Patients with type 2 diabetes have a substantial risk of developing cardiovascular disease. Phosphodiesterase 4 (PDE4) dysregulation is of pathophysiological importance in metabolic disorders. For determination of the role of PDE4 in diabetic cardiac dysfunction, mice fed with a high-fat diet (HFD) were treated by pharmacological inhibition of PDE4 or cardiac specific knocking down of PDE4D. Mice on HFD developed diabetes and cardiac dysfunction with increased cardiac PDE4D5 expression. PDE4 inhibitor roflumilast can reverse hyperglycemia and cardiac dysfunction, accompanied by the decrease of PDE4D expression and increase of muscle specific miRNA miR-1 level in hearts. Either cardiac specific PDE4D knockdown or miR-1 overexpression significantly reversed cardiac dysfunction in HFD mice, despite persistence of hyperglycemia. Findings of gain- and loss-of-function studies of PDE4D in cardiomyocytes indicated that inhibition of insulin-induced PDE4D protected cardiac hypertrophy by preserving miR-1 expression in cardiomyocytes through promoting cAMP-CREB-Sirt1 signaling–induced SERCA2a expression. We further revealed that insulin also induced PDE4D expression in cardiac fibroblasts, which causes cardiac fibrosis through TGF-β1 signaling–mediated miR-1 reduction. Importantly, the expression of PDE4D5 was increased in human failing hearts of individuals with diabetes. These studies elucidate a novel mechanism by which hyperinsulinemia-induced cardiac PDE4D expression contributes to diabetic cardiac remodeling through reducing the expression of miR-1 and upregulation of miR-1 target hypertrophy and fibrosis-associated genes. Our study suggests a therapeutic potential of PDE4 inhibitor roflumilast in preventing or treating cardiac dysfunction in diabetes in addition to lowering glucose.
Introduction
There is sufficient evidence that type 2 diabetes mellitus (T2DM) is a risk factor for incident heart failure (HF) (1,2). However, neither precise mechanism nor effective therapeutic remedy is available for diabetes-induced cardiac remodeling and contractile dysfunction. Some hypoglycemic drugs, while effectively lowering blood glucose, can paradoxically increase risk of HF (3). These findings raise the importance of choosing hypoglycemic agents according to their cardiovascular effect, independently of their effectiveness in reducing blood glucose levels.
Phosphodiesterases (PDE) represent a diverse family of phosphodiester hydrolyzing enzymes relevant for the metabolism of cyclic nucleotides such as cAMP and cyclic guanosine monophosphate (cGMP). Recent evidence suggests that PDE4 plays an essential role in glucose and fat metabolism (4,5). PDE4 is also one of the main PDE expressed in heart. The PDE4 family is encoded by four genes (PDE4A–PDE4D). A report demonstrated reduced levels of PDE4A and PDE4B in a rat model of cardiac hypertrophy, whereas PDE4D was unchanged (6). Human studies indicated that the expression of PDE4A and PDE4D is significantly lower in the explanted hearts from patients with idiopathic dilated cardiomyopathy compared with normal control subjects (7). However, heart tissues from patients with T2DM display increased PDE4D protein levels relative to patients without diabetes (8). In addition, long variants of PDE4D expressed in the heart include the variants PDE4D3, PDE4D5, PDE4D8, and PDE4D9. They differ in their regulatory properties and their subcellular localization, which is determined by variant-specific protein-protein interactions (7). Roflumilast, which is the first clinically approved oral selective PDE4 inhibitor for chronic obstructive pulmonary disease (COPD) treatment. Several human studies have also reported reduction of fat mass and improvement in insulin resistance after treatment with roflumilast (9,10). In an extensive clinical trial, treatment with roflumilast reduced major adverse cardiovascular events in patients with COPD (11). However, whether roflumilast protect heart in diabetes remains unknown. There is a need in relevant models for better understanding of the potential clinical implications of roflumilast in diabetes-associated cardiac dysfunction. Therefore, it will be very important to clarify the precise role of these PDE4D isoforms in diabetes-associated cardiac dysfunction.
PDE4 regulates the cAMP-dependent sarcoplasmic/endoplasmic reticulum calcium ATPase 2a (SERCA2a) activity in human hearts (12). Most functional studies indicate that reduced SERCA2a expression plays a pivotal role in contractile dysfunction of the diabetic heart (13,14). Conditional increase in SERCA2a expression has also been shown to reverse contractile dysfunction in preexisting diabetic cardiomyopathy in transgenic mice (15). It has been reported that SERCA2a gene therapy of failing hearts restores miRNA miR-1 expression, which is associated with improved cardiac function (16). Indeed, miR-1 was previously found to be downregulated in different models of cardiac hypertrophy and failure including diabetic heart (17).
Based on these findings, it is plausible to speculate that inhibition of PDE4 could be of potential benefit in diabetes-associated cardiac dysfunction. Here, we reveal that hyperinsulinemia-induced PDE4D expression leads to reduction of miR-1 expression in cardiomyocytes and cardiac fibroblasts, which causes diabetic cardiac hypertrophy and fibrosis in HFD mice. We further demonstrate that cardiac specific silencing of PDE4D or overexpression of miR-1 ameliorates progressive cardiac remodeling in HFD mice and identify the molecular pathways that mediate this. We also show that posttreatment with PDE4 inhibitor roflumilast reverses diabetic cardiac dysfunction in addition to lowering glucose.
Research Design and Methods
Human Samples
Human heart samples were obtained from patients with clinically diagnosed HF and diabetes undergoing cardiac transplantation and control heart samples from unmatched or rejected healthy donor hearts. All human heart studies were approved by the human ethics committee of Union Hospital of Huazhong University of Science and Technology (2017-S10005), and the subjects provided informed consent. Patient characteristics are summarized in Supplementary Table 1.
