RBM20^S639G mutation is a high genetic risk factor for premature death through RNA-protein condensates

Abstract

Dilated cardiomyopathy (DCM) is a heritable and genetically heterogenous disease often idiopathic and a leading cause of heart failure with high morbidity and mortality. DCM caused by RNA binding motif protein 20 (RBM20) mutations is diverse and needs a more complete mechanistic understanding. RBM20 mutation S637G (S639G in mice) is linked to severe DCM and early death in human patients. In this study, we generated a RBM20 S639G mutation knock-in (KI) mouse model to validate the function of S639G mutation and examine the underlying mechanisms. KI mice exhibited severe DCM and premature death with a ~50% mortality in two months old homozygous (HM) mice. KI mice had enlarged atria and increased ANP and BNP biomarkers. The S639G mutation promoted RBM20 trafficking and ribonucleoprotein (RNP) granules in the sarcoplasm. RNA Seq data revealed differentially expressed and spliced genes were associated with arrhythmia, cardiomyopathy, and sudden death. KI mice also showed a reduction of diastolic stiffness and impaired contractility at both the left ventricular (LV) chamber and cardiomyocyte levels. Our results indicate that the RBM20 S639G mutation leads to RNP granules causing severe heart failure and early death and this finding strengthens the novel concept that RBM20 cardiomyopathy is a RNP granule disease.

Graphical Abstract

Introduction

RNA binding motif protein 20 (RBM20) is an RNA binding protein and a splicing factor that is highly expressed in heart muscle [1, 2]. Genetic linkage analysis in human patients with dilated cardiomyopathy (DCM) identified mutations in Rbm20 in ~3% of idiopathic DCM patients [1, 3]. These findings have placed Rbm20 on the list of DCM-causing genes. DCM induced by RBM20 is now referred to as RBM20 cardiomyopathy. Loss-of-function studies using RBM20 knockout (KO) rats and mice revealed changes in splicing variants in over 30 genes including the myofilament gene TTN and calcium handling genes [2, 4–6]. Furthermore, complete depletion of RBM20 in rats and mice results in the development of DCM [2, 7]. Our recent study showed that adult Rbm20 KO rats develop a DCM-like phenotype at about 6-month-old [8]. However, deletion of only the RNA binding motif (RRM) domain did not lead to an overt DCM-like phenotype even though the mice with RRM deletion manifested reduced Frank-Starling Mechanism, and an increase in fibrosis [5]. The underlying mechanisms by which complete RBM20 depletion results in DCM in rats and mice are widely accepted to be mainly associated with 1) reduced diastolic stiffness through alternative splicing of TTN, encoding the giant myofilament protein titin that is responsible for ventricular wall stiffness, and 2) impaired cardiomyocyte contractility through alternative splicing of calcium handling genes such as Ryr2 and Camk2d [2, 4–6]. RBM20 depletion in rats leads to sudden death in about 17% of adult rats [2].

Recently, a Rbm20 mutation knock-in (KI) (Rbm20^R636S) pig model [9] and a KI mouse model (Rbm20^S637A) [10] were generated revealing that pigs develop severe DCM and fatal circulatory failure in the majority of neonatal homozygous pigs [9] and atrial fibrillation and arrythmias in mutated mice [10]. These findings are consistent with human genetic studies showing that Rbm20 mutations are associated with young age, end-stage heart failure and high mortality [1, 3, 11–17]. They also reported that the Rbm20^R636S and Rbm20^S637A mutations facilitated RBM20 re-localization from the nucleus to the cytoplasm and assembly of ribonucleoprotein (RNP) granules. These findings suggest that beyond gene splicing, RNP granules might also be responsible for severe DCM and early death by Rbm20 mutations. A search of the National Center for Biotechnology Information ClinVar database (https://www.ncbi.nlm.nih.gov/clinvar) using gene symbol Rbm20, reveals over 900 Rbm20 mutations in need of functional validation. Two validated mutations in pig and mouse models are located in the highly conserved arginine/serine (RS)-rich domain, suggesting mutations in the RS domain seem relevant to RBM20 function.

In this study, we chose another uncharacterized mutation S639G in the RS region found in human patients to study whether this mutation is pathogenic. We generated an RBM20^S639G mutation KI mouse model and examined the functional effects of the RBM20^S639G mutation. Immunofluorescent staining in both paraffin-embedded tissues and isolated single cardiomyocytes was performed to determine protein trafficking and condensates. Transcriptomic profiling with deep RNA sequencing was carried out to identify genes with altered expression and splicing. Our results expand the functional validation of Rbm20 mutations and further strengthens the novel concept that RBM20 cardiomyopathy is a protein condensate disease.

Results

RBM20^S639G knock-in (KI) mice display enlarged heart chambers and thinner ventricular walls at an early age.

Rbm20^S639G mutant KI mice were generated through CRSPR/Cas9 gene-editing technology. Sanger sequencing established the successful mutation at amino acid S639 and PCR confirmed the genotypes (Figure 1A). Both heterozygous (HT) and homozygous (HM) Rbm20^S639G breeders yielded an average litter size (6–10) that follow Mendelian genetics. Gross anatomy demonstrated that HT and HM hearts had thin myocardial walls and were flabby, flimsy and distinctive from healthy wild type (WT) mice (Figure 1B). The whole heart of HT and HM mice could not maintain a sphere-like shape, a phenomenon noted in two months old male and female mice (Figure 1B). Histological assessments with hematoxylin and eosin (H&E) (Figure 1C) and Masson’s trichrome staining (Figure 1D) revealed that both HT and HM mice displayed a DCM-like phenotype presenting as enlarged left ventricular chamber and thinner ventricular wall in two-month-old male and female mice. HM mice exhibited a more severe DCM-like phenotype (Figure 1C and 1D). However, histological analyses failed to reveal cardiac fibrosis in both HT and HM mice, as assessed by Masson’s trichrome staining (Suppl Figure S1A and S1B). To test whether the mutation affects the expression of RBM20 level, we performed western blotting and the results showed that the protein expression levels of RBM20 were significantly elevated in both HT and HM mice with further increase in HM mice (Figure 1E).

Figure 1. Rbm20^S639G KI mouse model, histological and gross anatomical assessments.

