Fig 2: Peri-portal ERG-deficient mouse fibrogenesis phenotype is SMAD3-dependent. a, b SMAD3 activity in portal tract regions was assess by immunofluorescence for pSMAD3 (white), VWF (green), SMA (red) and DAPI (blue) in ErgcEC-het and ErgiEC-KO mice, respectively, aged 8–10 weeks. Quantification of pSMAD3 area (three fields per mouse, n = 3). SB-431542 (10 mg kg-1, i.p) was administered in ErgcEC-het mice three times a week for 2 weeks. Scale bar 20 µm. c mRNA was isolated from CD31+ murine EC isolated from lung tissue and analysed by qPCR (n = 4). Data were compared to EC from Ergfl/fl littermate controls (*) or to DMSO treated ErgcEC-het (#) by one-way ANOVA and Bonferroni multiple comparison post-test. d Representative images of portal tract VWF (grey scale; green in merge), SMA (grey scale; red in merge), ERG (white), DAPI (blue) and merged panels (Scale bar 50 µm). e Quantification of SMA expression (5 fields per mouse, n = 4). f Quantitative analysis of Picro-sirius red positive tissue, using polarised light (more representative images in Supplementary Fig. 8B). Data were compared to Ergfl/fl littermate controls (*) or to DMSO treated ErgcEC-het (#) by one-way ANOVA and Bonferroni multiple comparison post-test. All graphical data are mean ± s.e.m., * or #P < 0.05, ** or ##P < 0.01, *** or ###P < 0.001

Fig 3: ERG is a key factor for endothelial specific inhibition of SMAD3-DNA binding. Putative ERG binding sites (grey bars) are located within the a TGβ2 and b CNN1 promoters upstream of the transcription start site (TSS) (arrow); ENCODE sequence conservation between 100 vertebrates across this region is shown. ENCODE ChIP-seq data profile for H3K4Me3, H3K27Ac, H3K9Ac and RNA polymerase II (RNA Pol2) in HUVEC indicate open chromatin and active transcription. Location of qPCR amplicon region R is indicated. ChIP-qPCR analysis of HUVEC was assessed following treatment with c and d TGFβ2 or PBS for 30 min, or e and f control or ERG siRNA. Chromatin was immunoprecipitated with c and e an α-ERG antibody, d and f α-SMAD3 antibody or control IgG. DNA was analysed by qPCR comparing specific primers against a negative control region for each target. Results are expressed as fold change compared to IgG (n = 3 for all experiments). Basal enrichment was analysed compared to negative control region (#). TGFβ2 or ERG siRNA treatment were compared to PBS treated or control siRNA (*) mean ± s.e.m., * or #P < 0.05, ** or ##P < 0.01, *** or ###P < 0.001 by unpaired t-test

Fig 4: ERG differentially regulates transcriptional activity of SMAD1 and SMAD3. ERG-dependent regulation of SMAD1 and SMAD3 activity were assessed by transactivation assays using either SMAD1 reporter pGL3-BRE (BRE reporter, a and c) or SMAD3 reporter pBV-SBE4 (SBE reporter, b and d). HUVEC were either co-transfected with Control siRNA (Con siRNA) or ERG siRNA for 24 h (a and b) or with pcDNA or ERG3 overexpression construct (ERG3) (c and d) prior to treatment with TGFß2 (red) and BMP9 (blue) or PBS (dashed line) for 18 h (data from pooled HUVEC in triplicate experiments). The ratio of luciferase to renilla from each transfection was normalised to PBS-treated control or to groups co-transfected with (a and b) Con siRNA or (c and d) pcDNA (n = 3). e Schematic of the inverse regulation of SMAD1 and SMAD3 by ERG. f Protein-protein interactions between ERG, SMAD1, SMAD2 and SMAD3 were assessed by Co-IP assay in whole cell lysate from HUVEC. SB-431542 (SB; 10 µM) treatment was performed for 1 h and TGFß2 (10 ng ml-1) or PBS treatments were performed 30 min prior to lysis. Lysates were immunoprecipitated with mouse IgG or mouse a-ERG and then immuno-blotted for a-ERG, a-SMAD1, a-SMAD2, a-SMAD3 and a-pSMAD3 (Ser423/425). Images representative of four experiments. g ERG-SMAD3 interaction was assessed by Co-IP in untreated HSEC. h Cellular localisation of the interaction between ERG with pSMAD3 was investigated by Proximity Ligation Assay (PLA) following 30 min TGFß2 treatment in HUVEC (n = 2). i HUVEC or j HSEC were pre-treated with SB-431542 or DMSO prior to transfection with either control siRNA or ERG siRNA for 48 h and analysed by qPCR (n = 3). Data were normalised to GAPDH and compared to control siRNA treated (*) or to ERG siRNA treated with DMSO (#) by unpaired t-test. All graphical data are mean ± s.e.m., * or #P < 0.05, ** or ##P < 0.01, *** or ###P < 0.001

Fig 5: Etanercept prevents ERG loss during murine liver injury. a C57/B6 mice were injected with mineral oil (vehicle; bi-weekly i.p) or with chronic CCl4 administration (8 weeks, bi-weekly i.p); a third group was subjected to a 3-weeks recovery period following CCL4 administration. Representative immunofluorescence of liver tissue samples stained for VWF (grey scale, green in merge), SMA (grey scale, red in merge) and ERG (white). Scale bar 50 µm. Quantification of b SMA expression and c ERG expression (three fields per mouse, n = 3). Data were compared to mineral oil controls (*) or to chronic CCL4 administration (#) by one-way ANOVA and Bonferroni multiple comparison post-test. d Representative immunofluorescence following acute (48 h) mineral oil, CCL4 or co-administration of etanercept and CCl4. Staining for ERG (white); merged images showing VWF (green), SMA (red) and DAPI (blue). e pSMAD3 (white; stars); other colours as in d; scale bar 20 µm. f–h: Quantification of f SMA, g ERG and h pSMAD3 expression (three fields per mouse, n = 3). Data were compared to mineral oil controls (*) or to 48 h CCL4 administration (#) by one-way ANOVA and Bonferroni multiple comparison post-test. i Co-administration of etanercept and CCL4 for 48 h was able to significantly reduce SMA expression in Ergfl/fl mice while it was ineffective in ErgcEC-Het mice. Representative immunofluorescence of liver tissue, colour scheme as above. Scale bar 50 µm. j Quantification of SMAD2/3 target gene expression by qPCR in liver tissue of Ergfl/fl mice injected with etanercept and CCL4 compared to CCL4 alone (*) or ErgcEC-Het treated with etanercept and CCL4 compared to CCL4 alone (#). k Quantification of SMA expression (three fields per mouse, n = 3). Data analysed by one-way ANOVA and Bonferroni multiple comparison post-test. All graphical data are mean ± s.e.m., * or #P < 0.05, ** or ##P < 0.01, *** or ###P < 0.001