Research ArticlePulmonary Arterial Hypertension

PPARγ agonist pioglitazone reverses pulmonary hypertension and prevents right heart failure via fatty acid oxidation

See allHide authors and affiliations

Science Translational Medicine  25 Apr 2018:
Vol. 10, Issue 438, eaao0303
DOI: 10.1126/scitranslmed.aao0303

PPARsing the role of lipid metabolism in PAH

During pulmonary hypertension, maladaptive right ventricular hypertrophy, altered mitochondrial metabolism, and occlusive pulmonary vascular remodeling can ultimately lead to heart failure. Here, Legchenko et al. show that activation of the peroxisome proliferator–activated receptor γ (PPARγ) via pioglitazone treatment protects against heart failure in the Sugen hypoxia rat model of pulmonary arterial hypertension. The differential expression of microRNAs in lung tissue and pulmonary vessels from patients with idiopathic pulmonary arterial hypertension was mirrored in the rodent model of heart failure, and cardiac lipid metabolism, genetic, and epigenetic changes associated with PAH were reversed with pioglitazone in the rodents. These findings suggest that targeting PPARγ activation to restore fatty acid oxidation could be therapeutic for pulmonary hypertension and other diseases with altered lipid metabolism.


Right ventricular (RV) heart failure is the leading cause of death in pulmonary arterial hypertension (PAH). Peroxisome proliferator–activated receptor γ (PPARγ) acts as a vasoprotective metabolic regulator in smooth muscle and endothelial cells; however, its role in the heart is unclear. We report that deletion of PPARγ in cardiomyocytes leads to biventricular systolic dysfunction and intramyocellular lipid accumulation in mice. In the SU5416/hypoxia (SuHx) rat model, oral treatment with the PPARγ agonist pioglitazone completely reverses severe PAH and vascular remodeling and prevents RV failure. Failing RV cardiomyocytes exhibited mitochondrial disarray and increased intramyocellular lipids (lipotoxicity) in the SuHx heart, which was prevented by pioglitazone. Unbiased ventricular microRNA (miRNA) arrays, mRNA sequencing, and lipid metabolism studies revealed dysregulation of cardiac hypertrophy, fibrosis, myocardial contractility, fatty acid transport/oxidation (FAO), and transforming growth factor–β signaling in the failing RV. These epigenetic, transcriptional, and metabolic alterations were modulated by pioglitazone through miRNA/mRNA networks previously not associated with PAH/RV dysfunction. Consistently, pre-miR-197 and pre-miR-146b repressed genes that drive FAO (Cpt1b and Fabp4) in primary cardiomyocytes. We recapitulated our major pathogenic findings in human end-stage PAH: (i) in the pressure-overloaded failing RV (miR-197 and miR-146b up-regulated), (ii) in peripheral pulmonary arteries (miR-146b up-regulated, miR-133b down-regulated), and (iii) in plexiform vasculopathy (miR-133b up-regulated, miR-146b down-regulated). Together, PPARγ activation can normalize epigenetic and transcriptional regulation primarily related to disturbed lipid metabolism and mitochondrial morphology/function in the failing RV and the hypertensive pulmonary vasculature, representing a therapeutic approach for PAH and other cardiovascular/pulmonary diseases.


Right ventricular (RV) heart failure is the leading cause of death in pulmonary arterial hypertension (PAH) (1, 2). RV dysfunction (RVD) is common in left heart failure with preserved ejection fraction (HFpEF) when it is associated with adverse outcome (3). Because of these findings, the dogma that RVD is a direct consequence of pulmonary arterial (PA) pressure elevation in PAH and that targeted therapy does not need to directly address RVD in PAH or left HFpEF has recently been challenged. The events that drive the vicious cycle of heart failure in PAH include maladaptive RV hypertrophy (RVH) and dilation, capillary rarefication, cardiac fibrosis, in some cases myocardial ischemia/hypoxia, and ultimately, RV failure (1, 4). Cardiovascular remodeling in pulmonary vascular disease (PVD), PAH, and RV failure largely relates to increased growth factor–mediated cell proliferation (5, 6), activation and recruitment of myofibroblasts, DNA damage/resistance to apoptosis (7), extracellular matrix remodeling and fibrosis, and inflammation and endothelial dysfunction (8), with a smaller contribution from vasoconstriction (9). Abnormalities in glucose and lipid metabolism and epigenetic dysregulation [microRNAs (miRNAs); histone deacetylases (HDACs) (10, 11)] are emerging mechanisms involved in both PAH/PVD and RV failure (12). The notion that several extrapulmonary organs (heart, skeletal muscle, and adipose tissue) show vascular and metabolic abnormalities suggests that PAH is a systemic rather than exclusively pulmonary hypertensive disease (1317). Dyslipidemia and insulin resistance are evident in PAH animal models (18) and human disease (15, 19) and are associated with worse clinical outcome (15, 19).

Mitochondrial fatty acid (FA) oxidation (FAO) produces ≈70% of the adenosine 5′-triphosphate (ATP) that the heart uses. The shift away from aerobic FAO toward glucose utilization in left heart failure has been interpreted as an oxygen-sparing mechanism; however, recent data from multiple genetic mouse models of nonischemic left ventricular (LV) failure (LV pressure overload) suggest that boosting cardiomyocyte FAO may be beneficial during the development of heart failure (20, 21). Although previous PAH studies have identified several regulatory miRNAs in rodent and human pulmonary artery smooth muscle cells (HPASMCs) (22, 23) and prognostic miRNAs in human PAH plasma, very little is known about miRNA/mRNA expression networks in the pressure-overloaded, failing RV [miR-208 (4) and miR-126 (24)] and their role in FA uptake, storage, transport, and oxidation.

Peroxisome proliferator–activated receptor γ (PPARγ) is a ligand-activated transcription factor that affects glucose and FA metabolism in adipose tissue, skeletal muscle, liver, and other organs. Growing evidence indicates that PPARγ is a potent, protective regulator in PAH (14), PASMC (6, 25), and pulmonary artery endothelial cells (PAECs) (8, 26). Bone morphogenetic protein 2 (BMP2)/BMP receptor 2 (BMPR2) activates PPARγ in HPASMC (6), and mRNA expression of both BMP2 (27) and PPARγ (28) is decreased in lung tissue from idiopathic PAH (IPAH) patients. Recently, we identified PPARγ as a master regulator of BMP2/BMPR2 and transforming growth factor–β1 (TGFβ1) pathways in vascular SMC, regulating several miRNAs, cell proliferation, and glucose metabolism (23, 25). However, PPARγ’s role in cardiac homeostasis and heart failure in murine models is unclear (29, 30). Nevertheless, deletion of the PPARγ cofactor and PPARγ coactivator 1α (PGC-1α) in nonischemic LV failure led to decreased FAO and lipid homeostasis, increased glucose oxidation (GO), and worsened cardiac function (31, 32), raising speculations about a potential beneficial role for PPARγ agonists in pressure-overload heart failure (14). The PPARγ agonist pioglitazone (Pio), a thiazolidinedione (TZD)–class insulin sensitizer, was suggested for “therapeutic revival” due to its revised, beneficial safety profile versus rosiglitazone (3335) and its ability to improve diastolic LV function (36).

Here, we demonstrate that targeted deletion of PPARγ in cardiomyocytes leads to cardiac dysfunction and enhanced intramyocellular lipid (IMCL) deposition in the absence of PAH. Treatment with pioglitazone reverses PAH and prevents RV failure in the SuHx rat model by directing distinct mRNA and miRNA networks, restoring mitochondrial function (FAO), and preventing intramyocyte lipid accumulation. Our study suggests that PPARγ activation can correct epigenetic and transcriptional abnormalities primarily related to disturbed lipid metabolism and transport that are evident also in human PAH/RV failure, representing a new treatment strategy for PAH and other diseases associated with proliferation, fibrosis, impaired FAO, lipotoxicity, and a metabolic switch toward glycolysis (Gly).


