Activin A–Endothelin-1 Axis Governs Pulmonary Vascular Remodeling: Mechanistic Basis for Emerging Therapies in PAH

Abstract

BACKGROUND:

Pulmonary arterial hypertension remains a life-threatening disease despite advances in vasodilator therapy. Vascular remodeling, partly driven by pulmonary artery endothelial cell dysfunction, is accompanied by vasoactive mediators imbalance such as ET-1 (endothelin-1). Although endothelin receptor antagonists alleviate vasoconstriction, they incompletely address the remodeling process. We previously reported how endothelial-derived activin A promotes vascular remodeling, leading to the clinical development of the activin signaling inhibitor sotatercept, which improves outcomes when added to endothelin receptor antagonists. As both activin A and ET-1 originate from endothelial cells and promote remodeling, we investigated whether activin A regulates ET-1 production and activity in pulmonary arterial hypertension.

METHODS:

In vitro, we used pulmonary artery endothelial cell models of activin A overabundance alone or cocultured with pulmonary artery smooth muscle cells. Cells were treated with either the activin A inhibitor FST (follistatin), the endothelin receptor antagonist bosentan, the FST/bosentan combination, or vehicle for analysis. In vivo, we exposed wild-type or endothelial-specific INHBA (inhibin β-A)-overexpressing mice (VEcadherin-INHBA-Transgenic/VEcad-INHBA-Tg) to chronic hypoxia pulmonary hypertension model, with the addition of FST, bosentan, FST and bosentan, or vehicle treatments after the first week of hypoxia exposure.

RESULTS:

Activin A upregulated ET-1 expression via canonical SMAD2/3 (small mother against decapentaplegic family member 2/3) signaling in pulmonary artery endothelial cells. This induction, as well as ET-1–driven downstream effects—including reduced eNOS (endothelial NO synthase), pulmonary artery smooth muscle cell phenotypic switching, oxidative stress, and endothelial-to-mesenchymal transition—was reversed by FST alone or in combination with bosentan. In vivo, FST-based therapy achieved greater hemodynamic, right ventricular remodeling, and vascular structural normalization in wild-type and VEcad-INHBA-Tg mice than bosentan alone, accompanied by stronger ET-1 suppression.

CONCLUSIONS:

We identified ET-1 as a downstream effector of activin A in pulmonary arterial hypertension development, supporting activin A blockade as a strategy to inhibit ET-1–mediated vasoconstriction and remodeling. This mechanistic link provides a rationale for the rapid clinical benefits observed with sotatercept and suggests its potential role earlier in the pulmonary arterial hypertension treatment paradigm.

What Are the Clinical Implications?

Activin A as an upstream orchestrator of ET-1 (endothelin-1)–mediated vascular remodeling during pulmonary arterial hypertension via canonical SMAD2/3 (small mother against decapentaplegic family member 2/3) signaling. Activin A blockade simultaneously blunted ET-1–related provasoconstriction and proremodeling pathways in the endothelial and smooth muscle cells with similar or better efficacy compared with endothelin receptor inhibition. In vivo blockade of activin A, alone, or combined with endothelin receptor antagonism, more effectively reduces ET-1 production and pulmonary hypertension severity compared with ET-1 inhibition alone. This activin A–ET-1 link provides a mechanistic basis for the rapid clinical efficacy of sotatercept observed in multiple clinical trials and studies while encouraging further clinical exploration of activin signaling inhibitors as earlier or even first-line therapy in pulmonary arterial hypertension.

Pulmonary arterial hypertension (PAH), classified as World Health Organization Group I pulmonary hypertension (PH), is a progressive and life-threatening disorder characterized by elevated pulmonary arterial pressure and extensive vascular remodeling.^1,2 Despite substantial advances in understanding the genetic, molecular, and pathophysiological basis of PAH, as well as the development of targeted therapies, the disease remains incurable, and survival is still markedly reduced compared with age-matched populations.^3–5 Current treatments—primarily endothelin receptor antagonists (ERAs), prostacyclin analogs, and phosphodiesterase-5 inhibitors—provide symptomatic relief and improve hemodynamics, yet they fail to fully halt or reverse the structural changes in the pulmonary vasculature.^1,6 The inability to durably control vascular remodeling remains a central obstacle to long-term disease resolution.

Pulmonary vascular remodeling in PAH is a multifactorial process involving endothelial cell (EC) dysfunction, smooth muscle cell proliferation, inflammatory activation, extracellular matrix deposition, in situ thrombosis, and the formation of complex plexiform lesions.^5,7 Among the numerous vasoactive mediators implicated, ET-1 (endothelin-1) stands out as one of the most potent and consistently elevated factors in PAH.⁸ Increased ET-1 levels correlate with disease severity and mortality, and beyond vasoconstriction, ET-1 drives pathological processes including oxidative stress, apoptosis, metabolic reprogramming, endothelial-to-mesenchymal transition, and impaired angiogenesis.^8–12 ERAs, now a mainstay of PAH therapy, target this pathway; however, their impact on the underlying structural pathology is incomplete.

