Pulmonary endothelial dysfunction associated to pulmonary arterial hypertension (PAH) is a key pathogenetic mechanism that could be detrimental for disease susceptibility and development of pulmonary vascular remodeling.^1–3 Despite recent progress in the treatment of PAH,⁴ most patients still die from the disease or fail to respond adequately to medical therapy with a 5-year survival of 59%.⁵ Current treatments can relieve some PAH symptoms and slow the progress of the disease in some patients, but they have a limited impact on the progressive pulmonary vascular remodeling that eventually culminates in right heart failure. Since 2000, heterozygous germline mutations in the Bone MorphogeneticProtein Receptor type 2 (BMPR2) gene have been identified as critical genetic factors predisposing to pulmonary vascular remodeling and PAH development with low penetrance.^6–8 Still, the specific contribution of the long-term loss of signal transduction triggered by this signaling pathway remains poorly characterized.

Editorial, see p 822

Meet the First Author, see p 818

Among BMPs, BMP9 and BMP10 are 2 high-affinity ligands for ALK1 (Activin receptor-like kinase 1) and BMPR2 present in a heterotetrameric complex on pulmonary endothelial cells (ECs),⁹ and thereby are key actors of vascular development and homeostasis.¹⁰ Perturbation in the BMP9/BMP10 signaling pathway has emerged as essential in endothelial (dys)function and vascular remodeling, in particular in PAH, and hereditary hemorrhagic telangiectasia.^11,12 In line with this notion, several works have highlighted a potential role for this pathway in the regulation of vascular tone¹³ and in the modulation of ET-1 (endothelin-1) ^14,15 and apelin, 2 potent vasoreactive mediators for the pulmonary vasculature.^16,17 Even if Bmp9-KO mice are viable and present no obvious defects, these mice exhibit an altered lymphatic maturation that leads to drainage deficiency.¹⁸ Although new biomedical therapeutics targeting BMPs or the BMP signaling pathway hold promise in future treatment strategies for PAH and other life-threatening diseases, our current understanding of the molecular signaling pathways activated by BMPs and their exact pathogenic roles remains limited.¹¹

Therefore, we hypothesized that long-term loss of endothelial BMP9 signaling in rodents could alter pulmonary vascular tone and remodeling, thereby reducing their susceptibility to pulmonary hypertension.

Methods

The authors declare that all supporting data are available within the article and its Online Data Supplement.

Because of space limitations, a detailed description of the Materials and Methods is presented in the Online Data Supplement.

Results

Mice Deficient in BMP9 (Bmp9^−/−) are Less Susceptible to Chronic Hypoxia-Induced Pulmonary Hypertension

To gain information about the role of BMP9 in the susceptibility of mice to develop pulmonary hypertension (PH), we used mice congenitally deficient in the expression of the BMP9 gene (Gdf2). The loss of active circulating BMP9 levels in Bmp9^−/− mice was confirmed using a BMP responsive luciferase reporter assay (Figure 1A). Consistent with these findings, a 2-fold decrease in phosphorylated (p)-Smad1/5/8 and p-P38 were found in lungs of Bmp9^−/− versus Bmp9^+/+ littermates and no compensatory upregulation of other BMP family members was present in these mice deficient in BMP9 (Online Figure I). Bmp9^−/− mice in the C57BL/6 background are viable and present no obvious defects except a defect in lymphatic maturation.^17,18 Our present findings indicate that Bmp9^−/− mice are less susceptible to pulmonary vascular remodeling than Bmp9^+/+ littermates when exposed to chronic hypoxia. Indeed, the elevations in right ventricular systolic pressure (Figure 1B) and right ventricular hypertrophy right ventricle (RV/left ventricle+septum weight ratio) (Figure 1C) following chronic hypoxia were lower in the Bmp9^−/− than the Bmp9^+/+ mice. No significant differences were observed in heart rate or systemic pressures between Bmp9^−/− and Bmp9^+/+ mice (heart rate: 320±34 versus 324 ±44 bpm, respectively, NS; systemic blood pressure: 110±20 versus 105±18, n=10, respectively, NS). Similar findings have also been observed with intravenous administration of the active monocrotaline metabolite (monocrotaline pyrrole) to Bmp9^−/− and Bmp9^+/+ mice (Online Figure II). Following chronic hypoxia, the less pronounced right ventricular systolic pressure elevation in the Bmp9^−/− was also associated with a decreased number and thickness of muscularized distal pulmonary arteries versus Bmp9^+/+ mice (Figure 1D). Consistent with these observations, we did not observe a significant increase in the number of PCNA⁺ (proliferating cell nuclear antigen positive) cells or an accumulation of F4/80⁺ cells (monocytes/macrophages) in the perivascular area of muscularized vessels in Bmp9^−/− mice under normoxic versus hypoxic conditions in contrast to Bmp9^+/+ mice (Figure 1E).

