Genotype-phenotype effects of Bmpr2 mutations on disease severity in mouse models of pulmonary hypertension

Abstract Abstract

More than 350 mutations in the type-2 BMP (bone morphogenetic protein) receptor, BMPR2, have been identified in patients with heritable pulmonary arterial hypertension (HPAH). However, only 30% of BMPR2 mutation carriers develop PAH, and we cannot predict which of these carriers will develop clinical disease. One possibility is that the nature of the BMPR2 mutation affects disease severity. This hypothesis has been difficult to test clinically, given the rarity of HPAH and the complexity of the confounding genetic and environmental risk factors. To test this hypothesis, therefore, we evaluated the susceptibility to experimental pulmonary hypertension (PH) of mice carrying different HPAH-associated Bmpr2 mutations on otherwise identical genetic backgrounds. Mice with Bmpr2ΔEx4–5 mutations (Bmpr2^+/−), in which the mutant protein is not expressed, develop less severe PH in response to hypoxia or hypoxia with vascular endothelial growth factor receptor inhibition than mice with an extracellular-domain Bmpr2ΔEx2 mutation (Bmpr2^ΔEx2/+), in which the mutant protein is expressed. This was associated with a marked decrease in stabilizing phosphorylation of threonine 495 endothelial nitric oxide synthase (pThr495 eNOS) in Bmpr2^ΔEx2/+ compared to wild-type and Bmpr2^+/− mouse lungs. These findings provide the first experimental evidence that BMPR2 mutation types influence the severity of HPAH and suggest that patients with BMPR2 mutations who express mutant BMPR2 proteins by escaping non-sense-mediated messenger RNA decay (NMD− mutations) will develop more severe disease than HPAH patients with NMD+ mutations who do not express BMPR2 mutant proteins. Since decreased levels of pThr495 eNOS are associated with increased eNOS uncoupling, our data also suggest that this effect may result from defects in eNOS function.

Patients with heritable pulmonary arterial hypertension (HPAH) inherit heterozygous mutations in the bone morphogenetic protein (BMP) type 2 receptor (BMPR2) locus.¹ BMPR2 mutations account for about 75% of patients with family histories of PAH and 25% of patients with sporadic disease. This establishes BMPR2 mutations as the major genetic determinant of HPAH.² BMPR2 mutation carriers with PAH tend to have an earlier age of diagnosis and more severe pulmonary hemodynamic parameters and are less likely to demonstrate vasoreactivity than noncarriers.^3-6 However, less than 30% of BMPR2 mutation carriers develop clinical disease,⁷ and while the disease is known to have a sex bias and a number of candidate disease-modifying genetic variants at other loci have been shown to influence disease penetrance,² we are still unable to predict which patients carrying BMPR2 mutations will develop overt disease or, if they do, how severe their disease will be.

One possibility is that the nature of the BMPR2 mutation may affect the penetrance and/or severity of disease. More than 350 independent BMPR2 mutations have been identified in patients with HPAH.² The majority of these mutations are non-sense or frame-shift mutations resulting in premature termination of the mutant RNA transcripts predicted to give rise to non-sense-mediated messenger RNA (mRNA) decay (NMD+ mutations), which results in haploinsufficiency.⁸ However, approximately 40% of HPAH-associated BMPR2 mutations are mis-sense or in-frame deletions that are predicted to produce stable mRNA transcripts and express mutant protein products (NMD− mutations).² Since the expressed protein product may have reduced signaling function, it is anticipated that some NMD− mutations may have dominant negative effects on the functional properties of the remaining wild-type allele and that patients carrying these mutations may therefore have higher disease penetrance and/or more severe PAH. This simple hypothesis is confounded by the fact that some of these NMD− mutations, for example mis-sense mutations in the C-terminal cytoplasmic tail domain of BMPR2, may have only minor effects on BMPR2 function,⁹ while others, including mis-sense and in-frame deletions in the extracellular domain of BMPR2, may have more profound effects on cellular function resulting from protein misfolding and retention in the endoplasmic reticulum (ER).^10-12 Despite this, there is some clinical evidence to support the hypothesis that HPAH patients with NMD− BMPR2 mutations have more severe disease than those with NMD+ mutations. Austin et al.¹³ evaluated disease penetrance and survival in HPAH patients in whom the NMD status was determined in cultured, patient-derived lymphoblasts. Patients with NMD− BMPR2 mutations developed clinical disease at an earlier age and had a worse clinical outcome (survival) than those with NMD+ mutations. These findings are supported by those from another study in Chinese HPAH patients¹⁴ but contrast with those from a larger French study, in which no differences in disease severity or penetrance were seen between patients with mis-sense mutations and those with non-sense mutations, splice sites, or large BMPR2 gene rearrangements.¹⁵ This discrepancy may have arisen because a proportion of BMPR2-mutation mRNA products predicted to undergo NMD, when tested in patient-derived cells, do not actually undergo NMD (∼7% from studies using patient-derived lymphoblasts¹³). However, given the rarity of the disease, the range of BMPR2 mutations, and the complexity of confounding genetic and environmental risk factors known or presumed to influence the severity and prevalence of disease in individuals carrying different BMPR2 mutations, it may not be possible to establish clear genotype-phenotype correlations associated with NMD+ versus NMD− BMPR2 mutations from clinical studies alone.

