|Home | About | Journals | Submit | Contact Us | Français|
We sought to identify a novel gene for dilated cardiomyopathy (DCM).
DCM is a heritable, genetically heterogeneous disorder that remains idiopathic in a majority of patients. Familial cases provide an opportunity to discover unsuspected molecular bases of DCM, enabling preclinical risk detection.
Two large families with autosomal dominant DCM were studied. Genome-wide linkage analysis was used to identify a disease locus, followed by fine mapping and positional candidate gene sequencing. Mutation scanning was then performed in 278 unrelated subjects with idiopathic DCM, prospectively identified at the Mayo Clinic.
Overlapping loci for DCM were independently mapped to chromosome 10q25-q26. DNA sequencing of affected individuals in each family revealed distinct heterozygous missense mutations in exon 9 of RBM20, encoding RNA binding motif protein 20. Comprehensive coding sequence analyses identified missense mutations clustered within this same exon in six additional DCM families. Mutations segregated with DCM (composite logarithm of the odds score >11.49), were absent in 480 control samples, and altered residues within a highly conserved arginine/serine (RS)-rich region. Expression of RBM20 messenger RNA was confirmed in human heart tissue.
Our findings establish RBM20 as a DCM gene and reveal a mutation hotspot in the RS domain. RBM20 is preferentially expressed in the heart and encodes motifs prototypical of spliceosome proteins that regulate alternative pre-mRNA splicing, thus implicating a functionally distinct gene in human cardiomyopathy. RBM20 mutations are associated with young age at diagnosis, end-stage heart failure, and high mortality.
Prevention of heart failure has been a major public health focus, founded on knowledge of pathogenic mechanisms and modifiable risk factors for hypertension and coronary artery disease (1). Heart failure remains an idiopathic condition, however, in 50% of adults (2) and 66% of children (3) referred to cardiologists, and end-stage idiopathic dilated cardiomyopathy (DCM) is the most common indication for cardiac transplantation (4,5). Indeed, onset of heart failure symptoms in DCM typically portends advanced myocardial disease and risk for sudden death (6) following years to decades of clinically silent but insidiously progressive myopathy. Even in children, this inherent delay in diagnosis and treatment of DCM accounts for 10-year transplantation-free survival of only 42% (3). Improved prediction, treatment, and prevention of DCM will require discovery of preclinical biomarkers, better tools for risk-stratification, and the molecular and cellular basis of disease to enable mechanism-based therapies (1).
Recognition of DCM as a familial disorder in 20–48% of cases (7–11) has provided a rationale for routine screening echocardiography in at-risk relatives to detect presymptomatic disease (12). Moreover, it has been the impetus for human genetics investigations to uncover the molecular basis of DCM (13–14). Since 1993, pathogenic mutations in over 20 genes encoding cytoskeletal, contractile, nuclear membrane, calcium-regulating, and ion channel proteins have been identified in patients with DCM (15). The majority of studies are hypothesis-based, targeting candidate genes like cardiac actin (16) that encode proteins with established function in the heart. By contrast, unanticipated DCM genes and insights into disease pathobiology have emerged from rare families suitable for whole genome mapping studies (17–20). Here, we used genetic linkage analysis in 2 large families with autosomal dominant DCM to map a disease locus, leading to discovery of a mutation hotspot within an RNA-binding protein gene associated with high morbidity and mortality.
Patients with DCM evaluated at the Mayo Clinic in the years 1987–1992 and 1999–2008 and their relatives were recruited and medical records were reviewed. We enrolled 280 unrelated probands; familial DCM was confirmed in 24% (DCM documented in ≥1 first degree relative) and suspected in 27% (based on history alone). Family history of sudden death was present in 18%. The 8 families described in the current study were white and of northern European ancestry by self-reporting. An ethnically-matched group of 480 control subjects with normal echocardiograms was randomly selected from a community-based cohort (21). Subjects provided written informed consent under research protocols approved by the Mayo Clinic institutional review board.
