The study design was reviewed and approved by the Animal Ethics Committee of Guangzhou Medical University (Guangzhou, China) according to the Guidelines for Animal Care and Experimentation with approval number GY 2018-104.

Male C57BL/6J mice (age: 6–8 weeks, weight: 20 ± 2 g) were housed in a controlled environment and provided the free access to food and water. For TAC procedure, all the mice were anesthetized with isofluorane. Then, the midline sternotomy was performed, the thymus was separated, and the aortic arch was exposed. Then, a 6.0-silk suture was wrapped around the arch between the innominate artery and the left carotid artery over a 26-gauge needle. Subsequently, the needle was removed, and the chest was sutured. The sham group mice underwent all the procedures except the aortic constriction. For analgesia, buprenorphine (2 mg/kg) was administered thrice after surgical incision. After 7 days of surgery, 0.5% CMC-Na as the vehicle and EMPA (10 mg/kg/day) were administrated via oral gavage for the duration of the study.

Mice with two loxP sites inserted in the β-catenin gene’s 1 and 6 introns position were donated by Dr. Rolf Kemler (Max-Planck Institute of Immunology, Germany) (Brault et al., 2001) nd housed at The Jackson Laboratory (Stock No.004152) (Bar Harbor, ME). Transgenic mice expressing Cre recombinase under the control of the α-myosin heavy chain promoter (αMHC-Cre) were donated by Dr. M D Schneider and maintained at The Jackson Laboratory (Stock No.011038) (Agah et al., 1997). The homozygous β-catenin loxP-floxed (β-catenin flox/flox) mice were mated with α MHC-Cre mice to produce cardiac-specific β-catenin knockdown mouse models. Because conventional and cardiomyocyte-specific β-catenin -knock-out mice are embryonically lethal (Ye et al., 2015), the heterozygous β-catenin deletion Het (β-cat flox/+, αMHC-Cre+) mice or WT (β-cat flox/+, αMHC-Cre-) mice were subjected to TAC surgery for subsequent experiments. Littermate wild-type mice without Cre recombinase expression were used as controls. Supplementary Table S2 enlists all the genotyping primers utilized.

Visual Sonics Vevo 2,100 system equipped with an MS-400 transducer (Visual Sonics, Inc.) was employed for echocardiography. Left ventricle ejection fraction (LVEF) and other systolic functions were obtained from the short-axis with M-mode scans, as indicated by the presence of papillary muscles. For anesthesia, 3% isoflurane was administered, and the lack of response to firm pressure under a body temperature-controlled pad indicated successful anesthesia. Isoflurane was reduced to 1.0%–1.5% and adjusted to maintain a heart rate between 350 and 550 (Beats per minute; BPM). The (Heart Rate; HR) (Left ventricular end-diastolic diameter; LVED; d), (Left ventricular end-systolic diameter; LVED; s) (Left ventricular internal systolic dimension; LVID; s, (Left ventricular internal diastolic dimension; LVID; d) (Left ventricular end-systolic anterior wall thickness; LVAW; s), (Left ventricular end-diastolic anterior wall thickness; LVAW; d (Left ventricular end-diastolic posterior wall thickness; LVPW; d), (Left ventricular end-systolic posterior wall thickness; LVPW; s) (Left ventricular fractional shortening; FS), and (Left ventricular ejection fraction; EF), were detected. All the mice recovered from anesthesia without difficulty. All parameter data were presented as a mean of at least 5x continuous image analysis.

After the final Echocardiograph analysis, the mice were weighed, anesthetized using isoflurane, and sacrificed by cervical dislocation. The hearts were dissected in cold PBS and weighed before the Ventricle and Atrium were separated. The Ventricle was snap-frozen in dry ice and transferred into a −80°C refrigerator. Tibias were also removed.

Heart tissue and cells were homogenized with RIPA lysis buffer (Thermo Fisher Scientific; Cat # 89901) containing halt protease and phosphatase inhibitor cocktail (Thermo Fisher Scientific; Cat # 78442), incubated in the shaker for 30 min at 4°C, and then centrifuged at 12,000 g (30 min) to collect supernatant for immunoblotting.

