Inhibition of Dipeptidyl Peptidase-4 Impairs Ventricular Function and Promotes Cardiac Fibrosis in High Fat–Fed Diabetic Mice

Patients with type 2 diabetes are at increased risk for the development of cardiovascular disease (CVD) (1,2), including peripheral vascular disease, coronary artery disease, and heart failure (HF) (3). Although the use of antidiabetic agents to lower glycemia results in the reduction of microvascular disease, the rates of macrovascular disease remain elevated and take much longer to decline with glucose-lowering therapies. Heightened concern over the possible cardiovascular side effects of thiazolidinediones, notably rosiglitazone, prompted the development of regulatory guidance for evaluation of the cardiovascular safety of glucose-lowering agents (4). Hence, sponsors now carry out large cardiovascular outcome studies in subjects with diabetes who have established CVD or are at high risk for CVD.

Two classes of antidiabetic agents, designated incretin therapies, exert their glucoregulatory effects through the potentiation of gut hormone action (5). Dipeptidyl peptidase-4 inhibitors (DPP4is) control glycemia by preventing the degradation of glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide, resulting in the stimulation of insulin and inhibition of glucagon secretion. GLP-1 receptor (GLP-1R) agonists potentiate GLP-1R signaling, leading to increased insulin levels; reduced glucagon secretion, gastric emptying, and appetite; and modest weight loss. Dozens of preclinical studies (6,7) have demonstrated that incretin therapies reduce atherosclerosis and vascular injury, decrease inflammation, and are cardioprotective in experimental models of CVD. Furthermore, incretin therapies reduce blood pressure and decrease postprandial lipemia in subjects with type 2 diabetes and exert cardioprotective actions in short-term clinical studies (6,7). Moreover, meta-analyses of clinical trials (6,7) in subjects with diabetes without established CVD suggest that these agents may reduce cardiovascular event rates.

Conversely, outcome studies (8–10) have demonstrated cardiovascular safety, but no evidence for the reduction of cardiovascular event rates, with three different DPP4is, saxagliptin, alogliptin, and sitagliptin. Unexpectedly, subjects treated with saxagliptin in the Saxagliptin Assessment of Vascular Outcomes Recorded in Patients with Diabetes Mellitus—Thrombolysis in Myocardial Infarction 53 (SAVOR-TIMI 53) study exhibited a small but significant increase in the rate of hospitalization for HF (9). Furthermore, patients receiving alogliptin in the Examination of Cardiovascular Outcomes with Alogliptin versus Standard of Care (EXAMINE) trial exhibited a small numeric excess of HF events (10,11).

We have now examined the cardiovascular responses of younger Dpp4^−/− mice and older, dysglycemic, high fat–fed mice after transverse aortic constriction (TAC) induction of experimental HF. Genetic reduction of DPP4 activity was cardioprotective in euglycemic Dpp4^−/− mice with experimental TAC-induced HF. In contrast, pharmacological DPP4 inhibition was not cardioprotective in older, dysglycemic, dyslipidemic mice. Unexpectedly, we detected evidence for increased inflammation in Dpp4^−/− hearts, and gene expression profiles consistent with both inflammation and fibrosis in hearts from mice treated with a selective DPP4i. Collectively, these findings reveal that DPP4 inhibition is not uniformly cardioprotective in animal studies and provide an experimental model for ongoing analysis of how the reduction of DPP4 activity may promote cardiac fibrosis and inflammation and impair ventricular function in preclinical studies of HF.

