SIRT6 Mitigates Heart Failure With Preserved Ejection Fraction in Diabetes

Abstract

RATIONALE:

Heart failure (HF) with preserved ejection fraction (HFpEF) is a growing health problem without effective therapies. Epidemiological studies indicate that diabetes is a strong risk factor for HFpEF, and about 45% of patients with HFpEF are suffering from diabetes. However, the molecular mechanisms for the association of diabetes and HFpEF are poorly understood.

OBJECTIVE:

This study was designed to explore whether the longevity gene sirtuin 6 (SIRT6) could regulate endothelial fatty acid (FA) transport and ameliorate HFpEF in diabetes.

METHODS AND RESULTS:

We first observed that endothelial SIRT6 expression was markedly diminished in cardiac tissues from HF patients with diabetes and diabetic mice with the myriad symptoms of HFpEF, which was induced by the combination of the long-term high-fat diet and one low-dose streptozocin challenge. We then generated a unique humanized SIRT6 transgenic mouse model, in which a single copy of human SIRT6 transgene was engineered at mouse Rosa26 locus and conditionally induced with the Cre-loxP technology. We found that genetically restoring endothelial SIRT6 expression in the diabetic mice ameliorated diastolic dysfunction concurrently with decreased cardiac lipid accumulation. SIRT6 gain- or loss-of-function studies showed that SIRT6 negatively regulated endothelial FA uptake. Mechanistically, SIRT6 suppressed endothelial expression of PPARγ through SIRT6-dependent deacetylation of H3K9 around PPARγ promoter region; and PPARγ downregulation mediated SIRT6-dependent inhibition of endothelial FA uptake. Importantly, oral administration of small molecule SIRT6 activator MDL-800 to diabetic mice mitigated cardiac lipid accumulation and diastolic dysfunction.

CONCLUSIONS:

The impairment of endothelial SIRT6 expression links diabetes to HFpEF through altering FA transport across the endothelial barrier. The genetic and pharmacological strategies that restore endothelial SIRT6 function in diabetes alleviated experimental HFpEF by limiting FA overtake and improving cardiac metabolism, thus warranting further clinical evaluation.

Graphical Abstract:

INTRODUCTION

Heart failure (HF) with preserved ejection fraction (HFpEF) is a growing public health problem and accounts for more than 50% of all HF cases^{1, 2}. HFpEF is a heterogenous syndrome and is commonly associated with a variety of comorbidities including diabetes mellitus (DM), obesity, and hypertension^2–4. Epidemiological studies reveal that DM is a strong risk factor for HFpEF^5–8. About 45% of HFpEF patients suffer from DM^{5, 9}, and the prevalence of HFpEF in diabetic patients is rising in parallel with the global occurrence of DM¹⁰. However, the molecular mechanisms linking diabetes to HFpEF are poorly understood, and no effective therapy for HFpEF is currently available.

HFpEF is characterized by left ventricular diastolic dysfunction and metabolic derangements^{5, 8}. Specifically, due to insulin resistance and other metabolic abnormalities under diabetic conditions, the heart shifts away from utilization of both glucose and fatty acid (FAs) to almost complete dependence on FAs as an energy source¹¹. Once taken up into cardiomyocytes, FAs can be transported into the mitochondria for oxidation or esterified to form triglycerides and stored in the form of lipid droplets (LDs)¹². Excessive LDs accumulation in the cytosol of cardiomyocytes causes structural and functional disturbances in the heart, resulting in cardiac lipotoxicity and finally HF^12–14. Further understanding of signaling pathways governing FA transport and metabolic remodeling in the heart under diabetic conditions may help to discover new therapeutic targets for diabetes-associated HFpEF.

Endothelium, a cell layer lining blood vessels, is an independent organ that functions as a barrier for nutrient shuttling^{15, 16}. The neglected role of the endothelium in controlling metabolic homeostasis is beginning to evolve¹⁷. Recent studies show that the FA transport across the endothelium can be modulated by CD36¹⁸, VEGF-B¹⁹, and the Notch signaling pathway²⁰. These studies suggest that proper control of nutrient transport across the endothelium is crucial for the normal function of the organs. However, the molecular mechanisms by which the endothelial cells (ECs) govern FA transport and cardiac function remain largely unknown.

Sirtuin 6 (SIRT6), a well-recognized longevity gene, regulates genome stabilization, DNA repair, inflammation, and metabolic homeostasis^{21, 22}. In this study, we reported that insufficiency of SIRT6 in the endothelium linked diabetes to HFpEF. Furthermore, we showed that both genetically and pharmacologically restoring endothelial SIRT6 activation ameliorated diastolic dysfunction and exercise intolerance in a preclinical mouse model of diabetes-associated HFpEF by limiting the FA transport across the endothelium into the myocardium.

METHODS

Data Availability

All data that support the findings of this study are available in the article and the Data Supplement. Detailed Materials and Methods are provided in the Data Supplement. Please see the Major Resources Table in the Data Supplement. The RNA-seq data have been deposited at the NCBI Gene Expression Omnibus (GEO) repository and will be freely accessible at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE213425. Additional technical information is available from the corresponding author upon reasonable request.

RESULTS

Endothelial SIRT6 Expression Is Reduced in the Tissues of Diabetic HF Patients and the Diabetes-associated HFpEF Mice

To investigate the relevance of endothelial SIRT6 in diabetic HF, we first conducted data mining in the publicly available Gene Expression Omnibus (GEO) database and analyzed SIRT6 expression in human aortic endothelial cells (HAECs) isolated from diabetes patients and healthy subjects. There was a decrease in endothelial SIRT6 in HAECs in diabetic patients compared to healthy human subjects (Figure S1). We next examined mRNA expression of SIRT6 and other sirtuin family members in the cardiac tissues from HF patients with diabetes undergoing ventricular assist device implantation^{23, 24}. There was a significant decrease in the mRNA level of SIRT6 and other sirtuin family members in the cardiac tissues of the diabetic patients compared with healthy controls (Figure S2A). Furthermore, SIRT6 protein expression was significantly decreased in the heart tissues of diabetic HF patients (Figure 1A). The co-immunostaining assay revealed that expression of endothelial SIRT6, but not endothelial SIRT1, in cardiac tissues from HF patients with DM was greatly reduced (Figure 1B and 1C; Figure S2B and S2C). Thus, we focused on investigating the potential role of endothelial SIRT6 in the regulation of diabetes-associated cardiac dysfunction.

Figure 1. Endothelial SIRT6 expression was reduced in the vasculature of humans and mice under diabetic conditions.

