Ectonucleoside triphosphate diphosphohydrolase-1 (CD39) impacts TGF-β1 responses: insights into cardiac fibrosis and function following myocardial infarction

Abstract

Extracellular purine nucleotides and nucleosides released from activated or injured cells influence multiple aspects of cardiac physiology and pathophysiology. Ectonucleoside triphosphate diphosphohydrolase-1 (ENTPD1; CD39) hydrolyzes released nucleotides and thereby regulates the magnitude and duration of purinergic signaling. However, the impact of CD39 activity on post-myocardial infarction (MI) remodeling is incompletely understood. We measured the levels and activity of ectonucleotidases in human left ventricular samples from control and ischemic cardiomyopathy (ICM) hearts and examined the impact of ablation of Cd39 expression on post-myocardial infarction remodeling in mice. We found that human CD39 levels and activity are significantly decreased in ICM hearts (n = 5) compared with control hearts (n = 5). In mice null for Cd39, cardiac function and remodeling are significantly compromised in Cd39^−/− mice following myocardial infarction. Fibrotic markers including plasminogen activator inhibitor-1 (PAI-1) expression, fibrin deposition, α-smooth muscle actin (αSMA), and collagen expression are increased in Cd39^−/− hearts. Importantly, we found that transforming growth factor β1 (TGF-β1) stimulates ATP release and induces Cd39 expression and activity on cardiac fibroblasts, constituting an autocrine regulatory pathway not previously appreciated. Absence of CD39 activity on cardiac fibroblasts exacerbates TGF-β1 profibrotic responses. Treatment with exogenous ectonucleotidase rescues this profibrotic response in Cd39^−/− fibroblasts. Together, these data demonstrate that CD39 has important interactions with TGF-β1-stimulated autocrine purinergic signaling in cardiac fibroblasts and dictates outcomes of cardiac remodeling following myocardial infarction. Our results reveal that ENTPD1 (CD39) regulates TGF-β1-mediated fibroblast activation and limits adverse cardiac remodeling following myocardial infarction.

NEW & NOTEWORTHY We show that CD39 is a critical modulator of TGF-β1-mediated fibroblast activation and cardiac remodeling following myocardial infarction via modulation of nucleotide signaling. TGF-β1-induced CD39 expression generates a negative feedback loop that attenuates cardiac fibroblast activation. In the absence of CD39 activity, collagen deposition is increased, elastin expression is decreased, and diastolic dysfunction is worsened. Treatment with ecto-apyrase attenuates the TGF-β1-induced profibrotic cardiac fibroblast phenotype, revealing a novel approach to combat post-myocardial infarction cardiac fibrosis.

INTRODUCTION

Over 1 million people per year suffer an acute myocardial infarction (MI) in the United States (1). Failure of adequate myocardial healing and progression of coronary artery disease results in debilitating chronic heart failure due to progressive fibrosis (2, 3). Therefore, understanding of the factors that contribute to cardiac fibrosis following MI is paramount to the development of novel therapies to preserve cardiac function.

Extracellular purine nucleotides and nucleosides influence every aspect of cardiac physiology and pathophysiology (4). In response to injury or stimulation, cells release nucleotides into the interstitial space where they act as autocrine and paracrine signaling molecules via activation of P2X and P2Y purinergic receptors. Ectonucleoside triphosphate diphosphohydrolase 1 (ENTPD1; CD39) is an important ectonucleotidase that hydrolyzes released nucleotides and thereby regulates the magnitude and duration of purinergic signaling. Although we (5–9) and others (10–15) have examined the role of CD39 in cardiac pathology, the impact of this key purinergic metabolizing enzyme on post-MI remodeling is incompletely understood (16).

We hypothesized that CD39 activity is a critical regulator of post-myocardial infarction remodeling. Here, we demonstrate a decrease in CD39 levels and activity in left ventricular tissues from patients with ischemic cardiomyopathy suggesting that dysregulation of extracellular purinergic metabolism may contribute to cardiac dysfunction. To understand how CD39 activity influences post-myocardial infarction remodeling, we examined the effect of ablation of CD39 expression in a model of ischemic heart disease in mice. Our data show for the first time, that CD39 activity protects against adverse cardiac remodeling following myocardial infarction. Importantly, this study establishes that CD39 is a key negative regulator of TGF-β1-mediated fibroblast activation.

METHODS

Human Left Ventricular Sample Collection

The human study was approved by the Vanderbilt University Institutional Review Board and conforms to the principles outlined in the Declaration of Helsinki (17). Patient demographic and clinical data were abstracted from the medical records and stored in a deidentified RedCap file. Left ventricular tissue was obtained from seven end-stage ischemic cardiomyopathy hearts and seven normal hearts not used for transplantation (these were from patients who had stroke or ICH). Samples were flash frozen in liquid nitrogen and stored at −80°C until processed.

Nucleotidase Activity

Nucleotidase activity was measured as previously published (5).

Mice

The investigation conformed to the Guidelines for the Care and Use of Laboratory Animals of the National Institutes of Health and was approved by the Vanderbilt University and the Ohio State University Institutional Animal Care and Use Committees. Mice deficient in Cd39 expression were generated by targeted disruption of the cd39 gene and backcrossed for more than ten generations into a C57BL/6 background (18). For the studies reported, 10- to 20-wk-old (25–30 g) male, Cd39-null (Cd39^−/−), and wild-type (WT) littermate control mice were used in all protocols. Male mice were used to examine the impact on cardiac rupture and remodeling given that male mice have a significantly higher incidence of cardiac rupture than females (19, 20). To examine CD39 expression, Entpd1^{tm1a(EUCOMM)Wtsi}/Ieg mice were obtained from the European Conditional Mouse Mutagenesis Program (21). By crossing these mice with global cre-expressing mice, we generated a stable β-galactosidase reporter allele that was used to examine the expression of CD39 in the post-MI heart and in cardiac fibroblast (Supplemental Fig. S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.19350896) (22).

Quantitative RT-PCR

Total RNA was extracted using QIAzol followed by RNeasy Mini Kit (Qiagen), according to the manufacturer’s protocols. cDNA was prepared with High-Capacity cDNA Reverse Transcription Kit (Life Technologies) and quantified by Quant-iT OliGreen ssDNA Assay Kit (Life Technologies). Fifteen nanograms of cDNA were amplified using Power SYBR Green Master Mix (Life Technologies). The efficiency of primers was calculated before assays to ensure their appropriateness for this method. The primers used are provided in Supplemental Tables S1 and S2 Real-time PCR reactions were performed on QuantStudio Real-Time PCR system (ThermoFisher Scientific). The mRNA levels normalized to Gapdh (mouse) were determined using the comparative CT method (23). For analysis of the response to TGF-β1 stimulation, values were normalized to untreated WT and Cd39^−/− fibroblasts and the fold-change following TGF-β1 stimulation was compared.

