FGF21-FGFR4 signaling in cardiac myocytes promotes concentric cardiac hypertrophy in mouse models of diabetes

Fibroblast growth factor (FGF) 21, a hormone that increases insulin sensitivity, has shown promise as a therapeutic agent to improve metabolic dysregulation. Here we report that FGF21 directly targets cardiac myocytes by binding β-klotho and FGF receptor (FGFR) 4. In combination with high glucose, FGF21 induces cardiac myocyte growth in width mediated by extracellular signal-regulated kinase 1/2 (ERK1/2) signaling. While short-term FGF21 elevation can be cardio-protective, we find that in type 2 diabetes (T2D) in mice, where serum FGF21 levels are elevated, FGFR4 activation induces concentric cardiac hypertrophy. As T2D patients are at risk for heart failure with preserved ejection fraction (HFpEF), we propose that induction of concentric hypertrophy by elevated FGF21-FGFR4 signaling may constitute a novel mechanism promoting T2D-associated HFpEF such that FGFR4 blockade might serve as a cardio-protective therapy in T2D. In addition, potential adverse cardiac effects of FGF21 mimetics currently in clinical trials should be investigated.

The global prevalence of type 2 diabetes (T2D) is growing rapidly and is expected to reach 300 million by 2025. Diabetes is a risk factor for congestive heart failure, and cardiovascular complications are the leading cause of diabetes-related morbidity and mortality 1 . Diabetic cardiomyopathy is termed as heart failure in the absence of other comorbid conditions and is associated with both diastolic and systolic cardiac dysfunction, as well as pathological cardiac myocyte hypertrophy 1 . Cardiac remodeling includes defects in calcium handling and myocyte contractility, altered myocyte metabolism, and an increase in oxidative stress and mitochondrial dysfunction, as well as induction of interstitial myocardial fibrosis. Over 40% of T2D patients have increased left ventricular (LV) mass and wall thickness, conferring additional risk of heart failure that is independent of the other features of metabolic syndrome (hypertension, hyperlipidemia, and obesity) often present in these patients [2][3][4] . Importantly, hyperglycemia and the associated metabolic abnormalities in T2D are considered major factors promoting heart failure with preserved ejection fraction (HFpEF) 5 . An area of intense current research, HFpEF represents about half of all patients with heart failure and is a syndrome for which therapeutic options are limited to the recent discovery of the efficacy of sodium-glucose cotransporter-2 (SGLT2) inhibitors 6 . Here we consider the potential role of FGF21 in the development of pathological concentric cardiac hypertrophy and diabetic cardiomyopathy.

FGF21 induces concentric cardiac hypertrophy in mice.
To test for possible effects on cardiac structure and function by elevated FGF21 levels, a FGF21 transgenic mouse (FGF21-Tg) was utilized in which FGF21 is systemically elevated by expression in the liver under the control of the apolipoprotein E (ApoE) promoter 39 . Consistent with the original report 39 , at 24 weeks of age these mice had reduced body weight and blood glucose levels, and a 200-fold increase in serum FGF21 levels ( Fig. 1a-c). While the analysis of cardiac function in FGF21-Tg mice was confounded by their significantly reduced body weight and apparent failure-to-thrive, by echocardiography these mice exhibited a persistent elevation in ejection fraction and relative LV hypertrophy, as indicated by increased LV mass to body weight ratio and relative wall thickness ("concentricity index") 40 Table 1). Relative cardiac hypertrophy in FGF21-Tg mice was confirmed by gravimetric measurement of heart weight to body weight ratio (Fig. 1g). Paradoxically, cross-sectional area of cardiac myocyte was not increased in FGF21-Tg mice, presumably lower in proportion to the decreased body weight (Fig. 1h,i). Histological and gene expression analyses did not show evidence of cardiac fibrosis in FGF21-Tg mice (Supplementary Fig. 2a-d). Taken together, overexpression of exogenous FGF21 in mice appeared to induce a relative concentric cardiac hypertrophy, although the decreased body weight of the transgenic mice precluded definitive conclusions about the cardiac phenotype.