A, Representative immunoblotting image and quantification of SIRT6 protein level in the cardiac tissues from diabetic patients with heart failure (n = 6) and healthy controls (n = 3). B, Immunofluorescence staining SIRT6 expression (red) and EC marker CD31 (cyan) on cardiac specimens from diabetic patients with heart failure and healthy controls. Scale bars, 30 μm. C, Quantification of SIRT6 positive endothelial cells per mm² of area based on Immunostaining in B using Image J software. n = 4 patients. D, Percentage of left ventricular ejection fraction (LVEF). E, Percentage of left ventricular fraction shortening (LVFS). F, Ratio between mitral E wave and A wave (E/A). G, Ratio between mitral E wave and E′ wave (E/E′). H, Isovolumic relaxation time (IVRT). I, Ratio between wet and dry lung weight. J, Running distance during exercise exhaustion test. n = 8 mice per group in D through J. K, Representative images of general view, hematoxylin and eosin (H&E), and Masson’s Trichrome staining of the left ventricle section. Scale bars, 1000 μm (general view), 600 μm (H&E), and 100 μm (Masson). L, SIRT6 mRNA expression in the fresh isolated aortic intima tested by real-time PCR. GAPDH is used as a loading control for mRNA measurement. n = 6 mice per group. M, Immunofluorescence staining of SIRT6 and CD31 on cross sections of the cardiac tissues from mice under HFD/STZ insult and mice under normal diet. Scale bars, 30 μm. N, Quantification of SIRT6 positive endothelial cells per mm² of area based on Immunostaining in M. n = 4 repeated experiments. O, mRNA level of SIRT6 in HUVECs treated with high glucose (HG) or normal glucose (NG) tested by real-time PCR. P, Representative image and quantification of SIRT6 protein in HUVECs treated with HG or NG for 24 h tested by Immunoblotting assay. Q, mRNA level of SIRT6 in HUVECs treated with palmitic acid (PA) or vehicle control (CTL) tested by real-time PCR. R, Representative image and quantification of SIRT6 protein in HUVECs treated with PA or CTL tested by Immunoblotting assay. n = 5 repeated experiments in P and R; n = 6 repeated experiments in O and Q. Statistical analysis was performed by unpaired Student’s t-test (I, J, L, O, Q) and non-parametric unpaired Mann-Whitney test (A, C, N, P, R), two-way ANOVA plus Bonferroni’s multiple comparisons test (D through H). ns, no significance.

Next, we examined the responsiveness of endothelial SIRT6 to diabetes-associated experimental HFpEF in mice. We established and characterized a diabetic mouse model by feeding C57BL/6 mice with a high-fat diet (HFD, 60 kcal %) combined with one bolus of low-dose streptozocin (STZ, 35 mg·kg⁻¹)^{25, 26} for sequential observation up to 28 weeks (Figure S3A). The HFD/STZ challenge induced diabetes in mice (Figure S3B and S3C). The echocardiographic evaluation revealed persistent preservation of the left ventricular ejection fraction (LVEF) and left ventricular fractional shortening (LVFS) in both diabetic and non-diabetic mice, which was confirmed by a time course observation of cardiac function until 28 weeks of the HFD/STZ treatment (Figure 1D and 1E and Figure S3D). However, HFD/STZ-induced diabetic mice displayed an obvious diastolic functional decline, manifested by several indicators of diastolic dysfunction including the decrease in E/A ratio, an increase in the E to E’ ratio, and extended isovolumic relaxation time (IVRT) (Figure 1F through 1H, and Figure S3E). Furthermore, HFD/STZ-induced diabetic mice exhibited a robust increase in lung weight (Figure 1I), exercise intolerance (Figure 1J), and cardiac fibrosis (Figure 1K). All those characteristics observed in HFD/STZ-induced diabetic mice are consistent with the symptoms reported in HFpEF patients²⁷. Thus, the HFD/STZ-challenged diabetic mice may represent a useful model of experimental HFpEF, especially to recapitulate the clinical phenotypes of HFpEF in DM. Consistent with the results observed in cardiac samples from the DM patients with HF, SIRT6 mRNA expression was significantly decreased in the freshly isolated aortic intima (the endothelium layer) in HFD/STZ-induced diabetic mice (Figure 1L). Notably, while SIRT6 was present in both ECs and cardiomyocytes (CMs) in the mouse hearts, the immunoreactivity of SIRT6 co-localized with CD31 was decreased in the cardiac tissues from HFD/STZ-induced diabetic mice (Figure 1M and 1N).

Moreover, we examined the changes in SIRT6 expression in cultured human ECs with various stimuli implicated in diabetes. We observed a significant reduction of SIRT6 expression in the cultured human umbilical vein ECs (HUVECs) under the stimulation of high glucose or palmitic acid (PA, common saturated FA found in the human body) (Figure 1O through 1R). Collectively, these results from in vivo and in vitro studies from humans to mice indicate that a reduction of endothelial SIRT6 is a prominent phenomenon under diabetic conditions, which might link diabetes to cardiac dysfunction.

Generation and Validation of An Endothelial-Specific Humanized SIRT6 Overexpression Transgenic Mouse Model

Given that endothelial SIRT6 was reduced under diabetic conditions, we next investigated whether restoring endothelial SIRT6 expression in diabetes could ameliorate diabetic cardiac dysfunction. To this end, a human Sirt6-TG^fl/fl mouse line was genetically engineered by inserting a single copy of the human SIRT6 gene into the mouse Rosa26 locus under chicken beta-actin promoter (CAG) promoter in a conditional inducible manner with the Cre/Loxp technology (Figure 2A). Then we crossed the Sirt6-TG^fl/fl mice with Cdh5-Cre mice²⁸ to generate endothelial-specific humanized SIRT6 overexpression (Sirt6^ecTG) mice. All the mice were fertile and appeared normal. Genotyping PCR assays detected Sirt6-TG^fl/fl and wild-type alleles from mouse tail DNA (Figure 2B). Immunostaining for SIRT6 and CD31 demonstrated a specific increase of endothelial SIRT6 expression in the coronary vasculature of the heart from the Sirt6^ecTG mice compared with that in the wild-type mice (Figure 2C). Moreover, quantitative PCR, immunoblotting, and immunocytochemistry staining analysis of SIRT6 mRNA and protein levels in mouse coronary endothelial cells (MCECs) isolated from the Sirt6^ecTG mice and the Sirt6^ecWT mice further validated the overexpression of SIRT6 in MCECs from the Sirt6^ecTG mice (Figure 2D through 2F). Taken together, the results showed that we successfully established an endothelial-specific humanized SIRT6 overexpression mouse model, which could be valuable to define the tissue-specific role of SIRT6 gain-of-function in normal and pathological settings.

