Exome Sequencing and Genome-Wide Linkage Analysis in 17 Families Illustrate the Complex Contribution of TTN Truncating Variants to Dilated CardiomyopathyClinical Perspective
Background—Familial dilated cardiomyopathy (DCM) is a genetically heterogeneous disease with >30 known genes. TTN truncating variants were recently implicated in a candidate gene study to cause 25% of familial and 18% of sporadic DCM cases.
Methods and Results—We used an unbiased genome-wide approach using both linkage analysis and variant filtering across the exome sequences of 48 individuals affected with DCM from 17 families to identify genetic cause. Linkage analysis ranked the TTN region as falling under the second highest genome-wide multipoint linkage peak, multipoint logarithm of odds, 1.59. We identified 6 TTN truncating variants carried by individuals affected with DCM in 7 of 17 DCM families (logarithm of odds, 2.99); 2 of these 7 families also had novel missense variants that segregated with disease. Two additional novel truncating TTN variants did not segregate with DCM. Nucleotide diversity at the TTN locus, including missense variants, was comparable with 5 other known DCM genes. The average number of missense variants in the exome sequences from the DCM cases or the ≈5400 cases from the Exome Sequencing Project was ≈23 per individual. The average number of TTN truncating variants in the Exome Sequencing Project was 0.014 per individual. We also identified a region (chr9q21.11-q22.31) with no known DCM genes with a maximum heterogeneity logarithm of odds score of 1.74.
Conclusions—These data suggest that TTN truncating variants contribute to DCM cause. However, the lack of segregation of all identified TTN truncating variants illustrates the challenge of determining variant pathogenicity even with full exome sequencing.
Whole exome–sequencing technologies are rapidly enabling the identification of novel rare variants in patients with cardiomyopathy, but assigning pathogenicity remains challenging. Truncating variants in TTN were recently observed in 25% of familial dilated cardiomyopathy (DCM) cases.1 DCM is genetically heterogeneous with rare variants in >30 disease genes, including TTN, previously indicated to cause DCM.2,3 Before the recent publication of TTN contributing to a major fraction of genetic DCM, the fraction of cases attributable to any single gene ranged from <0.5% to ≈6% per disease gene.4
Clinical Perspective on p 153
Discovery and incorporation into clinical tests of a single gene accounting for a large fraction of DCM cases could be helpful for presymptomatic diagnosis in at-risk family members, but the clinical translation of this finding is confounded by several factors. First, despite a significant excess of truncating variants in DCM cases, these variants also occur in ≈3% of controls.1 This is not unusual in complex trait analysis, where common genetic variants occur more frequently but not exclusively in cases compared with controls and increase disease risk or susceptibility. However, in the context of DCM, which has been categorized primarily as a rare-variant mendelian disease with marked locus and allelic heterogeneity,5 it is essential to know which truncating variants are pathogenic. Second, with >300 exons and >34 000 amino acids, TTN has the largest coding sequence in the genome, and the majority of the general population will have at least 1 rare (defined as a mean allele frequency <0.5%) missense or truncating variant at this locus. Next-generation sequencing now allows rapid variation analysis of the TTN gene despite its size. However, it relies on an economy of scale, and at a cost similar to that of sequencing TTN alone, next-generation sequencing can be used to sequence the entire coding sequence of the genome. This allows DCM patients to be screened for sequence variants in TTN in parallel with all other known DCM genes and the rest of the coding genome, leading to the third issue: The recent study of TTN truncating variants in DCM1 used a custom next-generation sequencing panel specific for TTN, and therefore neither the role of genetic variants in other known DCM genes nor segregating variants in novel DCM genes could be assessed. Detailed examination of TTN truncating variants in the context of all coding variants in known and potentially novel DCM genes is needed to assess variant pathogenicity.
