Cardiac Tissue Factor Regulates Inflammation, Hypertrophy, and Heart Failure in Mouse Model of Type 1 Diabetes

The prevalence of diabetes is on the rise, with a World Health Organization–estimated global prevalence of 330 million in 2025 (1). Cardiovascular complications are the leading cause of diabetes-related morbidity and mortality (2). Diabetes is a major risk factor for the development of heart failure (HF), causing a twofold increased risk of HF in men and fivefold increased risk in women (2,3). In addition to reduced survival, patients with diabetes also have a two- to threefold increased risk of developing HF after myocardial infarction (4). HF is divided into at least two distinct entities: HF with reduced ejection fraction and HF with preserved ejection fraction (HFpEF). Nearly 50% of patients with HF have HFpEF, and the prevalence of diabetes in HFpEF is 30–40%. HFpEF, believed to be an inflammatory cardiometabolic disease (5), remains a challenge presently, because no treatment can significantly reduce morbidity or mortality in these patients (6).

Tissue factor (TF), also known as thromboplastin or factor III, is a transmembrane glycoprotein receptor expressed on cardiac cells. It has been shown to play a role in various processes, including inflammation, angiogenesis, metastasis, and cell migration (7). TF is essential for maintaining hemostasis (8). In the adult heart, TF is highly expressed in the left ventricle (LV) by cardiomyocytes, fibroblasts, and adventitial and perivascular cells (9). TF levels are altered in different pathological conditions. For example, patients with dilated cardiomyopathy, sepsis, or endotoxemia have reduced cardiac TF expression (9–11). In ischemia-reperfusion injury, inhibition of cardiac TF levels reduces inflammation and myocardial injury (12). In patients with hypertension and ventricular hypertrophy, TF level was reduced in structurally altered ventricular myocardium (9). Mice expressing very low levels of human TF showed ventricular dysfunction and myocardial fibrosis with increasing age (13). Elevated levels of TF have been observed in blood from patients with diabetes, atherosclerosis, and acute coronary syndromes or coronary artery disease (14–17). Patients with type 2 or type 1 diabetes had elevated levels of whole blood TF procoagulant activity compared with healthy participants without diabetes (18). However, the role of cardiac TF in diabetes-induced inflammation, hypertrophy, and HF has not been studied. We hypothesized that TF function in the heart in diabetes may cause cardiac remodeling and dysfunction, the hallmark features of HF. The aim of the current study was to understand the TF-dependent mechanisms of HF in the mouse model of streptozotocin (STZ)–induced type 1 diabetes. Accomplishing this aim will help us understand the effects of cardiac TF on biological processes that cause cardiac remodeling and dysfunction in diabetes. Our in vivo data showed upregulation of TF in the diabetic hearts compared with nondiabetic controls, and this was associated with increased cardiac inflammation, hypertrophy, and HFpEF. In contrast to the wild-type (WT) diabetic hearts, low-TF hearts showed reduced inflammation, no hypertrophy, and improved cardiac function, suggesting that inhibition of TF can improve diabetes-induced cardiac dysfunction. Our study presents a novel mechanism of regulation of cardiac remodeling and dysfunction by TF in diabetes-induced HF.

In the current study, we used male WT, heterozygous (TF-Het; mTF^+/−hTF⁺), and low-TF mice (mTF^−/−hTF⁺) on a C57B/6J genetic background, maintained through inbreeding, as mice with diabetes (STZ) and control mice without diabetes. The low-TF mice, as described previously (13), expressed 1% of human TF in the presence of null mouse TF background, because the complete TF-knockout (KO) condition is embryonically lethal. Diabetes was induced by a moderate dose of STZ injection (75 mg/kg body weight) in 8-week-old fasting (6 h) mice on every alternate day for 3 days. The littermate control mice without diabetes received an equal volume of vehicle (sodium citrate). This study included WT control (n = 14) and STZ (n = 21), Het control (n = 13) and STZ (n = 12), and low-TF control (n = 12) and STZ (n = 14) mice. After 3 days from the last STZ injection, mice developed diabetes, with nonfasting blood glucose levels >180 mg/dL. Body weight and blood glucose levels were recorded randomly during the course of the study (Supplementary Fig. 1E and F). Cardiac function was measured in all mice at 12 weeks of diabetes, followed by harvesting of the heart tissues. The DukeNUS and SingHealth Institutional Animal Care and Use Committee approved all animal protocols.