Mice were housed under a 12-h light/dark cycle in the Toronto Centre for Phenogenomics facility and maintained on chow (Teklad 2018 [18% kcal from fat]; Harlan, Mississauga, Ontario, Canada) with free access to food and water, unless otherwise noted. Dpp4^−/− mice (Supplementary Fig. 1) were described previously (12); for doxorubicin experiments, n = 4–12 mice per group for vehicle controls and 9–12 mice per group for doxorubicin treatment. For TAC studies in Dpp4^+/+ vs. Dpp4^−/− vs. mice, n = 6–7 per group. All experiments (Supplementary Fig. 2) used age- and sex-matched littermates. C57BL/6J mice were from The Jackson Laboratory (Bar Harbor, ME). Male C57BL/6J mice, 8–12 weeks of age, were housed two to three per cage and were fed ad libitum for 12 weeks (n = 12–15/group) with a high-fat diet (HFD) (45% kcal fat, 35% kcal carbohydrate, 0.05% w/w cholesterol; catalog #D12451; Research Diets). After 12 weeks of HFD feeding, mice were fasted for 5 h and injected with a single dose of streptozotocin (STZ) (90 mg/kg i.p., freshly prepared solution in 0.1 mmol/L sodium citrate, pH 5.5). One week later, mice were stratified by glucose and body weight and were randomized to three groups that were all maintained on the HFD. Two groups were injected once daily with either saline or liraglutide (30 µg/kg; Novo Nordisk). A third group was administered the DPP4i MK-0626, (2S,3S)-1-(3,3-difluropyrrolidin-1-yl)-4-(dimethylamino)-1,4-dioxo-3-(4-[1,2,4]triazolo[1,5-A]pyridin-6-ylphenyl)butan-2-aminium chloride, designated compound 6b in the study by Edmondson et al. (13), mixed into the HFD (18 mg/kg diet or 3 mg/kg body wt; Merck Laboratories, Rahway, NJ) and also injected daily with saline solution. MK-0626 is from a related yet different structural class than sitagliptin or des-fluoro-sitagliptin (13). We chose to use MK-0626 because it is a highly selective α-amino amide–derived DPP4i (13) previously used for the analysis of cardiac function in rodents (14). Doses of MK-0626 and liraglutide were based on pilot studies (13,15) to achieve equivalent glucose control without reduction of food intake or body weight. Mice were fasted for 4 h prior to sacrifice. All experiments were approved by the Animal Care and Use Subcommittee at the Toronto Centre for Phenogenomics, Mount Sinai Hospital.

Buprenorphine was administered 1 h preemptively (0.1 mg/kg s.c.) and every 12 h for 3 days after surgery. Mice were anesthetized with 3% isoflurane and intubated via insertion of a 20-gauge intravenous catheter attached to a connector into the trachea. The transverse aortic arch was banded between the innominate artery and the left common carotid artery with a 27-gauge (diabetic C57BL/6J mice) or 28-gauge (Dpp4^+/+and Dpp4^−/−) needle using 7–0 silk suture. Sham-operated mice underwent the same procedure by moving the suture behind the transverse aortic arch but without tying it off.

Heart rate was assessed via implanted radiotelemetry devices (PA-C10; DSI) (16). Blood pressure was assessed in conscious mice using the CODA system (Kent Scientific, Torrington, CT) (17). Body composition was assessed by an EchoMRI quantitative nuclear magnetic resonance system (Echo Medical Systems, Houston, TX). Oral glucose tolerance tests were performed after an overnight fast, as previously described (16). Exercise capacity was performed by running animals on a calibrated, motor-driven treadmill (Columbus Instruments), as previously described (18).

Mice were killed by injection with avertin, and hearts were frozen in liquid nitrogen–cooled Wollenberger tongs (16). Cardiac blood was collected, and plasma was prepared and stored at −80°C. Tissues were removed, snap frozen in liquid nitrogen, and stored at −80°C. Plasma insulin level (ELISA; ALPCO Diagnostics, Windham, NH), HbA_1c level (DCA 2000+ Analyzer; Bayer, Toronto, Ontario, Canada), DPP4 activity (substrate 10 mmol/L H-Gly-Pro-AMC HBr [catalog #I-1225; BACHEM], standard AMC [catalog #Q-1025; BACHEM]), and active GLP-1 7-36 level (immunogenicity assay; Mesoscale) were determined as per manufacturer instructions. Plasma triglycerides (Roche Diagnostics, Laval, Quebec, Canada) and plasma cholesterol (Wako Chemicals, Richmond, VA) were measured enzymatically (19).