To determine if FGF21 elevation could induce concentric cardiac hypertrophy in the absence of confounding effects on body weight, adult wildtype BALB/cJ mice were injected intravenously (i.v.) twice daily with 40 μg/kg of recombinant FGF21 protein or isotonic saline for five days, followed by echocardiography and tissue collection on the sixth day. Echocardiographic analysis revealed that FGF21 increased LV anterior wall thickness, LV mass and concentricity index (  Table 2). Accordingly, gravimetric and histological analysis showed an elevation in indexed heart weight (Fig. 1l,m) and in the cross-sectional area of individual cardiac myocytes (Fig. 1n,o). These results demonstrate that elevated FGF21 can rapidly induce concentric cardiac hypertrophy in wildtype mice. FGF21 promotes the concentric hypertrophy of adult rat ventricular myocytes in vitro. Systemic FGF21 elevation in vivo resulted in concentric cardiac hypertrophy. To demonstrate that FGF21 can induce cardiac myocyte hypertrophy in a cell autonomous manner, primary adult rat ventricular myocytes (ARVM) were studied in vitro. Stimulation by 25 ng/ml FGF2, an FGF isoform known to induce myocyte hypertrophy 41 , but not stimulation by FGF21 induced the hypertrophy of myocytes cultured in minimal medium (data not shown). Given that elevated FGF21 serum levels are associated with diabetic cardiomyopathy 8  www.nature.com/scientificreports/ cyte width by 6%, while not affecting myocyte length, resulting in a significantly increased width:length ratio indicative of a mild "concentric" myocyte hypertrophy ( Fig. 2a-d). Remarkably, addition of FGF21 to medium containing high glucose further increased myocyte growth in width without affecting myocyte length, such that ARVM cultured in FGF21 and 15.6 mM glucose were 12% wider and had a width:length ratio 10% greater than control cells. We next considered that FGF21 might drive concentric remodeling in the presence of other stress stimuli. The α-adrenergic agonist phenylephrine induces a selective growth in width of cultured ARVM, while the β-adrenergic agonist isoproterenol induces growth in both width and length, resulting in a more symmetric ARVM hypertrophy 38 . Co-treatment with FGF21 did not significantly further increase the prominent growth in width induced by phenylephrine or isoproterenol (Fig. 2e-h). Remarkably, however, co-treatment with FGF21  www.nature.com/scientificreports/ inhibited the growth in length induced by isoproterenol, resulting in an increased width:length ratio comparable to that induced by phenylephrine and FGF21. These findings suggest that in the presence of hypertrophic stimuli, including elevated glucose levels, FGF21 will promote a "concentric" phenotype characterized by preferential growth in width of the cardiac myocytes. Concentric cardiac myocyte hypertrophy can be induced by activation of the ERK1/2 signaling pathway 36 . Addition of the MEK inhibitor PD98059 significantly inhibited the growth in width and increased width:length ratio of myocytes cultured in the presence of both FGF21 and high glucose ( Fig. 3a-c). Western blot analyses revealed that co-treatment of ARVMs with FGF21 and glucose significantly increased the levels of phosphorylated ERK1 (Fig. 3d-f). Similarly, cardiac tissue from FGF21-Tg mice showed increased levels of phosphorylated ERK1/2 (Fig. 3g), as well as elevated mRNA levels of Egr-1 and c-Fos (Fig. 3h,i) which are downstream targets of ERK1/2 signaling. ERK1/2 signaling is induced by FGF21 in cells expressing β-klotho and FGFRs 9 , and β-klotho (Klb) expression has been detected in ARVMs (Supplementary Table 3 and 30 ). In a binding assay using recombinant proteins, we found that in the presence of β-klotho, FGF21 bound the Fc-coupled ectodomain of FGFR4 preferentially to FGFR1c, while FGFR2c and FGFR3c did not detectably bind FGF21 at all (Fig. 3j). Interestingly, Fgfr4 expression in ARVM was induced synergistically by high glucose and FGF21 (Fig. 3k), providing a potential mechanism for the permissive effects of high glucose on FGF21 signaling. Importantly, using a blocking antibody for FGFR4 42 , the synergy between glucose and FGF21 in promoting adult myocyte growth in width was found to require FGFR4 activation ( Fig. 2b-d).
