Cardiac Myocyte KLF5 Regulates Ppara Expression and Cardiac Function

Abstract

Rationale

Fatty acid oxidation is transcriptionally regulated by peroxisome proliferator-activated receptor (PPAR)α and under normal conditions accounts for 70% of cardiac ATP content. Reduced Ppara expression during sepsis and heart failure leads to reduced fatty acid oxidation and myocardial energy deficiency. Many of the transcriptional regulators of Ppara are unknown.

Objective

To determine the role of Krüppel-like factor 5 (KLF5) in transcriptional regulation of Ppara.

Methods and Results

We discovered that KLF5 activates Ppara gene expression via direct promoter binding. This is blocked in hearts of septic mice by c-Jun, which binds an overlapping site on the Ppara promoter and reduces transcription. We generated cardiac myocyte-specific Klf5 knockout mice that showed reduced expression of cardiac Ppara and its downstream fatty acid metabolism-related targets. These changes were associated with reduced cardiac fatty acid oxidation, ATP levels, increased triglyceride accumulation and cardiac dysfunction. Diabetic mice showed parallel changes in cardiac Klf5 and Ppara expression levels.

Conclusions

Cardiac myocyte KLF5 is a transcriptional regulator of Ppara and cardiac energetics.

INTRODUCTION

Fatty acid oxidation (FAO) accounts for the production of approximately 70% of the ATP that the heart utilizes¹. Some forms of heart failure are due to perturbations in heart energetics and severe heart failure is associated with energy starvation and reprogramming of cardiac energetics². These metabolic changes occur regardless of whether the primary cause of cardiac dysfunction is metabolic disease, pressure overload or ischemia^{3, 4}. A dramatic example of cardiac dysfunction due to reduction in FAO⁵ and energy depletion occurs in sepsis^5–7. The transcriptional mechanisms that underlie inhibition of cardiac FAO and cardiac dysfunction during sepsis and other types of cardiac dysfunction are incompletely understood.

Cardiac FAO is regulated at several stages: FA uptake, triglyceride (TG) formation and storage in lipid droplets, TG lipolysis leading to release of unesterified fatty acids, and transfer of fatty acids into the mitochondria for FAO and ATP production. Most of the proteins that participate in this cascade are transcriptionally regulated by peroxisome proliferator-activated receptor α (PPARα)⁸. Although it is known that PPARα is activated by FAs that are released via lipolysis from the intracellular triglyceride pool^{9, 10}, the transcriptional regulation of Ppara is not fully elucidated. Various gain or loss of PPARα function animal models resulted in mixed outcome with either protective or aggravating roles of PPARα in cardiac function. Α variety of metabolic and pathological stress conditions influence cardiac PPARα expression in multiple ways, which are not fully defined.

Metabolism in several tissues is regulated by members of the Krüppel-like factor (KLF) protein family, which regulate proliferation, differentiation, development, and cell death¹¹. Thus far, 17 KLF isoforms have been identified in humans and mice, while several homologs were described in other species¹¹. Adipocyte KLF2¹², KLF3¹³ and KLF7¹⁴ inhibit adipose tissue development. On the other hand, KLF4¹⁵, KLF6¹⁶ and KLF15¹⁷ have the opposite effect in adipocytes, as they induce Pparg and adipogenesis. Hepatic KLF11 induces Ppara and FAO genes and prevents hepatic TG accumulation¹⁸. KLF15 promotes lipid utilization in the heart¹⁹ and skeletal muscle²⁰. Thus, several KLF isoforms have been implicated in the regulation of metabolic pathways in several organs including the heart.

KLF5 is involved in pressure overload-mediated cardiac hypertrophy, but its role in cardiac metabolism remains unknown. Heterozygote Klf5^+/− mice are protected from pressure overload cardiac hypertrophy²¹, due to reduced transforming growth factor (TGF)β production in cardiac fibroblasts and not because of changes in cardiac myocytes²². Heterozygote Klf5^+/− mice showed increased skeletal muscle FA consumption due to activation of PPARδ²³, suggesting that KLF5 is an inhibitor of lipid catabolism. Conversely, Klf5 deletion inhibited lipid production in lung surfactant²⁴, indicating that KLF5 is a positive regulator of lipid homeostasis in lungs. Thus, the actions of KLF5 in lipid metabolism vary depending on its site of expression.

We focused on the role of KLF5 in the regulation of cardiac metabolic gene expression. Unexpectedly, we first discovered that Klf5 gene expression was induced in energy-depleted hearts of mice treated with E. coli lipopolysaccharides (LPS) that had lower Ppara expression. Although this observation implicated cardiac KLF5 in Ppara and FAO inhibition, our subsequent studies showed the opposite. We created a cardiac myocyte-specific Klf5^−/− mouse and conducted gain-of-function experiments in cardiac myocytes that revealed KLF5 to be a transcriptional activator of Ppara. Klf5 ablation in cardiac myocytes reduced cardiac FAO and ATP content, increased TG accumulation and caused cardiac dysfunction. Furthermore, cardiac KLF5 was reduced in the early stages of Type 1 and in Type 2 diabetes mouse models along with Ppara gene expression. Thus, KLF5 is a novel regulator of Ppara and cardiac lipid utilization.

METHODS

Expanded Methods are presented in the Online Data Supplement. All animal studies were approved by the institutional animal care and use committees. Data are expressed as the mean ± SEM. Statistical significance was assessed with t-test or 1-way ANOVA followed by Bonferroni post hoc tests, performing all pairwise comparisons. A p-value of less than 0.05 was considered statistically significant.