Immunoblotting and histological analysis (a separate cohort n = 3–4) were performed as previously reported (16) using antibodies listed in Supplementary Table 1, with the exception of collagen type 1a1 (Col1a1), which was run under nonreducing and nondenaturing conditions. The heart was processed and cut into three equal sections across a vertical axis. Cardiac morphometry was performed with Aperio ImageScope Viewer software (Leica Biosystems) on all three sections of the heart using digital planimetry. For analysis of trichrome staining, settings were modified such that the Positive Pixel Algorithm was set to a hue value of 0.62, a hue width of 0.40, and a color saturation threshold of 0.005.

Gene expression was performed using primer probe sets manufactured by Applied Biosystems TaqMan Assays-on-Demand (ThermoFisher Scientific) (Supplementary Table 2). Protein levels of MCP-1 (mouse) and transforming growth factor-β1 (TGF-β1) (human) were quantified in heart extracts using Cytometric Bead Array assays (BD Biosciences) (20).

Data are presented as the mean ± SEM. A two-way ANOVA was conducted to test the interaction between surgery/doxorubicin and genotype/treatment, followed by a post hoc Holm-Šidák test or a one-way ANOVA was used with post hoc Tukey test. Statistical analysis was performed using SigmaPlot version 14.0. Data were considered statistically significant at P ≤ 0.05.

We first determined that our Dpp4^−/− mice used for cardiovascular studies retained the originally described (21) metabolic phenotypes of improved glucose tolerance, elevated GLP-1 levels, and reduced DPP4 activity (Supplementary Fig. 1A–D), without changes in fasting blood glucose levels (Supplementary Fig. 1E). Furthermore, Dpp4^−/− mice exhibited lower body weights and reduced adipose tissue mass, as previously reported (Supplementary Fig. 1F and G) (22). An ultrasound echocardiography assessment demonstrated normal baseline myocardial wall dimensions and cardiac function in young Dpp4^−/− versus Dpp4^+/+ littermate control mice (Supplementary Fig. 1H–K).

To ascertain the susceptibility of normoglycemic Dpp4^−/− mice to the development of HF, we studied mice with 1) pressure overload–induced hypertrophy via TAC surgery and 2) chemically induced cardiomyopathy via administration of the antineoplastic agent doxorubicin (Supplementary Fig. 2). TAC surgery was performed at 10–12 weeks of age. We did not observe differences in survival between Dpp4^−/− and Dpp4^+/+ littermate control mice after TAC surgery; the survival rate was >90%. At 8 weeks post TAC, left ventricular (LV) mass was increased to a greater extent in Dpp4^+/+ vs. Dpp4^−/− mice (Fig. 1A, P = 0.075). The reduction in LV ejection fraction (LVEF) and LV fractional shortening (Fig. 1B and C) and dilation of the LV cavity (Fig. 1D–G and Supplementary Table 3) were significantly greater in Dpp4^+/+ compared with Dpp4^−/− mice. No differences in lung weight/tibia length ratio (Fig. 1H) or cardiac levels of atrial natriuretic peptide (Nppa) and B-type natriuretic peptide (Nppb) mRNA transcripts (Fig. 1I and J) were detected after TAC. The preservation of cardiac function in Dpp4^−/− mice was associated with a significant improvement in distance and time run on a treadmill prior to exhaustion 8 weeks after TAC surgery (Fig. 1K and L).