Recent studies have identified excessive endothelial-derived activin A as a novel driver of vascular remodeling in PAH.^13–15 We and others have demonstrated that activin A overabundance promotes EC dysfunction, accelerates BMPRII (bone morphogenic protein receptor type II) degradation, indirectly suppresses canonical BMPRII signaling, and exacerbates pulmonary vascular pathology.^13,14 Activin A is part of the TGF-β (transforming growth factor beta) superfamily of peptides, specifically belonging to the activin and inhibin subfamily.^16,17 Activin A is derived from 2 INHBA (inhibin β-A) chains bound by a disulfide chain and is secreted extracellularly to exert its effects through the binding of TGF-β superfamily receptors, which can be inhibited through its natural inhibitor, such as inhibin A or FST (follistatin).^16,17 Importantly, FST strongly binds to activin A (and GDF-8 [growth differentiation factor] and GDF-11) extracellularly and prevents its signaling activity.¹⁸

In preclinical models, genetic (through the deletion of its gene INHBA) or pharmacological inhibition of activin A attenuates PH, findings that translated into the recent successful clinical trials of sotatercept, the first-in-class activin signaling inhibitor (ASI) which strongly binds to activin A extracellularly in addition to GDF-8 and GDF-11.^13,15,19 Notably, sotatercept, when added to standard vasodilator therapy including ERAs, yields additive clinical benefit, suggesting that activin A–driven remodeling operates through mechanisms distinct from, yet potentially convergent with, ET-1 signaling.^20–22

Given that both activin A and ET-1 are EC-derived mediators that promote pulmonary vascular remodeling, we hypothesized that activin A may directly regulate ET-1 expression and activity in PAH. In this study, we demonstrate that activin A upregulates ET-1 production in pulmonary artery ECs via canonical SMAD2/3 (small mother against decapentaplegic family member 2/3) signaling. Furthermore, we show that activin A inhibition by FST, alone or combined with bosentan, more effectively reduces ET-1 production, downstream signaling, and PH severity in vivo compared with bosentan alone. These findings reveal a mechanistic link between activin A and ET-1 pathways and suggest that dual pathway inhibition may offer a superior therapeutic strategy for targeting vascular remodeling in PAH.

Methods

Additional details are provided in the Supplemental Methods.

Data Availability

The data that support the findings of this study are available from the corresponding author on reasonable request.

Ethics Statement

All experiments were approved by the Ethics Review Committee for Experimentation of Kobe Pharmaceutical University (2024-001) and conducted in accordance with the Kobe Pharmaceutical University Animal Experimentation Regulation and the National Institutes of Health Guide for the Care and Use of Laboratory Animals (Publication No. 85–23, revised 1996). All animal experiments adhered to the Animal Research: Reporting of In Vivo Experiments guidelines.

Cell Culture and Treatments

Human pulmonary artery ECs (PAECs; no. C-12241; PromoCell) and pulmonary artery smooth muscle cells (PASMCs; no. C-12521; PromoCell) were cultured in HuMedia-EG2 (Kurabo) and smooth muscle cells growth medium 2 (PromoCell), respectively, at 37 °C and 5% CO₂. Medium was changed every 48 hours, and passages 6 to 8 were used with similar culture conditions and confluence to avoid differences in quiescence state.

For stimulation experiments, confluent PAECs were treated for 6 hours with recombinant human activin A (100 or 20 ng/mL in some cases; no. 338-AC-010; R&D Systems), recombinant human endothelin-1 (100 nmol/L; no. 4198-v; Peptide Institute), or vehicle. In some experiments, we treated PAECs with TGF-β (10 ng/mL; no. 240-B; R&D Systems), BMP4 (bone morphogenetic protein 4; 50 ng/mL; no. 314-BP; R&D Systems), or TNF-α (tumor necrosis factor-α; 10 ng/mL; no. 210-TA; R&D Systems). Cells were harvested for quantitative polymerase chain reaction or immunoblotting. In the case of immunoblotting, all membranes were cut to blot the target protein and the corresponding loading controls, unless otherwise indicated. Membranes subjected to stripping and reprobing are specifically labeled in the Supplemental Material.

Retroviral INHBA Overexpression

INHBA/pMSCVneo (plasmid murine stem cell virus with neomycin) and GFP (green fluorescent protein)/pMSCVneo plasmids (from our previous study¹³) were transfected into GP2-293 packaging cells with Lipofectamine 3000 (Thermo Fisher) together with the pVSVG (plasmid vesicular stomatitis virus G glycoprotein) envelope plasmid. Viral supernatants were collected at 48 and 72 hours, clarified, and stored at –80 °C.

PAECs at ≈70% confluence were infected for 24 hours with a 1:1 mix of viral supernatant and fresh medium containing polybrene (8 μg/mL; Sigma-Aldrich). Medium was replaced, and experiments were performed 48 hours later.

Chemical Inhibitors

For indicated experiments, PAECs were treated for 24 hours with FST (100 ng/mL; no. 4889-FN-025; R&D Systems), bosentan (10 μM; no. SML1265; Sigma-Aldrich), ALK (activin receptor–like kinase) 4/5/7 inhibitor SB505124 (5 μM; no. S2186; Selleck), ALK1/2/6 inhibitor K02288 (1 μM; no. S7359; Selleck), the activin receptor type IIA fusion protein ACTRIIA-Fc (activin receptor type IIA fusion protein; 2500 ng/mL), or vehicle.

Hypoxia Exposure

PAECs at 90% confluence were exposed to hypoxia (0.1% O₂, 24 hours) in half-volume medium with or without FST, bosentan, or SB505124. Cells were collected for quantitative polymerase chain reaction.