Wild type C57BL/6 mice were treated with E. coli LPS to mimic sepsis. Microarrays for cardiac mRNA of LPS-treated mice was performed by Ocean Ridge Biosciences. The data are deposited in the Gene Expression Omnibus database (GSE63920). HL-1 cells²⁵ were infected with adenoviruses expressing constitutively active c-Jun or KLF5 and were harvested 48h post-infection for gene expression and chromatin immunoprecipitation (ChIP).

We generated mice with cardiac myocyte-specific Klf5 gene deletion (aMHC-Klf5^−/−). Cardiac function was assessed by 2D echocardiography. The microarray analysis for cardiac mRNA obtained from αMHC-Klf5^−/− mice was performed by Arraystar (data deposited in GSE63839).

C57BL/6 mice were injected IP with streptozotocin (STZ) to mimic Type 1 diabetes (insulin dependent). Inhibition of sodium/glucose cotransporter (SGLT)2 in diabetic mice was performed either via treatment with dapagliflozin (drinking water) or with SGLT2 anti-sense oligonucleotides (ASO, ISIS Pharmaceuticals).

RESULTS

In silico Ppara promoter analysis identified two overlapping potential AP-1 and KLF binding sites

We showed previously that LPS-mediated activation of the c-Jun N-terminal kinase (JNK) signaling pathway reduces cardiac Ppara gene expression⁵. Aiming to identify potential binding sites for the substrate of JNK, c-Jun (AP-1 sites), on the mouse Ppara gene promoter we performed in silico promoter analysis (Genomatix software). This analysis identified two potential AP-1 sites in the anti-sense strand of the region −792/−772 bp (region A) and in the sense strand of the region −719/−698 bp (region B) (Figure 1A). Interestingly, both predicted AP-1 sites overlapped with potential KLF binding sites (Figure 1A).

Figure 1. Cardiac KLF5 is upregulated in sepsis.

(A) Predicted AP-1 (yellow) and KLF (framed) binding sites on mouse Ppara promoter. (B–C) Cardiac mRNA levels of Klf isoforms (B) and protein levels of KLF5 and β-actin (C) in 10–12-weeks old C57BL/6 mice treated with 5 mg/kg LPS or saline (CTRL) (n=4–5; *P<0.05; **P<0.01; ***P<0.001 vs CTRL). (D–E) Ppara, Klf5 and Klf6 mRNA levels in HL-1 cells (D) treated with 1µg/ml LPS or saline (CTRL) for 9h (n=6; *p<0.05 vs. CTRL) or in aMHC-Pparg mice (E) treated with 5mg/kg LPS or saline (CTRL) for 8–10h (n=5; *p<0.05; **p<0.01 vs. CTRL).

KLF5 expression is induced by LPS treatment

We next evaluated the expression profile of the 17 existing KLF isoforms in hearts from LPS-treated mice. We performed whole genome microarray analysis followed by qRT-PCR gene expression analysis. Among the 10 KLF isoforms that were detected in microarrays, Klf5 was the most profoundly upregulated (8-fold; Figure 1B). KLF5 was also increased at the protein level (Figure 1C & Online Figure 1). Although KLF6 mRNA levels increased 3-fold (Figure 1B), KLF6 protein was not increased significantly in LPS-treated hearts (Online Figure I).

In order to assess whether LPS-mediated downregulation of Ppara gene expression involved KLF5 or KLF6, we treated a mouse cardiac myocyte cell line (HL-1)²⁵ with LPS. Ppara expression was downregulated by 55% in LPS-treated cells (Figure 1D). This was associated with a Klf5 mRNA increase by 75%, while Klf6 gene expression was not altered (Figure 1D). These changes in Klf5 expression could be induced either directly by LPS-triggered signaling or indirectly due to cardiac dysfunction, which alters KLFs²⁶. In order to test this, we utilized mice with constitutive PPARγ expression in cardiac myocytes (αMHC-Pparg) that are resistant to LPS-mediated cardiac dysfunction⁶. Treatment of αMHC-Pparg mice with LPS reduced Ppara gene expression by 75% and increased Klf5 by 2.3-fold but not Klf6 (Figure 1E). Thus, LPS induces cardiac Klf5 expression directly.

KLF5 and c-Jun have opposite functions on PPARα gene expression and compete for binding to the Ppara promoter

In order to identify whether c-Jun and KLF5 modulate Ppara gene expression in a synergistic or competitive fashion we generated an adenovirus that expresses a constitutively active form of c-Jun (Ad-cjun^Asp) (Online Figure II). In this isoform Ser58, Thr62, Ser63, Ser73, Thr91 and Thr93 of the transactivation domain have been substituted with the phospho-mimetic aspartic acid. Therefore c-Jun is constitutively active without phosphorylation by JNK²⁷. Infection of HL-1 cells with Ad-cjun^Asp reduced Ppara mRNA levels by 40% (Figure 2A) and protein levels by 60% (Figure 2B and Online Figure III). On the other hand, treatment of HL-1 cells with adenovirus expressing KLF5 (Ad-KLF5) increased Ppara gene expression levels by 4-fold (Figure 2C) and protein levels by 2.5-fold (Figure 2D and Online Figure III). Thus, c-Jun and KLF5 have opposite regulatory roles on Ppara expression.

Figure 2. KLF5 and c-Jun have opposite effects on Ppara expression and compete for binding on Ppara promoter.