In these studies, we have tested the hypothesis that the nature of the BMPR2 mutation affects the severity of disease in patients with HPAH, using a more controlled experimental approach in mice. We compared the severity of experimentally induced pulmonary hypertension (PH) in mice on otherwise identical genetic backgrounds carrying one of two well-characterized heterozygous, splice-site, germ line Bmpr2 mutations that model the effects of known NMD+ and NMD− BMPR2 mutations in patients with HPAH: the out-of-frame Bmpr2ΔEx4–5 mutation^16-19 and the in-frame, extracellular-domain Bmpr2ΔEx2 mutation.^10,16,20 Previous studies have shown that mice carrying the Bmpr2^ΔEx4-5/+ mutation (which we refer to as Bmpr2^+/− mice) have increased severity of PH, compared with their wild-type littermates, in response to inflammatory stress^21,22 or a combination of serotonin and hypoxia treatment²³ but not hypoxia alone.^22,24 In contrast, we have previously shown that Bmpr2^ΔEx2/+ mutant mice have increased susceptibility to hypoxic PH,²⁵ and while these mice also have increased susceptibility to pulmonary vascular remodeling (but not PH) in response to chronic airway inflammation,²⁶ no other studies have evaluated PH susceptibility in Bmpr2^ΔEx2/+-mutant mice. These studies suggest that the two mutations might confer differential susceptibility to PH in response to chronic hypoxia. However, the two mouse lines were maintained on different backgrounds for those hypoxia studies (Bmpr2^+/− mice were maintained on a C57Bl/6 background and Bmpr2^ΔEx2/+ mice on a mixed Balb/cJx129ScJ background), and the studies were performed on both male and female mice.^24,25 Since the sex and genetic background of the mice influence susceptibility to PH,^27-30 it is possible that there are sex- and/or strain-dependent interactions with the Bmpr2 genotypes that were confounding these responses. Therefore, for these studies we compared susceptibility of the two Bmpr2 mutant lines to PH after backcrossing both strains onto the same C57Bl/6 background. This strategy ensures that phenotypic differences in PH responses between the mutant strains must have been the result of differences in their Bmpr2 mutation status. Our findings provide the first definitive evidence that different Bmpr2 mutations influence PH disease severity in vivo, and they also suggest that increased disease severity in HPAH patients with NMD− mutations in the extracellular domain of BMPR2 may result from defective endothelial nitric oxide synthase (eNOS) function in the vasculature.

Methods

Chemicals and reagents

The following primary antibodies were used for Western blots, at a 1∶1,000 dilution unless otherwise noted: mouse monoclonal anti-eNOS (BD Biosciences), anti-phospho-eNOS (Thr495; BD Biosciences), and anti-β-actin (Sigma-Aldrich); mouse monoclonal anti-BMPR2 antibodies raised against the C-terminal BMPR2 residues 803–996 (Clone 18, BD Biosciences); and rabbit polyclonal anti-phospho-eNOS (Ser1177). Anti-mouse horseradish peroxidase (KPL) and anti-rabbit horseradish peroxidase (Cell Signaling) secondary antibodies were used at a dilution of 1∶2,000. Primary antibodies mouse monoclonal anti-α-smooth muscle actin (α-SMA; Sigma-Aldrich; 1∶400 dilution), rabbit polyclonal anti-PCNA (proliferating cell nuclear antigen; Santa-Cruz; 1∶50 dilution), and anti–von Willebrand factor (Dako; 1∶50 dilution) were used to assess vessel proliferation and muscularization. Secondary antibodies used for these studies included donkey anti-rabbit-IgG-Rhodamine Red (Jackson Immunoresearch; 1∶300 dilution) and horse anti-mouse IgG-fluorescein (Vector Labs; 1∶200 dilution). ProLong Gold antifade reagent with DAPI mounting media (Life Technologies) was used. Avertin (Sigma-Aldrich) with 2-methyl-2-butanol (Sigma-Aldrich) was used as light anesthesia for mice. SU5416 (Tocris) was diluted in CMC solution (0.5% [w/v] carboxymethylcellulose sodium [CMC; Sigma-Aldrich], 0.9% [w/v] sodium chloride [Sigma-Aldrich], 0.4% [v/v] polysorbate 80 [Sigma-Aldrich], and 0.9% [v/v] benzyl alcohol [Sigma-Aldrich] in deionized water).

Mouse lines

Bmpr2^ΔEx2/+ mice have been described previously.^10,20,25 Mice were backcrossed onto a C57Bl/6J background for more than 9 generations. Genotyping was performed by polymerase chain reaction (PCR) from ear-punch DNA using the following primers: common forward: CCATGCTCTTTTGAAGATGG; wild-type reverse: GTCCCCTTTTGATTTCTCCCA, producing a 1-kb wild-type product; and mutant reverse: GGCCGCTTTTCTGGATTCATC, producing a 700-bp mutant product. Bmpr2^+/− mice have been described previously.^18,19,24 Genotyping was performed by PCR from ear-punch DNA using three primers: common forward: GCTAAAGCGCATGCTCCAGA CTGCCTTG; wild-type reverse: TCACAGCATGAACATGATGGAGGCG G, producing a 200-bp product; and mutant reverse AGGTTGGCCTGGAACCTGAGGAAATC, producing a 260-bp product. Mice used in studies were approved by the Vanderbilt University Institutional Animal Care and Use Committee under protocol M/11/015, and experiments were adherent with the National Institutes of Health guidelines for care and use of laboratory animals under Vanderbilt animal welfare assurance license A3227–01.

