Interaction between bone morphogenetic protein receptor type 2 and estrogenic compounds in pulmonary arterial hypertension

Abstract Abstract

The majority of heritable pulmonary arterial hypertension (HPAH) cases are associated with mutations in bone morphogenetic protein receptor type 2 (BMPR2). BMPR2 mutation carries about a 20% lifetime risk of PAH development, but penetrance is approximately three times higher in females. Previous studies have shown a correlation between estrogen metabolism and penetrance, with increased levels of the estrogen metabolite 16α-hydroxyestrone (16αOHE) and reduced levels of the metabolite 2-methoxyestrogen (2ME) associated with increased risk of disease. The goal of this study was to determine whether 16αOHE increased and 2ME decreased penetrance of disease in Bmpr2 mutant mice and, if so, by what mechanism. We found that 16αOHE∶2ME ratio was high in male human HPAH patients. Bmpr2 mutant male mice receiving chronic 16αOHE had doubled disease penetrance, associated with reduced cardiac output. 2ME did not have a significant protective effect, either alone or in combination with 16αOHE. In control mice but not in Bmpr2 mutant mice, 16αOHE suppressed bone morphogenetic protein signaling, probably directly through suppression of Bmpr2 protein. Bmpr2 mutant pulmonary microvascular endothelial cells were insensitive to estrogen signaling through canonical pathways, associated with aberrant intracellular localization of estrogen receptor α. In both control and Bmpr2 mutant mice, 16αOHE was associated with suppression of cytokine expression but with increased alternate markers of injury, including alterations in genes related to thrombotic function, angiogenesis, planar polarity, and metabolism. These data support a causal relationship between increased 16αOHE and increased PAH penetrance, with the likely molecular mechanisms including suppression of BMPR2, alterations in estrogen receptor translocation, and induction of vascular injury and insulin resistance–related pathways.

Introduction

Pulmonary arterial hypertension (PAH) is a disease characterized by progressively increasing pulmonary vascular resistance (PVR), probably caused by dropout or narrowing of small blood vessels in the lungs, eventually leading to right heart failure. Current best therapies do not reverse and may not slow progression of disease. The heritable form is usually associated with mutations in bone morphogenic protein (BMP) receptor type 2 (BMPR2).^1,2 The risk of developing PAH for those carrying a BMPR2 mutation is ∼10% for men and 30% for women.

Female sex is the strongest and best-established risk factor for PAH, with a female-to-male ratio of ∼3∶1 in many etiologies, including idiopathic, heritable, and scleroderma associated.^3-5 Recently, we found that the method of estrogen metabolism might be the key to understanding disease penetrance in human patients. Parent compound estrogens in both men and women are predominantly metabolized through hydroxylation at the 2 or 16 position. Most women predominantly metabolize estradiol (E2) to 2-estrogens; some predominantly metabolize E2 to 16-estrogens. In most women this is benign; metabolism to 2-estrogens may confer slight protection against cancer, while metabolism to 16-estrogens may confer slight protection against osteoporosis.^6,7 However, in individuals with BMPR2 mutation, development of disease was strongly correlated with having a high ratio of 16-estrogens to 2-estrogens,⁸ associated with low expression of the estrogen-metabolizing cytochrome cytochrome p450 1B1 (CYP1B1) in lymphocytes⁹ and functional promoter polymorphisms in CYP1B1.⁸

The data from humans thus present a clear association between increased estrogenic activity and disease penetrance. Understanding the mechanism is difficult, however, because in traditional rodent models of PAH, including hypoxic mice and monocrotaline-treated rats, estrogen is protective.^10,11 Beneficial effects of estrogen include suppression of inflammation, vasodilation, and antiproliferative effects.¹²

There are thus a number of broadly unanswered questions regarding how estrogens affect PAH. Does the correlation between penetrance and a high ratio of 16-estrogens to 2-estrogens indicate causation? If so, is 16αOHE damaging, is 2-methoxyestrone (2ME) protective, or both? If 16αOHE increases penetrance, why, given that effects of estrogens are protective in most models? Is estrogen signaling different in the context of reduced BMPR2 signal? The purpose of the present study was to address these questions.

Methods

Transgenic mice

We used the Rosa26-rtTA2 × TetO₇-Bmpr2^R899X FVB/N and Rosa26-rtTA2 × TetO₇-Bmpr2^delx4+ mice as described elsewhere,^13,14 called Rosa26-Bmpr2^R899X and Rosa26-Bmpr2^delx4+ for brevity. R899X is an arginine-to-termination mutation at amino acid 899 in the BMPR2 tail domain found in family US33.¹⁵ Delx4+ is a T insertion at base 504 in the kinase domain, resulting in a premature stop 18 amino acids into the kinase domain, found in family UK21. Expression of transgene occurs in all tissue types, but only after initiation of doxycycline.