(A–D) Ppara and Klf5 mRNA (A, C) and protein (B, D) levels in HL-1 cells treated with Ad-cJun^Asp (A, B) or Ad-KLF5 (C, D); (n=6; *p<0.05; **p<0.01; ***p<0.001 vs CTRL). (E–I) Enrichment of −792/−772 bp region (E, F) or −719/−698 bp region (G, H) of mouse Ppara promoter with c-Jun (E, G) or KLF5 (F, H) of chromatin samples from HL-1 cells treated with Ad-GFP (CTRL) and either Ad-cJun^Asp (E, G) or Ad-KLF5 (F, H); **p<0.01 vs CTRL. (I) Enrichment of −792/−772 bp region of mouse Ppara promoter with c-Jun or KLF5 of chromatin samples from HL-1 cells treated with 1 µg/ml LPS or saline (CTRL); *p<0.05 vs CTRL. Data for all bar graphs are represented as means ± SEM (statistical analysis: t-test).

We performed chromatin immunoprecipitation (ChIP) to investigate whether the inhibitory effect of c-Jun and the positive effect of KLF5 on Ppara gene expression are due to their direct binding on the Ppara promoter. Treatment of HL-1 cells with Ad-cjun^Asp increased enrichment of the region A of the Ppara promoter with c-Jun (Figure 2E). The same region was also occupied by KLF5 when HL-1 cells were treated with Ad-KLF5 (Figure 2F). Conversely, the region B of the Ppara promoter was occupied by c-Jun (Figure 2G) but not KLF5 (Figure 2H). As c-Jun and KLF5 have opposite effects on Ppara expression and are both activated in the hearts of LPS-treated mice, we treated HL-1 cells with LPS and assessed binding of endogenous c-Jun and KLF5 on the region A. LPS treatment promoted c-Jun binding, which in turn prevented KLF5 binding on the region A (Figure 2I).

Cardiac myocyte-specific Klf5 ablation alters cardiac function

Our in vitro findings implicated KLF5 in activation of Ppara expression. In order to assess this in vivo we generated a mouse line with cardiac myocyte-specific deletion of Klf5 (αMHC-Klf5^−/−) by crossing αMHC-Cre mice with mice that have Klf5 exons 2 and 3 flanked with loxP sites (floxed mice)²⁴. Cardiac Klf5 mRNA levels were reduced by approximately 40% in αMHC-Klf5^−/− mice (Figure 3A). The partial downregulation of Klf5 gene expression was likely due to the presence of other cell types in the heart, such as fibroblasts and endothelial cells. Therefore, we isolated primary mouse cardiac myocytes and assessed Klf5 mRNA levels, which were 85% lower compared to control cardiac myocytes isolated from floxed mice (Figure 3B). Klf5 expression was not reduced in skeletal muscle, intestine, kidney, white adipose tissue, or brain of αMHC-Klf5^−/− mice (Figure 3A).

Figure 3. Cardiac myocyte-specific ablation of KLF5 induces a distinct transcriptome profile.

(A, B) Klf5 mRNA in the heart, skeletal muscle, intestine, kidney, white adipose tissue, brain (A) and primary cardiac myocytes (B) of aMHC-Klf5^−/− mice (n=3; *p<0.05 vs floxed). (C) Hierarchical clustering for differentially expressed mRNAs detected by whole genome microarray analysis of cardiac mRNA obtained from aMHC-Klf5^−/− mice and control floxed mice. Red color indicates high relative expression and blue color indicates low relative expression. (D–G) Gene ontology analysis for classification of the downregulated (D) or upregulated (E) genes based on the metabolic process that they are associated with and pathway analysis for downregulated (F) and upregulated (G) genes detected with whole genome microarray analysis of cardiac mRNA obtained from aMHC-Klf5^−/− mice and control floxed mice. Data for all bar graphs are represented as means ± SEM (statistical analysis: t-test).

We then treated αMHC-Klf5^−/− mice with LPS. Cardiac myocyte-specific ablation of Klf5 only partially improved cardiac function, which was still worse compared to control αMHC-Klf5^−/− treated with saline (Online Figure IV-A). This partial improvement in cardiac function was independent from changes in cardiac Ppara gene expression (Online Figure IV-B).

Changes in cardiac gene expression in aMHC-Klf5^−/− mice

We next performed microarray analysis on hearts from αMHC-Klf5^−/− and floxed control mice. Hierarchical clustering of the microarray data revealed two distinct expression signatures (Figure 3C). We identified 228 up-regulated and 79 downregulated genes in aMHC-Klf5^−/− hearts, with at least 2-fold change (Online Figure V). Gene ontology analysis for classification of the differentially expressed genes based on the biological process suggested that 36 of the downregulated genes are involved in metabolic pathways (Figure 3D & Online Table I). Two of these genes, acyl-CoA thioesterase (Acot) 3 and Acot4, are related to FAO while 6 genes are associated with carbohydrate metabolism: serine (or cysteine) peptidase inhibitor, clade A, member 1A (Serpina1a), Serpina 1b, phosphorylase kinase α 1 (Phka1) and solute carrier family 3 member 2 (Online Table I). Gene ontology analysis for the upregulated transcripts showed 39 metabolism-related genes (Online Table II). Among these genes, 13 were associated with lipid metabolism, 2 with glucose metabolism and 2 genes were related to both glucose and lipid metabolism (Figure 3E & Online Table II). Pathway analysis indicated a strong association of downregulated genes with the pathways of complement and coagulation, FA elongation, biosynthesis of unsaturated FAs, linoleic metabolism and RNA degradation (Figure 3F). Pathway analysis showed that upregulated genes are involved in carbohydrate digestion and absorption, as well as in linoleic acid metabolism (Figure 3G).