Whole lysate protein concentration was measured by Pierce BCA Protein Assay Kit (Thermo Fisher Scientific; Cat # 23225). Then, an equal quantity of protein lysates was isolated on 8%–15% SDS-PAGE gel by electrophoresis, transferred to 0.45 μm PVDF membranes, blocked in 5% non-fat milk diluted in TBST for 45 min at room temperature, probed overnight with primary antibodies diluted in a blocking buffer at 4°C on a shaker. Then, the membranes were treated for 45 min with an HRP-conjugated secondary antibody diluted in a blocking buffer at room temperature, and the target proteins were detected using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific; Cat # 34580). Amersham Imager 600 was used for immunoblotting processing.

For immunoprecipitation, the extracts from the ventricular tissue were cleared using Protein A/G beads, and the mixtures were subsequently centrifuged at 1,000 g for 4 min at 4 °C. The supernatants were incubated with primary antibodies at 4 °C overnight. On the 2nd day protein A/G beads incubated for 4 h, and the beads were collected and washed three times with lysis buffer. After the final wash, the proteins were collected and subjected wo Western blot analysis using the antibodies listed in Supplementary Table S3.

The dissected mice hearts were put in ice-cold PBS, ventricle slice was acquired, fixed for 60 min in 4% paraformaldehyde in PBS, washed twice with cold PBS, placed in 30% sucrose in PBS for 30 min, embedded in optimal cutting temperature (OCT) compound, snap frozen in dry-ice, and stored at −80°C. The heart sections were cut into 5 μm slices using cryostat. The myocardial tissue sections were fixed, and the treated cardiomyocytes were washed with 0.1 mol/L glycine for 25 min to reduce the non-specific crosslinking activity of residual paraformaldehyde. The sections were then immersed in 10% normal goat serum for 60 min to block non-specific binding and incubated overnight at 4°C with Alexa Fluor 488 conjugated antibody for 1 h, followed by Alexa 550 conjugated α-actinin antibody for 1 h. The nucleus was stained with DAPI (Sigma-Aldrich, St. Louis, MO). Confocal images were collected with a Nikon A1R confocal upright microscope under uniform settings. From each mouse, 3 heart sections were utilized for this analysis to determine the percentage intensity of antibodies.

Adult mouse cardiac left ventricular myocytes were isolated by following the method described previously (Chen et al., 2022). Briefly, the hearts were perfused with perfusion medium (Gibco) containing 1.2 mmol/L MgSO₄, 10 mmol/L HEPES, 4.6 mmol/L NaHCO₃, 30 mmol/L taurine, 10 mmol/L 2,3-butanedione monoxime, and 5.5 mmol/L glucose. After perfusion, the heart was digested with the isolation buffer using the Langendorff system with type II collagenase (Worthington, Cat # LS004174, NJ, United States). Then, forceps and pipettes were used to dissect and mechanically dissociate left ventricles (LV) in a perfusion medium. Isolated cells were fixed with 4% paraformaldehyde or stored at −80°C for the subsequent experiments.

NRCMs were isolated from one to two-day-old Sprague Dawley rats, as described in our previous study (Chen et al., 2022). Briefly, the ventricles were homogenized using D-Hank’s balanced salt solution, trypsinized by 0.05% trypsin, followed by 0.2% collagenase (type II), and seeded into plates containing DMEM to remove fibroblasts and non-cardiomyocytes. The cells were cultured with DMEM augmented with 10% fetal bovine serum (FBS) and 0.1 mmol/L bromodeoxyuridine and then treated with vehicle alone (0.05% DMSO), Angiotensin II (AngII) (1 μM) + vehicle (0.05% DMSO), and AngII (1 μM) + EMPA (1, 5, 10 μM, respectively) for 24 h before collection.