Experimental Animals
The animal care and experimental protocols followed the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the institutional animal care and use committee of Tongji Medical College, Huazhong University of Science and Technology. Male C57BL/6J mice (5–6 weeks old) were purchased from HFK Bioscience (Beijing, China) and housed in temperature-controlled cages, fed water and food ad libitum, and maintained on a 12-h light/dark cycle. Mice (n = 73) were randomly assigned to four groups: NC (n = 16), HFD (n = 32), Roflumilast-Prevention (n = 15), and Roflumilast-Ctrl (n = 10). The diets used for these studies were from HFK Bioscience. The NC and Roflumilast-Ctrl group were fed a normal chow (10% of calories from fat, 20% from protein, and 70% from carbohydrate) (no. D12450B), while the other two groups were fed a matched high-fat diet (HFD) (60% of calories from fat, 20% from protein, and 20% from carbohydrate) (D12492) for 20 weeks. Each group was administered by daily oral gavage with equal volume vehicle or roflumilast (1 mg/kg; Selleck Chemicals, Houston, TX) for 20 weeks. For the treatment experiments, after 16 weeks of HFD feeding, HFD mice (n = 32) were randomly assigned to two groups according to cardiac function. Each group was administered vehicle or roflumilast (1 mg/kg) for 4 weeks. At the end of the experiment, mice were fasted overnight and anesthetized prior to sacrifice. Blood samples were collected; hearts were weighed, fixed, and stored at −80°C for further experimental analysis.
Adeno-Associated Virus Serotype 9 Administration
All adeno-associated virus serotype 9 (AAV9) vectors used in this study were synthesized by Vigene Biosciences (Shandong, China). The shRNA oligonucleotides were cloned into the AAV shRNA expression vector pAV-U6-RFP (red fluorescent protein). AAV9 carrying shRNA against PDE4D (AAV9-shPDE4D) were generated according to standard protocols (target sequence: 5′-GATCCGGGCTGATTCTCCAAGCAAACTTCAAGAGAGTTTGCTT GGAGAATCAGCCCTTTTTTA-3′). AAV9-U6-shRNA (scramble)-RFP were produced as scramble control (AAV9-RFP). Mice (n = 40) fed with HFD displaying cardiac dysfunction were randomly assigned to four groups according to cardiac function: NC-AAV9-RFP, HFD-AAV9-RFP, NC-AAV9-shPDE4D, and HFD-AAV9-shPDE4D. Mice received a single-bolus tail vein injection (5 × 1011 viral genomes per mouse) of AAV9 containing either shPDE4D or a control gene RFP. After 5 weeks from injection, cardiac function was determined by echocardiography and then the animals were sacrificed. The mouse miR-1 (miR-1a-3p; miRBase identifier MIMAT0000123) segment was subcloned into the pAV-U6-GFP (green fluorescent protein) for overexpression. After 16 weeks of HFD feeding, mice (n = 40) were randomly assigned to four groups according to cardiac function: NC-AAV9-GFP, HFD-AAV9-GFP, NC–AAV9–miR-1, and HFD–AAV9–miR-1. AAV9-GFP or AAV9–miR-1 was injected into tail vein of mice with 5 × 1011 viral genomes per mouse. After 4 weeks, mice were sacrificed after echocardiography.
Echocardiographic Evaluation
Echocardiography was performed with a Vevo 2100 imaging system from VisualSonics (Toronto, Ontario, Canada) as previously described (8). Left ventricular ejection fraction (EF), fractional shortening (FS), and the ratio of early (E wave) and late (A wave) left ventricular diastolic filling velocities (E-to-A ratio) were assessed.
Glucose Tolerance Test and Insulin Tolerance Test
Glucose (1 g/kg i.g.) and insulin (0.75 mU/g i.p.) tolerance tests were performed after 12 h and 4 h of fasting, respectively. Blood glucose was measured with Roche Accu-Chek glucometer at 0, 30, 60, 90, and 120 min after glucose administration.
Serum Glucose and Insulin Assay
Blood serum was collected for measurement of glucose and insulin with a glucose kit (Biosino Bio-Technology and Science, Beijing, China) or Mouse Ultrasensitive Insulin ELISA kit (ALPCO, Salem, NH), respectively.
ELISA Detection of PDE4, SERCA-ATPase Activity, and TGF-β1
The activity of a PDE4-specific enzyme, SERCA-ATPase, and TGF-β1 in heart tissues was assayed with Human PDE4 ELISA kit or Mouse PDE4 ELISA kit (Jianglai Biotech, Shanghai, China), SERCA-ATPase ELISA kit (Jianglai Biotech), and Mouse TGF-β1 ELISA kit (MultiSciences Biotech, Hangzhou, China), respectively.
Neonatal Cardiac Myocytes and Fibroblasts Culture and Treatments
Neonatal rat ventricular myocytes and fibroblasts were isolated from 1- to 3-day-old SD rats as previously described (18). Cells were switched to serum-free medium for at least 12 h before experiments, preincubated for 30 min with roflumilast (100 nmol/L), and then treated with 100 nmol/L insulin (Sigma-Aldrich) to mimic an in vivo model of hyperinsulinemia. For in vitro miR-1 overexpression, cells were transfected with either a scramble or a miRNA mimic. PDE4D in cells were knocked down transiently by PDE4D siRNA (siPDE4D), and the scramble siRNA (si-nc) was used as a negative control. For PDE4D5 overexpression, cardiac myocytes were infected with adenovirus expressing PDE4D5 (Ad-PDE4D5) (gifts from Dr. Yang K. Xiang, University of California, Davis), and adenovirus containing empty plasmid (Ad-Ctrl) served as control. The infected cardiomyocytes treated with or without SERCA2a inhibitor 2,5-di-t-butyl-1,4-benzohydroquinone (BHQ) (10 μmol/L; Selleck Chemicals) for 24 h. The CREB siRNA was used to knock out CREB, and scramble siRNAs were used as control. For shRNA-mediated Sirt1 knockdown, Sirt1-shRNA (Genechem, Shanghai, China) or scramble control shRNA was transfected in neonatal rat cardiac myocytes for a total of 72 h and coincubated in the presence or absence of PKA activator forskolin (10 μmol/L; LC Laboratories, Woburn, MA) for 24 h. Cardiac fibroblasts were treated with TGF-β1 (10 ng/mL; PeproTech) to induce fibrosis for 24 h. Cells were transfected with PDE4D5 plasmid (a gift from Dr. Yang K. Xiang) and then treated with TGF-β1 receptor inhibitor SB431542 (1 μmol/L; Selleck Chemicals) for 48 h, and the pcDNA vector was used as a negative control.
All the small RNAs used in the current study were purchased from Riobio Co., Ltd (Guangzhou, China).
