Research ArticleCardiology

HDAC inhibition improves cardiopulmonary function in a feline model of diastolic dysfunction

See allHide authors and affiliations

Science Translational Medicine  08 Jan 2020:
Vol. 12, Issue 525, eaay7205
DOI: 10.1126/scitranslmed.aay7205

Counteracting cardiopulmonary dysfunction

In patients with heart failure, impaired relaxation of the left ventricle (diastolic dysfunction) results in incomplete filling with blood and increased pressure. Wallner et al. treated a feline model of slow-progressive pressure-overload–induced diastolic dysfunction with the histone deacetylase inhibitor SAHA. SAHA increased myofibril relaxation, improved pulmonary function, reduced left ventricular filling pressures, and reduced left ventricle hypertrophy in cats. The authors determined that SAHA altered acetylation of mitochondrial enzymes, resulting in enhanced mitochondrial respiration. Results suggest that histone deacetylase inhibition could potentially be beneficial for improving cardiopulmonary function.


Heart failure with preserved ejection fraction (HFpEF) is a major health problem without effective therapies. This study assessed the effects of histone deacetylase (HDAC) inhibition on cardiopulmonary structure, function, and metabolism in a large mammalian model of pressure overload recapitulating features of diastolic dysfunction common to human HFpEF. Male domestic short-hair felines (n = 31, aged 2 months) underwent a sham procedure (n = 10) or loose aortic banding (n = 21), resulting in slow-progressive pressure overload. Two months after banding, animals were treated daily with suberoylanilide hydroxamic acid (b + SAHA, 10 mg/kg, n = 8), a Food and Drug Administration–approved pan-HDAC inhibitor, or vehicle (b + veh, n = 8) for 2 months. Echocardiography at 4 months after banding revealed that b + SAHA animals had significantly reduced left ventricular hypertrophy (LVH) (P < 0.0001) and left atrium size (P < 0.0001) versus b + veh animals. Left ventricular (LV) end-diastolic pressure and mean pulmonary arterial pressure were significantly reduced in b + SAHA (P < 0.01) versus b + veh. SAHA increased myofibril relaxation ex vivo, which correlated with in vivo improvements of LV relaxation. Furthermore, SAHA treatment preserved lung structure, compliance, blood oxygenation, and reduced perivascular fluid cuffs around extra-alveolar vessels, suggesting attenuated alveolar capillary stress failure. Acetylation proteomics revealed that SAHA altered lysine acetylation of mitochondrial metabolic enzymes. These results suggest that acetylation defects in hypertrophic stress can be reversed by HDAC inhibitors, with implications for improving cardiac structure and function in patients.


Heart failure (HF) with preserved ejection fraction (HFpEF) accounts for about 50% of all cases of HF, and its prevalence relative to HF with reduced ejection fraction (HFrEF) is growing by 10% per decade. The HFpEF prognosis is poor, with a 5-year survival rate as low as 50%. Furthermore, the hospitalization rate is high, and quality of life for patients is substantially compromised (1, 2). There are currently no U.S. Food and Drug Administration (FDA)–approved therapies for this condition.

Inhibitors of excess neurohormonal activity have been shown to decrease mortality in patients with HFrEF in large clinical outcome trials (3). However, the effects of these same neurohormonal inhibitors in patients with HFpEF have consistently failed to reach positive primary outcomes (4). The dichotomy in the responses of patients likely results from differences in HFpEF versus HFrEF pathophysiology and the absence of experimental HFpEF models that capture essential characteristics of this syndrome that could be used to test new HFpEF therapies (5). Although all HFpEF animal models have limitations, rodent models have inherent limitations due to their small size and differences in physiological properties compared with large mammals (6). New large-animal models that recapitulate critical features of human HFpEF might be more useful in identifying beneficial HFpEF therapies. Most studies using large-animal HFpEF models have mainly focused on cardiac function and structure, without assessing pulmonary changes (79), although dyspnea, particularly during exercise, is one of the cardinal symptoms observed in patients with HFpEF. Furthermore, pulmonary hypertension (PH) is common in patients with HFpEF (36 to 83%) and strongly associated with morbidity and mortality (1012). Our group previously established a feline model that has critical features of HFpEF, including left ventricular hypertrophy (LVH), elevated left ventricular (LV) filling pressures [left ventricular end-diastolic pressure (LVEDP)], left atrial (LA) enlargement, increased concentration of plasma N-terminal probrain natriuretic peptide (NT-proBNP), and impaired pulmonary mechanics that contribute to functional compromises and poor oxygenation. We recently reported an in-depth structural and functional cardiopulmonary characterization of the model, which supports its translational efficacy (13).

The aim of the current study was to determine whether inhibition of histone deacetylase (HDAC) catalytic activity could prevent and/or reverse the adverse structural and functional cardiopulmonary remodeling in this feline model of slow-progressive pressure overload. We used SAHA (suberoylanilide hydroxamic acid, vorinostat), a pan-HDAC inhibitor, that has FDA approval for the treatment of cutaneous T cell lymphoma (CTCL) (14). Histone deacetylation leads to compaction of chromatin, therefore repressing mRNA synthesis and altering gene expression. HDACs also catalyze removal of acetyl groups from ε-amino groups of lysine residues in a variety of nonhistone proteins, thus having a role in the regulation of posttranslational protein modification affecting transcription and other protein functions (15, 16). In experimental models of HF (pressure overload, ischemia-reperfusion), small-molecule inhibitors of HDAC catalytic activity have demonstrated potent cardioprotective and antiremodeling effects, as seen by a decrease in scar size, attenuated cardiac hypertrophy, and improved systolic function (1719). A recent report showed that HDAC inhibition in rat and mouse models of diastolic dysfunction improved diastolic function by potentiating cardiac myofibril relaxation (20). These studies suggest that HDAC inhibition promotes beneficial remodeling in cardiac disease phenotypes through a nongenomic mechanism. The present study tests the hypothesis that HDAC inhibition can reverse and/or prevent adverse cardiopulmonary remodeling in a large-animal model of diastolic dysfunction.