Indirect Coculture

Coculture was performed using 0.4 µm transwell inserts (Falcon). PAECs and PASMCs were first cultured separately to avoid cross-contamination, with PAECs cultured in transwell inserts while PASMCs were cultured in traditional multiwell dishes. PAECs in inserts were transduced with GFP or INHBA retrovirus for 24 hours, as explained in the retroviral INHBA overexpression subsection, before being returned to normal PAEC culture medium. Transduced PAECs were then cocultured with PASMCs already seeded in the lower chamber. After overnight coculture, FST, bosentan, or both FST and bosentan were added as indicated, either on the PAEC or the PASMC. After 48 hours, PASMCs were analyzed for ET-1 secretion and phenotype markers. For some experiments, PAECs were treated with recombinant activin A (100 ng/mL) for 24 to 48 hours instead. All numbers of samples (n) described in the figure legends represent the number of biologically independent cells.

siRNA Transfection

PAECs (60% to 70% confluence) were incubated with Opti-MEM (no. 31985-062; Gibco) for 30 min, then transfected for 6 hours with SMAD2 siRNA (small interfering RNA; 20 nmol/L; no. J-003561-05-0002; Dharmacon), SMAD3 siRNA (20 nmol/L; no. J-020067-05-0002; Dharmacon), dual SMAD2 and SMAD3 siRNA, or nontargeting control siRNA (no. D-001810-01) using Lipofectamine RNAiMAX (Thermo Fisher). After 48 hours, retroviral INHBA or GFP transduction was performed. Cells were analyzed 2 days later (4 days postsiRNA). In some experiments, PAECs were treated with activin A (100 ng/mL) for 6 hours after siRNA.

Animal Model

Endothelial-specific INHBA-overexpressing mice (VEcadherin-INHBA-Transgenic/VEcad-INHBA-Tg; TG) on a C57BL/6J background (from our previous study¹³) were maintained under standard housing (23 °C, 60% humidity, 12-hour light/dark cycle, kept in 3–5 mice/cage in a conventional, non-SPF (non-specific pathogen-free environment) with food (CRF-1, Charles River, purchased from Oriental Yeast Co, Ltd., Tokyo, Japan; major contents per 100 g of diet: moisture 8.1 g, crude protein 22.3 g, crude fat 5.3 g, crude ash 6.5 g, crude fiber 3.4 g, nitrogen-free extract 54.4 g, and calories 354.9 kcal) and water ad libitum. Mice were propagated as heterozygous TG animals by breeding with wild-type mice, and littermate wild-type was always used as control mice. Male mice (8–12 weeks) were exposed to chronic hypoxia (10% O₂, 3 weeks). Only male mice were used in this study due to the potential phenotypic differences between male and female mice in PH models, in addition to the potential effects of sex hormones on the phenotype, thus necessitating us to choose one sex for the study.^23–25 After 1 week, mice received intraperitoneal injections of vehicle, FST (8.5 μg/kg; no. 769-FS; R&D Systems), bosentan (30 mg/kg; no. 13760; ChemScene), or both, 5 d/wk for 2 weeks under ≈2% inhaled isoflurane anesthesia. All numbers of samples (n) described in the figure legends represent the number of biologically independent mice.

Hemodynamic Measurements

Right ventricular systolic pressure was measured via right heart catheterization using a 1.4F Millar Mikro-Tip transducer inserted through the right jugular vein under ≈2% inhaled isoflurane anesthesia. Right ventricular systolic pressure was averaged from 5 consecutive pressure waves.

Fulton Index

Hearts were fixed in 4% paraformaldehyde for 24 to 48 hours, dissected, and weighed. Fulton’s index was calculated as right ventricular weight/(left ventricle+septum weight).

Histology and Immunostaining

Paraffin lung sections underwent antigen retrieval (citric acid buffer, Vector Laboratories) and blocking (5% skim milk, 0.2% Triton X-100 in PBS). Sections were incubated overnight at 4 °C with fluorescein isothiocyanate–anti-α-smooth muscle actin (1:200; Sigma-Aldrich), anti–von Willebrand Factor (1:200; Abcam), anti–phospho-SMAD2/3 (1:100, CST), anti-F4/80 (1:100, Abcam), or anti-IL (interleukin)-6 (1:100, Abcam) followed by fluorophore-conjugated secondary antibodies (1:200; Abcam) and DAPI (4′,6-diamidino-2-phenylindole) mounting medium (Vector). Images were acquired on a Keyence BZ-X800 microscope. Pulmonary arteries <50 µm in diameter were classified as non-, partially-, or fully muscularized based on α-SMA coverage (<25%, 25% to 75%, >75% of diameter).

Isolation of Mouse Lung ECs

Lungs were dissociated enzymatically (Miltenyi Biotec), homogenized with a gentleMACS Dissociator, filtered, and incubated with FcR (Fc receptors) Block (Miltenyi) followed by CD146 (cluster of differentiation 146) MicroBeads (no. 130-092-007). Labeled cells were separated using LS (large selection) columns in a MACS (magnetic-activated cell sorting) separator, washed, and collected.

Functional Assays in PAECs

Tube Formation

PAECs (2×104 cells per well) were seeded on Matrigel-coated 96-well plates, incubated for 6 to 8 hours, and imaged. Chord length and the number of branch point was quantified with ImageJ.

