Splicing Factor RBM20 Regulates Transcriptional Network of Titin Associated and Calcium Handling Genes in The Heart

Abstract

RNA binding motif 20 (RBM20) regulates pre-mRNA splicing of over thirty genes, among which titin is a major target. With RBM20 expression, titin expresses a larger isoform at fetal stage to a smaller isoform at adult resulting from alternative splicing, while, without RBM20, titin expresses exclusively a larger isoform throughout all ages. In addition to splicing regulation, it is unknown whether RBM20 also regulates gene expression. In this study, we employed Rbm20 knockout rats to investigate gene expression profile using Affymetrix expression array. We compared wild type to Rbm20 knockout at day1, 20 and 49. Bioinformatics analysis showed RBM20 regulates fewer genes expression at younger age and more at older age and commonly expressed genes have the same trends. GSEA indicated up-regulated genes are associated with heart failure. We examined titin binding partners. All titin direct binding partners are up-regulated and their increased expression is associated with dilated cardiomyopathy. Particularly, we found that genes involving calcium handling and muscle contraction are changed by RBM20. Intracellular calcium level measurement with individual cardiomyocytes further confirmed that changes of these proteins impact calcium handling. Selected genes from titin binding partners and calcium handling were validated with QPCR and western blotting. These data demonstrate that RBM20 regulates gene splicing as well as gene expression. Altered gene expression by RBM20 influences protein-protein interaction, calcium releasing and thus muscle contraction. Our results first reported gene expression impacted by RBM20 with heart maturation, and provided new insights into the role of RBM20 in the progression of heart failure.

Introduction

Pre-mRNA splicing is an important biological process that contributes to protein diversity in eukaryotic organisms ¹. This process is determined by RNA-binding splicing factors to produce protein isoforms in a tissue-specific and developmental-regulated manner ². Tissue-specific alternative splicing is usually regulated by tissue-specific splicing factors or a combination with ubiquitously expressed splicing factors to influence spliceosome assembly at splice sites ³. Therefore, identification of tissue-specific splicing factors and their targets is essential to decipher the mechanism of pre-mRNA splicing. In muscle tissues, only a few splicing factors have been reported ¹, and RNA binding motif 20 (RBM20) is one of the muscle tissue specific splicing factors that has been identified recently ⁴. Protein RBM20 contains two highly conserved functional domains, an RNA-recognition domain 1 (RRM1) and a serine (S)/arginine (R) rich domain (SR-domain) ⁵. These structural features defined it as a serine/arginine (SR) protein that is essential for assembly in the spliceosome of pre-mRNA splicing ^{6, 7}. RBM20 is highly expressed in striated muscles, with the highest expression levels in cardiac muscle ⁴. Rbm20 deficiency in animal models leads to dilated cardiomyopathy, a major cause of heart failure and a significant source of mortality and morbidity worldwide ^{4, 8}, and mutations in RBM20 have been identified in human patients with familial dilated cardiomyopathy ^{4, 5, 9, 10, 11, 12, 13}. Given the role of RBM20 in the regulation of gene splicing, efforts toward the identification of its RNA targets have been made ^{4, 14} with a goal that these findings would shed light on human heart failure progression and provide strategies for molecular therapy. With these efforts, more than thirty RNA targets of RBM20 have been identified with a major target being titin ^{4, 14}.

Titin is a giant multifunctional protein in the sarcomere that plays a critical role in elastic recoil of cardiac myocytes and contributes to passive stiffness of the myocardium ^{15-16, 17-18}. It has been increasingly recognized as a major disease-causing gene ^{16, 19}. Titin runs through distinct sarcomere regions: the Z-band, the I-band, the A-band and the M-band regions ^{20, 21}. Over twenty proteins have been found directly bound to these regions of titin, and over fifty proteins are indirectly bound to titin, linking it to multiple potential mechanic signaling pathways ^{22, 23}. Cardiac titin is encoded by a single gene, and it has two major classes of isoforms resulting from alternative splicing. These isoforms have been termed “N2B” and “N2BA” ²⁴. The ratio changes of these two isoforms are a major determinant for diastolic passive stiffness during left ventricular filling ^{15-18, 25}. Aberrant ratio changes of titin isoforms have been found in both systolic and diastolic heart failure ^{1, 26-32}. Therefore, the role of RBM20 in the regulation of titin alternative splicing may be critical for correction of abnormal titin isoform changes, and thereby, the treatment of diastolic dysfunction. In addition to its major role of the regulation in titin splicing, it is unknown whether RBM20 also regulates gene expression profiles in the heart. In this study, we compared the gene expression profiles of wild type and the Rbm20 knockout rats, and demonstrated that calcium handling channel proteins are affected and regulated by RBM20.

