Exercise Training Alleviates Cardiac Fibrosis through Increasing Fibroblast Growth Factor 21 and Regulating TGF-β1-Smad2/3-MMP2/9 Signaling in Mice with Myocardial Infarction

At first, 10-month-old mice were subjected to four types of exercise training separately, including aerobic exercise training (AET), resistance exercise training (RET), whole body vibration exercise group (WBV), and skeletal muscle electrical stimulation group (ES). Compared with control mice, AET and RET increased the cross-sectional area (CSA) of cardiomyocytes (both p < 0.01), heart weight (both p < 0.01), and the ratio of heart weight/body weight (HW/BW, both p < 0.01) of mice in AET and RET groups; AET, RET, WBV and ES all increased the ratio of heart weight/tibial length (HW/TL, p < 0.01 for AET and RET, p < 0.05 for WBV and ES), left ventricular ejection fraction (EF) and left ventricular fractional shortening (FS, p < 0.01 for AET and RET, p < 0.05 for WBV and ES), reduced left ventricular internal diameter at end diastole (LVIDd) and left ventricular internal diameter at end systole (LVIDs) (p < 0.01 for AET and RET, p < 0.05 for WBV and ES) (Figure S1). According to these results, we chose AET and RET to intervene the MI animals.

We detected the cardiac function and fibrosis to verify the effects of exercise training in the mice with MI. Masson staining, sirius red staining and collagen content were detected to assess the degree of cardiac fibrosis. Major components of the myocardial ECM are collagen type I (COL-I) and type III (COL-III) [30]. Therefore, we detected the protein expressions of COL-I and COL-III. The results showed that compared with the Sham-operated (S) group, replacement fibrosis occurred in the infarcted heart of MI mice, accompanied with increased collagen volume fraction (CVF) level (p < 0.01) and the protein expressions of COL-I (p < 0.01) and COL-III (p < 0.01), meanwhile, LVIDd (p < 0.01) and LVIDs (p < 0.01) significantly increased, and EF (p < 0.01) and FS (p < 0.01) decreased in the MI-sedentary (MI) groups. Compared with the MI group, AET and RET reduced the CVF level (both p < 0.01) and the protein expressions of COL-I (both p < 0.01) and COL-III (both p < 0.01) as well as the levels of LVIDd (both p < 0.01) and LVIDs (both p < 0.01), increased EF (both p < 0.01) and FS (both p < 0.01) of the infarcted hearts in the MAE and MRE groups (Figure 2A–E,G–K). These results indicated that AET and RET inhibited cardiac fibrosis and improved cardiac function in mice with MI.

FGF21 exerts a protective effect in multiple disease models [22,23,24]. In this study, we detected the protein expression of FGF21 in the heart. In the normal mice, AET (p < 0.01), RET (p < 0.01), WBV (p < 0.05) and ES (p < 0.01) increased FGF21 protein expression significantly (Figure S2A). Moreover, Pearson correlation analysis showed significant correlations between FGF21 level and LVIDd (r = −0.553, p < 0.01), LVIDs (r = −0.734, p < 0.01), EF (r = 0.829, p < 0.01) and FS (r = 0.799, p < 0.01) (Figure S2B–E). In the injured heart, compared with the S group, the FGF21 protein expression increased (p < 0.01) significantly in the MI group. Compared with the MI group, AET and RET further up-regulated the FGF21 protein expression (both p < 0.01) in the infarcted heart (Figure 2F). Based on these results, we speculated that FGF21 would play an important role in the cardioprotective effects of exercise.

Then, we used Fgf21 KO mice to establish the MI model and evaluate the role of FGF21 in the cardioprotective effects of exercise. Recombinant adeno-associated virus (rAAV)-Cre-GFP was injected into Fgf21^loxp mice via caudal vein one month before the MI surgery to inhibit FGF21 expression in mice, as the Fgf21 KO mice. The validation of virus infection effect and KO model is shown in Figure S3. In our study, we performed Fgf21 KO mice with AET intervention (KME group). The mortality of the mice in the KME group was up to 60% (Figure S4), especially during the adaptive training. Compared with the MI with AET group (MAE), we hardly detected the protein expression of FGF21 in the Fgf21 KO mice (p < 0.01, Figure 2F). In addition, the CVF level (p < 0.01) and the protein expressions of COL-I and COL-III increased (both p < 0.01) significantly, in contrast, EF and FS decreased (both p < 0.01). These results showed that inhibition of FGF21 protein expression reduced the AET-induced improvement of cardiac function and fibrosis.