(A) Schematic diagram of Rbm20 gene structure and generation of a S639G mutation knock-in mouse (Rbm20^S639G) through CRISPR/Cas9 technology. The schematic shows the mouse Rbm20 gene and coded protein domains. The mutation hot spot of RS domain is indicated as “RSRSP”. The target amino acid is highlighted in yellow in “RSRSP”, and the point mutation site is in red. The sgRNA is underlined in black and the orange nucleotides are in the silent mutation site. The desired switch from AGT (S) to GGT (G) introduced an XhoI restriction site and facilitated the genotyping. The edited Rbm20 alleles in mice were confirmed by Sanger sequencing and PCR genotyping. (B) Gross anatomy comparing WT, HT and HM hearts in the top panels of both male and female hearts, and slice anatomy at the mid-ventricular level comparing WT, HT and HM in the bottom panels of male and female hearts (n=3, the number of animals for WT, HT, and HM of each gender, respectively). The black line at the bottom represents a 0.5 cm ruler. (C) Representative images of hematoxylin and eosin staining of cross-section of mouse hearts at the age of two months in both males and females (n=3, the number of animals for WT, HT, and HM of each gender, respectively). The scale bar at the bottom indicates a 0.5 cm ruler. (D) Representative images of Masson’s trichrome staining in both males and females. The scale bar at the bottom indicates a 0.5 cm ruler. (E) Western Blotting of RBM20 and quantification in WT, HT and HM. N=3, number of male animals used, duplicate of blotting were calculated. Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used. RRM, RNA binding motif; RS, arginine/serine; WT, wild type; HT, heterozygotes; HM, homozygotes.

Functional assessment in RBM20^S639G mutation KI mice.

Impaired cardiac function can cause and exacerbate atrial fibrillation (AF) and arrhythmia, and lead to heart failure. To assess the cardiac function of the Rbm20 mutant KI mice, we performed both conscious and unconscious echocardiography in two months old mice. Electrocardiogram (ECG) recording was performed in combination with unconscious echocardiography. From the M-mode image analysis, both HT and HM survivors under unconscious conditions showed lower left ventricular (LV) ejection fraction (%EF) compared to WT mice in both genders (WT female, 40.4 ± 1.7%, WT male, 39.0 ± 2.1%; HT female 19.6 ± 1.9%, HT male, 20.0 ± 1.8%; HM female, 19.4 ± 3.3%, HM male, 28.1 ± 1.4%) (Figure 2A). Left ventricular end diastolic diameter (LVDd) and systolic diameter (LVDs) were significantly larger in HT and HM survivors than WT mice in both males and females (Figure 2B and 2C). Similarly, fractional shortening (%FS) and relative wall thickness (RWT) of the LV (2*LVPW/LVDd), were decreased significantly in HT and HM mice compared to control mice (Suppl Table S2 and Figure 2D). We also observed a difference in LV diastolic parameters, such as mitral E/A ratios (the ratio of blood velocity through the mitral valve in early diastole (the E wave) to blood velocity in late diastole caused by atrial contraction (the A wave)) between genotypes and genders (Suppl Table S2). The mitral E/A ratio increased in the female HT mice; however, the ratio decreased in males HT compared to WT individuals (Figure 2E). E/E’ (correlates LV filling pressure) was significantly elevated in the female HT mice but was normal in the males when compared to WT (Suppl Table S2). Increased filling pressure is one cause of LA hypertrophy and dilation, and this is consistent with the increase in LA dimensions found in HM females but not in males (Suppl Table S2). There were no major differences in right ventricular parameters, except RV internal diameter at systole (RVIDs) in females HM. Similarly, tricuspid annular plane systolic excursion (TAPSE), used as a parameter of global RV function, was only decreased in females (Suppl Table S2). In conscious echo, we found similar changes, such as increased LVDs, decreased %EF and %FS (Suppl Table S2) of HT mice both in females and males. In addition, the pressure-volume loop (PV-loop) analysis demonstrated a decreased %EF and increased end-systolic volume (ESV) and end-diastolic volume (EDV), indicating a dilated LV chamber, which is consistent with echocardiography data. We also observed that the HT mice had an increased potential energy (PE), decreased efficiency (Eff) and reduced maximum derivative of pressure (dP/dtmax) (Suppl Table S2). These data suggest that the RBM20^S639G KI mice present a DCM phenotype, with enlarged LV chamber dimensions, decreased RWT, and reduced systolic function, with the severity and presentation differing between males and females.

Figure 2. In vivo cardiac functional assessment in Rbm20^S639G mutation KI mice and transcriptomic profiling.

(A)-(E) Transthoracic echocardiography was performed on WT (n=17, 10 females, 7 males), HT (n=13, 6 females, 7 males), and HM (n=13, 7 females, 6 males) mice at the age of two months. (A) Ejection fraction (EF). (B) Left ventricular diastolic diameter (LVDd). (C) Left ventricular systolic diameter (LVDs). (D) Relative wall thickness (RWT). (E) Mitral E/A ratio. Tukey’s multiple comparisons test was used. (F) Networks of differentially expressed genes (between WT, HT, and HM mice) clustering in cardiomyopathy related HPO terms. (G) Heatmap of the 15 most up-and down- regulated genes (HM vs. WT) within the 10 HPO terms selected in (F). (H) Validation of selected genes with quantitative real-time PCR (n=5, the number of animals). WT, wild type; HT, heterozygotes; HM, homozygotes; NS, non-significance. Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used.

RBM20^S639G mutation leads to altered gene expression linked to cardiac dysfunction.

We next performed transcriptomic profiling with deep RNA-sequencing (RNA Seq) to determine differentially expressed genes (DEGs) in RBM20^S639G KI male mice. As expected, HM LVs had the highest number of DEGs when compared to HT and WT (2302 DEGs in HM vs. WT, 1168 DEGs in HM vs. HT, 597 DEGs in HT vs. WT) (Suppl Figure S2A and S2B). The higher number of DEGs in HM mice is consistent with their more severe phenotype. Gene enrichment analysis based on Human Phenotype Ontology (HPO) terms indicated that DEGs were related to cardiac dysfunction including DCM, abnormal LV function, hypertrophic cardiomyopathy, and left atrial enlargement (Figure 2F, Suppl Table S3). These findings are consistent with the above described gross anatomical, histological, and echocardiographic results. DEGs involved in ten cardiac pathology GO terms and three KEGG terms were indicated in the gene network and heatmap (Suppl Figure S2C through S2F, and S3A through S3D). Top 15 up- or down- regulated DEGs were illustrated by gene heatmap (Figure 2G) and eight of them (Ankrd1, Ccn2, Enpp1, Lox, Fah, Slc40a1, Wnt5a, and Ryr1) were validated with qPCR (Figure 2H). The results are consistent with RNA-seq data illustrated in the heatmap (Figure 2G). Many of these genes have previously been linked to cardiac dysfunction [18–26].

RBM20^S639G mutation caused early death in HM mice.

The survival rate was calculated by Kaplan–Meier estimator (Figure 3A). The survival rate in HM mice between one- to two-month-old was about 52% with no significant sex difference. Only 3% of HT mice died within two months, and no deaths occurred in WT mice. Two-month-old HM mice had significantly increased heart to body weight ratio (Figure 3B). Enlarged atria were observed only in HM mice (23.1% of total cases) (Figure 3C). Most interestingly, autopsy of HM mice with premature death revealed that LA/RA contained white and solid fat-like deposition and histological staining demonstrated that all mice had enlarged atria (Figure 3D). We also observed that HT and HM individuals developed abnormal heart rhythm with irregular PR and RR intervals from the ECG recordings during echocardiographic imaging (Figure 3E). In particular, the morphology of P wave suggests aberrant atrial conduction. RNA seq data showed that both natriuretic peptide precursor A (Nppa) and B (Nppb) were upregulated in HM individuals, and this was validated with qPCR (Figure 3F). Natriuretic peptides (ANP and BNP) are considered predictors of AF [27].