For experiments on the effects of 16αOHE alone, adult male Rosa26-only, Rosa26-Bmpr2^R899X, or Rosa26-Bmpr2^delx4+ mice had transgenes activated with doxycycline at 0.2 mg/g in chow for 2 weeks and then were implanted with 4-week Alzet osmotic pumps delivering either vehicle alone (polyethylene glycol [PEG]) or 16αOHE at 1.25 μg/h. After an additional 4 weeks, mice underwent hemodynamic phenotyping, as described below.

For experiments on 2ME and its interaction with 16αOHE, adult Rosa26-only or Rosa26-Bmpr2^R899X mice had transgenes activated with doxycycline at 0.2 mg/g in chow for 2 weeks and then were implanted with 4-week Alzet osmotic pumps delivering vehicle alone (PEG), 2ME alone, 16αOHE alone, or both 2ME and 16αOHE at 1.25 μg/h. After an additional 4 weeks, mice underwent hemodynamic phenotyping, as described below.

The Institutional Animal Care and Use Committee at Vanderbilt University approved all animal studies.

Hemodynamic phenotyping

Two-dimensional echocardiography was performed using a Vevo 770 High-Resolution Image System (VisualSonics, Toronto). Echocardiograms, including B-mode, M-mode, and spectral Doppler images, were obtained the day prior to euthanasia under isoflurane anesthesia, as described elsewhere.¹⁶

Right ventricular systolic pressure (RVSP) was directly measured via insertion of a 1.4F Mikro-tip catheter transducer (Millar Instruments, Houston, TX) into a surgically exposed right internal jugular vein, as described elsewhere.¹⁵ PVR is calculated in units of dyn s/m⁵ as 80 × (mean pulmonary arterial pressure − mean pulmonary artery wedge pressure)/(cardiac output). Since many of these are not directly measurable in mice, we approximate this by PVR = 80 × [(RVSP/3)/(cardiac output)].

Transient transfection and dual-luciferase assay of murine pulmonary microvascular endothelial cells (PMVECs)

Immortomouse-derived wild-type, Bmpr2^R899X, or Bmpr2^Delx4 PMVECs were cultured with medium containing 300 ng/mL doxycycline at 37°C for 3 days to induce the expression of transgene. They were then seeded at a density of 60,000 cells/well into 24-well plates. Cells were transfected with estrogen response element (ERE) reporter containing a mixture of 40∶1 ERE luciferase and CMV Renilla (Qiagen, Valencia, CA), using Lipofectamine 2000 reagent (Invitrogen, Carlsbad, CA). Transfection solutions were replaced by phenol-free medium with 300 ng/mL doxycycline 4 h after transfection. Twenty-four hours after transfection, cells were treated with and without 100 nM 16αOHE or 100 nM E2. Cells were treated for the indicated time (6 or 24 h) prior to cell lysis and measurement of firefly and Renilla luciferase luminescence using the Dual-Luciferase Reporter Assay System (Promega, Madison, WI). Twenty microliters of extracts was used for dual-luciferase assay, with 100 μL of substrate. Activity was determined using a luminometer (BioTek Instruments, Winooski, VT) for 15 seconds for each sample. Firefly luciferase luminescence was quenched by means of Stop and Glo buffer (Promega) for 30 seconds before measuring Renilla activity. Luciferase activities of each transfection were normalized by Renilla activity. A value of 1.0 indicates the basal normalized activity without any added hormone, and values >1 represent the fold of induction for each reporter construct. Each experiment was performed three times to obtain means and standard errors of the mean.

Western blotting and immunohistochemistry

Mouse lungs were homogenized in radioimmunoprecipitation assay buffer (phosphate-buffered saline [PBS], 1% IGEPAL, 0.5% sodium deoxycholate, and 0.1% sodium dodecyl sulfate) with proteinase and phosphatase-inhibitor cocktails (Sigma-Aldrich, St. Louis, MO). Protein concentration was determined by Bradford assay. Primary antibodies used for Western blot included phospho-Smad1/5/8 (Cell Signaling 9511S), Dan (Abcam ab70066), Bmpr2 (BD Biosciences 612292), and β-actin (Abcam ab8227). The Bmpr2 antibody produces a band slightly higher than 130 kDa, which has previously been shown to be specific by small interfering RNA studies.

For immunohistochemistry, immorto-Bmpr2^R899X and immorto-control PMVECs were collected from adult mice as described elsewhere¹⁷ and were verified by staining for endothelial markers platelet endothelial cell adhesion molecule (Santa Cruz Biotechnology Sc-1506) and von Willebrand factor (Dako A0082). Cells were stained with antibodies against estrogen receptor α (ERα; Abcam ab32063) and tubulin (Abcam ab15246).