In this study, we used exome sequence data from 17 families, each with ≥3 members affected with DCM, in an effort to identify a genetic cause of DCM, because each family proband was negative for mutations in the coding regions of 16 DCM genes, as previously reported.6–11 From these exome sequences, we identified several families with TTN truncating variants. To more carefully assess TTN variants as a cause of DCM, we constructed a linkage map of common informative single-nucleotide variants (SNVs) from our exome data in all 17 families and performed linkage analysis across the genome. We hypothesized that if TTN variants were causative of 25% of DCM, we would observe a significant combined logarithm of odds [LOD] score across our 17 families at this locus compared with the rest of the genome. Second, we evaluated all other unbiased rare variations in the exome data to identify putative DCM causative variants in each family. We describe the variants identified in TTN in the context of other top-ranking variants across the exome sequence of each family. Third, we examined the nucleotide diversity at the TTN locus in the 5400 exome sequences available from the Exome Variant Server12 to determine whether the large amount of variation within this gene is accounted for by size or whether the TTN locus is more genetically diverse than other known DCM genes.
Materials and Methods
Written informed consent was obtained from all subjects, and the Institutional Review boards at the Oregon Health and Science University and the University of Miami approved the study. The investigation included 17 families, each with ≥3 members affected with DCM and with each proband already known to be point mutation negative for 16 known DCM genes.6–11 Genomic DNA was extracted from whole blood according to a standard salting-out procedure, as previously reported.6–11
Two-point and multipoint parametric linkage analyses were performed with the Merlin software13 program. We assumed an affected-only model with a disease allele frequency of 0.0001 and penetrance of 0.9. In addition to traditional LOD scores, a heterogeneity LOD score resulting from a test of linkage in the presence of genetic heterogeneity was calculated. A genome-wide linkage map of common informative markers was constructed by identifying all SNVs present from the exome data that overlap with known SNVs in 60 unrelated Europeans from the International HapMap. SNVs with minor allele frequency <1% or mendelian errors within the HapMap were excluded. The remaining markers were then pruned with the PLINK software14 using pairwise r2<0.1 in sliding windows of 50 SNVs, moving in intervals of 5 SNVs. This resulted in a final exome-wide marker set of 4601 SNVs. Marker allele frequencies for linkage analysis were determined by the frequency of each SNV in the Exome Sequencing Project (ESP) in the relevant ethnically matched population, either European (n=3499 individuals) or African American ancestry (n=1864 individuals).
Exome Sequencing and Analysis
Exome sequencing was performed at the University of Washington Genome Science Center across 17 families (48 individuals) with NimbleGen V2 in solution capture and Illumina HiSeq. Sequences were aligned with the Burrows-Wheeler Aligner,15 and realignment and single-nucleotide and insertion-deletion variants were called with GATK version 1.4 at the Hussman Institute for Human Genomics. Vcf files were then imported into an in-house database, Genomes Management Application (GEMapp), to facilitate storage, variant annotation, querying, and analysis. In addition to our exome data, GEMapp was used to store transcriptome data from the left ventricle of 4 unrelated individuals, 2 with DCM and 2 unaffected individuals, as previously published.11 This allowed us to filter variants mapped to genes expressed in heart tissue.
Using GEMapp, we queried each family to determine putative disease–causing variants that met the following criteria: variants with read depth ≥5 and quality scores ≥40; variants that were missense, nonsense, splice site, or a coding insertion or deletion; variants shared across all affected members of a family; variants with a frequency <0.5% in 5400 exomes from the Exome Variant Server (EVS); variants that have either a Phastcons16 score >0.4 or a Genomic Evolutionary Rate Profiling17 score >2; and variants with expression in our heart transcriptome data set with Reads Per Kilobase per Million mapped reads >3. These criteria were defined from our previous work on 197 variants in known DCM genes published as disease causing andthat was analyzed by our group.18 We also excluded filtered variants thatwere present in all 48 exomes and variants occurring in >1 family that did not segregate with disease status in at least 1 other family.
Copy Number Variation
Copy number variation in 48 DCM exomes was also assessed by the Structural Variant Working Group at the University of Washington using CoNIFER (Copy Number Inference From Exome Reads; http://conifer.sourceforge.net/).19 A total of 200 non-DCM exomes and 48 DCM exomes were used. Singular value decomposition transformation was used to remove systematic bias, removing 8 components. The final singular value decomposition -ZRPKM signal was then smoothed, and the duplication/deletion break points were found using a threshold of ±1.5 singular value decomposition -ZRPKM.