Figure 2. Generation and characterization of endothelial-specific human SIRT6 overexpression transgenic (Sirt6^ecTG) mice.

A, Construction of the humanized Sirt6^ecTG mice. The transgenic CTV vector with SIRT6 (human Sirt6 cDNA, blue box) was coupled to the Rosa26 locus. B, PCR of tail genomic DNA to differentiate Sirt6^ecWT and Sirt6^ecTG mice. C, Representative images of immunofluorescence staining of SIRT6 (red) and CD31 (cyan) on the heart tissues from the Sirt6^ecWT and Sirt6^ecTG mice. Scale bars, 20 μm. D, SIRT6 mRNA expression in the primarily isolated mouse coronary ECs (MCECs) of the Sirt6^ecWT and Sirt6^ecTG mice. GAPDH is used as a loading control for mRNA measurement. n = 6 repeated experiments. E, SIRT6 protein expression in the isolated MCECs of the Sirt6^ecWT and Sirt6^ecTG mice. n = 5 repeated experiments. F, Immunofluorescence staining of SIRT6 (red) and CD31 (cyan) in the isolated MCECs from the Sirt6^ecWT and Sirt6^ecTG mice. Scale bars, 20 μm. Statistical analysis was performed by unpaired Student’s t-test in D and non-parametric unpaired Mann-Whitney test in E.

Endothelial SIRT6 Overexpression in Mice Protects against Diabetes-Associated Diastolic Dysfunction

Next, we aimed to delineate the role of endothelial SIRT6 in diabetes-associated HFpEF. To this end, Sirt6^ecTG mice and their Sirt6^ecWT control littermates were subjected to the HFD/STZ challenges as described above, or the normal chow diet (ND) as non-diabetic controls (Figure 3A). Mouse body weight and blood glucose levels were increased after HFD/STZ challenge (Figure S4A through S4C). However, under HFD/STZ challenges, there was no significant difference in body weight, blood glucose, and insulin sensitivity (ITT) except for oral glucose tolerance (OGTT) between the Sirt6^ecTG mice and the Sirt6^ecWT mice (Figure S4A through S4E). Systolic and diastolic pressure measurements showed that while HFD/STZ challenge increased distolic blood pressure, SIRT6 overexpression did not affect the blood pressure of the mice either with HFD/STZ or ND diet (Figure S5A and S5B). In addition, heart rate was similar among all experimental groups (Figure S5C).

Figure 3. Endothelial SIRT6 overexpression protected against cardiac dysfunction in diabetic mice.

A, Schematic of the experimental setup. B, Representative left ventricular M-mode echocardiographic tracings in short-axis view. C, Percentage of LVEF. n = 8, 8, 11, 12 mice for Sirt6^ecWT ND, Sirt6^ecTG ND, Sirt6^ecWT HFD/STZ, and Sirt6^ecTG HFD/STZ, respectively. D, Percentage of LVFS. n = 8, 8, 11, 12 mice, respectively. E, Representative pulsed-wave Doppler tracings. F, Ratio between mitral E wave and E′ wave (E/E′). n = 8, 9, 10, 11 mice, respectively. G, Isovolumic relaxation time. n = 8, 9, 9, 9 mice, respectively. H, Ratio between wet and dry lung weight. n = 8, 8, 11, 11 mice, respectively. I, Running distance from exercise exhaustion tests. n = 6, 6, 8, 8 mice, respectively. J, Representative images of H&E and Sirius red (SR) staining in longitudinal sections of mouse left ventricle. Scale bars, 800 μm (H&E) and 100 μm (SR). K, Quantification of fibrosis by assessing the SR-positive areas. n = 5, 5, 8, 8 replicates, respectively. L, Ratio of heart weight to body weight (HW/BW). n = 6, 6, 9, 8 mice, respectively. M, Ratio of heart weight to tibia length (HW/TL). n = 6, 7, 10, 8 mice, respectively. N, Representative images of wheat germ agglutinin (WGA) staining. O, Quantification of cardiomyocyte cross-sectional area based on WGA staining. n = 5, 5, 6, 6 replicates, respectively, with 150–300 myocytes analyzed per image. Scale bars, 20 μm. Statistics were performed using two-way ANOVA plus Tukey’s multiple comparisons test for C, D, F, G, H, I, L and M, and non-parametric Kruskal-Wallis test with the Conover-Iman method for post hoc pairwise comparison and Benjamini-Hochberg correction for K and O.

To delineate cardiac phenotypes, we conducted echocardiographic and invasive hemodynamic assessments in all four groups of experimental male mice. There was persistent preservation of LVEF and LVFS in both diabetic and non-diabetic mice (Figure 3B through 3D, and Table S1). However, diabetic Sirt6^ecWT mice displayed a significant decline in diastolic function, manifested by an increase in the E to E’ ratio and IVRT (Figure 3E through 3G). Furthermore, diabetic Sirt6^ecWT mice exhibited a robust increase in lung weight (Figure 3H), elevated left ventricular end-diastolic pressure (LVEDP) (Figure S6A and Table S2), and exercise intolerance (Figure 3I), whereas these changes were attenuated in the Sirt6^ecTG mice. Moreover, diabetic Sirt6^ecTG mice displayed less cardiac fibrosis (Figure 3J and 3K) and less cardiac hypertrophy (Figure 3L through 3O). HFD/STZ challenge did not affect left ventricular end-systolic pressure (LVESP) in either Sirt6^ecWT mice or Sirt6^ecTG mice, confirming the preserved systolic function in this model (Figure S6B). The myocardial capillary density was reduced to a similar extent in both Sirt6^ecTG mice and Sirt6^ecWT mice under diabetic conditions (Figure S7). Moreover, to ask whether there is any gender difference, we evaluated the phenotypes of experimental female mice. We observed that endothelial SIRT6 overexpression also alleviated cardiac remodeling, diastolic dysfunction, and exercise intolerance in the HFD/STZ-challenged diabetic female mice (Figure S8). Taken together, our results indicate that endothelial-specific SIRT6 overexpression blunts diabetes-associated diastolic dysfunction.