Mice with targeted deletion of PPARγ in cardiomyocytes develop systolic cardiac dysfunction

cmPPARγ−/− mice had systolic RV and LV dysfunction at age 12 to 16 weeks but no ventricular hypertrophy compared to littermate controls, as assessed by cardiac magnetic resonance imaging (MRI) mass and volume analysis (Fig. 1, A to H, and table S1). The RV ejection fraction (RVEF) and LV ejection fraction (LVEF) were decreased to a similar extent in cmPPARγ−/− mice versus controls (Fig. 1, C and G). RV systolic pressure (RVSP) and pulmonary artery acceleration time (PAAT), surrogate measures for PA pressure and pulmonary vascular resistance (PVR), were not different (Fig. 1, I to K, and table S1). LV end-diastolic pressure (LVEDP) was also not different between the groups (Fig. 1L), suggesting that cmPPARγ−/− mice did not exhibit diastolic LV dysfunction. IMCL accumulation was enhanced in the RV anterior wall of the cmPPARγ−/− mice versus controls (Fig. 1M). mRNA expression of the FAO-driving genes Cpt1b and Fabp4 were reduced by 30 to 40% in both ventricles, although this trend did not reach significance (Fig. 1N). There was no evidence of RVH (fig. S1, A to D) or obvious cardiac fibrosis (fig. S1, E to H) in the mice of either genotype. These findings suggest that PPARγ is crucial for cardiac performance even in the absence of high RV pressure afterload (PAH) and point to impaired FAO and evident lipotoxicity as possible mechanisms of cardiac dysfunction.

Fig. 1 Targeted deletion of PPARγ in cardiomyocytes leads to biventricular systolic dysfunction in mice.

RVEDV and LVEDV (A and E), RV end-systolic volume (RVESV) and LVESV (B and F), and RVEF and LVEF (C and G) as measures of ventricular dilation and systolic function, as well as RV and LV mass over body weight (BW) (D and H), assessed by cardiac MRI in 12- to 16-week-old male and female cmPPARγ−/− mice (α-MHC Cre Pparγflox/flox) and littermate controls. (I) PAAT as a surrogate for pulmonary artery pressure was not significant between cmPPARγ−/− and control mice. ECHO, echocardiography. (J to L) Invasive hemodynamic measurements in cmPPARγ−/− mice and controls. No difference observed. Means ± SEM, n = 4 to 6, two-tailed t test, *P < 0.05, **P < 0.01. CATH, cardiac catheterization. (M) In vivo cardiac magnetic resonance single voxel 1H spectroscopy (MRS) analysis of IMCL accumulation in the RV anterior wall of the cmPPARγ−/− mice versus controls. Average MR spectra of four mice per group are shown. Black line for controls and red for cmPPARγ−/− mice. TMA, trimethylamine; Tau, taurine; ppm, parts per million. (N) Relative mRNA expression of the FAO-related genes Cpt1b and Fabp4 in the RV and LV of cmPPARγ−/− and control mice. Means ± SEM, n = 6, Student’s two-tailed t test, *P < 0.05, **P < 0.01.

Multimodality phenotyping of the SuHx rat model reveals severe PAH and RV failure

To model severe PAH and RV failure (fig. S2), we separated male Sprague-Dawley rats into three animal groups: untreated [control normoxia (ConNx)], injected with vehicle [dimethyl sulfoxide (DMSO)] [control/hypoxia (ConHx); 10% oxygen, 1× subcutaneously (sc) DMSO as vehicle], or treated with vascular endothelial growth factor receptor 2 (VEGFR2) inhibitor SU5416 [SuHx; 1× SU5416, 20 mg/kg per dose, sc] (Fig. 2A). Serial echocardiograms and [18F]fluordesoxyglucose (FDG) positron emission tomography (PET)/computed tomography (CT) studies performed 1 week after the end of 3 weeks of hypoxia confirmed that SuHx rats had severe PAH, RVH, and moderately increased glucose uptake, but no RV dilation and no RV or LV dysfunction 1 week after the end of chronic hypoxia (Fig. 2A, figs. S3A and S4, and table S2). In the subsequent 5 weeks, however, SuHx rats developed severe PAH, RVH, RV dilation, systolic and diastolic RVD, increased glucose uptake, and overt RV failure, thereby sustaining normal LV mass, volume and function, and hematocrit, as assessed by cardiac catheterization (Fig. 2, B to D, fig. S3B, and table S3A), echocardiography (ECHO) (Fig. 2, E to G, fig. S3C, and table S3A), and cardiac magnetic resonance tomography (MRI; Fig. 2, H to M, fig. S3D, and table S3B). Consistently, macroscopic inspection, Fulton’s (RV/LV + S) mass index (Fig. 3A), and histology (Fig. 3C) demonstrated RVH and dilation. FDG-PET/CT quantified a twofold increased RV glucose uptake in SuHx rats (Fig. 3B, fig. S4, and table S3B). The heightened RV glucose uptake likely indicates a metabolic switch from lipid toward glucose utilization, and thus inefficient energy metabolism (Gly >> GO). By 7 to 8 weeks after the end of hypoxia, we observed increased mortality in SuHx rats versus controls. Peripheral blood draws excluded hyperviscosity as a cause for PAH (hematocrit, 44 to 47% across the groups; table S3C).

Fig. 2 PPARγ agonist pioglitazone fully reverses PAH and prevents RV failure in the SuHx rat model.

(A) Experimental design. Four age-matched groups: (i) ConNx; (ii) ConHx [injected once subcutaneously with vehicle (DMSO; v/v), and then exposed to chronic hypoxia (FiO2, 0.1) for 3 weeks, followed by a 6-week period in room air (FiO2, 0.21)]; (iii) SuHx [injected with SU5416 (20 mg/kg per dose, sc), and then exposed to hypoxia (3 weeks), followed by 6 weeks of room air]; (iv) SuHx + Pio [treated orally with Pio (20 mg/kg per day) for 5 weeks]. po, per os. (B to D) Invasive hemodynamic measurements performed 6 weeks after the end of hypoxia to assess the RVSP, RVEDP, and systolic blood pressure (SAP). (E to G) PAAT as a surrogate of mean PA pressure and PVR, end-diastolic diameter of the RV free wall (RVAWD), and tricuspid annular peak systolic excursion (TAPSE) as a measure of longitudinal systolic RV function, assessed via ECHO. (H to M) RVEDV, RVESV, and RVEF as a measure of RV dilation and systolic function, RV mass, LVEDV, and LVEF were assessed by cardiac MRI. Means ± SEM, n = 3 to 12, analysis of variance (ANOVA)–Bonferroni post hoc test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Fig. 3 PPARγ agonist pioglitazone fully prevents (normalizes) RVH, glucose uptake, vessel loss, and cardiac fibrosis in hearts of SuHx-exposed rats.

(A) Representative images of explanted rat hearts and mass ratio of RV over LV plus septum (Fulton’s index, RV/LV + S). (B) FDG-PET/CT images of glucose uptake in rat hearts. SUV, standardized uptake value. (C) Representative hematoxylin and eosin (H&E) staining of the hearts at the papillary muscle level and NT-proBNP plasma concentrations as a marker of heart failure [bar graph in (C)]. (D) H&E images of RV anterior wall cardiomyocyte size as cross-sectional area (CSA) (μm2). (E) Tomato lectin staining of capillaries in RV tissue. (F) Picrosirius red staining of interstitial collagen in RV tissue. Means ± SEM, n = 3 to 7 individual animals, ANOVA-Bonferroni post hoc-test, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. Detailed statistics in the Supplementary Materials. (D to F) Scale bars, 50 μm. For experimental design, see Fig. 2A.