In Vivo Myocardial Infarction

The in vivo coronary artery ligation model was performed as previously described (8, 24). All work was in conducted in agreement with recently published guidelines for in vivo mouse models of myocardial infarction (25). In brief, mice were anesthetized with ketamine (55 mg/kg) plus xylazine (15 mg/kg). Animals were intubated and ventilated with room air (tidal volume, 250 μL, 120 breaths/min) with a mouse respirator (Harvard Apparatus, Holliston, MA). Temperatures were maintained at 37°C by a thermoregulated heating pad. Following thoracotomy, an 8-0 silk suture was placed under the left coronary artery for ligation. The mice thoracotomy site was closed and the mice allowed to recover for 72 h to 28 days. Mice were randomly assigned to sham or MI, and the cardiac surgeon was blinded to genotype for all studies.

Echocardiographic Measurements

Cardiac function was assessed at baseline, 3 days, and 28 days following coronary artery ligation using echocardiographic imaging system (Vevo 2100, VisualSonic, Toronto, Canada). In nonanesthetized mice, two-dimensional echocardiographic views of the midventricular short axis were obtained at the level of the papillary muscle tips below the mitral valve. LV internal dimensions were measured, and the LV fractional shortening (LVFS) was calculated as previously described (8, 24, 26).

In Vivo Hemodynamic Monitoring

To monitor left ventricular hemodynamics, a 1.4-Fr Millar Mikro-Tip catheter (SPR-839; Millar Instruments, Houston, TX) was inserted into the right carotid artery and advanced into the left ventricle to measure cardiac function as previously described (8). Continuous LV pressure measurements were recorded for the duration of the experiment on a PowerLab system and Chart version 7 software (ADInstruments, Colorado Springs, CO). Heart rate, Left ventricular (LV) systolic pressure (LVSP), LV end-diastolic pressure (LVEDP), LV developed pressure (LVDP), as well as LV maximum positive change in pressure over time (+dP/dt_max) and LV minimum negative changes in pressure over time (−dP/dt_min) were analyzed off-line. The aorta was next cannulated, and the heart was perfused with ice-cold saline to remove any blood. The left ventricle was isolated and snap frozen for subsequent analyses.

Tensile Measurements

Hearts from WT or cd39^−/− mice subjected to sham surgery or CAL surgery for 3 or 28 days were rapidly excised and placed into saline. The hearts were cut into 1-mm-thick transverse sections, and two midpapillary sections/heart were individually mounted on an Instron 5944 load frame equipped with vascular ring hooks to secure the cardiac tissue. The tensile strength of each section was measured by stretching at a rate of 0.3 mm/min. The full tension curve was measured, and the load required for tissue failure was analyzed.

Immunoblot Analysis

Human LV samples 30 mg were homogenized in NP-40 lysis buffer, consisting of 50 mM Tris·HCl (pH 7.5), 150 mM NaCl, 0.5% NP-40, protease, and phosphatase inhibitors, for immunoblot analysis as we have previously described (5, 27). Mouse hearts were excised, the aorta cannulated, perfused with ice-cold saline, and snap frozen for storage at −80°C until homogenized for immunoblot analysis as previously described (5, 27). Total protein levels were assessed by Pierce BCA protein assay. Twenty to fifty micrograms of protein were loaded on acrylamide gels (Bio-Rad), separated by electrophoresis, and transferred using miniprotean electrophoresis system (Bio-Rad). The specific antibodies used for immunoblot analysis of human heart samples were CD39 (Abcam; 108248), CD73 (Abcam; 133582), GAPDH (Cell Signaling Technology; 2118). The specific antibodies used for immunoblot analysis of mouse hearts were anti-CD39 (eBioscience, clone 24DMS1), anti-α-smooth muscle actin (Abcam; 5694), anti-collagen I (Abcam; 21286), anti-plasminogen activator inhibitor-1 (Abcam; clone ab66705), anti-matrix metalloproteinase-9 antibody (Abcam, clone ab38898), anti-elastin antibody (Abcam; ab21607), anti-SMAD2/3 (Cell Signaling Technology; D7G7), anti-phospho-SMAD2 (Cell Signaling Technology; 138D4), and anti-GAPDH (Cell Signaling Technology; 2118). Membranes were washed with 0.05% Tween in TBS, incubated with the appropriate secondary antibodies conjugated to HRP (Cell Signaling Technology), and SuperSignal (Thermo Scientific) used to visualize proteins. The blots were scanned and quantified using ImageJ 1.48 software (NIH).

Immunohistochemistry and Histology

For murine samples, formalin-fixed paraffin-embedded sections were dewaxed and rehydrated. Heat-activated antigen retrieval using Citra-plus buffer (Biogenex) was performed according to the manufacturer’s protocol. Sections were blocked and incubated overnight at 4°C with rabbit anti-mouse fibrin using a 1:1,000 dilution (28), sheep anti-mouse CD39 (R&D Systems, AF4398) and mouse anti-a-smooth muscle actin (Abcam, 7817). After washing, sections were incubated with either biotinylated anti-sheep antibody (Vector), Alexa Fluor 488 anti-mouse antibody (ThermoFisher) or Alexa595 goat anti-rabbit IgG secondary antibody (ThermoFisher) for 1 h at room temperature. Biotinylated secondary antibody was detected by neutrAvidin Dylight550 (ThermoFisher). Sections were mounted with an antifade mounting media with DAPI (Life Technologies) and photographed with a Zeiss Axioplan2 microscope equipped with a Hamamatsu orca ER camera using MetaMorph software (Molecular Diagnostics) and analyzed using ImageJ (NIH).

Formalin-fixed paraffin-embedded heart sections were stained with 0.1% Picrosirius Red according to a standard protocol to assess extracellular collagen deposition. Briefly, sections were dewaxed and rehydrated, nuclei were stained with Weigert’s hematoxylin and then slides were stained in saturated picric acid with 0.1% Sirius red F3BA (Direct Red 80; Sigma-Aldrich) overnight to ensure near-equilibrium staining. Sections were washed in 0.5% acetic acid, dehydrated, cleared, and mounted. Staining was visualized using a Leica DMIRB microscope equipped with cross polarizer and a Nikon DXM1200C camera using MetaMorph software program, and ImageJ was used to quantify images. Three to six images from each border zone, remote area, and scar were captured by a blinded observer and quantified for a pixel density of thresholded light intensity using ImageJ software.

