Cardiac Myocyte KLF5 Regulates Body Weight via Alteration of Cardiac FGF21

Abstract

Cardiac metabolism affects systemic energetic balance. Previously, we showed that Krüppel-like factor (KLF)-5 regulates cardiomyocyte PPARα and fatty acid oxidation-related gene expression in diabetes. We surprisingly found that cardiomyocyte-specific KLF5 knockout mice (αMHC-KLF5^−/−) have accelerated diet-induced obesity, associated with increased white adipose tissue (WAT). Alterations in cardiac expression of the mediator complex subunit 13 (Med13) modulates obesity. αMHC-KLF5^−/− mice had reduced cardiac Med13 expression likely because KLF5 upregulates Med13 expression in cardiomyocytes. We then investigated potential mechanisms that mediate cross-talk between cardiomyocytes and WAT. High fat diet-fed αMHC-KLF5^−/− mice had increased levels of cardiac and plasma FGF21, while food intake, activity, plasma leptin, and natriuretic peptides expression were unchanged. Consistent with studies reporting that FGF21 signaling in WAT decreases sumoylation-driven PPARγ inactivation, αMHC-KLF5^−/− mice had less SUMO-PPARγ in WAT. Increased diet-induced obesity found in αMHC-KLF5^−/− mice was absent in αMHC-[KLF5^−/−;FGF21^−/−] double knockout mice, as well as in αMHC-FGF21^−/− mice that we generated. Thus, cardiomyocyte-derived FGF21 is a component of pro-adipogenic crosstalk between heart and WAT.

1. INTRODUCTION

The heart is a central regulator of systemic metabolism as it both consumes a significant amount of fatty acids for ATP synthesis (1) and affects metabolism in other tissues, as well. Heart specific changes in lipolysis regulate circulating levels of triglycerides (2). The heart also functions as an endocrine organ. Atrial natriuretic peptide (ANP) and brain natriuretic peptide (BNP) (3) secreted from the heart promote lipolysis in adipose tissue (4–7). Metabolites originating from the heart, such as phospholipase A₂, also affect systemic metabolism (8, 9). A series of studies have identified MED13 as a cardiomyocyte regulator of systemic metabolism (10, 11). Mediator complex subunit 13 (MED13) transgenic expression reduced diet-induced obesity (DIO), whereas deletion of this factor increased obesity (10, 11). MED13 is a member of the mediator complex that controls DNA transcription (12) and this gene is regulated by miR-208a (11). How MED13 regulates systemic metabolism is unknown.

PPARγ is a central transcription factor that is essential for adipocyte development (13). Insulin and steroids regulate PPARγ expression (14, 15), and its activity is modulated by association with lipids such as certain unsaturated fatty acids and eicosanoids (16), and association with retinoid X receptor (RXR) (17). Moreover, the activity of PPARγ can be regulated by multiple post-translational modifications, such as phosphorylation, sumoylation (SUMO), ubiquitylation, and O-GIcNAcylation (18).

FGF21, another regulator of adipose development, is mainly produced in the liver (19), but is also expressed in other tissues like white adipose tissue (WAT) (20, 21), brown adipose tissue (BAT) (22, 23), skeletal muscle (24, 25), duodenum (23) pancreas (23, 26, 27), and the heart (23, 28, 29). FGF21 signals primarily through the β-Klotho/FGFR1c receptor complex (30, 31). Induction of FGF21 production, as well as exogenous FGF21 administration, stimulates lipolysis in WAT (32, 33), increases browning of WAT (34), and improves insulin sensitivity. Given these functions, FGF21 has been proposed as a therapeutic agent of diabetes and obesity, and increased circulating levels of FGF21 in obesity has been described as an FGF21-resistance state (35). A recent study showed that exercise reverses diet-induced FGF21 resistance via increased adipose PPARγ acitivity (36). Several studies show beneficial effects of FGF21 treatment on body weight, fat mass, and lipid and glucose metabolism of animal models and obese and diabetic patients (37–39). On the other hand, various studies have associated greater adiposity with FGF21 expression (40, 41) and have shown that FGF21 can also decrease lipolysis (42, 43). These FGF21 signaling actions in the WAT has been linked to inhibition of sumoylation of PPARγ at Lys107, resulting in increased PPARγ transcriptional activity that enhances adipogenesis (44–46). Thus, the final positive or negative effect of FGF21 in WAT expansion and obesity seems to depend on a multifactorial signaling network that has not been fully elucidated.

Our recent study associated Krüppel-like factor (KLF)-5 with transcriptional regulation of Ppara, a central regulator of cardiac fatty acid oxidation (FAO) (47). Others have shown that KLF5^+/− mice are protected from diet-induced obesity, because of increased PPARδ-mediated energy expenditure in skeletal muscle (48). In this report, we show that αMHC-KLF5^−/− mice fed with high fat diet (HFD) had increased body weight gain compared to floxed mice on HFD, associated with reduced PPARγ sumoylation. αMHC-KLF5^−/− mice had reduced cardiac Med13 expression and we found that KLF5 is a positive regulator of Med13 expression and causes increased heart expression of FGF21. Thus, heart production of FGF21 regulates adipose development.