Molecular docking was performed to predict the possible binding of EMPA (EMPA) with FZD. The crystal structures of human FZD4, human FZD5, and mouse FZD8 were acquired from the protein data bank (PDB) under the following codes: 5CM4, 5URZ, and 1IJY, respectively. The reported potential binding sites were chosen as docking centers. Docking was performed using Surflex-Dock GeomX (SFXC) in SYBYL-X 2.1.1 software (Tripos, United States), and the results were evaluated by total score (unit: log10). The highest score was selected as the binding affinity in the best binding mode. The Discovery Studio Visualizer 2020 (Dassault Systems BIOVIA, United States) was utilized to demonstrate the two-dimensional diagrams of the ligand-receptor interactions.

HEK293T cells were cultured in DMEM (Thermo Fisher Scientific, Waltham, MA, United States) supplemented with 10% FBS (Biochrom, Berlin, Germany) without antibiotics. The full-length human FZD4, human FZD5, and human FZD8 were cloned into pcDNA3.1 vectors (Youbio, Hunan, China.) Lipofectamine^® 3,000 reagent (Invitrogen; Thermo Fisher Scientific, Inc., CA, United States) was used for transfecting the FZDs or negative control (NC) plasmids, per the kit’s instruction. After 16 h of transfection, the cells were treated with Wnt3a (100 ng/mL, R and D Systems, MN, United States), or Wnt3a + EMPA (50 mM) for 48 h before collection for subsequent analyses.

Luciferase reporter assay was performed according to a standard protocol. HEK293T cells (1 ×10⁶/well) were cultured in 60 mm plates for 24 h and then transfected with plasmids (FZD4, FZD5, FZD8) and PC-DNA (5 μg/well) using Lipofectamine 3,000 reagent (Thermo Fisher Scientific, Waltham, MA, United States, Cat. No. L3000008). After 16 h of transfection, the culture medium was removed, and 5 mL/well of DMEM with 10% FBS containing Wnt3a (400 μg) and 50 mM EMPA was added for another 48 h. Then, luciferase and Renilla signals were determined by a Dual-Luciferase Reporter Assay Kit (V Cat. No. DL101-01; Vazyme, Nanjing, China) on a Wallac 1,420 Victor microplate reader (PerkinElmer, United States). The Firefly/Renilla signal ratios were recorded.

SPSS 18.0 and Prism GraphPad software were employed for all the statistical measurements, and the data are presented as the mean ± SEM. Statistical analyses included one-way ANOVA with a post hoc Tukey test, Two-way ANOVA with a post hoc Tukey test. Statistically, the p-value of <0.05 was deemed as the significance threshold.

In this study, a mouse cardiac hypertrophy model was established by surgical ligation of the transverse aortic. No significant differences were observed in the serum glucose level, total serum cholesterol (TC), triglyceride (TG), low-density lipoprotein (LDL), and renal function parameters among these four groups (Supplementary Table S1). Furthermore, high-density lipoprotein (HDL) was decreased, and glutamic oxaloacetic transaminase (AST) was increased in the TAC group, which was attenuated after EMPA treatment (p < 0.05, Supplementary Table S1).

The effects of EMPA on TAC-induced myocardial hypertrophy and ventricular remodeling were also assessed. The 10 mg/kg/day EMPA was administered for 1 week after the surgery (Figure 1A), and it was observed that compared with the vehicle treatment, EMPA attenuated TAC-induced cardiac hypertrophy (Figure 1B), heart weight (HW)/body weight (BW) ratio, and lung ventricle weight (LW)/body weight (BW) ratio (Figure 1C). In addition, EMPA treatment significantly reduced the cardiomyocyte’s cross-sectional area in TAC mice (Figure 1D) compared to the mice in the sham group. Furthermore, the cardiomyocytes were isolated from each mouse’s heart to measure their surface area, which revealed that EMPA treatment decreased the size of individual cardiomyocytes (Figure 1E). Since myocardial fibrosis is a key feature of cardiac remodeling, Masson trichrome analysis was performed on cardiac tissue sections to evaluate the myocardial interstitial and perivascular fibrosis. The EMPA treatment reduced cardiac fibrosis in TAC mice compared to the sham mice (Figure 1F). Overall, continuous administration of EMPA for 8 weeks significantly attenuated the cardiac hypertrophy manifestations and fibrosis index (Figures 1B–G). Moreover, the mRNA levels of hypertrophic genes, Nppa, Nppb, and Myh7 were upregulated in TAC mice’s hearts but significantly reduced in the EMPA-treated mice (Figure 1G).

