HDAC Inhibition Reverses Preexisting Diastolic Dysfunction and Blocks Covert Extracellular Matrix Remodeling

Abstract

Background: Diastolic dysfunction (DD) is associated with the development of heart failure and contributes to the pathogenesis of other cardiac maladies, including atrial fibrillation. Inhibition of histone deacetylases (HDACs) has been shown to prevent DD by enhancing myofibril relaxation. We addressed the therapeutic potential of HDAC inhibition in a model of established DD with preserved ejection fraction.

Methods: Four weeks after uninephrectomy and implantation with deoxycorticosterone acetate pellets, when DD was clearly evident, 1 cohort of mice was administered the clinical-stage HDAC inhibitor ITF2357/Givinostat. Echocardiography, blood pressure measurements, and end point invasive hemodynamic analyses were performed. Myofibril mechanics and intact cardiomyocyte relaxation were assessed ex vivo. Cardiac fibrosis was evaluated by picrosirius red staining and second harmonic generation microscopy of left ventricle (LV) sections, RNA sequencing of LV mRNA, mass spectrometry-based evaluation of decellularized LV biopsies, and atomic force microscopy determination of LV stiffness. Mechanistic studies were performed with primary rat and human cardiac fibroblasts.

Results: HDAC inhibition normalized DD without lowering blood pressure in this model of systemic hypertension. In contrast to previous models, myofibril relaxation was unimpaired in uninephrectomy/deoxycorticosterone acetate mice. Furthermore, cardiac fibrosis was not evident in any mouse cohort on the basis of picrosirius red staining or second harmonic generation microscopy. However, mass spectrometry revealed induction in the expression of >100 extracellular matrix proteins in LVs of uninephrectomy/deoxycorticosterone acetate mice, which correlated with profound tissue stiffening based on atomic force microscopy. ITF2357/Givinostat treatment blocked extracellular matrix expansion and LV stiffening. The HDAC inhibitor was subsequently shown to suppress cardiac fibroblast activation, at least in part, by blunting recruitment of the profibrotic chromatin reader protein BRD4 (bromodomain-containing protein 4) to key gene regulatory elements.

Conclusions: These findings demonstrate the potential of HDAC inhibition as a therapeutic intervention to reverse existing DD and establish blockade of extracellular matrix remodeling as a second mechanism by which HDAC inhibitors improve ventricular filling. Our data reveal the existence of pathophysiologically relevant covert or hidden cardiac fibrosis that is below the limit of detection of histochemical stains such as picrosirius red, highlighting the need to evaluate fibrosis of the heart using diverse methodologies.

Keywords: extracellular matrix; fibroblasts; fibrosis; histone deacetylases; mass spectrometry; microscopy, atomic force.

Figure 1.. HDAC inhibition reverses preexisting diastolic dysfunction.

A, Schematic representation of the in vivo efficacy study. B, Change in mouse body weight from baseline to the conclusion of the 8-week study. Serial echocardiographic assessment of C mitral inflow velocities (E/A), D septal mitral annulus velocities (E’/A’), and E ratios of mitral flow to annulus velocities (E/E’). F, Isovolumic relaxation time (IVRT). G, Representative echocardiographic images from a single mouse (mouse #925) illustrating the development of Stage I diastolic dysfunction (DD) 4 weeks post-UNX/DOCA, and normalization of diastolic function following ITF2357/Givinostat feeding. H, Representative echocardiographic images from a single mouse (mouse # 918) illustrating the development of Stage II/III DD with a restrictive filling pattern 4 weeks post-UNX/DOCA, and normalization of diastolic function after ITF2357/Givinostat feeding. I, Invasive left ventricular end diastolic pressure (LVEDP) measurements obtained at the conclusion of study. J, Tail cuff measurements of mean systemic pressure (MSP) demonstrating sustained hypertension in UNX/DOCA mice fed normal chow or chow containing ITF2357/Givinostat. K, LV weight-to-tibia length measurements of cardiac hypertrophy. L, Systolic function, as measured by ejection fraction (EF), was preserved in each mouse cohort throughout the experiment. For B, I and K, the data are presented as mean +SEM, with statistical analysis performed using one-way ANOVA with Tukey’s multiple comparisons test. For C through F, J and L, the data are presented as the mean ± SEM, with statistical analysis performed using a mixed-effects model with Tukey’s multiple comparisons test. *P<0.05 vs. UNX sham control and ^#P<0.05 vs. UNX/DOCA fed normal chow at a given time point. The number of animals analyzed by echocardiography at each time point is provided in Table II in the Supplement.

Figure 2.. No evidence of cardiac fibrosis in UNX/DOCA mice based on histological analysis.