Experimental PH

We used 8–12-week-old adult male Bmpr2^+/− and Bmpr2^ΔEx2/+ mice as well as wild-type littermate controls. To induce hypoxic PH, mice were maintained in room air (21% O₂) or in a normobaric hypoxia chamber (10% O₂, 90% N₂), and hemodynamics was evaluated after 3 weeks. To induce SU5416/hypoxic PH, mice were injected subcutaneously with SU5416 (Tocris) suspended in 0.5% CMC solution (0.5% [w/v] CMC [Sigma-Aldrich], 0.9% [w/v] sodium chloride [Sigma-Aldrich], 0.4% [v/v] polysorbate 80 [Sigma-Aldrich], and 0.9% [v/v] benzyl alcohol [Sigma-Aldrich] in deionized water) once per week at 20 mg/kg for 3 weeks, as previously described.^31,32 The animals were maintained in room air (21% O₂) or in a normobaric hypoxia chamber (10% O₂, 90% N₂) for the duration of the 3 weeks. Hemodynamics was evaluated 1 week after the last injection of SU5416. For this, mice were anesthetized with 375 mg Avertin (Sigma) per kg body weight. Right ventricular systolic pressure (RVSP) was measured directly by right heart catheterization with a PVR-1035 pressure/volume catheter (Millar Instruments) through the surgically exposed right jugular vein. Pressure data were recorded for 10–20 seconds with the MPVS-300 Powerlab System (ADInstruments) and analyzed with LabChart Pro7 software (ADInstruments). After completion of hemodynamic measurements, the chest cavities were opened and terminal blood samples collected by cardiac puncture for measurement of hematocrit. After this, the left lung was clamped, excised, and snap-frozen for later use in protein expression analysis. The right lung was inflated with 10% formalin, fixed overnight, and then used for histological analyses. The heart was excised for assessment of right ventricle (RV) hypertrophy. For this, the RV free wall was dissected from the left ventricle (LV) and septum, dried for 2 days at 55°C, and weighed. RV hypertrophy was determined by dividing the dry weight of the RV by the dry weight of the LV and septum.

Histological evaluation of pulmonary vascular remodeling

To assess the thickness of resistance-level pulmonary vessels, 5-μm lung tissue sections were stained with hematoxylin and eosin (H&E) or Van Gieson elastin stain, and images were captured digitally by researchers blinded to genotype and treatment group. Ten 40× fields from 4 mice per group were evaluated, and all studies were performed while blinded to treatment and genotype. Vessels measuring 20–100 μm in outer diameter that had a length less than twice the width and were associated with terminal bronchi were classified as peribronchial vessels. Smaller, rounded vessels with an outer diameter between 20 and 50 μm that were distal from the bronchioles were classified as intrapulmonary vessels. To measure vessel thickness, the external diameter was determined by measuring the transluminal distance between the external elastic laminae (ED). Measurements were taken at the widest diameter. In the same area as the ED measurement, the distance between the outer and inner elastic laminae was measured on each side of the vessel (OD and ID, respectively). Vessel thickness was then calculated as [(OD − ID)/ED] × 100 and expressed as percent of wall thickness. Co-immunofluorescence was used to assess vessel muscularization and proliferation on 5-μm lung tissue sections, as previously described.^25,33,34 Briefly, to assess muscularization of the peripheral vessels, lung tissue sections were stained with mouse anti-α-SMA to identify smooth muscle cells and with rabbit anti–von Willebrand factor to identify endothelial cells. Peripheral vessels were defined as small vessels less than 100 μm in diameter, distal to muscularized bronchioles. Vessels were classified as nonmuscularized (vessels without SMA stain) or muscularized (1 or more SMA-stained cells). Proliferation was assessed after 1 week of hypoxia-and-SU5416 treatment, because previous studies have shown that the 1-week time point has the highest rate of proliferation and little proliferation is observed after 3 weeks in this model. Double immunofluorescence staining with rabbit-anti-PCNA and mouse anti-α-SMA was used to determine the proliferation index of the pulmonary vasculature, as previously described.³⁵ Sections were counterstained with DAPI (Life Technologies). Proliferation index was determined by counting the proportion of PCNA-positive, α-SMA-positive cells, for smooth muscle cells, or PCNA-positive and α-SMA-negative cells internal to α-SMA-positive cell staining, for endothelial cells, in more than 40 vessels. Results are expressed as the mean of 4 animals per group, with 10 fields per section and 2 sections per animal.

Western blot analyses

Lung tissue was homogenized with a motorized disposable pestle system (Fisher Scientific) with 1.0-mm disruption beads (Fisher Scientific) in ice-cold lysis buffer (LB) made up of 25 mM HEPES, 150 mM NaCl, 5 mM EDTA (ethylenediaminetetraacetic acid), 1% Triton X-100, 10% glycerol with additional proteinase inhibitor cocktail and phosphatase inhibitor cocktails 2 and 3 (Sigma-Aldrich). After homogenization, lysate was rocked at 4°C for 30 minutes and centrifuged, and the supernatant was stored at −80°C. For cell culture studies, cells were lysed in ice-cold LB with protease and phosphatase inhibitors and centrifuged, and the supernatants were stored, as previously described.¹⁹ Protein concentration was measured with the DC Protein Assay (Bio-Rad). Primary antibodies were used at 1∶1,000 in 5% nonfat dry milk in TBST (25 mM Tris, 1 M NaCl, 1% Tween 20), except phosphospecific antibodies, which were diluted in 5% bovine serum albumin in TBST. Secondary antibodies were diluted 1∶2,000 in 5% nonfat milk in TBST.