Figure 3. Survival curve and transcriptomic profiling.

(A) Kaplan–Meier survival curves of WT (n=42), HT (n=30), and HM (n=31) mice. Statistical significance was determined by log-rank (Mantel–Cox) test with two degrees of freedom (P<0.0001) (n is the number of individual littermates of each genotype). (B) Heart weight vs. body weight of WT (n=32, 17 females, 15 males), HT (n=25,13 females, 12 males), and HM (n=13, 3 females, 10 males) mice at the age of two months. Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used. (C) Enlarged atria were observed in HM mice at the age of two months. The scale bar at the bottom indicates a 0.5 cm ruler. (D) Necropsy and hematoxylin and eosin staining of heart tissue of HM mice with sudden death. A WT mouse with similar age was sacrificed for control. The scale bar at the bottom indicated a 0.5 cm ruler. (E) Representative ECG images from WT, HT and HM mice. (F) Quantitative PCR verification of the heart failure markers Nppa and Nppb (n=3, the number of animals). Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used. (G) Networks of differentially expressed genes (between WT, HT, and HM mice) clustering in 10 arrhythmia related HPO terms. (H) Heatmap of the 15 most up- and down- regulated genes (HM vs. WT) within the 10 selected HPO terms selected in (G). (I) Validation of selected genes with quantitative real-time PCR (n=5, the number of animals). Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used. (J) Networks of differentially expressed genes (between WT, HT, and HM mice) clustering in abnormal atrioventricular conduction-related HPO terms. (K) Heatmap of the 15 most up-and down-regulated genes (HM vs. WT) within the 6 selected HPO terms selected from (K). (L) Validation of selected genes with quantitative real-time PCR (n=5, the number of animals). Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Tukey’s multiple comparisons test was used. WT, wild type; HT, heterozygotes; HM, homozygotes.

Further, gene enrichment analysis categorized the DEGs into ten arrhythmia related HPO terms and six abnormal conduction HPO terms representing 123 and 58 DEGs, respectively (Figure 3G, S4A, S4B, S4C, and 3J, S4D, S4E, S4F). The top 15 up- or down-regulated DEGs from these GO terms were clustered in gene heatmaps (Figure 3H and 3K). These data further support the observations of early premature death that is associated with malignant arrhythmia in KI mice. Genes selected from the top 15 up- or- down-regulated DEGs were validated through qPCR (Figure 3I and 3L).

RBM20^S639G mutation facilitates trafficking of RBM20 from the nucleus to the sarcoplasm with formation of protein condensates in the sarcoplasm.

Previous studies, including ours, have shown that a single amino acid substitution in the RS rich domain of RBM20 causes severe DCM [1–3]. However, it is unclear how a single amino acid substitution results in cardiac remodeling and pathogenesis. In neurodegenerative disease, mutated RNA binding proteins can assemble into ribonucleoprotein granules (RNP) or RNA-protein condensates [28, 29], which is the primary cause for pathogenesis. To test if this might also occur in heart muscle disease, immunofluorescent staining in both paraffin-embedded tissues and in isolated single cardiomyocyte was performed. This demonstrated formation of protein condensates at the nuclear periphery or nuclear membrane, whereas unmutated RBM20 was constrained in the nucleus where it forms protein condensates providing evidence that the S639G amino acid substitution facilitates RBM20 trafficking from the nucleus to the sarcoplasm (Figure 4A and 4B, Suppl Figure S9, Suppl Video 1 and 2). The RNP is assembled from mRNA transcripts and proteins, most of which are RNA binding proteins (RBPs) or RNA chaperones; a diverse number of intermolecular interactions involving potential RNA–protein, protein–protein, and RNA–RNA interactions play a key role in RNP congregation [30, 31], trapping both RNAs and RBPs in the cytoplasm. The GO enrichment analysis of DEGs (WT vs. HT vs. HM) for both Molecular Function and Biological Process from RNA-Seq data validated that GO terms associated with protein binding and regulation of localization have relatively lower P-values compared to other GO terms (Figure 4C, Suppl Tables S4 and S5). Besides, differentially alternative spliced (DAS) genes were mostly clustered in the GO terms (cellular component and biological process) related to intracellular organelle, organelle and cellular component organization (Suppl Figure S5). Taken together, these data suggest that the RBM20^S639G mutation may create the cellular and molecular microenvironment for RNA-protein assembly and condensation.

Figure 4. RBM20 trafficking, formation of RNA-protein condensates and GO analysis.

(A) Immunohistochemistry staining with paraffin section of WT and HM mouse hearts. Scale bar, 20μm. (B) Immunocytochemical staining of isolated WT and HT cardiomyocytes. Scale bar, 20μm. (C) GO enrichment analysis of DEGs (WT vs. HT vs. HM). Protein binding and localization relevant GO terms with significant P-values were shown. Numbers indicate the intersection size (number of differentially expressed genes) of each term. (D) Differentially expressed genes encoding stress granule components. Log2 fold change ≥ 0.5 and ≤ −0.5 were shown in the bar chart. (E) Differentially expressed genes encoding components involving translational regulation (Biological Process, term ID: 0006417). WT, wild type; HT, heterozygotes; HM, homozygotes; GO, gene ontology; F.C., fold change.

Considering that the formation of RNA-protein condensates is likely to disturb the process of mRNA translation, we examined the expression of genes that are involved in translational regulation and the previously reported components of stress granules [32–38], one category of RNP aggregates. We found that the components of stress granules were dysregulated in RBM20^S639G mice when compared to WT (Figure 4D), and genes associated with translational regulation were also differentially expressed in RBM20^S639G mice (Figure 4E). For instance, eukaryotic translation initiation factor 4E (eIF4E, Eif4e), which plays a critical role in the control of cap-dependent translation initiation [39, 40], was found decreased in brain ischemia induced formation of stress granule [41]. Meanwhile, eIF4E is an RNP component [37, 38] that also regulates the fate of RNP granule formation through multiple mechanisms [42]. Further, Hsp90aa1, encoding the human stress-inducible 90-kDa heat shock protein alpha (Hsp90A), is involved in the formation of RNP granules [43–45]. Both proteins were differentially expressed in the RBM20^S639G mice (Figure 4D and 4E). We further did co-staining of RBM20 and stress granule core proteins, such as G3BP1, TIA-1/TIAR in isolated cardiomyocytes. We did not observe the colocalization of RBM20 and stress granule markers. However, induced stress granules in sodium arsenite-treated RBM20^R636S mutation-transfected U2OS cells colocalized mutated RBM20 and stress granule markers (9), suggesting the RBM20 mutation-induced RNP granules do not completely share the same components within the stress granules (Suppl Figure S7).