2. METHODS

2.2. Mouse studies

All animal studies were approved by the institutional animal care and use committees of Temple University in Philadelphia PA, Columbia University in New York NY, or Yale University in New Haven CT and mice were cared for in accordance with NIH guidelines. Mice were housed three per cage and maintained under appropriate barrier conditions in a 12hr light-dark cycle and received food and water ad libitum. The αMHC-KLF5^−/− mice have been described before (47). αMHC-KLF5^−/− mice were crossed with floxed-FGF21 mice generating mice with cardiomyocytes-specific double Klf5 and Fgf21 gene deletion αMHC-[KLF5^−/−;FGF21^−/−]). We used 4 to 22 week old male αMHC-KLF5^−/−; αMHC-[KLF5^−/−, αMHC-[KLF5^−/;FGF21^−/−], or floxed (αMHC-KLF5^−/− mice, which were compared with floxed-KLF5 mice and αMHC-[KLF5^−/;FGF21^−/−] with floxed FGF21 or double floxed-FGF21;KLF5 mice) mice weighing ~9-28g. All analyses involving animals were performed with at least 3 mice per experimental group.

2.3. Cell lines

HL-1 mouse cardiac muscle cell line were maintained in Claycomb medium (Sigma-Aldrich #51800C, St. Louis, MO, USA) supplemented with 10% fetal bovine serum (Sigma-Aldrich #121030, St. Louis, MO, USA), 100 U/ml penicillin/streptomycin (Sigma-Aldrich #P4333, St. Louis, MO, USA), 0.1 mM norepinephrine (Sigma-Aldrich #A0937, St. Louis, MO, USA), and 2 mM l-glutamine (Sigma-Aldrich #G7513, St. Louis, MO, USA) at 5% C0₂ at 37°C (49).

2.4. Primary cardiomyocyte isolation

Adult mouse cardiomyocytes were isolated from ventricles of floxed control mice and αMHC-FGF21^−/− mice as described previously (50) with minor modifications. Hearts from heparinized mice (90 USP; ip) were cannulated through the aorta. Hearts were perfused with perfusion buffer (120.4 mM NaCI, 14.7 mM KCL, 0.6 mM NaH₂PO₄, 0.6 mM KH₂PO₄, 1.2 mM MgSO₄, 10 mM Na-Hepes, 4.6 mM NaHC0₃, 30 mM taurine, 10 mM BDM, 5.5 mM glucose; pH 7.0) for 3 min followed by perfusion buffer containing 285 units/ml Collagenese type II (Worthington), 0.1-0.12 mg/ml trypsin and 0.02 mM CaCI₂for 7 min. Ventricles were gently teared in small pieces, perfusion buffer containing 5 mg/ml BSA and 0.125 mM CaCI₂ was added and the cell suspension filtered with 100 pm nylon. The filtrate was pelleted by gravity for 5 min, centrifuged for 30 sec at 0.7 rpm and the pellet resuspended in perfusion buffer containing 5 mg/ml BSA and 0.225 mM CaCI₂. The cells were pelleted by gravity for 10 min, centrifuged for 30 sec at 0.7 rpm and the pellet resuspended in perfusion buffer containing 5 mg/ml BSA and 0.525 mM CaCI₂. The cells were pelleted by gravity for 10 min, centrifuged for 30 sec at 0.7 rpm and the pellet resuspended in perfusion buffer containing 5 mg/ml BSA and 1.025 mM CaCI₂. The cells were pelleted by gravity for 10 min, centrifuged for 30 sec at 0.7 rpm and the pellet resuspended in TRIzol reagent (Invitrogen) for RNA isolation.

2.5. High fat diet treatment

αMHC-KLF5^−/−, αMHC-[KLF5^−/;FGF21^−/−], or floxed mice were fed HFD (60 kcal% fat, D12492i; Open Source Diet) starting when mice were 4 to 23 weeks old. Mice were sacrificed after 6 to 8 weeks HFD following anesthesia with isoflurane inhalation. Blood was collected, mice were perfused with PBS via the apex, and heart, posterior subcutaneous WAT, BAT, kidney, skeletal muscle, and liver were harvested. Heart, WAT, BAT, kidney, skeletal muscle, and liver samples were flash frozen and WAT, BAT, and liver samples were put in tissue tek, and stored at −80°C until further use.

2.6. Metabolic cages analysis

In a subset of HFD-fed αMHC-KLF5^−/− and floxed mice major determinants of energy balance, including feeding, activity, energy expenditure by indirect calorimetry and drinking, were measured using the Columbus Labs Comprehensive Lab Animal Monitoring System (CLAMS) (51). Mice were individually housed for three to four days prior to metabolic cage study and acclimated within the cages for an additional 24 hours. Data were collected for 48 hours and are reported as the 24-hour average per mouse.

2.7. RNA purification and gene expression analysis

Total RNA was purified from hearts, WAT, BAT, skeletal muscle, kidney, and liver using the TRIzol reagent according to the instructions of the manufacturer (Invitrogen). DNase-treated (Invitrogen) RNA was used for cDNA synthesis using the Protoscript II First-Strand cDNA Synthesis kit (New England BioLabs). cDNA was analyzed with quantitative real-time PCR that was performed with SYBR Select Master Mix (Applied Biosystems). qRT-PCR was performed on an Applied Biosystems StepOnePlus Real-Time PCR system. Samples were normalized against 18S ribosomal RNA (18S), ribosomal protein lateral stalk subunit P0 (RplpO / 36B4), or beta-actin (Actb). The BioRad Fgf21 primer assay (10025636) was used for Fgf21 detection. The sequences of the other primers have been described previously (47, 52), or are described in Table S1. Expression levels of miR-208 was analyzed using a microRNA PCR kit (Exiqon) and samples were normalized against 5S ribosomal RNA (5S), small nucleolar ribonucleoprotein U6 and small nucleolar RNA, C/D box 65 (Snord65).