Mouse pulmonary endothelial cell (PEC) culture

We utilized frozen stocks of conditionally immortalized lines of PECs that were isolated from three wild-type (WT1, WT2, and WT3) mice, three Bmpr2^+/− (N1, N3, and N6) mice, and one Bmpr2^ΔEx2/+ mouse crossed with H-2Kb-tsA58 SV40 large T antigen transgenic mice (“Immortomouse,” Charles River), as previously described.^10,19 The endothelial cell phenotype of these PECs was confirmed by morphology, FACS (fluorescence-activated cell sorting) analysis for VCAM (vascular cell adhesion molecule) and EPCR (endothelial protein C receptor) expression, and three-dimensional culture tube formation.¹⁹ Cells were cultured in complete microvascular endothelial medium EGM-2MV (Lonza) and expanded under permissive conditions in EGM-2MV + 10 units/mL mouse interferon γ (IFN-γ; Thermo Fisher Scientific) at 33°C before transferring to 37°C without IFN-γ for 3 days to inhibit SV40 large T antigen activity before lysis.

Statistics

Results are expressed as means ± SEM. Statistical analyses were performed by one-way analysis of variance, with multiple between-group comparisons using post-hoc Bonferroni correction for individual comparisons, in GraphPad Prism 5 software. Significance is indicated when P < 0.05.

Results

Expression of Bmpr2-mutant allelic products in Bmpr2^+/− and Bmpr2^ΔEx2/+ mice

In order to determine whether Bmpr2^+/− and Bmpr2^ΔEx2/+ mice model the effects of NMD+ and NMD− BMPR2 mutations, respectively, we compared Bmpr2 protein expression in cultured PECs and whole-lung lysates from wild-type, Bmpr2^+/−, and Bmpr2^ΔEx2/+ mice (Fig. 1). Both Bmpr2-mutant lines had reduced expression of the 150-kDa wild-type Bmpr2 allelic product when compared with wild-type PECs, but only Bmpr2^ΔEx2/+ PECs expressed a 130-kDa band (Fig. 1A). Previous studies from our laboratory have confirmed the identity of this 130-kDa band as the Bmpr2ΔEx2 mutant product.¹⁰ Comparison of Bmpr2 expression in whole-lung lysates from the two Bmpr2-mutant mouse lines confirms that there is also reduced Bmpr2 expression in Bmpr2^+/− mouse lungs. However, while there is lower and more variable expression of the 150-kDa wild-type product in all of the lung samples when compared with PEC lysates, we were able to detect 130-kDa bands corresponding to the Bmpr2ΔEx2 mutant product in two-thirds of the Bmpr2^ΔEx2/+ mouse lung lysates (Fig. 1B, lanes 7 and 8). These findings indicate that while Bmpr2^+/− mice do not express detectable Bmpr2ΔEx4–5 mutant protein, Bmpr2^ΔEx2/+ mice express the Bmpr2ΔEx2 mutant protein both in vitro in cultured PECs and in vivo in the intact lung. Since mutations subject to NMD do not express mutant protein products,⁸ these findings indicate that Bmpr2^+/− mice model the molecular effects of HPAH-associated NMD+ BMPR2 mutations, while Bmpr2^ΔEx2/+ mice model the molecular effects of NMD− BMPR2 mutations in patients with HPAH.

Figure 1.

Expression of Bmpr2 mutant products in Bmpr2^+/− and Bmpr2^ΔEx2/+mutant mice. Western blots for Bmpr2 detected with a Bmpr2 monoclonal antibody raised against a C-terminal Bmpr2 peptide, in pulmonary endothelial cells (PECs; A) and whole-lung lysates (B) from wild-type, Bmpr2^+/−, and Bmpr2^ΔEx2/+ mice, as indicated. We used PECs isolated from 3 wild-type mice (WT1–3) and 3 Bmpr2^+/− mice (N1, N3, and N6) and triplicate cultures of PECs obtained from one Bmpr2^ΔEx2/+ mutant mouse (EC6). Whole-lung lysates were all obtained from different mice. Bmpr2: bone morphogenetic protein receptor 2.

Differential susceptibilities of Bmpr2^+/− and Bmpr2^ΔEx2/+ mouse mutants to PH

Since sex and genetic background have strong effects on susceptibility to PH in mice,^27-30 we first backcrossed both the Bmpr2^ΔEx2/+ and Bmpr2^+/− mutant lines for more than 9 generations onto the same C57Bl/6 background and performed all studies in male mice, since males show enhanced PH responses to chronic hypoxia. As previously reported,^23,25 there were no differences in RVSP, a measure of PH response in mice, between wild-type mice and either of the Bmpr2-mutant mice under normoxic conditions, but Bmpr2^ΔEx2/+ mice had increased RVSP after 3 weeks of 10% oxygen, compared with wild-type but not with Bmpr2^+/− mice (Fig. 2A). These data are consistent with a previous report from our laboratory indicating that Bmpr2^ΔEx2/+ mice have increased susceptibility to hypoxic PH compared with wild-type littermates,²⁵ but they do not establish whether this mutation increases susceptibility to PH when compared with the Bmpr2^+/− mutation.

Figure 2.