FGF21 has been reported to improve cardiac remodeling by inhibiting cardiac fibrosis [31,32]. However, the mechanism is still not well known. Fibrosis occurrence is closely related to the imbalance between proliferation and apoptosis of CFs and increased collagen secretion [33,34]. Therefore, we isolated and cultured the CFs from the neonatal mice hearts and explored the possible mechanism of FGF21 on inhibiting cardiac fibrosis. Oxidative stress is one of the causes of cardiac fibrosis in the injured heart [35]; we treated the CFs with H₂O₂ to mimic the ischemic myocardium and detected the effect of FGF21 on the activation of TGF-β1-Smad2/3 signaling in the CFs. Western blotting results showed that compared with control, H₂O₂ significantly increased the protein expression levels of TGF-β1, Smad2/3, MMP2 and MMP9 of CFs (all p < 0.01), which were inhibited by rhFGF21 intervention (all p < 0.01). AICAR, an activator of adenosine monophosphate (AMP)-activated protein kinase (AMPK), is often used to mimic the exercise effects in in vitro experiments. It has been demonstrated that AICAR could increase FGF21 production and release from cardiomyocytes [36]. In the present study, we found intervention of AICAR alone or combined with rhFGF21 reduced H₂O₂-induced increase of Smad2/3, MMP2 and MMP9 (all p < 0.01, Figure 3A,B). These results indicated both rhFGF21 and AICAR could significantly inhibit H₂O₂-induced activation of the TGF-β1-Smad2/3-MMP2/9 signaling pathway.

To verify the effects of FGF21 and AICAR on cell apoptosis, antioxidant capacity, and collagen synthesis, we performed the TUNEL staining and detected the levels of Malondialdehyde (MDA), SOD2, COL-I and COL-III in CFs with H₂O₂ treatment. As shown in Figure 4, compared with the control cells, the level of MDA content and the protein expressions of COL-I and COL-III increased significantly (all p < 0.01), in contrast, SOD2 level decreased significantly after H₂O₂ treatment (p < 0.01). rhFGF21 and/or AICAR treatment increased the number of TUNEL positive particles (all p < 0.01) and SOD2 protein expression (all p < 0.01), and reduced the levels of MDA (all p < 0.01), COL-I (all p < 0.01) and COL-III (all p < 0.01) significantly in the H₂O₂-treated CFs. These results showed that AICAR and/or rhFGF21 intervention significantly promoted the CFs apoptosis and antioxidant capacity, and reduced collagen production in CFs with H₂O₂ treatment.

Based on the results of in vitro experiment, we also evaluated the protein expression levels of TGF-β1, Smad2/3, MMP2 and MMP9 in vivo. As shown in Figure 5, the protein expression levels of TGF-β1, Smad2/3, MMP2, and MMP9 increased significantly in the MI group when compared with the S group (all p < 0.01). AET and RET down-regulated the protein expressions of TGF-β1 (both p < 0.01), Smad2/3 (both p < 0.01), MMP2 (both p < 0.01) and MMP9 (both p < 0.01) in the infarcted heart (all p < 0.01). In addition, compared with the MAE group, the protein expression levels of TGF-β1 (p < 0.01), Smad2/3 (p < 0.01), MMP2 (p < 0.01) and MMP9 (p < 0.01) increased in the KME group. These results indicated that exercise training partly inhibited the MI-induced activation of the TGF-β1-Smad2/3-MMP2/9 signaling pathway, and knockout of Fgf21 attenuated the effects of AET.