2.8. Plasma FGF21, adiponectin, leptin, triglyceride, and glucose levels

Plasma and liver TG levels were measured using Infinity Triglyceride Reagent (Thermo Scientific) according to the manufacturer’s instructions. Glucose levels were measured in plasma using a blood glucose meter (Contour next; Bayer). ELISA kits were used to measure plasma levels of leptin (EZML-82K; EMD Millipore), adiponectin (EZMADP-60K; EMD Millipore), BNP (EIAM-BNP; RayBio), and FGF-21 (MF2100; R&D systems) according to the manufacturer’s instructions.

2.9. Protein and post-translational modifications analysis

IP: Total protein was isolated from whole fat tissue after grinding in 300 μl of RIPA buffer. Supernatants were collected and placed in fresh tubes where protein concentration was determined via Bradford assay (BCA Protein Assay kit, Fisher). 4 μl of Anti-PPARγ antibody (D69, Cell Signaling) were then added to 200 μg the total protein supernatants with a final concentration of 1 mg/ml and incubated O/N with rotation at 4°C. Then, we added 30 μl of Protein A agarose beads (Cell Signaling, #9863) to each sample and incubated at 4°C for 3 hrs. The samples were then microcentrifuged for 30 sec at 4°C and washed 5× using 500 μl 1× RIPA buffer. Pellets were then re-suspended in 40 μl 6× SDS-sample buffer (Boston Bioproducts) denatured by boiling and used for western immunoblotting. Western Blotting: Proteins were separated by electrophoresis in a 10% polyacrylamide gel. Samples were electrophoresed at 100-150 V for 1.5 h and the separating gel was equilibrated in transfer buffer (20 mm Tris-HCI, 150 mm glycine, 20% methanol, and 0.1% sodium dodecyl sulfate) for 10 min. The proteins were then transferred to Immobilon-FL membranes (Millipore, Billerica, MA) at 4°C, O/N. The membranes were then blocked for 1 h at room temperature in protein-free (PBS) blocking buffer (Pierce) and incubated at a 1:1,000 dilution with a mouse primary antibody against SUMOylated proteins (SUMO-1 2A12, Cell Signaling, # 5718 in blocking buffer, O/N at 4°C. The membrane was then blotted with the IRDye® 800CW Goat anti-Mouse IgG secondary antibody 1:15,000 (LI-COR) in blocking buffer for 1 h and the proteins were visualized and quantified using the Odyssey CLx Imaging system (LI-COR).

2.10. Histology

Cryosections of WAT and BAT were stained with H&E staining and cryosections of liver were stained with Oil Red O staining by the immunohistochemistry core at NYU, New York. Of each section up to up to 10 pictures were taken randomly at 10× or 20x magnification. WAT adipocyte size in αMHC-KLF5 or floxed mice was measured in images of 10× magnification in all cells (at least 93 cells per mouse) using ImageJ software. WAT adipocyte size in αMHC-[KLF5^−/;FGF21^−/−] or floxed mice was measured in images of 20× magnification in all cells (at least 28 cells per mouse) using ImageJ software.

2.11. Adenoviral infection

The recombinant adenovirus that expresses KLF5 was kindly provided by Dr. Ceshi Chen (53). HL-1 cells were infected with ad-KLF5 or control adenovirus-expressing GFP (Ad-GFP) at a multiplicity of infection (MOI) of 10 in medium supplemented with 2% Heat-Inactivated Horse Serum and 1% Penicillin/Streptomycin for 6 hours. The medium was then replaced with 10% FBS-containing medium and cells were harvested after an additional 48 hours (49).

2.12. Chromatin immunoprecipitation (ChIP)

HL-1 cells were subjected to ChIP as previously described (54). The antibodies used for ChIP were anti-KLF5 (07-1580; Millipore), and anti-IgG. Precipitated DNA was further analyzed with RT PCR.

2.13. Statistics

Data are expressed as the mean ± SEM. Western Blots are quantified using the Odyssey CLx Imaging system (LI-COR). Statistical significance was assessed using GraphPad Prism 6 software with the appropriate test; 2 tailed t test, 1-way ANOVA, or 2-way ANOVA followed by Tukey’s multiple comparisons test. A P<0.05 was considered statistically significant. The values of n, statistical measures (mean ± SEM), and statistical significance are reported in the figures and figure legends.