Microtomography

Microfil (Flow Tech MV-122 yellow) was mixed at 2.5 mL of diluent, 2 mL of compound, and 0.225 mL of curing agent immediately before use. Immediately after euthanasia, the mouse’s chest wall was opened, the left atrium was removed, and PBS was perfused through the right ventricle and lungs until clear (∼5 mL) using a constant-volume pump (5 mL/min). A stopcock was switched to the microfil line, and 2 mL of microfil was infused at 0.25 mL/min. Pulmonary arteries and veins were clamped to prevent backflow, and the lungs were packed with ice for 30 minutes to allow the microfil to cure. After 30 minutes, lungs were inflated with 0.8% agarose through the trachea, and lungs were fixed in 10% formalin at 4°C.

Patient estrogen metabolite ratios

All urinary estrogen metabolite study samples were run together in a blinded manner. The samples were thawed at room temperature and centrifuged at 2,000 g for 10 minutes, and 10 mL of clear urine was extracted for each assay. Urinary estrogen metabolites were quantitated using a monoclonal antibody–based enzyme assay (Estramet 2/16TM; Immuna Care, Blue Bell, PA), as described elsewhere.⁸

Expression arrays

Mouse Genome 430 2.0 microarrays (Affymetrix, Foster City, CA) were performed at 6 weeks of gene activation in Rosa26-control and Rosa26-Bmpr2^R899X mice with normal RVSP that had received 4 weeks of either vehicle (PEG) or 1.25 μg/h 16αOHE, as described elsewhere.¹⁸ Each array consisted of a pool of 2–3 mice, and two arrays were used per condition. Array results were submitted to the National Center for Biotechnology Information gene expression and hybridization array data repository, Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/). Accession numbers are pending.

Statistics

Overrepresentation of Kyoto Encyclopedia of Genes and Genomes groups and gene ontology groups were determined using the hypergeometric test within the Webgestalt program.¹⁹ Principal components analysis and other statistical tests (one-way or two-way analysis of variance [ANOVA] with the post hoc Fisher least significant difference [LSD] test) were performed using the JMP program (SAS Institute, Cary, NC).

Results

16αOHE∶2ME ratio is increased in male HPAH patients compared with that in controls

In planning mouse experiments, we wished to use male mice, to avoid complications associated with ovariectomy or the native estrous cycle. However, to establish relevance we first wished to determine whether the ratio of 16αOHE to 2ME correlated with disease penetrance in male heritable PAH patients. We measured urinary estrogen metabolites in 10 male HPAH patients and 6 male healthy controls. We found no difference in absolute levels of creatinine, 2ME, or 16αOHE. However, we found that there was a significant () increase in the ratio of 16αOHE to 2ME in male HPAH patients compared with that in healthy controls (Fig. 1A). The ratio in control males was similar to the ratio we had previously measured in control females, but the groups were less well separated in men than they had been in women. This implies that a high 16αOHE∶2ME ratio is a risk factor in men but perhaps does not account for as much variability in penetrance as it does in women.

Figure 1.

A, Levels of urine estrogen metabolites 2-methoxyestrogen (2ME) and 16α-hydroxyestrone (16αOHE) are not statistically different between controls (gray lines) and heritable pulmonary arterial hypertension (HPAH) patients (black lines), although 2ME levels trend higher in controls (). Lines link levels of metabolite in a single patient; slope corresponds to the ratio of metabolites. B, 16αOHE∶2ME ratio measured in urine is significantly ( by unpaired t test with unequal variance) higher in male HPAH patients compared with male healthy controls. B, C, 16αOHE∶2ME ratio measured in urine (X-axis) correlates strongly with reduced bone morphogenetic protein 2 (BMP2) expression (, by correlation z test) and with increased expression of the intracellular WNT inhibitor PRICKLE1 (, ), as measured in patient fibroblasts.

For 4 of the patients and 1 of the controls from whom we obtained urine metabolites, we had existing gene expression data derived from cultured skin fibroblasts.²⁰ We tested whether the 16αOHE∶2ME ratio correlated with gene expression as indicated by these arrays. Because of the small numbers, it was not possible to overcome the multiple-comparisons problem, but we found that the ratio of estrogen metabolites correlated strongly to biologically relevant genes, including an inverse correlation to BMP2 expression (Fig. 1B) and a positive correlation to the intracellular WNT inhibitor PRICKLE1 (Fig. 1C).

16αOHE increases PVR

The above-described studies confirmed that high relative 16αOHE levels are correlated with disease in human male HPAH patients. To determine whether this correlation implies causation, we performed an experiment to determine whether addition of 16αOHE would increase penetrance in mice expressing Bmpr2 mutant transgenes. Male mice were used to avoid complications associated with the native estrous cycle.

