-00-0005) or ON-TARGETplus Non-targeting Control siRNA as a control, using Lipofectamin RNAiMAX (Thermo Fisher Scientific, 13778075) according to the manufacturer’s instructions with the following modifications: 30 nM siRNA (final concentration) was transfected with 2 µl of Lipofectamine RNAiMAX reagent in 1 ml of complete EGM-MV2 medium (final volume 1.4 ml) per well of a six-well plate overnight. The next day, cells were supplemented with fresh medium, followed by starvation in EGM-MV2 basal medium supplemented with 2% heat-inactivated FBS for 6–8 h before harvesting them at 60–72 h after transfection, for either RNA or protein analysis. For knockdown of ERG, STAT3 or VEGFR2, 1 nM or 15 nM siERG as indicated in the figure legends (siRNA1:CAGAUCCUACGCUAUGGAGUA, siRNA2:ACUCUCCACGGUUAAUGCAUGGUAG mixed 1:1, Sigma-Aldrich) siRNA, 7.5 nM STAT3 (CACAUGCCACUUUGGUGUUUCAUAA) siRNA or 30 nM VEGFR2 siRNA (Dharmacon/Horizon, L-003148-00-005) was used, respectively.
For ERG overexpression in HUVECs, HEK293T cells were transfected with 70% confluency in six-well plates with control plasmid or ERG overexpression plasmid (TRCN0000465638), obtained from the TRC3puro overexpressing library, along with 1 µg of CMVΔR8.74 packaging plasmid and 0.5 µg of VGV.G envelope plasmids using X-tremeGENE DNA transfection reagent (Roche, 6366236001) in 1 ml of complete EGM-MV2 medium (final volume 1.2 ml). The next day, an additional 1 ml of medium was added to the cells. Forty-eight hours after transfection, the viral supernatant was collected, and 100–250 µl of virus was used to transduce HUVECs at 70% confluency in six-well plates overnight in the presence of 8 µg ml−1 polybrene (Sigma-Aldrich, TR-1003-G). Seventy-two hours after transduction, HUVECs were harvested for RNA and protein analysis or used for the spheroid sprouting assay.
The spheroid sprouting assay was performed according to ref. 55 with the following modifications. Twenty-four hours after transfection of HUVECs, spheroids were formed in growth medium containing 20% methylcellulose solution (1.2% (w/v) methylcellulose in EGM basal medium) with the hanging drop method for the next 24 h. For embedding the spheroids in collagen gels, a 2.8 mg ml−1 collagen solution containing Rat Tail Collagen Type I (Corning, 354249), 10× M199 medium (Gibco, 11825015) and NaOH in a ratio 8:1:1 was mixed 1:1 with the methylcellulose solution containing 20% heat-inactivated FBS to obtain a final collagen concentration of 1.4 mg ml−1. Then, 1 ml of the prepared collagen/methylcellulose solution was mixed with approximately 60 spheroids containing the test substances and immediately transferred to a well in a pre-heated 24-well plate to polymerize for 30 min at 37 °C. After polymerization, 100 µl of basal medium was added on top of the collagen gels, and the spheroids were incubated for 24 h at 37 °C (5% CO2, 100% humidity). After fixation with 4% PFA, images were acquired with a Leica DMI8 microscope (×10 objective magnification), and the length of the sprouts was measured with FIJI/ImageJ software (calculated as cumulative sprout length). At least 10 spheroids were analyzed for each group.
For whole-cell protein isolation, cells were collected and lysed in RIPA buffer supplemented with protease and phosphatase inhibitor cocktail (Millipore, 535142 and 524632). Protein concentration was quantified using a Pierce BCA protein assay kit (Pierce Biotechnology, 23225).
For protein fractionation, cells were lysed on the plate in lysis buffer (10 mM Tris-HCl (pH 8.0), 10 mM NaCl, 0.2% NP-40, 1× protease inhibitor; approximately 50 µl per 1 × 106 cells) by incubation for 15 min on ice. After centrifugation at 2,500g for 5 min at 4 °C, the supernatant containing the cytoplasmic fraction was collected. The cell pellet was washed another three times with lysis buffer and re-suspended in NEB buffer (20 mM HEPES (pH 7.5), 400 mM NaCl, 1 mM EDTA, 1 mM DTT, 1 mM PMSF, 1× protease inhibitor; 25 µl per 1×1 06 cells), followed by 30-min incubation on ice. Next, the nuclei were centrifuged at 5,000g (4 °C), and the supernatant containing the soluble nuclear fraction was collected. The resulting cell pellet was washed again in NEB buffer, re-supended in CEB buffer (NEB buffer, 5 mM CaCl2, 150 U of MNase/1 × 106 cells; 25 µl per 1 × 106 cells) and incubated for 10 min at 37 °C to obtain the chromatin-bound fraction.
After separation via SDS-PAGE, proteins were transferred to nitrocellulose membranes (GE Healthcare, 10600002), blocked in 5% skim milk in PBST (0.1% Tween 20 in PBS) and incubated with the appropriate primary antibody followed by incubation with HRP-coupled secondary antibodies as indicated in Extended Data Table 1. Images were taken with an Amersham Imager 600 (GE Healthcare Life Sciences).
RNA was isolated using TRIzol RNA isolation reagent (Invitrogen, 15596018). For real-time PCR analysis, cDNA was synthesized with a High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, 4368813). Real-time PCR was performed using SYBR Green PCR Master Mix (Applied Biosystems, A25742) or qPCRBIO SyGreen Blue Hi-ROX (Nippon Genetics). Cycle numbers were normalized to those of α-tubulin (Tuba1a). Primers used for qPCR analysis are listed in Extended Data Table 2.