Sanger Sequencing Validation
All variants passing filter criteria and occurring within TTN were validated with Sanger sequencing and run on a 3130xl, as previously published.11 Primer sequences are shown in Table I in the online-only Data Supplement. Additional DNA samples (n=29) from affected and unaffected family members were also sequenced for these variants.
A total of 316 exons in TTN were targeted in our exome sequence. To assess the genetic variation at this locus, accounting for the large amount of coding sequence, we used the normalized number of variant sites, θ, as a measure of nucleotide diversity across the 5379 exomes available from the Exome Variant Server. θ was calculated as described by Cargill et al20 for all coding sequence and separately for both missense and truncating variants.
Exome-wide Linkage Analysis Across 17 Families With DCM
The maximum LOD score across the genome and the LOD score at the TTN locus are shown for each family in Table 1. In those families with TTN truncating variants identified by exome sequencing analysis, the maximum LOD score achieved across the genome for each family was comparable to the LOD score at the TTN locus (Table 1). The highest multipoint peak within the genome fell in the region spanning chromosome 9q21.11-q22.31 (hg19:71,862,987–95,840,256), producing a heterogeneity LOD score of 1.74. Overall, in the 17 families, the TTN locus was the second highest (heterogeneity LOD=1.59; Table 2).
Exome Sequence Analysis From 17 Families With DCM
Our criteria for putative DCM variants identified in the exome sequences were based on defined criteria (Methods) and as previously described.11 The number of shared variants meeting these criteria present in each family ranged from 1 to 80 (average, 28.1; median, 24). We had previously reported that of the 197 variants already published as causative of DCM, 16% were present in 2400 exomes from the ESP, and of those with functional data (and therefore presumed to be pathogenic variants of very low frequency), the median frequency in the ESP population was 0.04%.18 When this maximum 0.04% frequency criterion was applied to the present exome analysis, the number of shared filtered variants per family ranged from 1 to 49 (average, 16.8; median, 15). Copy number variant analysis using exome data did not identify any shared rare variants (frequency <1% in the ESP data set) across these families.
A total of 6 TTN truncating variants (2 frameshift, 3 nonsense, and 1 splice variant that occurred in 2 DCM families) were identified among the filtered candidates in 7 of our 17 families (41%; Table 3). Our approach to identifying which of these truncating variants were likely disease causing within these families was to genotype them in additional DNA samples in the extended families when possible to assess segregation of the variant with disease; we also consider them in the context of the additional shared variants in our filtered lists for each family. We further observed from our linkage data that those families with highly negative LOD scores at the TTN locus had no TTN truncating variants that passed our exome analysis filtering criteria. In addition, we screened against presence in the 1000 Genome data (which are independent of the EVS data set). None of the 6 truncating variants were present in this data set.
Clinical Characteristics Families With DCM and Segregating TTN Variants
Family A had DNA samples available for 3 additional members. Sanger sequencing showed that all 6 family members were heterozygous for the truncating variant. Subject III.3, a woman who carried the TTN variant, died at 69 years of age with mild systolic dysfunction (ejection fraction, 42%) but without left ventricular enlargement, having suffered a myocardial infarction in her 50s, and thus confounding assessment of whether the TTN variant, the myocardial infarction, or both contributed to her systolic dysfunction. Two subjects (IV.3, V.1), both mutation carriers in their 20s, had no evidence of DCM.
In addition to the 3 samples that had exome sequencing, family B had DNA samples available from 6 other family members. Sanger sequencing confirmed the nonsense variant as present in all affected family members. A female obligate carrier (II.2) died of cancer at 76 years of age without a cardiovascular history. Another female obligate carrier (II.5) had no cardiovascular history at 70 years of age. A male who carried the TTN variant at 69 years of age (II.6) had only borderline systolic dysfunction without left ventricular enlargement.