• Download figure
• Download PowerPoint

Administration of Neutralizing Anti-BMP9 Antibodies in Mice Prevents Chronic Hypoxia-Induced Pulmonary Hypertension

To further demonstrate the role of BMP9 in the susceptibility of mice to develop PH, we study the effect of suppressing BMP9 action in adult C57BL/6 mice (wild type) with neutralizing anti-BMP9 antibodies (kindly obtained from Dr M. Yan). Weekly administration of neutralizing anti-BMP9 antibodies (5 mg/kg, IP) after hypoxia exposure completely inhibited BMP circulating ALK1-dependent activity (Figure 2A). Consistent with our hypothesis and our findings obtained in Bmp9 deficient mice, we found that C57BL/6 mice chronically exposed to hypoxia and treated with neutralizing anti-BMP9 antibodies were less susceptible to the development of chronic hypoxia–induced PH than C57BL/6 mice treated with IgG. The severity of the disease was assessed by right ventricular systolic pressure measurement (Figure 2B), by right ventricular hypertrophy (Figure 2C), and by pulmonary distal arterial muscularization (Figure 2D). We also noted that cell proliferation and the degree of inflammatory cell infiltration were significantly decreased in lungs from C57BL/6 mice chronically exposed to hypoxia and treated with neutralizing anti-BMP9 antibodies when compared with lungs of C57BL/6 mice chronically exposed to hypoxia and treated with IgG (Figure 2E).

• Download figure
• Download PowerPoint

Efficacy of Treatment With the Soluble Extracellular ALK1 Domain (ALK1_ECD) on the Progression of Established PH in 2 Complementary and Well-Established Models of Severe PH in Rats

Next, these results prompted us to test the efficacy of the soluble extracellular ALK1 domain (ALK1_ECD), a ligand trap targeting ALK1’s ligands thus both BMP9 and BMP10,¹⁹ on the development of monocrotaline (MCT)-induced PH in Wistar rats. Treatment with ALK1_ECD (4 mg/kg, IP) was started 1 week after a subcutaneous MCT injection and was repeated 1 week later (3 mg/kg, IP; Figure 3A). We first validated that the weekly ALK1_ECD administration completely inhibited BMP circulating ALK1-dependent activity (Figure 3B). On day 21, in MCT-injected rats treated with human IgG, a marked increase in mean pulmonary arterial pressure, total pulmonary vascular resistance, RV/(left ventricle+septum) ratio (Figure 3C), percentage medial wall thickness, and numbers of muscularized distal pulmonary arteries (Figure 3D) were found compared with controls. However, MCT-injected rats receiving ALK1_ECD exhibited reduced mean pulmonary arterial pressure, total pulmonary vascular resistance, and RV/(left ventricle+septum) ratio as compared with MCT-injected rats receiving IgG (Figure 3C). Consistent with these results, the percentages of medial wall thickness and of muscularized distal pulmonary arteries were substantially decreased in MCT-injected rats receiving ALK1_ECD when compared with MCT-injected rats receiving IgG (Figure 3D). In addition, we found a substantial decrease in collagen deposition in the RV myocardium of MCT-injected rats receiving ALK1_ECD when compared with MCT-injected rats receiving IgG, an observation that is consistent with the more preserved cardiac function observed in this ALK1_ECD-treated rats (Online Figure IIIA-B). Interestingly, no significant differences were found in the number of PCNA+ and CD68+ (monocyte/macrophages) cells in pulmonary vessel walls in lungs of MCT-injected rats receiving ALK1_ECD as compared to lungs of control rats receiving IgG (Figure 3E).