A, Representative left ventricular (LV) cross-sections (post-8 weeks of UNX or UNX/DOCA) stained with picrosirius red (PSR) dye (scale bar = 0.5mm). B, Quantification of PSR signals failed to demonstrate changes in fibrillar collagens in the LVs of any mouse cohort. C, Representative 20x magnification images of the PSR-stained LVs shown in A; (top) bright-field microscopy and (bottom) polarized light microscopy of PSR birefringence (scale bar = 50 μm). D, Second harmonic generation (SHG) microscopy of LV cross-sections (scale bar = 50 μm). E, Quantification of the SHG signal across groups. F and G, Quantitative RT-PCR analysis of the indicated fibrosis-related mRNAs in whole LV homogenates. For B and E through G, data represent the mean +SEM. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test.

Figure 2.. No evidence of cardiac fibrosis in UNX/DOCA mice based on histological analysis.

A, Representative left ventricular (LV) cross-sections (post-8 weeks of UNX or UNX/DOCA) stained with picrosirius red (PSR) dye (scale bar = 0.5mm). B, Quantification of PSR signals failed to demonstrate changes in fibrillar collagens in the LVs of any mouse cohort. C, Representative 20x magnification images of the PSR-stained LVs shown in A; (top) bright-field microscopy and (bottom) polarized light microscopy of PSR birefringence (scale bar = 50 μm). D, Second harmonic generation (SHG) microscopy of LV cross-sections (scale bar = 50 μm). E, Quantification of the SHG signal across groups. F and G, Quantitative RT-PCR analysis of the indicated fibrosis-related mRNAs in whole LV homogenates. For B and E through G, data represent the mean +SEM. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test.

Figure 3.. ECM mass spectrometry uncovers ‘hidden’ cardiac fibrosis that is blocked by ITF2357/Givinostat.

A, Decellularization of left ventricular (LV) biopsy tissue yields an extracellular matrix (ECM) enriched fraction that is subsequently cleaved into peptides prior to analysis by mass spectrometry. B, Concentration of ECM-derived peptides from the indicated groups of mice, 8 weeks post-UNX or UNX/DOCA, after enzymatic digestion and prior to mass spectrometry; data represent the mean +SEM. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test. *P<0.05 versus UNX Sham; ^#P<0.05 versus UNX/DOCA + Normal Chow. C, Heat map of differentially expressed ECM proteins. Each column represents data from a single mouse LV, and each row represents an individual ECM protein, with collagens highlighted in red. The color scale bar represents relative expression of log-transformed, normalized expression with upregulated proteins shown in red and downregulated proteins in blue. D, Principal component analysis (PCA) reveals the variance in ECM protein distribution. E, Volcano plots show magnitude and significance of ECM proteins altered in LVs of UNX/DOCA + Normal Chow versus UNX + Normal Chow (left) and UNX/DOCA + ITF Chow versus UNX/DOCA + Normal Chow (right), with collagens highlighted. F, Volcano plot showing no significant changes in ECM protein expression in LVs of UNX/DOCA + ITF Chow versus UNX sham + Normal Chow. G, Tandem Mass Tag (TMT) intensity of collagen type I alpha 1 expression. Box: interquartile range; whiskers: min to max, with 4 individual animals per condition. The mass spectrometry data are deposited via PRIDE, as indicated in the online-only Supplemental Materials.

Figure 4.. Atomic force microscopy reveals LV stiffening that is blocked by ITF2357/Givinostat.

A, Force-distance curves were obtained by measuring cantilever deflection over controlled deformation at each pixel by QI™ mode. Left ventricular (LV) sections were analyzed in a liquid environment. Once the tip was in contact with the sample, forces between the tip and the sample led to a deflection of the cantilever, which was then monitored during the sample scan; this generated a force-distance curve at each scan pixel. B, Young’s Modulus heat map (log kPa) showing the stiffness distribution across groups (3 independent animals per group). Red-orange colors represent highest Young’s Modulus, and yellow colors represent the lowest Young’s Modulus, in kPa. C, Relative stiffness at the nanoscale of each group. Data represent the mean +SEM. Statistical analysis was performed using one-way ANOVA with Tukey’s multiple comparisons test. *P<0.05 versus UNX Sham; ^#P<0.05 versus UNX/DOCA + Normal Chow. D, Stiffness maps in kPa (upper panel) and their corresponding topography maps in μm (lower panel) of representative LV section measurements. Upper panel: light yellow colors represent highest Young’s Modulus in kPa; dark red colors represent the lowest Young’s Modulus. Lower panel: Lighter sections represent the highest Z-values in μm; darker sections represent the lowest Z-values. Scan area: 10 x 10 μm. Pixel resolution: 256 x 256.

Figure 5.. A transcriptomic signature of cardiac fibroblast activation in LVs of DOCA/UNX mice is attenuated by ITF2357/Givinostat.