PPARγ agonist pioglitazone fully reverses severe PAH in the SuHx rat model

We proposed PPARγ activation as potential PAH treatment because of the multiple beneficial effects of this transcription factor, including inhibition of PASMC proliferation and modulation of energy metabolism (14, 18). To determine whether pharmacological PPARγ activation would reverse PAH and PVD in an animal model that closely reflects human disease, we administered the PPARγ agonist Pio (20 mg/kg per day) orally to SuHx-exposed rats with PAH (SuHx + Pio; Fig. 2A). Pio fully reversed PAH (Fig. 2, B and E) and RVH (Figs. 2K and 3, A and C) in the SuHx rat model—an animal model resistant to most therapeutic interventions (12). Pio-mediated PPARγ activation decreased RVSP from 91 to 34 mmHg (ConNx, 29 mmHg; ConHx, 32 mmHg, not significantly different; Fig. 2B and table S3A) and decreased RV mass by more than 50% [RV mass by MRI (Fig. 2K); RV/LV + S mass ratio (Fig. 3, A and C)].

Pioglitazone prevents RV dilation and failure in RV pressure overload

To investigate whether PPARγ activation would ameliorate or prevent pressure-overloaded RV failure without toxicity, rats within the four experimental groups underwent right and left cardiac catheterization and advanced cardiac imaging (movies S1 to S4). Oral treatment with Pio prevented RV failure, as shown by normalized RV end-diastolic volume (RVEDV; 0.25 versus 0.27 versus 0.45 versus 0.29 ml; Fig. 2H and table S3B), end-systolic volume (Fig. 2I), and normalized RVEF (78 versus 75 versus 48 versus 75%; Fig. 2J and table S3B). Pio treatment also normalized RVEDP, a marker of RV filling pressure and diastolic RV function (3.4 versus 3.1 versus 15.6 versus 2.6 mmHg; Fig. 2C and table S3A). A macroscopic decrease in RV mass and volume in the Pio-treated SuHx rats (Fig. 3, A and C) was supported by the decreased plasma concentration of N-terminal prohormone of brain natriuretic peptide (NT-proBNP), a clinical heart failure marker (Fig. 3C, bar graph). The vasoprotective adipocytokine adiponectin (APN) was increased 3.8-fold by Pio, demonstrating efficient plasma drug concentrations and suggesting another possible mechanism for PAH reversal with Pio (Fig. 5A, bar graph).

Fig. 5 PPARγ agonist pioglitazone reverses obliterative pulmonary vascular remodeling in the SuHx rat model of PAH.

(A and B) Representative images of small, peripheral pulmonary arteries in H&E and Masson’s trichrome (M. trichrome) staining, respectively. Scale bars, 50 μm. Bar graph (A): Plasma adiponectin, measured by radioimmunosorbent assay, as an indicator of an appropriate Pio delivery, resorption, and bioavailability. Means ± SEM, n = 4 to 7, ANOVA-Bonferroni post hoc test, *P < 0.05, **P < 0.01. (C) Representative images of α-SMA staining of small, peripheral pulmonary artery muscularization. Scale bars, 25 μm. Means ± SEM, ANOVA-Bonferroni post hoc test, n = 119 to 265 vessels, ****P < 0.0001. (D) Representative images of immunofluorescence staining for vWF and quantification of small vessel loss in pulmonary hypertensive SuHx rats. Scale bars, 100 μm. Means ± SEM, n = 7 to 10 individual animals, 563 to 790 fields calculated, ANOVA-Bonferroni post hoc test, *P < 0.05, **P < 0.01, ***P < 0.001; ****P < 0.0001. (E to G) Representative images of stainings for vWF, CD45, and CD3, respectively. Scale bars, 50 μm. For experimental design, see Fig. 2A.

Pioglitazone alters cardiac metabolism, decreases cardiomyocyte size, and inhibits capillary rarefication and cardiac fibrosis in the pressure-overloaded RV

Glucose uptake is greatly increased in human and rodent (37) PAH-RVs, including the SuHx RV. Pio-treated SuHx rat RV glucose uptake was decreased and not different from controls (ConNx and ConHx) (Fig. 3B). Pio-treated rat hearts regained a crescent-shaped RV with an intraventricular septum convex from left to right, indicating subsystemic RV pressure (Fig. 3, B and C, and movie S1, A to D).

In accordance with the imaging and macroscopic findings on RV mass (RVH) and function, both the cardiomyocyte size (Fig. 3D) and the coronary capillary bed (Fig. 3E) were not different from control samples in the Pio-treated SuHx-rats. In addition, Pio prevented the development of cardiac fibrosis (Fig. 3F) in the SuHx-exposed RV. Pio prevented RV failure in the SuHx rat model of PAH by addressing key heart failure mechanisms such as cardiac hypertrophy and fibrosis, myocardial perfusion, and energy metabolism.

PPARγ activation prevents IMCL accumulation and mitochondrial disarray in the pressure-overloaded RV

By using ex vivo two-dimensional (2D) electron microscopy and 3D electron tomography, we found abnormal mitochondrial clustering and increased mitochondrial size heterogeneity in failing SuHx RV cardiomyocytes, with a decrease in mean minimal mitochondrial diameter. These features were absent in Pio-treated hearts (Fig. 4, A to C, and fig. S5A). Autophagosomes were frequently found within the mitochondrial clusters, indicative of damaged and potentially cytotoxic mitochondria, and cytoplasmic vacuoles associated with heart failure were present in SuHx RV cardiomyocytes (Fig. 4A and fig. S5B). Abnormal T-tubular morphology (Fig. 4D) was evident in the SuHx RV: absence of regular T-tubule/sarcoplasmic reticulum (SR) couplons at Z-disc, and presence of axial elements as well as orphaned SR, again indicative of pathological remodeling—a feature not present in the other experimental groups. Furthermore, electron microscopy imaging confirmed the presence of fibrosis in the SuHx RV (fig. S5C), but not in the SuHx + Pio RV or controls. In vivo cardiac magnetic resonance spectroscopy (MRS) revealed increased IMCL content, a surrogate of lipotoxicity, in the heart of SuHx rats with PAH and RV pressure overload that was largely prevented by Pio (Fig. 4E).

Fig. 4 Mitochondrial disarray, disturbed T-tubule/SR structure, and increased IMCL content in failing SuHx hearts are prevented by pioglitazone treatment.

(A) Representative 2D electron micrographs of RV cardiomyocytes from ConNx, ConHx, SuHx, and SuHx + Pio hearts. Green arrowheads indicate T-tubule/SR couplons; red arrowheads, abnormal T-tubule/SR couplon morphology; and yellow arrowheads, autophagosome (see also fig. S4A). n = 2-3 animals per group. Scale bars, 1 μm. (B) Representative dual-axis electron tomographic slices showing mitochondrial clustering and size heterogeneity (quantification in fig. S4B). n = 2 to 3 animals/12 to 16 dual-axis tomograms per group. Scale bars, 500 nm. (C) 3D segmented models based on (B). (D) Representative electron tomographic slices of T-tubule/SR remodeling in RV cardiomyocytes. Green arrowheads indicate T-tubule/SR couplons, and red arrowheads indicate abnormal T-tubule/SR couplon morphology. n = 2 to 3 animals/12 to 16 dual-axis tomograms per group. Scale bars, 500 nm. (E) In vivo cardiac MRS of IMCL content in rat hearts. Averaged 1H spectra with chemical shift selective (CHESS) water suppression for all groups. IMCL as area under the curve, normalized to water peak. The peak of non-IMCL lipids [“other lipid” (OL) content] is also shown. Average MR spectra of 3 to 5 rats per group. For experimental design, see Fig. 2A.