RBM20^S639G mutation regulates mRNA posttranscriptional process.

We have reported that loss-of-function mutations in RBM20 (depletion of RBM20 in cardiomyocytes) in rats results in the expression of larger titin isoforms, due to splicing deficiency of the titin (TTN) gene [2]. Re-localization of RBM20 at the nuclear periphery caused by Rbm20 mutation reduces the level of RBM20 in the nucleus, suggesting that Rbm20^S639G mutation could interfere with the mRNA posttranscriptional process, similarly to RBM20 depletion. To test this, we studied titin isoform expression using agarose gel electrophoresis. As expected, we observed large titin isoforms in HT and HM hearts similar to RBM20 KO hearts (Figure 5A). Further, splicing analysis with RNA-seq data revealed that 145 genes were differentially spliced in HT and 162 genes in HM mutated hearts (Suppl Figure S6A and Tables S6, S7). The volcano and violin plots showed the pattern of the differentially alternative spliced (DAS) genes in HT or HM with five alternative splicing events (SE, skipped exon; A5SS, alternative 5’ splice site; A3SS, alternative 3’ splice site; MXE, mutually exclusive exons; RI, retained intron) (Figure 5B and 5C). The SE event was the most frequent event of DAS genes (Figure 5B and 5C). In addition to TTN, two alternatively spliced genes Ryr2 and Camk2d in RBM20 KO rats and mice were also found differentially spliced in both HT and HM mutated hearts (Figure 5D, Suppl Figure S8, Suppl Tables S6, S7). In addition to those miss-spliced genes observed in RBM20 KO hearts, splice mismatches of additional genes were observed in Rbm20 mutations (Suppl Figure S6A and Tables S6, S7). Further, motif analysis showed that all 145 alternatively spliced HT genes and all 162 alternatively spliced HM genes contain potential RBM20 (“TCTT”/”UCUU”) motifs, indicating interaction of RBM20 with the alternatively spliced genes (Suppl Tables S6, S7). Introns adjacent to the alternatively spliced exon (“intron upstream”, ”intron downstream”) can be found in Table S9 by their isoform identifiers with positional information about RBM20 binding motifs. Six of these genes (Phldb1, Tpm2, Ambra1, Epb41, Mtg2 and Ppox) were validated by RT-PCR, which is consistent with the RNA-seq data (Figure 5D, Suppl Figure S8). HPO terms and GO terms of Cellular Component (CC) and Molecular Function (MF), and KEGG pathways of the DAS genes were summarized by Venn diagrams (Suppl Figure S6B through S6E). Gene enrichment analysis based on HPO indicated that DAS genes (HM vs. WT) were associated with atrial arrhythmia (P = 9.46×10⁻⁵), atrial fibrillation (P = 9.46×10⁻⁵), tachycardia (P = 0.016), and sudden death (P = 0.021) (Figure 5E and Suppl Table S8). The five alternative splicing events of the DAS genes representing the top 15 HP terms are shown here (Figure 5E).

Figure 5. Splicing variant analysis with RNA seq data and splicing events involving cardiac function.

(A) Titin isoform switching resulting from alternative splicing was detected in WT, HT, and HM heart tissues through vertical sodium dodecyl sulfate (SDS)-1% agarose gel electrophoresis (VAGE) system. N2BA-G, N2BA, and N2B represent different sizes of titin isoforms. (B) Volcano and violin plots of splicing events of the differentially alternative spliced genes in HT hearts. (C) Volcano and violin plots of splicing events of the differentially alternative spliced genes in HM hearts. (D) Validation of selected alternative spliced genes with RT-PCR and schematic diagram of selected genes with different splicing events. (E) DAS genes falling into the top 15 HPO terms related to cardiac function, and splicing events of selected genes are presented. The numbers indicated the intersection size of each term. (HM vs. WT FDR ≤ 0.05). GAPDH, loading control; SE, skipped exon; A5SS, alternative 5’ splice site; A3SS, alternative 3’ splice site; MXE, mutually exclusive exons; RI, retained intron; DAS, differentially spliced genes; WT, wild type; HT, heterozygotes; HM, homozygotes; bp, base pair; V, variant; T2, degraded titin.

RBM20^S639G mutation reduced cardiomyocyte diastolic stiffness and impaired cardiomyocyte contractility.

Titin plays a major role in diastolic function and ventricular filling [46]. Size switching of titin is one of the primary mechanisms modifying the elastic properties of titin and altering cardiomyocyte function [47–50]. To determine how RBM20^S639G mutation alters cardiac function at the cellular level, we performed experiments in loaded and unloaded intact cardiomyocytes. Due to the higher mortality rate in HM mice, only HT mice were used in this study. In the loaded intact myocyte study, the cell was glued to the force transducer and the length controller. Single cell force measurements were performed using a work loop protocol, which is the cellular equivalent of pressure-volume (PV) loop analysis at the LV level [51]. The slopes of ED-SSLR (end diastolic stress sarcomere length relation) and ES-SSLR (end systolic stress-sarcomere length relation) which reflect the cellular diastolic stiffness and systolic contractility, respectively, were obtained (Figure 6A, 6B and 6C). Representative cellular work loops are shown in Figure 6A. Intact cardiomyocytes from HT had decreased ED-SSLR and ES-SSLR slopes, compared to WT myocytes (Figure 6B and 6C). The maximal rate of stress rise (dS/dt_max) during isometric contraction is reduced in HT (Figure 6D), but the maximal rate of stress decline (dS/dt_min) is not different from WT (Figure 6E). These results indicate reduced cellular diastolic stiffness due to expression of large titin isoforms, and systolic contractility in HT cardiomyocytes. In the unloaded intact myocyte, sarcomere length (SL) shortening and re-lengthening were evaluated. The baseline SL of HT myocytes were observed to be shorter than WT (Figure 6F). In contrast, the shortening amplitude was larger in HT (Figure 6G). Additionally, the time to peak shortening was prolonged while the time from peak to 50% baseline was unaltered (Figure 6H and 6I), consistent with the reduced dS/dt_max and normal dS/dt_min observed in loaded cells (Figure 6D and 6E). Ca²⁺ transients were measured with Fura-2. There was no significant difference in diastolic baseline signal (Figure 6J), while the transient amplitude was larger in HT (Figure 6K). The time to peak transient was prolonged in HT mice (Figure 6L), while the time from peak to 50% decay was unaltered (Figure 6M). The delay in Ca²⁺ release contributed to the reduced contraction velocity (Figure 6H and 6L). Overall, the results demonstrate changes in Ca²⁺ release kinetics and contractile dysfunction of HT myocytes.