Diabetic BTBR ob/ob mice develop FGFR-dependent cardiac hypertrophy. As high serum FGF21 levels are present in diabetic cardiomyopathy 8 and FGF21 and high glucose directly promoted cardiac myocyte growth in width, we considered that FGF21/FGFR signaling might promote cardiac hypertrophy in this disease state. The genetic ob/ob mouse model lacking leptin is grossly obese and develops elevated blood glucose 43,44 and FGF21 levels 14,45 , while exhibiting cardiac hypertrophy at six months of age 46 . ob/ob mice in the BTBR genetic background, which develop severe diabetes 47 , were treated at three months of age with the pan-FGFR inhibitor PD173074 by daily intraperitoneal (1 mg/kg i.p.) injection for six weeks. At endpoint, the elevated body weight and blood glucose levels in ob/ob mice 47,48 were modestly reduced by PD173074 treatment, reaching significance only for body weight (Fig. 4a,b). Similarly, serum FGF21 levels were increased 15-fold in BTBR ob/ob mice compared to wild-type littermates and not significantly altered by PD173074 treatment (Fig. 4c). BTBR ob/ob mice developed cardiac hypertrophy without fibrosis ( Supplementary Fig. 2e,f), as indicated by increased LV wall    www.nature.com/scientificreports/ thickness (Fig. 4d,e) and cardiac myocyte cross-sectional area (Fig. 4f,g). Notably, the ob/ob-associated hypertrophy was inhibited significantly by PD173074 treatment. These results show that inhibition of FGFR signaling can diminish the development of cardiac hypertrophy despite minimally affecting hyperglycemia and obesity.
FGFR4 blockade attenuates cardiac hypertrophy in diabetic db/db mice. Our in vitro data suggest that FGFR4 is the primary FGFR isoform transducing hypertrophic FGF21 signals. This hypothesis was tested in vivo using the FGFR4 blocking antibody, in this case using db/db mice 49 , which lack the leptin receptor and develop a more severe cardiomyopathy that includes interstitial myocardial fibrosis. db/db mice, whether saline-injected controls or injected i.p. bi-weekly with FGFR4 blocking antibody (25 mg/kg) for 24 weeks starting at four weeks of age, exhibited elevated body weight, blood glucose and serum FGF21 levels ( Fig. 5a-c).
Notably, the progressive increase in LV wall thickness and LV mass in db/db mice was prevented by anti-FGFR4 treatment (Fig. 5d,e; Supplementary Fig. 1c). In addition, the increase in concentricity index at endpoint in db/ db mice (Supplementary Table 3), indicative of concentric cardiac hypertrophy, was prevented by treatment with the FGFR4 antibody. Gravimetric analysis confirmed that the blocking antibody inhibited db/db-induced cardiac hypertrophy (Fig. 5f). Histologically, db/db mice had increased cardiac myocyte cross-sectional area (Fig. 5g,h), as well as interstitial myocardial fibrosis ( Supplementary Fig. 2 g-j), that were both attenuated by the FGFR4 antibody. These results demonstrate that in diabetic db/db mice, FGFR4 activation promotes pathological cardiac remodeling, including concentric hypertrophy and interstitial fibrosis, providing further evidence for a critical role for FGF21-FGFR4 signaling in the cardiomyopathy associated with T2D.