Total RNA was isolated from heart LV samples using TRIzol (catalog number 15596-018; Life Technologies). The cDNA synthesis was performed from 1 μg RNA using random hexamers and the SuperScript III First-Strand Synthesis System (catalog number 18080-051; Life Technologies), and HOT FIREPol EvaGreen qPCR Supermix was used for quantitative RT-PCR using 7900HT Sequence Detection System. The results were analyzed by SDS 2.0 software (Applied Biosystems), and gene expressions were normalized to corresponding β-actin controls.

Snap-frozen diabetic and nondiabetic control heart (LV) tissues from WT, TF-Het, and low-TF mice were homogenized by using radioimmunoprecipitation assay lysis and extraction buffer (catalog number 89901; Thermo Fisher Scientific) along with protease (catalog number P8340; Sigma-Aldrich) and phosphatase inhibitors (catalog number 88667; Thermo Fisher Scientific), and total protein was quantified using the Pierce BCA Kit (catalog number 23225; Thermo Fisher Scientific). For cell culture samples, the cells were washed with cold Dulbecco’s PBS (catalog number 17-512F; Lonza) and lysed with radioimmunoprecipitation assay buffer. All experiments were repeated three times and quantified. Equal amounts of protein (30–50 μg) were loaded in 6–10% SDS-PAGE gels, and the blots were developed using the WEST-Queen Western Blot Detection System (catalog number 16026; iNtRON Biotechnology). The primary antibodies used for immunoblot analyses were TF (catalog number sc-374441; Santa Cruz Biotechnology), interleukin-6 (IL-6) (catalog number sc-57315; Santa Cruz Biotechnology), Toll-like receptor 4 (TLR4) (catalog number sc-293072; Santa Cruz Biotechnology), Bcl-2 (catalog number CPTC-BCL2L2-1C; Developmental Studies Hybridoma Bank), phosphorylated extracellular signal–regulated kinase 1/2 (pERK1/2) and ERK1/2 (catalog numbers 9101S and 9102S, respectively; Cell Signaling), phosphorylated STAT3 (pSTAT3) and STAT3 (catalog numbers sc-8059 and sc-8019, respectively; Santa Cruz Biotechnology), phosphorylated p38–mitogen-activated protein kinase (MAPK) and p38-MAPK (catalog numbers 4631 and 9212, respectively; Cell Signaling Technology), GAPDH (catalog number sc-20357; Santa Cruz), β-actin (catalog number sc-47778; Santa Cruz Biotechnology), and vinculin (catalog number V9131; Sigma-Aldrich).

Histology and immunohistochemistry were performed as described previously (19,20). In brief, hearts were harvested from mice with diabetes and control mice. The samples were processed for paraffin embedding and sectioned at 5–10 μm thickness. For immunohistochemistry, the sections were incubated with primary antibodies (anti-TF mouse monoclonal antibody [catalog number sc-374441; Santa Cruz], normal mouse IgG [catalog numbersc-2025; Santa Cruz Biotechnology], anti-TLR4 mouse monoclonal antibody [catalog number sc-293072; Santa Cruz Biotechnology], and anti–Bcl-xL mouse monoclonal antibody [catalog number sc-8392; Santa Cruz Biotechnology]) at 4°C overnight and were also counterstained with hematoxylin. Detection of horseradish peroxidase activity was performed using a DAB kit (catalog number SK-4100; Vector Laboratories). For wheat germ agglutinin (WGA) immunofluorescence, the frozen sections were permeabilized and washed with 1% Triton X-100 in PBS. Sections were incubated with WGA antibody (1:50 dilution) for 3 h followed by nuclear staining with DAPI for 30 min, followed by mounting with 80% glycerol. Hearts from four to six mice from each group (diabetes vs. nondiabetes) for each genotype were used, and at least five different regions of each section were studied. All images were quantified using ImageJ software, and the cardiomyocyte cross-sectional areas (CSA) for WGA staining were analyzed using GraphPad Prism version 8.0.