A, Heat map of genes differentially expressed in the hearts UNX Sham, UNX/DOCA + Normal Chow or UNX/DOCA + ITF Chow animals 8 weeks after DOCA pellet implantation. Each column represents data from an individual animal; there are 4 replicates in the UNX Sham and UNX/DOCA + ITF Chow groups, and 3 replicates in the UNX/DOCA + Normal Chow group. Each row represents an individual gene, and the color bar represents relative expression of log-transformed, normalized counts with upregulated genes shown in red and downregulated genes in blue. B, Volcano plots of normalized enrichment scores of gene sets versus the -log10(P-value) for all gene set pathways in each pairwise comparison. The gene sets of interest are highlighted to visualize their presence in each group and their location based on an FDR cutoff value of 0.05 (dashed line). C, Boxplots represent Log₂ fold change of expression levels for all enriched genes from the Lian Lipa Targets 3M, Jechlinger EMT Up and Poola Invasive Up gene sets across pairwise comparisons. Total genes on average are more similar but core enriched genes are depicted by the subtle differences between the boxplot groups as enriched genes may be a smaller fraction of the total genes that are involved in each gene set. D, DOCA. *P < 0.01; **P < 0.001; ***P < 0.0001.

Figure 6.. HDAC inhibition suppresses TGF-β-induced cardiac fibroblast activation.

A, Adult rat ventricular fibroblasts (ARVFs) were cultured in low-serum medium in the absence or presence of ITF2357 (500 nM) and/or TGF-β (10 ng/ml) for 48 hours. Indirect immunofluorescence staining of smooth muscle α-actin (α-SMA; upper panel) and cellular fibronectin (lower panel) in ARVFs co-stained with DAPI to label nuclei. Scale bars = 50 μm. B, α-SMA protein expression in ARVFs, cultured as described in A, was determined by immunoblotting; α-Tubulin (α-Tub) served as a loading control. Each lane represents protein from an independent plate of cells. C, ARVF were cultured as described in A, and Acta2 (α-SMA) and Fn1 (Fibronectin-1) transcript levels were assessed by qPCR. D, Representative images of contraction of collagen gels on which ARVFs were seeded for 72 hours after detachment; gels were detached at the time of TGF-β stimulation. E, Percent contraction was measured at 24 hour intervals; n = 3 gels per condition. F, Normal human ventricular fibroblasts (NHVFs) were cultured in low-serum medium in the absence or presence of ITF2357 (500 nM) and/or TGF-β (10 ng/ml) for 48 hours. Indirect immunofluorescence staining of α-SMA in NHVFs co-stained with DAPI to label nuclei; scale bars = 50 μm. G, NHVFs were cultured as described in F, and ACTA2 (α-SMA) transcript expression was assessed by qPCR. Statistical analysis in C and G utilized two-way ANOVA with Tukey’s multiple comparisons test; *P<0.05 versus vehicle only (−/−); ^#P<0.05 versus TGF-β alone, comparing only the last two bars. For E, two-way repeated measures ANOVA and Tukey’s post-hoc test was employed. *P<0.05 versus vehicle; ^#P<0.05 versus TGF-β alone.

Figure 7.. HDAC inhibition blunts TGF-β-mediated recruitment of BRD4 to pro-fibrotic promoter and enhancer elements.

A, A schematic of the rat Sertad4 locus in cardiac fibroblasts depicting previously identified BRD4 peaks at the promoter and six distinct enhancers (E1-E6); E1 and E2 are defined as super-enhancers (SEs) based on the high level of BRD4 binding to these regions. B, Adult rat ventricular fibroblasts (ARVFs) were cultured in low-serum medium in the absence or presence of ITF2357 (500 nM) and/or TGF-β (10 ng/ml) for 48 hours. Sertad4 mRNA expression was determined by qPCR. C, Schematic representation of the chromatin immunoprecipitation (ChIP)-qPCR experiment. BRD4 binding to the Sertad4 gene regulatory elements D Super enhancer 1, E Super enhancer 2, F Enhancer 3, G Enhancer 4, H Enhancer 5, I Enhancer 6, and J Promoter, as determined by ChIP-qPCR analysis. For B and D through J, all data represent the mean +SEM, and statistical analysis was performed with two-way ANOVA with Tukey post-hoc analysis. *P<0.05 versus vehicle; ^#P<0.05 versus TGF-β alone. K, ARVFs were cultured as described in B, and BRD4 was detected by indirect immunofluorescence. Cells were co-stained with CellMask and DAPI. Representative confocal microscopy images are shown; scale bars = 20 μm. L, ARVFs were cultured as described in B, and BRD4 protein expression was determined by immunoblotting; α-Tubulin (α-Tub) served as a loading control.

Figure 8.. A model for HDAC inhibitor-mediated inhibition of cardiac fibroblast activation and reversal of established diastolic dysfunction.

In activated cardiac fibroblasts, BRD4 stimulates pro-fibrotic gene expression through association with specific enhancer and promoter elements, leading to fibrosis and diastolic dysfunction. HDAC inhibition prevents recruitment of BRD4 to these elements, presumably through the creation of spurious acetyl-histone marks throughout the cardiac fibroblast genome. This blunts pro-fibrotic gene expression and fibrosis, and leads to improved diastolic function.