3. RESULTS

3.1. Cardiomyocyte-specific KLF5 deletion accelerates DIO

We previously showed that cardiomyocyte KLF5 is a positive transcriptional regulator of Ppara, as well as that KLF5 inhibition decreases FAO and eventually leads to cardiac dysfunction (47). As previous studies have correlated inhibition of cardiac fatty acid utilization with altered systemic lipid metabolism (55), we treated αMHC-KLF5^−/− mice with HFD for 6 weeks. Surprisingly, αMHC-KLF5^−/− mice showed accelerated DIO (Figure 1A and 1B), compared to HFD-fed control floxed mice. The percentage of body weight gain over 6 weeks of HFD-fed αMHC-KLF5^−/− mice was 1.55 times greater than HFD-fed control floxed mice (20% vs 31%) (Figure 1C). The increased body weight gain was associated with a trend (p = 0.12) for increased WAT weight (Figure 2A), and increased mRNA levels of proteins that are positively correlated with adipocyte lipid metabolism: Pparg1, Pparg2, lipoprotein lipase (Lpl), Cd36, diacylglycerol o-acyltransferase 2 (Dgat2), and glucose transporter 4 (Glut4) (Figure 2B). Histological analysis (H&E staining) of WAT and BAT showed larger adipocytes in WAT and more lipid accumulation in BAT obtained from HFD-fed αMHC-KLF5^−/− mice (Figure 2C and S1A). Also, Oil-Red-O staining and biochemical analysis indicated increased hepatic triglyceride (TG) content in HFD-fed αMHC-KLF5^−/− mice (Figure 2D and S1B). Plasma TG and glucose levels were similar in HFD-fed wild type mice and HFD-fed αMHC-KLF5^−/− mice (Figure S1C and S1D).

Figure 1: Cardiomyocyte-specific KLF5 deletion accelerates diet-induced obesity –

A: Representative pictures of control floxed or αMHC-KLF5^−/− mice after 6 weeks of HFD. B-C: Cumulative body weight gain (g) (B), body weight gain (%) (C) of control floxed (n = 6) and αMHC-KLF5^−/−(n = 13) mice after 6 weeks HFD (*p<0.05, **p<0.01, ***p<0.001 vs floxed with 2way ANOVA (B) or Student’s T-Test (C)). Data are presented as mean ± SEM. (See also Figure S1A–D) HFD: high fat diet, KLF: Krüppel-like factor.

Figure 2: Cardiomyocyte-specific KLF5 deletion promotes adipose tissue expansion and hepatic lipid accumulation –

A: posterior subcutaneous WAT weight of control floxed (n = 4) and αMHC-KLF5^−/− (n = 7) mice after 6 weeks HFD. B: mRNA levels of lipid metabolism markers Pparg1, Pparg2, Lpl, Cd36, Dgat2 and Glut4 in WAT isolated from control floxed or αMHC-KLF5^−/− mice after 6 weeks of HFD (n=3; **p<0.01, ***p<0.001 with Student’s T-Test). C-D: Representative pictures of H&E staining of WAT and BAT (C), and Oil Red O staining of liver (D) of control floxed and αMHC-KLF5^−/− mice after 6 weeks HFD. Data are presented as mean ± SEM. BAT: brown adipose tissue, Dgat: diglyceride acyltransferase, Glut: glucose transporter, HFD: high fat diet, KLF: Krüppel-like factor, Lpl: lipoprotein lipase, WAT: white adipose tissue.

We next evaluated the food intake and activity levels of the mice as a possible explanation of the difference in weight gain. Metabolic cage analysis did not show significant differences in respiratory exchange ratio, caloric intake, activity, and energy expenditure between HFD-fed αMHC-KLF5^−/− and control floxed mice fed with HFD (Figure 3A–D). Thus, the differences in these parameters were not great enough to detect with this method.

Figure 3: Cardiomyocyte-specific KLF5 deletion does not alter significantly respiratory exchange ratio, caloric intake, activity, and energy expenditure at the early stage of HFD –

Respiratory exchange ratio (A), feeding timecourse (B), activity time course (C), and energy expenditure (D) from metabolic cage analysis of control floxed (n = 6) and αMHC-KLF5^−/− (n = 5) mice on HFD for 3 weeks. Data are presented as mean ± SEM. HFD: high fat diet, KLF: Krüppel-like factor.

3.2. KLF5 is a miR-208-independent, direct positive regulator of cardiac MED13.

As the effect of cardiomyocyte-specific deletion of KLF5 in the expansion of WAT resembles the phenotype of HFD-fed αMHC-MED13^−/− mice (11), we assessed whether KLF5 regulates Med13 expression. The αMHC-KLF5^−/− mice have reduced cardiac Med13 mRNA levels (Figure 4A), which may account for DIO. However, miR-208, which targets Med13 transcript (11), was not increased, on the contrary it showed a trend (p = 0.29 for male mice and p = 0.13 for female mice) to decrease (Figure 4B). Then, we treated HL-1 cells with adenovirus expressing KLF5 (ad-KLF5) or GFP (ad-GFP) as control. Overexpression of KLF5 significantly increased Med13 expression (Figure 4C). Furthermore, in silico promoter analysis (Genomatix software) followed by alignment (CLUSTAL O - 1.2.0) of mouse and human Med13 promoters (obtained from the UCSC Genome Browser) identified two potential KLF5 binding sites in the - 730/−713 bp region and in the −142/−125 bp region (Figure 4D). Chromatin immunoprecipitation in HL-1 cells treated with ad-KLF5 showed enrichment of the −730/−713 bp region (Figure 4E) but not the −142/−125 bp region (Figure 4F) of Med13 promoter with KLF5. Thus, KLF5 is a positive regulator of Med13 gene expression.