The experimental protocol is displayed in Fig. 1A. Subcutaneous injections of SAHA and vehicle were administered daily and were well tolerated. SAHA did not alter total counts of white blood cells, red blood cells (RBC), and platelets, or result in changes to RBC or platelet size (table S1). There was no difference in plasma concentrations of blood urea nitrogen between groups, suggesting that kidney function was not negatively affected by slow-progressive pressure overload or SAHA treatment (fig. S1A). Furthermore, SAHA treatment did not cause QT interval prolongation (fig. S1B), which had been previously described in a phase 1 trial (21). The treatment strategy used in this study did not induce any known toxic side effects of SAHA. As previously reported (13), banded animals displayed symptoms of HF such as tachypnea and diaphragmatic breathing, particularly in response to physical exertion.

Fig. 1 Experimental study design and remodeling effects of SAHA.

(A) Baseline measurements were collected, and felines then underwent the aortic banding or sham procedure. After 2 months, banded cats were assigned to either daily subcutaneous injection of vehicle (veh) or SAHA (10 mg/kg) for 2 months. At 4 months after banding, invasive hemodynamic studies were performed. Conventional echocardiography data showing (B) left ventricular wall thickness, (C and D) LA size, (E) LA function, (F) LV diastolic function, (G) LV EF, and (H) LVEDD in sham, vehicle-treated, and SAHA-treated animals. (I) Speckle-tracking–based strain analysis was quantified from sham, vehicle-treated, and SAHA-treated animals. (J) Parasternal long-axis images showing septal wall thickness and LA size. (K) Parasternal short-axis images with superimposed regional radial strain values. Data shown are means ± SEM. Statistical significance was determined by linear mixed-effects models. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 between b + vehicle and sham. §P < 0.05, §§P < 0.01, §§§P < 0.001, §§§§P < 0.0001 between b + SAHA and sham. #P < 0.05, ##P < 0.01, ###P < 0.001, ####P < 0.0001 between b + SAHA and b + vehicle. NS, not significant. Ao, aorta; BGA, blood gas analysis; BL, baseline; E, peak early diastolic transmitral velocity; e′, spectral tissue Doppler-derived peak early diastolic velocity; Echo, echocardiography; EF, ejection fraction; HFpEF, heart failure with preserved ejection fraction; LA, left atrium; LA EF, left atrial ejection fraction; LAVES, left atrial end-systolic volume; LV, left ventricle; LVEDD, left ventricular end-diastolic diameter; PFT, pulmonary function testing; SAHA, suberoylanilide hydroxamic acid. The number of animals included in each parameter is reported in table S7.

SAHA partially rescues HFpEF phenotype

At 2 months after banding, before starting SAHA treatment, all banded animals exhibited a similar cardiac phenotype with significant LVH (P < 0.0001), LA enlargement (P < 0.001), and reduced radial strain values (P < 0.05), but with preserved LV ejection fraction (EF) when compared with sham-operated animals (Fig. 1). SAHA treatment prevented progression of LV hypertrophy, as reflected by a decrease in LV wall thickness [4 months: 4.5 ± 0.1 mm (SAHA) versus 6.1 ± 0.2 mm (vehicle); P < 0.0001] (Fig. 1B) and significantly reduced LA size, reflected by a decrease in LA volume (P < 0.0001; Fig. 1C) and left atrial–to–aortic root size ratio (LA/Ao) (P < 0.0001; Fig. 1D). After 2 months of SAHA treatment, LA size and LV wall thickness did not differ remarkably between SAHA-treated and sham animals, whereas vehicle-treated animals had progressive deterioration of these parameters. LA EF, a parameter of LA function, also improved with SAHA treatment and was significantly greater than in vehicle-treated animals (P = 0.0011; Fig. 1E). Furthermore, SAHA treatment attenuated an increase in peak early diastolic transmitral velocity/spectral tissue Doppler-derived peak early diastolic velocity (E/é) compared with vehicle-treated animals at 4 months, indicative of lower LV filling pressures (Fig. 1F) (22). LV EF (Fig. 1G) and LV end-diastolic diameter (LVEDD) (Fig. 1H) were not affected by aortic banding and did not differ between groups at any time point, suggesting normal LV capacitance. However, speckle tracking–based strain analysis, a sensitive method used to assess myocardial function (23), revealed a significant impairment in all banded animals at 2 months compared with sham-operated animals (P < 0.05). By 4 months, the global radial strain further decreased in the vehicle-treated animals but did not decline in the SAHA group (Fig. 1I). Representative parasternal long-axis b-mode echocardiogram (Echo) images (Fig. 1J) display LV wall thickness and LA size at 4 months. Representative short-axis b-mode images with superimposed regional radial strain values (Fig. 1K) show differences in wall thickness and myocardial contractility. These data document that SAHA treatment in banded animals had strong anti-remodeling effects (LV wall thickness and LA size) and improved LV systolic and diastolic function.

SAHA improves diastolic function and decreases LV filling pressures

Invasive hemodynamic measurements were performed at 4 months after banding. Mixed-venous and arterial blood gas analysis (BGA), cardiac output (CO), pulmonary and cardiac pressures were recorded before (dob−) and after (dob+) intravenous infusion of dobutamine (5 μg kg−1 min−1). Pressure gradients across the aortic band were measured using the fractional flow reserve technique, which is conventionally used to assess coronary physiology in humans. The mean systolic aortic pressure gradient across the band did not differ between SAHA- (33 ± 6 mmHg) and vehicle-treated (38 ± 9 mmHg, P = 0.6) animals (Fig. 2A), indicating comparable pressure overload between banded groups. CO (Fig. 2B) and heart rate (HR) (Fig. 2C) did not differ significantly between groups, although SAHA-treated cats tended to have lower HRs. dP/dtmax was comparable between groups before dobutamine infusion; however, the increase in dP/dtmax with dobutamine was significantly less in vehicle-treated (4947 ± 598 mmHg/s) compared with SAHA-treated animals (7690 ± 723 mmHg/s, P < 0.05) (Fig. 2D), suggesting depressed contractile reserve in vehicle-treated banded animals versus shams and improved inotropic reserve after SAHA treatment. dP/dtmin (Fig. 2E) was reduced in both banded groups compared with sham. However, the ratio between diastolic time interval and LV isovolumic relaxation constant (t-dia/τ), an index of relaxation, was significantly smaller in vehicle-treated (3.9 ± 0.3, P < 0.01) but not in SAHA-treated banded animals (5.2 ± 0.4, P = 0.16) when compared with shams (6.4 ± 0.7) (Fig. 2F). SAHA treatment also significantly reduced LVEDP (P < 0.01) and mean pulmonary arterial pressure (mPAP) (P < 0.01) compared with vehicle-treated banded animals (Fig. 2, G and H). These results show that SAHA treatment improves both systolic and diastolic function in banded animals.

