Repression of the Central Splicing Regulator RBFox2 Is Functionally Linked to Pressure Overload-Induced Heart Failure

Abstract

Heart failure is characterized by the transition from an initial compensatory response to decompensation, which can be partially mimicked by transverse aortic constriction (TAC) in rodent models. Numerous signaling molecules have been shown to be part of the compensatory program, but relatively little is known about the transition to decompensation that leads to heart failure. Here, we show that TAC potently decreases the RBFox2 protein in the mouse heart, and cardiac ablation of this critical splicing regulator generates many phenotypes resembling those associated with decompensation in the failing heart. Global analysis reveals that RBFox2 regulates splicing of many genes implicated in heart function and disease. A subset of these genes undergoes developmental regulation during postnatal heart remodeling, which is reversed in TAC-treated and RBFox2 knockout mice. These findings suggest that RBFox2 may be a critical stress sensor during pressure overload-induced heart failure.

Figure 1. Essential requirement for RBFox2 to prevent heart failure

(A) Western blotting analysis of RBFox2 in sham- and TAC-treated mice after 5 weeks of surgery. GADPH served as a loading control. (B) Kaplan-Meier survival curves of sham and TAC littermate male mice. (C) Diagram of conditional RBFox2 knockout and generation of heart-specific knockout by crossing with an Mlc2v-Cre transgenic mouse. (D) Western blotting analysis of RBFox1 and RBFox2 in wild-type (WT) and RBFox2 knockout (KO) cardiomyocytes at three postnatal stages. Triplicate results were quantified relative to WT, as indicated below each gel. (E) Survival curves of WT and RBFox2 KO littermate mice in both sexes.

Figure 2. General heart failure phenotype induced in RBFox2 ko mice

(A) Examples of cardiac echocardiography of WT and RBFox2 KO mice acquired at 9 weeks of age. (B) Key echocardiography parameters of WT and RBFox2 KO mice at different ages. LVID-D: left ventricular internal dimension values at end-diastole; LVID-S: left ventricular internal dimension values at end-systole. The data in each group were collected from 5 to 8 mice and expressed as mean±SEM. *P <0.05; **P <0.01. (C) H&E staining of the coronal sections of WT and RBFox2 KO hearts at different ages. (D) Size and mass of WT and RBFox2 KO hearts at 9 weeks of age. Left panel: Side by side comparison between WT and KO hearts at 9 weeks. Right panel: Statistics of the heart-to-body weight ratio of WT and KO hearts. The data in each group were collected from 3 mice and expressed as mean±SEM. (E) Size of cardiomyocytes isolated from WT and RBFox2 KO hearts at 9 weeks of age. Left panel: Transmission images of WT and RBFox2-deleted cardiomyocytes. Right panel: Statistics of the cell surface area of WT and KO cardiomyocytes. The data in each group were based on analysis of 23 to 25 cells from 3 mouse hearts and expressed as mean±SEM. (F) Statistics of fibrosis from Trichrome-stained WT or RBFox2 KO hearts at different development stages. The data in each group were based on analysis of 5 to 6 images from 2 mice and expressed as mean±SEM. **P <0.01. (G) Trichrome staining for fibrosis and myofibril disorder in RBFox2 KO heart by 22 weeks. See also Figure S1.

Figure 3. Defective EC coupling in RBFox2 knockout cardiac muscle

(A) Confocal images of intracellular calcium transients and contraction. Isolated cardiomyocytes from WT and KO mice at 9 weeks of age were paced electrically at 1.0 Hz. Lower panels show the traces of spatially averaged calcium transient (top) and the corresponding cell shortening (bottom). (B and C) Statistics of calcium transient peak (ΔF/F_0, which is defined as the ratio between the signal and baseline) and twitch amplitude (TA) in cardiomyocytes from WT and KO mice at two different ages. The data were based on analysis of 13 to 20 cells from three hearts and expressed as mean±SEM. **P <0.01. (D) Confocal measurement of spontaneous calcium sparks in 9 weeks WT or KO cardiomyocytes at rest. (E) Statistics of the frequency of calcium sparks in cardiomyocytes from WT or KO mice at two different ages. The data were based on analysis of 14 to 20 cells from three hearts and expressed as mean±SEM. **P <0.01. See also Figure S2 on the results observed on RBFox2 siRNA-treated cardiomyocytes.

Figure 4. Primary and accumulated defects in disease progression

(A) The heatmap of the splicing profiling data by RASL-seq among three different disease stages. The data were sorted by the mean value of the three stages groups analyzed. Green: induced inclusion; red: induced skipping. Genes that were previously linked to dilated cardiomyopathy are highlighted on the right. (B) Number of > 2-fold significant events among disease stages. (C) The genomic distribution of RBFox2 CLIP-seq reads. The top enriched motif and the percentage of the motif on total CLIP peaks are shown on the lower panel. (D and E) The RNA maps of RBFox2 binding and splicing responses based on those with >2-fold change relative to WT consistently detected among all three stages. (D) versus changes detected only in mice of week 9 and 18 (E). See also Figure S3 and Table S2.

Figure 5. Critical splicing changes linked to specific defects in the contractile apparatus

(A) Induced exon inclusion in Pdlim5. Top: The gene structure, illustrating the alternative exon 5a and 5b. Yellow and red lines indicate the locations of non-conserved and conserved (U)GCAUG motifs, respectively. Middle: The RBFox2 CLIP-seq signals. Bottom: the RT-PCR results and calculated percentages of Splice-In (included exon divided by both isoforms) in each sample. (B) Induced exon skipping in Ldb3. The gene structure, motif distribution, and CLIP-seq signals were similarly annotated as in A. (C) Induced exon skipping in Mef2a. The gene structure, motif distribution, and CLIP-seq signals were similarly annotated as in A. (D) PCR validation of 11 RASL events at three different disease stages. Black line shows the correlation of all 33 PCR results. (E) Western blotting analysis of PDLIM5 and LDB3 in WT and KO cardiomyocytes. Note that both PDLIM5 and LDB3 have long and short isoforms resulting from alternative polyadenylation. In both cases, the long isoform contains the alternatively spliced products. The percentages of exon inclusion based on the intensity of protein isoforms on the gel. The results had been repeated on different sets of WT and KO hearts. (F) Western blotting analysis of multiple MEF2 family proteins. α-ACTININ and c-TNT served as loading controls. (G) Percentage of directly RBFox2 binding events in different RASL-seq data groups. (H) Gene ontology analysis of direct RBFox2 target genes in biological processes based on the DAVID database. See also Figure S4.

Figure 6. RBFox2 ablation induce alternative splicing events in the opposite directions during heart remodeling

(A) Positive correlation between significantly changed splicing events detected response to RBFox2 KO and TAC surgery. Green-colored dots show changes in the same directions; Orange-colored dots indicate changes in the opposite directions, as summarized in the right panel. (B) Negative correlation between 54 events commonly induced by RBFox2 KO and TAC and those detected during postnatal heart modeling from day 1 to day 28. Green-colored dots show changes in the same directions; Orange-colored dots indicate changes in the opposite directions. (C) Heatmap of 36 regulated splicing events that show opposite changes during heart remodeling relative to RBFox2 KO and TAC surgery. Red-labeled on the right are genes previously implicated in heart function and/or disease. (D) Diagram that highlights 15 genes (from red-labeled ones in C) with diverse regulatory functions in cardiomyocyte. IR: insulin receptor. See also Figure S5 and Table S3.