Three different genotypes of mice were used: mice with the universally expressed transactivator Rosa26-rtTA2 only (controls) or mice with Rosa26-rtTA2 crossed to either the TetO₇-Bmpr2^delx4+ or the TetO₇-Bmpr2^R899X transgene, which express a kinase domain or cytoplasmic tail domain truncation, respectively. A total of 66 male mice were used, split roughly evenly across the genotypes. Transgenes were activated for 6 weeks in adult animals, accompanied by osmotic pumps delivering either vehicle only (PEG) or 16αOHE for the last 4 weeks.

We found that 16αOHE significantly worsened PVR, by 25% on average in Bmpr2^delx4+ mice and by 38% in Bmpr2^R899X mice (Fig. 2A). Using a PVR of 150 dyn s/cm⁵ as the cutoff for elevated PVR, penetrance roughly doubled in both Bmpr2^delx4+ and Bmpr2^R899X mice, going from 3 of 12 to 6 of 13 and from 3 of 12 to 6 of 12, respectively.

Figure 2.

A, 16α-Hydroxyestrone (16αOHE) treatment worsens pulmonary vascular resistance in bone morphogenetic protein receptor type 2 (Bmpr2) mutant mice ( by multiple-factor analysis of variance [ANOVA]), which was already elevated compared with that in controls ( by multiple-factor ANOVA). B, 16αOHE causes a significant drop in cardiac output in Bmpr2^R899X mice but not in other genotypes (drug effect: , differential effect by genotype; by multiple-factor ANOVA; significance bars in figure were determined by the Fisher least significant difference test). C, Cardiac output and weight gain are only weakly correlated in control mice (left panel, not significant for either z correlation or effect of treatment on weight or cardiac output) but are highly correlated in Bmpr2^R899X mice (right panel, , by correlation z test). Each symbol is 1 mouse: open and filled symbols indicate mice receiving vehicle and 16αOHE, respectively. D, 16αOHE treatment produces a qualitative decrease in vessels perfusable by microfill in Bmpr2 mutant mice only. ns: not significant.

In Bmpr2^R899X mice, the effect of 16αOHE on PVR was primarily driven by a drop in cardiac output from 7.2 to 4.3 mL/min (Fig. 2B), while in Bmpr2^delx4+ mice cardiac output already trended lower than that in controls (5.5 as opposed to 6.7 mL/min; ), but neither controls nor Bmpr2^delx4+ mice had significant changes in cardiac output when treated with 16αOHE. The drop in cardiac output in Bmpr2^R899X mice was driven by a ∼40% drop in stroke volume ( by the Fisher LSD test after multiple-factor ANOVA), from 19.3 to 11.9 μL/stroke.

There was a significant relationship in Bmpr2^R899X mice between cardiac output and weight gain during the experiment (, by correlation z test; Fig. 2C, right panel). This effect was much larger than one would expect from the simple fact that larger mice have more cardiac output—see for comparison control mice (Fig. 2C, left panel), in which the correlation was much weaker (, ). There was also no significant correlation in Bmpr2^delx4+ mice (, ; not shown). It is possible that 16αOHE treatment has a direct effect on weight gain, but a simpler explanation is that mice with extremely low cardiac output are too sick to gain weight. Structurally, 16αOHE delivery results in the appearance of qualitatively fewer perfusable small vessels in Bmpr2 mutant mice that received 16αOHE (Fig. 2D) compared with our previous results.¹⁴ This functions as a significant “second hit” to the already compromised RV in Bmpr2 mutant mice.

2ME treatment does not prevent Bmpr2-related PAH

The above-described experiments indicate that increased 16αOHE in the context of Bmpr2 mutation results in increased penetrance and PVR. However, increased 16αOHE is only half of the 16αOHE∶2ME ratio; we thus also sought to determine whether increased 2ME was protective.

This experiment involved a total of 86 age-matched mice, split into 6 groups: control mice with and without 2ME, Bmpr2^R899X mice with and without 2ME, and Bmpr2^R899X mice with 16αOHE and without 2ME. Mice were treated with doxycycline to activate transgene and estrogen or vehicle in osmotic pumps for 6 weeks, followed by phenotyping.

2ME dropped systemic systolic blood pressure, as measured by tail cuff, from 110 to 86 mmHg ( by multiple-factor ANOVA); this effect was not dependent on the Bmpr2 mutation. 2ME had no effect on cardiac output or cardiac index. Presence of Bmpr2 mutation increased both RVSP and PVR, and as before 16αOHE caused increased RVSP and PVR, but 2ME treatment could not prevent an increase in RVSP, whether used alone or in combination with 16aOHE (Fig. 3). There was a nonsignificant trend toward worsened RVSP and PVR with 2ME alone and a nonsignificant trend toward improved RVSP and PVR with 2ME in combination with 16αOHE.