Echocardiograms in relatives were performed for clinical indications or under the auspices of the research study. Diagnostic criteria for DCM were: lack of an identifiable cause for disease, left ventricular diastolic and/or systolic dimensions >95th percentile indexed for body surface area (22), and left ventricular ejection fraction <50%. Subjects with normal echocardiograms were classified as “unaffected” and those with equivocal or insufficient data were classified as “uncertain.” Genomic DNA was isolated from peripheral-blood white cells (Puregene Blood Kit, Gentra/QIAGEN, Valencia, California) or from paraffin-embedded tissue (QIAamp DNA FFPE Tissue Kit, QIAGEN).
Genome-wide linkage analysis was performed with the ABI PRISM Linkage Mapping Set MD10, version 2.5 (Applied Biosystems, Foster City, California), consisting of polymerase chain reaction (PCR) primer pairs for 400 short tandem repeat markers. After PCR amplification of DNA samples, fragments were resolved on an ABI PRISM 3130xl and genotypes were scored with GeneMapper Software. Two-point and multipoint linkage analyses were performed with use of the FASTLINK program and specification of the following variables: a phenocopy rate of 0.001, equal marker allele frequencies, and dichotomous liability classes (“affected” and “unaffected”). For mutations, a frequency of 0.001 was specified. Logarithm of the odds (LOD) scores were determined for affected subjects only and for 80% and 100% penetrance models at recombination frequencies of 0.0 to 0.4.
Fine locus mapping was performed with microsatellite markers on physical maps, accessible on the Web site of the National Center for Biotechnology Information (NCBI; www.ncbi.nlm.nih.gov). Genotyping was accomplished by PCR amplification of DNA radiolabeled with [α-32P] deoxycytidine triphosphate, resolution of alleles by polyacrylamide-gel electrophoresis, and visualization by autoradiography. Scored genotypes were assembled as haplotypes to define the critical region.
Expression profiles of candidate genes, derived from Affymetrix GeneChip array data for 12 normal human tissues (accession GDS424) or 61 normal mouse tissues (accession GDS592), were assessed by searching the Gene Expression Omnibus (GEO) link on the NCBI Web site (23). The genomic structure of RBM20 was based on predicted reference mRNA sequence (accession NM_001134363.1), retrieved from NCBI. Primer pairs were designed for genomic DNA PCR-amplification of the coding regions of the 14 predicted exons (Online Table 2), using Oligo Primer Analysis Software, version 6.71 (Molecular Biology Insights, Cascade, Colorado). For sequencing, amplified products were treated with ExoSAP-IT (USB Corp, Cleveland, Ohio) and sequenced by the dye-terminator method with use of an ABI PRISM 3730xl DNA Analyzer (Applied Biosystems). DNA sequences were viewed and analyzed using Sequencher, version 4.5 DNA analysis software (Gene Codes Corp, Ann Arbor, Michigan). The reference mRNA and derived protein sequence (accession NP_001127835.1) were used for annotation of identified mutations.
Denaturing high-performance liquid chromatography (DHPLC) heteroduplex analysis (WAVE DHPLC System, Transgenomic, Omaha, Nebraska) was used to screen for sequence variants in our DCM cohort and control samples. Ideal buffer gradients and column melting temperatures were determined using Transgenomic Navigator™ software version 1.7.0 Build 25 and subsequent optimization (Online Table 2). Chromatographic elution profiles of amplified fragments were compared against the wild-type homoduplex pattern; samples yielding anomalous traces were selected for sequencing. To test for a common founder among families with the same RBM20 mutation, haplotypes for mutant alleles were constructed from an intra-genic tetranucleotide-repeat sequence and single nucleotide polymorphisms, identified by sequencing family members.