Materials and methods

Animals and Sample preparation

Sprague-Dawley strain rats were crossed with Fisher 344 inbreeds. Both strains were obtained from Harlan Sprague Dawley, Indianapolis, IN. Animals were maintained on standard rodent chow using protocols approved by the University of Wisconsin-Madison and University of Wyoming Animal Use and Care Committees. Hearts were removed immediately after euthanasia and the left ventricle, right ventricle, and atria were separated. Samples were obtained from animals ranging from 16 days fetuses (E16) to 1 year after birth. Tissues were snap-frozen in liquid nitrogen and stored in -80°C freezer prior to gel sample and RNA preparation. All samples for protein gel electrophoresis were prepared as described previously ³³.

Gel electrophoresis

Titin isoforms were resolved using a previously published method using a vertical sodium dodecyl sulfate (SDS)-1% agarose gel electrophoresis (VAGE) system ³³. The frozen tissue samples from E16, E18 (18 days fetuses), day (d) 1, 10, 20, 30, 60, 180, and 1 year after birth were dissolved in urea-thiourea-SDS-dithiothreitol sample buffer using a small Dounce homogenizer or a MiniBeadBeater (Biospec Products, Bartlesville, OK) and were heated for 10min at ~65℃. Silver stained agarose gels were dried between sheets of mylar and dried gels were scanned. Human cardiac titin N2B (or the identically sized rat N2B) were used as standards to estimate molecular weights of the rat N2BA titin isoform. Full details have been discussed in a previous published report ³³.

Microarray hybridization and data analysis

Total RNA was extracted in 1ml TRIzol (Invitrogen, Frederick, MD), following manufacturer instructions. RNA was precipitated from the TRIzol reagent with isopropanol and resuspended at 0.5-1.0mg/ml using RNA secure solution (Ambion, Austin, TX). RNA samples were treated with DNaseI (DNA-free; Ambion, Austin, TX). After extraction, all samples were stored at -80 degree. RNA concentration was measured with a NanoDrop ND 1000 spectrometer (NanoDrop Technologies, Wilmington, DE), and RNA integrity was assessed on an Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA). RNA was further purified by the RNeasy mini kit (Qiagen, Valencia, CA). Double stranded cDNA was synthesized from total RNA (SuperScript II system; Invitrogen). An in vitro transcription reaction was then performed to obtain biotin-labeled cRNA from the double-stranded cDNA (Enzo BioArray High Yield RNA Transcript Labeling kit; Enzo Diagnostics, Farmingdale, NY). The cRNA was fragmented before hybridization, and then mixed in a hybridization mixture containing probe array controls, BSA, and herring sperm DNA ³⁴. A cleanup procedure was performed on the hybridization cocktail using an RNeasy spin column (Qiagen), after which it was applied to the Affymetrix Rat 230 2.0 probe array. A total of eighteen hybridization experiments were performed in which each stage (d1, d20 and d49) was represented by three wild type (Wt) and three RBM20 homozygous knockout (Hm) individual ventricular RNA extracts (Fig. 2A). Hybridization was allowed to continue for 16 h at 45°C in a GeneChip 640 hybridization oven, after which the arrays were washed and stained with phycoerythrin-conjugated streptavidin (Molecular Probes, Eugene, OR). Images were scanned using a GeneArray scanner (Agilent Technologies, Palo Alto, CA) and GeneChip cel files were subsequently processed by the log scale robust multi-array analysis (RMA). The RMA estimates were based upon a robust average of log2 (B (PM)), where B (PM) is background corrected perfect match intensities. Compared with the GCOS method the RMA method has better precision, provided more consistent estimates of fold change and provided higher specificity and sensitivity when using fold change analysis to detect differential expression ³⁵.

Figure 2.

Gene expression profile analysis with gene-chip data. (A), Biological replicates of heart tissues of Wt and Hm with development; (B), Venn Diagram analysis of differentially expressed genes at three ages; (C), Percentage of DEG with up- and down-regulated genes at three ages; The Heatmap indicated the up- and down-regulated genes at three ages between Wt and Hm. Wt, wild type; Hm, homozygous Rbm20 knockout; D, day; DEG, differentially expressed genes; Up, up-regulation; Down, down-regulation.