Transthoracic echocardiography was performed on all mice using Vevo 2100 (Visual Sonics, Toronto, Ontario, Canada) with an MS400 linear array transducer at 18 to 38 MHz. All animals were anesthetized using 5% isoflurane and maintained at 0.6–1% isoflurane. An average of 10 cardiac cycles with standard two-dimensional parasternal long and short axes (mid–papillary muscle level), as well as pulsed-wave Doppler recording of the mitral inflow, were obtained and stored for subsequent offline analysis. LV end-diastolic volume and end-systolic volume were measured at the parasternal long axis. Stroke volume was then calculated by subtracting end-systolic volume from end-diastolic volume. LV internal diameter at end of systole and end of diastole were measured from the two-dimensional guided M mode of the short axis at the mid–papillary muscle level. Subsequently, LV ejection fraction (LVEF) and fractional shortening (FS) were calculated using the modified quinone method (21,22). Isovolumic relaxation time (IVRT) was measured using pulsed-wave Doppler of the mitral valve at apical four-chamber view and corrected for heart rate according to the formula IVRT = IVRT/√RR%. All measurements were averaged on three cardiac cycles for three or more mice per group from each genotype by a trained blinded operator. These studies were performed at SingHealth Experimental Medicine Centre in Singapore.

H9C2 rat cardiomyoblast cells were used for in vitro experiments. Cells were cultured and expanded in DMEM (catalog number 11995-065; Thermo Fisher Scientific) containing 4.5 g/L d-glucose, 1% penicillin/streptomycin, 15% FBS, and 1% nonessential amino acids (catalog number 11140050; Thermo Fisher Scientific). For hyperglycemia (HG) studies, 24 h after seeding (1.5 × 10⁵ cells/well in a six-well plate), the culture medium was replaced with low-glucose DMEM (containing 1 g/L glucose) (catalog number 08456-36; Nakalai Tesque), 1% penicillin/streptomycin, and 1% FBS. We used d-glucose (product number G8270; Sigma-Aldrich) at 400 mg/dL concentration for 48 h for HG. Normoglycemia was maintained for control wells by adding 400 mg/dL d-mannitol (product number M9546; Sigma-Aldrich) as osmotic control. After 48 h, cells were washed with PBS and harvested with TRIzol for RNA analysis.

TF-siRNA was purchased from Thermo Fisher Scientific (Silencer Select siRNA for Rat F3 gene ID 25584; catalog numbers 4390771 and s130191). H9C2 cells were cultured as per the protocol described above. TF-siRNA (50 pmol), negative control siRNA (catalog number AM4611; Life Technologies), and Lipofectamine RNAiMAX (catalog number 13778030; Life Technologies) reagents were used as per manufacturer instructions, and cells were treated in the presence of HG (400 mg/dL glucose) for 48 h. The chemical inhibitors for the ERK (GDC-0994; catalog number HY-15947; Med Chem Express), p38-MAPK (SB203580; catalog number s1076; Selleck Chemicals), TLR4/TAK1 (5z-7-oxozeaenol; catalog number 3604; Tocris Bioscience), and STAT3 (inhibitor VI S3I-201; catalog number sc-204304; Santa Cruz) pathways were used in final concentrations of 1, 5, 1, and 50 μmol/L, respectively, along with vehicle (DMSO) control in the presence of HG for 48 h. After 48 h, cells were washed with PBS and harvested with TRIzol for RNA isolation and quantitative PCR.

At 12 weeks of diabetes, just before harvesting, the mice with diabetes and the control mice were euthanized, and blood was collected by cardiac puncture into collection tubes containing 3.8% sodium citrate as anticoagulant and centrifuged at 2,500 rpm for 15 min for plasma separation. Plasma samples were used to measure the B-type natriuretic peptide (BNP) and triglyceride levels, using the mouse BNP Enzyme Immunoassay Kit (catalog number EIAM-BNP-1; RayBiotech) and the mouse Triglycerides ELISA Kit (catalog number MBS1601281; My Biosource).

Statistical analyses were performed using two-tailed paired and unpaired t tests for comparison between two groups (control mice without diabetes vs. mice with diabetes/STZ) in cases of in vitro (cell-based) and in vivo (WT, TF-Het, and TF-low mice) data, respectively. Data are expressed as mean ± SEM. Differences were considered significant if the P value was <0.05. One-way ANOVA was used for multiple comparisons with the post hoc Bonferroni, Dunnett, or Tukey method depending on the nature of the comparison. Statistical analyses were performed using GraphPad Prism version 8.0.