• Download figure
• Download PowerPoint

To validate our findings obtained in the MCT rat model, weekly treatments of Sugen-hypoxia (SuHx) rats with ALK1_ECD or IgG were next performed (Figure 4). Treatment with ALK1_ECD (4 mg/kg, IP) was started 5 weeks post-SU5416 injection and was repeated 1 and 2 weeks later (3 mg/kg, IP; Figure 4A). We first validated that the weekly ALK1_ECD administration completely inhibited BMP circulating ALK1-dependent activity (Figure 4B). Eight weeks post-SU5416 injection, SuHx rats develop severe experimental PH, as reflected by a marked increase in values of mean pulmonary arterial pressure total pulmonary vascular resistance, and Fulton index (Figure 4C). Although only a trend towards lower mean pulmonary arterial pressure values was found, SuHx rats treated with ALK1_ECD exhibit less severe PH when compared with SuHx rat-treated with IgG, as reflected by significantly lower values of total pulmonary vascular resistance and Fulton index (Figure 4C). Consistent with these findings and those obtained in the MCT rat model, we also noted a beneficial effect of ALK1_ECD treatments on the pulmonary arterial muscularization (Figure 4D), numbers of PCNA+ and CD68+ cells per vessel (Figure 4E), as well as on RV function and collagen deposition in the RV myocardium (Online Figure IIIC-D) when compared with SuHx rat-treated with IgG.

• Download figure
• Download PowerPoint

BMP9 Regulates In Vitro and In Vivo the Balance of Key Vascular Tone Regulators

BMP9 has been described as a potential vasoconstriction factor in the chick chorioallantoic membrane assay¹³ and as a BMP ligand that can down-regulate the expression of the vasodilator factor apelin^16,17 and increase the vasoconstrictor peptide ET-1.^14,15 An in silico reanalysis of the microarray data (accession number E-MTAB-2495) obtained from Long et al²⁰ in which, human pulmonary ECs were stimulated with 1 ng/mL of recombinant BMP9 for 4 hours, revealed 2 key pathways that are susceptible to regulate vascular contraction, namely the cGMP-PKG signaling pathway” and the “Vascular smooth muscle contraction pathway. Based on these sets of genes and their known effectors, we set-up a list of genes that allowed us to identify 23 genes that were significantly regulated by BMP9 (Online Figure IV), including apelin and ADM (adrenomedullin), 2 potent vasodilator factors. We thus investigated whether recombinant human BMP9 could alter the balance between pulmonary endothelium-derived relaxing factors and contracting factors in vitro using freshly isolated pulmonary ECs established from lung tissue obtained from control and PAH patients (both idiopathic and BMPR2 mutation carriers). Interestingly, we found that recombinant human BMP9 induced a dose-dependent production of ET-1, and a dose-dependent decrease in the production of apelin and ADM in control pulmonary ECs (Figure 5A). A similar profile was observed in pulmonary ECs derived from PAH patients except for ADM level that was already significantly reduced at the basal level in comparison to control cells (Figure 5A). Based on these in vitro findings, we then assessed the levels of the synthesis of potent vasoreactive factors in lungs of Bmp9^−/− and Bmp9^+/+ mice under normoxic and hypoxic conditions. Consistent with our findings showing that Bmp9^−/− mice are less susceptible to chronic hypoxia-induced PH, we found a decreased ET-1 mRNA level and an increased apelin and ADM mRNA levels in Bmp9^−/− mice chronically exposed to hypoxia compared with Bmp9^+/+ mice chronically exposed to hypoxia (Figure 5B). Since these mediators can regulate the function and the thickness of the underlying smooth muscle cells and because myosin light chain is a major regulatory molecule for smooth muscle contraction, we next studied its expression in lungs of Bmp9^−/− mice under normoxic and hypoxic conditions. Interestingly, we found that myosin light chain protein expression is decreased in lungs of Bmp9^−/− mice chronically exposed to hypoxia compared with Bmp9^+/+ mice chronically exposed to hypoxia (Figure 5C and 5D), a phenomenon that was associated with decreased activity of the ROCK (Rho/Rho-associated protein kinase) signaling pathway (Online Figure V). Consistent with these findings obtained in lungs of Bmp9^−/− mice, a substantial decrease in myosin light chain protein expression was found in lungs of MCT- and SuHx rats receiving ALK1_ECD when compared with lungs of MCT- and SuHx rats receiving IgG (Figure 5E and 5F).

• Download figure
• Download PowerPoint

Discussion

Our study identifies BMP9, also known as GDF2 (growth and differentiation factor 2), as a major factor for controlling functions of pulmonary vascular cells (Figure 6). Using 3 different approaches of suppressing BMP9 action in rodents (Bmp9^−/− mice, administration of neutralizing anti-BMP9 antibodies or of ALK1_ECD ligand trap for BMP9 and BMP10 in rats) in different complementary and well-established animal models of PH, we demonstrate that the selective loss of BMP9 partially protects against experimental pulmonary hypertension.