P7 retinas were dissected and digested with a collagenase enzyme solution (1 mg ml−1 Collagenase I (Worthington, LS004196), 0.1 mg ml−1 DNAse I (Merck, 10104159001) in HBSS) for 1 h at 37 °C followed by filtering the cell suspension through a 70-µm cell strainer with basal medium (DMEM high glucose, 20% FBS, 1× penicillin–streptomycin, 1× L-glutamin, 25 mM HEPES). After centrifugation and washing, cells were stained with a CD31-APC coupled antibody for 30 min and sorted with a BD FACSAria IIu (BD Biosciences) for living, CD31-positive cells. Total RNA was isolated using an RNeasy Universal Mini Kit (Qiagen, 73404). The integrity of the RNA was assessed on a Bioanalyzer 2100 (Agilent) using an RNA 6000 Pico Kit (Agilent, 5067-1513). Low-input RNA library preparation (smart-seq2) and sequencing were performed on a BGISEQ-500 platform.
For RNA-seq of control and RNF20-deficient HUVECs, total RNA of siRNA-transfected HUVECs was isolated 60 h after transfection using the RNeasy Universal Mini Kit. The RNA integrity was assessed on a Bioanalyzer 2100 using an RNA 6000 Nano Kit (Agilent, 5067-1511), and standard library preparation and sequencing were performed on a BGISEQ-500 platform.
RNA-seq reads were trimmed of adapters using Trimmomatic (version 0.39) and mapped to the mm10 reference genome using STAR56 (version 2.7.3a) (−alignIntronMin 20 −alignIntronMax 500000). Read quality was controlled by the MultiQC tool (version 1.14), and reads were counted using the analyzeRepeats.pl function (rna mm10 –count exons –strand both –noadj) from HOMER (version 4.11.1) after creating the tag directories with makeTagDirectory. Differential expression was quantified and normalized using DESeq2 (version 1.40.0). Reads per kilobase per million mapped reads (RPKM) was determined using rpkm.default from EdgeR (version 3.42.4). Differentially regulated genes were filtered as follows (retinal EC: log2 fold change ≤ −1, ≥1; P < 0.05; HUVECs: log2 fold change ≤ −0.58, ≥0.58; P < 0.05). GO and pathway analysis were performed using Metascape (version 3.5)57. Heatmaps were created using d3Heatmap (version 0.9.0) (http://heatmapper.ca/expression/). All heatmaps represent the row-based z-scores calculated from trimmed mean of M-values (TMM). The principal component analysis (PCA) plots were obtained using prcomp function from the stats package into a custom R script (version 4.3.0), and volcano plots were obtained using the EnhancedVolcano R package (version 1.18.0). All data, including publicly available data, were normalized with the same parameters.
Differential spliced genes were detected by using the DEXSeq58 R package (version 1.40.0) with P < 0.05 and log2 fold change ≤ −1, ≥1.
At P7, mice were euthanized and retina were dissected, followed by digestion with enzyme solution (1 mg ml−1 Collagenase I (Worthington, LS004196), 0.1 mg ml−1 DNAse I (Merck, 10104159001) in HBSS) for 45 min at 37 °C with agitation. The digestion was stopped by adding basal medium (DMEM high glucose, 20% FBS, 1× penicillin–streptomycin, 1× L-glutamin, 25 mM HEPES) and filtering the cell suspension through a 70-µm cell strainer. After centrifugation, the cell pellet was resuspended in 1 ml of basal medium and bound with CD31-coupled magnetic Dynabeads (Thermo Fisher Scientific, 11035) for 30 min at room temperature with agitation. After five washes, ECs were counted, and 5,000 cells were used for ATAC-seq protocol as described in ref. 59 with the following modifications. For 5,000 cells, the amount of the Tn5 enzyme from the Illumina tagmentation kit (cat. no. 20034211) was scaled down to 1 µl per 50-µl reaction.
ATAC-seq raw reads processing, mapping and peak calling were done according to the ENCODE ATAC-seq pipeline. Differential ATAC peaks were detected using the Diffbind R package31 (version 3.4.11). Differentially ATAC peaks were filtered as follows (log2 fold change ≤ −0.58, ≥0.58; P < 0.05). Footprinting and TF network analysis were done using the TOBIAS footprinting package31 (version 0.14.0), and the results were plotted using the EnhancedVolcano R package (version 1.18.0). TF network analysis was performed with Cytoscape (version 3.6.1) on STAT3, ERG and Fosl2 target gene networks detected in the footprinting analysis, which showed increased binding score (log2 fold change > 0.58), upregulation and involvement in the angiogenesis-related pathways.
For H2Bub ChIP, 500,000 control and RNF20 siRNA-transfected HUVECs were harvested and fixed for 10 min in 1% formaldehyde per each replicate. Fixation was stopped by adding glycine to a final concentration of 0.125 M. Cells were then washed twice in cold PBS and lysed in Lysis Buffer I (10 mM Tris HCl (pH 8.0), 10 mM NaCl, 0.5% NP-40) (50 µl per 500,000 cells) for 15 min at 4 °C while rotating.