Flow Cytometric Analysis

At the designated times, hearts were excised and perfused with ice-cold saline. Inflammatory cells were isolated using a modified protocol based on previously published methods (29). For analysis of cardiac fibroblasts, CD39 expression on CD140a⁺, Sca1⁺ cells was measured. A Becton Dickinson LSRII was used to capture fluorescent intensities of all events and flow cytometric analysis was performed using FlowJo software (FlowJo, LLC).

β-Galactosidase Reporter Measurement

Whole hearts were perfused with PBS, dissected out and fixed in a solution of 1% formaldehyde, 0.2% glutaraldehyde, 2 mM MgCl_2, 5 mM EGTA, and 0.1 M phosphate buffer (pH 7.3). Fixed whole hearts were washed with PBS and immersed into staining solution containing 5 mM K₃Fe(CN)₆, 5 mM K₄Fe(CN)₆, 2 mM MgCl₂, 0.01% Na-deoxycholate, 0.02% NP-40, 1 mg/mL X-Gal and incubated at 37°C overnight. After being washed with PBS, hearts were photographed.

TGF-β1 Measurement

Active and latent TGF-β1 levels were measured from baseline and post-MI infarct and remote cardiac tissue by ELISA according to the manufacturer’s protocol (R&D Systems).

Cardiac Fibroblast Isolation

Cardiac fibroblasts were isolated using modifications of previously published methods (30). Hearts were perfused with cold saline, ventricles removed, and minced. Minced tissue was incubated with collagenase II (10 mg/mL, Worthington Biochemical Corp.), Dispase II (2.5 U/mL, Roche) 2.5 mM CaCl₂ in DMEM media for 40 min at 37°C with shaking. Single-cell suspensions were prepared by passing tissue fragments through the mesh of 70-μm sterile filters (Fisherbrand). Cells were resuspended in Ham's F12:DMEM media containing 10% FBS and 1× antibiotic-antimycotic solution (Sigma) and seeded onto six-well plates. Nonadherent debris was discarded by replacement media every day for 3 days. Cells were maintained in same media, after first passage and the purity of population was confirmed by FACS analysis; cells were negative both for CD45 and CD31 and were positive for CD90, CD140a, and Sca1. Passages 3–6 were used for experiments. Fibroblasts were treated with TGF-β1 (2 ng/mL; R&D Systems) alone or with apyrase (2 U/mL; Sigma). TGF-β1 was purchased from R&D Systems. Apyrase was purchased from Sigma.

Extracellular ATP Quantification

Media (100 μL) from treated and untreated cardiac fibroblasts was carefully removed to avoid perturbation of the cells. A luciferase-based ATP assay kit was used to measure extracellular ATP concentration (ENLITEN ATP Assay; Promega, Madison, WI), and the ATP levels were quantified with a Synergy Mx Microplate Reader (BioTek Instruments). Standard curves using media supplemented with known ATP concentrations were generated and used to quantify ATP concentration in the experimental samples. Assays were conducted at 37°C.

Statistical Analysis

The results of experiments were analyzed using GraphPad Prism, version 6.0. Results are expressed as a means ± SE. For comparison between two groups, significance was determined by paired or unpaired Student’s t test with Welch’s correction. For comparison of multiple groups, multifactorial ANOVA with post hoc comparison of the means with Tukey correction was used to determine statistical significance. For interpretation and discussion, the authors used a threshold of P < 0.05 to indicate evidence of a significant effect. By the recent American Statistical Association statement on statistical significance and P values, we provide P values so the readers can judge the significance of the reported effects (31).

RESULTS

Decreased CD39 Levels and Activity in End-Stage Ischemic Cardiomyopathy in Humans

Table 1 provides the demographic and clinical parameters of the patient cohort. The control group consisted of hearts from patients with either stroke or ICH that could not be matched for cardiac transplantation. When compared with the control group, the ICM group is older (control, 41.8 ± 12 versus ICM, 57 ± 8.2; P = 0.04). The systolic blood pressure is also lower in the ICM group compared with the control group; however, this difference did not reach statistical significance. As expected, the ejection fraction of the ICM group is significantly lower than the control group (control, 58 ± 13 versus ICM, 22 ± 14; P = 0.016). The use of ACE inhibitors, β-blocker therapy, statin therapy, aspirin, and P2Y₁₂ antagonists is higher in the ICM group compared with the control group. In addition, the use of inotropic support is higher in the ICM group compared with the control group.

Table 1.

Patient demographic and clinical data

	Patient Population		P
Demographic and Clinical Parameters	Control	ICM
n	5	5
Age, yr	41.8 ± 12	57 ± 8.2	0.04
Sex, %male	80	60	0.492
Systolic BP, mmHg	115 ± 14	107 ± 12	0.46
Diastolic BP, mmHg	68 ± 8	70 ± 9	0.952
Heart rate, beats/min	118 ± 24	79 ± 9	0.016
Weight, kg	83.4 ± 20	84 ± 5	0.841
Height, cm	164 ± 14	167 ± 7	0.897
Ejection fraction, %	58 ± 13	22 ± 14	0.016
ACE inhibitor use, %	20	60	0.524
β-Blocker use, %	20	80	0.206
Inotrope support, %	0	80	0.048
Stroke or ICH, %	100	0	0.008
Statin use, %	20	100	0.048
Aspirin use, %	20	100	0.048
P2Y12 antagonist use, %	0	100	0.008

Values are means ± SE. ACE, angiotensin-converting enzyme; BP, blood pressure; ICM, ischemic cardiomyopathy; ICH, intracerebral hemorrhage.

Immunoblot analysis demonstrates a marked reduction in CD39 protein levels in ICM hearts compared with control hearts (Fig. 1, A and B; P = 0.008). Consistent with the changes in protein levels observed by immunoblot analysis, ATPase (Fig. 1D; P = 0.008) and ADPase (Fig. 1E; P = 0.008) activities were also significantly decreased in whole heart homogenate from ICM compared with control hearts. These data demonstrate altered expression of the extracellular purinergic-metabolizing molecule CD39 in the hearts of patients with end-stage ischemic cardiomyopathy. The impact of alterations of CD39 expression is the focus of our animal studies.

Figure 1.

CD39 is decreased in human ischemic cardiomyopathy hearts. A: representative immunoblot analysis of CD39 and GAPDH in control and ischemic cardiomyopathy (ICM) hearts from 3 of the 5 samples/group. B: quantification of CD39 levels in control and ICM hearts. C: quantification of ATP hydrolysis in control and ICM hearts. D: quantification of ADP hydrolysis in control and ICM hearts. n = 5 subjects/group. Means ± SD are plotted.P values from Student’s t test are shown.