The complete data set is publicly available in the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/; accession numbers GSE11137). Differential gene expression analysis was performed with Limma (R/Bioconductor) Gene set and pathway enrichment analysis was performed using DAVID ³⁶ (http://david.niaid.nih.gov./david/beta/index.htm) and GSEA (http://software.broadinstitute.org/gsea/index.jsp)

Cardiomyocyte isolation and intracellular calcium measurements

Rats were injected with heparin (1000U/kg) 20mins before anesthetized with 60mg/kg ketamine and 8mg/kg xylazine (i.p.). Myocytes were isolated as described previously ¹¹. In brief, under anesthesia, the heart was dissected and cannulated via the aorta. The heart was perfused at 37°C for 4 min with perfusion buffer (in mM: 113 NaCl, 4.7 KCl, 0.6 Na₂HPO₄, 1.2 MgSO₄, 12 NaHCO₃, 10 KHCO₃, 10 HEPES, and 30 taurine, pH 7.4) equilibrated with 5%CO₂-95% O₂, followed by perfusion with digestion buffer (perfusion buffer plus 0.45 mg/ml collagenase II [240 U/mg; Worthington Biochemical Corporation], 0.13 mg/ml trypsin, and 25 µM CaCl₂) for 8-10 min. When the heart was flaccid, digestion was halted and the heart was placed in myocyte stopping buffer 1 (perfusion buffer plus 0.04 ml bovine calf serum [BCS]/ml buffer and 5 µM CaCl₂). The left ventricle (LV) was cut into small pieces, and the rest of the heart was discarded. The small pieces were then triturated several times with a transfer pipette and then filtered through a 300µm nylon mesh filter. After this, the cells were gravity pelleted and the supernatant was discarded. Next, 10 ml of myocyte stopping buffer 2 was added (perfusion buffer plus 0.05 ml BCS/ml buffer and 12.5 µM CaCl₂), and Ca²⁺ was reintroduced to a final concentration of 1.8 mM. Then, myocytes were loaded with Fura-2 AM (0.5 μM) for 10 min, and fluorescence measurements were recorded with a dual-excitation fluorescence photomultiplier system (IonOptix). Myocytes were placed on an Olympus IX-70 inverted microscope stage and imaged through a Fluor 40× oil objective. Myocytes were exposed to light emitted by a 75-W lamp and passed through either a 360- or a 380-nm filter while being stimulated to contract at 0.5 Hz. Fluorescence emissions were detected between 480 and 520 nm by a photomultiplier tube after first illuminating the myocytes at 360 nm for 0.5 sec and then at 380 nm for the duration of the recording protocol (333Hz sampling rate). The excitation scan at 360 nm was repeated at the end of the protocol and qualitative changes in intracellular Ca²⁺ concentration were inferred from the ratio of Fura-2 fluorescence intensity (FFI) at the two wavelengths (360/380). Fluorescence decay time was measured as an indication of the intracellular Ca²⁺ clearing rate. Single exponential curve fit programs were used to derive the intracellular Ca²⁺ decay constant ³⁷.

Quantitative real time PCR (QPCR)

Total RNA was extracted using TRIzol, as described above. For reverse transcription, 60 ng of RNA were mixed with 5 uM random hexamers, 1 mM each dNTP, 7.5 mM MgCl2, 40 U RNasin (Promega, Madison, WI), 1X PCR buffer II (Applied Biosystems, Foster City, CA) and 250 U of SuperScript II reverse transcriptase (Invitrogen). The reaction mixture was incubated at 25°C for 10 min, 48°C for 45 min, and 95°C for 5 min, then cooled down to 4°C. Two volumes of ethanol were added to the cDNA and evenly mixed, and then stored at -20°C for 30 minutes, centrifuging for 15 minutes at 13,000 rpm, washed with 75% ethanol and dried. The cDNA was resuspended in distilled water and used as a template. For SYBR Green quantitative real-time PCR, all primers were designed from the selected genes (Supplement table S1). All primer pairs produced a single PCR product as determined by the dissociation curve and gel analysis. Real-time PCR was performed in a 20ul reaction, 96-well format and 1 X SYBR PCR Master Mix (Biorad). The reaction was incubated in a CFX96 real-time PCR system (Biorad) for 40 cycles consisting of denaturation at 95°C for 15 s and annealing/extension at 58-60°C for 1 min. Three samples were measured in each experimental group in quadruplicate, with a minimum of two independent experiments. The relative amount of target mRNA normalized to GAPDH was calculated according to the method described previously ³⁸.

Western Blot

Total protein was separated by SDS-PAGE gel, and transferred onto a PVDF membrane. The membrane was probed with antibodies against FHL (Santa Cruz), CRP2 (Santa Cruz), IGFBP8/CTGF (Abcam), KCNQ1 (Santa Cruz), TGF β2 (Cell Signaling), RyR2 (Abcam), SERCA2A (Santa Cruz), TRDN (Santa Cruz), CASQ1 (Santa Cruz), PLB (Abcam), Rabbit anti-mouse IgG-conjugated with horseradish peroxidase (Fisher Scientific) was served as the secondary antibody. GAPDH (Santa Cruz) was served as the protein loading control. The detailed procedure was described previously ³⁴.