PPARγ agonist pioglitazone dampens pulmonary hypertensive vascular remodeling

We next sought to understand how PPARγ activation reverses the PAH phenotype in the lung at the microscopic level. Serial rat lung sections were stained with H&E, Masson’s trichrome, and antibodies against α-smooth muscle actin (α-SMA), von Willebrand factor (vWF), CD45, and CD3. The SuHx lung vasculature was characterized by concentric medial hypertrophy of peripheral PAs (Fig. 5A) with increased perivascular collagen content (Fig. 5B), increased muscularization of small- and medium-sized PAs (Fig. 5C), and occurrence of plexiform lesions (PLs; Fig. 5, A and B). The number of small vWF-positive vessels was decreased in the SuHx versus ConHx lung (Fig. 5, D and E), indicating PA pruning (vessel loss). We also found a perivascular accumulation of CD3+/CD45 T lymphocytes in the SuHx lung (Fig. 5, F and G), which we had described before in human PLs (38). These histological features of PVD were dampened by Pio treatment.

Pioglitazone regulates miRNAs involved in fatty acid metabolism in the failing rat and human RV and in human IPAH pulmonary arteries

The key mechanisms of heart failure in RV pressure overload include maladaptive RVH (dilation), myocardial ischemia/hypoxia due to decreased coronary perfusion pressure, impaired angiogenesis/coronary capillary rarefication, cardiac fibrosis, alterations in energy metabolism (Gly >> GO, inefficient ATP production), and gene transcription/epigenetic regulation by miRNAs. PPARγ expression is inhibited by miRNA-130/301 in PASMC (25, 39). However, whether and how PPARγ regulates or is controlled by other noncoding RNAs (4042) in the pressure-overloaded RV or hypertensive lung (42) and how these systems relate to cytoplasmic events and/or metabolic-mitochondrial function (13) remain poorly understood. Using unbiased quantitative polymerase chain reaction (qPCR) arrays and single qPCR miRNA/mRNA expression studies to explore the effects of Pio on altered miRNA expression in SuHx versus ConHx rats, we identified several miRNAs that were altered in the SuHx RV versus ConHx RV and were inversely regulated by Pio. miR-197 expression was up-regulated twofold in the SuHx-RV, but down-regulated with Pio treatment (Fig. 6A). miR-146b was up-regulated in the failing SuHx-RV and down-regulated by Pio, whereas miR-133b was down-regulated in SuHx-RV versus ConHx-RV, but not different from either control in the Pio-treated group (Fig. 6A). miR-491 expression was up-regulated upon Pio treatment (Fig. 6A).

Fig. 6 Differential miRNA expression profiles in hearts of SuHx rats with PAH and RV failure are recapitulated in human end-stage PAH cardiac tissue and regulated by PPARγ agonist pioglitazone.

(A) Results of microRNA profiling of rat RVs using 384-well array cards. Relative to global mean expression profiles of four novel miRNAs in the SuHx model. Means ± SEM, n = 3 individual animals, two-tailed, two-sample, unequal variance t test, *P < 0.05, **P < 0.01. (B) Expression profiles of three novel miRNAs in human explanted hearts (RV and LV, IPAH, and healthy donors). Means ± SEM, n = 6 to 7, unpaired two-tailed t test, *P < 0.05, **P < 0.01. (C) Representative images of in situ hybridization (ISH) of miR-197 in rat RV tissue. Scale bars, 200 μm. The inset was captured at ×400 magnification. (D) Expression profiles of novel miRNAs in human whole-lung tissue (IPAH versus downsizing lung). Means ± SEM, n = 7 to 10, unpaired two-tailed t test. PA-control (Con), PA-IPAH, PL-IPAH (n = 5). Expression profiles of novel miRNAs in laser-captured microdissected PLs (PL-IPAH) of IPAH patients, microdissected pulmonary arteries of IPAH patients <500-μm inner diameter, characterized by concentric medial hypertrophy but no PLs (PA-IPAH), and pulmonary arteries <500 μm of downsizing lungs (PA-control). PA-control, PA-IPAH, and PL-IPAH (n = 5). Means ± SEM, ANOVA-Bonferroni post hoc test, **P < 0.01, ****P < 0.0001. For experimental design, see Fig. 2A.

We recapitulated these major miRNA findings in human end-stage IPAH (table S4). In the pressure-overloaded failing RV (SuHx group), we found up-regulation of miR-197 and miR-146b but no change to miR-133b (Fig. 6, B and C). In peripheral pulmonary arteries, miR-146b was increased and miR-133b was decreased (Fig. 6D), whereas in PLs, miR-133b was up-regulated and miR-146b was down-regulated (Fig. 6D). Searching miRDB, an online database of miRNA target prediction, revealed that miR-491 aligns to the mRNA of MGLL, the gene encoding for monoacylglycerol lipase—a key player in lipid metabolism (fig. S6). Additional in vivo experiments in mice using alveolar hypoxia (10% oxygen for 5 days) as a surrogate for myocardial hypoxia in the absence of RV failure revealed that none of the aforementioned miRNAs were up-regulated in the RV in response to alveolar hypoxia; miR-197 tended to be down-regulated after 5 days of hypoxia (fig. S7). Thus, the heightened expression of miR-197 and miR-146b and the changes in their target mRNA transcripts are not triggered by hypoxia and seem to be specific for human and rat RV failure in PAH.

The lung and pulmonary arteries of IPAH patients exhibit differential expression of miR-197, miR-146b, and miR-133b

Although miR-197 and miR-146b were increased in the failing RV of IPAH patients, only miR-197 was increased in whole human IPAH lung versus controls (Fig. 6D). To explore whether there is differential expression of miRNAs in peripheral PAs (inner diameter, <500 μm) and/or PLs of IPAH patients versus controls (downsizing donor lung), we performed laser capture microdissection on the PA media layer. There was no significant difference in miR-197 expression, although it tended to be increased in PL-IPAH (Fig. 6D). miR-146b expression was increased by nearly threefold in PA-IPAH but decreased in PL-IPAH (Fig. 6D). miR-133b expression was higher in PL-IPAH than in PA-IPAH (Fig. 6D). These differential and inverse miRNA expression patterns in small PAs and PLs in IPAH lungs are likely dependent on the global and segmental disease stage.