Apoptosis

Cells were serum-starved for 24 hours, fixed, permeabilized, and stained with the TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling) kit (Roche). Apoptotic cells were expressed as % of total nuclei.

Proliferation

PAECs (0.5×104 cells per well) were seeded in 96-well plates, and proliferation was measured with WST-1 (water-soluble tetrazolium-1) reagent (Roche) at specified time points.

Reactive Oxygen Species Production

PAECs (2.5×104 cells per well) were incubated with DCFDA (2′,7′–dichlorofluorescin diacetate) reagent (Abcam), and fluorescence was recorded at baseline and after 4, 6, 16, and 24 hours of treatment.

Statistical Analysis

The Shapiro-Wilk test was used to test for the normality assumption. Data are presented as mean±SEM, with individual data points shown. Sample size (n) is given in the figure legends. Two-tailed unpaired Student t test was used for 2-group comparisons. One-way or 2-way ANOVA with Tukey post hoc test was used for ≥3 groups. P<0.05 was considered statistically significant. Analyses were performed with GraphPad Prism v10.5 (GraphPad Software).

Results

Activin A Regulates ET-1 Production in Human PAECs but Not Vice Versa

To examine the relationship between activin A and ET-1 in pulmonary ECs, we used our established system of INHBA overexpression and recombinant activin A treatment in human PAECs. Both activin A stimulation (Figure 1A and 1B) and INHBA overexpression (Figure 1C and 1D) markedly increased ET-1 mRNA expression and secretion, whereas it did not change the expression of GDF-11; GDF-8 expression was not detectable in PAEC (Figure S1A). In contrast, recombinant ET-1 failed to alter INHBA expression or activin A secretion, as did the inflammatory cytokine TNF-α (Figure 1E and 1F; Figure S1B), indicating that activin A lies upstream of ET-1. Further, direct treatment of other TGF-β superfamily ligands, such as TGF-β1 or BMP4, and TNF-α did not increase ET-1 expression (Figure S1B and S1C). Inhibition of activin A with FST reduced ET-1 induction to a degree comparable to bosentan, and combined FST+bosentan treatment yielded no further suppression (Figure 1G and 1H). Similarly, inhibition of activin A with the Sotatercept analog, ACTRIIA-Fc, could inhibit the increase in ET-1 levels (Figure 1I and 1J). Lastly, the increase in ET-1 expression increases gradually with activin A concentration (Figure S1D). These findings identify activin A as a direct upstream regulator of ET-1 in PAECs.

Figure 1.

Activin A induces excessive ET-1 (endothelin-1) production in pulmonary artery endothelial cells (PAECs), reversible by follistatin or bosentan. A and B, Quantitative real-time polymerase chain reaction analysis of ET-1 mRNA expression (A, n=6 biologically independent samples per group) and ELISA measurement of ET-1 concentration in culture medium (B, n=3–4) from PAECs treated for 6 hours with recombinant activin A (100 ng/mL) or vehicle. C and D, ET-1 mRNA expression (C, n=6) and secreted ET-1 concentration (D, n=4-5) in PAECs 48 hours after INHBA (inhibin β-A) overexpression (OE) or GFP (green fluorescent protein) control retroviral transfection. E and F, INHBA mRNA expression (E, n=6) and activin A concentration in culture medium (F, n=4) in PAECs treated for 6 hours with recombinant ET-1 (100 nmol/L) or vehicle. G, ET-1 mRNA expression in PAECs treated for 6 hours with vehicle or recombinant activin A in the presence of vehicle (VEH), FST (follistatin; 100 ng/mL), bosentan (BOS; 10 μM), or both (FST+BOS; n=4). H, ET-1 mRNA expression in PAECs 48 hours after INHBA OE or GFP transfection, followed by 24 hours of treatment with VEH, FST (100 ng/mL), BOS (10 μM), or FST+BOS (n=4). I, ET-1 mRNA expression in PAECs treated for 6 hours with vehicle or recombinant activin A in the presence of VEH or ACTRIIA-Fc (activin receptor type IIa fusion protein; 2500 ng/mL; n=3-4). J, ET-1 mRNA expression in PAECs 48 hours after INHBA OE or GFP transfection, followed by 24 hours of treatment with VEH or ACTRIIA-Fc (2500 ng/mL; n=4). Data are mean±SEM. P<0.05 is deemed statistically significant. Statistical tests: 2-sided Student t test for A through F; 1-way ANOVA with Tukey post hoc test for G and H.

Inhibition of Activin A Improves Human PAEC Function as Effectively as ET-1 Blockade

We next compared the functional consequences of activin A and ET-1 inhibition in INHBA overexpression PAECs. Bosentan treatment partially restored angiogenesis but did not significantly reduce apoptosis (Figure 2A through 2D). In contrast, FST, alone or combined with bosentan, significantly improved both tube formation and survival. Similarly, activin A stimulation or INHBA overexpression increased PAEC proliferation, which was suppressed by FST or bosentan (Figure 2E; Figure S1E). These results indicate that activin A drives the same pathogenic endothelial phenotypes as ET-1, and its blockade achieves equal or superior rescue.

Figure 2.