Fig. 2 SAHA improves diastolic function.

(A) Aortic band pressure gradient measured in banded animals treated with SAHA or vehicle. Invasive hemodynamics showing (B) CO, (C) heart rate (HR), (D) dP/dtmax, (E) dP/dtmin, (F) t-dia/τ, (G) LVEDP, and (H) mPAP (three banded cats added to b + veh) before and after dobutamine infusion. Data shown are means ± SEM. Statistical significance was determined by linear mixed-effects models. *P < 0.05, **P < 0.01, ***P < 0.001 between b + veh and sham. §P < 0.05, §§P < 0.01 between b + SAHA and sham. #P < 0.05, ##P < 0.01 between b + SAHA and b + veh. Dob−/+, before and after dobutamine; dP/dtmax, maximum rate of pressure rise; dP/dtmin, minimum rate of pressure decay; LVEDP, left ventricular end-diastolic pressure; mPAP, mean arterial pulmonary pressure; τ, isovolumic relaxation constant; t-dia, diastolic time interval. The number of animals included in each parameter is reported in table S7.

SAHA improves myofibril relaxation and decreases cardiomyocyte size

To determine whether in vivo diastolic dysfunction in vehicle-treated animals was related to HDAC-dependent alterations in intrinsic cardiomyocyte relaxation, myofibril mechanical studies were performed (20). Samples of myofibrils isolated from homogenized LV tissue were mounted on a force transducer and rapidly shifted from high (to measure Ca2+-activated force) to low Ca2+-containing solution (to measure relaxation kinetics). The maximal calcium-activated tension generation (myofibril maximal tension) was not different between groups (Fig. 3A). However, myofibrils from vehicle-treated banded animals exhibited a significant prolongation in the duration of the linear phase of relaxation when compared with sham (P < 0.01), which was significantly reversed by SAHA (P < 0.05; Fig. 3B). The linear phase of relaxation represents inactivation of thin filament regulatory proteins upon unbinding of Ca2+. Linear relaxation duration significantly correlated with invasively measured t-dia/τ (P = 0.02; Fig. 3C) and LVEDP (P = 0.0061; Fig. 3D) but not with LV wall thickness (P = 0.28; Fig. 3E). These findings show that myofibril-autonomous impairment in banded animals correlates with in vivo diastolic dysfunction. These results also show that HDAC inhibition improved myofibril relaxation independent of the antihypertrophic effects of the treatment.

Fig. 3 SAHA improves myofibril relaxation and exerts antihypertrophic effects.

Ex vivo myofibril mechanical studies on LV tissue collected at 4 months after banding showing (A) myofibril maximal tension and (B) linear relaxation duration. Correlations of ex vivo linear relaxation duration with in vivo (C) t-dia/τ, (D) LVEDP, and (E) LV wall thickness. LV hypertrophy was assessed by (F) HW/BW ratio and (G) LV cardiomyocyte cross-sectional area. (H) Confocal micrographs of LV tissue section stained for WGA (wheat germ agglutinin) (green) and DAPI (4′,6-diamidino-2-phenylindole; blue) show cardiomyocyte size. Scale bars, 50 μm. Assessment of (I) LV fibrosis in subendocardial and subepicardial layers. Quantified plasma concentrations of (J) NT-proBNP. Data shown are means ± SEM. Statistical significance was determined by (A, B, F, G, and I) Kruskal-Wallis tests with multiple comparisons using two-stage linear step-up procedure, (C to E) Pearson correlation, and (J) linear mixed-effects models. *P < 0.05, **P < 0.01, ***P < 0.001 between b + veh and sham. §P < 0.05 between b + SAHA and sham. #P < 0.05 between b + SAHA and b + veh. †P < 0.05 versus subendo b + veh. @P < 0.05 versus subendo b + SAHA. BW, body weight; CM, cardiomyocyte; CSA, cross-sectional area; HW, heart weight; NT-proBNP, N-terminal probrain natriuretic peptide; τ, isovolumic relaxation constant; t-dia, diastolic time interval. The number of animals included in each parameter is reported in table S7.

SAHA-treated animals (3.4 ± 0.1 kg) gained less weight than untreated sham-operated animals (4.2 ± 0.1 kg), and this difference was significant at 4 months (fig. S1C) when compared with vehicle-treated animals (3.8 ± 0.1 kg, P < 0.05). Body weight was lower in the SAHA group, and the heart weight–to–body weight (HW/BW) ratio was also significantly decreased in SAHA-treated compared with vehicle-treated animals (P < 0.05; Fig. 3F). Histological analysis showed that cardiomyocyte cross-sectional area in SAHA-treated animals was significantly smaller than vehicle-treated animals (P < 0.05) but still increased compared with shams (P < 0.05; Fig. 3G). Representative confocal micrographs show LV cardiomyocyte cross sections (Fig. 3H) and demonstrate that decreased myocyte size is responsible for the reduced LV wall thickness in SAHA-treated banded animals shown by Echo analysis.

LV cross sections were stained with Masson’s trichrome, and the aniline blue–stained area (collagen) was measured. Subendocardial and subepicardial layers were analyzed separately. Only the subendocardial layer from vehicle-treated banded animals showed an increase in LV fibrosis (Fig. 3I). SAHA-treated animals had greater fibrotic area than shams and less than vehicle-treated animals, but these differences were not significant in any specific region of the left ventricle wall.

Plasma concentration of NT-proBNP, an HF biomarker, was significantly higher in banded animals at 4 months when compared with shams (P < 0.001). SAHA-treated animals had lower plasma concentrations of NT-proBNP compared with vehicle-treated animals (Fig. 3J).