Figure 4: KLF5 is a miR-208-independent, direct positive regulator of cardiac MED13 –

A-B: Cardiac Med13 mRNA levels (n=4-6; ***p<0.001 with Student’s T-Test) (A) and miR-208a levels (n=3-4) (B) in male and female control floxed and αMHC-KLF5^−/− mice. C: Klf5 and Med13 mRNA levels in HL-1 cells treated with Ad-GFP or Ad-KLF5 (n=4, *p<0.05; **p<0.01 vs ad-GFP with Student’s T-Test). D: Predicted KLF-binding sites by in silico promoter analysis on aligned mouse and human Med13 promoters (highlighted in yellow). E-F: Enrichment of −730/−713 bp region (E) or −142/−125 bp region (F) (highlighted in yellow) of mouse Med13 promoter with KLF5 of chromatin samples from HL-1 cells treated with Ad-GFP or Ad-KLF5 (n=3; ***p<0.001 vs Ad-GFP with Student’s T-Test). Data are presented as mean ± SEM. Ad: adenovirus, KLF: Krüppel-like factor, Med13: mediator complex subunit 13.

3.3. αMHC-KLF5^−/− mice on HFD have increased FGF21 signaling

We then investigated what mechanism may underlie the cross-talk between cardiomyocytes and WAT expansion. Plasma leptin and adiponectin levels are not changed in HFD-fed αMHC-KLF5^−/− mice compared to control HFD-fed floxed mice (Figure 5A and 5B). We then assessed the cardiac endocrine factors ANP and BNP that have lipolytic properties in adipose tissue (4, 56), and are inversely associated with obesity (57–59). We have previously shown that cardiac Bnp expression is increased in chow αMHC-KLF5^−/− mice compared to floxed mice (47), and here we show that HFD-fed αMHC-KLF5^−/− mice have increased cardiac Bnp expression compared to HFD-fed floxed mice (Figure 5C). Plasma BNP levels were not increased in HFD-fed αMHC-KLF5^−/− mice (Figure 5D). Accordingly, cardiac Anp expression levels were increased in HFD-fed αMHC-KLF^−/− mice (Figure 5C).

Figure 5: αMHC-KLF5^−/− mice on HFD have neither higher leptin and adiponectin nor lower natriuretic peptide levels –

A-B: Leptin (A) and adiponectin (B) levels in plasma obtained from control floxed and αMHC-KLF5^−/− mice after 6 weeks of HFD (n=3-5). C: Cardiac Anp and Bnp mRNA expression from control floxed and αMHC-KLF5^−/− mice after 6 weeks HFD diet (n=3-5, ***p<0.001 with Student’s T-Test). D: BNP levels in plasma from control floxed and αMHC-KLF5^−/− mice after 6 weeks in HFD (n=3-9). Data are presented as mean ± SEM. ANP; atrial natriuretic peptide, BNP: brain natriuretic peptide, HFD: high fat diet, KLF: Krüppel-like factor. (See also Figure S3)

Based on observations that have associated FGF21 with fat content either in a positive (37–39) or negative way (40, 41), we measured cardiac gene expression and plasma FGF21 levels, which were increased in HFD-fed αMHC-KLF5^−/− mice (Figure 6A and 6B). Accordingly, mRNA levels of Fgfr and Klotb that transmit FGF21 signaling were increased in WAT obtained from HFD-fed αMHC-KLF5^−/− mice (Figure 6C). Increased FGF21 signaling has been associated with inhibition of PPARγ SUMOylation (44). To assess sumoylation of PPARγ, we immunoprecipitated PPARγ from WAT protein and then performed western blots with pan-SUMO antibody. WAT of HFD-fed αMHC-KLF5^−/− mice showed decreased sumo-PPARγ to total PPARγ ratio (Figure 6D). The expression of SUMO-specific proteases (SENPs), which may regulate sumoylation and desumoylation was not altered in WAT of αMHC-KLF5^−/− mice on HFD (Figure S2).

Figure 6: αMHC-KLF5^−/− mice on HFD have increased FGF21 signaling –

A-D: Cardiac Fgf21 mRNA expression (n=4-5, ***p<0.001 with 2way ANOVA) (A), and plasma FGF21 levels (n=8, *p<0.05 with Student’s T-Test; 95% CI 69.7-138.9 pg/ml in control floxed and 118.7-233.5 pg/ml in αMHC-KLF5^−/− mice) (B) in control floxed and αMHC-KLF5^−/− mice after 6 weeks in HFD. C: Fgf21r and Klotb mRNA levels in WAT obtained from control floxed and αMHC-KLF5^−/− mice after 6 weeks in HFD (n=3, **p<0.01, ***p<0.001 vs floxed with Student’s T-Test). D: Representative western blot image and quantitative analysis of sumo-PPARγ and PPARγ protein levels obtained with immunoprecipitation from WAT isolated from control floxed and αMHC-KLF5^−/− mice after 6 weeks in HFD (n=3-4, *p<0.05 with Student’s T-Test). The lanes were run on the same gel but were noncontiguous. Data are presented as mean ± SEM. HFD: high fat diet, KLF: Krüppel-like factor.