For ChIP of RNA Pol II, ERG and NICD, siRNA-transfected HUVEC cells in one full six-well plate were fixed on the plate per each replicate. Fixation was performed for 10 min in 1% formaldehyde with gentle rotation at room temperature and then stopped by adding 0.125 M glycine for 5 min, followed by three washes with ice-cold PBS. Cells were harvested in Lysis Buffer I (10 mM Tris (pH 8.0), 10 mM NaCl, 0,5% NP-40, 1× protease inhibitor, 1× phosphatase inhibitor) and harvested by scraping.
For all ChIP samples, nuclei were spun down at 2,500g for 5 min and resuspended in Lysis Buffer II (50 mM Tris (pH 8.0), 10 mM EDTA, 0,5% SDS, 1× protease inhibitor, 1× phosphatase inhibitor) at a concentration of 1 million cells per 40 µl. Chromatin was sheared with Covaris in Lysis Buffer II containing 0.1% SDS to obtain fragments between 200 bp and 400 bp. After centrifugation at 16,000g for 15 min, the supernatant was transferred into a new tube and diluted with the same volume of 2× immunoprecipitation (IP) buffer (100 mM Tris (pH 8.0), 2 mM EDTA, 300 mM NaCl, 2% Triton X-100, 0.2% SDS, 0.2% sodium deoxycholate monohydrate, 1× protease inhibitor, 1× phosphatase inhibitor). The samples were pre-cleared with 100 µl of Sepharose Fast Flow 4 beads for 1.5 h at 4 °C under gentle rotation and incubated with 1 µg of total Pol II (Rpb1 NTD, 14958S, Cell Signaling Technology; 1 µg/IP), 0.5 µg of H2Bub1 (5546, Cell Signaling Technology; 0.5 µg/IP), 1 µg of ERG (ab92513, Abcam; 1 µg/IP) or 1 µg of NICD (4147, Cell Signaling Technology; 1 µg/IP) overnight at 4 °C with rotation. The next day, 50 µl of Sepharose Fast Flow 4 beads were added and incubated for two additional hours at 4 °C. The samples were then washed two times with low-salt buffer (20 mM Tris (pH 8.0), 2 mM EDTA, 150 mM NaCl, 1% Triton X-100, 0.1% SDS, 1× protease inhibitor, 1× phosphatase inhibitor), one time with high-salt buffer (20 mM Tris (pH 8.0), 2 mM EDTA, 500 mM NaCl, 1% Triton X-100, 0.1% SDS, 1× protease inhibitor, 1× phosphatase inhibitor), one time with LiCl wash buffer (10 mM Tris (pH 8.0), 1 mM EDTA, 250 mM LiCl, 1% NP40, 1% sodium deoxycholate monohydrate, 1× protease inhibitor, 1× phosphatase inhibitor) and two times with TE buffer (10 mM Tris (pH 8.0), 1 mM EDTA, 1× protease inhibitor, 1× phosphatase inhibitor). Samples were eluted in 200 µl of elution buffer (0.6% SDS, 10 mM Tris (pH 8.0), 0.5 mM EDTA, 300 mM NaCl) and treated with RNase A, followed by proteinase K treatment and reverse crosslinking overnight at 65 °C. The DNA was purified with a Zymo kit (D5205).
For ChIP–qPCR, 0.1 ng of DNA was used per reaction.
For ChIP-seq, libraries were prepared with a NEBNext Ultra II DNA Library Prep Kit for Illumina (New England Biolabs, E7645S/L and E7103S/L) according to the manufacturer’s instructions.
ChIP-seq reads were trimmed using Trimmomatic (version 0.39), minimum length of 60 bp and a quality score of a minimum of 15. Trimmed reads were mapped to mouse genome mm10 (UCSC assembly) with using Bowtie2 (version 2.4.4) (default settings). SAMtools view (version 1.7.) was used to convert SAM files to BAM format. PCR artifacts were removed by the MarkDuplicates.jar function from Picard (version 1.119). Individual BAM-mapped files were merged using BamTools (version 2.5.1) merge (default settings). Merged BAM files were converted to BigWig format using BamCoverage from deepTools (version 3.5.1) (-b 20 -smooth 40 --normalizeUsing RPKM -e 150). Peaks were called using MACS2 (version 2.2.7.1) (callpeak, --broad --nomodel -g mmu -p 0.01 --shift −75 --extsize 150 --keep-dup all). Peaks overlapping the blacklist defined by ENCODE were removed. ERG-ChIP-seq enrichment of Rnf20 deregulated genes was performed using plotProfile from deepTools. ngs.plot (version 2.41.4) was used to plot the H2Bub-ChIP-seq enrichment genome wide and at Notch targets.
PI was calculated by dividing the normalized count per million reads on the TSS area (−50 bp to 300 bp) by those on the gene body plus 3 kb after the TTS. For the calculation, the GitHub repository code (https://github.com/MiMiroot/PIC) was used with settings (--gtf hg38.gtf --TSSup 50 --TSSdown 300 --GBdown 3000) and Ensembl hg38, version 108. Highly paused genes were defined as the genes with PI > quartile 3 (Q3) in the control samples. P values were calculated using the Wilcoxon test from the library (rstatix) (version 0.7.2) from the R language.
To calculate differential pausing, we used the t-test function from the rstatix package on PI values from replicates of control and siRNF20 or ERG-overexpressing HUVECs. Genes with decreased PI were identified as those with a log2 PI < 0.5 and a false discovery rate (FDR) < 0.25. All scripts for processing the pausing matrices are available in the repository at https://github.com/jcorderJC12/02PAUSING_P2t.