Figure 3.

A, Bone morphogenetic protein receptor type 2 (Bmpr2) mutation increases right ventricular systolic pressure (RVSP; by multiple-factor analysis of variance [ANOVA]). Treatment with 2-methoxyestrogen (2ME) trends toward increased RVSP when used alone and does not significantly decrease RVSP when used in combination with 16α-hydroxyestrone (16αOHE; ). Circles represent measurements from individual mice, with averages indicated by column graph; error bars show standard error of the mean. Numbers in each group are listed at the bottom of each column. B, Bmpr2 mutation increases pulmonary vascular resistance (PVR; by multiple-factor ANOVA). Treatment with 2ME trends toward increased PVR when used alone and does not significantly decrease PVR when used in combination with 16aOHE (). ns: not significant.

16αOHE decreases SMAD signaling in control mice but not in Bmpr2 mutant mice

The above-described results suggest that increased 16αOHE is the important part of the 16αOHE∶2ME ratio and that low 2ME is important only because in humans it indicates high 16αOHE. Our next question was how 16αOHE worsened PAH. An obvious first place to look was its impact on the BMP pathway.

To determine whether 16αOHE had a direct effect on the BMP pathway, we performed quantitative reverse-transcription polymerase chain reaction on lung tissue from control mice treated with either vehicle or 16αOHE, derived from the above-described experiment. All mice used had normal RVSP. We found that most pathway components were not changed. However, there was a more than twofold decrease in Bmpr1b and a more than fivefold increase in expression of the secreted inhibitor Dan (Fig. 4A).

Figure 4.

A, Quantitative reverse-transcription polymerase chain reaction (PCR) on lungs from control mice shows a decrease in bone morphogenetic protein receptor type 1b (Bmpr1b) and in increase in Dan expression with chronic 16α-hydroxyestrone (16αOHE) treatment. Each point is an average of PCR results for 3 mice; error bars show standard error of the mean. Copies per 100 hypoxanthineguanine phosphoribosyltransferase (HPRT) for the vehicle-treated group are indicated. Open circles indicate vehicle-treated groups (normalized to 1), and filled circles indicate 16αOHE-treated groups. B, 16αOHE treatment results in reduced Smad phosphorylation in whole lungs from control mice; Dan protein also appears to be reduced. Numbers are densitometry, normalized to β-actin. Total SMAD levels, normalized to β-actin (adjacent), did not change with 16αOHE treatment. C, 16αOHE treatment also results in reduced Smad1/5/8 phosphorylation in whole kidney from control mice (normalized to β-actin). Total SMAD protein levels were increased, in contrast to phosphorylated Smad1/5/8 levels. D, 16αOHE suppresses bone morphogenetic protein receptor type 2 (Bmpr2) protein expression in wild-type mice but not in Bmpr2 mutant mice and does not further reduce Smad phosphorylation in lungs from Bmpr2 mutant mice.

Consistent with this, there was a more than twofold decrease, on average, in Smad1/5/8 phosphorylation in the lungs of control mice treated with 16αOHE (Fig. 4B). To our surprise, however, we did not detect an increase in Dan protein consistent with the increase in expression (Fig. 4B); one explanation is that these proteins are highly modified posttranslationally, and it is possible that the antibody we used, made to a small bacterially produced epitope, was not capable of detecting the modified protein.

To determine whether the 16αOHE-mediated suppression of Smad phosphorylation was lung specific, we also probed protein derived from kidneys. We found that while there was some suppression of Smad1/5/8 phosphorylation in kidneys, it was substantially less than that in lungs (∼30% compared with ∼60%; by the Fisher LSD test after multiple-factor ANOVA; Fig. 4C).

We also examined Smad 1/5/8 phosphorylation in Bmpr2 mutant mice with and without 16αOHE treatment. As before, Smad 1/5/8 phosphorylation was suppressed by 16αOHE in control mice. In Bmpr2^R899X mice, however, already-low Smad phosphorylation was not further suppressed by 16αOHE (Fig. 4D). Finally, although there was no change in expression levels of Bmpr2, there is some literature indicating posttranslational regulation of protein levels, so we checked Bmpr2 protein levels using an antibody to the amino terminus of the protein that recognizes the native Bmpr2 but not the transgenic Bmpr2. We found that either 16αOHE treatment or expression of the Bmpr2^R899X transgene suppressed native Bmpr2 protein levels (Fig. 4D). Expression of Bmpr2 has a 0.85 correlation coefficient with Smad1/5/8 phosphorylation ( by correlation z test), suggesting that the suppression of Smad1/5/8 may be directly caused by Bmpr2 protein suppression. Note that the R899X mutation, which leaves the kinase domain intact, does not suppress Smad signaling in cell culture or after only a few days of activation in live mice¹⁵ and is thus not directly responsible for the suppression of Smad signaling. In light of these data, it appears that chronic expression of the R899X transgene suppresses expression of the native Bmpr2, resulting in suppression of Smad phosphorylation. In addition, note that 16αOHE does not suppress and may increase Bmpr2 protein levels in the context of Bmpr2 mutation. Overall, these data suggest that the mechanism of 16αOHE signaling is different in the context of Bmpr2 mutation.