CD39 Levels Increase in the Heart following Myocardial Infarction

First, to determine the temporal expression of CD39 during recovery from a myocardial infarction (MI), we subjected Cd39-β-galactosidase reporter mice, that express LacZ under the native CD39 promoter (Fig. 2), to permanent left coronary artery ligation (CAL). In noninjured hearts, CD39 expression was limited to the vasculature (sham, Fig. 2A, left arrow). After myocardial infarction, CD39 expression was increased in the heart at 7 days and 28 days post-MI (Fig. 2A, middle and right, respectively) and, coincided with the ischemic areas and the scar tissue. CD39 expression on cardiac fibroblast was confirmed at day 7 post-MI as measured by Cd39-β-galactosidase reporter analysis (Fig. 2B). These data demonstrate for the first time a temporal increase in CD39 expression on cardiac fibroblasts in the post-MI heart.

Figure 2.

CD39 is increased in the postmyocardial infarct (MI) hearts and cardiac fibroblasts. A: representative β-galactosidase staining of Cd39 reporter mice that have the lacZ gene under the control of the native Cd39 promoter were subjected to coronary artery ligation and at the designated times stained for β-galactosidase activity as an indicator of Cd39 expression. Representative hearts at baseline and at post-MI days 7 and 28 are shown. B: β-galactosidase CD39 reporter expression at day 7 in sham and post-MI mouse cardiac fibroblasts by FACS. MFI ± SD provided. n = 5–7 mice. P values from ANOVA analysis.

CD39 Limits the Progression of Heart Failure

In agreement with prior work (15), we found increased survival in mice devoid of CD39 expression and a decreased in cardiac rupture (Supplemental Fig. S2). To understand the role of CD39 activity in modulating post-MI cardiac recovery more fully, we examined infarct size in wild-type (WT) mice and Cd39-null (Cd39^−/−) mice subjected to coronary artery ligation (CAL). Following CAL, infarct size was comparable in WT and cd39^−/− mice at day 3 postinfarction (Fig. 3A, Supplemental Fig. S3). Echocardiographic measurements of cardiac function at day 3 postinfarction were similarly impacted in WT and Cd39^−/− mice (Fig. 3, B–D). When compared with baseline measurements, fractional shortening is also significantly, but comparably, reduced from baseline at day 3 following myocardial infarction in WT and cd39^−/− mice (Fig. 3B). EF showed comparable reductions (Supplemental Fig. S4A). Consistent with this finding, the left ventricular internal systolic diameter (Fig. 3C) and left ventricular internal diastolic diameter (Fig. 3D) were equally and significantly increased at day 3 post-MI in WT and Cd39^−/− mice.

Figure 3.

CD39 reduces cardiac dysfunction post-myocardial infarction (MI). A: quantification of myocardial infarct size at day 3 post-MI in WT and Cd39^−/− mice. B: quantification of baseline and day 3 post-MI fractional shortening. C: quantification of baseline and day 3 post-MI let ventricular (LV) systolic dimension. D: quantification of baseline and day 3 post-MI LV diastolic dimension. E: fractional shortening at baseline and day 28 post-MI. F: rate of left ventricular pressure rise (top; dP/dt_max) and pressure fall (bottom; dP/dt_min) in sham and day 28 post-MI hearts. G: left ventricular end-diastolic pressure in sham and day 28 post-MI hearts. H: isovolumic relaxation constant (Tau; Glantz method; n = 3–5/group). P values from Student's t test (A) and ANOVA analyses (B–H) are shown. WT, wild type.

Our in-depth analysis revealed profound differences related to parameters of cardiac function and fibrosis. At day 28 post-MI, there were no significant differences in body, heart, body/heart, lung, and liver weights between WT and Cd39^−/− mice (Table 2). However, fractional shortening in Cd39^−/− mice was significantly reduced compared with WT mice Fig. 3E; P = 0.046). Ejection fraction was comparably decreased at day 28 post-MI in WT and Cd39^−/− mice (Supplemental Fig. S4B). In vivo hemodynamic measurements reveal compromised contractile function, as the maximal rate of the left ventricular pressure change (dP/dt_max) is equally decreased in both WT and Cd39^−/− mice (Fig. 3F, top; P = 0.709). However, there is a significant decrease in diastolic function, measured as the rate of left ventricular pressure change during myocardial relaxation (lusitropy; dP/dt_min; Fig. 3F, bottom; P = 0.012) in Cd39^−/− compared with WT mice. Although at baseline, there was no difference in LV end-diastolic pressures, at 28 days post-MI there was a significant increase in left ventricular end-diastolic pressure in Cd39^−/− compared with WT mice (Fig. 3G; P < 0.0001). Importantly, Tau, the preload-independent measure of isovolumic relaxation, did not differ at baseline between WT and Cd39^−/− mice (Fig. 3H; P = 0.964). However, at day 28 post-MI, Tau was significantly increased in Cd39^−/− mice, compared with WT mice (Fig. 3H; P < 0.0001). Also consistent with compromised diastolic function, circulating B-type natriuretic peptide (BNP) was elevated in post-MI Cd39^−/− compared with WT mice (21.5 ± 1.17 versus 14.5 ± 2.24 ng/mL.; P = 0.033). Together, these data demonstrate a severely compromised diastolic function at 28-day post-MI in the mice devoid of CD39.

Table 2.

Necropsy comparison of organ measures in WT and Cd39^−/− mice at day 28 post-MI

	Genotype
	WT	Cd39−/−	P Value
n	14	21
Body weight, g	21.68 ± 3.82	22.92 ± 4.93	0.433
Heart weight, mg	166.3 ± 52.14	173.8 ± 49.0	0.654
Heart wt/body wt, mg/g	6.61 ± 1.38	7.83 ± 1.83	0.057
Lung weight, mg	173.2 ± 7.57	151.1 ± 6.37	0.340
Liver weight, mg	846.9 ± 21.95	882.9 ± 25.76	0.657
Tibia length, mm	17.56 ± 1.03	17.45 ± 1.14	0.767

Values are means ± SE. P value of t test comparison between wild-type (WT) and Cd39-null (Cd39^−/−) mice. MI, myocardial infarction.