Data and materials are available upon request to the corresponding author.

We determined the changes in TF mRNA and protein levels in LV tissues from WT (mTF^+/+) mice with diabetes and controls without diabetes. There was significant upregulation in both cell surface and alternatively spliced TF mRNA levels (Fig. 1A). The TF protein levels were also increased in diabetic hearts compared with nondiabetic control hearts (Fig. 1B). Similar to WT mice, the TF levels were also elevated (Fig. 1C and D) in TF-Het (mTF^+/−hTF⁺) mice, which are phenotypically normal and carry a human TF transgene and were used as a control for the low-TF mice (mTF^−/−hTF⁺). However, we did not see any changes in TF levels in the diabetic hearts from low-TF mice compared with their nondiabetic controls (Supplementary Fig. 1A and B). There was increased TF expression on immunostaining in cells of the diabetic WT hearts (Fig. 1E). These findings suggest that TF is upregulated in type 1 diabetic mouse hearts. In a separate study in 12-week-old mice with type 2 diabetes (C57BL/6J), we found significant upregulation of both membrane-bound and alternatively spliced TF mRNA levels in heart LV samples compared with respective control mice without diabetes (Supplementary Fig. 3A). This further suggests that TF is upregulated in both type 1 and 2 diabetes and may have an important role irrespective of the pathophysiology of diabetes.

We examined the expression levels of inflammatory genes in the diabetic WT, TF-Het, and low-TF hearts compared with respective nondiabetic controls. We observed a significant increase in the mRNA levels of IL-6, IL-33, and TLR4 in the diabetic WT and TF-Het hearts compared with their respective controls (Fig. 2A). However, the expression of IL-6, IL-33, and TLR4 in the diabetic low-TF hearts remained unchanged compared with their nondiabetic controls (Fig. 2A). We also observed a significant increase in the protein levels of IL-6 and TLR4 in diabetic WT and TF-Het hearts compared with respective controls (Fig. 2B). However, reduced IL-6 and TLR4 protein expression was observed in diabetic low-TF hearts versus controls (Fig. 2B). Consistent with this, TLR4 immunohistochemistry showed an upregulation in diabetic WT and TF-Het hearts and reduced expression in low-TF hearts (Fig. 2C). We did not find any significant changes in the expression of other inflammatory genes (except nuclear factor-κB p65 and interferon-γ, which were down regulated), including PAR1 and PAR2 in the WT diabetic hearts (Supplementary Fig. 1C). These results indicate that cardiac TF may regulate inflammation in diabetic mouse hearts in a PAR1/2-independent manner.

Cardiac hypertrophy, mainly LV hypertrophy, is common in diabetic HF (23,24). In this study, cardiac hypertrophy was corroborated by higher mRNA expressions of β-myosin heavy chain (β-MHC) (Fig. 3A) and corresponding protein (myosin heavy chain 7) (Fig. 3B) in diabetic WT hearts (LV) compared with nondiabetic controls. Cardiac hypertrophy was further characterized by measuring the ratio of heart to body weight (HW/BW), which indicated hypertrophy in diabetic WT hearts compared with nondiabetic controls (Fig. 3C). In diabetic TF-Het hearts, there was increased β-MHC expression, but the HW/BW ratio did not suggest any hypertrophy. On the contrary, low-TF mice did not present any elevation in β-MHC mRNA levels in diabetic hearts versus nondiabetic controls, and the HW/BW ratio was also significantly reduced (Fig. 3A and C). However, among the mice with diabetes only, WT mice had a significantly increased HW/BW ratio compared with the TF-Het and low-TF mice (Fig. 3C, right panel). Next, we performed WGA staining on heart sections to determine cardiomyocyte sizes and measured the cardiomyocyte CSA from the hearts of all mice (Fig. 3D). Based on this, cardiac hypertrophy was evidenced in WT diabetic hearts, but no hypertrophy occurred in TF-Het or low-TF mice; rather, the CSA was reduced in low-TF hearts (Fig. 3D and E). These data together show that TF regulates cardiac hypertrophy in diabetes.