Gene transcript alterations in the failing RV are related to energy metabolism and regulated by PPARγ activation

To unravel distinct mRNA expression profiles in both RV failure and preventative Pio treatment, we performed RNA sequencing (RNA-seq) on rat RV and LV samples. After pairwise comparison of failing versus nonfailing RVs (SuHx versus ConHx), we discovered 103 genes that were statistically significantly differentially expressed in the diseased group [false discovery rate (FDR) < 5%; Fig. 7, A and C]. Genes associated with cardiac hypertrophy (Thbs4, eightfold), cardiac fibrosis (Ctgf and Nppb, threefold), and TGFβ signaling (Tgfb2, twofold; Ltbp2, 6.8-fold) were highly up-regulated. Conversely, sarcolipin (Sln, 0.01-fold), whose ablation also leads to cardiac hypertrophy and decreased contractility (43), and Cyp2e1 (0.1-fold) were down-regulated. Significantly differentially expressed genes related to Pio treatment (SuHx + Pio versus SuHx) are shown in Fig. 7 (B and D). This Pio-PPARγ–regulated group consists of 66 genes, including genes associated with cardiac hypertrophy, fibrosis, fetal gene expression profile (Thbs4, Cacna1h, Nppa, Slc4a1, and Nppb, all down-regulated by 0.2- to 0.4-fold), and FA metabolism [Mgl and FA binding protein 4 (Fabp4), up-regulated threefold]. Pio also induced PPARγ-mRNA expression (threefold). The transcription of 21 genes reversed direction with Pio treatment (SuHx + Pio versus SuHx, change from up-regulation to down-regulation and vice versa), including Thbs4, Nppa, Nppb, and Aqp7 (fig. S8). The aforementioned 21 Pio-regulated genes and gene products (table S5) are thus potential targets for novel PAH therapies. Figure S9 shows a model of molecular interactions involved in the prevention of RV failure based on the observed gene expression patterns. We performed comprehensive analyses of miRNA expression (Fig. 6A) and RNA-seq (Fig. 7) in the identical (same) rat heart samples (RV and LV), as well as miRNA quantification by single qPCR in the failing RV of IPAH patients. Subsequent network exploration (Ingenuity Pathway Analysis) revealed Pio-PPARγ–mediated, distinct epigenetic regulation of RV miRNA/mRNAs (miR-146b, miR-133b, miR-491/MGLL, and miR-197) previously not known to be associated with PAH/RVD (figs. S6, S8, and S9 and table S5).

Fig. 7 Differential cardiac mRNA expression in the RV and LV of controls, SuHx-exposed rats with RV failure, and SuHx rats treated with pioglitazone (RNA-seq).

(A) All differentially expressed genes in the SuHx versus ConHx group comparison in RVs, ranging from the most up-regulated (red) to the most down-regulated (blue), and their counterparts in the other comparison groups. Each group comparison contains two columns: Log2 of fold change (FC) and the corresponding FDR-adjusted P values (q values) represented as −log10(q) for up-regulated genes and log10(q) for down-regulated genes. (B) All differentially expressed genes in the RVs of SuHx-exposed rats (SuHx + Pio) treated orally with Pio versus SuHx group and their counterparts in the other comparison groups. (C) Volcano plot of the SuHx versus ConHx comparison in RV (q values are capped at 10). (D) Volcano plot of the SuHx + Pio versus SuHx comparison in RV (q values are capped at 15). n = 3 individual animals in each group. For experimental design, see Fig. 2A.

Collectively, pioglitazone conducts RV-specific, distinct epigenetic-transcriptional networks strongly related to lipid metabolism, FA transport, and mitochondrial recovery of FAO. mRNA network analysis of statistically significantly differentially expressed genes in the RV of SuHx-exposed and Pio-treated rats (SuHx + Pio/no RV failure versus SuHx/RV failure) revealed the up-regulation of several metabolic pathways; the latter included Pio-induced inhibition of GO, induction of lipolysis and FAO, and inhibition of macrophage activation/reactive oxidative stress (ROS) (Fig. 8, A to D, and fig. S9). Moreover, Pio treatment resulted in reduced expression of cyclooxygenases 1 and 2, which have been linked to pathological cardiovascular processes such as vasoconstriction, cardiac hypertrophy, fibrosis, and angiogenesis (fig. S9) (44, 45).

Fig. 8 PPARγ agonist pioglitazone induces FAO, ATP production, and FAO-driving genes in cardiomyocytes, whereas FAO genes are repressed in human IPAH.

(A and B) Oxygen consumption rates (OCR) in the setting of FAO of RV and LV NRCMs, respectively, with and without 10 μM Pio. Number of biological replicates, n = 3; number of technical replicates, n = 5; number of each measurement, n = 3; unpaired t test, ****P < 0.0001. (C) Bar graph showing percentage of dependency (reliance to maintain baseline respiration) and flexibility (ability to increase oxidation of the substrate to compensate for inhibition of alternative substrate pathways) of NRCMs to oxidize FAs (as a percentage of oxidation of glucose, FAs, and glutamate). Number of biological replicates, n = 3; number of technical replicates, n = 5; number of each measurement, n = 3; unpaired t tests, **P < 0.01. GLC, glucose; GLN, glutamine. (D) Relative extracellular ATP production of NRCMs induced by Pio. Number of biological replicates, n = 3; number of technical replicates, n = 16; means ± SEM, unpaired two-tailed t test, **P < 0.01. (E) Relative expression of FAO-related genes in NRCMs with and without 10 μM Pio. Means ± SEM, unpaired two-tailed t tests. Number of biological replicates, n = 3; number of technical replicates, n = 6; *P < 0.05, **P < 0.01, ****P < 0.0001. (F) Relative expression of FAO-related genes in human explanted hearts (RV, IPAH, and healthy donors). Means ± SEM; control RV, n = 6, IPAH RV, n = 7, unpaired two-tailed t-test, *P < 0.05, **P < 0.01. (G and H) Relative expression of FAO-related genes in NRCMs after 72 hours of preincubation with pre-miR-197 or pre-miR-146b (each 30 μM), respectively. Number of biological replicates, n = 3; number of technical replicates, n = 6; means ± SEM, unpaired two-tailed t tests, *P < 0.05, **P < 0.01, *** P < 0.001. Means ± SEM, n = 6.

A subgroup of RNA-seq analysis found that Pio induced five angiogenesis-related genes (Egr3, Cyr61, Thbs4, Nppb, and Cyr1b1), consistent with the coronary capillary loss in SuHx RV and its prevention by Pio in SuHx + Pio RV (table S6). Three of these five genes have been reported as positive regulators of angiogenesis (Egr3, Cyr61, and Cyp1b1), and the other two are antiangiogenic (Thbs4 and Nppb). Egr3 and Cyr61 were down-regulated, whereas Nppb and Thbs4 were up-regulated in the failing RV. The observed reversal of the direction of mRNA expression of Egr3, Cyr61, Thbs4, and Nppb with Pio elucidates a possible mechanism of Pio-mediated prevention of coronary capillary rarefication.

Pioglitazone induces FAO and ATP production in cardiomyocytes, and related gene transcripts are repressed in the human failing IPAH-RV

To study FAO and mitochondrial respiration directly, we first measured the oxygen consumption rate in neonatal rat cardiomyocytes (NRCMs) (fig. S10A). Pio induced cardiac FAO, with similar effects on cultured NRCM from the RV and LV (Fig. 8, A and B). In contrast to our ex vivo experiments, there were no significant differences between cultured RV and LV cardiomyocytes in terms of gene expression and functional assays; all subsequent in vitro culture experiments were performed in pooled RV and LV NRCM. Pio increased flexibility and decreased dependency of NRCM metabolism on glucose as the fuel source (Fig. 8C), which was associated with heightened ATP production (Fig. 8D). Consistently, Pio induced mRNA expression of Cpt1b in NRCM (Fig. 8E)—an enzyme well known to also drive FAO and, thus, ATP production in cardiomyocytes. Expression of Fabp4, known to regulate free FA transport and lipid storage, was increased with Pio administration (Fig. 8E). In the failing RV of IPAH patients, mRNA expression of PPARG, CPT1B, and FABP4 was significantly decreased (9.7-, 4.3-, and 2-fold, respectively) compared to donor controls (Fig. 8F), whereas all three genes were induced by Pio in NRCM (Fig. 8E). Additional mechanistic studies using NRCM identified a direct link between overexpression of miR-197 and miR-146b and the suppression of genes that drive FAO: Preincubation with pre-miR-197 (Fig. 8G) or pre-miR-146b (Fig. 8H) decreased mRNA expression of Cpt1b by 2-fold (each pre-miR) and Fabp4 by 3-fold and 1.7-fold, respectively. These findings support our overall hypothesis that decreased mitochondrial FAO and subsequent inefficient ATP production ultimately lead to failure of the RV in PAH, which can be prevented by the PPARγ agonist Pio. On the basis of our results, Pio down-regulates several drivers of RV inflammation, hypertrophy, and fibrosis and induces FAO, which then becomes the primary mechanism for ATP production as opposed to GO/Gly (Fig. 8 and fig. S8). Together, our results in human end-stage IPAH cells and tissues recapitulated the major pathogenic miRNA findings obtained in vivo in the SuHx rat model.