Array analysis

Since 16αOHE does not appear to directly suppress BMP signaling further in Bmpr2^R899X mice, we turned to array analysis to determine the molecular mechanism by which 16αOHE causes increased PVR. We used Affymetrix Mouse Gene 1.0 arrays, with two arrays per group and four groups (with and without the Bmpr2^R899X mutation and with and without 16αOHE).

There were 3,516 noncontrol probe sets with a median absolute magnitude of >7 and a range from maximum to minimum across arrays of >0.4. These cutoffs were chosen to minimize noise and eliminate probe sets that did not change at all across samples. Principal components analysis found that the first two principal components together explained 60% of the variability across samples. The F1 component roughly corresponded to 16αOHE treatment, while the F2 component roughly corresponded to the effect of Bmpr2 mutation (Fig. 5A). In keeping with the results described above, 16αOHE treatment moves the control mice on the F2 axis substantially more than it moves the Bmpr2^R899X mice—since 16αOHE is suppressing BMP signaling, but only in control mice. However, 16αOHE treatment moves Bmpr2^R899X mice on the F1 axis more than control mice.

Figure 5.

A, Principal components analysis on expression arrays shows differential effect of 16α-hydroxyestrone (16αOHE) on control and bone morphogenetic protein receptor type 2 (Bmpr2) mutant mice; 16αOHE primarily causes movement on the F2 axis for controls and on the F1 axis for Bmpr2 mutants. B–G, Selected functional pathways from expression arrays show that 16αOHE suppresses cytokine and lectin signaling, implying reduced conventional inflammation, while causing alterations in thrombosis, angiogenesis, insulin, and Wnt pathway genes, suggesting an alternate injury. Error bars show standard error of the mean. In all parts, open symbols indicate vehicle-treated mice, and filled symbols indicate 16αOHE treatment; circles indicate control genotypes (wild type [WT] or Rosa26 only), and squares indicate Rosa26-Bmpr2^R899X mice.

Analyzing the 400 probe sets most strongly associated with the F1 (16αOHE-related) component resulted in 225 unique Entrez IDs, which were subjected to enrichment analysis. Significantly overrepresented gene ontology groups were overwhelmingly those related to immune and stress response, taxis, and adhesion. Classic cytokines were generally downregulated by 16αOHE treatment, and to a greater extent in Bmpr2 mutants than in control animals (Fig. 5B). This was also true for lectins (Fig. 5C); while lectins have diverse functions, most were associated with inflammatory cell recruitment.^21,22 There was an increase in genes related to platelet adhesion (Fig. 5D), including two platelet glycoproteins (GP5 and GP9), the megakaryocyte recruitment gene mesothelin, and the crucial hemostasis gene von Willebrand factor. This increase in platelet adhesion gene expression implies vascular damage; supporting this are increases in a variety of vascular injury or angiogenesis-related genes (Fig. 5E). Next, there were alterations in metabolic pathways (Fig. 5F). These include a nearly threefold increase in a murine homolog of the resistin gene (Retnla), which drives insulin resistance but is also associated with alternative macrophage activation.²³ There were also increases in two insulin-like growth factor binding proteins and gastrin-releasing peptide. Superoxide dismutase 3 was decreased, lowering the cell’s ability to handle oxidative stress. Multiple additional genes related to energy metabolism were also altered. Finally, there were alterations in Wnt pathway signaling (Fig. 5G), in this context probably reflecting changes in planar polarity.

Overall, then, the pattern of gene expression changes caused by 16αOHE in the context of Bmpr2^R899X mutation suggests that it is suppressing classic cytokines and inflammatory pathways, which is in keeping with its reported effects in the literature, but is still promoting vascular injury through alternate means, as suggested by increased platelet-related, angiogenesis and injury-sensing, and insulin-resistance-related genes.