Discussion
The development of concentric cardiac hypertrophy is a key feature of the cardiomyopathy associated with diabetes and HFpEF 5 . Here we provide evidence that elevation in systemic FGF21 levels and the associated cardiac myocyte signaling constitutes a previously unrecognized mechanism for the induction of pathological concentric remodeling, thereby defining a new candidate target for therapeutic intervention in T2D. Blockade of FGFR signaling using both a pan-FGFR small chemical inhibitor and a FGFR4-specific blocking antibody inhibited the concentric hypertrophy associated with diabetic cardiomyopathy in mice. In addition, inhibition of FGFR4 signaling using the blocking antibody and a MEK inhibitor prevented the selective growth in width of adult myocytes in vitro induced synergistically by high glucose concentrations and FGF21. Thus, both gain and loss of FGF21-FGFR4 signaling in mouse models, as well as in vitro studies using adult cardiac myocytes, suggest that elevated FGF21 signaling promotes pathological concentric cardiac hypertrophy. Taken in the context of previously published human studies showing elevated FGF21 levels in patients with metabolic syndrome and HFpEF 9,23,24 , we proposed that FGF21 signaling is a conserved critical mechanism contributing to pathological concentric hypertrophy and heart failure. We found that FGF21 induces myocyte hypertrophy via FGFR4. While others have shown that FGF21 could bind FGFR4-β-Klotho heterodimers 50 , as we show here, it has been suggested that FGF21 cannot activate downstream signaling via FGFR4 51,52 . Instead, we found that in adult cardiac myocytes, induction of hypertrophy by FGF21 could be inhibited by a FGFR4-specific blocking antibody. FGF21 is not the only member of the FGF family that can induce cardiac hypertrophy via ERK1/2 signaling. For example, presumably through activation of FGFR1, the paracrine factor FGF2 promotes cardiac hypertrophy under conditions of excess catecholamine stimulation and renin-angiotensin system activation 53 . In addition, we have shown that FGF23-FGFR4 signaling can induce cardiac hypertrophy in chronic kidney disease (CKD) 54,55 . However, of the endocrine FGFs, the association between elevated FGF21 systemic levels and diabetes and obesity is most well established 8 . Whether FGF23 is increased in human obesity is controversial 56 , and we found that FGF23 was not elevated in either the ob/ob or db/db mouse (data not shown). In addition, FGF19 levels are reduced in human obesity 8 . Thus, while FGF23 may be an important mediator of cardiac hypertrophy in CKD, FGF21 may have an especially important role in the induction of concentric hypertrophy in diabetic cardiomyopathy.
The identification of FGF21 as a mediator of pathological concentric hypertrophy stands in contrast to a literature purporting the beneficial effects of FGF21 on the cardiovascular system. While not excluding other mechanisms, including indirect systemic effects, many of these differences may be explained by ERK1/2-mediated differential regulation of concentric and eccentric cardiac hypertrophy and ERK1/2-mediated cardioprotection 36,57 . As observed in vitro for isoproterenol-treated myocytes, FGF21 may promote concentric, but oppose eccentric hypertrophy in vivo, regulating the balance between myocyte growth in width and length like ERK1/2 signaling 36 . Both ERK1/2 activation by expression of a constitutive MEK1 transgene and ERK1/2 (Mapk1/Mapk3) double knock-out resulted in cardiac hypertrophy; however, the former results in a non-fibrotic concentric hypertrophy with thickened myocytes 58 , while the latter resulted in ventricular dilation and elongated myocytes 36 . This model can explain the exaggerated eccentric hypertrophy in Fgf21 targeted mice subjected to chronic isoproterenol infusion that promotes eccentric hypertrophy 26 . Likewise, the absence of fibrosis in wildtype mice given pharmacologic doses of FGF21 phenocopies MEK1 transgenic mice 58 .
Presumably via different downstream effectors that regulate myocyte apoptosis, MEK1 transgenic mice and ERK2 hemizygous knock-out mice exhibit decreased and increased infarct size, respectively, after ischemia-reperfusion 57 . Likewise, gain and loss of ERK1/2 signaling may explain the beneficial and deleterious effects on myocyte survival and infarct size of exogenous FGF21 and Fgf21 knock-out, respectively, in mice subjected to ischemia-reperfusion injury or permanent coronary artery ligation 27,28 , although other ERK1/2-independent pathways may also be involved 31,33,34 . This pro-survival signaling could complicate the targeting of FGF21-FGFR4 signaling in diabetic cardiomyopathy. However, while Fgf21 deletion increased myocyte apoptosis in streptozotocin-induced diabetic mice, it did not further worsen cardiac dysfunction 59 . www.nature.com/scientificreports/ The identification of FGF21-FGFR4 signaling as a key regulator of concentric cardiac hypertrophy might provide a novel therapeutic opportunity to prevent or treat heart failure in diabetes. Targeting this pathway might complement the use of SGLT2 inhibitors that have recently been shown to be efficacious in clinical trials for diabetic cardiomyopathy and HFpEF 60,61 . While concentric cardiac hypertrophy can in theory reduce ventricular wall stress (Law of LaPlace), left ventricular hypertrophy is a major risk factor for future heart failure, such that extensive preclinical evidence suggests that preventing concentric hypertrophy even in the presence of persistently increased afterload (hypertension) would be ultimately beneficial 62 . In the context of diabetic cardiomyopathy, concentric hypertrophy predisposes the heart to significant interstitial fibrosis and diastolic dysfunction 63 . As shown here, inhibited FGFR4 signaling attenuated both cardiac hypertrophy and interstitial  www.nature.com/scientificreports/ myocardial fibrosis in db/db mice. Similarly, inhibition of FGFR4 is a potential approach that might be useful for the cardiomyopathy associated with CKD 54 . Besides the potential use of biologic agents such as blocking antibodies, there are several recently developed small molecule FGFR4-specific inhibitors being considered for cancer indications in patients 64 . Alternatively, FGF21 neutralizing antibodies might be developed to inhibit FGF21 function in diabetic cardiomyopathy. Whether non-cardiac effects of FGF21-FGFR4 inhibition would be tolerated in the context of a chronic cardiovascular therapy will determine the potential of this approach. The lack of a severe phenotype in mice with global Fgfr4 deletion is encouraging 65,66 . Regardless, readily reversible therapies would be desirable given the beneficial effects of FGF21 in ischemic heart disease, especially since those with diabetes are at risk for a coronary event.