Figure 6. Cardiomyocyte diastolic stiffness, contractility and calcium transient.

Cardiomyocytes were stimulated at 4 Hz in the solutions at physiological temperature (37° C). (A)-(E) Loaded intact myocyte force measurement. WT, n= 5; HT, n= 7 mice, and 4–10 cells per mouse. (A) Representative examples of work loop series of WT and HT cells. (B) The slope of ED-SSLR, reflecting cellular diastolic stiffness. (C) The slope of ES-SSLR, reflecting cellular contractility. (D) The maximal rate of stress rises during isometric contraction. (E) The maximal rate of stress decline. (F)-(I) Unloaded intact myocyte SL shortening-relengthening. n= 5 mice, and ~50 cells per mouse. (F) The diastolic (baseline) sarcomere length of WT and HT cardiomyocytes. (G) The shortening amplitude of WT and HT cardiomyocytes. (H) The time to peak shortening. (I) The time from peak to 50% baseline. (J)-(M) Unloaded intact myocyte Ca²⁺ release-reuptake. Ca²⁺ transient of isolated intact cardiomyocytes was measured with Fura-2 340/380 signal. (n= 5 mice, and ~50 cells per mouse). (J) Diastolic baseline signal. (K) The transient amplitude. (L) The time to peak transient. (M) The time from peak to 50% transient decay. WT, wild type; HT, heterozygotes; ED-SSLR, end diastolic-stress sarcomere length relation; ES-SSLR end systolic-stress sarcomere length; SL, sarcomere length. Error bars indicate mean ± SEM and statistical significance is indicated as *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Nested t-test was used.

Discussion

Our findings show that RBM20^S639G mutation results in severe DCM, atrial dysfunction, arrythmia and a high mortality rate in young mice. In addition, RBM20 S639G mutation promotes protein-RNA condensates in the sarcoplasm. These findings are consistent with observations in human genetic studies showing human subjects carrying RBM20 mutations suffer from severe heart failure and early death [1, 3, 11–17]. There are over 900 RBM20 mutations listed in the NCBI ClinVar database. Recently, two mutations (R636S and S637A) located in the RS domain have been validated in pig and mouse model respectively [9, 10]. Both gene-edited mutant KI pigs (RBM20^R636S) and mice (RBM20^S637A) display an accumulation of stress granule-like structures in the sarcoplasm. Homozygous piglets are more severely affected than heterozygous piglets. All neonatal homozygous piglets develop a moderate-to-severe DCM phenotype with poor growth and low survival rate [9]. Rbm20^S637A mice also exhibit severe DCM phenotypes with a high prevalence of atrial fibrillation and ventricular tachycardia and an increased early death rate, mimicking the clinical observations in human patients carrying S637A mutation [10]. In the present study, we selected the S639G mutation that is also located in the RS domain. The Rbm20^S639G KI mouse model demonstrated a phenotype similar to the RBM20^R636S pig and RBM20^S637A mouse models. Mice homozygous for RBM20^S639G mutation exhibit ~50% mortality by ~2 months of age, which is even worse than that in Rbm20^S637A mice. Our results further establish that mutations in the RS domain of RBM20 contribute to RBM20 cardiomyopathy. Mutations outside the RS domain may also cause RBM20 cardiomyopathies, whether this is indeed the case needs to be addressed in future studies.

A decade ago, RBM20 was identified as a regulator of the TTN gene alternative splicing, encoding the giant sarcomeric protein titin that is responsible for ventricular wall stiffness [2, 46] and 30 other genes including calcium handling genes [4, 6]. Subsequent studies using RBM20 KO rats and mice revealed that RBM20 loss-of-function leads to reduced diastolic stiffness, DCM phenotype, arrhythmia and sudden death [2, 5–7]. Therefore, it is widely accepted that RBM20 cardiomyopathy is caused by altered splicing variants of TTN and 30 more genes. Recent studies in RBM20 mutation KI animal models, including ours, indicated that RBM20 mutations lead to a cardiac phenotype similar to, but more severe than RBM20 KO, suggesting RBM20 mutations result in a gain-of-function. The new function gained from RBM20 mutations is associated with RBM20 re-localization and stress granule-like structure formation in the cytoplasm. Why do RBM20 mutations regulate splicing variants of the same genes as in RBM20 KO? RBM20 is a nuclear protein and when RBM20 is knocked out it would eliminate RBM20 from the entire cell, including the nucleus, resulting in abnormal posttranscriptional regulation. However, RBM20 mutations may facilitate RBM20 shuttling from the nucleus to the sarcoplasm, which depletes RBM20 from the nucleus and causes an accumulation in the sarcoplasm. This may explain why RBM20 mutated animals exhibit a similar cardiac phenotype as the RBM20 KO animals. Furthermore, protein condensates in the sarcoplasm promoted by RBM20 mutations may interrupt the translational process by trapping RNAs and proteins in the stress granule-like structure, thereby changing the RNA concentration and intermolecular RNA-RNA interactions [31]. In addition, RNP granules are enriched with RNA binding proteins (RNPs) that may not be released to facilitate normal transcriptional and posttranscriptional regulation. Thus, these trapped RNPs interfere with these important biological processes [31, 52]. The alterations in transcriptional and post-transcriptional regulation may also explain why we found so many additional differentially expressed and spliced genes in hearts from mice with RBM20 mutations, and a severe phenotype such as atrial enlargement and high mortality at young age, compared to Rbm20 KO rodent models.

Lastly, our results with RBM20^S639G KI mice and recently published data with RBM20^R636S pigs and RBM20^S637A mice suggest that RBM20 mutations place patients at high risk for severe cardiac dysfunction and premature death. RNP granules caused by mutations in RNA binding proteins (RBPs) have been extensively studied in neurodegenerative diseases [53]. Mutations in the TARDBP gene, encoding the RNA binding protein TDP43 promote RNP granules in the neurons and result in neurodegenerative diseases [54, 55]. In this regard, RBM20 cardiomyopathy caused by RBM20 mutations might be analogous to RNP granule disease. Future studies on mechanisms of RBM20 trafficking and the formation of RNA-protein condensates in the sarcoplasm are warranted. Answers to these questions will allow us to develop therapeutic strategies to prevent or reverse DCM caused by RBM20 mutations. In addition, posttranslational modifications (PTMs) in the RS domain of RBPs are critical for posttranscriptional regulation and nucleocytoplasmic transportation [56, 57]. Our previous mass spectrometry study demonstrated that RBM20 is highly phosphorylated in the RS domain [58]. Hence, future studies should also address the role of PTMs in cardiac pathogenesis induced by RBM20 mutations.