• Download figure
• Download PowerPoint

These results seem at first glance counter-intuitive, as it would have been expected that less ligands for the BMPR2 receptor should mimic the loss of BMPR2 signaling found in PAH patients. In accordance with this hypothesis, it was recently proposed to enhance this pathway by exogenous BMP9 in PAH.²⁰ Even if there are some arguments that support this notion,^21–24 our functional data indicate the opposite and further illustrate that BMP signaling is complex and involves many ligands and many receptor combinations that could explain these unexpected results, especially in the context of PAH. Indeed, it has been found that BMPR2 mutations could alter BMP signaling in a BMP ligand- or receptor-specific manner.^25–27 During the writing of this article, it was published that BMP9 could be a sensitive biomarker that segregates portopulmonary hypertension from other forms of PAH.²⁸ Together, these results clearly support that BMP9 signaling pathway is involved in PAH development but also that the molecular mechanisms involved in PAH are complex.

BMP9 is mainly produced by the liver and circulates in blood under a biologically active form and has been proposed to act as a vascular quiescence factor.¹³ Several works also support a potential role for this pathway in the regulation of vascular tone. Indeed, it was shown by our group and others that BMP9 induces the expression of ET-1, a strong vasoconstrictor,^14,15 and decreases the expression of apelin, a potent vasodilator for the pulmonary vasculature.^16,17 We also previously showed that addition of BMP9 induced vasoconstriction in the chicken chorionallantoic membrane.¹³ So, one hypothesis that could explain why the loss of BMP9 protects from PH development is that BMP9 could act as a potent modulator of vasoconstriction in the pulmonary circulation. This hypothesis is consistent with the dilation phenotype observed in Alk1 ± or Eng ± mice^29,30 or Alk1 mutants, vbg, in zebrafish.³¹ This hypothesis is further supported by the fact that we found that Bmp9^−/− pups presented a partial reopening of the ductus arteriosus just after birth that could because of a defect in vasoconstriction.³² Interestingly, patent ductus arteriosus is a serious congenital heart disease that can lead to irreversible pulmonary hypertension.³³ We confirm here the downregulation of apelin and show, for the first time, the downregulation of adrenomedullin by BMP9, 2 known critical actors that can impair pulmonary endothelial function.^34,35

We also confirm here that BMP9 induces a dose-dependent ET-1 expression in human pulmonary ECs and regulate its lung production in rodents. However, although BMP9 and other BMPs have been shown to induce ET-1,^14,15 unexpectedly BMPR2 downregulation is also associated with an increase in ET-1 expression,^15,36 supporting the notion that BMP9 stimulated ET-1 production by pulmonary ECs is not sufficient alone to lead to PH development and strengthening the evidence for the importance of BMP receptor balance in the control of vascular tone.

The role of BMP9 has recently been studied in a bronchopulmonary dysplasia model induced by chronic exposure to hyperoxia in neonatal rats. Consistent with our present findings, BMP9 treatment has been found to improve aberrant alveolar development and reduced lung inflammation and fibrosis but did not improve aberrant vascularization (arterial medial wall thickness and muscularization) and right ventricular hypertrophy.³⁷ In line with this notion, BMP9 and BMP10 can synergize with TNF-α (tumor necrosis factor α) to induce the upregulation of endothelial selectins and adhesion molecules and the secretion of various key proinflammatory mediators,^38,39 and are, therefore, likely to contribute to the proinflammatory signature of the dysfunctional endothelium in PAH.⁴⁰

In the present study, we provide definitive evidence that the selective BMP9 loss or inhibition partially prevent and protects against experimental pulmonary hypertension. In addition, this study offers important physiopathological insights into the role of BMP9 in the synthesis of potent vasoreactive factors by the pulmonary endothelium in vitro and in vivo, namely ET-1, apelin, and adrenomedullin and may have important implications for human PAH.

Acknowledgments

We thank Dr M. Yan (Genentech Inc, San Francisco) for kindly providing anti-BMP9 and ALK1_ECD reagents. We also thank Professor F. Soubrier and Dr M. Eyries (Laboratoire d’Oncogénétique et Angiogénétique Moléculaire, Groupe Hospitalier Pitié-Salpétrière, Paris, France) for the genetic analysis of pulmonary arterial hypertension patients.

Footnotes