Only 3 DNA samples were available for this family and were used for exome sequencing. Subject III.3, who died of DCM, was an identical twin by family history and thus may have been an obligate carrier. Another obligate carrier (III.3) had no known DCM but according to the death certificate died of ventricular tachycardia and coronary artery disease. None of the additional 12 variants passing filtering criteria (8 under more stringent filtering of population frequency <0.05% in the ESP exome data set) occurred in known DCM or other cardiomyopathy-associated genes.
This family of European ancestry carried the same TTN splice variant identified in family C. Of the additional 32 variants also identified as putative disease causing in this family, only 1 variant occurred in a known DCM gene, a missense variant in TTN, chr2:179,410,975, NM_133378.4, Gly29127Arg.
This family carried a single base insertion in TTN, resulting in a frameshift mutation. Of the additional 26 segregating variants also identified in this family, the only variant in a gene with a reported association with hypertrophic cardiomyopathy occurred in SOS1.21 One of the affected children (II.2) carried a rare variant in MYBPC3 inherited from his mother (I.2), previously reported by us as likely disease causing 9 but suggested to be of unknown significance on the basis of a subsequent study22; we also note that this variant is present at a frequency of 0.12% in the EVS, making it more common than many DCM rare variants.18
A 4-bp deletion in TTN resulted in a frameshift mutation. DNA was available from 1 additional affected member and 3 unaffected family members, and the TTN variant was confirmed to be present in all 4 affected members by Sanger sequencing and was not present in the 3 unaffected members. None of the other 23 segregating variants identified occurred in known cardiomyopathy-associated genes.
No DNA samples beyond those used for exome sequencing were available to assess segregation in this family. Of the additional 23 variants that segregated, none occurred in other cardiomyopathy-associated genes.
Segregating Missense Variants at the TTN Locus
We also considered the implication of segregating TTN missense variants passing our exome-filtering pipeline as a class of variants that were not discussed in the recent TTN study.1 Determination of pathogenicity of these variants will be extremely challenging because of the large number of coding exons. In the 5400 ESP exome data sets, the average number of TTN missense variants per individual was 23.3, ranging 6 to 55. These results were comparable to those observed in the exome sequences of our DCM families, with the average number of TTN missense variants per DCM individual at 22.75 (range, 11–43). Application of our DCM filtering criteria to missense variants in the 5400 ESP exomes (frequency <0.5% and either a PhastCons score >0.4 or a GERP score of >2) resulted in an average of 1.91 missense variants per individual (range, 0–23). Five TTN missense variants passed our exome-filtering approach (that segregated with all individuals affected with DCM in a family): 1 each in 2 DCM families who also had segregating truncating variants (Table 3) and the others in 2 families, each with high-quality candidates in known cardiomyopathy genes, so they were not further prioritized. The average number of TTN missense variants without regard to sharing, that is, an analysis of only 1 individual from each of the 17 families, a less stringent approach and similar to the analysis of all missense TTN variants conducted for the EVS, was 1.88.
Nonsegregating, Truncating Variants
Given the observed excess of TTN truncating variants in both familial and in sporadic DCM versus controls,1 we thought it also relevant to report the number of nonsegregating truncating variants in TTN identified in the exome sequences of these 48 individuals with DCM, because they may be potential susceptibility variants. We observed 2 nonsegregating TTN truncating variants that were validated with Sanger sequencing. First, a C insertion at hg19 chr2:179,426,992, generating a frameshift in 1 of 3 family members with DCM who underwent exome sequencing (family 14; Table 1) and a nonsense variant at hg19 chr2:179,605,218, NM_003319.4 Gln3885stop in 2 of 3 family members (family 17; Table 1). Neither variant was observed in the 5400 ESP exome sequences or in the 1000 Genomes data, making them potential susceptibility variants. In the case of the frameshift variant, this family had already been shown to segregate a variant published as disease causing accompanied by functional data,6 and in family 14 with the nonsense variant, a total of 43 segregating variants were identified by our exome-filtering pipeline (Table 1), none of which were in previously published cardiomyopathy genes.
