Dysregulation of RBFOX2 Is an Early Event in Cardiac Pathogenesis of Diabetes

Abstract

Alternative splicing (AS) defects that adversely affect gene expression and function have been identified in diabetic hearts; however, the mechanisms responsible are largely unknown. Here, we show that the RNA-binding protein RBFOX2 contributes to transcriptome changes under diabetic conditions. RBFOX2 controls AS of genes with important roles in heart function relevant to diabetic cardiomyopathy. RBFOX2 protein levels are elevated in diabetic hearts despite low RBFOX2 AS activity. A dominant-negative (DN) isoform of RBFOX2 that blocks RBFOX2-mediated AS is generated in diabetic hearts. DN RBFOX2 interacts with wild-type (WT) RBFOX2, and ectopic expression of DN RBFOX2 inhibits AS of RBFOX2 targets. Notably, DN RBFOX2 expression is specific to diabetes and occurs at early stages before cardiomyopathy symptoms appear. Importantly, DN RBFOX2 expression impairs intracellular calcium release in cardiomyocytes. Our results demonstrate that RBFOX2 dysregulation by DN RBFOX2 is an early pathogenic event in diabetic hearts.

Figure 1. 73% of transcripts aberrantly spliced in diabetic mouse hearts have RBFOX2 binding sites

A) Pie chart of RBFOX2 binding clusters in 967 transcripts that are mis-spliced in Type 1 diabetic mouse (T1D) hearts. Alternative splicing (AS) events affected in diabetic hearts with at least one CLIP peak at a Bayes factor ≥ 1 was designated as “significant”, <1 as “not significant”. B) RBFOX2 binding site distribution among 1417 RBFOX2 binding clusters within 704 pre-mRNAs aberrantly spliced in diabetic hearts. (See Fig. S1) C) Top gene ontology categories for RBFOX2 regulated genes affected in diabetic hearts. DAVID server was used to categorize genes based on biological or molecular function. P values represent statistical significance when compared to the whole genome, which was used as a background control. D) Effect of diabetes induced AS changes on gene expression and function. E) Schematic of intronic RBFOX2 binding sites obtained from human ES cell RBFOX2 CLIP-seq data with CLIP clusters and consensus RBFOX2 binding site motif (U)GCAUG on target pre-mRNAs mis-spliced in diabetic hearts.

Figure 2. Alternative splicing of RBFOX2 targets is disrupted in diabetic hearts independent of hypertension and obesity

A) Percent inclusion of MTMR3 exon 16, FXR1 exon 15+16, MEF2A exon 9 and PBX3 exon 7 as determined by quantitative RT-PCR (qRT-PCR) in non-diabetic (control) or Type 2 diabetic (T2D) human left ventricles (n≥3). B) AS analysis of Mtmr3 exon 16 in hypertensive, obese and obese-hypertensive mouse models (See also Fig. S2) (n=3). C) Body weight of mock treated control, angiotensin II treated hypertensive, obese, and angiotensin II treated obese mice. D) Heart to body weight ratio of mock treated control mice or angiotensin II treated hypertensive mice. E) AS analysis in chagasic mice with severe cardiomyopathy and heart failure. Mtmr3 exon 16, Fxr1 exon 15+16, Mef2a exon 9 and Pbx3 exon 7 AS was determined in non-infected (control) or chronically infected chagasic mice with cardiomyopathy (cardiomyopathy) (n≥3). Data represent means ± SD. Statistical significance was calculated using unpaired t-test for two group comparisons or using one-way ANOVA to compare four different groups. P-values are represented as **** < 0.0001, *** < 0.001, ** < 0.01, * < 0.05.

Figure 3. RBFOX2 regulates alternative splicing events altered in diabetic hearts

A) Schematic of Rbfox2 pre-mRNA (only exons 3 to 7) regions targeted by siRNA and shRNA. Introns were not drawn to scale. B) Total protein lysates from siRNA treated H9c2 cells analyzed by Western blotting (WB) using an antibody (Ab) against RBFOX2 to determine the depletion efficiency of Rbfox2 siRNA. GAPDH WB was performed to confirm even protein loading. C) RBFOX2 protein levels determined in shRNA expressing H9c2 cells by WB using an anti-RBFOX2 Ab. GAPDH WB was used as a loading control. D) AS analysis of Mtmr3 exon 16, Fxr1 exon 15+16, Mef2a exon 9, and Pbx3 exon 7 in H9c2 cells transfected with scrambled (control) or Rbfox2 specific siRNA (targets exon 6) (n≥8). E) AS analysis in H9c2 cells transfected with scrambled (control) or Rbfox2 specific shRNA (targets exon 3) (n≥6). In AS gel figures +E# = Exon # inclusion and −E# = Exon # exclusion. Data represent the means ± SD. The unpaired t-test was used to calculate statistically significant differences between two sample groups from 3 independent experiments. P-values are indicated as **** < 0.0001, *** < 0.001.