We investigated the effect of HG and diabetes on cardiac cell survival, fibrosis, and apoptosis by analyzing the expression of marker genes for these events. There was a significant downregulation in the mRNA expression of fibrotic genes (collagen 1A1, collagen 3A1, and MMP2) found in diabetic WT hearts compared with controls (Supplementary Fig. 1D). However, we found increased expression of Bcl-2 and Bcl-xL genes implicated in the activation of the prosurvival/antiapoptosis pathway in diabetic WT and TF-Het hearts versus respective controls (Fig. 4A). Consistently, Bcl-2 protein levels were also elevated (Fig. 4B). Surprisingly, in diabetic low-TF hearts, expression of Bcl-2 and Bcl-xL was either unchanged or downregulated compared with their nondiabetic controls (Fig. 4A–C). We also found that the mRNA expression of caspase-3 was unchanged in diabetic WT hearts but elevated in diabetic TF-Het and low-TF hearts compared with respective controls (Fig. 4A, bottom panel). These data suggest that TF may play a role in suppressing apoptosis and promoting survival of cardiomyocytes in diabetic hearts.

We measured cardiac function using echocardiography to assess the systolic and diastolic function in WT, TF-Het, and low-TF mice with and without diabetes. All mice with diabetes showed no change in LVEF or FS compared with respective controls without diabetes (Fig. 5A and B). The LVEF remained >50% (Fig. 5A), and FS levels remained >30% (Fig. 5B), showing preserved ejection fraction phenotype with no systolic dysfunction. Cardiac LV mass (indexed to body weight) as measured by echocardiography suggested a hypertrophic trend in WT mice with diabetes compared with controls without diabetes (Fig. 5C, left panel). However, among those with diabetes only, the LV mass was significantly higher in WT mice than in others (Fig. 5C, right panel). This suggests that a deficiency of TF may be protective against hypertrophic remodeling of diabetic hearts. These results are consistent with our findings from direct measurements of the HW/BW ratio and WGA staining on cardiac hypertrophy. We found a significantly decreased ratio of early to late atrial diastolic transmitral flow velocity in WT and TF-Het mice with diabetes versus respective controls without diabetes (Fig. 5D). Additionally, IVRT corrected for heart rate (IVRTc) was elevated in both WT and TF-Het mice with diabetes compared with their respective controls, but no significant increase was seen in low-TF mice with diabetes (Fig. 5E). The Tei index or the myocardial performance index showed results similar to IVRTc (Fig. 5F), suggesting the presence of diastolic dysfunction in diabetic WT and TF-Het hearts. When compared among the mice with diabetes only, the IVRTc and Tei levels were higher in WT and TF-Het compared with low-TF mice (Supplementary Fig. 2D and E). These data together suggest that WT mice with diabetes had HFpEF and diastolic dysfunction, unlike low-TF mice with diabetes.

We found a marked increase in BNP mRNA levels in diabetic hearts compared with nondiabetic hearts from all the three genotypes (Fig. 6A). However, the increased BNP mRNA expression in diabetic low-TF hearts was comparable only to the levels in nondiabetic WT control hearts. The WT mice with diabetes compared with controls without diabetes also had significantly increased plasma BNP protein, which is an established biomarker of HF. However, no significant increase in BNP protein levels was found in TF-Het and low-TF mice with diabetes versus their controls (Fig. 6B).

We cultured H9C2 (rat cardiomyocytic) cells and subjected them to HG with 400 mg/dL glucose for 48 h. We observed increased TF mRNA expression in hyperglycemic cells compared with cells grown under control normoglycemia (400 mg/dL mannitol) (Fig. 7A). Additionally, there was upregulation of inflammation, evidenced by higher mRNA levels of IL-6 and TLR4, as well as upregulation of prosurvival-related Bcl-2 mRNA in these cells (Fig. 7B). Upon treatment with TF-siRNA in the presence of HG, these effects were significantly reduced (Fig. 7C and D), as evidenced by downregulation of IL-6, TLR4, and Bcl-xL mRNA levels (Fig. 7D). We did not find detectable expression levels for hypertrophic markers like ANP and BNP in the in vitro system.