3.4. FGF21 mediates increased weight gain of ^−/−MHC-KLF5^−/− mice on HFD.

In order to assess whether cardiomyocyte-derived FGF21 has a distinct effect in mediating the obesogenic effect of Klf5 ablation, we generated mice with cardiomyocyte-specific double knockout of Klf5 and Fgf21 (αMHC-[KLF5^−/−;FGF21^−/−]) and fed them with HFD. The αMHC-[KLF5^−/−;FGF21^−/−] mice have reduced cardiac Fgf21 mRNA levels by 95% (Figure 7A). When these mice were fed with HFD, they did not show increased body weight gain (Figure 7B and 7C), or WAT weight (Figure 7D) compared to control floxed mice on HFD. Plasma FGF21 levels in HFD-fed αMHC-[KLF5^−/−;FGF21^−/−] mice were decreased compared to control floxed mice on HFD (Figure 7E). The double cardiomyocyte-specific knockout of Klf5 and Fgf21 also prevented the increased expression of lipid metabolism-related gene expression markers in WAT (Figure 7F, 1A), the increase in adipocyte size in WAT, and the increased lipid accumulation in BAT and liver that was found in αMHC-KLF5^−/− mice on HFD (Figure 8A, 8B, 2C, 2D, S1E). Plasma TG levels were normal and glucose levels are slightly decreased (15%) in αMHC-[KLF5^−/−;FGF21^−/−] mice on HFD compared to control floxed mice on HFD (Figure S1F and S1G). HFD-fed αMHC-[KLF5^−/−;FGF21^−/−] mice had no change in PPARγ sumoylation in WAT (Figure 8C). Cardiac Med13 expression was still lower in HFD-fed αMHC-[KLF5^−/−;FGF21^−/− than control floxed mice (Figure 8D). Thus, cardiomyocyte-specific deletion of Fgf21 prevents accelerated DIO driven by cardiac Klf5 ablation despite downregulation of MED13, although downregulation of Med13 expression in aMHC-[KLF5−/−;FGF21−/−] male mice seems to be less robust (−25%) than the respective change in αMHC-KLF5−/− mice of the same gender (−35%).

Figure 7: Cardiomyocyte FGF21 is associated with increased weight gain of HFD-fed αMHC-KLF5^−/− mice –

A: Cardiac Fgf21 mRNA levels in control floxed and αMHC-[KLF5^−/−;FGF21^−/−] mice (n= 8, *p<0.05 vs floxed with Student’s T-Test). B-D: Cumulative body weight gain (B), body weight gain (C), and posterior subcutaneous WAT weight (D) of control floxed (n = 21) and αMHC-[KLF5^−/−;FGF21^−/−] (B and C: n = 9, D: n = 5) mice after 6 weeks on HFD. E: Plasma FGF21 levels in control floxed (n = 34; 95% CI 102.4-225.2 pg/ml), and αMHC-[KLF5^−/−;FGF21^−/−] (n = 8; 95% CI 1.0-115.4 pg/ml) mice after 6 weeks HFD (*p<0.05 vs floxed with Student’s T-Test). F: mRNA levels of lipid metabolism markers Pparg1, Pparg2, Lpl, Cd36, Dgat2 and Glut4 in WAT obtained from control floxed (n = 25) and aMHC-[KLF5^−/−;FGF21^−/−] (n = 6) mice. Data are presented as mean ± SEM. cmDKO is αMHC-[KLF5^−/−;FGF21^−/−]. Dgat: diglyceride acyltransferase, Glut: glucose transporter, HFD: high fat diet, KLF: Krüppel-like factor, Lpl: lipoprotein lipase, WAT: white adipose tissue. (See also Figure S1E–G, and Figure S3)

Figure 8: Cardiomyocyte FGF21 ablation negates the proadipogenic effect of KLF5 deletion –

A: Representative pictures of H&E staining of WAT and BAT, and Oil Red O staining of liver of control floxed and aMHC-[KLF5^−/−;FGF21^−/−] mice after 6 weeks HFD. B: Quantification of WAT adipocyte size (n = 4) of control floxed and aMHC-[KLF5^−/−;FGF21^−/−] mice after 6 weeks HFD. C: Western Blotting analysis for sumo-PPARγ and PPARγ protein levels obtained with immunoprecipitation from WAT isolated from control floxed and aMHC-[KLF5^−/−;FGF21^−/−] mice after 6 weeks in HFD (n=3-6). D: Cardiac Med13 mRNA levels in HFD-fed control floxed and aMHC-[KLF5^−/−;FGF21^−/−] mice (n=4, *p<0.05 vs floxed with Student’s T-Test). Data are presented as mean ± SEM. cmDKO is aMHC-[KLF5^−/−;FGF21^−/−]. BAT: brown adipose tissue, HFD: high fat diet, KLF: Krüppel-like factor, WAT: white adipose tissue.

To further assess the potential role of cardiomyocyte-derived FGF21 in DIO, independent from KLF5, we generated mice with cardiomyocyte-specific Fgf21 deletion (αMHC-FGF21^−/−). These mice have decreased FGF21 expression (70%) in isolated cardiomyocytes (Figure S3A). Treatment of αMHC-FGF21^−/− mice with HFD did not change the cumulative body weight gain compared to control floxed mice on HFD, but the relative (%) body weight gain was decreased (Figure S3B and S3C). Thus, the body weight gain of the αMHC-FGF21^−/− mice on HFD was lower compared to HFD-fed control floxed mice. Plasma FGF21 levels in HFD-fed αMHC-FGF21^−/− mice were not changed (Figure S3D). Fgf21 mRNA expression in liver, skeletal muscle, WAT, BAT, and kidney were similar between control floxed, αMHC-KLF5^−/−, αMHCFGF21^−/−, and αMHC-[KLF5^−/−;FGF21^−/−] mice on HFD (Figure S3E), suggesting that FGF21 derived from other tissues does not contribute to the changes in plasma FGF21 levels between the different strains.