Histology and Cell Size Evaluation
Mouse hearts were perfused with 4% paraformaldehyde and fixed in paraformaldehyde for 24 h. Fixed hearts were paraffin embedded and serially sectioned at 5 μm on a microtome. Tissue sections were stained with wheat germ agglutinin to measure surface area of cardiomyocytes. Masson trichrome staining was used to examine myocardial fibrosis. Sirius red staining was used to examine cardiac collagen deposition. The tissue sections were visualized by microscope and four to five fields per heart randomly selected for the statistical analysis with ImageJ program (National Institutes of Health, Bethesda, MD) or Image-Pro Plus (Media Cybernetics, Bethesda, MD), respectively.
Immunofluorescence
The heart tissue sections were subjected to deparaffinization and antigen unmasking using citrate buffer for immunofluorescence staining. The tissue sections and cell slides were then blocked with 3% BSA, incubated overnight at 4°C with primary antibodies, and further incubated with corresponding fluorescent secondary antibodies. Nucleus was stained with DAPI. The primary antibodies used for immunofluorescence are listed in Supplementary Table 3.
Dual-Luciferase Reporter Gene Assays
The CREB plasmid (200 ng; Genechem), Sirt1 reporter (−1384/+277 relative to the transcription start site), or SERCA2a reporter plasmid (−1416/+251 relative to the transcription start site) (100 ng) or pRL-TK (10 ng) was cotransfected into 293T cells with use of Lipofectamine 3000 reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions. After transfection for 48 h, the cells were collected and treated with the Dual-Luciferase Reporter Assay kit (E1910; Promega, Madison, WI). GloMax (GloMax 20/20) (Promega) was used to detect the luciferase activity and calculate the ratio of firefly luciferase and renilla luciferase.
The amplify primer sequences were listed as follows: Sirt1, forward 5′-ATCACGCGTGCCAGAATTTCAGGGA-3′, reverse 5′-TCCGCTCGAGGCCGGGACCATC-3′, and SERCA2a, forward 5′-CGACGCGTCCCACCGTGCACTCACGTG-3′, reverse 5′-GGAAGATCTCCGAGCCCCCTCCCCGCTCTCTT-3′.
Real-time PCR
Total RNA from heart tissues or cells was extracted with use of RNAiso Plus kit (Takara Bio, Kusatsu, Japan), and cDNA preparation was performed with Prime Script RT Master kit (Takara Bio). Quantitative PCR was performed with TB Green Premix Ex Taq (Tli RNaseH Plus; Takara Bio). The relative expression level of specific mRNA was determined with the comparative cycle threshold (CT) method (2-ΔΔCT), normalized to the endogenous control gene 18s. U6 was used as an internal control for miRNAs template normalization. All of the primer sequences are listed in Supplementary Table 2.
Immunoprecipitation and Western Blot
Immunoprecipitation and Western blot were performed as previously described (18). Quantification of phosphorylation is based on the ratio of phosphorylated protein levels to total protein levels. The antibodies used for Western blot or immunoprecipitation are listed in Supplementary Table 3.
cAMP Detection
cAMP levels in heart tissues were measured with the cAMP-Glo Assay kit (Promega).
Statistical Analysis
All data are expressed as mean ± SEM. Differences between two groups were analyzed with an unpaired two-tailed t test. The comparisons among multiple groups were performed with one- or two-way ANOVA, followed by Tukey test for multiple comparisons. All P values considered statistically significant were <0.05. All statistical analysis were performed with GraphPad Prism 8.0 software (GraphPad Software, La Jolla, CA).
Data Resource and Availability
The data sets generated during or analyzed during the current study are available from the corresponding author on reasonable request.
Results
PDE4 Inhibitor Roflumilast Reverses HFD-Induced Hyperglycemia and Cardiac Dysfunction
C57BL/6 mice fed an HFD developed markedly obesity, hyperglycemia, hyperinsulinemia, and insulin resistance at 20 weeks (Supplementary Fig. 1A–E). HFD mice also displayed a time-dependent reduction in cardiac EF and FS (Supplementary Fig. 1F) associated with cardiac remodeling in the form of increased hypertrophy, fibrosis, and heart weight–to–tibia length ratio (Fig. 1A–D). Cotreatment with PDE4 inhibitor roflumilast for 20 weeks induced weight loss in HFD mice (Supplementary Fig. 1A) and markedly ameliorated hyperglycemia, hyperinsulinemia (Supplementary Fig. 1B and C), glucose intolerance, and insulin tolerance (Supplementary Fig. 1D and E) as well as cardiac dysfunction (Fig. 1A and Supplementary Fig. 1F). Roflumilast did not affect the cardiac function in mice fed normal chow (Fig. 1A).
For evaluation of whether HFD-induced cardiac dysfunction is reversed by PDE4 inhibitor, mice fed with HFD for 16 weeks displaying cardiac dysfunction (Supplementary Fig. 1F) were subjected to therapy with roflumilast for 4 weeks. Over the 4 weeks, posttreatment with roflumilast induced slight weight loss (Supplementary Fig. 1A) and ameliorated hyperglycemia, hyperinsulinemia, and glucose and insulin intolerance in HFD mice compared with vehicle-treated HFD mice (Supplementary Fig. 1B–D). Importantly, roflumilast therapy significantly reversed cardiac dysfunction including preserving EF, FS, and E-to-A ratio (Fig. 1A). The cardiac protective effect of roflumilast therapy was comparable with cotreatment with roflumilast (Fig. 1A). Moreover, roflumilast therapy reversed cardiac hypertrophy and fibrosis in comparison with vehicle-treated HFD mice (Fig. 1B–D). The mRNA levels of these cardiac hypertrophy–, fibrosis-, and oxidative stress–related genes also returned to normal in roflumilast-treated HFD mice (Fig. 1E–G). As detected with Sirius red staining, HFD mice exhibited increased collagen deposition of the heart, which was ameliorated by roflumilast treatment (Fig. 1C). In addition, HFD hearts had increased expression of collagen degrading matrix metalloproteinase 2 (MMP2) and tissue inhibitor of metalloproteinase 2 (TIMP2) compared with control hearts, while the expression of MMP9 and TIMP1 remained unchanged (Supplementary Fig. 2). In the roflumilast therapy group, MMP9 expression was increased, MMP2 and TIMP1 expression remained unchanged, and TIMP2 expression markedly decreased compared with HFD group (Supplementary Fig. 2).