In addition to gene ontology analysis, we performed ingenuity pathway analysis for the whole genome microarray dataset to predict transcriptional networks and molecular relationships among genes. This analysis also revealed that many of the downregulated cardiac genes of the aMHC-Klf5^−/− mice affect lipid metabolism (Figure 4A & Online Table III).

Figure 4. Cardiac myocyte-specific ablation of Klf5 inhibits the expression of genes that are associated with FA metabolism.

(A) Ingenuity pathway analysis of genes regulated over 2-fold in the aMHC-Klf5^−/− mouse array that are related to FA metabolism. (B) Cardiac Klf5 and Ppara mRNA levels of 10- to 12-week-old aMHC-Klf5^−/− male and female mice (n=5; **p<0.01; ***p<0.001 vs same gender floxed mice). (C) Cardiac PPARα and β-actin protein levels of 10- to 12-week-old floxed and aMHC-Klf5^−/− male mice. (D–F) Cardiac mRNA levels for FA oxidation- (Ppargc-1a, Ppargc-1β, Pparg, Ppard, Acox and Cpt1b) (D), lipid uptake- (Cd36, Lpl and Angptl4) (E) and lipid storage-related genes (Dgat1, Dgat2, Plin2, Plin5) (F) (n=5; *p<0.05, **p<0.01, ***p<0.001 vs same gender floxed mice). (G) Cardiac PGC-1, CPT-1, DGAT-1, ATGL, phosphorylated AMPK, total AMPK, and GAPDH protein levels of 10- to 12-week-old floxed and aMHC-Klf5^−/− male mice. (H, I) [¹⁴C]-Palmitic acid (H) and [¹⁴C]-Glucose (I) oxidation levels in cardiac muscle of 10- to 12-week-old floxed and aMHC-Klf5^−/− male mice (n=4–5; *p<0.05; **p<0.01 vs floxed mice). Data for all bar graphs are represented as means ± SEM (statistical analysis: t-test).

αMHC-Klf5^−/− mice have reduced cardiac Ppara and FAO

We confirmed with RT-PCR that deletion of cardiac myocyte Klf5 was associated with reduced cardiac Ppara expression levels in both male (30%) and female (45%) mice (Figure 4B). Cardiac PPARα protein levels were reduced by approximately 60% in αMHC-Klf5^−/− mice (Figure 4C and Online Figure III). Similarly, the expression of a broad spectrum of FAO-related genes was reduced in the hearts of aMHC-Klf5^−/− mice (Figure 4D). The expression of PPARγ coactivator (Pgc)1α was reduced by 45% in male and 25% in female mice, acyl-CoA oxidase (Acox) was reduced by 30% in male and 35% in female and carnitine palmitoyl-transferase (Cpt)1b was reduced by 45% in male and 35% in female mice. On the other hand, cardiac Pgc1b, Ppard and Pparg mRNA levels were not altered significantly (Figure 4D). The αMHC-Klf5^−/− hearts had a trend for reduced FA uptake (Online Figure VI-A, B), and the expression of lipid uptake-related genes was reduced (Figure 4E). Cardiac expression of cluster of differentiation (Cd)36 was reduced by 40% in male and 35% in female αMHC-Klf5^−/− mice, and lipoprotein lipase (Lpl) was reduced by 20% in male and female αMHC-Klf5^−/− mice. The gene expression levels of enzymes that catalyze cardiac TG formation were also reduced. Diacylglycerol acyltransferase (Dgat)1 expression was reduced by 25% in male and 30% in female αMHC-Klf5^−/− mice. Cardiac Dgat2 mRNA levels were 45% lower in both male and female αMHC-Klf5^−/− mice (Figure 4F). These changes were consistent with reduced (25%) cardiac TG content in αMHC-Klf5^−/− mice (Online Figure VII). Gene expression levels of lipid droplet-associated proteins, perilipin (Plin) 2 and Plin5, were not altered (Figure 4F). The downregulation in Pgc1 expression was also seen at the protein level, which was 40% lower in αMHC-Klf5^−/− mice, while CPT1, DGAT1 and ATGL protein showed a trend towards lower levels in αMHC-Klf5^−/− mice (Figure 4G & Online Figure III). Reduced PPARα expression levels were associated with increased (60%) phosphorylation of adenosine monophosphate kinase (AMPK) (Figure 4G & Online Figure III), reduced cardiac FAO (Figure 4H) and increased glucose oxidation levels (Figure 4I) although glucose uptake was not altered in αMHC-Klf5^−/− hearts (Online Figure VI-C, D). These findings suggest that cardiac FA metabolism is suppressed in αMHC-Klf5^−/− mice, which is consistent with cardiac Ppara downregulation.

Older αMHC-Klf5^−/− mice have lower cardiac ATP content and increased TG accumulation

The reduced FAO-related gene expression profile of the αMHC-Klf5−/− mice was associated with reduced ATP content (Figure 5A). Reduced cardiac ATP content in 9–12 months old αMHC-Klf5^−/− mice was accompanied by increased accumulation of cardiac TG (Figure 5B). Increased cardiac TG content in older αMHC-Klf5^−/− mice was associated with reversal of the expression of Dgat1, as well as of other fatty acid metabolism-related genes, such as Ppara, Cd36, Pgc1a and Aox (Online Figure VIII). Mitochondrial DNA content (Figure 5C), complex I activity (Figure 5D) and complex IV activity (Figure 5E) were not altered in the hearts of αMHC-Klf5^−/− mice.

Figure 5. Cardiac myocyte-specific ablation of Klf5 reduces ATP content and promotes cardiac TG accumulation.