All experiments were performed at least three independent times in biological replicates, and the respective data were used for statistical analyses.
For animal experiments, mice (C57BL/6J) were grouped into control (C57BL/6_PdgfbiCre2neg_Rnf20tm1c(EUCOMM)Wtsiflox/flox) and Rnf20iECKO (C57BL/6_PdgfbiCre2pos_Rnf20tm1c(EUCOMM)Wtsiflox/flox) cohorts, irrespective of their sex. The data presented were obtained from retinas isolated from individual mice. Quantifications in Figs. 1d,f,h,j,l,n,p,s and 7h,i and Extended Data Figs. 1e,f,h,i,j and 6f,h were performed on mice at P7, whereas quantifications in Fig. 2f were conducted on mice at P28. Differences between groups were assessed using an unpaired two-tailed Student’s t-test or ANOVA multiple comparisons test as indicated in the figure legends. For all tests, P < 0.05 was considered statistically significant. Bar plots and box plots were produced with GraphPad Prism 10 software. Numeric P values are shown within the figures.
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.
Sequencing data generated in this study have been deposited in the Gene Expression Omnibus (accession number GSE212524). Processed data are included in the source data for each figure. Single-cell data from neonatal mouse retinas (GSE175895), ChIP-seq of ERG (GSE124893) and ETS1 (GSM2442778) in HUVECs were retrieved from previously published studies. Figures that have associated raw data include Figs. 3a–d,f–k, 4a–c,f, 5h,i,k and 6a–g and Extended Data Figs. 2a–g,i,j, 3a,b, 4a–d,g,h and 5a–c. Source data are provided with this paper.
Blanco, R. & Gerhardt, H. VEGF and Notch in tip and stalk cell selection. Cold Spring Harb. Perspect. Med. 3, a006569 (2013).
Article PubMed PubMed Central Google Scholar
Eilken, H. M. & Adams, R. H. Dynamics of endothelial cell behavior in sprouting angiogenesis. Curr. Opin. Cell Biol. 22, 617–625 (2010).
Article CAS PubMed Google Scholar
Hasan, S. S. et al. Endothelial Notch signalling limits angiogenesis via control of artery formation. Nat. Cell Biol. 19, 928–940 (2017).
Article CAS PubMed PubMed Central Google Scholar
Hellstrom, M. et al. Dll4 signalling through Notch1 regulates formation of tip cells during angiogenesis. Nature 445, 776–780 (2007).
Jakobsson, L. et al. Endothelial cells dynamically compete for the tip cell position during angiogenic sprouting. Nat. Cell Biol. 12, 943–953 (2010).
Article CAS PubMed Google Scholar
Lee, H. W. et al. Role of venous endothelial cells in developmental and pathologic angiogenesis. Circulation 144, 1308–1322 (2021).
Article CAS PubMed PubMed Central Google Scholar
Pitulescu, M. E. et al. Dll4 and Notch signalling couples sprouting angiogenesis and artery formation. Nat. Cell Biol. 19, 915–927 (2017).
Article CAS PubMed Google Scholar
Pontes-Quero, S. et al. High mitogenic stimulation arrests angiogenesis. Nat. Commun. 10, 2016 (2019).
Article PubMed PubMed Central Google Scholar
Potente, M. & Makinen, T. Vascular heterogeneity and specialization in development and disease. Nat. Rev. Mol. Cell Biol. 18, 477–494 (2017).
Article CAS PubMed Google Scholar
Stewen, J. et al. Eph-ephrin signaling couples endothelial cell sorting and arterial specification. Nat. Commun. 15, 2539 (2024).
Article CAS PubMed PubMed Central Google Scholar
Xu, C. et al. Arteries are formed by vein-derived endothelial tip cells. Nat. Commun. 5, 5758 (2014).
Article CAS PubMed Google Scholar
Udan, R. S., Vadakkan, T. J. & Dickinson, M. E. Dynamic responses of endothelial cells to changes in blood flow during vascular remodeling of the mouse yolk sac. Development 140, 4041–4050 (2013).
Article CAS PubMed PubMed Central Google Scholar
Hargreaves, D. C., Horng, T. & Medzhitov, R. Control of inducible gene expression by signal-dependent transcriptional elongation. Cell 138, 129–145 (2009).
Article CAS PubMed PubMed Central Google Scholar
Levine, M. Paused RNA polymerase II as a developmental checkpoint. Cell 145, 502–511 (2011).
Article CAS PubMed PubMed Central Google Scholar
Liu, X., Kraus, W. L. & Bai, X. Ready, pause, go: regulation of RNA polymerase II pausing and release by cellular signaling pathways. Trends Biochem. Sci. 40, 516–525 (2015).
Article PubMed PubMed Central Google Scholar
Adelman, K. & Lis, J. T. Promoter-proximal pausing of RNA polymerase II: emerging roles in metazoans. Nat. Rev. Genet. 13, 720–731 (2012).
Article CAS PubMed PubMed Central Google Scholar
Galbraith, M. D. et al. HIF1A employs CDK8-mediator to stimulate RNAPII elongation in response to hypoxia. Cell 153, 1327–1339 (2013).
Article CAS PubMed PubMed Central Google Scholar
Day, D. S. et al. Comprehensive analysis of promoter-proximal RNA polymerase II pausing across mammalian cell types. Genome Biol. 17, 120 (2016).
Article PubMed PubMed Central Google Scholar
Kim, J. et al. RAD6-mediated transcription-coupled H2B ubiquitylation directly stimulates H3K4 methylation in human cells. Cell 137, 459–471 (2009).