CD39 Activity Is Associated with Profibrotic Pathways

Prior work suggests that matrix metalloproteinase-9 (MMP-9) and plasminogen activator inhibitor-1 (PAI-1) are critical early regulators of post-MI cardiac remodeling (32–38). Although at baseline, there was no difference in MMP-9 and PAI-1 levels in WT and Cd39^−/− hearts (Supplemental Fig. S5), at day 7 post-MI, MMP-9 levels were decreased in Cd39^−/− hearts when compared with WT hearts (Fig. 4, A and B; P = 0.046). In contrast, PAI-1, which regulates fibrin deposition (39), was increased at day 7 post-MI in Cd39^−/− hearts compared with WT hearts (Fig. 4, A and C; P = 0.02). Concordant with the increase in PAI-1, fibrin deposition was an increased in Cd39^−/− hearts compared with WT hearts (Fig. 4, D and E; P = 0.02). Together these data suggest that CD39 activity affects the expression of matrix metabolizing proteins in the early post-MI heart. Given that increased PAI-1 expression is associated with increased cardiac fibrosis in models of cardiac injury (40), we next examined how ablation of Cd39 impacts late post-MI cardiac remodeling and the development of heart failure.

Figure 4.

CD39 modulates the extracellular matrix metabolizing system post-myocardial infarction (MI). A: representative immunoblot analysis of MMP-9, PAI-1, and GAPDH at day 7 post-MI. B: quantification of MMP-9 levels at day 7 post-MI. C: quantification of PAI-1 levels at day 7 post-MI. D: representative immunohistochemistry staining of fibrin at the border zone of WT and Cd39^−/− hearts at day 7 post-MI (×20). E: quantification of fibrin staining at day 7 post-MI. N = 3–5 mice/group. P values from Student’s t test analysis are shown.

CD39 Activity Is Necessary to Resolve Fibrogenesis

Picrosirius red staining of hearts at day 28 post-MI demonstrated increased collagen deposition in Cd39^−/− versus WT hearts in both the border zone and the remote myocardium (Fig. 5, A and B; P = 0.002), consistent with increased fibrogenesis in mice devoid of CD39.

Figure 5.

CD39 limits post-myocardial infarction (MI) fibrosis. A: representative picrosirius staining in remote and border zone in WT and Cd39^−/− hearts at day 28 post-MI (bar = 250 µm). B: quantification of Picrosirius red (type I: thick fibers) and green (type III; thin fibers) birefringence in the remote and border zone at day 28 post-MI (n = 10–12 mice/group). C: representative immunoblot analyses of CD39, αSMA, elastin, collagen I, and GAPDH at day 28 post-MI (n = 3 mice/group). D–F: quantification of αSMA (D), collagen I (E), and elastin levels (F) in WT and Cd39^−/− hearts at day 28 post-MI (N = 3–4 mice/group). G: quantification of load required for rupture in sham and day 28 post-MI WT and Cd39^−/− hearts (n = 6–10 mice/group). H: quantification of active TGF-β1 levels in sham, day 7 post-MI, and day 28 post-MI WT and Cd39^−/− hearts (n = 4–5 mice/group). P values from Student’s t test are shown. WT, wild type.

Prior work suggests that the persistence of myofibroblasts in the heart correlates with adverse remodeling and fibrosis (41). At day 28 post-MI, immunoblot analysis demonstrated increased aSMA (Fig. 5, C and D; P = 0.028) and collagen-I (Fig. 5, C and E; P = 0.003) levels, but decreased elastin levels (Fig. 5, C and F; P = 0.047), in Cd39^−/− versus WT hearts. Prior work suggests that the balance between collagen-I and elastin levels is critical to maintaining tissue compliance and tensile strength (42). Analysis of the tensile strength of sham hearts revealed no difference in the load necessary for mechanical tissue failure in sham-treated Cd39^−/− hearts compared with WT hearts (Fig. 5, G and F; P = 0.47). However, at day 28 post-MI, myocardial tensile strength was significantly increased in Cd39^−/− hearts compared with day 28 post-MI WT hearts (Fig. 5G; P = 0.002). This increase in tensile strength demonstrates that the post-MI Cd39^−/− myocardium is more stiff than the post-MI WT myocardium, consistent with our in vivo findings of impaired diastolic function (Fig. 3, F and H) and our ex vivo findings of increased fibrosis (Fig. 5A).

Given the central role of TGF-β1 in mediating fibroblast function and post-MI fibrosis, we measured the levels of TGF-β1 in WT and Cd39^−/− hearts. Although active TGF-β1 levels were low in WT and Cd39^−/− sham hearts (Fig. 5H; P = 0.996), active TGF-β1 levels were significantly elevated in Cd39^−/− hearts compared with WT hearts at both day 7 (Fig. 5H; P = 0.012) and day 28 post-MI (Fig. 5H; P = 0.007). These data suggest that an absence of CD39 activity results in the failure to resolve the factors that drive fibrogenesis in post-MI hearts.

CD39 Activity Modulates the Response of Cardiac Fibroblasts to TGF-β1

TGF-β1 stimulation transforms cardiac fibroblasts (CFs) to myofibroblasts, critical mediators of post-MI fibrosis. Therefore, to understand the interaction between TGF-β1 signaling and extracellular nucleotide metabolism, we examined the impact of CD39 activity on the TGF-β1-induced expression of specific fibrosis-associated genes in untreated and TGF-β1-treated WT and Cd39^−/− CFs (Fig. 6). Following TGF-β1 stimulation, when compared with WT CFs, Cd39^−/− CFs expressed increased levels of Collagen-Ia (Fig. 6A; P = 0.049). Prolyl-4-hydroxylase-1 (P4HA1), the enzyme that catalyzes the formation of 4-hydroxyproline in collagens (Fig. 6B; P = 0.039), and connective tissue growth factor (CTGF), a matricellular protein of the CCN family of extracellular matrix-associated heparin-binding proteins that is a central mediator of tissue remodeling and fibrosis, (Fig. 6C; P < 0.0001) (43). Plasminogen activator inhibitor-1 (PAI-1), which inhibits fibrinolysis by inhibiting plasminogen activators tissue plasminogen activator (tPA) and urokinase (uPA) (40), was also increased in Cd39^−/− CFs (Fig. 6D; P = 0.002). Similarly, there was higher expression of tenascin-C (TNC), an extracellular matrix protein that can drive persistence of organ fibrosis (Fig. 6E; P = 0.042) (43), in TGF-β1 stimulated Cd39^−/− CFs. Thrombospondin-1 (THBS1), which interacts with latency-associated peptide (LAP) to release TGF-β1 (Fig. 6F; P = 0.039), and Yes-associated protein (YAP-1), a transcriptional effector of the Hippo pathway that is a key coordinator of fibroblast activation and matrix synthesis (Fig. 6G; P = 0.028) (44), were also increased in TGF-β1 stimulated Cd39^−/− CFs. α-SMA, a marker of myofibroblast transformation, and lysyl oxidase (LOX), the enzyme responsible for the formation of collagen cross-links, trended higher in Cd39^−/− CFs compared with WT CFs (Fig. 6H; P = 0.075 and Fig. 6I; P = 0.165, respectively). However, neither secreted protein acidic and rich in cysteine (SPARC, osteonectin), a collagen-binding protein shown to be antiproliferative and counter adhesive (Fig. 6J; P = 0.811) (45), nor decorin (DCN), a leucine-rich proteoglycan that regulates collagen fibrillogenesis (Fig. 6K; P = 0.811) (46), differed between WT and Cd39^−/− TGF-β1-stimulated CFs. Interestingly, TGF-β1-induced elastin expression, critical for shape resumption after stretching or contracting of the heart, was impaired in Cd39^−/− CFs (Fig. 6L; P = 0.055). Importantly, there was no difference in the level of pSMAD2 following TGF-β1 treatment between WT and Cd39^−/− CFs, demonstrating that the observed transcriptional changes are not due to alterations in canonical TGF-β1 signaling in Cd39^−/− CFs (Fig. 7, A and B).