Total RNA was extracted from frozen human heart tissue (RNeasy Fibrous Tissue Midi Kit, QIAGEN) and 1.0 μg was reverse transcribed with an oligo(dT) primer to produce complementary deoxyribonucleic acid (cDNA) from messenger RNA (mRNA) (SMART RACE cDNA Amplification Kit, Clontech, Mountain View, California). Primers cDNA-F (CCTACCCCAGATCATCCAAAATGC) and cDNA-R (AACAAACACTTTGCAGTCAGTTATACA) were designed to PCR amplify and sequence 5'-RACE-Ready cDNA, spanning the RBM20 region containing the identified mutations. A subsequent nested reaction utilizing primers cDNA-2F (GAACCCATTCTCGGTCAGTAACCC) and cDNA-2F/3'UTR-R (TCTCTCTGCCCTTCCTCCATTAGT) was performed to provide optimal sequence quality. To identify conserved structural domains, RBM20 reference protein sequence was subjected to a Conserved Domain Database search performed with BLASTP, accessed on the NCBI Web site. Conservation of amino acids altered by missense RBM20 mutations was investigated by aligning our translated RBM20 cDNA sequence with RBM20 protein sequences of other species.
Clinical data and DNA samples were collected from 2 large families in which a clinically aggressive form of DCM segregated as an autosomal dominant trait (Figure 1, Table 1). Kindred DC-12 was recruited for the study in 1991, when an unaffected family member sought medical genetics consultation. The patriarch (Figure 1A: I.1) was of Scottish ancestry and died suddenly at age 39 years. Ten family members developed documented DCM, 2 as young children (mean age at diagnosis = 30.0 years). Two underwent cardiac transplantation as young adults and all but 3 have died of their disease (mean age at death = 37.7 years). Kindred DC–35 was recruited in 2005, following a diagnostic screening echocardiogram in the proband (Figure 1B: III.17) whose father died suddenly at age 29 years. The family was of Norwegian ancestry and comprised of 12 relatives with documented DCM (mean age at diagnosis = 41.3 years) and 5 others with DCM and/or sudden death by history alone. Seven family members with confirmed or suspected DCM died at a mean age of 45.7 years. Five living relatives with DCM had received implantable cardioverter defibrillators (ICDs).
Genome-wide linkage analyses, followed by regional high-density genotyping on chromosome 10, identified a peak two-point LOD score of 3.55 at marker D10S1269 in DC-12 and 4.55 at marker D10S221 in DC-35. Linkage to other regions of the genome with two-point LOD scores >1.0 was excluded by multipoint and/or haplotype analyses with additional markers (data not shown). Fine mapping in DC-12 identified a disease-associated haplotype on chromosome 10q25.1–q26.2 (Figure 1A), a region spanning 19.3 Mb, which was inherited by all affected subjects (peak multipoint LOD score 3.62 for all subjects, assuming 100% mutation penetrance, and 2.67 for affected subjects only). A recombination event within this interval occurred in a 43-year-old female with a normal echocardiogram (III.14). Assuming she did not inherit the disease-associated mutation, the critical region narrowed to 4.6 Mb. Fine mapping in DC-35 identified an overlapping disease-associated haplotype (Figure 1B) spanning 22.8 Mb (peak multipoint LOD score 4.89 for all subjects, assuming 100% mutation penetrance, and 3.58 for affected subjects only). The haplotypes were different for each family suggesting they did not share common ancestry, yet the overlapping disease loci raised the possibility of a shared DCM gene.
Candidate genes were selected from the 19.3 Mb critical region in DC-12, comprised of more than 150 genes, based on cardiac expression and/or physiologic rationale. Mutations within exons of 25 genes were excluded by DNA sequencing (Online Table 1). RBM20, a gene with unknown function, was included based on its genomic location and expression pattern. Among 12 human tissues, RBM20 is most highly expressed in the heart with 4-fold greater transcript abundance in cardiac than skeletal muscle according to GEO array data. Moreover, it is one of only 19 genes with a mean expression in heart >8-fold higher than the combined mean expression in 11 other tissues. Similarly, among 61 murine tissues it is most highly expressed in heart (5-fold greater than skeletal muscle). Sequencing of the 14 exons of RBM20 identified a distinct heterozygous missense mutation in exon 9 in each family, resulting in a P638L substitution in DC-12 and a R634Q substitution in DC-35 (Figures 1, ,3A).3A). Mutations cosegregated with the disease phenotype and were absent in unaffected family members and 480 ethnically-matched control subjects.