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 a Tukey post test, or a 2-tailed t-test for comparison of two groups. The significance level was P<0.05, P<0.01 and P<0.001.

Results and Discussion

RBM20 knockout Rat model characterization

A major RBM20 target, titin is a striated muscle-specific giant elastic protein and largely responsible for the generation of the diastolic force in the cardiac myocyte ^15-18. Titin has two classes of isoforms with one smaller isoform N2B that is stiffer, and the larger isoforms N2BAs that are more compliant ^24-25. Cardiac titin undergoes a series of isoform transitions during maturation in wild type rat (Wt) heart as observed by SDS agarose electrophoresis (embryonic day (E) 16, E18, day (d) 1, d10, d20, d30, d60, d180 and year (yr) 1) (Fig. 1A). Larger fetal titin N2BA isoform is predominantly expressed before E18, and decreases in sizes and amounts with maturation, while smaller N2B isoform starts to be present after E18 and dominant after d10 (Fig. 1A and 1C). Percentage of N2B to total titin (N2B+N2BA) changes from about 2% at E18 to approximately 85% at adult (Fig 1C). However, in Rbm20 knockout rats (Hm), titin is expressed as a constant size that is similar to the larger fetal titin N2BA isoform through all ages when compared to Wt rats (Fig. 1B and 1C). This constantly expressed larger titin isoform in Rbm20 Hm rats leads to reduced ventricular wall stiffness and has been associated with heart failure progression as described previously ²⁰. Immunocytochemistry staining of cardiomyocytes isolated from adult rat hearts with RBM20 antibody indicated that RBM20 is located in splicing speckles in the nucleus with two spots in Wt cardiomyocytes and completely depleted in Hm cardiomyocytes (Fig 1D). Splicing speckles are nuclear bodies that are important for splicing factor storage and modification and contain components of the splicing machinery and polyA RNA ^39-40.

Figure 1.

Rbm20 deficient rat model. (A and B), Agarose gel showing titin protein isoform changes with development in wild type (Wt) and homozygous Rbm20 knockout (Hm); (C), Percentage of N2B with development in Wt and Hm; (D), Rbm20 location and expression in Wt and Hm rat adult cardiomyocytes. The arrows indicate the Rbm20 localization in the nucleus. Wt, wild type; Hm, homozygous Rbm20 knockout; E, embryonic day; D, day; yr, year.

Our previous study has shown that Rbm20 deficiency in the rat model leads to heart failure with reduced ejection fraction, enlarged ventricular chamber and thinner ventricular walls ⁴. Additionally, Ours and other studies have identified over thirty RBM20 target genes that undergo alternative splicing ^{4, 14}. However, it is unknown whether Rbm20 deficiency leads to gene expression changes in the heart. Therefore, this study employs the Rbm20 knockout rat model to investigate gene expression profile using Affymetrix gene expression microarray and determine which genes or signaling pathways are associated with heart failure progression in Rbm20 knockout rats.

Gene expression profile indicated that developmental changes of gene expression are consistent with titin splicing pattern regulated by RBM20

In order to examine the effect of RBM20 on gene differential expression in heart muscle at distinct developmental stages, we selected d1, d20 and d49 as representative maturation time courses according to titin isoform expression pattern as indicated in Fig. 1. The experimental design is shown in Fig 2A. Gene expression array data were analyzed based on student t test between Wt and Hm with p value cutoff less than 0.05 and fold change more than 1.5. Out of a total 31099 probes on the Affymetrix Rat 230 2.0 Chip, 418 probes representing 381 genes including 21 unannotated genes were differentially changed when comparing Hm to Wt rats (Supplement table S2). Among 381 genes, 77 of them are differentially expressed at d1 (Supplement table S3), 121 at d20 (Supplement table S4), and 227 at d49 (Supplement table S5). Venn diagram analysis of differentially expressed genes at three ages demonstrated that four genes were commonly expressed at d1 and 20, seven at d1 and 49, forty-nine at d20 and 49, and ten at all three ages (Fig. 2B). These data hint that most of differentially expressed genes regulated by RBM20 are development dependent. RBM20 regulates the expression of fewer genes at early developmental stages when compared to late stages, which matches the pattern of titin size changes from the larger ones to smaller during maturation. Interestingly, we found that almost 99% of differentially expressed genes are up-regulated at early maturation stage, and percentage of up-regulated genes decreases at d20 to approximately 68% and at d49 to 57% (Fig. 2C). These data suggest that genes regulated by RBM20 are fewer at d1 with mostly elevated expression by comparing to late maturation stages d20 and d49 with more down-regulated genes. However, gene expression pattern regulated by RBM20 at d20 and d49 are similar, which is consistent with titin splicing pattern with maturation (Fig. 1A).