Figure 6.

CD39 activity regulates the response of cardiac fibroblasts to TGF-β1. A–L: relative fold increase from basal expression induced by TGF-β1 stimulation in WT and Cd39^−/− cardiac fibroblasts of collagen I (Col-I; A), prolyl-4-hydroxylase-1 (P4HA1; B), connective tissue growth factors (CTGF; C), plasminogen activator inhibitor-1 (PAI-1; D), tenascin-C (TNC; E), thrombospondin-1 (THBS1; F), Yes-associated protein (YAP-1; G), α-smooth muscle actin (aSMA; H), secreted protein acidic and rich in cysteine (SPARC; I), decorin (DCN; J), lysyl oxidase (LOX; K), and elastin (L). n = 3–4 replicates/group. P values from Student’s t test are shown. WT, wild type.

Figure 7.

SMAD-2 and phospho-SMAD-2 levels do not differ between WT and Cd39-null cardiac fibroblasts. A: representative immunoblot analysis of phospho-SMAD2 and total SMAD2. B: quantification of phospho-SMAD2/SMAD2 levels in untreated and TGF-β1-treated WT and Cd39^−/− cardiac fibroblasts. n = 3 replicates/group. P values from ANOVA analysis are provided. WT, wild type.

The observed impact of CD39 knockout on the cardiac fibroblast transcriptome suggested that TGF-β1 might stimulate nucleotide release. Following stimulation of murine CF with TGF-β1 we found a significant increase in ATP release (Fig. 8A). Importantly, there was no difference in the release of ATP from Cd39^−/− CFs, and that the levels of ATP may trend higher given the lack of extracellular hydrolysis in the absence of CD39. Together, these data demonstrate that TGF-β1-induced ATP release is an autocrine and/or paracrine profibrotic signaling loop that potentiates CF activation. Indeed, ATP potentiates a variety of signaling cascades, most notably, in the nerve termini where it amplifies adrenergic signaling (47). Furthermore, TGF-β1 stimulation of WT CFs increased both CD39 expression (Fig. 8, B and C) and activity (Fig. 8D), thereby promoting nucleotide degradation, and limiting P2 receptor signaling. Except for the P2ry6 receptor, which was robustly expressed on both WT and Cd39^−/− CFs, there were no differences in the basal expression levels of P2-puriergic receptors (Supplemental Fig. S6). Together these data suggest that TGF-β1-mediated upregulation of CD39 expression and activity is a critical negative regulator of TGF-β1-induced nucleotide potentiation of fibroblast activation.

Figure 8.

TGF-β1 stimulation increases ATP release and CD39 expression. A: extracellular ATP levels at baseline and following TGF-β1 stimulation of WT and Cd39^−/− cardiac fibroblasts. B: representative fluorescence-activated cell sorter histograms of cell surface expression of CD39 at baseline and following TGF-β1 stimulation of WT cardiac fibroblasts. C: quantification of mean fluorescence intensity of CD39 expression at baseline and following TGF-β1 stimulation of WT cardiac fibroblasts. D: extracellular nucleotide hydrolysis at baseline and following TGF-β1 stimulation of 1.5 × 10⁴ WT cardiac fibroblasts. n = 4–5 replicates/group. P values from ANOVA analysis (A) and Student’s t test (C and D) are shown. WT, wild type.

Ecto-Nucleotidase Treatment Rescues Cd39-Null Fibroblasts

To determine if exogenous ecto-nucleotidase activity influences the cardiac fibroblast response to TGF-β1 stimulation, we examined if treatment of TGF-β1 stimulated Cd39^−/− CFs with apyrase rescues their profibrotic phenotype (Fig. 9). Apyrase treatment of TGF-β1-stimulated Cd39^−/− CFs decreased expression of collagen-Ia (Fig. 9A; P = 0.03) and plasminogen activator inhibitor-1 (PAI-1; Fig. 9B; P < 0.001), establishing that nucleotide signaling contributes to their expression. However, there was a nonsignificant reduction in the expression of aSMA (Fig. 9C; P = 0.253), CTGF (Fig. 9D; P = 0.374), and YAP (Fig. 9E; P = 0.434). Critically, elastin expression increased with apyrase treatment of TGF-β1-stimulated Cd39^−/− CFs (Fig. 9F; P = 0.002), demonstrating that nucleotide-mediated signaling events appear to modulate elastin gene expression in CFs. Together these data establish that TGF-β1-induced Cd39 activity is necessary to terminate nucleotide-mediated potentiation of fibroblast activation and thereby modulate fibrosis.

Figure 9.

Exogenous ectonucleotidase treatment rescues the profibrotic phenotype of TGF-β1-stimulated CD39-null cardiac fibroblasts. A–F: relative expression of collagen1a (A), plasminogen activator inhibitor-1 (PAI-1; B), α-smooth muscle actin (aSMA; C), connective tissue growth factors (CTGF; D), Yes-associated protein-1 (YAP-1; E), and elastin (F) in Cd39^−/− CFs at baseline, following TGF-β1 stimulation and following TGF-β1 stimulation in the presence of apyrase. n = 3–4 replicates/group. P values from ANOVA analysis are shown.

DISCUSSION

CD39 is a key cell surface ecto-nucleotidase that hydrolyzes extracellular ATP or ADP, thereby extinguishing P2 receptor-mediated purinergic signaling. Here, we report for the first time that in human end-stage ischemic cardiomyopathy, CD39 expression and activity is reduced in the heart. To begin to understand how differences in CD39 levels impact in cardiac remodeling, we examined a mouse model of ischemic heart disease. Our data demonstrate for the first time that CD39 is a critical regulator of TGF-β1-mediated fibroblast activation and adverse cardiac remodeling following myocardial infarction. Mice devoid of CD39 have increased fibrosis associated with compromised cardiac function. Furthermore, in isolated cardiac fibroblasts, we found that Cd39 expression regulates the TGF-β1-induced profibrotic transcriptome. Together our findings demonstrate that CD39 activity is a negative regulator of cardiac fibrosis.