To further understand the downstream signaling involved in TF-regulated mechanisms of inflammation and other processes, we assessed the plausible downstream signaling pathways involved. The ERK and STAT3 pathways were activated, as evidenced by increased levels of pERK and pSTAT3 in diabetic WT hearts (Fig. 8A and C). However, this activation was abolished, evidenced by reduced levels of pERK (Fig. 8A) and pSTAT3 (Fig. 8C) in diabetic low-TF hearts versus nondiabetic controls. The total ERK levels, however, remained unchanged in all mice with diabetes compared with their respective controls (Fig. 8A). However, total STAT3 levels were also higher in WT mice with diabetes compared with their respective controls. To further confirm whether these pathways were downstream of TF, when H9C2 cells were cultured in the presence of ERK (GDC0994) and STAT3 (S3I-201) inhibitors separately under HG conditions, the mRNA expression levels of inflammatory (IL-6 and TLR4) and prosurvival (Bcl-2 and Bcl-xL) genes were decreased in inhibitor-treated samples compared with DMSO-treated controls (Fig. 8B and D). To further investigate the signaling pathways involved in TF-regulated inflammation, we treated H9C2 cells with a p38-MAPK inhibitor (SB203580) and found that only IL-6 levels were inhibited (Fig. 8E). The in vivo Western blot data on the p38-MAPK protein in WT, TF-Het, and low-TF mice showed only slight increases in phosphorylated p38 levels in WT and TF-Het mice with diabetes but not in low-TF mice with diabetes (Fig. 8E). To further understand the nature of the interaction between TF and TLR4 and its role in inflammation, we used an inhibitor of TAK1, which is an important mediator of the TLR4 pathway, and found that neither the inflammation nor prosurvival pathway was affected (Fig. 8G). A thorough investigation of transcription factors involved in the abovementioned pathways in diabetic hearts revealed upregulation of mRNA expression of c-Myc, c-JUN, SOCS, STAT3, and Myd88 genes (Fig. 8H) in WT and TF-Het mice but not in low-TF mice. Together, our in vivo and in vitro data demonstrate that TF regulation of cardiac inflammation is mainly dependent on ERK (IL-6 and TLR4), STAT3 (TLR4), and p38-MAPK (IL-6 only) pathways, whereas the prosurvival pathway is mainly ERK and STAT3 dependent (Fig. 8I).

The underlying mechanisms of inflammation and hypertrophy, which eventually lead to pathological remodeling and cardiac dysfunction in diabetes, are unknown. This further encompasses the unique and challenging HFpEF phenotype with diastolic dysfunction. In the current study, using a mouse model of type 1 diabetes, we demonstrate the role of TF, a receptor expressed on cardiac cells, in regulating diabetes-induced cardiac inflammation, hypertrophy, and HFpEF.

We observed upregulation of TF in the hearts of a type 1 diabetic mouse model. Although the TF upregulation in whole blood in HG and diabetes has been reported (14,15,18), little is known about alterations in TF in the heart. Diabetes is considered an inflammatory disease (25,26), and the role of TF in inflammation, although reported in other pathological conditions (12,27,28), has only recently started to unfold in diabetes (29). However, there remains a dearth of knowledge on the role of TF in regulating inflammation in diabetic cardiomyopathies. Therefore, our study provides new insights into cardiac inflammation regulation by TF primarily in the HG-driven mouse model of diabetes (type 1 diabetes). Additionally, we provide evidence that TF is upregulated in a type 2 diabetes model as well, with a concomitant increase in inflammation in type 2 diabetic mouse hearts compared with nondiabetic controls (Supplementary Fig. 3). In the current study, we observed increased levels of IL-6, IL33, and TLR4 in diabetic WT and TF-Het hearts, and TF deficiency abolished these changes, suggesting an important role of TF in regulating diabetes-induced cardiac inflammation. In the mouse model of endotoxemia, low-TF mice showed reduced levels of IL-6 compared with control mice (30). Inhibition of TF in a sepsis model reduced circulating levels of the proinflammatory cytokines IL-6 and IL-8 (31). TF enhanced inflammation associated with ischemia-reperfusion injury, and inhibition of TF reduced inflammation as well as infarct size (12). TF can modulate inflammation in a coagulation-dependent (involving PAR1/2) or -independent manner. A recent study demonstrated that endothelial cell–specific deletion of TF significantly attenuated plasma IL-6 levels but had no effect on thrombin generation, suggesting that endothelial TF is primarily involved in inflammation rather than coagulation (28). Studies of IL-33 and TF interplay have demonstrated that proinflammatory cytokine IL-33 induced TF mRNA and protein levels on both monocytes and endothelial cells (32,33). Moreover, previous studies have reported upregulation of TLR4 by HG in both in vitro (monocytes) and in vivo (myocardial tissue of mice with diabetes) setups (34,35), and there exists a link between TF and TLR4 in diabetes (18). In the current study, however, we did not find any parallel or independent activation of the TLR4 pathway; rather, TLR4 may be transcriptionally regulated by TF via ERK and STAT3 pathways downstream of TF. There was significant downregulation of TLR4 mRNA levels upon TF, ERK, and STAT3 inhibition (Figs. 7D and 8B and D), respectively. But inhibition of TAK1 (downstream marker of the TLR4-Myd88 pathway) did not result in any changes in inflammation in H9C2 cells (Fig. 8G). Previous studies have also shown transcriptional regulation of TLR4 in macrophages (36) in inflammation and in intestinal epithelial cells (37) in homeostasis.