Nucleotide Diversity of TTN
The NimbleGen V2 in solution capture target included 315 discrete exons from 6 TTN transcripts (NM_001256850.1, NM_133432.3, NM_133378.4, NM_003319.4, NM_133437.3, and NM_133379.3), totaling 110 459-bp coding sequence. Our exome pipeline identified TTN truncating variants in 7 DCM families. This could simply be a result of the large number of exons. Hence, we investigated the nucleotide diversity at the TTN locus in a non-DCM population. The EVS contains annotated exome sequence from 5379 individuals at the TTN locus, totaling 2425 and 25 discrete missense and nonsense variants, respectively. Across 5379 EVS individuals, there are a total of 125 575 missense alleles and 77 nonsense alleles, averaging 23 and 0.014 per individual, respectively. We calculated the normalized number of variant sites, θ,20 accounting for sample size and the number of coding bases in the EVS individuals at the TTN locus to be 2.23×10−3 and 2.3×10−5 for missense and nonsense variants, respectively. The same calculation across 5 other known DCM genes (MYBPC3, TNNC1, TNNI3, MYH6, and TPM1) in the EVS data gave results comparable to those observed in TTN (for missense variants, θ ranged from 6.9×10−4 at TPM1 to 1.05×10−3 at TNNC1 and for truncating variants, from 0 at TNNC1 and TNNI3 to 1.29×10−4 for MYBPC3), suggesting that the excess of shared truncating variants in our DCM families is not attributable to the large number of exons alone and that nucleotide diversity at TTN is comparable to that of other known DCM genes.
This is the first independent replication study of TTN truncating variants as frequently involved in the pathogenesis of familial DCM. Herman et al1 recently identified TTN truncating mutations in 25% of familial DCM and 18% of sporadic DCM, a significant excess compared with 3% of controls. The authors concluded that truncating mutations in TTN are a frequent cause of DCM because all previous reports of unselected patients with DCM of unknown cause ranged from <<1% to 5% to 8%.4 However, 3% of controls in the Herman et al1 study also were observed to have TTN truncating variants, suggesting that the interpretation of specific TTN truncating variant pathogenicity would be challenging, especially in simplex cases. Analysis across 19 DCM families, segregating rare TTN truncating variants in the Herman et al1 study, yielded a combined LOD score of 11.1, providing strong evidence that the truncating variants in those families were pathogenic. However, in that study, TTN was sequenced in isolation, so the relevance of the linkage evidence in the TTN region could not be compared with the rest of the genome.
We hypothesized that if rare variants in TTN indeed account for one quarter of familial DCM, this locus should also be detected using an unbiased genome-wide linkage approach across our 17 DCM families because they should be enriched for causative variants at this locus, especially because the DCM families in this study were selected for exome sequencing, because they were already known to be point mutation negative for 16 other known DCM genes6–11 (except for 1 family segregating a previously described variant6 in a gene attributing ≈0.5% of DCM). Genome-wide linkage analysis yielded the second most significant evidence of linkage at the TTN locus compared with other regions in the genome, which we interpret as evidence of the TTN locus in DCM pathogenesis.
Next, we identified those nonsense, missense, splice, and frameshift variants in the exome sequences meeting our filtering approach that included conservation and myocardial expression, which segregated with DCM affection status in each family. Seven of 17 families (41%) had segregating TTN truncating variants identified in their filtered exome variants. Using common informative SNVs within the exome sequences, we obtained a combined LOD score at the TTN locus for these 7 families of 2.99, and in each family, the maximum LOD score at the TTN locus was either the maximum LOD score achieved in that family across the whole exome or comparable to the maximum observed LOD score at any other locus. We interpret these data as replication of TTN truncating variants as frequently linked with DCM.