To our knowledge, this is the first evaluation of the purinergic pathway in end-stage ischemic patients with heart failure compared with control subjects. Although two prior studies have reported the expression of CD39 in relation to human heart disease, neither quantified protein or activity (48, 49). The first study demonstrated colocalization of CD39 and caveolin in human right atrial muscle obtained from 17 patients with ischemic heart disease undergoing CABG surgery (48). In contrast to our work using left ventricular muscle, increased ATPase activity was observed in these atrial samples. However, clinical data regarding the ejection fraction and the stage of heart failure were not provided, thereby preventing a comparison of the patient population examined to our study population. The second report measured leukocyte mRNA levels as a surrogate for messenger RNA levels in the heart. No change in the level of CD39 in leukocytes from patients with stage I–IV heart failure was reported (49). In our study, we measured the level of CD39 and the ectonucleotidase activity in human left ventricle samples revealing that CD39 activity is decreased in end-stage heart failure is intriguing. Although there was a small difference in the average age between the two groups examined, there are no data, to date, that suggests a difference in the level of CD39 expression with age. Therefore, a disease-associated decrease in CD39 activity could not only result in increased ATP levels in the interstitium, but also in decreased adenosine levels. Such a scenario could create a proinflammatory and profibrotic microenvironment in the ischemic heart that contributes to progressive heart failure and fibrosis. Indeed, comparable endogenous crosstalk pathways have been described for neutrophil-endothelial interactions (50). Therefore, targeting the purinergic metabolizing system may provide a unique approach to attenuate cardiac fibrosis.

Although end-stage human hearts displayed decreased levels of CD39, in the acute post-MI setting in mice, we found that CD39 levels and activity were actually iniitially increased in hearts following MI in interstitial fibroblasts.

To understand the role of CD39 activity in modulating post-MI cardiac recovery, we subjected wild-type (WT) mice and Cd39-null (Cd39^−/−) mice to coronary artery ligation (CAL). We chose this model because we have previously demonstrated that increased expression of CD39 reduces myocardial infarct size in ischemia-reperfusion models in mice (5) and pig (7) using 60 min of ischemia and 24 h of reperfusion. Furthermore, one report suggests that knockout of cd39 increases infarct size at 4 h following reperfusion (10). Thus, we chose the CAL-induced MI model, rather than an ischemia-reperfusion (IR) model, so that we could compare mice with uniform infarcts given that genotype-driven differences in the infarct size can affect inflammation, fibrosis, and cardiac function (25, 51). Furthermore, this model is clinically relevant; up to 12.9% of patients with an MI achieve inadequate microvascular reperfusion (TIMI flow < 3), despite early percutaneous reperfusion, resulting in worse clinical outcomes (52).

Interestingly, we found that a lack of CD39 activity increases cardiac fibrosis. The mechanism for the increased cardiac fibrosis is due, in part, to effects on matrix metabolizing proteins, as MMP-9 levels are decreased in Cd39-null hearts post-MI. Previous studies have also demonstrated that inhibition of MMPs and plasminogen activators can lead to cardiac failure (38, 53). In Cd39-null hearts, not only were MMP-9 levels decreased but PAI-1 levels were also significantly increased, and there was an increase in fibrin deposition. Plasmin, a serine protease, plays a critical role in the degradation of extracellular matrix directly and also in the activation of pro-MMPs in cardiac tissue (54). Increased PAI-1 inhibits the activation of plasmin and matrix metalloproteinases and thereby impacts the development of tissue fibrosis. Our findings that PAI-1 is increased in Cd39-null hearts suggested to us that CD39 activity can regulate PAI-1 levels in cardiac fibroblasts. Indeed, prior studies have suggested that ATP stimulates a transient upregulation of PAI-1 on cardiac fibroblasts, however, uridine nucleotides increase PAI-1 for a more prolonged period of time (13, 55, 56). Importantly, the upregulation of PAI-1 by nucleotides was much less than that induced by TGF-β1, the primary stimulator of PAI-1 expression in fibroblasts (57).

Post-MI, Cd39-null hearts had increased levels of active TGF-β1. Furthermore, TGF-β1 stimulation of Cd39-null cardiac fibroblasts resulted in a greater increase in PAI-1 and collagen expression, but a decrease elastin expression, compared with CD39-expressing cardiac fibroblasts. Our findings demonstrate that the absence of CD39 activity promotes a more profibrotic cardiac fibroblast phenotype. In support of this tenet, additional profibrotic markers including, tenascin-C and thrombospondin were also significantly upregulated in Cd39-null cardiac fibroblasts. Tenascin-C is an extracellular matrix protein, which drives persistent organ fibrosis and aggravates left ventricular fibrosis (43, 58). Thrombospondin is a well-established activator of latent TGF-β1 (59). Therefore, an increase in thrombospondin expression by Cd39-null cardiac fibroblasts would favor TGF-β1 activation. Indeed, we found that in post-MI Cd39-null hearts activated TGF-β1 levels were increased. Together these data support our hypothesis that CD39 activity regulates the profibrotic transcriptome in TGF-β1-stimulated cardiac fibroblasts. Importantly, our ex vivo studies comparing the profibrotic phenotype of wild-type and Cd39-null cardiac fibroblasts were conducted in the absence of exogenous nucleotides, indicating that autocrine purinergic stimulation potentiates the specific up- or downregulation of fibrotic gene expression.

To understand the relationship between TGF-β1 and nucleotide signaling, we first measured the release of ATP after TGF-β1 stimulation. Critically, we found that TGF-β1 stimulation of mouse cardiac fibroblasts increases ATP release. These data suggest a novel signal-potentiating loop between ATP and TGF-β1 (Fig. 10). Indeed, this tenet is supported by established biology; ATP potentiates a variety of signaling cascades, most notably, in the nerve termini where it amplifies adrenergic signaling (47). Our findings that the expression of several TGF-β1-regulated genes is comodulated by CD39 activity and purinergic receptor signaling are novel. Our data suggest that limiting nucleotide-mediated signaling may be a novel approach to prevent excessive post-MI fibrosis. In addition, the finding that TGF-β1 stimulation increases CD39 expression and activity on cardiac fibroblasts, suggests that the upregulation of CD39 activity regulates nucleotide-mediated autocrine signaling in TGF-β1-stimulated cardiac fibroblasts. Prior studies have shown that other factors also modulate CD39 expression. SP1 and Hypoxia Inducible Factor (HIF)-mediated pathways appear to modulate CD39 expression in other cells during ischemia or hypoxia (60–62). Our data establish that TGF-β1 upregulation of CD39 expression and activity is a key feedback inhibitory loop of TGF-β1-induced nucleotide potentiation of fibroblast activation (Fig. 10).