After doxorubicin administration, we did not observe differences in survival (≥80% in all groups); however, greater weight loss was observed in Dpp4^+/+ compared with Dpp4^−/− mice (Supplementary Fig. 3A). Doxorubicin significantly decreased heart weight and LV mass (Supplementary Fig. 3B and C) and reduced heart dimensions (LV systolic dimension [LVDs], LV diastolic dimension [LVDd]) in Dpp4^+/+ mice, but not in Dpp4^−/− mice (Supplementary Fig. 3D and E). Significant elevations in the levels of cardiac mRNA transcripts encoding Nppb, Nppa (Supplementary Fig. 3F and G), TGF-β2 (Tgfb2), matrix metallopeptidase 9 (Mmp9), and tissue inhibitor of metalloproteinase 1 (Timp1) and diminished levels of Col1a1 expression were observed in hearts from both Dpp4^+/+and Dpp4^−/− mice after doxorubicin treatment (Supplementary Fig. 4A–D). Reductions in the phosphorylation of extracellular signal–related kinase (ERK) 1/2 and increased levels of phosphorylated Akt (after doxorubicin) were similar across genotypes (Supplementary Fig. 4E and F).

Masson trichrome staining intensity trended lower, and ventricular protein levels of Col1a1 were significantly decreased in hearts from Dpp4^−/− mice post TAC surgery (Fig. 2A and B). No perturbation in fibrosis-related gene expression profiles was detected (Fig. 2C–G). Unexpectedly, levels of gene products expressed by cardiac fibroblasts such as vimentin (Vim) and the discoidin domain receptor family 2 (Ddr2) (Fig. 2H and I) were increased in hearts of Dpp4^−/− mice.

Despite improved cardiac function and reduced fibrosis, we detected significant increases in the expression of genes encoding regulators and effectors of the immune response regulating cardiac inflammation in Dpp4^−/− mouse hearts, including macrophage chemoattractant proteins 1 and 2 (Ccl2 and Ccl8, respectively), ventricular MCP-1 protein (Fig. 3A–C), interleukin-1β (Il1b), growth differentiation factor 5 (Gdf5), and growth differentiation factor 15 (Gdf15) (Fig. 3D–F). No differences were observed for Tnf (Fig. 3G). Furthermore, Dpp4^−/− hearts exhibited increased mRNA levels of 1) the macrophage markers cluster of differentiation (Cd11b-Itgam) and Cd68 (P = 0.06) (Fig. 3H and I), and 2) markers of infiltrating hematopoietic cells (Cd34) (Fig. 3J). Histological staining for CD34 demonstrated increased infiltration of cells of hematopoietic origin in Dpp4^−/− hearts (Fig. 3K). No differences in the expression of hypoxia markers, including heme oxygenase-1 (Hmox1), inducible nitric oxide synthase (Nos2), endothelial nitric oxide synthase (Nos3), hypoxia-inducible factor 1α (Hif1a), vascular endothelial growth factor α (Vegfa), or interleukin-6 (Il6) were evident in Dpp4^−/− versus Dpp4^+/+ hearts (Supplementary Fig. 5A–F).

To verify the selectivity of MK-0626 (13), we performed an oral glucose tolerance test (Supplementary Fig. 6). Glucose excursions were reduced by 50% in Dpp4^+/+ mice, but not in Dpp4^−/− mice. We next examined the cardiac responses to TAC surgery in HFD-fed mice with low-dose STZ-induced diabetes after prolonged pharmacological DPP4 inhibition or GLP-1R agonism (Supplementary Fig. 2). Mice treated with liraglutide or MK-0626 had improved glucose tolerance, decreased HbA_1c and fasting glucose levels, and improved glucose-stimulated insulin levels (Supplementary Fig. 7A–D). MK-0626 markedly diminished plasma DPP4 activity and increased active GLP-1 levels assessed before and after an oral glucose load (Supplementary Fig. 7E and F). Plasma triglyceride levels, but not cholesterol levels, were reduced by liraglutide and MK-0626 (Supplementary Fig. 7G and H). Unexpectedly, mice treated with MK-0626 demonstrated increased body weight (Supplementary Fig. 7), food intake (Supplementary Fig. 7J), and fat mass (Supplementary Fig. 7K and L).