Despite the previous evidence1 and our findings presented here, all of which collectively support the concept that TTN truncating variants are highly relevant for DCM pathogenesis, determining the pathogenicity of any specific variant remains extremely challenging. We interpret the 2 truncating variants not shared by all affected family members in 2 families (families 14 and 17) as unlikely to be causative of DCM. We also note that in the 7 families in whom all those affected with DCM carried a truncating variant, some unaffected members at older ages also carried the same truncating variant. This observation confounds pedigree analysis, although it is consistent with reduced penetrance, which is commonly observed with familial DCM. Furthermore, the plethora of TTN missense variants observed in all individuals, whether from control or DCM cohorts, further complicates TTN variant interpretation. The available evidence from the ESP data has shown that most individuals will carry numerous TTN missense variants, some even very rare, and even if such variants segregate with DCM in a family, it may occur as a play of chance. This concept may also apply to truncating variants.
These issues raise 2 central questions of TTN biology in DCM. Which specific variants, whether truncating or missense, play a role in DCM pathogenesis? Do TTN variants include causative and risk alleles? Titin splicing and titin biology are exceedingly complex,23,24 and penetrance is well known to be an incomplete and expressivity variable in familial DCM,4,5 so it is possible if not likely that some TTN variants, whether missense or truncating, may also modulate penetrance and expressivity in DCM. Addressing these questions will require much larger DCM cohorts with detailed phenotypic data, ideally with knowledge of extended family structure (including presymptomatic DCM), genome-wide sequence data, and comprehensive, insightful analysis of the pathophysiological effects of TTN variants.
Although linkage analysis has been used less frequently in the genome-wide association study era, our study highlights the importance of coupling linkage information with sequence data. This provides us both a measure of evidence for a cumulative effect of rare variants, because linkage is not compromised by allelic heterogeneity (ie, multiple rare disease variants within a gene), and an assessment of the evidence in the context of the rest of the genome. Together, linkage analysis and sequencing provide complementary evidence that can improve the efficiency of gene discovery in sequencing studies.
We observed that 41% (7/17) of families carried TTN variants that segregated with DCM, which is higher than the 25% frequency observed in the Herman et al1 study. This most likely resulted from a sample bias in our study because our families were already known to be point mutation negative for 16 other known DCM genes and thus were likely enriched for TTN variants.
We also examined TTN missense variants in our 17 families using an unbiased approach to exome analysis and additional data from >300 exomes with neurological phenotypes collated in our in-house database, GEMapp. Our calculation of nucleotide diversity in the ESP data set for this gene suggested that diversity at the TTN locus was comparable to other known DCM genes relative to the number of coding nucleotides. However, the >300 TTN exons resulted in a very large number of missense variants identified in both the exome sequence from individuals with DCM and individuals in the ESP data set. Two TTN missense variants from 2 DCM families, both also carrying TTN truncating variants, passed our exome-filtering criteria. We considered these 2 missense variants (Gly29127Arg and Ile2685Val) of unknown significance because each met our stringent exome–filtering criteria and were not present in >300 other exomes in GEMapp but occurred in families B and D, who also had TTN truncating variants.
We also note that the most highly linked region in this study on chromosome 9q21.1-q22.31 did not contain any known DCM genes and that none of our filtered genes mapped to this region. The positive linkage at this region could be a chance finding. Alternatively, this could represent a region of the genome containing a novel DCM gene missed by our exome pipeline for 2 possible reasons. Our rare variant exome analysis approach is based on assumptions based on our previous work.18 First, we assumed that causative rare variants were missense, nonsense, splice, or frameshift and that the allele frequency of these variants would be <0.5%. We note here that the majority of known (published) DCM variants are significantly less frequent than 0.5% in the general population,18 but there are examples of known DCM variants with convincing functional data in which the variant frequency is very close to this cut point (CSRP3 Trp4Arg variant25 has a frequency of 0.35% in the European ancestry EVS data set). Although this variant would have been identified in our pipeline, it is possible that other pathogenic variants have frequencies slightly greater than this. Second, our genome-wide linkage approach could also have identified regions containing common susceptibility or modifying variants, again, that would not have been detected by our exome analysis approach. We note recent evidence of linkage to congenital heart defects and low atrial rhythm to this region,26 suggesting the possibility of cardiovascular modifying variants located here.