Figure 10.

TGF-β1 and purinergic signaling in the regulation of cardiac fibrosis. TGF-β1 stimulation induces the release of ATP, which forms a feed-forward loop through purinergic receptors to potentiate a profibrotic fibroblast phenotype. TGF-β1 also induces the expression of CD39, which forms a critical inhibitory loop, hydrolyzing extracellular ATP, and thereby terminating nucleotide-mediated potentiation of fibroblast activation. Loss of CD39 expression results in increased expression of collagen but a decrease in elastin expression following TGF-β1 stimulation of cardiac fibroblasts, thereby contributing to increased cardiac stiffness following myocardial infarction.

Prior studies have examined the direct impact of ATP stimulation on rat cardiac fibroblast activation (63–65). These studies suggest that ATP increases fibroblast proliferation (66), collagen synthesis (56), and α-SMA expression, a marker of myofibroblast differentiation (67, 68). Also, treatment of ATP-stimulated rat cardiac fibroblasts with apyrase decreases collagen synthesis and increased fibroblast migration (56, 69). However, we have found that the induction of collagen expression by ATP alone is much lower than that induced by TGF-β1 (data not shown). Prior work also suggests that fibroblast-expressed CD39 establishes the set point for ATP induction of a profibrotic phenotype (13). In contrast, our results have revealed that TGF-β1 increases both ATP release and CD39 expression and activity. Furthermore, the inability to upregulate Cd39 expression (Cd39^−/−) in response to TGF-β1 stimulation results in enhanced expression of key profibrotic genes. This finding is quite novel as it suggests that nucleotide signaling potentiates the expression of certain elements of the profibrotic transcriptome. We have not examined the impact of TGF-β1 on the expression of purinergic receptors as this is beyond the current study, however, it should be noted that purinergic receptors have been implicated in modulating cardiac physiology and the response to myocardial injury (70–72). Finally, our data show that exogenous ecto-nucleotidase treatment rescues the hyperfibrotic phenotype induced by TGF-β1 treatment of CD39-null cardiac fibroblasts, thereby identifying a novel mechanism by which post-MI adverse cardiac remodeling can be targeted. Together our studies establish that CD39 is a critical negative modulator of TGF-β1-induced profibrotic signaling pathways (Fig. 10).

The use of global knockout animals limits the ability to determine the impact of cell-specific expression of CD39 on the observed pathophysiology. Our findings confirm prior reports that suggest an increase in cardiac survival in mice devoid of CD39 expression (15). However, the improved survival comes at a cost of increased fibrosis and worsened diastolic function. Our studies support that fibroblast-expressed CD39 modulates the TGF-β1 pathways that contribute to fibrosis. Future studies using cell-specific knockout of CD39 are in progress to address the contributions of fibroblast-specific expression of CD39 on cardiac fibrosis. Importantly, the impact of CD39 activity on cardiac fibrosis provides an intriguing hypothesis for translational studies. Whether treatment with soluble CD39 or CD39-like molecules during myocardial healing can prevent excessive cardiac fibrosis is the focus of ongoing studies in our laboratory.

Conclusions

CD39 activity checks cardiac fibrosis. CD39 is a critical modulator of TGF-β1-mediated fibroblast activation and cardiac remodeling following myocardial infarction via modulation of nucleotide signaling. TGF-β1-induced CD39 expression generates a negative feedback loop that attenuates cardiac fibroblast activation and restricts the in vivo fibrotic response following myocardial infarction to limit collagen deposition and maintain elastin levels (Fig. 10) resulting in compensated cardiac function. In the absence of CD39 activity, diastolic dysfunction is worsened. Critical to the translation of these data is the fact that treatment with the CD39-like molecule, ecto-apyrase, attenuates the TGF-β1-induced profibrotic phenotype of cardiac fibroblasts. Therefore, treatment with soluble or liposomal preparations of CD39 may represent a novel approach to combat post-MI cardiac fibrosis.

SUPPLEMENTAL DATA

Supplemental Figs. S1–S6 and Supplemental Tables S1 and S2:https://doi.org/10.6084/m9.figshare.19350896.

GRANTS

This work was supported in part by National Institutes of Health (NIH) Training Grant 5T32HL007411-35 (to R.C.) and NIH Grants P01 HL091799 (to W.J.K.); P01 HL107152, P01 HL087203, and R01DK103723 (to S.C.R.); and R01-HL127442-A1, R21-HL096038, and K08-HL094703 (to R.J.G.) and the Robert J. Anthony Fund for Cardiovascular Research (to R.J.G.).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

DISCLAIMERS

The content is solely the responsibility of the authors and does not necessarily represent the official views of any of the funding agencies. All authors have reviewed and approved the article.

AUTHOR CONTRIBUTIONS

T.N., R.C., D.G.W., K.M.D., P.J.C., Y.R.S., T.S.A., L.M.B., W.J.K., S.B., I.F., S.C.R., E.G., and R.J.G. conceived and designed research; T.N., S.N., R.C., D.G.W., E.C., O.B-B., Z.X., B.B., H.H., S.N.M., I.F., S.C.R., E.G., and R.J.G. performed experiments; T.N., R.C., D.G.W., E.C., O.B-B., Z.X., B.B., H.H., S.N.M., T.S.A., J.S., L.M.B., S.B., I.F., S.N., S.C.R., E.G., and R.J.G. analyzed data; T.N., S.N., R.C., D.G.W., Z.X., B.B., S.N.M., P.J.C., Y.R.S., J.S., S.B., I.F., S.C.R., and R.J.G. interpreted results of experiments; T.N., S.N., R.C., D.G.W., E.C., O.B-B., and R.J.G. prepared figures; T.N., R.C., D.G.W., B.B., T.S.A., J.S., L.M.B., S.C.R., E.G., and R.J.G. drafted manuscript; T.N., S.N., R.C., D.G.W., E.C., O.B-B., Z.X., B.B., H.H., S.N.M., K.M.D., P.J.C., Y.R.S., T.S.A., J.S., L.M.B., W.J.K., S.B., I.F., S.C.R., E.G., and R.J.G. edited and revised manuscript; T.N., R.C., D.G.W., E.C., Z.X., B.B., H.H., S.N.M., K.M.D., P.J.C., Y.R.S., T.S.A., J.S., L.M.B., W.J.K., S.B., I.F., S.C.R., E.G., and R.J.G. approved final version of manuscript.