Gene functional analysis showed that up-regulated genes are associated with heart failure and gene expression pattern is similar at later maturation stages

Next we did further data mining for the relevant function of these differentially expressed genes. Gene Set Enrichment Analysis (GSEA) is a computational method that determines whether a priori defined set of genes shows statistically significant, concordant differences between two biological states ⁴¹. In order to interpret gene expression data by analyzing gene sets, GSEA analysis indicated that gene sets associated with failing heart were up-regulated at all three ages (Fig. 3A, 3B and 3C). Gene Ontology (GO) analysis indicated that most of differentially expressed genes were associated with protein binding and cell signal transduction (Fig. 3D). Gene cluster analysis suggested that the expression pattern of d20 and d49 between Wt and Hm was similar, but it was different from d1. GO and gene cluster analysis further confirmed that gene expression pattern regulated by RBM20 was consistent with titin splicing pattern changes in the heart of Wt rats which was also modulated by RBM20 (Fig. 3E, 1A and 1B).

Figure 3.

Gene functional annotation and heatmap of pairwise correlations between Wt and Hm. (A, B and C), Gene Set Enrichment analysis indicated that gene sets associated with heart failure are up-regulated at all three ages; (D), A cluster of differentially expressed gene between Wt and Hm at three ages; (E), Molecular function of differentially expressed genes with Gene Ontology (GO) analysis. D, day; Wt, wild type; Hm, homozygous Rbm20 knockout; GO, Gene Ontology.

Differentially expressed genes associated with titin direct and indirect binding and these binding partners play diverse roles that may together contribute to heart failure progression induced by RBM20

Titin binds over 70 proteins directly and indirectly ²³. It is unknown whether differentially expressed genes regulated by RBM20 contain titin binding partners. We searched the list of differentially expressed genes with known titin partners, and found that among these genes, proteins encoded by five of them directly bound to titin and twenty-two of them were indirectly associated with titin (Fig. 4). Five direct binding proteins were MyBP-C, Fhl1, Carp, Darp, and β-myosin. MyBPC has three isoforms: slow skeletal, fast skeletal and cardiac isoform that are encoded by three different genes ^42-45. Gene MyBPC2 encodes a fast skeletal isoform (fMyBP-C) that is expressed primarily in fast skeletal muscles ⁴⁵. However, it can also be expressed in cardiac muscle, and its increased expression in the heart is associated with severe dilated cardiomyopathy ⁴⁶. Mammalian cardiac muscle contains two myosin isoforms: α-myosin and β-myosin encoded by two distinct genes ⁴⁷. In adult human cardiac muscle, the major isoform is β-myosin ⁴⁸. Elevated β-myosin level has been found in human failing heart myocardium, cardiomyopathy, aging and pressure overload ^49-51. Four-and-a-half Lim domain protein-1 (Fhl1) is a member of Fhl family structurally characterized by four complete LIM domain and a N-terminal half LIM domain ⁵², which has been found associated with titin N2B domain ⁵³ and the mitogen-activated protein kinase signaling cascade (Raf1/Mek1/2/Erk2) ⁵⁴. Fhl1 plays an important role in sensing biomechanical hypertrophic stress responses in the sarcomere via titin N2B, which in turn regulates MAPK/hypertrophic signaling and muscle compliance ^54-55. Carp and Darp are two members of the three homologous muscle-ankyrin-repeat proteins (MARPs) that interact with titin I-band N2A region ^56-57. They are associated with muscle stress signaling and up-regulated when muscles undergo mechanical or metabolic challenge ⁵⁶. Carp and Darp exist both in the nucleus and in the sarcomeric I-band, and show increased expression levels in end-stage failing human dilated cardiomyopathy hearts ³². All of these five direct titin binding partners were up-regulated in Rbm20 knockout rats. According to previous studies described above, up-regulation of these five genes in human heart or animal models leads to heart failure with preserved ejection fraction ^{46,49, 50, 55, 56}, which was consistent with the phenotype of Rbm20 knockout rats and human patients with Rbm20 loss-of-function mutations ⁴.

Figure 4.