Activin A–driven ET-1 (endothelin-1) contributes to pulmonary artery endothelial cell (PAEC) dysfunction. A and B, Representative images (A) and quantification (B) of chord length and number of branching points in a Matrigel tube formation assay (n=3–4) using PAECs transfected with GFP (green fluorescent protein) or INHBA (inhibin β-A) overexpression (OE), treated with vehicle (VEH), FST (follistatin; 100 ng/mL), bosentan (BOS; 10 μM), or both (FST+BOS). C and D, Representative images (C) and quantification (D) of apoptotic cells assessed by TUNEL (terminal deoxynucleotidyl transferase dUTP nick-end labeling) staining (n=3–4) under serum starvation. Apoptotic cells are indicated by white arrows. E, Cell proliferation measured by WST-1 (water-soluble tetrazolium-1) assay in GFP- or INHBA OE–transfected PAECs treated with VEH, FST, BOS, or FST+BOS (n=3). Data are mean±SEM. P<0.05 is deemed statistically significant. Statistical test: 1-way ANOVA with Tukey post hoc test for B, D, and E. AU indicates absorbance units; and DAPI, 4′,6-diamidino-2-phenylindole.

Activin A–Derived ET-1 Alters Vasoconstrictive Properties of the Pulmonary Vasculature

Activin A overexpression in PAECs reduced eNOS (endothelial NO synthase) expression and phosphorylation, effects reversed by FST or bosentan (Figure 3A and 3B). Another cause of the vasoconstricting nature of PAH is the shift in PASMCs’ phenotype from the contractile phenotype to the synthetic/proliferative phenotype.²⁶ This contractile-to-synthetic switch is also promoted by excess ET-1 activity in PASMCs.⁹ In PASMC cocultures, activin A–stimulated PAECs slightly but significantly increased both ET-1 and INHBA mRNA levels in PASMCs while also contributing to the increase in ET-1 levels in the PASMCs culture medium (Figure 3C and 3D; Figure S2A and S2B) and induced a phenotypic switch toward a synthetic, less contractile state, as evidenced by increased PCNA (proliferating cell nuclear antigen) and fibronectin and decreased MYH11 (myosin heavy chain 11), SM22α (smooth muscle protein 22-α), and α-SMA (α-smooth muscle actin; Figure 3E and 3F). However, direct activin A treatment in PASMCs failed to increase ET-1 expression (Figure S2C), and direct treatment of FST or bosentan to the cocultured PASMCs could not completely recapitulate the amelioration seen when the INHBA-overexpression PAECs were treated instead (Figure S2D). Treatment of PAECs with FST or bosentan during coculture prevented these changes. Thus, endothelial activin A excess promotes a vasoconstrictive PASMC phenotype partly via ET-1.

Figure 3.

Endothelial cell (EC)–derived activin A–induced ET-1 (endothelin-1) alters vasoconstrictive properties of the pulmonary vasculature. A, eNOS (endothelial NO synthase) mRNA expression in pulmonary artery endothelial cells (PAECs) 48 hours after GFP (green fluorescent protein) or INHBA (inhibin β-A) overexpression (OE), treated with vehicle (VEH), FST (follistatin; 100 ng/mL), bosentan (BOS; 10 μM), or FST+BOS for 24 hours (n=3). B, Representative immunoblots and quantification of p-eNOS (phosphorylated eNOS) and total eNOS under the same conditions (n=3–4). C, ET-1 concentration in pulmonary artery smooth muscle cell (PASMC) culture medium after coculture with PAECs treated with recombinant activin A for 24 or 48 hours (n=3). D, ET-1 and INHBA mRNA expression in PASMCs cocultured for 48 hours with GFP- or INHBA OE–PAECs, with VEH, FST, BOS, or FST+BOS added for the last 24 hours (n=3–4). E, PASMC mRNA expression of PCNA (proliferating cell nuclear antigen), fibronectin, SM22α (smooth muscle protein 22-α), and α-SMA (α-smooth muscle actin) under the same conditions (n=3–4). F, Representative immunoblots and quantification of MYH11 (myosin heavy chain 11), SM22α, and MMP2 (matrix metalloproteinase-2) in PASMCs cocultured with GFP- or INHBA OE–PAECs, treated as in E (n=3). Data are mean±SEM. P<0.05 is deemed statistically significant. Tests: 2-sided Student t test for C; 1-way ANOVA with Tukey post hoc test for A, B, D through F.

Activin A Inhibition More Effectively Reverses Proremodeling Signaling Than ET-1 Blockade

INHBA overexpression in PAECs induced endothelial-to-mesenchymal transition, with downregulation of VE-cadherin and upregulation of fibronectin, vimentin, BMP4, SNAIL (snail family transcriptional repressor 1), and SLUG (snail family transcriptional repressor 2), with an increase in colocalized endothelial and mesenchymal markers in PAECs (Figure 4A and 4B; Figure S3A). FST, alone or with bosentan, more effectively attenuated endothelial-to-mesenchymal transition than bosentan alone. Increased oxidative stress in the vasculature is another mark of the remodeling process in PAH, and we found that activin A overabundance also elevated the central oxidative stress response protein NRF2 (nuclear factor erythroid 2-related factor 2), suppressed the superoxide dismutase enzyme SOD2 (superoxide dismutase 2), and increased total reactive oxygen species production (Figure 4B and 4C; Figure 3B). Overall, FST-based treatments produced greater restoration toward control levels than bosentan, suggesting broader antiremodeling effects. On the contrary, although we found that MCP-1 (monocyte chemoattractant protein 1) expression, which promotes monocyte/macrophage infiltration and could be triggered by ET-1, was increased with excess activin A, this effect is only reversible with FST-based treatment and not with bosentan-only treatment (Figure S3C), indicating that the blockade of ET-1 alone might not be sufficient to prevent this particular proremodeling aspect.