Echocardiography performed 8 weeks post TAC surgery revealed significant increases in LV mass and heart weight only in mice treated with MK-0626 (Fig. 4A and B); however, survival was not different among groups, and mortality was <20%. Lung weight/tibia length ratios were not significantly increased in MK-0626–treated mice (P = 0.09) (Fig. 4C). LVEF was significantly lower in MK-0626–treated mice but not in mice treated with liraglutide (Supplementary Table 4 and Fig. 4D). Gene expression analysis demonstrated a relatively greater increase in levels of cardiac Nppa mRNA transcripts in MK-0626–treated mice, whereas levels of Nppb were significantly elevated only in saline-treated mice after TAC surgery (Fig. 4E and F). No changes in mean tail-cuff blood pressure were observed among any of the groups (Fig. 4G). Mice treated with saline or liraglutide, but not MK-0626, demonstrated increased exercise tolerance, running for greater distances (and durations) prior to exhaustion (Fig. 4H and I). Consistent with previous findings (16), heart rate was elevated after liraglutide injection and was also higher in mice chronically treated with MK-0626 (Fig. 4J).

Surprisingly, levels of cardiac mRNA transcripts for collagen type I, alpha 1 (Col1a1); collagen type III, alpha 1 (Col3a1); and fibronectin (Fn) were elevated in hearts from MK-0626–treated mice (Fig. 5A–C). Hearts from MK-0626–treated mice also demonstrated nonsignificant increases in trichrome staining and ventricular Col1a1 protein levels compared with saline-treated controls (Fig. 5D and E). α-Actin (Acta2) levels were elevated in all TAC surgery groups, whereas levels of gene products associated with collagen turnover, Mmp2 and Timp1, were significantly increased in hearts from MK-0626–treated mice (Fig. 5F–H). Mmp9 expression was reduced in all groups in response to TAC surgery (Supplementary Fig. 8A).

Key inducers of cardiac fibrosis, such as Tgfb1, Tgfb2, and angiotensinogen (Agt) (Fig. 6A and B and Supplementary Fig. 8B) were significantly increased in hearts from mice treated with MK-0626. Because TGF-β is secreted as a latent inactive complex, we also measured activated ventricular TGF-β1 protein, which trended higher in hearts from MK-0626–treated mice (Fig. 6C). No differences were observed in cardiac expression of ACE 2 (Ace2), endothelin-1 (Edn1), and cardiotrophin-1 (Ctf-1) (Supplementary Fig. 8C–E). Hearts from MK-0626–treated mice exhibited increased cardiac levels of phosphorylated glycogen synthase kinase (GSK)-3β (Fig. 6D and M) and a nonsignificant increase in both phosphorylated ERK and Akt (Fig. 6E, F, and M).

As cardiac hypertrophy is associated with nuclear localization of Yes-associated protein (YAP), we assessed both the nuclear localization of YAP and the phosphorylation of YAP on residue Ser127, which results in cytosolic retention; changes in these parameters were not statistically significant between groups (Supplementary Fig. 9 and Fig. 6G and M). Other known TGF-β and Hippo signaling targets, including phosphorylation of small mother against decapentaplegic (Smad2) (Ser 465/467) and stabilization of β-catenin, were unchanged between groups (Fig. 6H and M). Conversely, the expression of some gene targets of these signaling pathways, specifically ankyrin repeat domain–containing protein 1 (Ankrd1), cysteine-rich angiogenic inducer (Cyr61), and connective tissue growth factor (Ctgf) were elevated in hearts from dysglycemic mice treated with MK-0626 compared with liraglutide (Fig. 6J–L).