SAHA improves pulmonary function and preserves lung structure

Pulmonary mechanics were assessed at baseline (BL), 1, 2, 3, and 4 months after banding using pneumotachography in spontaneously breathing animals. Banded animals had significantly lower lung compliance at 2 months after banding compared with sham-operated animals (P < 0.05). Over the next 2 months, lung compliance deteriorated in vehicle-treated animals, but SAHA treatment prevented further stiffening of the lungs, and at the end of the treatment period, there was no difference compared with sham animals (Fig. 4A). Lung compliance measured during ventilation was consistent with the data collected from spontaneously breathing cats (fig. S1D). Gas exchange parameters at 4 months demonstrated that vehicle-treated banded animals had a significant decrease in PaO2/FIO2 (P < 0.01; Fig. 4B), greater alveolar-arterial oxygen difference (A-aDO2) (P < 0.01; Fig. 4C), and greater intrapulmonary shunt fraction (P < 0.01; Fig. 4D) compared with sham and SAHA-treated animals. Except for PaO2, parameters derived from the arterial BGA were comparable between groups (table S2). These findings suggest that SAHA treatment in animals with pressure overload improves pulmonary function and oxygenation. Lung compliance negatively correlated with LV wall thickness and LA size and positively correlated with PaO2/FIO2 (Fig. 4, E to G), suggesting a link between the development of LV-filling defects and impaired lung function.

Fig. 4 SAHA improves lung mechanics, gas exchange, and attenuates lung morphological changes.

Lung function was assessed by measuring (A) lung compliance, (B) PaO2/FIO2 ratio (FIO2 = 1), (C) A-aDO2, and (D) intrapulmonary shunt fraction. Correlation of lung compliance with (E) LV wall thickness, (F) LA size, and (G) PaO2/FIO2. Quantification of (H) alveolar-capillary wall thickness, (I) alveolar area, (J) expansion index (the ratio of volume of gas exchange to parenchymal space), and (K) ratio of cuff/vessel area of extra-alveolar vessels. (L) Bright-field micrographs of H&E-stained lung sections. Scale bars, 250 μm. Data shown are means ± SEM. Statistical significance was determined by (A) linear mixed-effects models, (B to D and H to K) Kruskal-Wallis tests with multiple comparisons using two-stage linear step-up procedure, and (E to G) Pearson correlation. *P < 0.05, **P < 0.01, ***P < 0.001 between b + veh and sham groups. §P < 0.05 between b + SAHA and sham. #P < 0.05, ##P < 0.01 between b + SAHA and b + veh. Open symbols in (E) and (F) indicate baseline data points. A-aDO2, alveolar-arterial oxygen difference; LA/Ao, left atria to aortic root ratio; PaO2, partial pressure of oxygen in arterial blood; FIO2, fraction of inspired oxygen. *, bronchioles; †, extra-alveolar vessel; arrow, fluid cuff around extra-alveolar vessels. The number of animals included in each parameter is reported in table S7.

We have previously shown that slow-progressive pressure overload in this model induces lung remodeling that explains the altered lung function (13). Quantitative histomorphological analyses were performed to determine whether the improved pulmonary function in SAHA-treated animals was associated with prevention or reversal of morphological changes in the lung. SAHA significantly reduced the alveolar-capillary wall thickness (P < 0.05; Fig. 4H) and increased both the alveolar area (P < 0.05; Fig. 4I) and expansion index (ratio of volume of gas exchange to parenchymal space) (P < 0.01; Fig. 4J) compared with vehicle-treated cats. Assessment of perivascular fluid cuff formation (Fig. 4K) around extra-alveolar blood vessels revealed a significant increase in vehicle-treated compared with SAHA-treated (P < 0.05) and sham animals (P < 0.05). No statistically significant differences were observed between SAHA-treated and sham animals in any histomorphological indices. Representative bright-field micrographs of hematoxylin and eosin (H&E)–stained lung sections are displayed in Fig. 4L. These results show that SAHA treatment in this animal model can improve the lung remodeling that underlies abnormal lung function, likely by normalizing LV filling pressures.

SAHA induces a switch in skeletal muscle fiber composition and improves mitochondrial function

O2 consumption was assessed during hemodynamic studies by performing arterial and mixed venous BGA and CO measurements. Vehicle-treated banded animals had lower O2 consumption compared with SAHA-treated banded and sham-operated animals (Fig. 5A). Furthermore, SAHA-treated animals had a significantly lower venous oxygen return (SvO2%, P < 0.01; PvO2, P < 0.01; oxygen content, P < 0.05) compared with vehicle-treated animals, consistent with increased O2 extraction (table S3). The root cause of these changes was further investigated by examining whether changes in metabolic genes could explain the differences in O2 consumption between different groups of animals. In skeletal muscle (gastrocnemius muscle) of SAHA-treated animals, mRNA expression of a transcription factor (Tfam) and coactivators (Pgc-1α, Pgc-1β, P < 0.05) that regulate mitochondrial function, as well as genes involved in glucose metabolism (Glut4, P < 0.05) and the tricarboxylic acid cycle (TCA) cycle (Idh3α, P < 0.01), were significantly increased compared with the vehicle-treated animals. mRNA expression of all genes tested was slightly decreased in vehicle-treated animals compared with sham animals, but these changes did not reach statistical significance (Fig. 5B). Furthermore, SAHA treatment led to a switch in skeletal muscle fiber composition with a significant increase in the percentage of type 1 muscle fibers compared with vehicle-treated animals (P < 0.05) (Fig. 5C). Type 1 fibers have a higher oxidative capacity compared with type 2 fibers and are known to be reduced in patients with HFpEF (24). Type 2 skeletal muscle fibers can be further categorized as type 2a or type 2b. There was no difference in the percentage of type 2a fibers between groups. These findings suggest increased mitochondrial biogenesis and oxidative capacity in skeletal muscle after SAHA treatment.

Fig. 5 SAHA induces switch in skeletal muscle fiber composition and improves mitochondrial function.

