High Shear Stress Reduces ERG Causing Endothelial-Mesenchymal Transition and Pulmonary Arterial Hypertension

Abstract

Background: Computational modeling indicated that pathological high shear stress (HSS; 100 dyn/cm²) is generated in pulmonary arteries (PAs; 100-500 µm) in congenital heart defects causing PA hypertension (PAH) and in idiopathic PAH with occlusive vascular remodeling. Endothelial-to-mesenchymal transition (EndMT) is a feature of PAH. We hypothesize that HSS induces EndMT, contributing to the initiation and progression of PAH.

Methods: We used the Ibidi perfusion system to determine whether HSS applied to human PA endothelial cells (ECs) induces EndMT when compared with physiological laminar shear stress (15 dyn/cm²). The mechanism was investigated and targeted to prevent PAH in a mouse with HSS induced by an aortocaval shunt.

Results: EndMT, a feature of PAH not previously attributed to HSS, was observed. HSS did not alter the induction of transcription factors KLF (Krüppel-like factor) 2/4, but an ERG (ETS-family transcription factor) was reduced, as were histone H3 lysine 27 acetylation enhancer-promoter peaks containing ERG motifs. Consequently, there was reduced interaction between ERG and KLF2/4, a feature important in tethering KLF and the chromatin remodeling complex to DNA. In PA ECs under laminar shear stress, reducing ERG by siRNA caused EndMT associated with decreased BMPR2 (bone morphogenetic protein receptor 2), CDH5 (cadherin 5), and PECAM1 (platelet and EC adhesion molecule 1) and increased SNAI1/2 (Snail/Slug) and ACTA2 (smooth muscle α2 actin). In PA ECs under HSS, transfection of ERG prevented EndMT. HSS was then induced in mice by an aortocaval shunt, causing progressive PAH over 8 weeks. An adeno-associated viral vector (AAV2-ESGHGYF) was used to replenish ERG selectively in PA ECs. Elevated PA pressure, EndMT, and vascular remodeling (muscularization of peripheral arteries) in the aortocaval shunt mice were markedly reduced by ERG delivery.

Conclusions: Pathological HSS reduced lung EC ERG, resulting in EndMT and PAH. Agents that upregulate ERG could reverse HSS-mediated PAH and occlusive vascular remodeling resulting from high flow or narrowed PAs.

Keywords: ERG; aorto-caval shunt mice; endothelial cells; endothelial-mesenchymal transition; pulmonary arterial hypertension; shear stress.

Figure 1:. High shear stress induces Endothelial-Mesenchymal transition.

Human donor PAEC exposed to high shear stress (HSS, 100 dyn/cm²), or laminar shear stress (LSS,15 dyn/cm²). Static condition (ST). (A) Representative immunoblot and densitometric analysis of EndMT markers relative to ACTB loading control. (B) Representative immunofluorescent staining of PECAM1 and CDH5 (green), FSP1 (S100A4) and ACTA2 (red), and nuclei (DAPI, blue). Quantification of intensity for each marker is shown as an average of 5 fields of view per each sample. Scale bar, 50 μm. (C) BMPR2 mRNA in PAEC. (D) Representative immunoblot and densitometric analysis of BMPR2. (E) KLF2 and KLF4 mRNA. (F) Representative immunoblot and densitometric analysis of KLF4. n=4 biological replicates. In all panels of Figure 1 (A–F) Paired t-test was used to compare HSS vs. LSS only; the ST levels are shown as a reference only. *p<0.05; **p<0.01 vs. LSS, by paired Student t test.

Figure 2:. HSS reduces H3K27 acetylation at ERG motifs and decreases ERG.