To determine if RBM20 mutations were present in other cases of DCM, we screened the 14 coding exons in our remaining cohort of 278 subjects using DHPLC. Three unique heterozygous missense mutations - R636S, R636H, and S637G - were identified in 6 other families, all clustered within exon 9 (Figures 2, ,3A).3A). Among the 8 families with RBM20 mutations, 2 had an identical mutation resulting in P638L substitution and 3 had an identical mutation resulting in R636S substitution. Haplotype analysis (Online Table 3) excluded a common ancestral founder for the P638L substitution. While the disease-associated haplotypes were the same in the three families with an R636S substitution, the majority of individual alleles comprising the haplotype are the most common variants within a white European population. Consequently, a founder effect could not be conclusively established. Mutations were absent in control samples and cosegregated with DCM in the 7 families where DNA samples were available from 2 or more affected subjects. Combined peak two-point LOD scores for mutations versus DCM in the 4 largest families (DC-12, DC-35, DC-27, DC-50) ranged from 8.02 (affected subjects only) to 11.49 (all subjects, assuming 100% mutation penetrance).
Based on the predicted reference cDNA (mRNA), RBM20 is comprised of 14 exons (Figure 3B). Portions of exons 2 and 14 and all of exons 3 through 13 were verified in a single open reading frame cDNA derived from oligo(dT)-primed heart RNA (Figure 3B). This confirmed that these exons are transcribed and spliced into messenger RNA in the heart, including exon 9 which contained the cluster of identified RBM20 mutations. A Conserved Domain Database search of the translated reference RBM20 cDNA indicated homology to an RNA Recognition Motif 1 Superfamily domain spanning exons 6–7 (e-value = 0.005) and a U1 zinc finger domain (e-value = 2e−4) spanning exons 13–14. Additionally, exon 9 encodes an RS-rich domain, which is disrupted by the 5 identified missense mutations. Each resultant amino acid substitution alters a residue in RBM20 conserved among diverse species (Figure 3C).
RBM20 mutations were associated with clinically aggressive DCM. Collectively, the 39 subjects in our 8 families with a mutation and confirmed DCM were diagnosed 9 years earlier than a comparable series of patients with sporadic and familial DCM who underwent family screening (mean age at diagnosis 35.9 versus 45.2 years) (7). Death occurred in 11 (mean age = 45.2 years) and was deemed sudden in 3; 4 underwent cardiac transplantation (mean age = 28.5 years) and 8 ICD implantation. Subjects who enrolled in our study did not, however, fully represent the malignant nature of their familial disease as revealed by their pedigrees. Among the 32 additional relatives with suspected DCM by family history, for whom medical records were unavailable and/or mutation status could not be determined, 13 died suddenly (mean age = 32.7 years), 3 underwent cardiac transplantation, and 3 had ICD implantation. There were no consistent electrocardiographic features in subjects with an RBM20 mutation; 9 had ventricular tachycardia. Variable degrees of myocyte hypertrophy and interstitial fibrosis were observed on histopathological analysis. Most enrolled subjects with accessible follow up data had advanced disease and exhibited minimal improvement or further deterioration on medical treatment, albeit drug therapy was highly variable. Correlation between RBM20 mutations and phenotype was not without exception, however. There were 5 female subjects who inherited a mutation but did not fulfill diagnostic criteria for DCM: 1 subject in DC-35 (age 24 years) and 3 subjects in DC-27 (ages 15, 39, and 64 years) had left ventricular enlargement with normal ejection fraction; 1 subject in DC-9 (age 27 years) had a normal echocardiogram. No overt non–cardiac phenotypes were evident among subjects with RBM20 mutations.