Figure 4.

Activin A–derived ET-1 (endothelin-1) drives multiple proremodeling pathways in pulmonary artery endothelial cells (PAECs). A, mRNA expression of BMP4 (bone morphogenetic protein 4), SLUG (snail family transcriptional repressor 2), SNAIL (snail family transcriptional repressor 1), VE-cadherin (vascular endothelial cadherin), fibronectin, and SOD2 (superoxide dismutase 2) in GFP (green fluorescent protein)- or INHBA (inhibin β-A) overexpression (OE)–PAECs treated with vehicle (VEH), FST (follistatin), bosentan (BOS), or FST+BOS for 24 hours (n=3–4). B, Representative immunoblots and quantification of vimentin, SOD2, and NRF2 (nuclear factor, erythroid 2-related factor 2) under the same conditions (n=3). C, Reactive oxygen species (ROS) production in GFP- or INHBA OE–PAECs treated as in A, measured at 4, 6, 16, and 24 hours (n=3–4). Data are mean±SEM. P<0.05 is deemed statistically significant. Tests: 1-way ANOVA with Tukey post hoc test for A and B; 2-way ANOVA with Tukey post hoc test for C.

SMAD2/3 Signaling Mediates Activin A–Induced ET-1 Expression

INHBA overexpression activated canonical SMAD2/3 signaling (Figure 5A; Figure S3D). Pharmacological SMAD2/3 inhibition with SB505124 (ALK 4/5/7 inhibitor) suppressed ET-1 induction by activin A (Figure 5B and 5C), something not seen after inhibition with the ALK 1/2/6 inhibitor K02288 (Figure S3E), and restored eNOS, SOD2, and SNAIL expression (Figure S3F). Additionally, treatment using ACTRIIA provided similar results in suppressing SNAIL (Figure S3G). SMAD2 or SMAD3 knockdown alone reduced, but did not abolish, ET-1 upregulation, while the double knockdown of both SMADs successfully abolished ET-1 expression, implicating both SMAD2 and SMAD3 (Figure 5D and 5E). Hypoxia increased both INHBA and ET-1 expression in PAECs, and FST or SB505124 prevented ET-1 induction without altering INHBA levels (Figure 5F through 5H; Figure S3H). This SMAD2/3-activating effect of activin A goes in concert with the BMPRII-degrading effect that we have previously demonstrated (Figure S3I). These data establish the ACTRIIA-ALK4/5/7 complex-driven SMAD2/3 as a critical link between activin A and ET-1 production.

Figure 5.

Canonical SMAD2/3 signaling mediates activin A–induced ET-1 (endothelin-1) expression. A, Representative immunoblots and quantification of p-SMAD2/3 (phosphorylated SMAD2/3; small mother against decapentaplegic family member 2/3) and total SMAD2/3 in GFP (green fluorescent protein)- or INHBA (inhibin β-A) overexpression (OE)–pulmonary artery endothelial cells (PAECs) treated with vehicle (VEH), FST (follistatin), bosentan (BOS), or FST+BOS for 24 hours (n=3). B, Immunoblots showing p-SMAD2/3 and total SMAD2/3 in GFP- or INHBA OE–PAECs treated with SB505124 (5 μM) or vehicle for 24 hours. C, ET-1 mRNA expression in GFP- or INHBA OE–PAECs treated with SB505124 or vehicle for 24 hours (left, n=3) and in PAECs treated for 6 hours with activin A±SB505124 (right, n=3–4). D, Immunoblots showing SMAD2/3 and β-actin in PAECs pretreated with SMAD2 siRNA (small interfering RNA; siSMAD2), SMAD3 siRNA (siSMAD3), dual SMAD2-SMAD3 siRNA (dual siSMAD), or control siRNA (siNC [siRNA negative control]). E, ET-1 mRNA expression in GFP- or INHBA OE–PAECs pretreated with siSMAD2, siSMAD3, dual siSMAD, or siNC (n=3–4). F, INHBA and ET-1 mRNA expression in PAECs exposed to normoxia or hypoxia (0.1% O₂, 24 hours; n=6). G and H, ET-1 mRNA expression in PAECs under hypoxia treated with FST (G) or SB505124 (H; n=3). Data are mean±SEM. P<0.05 is deemed statistically significant. Tests: 1-way ANOVA with Tukey post hoc test for A, C, E, G, and H; 2-sided Student t test for F.