Bmpr2 mutant cells are insensitive to canonical estrogen signaling

The final question we wished to address was why Bmpr2 mutants seem to respond to estrogen signaling differently than do control animals. We have previously published data indicating that Bmpr2 mutant cells and animals have a combination of insensitivity to and constitutive activation of a different steroid hormone receptor, the glucocorticoid receptor (GR), likely related to defects in microtubule trafficking.^14,16 We hypothesized that this defect in nuclear trafficking of steroid hormone receptors may apply across the class, so we examined trafficking of ERα in murine PMVECs. ERα is normally evenly distributed throughout the cell but is transported to the nucleus in the presence of estrogen. However, as was true for the GR, ERα transport in Bmpr2 mutant PMVECs appears to be defective, with localization frequently perinuclear even in the absence of estrogen and little motion or movement to the nucleus in the presence of estrogen (Fig. 6A). Localization of ERα in Bmpr2 mutant cells is usually closely perinuclear, but there is an occasional cell in which the ring of ERα is slightly removed from the nucleus; this ring corresponds to a failure of the microtubule network, as we have previously seen (Fig. 6B).^14,16

Figure 6.

A, Pulmonary microvascular endothelial cells (PMVECs) derived from bone morphogenetic protein receptor type 2 (Bmpr2) mutant mice have defects in estrogen receptor α (ERα; red) trafficking, with little or incorrect relocalization after estrogen treatment. B, Failure of ERα (red) translocation correlates with defects in the microtubule network (green); 4′,6-diamidino-2-phenylindole dihydrochloride (Dapi) nuclear staining is blue. C, Bmpr2 mutant PMVECs have reduced estrogen response element–driven luciferase activity relative to control PMVECs, whether induced with estrogen or with 16α-hydroxyestrone (16αOHE). By two-way analysis of variance, for genotype and estrogen effect. By the post hoc Fisher least significant difference test, for the difference between control and Bmpr2 mutants at the 24-hour time point with either E2 or 16αOHE treatment.

To quantify this, we transfected PMVECs derived from control or Bmpr2 mutant mice with a luciferase reporter construct driven by a canonical ERE. We found that both classes of Bmpr2 mutant were insensitive to estrogen signaling with either E2 or 16αOHE, compared with control cells (Fig. 6C). In response to E2, control cells had a 60% increase in signal at 24 hours, compared with no change in Bmpr2^R899X cells and a 12% increase in Bmpr2^delx4+ cells. In response to 16αOHE, control cells had a 150% increase in ERE signal, compared with a 20% increase in Bmpr2^R899X cells and a 35% increase in Bmpr2^delx4+ cells. These findings are consistent with the reported increased avidity of 16αOHE compared with other estrogenic compounds.²⁴

Discussion

Several recent publications, both from our group and from others, have shown that estrogen metabolism is correlated with disease penetrance in PAH.^8,9,25 In this study, we show that this is also true in HPAH males (Fig. 1). One of the primary findings of this study is that increased 16αOHE is causal for increased penetrance in Bmpr2 mutant mice (Fig. 2) but that 2ME is not significantly protective (Fig. 3). We showed that 16αOHE suppresses both Bmpr2 protein expression and SMAD signaling downstream, but only in control mice (Fig. 4). Expression arrays showed that 16αOHE has the expected protective effect against cytokine-based inflammation but causes dysregulation of alternate pathways associated with vascular injury, including angiogenesis, metabolism, and planar polarity (Fig. 5). Finally, we showed that part of the reason 16αOHE appears different in Bmpr2 mutant mice is that Bmpr2 mutant cells have become estrogen insensitive, at least through canonical signaling pathways (Fig. 6).

Several things regarding the hemodynamic data in this study bear comment. First, there is wide spread in the PVR compared with that in many other mouse models. This has been true of our Bmpr2 mutant models historically,^26,27 as well as other pure adult-onset Bmpr2 models,²⁸ and is likely due to this model exhibiting more vascular occlusion and narrowing than increased tone.^15,29 Second, much of the increased PVR is driven by decreased CO. We have studied this phenomenon more extensively and have manuscripts in preparation showing that either the presence of a Bmpr2 mutation in the heart or treating wild-type mice with pulmonary artery banding together with 16αOHE delivered by osmotic pump will cause the right ventricle to decompensate and fail rather than remodel. While the reduction in CO does indicate a problem with the heart in our model, it also reflects a real increase in PVR.

White et al.²⁵ recently published data indicating that 4 weeks of daily injections of 16αOHE caused a slight increase in RVSP in wild-type mice, from 22 to 28 mmHg. We saw no change in RVSP in wild-type mice (averaging 23.6 mmHg for vehicle and 22.9 mmHg for those receiving 16αOHE). However, we did see a comparable (although not statistically significant) rise in PVR, driven by a drop in cardiac output (consider Fig. 2A and 2B). We suspect that the difference is primarily due to strain differences; the C57BL/6 mice used by White and colleagues have a much stronger remodeling response in the right heart than do the FVB/N mice we use³⁰ and so responded to increased PVR with increased RVSP rather than decreased cardiac output.