Human donor PAEC cultured under ST, LSS and HSS as in Figure 1, and H3K27ac measured using CUT&RUN. (A) Representative H3K27ac track showing increases in H3K27 at the SNAI2 locus under HSS. Red boxes, peaks increased in HSS vs. LSS, FDR<0.05. (B) Volcano plot of H3K27ac peaks with HSS vs. LSS with dots representing significant changes. P-values by a two-tailed Wald test with Benjamini-Hochberg adjustment. (C) Motif enrichment of peaks with increased H3K27ac in LSS vs HSS, analyzed using Homer. P-values by binomial test. Motifs colored by transcription factor family. (D) Normalized H3K27ac read densities for HSS, LSS, and ST across peaks containing predicted ERG motifs or across all peaks. n=3 technical replicates. (E) Representative images of Proximity Ligation Assays of PAEC, showing interaction of KLF4 with ERG under LSS and HSS (magenta). Foci per nucleus quantified in 6 non-overlapping random fields of view per replicate (right panel). Data shown as mean±SEM, n = 4 biological replicates, *p<0.05 by Mann Whitney test. Scale bar, 10 μm. (F) ERG mRNA by RT-qPCR. (G) Representative immunoblot and densitometric analysis of ERG. (H) Representative immunofluorescent staining of ERG and DAPI. Intensity quantified as an average of 5 fields of view per sample. Scale bar, 50 μm. (F, G, H), n=4 biological replicates. **p<0.01 by paired t test.

Figure 3:. ERG is reduced in PAH; decreasing ERG under LSS recapitulates EndMT.

(A) Confocal microscopic images of lung tissue sections from 4 patients with PAH and a congenital heart defect (APAH-CHD), 4 patients with Idiopathic or Familial PAH (I/FPAH, 2 each), or healthy controls (Donor, n=4). ERG (red), EC marker vWF (green) and nuclei stained with DAPI (blue). Arrows in magnified images (right column) point to EC with ERG in the nucleus. Scale bars, 40 μm in the three left columns, 15 μm in the magnified images. Quantification shows the average percentage of ERG-positive EC per small pulmonary artery (SPA) per patient/control, *p<0.05 vs Donor by One-way ANOVA and Kruskal-Wallis multiple comparisons test. (B-E) Human donor PAEC transfected with ERG siRNA (siERG) or nontargeting siRNA (siCON) followed by 48 h LSS (15 dyn/cm²). (B) Representative immunoblot and densitometric analysis. (C) Representative immunofluorescent staining shown with intensity quantified in an average of 5 fields of view per sample. Decreasing ERG leads to reduced PECAM1 (top left), reduced CDH5 (lower left), and increased ACTA2 (Right). Scale bars, 50 μm. (D) BMPR2 mRNA by qRT-PCR. (E) Representative immunoblot and densitometric analysis of BMPR2. n=4 biological replicates, *p<0.05; **p<0.01; ***p<0.001 vs. siCON under LSS, by paired Student t test.

Figure 4.. Replenishing ERG prevents EndMT under HSS.

Transfecting ERG by lentiviral vector (LV-ERG) vs. control lentiviral vector carrying EGFP (LV-CON) in human PAEC exposed to HSS. (A) Representative immunoblot and densitometric analysis of ERG and EndMT markers, normalized to ACTB. (B) Representative immunofluorescent staining. Top left: Replenishing ERG (red) increases PECAM1 (green) and Bottom left: CDH5 (green). Right: Gain of ERG (green) leads to downregulation of ACTA2 (red). Quantification of intensity per nuclei stain (DAPI) shows an average of 5 fields of views per sample. Scale bar, 50 μm. (C) Gene expression levels of BMPR2 in PAEC, assessed by qRT-PCR. (D) Representative immunoblot and densitometric analysis of BMPR2. Individual data points are shown, for n=4 biological replicates. *p<0.05; **p<0.01 vs. LV-CON under HSS, by paired Student t test.

Figure 5.. Aorto-Caval (AV) shunt in mice treated with AAV2 encoding Erg vs. Luciferase (control).