Intriguingly, Tnf was upregulated in hearts from MK-0626–treated diabetic mice (Fig. 7A), whereas both MK-0626 and liraglutide lowered cardiac Il6 expression compared with saline-treated mice (Fig. 7B). In contrast with findings in Dpp4^−/− mice, we did not observe differences in cardiac Il1b, Ccl2, Ccl8, and Gdf15 expression (Supplementary Fig. 10A–D). We detected increased expression of the cardiac fibroblast collagen receptor Ddr2, but not Vim (Fig. 7C and Supplementary Fig. 10E), and increased levels of Itgam (P = 0.12), Cd34, and Cd68 were also observed in cardiac RNA from MK-0626–treated mice (Fig. 7D–F). Furthermore, increased expression of Hif1a, Nos3, Hmox1, Vegfa, and Nos2, genes transducing the response to hypoxia, was evident in hearts after MK-0626 treatment (Supplementary Fig. 10F–J).

The unexpected report of increased hospitalization for HF in a clinical trial assessing the cardiovascular safety of saxagliptin (9) prompted us to assess the consequences of genetic Dpp4 elimination or selective DPP4 inhibition in murine models of HF. Previous studies (12) demonstrated that Dpp4^−/− mice exhibit a cardioprotective phenotype, encompassing improved survival and improved ventricular function, in response to coronary artery ligation or ischemia-reperfusion injury ex vivo. The current findings demonstrate that whole-body, germline disruption of the Dpp4 gene in young, healthy, euglycemic mice attenuates the extent of adverse LV remodeling and preserves ventricular function after TAC-induced pressure overload HF or in response to doxorubicin administration. Whether the cardiovascular benefits attributable to genetic loss of DPP4 activity in preclinical studies are secondary to direct or indirect effects of reduced DPP4 activity and modulation of DPP4 substrate activity in the cardiovascular system (6) or, alternatively, are influenced in part by the favorable metabolic phenotypes arising in Dpp4^−/− mice (21,22) has not yet been ascertained.

In contrast to the cardioprotective phenotypes exhibited by euglycemic Dpp4^−/− mice, prolonged treatment of older HFD-fed diabetic mice with MK-0626 and subjected to TAC did not improve cardiac function. Myocardial dysfunction and cardiomyocyte injury are commonly detected after the administration of STZ; however, whether this reflects the direct effects of STZ on the myocardium or arises independently from changes in the metabolic environment remains uncertain (23,24). MK-0626 is a highly selective orally available DPP4i (13) that produces sustained reduction of DPP4 activity in vivo. Our current findings were unexpected, as long-term therapy with MK-0626 in mice fed a HFD/high-fructose diet preserved cardiac structure and function, in association with decreased cardiac fibrosis (14). Notably, in a similar model, albeit with the added superimposed stress of TAC-induced HF, we observed the following: 1) relative impairment of ventricular function, 2) evidence for greater ventricular hypertrophy, and 3) increased interstitial fibrosis in diabetic mice undergoing long-term treatment with MK-0626 post TAC surgery. Furthermore, mice treated with saline or liraglutide, but not MK-0626, exhibited increased functional exercise capacity (distance and time, Fig. 4H and I) after TAC through mechanisms that require further elucidation. The increase in cardiac hypertrophy in hearts from MK-0626–treated mice was accompanied by a significant induction of cardiac Nppa and an increase in the heart weight/tibia length ratio. In contrast, mice treated with liraglutide had no impairment of contractile function and no induction of fibrosis-related gene expression in response to pressure overload. Hence, the potentiation of GLP-1R signaling likely does not account for the adverse cardiovascular actions of MK-0626 in our studies.

The lack of improvement in ventricular function after treatment with a DPP4i in our current studies is not unique to experiments using MK-0626. Indeed, long-term administration of vildagliptin to normoglycemic rats both prior to and after the induction of myocardial infarction and HF did not attenuate the reduction of LVEF or modify cardiac remodeling (25). Similarly, the administration of saxagliptin to nondiabetic mice with HF secondary to cardiac-specific overexpression of Cre recombinase did not improve ventricular function or modify survival (26). Nevertheless, previous preclinical studies have not revealed a significant impairment of ventricular function after prolonged DPP4 inhibition.