Differential susceptibility of Bmpr2^+/− and Bmpr2^ΔEx2/+ mutant mice to pulmonary hypertension. A, B, Right ventricular systolic pressure (RVSP) measurements in male wild-type (WT), Bmpr2^+/− (+/−), and Bmpr2^ΔEx2/+ (ΔE2/+) mutant mice all maintained on a pure C57Bl/6 background (>9 generations) after exposure to normoxia or 10% oxygen for 3 weeks (A) or treatment with the VEGFR2 (vascular endothelial growth factor receptor 2) antagonist SU5416 (or vehicle) subcutaneously at 20 mg/kg weekly ± 10% oxygen for 3 weeks, as indicated (B). Individual data points are shown, with means ± SEM indicated. C–E, Physiological characteristics of WT, Bmpr2^+/−, and Bmpr2^ΔEx2/+ mice exposed to normoxia or weekly SU5416 ± 10% oxygen, as indicated. Values shown are means ± SEM. C, Average heart rate recorded during measurement of RVSP over 5–10 seconds in 9 normoxic control, 8 WT, 9 Bmpr2^+/−, and 8 Bmpr2^ΔEx2/+ mice treated with 10% oxygen and SU5416 for 3 weeks. D, Fulton index (right ventricle weight/(left ventricle + septum weight)) measured immediately after RSVP studies in 8 mice per group. E, Hematocrit from terminal cardiac punctures obtained after RSVP studies in 8 mice per group. One-way analysis of variance with Bonferroni correction for between-group comparisons: *P < 0.05; **P < 0.01; ***P < 0.005; ^#P < 0.0001. Bars indicate statistically significant between-group differences. For example, in A, Bmpr2^+/− and Bmpr2^ΔEx2/+ mice under normoxic conditions are both independently significantly different from WT, Bmpr2^+/−, and Bmpr2^ΔEx2/+ mice exposed to 10% oxygen for 3 weeks with a probability of P < 0.05 after Bonferroni correction.

Since the PH response to hypoxia in mice is relatively mild, studies focusing only on hypoxia may fail to detect minor differences in RVSP between genotypes; therefore, we also evaluated a mouse model of PH in which exposure to hypoxia was combined with the use of the VEGFR2 (vascular endothelial growth factor receptor 2) inhibitor SU5416.³¹ Under these conditions, the majority of mice had RSVP values higher than 45 mmHg, and while there was overlap between genotypes, there was a marked increase in RVSP in Bmpr2^ΔEx2/+ mice, compared with wild-type and Bmpr2^+/− mice (Fig. 2B). This was associated with increased RV hypertrophy in Bmpr2^ΔEx2/+ mice, determined by the RV-to-(LV + septum) weight ratios (Fig. 2D), but there were no differences in heart rates or hematocrit levels between genotypes (Fig. 2C, 2E). These data indicate that Bmpr2^ΔEx2/+ mice develop more severe PH and RV hypertrophy than Bmpr2^+/− mice after treatment with SU5416 and hypoxia.

Increased pulmonary vascular remodeling in Bmpr2^ΔEx2/+ mice

SU5416 plus hypoxia induces marked pulmonary vascular remodeling in mice.³¹ We therefore evaluated whether genotype-dependent effects on PH responses in this model are associated with differences in pulmonary vascular remodeling. Resistance-level peribronchiolar vessel walls showed progressively increased thickening in Bmpr2^+/− and Bmpr2^ΔEx2/+ mice exposed to hypoxia and SU5416 for 3 weeks (Fig. 3A), but vessel wall thickness in small intrapulmonary vessels was significantly increased in Bmpr2^ΔEx2/+ mice compared with that in both wild-type and Bmpr2^+/− mice (Fig. 3B). This was associated with a right shift in the distribution of wall thicknesses of these small intrapulmonary vessels in Bmpr2^ΔEx2/+ mice (Fig. 3C). Of the intrapulmonary vessels from Bmpr2^ΔEx2/+ mice exposed to hypoxia and SU5416, 15.6% showed more than 50% occlusion, compared with 2.9% in Bmpr2^+/− and none in wild-type mice. Vessel wall thickening was associated with expansion of the medial layer, with no evidence of neointimal thickening or vascular lumen obstruction on elastin- or H&E-stained tissue sections (Fig. 3D–3K). In addition, there was no difference in small intrapulmonary vessel numbers between wild-type and either of Bmpr2 mutant mouse lines under normoxic conditions or after treatment with SU5416 and hypoxia (Fig. S1), suggesting that this model does not result in vessel “drop-out.”

Figure 3.