Endothelial SIRT6 Overexpression Reduces the Excessive Accumulation of Lipid in the Diabetic Heart

Given the pivotal role of cardiac lipotoxicity in the pathogenesis of HF in obesity and diabetes^{8, 12, 29}, we then investigated the effect of endothelial SIRT6 overexpression on the accumulation of LDs in the heart. Oil Red O staining showed that the cardiac LD accumulation was significantly increased in the diabetic mice; however, diabetic Sirt6^ecTG mice displayed diminished LD accumulation (Figure 4A and 4B) compared with diabetic Sirt6^ecWT mice. Electron microscopy assays confirmed that the number and the size of LDs were significantly decreased in the hearts of diabetic Sirt6^ecTG mice (Figure 4C and 4D). Moreover, direct measurement of heart lipid contents showed that diabetic Sirt6^ecTG mice had a reduction of FAs and TGs (Figure 4E). In contrast, there was no significant difference between TGs and FAs in the liver (Figure 4F), an organ with a unique discontinuous fenestrated sinusoidal ECs that are not a barrier to FA uptake by the hepatocytes^{16, 30}. In addition, the diabetic Sirt6^ecTG mice displayed higher plasma TGs and FAs (Figure 4G). These results indicate that endothelial SIRT6 overexpression specifically corrected excessive accumulation of lipid into the cardiac tissues in the HFD/STZ-induced diabetic mice.

Figure 4. Endothelial SIRT6 overexpression reduced cardiac lipid accumulation in diabetic mice.

A, Representative images and quantification of Oil Red O-staining of the cardiac sections of mouse left ventricle. Scale bars, 50 μm. B, Quantification of Oil Red O positive area in the heart sections (expressed as a percentage) measured by Image J software. n = 6 repeated experiments per group. C, Representative images of electron micrographs (original magnification ×15,000) of myocardial tissue showing lipid droplets (red arrows) within the sarcoplasm of cardiomyocytes. D, Lipid droplet size (expressed as μm²) and lipid droplet numbers (expressed per 100 μm²) in heart sections. n = 3 replicates per group. Scale bars, 1 μm. E, Cardiac TG or FA contents in diabetic Sirt6^ecWT and Sirt6^ecTG mice. n = 6 mice per group. F, Liver TG or FA contents in diabetic Sirt6^ecWT and Sirt6^ecTG mice. n = 6, 8 mice for Sirt6^ecWT and Sirt6^ecTG, respectively. G, Plasma TG or FA levels in the diabetic mice fasted for 12 h. n = 6, 8 mice, respectively. Statistical analysis was performed by two-way ANOVA plus Tukey’s multiple comparisons test (B), non-parametric unpaired Mann-Whitney test (D), and unpaired Student’s t-test (E through G).

To examine whether SIRT6 overexpression affects the acute clearance of circulating TGs, we challenged the mice with an overdose of olive oil by gavage, which mainly consists of TGs that are glycerol esters of FAs. The levels of plasma TGs at 2, 4, and 6 hours after olive oil gavage in Sirt6^ecTG mice were not significantly different from those in Sirt6^ecWT mice (Figure S9), suggesting that endothelial SIRT6 is not involved in the rapid clearance of circulating TGs.

Endothelial SIRT6 Epigenetically Represses PPARγ Transcription under Diabetic Conditions

SIRT6 is an NAD⁺-dependent deacetylase that inhibits the acetylation of histone 3 lysine 9 (H3K9), an epigenetic marker for active gene transcription^{31, 32}. Thus, we asked whether the protective phenotype of the endothelial SIRT6 transgenic mice is resulting from alterations of gene expression in ECs. To this end, we performed RNA-sequencing analysis for SIRT6-dependent transcriptome profiling in the cultured HUVECs with adenovirus-mediated overexpression of SIRT6 (Figure S10A). Among the differentially expressed genes, 565 genes were downregulated, and 880 genes were upregulated, which was at least 2-fold compared with control (P<0.05) (Figure 5A). Gene Ontology (GO) enrichment analysis of the metabolism-relevant genes showed that long-chain FA transport and lipid transporter activity was influenced by SIRT6 overexpression (Figure 5B). The differentially expressed genes related to FA uptake or storage were selected, including PPARγ, a crucial regulator of FA handling and TG metabolism³³ (Figure 5C). Furthermore, a real-time PCR assay validated that PPARγ and CD36, (a gene encoding the protein that facilitates FA uptake), were reduced in the ECs treated with SIRT6 overexpression (Figure 5D). We also observed that PPARγ expression was increased, while SIRT6 was decreased, in the ECs treated with different concentrations of PA (Figure S11). This response was exacerbated with SIRT6 knockdown by siRNA treatment but mitigated by adenoviral SIRT6 overexpression in ECs (Figure 5E through 5H, and Figure S10B). Moreover, the mRNA levels of PPARγ-dependent downstream genes (CD36, FABP4, and FABP5), which are involved in FA uptake and transport, were increased with SIRT6 knockdown but decreased with SIRT6 overexpression in ECs (Figure S12). In line with those experimental observations, PPARγ expression was also increased in the heart tissue of the diabetic HF patients (Figure S13), paralleled by a decrease of endothelial SIRT6 (Figure 1) and an increase in cardiac TGs and FAs (Figure S13C). Together, these results indicate that endothelial SIRT6 represses the expression of PPARγ and inhibits PPARγ downstream FA transport-related genes.

Figure 5. SIRT6 epigenetically repressed PPARγ transcription via deacetylating H3K9.

A, Volcano plot of differentially expressed genes. Downregulation and upregulation were shown in the blue and red dots, respectively. Data were submitted to GEO database (GSE 213425). B, Gene Ontology (GO) enrichment analysis of 67 altered metabolism-relevant genes showing the top 20 regulated GO terms in SIRT6 overexpressed-HUVECs. The significantly enriched GO terms in differentially expressed genes compared to the genome background were defined by hypergeometric test followed by Benjamini-Hochberg correction. P values were shown by a different color, the size of the bubble indicates the gene count of each term. C, Heat map of representative differentially expressed metabolic genes in SIRT6 overexpressed HUVECs. D, Validation of RNA-seq data by real-time PCR in the SIRT6-overexpressed HUVECs compared with that of Ad-LacZ treated cells, normalized to the loading control GAPDH. E and F, Real-time PCR and Western blots showing PPARγ expression in PA-treated HUVECs with or without SIRT6 knockdown by small interfering RNA. G and H, The expression of PPARγ in PA-treated HUVECs with or without SIRT6 overexpression. I, ChIP assays with a SIRT6-specific antibody or IgG control in the HUVECs detected SIRT6 binding to the promoter of PPARγ. J, ChIP assay detected SIRT6 binding to the promoter of PPARγ in the SIRT6-overexpressed or -depleted HUVECs. The occupancy of SIRT6 to promoters was shown relative to background signals with IgG control. K, ChIP analysis detected H3K9 acetylation at the promoter of PPARγ in HUVECs compared with IgG control. L, ChIP analysis detected H3K9 acetylation at the promoter of PPARγ in the SIRT6-overexpressed or -deleted HUVECs. n = 5 (D) and n = 4 (E-L) repeated experiments. Statistical analysis was performed by non-parametric unpaired Mann-Whitney test (D, I, J, K and L), and non-parametric Kruskal-Wallis test with Conover-Iman method for post hoc pairwise comparison and Benjamini-Hochberg correction (E, F, G, and H).