PDE4D5 Expression Is Increased in Human Failing Diabetic Hearts and Hearts of Mice Fed HFD
Our recent results revealed that expression of PDE4D is elevated in human diabetic hearts (8). We further found that PDE4D5 expression was significantly elevated in cardiac tissues from HF patients with diabetes in comparison with donors without diabetes (Fig. 2A), and the other isoform, PDE4D3, was mildly increased and PDE4D9 showed a trend to increase in human failing diabetic hearts. Accordingly, the PDE4 activity was increased, while the cAMP level was reduced, in human failing diabetic hearts (Fig. 2B and C). Consistent with this, both mRNA and protein expression of PDE4D were increased in HFD mouse hearts (Fig. 2D and E), and PDE4D5 expression was also significantly increased (Fig. 2E), with a nearly twofold increase in PDE4 activity and an ∼40% decrease in cAMP content in HFD mouse hearts compared with those fed with normal chow (Fig. 2F and G). The phosphorylation of PDE4D was also increased in HFD mouse hearts, which might due to the increased expression of PDE4D (Supplementary Fig. 3A). In vitro study showed that acute insulin stimulation induced PDE4D phosphorylation in cardiomyocytes and cardiac fibroblasts (Supplementary Fig. 3B and C). Moreover, insulin also induced PDE4D expression in cardiomyocytes at 24 h (Supplementary Fig. 3B), while chronic insulin stimulation–induced PDE4D expression in cardiac fibroblasts was observed up to 72 h (Supplementary Fig. 3D). These findings suggest that PDE4D plays an important role in the heart during the early and late stages of diabetes. Roflumilast treatment attenuated PDE4D and PDE4D5 induction (Fig. 2D and E) and returned PDE4 activity and cAMP content to normal levels in HFD mouse hearts in comparisons with HFD mice treated with vehicle alone (Fig. 2F and G). In contrast, the mRNA level of other PDE4 subtypes was not significantly modified by roflumilast (Fig. 2D). These results indicated that PDE4 inhibitor may exert a direct cardiac effect to protect against cardiac dysfunction in HFD mice via inhibition of PDE4D.
Cardiac PDE4D Suppression Reverses HFD-Induced Cardiac Dysfunction and Remodeling
For exclusion of the possibility that the beneficial cardiac effects observed from PDE4 inhibitor treatment are due to the improvement of glucose metabolism, mice fed with HFD displaying cardiac dysfunction were subjected to 5 weeks’ therapy with AAV9-shPDE4D, which mediated cardiac gene transfer of shRNA to knock down expression of PDE4D in the heart. HFD mice injected via the tail vein with AAV9-shPDE4D or AAV9-RFP (control scramble) developed similar obese phenotype, evidenced by comparable body weight gain, fasting blood glucose level, and glucose tolerance (Supplementary Fig. 4A–C). PDE4D protein expression in hearts was markedly decreased in HFD-fed AAV9-shPDE4D mice (Fig. 3A). In addition, HFD was able to increase expression of PDE4D in skeletal muscle or liver tissues, while PDE4D expression in skeletal muscle or liver tissues did not differ between HFD-fed AAV9-shPDE4D and HFD-fed AAV9-RFP mice (Supplementary Fig. 4D). The increased PDE4 activity and decreased cAMP content in HFD hearts also returned to normal level with AAV9-shPDE4D (Fig. 3B and C). In concert with this, AAV9-shPDE4D significantly increased EF, FS, and E-to-A ratio in comparison with HFD mice treated with AAV9-RFP (Fig. 3D). These observations were accompanied by ameliorated cardiac hypertrophy and fibrosis in HFD-fed AAV9-shPDE4D mice compared with HFD-fed AAV9-RFP mice (Fig. 3E–G). The expression of hypertrophy-, fibrosis-, and oxidative stress–related genes also returned to normal levels in HFD-fed AAV9-shPDE4D mice compared with AAV9-RFP HFD mice (Supplementary Fig. 5A–C). It is noteworthy that AAV9-shPDE4D did not ameliorate hyperglycemia or glucose intolerance in HFD mice (Supplementary Fig. 4B and C), suggesting that inhibition of PDE4D in heart plays a direct beneficial role in the improvement of diabetes-associated cardiac dysfunction.
PDE4D Inhibition Improves Cardiac Dysfunction via Preserving SERCA2a Expression
Given that PDE4D fine-tunes the cAMP-dependent SERCA2a activity in human hearts and PDE4 inhibitor–induced positive inotropic effect by increasing the activity of SERCA2a (12,19), we detected the expression and activity of SERCA2a in HFD hearts. HFD-fed mice had lower cardiac SERCA2a expression and phospholamban phosphorylation compared with the mice that received a normal diet, which was reversed by roflumilast treatment for 4 weeks (Fig. 4A). Furthermore, roflumilast therapy increased SERCA2a activity in HFD hearts (Fig. 4B). Likewise, suppression of cardiac PDE4D through AAV9-shPDE4D significantly enhanced not only the activity but also the mRNA and protein expression of SERCA2a in HFD hearts (Fig. 4C–E). Notably, in vitro study showed that overexpression of PDE4D5 can reduce SERCA2a expression in cardiomyocytes (Fig. 4F). Treatment with SERCA2a inhibitor BHQ augmented expression of cardiac hypertrophy and oxidative stress–related genes in cardiomyocytes, which was inhibited by silencing of PDE4D (Fig. 4G). Together, these data suggested that PDE4 inhibitor–ameliorated cardiac dysfunction in HFD hearts was associated, at least in part, with restoration of SERCA2a expression. We then sought to examine which signaling branch is involved in PDE4D-mediated reduction of SERCA2a expression.