Article CAS PubMed PubMed Central Google Scholar
Kim, J., Hake, S. B. & Roeder, R. G. The human homolog of yeast BRE1 functions as a transcriptional coactivator through direct activator interactions. Mol. Cell 20, 759–770 (2005).
Article CAS PubMed Google Scholar
Pavri, R. et al. Histone H2B monoubiquitination functions cooperatively with FACT to regulate elongation by RNA polymerase II. Cell 125, 703–717 (2006).
Article CAS PubMed Google Scholar
Shema, E., Kim, J., Roeder, R. G. & Oren, M. RNF20 inhibits TFIIS-facilitated transcriptional elongation to suppress pro-oncogenic gene expression. Mol. Cell 42, 477–488 (2011).
Article CAS PubMed PubMed Central Google Scholar
Zaidi, S. et al. De novo mutations in histone-modifying genes in congenital heart disease. Nature 498, 220–223 (2013).
Article CAS PubMed PubMed Central Google Scholar
Zhang, M. et al. Endothelial cells regulated by RNF20 orchestrate the proliferation and differentiation of neural precursor cells during embryonic development. Cell Rep. 40, 111350 (2022).
Article CAS PubMed Google Scholar
Zarkada, G. et al. Specialized endothelial tip cells guide neuroretina vascularization and blood-retina-barrier formation. Dev. Cell 56, 2237–2251 (2021).
Article CAS PubMed PubMed Central Google Scholar
del Toro, R. et al. Identification and functional analysis of endothelial tip cell-enriched genes. Blood 116, 4025–4033 (2010).
Article PubMed PubMed Central Google Scholar
Brash, J. T. et al. Tamoxifen-activated CreERT impairs retinal angiogenesis independently of gene deletion. Circ. Res. 127, 849–850 (2020).
Article CAS PubMed PubMed Central Google Scholar
Bentley, D. L. Coupling mRNA processing with transcription in time and space. Nat. Rev. Genet. 15, 163–175 (2014).
Article CAS PubMed PubMed Central Google Scholar
Moyal, L. et al. Requirement of ATM-dependent monoubiquitylation of histone H2B for timely repair of DNA double-strand breaks. Mol. Cell 41, 529–542 (2011).
Article CAS PubMed PubMed Central Google Scholar
Nakamura, K. et al. Regulation of homologous recombination by RNF20-dependent H2B ubiquitination. Mol. Cell 41, 515–528 (2011).
Article CAS PubMed Google Scholar
Bentsen, M. et al. ATAC-seq footprinting unravels kinetics of transcription factor binding during zygotic genome activation. Nat. Commun. 11, 4267 (2020).
Article CAS PubMed PubMed Central Google Scholar
Bray, S., Musisi, H. & Bienz, M. Bre1 is required for Notch signaling and histone modification. Dev. Cell 8, 279–286 (2005).
Article CAS PubMed Google Scholar
Birdsey, G. M. et al. The endothelial transcription factor ERG promotes vascular stability and growth through Wnt/β-catenin signaling. Dev. Cell 32, 82–96 (2015).
Article CAS PubMed Central Google Scholar
Kalna, V. et al. The transcription factor ERG regulates super-enhancers associated with an endothelial-specific gene expression program. Circ. Res. 124, 1337–1349 (2019).
Article CAS PubMed PubMed Central Google Scholar
Shah, A. V. et al. The endothelial transcription factor ERG mediates Angiopoietin-1-dependent control of Notch signalling and vascular stability. Nat. Commun. 8, 16002 (2017).
Article CAS PubMed PubMed Central Google Scholar
Fish, J. E. et al. Dynamic regulation of VEGF-inducible genes by an ERK/ERG/p300 transcriptional network. Development 144, 2428–2444 (2017).
CAS PubMed PubMed Central Google Scholar
Sissaoui, S. et al. Genomic characterization of endothelial enhancers reveals a multifunctional role for NR2F2 in regulation of arteriovenous gene expression. Circ. Res. 126, 875–888 (2020).
Article CAS PubMed PubMed Central Google Scholar
Chen, J. et al. VEGF amplifies transcription through ETS1 acetylation to enable angiogenesis. Nat. Commun. 8, 383 (2017).
Article CAS PubMed PubMed Central Google Scholar
Muniz, L., Nicolas, E. & Trouche, D. RNA polymerase II speed: a key player in controlling and adapting transcriptome composition. EMBO J. 40, e105740 (2021).
Article CAS PubMed PubMed Central Google Scholar
Fuchs, G., Hollander, D., Voichek, Y., Ast, G. & Oren, M. Cotranscriptional histone H2B monoubiquitylation is tightly coupled with RNA polymerase II elongation rate. Genome Res. 24, 1572–1583 (2014).
Article CAS PubMed PubMed Central Google Scholar
Mineur, P. et al. Newly identified biologically active and proteolysis-resistant VEGF-A isoform VEGF111 is induced by genotoxic agents. J. Cell Biol. 179, 1261–1273 (2007).
Article CAS PubMed PubMed Central Google Scholar
Olsson, A. K., Dimberg, A., Kreuger, J. & Claesson-Welsh, L. VEGF receptor signalling—in control of vascular function. Nat. Rev. Mol. Cell Biol. 7, 359–371 (2006).
Article CAS PubMed Google Scholar
Simons, M., Gordon, E. & Claesson-Welsh, L. Mechanisms and regulation of endothelial VEGF receptor signalling. Nat. Rev. Mol. Cell Biol. 17, 611–625 (2016).