We also detected the levels of oxidative stress and cell apoptosis in the mice with MI. The results showed that compared with the S group, the number of TUNEL positive cells (p < 0.01) and MDA content (p < 0.01) increased significantly, while protein expression of SOD2 (p < 0.01) and catalase (CAT) activity (p < 0.01) decreased in the MI group. Compared with the MI group, AET and RET reduced the number of TUNEL positive cells (all p < 0.01) and MDA content (all p < 0.01), and increased the expression of SOD2 protein (all p < 0.01) and CAT activity (all p < 0.01) in the infarcted heart. Moreover, compared with the MAE group, the number of TUNEL positive cells (p < 0.01) and MDA content (p < 0.01) increased, and SOD2 level (p < 0.05) and CAT activity (p < 0.01) decreased in the KME group (Figure 6A–E). These results pointed out that exercise can inhibit MI-induced oxidative stress and apoptosis, and knockout of Fgf21 inhibited the protective effects of AET. We speculated that FGF21 would play a key role in exercise-induced cardioprotective effects following MI.

Wildtype male C57BL/6J mice, three months old, were purchased from the laboratory animal center of the Xi’an Jiaotong University (Xi’an, China; No: SCXK (Shan) 2017-003). Fgf21^loxP mice on the C57BL/6 background were purchased from The Jackson Laboratory (B6.129S6(SJL)-Fgf21tm1.2Djm/J, Bar Harbor, ME, USA). For inducing Fgf21 KO mice, Fgf21^loxP mice were injected with rAAV-CMV-EGFP-P2A-CRE-WPRE-bGHpA (rAAV-Cre, titer: 5.73 × 10¹² vg/mL, 15 µL/20 g, BrainVTA Co., Ltd., Wuhan, China) via caudal vein four weeks before the establishment of the MI model. The genotype identification results are shown in Figure S1.

All mice were housed in the animal room of the institute of sports biology, Shaanxi Normal University. The temperature of the animal room was controlled at 25 °C and mice were given access to food and water freely under 12 h light/dark cycles. All mice were fed to 10 months old for experimental intervention. All surgical procedures and experimental protocols were performed with the Guide for Using Animal Subjects, and approved by the ethical committee of Shaanxi Normal University (approval number: 201916003; approved on 7 July 2019).

MI model was established by ligation of the left anterior descending (LAD) coronary artery. In brief, the 10-month-old mouse was anesthetized with isoflurane, fixed in the supine position and the chest was opened to expose the heart. LAD was ligated with 6.0 silk suture at the position approximately 2 mm under the junction of the left atrial appendage and pulmonary conus. Electrocardiogram (ECG) was used in the whole process to monitor the surgery, and ST-segment elevation was recognized as the sign of a successfully established model. On the seventh day after surgery, the echocardiography was performed, mice with similar infarct degrees were used in this study and divided into four groups: MI-sedentary group (MI, n = 8), MI with aerobic exercise training (AET) group (MAE, n = 8), MI with resistance exercise training (RET) group (MRE, n = 8) and FGF21 KO with AET group (KME, n = 4). Sham-operated mice without ligation were used as a control group (S, n = 8).

After seven months feeding, WT mice were randomly divided into the control group (CON, n = 8), AET group (AET, n = 8), RET group (RET, n = 8), WBV group (WBV, n = 8), and ES group (ES, n = 8). Exercised mice were subjected to five weeks of different types of exercise training, including one-week adaptive training.

The AET protocol was performed on an animal treadmill (Zhenghua Technology Co., Hefei, China) as previously described [52]. On the first day, the training started at a speed of 8 m/min, for 10 min, and the speed increased to 12 m/min for 60 min on the fifth day. From the second week, the speed was kept at 12 m/min (oxygen consumption level is estimated at 76% of VO_2max) for 60 min, five days per week for four weeks. Before and after each training, warm-up and cool down were carried out by running at a pace of 6 m/min for 5 min.

The RET program was modified from a published article [53,54]. Maximum carrying load was evaluated for each mouse by using a vertical ladder (1.0 m height, 1.0 cm intervals and 80° incline). Maximum carrying load was checked by ladder-climbing nine times with progressive loads. The heaviest load which mice could carry to climb successfully was viewed as the maximum carrying load. On the first day of the adaptive training, mice climbed the ladder with no load. The load increased with 15% maximum carrying load every day, and at the fifth day, up to 60% of the maximum load, climbing one time per set, nine sets per day with 1 min rest between sets. From the second week, the load was 75% of the maximum load, and this load was kept until the end of the training, three times per set, nine sets per day with 1 min rest between sets, for four weeks.