PDE4D Inhibition Enhances SERCA2a Expression in Cardiomyocytes via cAMP/PKA/CREB/Sirt1 Signaling Pathway
Because protein deacetylase sirtuin 1 (Sirt1) acts as a transcriptional activator of SERCA2a gene expression in high-glucose conditions (20) and Sirt1 is a direct transcriptional target of cAMP responsive element binding protein (CREB), which is activated by cAMP-PKA (21), we assessed whether PDE4 inhibitor restores SERCA2a expression through activation of cAMP-PKA-CREB-Sirt1 signal pathway. In line with previous research (21), forskolin, a PKA activator, could facilitate CREB phosphorylation on serine 133, which in turn heightened Sirt1 and SERCA2a expression (Supplementary Fig. 6A). As expect, knocking down Sirt1 significantly inhibited forskolin-induced SERCA2a expression and CREB phosphorylation in cardiomyocytes (Supplementary Fig. 6A). Overexpression of CREB significantly upregulated Sirt1 promoter and SERCA2a promoter activity compared with the control (Supplementary Fig. 6B and C). In turn, knocking down of CREB led to a reduction in Sirt1 and SERCA2a expression (Supplementary Fig. 6D). These data established the involvement of cAMP-PKA-CREB-Sirt1 signal pathway in the regulation of SERCA2a expression in cardiomyocytes. Consistent with this, both CREB phosphorylation on serine 133 and Sirt1 expression were decreased in hearts of HFD group (Fig. 5A and B), which were reversed by suppression of cardiac PDE4D or roflumilast therapy (Fig. 5A and B). Sirt1 mRNA was also drastically reduced in the HFD group and reversed by silencing of PDE4D, but no difference was observed in CREB mRNA among groups (Supplementary Fig. 6E). Moreover, cardiac specific silencing of PDE4D inhibited acetylation of SERCA2a in HFD hearts (Supplementary Fig. 6F).
Our in vitro data further support these hypotheses. Consistent with previous study (8), insulin induced PDE4D protein in isolated ventricular myocytes (Fig. 5C). Roflumilast not only returned the PDE4D expression to normal levels but also enhanced CREB phosphorylation and expression of Sirt1 and SERCA2a in insulin-treated cardiomyocytes (Fig. 5C). Furthermore, PDE4D5 overexpression reduced CREB phosphorylation and expression of Sirt1 and SERCA2a in cardiomyocytes (Fig. 4F). In contrast, knocking down Sirt1 attenuated the restoration effect of silencing of PDE4D on SERCA2a expression in insulin-treated cardiomyocytes (Fig. 5D). Together, these results suggest that PDE4 inhibition activated cAMP-PKA-CREB signaling pathway, which increased the expression of Sirt1, and subsequently promoted SERCA2a expression and deacetylation, leading to increased activity of SERCA2a in HFD hearts.
PDE4D Inhibition Reverses Cardiac Dysfunction Through SERCA2a-Mediated miR-1 Restoration
SERCA2a gene therapy protects cardiac dysfunction in HF through restoring miR-1 expression via an Akt/FoxO3a-dependent pathway (16). Consistently, we found that the decrease of SERCA2a expression in HFD hearts was accompanied by increased phosphorylation of Akt and FoxO3a, which was reversed by roflumilast treatment (Fig. 4A). Meanwhile, miR-1 expression was also reduced in HFD hearts, which was rescued by both roflumilast (Supplementary Fig. 7A) and AAV9-shPDE4D (Fig. 6A) therapy. We further examined the other diabetic cardiomyopathy associated–miRNA expression in heart tissues from HFD mice. The results showed that miR-141 expression was significantly increased during HFD-induced cardiac dysfunction; however, the increased level of miR-141 was not affected by roflumilast treatment (Supplementary Fig. 7A).
Then we examined whether the cardioprotective effect of PDE4D inhibition is associated with SERCA2a–miR-1 signal pathway in cardiomyocytes. Here we show that silencing of PDE4D reversed the reduced miR-1 expression caused by SERCA2a inhibitor BHQ (Fig. 6B) and BHQ exacerbated the inhibition effect of PDE4D5 overexpression on miR-1 level in cardiomyocytes (Fig. 6C). Furthermore, the expression of cardiac hypertrophy– and oxidative stress–related genes was significantly upregulated by PDE4D5, and these increases were attenuated by miR-1 (Fig. 6D). We next used AAV9 expressing miR-1 (AAV9-miR-1) to assess the role of miR-1 in HFD-induced cardiac dysfunction in vivo. As shown in Supplementary Fig. 7B, miR-1 was overexpressed successfully in mouse hearts. The results indicated that AAV9-miR-1 not only significantly restored miR-1 expression in HFD hearts (Fig. 6E) but also ameliorated HFD-induced cardiac dysfunction (Fig. 6F), cardiac hypertrophy, and fibrosis (Fig. 6G–I). The mRNA level of cardiac hypertrophy–, fibrosis–, and oxidative stress–related genes also returned to normal in AAV9-miR-1–treated HFD mice (Supplementary Fig. 7C), as was seen in the HFD AAV9-shPDE4D group. Previous studies identified that miR-1 attenuates pathological cardiac remodeling by negatively regulating MEF2a, GATA4, and fibulin 2 (Fbln2) (22–24). In line with this, the expression of GATA4, MEF2a, and Fbln2 was significantly increased in HFD mice, which was attenuated by AAV9-miR-1, PDE4 inhibitor roflumilast, and AAV9-shPDE4D (Supplementary Fig. 8A–C), suggesting that these miR-1 target genes are involved in the protective effects of miR-1 and PDE4 inhibition. However, body weight, fasting glucose level, and glucose tolerance were not ameliorated by AAV9-miR-1 (Supplementary Fig. 7D–F). Together, these data indicated that PDE4 inhibition alleviated HFD-induced cardiac dysfunction that was partially associated with SERCA2a–miR-1 signal pathway in cardiomyocytes.