We have previously shown that the BMPR2 promoter has a functional estrogen receptor binding site and is suppressed transcriptionally by estrogens.³¹ Our finding that, in vivo, 16αOHE suppresses BMPR2 expression at the protein level but not at the RNA level (Fig. 4) is thus moderately surprising. However, previous studies have shown that steady-state BMPR2 protein levels are regulated by stress-induced lysosomal degradation in vascular cells.³² Estrogens have also been shown to regulate stress-induced lysosomal regulation through a nontranscriptional mechanism,³³ and so the simplest explanation for our data is that 16αOHE regulates BMPR2 protein levels nontranscriptionally through lysosomal activation.

One of the strengths of this study is in its unbiased discovery approach to determining the mechanisms by which 16αOHE impacts PAH development. The relationship between estrogen and PAH has been a puzzle for decades; while the epidemiologic data in patients have been clear, in common rodent models estrogen was protective. Estrogen is vasodilatory and anti-inflammatory and thus suppressed hypoxic and monocrotaline-induced PAH.^10-12 In our data, these positive attributes are preserved; 16αOHE did suppress a number of cytokines that were upregulated in Bmpr2^R899X mice (Fig. 5B), and 2ME did reduce systemic pressures significantly, from 110 to 86 mmHg. White and colleagues hypothesized that the effect of 16αOHE was proproliferative, and they showed this in cell culture. In our in vivo experiments, there was no evidence of a proproliferative phenotype—no groups of genes directly associated with cell cycle or apoptosis were affected. Our findings instead suggested that the pathogenic effect of 16αOHE occurred through alterations in thrombosis, angiogenesis, metabolism, and WNT pathway/planar polarity (Fig. 5D–5G). Taken together, these suggest that 16αOHE causes perceived injury to the vasculature through noncytokine pathways.

Although most of these estrogen effects have not been considered in PAH before, many of the interactions have been previously reported in the literature. The Wnt pathway has previously been shown to be both subject to noncanonical estrogen signaling³⁴ and an important downstream mediator of BMPR2 function.³⁵ E2 has previously been shown to drive von Willebrand factor expression,³⁶ and inhibiting estrogen receptors has been shown to inhibit angiogenesis.³⁷ The prothrombotic effects of estrogens are well recognized.^38,39 Previous reports have shown that some estrogens (progesterone⁴⁰ and octylphenol⁴¹) induce and some estrogens (estradiol) inhibit the insulin-resistance protein resistin. Coupled with our previous report¹⁶ that insulin-resistance may be driving Bmpr2-related PAH, we hypothesize on the basis of these data that 16αOHE may drive vascular injury through exacerbation of insulin resistance.

Our studies show defects in estrogen sensitivity in Bmpr2 mutant cells (Fig. 6). This matches glucocorticoid insensitivity, which we have previously published¹⁶ and which has been shown to be functionally relevant in humans.⁴² However, because estrogen has noncanonical signaling pathways, we do not have direct evidence for how this insensitivity changes estrogen function in vivo.

16αOHE, a 16-estrogen, is highly estrogenic, with a greater capacity to activate the estrogen receptors than E2 itself, although the pharmacology is quite complex.²⁴ By classical methods to assess estrogen receptor binding affinity, 16αOHE possesses only a fraction of the binding affinity of E2, with greater affinity for ERβ.⁴³ However, 16αOHE has been shown to bind covalently (i.e., irreversibly) to estrogen receptors in a time-dependent fashion by way of the Heyns rearrangement.^44,45 The presence of other estrogens, such as 2ME, could prevent the covalent binding of 16αOHE. This could make 2ME appear functionally equivalent to a partial agonist compared with 16αOHE over the time course of these experiments.

This study has several additional limitations. Inducible expression of Bmpr2 mutants overcomes many of the barriers that have made in vivo study difficult, but it results in a nonphysiologic stoichiometry. For instance, it is possible that in humans 16αOHE suppresses Bmpr2 expression even in those patients who carry a mutation, since native Bmpr2 expression has not been previously suppressedby overexpression of a third allele. Furthermore, many of these studies are clearly a first step to more extensive studies. Does 16αOHE regulate Bmpr2 protein levels through lysosomal degradation? Is 16αOHE effect on planar polarity, metabolism, and angiogenesis mediated through nuclear ER signaling, nonnuclear ER signaling, or non-ER-mediated signaling? Although PAH patients have both increased insulin resistance and increased 16αOHE, we do not yet know whether these two are correlated in individual patients, let alone causative.

We conclude from these studies that 16αOHE is causative for increased penetrance in PAH, through regulation of BMPR2 protein levels and through induction of angiogenesis/vascular injury–related pathways.

Source of Support: This project was supported by NIH R01 HL 095797 and P01 HL108800.

Conflict of Interest: None declared.