Next, we explored the underlying mechanisms whereby SIRT6 modulates PPARγ gene transcription in ECs. Using chromatin immunoprecipitation (ChIP) assay, we found that SIRT6 bound to the PPARγ promoter region, but not the coding region (Figure 5I, Figure S14 and S15A). SIRT6 overexpression increased SIRT6 binding to the PPARγ promoter region, whereas siRNA-mediated SIRT6 knockdown reduced the binding with the PPARγ promoter region (Figure 5J). SIRT6 deacetylates H3K9 and thereby regulates gene expression by modifying chromatin structure³¹. Consistent with the notion, we observed that H3K9 acetylation was enriched around the PPARγ promoter region (Figure 5K and Figure S15B). Moreover, SIRT6 overexpression decreased, while SIRT6 knockdown increased, and the level of H3K9 acetylation of the PPARγ promoter region (Figure 5L). The luciferase reporter assay further substantiated that SIRT6 negatively regulated PPARγ promoter activity (Figure S16). These results reveal that SIRT6 specifically suppresses PPARγ expression in ECs by deacetylating H3K9 at the PPARγ promoter region.

SIRT6 Orchestrates Endothelial Fatty Acid Uptake

We further examined the role of SIRT6 in the regulation of endothelial FA uptake in an in vitro setting to exclude in vivo unknown factors. Using a fluorescent palmitate analog to trace FA uptake, we found that SIRT6 knockdown increased FA uptake into HUVECs in response to PA (Figure 6A), whereas SIRT6 overexpression inhibited endothelial FA uptake (Figure 6B). Moreover, mouse ECs isolated from the diabetic Sirt6^ecTG mice showed much lower FA uptake compared to that in ECs from the diabetic Sirt6^ecWT mice (Figure 6C). Consistently, expression of PPARγ, FABP4, and CD36 in ECs isolated from the diabetic Sirt6^ecTG mice was significantly decreased (Figure S17). Furthermore, the aortic intima was isolated from diabetic mice to directly test whether FA-related genes and the transendothelial FA transport were altered in the diabetic Sirt6^ecTG mice ex vivo. The mRNA levels of FA uptake-related genes (PPARγ, FABP4, and CD36) were significantly decreased in the diabetic Sirt6^ecTG mice compared with those in the diabetic Sirt6^ecWT control mice (Figure S18). In agreement with the gene expression profile, en face staining showed that the FA uptake to aortic endothelium was significantly decreased in the diabetic Sirt6^ecTG mice ex vivo (Figure 6D).

Figure 6. Endothelial SIRT6 inhibited FA uptake both in vitro and in vivo.

A, Representative images and quantification of BODIPY-C16 493/503 uptake in the SIRT6-depleted HUVECs in the presence or absence of palmitic acid (PA). Scale bars, 20 μm. B, Representative images and quantification of BODIPY-C16 493/503 uptake in the SIRT6-overexpressed HUVECs in the presence or absence of PA. Scale bars, 20 μm. C, Representative images and analytical results of BODIPY-C16 493/503 uptake in the lung microvascular ECs isolated from the Sirt6^ecWT or Sirt6^ecTG mice with or without HFD/STZ insult. Scale bars, 20 μm. n = 6 repeated experiments in A through C. D, En face immunofluorescence imaging of the thoracic aortae freshly isolated from Sirt6^ecWT or Sirt6^ecTG mice with HFD/STZ insult and incubated with BODIPY-C16 493/503 for 30 min. Scale bars, 10 μm. n = 3 mice. E and F, Uptake of oleic acid in Sirt6^ecWT and Sirt6^ecTG mice after HFD/STZ insult. Mice were fasted for 12 hours and then given a tail i.v. injection of [³H] oleic acid. E, Plasma [³H] radioactivity in Sirt6^ecWT and Sirt6^ecTG mice were measured at the indicated time points after injection. F, [³H] content in the different tissues 5 min after the injection of [³H] oleic acid. n = 6 mice per group in E and F. Statistical analysis was performed by two-way ANOVA plus Tukey’s multiple comparisons test in A-C, E, and F, non-parametric Kruskal-Wallis test followed by Dunn’s multiple comparison tests in D. * P=0.025, # P=0.0089 Sirt6^ecWT HFD/STZ vs. Sirt6^ecWT ND in E.

Restoring Endothelial SIRT6 Alleviates Excessive Uptake of FAs in the Diabetic Heart

Next, we examine whether endothelial SIRT6 overexpression in diabetic mice could correct excess FA uptake in vivo. To this end, we injected [³H] oleic acid (one type of FAs commonly found in animal and plant products) into 12-hour-fasted mice with or without HFD/STZ challenge and then assessed their tissue uptake of FAs. There was no significant difference in plasma clearance of oleic acid between diabetic Sirt6^ecTG mice and Sirt6^ecWT controls (Figure 6E), consistent with the finding of the TG clearance test (Figure S19). Tissue-specific oleate acid uptake in the Sirt6^ecTG mice was not significantly different from that in Sirt6^ecWT mice under non-diabetic conditions. However, under HFD/STZ-induced diabetic condition, there was a significant reduction of oleate acid uptake in the heart, quadriceps muscle, and brain in the diabetic Sirt6^ecTG mice. In contrast, the uptake of [³H] oleate in the liver was unchanged (Figure 6F).