Echocardiography of each group of mice was performed every 2 weeks after EMPA administration to evaluate the function and structure of heart ventricles. Left-ventricular systolic function was indicated by LVEF and LV fractional shortening (LVFS). Compared to the sham group, LVEF and LVFS significantly increased from 2 weeks after TAC surgery and peaked at 4 weeks, which decreased 6 weeks thereafter (Figures 2A, B). Moreover, LVAW and LVPW at both systole and diastole in the TAC group increased gradually compared with the sham group. However, 4 weeks of EMPA treatment significantly improved the above echocardiographic parameters.

Recently, Speckle Tracking Echocardiography (STE) has emerged as an important diagnostic modality in cardiac diagnostic imaging. Global circumferential strain (GCS) from STE is a novel index of LV function and has been used to estimate LV filling pressures and predict outcomes in several disease states. Therefore, in this study, the LV systolic function was assessed with GCS from STE. Consistent with the changes of LVEF and LVFS, EMPA treatment markedly attenuated the TAC-induced LV dysfunction (Figure 2C).

The canonical Wnt/β-catenin pathway has been observed to be associated with cardiac hypertrophy. Previous studies have indicated that cardiac β-catenin deletion can attenuate stress-induced cardiac hypertrophy within 4 weeks. To assess if EMPA can inhibit the canonical Wnt/β-catenin pathway in TAC mice, GSK3β phosphorylation levels and the protein expression of Non-phospho (active) β-catenin and total-β-catenin were identified by Western blotting using the heart tissues from each group. The phosphorylation levels of GSK3β′s Ser9 residue, a key regulator of cytoplasmic β-catenin level, as well as the Non-phospho (active) β-catenin level, were significantly increased in TAC mouse hearts compared with the vehicle (Figures 3A, B). These results were consistent with previous findings. Furthermore, TCF7L2, a key transcriptional factor in nuclei of β-catenin, increased dramatically. The Wnt/β-catenin pathway’s downstream target, c-Myc, was also observed to be notably expressed. However, the EMPA treatment markedly reduced the levels of GSK3β phosphorylation, active-β-catenin, TCF7L2, and c-Myc (Figures 3A, B). The landmark of canonical Wnt/β-catenin pathway activation is β-catenin nuclear accumulation. Therefore, immunofluorescence and cellular fractionation analyses were performed to analyze the β-catenin location. Immunofluorescence and Western blot analyses confirmed that both active- and total-β-catenin were accumulated in the nucleus of the TAC mice heart cells (Figures 3C, E). These changes were repressed by EMPA treatment (Figures 3C, E). Immunofluorescence assay in adult cardiomyocytes isolated from each group of mice hearts further validated these findings (Figure 3D). Moreover, the co-immunoprecipitation analysis revealed that TAC significantly promoted the β-catenin and TCF7L2 co-localization, whereas the EMPA treatment had opposite effects (Figure 3F). These data suggested that the inhibition of the canonical Wnt/β-catenin pathway may play a critical role in the improvement of cardiac hypertrophy by EMPA.

These data suggested that TAC-induced reduction of the β-catenin/TCF7L2 pathway might be associated with the targeted mechanism of EMPA, which reduces the primary composite endpoint of major adverse cardiovascular events, cardiovascular death, and hospitalization rate for HF, as observed in EMPA-REGOUTCOME trial.

An in vitro AngII treatment model was employed to assess whether EMPA reduced cardiomyocyte hypertrophy by inhibiting the canonical Wnt/β-catenin pathway. The results in NRCMs were consistent with the findings in mouse hearts. After 24 h of co-treatment, EMPA significantly blocked the AngII (1 μM)-induced cardiomyocyte hypertrophic manifestations, including increased cell surface area and ANP expression (Figures 4A, B). Furthermore, EMPA dose-dependently repressed the AngII (1 μM)-induced active-β-catenin and c-Myc protein levels in the Wnt/β-catenin signaling (Figure 4C). Moreover, EMPA treatment reduced TCF7L2 expression, but the decrease was not statistically significant (Figure 4C).