Consistent with reduced interstitial fibrosis in hearts of MK-0626–treated C57BL/6J mice fed a HFD/high-fructose diet for 16 weeks (14), euglycemic Dpp4^−/− mice also exhibited a significant reduction in cardiac fibrosis and no induction of fibrosis-related gene expression post TAC surgery. Considerable evidence also links the attenuation of DPP4 activity to improvement in ventricular function and reduced systemic and cardiac inflammation (6,27–29). Hence, we were surprised to detect increased cardiac expression of mRNA transcripts encoding multiple effectors of the immune response in normoglycemic mice with reduced DPP4 activity. Analysis of RNA from euglycemic Dpp4^−/− mouse hearts revealed elevated levels of inflammation-related mRNA transcripts, including the macrophage chemokines Ccl2, Ccl8, and Il-1b, as well as members of the TGF-β chemokine family, Gdf5 and Gdf15, but no differences in Tnf were observed. Furthermore, we observed increased levels of mRNAs corresponding to 1) markers of cardiac fibroblasts, including Ddr2 (30) and Vim (31), and 2) infiltrating cells of bone marrow origin, such as Cd34, Cd68, and Itgam. Protein levels of MCP-1 and CD34 were also elevated in Dpp4^−/− mice compared with littermate controls.

Although the precise mechanisms linking DPP4 inhibition to the development of cardiac fibrosis in our mouse model are incompletely understood, we hypothesize that the combination of hyperglycemia, an HFD, prolonged DPP4 inhibition in older mice, and the effects of pressure overload may combine to alter the milieu of growth factors, cytokines, and neurohumoral factors to promote a fibrogenic gene expression program. Although the clinical relevance of a single dose of STZ and 5 months of high-fat feeding in mice is questionable, diabetes is a significant risk for the development of organ fibrosis (32), and our model replicates the development of diabetes-associated cardiac fibrosis within a reasonable time frame.

Cardiac fibroblasts are responsible for the deposition of proteins encoded by Col1a1, Col1a3, and Fn1 to maintain the integrity of the heart. All three gene products were elevated in diabetic MK-0626–treated mice, as was the cardiac fibroblast gene product Ddr2. Elevated levels of TGF-β are associated with both fibrosis and cardiac hypertrophy (33,34). We evaluated signaling through canonical and noncanonical mediators of fibrosis and hypertrophy in response to TGF-β, including ERK, GSK-3β/β-catenin (35), pSMAD2/SMAD2 (36), Akt (37), and YAP (38). Although we did not observe any robust changes in these effectors, these pathways are known to be activated in the short term in response to pressure overload and remain only moderately elevated at later time points (34,39). Therefore, in our assessment of hearts 12 weeks post TAC surgery, these pathways may not remain significantly perturbed. However, levels of direct transcriptional targets that have a known role in the progression of HF (Col1a1, Col3a1, Fn1, Ankrd1, Cyr61, and Ctgf) were all elevated, suggesting a role for TGF-β/Hippo pathway signaling in mediating the aberrant cardiac remodeling observed in mice treated with MK-0626.

Intriguingly, multiple DPP4 substrates increase cell proliferation and promote fibrosis, including neuropeptide Y, protein YY (40,41), and stromal-derived factor-1 (42). Furthermore, exposure to high glucose upregulates a transcriptional program enabling cardiac fibroblast proliferation, collagen synthesis, and cardiac fibrosis (43–45); upregulation of local angiotensin-II expression has also been linked to cardiac fibrosis (46). Moreover, high glucose levels and increased levels of free fatty acids exert complementary effects to promote cardiomyocyte Ctgf expression, which in turn may independently promote cardiac fibrosis. The plausibility of DPP4 activity controlling the extent of cardiac fibrosis is further supported by data demonstrating that DPP4 inhibition potentiated the neuropeptide Y–dependent and protein YY–dependent regulation of cell proliferation and collagen synthesis in cardiac fibroblast cultures (41). Hence, we hypothesize that, under conditions characterized by sustained hyperglycemia, dyslipidemia, and pressure overload in older animals, reduction of DPP4 activity may further accelerate the development of cardiac fibrosis.