Article CAS PubMed Google Scholar
Munoz, M. J. et al. DNA damage regulates alternative splicing through inhibition of RNA polymerase II elongation. Cell 137, 708–720 (2009).
Article CAS PubMed Google Scholar
Delcombel, R. et al. New prospects in the roles of the C-terminal domains of VEGF-A and their cooperation for ligand binding, cellular signaling and vessels formation. Angiogenesis 16, 353–371 (2013).
Article CAS PubMed Google Scholar
Ong, C. T. & Corces, V. G. Enhancer function: new insights into the regulation of tissue-specific gene expression. Nat. Rev. Genet. 12, 283–293 (2011).
Article CAS PubMed PubMed Central Google Scholar
Krebs, L. T., Deftos, M. L., Bevan, M. J. & Gridley, T. The Nrarp gene encodes an ankyrin-repeat protein that is transcriptionally regulated by the Notch signaling pathway. Dev. Biol. 238, 110–119 (2001).
Article CAS PubMed Google Scholar
Lamar, E. et al. Nrarp is a novel intracellular component of the Notch signaling pathway. Genes Dev. 15, 1885–1899 (2001).
Article CAS PubMed PubMed Central Google Scholar
Phng, L. K. et al. Nrarp coordinates endothelial Notch and Wnt signaling to control vessel density in angiogenesis. Dev. Cell 16, 70–82 (2009).
Article CAS PubMed PubMed Central Google Scholar
Fischer, A. & Gessler, M. Delta–Notch—and then? Protein interactions and proposed modes of repression by Hes and Hey bHLH factors. Nucleic Acids Res. 35, 4583–4596 (2007).
Article CAS PubMed PubMed Central Google Scholar
Nakayama, M. et al. Spatial regulation of VEGF receptor endocytosis in angiogenesis. Nat. Cell Biol. 15, 249–260 (2013).
Article CAS PubMed PubMed Central Google Scholar
Sethi, G., Shanmugam, M. K., Arfuso, F. & Kumar, A. P. Role of RNF20 in cancer development and progression - a comprehensive review. Biosci. Rep. 38, BSR20171287 (2018).
Article PubMed PubMed Central Google Scholar
Claxton, S. et al. Efficient, inducible Cre-recombinase activation in vascular endothelium. Genesis 46, 74–80 (2008).
Article CAS PubMed Google Scholar
Pitulescu, M. E., Schmidt, I., Benedito, R. & Adams, R. H. Inducible gene targeting in the neonatal vasculature and analysis of retinal angiogenesis in mice. Nat. Protoc. 5, 1518–1534 (2010).
Article CAS PubMed Google Scholar
Heiss, M. et al. Endothelial cell spheroids as a versatile tool to study angiogenesis in vitro. FASEB J. 29, 3076–3084 (2015).
Article CAS PubMed Google Scholar
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Article CAS PubMed Google Scholar
Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat. Commun. 10, 1523 (2019).
Article PubMed PubMed Central Google Scholar
Anders, S., Reyes, A. & Huber, W. Detecting differential usage of exons from RNA-seq data. Genome Res. 22, 2008–2017 (2012).
Article CAS PubMed PubMed Central Google Scholar
Buenrostro, J. D., Wu, B., Chang, H. Y. & Greenleaf, W. J. ATAC-seq: a method for assaying chromatin accessibility genome-wide. Curr. Protoc. Mol. Biol. 109, 21.29.21–21.29.29 (2015).
We would like to thank S. Uhlig, the FlowCore Mannheim and the Core Facility Preclinical Models for excellent support. G.D. was supported by the Collaborative Research Center 1366 (project A03), funded by the German Research Foundation (DFG); the German Centre for Cardiovascular Research (DZHK) (81Z0500202 and 81X2500216), funded by the Federal Ministry of Education and Research (BMBF); and the Baden‐Württemberg Foundation special program ‘Angioformatics Single Cell Platform’. M.P. was supported by a European Research Council Consolidator Grant (EMERGE-773047), the DFG (project number 456687919 – SFB 1531) and a Leducq Foundation grant.
Department of Cardiovascular Genomics and Epigenomics, European Center for Angioscience (ECAS), Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
Nalan Tetik-Elsherbiny, Adel Elsherbiny, Aadhyaa Setya, Yongqin Tang, Yanliang Dou, Heike Serke, Julio Cordero & Gergana Dobreva
Cardiovascular Pharmacology, European Center for Angioscience, Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
Johannes Gahn, Purnima Gupta & Roxana Ola
German Centre for Cardiovascular Research (DZHK), Mannheim, Germany
Heike Serke, Thomas Wieland, Joerg Heineke, Michael Potente, Julio Cordero, Roxana Ola & Gergana Dobreva
Experimental Pharmacology, European Center for Angioscience (ECAS), Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
Research Center, CHU St. Justine, Montreal, Quebec, Canada
Department of Cardiovascular Physiology, European Center for Angioscience (ECAS), Medical Faculty Mannheim, Heidelberg University, Mannheim, Germany
Angiogenesis & Metabolism Laboratory, Center of Vascular Biomedicine, Berlin Institute of Health at Charité—Universitätsmedizin Berlin, Berlin, Germany
Max Delbrück Center for Molecular Medicine in the Helmholtz Association, Berlin, Germany
Helmholtz-Institute for Translational AngioCardioScience (HI-TAC) of the Max Delbrück Center for Molecular Medicine in the Helmholtz Association (MDC) at Heidelberg University, Heidelberg, Germany
Michael Potente & Gergana Dobreva
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
You can also search for this author in PubMed Google Scholar
N.T.-E., A.E., A.S., J.G., Y.T. and P.G. performed the experiments. N.T.-E., A.E., A.D., Y.D. and J.C. performed the bioinformatic analysis. M.P., A.D., H.S., T.W. and J.H. provided reagents and valuable intellectual input. N.T.-E., J.C., R.O. and G.D. designed the experiments, analyzed the data and wrote the paper. All authors discussed the results and commented on the paper.