(A–E) ATP levels in cardiac muscle (A), cardiac TG levels normalized to tissue weight (B), ratio of cardiac mitochondrial gene DNA (ATPase6) to nuclear gene DNA (β-actin) (C), complex I:citrate synthase activity ratio (D) and complex IV:citrate synthase activity ratio (E) normalized citrate synthase activity in 8–12 months old αMHC-Klf5^−/− and floxed (WT) mice (n=5–7; *p<0.05). Data for all bar graphs are represented as means ± SEM (statistical analysis: t-test).

αMHC-Klf5^−/− mice develop cardiac dysfunction

Because cardiac myocyte-specific deletion of Klf5 reduces Ppara, FAO and cardiac ATP levels, we hypothesized that αMHC-Klf5^−/− mice would have compromised cardiac function. Indeed, although young mice (2–3 months old) had normal cardiac function (Figure 6A–C), mice begin to develop cardiac dysfunction at the age of 6 months (Figure 6D–F). As the mice aged further (8–12 months old), the dysfunction progressed (Fig. 6G) with signs of dilated cardiomyopathy, as shown by reduced fractional shortening (Fig. 6H) and increased left ventricular internal dimension during both diastole (Fig. 6I) and systole (Fig. 6J). Left ventricular posterior wall thickness during either diastole or systole was not different in 6 (Online Figure IX) and 8–12 months old (Figure 6K, 6L) αMHC-Klf5^−/− mice. Heart weight/tibia length ratio was slightly increased (Fig. 6M) in αMHC-Klf5^−/− mice. The mRNA levels of heart failure marker brain natriuretic peptide (BNP) were increased in both young (Fig. 6N) and old αMHC-Klf5^−/− mice (Fig. 6O), while atrial natriuretic factor (ANF) gene expression was increased in old mice (Fig. 6O). The expression of aMHC and bMHC was not altered either in young or old αMHC-Klf5^−/− mice (Fig. 6N, 6O). Thus, αMHC-Klf5^−/− mice have lower cardiac FAO-gene expression profile, increased lipid accumulation and dilated cardiomyopathy.

Figure 6. Cardiac myocyte-specific ablation of Klf5 impairs cardiac function.

(A–F) Fractional shortening (A, D), left ventricular internal dimension during diastole (B, E), left ventricular internal dimension during systole (C, F), in 2–3 months old (A–C) and 6 months old (D–F) αMHC-Klf5^−/− and floxed (WT) mice. (G–M) Photographs of echocardiograms (G), fractional shortening (H), left ventricular internal dimension during diastole (I), left ventricular internal dimension during systole (J) left ventricular posterior wall during diastole (K) left ventricular posterior wall during systole (L), and heart weight/tibia length ratio (M) in 8–12 months old αMHC-Klf5^−/− and floxed (WT) mice (n=7–8; *p<0.05). (N, O) Cardiac mRNA levels for Bnp, Anf, αMHC and βMHC genes in 2–3 months old (N) and 11–12 months old (O) male floxed and αMHC-Klf5^−/− mice (F) (n=5; *p<0.05, **p<0.01 vs floxed mice).

KLF5 mediates altered cardiac Ppara gene expression in diabetes

Insulin deficient diabetes alters lipid metabolism of the heart and other organs. Ingenuity pathway analysis of the whole genome microarray data indicated that several downregulated genes in the hearts of αMHC-Klf5^−/− mice are involved in the post-translational regulation of insulin signaling proteins^{28, 29}, such as AKT, phosphoinositide 3-kinase (PI3K), p38 MAPK and extracellular signal-regulated kinase (ERK)1/2 (Figure 7A). To test whether KLF5 mediates cardiac metabolism-related gene expression changes in diabetes, we induced diabetes in wild type mice (C57BL/6) with STZ intraperitoneal (ip) injections. Six weeks post-STZ administration, diabetic mice with increased plasma glucose levels (CTRL: 149 ± 6.9 mg/dl, STZ: 450 ± 61 mg/dl; p<0.001; n=6) had mild cardiac dysfunction as shown by reduced fractional shortening levels (Figure 7B). Hyperglycemia reduced cardiac KLF5 protein levels (Figure 7C), as well as Klf5 and Ppara gene expression levels by 35% and 55%, respectively (Figure 7D). Induction of hyperglycemia in αMHC-Klf5^−/− mice (Online Figure X-A) with STZ administration did not reduce further cardiac Ppara gene expression levels (Online Figure X-B), neither did it worsen cardiac function, compared to diabetic wild type mice (Online Figure X-C).

Figure 7. Diabetes inhibits cardiac Klf5 and Ppara gene expression.

(A) Ingenuity pathway analysis of cardiac genes regulated over 2-fold in the aMHC-Klf5^−/− mouse array that have direct or indirect association with insulin signaling and glucose metabolism proteins. Highlighted with bold fonts within the diagram are proteins that modulate insulin signaling. (B) Fractional shortening of C57BL/6 mice 6 weeks post-STZ or saline (CTRL) administration (n=5; *p<0.05 vs CTRL). (C) Western blot analysis for cardiac KLF5 and β-actin protein levels in C57BL/6 mice 6 weeks post-STZ administration (n=3; ***p<0.001 vs CTRL). (D) Cardiac Klf5 and Ppara mRNA levels in floxed and aMHC-Klf5^−/− mice 6 weeks post-STZ administration (n=5; *p<0.05, **p<0.01 vs CTRL). (E) Cardiac Klf5 and Ppara mRNA levels in 12 weeks old ob/ob mice compared with wild type C57BL/6 mice (n=4–5, *p<0.05, ***p<0.001 vs wt). (F–I) Plasma glucose levels (F, G) and cardiac Klf5 and Ppara mRNA levels (H, I) in wild type mice treated with STZ (6 weeks prior to glucose measurement), dapagliflozin (F, H), antisense oligonucleotides against SGLT2 (SGLT2-ASO) (G, I) and combination of either STZ with dapagliflozin (F, H) or STZ with SGLT2-ASO (G, I) (n=5, **p<0.01, ***p<0.001 vs CTRL).