FST Confers Greater In Vivo Benefit Than Bosentan Alone

In chronic hypoxia–induced PH, FST treatment, alone or with bosentan, reduced right ventricular systolic pressure more effectively than bosentan in VEcad-INHBA-Tg mice and achieved comparable benefit in wild-type mice (Figure 6A and 6B; Figure S4A). All treatments improved right ventricular hypertrophy (Figure 6C), but FST-based regimens more effectively reduced pulmonary vascular remodeling and medial thickening (Figure 6D through 6F). Lung ET-1 expression was elevated in Tg mice and suppressed by all treatments, especially with FST-based therapy (Figure 6G). In addition, similar ET-1–downstream pathways were also altered in the isolated lung ECs of Tg mice (Figure 6H), while phosphorylation of SMAD2/3 was also increased in the endothelium of Tg lungs (Figure S4B). Immune cells such as macrophages also play an important part in accelerating vascular remodeling, partly by producing proinflammatory cytokines such as IL-6. With the increase of the monocyte/macrophage-recruiting MCP-1 in the ECs during high activin A, we found that in the lungs of VEcad-INHBA-Tg mice, there was an increase in F4/80+-macrophage cell numbers that colocalized with IL-6 (Figure S4C and S4D). Concurrently, whole-lung mRNA level of IL-6 was also increased in VEcad-INHBA-Tg mice, and only FST-based therapy was able to improve this, something not seen in the bosentan-only group (Figure S4E). To conclude, these findings support activin A inhibition as a potent strategy to attenuate ET-1–driven vascular remodeling in PAH.

Figure 6.

In vivo activin A inhibition improves pulmonary hypertension (PH) phenotype comparably or more than ET-1 (endothelin-1) blockade. A, Experimental design: wild-type (WT) and VE-cadherin (vascular endothelial cadherin)–INHBA (inhibin β-A)-Tg (TG/transgenic) mice were exposed to hypoxia (10% O₂) for 3 weeks, with vehicle (VEH), FST (follistatin; 8.5 μg/kg), bosentan (BOS; 30 mg/kg), or FST+BOS administered during the final 2 weeks. B, Right ventricular systolic pressure (RVSP; n=4–9). C, Fulton index (RV/[LV+S] [right ventricle to left ventricle plus septum] ratio; n=4–8). D, Representative hematoxylin and eosin–stained lung sections. Blue arrows indicate vessels. E, Representative immunofluorescent staining of α-SMA (α-smooth muscle actin protein; green, SMC [smooth muscle cell] marker), vWF (von Willebrand Factor; red, endothelial cell [EC] marker), and DAPI (4′,6-diamidino-2-phenylindole; blue, nuclei). White arrows indicate vessels. F, Quantification of pulmonary artery muscularization (non-, partial-, full; n=12–15 fields from 3–4 mice). PA indicates pulmonary artery. G, Lung ET-1 mRNA expression (n=3–4). H, mRNA expression of INHBA, ET-1, eNOS (endothelial NO synthase), SOD2 (superoxide dismutase 2), fibronectin, SLUG (snail family transcriptional repressor 2), CD31 (cluster of differentiation 31), and BMP4 (bone morphogenetic protein 4) in lung ECs isolated from WT and TG mice (n=3–4). Data are mean±SEM. P<0.05 is deemed statistically significant. Tests: 1-way ANOVA with Tukey post hoc test for B, C, and G; 2-way ANOVA with Tukey post hoc test for F; 2-sided Student t test for H.

Discussion

This study identifies ET-1 as a downstream effector of endothelial-derived activin A in PAH and demonstrates that blocking activin A signaling yields equal or superior reversal of proremodeling phenotypes compared with ET-1 receptor antagonism. Mechanistically, activin A upregulates ET-1 via canonical SMAD2/3 signaling, driving endothelial dysfunction, PASMC phenotypic switching, vasoconstrictive imbalance, and profibrotic and pro-oxidative pathways. These findings position activin A as an upstream orchestrator of ET-1–mediated vascular pathology. It is interesting to note that blocking ET-1 activity via ERA can decrease ET-1 mRNA expression. This is attributable to one of ET-1’s regulatory mechanisms, where there is an autoinduction of ET-1 mRNA level through ET_B (endothelin receptor type B).²⁷ We think that the observed reduction in ET-1 mRNA level after ERA administration is caused by the blockade of this ET-1 expression autoinduction mechanism.

Our work also adds to the growing evidence that activin A sits at a critical intersection of TGF-β superfamily signaling, capable of shifting the balance between SMAD2/3 and SMAD1/5/8 pathways.^17,28 This duality may explain its broad influence on vascular cell phenotypes, highlighting the complexity of selectively modulating this axis. Further mechanistic dissection of activin A’s receptor interactions and downstream effectors is warranted.

From the in vitro studies on FST, bosentan, and their combination, we found that not only is the excessive production of ET-1 in PAECs during pathological conditions regulated by activin A, but activin A can also affect similar downstream pathways to ET-1 in PAECs. One such notable example is the vasodilator eNOS, a known downstream of ET-1.²⁹ We demonstrated that blocking activin A can achieve a similar effect to that of ERAs in restoring the eNOS level in PAECs. Combined with how the INHBA/activin A–overexpressing PAECs can affect the PASMCs’ phenotype and how ET-1 itself could lead to a defective and imbalanced vasoconstriction in the pulmonary arteries, the fact that blocking activin A could lead to a beneficial effect in the vasoactive properties of the pulmonary arteries bodes well in the sense that it could potentially give additive or similar effect to the vasodilators.