Correspondence to Julio Cordero, Roxana Ola or Gergana Dobreva.
The authors declare no competing interests.
Nature Cardiovascular Research thanks the anonymous reviewer(s) for their contribution to the peer review of this work.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
a Dot plot of Rnf20 expression level and frequency in EC from P6 and P10 retinas25. b Schematic diagram of the Rnf20 targeting construct and the crosses used for the generation of inducible (floxed) allele in ECs. c, d mRNA expression of Rnf20 (c) and WB using anti-RNF20 antibody and VE-cadherin (Cdh5) as an EC loading control (d) of ECs isolated from lungs of control and Rnf20iEC-KO mice. n = 4 Control; n = 6 Rnf20iEC-KO. e Body weight of control and Rnf20iEC-KO mice. Dots represent individual pups (n = 30 control, n = 36 Rnf20iECKO). f Quantification of vessel length normalized to radial length (μm); n = 10 control and 10 Rnf20iEC-KO retinas; n = 35 Control arteries, n = 32 Rnf20iEC-KO arteries, n = 32 Control veins; n = 30 Rnf20iEC-KO veins. g Immunostaining for ESM1, ERG and CD31 showing excessive tip cell formation in Rnf20iEC-KO. h Quantification of the radial length in Pdgfb-CreERT2negRnf20+/fl (n = 4); Pdgfb-CreERT2posRnf20+/fl (n = 4) and Pdgfb-CreERT2posRnf20fl/fl (n = 3) retinas; each data point represents one leaflet. i Quantification of the length of the vascular front in Pdgfb-CreERT2neg and Pdgfb-CreERT2pos retinas (n = 9 for each group). j Quantification of the length of arteries and veins normalized to radial length in Pdgfb-CreERT2neg and Pdgfb-CreERT2pos retinas (each data point represents one leaflet of n = 8 retinas for each group). k Immunostaining for ESM1 and Pecam1 of Pdgfb-CreERT2neg and Pdgfb-CreERT2pos retinas. Data in e, f, i and j represent mean ± SEM; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Data in h represent mean ± SEM; differences between groups were assessed using one-way ANOVA multiple comparisons test. ns: not significant. Numeric p-values are shown within the figures.
a Principal component analysis (PCA) of genome-wide gene expression variation of sorted control and Rnf20iEC-KO retinal ECs (n = 3). b Correlation heatmap of RPKM normalized RNA-seq reads in control and Rnf20iEC-KO retinal ECs. c Principal component analysis (PCA) of genome-wide gene expression variation of HUVECs transfected with control or RNF20 siRNAs for 60 h (n = 3). d Correlation heatmap of RPKM normalized RNA-seq reads in control and RNF20 knockdown HUVECs. e Volcano plot showing the distribution of differentially expressed genes in HUVECs transfected with control and RNF20 siRNAs. n = 3; log2 fold change ≤ −0.58, ≥0.58; p-value < 0.05. Differential expression was performed with Deseq2 (v1.40.1). f Heatmap representation of cell cycle related genes downregulated in HUVECs upon RNF20 loss. g qPCR validation of RNF20-dependent transcriptional alterations in HUVECs transfected with control and RNF20 siRNA (n = 12). h Principal component analysis (PCA) of genome-wide Pol II binding variation in HUVECs transfected with control and RNF20 siRNAs (n = 3 independent ChIP-Seq experiments). i Genome tracks of merged total Pol II ChIP-Seq reads at CCN1, SRSF1, GADD45A and JUNB loci. Data in g represent mean ± SEM; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Numeric p-values are shown within the figures.
a Genome tracks of merged RNA-Seq reads at the Ezh2, Vegfr3 (Flt4), Vegfc and Bmpr2 gene loci in control and Rnf20iEC-KO retina ECs, showing alternative splicing in Rnf20iEC-KO retina ECs. b Top GO terms of DSG (DSG, p ≤ 0.05, log2 fold change ≤ −1, ≥1, n = 3) with no change in expression levels in Rnf20iEC-KO versus control retina ECs. GO term enrichment analysis was performed on differentially spliced genes using Metascape (v.3.5). The bars represent the -log10(p-value) for each enriched GO term. c Western blot analysis for γH2AX, a sensitive DNA damage marker, in chromatin bound fractions from HUVECs transfected with control siRNA and siRNA against RNF20. H2A served as a loading control.