Titin direct and indirect binding partners. The central rectangle depicts the sarcomere positions of different titin segments. The direct titin binding partners are orange ovals; the indirect titin binding partners are yellow rectangles; the direct and indirect titin binding partners with differential expression are red ovals and rectangles. sANK1, small ankyrin 1 isoform; p94, calpain-3; FHL1/2, four-and-a-half-LIM-domain-protein; MyBPC, myosin binding protein C; Nbr1, neighbor-of-BRCA1-gene-1; MURF, muscle-specific-RING-finger-protein; MyHC, myosin-heavy-chain; MyLC, myosin-light-chain; TPM, tropomyosin; CARP, cardiac-ankyrin-repeat-protein; DARP, diabetes-related-ankyrin-repeat-protein; ARPP, ankyrin-repeat-domain-protein-2; TMOD, tropomodulin; MYOT, myotilin; FLNC, Υ-filamin; T-cap, telethonin; minK, potassium-channel subunit; MLP (CRP3), muscle-LIM-protein (cysteine-rich-protein-3); CRP2, muscle-LIM-protein (cysteine-rich-protein-2); ENAH, protein enabled homolog; ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; KCND2/3, potassium voltage-gated channel subfamily D member 2/3; KCNK2, potassium channel subfamily K member 2; KCNQ1, potassium voltage-gated channel subfamily Q member 1; TNFa-CE (ADAM-17), tumor-necrosisfactor-alpha-converting-enzyme (A-disintegrin-and metalloprotease-17 protein); ADAM33, disintegrin-and metalloprotease-33 protein; IGFBP, insulin growth factor binding protein; ZNF, zinc finger protein; ERK2, extracellular-signal-regulated-kinase-2; ANK2, ankyrin 2; YB-1, Y-box-binding-protein-1; ISOT-3 (USP13), isopeptidase-T-3; USP24, ubiquitin specific peptidase 24; UBC9, SUMO-conjugating enzyme UBC9; UBCH10, SUMO-conjugating enzyme E2C.

In addition to titin direct binding partners, we found more titin indirect binding partners that are regulated by RBM20. These indirect binding partners play diverse roles in different physiological function. For example, Kcnq1, Kcnd2, Kcnd3 and Kcnk2 are potassium voltage-gated channels required for cardiac action potential, cellular excitability and heart function ^58-61. The cardiac gene expression and plasma level of ANP and BNP are in response to changes of diuresis, natriuresis and systemic blood pressure. Increased levels of ANP and BNP are associated with chronic heart failure, and can be predictors for left ventricular ejection fraction, mortality and morbidity in human patients ^62-63. Another example is Myl4 that encodes a protein called atrial light chain 1 (ALC-1). Increased expression of ALC-1 was found in patients with hypertrophic obstructive cardiomyopathy, and in ischemic and idiopathic cardiomyopathy with larger individual differences ^{51, 64-65}. RBM20-regulated genes involving diverse functions indicate that Rbm20 deficiency leads to heart failure through cooperation of multiple biological processes instead of single gene effect such as titin, though they may play minor roles when compared to titin.

Validation of titin binding partners and non-titin binding partners by QPCR and western blotting

We next validated gene expression resulting from DNA microarray by QPCR and western blotting. We selected all five genes (Myosin, Carp, Darp, Mybpc2 and Fhl1) from direct titin binding partners and five genes (Myot, Anp, Myl4, Crp2 and Kcnq1) from indirect titin binding partners. We also selected six genes (Tgfβ2, Triadin/Trdn, Akip1, Eif4ebp, Igfbp8/Ctgf and Casq1) that haven't been identified associated with titin binding for validation at d1, d20 and d49. Comparisons were made between gene expression levels from DNA microarray and QPCR at individual ages. For titin direct binding partners, we found that all five genes were up-regulated and validated by QPCR at three ages which is consistent with DNA microarray data (Fig. 5A, 5B and 5C). For titin indirect binding partners, DNA microarray data indicated that all five genes were not increased by comparing to Wt at d1, while one gene Anp was validated to increase by QPCR at d1 (Fig. 5D). All five genes were significantly increased at d20 and d49 from both DNA microarray and QPCR (Fig. 5E and 5F). For other genes which are not related to titin binding, the expression level of genes Akip1 and Eif4ebp was similar to Wt at d1 from both DNA microarray and QPCR data. Expression of Igfbp8 was slightly lower by comparing to Wt at d1 from DNA microarray data, but it had no significant difference validated from QPCR. Gene Casq1 was slightly increased, but no significant difference at d1 from DNA microarray data, while QPCR results indicated significant increase. The expression level of Tgfβ2 and Trdn validated by QPCR was consistent with DNA microarray data. Interestingly, Genes Trdn and Eif4ebp were significantly decreased and Tgfβ2, Akip1, Igfbp8 and Casq1 were significantly increased by comparing to Wt at d20 and d49 from DNA microarray, which were confirmed by QPCR (Fig. 5G, 5H and 5I). We then selected total six proteins (Fhl1, Darp, Crp2, Igfbp8, Kcnq1 and Tgfβ2) from direct and indirect titin binding partners and non-titin binding partners to validate in protein level with western blotting at three ages (Fig. 5J). We found that proteins Fhl1, Darp, Crp2 and Tgfβ2 were consistent with DNA microarray and QPCR data. Protein Igfbp8 was slightly reduced in d1 Hm which was consistent with DNA microarray data and significantly increased at d20 and 49 that was consistent with both DNA microarray and QPCR data. Protein level of Kcnq1 was not changed at three ages which was consistent with DNA microarray and QPCR data at d1, but not d20 and 49 (Fig. 6J). These data suggest that most of genes validated by QPCR and western blotting have consistency between mRNA level and protein level, and they are consistent with DNA microarray data too, confirming the gene expression changes obtained from DNA microarray data are reliable.