Whether or not FGF21-FGFR4 signaling will prove to be a successful target for intervention against diabetic cardiomyopathy, the observation that FGF21 promotes concentric cardiac hypertrophy should be considered by the community actively developing FGF21 analogs and mimetics as agents to promote weight loss, hyperglycemia and dyslipidemia 8 . Besides the potential for bone loss, as well as the possibility that there may be "resistance" to the metabolic effects of elevated FGF21 in T2D 7 , further elevation in FGF21's systemic activity may exacerbate the development of diabetic cardiomyopathy and other conditions characterized by concentric cardiac hypertrophy. (human monoclonal, U3-11) was isolated by U3 Pharma/Daiichi-Sankyo (Germany) as described before 67 . This antibody is specific for FGFR4 and does not inhibit the activation of FGFR1-3 42 . We used a horseradish peroxidase-conjugated anti-human antibody (109035098, Jackson Immunolabs) for the plate-based binding assay. We used anti-ERK1/2 (9102 and 4695, Cell Signaling), anti-phospho-ERK1/2 (Thr202/Tyr204) (9101, Cell signaling) and anti-GAPDH (CB1001, EMD Millipore) for Western blot analyses.

Isolation and analyses of adult rat ventricular myocytes. Eight-week-old male Sprague-Dawley
rats were anesthetized with Ketamine (40-100 mg/kg i.p.) and Xylazine (5-13 mg/kg i.p.). Hearts were extracted and perfused using a Langendorff system with calcium depletion perfusion buffer containing 120 mM NaCl, 5.4 mM KCl, 1.2 mM NaH 2 PO 4 ·7H 2 O, 20 mM NaHCO 3 , 1.6 mM MgCl 2 ·6H 2 O, 5 mM taurine, and 5.6 mM glucose, 10 mM 2,3-butanedione monoxime (BDM), followed by type II collagenase digestion at 37 °C. The ventricular tissue was cut using sterilized scissors, and the cells were disassociated by pipetting and then filtered using 200 µm mesh. Ventricular myocytes (ARVMs) were collected by centrifugation at 50g for 1 min. ARVMs were then washed and purified using perfusion buffer containing 0.1% fatty acid-free BSA, 1 mM ATP (pH 7.2), 1 mM glutamine with CaCl 2 added gradually until reaching a Ca 2+ concentration of 1.8 mM. The purified ARVMs were then resuspended in a defined Minimal Medium [Medium 199 (11150059, Thermo Fisher, United States) supplemented with 5 mM creatine, 2 mM l-carnitine, 5 mM taurine, 25 mM HEPES, 0.2% fatty acid-free BSA, 10 mM BDM and insulin-transferrin-selenium (ITS)] and seeded onto square cover slips (21 mm × 21 mm) coated with mouse laminin (10 μg/mL diluted in PBS) in 6-well plates (10,0000 cells per well). One hour after isolation, the culture medium was removed, and ARVMs were washed with 2 mL medium once, then cultured for 48 h in Minimal Medium containing recombinant FGF21 (25 ng/mL, dissolved in 0.1% BSA) or 0.1% BSA control with and without phenylephrine (20 μM), isoproterenol (10 μM) or additional glucose (10 mM, final 15.6 mM) as indicated. To block FGFR4, isolated ARVMs were pre-treated with Minimal Medium containing FGFR4 blocking antibody (10 mg/mL human monoclonal U3-11; U3Pharma) for one hour before addition of FGF21 (25 ng/mL) and glucose (15.6 mM final). For morphometric analysis, ARVMs were washed twice with PBS and fixed with 3.7% formaldehyde solution (diluted in PBS) for one hour. Coverslips were mounted onto slides, and nine images per slide were obtained using a Leica DM4000 Microscope. The width and length of 100 ARVMs/image were measured by Leica LAX software. ARVM length and width were defined by the maximum dimensions of the cell either parallel and perpendicular to sarcomeric striations, respectively.