The WBV program was performed by using a vibration table as previously described and modified [55]: the frequency was 13 Hz, and the peak amplitude was 2.0 mm. In the first week, the time of vibration training was 10 min per day and up to 15 min per day from the second week to the end.

The skeletal muscle ES program was performed as described [56]: with anesthesia, the mice were fixed in a lying position. The needle was connected to the SDZ-II electronic acupuncture instrument, using continuous pulses with an electrical frequency of 20 Hz and a current of 1 mA. In the first week, the electrical stimulation program was performed 10 min per day, and from the second week, up to 15 min per day for four weeks.

According to the effects of different types of training on WT mice, MI mice were subjected to five weeks of AET or RET. The programs were performed as previous results [53,54,57].

AET program for MI mice: The first day of adaptive training was 10 m/min for 10 min, and increased to 10 m/min for 50 min on the fifth day. From the second week, the training started at 10 m/min for 60 min per day, for four weeks.

RET program for MI mice: the adaptive training is the same as normal mice. From the second week, the load was 75% of the maximum load and the load was kept until the training ended, climbing one time per set, nine sets per day for four weeks.

Primary CFs were isolated from neonatal mice 1–3 days post-birth (dpb) according to the literature [58,59]. After being disinfected with alcohol, the heart was taken out from mice and the ventricular tissue was cut into small pieces and digested with digestion buffer (0.04% trypsin, 20 mM 2,3-Butanedione 2-monoxime (BDM), 0.08% Type II collagenase, phosphate-buffered saline (PBS)) at 37 °C. The cell suspension was collected and added an equal volume of complete medium to terminate digestion. After centrifugation at 1200 rpm/min, we discarded the supernatant and resuspended the cell precipitate in complete medium, and cultured the cells in the incubator (Thermo Model 371, Marietta, OH, USA) for 90 min. Cardiomyocytes and CFs were collected by differential centrifugation. We collected the CFs, and cultured them with 10% FBS-DMEM. The morphology of myocardial fibroblasts should be fusiform, triangular and polygonal.

We collected cells and inoculated them into 6-well plates. When cell density reached to 70–80%, CFs were treated with H₂O₂ (100 μM, 6 h, to establish a model of oxidative stress-induced injury) [60], 5-aminoimidazole-4-carboxamide-1-β-d-ribofuranoside (AICAR, Selleck Chemicals, Houston, TX, USA, 1 mM, 15 h, to mimic the exercise effect) [49], and FGF21 Recombinant protein (rhFGF21, Selleck Chemicals, Houston, TX, USA, 100 ng/mL, 15 h) [61].

Before and after the whole exercise process, echocardiography was detected using an ultrasound cardio tachograph (VINNO 6 VET, VINNO, Suzhou, China) to assess cardiac functions. The mouse was fixed in the supine position and inhalation anesthetized with isoflurane mixed with oxygen (1:5). LVIDs, LVIDd and EF were recorded. The FS was calculated as (LVIDd−LVIDs)/LVIDd × 100%. Then, mice were sacrificed, the heart was quickly collected, and fixed in cold 4% formaldehyde or liquid nitrogen for subsequent experiments.

The heart tissue, fixed with paraformaldehyde for 48 h, was used for histological staining. After being washed with water, heart samples went through dehydration, transparent and paraffin embedding, then were cut into 5 μm thick microtome sections. Masson staining and Sirius red staining were performed on myocardial tissue sections to evaluate the degree of myocardial fibrosis. The collagen volume fraction was analyzed by Image Pro Plus analysis software (IPWIN Media Cybernetics, Inc., Rockville, MD, USA), and the percentage of CVF was calculated as collagen area/total area of myocardial tissue × 100%.