Under healthy conditions, the heart receives approximately 70% of its energy from fatty acid oxidation (FAO) and 30% from the metabolism of carbohydrates and other substrates¹². Due to insulin resistance and the imbalance between FAs uptake and its oxidation, the diabetic heart is characterized by an abnormal metabolic shift to almost 90% reliance on FAO, which increases mitochondrial oxidative stress and lipid peroxidation as well as LD accumulation in the cardiomyocytes, ultimately resulting in cardiac lipotoxicity³⁴. To ask whether endothelial SIRT6 overexpression amends abnormal metabolic shifts in the diabetic heart, we first examined the expression of the metabolic genes that are involved in FAs and glucose metabolism. The mRNA level of FAs metabolism genes in the heart was upregulated in diabetic Sirt6^ecWT mice, whereas these changes were attenuated in diabetic Sirt6^ecTG mice (Figure S9A). In contrast, several glucose metabolism-related genes were upregulated in diabetic Sirt6^ecTG mice (Figure S19B). To quantify metabolic changes at the cellular level, we used a Seahorse extracellular flux analyzer to assess FAO in the isolated neonatal cardiomyocytes from both Sirt6^ecWT mice and Sirt6^ecTG mice. Etomoxir, a Cpt1 inhibitor, which can inhibit long-chain FAO, was used to evaluate the mitochondrial FAO in cardiomyocytes^{35, 36}. The maximal respiratory capacity was increased in the diabetic mice indicating an enhanced mitochondrial function in both Sirt6^ecWT mice and Sirt6^ecTG mice under diabetic conditions (Figure S19C and S19D). However, after FAO was inhibited by etomoxir, the maximal respiratory OCR was significantly decreased in the diabetic mice (Figure S19E), indicating that the FAO rate was increased under diabetes. Intriguingly, the increased FAO under diabetic conditions was attenuated in the cardiomyocytes from Sirt6^ecTG mice (Figure S19F and S19G). Moreover, we analyzed lipid peroxidation and mitochondria superoxide production in the cardiac tissues by 4-hydroxynonenal (4-HNE) staining and MitoSOX staining, respectively (Figure S19H through S19K). Diabetic Sirt6^ecTG mice displayed reduced 4-HNE and MitoSOX signals when compared with those of Sirt6^ecWT mice. Taken together, these data suggest that endothelial SIRT6 overexpression limits excess FA uptake to the diabetic heart and thus corrects abnormal metabolic shifts in diabetes.

SIRT6 Orchestrates Endothelial FA Uptake through A PPARγ-Dependent Pathway

We further ask whether PPARγ mediates SIRT6 action on FA uptake in ECs. To this end, we infected ECs with PPARγ adenovirus or treated ECs with a PPARγ agonist (rosiglitazone) to see whether PPARγ gain-of-function could reverse the reduced FAs uptake observed under SIRT6-overexpressed conditions. Both adenoviral PPARγ overexpression and rosiglitazone treatment abrogated the decreased FA uptake in SIRT6-overexpressed ECs (Figure 7A through 7D). Consistently, PPARγ overexpression rescued the reduced levels of CD36 and FABP4 mRNA expression in SIRT6-overexpressed ECs (Figure 7E and 7F). Furthermore, we observed that siRNA-mediated PPARγ knockdown attenuated FA accumulation in the SIRT6-depleted ECs (Figure 7G and 7H). Consistently, PPARγ knockdown abrogated the increase of FA transport-related genes in the SIRT6-depleted ECs (Figure 7I and 7J). Collectively, these results support that SIRT6 controls endothelial FA uptake in a PPARγ-dependent mechanism.

Figure 7. SIRT6 orchestrated endothelial fatty acid uptake in a PPARγ-dependent pathway.

A and B, Representative images and quantification results of FAs uptake in the SIRT6-overexpressed HUVECs in the presence or absence of Ad-PPARγ infection. Scale bars, 20 μm. n = 4 repeated experiments. C and D, Representative images and quantification results of FAs uptake in the SIRT6-overexpressed HUVECs in the presence or absence of rosiglitazone. Scale bars, 20 μm. n = 4 repeated experiments. E, CD36 mRNA expression in the SIRT6-overexpressed HUVECs in the presence or absence of Ad-PPARγ measured by real-time PCR. n = 5 repeated experiments. F, FABP4 mRNA expression in the SIRT6-overexpressed HUVECs in the presence or absence of Ad-PPARγ measured by real-time PCR. n = 5 repeated experiments. G and H, Representative images and quantification results of FAs uptake in the SIRT6-depleted HUVECs with or without PPARγ depletion by siRNA. Scale bars, 20 μm. n = 4 repeated experiments. (I) CD36 and (J) FABP4 mRNA expression in the SIRT6-depleted HUVECs with or without PPARγ depletion by siRNA measured by real-time PCR. n = 4 repeated experiments I and J. Statistical analysis was performed non-parametric Kruskal-Wallis test with the Conover-Iman method for post hoc pairwise comparison and Benjamini-Hochberg correction.

SIRT6 Activator MDL-800 Reverses Diabetes-Associated Diastolic Dysfunction in the Mouse Model of Experimental HFpEF

To translate our findings into a clinically relevant setting, we evaluate the potential therapeutic effects of MDL-800, a newly discovered SIRT6 activator³⁷, on HFD/STZ-induced HFpEF in mice. Briefly, after the establishment of HFpEF phenotypes with HFD/STZ challenges for 24 weeks, mice were subjected to daily orally gavage of MDL-800 (50 mg/kg) or vehicle control for another 4 weeks concurrent with HFD feeding (Figure 8A). We observed that MDL-800 administration significantly ameliorated the decline of diastolic function in HFD/STZ challenge-induced diabetic mice (Figure 8B through 8E, and Table S3). Furthermore, MDL-800 protected against cardiac dysfunction, as evidenced by a significant relief of exercise intolerance (Figure 8F), pulmonary edema (Figure 8G), and cardiac hypertrophy (Figure 8H and 8I). MDL-800 treatment also diminished TG and FA contents in the heart tissues (Figure 8J and 8K), without an obvious effect on hepatic lipids (Figure S20A). In addition, MDL-800 treatment decreased serum TG content but not FA level (Figure S20B). Oil Red O staining showed that diabetes-induced cardiac LD accumulation was significantly reduced by MDL-800 treatment (Figure 8L and 8M). Collectively, these results demonstrate that MDL-800 reduces cardiac LDs accumulation and mitigates diastolic dysfunction in a mouse model of diabetes-associated HFpEF.

Figure 8. SIRT6 activator MDL-800 alleviated diabetes-associated HFpEF in the murine model.

A, Schematic of the experimental setup. After HFD/STZ challenge for 24 weeks, MDL-800 (50 mg/kg, p.o.) or vehicle was orally given per day concurrently with continuous HFD feeding for 4 weeks. B, Percentage of LVEF, n = 8, 7, 7 mice per group. C, Percentage of LVFS, n = 8, 7, 7 mice per group. D, Representative pulsed-wave Doppler tracing images. E, Ratio between mitral E wave and E’ wave, n = 8 mice per group. F, Running distance from exercise exhaustion tests. n = 7 mice per group. G, Ratio between wet and dry lung weight, n = 7 mice per group. H, Ratio of heart weight to body weight (HW/BW), n = 7 mice per group. I, Representative images of H&E in longitudinal sections of left ventricles from mice. Scale bars, 800 μm. J and K, TG, or FA contents in the cardiac tissues. n = 6 mice per group. L and M, Representative images and quantification of Oil Red O-staining of the longitudinal sections of the left ventricles. n = 4, 6, 6 repeated experiments, respectively. N and O, Representative images and analytical results of BODIPY-C16 493/503 uptake in HUVECs in the presence or absence of MDL-800 (20 μM), with or without PA insult. Scale bars, 20 μm, n = 6 repeated experiments. P, Tissue [³H] oleic acid uptake in mice after HFD/STZ insult with or without MDL-800 treatment. n = 6 mice per group. Statistical analysis was performed one-way ANOVA plus Tukey’s multiple comparisons test (B, C, E, F, G, H, J and K), non-parametric Kruskal-Wallis test with the Conover-Iman method for post hoc pairwise comparison and Benjamini-Hochberg correction (M), and two-way ANOVA plus Tukey’s multiple comparisons test (O and P).