a PCA of genome-wide gene chromatin accessibility variation of sorted control and Rnf20iEC-KO retina ECs (n = 2). b, c Venn diagram showing overlap between genes with decreased chromatin accessibility and decreased expression in Rnf20iEC-KO retinal ECs (b) and GO analysis of genes within the overlap (c). GO term enrichment analysis was performed on differentially spliced genes using Metascape (v.3.5). The bars represent the -log10(p-value) for each enriched GO term. d Motifs of transcription factors enriched in control (left) or in Rnf20iEC-KO retina ECs (right). e Quantification of N1ICD levels in three independent preparations of chromatin bound fractions of control and RNF20 siRNA transfected HUVECs (n = 3 biological replicates for each group). H2B served as a loading control. f Average H2Bub1 ChIP-Seq profile genome-wide and at Notch targets in HUVECs transfected with control siRNA and siRNA against RNF20; the line plots represent mean values, with shaded areas representing the ± SEM across n = 3 biological replicates for each froup. g Transcription factor-gene regulatory network build on the basis of ERG, STAT3 and FOSL2 footprinting analysis. h Motifs of transcription factors enriched in both HUVECs transfected with RNF20 siRNAs versus control siRNA and in Rnf20iEC-KO versus control retina ECs. Data in e represent mean ± SEM; differences between groups were assessed using an unpaired two-tailed Student’s t-test.
a Genome tracks of ERG (light green), ETS1 (grey) ChIP-Seq reads and merged total Pol II ChIP-Seq reads (n = 3) in HUVECs transfected with control (black) or siRNF20 (red). b GO analysis of genes bound only by ERG or only by ETS1 with genes showing decreased PI upon RNF20 loss. c GO analysis of genes bound only by ERG with genes showing decreased PI upon ERG overexpression. d Relative RNF20 mRNA levels in control and ERG OE HUVECs (n = 3 siControl n = 4 siRNF20). e qPCR for DLL4 and NRARP in HUVECs transfected with control, RNF20 siRNAs alone or together with ERG siRNA (left, n = 8) or HUVECs transfected with control, RNF20 siRNAs alone or together with ETS1 siRNA (right, n = 4 for all groups). Data in d represent mean ± SEM; differences between groups were assessed using an unpaired two-tailed Student’s t-test. Data in e represent mean ± SEM; differences between groups were assessed using one-way ANOVA multiple comparisons test. GO term enrichment analysis in b and c were performed on differentially spliced genes using Metascape (v.3.5). The bars represent the -log10(p-value) for each enriched GO term. Numeric p-values are shown within the figures.
a, b WB quantification for pERKThr202/Tyr204, DLL4 and VEGFR3 protein levels normalized to α-tubulin of total protein extracts of HUVECs transfected with control, VEGFR2 and RNF20 siRNA; (a) or of HUVECs transfected with control or RNF20 siRNA and either treated with DMSO or MEK inhibitor PD0325901 (b) for 24 h; n = 4 biological replictaes for control for each group, 4 biological replictaes for Rnf20iEC-KO for each group; figures represent mean ± SEM; statistics were quantified with two-way-ANOVA with Tukey’s correction. c, d Sprouting assay with HUVECs transfected with control or RNF20 siRNA and treated with DMSO, the VEGFR inhibitor Vandetanib (ZD6474) or the MEK inhibitor PD0325901 for 24 h, with VEGFA supplementation (c) and quantification of the cumulative sprout length (d). (n = 8 for siControl, n = 9 siControl+Vandetanib, n = 7 siControl + MEKI, n = 9 for siRNF20, n = 11 for siRNF20+Vandetanib, n = 7 siRNF20 + MEKI). e, g Immunostaining for CD31 in control and Rnf20iEC-KO retinas treated either with PBS or DC101 (e) or oil and the MEK inhibitor SL327 (g). f, h Quantification of filopodia length in control and Rnf20iEC-KO retina ECs treated either with PBS or DC101 (f) or oil and the MEK inhibitor SL327 (h); each data point represents one leaflet of n = 4 control and 4 Rnf20iEC-KO retinas treated with PBS or DC101; n = 4 control+oil, 4 Rnf20iEC-KO+oil, 2 control+SL327, 4 Rnf20iEC-KO + SL327 retinas (h); figures represent mean ± SEM; statistics were quantified with two-way-ANOVA with Tukey’s correction. Numeric p-values are shown within the figures.
In wild-type retina endothelial cells, RNF20 promotes RNA Pol II pausing at highly paused genes involved in shear stress response, cell cycle regulation, mRNA splicing, and VEGFA signaling. Additionally, it interacts with N1ICD to regulate Notch-dependent transcriptional program by monoubiquitination of histone H2B. In retina ECs lacking RNF20 (Rnf20iEC-KO), highly paused genes, including Efnb2, genes related to VEGFA signaling, as well as splicing regulators, are released for active elongation. This release is mediated, at least in part, by elevated ERG levels. Importantly, increased ERG levels can also induce Pol II pause release of highly paused genes by displacing RNF20 from chromatin. Further, RNF20 loss leads to widespread changes in splicing patterns, including splicing of VEGFA, resulting in an increase in the proteolytically stable VEGFA111 isoform. Exacerbated VEGF-VEGFR signaling and decreased Notch signaling, ultimately leads to uncontrolled specification of tip cells and compromised vessel growth upon RNF20 loss. The figure was designed with BioRender.
Statistical source data and unprocessed western blots.
Statisticalsource data and unprocessed western blots.
Statistical source data and unprocessed western blots.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Tetik-Elsherbiny, N., Elsherbiny, A., Setya, A. et al. RNF20-mediated transcriptional pausing and VEGFA splicing orchestrate vessel growth. Nat Cardiovasc Res (2024). https://doi.org/10.1038/s44161-024-00546-5
DOI: https://doi.org/10.1038/s44161-024-00546-5
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Nature Cardiovascular Research (Nat Cardiovasc Res) ISSN 2731-0590 (online)
anti pcsk9 antibody Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.