To further investigate the role of β-catenin in TAC-induced cardiac pathological remodeling. Mice with cardiomyocytes-specific heterozygous deletion of the β-catenin coding gene; Het (β-catenin flox/+; α-MHC-Cre+) or WT (β-catenin flox/+, αMHC-Cre-) were stimulated with TAC and then treated with EMPA. The mice underwent TAC surgery or not before the NS or EMPA interventions. Compared to the wild-type (WT) mice, EMPA-treated mice had reduced TAC-induced cardiac hypertrophy (Figure 5A). Furthermore, Het mice indicated reduced levels of TAC-induced cardiac hypertrophy and diminished interstitial fibrosis (Figure 5B C and D). Het mice had decreased β-catenin protein expression in the heart (Figures 5E, F). Moreover, EMPA treatment inhibited the TAC-induced β-catenin/TCF7L2 axis activation (Figures 5E, F). These data suggest that EMPA reduces the TAC-induced β-catenin nuclei accumulation and its downstream transcriptional regulation to attenuate pathological cardiac remodeling. In summary, it was inferred that EMPA treatment attenuates the TAC-induced cardiac hypertrophy by reducing the β-catening/TCF7L2 pathway.

The above results showed the inhibitory effect of EMPA on the Wnt/β-catenin pathway. However, the mechanisms underlying this inhibitory effect remain poorly understood. The Wnt/β-catenin pathway is activated when Wnt, FZD, and LRP5/6 form a complex, which stabilizes the β-catenin. A cysteine-rich domain (CRD) is in the extracellular region of FZD proteins, which has been implicated as the Wnt binding domain. Wnt binding to CRD of FZD activates the Wnt/β-catenin pathway. Therefore, in this study, molecular docking analysis was performed to predict the binding mode and affinity of EMPA with the CRD of FZD. Docking results showed that EMPA had a high affinity with FZD4 (binding score = 7.3455), FZD5 (binding score = 7.0631), and FZD8 (binding score = 7.3865) (Figures 6A–D). Furthermore, the best binding modes and affinity were observed when EMPA was docked into the middle of the FZD dimer. EMPA could bind to both mouse and human FZDs. Based on these data, it was hypothesized that EMPA inhibited the Wnt/β-catenin pathway via FZDs. To verify if EMPA directly binds to FZD, HEK293T cells were stably transfected with FZD-4, -5, and -8 plasmids, respectively. The cells were then stimulated with Wnt3a, in the presence of EMPA or without EMPA. The EMPA and Wnt3a co-treatment significantly inhibited the expression of β-catenin protein, which was further upregulated by Wnt3a treatment in FZD4-transfected cells, but EMPA treatment suppressed the β-catenin upregulation (Figure 6E). However, when FZD-5 and -8 transfected cells were treated with Wnt3a, no changes in the expression of β-catenin were observed.

This study also assessed the transcriptional activity of β-catenin via the TOP/FOP dual-luciferase reporter gene system, and the results were consistent with Figure 6E. When FZD4 transfected 293T cells were treated with Wnt3a, β-catenin transcriptional activity was increased, whereas EMPA treatment significantly inhibited this increase. Moreover, FZD-5 and -8 transfected cells treated with Wnt3a indicated no significant effect on the transcriptional activity of β-catenin. The luciferase reporter assay results indicated that EMPA inhibits FZD4-mediated TOPFLASH luciferase activity (Figure 6F).

The Wnt/β-catenin pathway is involved in the pathogenesis of cardiac diseases, especially cardiac hypertrophy. This study revealed that EMPA treatment can alleviate TAC-induced cardiac hypertrophy by directly inhibiting the FZD4/Wnt/β-catenin/TCF7L2 pathway in cardiomyocytes (Figure 6G).