We then asked whether MDL-800 regulates FA uptake in vitro and in vivo. Using a cell culture system, we showed that MDL-800 attenuated the fluorescent-labeled FA uptake into ECs (Figure 8N and 8O). To further address whether MDL-800 regulates the tissue FA uptake in vivo, we injected [³H] oleic acid into 12-hour-fasted mice with or without MDL-800 treatment and then assessed the tissue FA uptake. MDL-800 gavage decreased FA uptake in the tissues of the heart and muscle, but not the liver (Figure 8P). These results are generally consistent with the results observed in the Sirt6^ecTG mice. Taken together, our results indicate that both genetic and pharmacological activation of SIRT6 inhibits FA uptake in vitro and in vivo.

DISCUSSIONS

Our study reveals a previously unrecognized role of SIRT6 in controlling endothelial FA transport to the heart and hence protecting against HFpEF under diabetic conditions. Firstly, the data observed from human cardiac specimens imply that a deficiency of endothelial SIRT6 may link diabetes to cardiac dysfunction. This notion is further supported by the results that genetically restoring endothelial SIRT6 expression in diabetic mice alleviated diastolic dysfunction in a mouse model of experimental diabetes-associated HFpEF. Secondly, our results reveal that SIRT6 negatively regulates FAs shuttling through suppressing PPARγ transcription by deacetylating H3K9 and hence inhibiting expression of PPARγ-dependent FA uptake-related genes (i.e., CD36, FABP4). Finally, we observed that small molecule SIRT6 activator MDL-800 attenuated excessive FAs uptake and ameliorated experimental HFpEF in diabetic mice, suggesting the therapeutic potential of MDL-800 for diabetes-associated HFpEF. Collectively, our results demonstrate that endothelial SIRT6 functions as a gatekeeper for FAs transport across the endothelium, and restoring endothelial SIRT6 expression combats diastolic dysfunction in diabetes via limiting overly uptake of FAs.

The phenotypic heterogeneity and the largely undefined pathogenesis of HFpEF are major reasons for the lack of evidence-based therapies^38–40. One of the critical obstacles to understanding the mechanistic underpinnings of HFpEF is the lack of good animal models that accurately recapitulate the complexities of human HFpEF⁴¹. HFpEF patients usually suffer from the comorbidities of hypertension, obesity, diabetes, and other metabolic disarrangement^2–4. Recently, several groups have attempted to generate animal models of experimental HFpEF to recapture the multiple comorbidities observed in HFpEF patients^42–45. For example, it was reported that concomitant metabolic stress (HFD feeding) and hypertensive stress (inhibition of constitutive nitric oxide synthase) in mice recapitulates several features of human HFpEF^{45, 46}, supporting the importance of metabolic stress in HFpEF etiology. However, a diabetic mouse model of experimental HFpEF is still lacking. We used the combination of HFD feeding and a single injection of low-dose STZ to induce diabetes in mice^{25, 26, 47, 48}, which featured insulin resistance and beta-cell dysfunction that are observed in the progression of metabolic syndrome to type 2 diabetes (T2D) in human patients^49–51. Our detailed analysis showed that HFD/STZ challenge-induced diabetes mouse model exhibited multiple cardiometabolic features of HFpEF in humans with the late stage of T2D, including obesity, mild hyperglycemia, diastolic dysfunction, pulmonary congestion, exercise intolerance, cardiac hypertrophy, and myocardial fibrosis^{5, 52}. Therefore, HFD/STZ challenge-induced diabetic mice may represent a useful animal model to investigate diabetes-associated HFpEF.

The presence of diabetes in our mouse model of experimental HFpEF might identify a unique phenotype of HFpEF, with implications for therapeutic strategies. In this study, we investigated the role of endothelial SIRT6 in diabetes-associated HFpEF. We observed that endothelial SIRT6 expression in the heart was significantly reduced in the cardiac tissues of diabetic HF patients. Similar results were observed in HFD/STZ-induced diabetic mice with experimental HFpEF. These results suggest that a deficiency of endothelial SIRT6 is implicated in diabetes-associated cardiac dysfunction. Moreover, genetically restoring endothelial-specific SIRT6 expression in HFD/STZ-induced diabetic mice protected against HFpEF. These results provide the in vivo evidence to support a crucial role of endothelial SIRT6 in diabetes-associated HFpEF. To further explore the clinical relevance of our findings, we applied the small molecule SIRT6 activator MDL-800 to treat the established HFpEF in HFD/STZ-induced diabetic mice. We found that oral gavage of MDL-800 for diabetic mice alleviated HFpEF phenotypes. Of note, our study focused on the role of endothelial SIRT6 in the regulation of HFpEF is distinguished from previous reports showing the protective effect of cardiomyocyte SIRT6 on cardiac hypertrophy^{53, 54} and doxorubicin-induced cardiac apoptosis⁵⁵. Collectively, our findings reveal that a deficiency of endothelial SIRT6 links diabetes to HFpEF and endothelial SIRT6 activation protects against diabetes-associated HFpEF in a preclinical setting.

It has recently been proposed that the abnormalities of the vasculature could potentially contribute to the pathophysiology of HFpEF^{3, 56, 57}, however, the underlying molecular mechanisms remain elusive. Endothelium, as an independent organ, governs nutrient shuttling^{33, 58}. Specifically, FAs are delivered to the cardiomyocytes via a dense network of non-fenestrated capillaries, whose interior is lined by a continuous layer of ECs specialized for FAs transport^{15, 30}. In this study, we identified SIRT6 as a molecular gatekeeper for FAs transport across the endothelium, orchestrating the endothelial FAs shuttling to the heart. Our results showed that endothelial SIRT6 overexpression in diabetic mice diminished FAs uptake in the heart through crossing coronary endothelium and diminished LDs accumulation in the heart tissue, without altering the hepatic FAs uptake and plasma TGs clearance. Unlike coronary endothelium which provides a physical barrier to delicately control transcellular transport of FAs, discontinuous fenestrated hepatic sinusoidal ECs are not a barrier to FAs transport to the hepatocytes in the liver³⁰. Consistent with our observations, it has been shown that endothelial Notch inhibition impairs FA transport across the endothelium in the heart but not in the liver²⁰. Moreover, endothelial (but not myocardial)-specific CD36 deletion in mice decreased FA uptake in the heart, whereas FAs uptake in the liver was not altered¹⁸. Together, these results indicate proper control of endothelial FA transport is particularly important for cardiac function.