4. DISCUSSION

Various organs, such as the heart, liver (60), skeletal muscle (24, 61), bones (62) and gut microbiota (63) affect WAT expansion. Specifically, skeletal muscle (64, 65), gut (66), bone (62), liver (60, 67), and the heart (4–7, 10, 11, 68, 69) secrete factors that affect systemic lipid metabolism and adipose tissue. A previous study has shown that reduced cardiac Med13 accelerates DIO and overexpression of this gene reduces DIO (10, 11). The reasons for this are unknown but suggest that Med13 regulates production of a secreted factor that affect adipose biology. Parabiosis experiments are consistent with this conclusion (10). We found that KLF5 is a positive regulator of Med13 expression.

Here, we show that cardiomyocyte Klf5 ablation increases body weight gain rate, as well as that this effect of cardiomyocyte KLF5 is abrogated upon deletion of cardiomyocyte-derived FGF21. Increased plasma FGF21 levels that are observed in αMHC-KLF5^−/− mice result in PPARγ activation in WAT causing increased adiposity. This is the first time an endocrine action of cardiomyocyte-derived FGF21 is demonstrated. The cardiac KLF5-FGF21 signaling pathway found in this study might play a role in diabetes as we have previously shown in diabetic animal models of both insulin-deficient and Type 2 diabetes that Klf5 expression is downregulated at the early stage of diabetes and increased as the disease progresses (47). In this previous study we also showed that KLF5 is a positive regulator of Ppara gene transcription. Others have shown that cardiac specific PPARa overexpression leads to hepatic insulin resistance and future studies will need to show whether cardiac KLF5-mediated changes in PPARa signaling and altered insulin sensitivity in other tissues contribute towards the DIO of αMHC-KLF5−/− mice (70). Interestingly, in both αMHC-Klf5^−/− and the αMHC-[KLF5^−/−;FGF21^−/−] mice, we observed a small trend for reduced plasma glucose levels.

HFD-fed αMHC-[KLF5^−/−;FGF21^−/−] mice also have lower cardiac Med13 mRNA levels compared to HFD-fed control floxed mice (Figure 8D) but normal body weight gain rate. Whether αMHC-KLF5^−/− mice on HFD also have a cardiac phenotype and develop cardiomyopathy will be subject of future study. The interrelationship between cardiomyocyte Med13 and Fgf21 expression, as well as whether cardiomyocyte-derived FGF21 may be involved in the induction of the obesogenic phenotype of αMHC-MED13^−/− mice remains to be investigated (10). But our data suggest that not all sources of FGF21 have identical actions, and that cardiac FGF21 works different from hepatic FGF21.

In our study the increased cardiac FGF21 expression levels in HFD-fed αMHC-KLF5^−/− mice are associated with increased plasma levels of FGF21. Although plasma FGF21 is not decreased in αMHC-FGF21^−/− mice, which may be due to compensatory FGF21 production from other tissues or due to the limits of detection sensitivity, plasma FGF21 levels are decreased in αMHC-[KLF5^−/−;FGF21^−/−] mice. Cardiomyocyte-specific overexpression of FGF21 has been shown to increase circulating levels of FGF21 and a moderate decreased body weight due to a reduction in lean body mass while fat mass was increased (41). Although increased circulating levels of FGF21 during obesity, fasting, and refeeding have been shown to originate from the liver (71), mitochondrial dysfunction and stress increase cardiac and circulating levels of FGF21 without an increase in hepatic FGF21 levels (72).

The role of FGF21 in regulating body weight has been controversial and it seems to rely on the source of FGF21, as well as on acute or chronic administration. Previous reports have associated FGF21 with anti-diabetic and weight-lowering effects (37, 38). Various studies have demonstrated beneficial metabolic effects of systemic administration of exogenous FGF21, and reduction of body weight and plasma lipid and glucose levels in obesity (37, 38, 73–76). Adipose tissue-specific constitutive expression of the FGF21 co-receptor, β-Klotho, in mice that underwent exogenous administration of human FGF21 also lowered body weight gain rate compared to HFD-fed control mice (77). In addition, treatment of obese non-human primates and obese and diabetic patients with FGF21 mimetics show beneficial effects on lipid profile and body weight (39, 78). Chronic treatment of HFD-fed mice with FGF21 decreased body weight, fat mass, and liver steatosis (79). Fgf21^−/− mice have increased body weight and fat mass on normal chow or ketogenic diet, but similar body weight as wild type mice on HFD (26, 35, 80). A recent study using tissue specific β-klotho KO mice shows that the body-lowering effect of FGF21 administration on weight loss is mediated primarily by the central nervous system (81). Other effects of FGF21 to reduce glucose and lipid levels are found with exogenous administration or transgenic overexpression and 5-fold or higher increases of FGF21 levels (41, 75), which are much greater than those found in our mice. In contrast to these findings, FGF21 signaling in adipose tissue improves insulin-sensitivity (71). Insulin drives pre-adipose differentiation and greater adipose mass (82). FGF21 decreases lipolysis in human adipocytes in vitro (Figure 1A in (42)) and in mouse primary adipocytes (Figure 2B in (43)). Cardiomyocyte-specific overexpression of FGF21 has been shown to increase circulating levels of FGF21 and a moderate decreased body weight due to a reduction in lean body mass while fat mass was increased (Figure 4E in (41)). Krievina et al. showed that accumulation of adipose tissue in the renal sinus is associated with the visceral adipose amount and increased circulating FGF21 (Table 3 of (40)).