Methods

Generation of Rbm20^S639G mutation knock-in mice and tissue sample preparation

The Rbm20^S639G mouse (BL6J) was generated via CRISP/Cas9 system at the Genetically Engineered Mouse Models Core Facility of the University of Arizona using the CRISPR/Cas9 system. The CRISPR guide RNAs (gRNAs), used for mouse Rbm20^S639G knock-in were designed using the CRISPOR web tool (http://crispor.tefor.net/). The gRNAs CTCATTGGACTTCGAGAACGTGG (gRNA1) and CTTCGAGAACGTGGCCGCTCTGG (gRNA2) (PAM sequence was underlined) were designed for the CRISPR knock-in experiments. Oligos with 40–50 bp homology to sequences on each side of each gRNA-mediated double-stranded break were designed. The mixture of Cas9 protein with sgRNA complexes and oligo were injected into the pronucleus of 1-cell zygotes from the fertilized eggs collected from the oviducts of super ovulated BL6/NJ females. The genetically edited mice were identified by PCR genotyping, as well as Sanger sequencing. Two screening primers: forward, 5’- TGTGTGTCTCCATCTGGGTG and reverse, 5’- CTTTGTCCAGCTGCCTAGGA, producing 307 bp band for the wild type and two additional bands of 207bp and 100bp in the positive mice (heterozygous, HT) when restricted with XhoI.

Animals were maintained on standard rodent chow. Hearts were removed immediately after euthanasia and the left ventricle, right ventricle, and atria were separated. Samples were collected from 8-week-old mice. Tissues were separated for paraffin-embedding and snap-frozen in liquid nitrogen and stored in a −80°C freezer until use. All animal experiments were in accordance with the guidelines of the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Wisconsin-Madison (#A006262) and the University of Arizona (#09–056).

Histological assessments

Whole hearts from two-months-old mice were isolated and fixed in 4% paraformaldehyde, paraffin-embedded, and stained with Hematoxylin and Eosin (H&E) (n=3), Masson’s trichrome. Stained sections were photographed (BZ-X800, Keyence), and the fibrotic area was quantified by using ImageJ software (Fiji) [59]. The proportion of fibrotic area was calculated as a ratio of the fibrotic area to the total cross-sectional area.

Immunochemistry staining and cell imaging

Isolated cardiomyocytes were fixed with methanol for 15 min on ice and blocking/permeabilized with 10% goat serum (Sigma-Aldrich; Cat#: G6767), 0.2% Triton-X100 (Sigma-Aldrich; Cat#: X100) in TBS for 1 hour at room temperature (RT). Then cells were incubated with primary antibodies, rabbit anti-RBM20, CoraLite®594-conjugated G3BP1 Monoclonal antibody (Cat#: CL594–66486, Proteintech), TIA-1/TIAR (D-9) mouse mAb (Cat#: sc-48371, Santa Cruz Biotechnology), mouse monoclonal anti-alpha-actinin antibody (Cat#: A7811, Sigma-Aldrich), in blocking/permeabilized buffer for 1 hour at RT or overnight at 4°C. After washing, cells were spined down and incubated with Alexa flour-conjugated secondary antibody (Goat anti-rabbit, Alexa Fluor 488, Cat#: A32731, Invitrogen; Goat anti-mouse, Alexa Flour 594, Cat#: 8890, Cell Signaling Technology) for 1 hour at RT, washed with TBST, and mounted in SlowFade™ Gold Antifade Mountant with DAPI (Invitrogen; Cat#S36938). Cells were visualized using Keyence BZ-X800 microscope (Osaka, Japan).

The antibodies and procedure for the immunostaining of the paraffin embedded tissue is the same after deparaffinizing and rehydrating, permeabilization, and epitope retrieval.

Western blotting and subcellular protein fractionation

Total proteins were lysed from heart LV tissues as described previously [8]. Briefly, proteins were separated by SDS-PAGE gel and transferred onto a PVDF membrane. The membrane was probed with primary antibodies against RBM20 (1:1000, rabbit), GAPDH (1:1500, rabbit; Cell Signaling Technology, Danvers, Massachusetts, USA; Cat#2118S) was served as the protein loading control. Then the membrane was probed with horseradish peroxidase-coupled secondary antibodies (1:3000, anti-rabbit; Promega; Cat#W4011) for 1 hour. Chemiluminescence images were taken by ChemiDoc system (Bio-Rad, Hercules, CA).

Subcellular proteins were extracted from LV at two-months-old mice followed the instruction (Thermo Fisher Scientific, Waltham, MA; Cat# 87790). Cytoplasmic extract and nuclear extract were loaded into SDS-PAGE gel and transferred onto PVDF membrane. The membrane was probed with primary antibodies against RBM20, HSP90 (1:1000, rabbit; Cell Signaling Technology, Danvers, Massachusetts, USA; Cat# #4874), GAPDH and Histone H3 (1:1500, rabbit; Cell Signaling Technology, Danvers, Massachusetts, USA; Cat# #4499). Incubation with secondary antibody and taken chemiluminescence images were descripted above.

Echocardiography

To assess cardiac function in vivo, wild type (WT), heterozygous RBM20^S639G mutant (HT), and homozygous RBM20^S639G mutant (HM) male and female mice at the age of two months were used. Echocardiographic analysis has been done in two different laboratories independently. Transthoracic echocardiography was performed using a Visual Sonics Vevo 3100 ultrasonography outfitted with the MX400 transducer, ~30-MHz, as detailed previously [60]. To acquire of two-dimensional guided M-mode images at the tips of papillary muscles and Doppler studies, mice were sedated by facemask administration of 1% isoflurane, hair removed, and maintained on a heated platform. Blood velocities across the mitral, aortic and pulmonary valves were measured using Doppler pulse wave imaging angling the probe to obtain a nearly parallel orientation to the blood flow.

End diastolic and systolic left ventricular (LV) diameter and anterior and posterior wall (AW and PW, respectively) thickness were measured online from M-mode images obtained in a parasternal short-axis view using the leading edge-to-leading edge convention. All parameters were measured over at least three consecutive cardiac cycles and averaged. Left ventricular fractional shortening were calculated as [(LV diameter_diastole −LV diameter_systole)/LV diameter_diastole] × 100; ejection fraction [(7.0/ (2.4 + LV diameter_diastole) (LV diameter_diastole)³ − (7.0/ (2.4 + LV diameter_systole) (LV diameter_systole)³/ (7.0/ (2.4 + LV diameter_diastole) (LV diameter_diastole)³ × 100 and LV mass was calculated by using the formula [1.05 × ((Posterior Wall _diastole+Anterior Wall_diastole+LV diameter_diastole)³ − (LV diameter_diastole)³)]. Right ventricular wall thicknesses were measured as above from M-mode images obtained in a right parasternal long-axis window. Pulmonary artery (PA) diameter was measured just distal to the pulmonary valve. Care was taken to measure PA diameter during the widest part of the ejection cycle. The aortic diameter was measured just distal to the aortic valve. Heart rate was determined from at least three consecutive intervals from the pulse wave Doppler tracings of the LV outflow tract. Ejection time was measured from the same outflow track tracings from the onset of flow to the end of flow. Isovolumic relaxation time was measured as the time from the closing of the aortic valve to the opening of the mitral valve from pulse wave Doppler tracings of the LV outflow tract and mitral inflow region. All images and measures were obtained by the same observer.