For Western blot analyses, ARVMs were washed with cold PBS and lysed in buffer (20 mM Tris-HCl, pH 7.6, 150 mM NaCl, 2 mM Na 2 EDTA, 1% NP-40, 1% sodium deoxycholate, 0.1% SDS, 2.5 mM sodium pyrophosphate, 1 mM beta-glycerophosphate, 1 mM Na 3 VO 4 , 5 mM NaF, and protease inhibitors). Protein concentrations were measured by Bio-Rad Protein Assay. Proteins were separated by SDS-PAGE electrophoresis. Precision Plus Protein marker was used to identify protein sizes. Equal loading was confirmed by Ponceau S total protein staining. Blots were incubated with primary antibody followed by donkey anti-IgG horseradish peroxidase antibody conjugates (Jackson Immune Research) secondary antibody. Super-Signal West Chemiluminescent Substrates (Thermo Scientific) and an Amersham Imager 600 Imaging System (GE Healthcare Life Sciences) were used to detect signals. ImageQuant™ TL software was used to quantify signals.
Plate-based binding assay. To measure FGF21 binding to FGFRs, 96-well plates (Thermo Fisher; #44-2404-21) were coated with 200 ng of human FGF21 protein in 100 μL of coating buffer (E107, Bethyl) per well at 4 °C overnight. Plates were washed 5× with 350 μL of assay buffer (50 mM Tris pH 7.4, 200 mM NaCl, 0.01% Tween 20) on a 50TS microplate washer (BioTek, United States). Plates were blocked for 1 h in 200 μL assay buffer with 0.5% BSA (Rockland; BSA-50). Plates were washed 5× with assay buffer and incubated with 100 μL assay buffer with 0.5% BSA in combination with PBS or 1 μg of human soluble β-klotho at room temperature. After 1 h, plates were washed, and 500 ng of FGFR-Fc in 100 μL volume of assay buffer with 0.5% BSA were added per well. After 1 h at room temperature, plates were washed 5× in assay buffer and incubated with 100 μL RNA isolation and real-time PCR. RNA from total mouse hearts was extracted using Trizol (15596026, Invitrogen, United States). 2 μg total RNA was reverse transcribed using Qscript (95048, Quanta Biosciences, United States). Quantitative PCR reactions were carried out in the StepOne plus Real-Time PCR System (Applied Biosystems, United States) using FAST SYBR Green Master Mix (4385610, Applied Biosystems). Raw data was quantified via StepOneTM software v2.3 from life Technologies. Relative gene expression was normalized to expression levels of GAPDH and evaluated using the 2 −ΔΔCt method. ARVM RNA was extracted using RNeasy mini kit (74104, Qiagen, United States) according to manufacturer's instructions. 500 ng total RNA was reverse transcribed using a Reverse Transcription Kit (4368814, Applied Biosystems). Quantitative PCR reactions were performed using the QuantStudio 5 System (Thermo Scientific, United States) using Powerup SYBR Green Master Mix (A25742, Thermo Scientific). Raw data was analyzed using QuantStudio 5 qPCR Data Analysis Software (Thermo Scientific). Relative gene expression was normalized to expression levels of Hprt1 and evaluated using the 2 −ΔΔCt method. Primer sequences are presented in Table 1.
FGF21 transgenic mice. FGF21 transgenic mice expressing a mouse FGF21 cDNA under the control of the apolipoprotein (ApoE) promoter (C57BL/6-Tg(Apoe-Fgf21)1Sakl/J; The Jackson Laboratory Stock No: 021964) were as previously described 39 . Eleven hemizygous and 10 wild type mice, male and female, were analyzed by echocardiography at 4 and 6 months of age before euthanasia for further analysis. For molecular analysis of ERK1/2 signaling, RNA and protein was isolated from heart tissue of 8-12 week-old male mice. Expression levels of c-Fos and Egr-1 were determined by qPCR and the activation/phosphorylation of ERK1/2 by immunoblotting.