Heart tissue fixed with 4% paraformaldehyde was also used to perform frozen section (10 μm). After being fixed in cold acetone for 5–10 min, the frozen sections were washed with PBS (pH = 7.2), then incubated with 5% bovine serum albumin (BSA) for 1 h. Then, sections were incubated with rabbit polyclonal antibody Laminin (1:1000, Abcam, Waltham, MA, USA) at 4 °C overnight. On the next day, after washing with PBS, the sections were incubated with the tetramethyl rhodamine isothiocyanate (TRITC)-labeled goat anti-rabbit antibody (1:100, Jackson Immunoresearch, West Grove, PA, USA) at room temperature in darkness for 1.5 h. Then, they were washed with PBS and sealed with an anti-fluorescence quench agent. The images were observed and collected by using a fluorescence microscope (Nikon Eclipse 55i, Tokyo, Japan). Three sections of each sample were scanned, and 20 fields per section were viewed under a microscope. Image J software (National Institutes of Health, Baltimore, MD, USA) was used to analyze the cross-sectional area (CSA) of cardiomyocytes.

Cell apoptosis in heart tissues and CFs was detected by using a terminal deoxynucleotidyl transferase (TdT) dUTP nick-end labeling (TUNEL) assay (Beyotime, Shanghai, China). Sections of paraffin embedded tissue were dewaxed to water, incubated with protease K at 37 °C for 30 min, and washed with PBS. Then, sections were incubated with prepared TUNEL solution (TdT enzyme: fluorescent labeling solution = 1:9) at 37 °C for 60 min. The nuclei were stained with 4′,6-Diamidino-2-phenylindole (DAPI, 1:800, US EVERBRIGHT, Suzhou, China). After being washed with PBS, sections were sealed, and the images were captured using the fluorescence microscope (Nikon Eclipse 55i, Tokyo, Japan).

For CFs, the cell climbing sheets were fixed with 4% paraformaldehyde for 30 min, then incubated with 0.3% Triton X-100 solution for 5 min, and washed with PBS. The follow-up operation is the same as paraffin section.

MDA content and CAT activity in myocardial tissue homogenate and CFs cells were detected by using the assay kits (Jiancheng Biotech, Nanjing, China) according to the manufacturer’s protocols.

The heart samples and CFs harvested were lysed in the lysate mixture (RIPA: PMSF: phosphatase inhibitor = 100:1:1) and crushed by tissue homogenizer (FLUKO, Shanghai, China) and ultrasonic cell disruptor (Jining Tianhua Ultrasonic Electronic instrument Co., Jining, China). The supernatant was extracted and followed by protein quantification and denaturation. The total protein was separated by SDS-PAGE and then transferred to nitrocellulose membranes (Millipore, Bredford, MA, USA). The membranes were incubated in 3% BSA at room temperature for 1.5 h and then incubated with primary antibodies: FGF21 (1:1000 dilution, Abcam, Waltham, MA, USA), TGF-β1 (1:1000, Proteintech, Rosemont, IL, USA), Smad2/3 (1:1000, Abcam, Waltham, MA, USA), MMP2 (1:1000, Proteintech, Rosemont, IL, USA), MMP9 (1:1000, Proteintech, Rosemont, IL, USA), COL-I (1:500, Affinity Biosciences, Cincinnati, OH, USA), COL-III (1:500, Proteintech, Rosemont, IL, USA), SOD2 (1:1000, GeneTex, Irvine, CA, USA). GAPDH was used as a loading control for protein normalization. The next day, the membranes were washed with Tris-Buffered Saline and Tween 20 (TBST) three times, and then incubated with the horseradish-peroxidase (HRP)-conjugated secondary antibody at room temperature for 1 h. Reactive bands were detected using enhanced chemiluminescence reagent (Bio-Rad, Berkeley, CA, USA), analyzed using a digitalized Bio-Rad ChemiDocTM MP Imaging system (Universal Hood III, Bio-Rad, Berkeley, CA, USA), and quantified using Imagelab software 5.1 (Bio-Rad, Berkeley, CA, USA).

Image Lab Software 5.1 (Bio-Rad, CA, USA) was used to analyze the Western blotting results, GraphPad Prism 8.0.2 (GraphPad Software, La Jolla, CA, USA) was used for drawing the graphs. SPSS 21.0 statistical software (IBM Company, Armonk, NY, USA) was used for data analysis, and the differences between groups were determined by one-way analysis of variance (ANOVA). Values were expressed as mean ± Standard Deviation (X¯ ± SD) and significance levels were set at p < 0.05 and p < 0.01.