PDE4D Inhibition Prevents TGF-β1–Driven miR-1 Reduction and Fibrosis in Cardiac Fibroblasts
In a recent study, miR-1 expression and regulation were demonstrated in adult ventricular fibroblasts, where it acts as a novel negative regulator of TGF-β1–induced adult cardiac fibroblast proliferation (25). As shown in Supplementary Fig. 9A and B, roflumilast reduced PDE4D expression not only in cardiomyocytes but also in cardiac fibroblasts of HFD mice. In view of both PDE4D inhibition and miR-1 overexpression displaying protective effects on cardiac fibrosis in HFD mice (Figs. 1C, 3F, and 6H), and upregulation of TGF-β1 relating to fibrosis in diabetic heart (26), we aimed then to investigate whether the antifibrosis effect of PDE4D inhibition is also associated with miR-1 restoration in cardiac fibroblasts. In line with previous studies (26), mice fed HFD, as compared with chow, displayed increased TGF-β1 expression and activity in heart tissues. The increased TGF-β1 expression and activity were attenuated by both PDE4 inhibitor and AAV9-shPDE4D (Fig. 1F and Supplementary Figs. 5B and 10A and B). As shown in Fig. 7A, miR-1 was downregulated in response to TGF-β1 stimulation in cultured cardiac fibroblasts, which was significantly reversed by miR-1 mimic. miR-1 mimic also inhibited TGF-β1–induced fibrosis-related genes expression (Fig. 7B), together with miR-1–attenuated HFD-increased expression of MEF2a and fibulin 2 (Supplementary Fig. 8A), which were considered as critical regulators of TGF-β in cardiac fibroblast (23,27), indicating that miR-1 negatively regulates TGF-β1–induced cardiac fibrosis through inhibition of its direct target genes MEF2a and fibulin 2. Given that insulin enhances TGF-β1 responsiveness and autocrine TGF-β1 signaling in fibroblasts and epithelial cells (28) and hyperinsulinemia contributes to HF by increasing PDE4D in cardiomyocytes (8), cardiac fibroblasts were treated with insulin for examination of whether PDE4D inhibition protects against HFD-induced cardiac fibrosis via regulation of insulin–TGF-β1–miR-1 signaling. As with the cardiomyocytes, insulin also induced PDE4D expression in cardiac fibroblasts (Supplementary Fig. 3D and Fig. 7C). More than that, insulin treatment increased active TGF-β1 expression, consequently enhancing activation of Smad2, a key intracellular mediator of TGF-β1 signaling, and collagen deposition in cardiac fibroblasts (Fig. 7C). Accordingly, miR-1 expression was reduced in insulin-treated cardiac fibroblasts (Fig. 7D). Such effects of insulin on TGF-β1–mediated collagen deposition and miR-1 reduction in cardiac fibroblasts were attenuated by PDE4 inhibitor roflumilast (Fig. 7C and D). Consistently, silencing of PDE4D abolished insulin-induced Smad2 activation and collagen deposition in cardiac fibroblasts (Fig. 7E). Consistent with HFD mouse hearts, in vitro study showed that PDE4D5 was upregulated by insulin in cardiac fibroblasts, which was attenuated by roflumilast (Supplementary Fig. 11). Furthermore, overexpression of PDE4D5 resulted in increased expression of TGF-β1 and fibrosis-related genes in cardiac fibroblasts accompanied by decreased miR-1 expression (Fig. 7F). SB431542, a selective TGF-β1 receptor inhibitor, not only inhibited PDE4D-induced TGF-β1 and fibrosis-related genes expression but also attenuated PDE4D-mediated miR-1 reduction in cardiac fibroblasts (Fig. 7F). These data strongly suggest that PDE4D participates in and contributes to insulin–TGF-β1–miR-1 signaling–mediated fibrosis.
Discussion
In the current study, we find that cardiac PDE4D5 expression is significantly increased in diabetes-associated HF patients. Consistently, mice on HFD developed diabetes and cardiac dysfunction with cardiac PDE4D5 upregulation. We describe a novel mechanism by which hyperinsulinemia induced PDE4D expression in cardiomyocytes, leading to reduction of cAMP-CREB-Sirt1 signaling–mediated SERCA2a expression, which in turn results in miR-1 downregulation and cardiac hypertrophy through inhibition of miR-1 target hypertrophy-associated genes. In addition, insulin also induced PDE4D expression in cardiac fibroblasts, causing cardiac fibrosis through TGF-β1 signaling–mediated miR-1 reduction in cardiac fibroblasts. We further demonstrate that cardiac specific silencing of PDE4D or overexpression of miR-1 reversed preexisting pathological remodeling in HFD mice. Moreover, pharmacological inhibition of PDE4 with roflumilast not only effectively reverses progressive deterioration of cardiac function and remodeling in mice but also ameliorates hyperglycemia.
Phosphodiesterases play essential roles in modulating SERCA2 pump, which controls excitation-contraction coupling in cardiomyocytes (12,29). Previous studies suggest that PDE4D coassembles with SERCA2a in both murine and failing human hearts (12) and fine-tunes the cAMP-dependent SERCA2a activity (7,12). PDE4 inhibition resulted in a positive inotropic effect by activating SERCA2a (19). These findings support the possibility that selective PDE4D inhibition could serve as a target for the treatment of HF by specifically activating SERCA2a. Previous studies have demonstrated that hyperinsulinemia-induced PDE4D expression contributes to impaired inotropic signaling in isolated HFD cardiomyocytes (8). In our current study, PDE4 inhibitor reversed the decrease of SERCA2a expression and cardiac dysfunction elicited by high-fat intake in vivo. We also observed restoration of cardiac function and SERCA2a expression in HFD mice after silencing of cardiac PDE4D, despite persistence of hyperglycemia. We noted an increase in SERCA2a mRNA level was in accordance with the protein level through inhibition of PDE4D. These findings suggest that the beneficial effects of PDE4D inhibition are at least in part associated with restoration of SERCA2a expression at the transcriptional level.
Sirt1, the founding member of a large family of class III histone deacetylases, regulates a wide variety of cellular processes such as gene transcription (20). Sirt1 activation significantly reduced acetylation and restored the SERCA2a function, resulting in beneficial outcomes under cardiac insults (30). In a previous study it was reported that resveratrol enhances SERCA2a expression and improves cardiac function in diabetic cardiomyopathy through activation of Sirt1 (20). Moreover, resveratrol activates Sirt1 in vivo due to its effect on cAMP signaling by directly inhibiting PDE4 (4). It also has been reported that cAMP increased the protein level of sirtuins and Sirt1 is a direct transcriptional target of CREB, which is activated by cAMP-PKA in the nucleus (21,31). Given that PDE4D5 appears to form puncta in nucleus and plays a critical role in the functional nuclear cAMP-PKA signaling (32,33) and the expression of cardiac PDE4D5 was increased in human failing hearts with diabetes and HFD mice, we demonstrated that PDE4D5 overexpression reduced CREB phosphorylation, leading to decreased expression of Sirt1 and SERCA2a. In line with these studies, our results further revealed that PDE4D inhibition significantly reversed insulin-induced SERCA2a downregulation in a Sirt1-dependent manner. These results indicated that PDE4D inhibition increased SERCA2a expression in a cAMP-PKA-CREB-Sirt1–dependent manner.