The majority of known DCM genes encode cytoskeletal or contractile proteins of cardiac myocytes, with direct roles in the generation and/or transmission of contractile force through protein-protein interactions (14). An expanded understanding of the pathobiology of DCM has emerged from identification of mutations that perturb myocardial function via impaired calcium (24), potassium (25), or sodium ion homeostasis (18,19). Collectively, these molecular genetic etiologies for DCM reveal a fundamental defect in excitation-contraction coupling and the heart's capacity to perform under physiologic and stress conditions. Notable exceptions to this paradigm have been revealed through discovery of unsuspected DCM genes, like LMNA and EYA4, in large families suitable for linkage analysis. LMNA encodes lamin A/C, a ubiquitously expressed nuclear membrane protein. By unknown mechanisms, mutations in LMNA cause DCM and conduction system disease (17) or a spectrum of non-cardiac disorders. EYA4 encodes a transcriptional coactivator, which interacts with DNA-binding transcription factors. Mutations in EYA4 are predicted to alter cochlear and cardiac gene expression, causing a syndrome of DCM and sensorineural hearing loss (20). RBM20, here identified as a gene for familial DCM, suggests perturbation of post-transcriptional pre-mRNA processing as a distinct molecular basis for the disorder.
RBM20 encodes RNA binding motif protein 20, with a prototypical RNA-recognition motif followed by an RS domain (26). These structural features are characteristic of a family of RNA-binding SR proteins that assemble in the spliceosome, a large multi-protein complex that orchestrates constitutive and alternative splicing of pre-messenger RNA (27). Indeed, over 70% of human genes express multiple mRNA transcripts via alternative splicing of exons, conferring vast diversity to the proteome (28). Heritable diseases are frequently attributable to cis-acting mutations, which disrupt normal splicing of the gene in which the mutation occurs. However, trans-acting mutations within spliceosome protein genes have been identified in only three human disorders - spinal muscular atrophy, retinitis pigmentosa, and Prader-Willi syndrome (27). Such mutations have the potential to impair normal splicing of multiple genes, as recently demonstrated by exon microarray analysis in a mouse model of spinal muscular atrophy (29). The specific function of RNA binding motif protein 20 in the human heart and the downstream effects of the identified RBM20 mutations that cause DCM remain unknown. However, a pathogenic link between genetic disruption of alternative splicing-regulating SR proteins of the spliceosome and DCM has now been established in mouse models (30).
Since the first DCM-associated gene was identified by linkage analysis 15 years ago (31,32), clinical application of research findings has proved challenging due to the marked genetic heterogeneity of DCM. While routine genetic testing may be practical in certain heritable cardiac disorders (33), no single gene or mutation for DCM has emerged as common (15). Targeted genetic testing may be practical, however, in clinically defined subgroups. For example, mutations in LMNA and SCN5A have been associated with a cardiac syndrome of DCM, impaired automaticity and conduction, and atrial fibrillation (17–19). By use of genome-wide linkage analysis, the present study further expands the spectrum of DCM genes. Remarkably, the 5 unique RBM20 mutations identified in 8 families are clustered within a single exon that encodes an RS-rich domain. In our cohort, this mutation hotspot accounted for 3% (8/280) of all DCM cases, 5% (8/151) of confirmed or suspected familial cases, and 13% (7/54) of cases with a history of sudden death.
Our study highlights the importance of family screening to detect presymptomatic DCM (7,12). Indeed, 68% (43/63) of the subjects in our 8 families were asymptomatic and first diagnosed with DCM on the basis of a screening echocardiogram. Despite the lack of symptoms, the RBM20 mutations we identified were highly penetrant and only 5 of 44 individuals with a mutation did not fulfill diagnostic criteria for DCM. In fact, 4 of these 5 subjects had left ventricular dilation, a known precursor to overt DCM (7,10). Penetrance of familial DCM is, however, age-dependent and the majority of subjects who enrolled in our study were adults. Discovery of the genetic basis for DCM in these families now enables a preclinical diagnosis in at-risk children and young adults. Given the malignant nature of RBM20 mutations, this knowledge would justify closer clinical follow up, meticulous attention to coexistent modifiable risk factors, and earlier institution of therapies proven to alter the natural history of heart failure (34) and decrease risk of sudden death (6).
We gratefully acknowledge the patients and families who participated in this study and the physicians who referred them. We thank Jeanne L. Theis, PhD for critical review of the manuscript.
This work was supported by the National Heart, Lung, and Blood Institute, National Institutes of Health (R01 HL071225) and the Marriott Program for Heart Disease Research.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.