In vivo measurements of (A) oxygen consumption. (B) mRNA expression of genes related to mitochondrial biogenesis (Pgc-1α, Pgc-1β, and Tfam), glucose metabolism (Glut4), and TCA cycle (Idh3α) in skeletal muscle. (C) Skeletal muscle fiber composition. Assessment of (D) mitochondrial membrane potential, (E) mitochondrial structure, and (F) mitochondrial calcium uptake before and after histamine administration in HeLa cells. Scale bars, 3 μm. (G) Oxygen consumption rate and extracellular acidification rate were measured using Seahorse Technology: basal (1) and maximal (3) mitochondria respiration, uncoupling (4), and proton leak (2) in HeLa cells. Data shown are means ± SEM. Statistical significance was determined by (A to C) Kruskal-Wallis tests with multiple comparisons using two-stage linear step-up procedure and (D, F, and G) analysis of variance (ANOVA) followed by Bonferroni post hoc testing to compare mean values between groups. *P < 0.05, **P < 0.01 between b + veh and sham groups. §§P < 0.01 between b + SAHA and sham. #P < 0.05, ##P < 0.01 between b + SAHA and b + veh. @P < 0.05 between Control and SAHA. The number of animals included in each parameter is reported in table S7.

The effects of SAHA on mitochondria function and structure were assessed in vitro using HeLa cells. There was no difference in basal or histamine-triggered cytosolic Ca2+ content observed (fig. S1, E and F), indicating that SAHA does not affect the integrity of the endoplasmic reticulum (ER). SAHA induced slight depolarization of the mitochondrial membrane (Fig. 5D) and elongation of the mitochondria structure (Fig. 5E). Basal mitochondrial Ca2+ was reduced with SAHA, but when treated with histamine, the mitochondrial uptake of Ca2+ significantly increased (P < 0.05; Fig. 5F). Furthermore, SAHA treatment led to an elevation in both basal and maximal mitochondria respiration, uncoupling, and proton leak (Fig. 5G). These results support the idea that SAHA directly affects mitochondrial function.

To determine whether the SAHA-induced changes in mitochondrial function occur in cardiac myocytes, experiments were performed using isolated adult feline ventricular cardiomyocytes (AFVM). There was no difference in oxygen consumption rate during basal respiration between SAHA- and vehicle [dimethyl sulfoxide (DMSO)]–treated AFVMs, but SAHA treatment led to a significant increase in the oxygen consumption rate during maximal respiration (P < 0.05; fig. S1G).

Mitochondrial metabolism changes induced by SAHA are associated with posttranslational modifications

To determine whether SAHA treatment alters the concentration of acetyl–coenzyme A (CoA) and protein acetylation patterns, mass spectrometry was performed on LV tissue from SAHA-treated and control sham-operated animals. Sham (not banded) animals were used to determine the impact of SAHA on acetylation independent of any influence from the aortic banding procedure. The concentration of acetyl-CoA was not altered (Fig. 6A) by SAHA treatment, whereas protein acetylation was altered substantially (table S4). We found 51 acetylation sites out of 577 to be altered by SAHA when allowing two miscleavages by trypsin, and another 17 out of 589 acetylation sites altered when allowing three miscleavages by trypsin. Most of the differentially acetylated proteins could also be quantified at the level of total protein by performing label-free quantitative proteomics on the nonacetylated peptides in the flow-through of the immunoprecipitations of acetylated peptides, revealing no significant changes in their abundance (table S5). These findings suggest that SAHA can exert changes in protein function via posttranslational modifications.

Fig. 6 SAHA mediates mitochondrial metabolism via posttranslational modifications.

Assessment of (A) acetyl-CoA concentrations and (B to F) protein acetylation patterns measured using mass spectrometry of LV samples from sham and sham + SAHA animals. Acetylation patterns on (B) histone 3, proteins of the (C) electron transport chain (ETC), proteins of the (D) malate-aspartate shuttle, proteins of the (E) TCA cycle, and proteins involved in (F) mitochondrial fatty acid oxidation. Data shown are means ± SEM. Statistical significance was determined by two-tailed Student’s t test with subsequent multiple testing correction by permutation-based false discovery rate method (P < 0.05). The number of animals included in each parameter is reported in table S7.

The most up-regulated acetylation sites were lysine 24 (>3000-fold), 19 (>3000-fold), 15 (>100-fold), and 10 (>100-fold) on histone 3 (Fig. 6B). These sites are conserved in various species and known to activate genes (25). The different histone H3 variants could not be discriminated because they share an identical N-terminal sequence. Proteins of complexes I, II, and V of the electron transport chain (ETC) were differentially acetylated (Fig. 6C). NDUB3 of complex I could not be quantified in all samples (marked by an asterisk in Fig. 6C), but all other proteins were unchanged in terms of total protein abundance. However, sham + SAHA samples had both hyper- and hypoacetylation of individual ETC components, suggesting that SAHA also indirectly affects protein acetylation in addition to inhibiting lysine deacetylases. Furthermore, components of the malate-aspartate shuttle were also highly regulated by SAHA, with changes in acetylation patterns (Fig. 6D) but not total protein abundance (table S5). We discovered three proteins that were differentially acetylated on six lysines.

We also investigated the effect of SAHA treatment on enzymes involved in the TCA cycle. There was an increase in the acetylation of fumarate hydratase on lysine 292 (about 3000-fold), whereas acetylation of others was slightly down-regulated (Fig. 6E). Moreover, seven sites on five proteins involved in mitochondrial fatty acid oxidation were less acetylated in SAHA-treated hearts (Fig. 6F). Together, many key mitochondrial enzymes had acetylation pattern alterations, which suggests that SAHA affects mitochondrial metabolism via posttranslational modifications.


Using a large-animal model with critical cardiac characteristics (long action potential duration and b-myosin heavy chain as the primary myosin isoform) that are identical to humans and not found in rodents (26, 27), we showed that slow-progressive pressure overload causes the development of dysfunctional cardiopulmonary features also seen in humans with HFpEF. This model was used to test the hypothesis that HDAC inhibition improves/reverses these features. Our study found that the pan-HDAC inhibitor SAHA reversed LVH, reduced LA enlargement, improved LV systolic and diastolic function, and normalized LVEDP. SAHA treatment also restored lung compliance and structure and improved oxygenation. The enhanced LV filling characteristics resulted from improved relaxation and were responsible for lower LV filling pressures. We also demonstrated that HDAC inhibition augmented the relaxation properties of myofibrils isolated from the LVs of pressure-overloaded hearts and that this mechanism likely contributes to improved diastolic function in vivo. The findings from the current study are compared with and contrasted with previously published reports that used HDAC inhibitors in animal models of pressure overload and/or HF in table S6.