Mice were anesthetized under 1–3% isoflurane (USP, Phoenix) in oxygen mixture, then placed in dorsal recumbence on a heated platform for echocardiography. Transthoracic echo images were obtained with a Vevo 3100 High Resolution Imaging System (Visual-Sonics, Toronto, Canada) using the model Mx550D scan head designed for murine cardiac imaging. Body temperature was maintained at 37°C. Imaging was performed at a depth setting of 1 cm. Images were collected and stored as a digital cine loop for off-line calculations. Standard imaging planes, M-mode, Doppler, and functional calculations were obtained according to American Society of Echocardiography guidelines. The parasternal long-axis view and mid-wall cross sectional view of the left ventricle (LV) were used to calculate percentage fractional shortening, percentage ejection fraction, and ventricular dimensions and volumes. In addition, the left atrial dimension was measured in the short-axis view directly below the aortic valve leaflets. LV filling parameters (E and A wave, E’ and A’, and deceleration time) were acquired from the images of mitral valve Doppler flow or mitral annulus tissue Doppler from an apical view. A sweep speed of 100 mm/sec was used for M-mode and Doppler studies. The heart rate of animals during echocardiographic study were maintained in the range of 500 −550 beats/min for M-mode, 500– 550 beats/min for B-mode and 400 to 500 beats/min for Doppler studies.

Conscious echocardiography was performed in animals restrained through scruffing the skin at the nape of the neck. Short axis M-mode cine loops at the level of the mid-papillary were recorded for evaluation of morphometry and systolic LV function, as described in the anesthetized echocardiography.

Pressure-Volume analysis

In vivo pressure volume analysis was performed using a SciSense Advantage Admittance Derived Volume Measurement System and 1.2F catheters with 4.5 mm electrode spacing (SciSense, London, Ontario, Canada). Mice were anesthetized and ventilated with 1–3% isoflurane using an SAR-1000 Ventilator (CWE Inc) and body temperature maintained at 37°C using a MouseMonitor S platform (Indus Instruments). Mice were secured and the abdomen was opened below the sternum. The apex of the LV was punctured using a 28G needle and the catheter was advanced into the LV. The IVC was located and occluded during a pause in ventilation to acquire load-independent indexes. Data acquisition and analysis was performed in LabScribe3 (iWorx, Dover NH). The end-systolic pressure-volume relationship was fit to a linear equation (P = E_es∙V + b). The end-diastolic pressure-volume relationship was analyzed using a monoexponential fit (P = Ae^βV+C) with the exponent (β) reported as the stiffness.

Adult mouse intact cardiomyocyte isolation

Cardiomyocytes were isolated as described previously [61]. Briefly, mice were anesthetized under isoflurane. The heart was removed and cannulated via the aorta with a blunted 21-gauge needle for antegrade coronary perfusion. The heart was perfused for 4 min with perfusion buffer ([in mmol/L] 90 NaCl, 34.7 KCl, 0.6 KH2PO4,0.6 Na2HPO4, 1.2 MgSO4, 12 NaHCO3, 10 KHCO3, 10 HEPES, 10 taurine, 5.5 glucose, 5 BDM, 20 creatine, pH 7.4), followed by digestion buffer (perfusion buffer plus 0.05 mg/ml Liberase TM research grade; Roche Applied Science, and 13 μM CaCl2) for 20 min. Then, the heart was placed in myocyte stopping buffer (perfusion buffer plus bovine calf serum 0.08 [BCS]/ml and 8 μM CaCl2) with protease inhibitors ([in mmol/L] 0.4 Leupeptin, 0.1 E64, and 0.5 PMSF (Peptides International, Sigma-Aldrich)). The left ventricle was cut into small pieces and triturated several times with a transfer pipette and then filtered through a 100 μm nylon mesh filter. Then Ca²⁺ was reintroduced to cardiomyocyte suspension to a final concentration of 1 mM.

Cellular work loop force measurement

All intact cell experiments were performed at temperature 37°C in Medium199 (M5017, Sigma-Aldrich) plus 10 μg/mL insulin (I9278, Sigma-Aldrich). Cells were field stimulated at 4 Hz frequency by MyoPacer stimulator (IonOptix Co, MA). An inverted microscope (IX-70; Olympus) was used with a chamber that had platinum electrodes to electrically stimulate cells and a perfusion line with heater control and suction out to maintain a flow rate of 2 ml/min. All images were recorded with 40X objective lens. Data were collected using an IonOptix FSI A/D board and IonWizard 6.3 software (IonOptix Co, MA) with SarcLen and SoftEdge modules to determine sarcomere length (SL) and cell width. The sampling frequency of the system was sufficient to measure force (1,000 Hz) and SL (250 Hz) simultaneously. The myocyte was glued with Myotak (IonOptix LLC) at one end to a glass rod that attached to the force transducer (OFT200, OptiForce transducer, IonOptix LLC) and the other end to a glass rod that attached to the piezo translator (Mad City Lab). Cross sectional area of intact cell was obtained from the measured cell width assuming that the cross section of the cell was an ellipse [62]. All forces were normalized to stress.

The cell work loop algorithm was applied through the interface box of the IonOptix system which contains a field programable gate array (FPGA) that digitizes the force signal, and the preload and after load value programmed through the IonWizard software 6.3, as described in [63]. The output from the FPGA drove the piezo translator to adjust cell length through a feedback control system based on developed force, implementing preload and afterload. The system constructs force-SL loops by modulating cell length, generating 4 phases that mimic isovolumic contraction, systolic ejection, isovolumic relaxation and diastolic filling phases of the LV chamber²⁰. Data analysis was performed in LabScribe3 (iWorx, Dover, NH). The ED-SSLR (end diastolic stress-sarcomere length relation) and ES-SSLR (the end systolic stress-sarcomere length relation) were fit with linear relation.

Unloaded intact cardiomyocytes

Measurement of Ca²⁺ release-reuptake and SL shortening-relengthening in unloaded intact cardiomyocytes was performed as follows. Isolated LV cardiac myocytes were incubated with 1 uM Fura-2 AM (F-1225, Life Technologies). Fluorescence was measured ratiometrically with the IonOptix photometry MultiCell High Throughput system (IonOptix Co, MA). Fura-2 was excited alternately at 340 and 380 nm, and emission was recorded at 510 nm. Background fluorescence was subtracted for each excitation wavelength. Ratio of fluorescence intensities excited at 340 nm and 380 nm was used as a relative measurement of cytoplasmic Ca²⁺, the ratio transient was fitted by monotonic transient analysis software (IonWizard Transient Analysis Tool 1.2.97.0). All measurements were carried out at 37°C. The SL shortening-relengthening was measured in cells that were not loaded with Fura-2.