Increased pulmonary vascular remodeling in Bmpr2^ΔEx2/+ mice with pulmonary hypertension. Data obtained from wild-type (WT), Bmpr2^+/− (+/−), and Bmpr2^ΔEx2/+ (ΔE2/+) mice under normoxic conditions or after treatment with SU5416 and 10% oxygen, as indicated. All results are expressed as means ± SEM (A, B, L–N) or as individual data points (C). A–K, Vessel wall thickness. A–C, Vessel wall thickness (mean of two orthogonal outer vessel diameter − inner vessel diameter measurements expressed as the percentage of outer diameter) was measured in 10 round or oval sections of 20–100-μm peribronchiolar vessels (A), and 10 20–50-μm-diameter intrapulmonary vessels (B, C) per mouse in 4 WT controls and 6 WT, 7 Bmpr2^+/−, and 5 Bmpr2^ΔEx2/+ mice treated with SU5416 and hypoxia for 3 weeks. C indicates the range of vessel wall thicknesses measured under different conditions. Individual data points represent the percentage of total vessels measured in each group, with vessel wall thicknesses within the indicated ranges (0–9, 10–19, 20–29, 30–39, 40–49, 50–59, 60–69, 70–79, 80–89, 90–100). D–K, Representative images from hematoxylin and eosin (D–G)– or Van Gieson (elastin; H–K)-stained lung tissue sections, showing increasing small-interacinar-vessel wall thickening in WT (E, H), Bmpr2^+/− (F, J), or Bmpr2^ΔEx2/+ (G, K) mice exposed to SU5416 and 10% oxygen for 3 weeks compared with walls in WT normoxic controls (D, H). Scale bars: 50 μm. L, Peripheral muscularization after 3 weeks of treatment with SU5416 and 10% oxygen. Lung sections underwent two-color immunofluorescence staining for Von Willebrand factor (endothelial cells) and α-SMA (smooth muscle cells). The percent of the circumference of vessels covered with smooth muscle cells was determined in 20 round or oval sections of 20–50-μm-diameter interacinar vessels per mouse in 4 WT control, 6 wild-type, 6 Bmpr2^+/−, and 5 Bmpr2^ΔEx2/+ mice treated with SU5416 and 10% oxygen for 3 weeks. M, N, Pulmonary vascular proliferation after 1 week of treatment with SU5416 and 10% oxygen. Lung sections were colabeled with anti-PCNA and α-SMA antibodies. Proliferating endothelial and vascular smooth muscle cells were identified as internal to or within α-SMA domains, respectively. Proliferation index (PCNA/DAPI), defined as the proportion of each cell type staining with PCNA, was determined in 3 WT control, 5 WT, 4 Bmpr2^+/−, and 4 Bmpr2^ΔEx2/+ mice treated with SU5416 and hypoxia for 1 week. One-way analysis of variance with Bonferroni correction for between-group comparisons: *P < 0.05; **P < 0.01; ***P < 0.005; ^#P < 0.0001. Bars indicate statistically significant between-group differences. α-SMA: α–smooth muscle actin; PCNA: proliferating cell nuclear antigen.

Peripheral muscularization, another measure of pulmonary vascular remodeling, was evaluated by determining the percent coverage of small intrapulmonary vessels with α-SMA-positive smooth muscle cells. There was an increase in the percent muscularization of small intrapulmonary vessels from Bmpr2^ΔEx2/+ mice compared with that in wild-type mice treated with SU5416 and hypoxia but no differences in peripheral muscularization between Bmpr2^+/− and Bmpr2^ΔEx2/+ mice under the same conditions (Fig. 3L). This was associated with an increase in pulmonary vascular smooth muscle but not endothelial cell proliferation in Bmpr2^ΔEx2/+ mice, compared with Bmpr2^+/− mice treated with SU5416 and hypoxia (Fig. 3M, 3N). These data indicate that Bmpr2^ΔEx2/+ mice have increased pulmonary vascular remodeling compared with wild-type and Bmpr2^+/− mice.

Differential regulation of eNOS phosphorylation sites in Bmpr2^+/− and Bmpr2^ΔEx2/+ mouse lungs

One of the features of endothelial dysfunction in both experimental PH and patients with PAH is the impaired production of nitric oxide (NO) by eNOS.^36-39 This is associated with uncoupling of eNOS, resulting not only in reduced NO production but also in increased production of pathogenic reactive oxygen species.⁴⁰ Since the balance of activating phosphorylation of Ser1177 (pSer1177) and inhibitory phosphorylation of Thr495 (pThr495) in eNOS regulates eNOS activity,^41-43 we compared the levels of eNOS, as well as pSer1177 and pThr495 eNOS levels, in wild-type, Bmpr2^+/−, and Bmpr2^ΔEx2/+ mouse lungs (Fig. 4). Unexpectedly, while activating pSer1177 eNOS levels were reduced in both Bmpr2^+/− and Bmpr2^ΔEx2/+ mouse lungs (Fig. 4A, 4B), this was associated with a marked reduction in pThr495 eNOS in Bmpr2^ΔEx2/+ mouse lungs only (Fig. 4A, 4C). There was no significant difference in total eNOS expression levels between genotypes (Fig. 4A, 4D).

Figure 4.

Regulation of endothelial nitric oxide synthase (eNOS) in mouse lungs from wild-type (WT), Bmpr2^+/− (+/−), and Bmpr2^ΔEx2/+ (ΔE2/+) mice. A, Western blots of lung lysates, demonstrating phosphorylated S1177 and T495 (pS1177 and pT495, respectively) and total eNOS levels and β-actin. Molecular weight markers are indicated. B–D, Quantification of band densities corrected for total protein or β-actin loading, as indicated. Results are expressed as means ± SEM, 4 mice/group, normalized to WT controls. One-way analysis of variance with Bonferroni correction for between-group comparisons: *P < 0.05; **P < 0.01; ^***P < 0.005. Bars summarize the levels of significance for between-group differences.