The dose of SAHA used was based on previous studies conducted in experimental models of HF (18, 28, 29). McDonnel et al. (28, 30) showed that both intravenous (2 mg/kg) and oral (17 mg/kg) administration of SAHA were well tolerated in felines and produced the predicted pharmacodynamic effect of histone acetylation. They also found that oral bioavailability of SAHA in their animals was generally poor. Therefore, they concluded that an alternative route of administration, such as subcutaneous injection, may provide more optimal drug delivery in these animals. Our experiments documented that this dose of SAHA (10 mg/kg) was well tolerated and did not induce hematological toxicity or QT interval prolongation, which are both potential adverse effects that have been described previously (21, 31). In our study, QTc was shortened in banded animals treated with SAHA. This effect is likely related to the antihypertrophic effects of SAHA and the known relationship between LVH, QTc prolongation, and sudden cardiac death (32). In humans, reported adverse side effects of SAHA treatment include leuko- and thrombocytopenia, bone marrow suppression, and gastrointestinal symptoms (31). We found no evidence for any of these adverse effects in our studies. In addition, efficacious doses of HDAC inhibitors will likely be significantly lower for HF treatment than those required for cancer therapy (16).

We found that SAHA can blunt and reverse LVH caused by pressure overload. Consistent with our findings, pan-HDAC inhibitors in rodent HF models attenuated cardiac hypertrophy induced by infusion of angiotensin II (19) and isoproterenol (33). HDAC inhibition was also shown to reverse established cardiac hypertrophy after transverse aortic constriction in rodent models (19). A combination of chromatin and nonchromatin-based direct effects on protein activity has been proposed as important mechanisms by which HDAC inhibitors counteract pathological LV hypertrophy (16). In cultured myocytes, the transcription factor Krüppel-like factor 4 (KLF4) increased after HDAC inhibition and blocked phenylephrine-induced hypertrophy (16). The antihypertrophic effect observed in the current study was still evident, although the aortic band had not been removed and chronic pressure overload was present throughout the entire study. The notable decrease in LA size in SAHA-treated cats is likely a consequence of the regression of LVH, improvement in diastolic function, and the resulting reduction in LVEDP. A decrease in LV filling pressures leads to reduced pressures within the LA and, thus, reduced atrial remodeling. Reduced LV filling pressures also explain the improvement of the pulmonary phenotype.

Patients with HFpEF have EFs within the normal range (34). In our study, LV EF did not differ between groups at any time point. EF is highly load dependent, making it a measure of LV capacitance rather than contractility, and EF is overestimated in the presence of LVH due to increased myocardial thickening (2). More sensitive imaging techniques, such as speckle-tracking–based strain analysis, are powerful tools for exploring the idea that the HFpEF phenotype is accompanied by early systolic dysfunction despite preserved LV EF (23). In line with these reports, we found a significant impairment in global radial strain before starting SAHA treatment at 2 months after banding. Radial strain values gradually deteriorated in vehicle-treated animals over time, whereas SAHA prevented this deterioration. In addition, the increase in dP/dtmax in response to dobutamine was significantly blunted in vehicle-treated versus sham animals, and SAHA treatment significantly improved the response to dobutamine. Collectively, these findings document that animals with slow-progressive pressure overload exhibit impaired systolic reserve, whereas their EF remained unchanged. An impaired systolic reserve also affects diastolic function because recoil and suction forces during early diastole are attenuated (35). These findings support the idea that SAHA improves systolic contractile reserve in this model.

Previous animal studies have tested pan-HDAC inhibitors in HF models with reduced EF (transverse aortic constriction, ischemia-reperfusion in rodents) (1719). Our study tested HDAC inhibition in a large-animal model exhibiting key clinical characteristics of HFpEF, including diastolic dysfunction and pulmonary impairment confirmed by invasive hemodynamics and state-of-the-art pulmonary function testing. We showed that banded animals exhibited diastolic dysfunction (elevated LVEDP, decreased t-dia/τ ratio) before starting treatment and that SAHA treatment significantly decreased LVEDP and improved isovolumic relaxation. A delay in LV isovolumic relaxation limits LV filling, impairs diastolic suction, and increases end-diastolic pressures because relaxation remains incomplete (36, 37). Increased LV filling pressures are the ultimate downstream expression of impaired LV diastolic function and provide sufficient evidence to secure an HFpEF diagnosis in symptomatic patients (38).

Although exercise intolerance is a hallmark clinical feature of HFpEF, full exercise stress testing was beyond the scope of the current study. As an alternative to controlled exercise, we used pharmacologic stress testing. Dobutamine infusion improved indices of diastolic function, which was expected because ß-adrenoreceptor signaling accelerates relaxation and therefore counteracts acquired diastolic dysfunction (39). These findings are also consistent with a previously published hemodynamic study performed in patients with HFpEF in whom dobutamine infusion decreased LVEDP and τ and increased dP/dtmin, whereas hand-grip exercises increased LVEDP and caused a prolongation of τ (40).

In a recent study, Jeong et al. (20) reported a nongenomic role for HDACs in the direct control of myofibril relaxation in murine models of diastolic dysfunction with preserved EF. The authors showed prolonged kinetics of the linear phase of relaxation in Dahl salt-sensitive (DSS) rats and aged mice. After a 10-week treatment period with the pan-HDAC inhibitor ITF2357, myofibril relaxation was significantly enhanced in DSS rats independent of myofibril composition, calcium sensitivity, titin function, and cardiac fibrosis. Similar findings were obtained in aged mice treated with ITF2357 for 11 months. Furthermore, this study demonstrated abnormal myofibril relaxation in samples collected from patients with restrictive cardiomyopathy, emphasizing the translational significance of this finding. In the present experiments, myofibril relaxation was significantly impaired in banded animals, and this abnormality was reversed by SAHA treatment. Myofibril relaxation correlated with in vivo LVEDP and t-dia/τ, a marker of active relaxation, but not with LV wall thickness, indicating a mechanism of action that is independent of the antihypertrophic effects of SAHA.