Pioglitazone and rosiglitazone do not exert endothelial toxicity

Rosiglitazone was reported to disrupt the PPARγ–β-catenin complex in cultured endothelial cells (46); however, this has not been shown for Pio, and direct toxic effects of TZD-class PPARγ agonists on endothelial cells have not been demonstrated. We did not find any toxicity or change in GO or FAO in human PAECs isolated from non-PAH and IPAH patients when treated with rosiglitazone or Pio (fig. S10, B to E, and table S7). In addition, even high-dose Pio had no effect on cardiomyocyte survival (fig. S10F). Thus, we cannot confirm the previously proposed endothelial cell toxicity of TZD-type PPARγ agonists. Moreover, in vivo blood glucose concentration was similar in controls, SuHx, and SuHx + Pio rats (fig. S11).


Here, we first demonstrated that targeted deletion of PPARγ expression in cardiomyocytes directly leads to ventricular systolic dysfunction in the absence of pulmonary hypertension/RV pressure load and is associated with impaired FAO and IMCL in vivo, indicating that PPARγ and FAO are crucial for cardiac performance. Multimodality phenotyping in the most widely accepted, therapy-resistant PAH model (SuHx rat) demonstrated that it closely resembles human progressive PVD and RV failure (12, 47). Oral treatment with the PPARγ agonist pioglitazone fully reversed severe PAH and PVD and completely prevented RV failure through induction of FAO and maintenance of mitochondrial morphology, arrangement, and function, suggesting a new avenue for clinical PAH/RV failure therapy.

Published research has rarely addressed the differential expression of mRNA and miRNA in the RV versus LV under normal and pathological conditions, despite the differences in the ventricles’ developmental origins, morphology, and mechanics. Very little is known about the dynamic expression of mRNA transcripts and miRNA (48) in the failing, pressure-overloaded RV [miR-28, miR-34a, miR-93, miR-148a (48, 49), and miR-208 in the monocrotalin rat model (4)] despite their impact on cardiac performance (50). To the best of our knowledge, ours is one of the first studies that systematically explored the differential expression of the transcriptome and relevant miRNAs in RV, LV, and lung, in a solid PAH in vivo model (with/without intervention), in human cultured cardiovascular cells, and in human end-stage PAH heart-lung tissues.

Although PPARα and PPARβ/δ have been shown to regulate cardiac glucose and lipid metabolism (FAO++), and LV mass and function (51), the effect of PPARγ activation on myocardial FAO and ventricular hypertrophy and function was obscure, due to ambiguous results in genetically modified mice and with rosiglitazone treatment (29, 30). Here, we show that deletion of PPARγ in cardiomyocytes leads to RV and LV systolic dysfunction in the absence of PAH. Our data demonstrate that although myocardial PPARγ expression is known to be lower than that of PPARα and PPARβ/δ, its deletion affects cardiac performance. Mice with targeted deletion of PPARγ in cardiomyocytes (α-MHC Cre PPARγflox/flox) were reported to develop LV hypertrophy with age (29), which we did not see in the younger animals we studied (3 months). Adult mice with inducible, conditional PPARγ knockout from cardiomyocytes [α–myosin heavy chain (α-MHC) MerCreMer] have impaired myocardial FA uptake and oxidation (CD36, FABP, and CPT1) in the LV and systolic LV dysfunction in the isolated working heart model (52). In contrast, mice with overexpression of PPARγ in cardiomyocytes [cmPPARγ–transgenic (Tg); α-MHC Cre] were protected from lipopolysaccharide-induced cardiac dysfunction (53). Negative cardiac effects of TZDs (induction of LV hypertrophy) in these mouse models have been reported for rosiglitazone but not Pio and were largely PPARγ-independent (29).

In our current work, we demonstrate that the PPARγ agonist pioglitazone directs RV-specific, distinct transcriptional networks strongly related to lipid metabolism, FA transport and FAO in the RV, and epigenetic regulation of RV miRNAs previously not known to be involved in PAH/RVD: In unbiased RV/LV miRNA arrays, RNA-seq, and lipid metabolism studies in the failing rat RV, expression of several genes associated with cardiac hypertrophy (Nppb and Thbs4), fibrosis (Ctgf, Nppb, Ltbp2, and Tgfb2), and fetal gene programming (Nppa and Nppb) was highly up-regulated. In contrast, both Sln, whose ablation also leads to cardiac hypertrophy and decreased myocardial contractility (43), and Cyp2e1 that is involved in arachidonic acid metabolism, were greatly repressed. Angiogenesis-related genes (Egr3 and Cyr61) were also repressed in the SuHx RV. PPARγ activation predominantly regulated RV-specific and distinct epigenetic-transcriptional networks related to FA transport/lipid storage and lipid metabolism, including mitochondrial recovery of FAO (Cpt1b and Mgll). Activation of PPARγ suppressed transcription of genes related to cardiac fibrosis and hypertrophy. We unraveled the distinct epigenetic regulation of RV miRNA/mRNAs (miR-197, miR-146b, miR-133b, and miR-491/MGLL) previously not known to be involved in PAH/RVD and their regulation by Pio/PPARγ. We recapitulated our major pathogenic findings in human IPAH, particularly in the pressure-overloaded failing RV and in obliterated pulmonary arteries. The miRNA signature in the RV of the SuHx rat was likely specific to RV failure because it was not recapitulated in the hypoxic, nonfailing murine RV. In additional pre-miR mechanistic studies in primary cardiomyocytes, we could demonstrate a direct link between overexpression of miR-197 and miR-146b and the suppression of genes that drive FAO (CPT1B and FABP4).

Our results are in line with recent reports on the impact of dysfunctional FA uptake and oxidation in the pressure overloaded RV: In a very small study on severe PAH/RV failure in the SuHx rat model, RV FA uptake was decreased 2.1-fold and corroborated by decreased expression of FA transporters and enzymes in RV tissue (37). Hence, it was proposed that cardiac FA uptake increases in early disease stages and then decreases when PAH progresses (37), so that FAs are lost as fuel source for myocardial ATP production. We found heightened glucose uptake by FDG-PET in the failing RV of the SuHx rat that was prevented by Pio without an impact on total blood glucose levels. Because FA uptake is decreased in heart failure (37), metabolism then most likely switches to glucose (Gly/GO > FAO), and thus inefficient ATP production.

Pressure-overloaded hearts with pathological hypertrophy revert to a fetal transcriptional and metabolic program, demonstrating increased reliance on glucose (GO >> FAO) and reduced oxidative capacity (51). Mitochondrial fragmentation that occurs in response to cellular stress in PAH/RV pressure overload induces a metabolic switch from FAO to GO in the heart; reversing this “FAO-to-GO switch” and restoring normal cardiac metabolism were sufficient to preserve LV function despite mitochondrial fragmentation (21). The latter finding indicates that the switch in fuel usage (GO >> FAO) in the failing adult ventricle may be maladaptive and likely contributes to the pathogenesis of heart failure (21).