Figure 5.

Validation of differentially expressed genes with QPCR and western blot. (A, B and C), Expression level of selected titin direct binding partners with DNA microarray and QPCR; (D, E and F), Expression level of selected titin indirect binding partners with DNA microarray and QPCR; (G, H and I), Expression level of selected genes that are not titin binding partners with DNA microarray and QPCR; (J), Validation of selected proteins from titin direct and indirect partners and non-titin binding partners with western blot. D, day; Wt, wild type; Hm, homozygous Rbm20 knockout; GAPDH, protein loading control. Mean±SEM (n=3); *P<0.05; **P<0.01; ***P<0.001; ^#P<0.05; ^##P<0.01; ^###P<0.001.

Figure 6.

Transcriptional network of calcium handling genes and intracellular calcium changes. (A), Schematic diagram of calcium release signaling and sarcomere; (B), validation of key calcium handling proteins with western blotting; (C), Intracellular calcium concentration changes with cardiomyocytes isolated from the rat heart at d49. Resting FFI, basal intracellular calcium level; (D), ΔFFI, Rise in intracellular calcium in response to electrical stimulus shown as changes in Fura-2 fluorescence intensity; (E), D, day; Wt, wild type; Hm, Rbm20 homozygous knockout; GAPDH, protein loading control. Mean ± SEM (n=30), *P < 0.05, **P<0.01.

RBM20 regulates a transcriptional network of calcium handling genes in the heart and alters intracellular calcium levels during contraction

We next studied and validated genes involving in calcium release signaling. From our DNA microarray data, we found that genes Trdn, Casq1, Cacna1c, Tnnt1, MyHC involved in calcium handling and sarcomere contractility were differentially expressed. These genes were displayed in schematic diagram of calcium handling and sarcomeric structure (Fig. 6A). We validated calcium handling genes Trdn and Casq1 with fold changes, while other genes (RyR2, Serca2a and phospholamban/PLN) that have no fold changes in calcium handling from DNA microarray. Western blotting analysis demonstrated that Trdn was down-regulated and casq1 was up-regulated in Hm at d1, d20 and d49, while other genes (RyR2, Serca2a and PLB) had no changes (Fig. 6B). These results are consistent with the DNA microarray data. Interestingly, western blot indicated that Trdn and casq1 protein bands had migration difference between Wt and Hm. This could be caused by alternative splicing that is regulated by RBM20 as reported previously ⁴. Given that expression profile changes of some calcium handling genes, we assumed that intracellular calcium concentration may be subjected to changes during muscle contraction. Thus, we measured the calcium transient with cardiomyocytes isolated from the heart at d49 using a calcium contractility system (IonOptix). The results indicated that basal level of the intracellular calcium concentration was significant higher in Hm cardiomyocytes than that in Wt (Fig. 6C), while, in response to electrical stimulus, intracellular calcium change was lower in Hm cardiomyocytes than that in Wt (Fig. 6D). However, intracellular calcium decay rate had no significant difference between Wt and Hm cardiomyocytes (Fig. 6E). These data suggest that calcium releasing signaling could be impaired, leading to calcium leaking, while calcium ATPase pump system was physiologically functional and could normally pump calcium from cytosol to sarcoplasmic reticulum (SR) for calcium uptake. Calcium releasing from SR is through a multi-molecular signaling complex channel includeing RyR, Trdn, Junctin and Casq that plays a key role in excitation-contraction coupling (EC-coupling). The calcium enters the cytosol through plasmalemmal voltage-gated calcium channels and activates RyR channel to release more calcium from SR for muscle contraction ^66-68. However, the RyR channel is also positively controlled by the luminal side of the channel ^{66, 69-70}. Auxiliary proteins (Trdn, Junctin and Casq) with luminal localization mediate the effect of luminal calcium on RyR ^71-72. Casq is a major calcium binding protein and changes conformation upon calcium binding, resulting in interaction with RyR to sense luminal calcium ^73-76. Trdn and Junctin mediate the inhibitory interactions between Casq and the RyR channel ⁷³. Therefore, down-regulated Trdn and up-regulated Casq1 may cooperate for calcium releasing. Since RyR2 is not changed based on our data, it still remains unknown how calcium leaking occurs in Rbm20 Hm rats. Previous study indicated that RBM20 regulates alternative splicing of RyR2, leading to a 24bp insertion in Rbm20 Hm rats ¹⁴, which could be a cause for calcium leaking. However, further study is needed to confirm this hypothesis. Nevertheless, expression changes of Trdn and Casq1 may play a role in the impaired calcium release signaling in Rbm20 Hm rats. On the other hand, SR calcium uptake is mediated by a calcium ATPase Serca2, whose activity is reversibly regulated by PLN ⁷⁷. Since no changes of these two proteins were found in Rbm20 Hm rats, calcium uptake appears normal, which was consistent with the measurement of intracellular calcium decay rate (Fig. 6E).