Recent studies have shown that miRNAs play important roles in cardiac physiology and pathology (34,35). Altered expression of several miRNAs has been shown in diabetic cardiomyopathy; e.g., miR-1 was markedly depressed in the diabetic cardiomyocytes (36–38). It was reported that miR-1 overexpression inhibited cardiomyocyte hypertrophy induced by leptin in vitro (39) and prevented pressure overload–induced cardiac hypertrophy and fibrosis in vivo (24). Findings of previous studies showed that miR-1 attenuated cardiomyocyte hypertrophy by negatively regulating hypertrophy-associated genes MEF2a and GATA4 (22). In addition, it has been reported that MEF2a silencing significantly reduced hyperglycemia-induced cardiac fibroblast proliferation and attenuated diabetes-induced myocardial fibrosis and cardiac dysfunction (23). miR-1 may also be implicated in the regulation of fibrosis by targeting Fbln2, a secreted ECM protein that plays an important role during adverse tissue remodeling (24). However, the mechanism underlying the contribution of the change in miR-1 expression to the progression of diabetic cardiomyopathy remains largely unknown. The miR-1 promoter has consensus sequences for the FoxO3a transcription factor (40). FoxO3a phosphorylation leads to FoxO3a export from the cardiomyocyte nucleus, thus resulting in decreased miR-1 transactivation (16). Kumarswamy et al. (16) reported that miR-1 is downregulated in a chronic HF model and its expression is restored to normal levels during reverse remodeling by AAV9-SERCA2a treatment via reversal of the phosphorylated Akt–FoxO3a axis. In line with these findings, our present study showed that an HFD induced a significant decrease of miR-1 expression in hearts. Of mention, PDE4 inhibitor or cardiac suppression of PDE4D effectively reversed cardiac dysfunction and remodeling and rescued the expression of SERCA2a and miR-1 in HFD hearts accompanied by restoration of Akt and FoxO3a phosphorylation. Moreover, the results described herein demonstrated that AAV9-miR-1, roflumilast, and AAV9-shPDE4D reversed cardiac remodeling and downregulated hypertrophy-associated genes MEF2a and GATA4 compared with control treatment HFD mice. Consistently, in vitro study showed that PDE4D decreased miR-1 expression and increased the expression of hypertrophy and oxidative-related genes in primary cultured cardiomyocytes, which was attenuated by miR-1 mimic. Furthermore, SERCA2a inhibitor induced miR-1 reduction in cardiomyocytes was reversed by silencing of PDE4D and was exacerbated by PDE4D5 overexpression. These finding are consistent with the other studies with results showing that impaired SERCA2a activity and miR-1 downregulation are associated with oxidative stress, which leads to pathological heart conditions such as HF (38,41). Together these data suggest that suppression of PDE4D rescued SERCA2a expression, leading to miR-1 restoration in cardiomyocytes, which may contribute to improvement of cardiac function and hypertrophy through inhibition of MEF2a and GATA4 (Fig. 8). Cardiac fibrosis is one of the predominant pathological features of diabetic cardiomyopathy. A dysregulation of MMPs and TIMPs is supposed to be a hallmark for myocardial fibrosis in diabetes. However, there are conflicting reports regarding their expression levels in diabetic hearts (42–45). From previous study investigators reported that reduced MMP-2 activity contributes to cardiac fibrosis in diabetic cardiomyopathy (43). In the current study, the expression of MMP2 and TIMP2 was increased in HFD heart, while roflumilast treatment reduced TIMP2 expression and had no effect on MMP2 expression compared with the HFD group, indicating that roflumilast increased MMP2 activity in HFD heart as TIMP2 effectively inhibits MMP2. In addition, roflumilast ameliorated collagen deposition and reduced collagen 1 expression in HFD hearts. Furthermore, our present study demonstrated that both cardiac silencing of PDE4D and overexpression of miR-1 correlated with improvement in HFD-induced cardiac fibrosis and downregulation of fibrosis-associated genes MEF2a and Fbln2 (23,27), indicating a potential novel role of PDE4D and miR-1 in cardiac fibroblasts, although there are very few reports regarding either PDE4 or miR-1 in cardiac fibroblasts. miR-1 has been viewed as a muscle-specific miRNA and extensively studied in cardiomyocytes (46). However, miR-1 is critical to inhibit not only myocyte hypertrophy but also extracellular matrix deposition (47). Recently, miR-1 was also found to be expressed in cardiac fibroblasts and plays an important role in TGF-β1–induced fibroblast proliferation (25). The miR-1 target genes MEF2a and Fbln2 also prevent fibrosis by blocking TGF-β1 in cardiac fibroblasts (23,27). It is well established that TGF-β1 is one of the molecular mediators implicated in the progression of fibrogenesis. In diabetes, hyperglycemia can cause the expression changes of TGF-β that lead to diabetic cardiomyopathy through the SMAD-dependent and -independent pathways (28,48). In line with these findings, in comparisons with the normal chow mice group, the increased TGF-β1 activity in hearts was associated with cardiac fibrosis and a significant reduction of miR-1 level in the HFD mice group. Our in vitro data showed that miR-1 was downregulated in response to TGF-β1 stimulation in primary cultured cardiac fibroblasts. Moreover, TGF-β1 induced expression of fibrosis-associated genes in cardiac fibroblasts was attenuated by miR-1 mimic, which is consistent with other study (25), suggesting that TGF-β1–mediated miR-1 reduction in cardiac fibroblast plays an essential role in HFD-induced cardiac fibrosis. Given that cAMP in fibroblasts suppressed the inductive activity of TGF-β1 and abolished TGF-β1–induced interaction of Smad2/3 with its transcriptional coactivator CREB binding protein (49,50), we suggested that PDE4D decreased the expression of miR-1 in cardiac fibroblasts through inhibiting cAMP-mediated TGF-β1 signaling inhibition, which is distinct from cardiomyocytes that is mainly dependent on SERCA2a. Thus, although most studies have focused on PDE4D in cardiac myocytes and its critical role in regulating myocyte contractility and arrhythmogenesis, the present in vitro study further depicted that insulin also induced PDE4D in cardiac fibroblasts, which is necessary for activation of TGF-β1 and TGF-β1–miR-1–mediated profibrotic signaling. Taken together, these findings suggest that hyperinsulinemia-increased PDE4D in cardiac fibroblasts leads to cardiac fibrosis via TGF-β1 signaling–mediated miR-1 reduction and upregulation of miR-1 target fibrotic genes MEF2a and Fbln2 in cardiac fibroblasts (Fig. 8).