Beyond its effect on vasodilators and PASMCs, other pathways important in promoting vascular remodeling in PAECs are also altered by the overabundance of activin A and ameliorated by FST or bosentan, such as SOD2 and NRF-related reactive oxygen species scavenging and response, or the Snail/Slug-mediated endothelial-to-mesenchymal transition, showing that FST can also affect auxiliary, nonvasoconstriction signaling pathways affected by ET-1, which is in line with our experiments’ aim.^30,31 Despite this, our PAEC functional analysis suggests that Bosentan-mediated ET-1 signaling blockade cannot totally ameliorate the activin A–induced angiogenic capacity decrease, which is corroborated by the failure of bosentan to ameliorate the increase in endothelial MCP-1 and, subsequently, macrophage infiltration and IL-6 production. Although a <100% ET-1 blockade is a possibility, a more likely explanation is that (1) there are more unexplored downstream signaling pathways of activin A, such as the immune cell–mediated inflammatory pathways, and (2) ET-1 blockade alone is not enough to ameliorate the remodeling process as a whole, plausible when connecting our results with the clinical findings over recent years.³² With activin A overabundance increasingly seen as central to the pathophysiology of PAH, further research is needed to fully elucidate its effect in promoting vascular remodeling in PAH.⁵

The clinical implications are significant in the context of the recent approval of sotatercept—the first-in-class ASI—after the STELLAR (phase 3 study of sotatercept for the treatment of pulmonary arterial hypertension), PULSAR (a study of sotatercept for the treatment of pulmonary arterial hypertension), ZENITH (sotatercept in patients with pulmonary arterial hypertension at high risk for death), and HYPERION (sotatercept for pulmonary arterial hypertension within the first year after diagnosis) trials.^20–22 Sotatercept is currently indicated as an add-on therapy to established vasodilator regimens, including ERAs.⁶ Our data provide a mechanistic rationale for this benefit: by acting upstream of ET-1, activin A blockade may simultaneously blunt ET-1–dependent and ET-1–independent remodeling pathways, potentially offering broader protection than receptor blockade alone. As noted, FST-mediated activin A inhibition matched or exceeded bosentan in restoring eNOS expression, normalizing PASMC phenotype, reducing oxidative stress, and limiting endothelial-to-mesenchymal transition—effects that extend beyond vasodilation.^11,29,33,34 As shown in our data, preventing excessive ET-1 production (via activin A blockade) and inhibiting its activity (through ERA) simultaneously may have an additive effect in certain cases. Some of the possibilities that could cause this effect include the fact that ERA could give additional benefit via the blockade of ET-1 naturally produced from a nonactivin A–dependent pathway in the PAECs, or the excess ET-1 produced in non-EC cells.

Interestingly, in clinical trials, the hemodynamic and exercise capacity benefits of sotatercept emerged early after treatment initiation—within a timeframe generally considered too short for substantial reverse remodeling of pulmonary vessels to occur.^21,22,35,36 This has led to the hypothesis that sotatercept may also exert acute vasodilatory effects, in addition to its remodeling benefits.^22,35,36 Our findings provide a plausible mechanistic basis for this early benefit: by suppressing ET-1 production and downstream signaling, activin A inhibition could relieve vasoconstrictive tone in the pulmonary circulation, thereby contributing to rapid improvements in vascular resistance and right ventricular afterload.³⁷ Such an effect, if confirmed in clinical studies, would expand the therapeutic profile of sotatercept to encompass both immediate functional relief and long-term structural repair.

These results raise the possibility that ASIs could be considered earlier in the treatment algorithm for PAH, either as monotherapy in select patients or in combination with vasodilators to achieve additive benefits, a theory which is strongly boosted by the recent results of HYPERION.³⁵ Such a strategy could target the root drivers of vascular remodeling rather than only its downstream mediators.³⁸ Although this proposition is premature for immediate clinical adoption, it underscores the need for comparative and combinatorial studies of ASIs and ERAs in both preclinical and clinical settings.

Limitations include the use of a chronic hypoxia mouse model, which represents a milder PAH phenotype compared with models, such as Sugen5416/hypoxia, and the absence of validation in human pulmonary vascular tissue. The translational relevance would be strengthened by correlating activin A and ET-1 levels in human PAH samples and by testing other ERAs (eg, macitentan, ambrisentan) and ASIs (eg, sotatercept) in head-to-head preclinical comparisons.

In conclusion, we demonstrate that activin A is a master regulator of ET-1 production and its pathological consequences in PAH. Inhibiting activin A not only suppresses ET-1–driven remodeling but also attenuates additional pathogenic pathways, supporting its potential as a foundational therapeutic target. These findings provide a mechanistic basis for the immediate clinical efficacy of sotatercept and encourage further exploration of ASIs as earlier or even first-line therapy in PAH.

ARTICLE INFORMATION

Acknowledgments

The authors thank Drs Satrio Adi Wicaksono, Arinal Chairul Achyar, Aditya Adinata, Saiful Hidayat, and Andreas Billy Yoel Turnip (Kobe University Graduate School of Medicine) for their excellent technical assistance, and Dr Ria Lupitasari for the excellent assistance in preparing the Graphical Abstract.

Disclosures

None.

Supplemental Material

Supplemental Methods

Tables S1–S2

Figures S1–S4

Major Resource Table

Uncut Blots

ARRIVE Checklist

Supplementary Material

atv-46-e323681-s001.pdf ^{(2.5MB, pdf)}

atv-46-e323681-s002.pdf ^{(148.7KB, pdf)}

Funding Statement

None.