SIRT6 is an NAD⁺-dependent deacetylase that targets the acetylation of H3K9, an epigenetic marker for active gene transcription^{31, 32, 59}. Through the analysis of SIRT6-dependent gene transcriptome in ECs, we identified PPARγ as a SIRT6 target gene in ECs. PPARγ is a master regulator for lipid metabolism and FAs transport. Mounting clinical evidence indicates that the use of PPARγ agonists such as thiazolidinediones (TZDs, rosiglitazone, and pioglitazone) as anti-diabetic drugs often triggers HF in diabetic patients^60–62. However, the reasons for such a detrimental outcome remain unclear. The results from this study may provide a reasonable explanation, in which overly activation of PPARγ by its agonists and subsequent excessive FA uptake would aggravate cardiac lipid accumulation and cardiac dysfunction. Thus, the clinical application of thiazolidinediones in T2D patients with a high risk of cardiovascular disease should be cautious. In addition, our study revealed that SIRT6 represses CD36 expression and limits endothelial FA uptake in a PPARγ-dependent manner. Mechanistically, we showed that SIRT6 directly deacetylases H3K9 at the PPARγ promoter region and thus represses PPARγ expression, resulting in decreased expression of the PPARγ downstream genes CD36, Fabp4, and Fabp5 in ECs. Of clinical relevance, we also observed that the SIRT6/PPARγ signaling was altered in the cardiac specimens of DM patients with HF and was associated with increased myocardial FA and TG contents. In line with our results showing the SIRT6-dependent regulation of endothelial FAs transport via PPARγ, it has been reported that deletion of endothelial PPARγ increased circulating FA and TG, and decreased skeletal muscle TG accumulation in response to HFD feeding. Collectively, our results suggest that proper control of PPARγ activity and tailoring nutrient delivery by SIRT6 could be beneficial for mitigating cardiac metabolic stress and cardiac dysfunction in diabetic conditions.

A few limitations should be taken into consideration when interpreting the results reported herein. First, we cannot rule out the possibility that SIRT6 activation improves diastolic function through additional mechanisms, besides altering endothelial FA transport. We observed that endothelial SIRT6 overexpression also alleviated cardiac hypertrophy and fibrosis; however, the underlying mechanisms remain unclear. It could be secondary effects resulting from SIRT6-mediated mitigation of cardiac lipid accumulation. Alternatively, endothelial SIRT6 overexpression may alter endothelial secretome including growth factors and cytokines that might affect cardiac signaling and function. Second, our data indicated that both genetic and pharmacological activation of SIRT6 protects against HFD/STZ challenge-induced HFpEF; however, we cannot conclude that every HFpEF associated with various conditions of metabolic disorders would benefit from endothelial SIRT6 activation. Specifically, it is unclear whether endothelial SIRT6 activation would improve cardiac dysfunction with type-1 diabetes or with obesity alone since we did not test the effect of endothelial SIRT6 activation on cardiac function in the group fed HFD not receiving STZ and/or an STZ-injected group fed a normal diet. Understanding of comprehensive mechanisms whereby SIRT6 activation improves diastolic dysfunction warrants further investigation, and additional studies are needed to extend our concept to other types of HFpEF models and their potential clinical applications.

In conclusion, our study unveils a critical role of endothelial SIRT6 as a novel regulator of FAs transport and diastolic dysfunction under diabetic conditions. Our data support the concept that the endothelium is a pivotal independent organ in maintaining tissue lipid homeostasis. This study also improves our understanding of the pathogenesis of diabetes-associated HFpEF, particularly the reduction of endothelial SIRT6 expression and the impairment of endothelial FAs gateway function in cardiac metabolic abnormalities and diastolic dysfunction. Our findings also suggest that agents, which target SIRT6 activation and FAs transport, may hold promise for the treatment of diabetes-associated HFpEF, and the therapeutic potential of the SIRT6 activator MDL-800 in human HFpEF warrants further clinical evaluation.

What Is Known?

• HFpEF is a growing health problem without effective therapies.

• Diabetes is a strong risk factor for HFpEF.

• Fatty acids are important energy resources for the healthy heart. Under diabetic conditions, however, increased free fatty acid levels are associated with cardiac dysfunction.

• Endothelium is a gatekeeper of fatty acid transport from blood to various tissues.

What New Information Does This Article Contribute?

• This study shows that a reduction of endothelial SIRT6 is a prominent phenomenon under diabetic conditions, which might link diabetes to cardiac dysfunction.

• Endothelial SIRT6 overexpression in mice protects against HFpEF under diabetic conditions.

• SIRT6 limits endothelial fatty acid uptake via epigenetically repressing PPARγ transcription in endothelial cells.

• SIRT6 activator MDL-800 improves diabetes-associated diastolic dysfunction in the experimental HFpEF mouse model.

HFpEF is a growing health problem without effective therapies. Diabetes mellitus is a strong risk factor for HFpEF. This study establishes endothelial SIRT6 as a key regulator of diabetes-associated HFpEF. SIRT6 acts as a gate keeper for endothelial fatty acid transcript through controlling metabolic genes such as PPARγ. Gain-of-function of SIRT6 ameliorated diastolic dysfunction concurrently with decreased cardiac lipid accumulation. Importantly, oral administration of small molecule SIRT6 activator MDL-800 to diabetic mice mitigated cardiac lipid accumulation and diastolic dysfunction. This study unveils a critical role of endothelial SIRT6 as a novel regulator of fatty acid transport and diastolic dysfunction under diabetic conditions. This study also suggests that the SIRT6 activator MDL-800 may have the therapeutic potential to combat human heart failure with preserved ejection fraction.

Nonstandard Abbreviations and Acronyms

ChIP

chromatin immunoprecipitation

CM

cardiomyocyte

DM

diabetes mellitus

EC

endothelial cell

FA

fatty acid

FABP4

fatty acid-binding protein 4

FABP5

fatty acid-binding protein 5

FAO

fatty acid oxidation

H3K9

histone 3 lysine 9

HF

heart failure

HFD

high-fat diet

HFpEF

heart failure with preserved ejection fraction

LD

lipid droplet

PA

palmitic acid

PPAR

peroxisome proliferator-activated receptor

siRNA

small interfering RNA

SIRT6

sirtuin 6

STZ

streptozocin

TG

triglyceride