Discussion

These studies provide the first definitive evidence that the nature of heterozygous germ line Bmpr2 mutations influences the severity of PH in response to different stimuli. We show that mice carrying the NMD− Bmpr2^ΔEx2/+ mutation have increased susceptibility to PH in response to chronic hypoxia and that they develop more severe PH in response to chronic hypoxia and VEGFR2 blockade than mice carrying the heterozygous null, NMD+ Bmpr2^ΔEx4-5/+ mutation (referred to as Bmpr2^+/−). These findings provide clear experimental evidence that supports clinical observations that HPAH patients with NMD+ BMPR2 mutations develop less severe PAH than patients carrying NMD− BMPR2 mutations.^13,14 These findings also suggest that the failure to detect differences in disease severity according to BMPR2 mutation type in another clinical study¹⁵ is likely to reflect differences in patient ascertainment and/or the stringency by which the NMD status of these mutations was defined in different studies. These findings also have implications for future clinical-outcome studies and studies of disease-modifying factors, treatment, and genetic variants, since they indicate the importance of classifying HPAH patients according to their BMPR2-mutation NMD status, as recently described.⁴⁴

Cultured PECs from Bmpr2^+/− mice express lower levels of the wild-type Bmpr2 protein than those from their wild-type littermates,¹⁹ and we have used specific antibodies against the peptide sequence encoded by exon 2 of Bmpr2 to show that PECs from Bmpr2^ΔEx2/+ mice express an additional lower-molecular-weight Bmpr2 band corresponding to the expressed Bmpr2ΔEx2 mutant protein product.¹⁰ Our studies using cultured PECs confirm these findings in vitro, and in vivo we show that two-thirds of the Bmpr2^ΔEx2/+ mice also express a 130-kDa band on Bmpr2 Western blots that corresponds to the Bmpr2ΔEx2 mutant product. Since the Bmpr2^+/− mouse mutation corresponds to an out-of-frame, NMD+ BMPR2ΔEx4–5 mutation described in patients with HPAH^16-19 and the Bmpr2^ΔEx2/+ mouse mutation corresponds to an in-frame, extracellular domain, NMD− BMPR2ΔEx2 mutation also described in patients with HPAH,^10,16,20 these findings support our hypothesis that Bmpr2^+/− and Bmpr2^ΔEx2/+ mice model the molecular effects of NMD+ and NMD− BMPR2 mutations, respectively, in patients with HPAH. Our data also show that the level of expression of Bmpr2 in mouse lungs is lower than that in cultured PECs, suggesting that PECs, which constitute only a fraction of the total cells in whole lung, express higher levels of Bmpr2 than the other cellular compartments. This is consistent with the hypothesis that the pulmonary endothelium is the primary target of vascular injury in HPAH patients and suggests that the selective changes in the levels of inhibitory, stabilizing eNOS phosphorylation in Bmpr2^ΔEx2/+ mouse lungs that we observed may reflect Bmpr2-dependent defects on pulmonary endothelial function in vivo.

To determine whether these Bmpr2 mutations alter susceptibility to PH, we exposed mice to two different PH induction protocols. Chronic hypoxia is a contributing factor to PH associated with chronic lung disease and hypoxia in humans (group 3 PH).⁴⁵ In mice, exposure to prolonged hypoxia causes mild PH and reversible pulmonary vascular remodeling. We have also shown that chronic hypoxia activates BMP signaling and induces expression of the Bmpr2 ligands BMP2 and BMP4 in the pulmonary vasculature,^34,35 so that this model of PH, while an imperfect model of human disease,⁴⁶ is likely to be influenced by defects in BMP signaling and may provide insight into differential effects of the two mutations on this BMP-dependent response. Our findings that Bmpr2^ΔEx2/+-mutant mice develop slightly more severe hypoxic PH than their wild-type littermates are consistent with those of our earlier studies.²⁵ In contrast, and consistent with earlier reports,^22,24 we saw no significant increase in PH between wild-type and Bmpr2^+/− mice exposed to hypoxia. This supports the hypothesis that expression of the mutant allele in Bmpr2^ΔEx2/+ mice increases susceptibility to PH when compared with that in Bmpr2^+/− mice, which are heterozygous null and cannot express any mutant allelic product. However, RSVP measurements were mildly increased in hypoxic Bmpr2^+/− mice compared with their wild-type littermates. This raised the question as to whether, because of the relatively mild PH induced by exposure to prolonged hypoxia, these studies might have failed to detect minor but biologically significant effects of the Bmpr2^+/− mutation on PH responses.

To address this, we exposed mice to a second PH-inducing protocol: chronic hypoxia combined with VEGFR2 inhibition using the small-molecule inhibitor SU5416. This model was originally described as an angio-obliterative model of severe PH in rats that models many of the features of human PAH.^47,48 More recently, this model has been developed in mice.³¹ While mice do not develop the severe angio-obliterative PH seen in rats, a number of labs have shown that they develop moderate obstructive vasculopathy and RVSPs substantially higher than those seen after hypoxia alone.^31,49-53 Our data confirm that mice develop consistently higher RVSPs with SU5416 and hypoxia than in response to hypoxia alone, and they now reveal a clear increase in PH responses in Bmpr2^ΔEx2/+ mice when compared with both wild-type mice and Bmpr2^+/− mutants. While there was a spread in PH responses between mice, Bmpr2^ΔEx2/+ mice showed the highest RVSP measurements: 3 of 9 Bmpr2^ΔEx2/+ mice had mean RVSPs of >80 mmHg, similar to those found in patients with PAH. Pulmonary vascular remodeling, as measured by small-vessel wall thickness and peripheral muscularization, was also increased in Bmpr2^ΔEx2/+ mice, and there was increased vascular smooth muscle cell proliferation in Bmpr2^ΔEx2/+ mice. In contrast with findings in the rat model of SU5416 and hypoxia, we were unable to see any evidence of neointimal thickening or obstructive vasculopathy. However, 15% of small intrapulmonary vessels in Bmpr2^ΔEx2/+ mice treated with SU5416 and hypoxia had more than 50% luminal occlusion, compared with only 3% of Bmpr2^+/− mice and none of the wild-type mice under the same conditions, suggesting that the degree of structural remodeling might be sufficient to account for differences in pulmonary vascular resistance between genotypes.