Given that more intensive glycemic control in patients with established T2DM does not reduce occurrence of HF events (51,52), it is of great clinical importance to explore a more comprehensive treatment approach in T2DM management that addresses both hyperglycemia and the associated risk of cardiovascular morbidity and mortality. The results of our present study demonstrate that roflumilast not only improves HFD-induced cardiac dysfunction and remodeling but also ameliorates hyperglycemia, glucose intolerance, and insulin resistance. These results are consistent with other human and animal studies showing that roflumilast exhibits cardioprotective effects (11) and may help to reduce hyperglycemia and insulin resistance (9,10,53–55). Given that HF is associated with states of insulin resistance like T2DM, these findings indicate that the potential benefits of roflumilast in diabetes-associated cardiac dysfunction may also in part be related to the improvement of glucose metabolism and insulin sensitivity. Thus, the current study has important translational implications for the use of PDE4 inhibitor roflumilast in T2DM management and opens up a new therapeutic role for an existing drug that is clinically approved for COPD treatment.
However, PDE4D gene inactivation in mice can trigger arrhythmias and lead to the development of HF after myocardial infarction, raising the question of whether PDE4 inhibitor treatment may represent a risk factor for cardiac disease (7). Considered in the context of our observation of the cardioprotective effects of PDE4D inhibition by AAV9 injection or pharmacological inhibitor roflumilast, this discrepancy may have emerged for several reasons. The detrimental effect of PDE4D gene inactivation was observed in myocardial infarction–induced HF, which is different from HF simulated by HFD-induced diabetes in our study. Moreover, drug intervention and AAV9 techniques were shown to inhibit PDE4D expression but not to the same extent as gene inactivation. Karam et al. (56) uncovered that moderate overexpression of PDE4B by AAV9 is cardioprotective, while overexpression of PDE4B with a higher level led to maladaptive remodeling. Similarly, previously studies reported that the effects of cAMP on hypertrophy differ depending on the cAMP concentration (57). cAMP has two downstream targets, exchange protein directly activated by cAMP (Epac) and protein kinase A (PKA). mAKAP organizes a cAMP-responsive network containing PDE4D3, Epac1, and ERK5 that modulates cardiomyocyte size. PKA responds to nanomolar concentrations of cAMP and represses the Epac-mediated block of ERK5, allowing cardiac hypertrophy. However, Epac1 would only become activated once cAMP concentrations reached micromolar levels. Epac1 activator and PDE4 inhibitor rolipram could block cardiac hypertrophy induced by ERK5 activation (57). Together with previous study reporting the bidirectional roles of β2 adrenergic receptor at different expression levels (58), our study suggests that a beneficial or detrimental effect of PDE4D during HF may depend on the expression level of PDE4D. To what extent PDE4D inhibition applies to HF is an important area for further investigation. In addition, PDE4D3 expression is reduced in failing human hearts (59). The PDE4D3 was found in the cardiac ryanodine receptor (RYR2)/calcium release channel complex, which is required for excitation-contraction coupling in heart muscle (59). The reduced levels of PDE4D3 associated with the RYR2 may be responsible for this reduction in PDE4D protein expression in failing human hearts, resulting in defective RYR2 channel function associated with HF and arrhythmias (59). Different from this, PDE4D5 expression increased to a greater extent than PDE4D3 expression in failing diabetic hearts, indicating that PDE4D5 is more important than PDE4D3 in the pathological development of diabetes-associated HF. Furthermore, disruption of PDE4D5-HSP20 complex attenuates the β-agonist–induced hypertrophic response in cardiac myocytes (60). Thus, these PDE4D isoforms serve different biological roles in various types of HF.
Certainly, our study had some limitations. Due to the fact that PDE4 is widely expressed in inflammatory and immune cells, inhibition of PDE4 is an effective way to reduce the activation and recruitment of inflammatory cells and the release of various cytokines (61). Furthermore, PDE4 inhibitor rolipram can protect mice on an HFD against obesity and improve glucose tolerance by activating AMPK and increasing mitochondrial function (4). We cannot exclude the possibility that roflumilast could directly affect signaling pathways in cardiac energy metabolism and in other cell types such as fibroblasts, endothelial cells, adipocytes, and immune cells. More focused mechanistic studies with cardiomyocytes and cardiac fibroblast specific knockout PDE4D mice would also be required for understanding of the interactions between insulin-PDE4D signaling and miR-1 expression in cardiac fibroblasts. Remarkably, there is a paucity of data available on direct actions of PDE4 inhibitors in cardiac function and remodeling in patients with diabetes and HF.
In conclusion, in addition to highlighting a novel mechanism of cardiac dysfunction in diabetes, our data provide a strong rational for targeting PDE4D and miR-1 as an attractive therapeutic modality to treat diabetes-associated cardiac dysfunction, and the findings of this study also support a potential therapeutic role for selective PDE4 inhibitor roflumilast in patients with diabetes with impaired cardiac function.
This article contains supplementary material online at https://doi.org/10.2337/figshare.19723612.
R.X. and J.F. contributed equally to this work.
Article Information
Acknowledgments. The authors thank Dr. Yang K. Xiang (Department of Pharmacology, University of California at Davis) for kind provision of PDE4D5 adenovirus and plasmids.
Funding. This study was supported by National Natural Science Foundation of China grants 81773730 and 81729004 to Q.F.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. Q.F. designed the study and wrote the manuscript. R.X., J.F., Y.H., and X.T. performed experiments in the animal model. R.X. and J.F. conducted experiments, performed data analyses, and wrote the manuscript. X.Y. analyzed data and provided critical insights. L.C. conducted cardiac function measurements. K.H. provided the human samples and critical insights. All authors approved the manuscript. Q.F. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.