DPP4 is known to bind to the extracellular matrix and cleave collagen (47,48); however, the importance of DPP4 catalytic activity for collagen degradation has not been extensively studied in the heart. Although short-term systemic reduction of DPP4 activity altered the metabolism of circulating collagen-derived peptides in rats in vivo (49), whether DPP4 activity controls collagen metabolism in a heart subjected to metabolic stress and pressure overload has not been determined. Furthermore, DPP4, together with seprase, regulates fibroblast migration and invasion in the presence of a collagen matrix, with proteolytic activity of the DPP4-seprase complex modulating gelatin degradation and wound healing in a cell culture model ex vivo (50). We also detected reductions in cardiac Mmp9 expression in mice after TAC surgery, and elevations in Mmp2 and its inhibitor Timp1 in hearts from MK-0626–treated mice that together may promote an imbalance in the degradation of extracellular matrix.

Our findings establish a preclinical model, encompassing long-term exposure to an HFD, experimental diabetes, and pressure overload–induced cardiac dysfunction, for understanding how the reduction of DPP4 activity may promote cardiac fibrosis and inflammation and impair ventricular function in older diabetic rodents. A major limitation of the current study is the use of only one DPP4i, MK-0626, in a single mouse model of cardiac dysfunction. Hence, our results cannot be generalized or extrapolated to other clinically used DPP4is. Our data provide an experimental preclinical model for further investigation of the pleiotropic cardiovascular mechanisms of actions of selective DPP4is in diabetic mice with impaired cardiac function.

Acknowledgments. The authors thank Dawei Qu and the Centre for Modeling Human Disease at the Lunenfeld-Tanenbaum Research Institute and Cindy Sawyez and the Metabolic Phenotyping Laboratory Core at the Robarts Research Institute for their excellent technical support. The authors also thank Dr. Masahiro Narimatsu and Dr. Jeff Wrana at the Lunenfeld-Tanenbaum Research Institute for helpful discussions.

Funding. This work was funded by the Heart and Stroke Foundation of Canada grant G-14-0005953. E.E.M. has received fellowship funding from the Canadian Diabetes Association and the Canadian Institutes of Health Research. J.R.U. was supported by fellowships from the Canadian Institutes of Health Research and the Alberta Innovates–Health Solutions. J.E.C. has received fellowships from the Banting & Best Diabetes Centre, the University of Toronto, and the Canadian Institutes of Health Research. D.J.D. is supported by a Canada Research Chair in Regulatory Peptides and a Banting & Best Diabetes Centre Novo Nordisk Chair in Incretin Biology.

Duality of Interest. J.R.U. has received a speaker's honorarium for symposia sponsored by Novo Nordisk. L.L.B. has served as a panel member and/or received a speaker's honorarium for symposia sponsored by Novo Nordisk, Merck Frosst, and GlaxoSmithKline LLC. D.J.D. has served as an advisor or consultant to Arisaph Pharmaceuticals, Intarcia, Merck Research Laboratories, MedImmune, Novo Nordisk, Receptos Inc., and Sanofi Inc. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. E.E.M., E.M.V., J.R.U., J.E.C., and L.L.B. designed and carried out the experiments, analyzed the data, and wrote the article. K.W.A.B. and T.A. designed and carried out the experiments and analyzed the data. D.J.D. designed and analyzed the experiments and wrote the article. D.J.D. 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.