Titin gel electrophoresis

Titin isoforms were resolved by a previously published method using a vertical sodium dodecyl sulfate (SDS)-1% agarose gel electrophoresis (VAGE) system [64, 65]. The frozen left ventricle samples from 2-month-old mice were homogenized in urea-thiourea-SDS-dithiothreitol sample buffer using a small Dounce homogenizer or Geno/Grinder (SPEX SamplePrep, Metuchen, NJ) at 1500 strokes/min for 1 min, repeat for 5–7cycles and were heated at 65°C for 10 min. The denatured protein samples were loaded into 1% SDS-agarose gel, which is run at 30 mA constant current for 3.5 hours. The agarose gel was fixed in 50% methanol,12% glacial acetic acid, and 5% w/v glycerol for 1 hour and then dried overnight at 37°C. The dried gel was scanned by the silver staining method. Full details have been discussed in a previous published report [64, 65].

RNA preparation and RNA-Seq

Total RNA was extracted from left ventricles of male mice at the age of two months (n=3) using TRIzol Reagent (Life Technologies, Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s protocol. DNase I (#D9905K, Lucigen, Middleton, WI) treatment was applied to clean up genomic DNA before RNA-seq. The quantity and quality were determined by Nanodrop One (Thermo Fisher Scientific, Waltham, MA) and electrophoresis. The RNA-seq procedure was performed at the core facility of University of Wisconsin-Madison. A quality check of the raw sequencing data was performed using FastQC software [66]. Low quality reads and adapter sequences were trimmed using Trimmomatic [67]. Trimmed reads were then aligned to the mouse reference genome (NCBI Mus musculus GRCm39) using STAR [68]. The full dataset was uploaded to NCBI database (GSE180216, https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE180216)

Differential expression and alternative splicing analyses

For gene-level expression, gene counts were estimated using “--quantMode GeneCounts” option in STAR [68]. The R package edgeR [69] was used to normalize gene counting based on trimmed mean of M-values (TMM) method [70]. Only expressed genes with at least 15 counts in more than three samples were evaluated, resulting in 14,648 genes for further analysis. Differential expression analysis were carried out for all three pairwise comparisons between WT vs. HT, WT vs. HM, HT vs. HM (three samples per group) based on negative binomial generalized linear model using edgeR package [69]. The statistical tests were corrected for multiple testing and only genes with a false discovery rate (FDR) less than 0.05 were considered significant [71]. Heatmaps were plotted using ‘qplots’ R package [72] and gene networks were generated using Cytoscape [73].

Alternative splicing analysis was performed for the three pairwise comparisons, i.e. WT vs. HT, WT vs. HM, and HT vs. WT, using Multivariate Analysis of Transcript Splicing software for replicates, rMATS [74]. Five types of alternative splicing events were analyzed, including skipped exon (SE), alternative 5’ splice site (A5SS), alternative 3’ splice site (A3SS), mutually exclusive exons (MXE), and retained intron (RI). The output from rMATS was filtered using FDR ≤ 0.05 and absolute value of ΔPSI ≥ 10% as the cut-off criterion to identify significant alternative splicing events. Splicing gene figures were generated based on ‘plotGenes’ function from the ‘Sushi’ R package functions [75].

Gene functional annotation

Functional enrichment analysis using differential expressed and spliced genes were performed using g:Profiler [76], separately, based on Gene Ontology (Biological Process, Molecular Function, and Cellular Component), KEGG pathways and Human Phenotype Ontology (HPO). Only enriched terms with FDR more than 5% were considered significant.

Reverse transcription and quantitative real-time PCR for RNA-Seq validation

The cDNA was synthesized by iScript™ Reverse Transcription Supermix (#1708841, Bio-Rad, Hercules, CA) according to the manufacturer’s protocol. Real-time PCR was performed in a 10ul reaction, 384-well format and SsoAdvanced™ Universal SYBR® Green Supermix (#1725272, Bio-Rad). The reaction was incubated in a CFX384™ Real-time system (Bio-Rad) for polymerase activation at 95°C for 2 min, and then 40 cycles consisting of denaturation at 95°C for 15 s and annealing/extension at 60°C for 30 s; finally, the melt curve analysis was performed as default. Five samples were measured in each experimental group in duplicate, with a minimum of two independent experiments using primers listed in Table S1. The relative amount of target mRNA normalized to GAPDH was calculated according to the 2 ^−ΔΔCt method [77].

Motif analysis

Intronic coordinates for all annotated isoforms from alternatively spliced genes were extracted using R’s “GenomicFeatures” package [78]. Intronic sequences were retrieved from a database generated with “BSgenome” (genome version GRCm39/NCBI) [79]. All transcript isoforms containing alternatively spliced exons were extracted by matching exon coordinates. Exon positions were used to determine the closest upstream- and downstream introns. Positions of “TCTT” motifs were determined by pattern matching and the number of matches was counted separately for each intron. Genes encoding at least one isoform with “TCTT” introns were counted as potentially regulated by RBM20.

Statistical Analysis

GraphPad Prism software was used for statistical analysis. Results were expressed as means ± SEM. Statistical significance between groups was determined using one way ANOVA with Tukey’s multiple comparisons test, or Two-way ANOVA. A nested t-test was used for intact cardiomyocyte experiments. The significance levels were *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Highlights:

• An RBM20 mutation S639G leads to dilated cardiomyopathy and high mortality rate.

• Mutation results in RBM20 nucleocytoplasmic transport and protein condensates.

• Protein condensates caused by RBM20 mutation are a new causative mechanism of DCM.

Non-standard Abbreviations and Acronyms

ANP

atrial natriuretic peptide

BNP

brain natriuretic peptide

DCM

dilated cardiomyopathy

FS

fractional shortening

FFI

fura-2 fluorescence intensity

RBP

RNA binding protein

RBM20

RNA binding motif protein 20

RMA

robust multi-array analysis

RNP

ribonucleoprotein

LV

left ventricle

LVDd

left ventricular end diastolic dimension

LVDs

left ventricular end systolic dimension

LVPWd

left ventricular end diastolic posterior wall dimension

LVPWs

left ventricular end systolic posterior wall dimension

GO

Gene Ontology

SDS-VAGE

sodium dodecyl sulfate vertical agarose gel electrophoresis

WT

wild type

HT

heterozygous mutation

HM

homozygous mutation

H&E staining

Hematoxylin and eosin staining

RS domain

arginine and serine-rich domain

ED-SSLR

end diastolic-stress sarcomere length relation

ES-SSLR

end systolic-stress sarcomere length

SL

sarcomere length