Figure 4. Dominant negative RBFOX2 is generated in diabetic hearts at early stages

A) WB analysis of RBFOX2 protein in non-diabetic control or Type 2 diabetic (T2D) human left ventricles using an anti-RBFOX2 Ab. α-TUBULIN WB was used as a loading control. Fold change in RBFOX2 protein levels were determined after normalizing protein levels to α-TUBULIN, the control averages were set to 1 and patient averages were summarized below. B) Relative RBFOX2 mRNA levels in non-diabetic control and T2D human patients’ hearts, determined by qRT-PCR (n≥5). C) Schematic of dominant negative (DN) RBFOX2 generation via exon 6 exclusion (RRM: RNA recognition motif). Introns were not drawn to scale. D) DN RBFOX2 mRNA levels were determined by RBFOX2 exon 6 exclusion using qRT-PCR from left ventricles of T2D human patients or non-diabetic controls (n≥5). E) DN Rbfox2 expression in hearts from non-diabetic control mice or non-obese diabetic (NOD:T1D) mice (n≥4). F) DN Rbfox2 expression in hearts from mock injected control or streptozotocin induced T1D (STZ:T1D^3wks) mice hyperglycemic for 3 weeks (n≥3), G) Cardiac function of mice before (Pre-STZ) or 3 weeks after hyperglycemia (STZ:T1D^3wks) was assessed by M-mode echocardiography measuring fractional shortening (%), H) stroke volume (Vol.), I) ventricular internal diameter end at diastole (LVIDd) (mm) (n≥3). J) DN Rbfox2 expression in left ventricles from chagasic mice with severe cardiomyopathy or mock infected controls (n≥3). K) DN Rbfox2 expression pattern in left ventricles of obese, hypertensive and obese/hypertensive mouse left ventricles (n=3). Data represent the means ± SD. Statistical significance between two groups was calculated using the unpaired t-test, and between four groups using one-way ANOVA. P-values are indicated as *** < 0.001, ** < 0.01.

Figure 5. Increased RBFOX2 protein with RNA binding capability is important for generation of DN Rbfox2

A) WB analysis of RBFOX2 in H9c2 cells expressing empty vector (Control), RBFOX2^WT, or RBFOX2^RRM (RNA binding deficient mutant). B) Endogenous DN Rbfox2 expression by exclusion of exon 6 in H9c2 cells expressing control, RBFOX2^WT or RBFOX2^RRM determined by qRT-PCR (n=4 from 3 independent experiments). Data represent the means ± SD. C) RBFOX2 WB in H9c2 cells expressing GFP or increasing levels of GFP-tagged RBFOX2 using anti-GFP or anti-RBFOX2 Ab. Ponceau S staining was used to assess even protein loading. D) AS of Mef2a, Mtmr3 and E) Mbnl2 and expression levels of DN Rbfox2 in H9c2 cells expressing increasing levels of RBFOX2^WT-GFP (See Fig. S3). To equalize the amount of DNA per plate to 2 μg, different concentrations of vector DNA were transfected together with RBFOX2^WT-GFP. Quantifications were performed using samples from at least four independent experiments (n≥4). Data represent the means ± SE. Statistical significance was calculated using one-way ANOVA to compare three or more different groups. P-values are indicated as **** < 0.0001, *** < 0.001, ** < 0.01 compared to the GFP control (0 μg RBFOX2 DNA). P-values indicated as #### < 0.0001 represents comparisons to the lowest RBFOX2^WT-GFP level (0.2 μg RBFOX2 DNA).

Figure 6. Dominant negative RBFOX2 interacts with wild type RBFOX2 and inhibits AS of RBFOX2 regulated splicing

A) WB analysis of DN RBFOX2 protein in H9c2 cells expressing empty vector (Control) or FLAG-RBFOX2^DN. B) AS analysis of Mtmr3, Fxr1, and Mef2a in H9c2 cells transfected with empty vector (Control) or FLAG-RBFOX2^DN (n≥4). C) Interactions of FLAG tagged RBFOX2^WT, RBFOX2^DN, and RBFOX2^RRM (RNA binding deficient mutant) with U1C protein in COS M6 cells determined by FLAG IP followed by WB using FLAG and U1C-specific Abs. D) Interactions of FLAG-RBFOX2^DN with GFP-RBFOX2^WT in H9c2 cells. Pulled down proteins were analyzed by WB using HRP conjugated anti-FLAG and anti-GFP Abs. Control lysates expressing empty vector was used as a negative control for non-specific protein binding to the FLAG-antibody coated beads (FLAG-IP control). Input lysates represents 1/20 of total protein loaded onto the FLAG-beads. Results are representative of three independent experiments.

Figure 7. Consequences of aberrant AS of RBFOX2 targets

A) WB analysis of TALIN2 (TLN2), RBFOX2, and GAPDH in H9c2 cells treated with scrambled siRNA (control) or Rbfox2 specific siRNA. Fold changes in TLN2 protein levels were determined after normalizing to GAPDH protein levels (n≥9 from 3 independent experiments). B) Representative WB analysis of TLN2 protein levels in STZ:T1D mice. α-TUBULIN WB was used as a loading control. Fold changes in TLN2 levels were determined after quantifying protein levels and normalizing to α-TUBULIN levels. TLN2 levels in control mice were set as 1. Average values were represented as means ± SD (n=4). C) Representative tracing of calcium transients from human primary cardiomyocytes infected with AAV9-CMV-Null (Control) and AAV9-CMV-DN RBFOX2 (DN RBFOX2). Quantification of D) time to peak and E) amplitude of calcium transients. The numbers in the bar graphs represent the number of cardiomyocytes used for quantification from 3 independent experiments. Statistical significance was calculated using the unpaired t-test to compare 2 groups. P-values are indicated as **** < 0.0001, ** < 0.01, * < 0.05.