We focused on FAO and ATP production, but other mechanisms such as cardiac fibrosis and lipid toxicity may also be involved in the late-stage RV failure in SuHx rats. The substantial interstitial fibrosis, a disarray of cardiomyocyte mitochondria, disturbed T-tubule/SR structure, presence of autophagosomes, abnormal energy phosphate transport/storage, disturbed FAO, and intramyocyte lipid accumulation we found in the failing SuHx RV were largely prevented by pharmacological PPARγ activation. Autophagosomes within the mitochondrial clusters, indicating damaged and potentially cytotoxic mitochondria (54), and cytoplasmic vacuoles associated with heart failure, were present in SuHx RV cardiomyocytes but not in control (ConNx and ConHx) or SuHx + Pio RV cardiomyocytes.

The lipid deposition in the pressure-overloaded SuHx rat heart may be due to a major block in FAO and/or disturbed lipid export/storage. The specific lipids that mediate cardiac toxicity in heart failure have not been identified. Ceramides have been suggested to accumulate in the RV and contribute to lipid toxicity in BMPR2 mutant mice and postmortem RVs of deceased PAH patients (55, 56). However, a comprehensive analysis in murine sepsis-related cardiac dysfunction did not indicate any association with alteration of ceramide in the heart (53). Cardiac lipotoxicity (56) may directly arise from decreased FAO, and thus, lipotoxicity could be prevented with Pio treatment.

Previously, we had discovered a vasoprotective BMP2/BMPR2-PPARγ axis in HPASMC (6) and identified PPARγ as an important link between vasoprotective BMP2/BMPR2 and detrimental, proliferative/profibrotic TGFβ signaling in vascular SMC (25). Although the vasoprotective role of PPARγ and its target adipocytokine APN in pulmonary vascular remodeling is well known (17), their roles in the healthy and stressed heart were not defined. A possible mechanism of the reversal of PAH/PVD and prevention of RV failure with Pio in the SuHx rat is the induction of circulating APN. Secretion of APN from epicardial fat cells and inhibition of macrophage activation through Toll-like receptor 4 (57) may contribute to the Pio-induced reverse remodeling we observed in the lung and RV of SuHx + Pio rats. APN is not only protective in the vasculature (17, 18, 58) but also appears to represent a defense mechanism against ROS in the stressed human heart when released from epicardial adipose tissue (59). Along these lines, Pio improved LV diastolic function and altered myocardial substrate metabolism without affecting cardiac triglyceride accumulation and high-energy phosphate metabolism in patients with well-controlled type 2 diabetes mellitus (36, 60). Moreover, Pio, in combination with the tyrosine kinase inhibitor imatinib (an experimental adjunct PAH agent), has been shown to erode the chronic myeloid leukemia stem cell pool by decreasing STAT5 (signal transducers and activators of transcription 5) expression, inducing apoptosis (61).

PPARγ agonists of the TZD class (rosiglitazone, pioglitazone) received “bad press” from 2007 to 2011 for potential cardiovascular adverse effects—an interpretation that was later substantially corrected after readjusting the RECORD trial (33). Overall, Pio has a positive risk-benefit ratio, with a safety profile that compares favorably with rosiglitazone (33, 62). Pio was proposed for therapeutic revival because it is a safe drug with strong therapeutic potential beyond insulin resistance/diabetes and a benefit of 52 fewer cardiometabolic deaths per 100,000 population (33). TZDs are not thought to cause cardiac dysfunction directly but rather to exacerbate heart failure via fluid retention in patients with multiple systemic cardiovascular risk factors (such as diabetes) (33). The clinical effects of Pio on the heart in the presence or absence of insulin resistance are unclear (33) but have been partly addressed in the large randomized controlled trial IRIS (n = 3876) (35): There was no significant between-group difference in the number of patients with heart failure (74 versus 71; P = 0.80) or in the number of patients hospitalized for heart failure (51 versus 42; P = 0.35). Consistent with this, we did not find any toxicity for either Pio or rosiglitazone in healthy or IPAH-derived human PA endothelial cells or cardiomyocytes in culture.

To the best of our knowledge, oral pioglitazone is the first agent that prevents RV failure in the SuHx rat model of PAH and reverses PAH and PVD, characterized by concentric-hypertrophic and plexiform lesions. Given that PAH/RV failure leads to death in 25 to 50% of patients within 5 years of PAH diagnosis even when treated according to the guidelines (1, 63), advanced PPARγ agonistic drugs (33) including Pio should undergo a timely risk-benefit analysis. Phase 1/2 studies with PPARγ-activating drugs may be started in PAH patients, regardless of whether risk factors such as insulin resistance and/or BMPR2 mutations are present. In addition to promising preclinical and clinical findings (14, 17, 35, 64, 65), our study provides proof of concept for the treatment of PAH/RV failure with PPARγ-activating agents that can rescue the disturbed glucose (25) and lipid metabolism and direct miRNA/mRNA networks in the pressure-overloaded heart and lung. Given that FAO also increases oxygen consumption while producing more ATP, therapeutic PPARγ activation may require sufficient oxygen supply to the myocardium and thus may not be started in end-stage PAH with low cardiac output and high end-diastolic ventricular filling pressures.

Together, PPARγ-activating drugs, such as pioglitazone, are promising agents for the treatment and prevention of PAH/RV failure and other conditions associated with altered lipid/glucose metabolism and increased TGFβ signaling, which are strongly linked to cardiopulmonary remodeling/dysfunction (cardiac allograft vasculopathy, septic heart failure, and pulmonary fibrosis) and cancer (NCT02730195).


Study design

All animal experiments were conducted under the approval of the Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit (LAVES; #15/2022 and #13/1328) and the Animal Care and Use Committee (#12-10-2234R), Boston Children’s Hospital, USA. Human formalin-fixed paraffin-embedded (FFPE) specimens and fresh lung tissue for cell extractions were handled anonymously according to the principles expressed in the Declaration of Helsinki and following the requirements of the ethics committee of Hannover Medical School (ethical committee votes #1691-2013 and #3381-2016). Additional materials and methods details can be found in the Supplementary Materials.

Animal studies

Mice with targeted deletion of PPARγ in cardiomyocytes. We bred α-MHC promoter–driven Cre-Tg mice [B6.FVB-Tg(Myh6-Cre)2182Mds/J] with PPARγ homozygous floxed mice (B6.129-Ppargtm2Rev/J). Both strains were purchased from the Jackson Laboratory. The cross resulted in α-MHC Cre PPARγflox/flox (cmPPARγ−/−) mice; 12- to 16-week-old mice (cmPPARγ−/− and littermate controls) of both genders were taken for cardiac MRI, ECHO, and cardiac catheterization.

SuHx rat PAH model. Six- to 8-week-old male Sprague-Dawley rats weighing ≈180 to 200 g were purchased from Charles River and divided into four age-matched groups according to the experimental design (Fig. 1A): (i) ConNx; (ii) ConHx [injected once subcutaneously with vehicle (DMSO; v/v), and then exposed to chronic hypoxia (FiO2, 0.1; CO2, <10.000 ppm) for 3 weeks, followed by a 6-week period in room air (FiO2, 0.21)]; (iii) SuHx [injected with the VEGFR2 inhibitor SU5416 (Sigma), 20 mg/kg per dose sc dissolved in DMSO and subsequently exposed to chronic hypoxia (3 weeks), followed by 6 weeks of room air]; (iv) Sugen/chronic hypoxia (SuHx + Pio) [treated orally with Pio (ACTOS, Takeda) 20 mg/kg per day, incorporated into the diet, starting 1 week after the end of chronic hypoxia (recovery), for the subsequent 5 weeks]. Noninvasive imaging and hemodynamic assessments were performed at two points: 3 + 1 weeks (3 weeks hypoxia plus 1 week recovery in room air before the optional start of Pio treatment) and 3 + 6 weeks (3 weeks hypoxia plus 6 weeks in room air).