Previous studies^{30, 31} have shown that cardiac Ppara expression is reduced in leptin-deficient B6.V-Lep^ob/J (ob/ob) mice, which is a model of Type 2 diabetes. Therefore, we tested whether this was associated with decreased Klf5 expression. We measured gene expression levels in the hearts of hyperglycemic 12-weeks old ob/ob mice. Indeed, these mice had reduced cardiac Klf5 (60%) and Ppara (70%) mRNA levels (Figure 7E).

Correction of hyperglycemia restores cardiac Klf5 and Ppara gene expression levels

We reduced plasma glucose levels in diabetic mice using pharmacologic or ASO-mediated inhibition of the SGLT2³². Both dapagliflozin (Figure 7F) and SGLT2 ASO (Figure 7G) administration corrected plasma glucose levels and normalized cardiac Klf5 and Ppara expression levels in STZ-treated wild type mice compared to mice treated with STZ alone (Figures 7H and 7I). Restoration of plasma glucose and cardiac Klf5 and Ppara gene expression levels was accompanied by prevention of cardiac dysfunction that was observed in wild type mice treated with STZ alone (Online Figure XI-A). On the other hand, despite normalization of plasma glucose levels (Online Figure XI-B), neither cardiac function (Online Figure XI-A) nor cardiac Ppara gene expression levels (Online Figure XI-C) were improved in αMHC-Klf5^−/− mice treated with STZ and dapagliflozin.

Klf5 and Ppara gene expression levels are increased in later stages of diabetes

Our data indicate that cardiac Klf5 and Ppara gene expression levels are reduced in the early stage of diabetes (6 weeks post-STZ administration). Although, our findings are consistent with previous studies showing downregulation^{33, 34} or lack of change³⁵ in cardiac Ppara gene expression during diabetes, other reports have associated diabetes with increased cardiac Ppara expression in mice³⁶ and increased cardiac FAO in diabetic humans³⁷. As it has been suggested that differential cardiac Ppara gene expression in diabetes may reflect differences in disease severity or duration³⁵, we measured cardiac Klf5 and Ppara expression levels in mice 12 weeks post-STZ administration. We found that both Klf5 (Online Figure XII-A) and Ppara (Online Figure XII-B) gene expression levels were increased at this later stage of diabetes. Accordingly, we measured cardiac Klf5 and Ppara gene expression levels in the early stages of hyperglycemia in db/db mice. Hyperglycemia begins in db/db mice at the age of 5 weeks and increases further 1 week later (Online Figure XII-C). Cardiac Klf5 (Online Figure XII-D) and Ppara (Online Figure XII-E) expression levels were reduced in 5 weeks old and increased in 6 weeks old db/db mice. Thus, Klf5 gene expression parallels both downregulation of Ppara in the early stages of diabetes, as well as its upregulation in the late stages of the disease.

DISCUSSION

We identified KLF5 as a regulator of cardiac Ppara that controls cardiac FAO^{8, 38}. We also showed that cardiac myocyte-specific Klf5 ablation reduces cardiac FAO and ATP content and leads to cardiac dysfunction. Moreover, we suggest that KLF5 regulates cardiac Ppara expression in Type 1 diabetes.

Balanced FA metabolism is critical for normal cardiac function. Either deficiency or excess of cardiac FAO can lead to organ dysfunction characterized either as energetic starvation that occurs in heart failure^{3, 39} and sepsis^5–7 or lipotoxicity that is observed in obesity and diabetes⁴⁰. Here we report that KLF5 is a transcriptional regulator of Ppara, which implicates it as a novel regulatory protein for cardiac FA metabolism.

KLFs regulate proliferation, differentiation, development, and cell death¹¹. KLF5 either increases or reduces lipid metabolism in several tissues, such as adipose tissue⁴¹, skeletal muscle²³ and lung²⁴. However, the role of KLF5 in cardiac myocyte metabolism has not been examined. A previous study²² showed that cardiac myocyte-specific Klf5 ablation did not prevent pressure overload cardiac hypertrophy. Cardiac myocyte Ppara downregulation, which occurs in our αMHC-Klf5^−/− mice, may account for the lack of protection from hypertrophy in these mice, as shown by increased cardiac hypertrophy in Ppara^−/− mice⁴². As homozygous Klf5^−/− mice are lethal, in vivo systemic metabolic studies were performed in global heterozygote Klf5^+/− mice²³. Klf5^+/− mice have defects in development of WAT, as KLF5 is crucial for adipocyte differentiation⁴¹. Thus, Klf5^+/− mice are resistant to high-fat diet-induced obesity, hypercholesterolemia and glucose intolerance, although they consume more food compared to wild type mice²³. However, the effects of haploinsufficiency on cardiac FAO were not studied.