Conclusion

Our results reveal that the muscle tissue specific splicing factor RBM20 not only regulates gene alternative splicing but also gene expression in the heart. Previous studies have identified approximately thirty genes that are undergoing alternative splicing regulated by RBM20. In this study, we have demonstrated that RBM20 also modulates gene expression of about four hundred genes with heart maturation. Intriguingly, we found that for certain genes such as Trdn and MyHC, RBM20 regulates their splicing as well as their expression level. Rbm20 deficiency in animal models leads to dilated cardiomyopathy and heart failure, and mutations found in RBM20 are also associated with dilated cardiomyopathy in human patients. However, the detailed mechanisms of how RBM20 causes heart dysfunction remain unclear. Our data suggest that the mechanisms of both gene splicing and gene expression regulated by RBM20 will require more detailed study in the future. Particularly, whether gene expression is directly regulated by RBM20 or is a secondary effect remains unknown. At least, previous studies have shown that pre-mRNA splicing can be coupled to transcription and crosstalk and impact each other ^78-79, and trans- and cis-acting factors can regulate RNA polymerase II elongation ^{80, 81}. However, whether RBM20-mediated transcription is involved in coupling to RBM20-mediated splicing is very interesting to be further studied. In conclusion, our results provide new insights into the mechanism of RBM20-induced heart disease.

Supplementary Material

Supplementary tables.

Abbreviation

RBM20

RNA binding motif 20

RRM

RNA-recognition domain

SR

serine /arginine

E16

16 days fetuses

SDS-VAGE

sodium dodecyl sulfate agarose gel electrophoresis

Wt

wild type

Hm

homozygous knockout

RMA

robust multi-array analysis

LV

left ventricule

FFI

Fura-2 fluorescence intensity

d

day

yr

year

GO

Gene Ontology

GSEA

Gene Set Enrichment Analysis

Fhl1

four-and-a-half Lim domain protein-1

MARPs

muscle-ankyrin-repeat proteins

SR

sarcoplasmic reticulum

EC-coupling

excitation-contraction coupling

sANK1

small ankyrin 1 isoform

p94

calpain-3

MyBPC

myosin binding protein C

Nbr1

neighbor-of-BRCA1-gene-1

MURF

muscle-specific-RING-finger-protein

MyHC

myosin-heavy-chain

MyLC

myosin-light-chain

TPM

tropomyosin

CARP

cardiac-ankyrin-repeat-protein

DARP

diabetes-related-ankyrin-repeat-protein

ARPP

ankyrin-repeat-domain-protein-2

TMOD

tropomodulin

MYOT

myotilin

FLNC

Υ-filamin

T-cap

telethonin

mink

potassium-channel subunit

MLP (CRP3)

muscle-LIM-protein (cysteine-rich-protein-3)

CRP2

muscle-LIM-protein (cysteine-rich-protein-2)

ENAH

protein enabled homolog

ANP

atrial natriuretic peptide

BNP

brain natriuretic peptide

KCND2/3

potassium voltage-gated channel subfamily D member 2/3

KCNK2

potassium channel subfamily K member 2

KCNQ1

potassium voltage-gated channel subfamily Q member 1

TNFa-CE (ADAM-17)

tumor-necrosisfactor-alpha-converting-enzyme (A-disintegrin-and metalloprotease-17 protein)

ADAM33

disintegrin-and metalloprotease-33 protein

IGFBP

insulin growth factor binding protein

ZNF

zinc finger protein

ERK2

extracellular-signal-regulated-kinase-2

ANK2

ankyrin 2

YB-1

Y-box-binding-protein-1

ISOT-3 (USP13)

isopeptidase-T-3

USP24

ubiquitin specific peptidase 24

UBC9

SUMO-conjugating enzyme UBC9

UBCH10

SUMO-conjugating enzyme E2C.