TF has been implicated in the maintenance of the structural integrity of cardiac muscles (9,38) because of its localization and interaction with structural proteins in the cardiomyocytes. However, its role in diabetes-induced cardiac hypertrophy is not known. In the current study, we found upregulation of hypertrophic markers (β-MHC) and an increased HW/BW ratio, in addition to an increased size of cardiomyocytes, suggesting cardiac hypertrophy in WT mice with diabetes versus controls without diabetes. However, TF deficiency protected the hearts from diabetes-induced hypertrophy. In a rat model of angiotensin II–induced hypertension and cardiac hypertrophy, cardiac TF levels were upregulated, and inhibition of the angiotensin II receptor ameliorated cardiac hypertrophy and reduced TF levels (39). Furthermore, in sickle cell disease, the reduction of TF resulted in the attenuation of cardiac hypertrophy. These studies, similar to our study, showed positive correlation between TF expression and cardiac hypertrophy. However, there exists a negative correlation between TF expression and cardiac dysfunction in certain pathological conditions. For example, TF levels are significantly reduced in the ventricular septa of patients with dilated cardiomyopathy compared with healthy controls (10). In the current study, increased cardiomyocyte CSA in diabetic WT hearts suggests a prohypertrophic role of TF. Additionally, upregulation of Bcl-2 and Bcl-xL, the essential inhibitors of apoptosis (40), in diabetic WT and TF-Het hearts and no such effects in low-TF hearts suggests a clear role of cardiac TF in the activation of the prosurvival pathway in diabetes. Previous studies by Boltzen et al. (41) demonstrated that TF protects cardiomyocytes against tumor necrosis factor-α–induced apoptosis via upregulation of the antiapoptotic genes Bcl-2 and Bcl-xL. However, the TF-deficient cardiomyocytes showed reduced expression of Bcl-2 and Bcl-xL, with increased sensitivity to tumor necrosis factor-α–induced apoptosis (41). Similarly, transgenic mice with a heart-specific overexpression of Bcl-2 exhibited reduced infarct size and fewer apoptotic cells (42). The TF-induced expression of Bcl-2 family members may thereby prevent the activation of apoptotic marker caspase-3. In agreement with this, our diabetic low-TF and TF-Het hearts showed an upregulation of caspase-3, but no such increase was found in diabetic WT hearts. These findings suggest that TF may be protective against diabetes-induced apoptosis, and it may impart better survival of cardiomyocytes in diabetes via upregulation of Bcl-2 and Bcl-xL. On the basis of these findings, we postulate that TF-dependent cardiac hypertrophy in diabetes may be an adaptive response for better cardiomyocyte survival; however, this may eventually lead to cardiac dysfunction, thus causing HF.