For Western blot analyses, hearts were lysed in RIPA-based buffer (50 mM sodium phosphate pH 7.5, 200 mM NaCl, 1% Triton X-100, 0.25% deoxycholic acid) with addition of protease inhibitors (11836153001, Roche) and phosphatase inhibitors (P5726, P0044, Sigma-Aldrich) for 30 min. The lysate was centrifuged at 20,000g for 60 min to remove tissue debris. Samples were boiled in Laemmli sample buffer (1610737, Biorad) with 1.42 M 2-mercaptoethanol for 5 min. 20 μL of samples were loaded onto 12% SDS-PAGE gels and analyzed by Western blotting. FGF21 serial injections. FGF21 injections were conducted following a similar protocol as previously established for FGF23 by us 68 and others 69,70 . Briefly, 12-week-old, male and female BALB/cJ mice (Stock No: 000651; The Jackson Laboratory, United States) underwent tail vein injections. The day before the experiment, mice underwent echocardiographic analysis, as described below. Mice were anesthetized using 2.5% isoflurane and placed on a heat pad. Mice were divided into 2 groups of 10, receiving either vehicle solution (isotonic saline) or mouse FGF21 protein. Per injection, we used 40 μg/kg of FGF21 dissolved in 200 μL of isotonic saline, with 8 h between injections for a total of 5 consecutive days. All injections were performed via the lateral tail vein. On the morning of the 6th day, 16 h after the final tail vein injection, animals underwent echocardiographic analysis and were sacrificed. ob/ob mice. ob/ob mice lacking leptin 43 were studied in the genetic BTBR background (BTBR.Cg-Lep ob / WiscJ; Stock No: 004824; The Jackson Laboratory) 47 . At 3 months of age, 5 ob/ob and 5 wild type mice, male and female, were i.p. injected daily with PD173074 (P2499, Sigma Aldrich) at 1 mg/kg and 5 ob/ob and 4 wild type mice with saline for 6 weeks. Animals were sacrificed and samples prepared as described here. Echocardiography of mouse hearts. For mice receiving serial injections of recombinant FGF21 protein, transthoracic echocardiographic analysis was performed on day 6 of the experiment, at 13 weeks of age using a Vevo 770® High-Resolution Micro-Imaging System (FUJIFILM VisualSonics), equipped with an 707B-253 transducer. Animals were minimally anesthetized with 1-1.5% isoflurane, and normal body temperature was maintained using a rectal probe for temperature monitoring. For analysis, both B-and M-mode images were obtained in the short and long axis view. Correct positioning of the transducer was ensured using B-mode imaging in the long axis view, before switching to the short axis view. Parameters were calculated from at least three cardiac cycles by M-mode echocardiographs. All parameters were measured or calculated using Vevo® 770 Workstation Software (FUJIFILM VisualSonics). Serial transthoracic echocardiography was performed in FGF21 Tg mice at the age of 16 and 24 weeks, and in db/db mice every 4 weeks from 4 weeks of age, using a Vevo 2100 High-Resolution Imaging System (FUJIFILM VisualSonics). Animals were minimally anesthetized with 1-1.5% isoflurane, and normal heart rate and body temperature were monitored and maintained throughout the procedure. Parameters were calculated from at least three cardiac cycles by M-mode echocardiographs. All parameters were measured or calculated using the VevoLAB software (FUJIFILM VisualSonics). Measurement definitions of short axis and calculations (M-Mode) are presented in Tables 2 and 3.

Statistics.
Values are expressed as mean ± SEM if not otherwise indicated. Comparisons between 3 or more groups were performed by one-way ANOVA followed by post-hoc Tukey test or by 2-way ANOVA with posthoc Sidak's multiple comparison test. Comparisons between 2 groups were performed by two-tailed t-tests. A significance level of P ≤ 0.05 was accepted as statistically significant. No statistical method but experience from previous publications was used to predetermine sample size. No formal randomization was used in any experi-