LV fibrosis is thought to be involved in HFpEF-related diastolic dysfunction, but neither biopsies from patients with HFpEF nor animal models strongly support this hypothesis (2, 20). SAHA- and vehicle-treated animals had an increase in LV fibrotic area, which was more pronounced in the subendocardium compared with the subepicardium. This gradient could be related to a limited subendocardial coronary reserve (41), but this idea was not explored. Although HDAC inhibitors have been previously reported as having antifibrotic effects (17, 19, 29), we did not observe a significant reduction in fibrosis with SAHA treatment. Therefore, the improved LV diastolic function in the SAHA group was not strongly linked to collagen deposition. Note that this model has a modest fibrotic phenotype, so SAHA treatment in a condition with a more robust presentation of fibrosis could produce an antifibrotic effect with functional significance.

The basis of PH in HFpEF is still not well established, although it is a common and life-threating complication associated with poor prognosis (42). We recently reported pulmonary arterial hypertension in animals with slow-progressive pressure overload due to a passive backward transmission of elevated LV filling pressures (13). In the present study, mPAP was also significantly increased in banded animals compared with shams. SAHA treatment normalized pulmonary arterial pressures, preserved lung structure, and improved lung compliance and function (PaO2/FIO2, A-aDO2, and intrapulmonary shunt). The beneficial effects of SAHA on lung function are best explained as a result of the reduction in LV filling pressures, which subsequently lowers pulmonary pressures, thus preventing pulmonary remodeling and dysfunction. Vehicle-treated animals showed significant perivascular fluid cuff formation, which we previously reported for the first time in HFpEF (13). Perivascular fluid cuffs are suggestive of capillary endothelial disruption and impaired permeability due to elevated hydrostatic pressures (alveolar-capillary stress failure). In a lung injury model, fluid cuffs decreased lung compliance by increasing tissue resistance and impairing mechanical coupling between the lung parenchyma and bronchovascular bundle (43). The effects of SAHA that we observed in the present study are likely the downstream result of reduced pulmonary pressure. Because SAHA has been shown to have a direct beneficial effect on the lungs in rodent models of induced PH (44), it is possible that the improvements observed in our study are partially due to direct effects of SAHA on the lung.

The enhanced mitochondrial oxidative phosphorylation and increased percentage of type 1 skeletal muscle fibers in SAHA-treated animals could underlie the increase in in vivo O2 consumption, decreased venous O2 content, and smaller body weight. Type 1 and 2a skeletal muscle fibers rely on oxidative metabolism as their primary source of energy, whereas type 2b fibers rely on glycolytic metabolism (45). There were no differences in the percentage of type 2a skeletal muscle fibers between groups. In addition, we found increased mRNA expression of genes responsible for mitochondrial biogenesis in skeletal muscle of SAHA-treated animals. SAHA had no effect on the Ca2+ content of the ER and did not induce ER Ca2+ leakage, indicating that SAHA does not affect the integrity of the ER. However, SAHA increased mitochondrial Ca2+ uptake, although the mitochondrial membrane potential and resting Ca2+ were reduced, which indicates that mitochondrial Ca2+ sequestration was not compromised. The underlying mechanism for increased mitochondrial Ca2+ uptake and respiratory activity in the presence of normal intracellular Ca2+ is unclear but could potentially result from enhanced ER-mitochondria tethering or functional changes to mitochondrial Ca2+ uptake machinery (46). SAHA may also affect mitochondrial fission machinery, reflected by elongated mitochondrial structures. Our in vitro results show that SAHA treatment can directly modify mitochondria function, suggesting that modification of mitochondria function by SAHA could explain aspects of our in vivo findings.

Using mass spectrometry, we found that SAHA-induced changes in mitochondrial metabolism correlated with associated posttranslational modifications. Succinate dehydrogenase A (SDHA), which is a subunit of complex II in the ETC, was three times less acetylated on lysine 547 in SAHA-treated sham animals. This lysine residue is conserved in both mice and humans, and hyperacetylation of this site has previously been shown to negatively affect complex II activity (47). Thus, our finding of decreased acetylation at lysine 547 may lead to increased complex II activity. Consistent with our finding, Herr et al. (48) reported that selective HDAC1 inhibition led to a decrease in SDHA activity and subsequent reactive oxygen species (ROS) production in an ex vivo model of ischemia-reperfusion. Similarly, acetylation of lysine 259 of adenosine triphosphate (ATP) synthase subunit β (ATPB), the catalytic subunit of complex V, was reduced by two-thirds. This site is conserved in both mammals and flies, and hyperacetylation has been shown to inhibit ATP synthase activity (49). Thus, ATP synthase activity may be modulated by SAHA treatment.

Hyperacetylation of mitochondrial aspartate aminotransferase (AATM) and malate dehydrogenase (MDHM) has been described to improve mutual interaction and activity between AATM and MDHM, resulting in increased transportation of NADH (reduced form of nicotinamide adenine dinucleotide) into mitochondria and enhanced ATP synthesis (50, 51). We found that acetylation of AATM was about 2000-fold higher on lysine 363 and about 500-fold higher on lysine 234 in SAHA-treated sham animals. Acetylation was increased about 2000-fold on lysine 301 of MDHM. In contrast, acetylation of lysine 345 of AATM and lysine 73 of cytosolic malate dehydrogenase (MDHC) were reduced about 50 and 40%, respectively. AATM and MDHM lysine acetylation is altered by SAHA, but the effect of these changes on enzyme interaction and activity was beyond the scope of the study.

Consistent with our findings, Galmozzi et al. (52) reported that HDAC inhibition increased O2 consumption and oxidative capacity in skeletal muscle in a mouse model of obesity and diabetes. HFpEF is associated with impaired mitochondrial oxidative capacity, rapid depletion of high-energy phosphate, increased formation of reactive oxygen species, and a reduction in the number of type 1 skeletal muscle fibers, leading to an imbalance in energy supply and demand. The observed effects of SAHA on mitochondrial function may help restore energy supply in HF, which is in contrast to current HF drugs that aim to reduce energy demand to match the reduced supply (24, 53).