Our initial findings linking cardiac Klf5 upregulation with inhibition of Ppara and cardiac energy production in LPS-treated mice and cardiac myocyte cell lines⁵ led us to the hypothesis that KLF5 is a repressor of Ppara expression. However, we found that KLF5 activates cardiοmyocyte Ppara. The observed Klf5 upregulation in the hearts of LPS-treated mice may be attributed to other mechanisms, which pertain to inflammatory pathways that are activated in LPS-treated animals. Nevertheless, LPS increases Klf5 expression, which induces NF-κB and inflammation in intestinal epithelial cells⁴³. Accordingly, the increased levels of cardiac Klf5 expression that we measured during later stages of diabetes (Online Figure XII) may be associated with increased inflammation that develops during progression of the disease^{44, 45}. Binding of c-Jun in the Ppara promoter region that KLF5 could also occupy may constitute a critical event that explains the lack of upregulation of Ppara expression despite the increased levels of KLF5 in LPS-treated mice. Moreover, the increased levels of KLF5 that we observed in the hearts of LPS-treated mice and cardiac myocytes may represent a compensatory response of cardiac myocytes to the suppression of Ppara gene expression and the energetic deficiency.

Cardiac KLF5 is a positive regulator of Ppara and changes in Klf5 expression induce parallel changes in Ppara gene expression in non-LPS treated mice. Cardiac myocyte-specific Klf5 ablation downregulated the expression of Ppara and a broad range of cardiac FAO-related genes, leading to reduced cardiac FAO, lower ATP levels, increased cardiac TG accumulation, and cardiac dysfunction. This finding is consistent with previous studies^46–49 showing that excessive cardiac lipid accumulation can lead to dilated cardiomyopathy in animal models^{46, 49} and humans⁴⁸. Interestingly, cardiac Ppara gene expression reverses as the αMHC-Klf5^−/− mice become older and TG accumulate in their hearts. Cardiac Ppara expression reduces with aging⁵⁰. Thus, our observation may indicate potential non-KLF5-dependent mechanisms that are triggered by cardiac TG expansion and influence PPARα activation as described before¹⁰. The increased glucose oxidation levels that we observed in αMHC-Klf5^−/− mice may also exacerbate further cardiac dysfunction, as shown previously⁵¹. The method we used to assess cardiac fatty acid and glucose oxidation is based on incubation of whole hearts obtained from mice in a buffer that contains radiolabeled (¹⁴C) palmitate or glucose and measurement of ¹⁴CO₂ that is released. The data we obtained with this method are consistent with the decreased cardiac fatty acid metabolism-related gene expression profile and a trend for reduced cardiac fatty acid uptake that we observed. Indeed, this is an in vitro approach that is performed in non-working hearts and may not sense fully the metabolic alterations that occur in a working heart. Thus, future studies with isolated working hearts perfused via the coronary arteries will help to complete the assessment of the cardiac metabolic effects of cardiac myocyte Klf5 ablation.

Changes in cardiac energetics and PPARα vary depending on the type of cardiac disease. Failing hearts show deficit in energy production and alterations in the source of energy substrates. Ppara^−/− mice have decreased cardiac fatty acid metabolism^{52, 53} and decreased longevity⁵⁴ but normal cardiac function at baseline⁵⁵. However, starvation causes contractile dysfunction⁵⁶ and lower cardiac ATP levels in Ppara^−/− mice, primarily due to defective fatty acid uptake⁵³. Heart failure is accompanied by lower mitochondrial oxidative metabolism and ATP levels³. The extent of the energetic deficiency during heart failure is influenced by the stage and cause of heart failure⁴⁰. Pressure overload-induced cardiac hypertrophy is accompanied by reduced PPARα levels⁴⁰, lower FAO and elevated glycolysis. Pharmacological PPARα activation with fenofibrate administration attenuates cardiac hypertrophy in mice⁵⁷ and rats⁵⁸. Increased flux of FA from intracellular TG to mitochondria for FAO also improves contractility in hearts with pressure overload¹⁰. On the contrary, Ppara haploinsufficiency attenuates pressure overload hypertrophy and failure⁵⁹. Ischemia-related heart failure either activates or inhibits PPARα, depending on the timing and experimental model. Mitochondrial oxidative metabolism is reduced during ischemia due to lower oxygen supply. However, cardiac FAO is induced during reperfusion of the ischemic heart⁶⁰. Pharmacological activation of PPARα 8–12 weeks post-myocardial ischemia aggravates cardiac hypertrophy⁶¹. Similarly, isolated hearts from either αMHC-Ppara mice or mice that are subjected to ischemia/reperfusion and are treated with a PPARα agonist have lower cardiac power⁶² and increased infarct size⁶³. On the other hand, PPARα activation improves contractile function and reduces infarct size in isolated perfused rat hearts following ischemia/reperfusion⁶⁴. Beneficial effects on infarct size and cardiac performance have also been reported in rats⁶⁵ following in vivo ischemia/reperfusion and treatment with PPARα agonist. Thus, PPARα can have either protective or aggravating effects in cardiac function.

We assessed Klf5 and Ppara expression in diabetic animal models. In agreement with other studies, we found that in animal models of both insulin-deficient^{33, 34} and Type 2 diabetes^{30, 31} cardiac Ppara mRNA levels were reduced in the early stage of the disease. In both cases this was associated with reduced Klf5 gene expression. KLF5-mediated regulation of Ppara expression was dependent on glucose changes. This was demonstrated by correction of hyperglycemia with SGLT2 inhibition that restored normal cardiac Klf5 and Ppara gene expression levels and prevented cardiac dysfunction in a Klf5-dependent manner. Cardiac cells can take up glucose via both the insulin-dependent GLUT4, the insulin-independent GLUT1, and non-specific pinocytosis. Glucose uptake is not completely inhibited in Type 1 diabetes animal models as they can still take up glucose via GLUT1, while membranous GLUT4 protein levels are reduced by approximately 35%⁶⁶. Therefore, the observed reduction in Klf5 and Ppara during the early stages of insulin-dependent diabetes may reflect increased utilization of glucose for energy production in cardiac myocytes instead of the preferred FAs. Further studies remain to be performed in order to elucidate whether downregulation of cardiac Ppara gene expression in the early stage of diabetes is accompanied by concomitant reduction of cardiac fatty acid oxidation, which reverses afterwards.