Diabetes is highly prevalent in patients with HFpEF, which is characterized by presence of LV hypertrophy (43), with normal systolic function (preserved LVEF) but diastolic dysfunction. The underlying mechanisms of HFpEF remain poorly understood. Although based on echocardiographic studies we observed only a trend toward LV hypertrophy in WT mice with diabetes, it was well established via other parameters (HW/BW ratio and WGA staining), but not in others (TF-Het and low-TF mice). However, in mice with diabetes only, WT mice had greater LV mass than both TF-Het and low-TF mice, suggesting that loss of TF is protective against diabetes-induced hypertrophic remodeling of the heart. Furthermore, the unchanged LVEF and FS values signified no systolic dysfunction in any mice. However, a significantly decreased ratio of early to late atrial diastolic transmitral flow velocity in WT and TF-Het mice suggested the presence of diastolic dysfunction. In contrast to low-TF mice, diastolic dysfunction was further established by significantly elevated IVRTc and Tei index values in both WT and TF-Het mice with diabetes. The increased levels of both IVRTc and Tei index values indicate a problem with the relaxation mechanism of the heart, suggesting diastolic dysfunction. These results together show that loss of TF causes better cardiac function in a diabetic heart. Higher Tei index has been previously reported in patients with diabetes with HF (44), and it has good clinical application in patients with various etiologies of HF (45,46). In addition to the changes in functional parameters suggesting HF, we also found increased mRNA and plasma protein levels of BNP, an established marker of HF (47), in diabetic WT hearts compared with controls. Although the BNP mRNA levels were also elevated in diabetic TF-Het and low-TF hearts, which could be attributed to other regulatory mechanisms, like neurohumoral interactions (endothelin, angiotensin) or local hypoxia resulting from diminished coronary flow (48), no change in plasma BNP protein levels suggests no HF. Together these results demonstrate an important role of TF in diabetes-induced HFpEF.

When we modulated the ERK, p38-MAPK, and STAT3 pathways to determine the downstream mechanisms of TF function, we found that TF was the upstream regulator of both ERK and STAT3 pathways in diabetes-induced cardiac remodeling. This was also evident by increased pERK and pSTAT3 levels in WT diabetic hearts, but not in diabetic low-TF hearts. We observed the downregulation of both inflammation and prosurvival pathways upon treatment with ERK and STAT3 inhibitors separately in the presence of HG. However, there was only a downregulation of IL-6 mRNA levels upon inhibition of the p38-MAPK pathway, suggesting TF may regulate inflammation partially through the p38-MAPK pathway in diabetic hearts. These results suggest that the TF-regulated mechanisms of inflammation, prosurvival, and hypertrophic remodeling in diabetic hearts is predominantly ERK/p38-STAT3 pathway dependent. The TF regulation of inflammation and cell survival via some of these pathways has also been reported in previous studies (49,50).

A limitation of our study is the use of a global low-TF mouse model instead of a cardiomyocyte-specific TF-KO model. However, cardiomyocyte-specific TF-KO mice have similar phenotypes to global low-TF mice, except that the phenotype (hemosiderin deposition and fibrosis) appears at a later stage (9 months old instead of 5 months old in low-TF mice) (51). Because our study used these mice of up to 5 months old only, we did not encounter any fibrosis or hemosiderin deposition in our STZ or control mice. Therefore, we believe that the use of cardiomyocyte-specific TF-KO mice would not have resulted in any significant differences in the findings of the current study. Furthermore, on the basis of the data presented in this study, we believe that complete loss of TF in low-TF mice may not have produced any systemic effect and therefore could not have affected the cardiac inflammation, hypertrophy, or dysfunction reported in this study.

In conclusion, TF plays an important role in type 1 diabetes–induced cardiac inflammation, prosurvival, and hypertrophy leading to HF with preserved ejection fraction (Fig. 8I). The current study demonstrates new roles of TF in diabetes-induced cardiac defects. In the future, it would be worth extending these studies to a type 2 diabetes model in detail and in patients to find better treatments for HF in diabetes.

Acknowledgments. The authors thank Lim Sze Yun from the National Heart Centre, Singapore, and Shamini Guna Shekeran and Sandip Chorghade from Duke–NUS Medical School, Singapore, for technical assistance.

Funding. This work was supported by funds from the Goh Foundation and the Duke–NUS Medical School to M.K.S.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. D.M.C., R.S., H.B., and S.A.B.A.G. designed and performed experiments and analyzed data. N.T. performed echocardiographic studies and analyzed data. B.K.S. provided type 2 diabetes and respective control mouse heart samples and reviewed relevant data. N.M. provided the low-TF mice and edited the manuscript. M.K.S. edited the manuscript and oversaw the project. A.S. designed and performed experiments, analyzed data, wrote the manuscript, conceived the original idea, and supervised and oversaw the entire project. A.S. 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.

Prior Presentation. A subset of data was presented as an abstract at the American Heart Association Scientific Sessions, Philadelphia, PA, 16–18 November 2019.