FVB mice in hypoxia. Nine-week-old male FVB mice were subjected to hypoxia (FiO2, 0.1; CO2, <10.000 ppm; see details above) for 5 days; control mice were kept in room air. At the end of the study, mice were anesthetized, and tissue sampling was performed as described below.

Human tissue specimens

For microvascular endothelial cell isolation, fresh lung explant tissue (IPAH, n = 7; controls chronic obstructive pulmonary disease/bronchiolitis obliterans syndrome/healthy donor, n = 6) was obtained immediately after lung transplantation (LuTx) and further processed as described below. For laser capture microdissection experiments, 5-μm-thick FFPE sections from three different lobes of patients who had undergone heart-lung transplantation for severe PAH [n = 5; New York Heart Association (NYHA) functional class IV] were selected. As a reference, downsizing samples (n = 5) from healthy donor organs, sampled directly before transplant, were chosen. Heart tissue controls were obtained from donor organs that had been explanted to harvest valvar homografts (n = 6).


Materials and Methods

Fig. S1. Mice with targeted deletion of PPARγ in cardiomyocytes (cmPPARγ−/−), in the absence of PAH, do not develop cardiac hypertrophy or fibrosis at the age of 12 to 16 weeks.

Fig. S2. Schematic depicting the vicious cycle of RV failure in PAH.

Fig. S3. PAH but no RV failure is evident 1 week after the end of hypoxia (3 + 1 weeks), and RV failure develops by week 6 after the end of hypoxia in SuHx-exposed rats.

Fig. S4. RV glucose uptake increases with chronic RV pressure afterload and correlates with RV systolic function 6 weeks after the end of hypoxia in SuHx-exposed rats.

Fig. S5. Decrease in minimal mitochondrial diameter, presence of autophagosomes and cytoplasmic vacuoles, and collagen deposits indicating RV failure and fibrosis are present in SuHx RVs but not SuHx + Pio RVs.

Fig. S6. In silico predicted miRNA/mRNA pairing of miR-491 with the mRNA of monoacylglycerol lipase (MGLL).

Fig. S7. Expression of miRNAs that are altered in rat and human RV failure is not changed by hypoxia in the RV of FVB mice.

Fig. S8. mRNA/miRNA expression signatures and networks in the failing RV (SuHx, on the left) and the PPARγ-mediated effects in the RV of the SuHx rat PAH model (SuHx + Pio, on the right).

Fig. S9. Important molecular interactions based on differential gene expression (mRNA) analysis in the RV of SuHx + Pio versus SuHx rats.

Fig. S10. Pioglitazone has no negative effects on survival, GO, or FAO in human PAECs and no impact on cardiomyocyte survival.

Fig. S11. Neither VEGFR2 blockade nor oral pioglitazone treatment cause any significant changes in blood glucose in rats.

Table S1. Cardiac MRI, ECHO, and cardiac catheterization hemodynamic and morphological data obtained in cmPPARγ−/− mice and littermate controls at 12 to 16 weeks of age.

Table S2. Echocardiographic measurements in rats 1 week after the end of hypoxia (=SuHx rat with PAH but no RV failure yet, 3 + 1 weeks).

Table S3A. Invasive hemodynamic and echocardiographic measurements in rats at 3 + 6 weeks (SuHx rat PAH/RV failure model).

Table S3B. Cardiac MRI, 18FDG-PET/CT, and heart weight measurements in rats of the SuHx study at 3 + 6 weeks.

Table S3C. Complete blood count, plasma NT-proBNP, and plasma APN in rats of the SuHx study at 3 + 6 weeks.

Table S4. Human heart and lung tissue specimens used for laser capture microdissection or whole-tissue gene expression assays.

Table S5. List of significantly differentially expressed genes (FDR < 5%) based on inversion of RV mRNA expression patterns in the SuHx PAH/RV failure model (RNA-seq) with PPARγ agonist pioglitazone, first up- or down-regulated in PAH/RV failure (SuHx) and inversely regulated with pioglitazone treatment (SuHx + Pio).

Table S6. List of mRNA transcripts related to angiogenesis identified by RNA-seq in the RV of control rats (ConHx), rats with PAH and RV failure (SuHx), and Pio-treated rats (SuHx + Pio).

Table S7. Fresh human lung tissue specimens obtained during lung transplantation.

Movie S1. Cardiac MRI ConNx.

Movie S2. Cardiac MRI ConHx.

Movie S3. Cardiac MRI SuHx.

Movie S4. Cardiac MRI SuHx + Pio.

References (6681)


Acknowledgments: We thank L. Kraus and C. Fiedler for the excellent technical assistance and S. Engeli for providing the Seahorse equipment. Funding: This study was supported by the German Research Foundation (DFG; HA4348/6-1 KFO 311), Kinderherzen (W-H-001-2014), Stiftung KinderHerz (2511-6-13-011), and the European Pediatric PVD Network (; all grants to G.H.). The electron microscopy studies were funded through the European Research Council Advanced Grant (CardioNECT 20120314). Author contributions: E.L. performed in vitro, ex vivo, and in vivo experiments, conducted data analysis, wrote parts of the manuscript, and edited and revised the manuscript for important intellectual content. P.C. performed the full bioinformatics analysis of the RNA-seq data, and partly the analysis of the miRNA qPCR arrays, including heat maps and volcano plots, and revised the manuscript for important intellectual content. P.B. performed ex vivo experiments and data analysis. A.F.-G. assisted in the monitoring of both hypoxic and normoxic animals and the processing of tissues harvested from animals, performed immunofluorescence staining (vWF) and morphometric analysis in a subset of experiments, and revised the manuscript for important intellectual content. E.S. performed the PET/CT in vivo experiments together with G.H. and conducted the related data analysis. M.M. performed cardiac MRS experiments and data analysis and revised the manuscript for important intellectual content. S.A.M. assisted in the monitoring of hypoxic animals and revised the manuscript for important intellectual content. E.A.R.-Z. performed 2D/3D electron microscopy and data analysis and revised the manuscript for important intellectual content. S.K. provided equipment and critically reviewed and revised the manuscript for important intellectual content. L.M. and D.J. histologically evaluated rat and human PAH lung tissue samples, supervised certain ex vivo experiments (laser capture microdissection, histological stainings), and revised the manuscript for important intellectual content. G.H. generated the hypotheses, developed the experimental design and concept of the study, performed in vivo and ex vivo experiments, supervised the experimental work, obtained funding, and wrote the manuscript. Competing interests: The authors declare that they have no competing interests. G.H. is the sponsor of an Investigational New Drug (IND) application/U.S. Food and Drug Administration acknowledgement related to pharmacotherapy of PAH with pioglitazone (IND no. 105,428). G.H. has two patent applications related to pulmonary hypertension, PH pharmacotherapy, and cellular biomarkers (United States Patent and Trademark Office) no. 1289344; and international application no. PCT/US12/23859, filed 3 February 2012]. Data and materials availability: The RNA-seq data generated and analyzed during the current study are available from the Sequence Read Archive (accession: SRP137055).

Stay Connected to Science Translational Medicine

Navigate This Article