Our study had some limitations. Invasive hemodynamics were performed in anesthetized cats in the supine position, which is an unusual posture for quadrupeds, and pharmacological stress testing was performed using dobutamine, which does not replicate physical exercise. Because of these factors, the results must be interpreted with caution. Note that the pericardial sack was dissected during both the sham and aortic constriction procedures, which could have potentially resulted in an underestimation of the left-sided filling pressures (54). Furthermore, we used only male animals for this study and did not address sex-related differences. Right ventricular function is known to be impaired in HFpEF but was not assessed is this study. Echo-based imaging parameters (tricuspid annular plane systolic excursion, strain) that are used to assess right ventricular function in patients are not well established and validated in cats. Only parasternal short-axis strain analysis is reported because acquiring the views for apical four-chamber and parasternal long-axis views is inconsistent in cats, making it an unreliable parameter to assess subtle differences between treatment groups. The animals were young and did not have any comorbidities or other disease background, which is in contrast with to the clinical presentation of many patients with HFpEF. It is likely that in contrast to young animals, aged animals would have shown impairments in cardiovascular reserve. No single animal model is likely to recapitulate the complex phenotype of human HFpEF, but the large-animal model used in the present study could serve as a useful platform for testing therapies for select populations of patients. Acetylation experiments were performed using only LV tissue from sham animals treated with SAHA to confirm that SAHA treatment induced the predicted pharmacodynamic effect of protein and histone acetylation without any other confounding influence from the aortic banding procedure. Performing the acetylation studies on other groups (b + SAHA or b + veh) was beyond the scope of the current study.

In summary, this study shows the efficacy of an FDA-approved HDAC inhibitor in a preclinical large-animal model of diastolic dysfunction. We demonstrated that SAHA exerts robust antihypertrophic effects, improves diastolic function, reduces LVEDP, and ultimately improves lung compliance and pulmonary function. In addition, we showed that impaired myofibril relaxation is an HDAC-dependent mechanism for diastolic dysfunction, which can be restored by SAHA. Furthermore, SAHA treatment induced metabolic effects that could restore energy supply in the failing heart. Given the promising preclinical results and the unmet need for effective therapeutics to treat the growing HFpEF population, our results suggest that SAHA treatment could potentially improve cardiopulmonary dysfunction in patients with features of the model system we used.


Study design

The aims of this study were to assess whether HDAC inhibition could improve cardiopulmonary function and preserve cardiopulmonary structure in a feline model of diastolic dysfunction. State-of-the-art invasive hemodynamics, echocardiography, immunohistochemistry, histology, tissue proteomics, mitochondrial function, and myofibril mechanic studies were performed. All animals were randomly assigned to treatment arms, and sample size was determined by power calculation (α = 0.05, β = 0.8, effect size of 1.5 SD) to assess the antihypertrophic effects of HDAC inhibition. All analyses were performed by investigators blinded to treatment or banding. All samples were given a code that was not linked to study group until analyses were completed. The number of animals included in each experimental parameter is listed in table S7. Primers for polymerase chain reaction (PCR) analysis are listed in table S8. Please refer to the Supplementary Materials and Methods for a detailed description of experimental methods and proteomic data (tables S4 and S5). Individual subject level data are reported in data file S1.

Statistical analysis

For echocardiography and pulmonary function test parameters with repeated measures, linear mixed-effects models were used to determine predicted mean values at each assessment point (BL, 1, 2, 3, and 4 months) and test group (sham, b + veh, b + SAHA) differences. In each linear mixed-effects model, time and treatment group were included as fixed effects. Similarly, linear-mixed effect models were used to estimate and compare before (dob−) and after (dob+) intravenous infusion and treatment differences for hemodynamics. Pearson correlation coefficients were calculated between two variables such as t-dia/τ and linear relaxation duration. Kruskal-Wallis tests with multiple comparisons using two-stage linear step-up procedure were used to evaluate the differences for all other parameters. For the in vitro mitochondrial experiments, each “n” value refers to the number of individual experiments performed in triplicate. Analysis of variance (ANOVA) with post hoc analysis was used when comparing more than two groups, and statistical significance of differences between means was adjusted by Bonferroni correction. Two-sample Student’s t test was used to compare means of two independent samples. Statistical analyses were performed using SAS 9.4 (SAS Institute) and GraphPad Prism 7.04. Values are expressed as mean ± SEM. Two-sided testing was used for all statistical tests. A P value of ≤0.05 was used to determine significance for all statistical tests.


Materials and Methods

Fig. S1. SAHA treatment in a feline model of diastolic dysfunction.

Table S1. Assessment of blood count.

Table S2. Arterial BGA.

Table S3. Venous BGA.

Table S4. Protein acetylation (separate file).

Table S5. Quantitative proteomics (separate file).

Table S6. Findings from the current study versus those from previously published papers.

Table S7. Animal numbers.

Table S8. Primers for PCR analysis.

Data file S1. Individual subject-level data (separate file).

References (5683)


Funding: This work was supported by the National Institutes of Health (HL147558 to S.R.H. and T.A.M., HL33921 to S.R.H., HL116848 and HL127240 to T.A.M., and R56-HL137850 and RO1-HL137850 to S.M.); Department of Defense/Office of Naval Research (VA—N000141210810 to M.R.W); American Heart Association (16POST30960017 to Y.H.L., 16SFRN31400013 to T.A.M., and SDG-15SDG25550038 to S.M.); Colorado Clinical and Translational Science Institute (5KL2TR001080-02 to M.Y.J.); Medical University of Graz–Start Funding Program (M.W.) and Stadt Graz (M.W.); and Austrian Science Fund (KLI425, KLI645, W1226, and F73 to R.B.-G.). Author contributions: S.R.H., M.W., M.R.W., T.A.M., and D.v.L. conceived and designed the research. M.W., D.M.E., R.M.B., J.W., M.Y.J., Y.H.L., G.B., S.T.B., L.L., M.S., J.G., H.B., S.B., W.F.G., J.P., and R.B.-G. acquired the data. H.Z. performed statistical analyses. M.W. and D.M.E. drafted the manuscript. S.R.H., M.R.W., T.A.M., D.v.L., S.M., P.P.R., A.Z., W.F.G., R.B.-G., and M.W. made critical revision of the manuscript for key intellectual content. All authors discussed the results and commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data associated with this study are in the paper or the Supplementary Materials. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium ( via the PRIDE partner repository (55) with the dataset identifier PXD013663.

Stay Connected to Science Translational Medicine

Navigate This Article