In conclusion, our data show that TTN was the only gene with implicated rare variants that occurred in multiple DCM families and hence have replicated the previous finding1 that TTN truncating variants contribute frequently to DCM pathogenesis. We reiterate that TTN analysis for DCM causation should be considered within the context of the genome. Whereas interpreting individual TTN truncating or missense variants will remain challenging because of the complexity of TTN biology, the availability of sequencing data from known DCM genes and variants at other exomic loci will assist in categorizing the pathogenicity of these variants.
We thank all the family members who participated, without whom this study would not be a success. We would like to thank the National Heart, Lung, and Blood Institute GO ESP, which produced exome variant calls for comparison; the Lung GO Sequencing Project (HL-102923); the Women’s Health Initiative (WHI) Sequencing Project (HL-102924); the Broad GO Sequencing Project (HL-102925); the Seattle GO Sequencing Project (HL-102926); and the Heart GO Sequencing Project (HL-103010). We would would like to thank the Exome Sequencing Project Family Studies Project Team: Sek Kathiresan, Jay Shendure, Mike Bamshad, Weiniu Gan, Rebecca Jackson, Ani Manichaikul, Christopher Newton-Cheh, Debbie Nickerson, Stephen Rich, Jerry Rotter, and James Wilson.
Sources of Funding
This work was supported by National Institutes of Health awards HL58626 (Dr Hershberger) and HL094976 (Dr Nickerson, Seattle Seq and the Stanley. J. Glaser Award (Dr Norton).
The online-only Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.111.000062/-/DC1.
- Received June 28, 2012.
- Accepted January 31, 2013.
- © 2013 American Heart Association, Inc.
- Hershberger RE,
- Norton N,
- Morales A,
- Li D,
- Siegfried JD,
- Gonzalez-Quintana J
- 12.↵Exome Variant Server, NHLBI Exome Sequencing Project (ESP). http://evs.gs.washington.edu/EVS/. Accessed August 8, 2012.
- Li H,
- Handsaker B,
- Wysoker A,
- Fennell T,
- Ruan J,
- Homer N,
- et al
- Siepel A,
- Bejerano G,
- Pedersen JS,
- Hinrichs AS,
- Hou M,
- Rosenbloom K,
- et al
- Cooper GM,
- Stone EA,
- Asimenos G,
- Green ED,
- Batzoglou S,
- Sidow A
- Norton N,
- Robertson PD,
- Rieder MJ,
- Züchner S,
- Rampersaud E,
- Martin E,
- et al
- Krumm N,
- Sudmant PH,
- Ko A,
- O’Roak BJ,
- Malig M,
- Coe BP,
- et al
- Guo W,
- Bharmal SJ,
- Esbona K,
- Greaser ML
- Patel JR,
- Pleitner JM,
- Moss RL,
- Greaser ML
Rare variants that cause familial dilated cardiomyopathy (DCM) have been discovered in >30 genes but account for only 40% to 50% of cases. Exome sequencing, in which ≈19 000 coding genes in the human genome are sequenced simultaneously, provides a dramatically more powerful approach for DCM gene discovery. To this end, exome sequencing was undertaken in 3 or more affected members in 17 DCM families. Truncating rare variants in TTN, encoding the giant protein titin, were identified in 7 families and segregated with DCM and were not present in 5400 DNA exome sequences from the Exome Sequencing Project or the 1000 Genomes Project. A recent report suggested that rare TTN truncating variants cause up to 24% of familial dilated cardiomyopathy, but in that study, 3% of controls had TTN truncating variants, complicating assignment of pathogenicity to any specific TTN variant, and only TTN was sequenced (so in some cases mutations in other DCM genes that were not sequenced may have been relevant). The data from the present study strengthen the previous conclusion that TTN truncating variants are in general relevant for the pathogenesis of DCM: The genome-wide approach here excluded mutations in other known and novel DCM genes, and linkage analysis supported the pathogenicity of the TTN truncating variants in the 7 families. Nevertheless, 2 other TTN truncating variants did not segregate in 2 DCM families. Overall, our study suggests that TTN truncating variants contribute to the pathogenesis of DCM, but the lack of segregation of all identified TTN truncating variants illustrates the challenge of determining TTN variant pathogenicity even with exome sequencing.