(A) Experimental design: Mice were treated with AAV2 encoding Erg (AAV-Erg) or luciferase (AAV-Con). One week later, we performed aorto-caval shunt (AV shunt) or sham surgery. Shunt patency was evaluated by color Doppler examination one week and eight weeks after surgery. At 8 weeks, mice were evaluated by MRI and echocardiography, and hemodynamic measurements (shown in Figure 6 and in Supplementary Figure S4) prior to sacrifice. Explanted heart was weighed, and the lungs evaluated by immunohistochemistry, immunofluorescence and Western blot analyses (Figure 7). (B) Patent shunt shows a mosaic Doppler color flow pattern in the IVC at the puncture site, indicating disturbed flow from aorta. Sham-operated or mice with a closed shunt show uniform color or no color at the IVC. Yellow arrow, IVC; white arrow, abdominal aorta. (C) Efficacy of targeting PAEC with the AAV2 vector, monitored by luciferase activity. Left, representative images of AAV2 luciferase treated mice at baseline and at 8 weeks, with quantification of luminescence on the right. Each data point represents a mouse with mean±SEM, n=10 mice per group. **p<0.01 vs. respective group at week=0, by two-way ANOVA followed by Tukey’s multiple comparisons. On the far right, a representative image of a distal PA in a mouse treated with AAV2 luciferase (AAV-Con) showing co-localization of luciferase (red) with the pan-endothelial cell marker MECA-32 (green). Scale bar, 20 μm. (D) Representative MRI images 8 weeks post-surgery, showing expansion of right and left ventricles (left), and expanded IVC (arrow), in the shunt groups with AAV2-luciferase and AAV2-Erg treatment vs. sham operation (right).

Figure 6.. AV shunt induced PAH in mice is attenuated by replenishing PAEC ERG.

Mice were treated as described in Figure 5. (A) Cardiac output and (B) partial pressure of oxygen (pO₂) in the IVC 8 weeks following surgery indicates AV shunt patency in AAV-Erg and AAV-Con groups. (C) Aortic systolic pressure, (D) right ventricular systolic pressure (RVSP), and (E) LV end-diastolic pressure (LVEDP) determined by heart catheterization. (F) Pulmonary vascular resistance (PVR) calculated using catheter and echocardiography data. (G) Simulated values of wall shear stress (WSS) with decreasing PA vessel order generated with a reduced order computational model of blood flow (see Methods). Each data point represents a mouse with mean±SEM, n=7–10 mice per group. *p<0.05, ***p<0.001, ****p<0.0001 vs. respective Sham group; ^†p<0.05, ^††p<0.01, ^†††p<0.001, ^††††p<0.0001 vs. AV shunt with AAV-Con by two-way ANOVA followed by Tukey’s multiple comparisons.

Figure 7.. AV shunt in mice induces EndMT and remodeling of the distal PAs, which are rescued by overexpression of Erg in PAEC.

Shunt or sham operated mice and hemodynamic parameters determined as in “Methods” and in Figure 5. (A) Weight ratio of the right ventricle (RV) and left ventricle (LV)+septum (S) to body weight, and the weight ratio of the RV/LV+S, eight weeks after AV shunt surgery. Each data point represents a mouse with mean±SEM, n=10 mice per group. (B) Examples are given of fully-, partially-, and non-muscular peripheral arteries. (C) Quantification of fully muscularized PAs per 100 alveoli in mice. Scale bars, 20 μm. n=9 mice per group. (D) Immunofluorescence images show vessels with EndMT, where ACTA2 (green) is localized to tdTomato (red) positive PAEC (arrowhead), in PAs from endothelial-specific inducible tdTomato (VE-Cre/Tomato) mice with quantification of percentage of EndMT vessels in (E). Scale bars, 20 μm. An average of 30 PAs were evaluated per mouse, in n=3 mice per condition. (F) Representative immunoblot and densitometric analysis of ERG, CDH5, BMPR2 and GAPDH from lung tissue of mice. Mean±SEM shown for n=6 mice per group. In (A, C, E, F), **p<0.01, ***p<0.001, ****p<0.0001 vs. respective Sham; ^†p<0.05, ^††p<0.01, ^†††p<0.001 vs. AV Shunt +AAV-Con by two-way ANOVA and Tukey’s multiple comparisons test.