How might FGF21 regulate adipose mass? A previous study showed that FGF21 stimulates both PPARγ expression (83) and PPARγ transcriptional activity by inhibiting sumoylation of the latter (44). However a subsequent study could not reproduce the finding that FGF21 inhibits PPARγ sumoylation (84). It has been proposed that the discrepancy between those two studies may be due to different Fgf21^−/− strains (85). Therefore, it remains elusive how PPARγ mediates the effects of FGF21. Our findings indicate increased PPARγ expression and reduced SUMO-PPARγ in WAT in HFD-fed αMHC-KLF5^−/− mice, which are in accordance with increased adiposity.

Our data demonstrate a role for cardiomyocyte KLF5 in regulating body weight. The mechanism via which KLF5 regulates cardiac Fgf21 expression and whether increased FGF21 levels mediate the effects of MED13 in adiposity remain to be elucidated. In this study, we demonstrate that KLF5 inhibition promotes DIO via a distinct effect of cardiomyocyte FGF21. Moreover, we have uncovered a specific pathway leading from the heart to regulation of body and show that KLF5 is involved in a complex metabolic network that controls WAT development and body weight gain. Overall, our data show a novel hormonally regulated inter-organ cross talk between the heart and systemic metabolism.

Supplementary Material

2

Figure S1: Plasma and hepatic triglyceride and plasma glucose levels of control floxed, αMHC-KLF5^−/− mice, and αMHC-[KLF5^−/−;FGF21^−/−] mice on HFD (related to Figures 1 and 3) – A: Quantification of WAT adipocyte size (n = 4-5) of control floxed and αMHC-KLF5^−/− mice on HFD. B: Hepatic TG levels of control floxed and αMHC-KLF5^−/− mice on HFD (n=5-7, **p<0.01 vs floxed with Student’s t-Test). C-D: Plasma TG and glucose levels of control floxed and αMHCKLF5^−/− mice on HFD (n=5-7). E: Hepatic TG levels of control floxed and αMHC-[KLF5^−/−;FGF21^−/−] mice on HFD (n=6-8). F-G: Plasma TG and glucose levels of control floxed (n = 36) and αMHC-[KLF5^−/−;FGF21^−/−] (n = 9) mice on HFD after 6 weeks HFD. Data are presented as mean ± SEM. cmDKO is αMHC-[KLF5^−/−;FGF21^−/−], HFD: high fat diet, KLF: Krüppel-like factor, TG: triglyceride, WAT: white adipose tissue.

3

Figure S2: Senp mRNA expression are not altered in HFD-fed αMHC-KLF5^−/− mice (related to Figure 3H) – SENPs mRNA expression in WAT of control floxed and αMHC-KLF5^−/− mice after 6 weeks HFD (n=3-5). Data are presented as mean ± SEM. HFD: high fat diet, KLF: Krüppel-like factor, Senp: SUMO specific proteases, WAT: white adipose tissue.

4

Figure S3: Ablation of cardiomyocyte-derived FGF21 attenuates DIO (related to Figure 4) A: Cardiomyocyte Fgf21 mRNA levels in control floxed and αMHC-FGF21^−/− mice (n=3-6, *p<0.05 vs floxed with Student’s T-Test). B-C: Cumulative body weight gain (B), and body weight gain (C) in control floxed (n=17) and αMHC-FGF21^−/− (n = 13) mice after 6 weeks HFD (*p<0.05 vs floxed with Student’s T-Test). D: Plasma FGF21 levels in control floxed (n = 34; 95% CI 102.4-225.2 pg/ml) and αMHC-FGF21^−/− (n = 29; 95% CI 60.8-241.2 pg/ml) mice after 6 weeks HFD. E: Liver (Li), skeletal muscle (SM), WAT, BAT, and kidney (Ki) Fgf21 mRNA levels in control floxed (n = 31 (Li), 5 (SM), 5 (WAT), 7 (BAT), 5 (Ki)), αMHC-KLF5^−/− (n = 3), αMHCFGF21^−/−(n = 16 (Li), 6 (SM), 6 (WAT), 6 (BAT), 5 (Ki)), and⁻ αMHC-[KLF5^−/−;FGF21^−/−] (n = 8 (Li), 6 (SM), 5 (WAT), 4 (BAT), 4 (Ki)) mice (*p<0.05 vs floxed with 1way ANOVA). Data are presented as mean ± SEM. cmDKO is αMHC-[KLF5^−/−;FGF21^−/−]. DIO: diet-induced obesity, HFD: high fat diet, KLF: Krüppel-like factor.

Highlights.

• Cardiomyocyte-specific KLF5 deletion accelerates diet-induced obesity.

• KLF5 is a positive regulator of Med13 expression.

• The proadipogenic effect of KLF5 relies also on non-MED13 mechanisms

• Cardiomyocyte KLF5 regulates systemic metabolism and adiposity via FGF21.

Abbreviations

18S

18S ribosomal RNA

ANP

atrial natriuretic peptide

Actb

beta-actin

BNP

brain natriuretic peptide

BAT

brown adipose tissue

DIO

diet-induced obesity

FAO

fatty acid oxidation

HFD

high fat diet

KLF

krüppel-like factor

MED13

mediator complex subunit 13

RXR

retinoid X receptor

Rplp0

ribosomal protein lateral stalk subunit P0

Snord65

small nucleolar RNA, C/D box 65

SUMO

sumoylation

WAT

white adipose tissue