Our findings linking hyperglycemia in the early stage of diabetes with reduced Ppara mRNA levels deviate from previous studies linking diabetes with increased cardiac Ppara mRNA or protein levels in STZ-treated wild type mice³⁶ or diabetic mice with mutated leptin receptor (db/db)^{30, 67}. Another study also showed that diabetic patients have increased myocardial FAO³⁷. On the other hand, our data are consistent with studies showing that cardiac Ppara expression are reduced in the diabetic Akita mice (C57BL/6 background)³³, as well as in rats⁶⁸ and in isolated hearts obtained from STZ-treated wild type mice³⁴. The discrepancy between these studies may be attributed to different experimental conditions, such as the age of mice, genetic background or duration of hyperglycemia. Nevertheless, it has been proposed that the intensity of diabetes and availability of plasma lipids may account for differential effects of the disease on cardiac Ppara expression levels³⁵. Our STZ experiments were performed in 10 weeks old C57BL/6J mice. Animals in previous studies that showed increased cardiac Ppara expression³⁶ were of mixed C57BL/6 × CBA/J³⁶ or C57BLKS⁶⁷ background. When the duration of hyperglycemia was extended to 3 months in our experiments, cardiac Klf5 and Ppara gene expression levels were increased. Accordingly, early stage of hyperglycemia in db/db mice was accompanied by reduced cardiac Klf5 and Ppara expression, which is reversed afterwards. Thus, inhibition of Klf5 and Ppara gene expression levels may represent an early cardiac response during diabetes that is followed by reversal of both Klf5 and Ppara expression at a later stage depending on the intensity of hyperglycemia and availability of fatty acids. Nevertheless, changes in cardiac Klf5 and Ppara gene expression are parallel during both early and late stage of diabetes.

In summary, our findings show that KLF5 is a positive transcriptional regulator of cardiac Ppara and its inhibition leads to cardiac dysfunction. Moreover, Klf5 expression changes parallel those of cardiac Ppara in diabetes. Thus, cardiac KLF5 emerges as a potential therapeutic target for several types of cardiac dysfunction that are associated with PPARα alterations and energetic deficiency.

NOVELTY AND SIGNIFICANCE.

What Is Known?

• Sepsis downregulates cardiac peroxisome proliferator-activated receptor (Ppar) α gene expression.

• Cardiac PPARα is upregulated in diabetic cardiomyopathy and is considered as the major driver of increased cardiac fatty acid oxidation.

What New Information Does This Article Contribute?

• Krüppel-like factor (KLF) 5 is a positive regulator of Ppara gene transcription.

• LPS-induced sepsis prevents binding of KLF5 in the Ppara promoter.

• Klf5 and Ppara gene expression are downregulated at the early stage of diabetes and increased as the disease progresses.

Cardiac fatty acid oxidation accounts for 70% of the ATP that is produced in the heart. PPARα is a nuclear receptor with a central role in the transcriptional regulation of various proteins that contribute in cardiac fatty acid metabolism. Sepsis, systemic inflammatory response that follows bacterial infection, suppresses cardiac fatty acid oxidation, which is accounted for by downregulation of cardiac myocyte PPARα. This study identified KLF5 as a positive regulator of Ppara gene expression. During sepsis KLF5 binding on Ppara gene promoter is prevented. Cardiac myocyte-specific Klf5 ablation reduced cardiac fatty acid oxidation and led to cardiac dysfunction in older mice. Ppara gene expression is also altered in diabetes. This study showed that a dynamic oscillation of cardiac Ppara gene expression occurs in diabetes. The initial response is constituted by downregulation of cardiac Ppara and is followed by increased Ppara expression as the disease progresses. Both early downregulation and late upregulation of Ppara are associated with respective changes of cardiac Klf5 gene expression. Further delineation of the metabolic profile of the early cardiac response in diabetes may indicate new targets for treating diabetic cardiomyopathy.

Nonstandard Abbreviations and Acronyms

FAO

Fatty acid oxidation

FA

fatty acid

TG

triglycerides

PPARα

peroxisome proliferator-activated receptor α

KLF

Krüppel-like factor

TGF

transforming growth factor

LPS

lipopolysaccharides

ChIP

chromatin immunoprecipitation

SGLT

sodium/glucose cotransporter

ASO

anti-sense oligonucleotides

JNK

c-Jun N-terminal kinase

αMHC

alpha myosin heavy chain

Acot

acyl-CoA thioesterase

Serpina

serine (or cysteine) peptidase inhibitor, clade A

Phka1

phosphorylase kinase α1

PGC

PPARγ coactivator

Acox

acyl-CoA oxidase

Cpt1b

carnitine palmitoyl-transferase

CD

cluster of differentiation

Lpl

lipoprotein lipase

Dgat1

diacylglycerol acyltransferase

Plin

perilipin

PI3K

phosphoinositide 3-kinase

ERK

extracellular signal-regulated kinase

STZ

streptozotocin

Lcad

long chain acyl-CoA dehydrogenase

Vlcad

very long chain acyl-CoA dehydrogenase

Glut

glucose transporter

PDK

pyruvate dehydrogenase kinase

UCP

uncoupling protein