To explore the molecular mechanisms by which the Bmpr2^ΔEx2/+ mutation might increase the severity of PH in response to different stimuli, we compared expression levels of eNOS as well as levels of the dominant activating and inhibitory eNOS phosphorylation events on Ser1177 and Thr495 in lung lysates. Previous studies from our laboratory have shown that there is a reduction in endothelial-dependent vasodilation in small, resistance-level intrapulmonary artery preparations associated with reduced activating phosphorylation of eNOS at Ser1177 under basal conditions as well as reduced levels of hypoxia-induced eNOS in Bmpr2^ΔEx2/+ mouse lungs compared with those of their wild-type littermates.²⁵ In these studies, we have shown that reduced levels of pSer1177 eNOS are found in both Bmpr2^+/− and Bmpr2^ΔEx2/+ mouse lungs, compared with those of wild-type littermates. These findings are consistent with the observation that BMP ligands enhance NO production and induce phosphorylation of Ser1177 eNOS in cultured PECs^34,54 and that there are decreased eNOS activity and pSer1177 eNOS levels in the pulmonary vasculature of Bmpr2^ΔEx2/+ mutant mice and in PECs from HPAH patients carrying BMPR2 mutations.^25,54 However, these findings also suggest that the differential regulation of pSer1177 eNOS is unlikely to account for differences in PH severity in Bmpr2^+/− and Bmpr2^ΔEx2/+ mice. However, unlike Bmpr2^+/− mouse lungs, there was also a profound reduction in inhibitory pThr495 eNOS in Bmpr2^ΔEx2/+ mouse lungs. Since persistent dephosphorylation of Thr495 also results in eNOS uncoupling, which is associated with reduced NO production and increased production of reactive oxygen species,⁵⁵ increased severity of PH in Bmpr2^ΔEx2/+ mice could result from dephosphorylation of Thr495 eNOS. These findings contrast with a recent human study in which pThr495 eNOS levels were shown to be increased in cultured PECs from patients with PAH.⁵⁶ However, none of the patients in these studies had defined BMPR2 mutations that are predicted behave like the Bmpr2ΔEx2 mutation. In addition, the authors saw increased pThr495 eNOS levels only in plexiform lesions in the lungs of patients with PAH,⁵⁶ and while plexiform lesions are characteristic of PAH lungs in humans,⁵⁷ we found no evidence of these proliferative endothelial lesions in the lungs of wild-type or Bmpr2-mutant mice after treatment with SU5416 and hypoxia.

The mechanism by which the Bmpr2ΔEx2 mutant product interferes with inhibitory Thr495 eNOS phosphorylation remains to be determined. However, we have shown that PECs from Bmpr2^ΔEx2/+ mice express a shortened Bmpr2ΔEx2 product lacking exon 2 and that this protein is mislocalized and retained in the ER.¹⁰ Since eNOS colocalizes and interacts with BMPR2 in cultured endothelial cells,⁵⁴ one possibility, therefore, is that binding of eNOS to the misfolded Bmpr2ΔEx2 mutant product results in abnormal sequestration of eNOS in the ER, where it is no longer accessible to the kinases and phosphatases that normally regulate eNOS phosphorylation.^43,58 This hypothesis might account for the more profound effects of the Bmpr2^ΔEx2/+ mutation on inhibitory Thr495 eNOS phosphorylation, compared with the Bmpr2^+/− mutation.

In summary, our studies provide the first experimental evidence of a genotype-phenotype relationship between disease severity and Bmpr2 mutations in experimental PH. Mice carrying a Bmpr2 mutation that is expressed as a misfolded mutant protein develop more severe PH than mice with a Bmpr2 mutation that does not express as a mutant protein product. Molecular analyses suggest that these effects may result from mutation-specific effects on inhibitory phosphorylation of eNOS predicted to promote eNOS uncoupling and increase oxidative stress. Since we and others have shown that treatment with chemical chaperones increases trafficking of the misfolded BMPR2 mutant products from the ER to the plasma membrane,^10,12 we anticipate that therapeutic interventions designed to enhance protein folding may improve PEC function and PH severity in patients carrying these mutations.

Appendix.

Figure S1.

Pulmonary vessel density in wild-type (WT), Bmpr2^+/− (+/−), and Bmpr2^ΔEx2/+ (ΔE2/+) mice treated with SU5416 and 10% oxygen for 3 weeks; two-color immunofluorescence staining of lung sections for Von Willebrand factor and α-smooth muscle actin. A, Alveolar density, determined by increasing the FITC (fluorescein isothiocyanate) channel fluorescence gain and counting the average number of alveolae per HPF (high-power field). B, Vessel density. Numbers of both muscularized and nonmuscularized 20–50-μm round or oval intra-acinar vessels were counted per HPF and presented as the ratio of vessel/alveola numbers. All results are expressed as means ± SEM. There were 3 normoxic controls as well as 5 WT, 6 Bmpr2^+/−, and 4 Bmpr2^ΔEx2/+ mice treated with SU5416 and 10% oxygen for 3 weeks. One-way analysis of variance shows no significant between-group differences.

Source of Support: This research was supported by National Institutes of Health grant R01-HL093057 and American Heart Association grant 16GRNT31310011.

Conflict of Interest: None declared.

Supplements

• Appendix ^{(512.4KB, pdf)}