Exome-wide association study reveals novel susceptibility genes to sporadic dilated cardiomyopathy

Abstract

Aims: Dilated cardiomyopathy (DCM) is an important cause of heart failure with a strong familial component. We performed an exome-wide array-based association study (EWAS) to assess the contribution of missense variants to sporadic DCM.

Methods and results: 116,855 single nucleotide variants (SNVs) were analyzed in 2796 DCM patients and 6877 control subjects from 6 populations of European ancestry. We confirmed two previously identified associations with SNVs in BAG3 and ZBTB17 and discovered six novel DCM-associated loci (Q-value<0.01). The lead-SNVs at novel loci are common and located in TTN, SLC39A8, MLIP, FLNC, ALPK3 and FHOD3. In silico fine mapping identified HSPB7 as the most likely candidate at the ZBTB17 locus. Rare variant analysis (MAF<0.01) demonstrated significant association for TTN variants only (P = 0.0085). All candidate genes but one (SLC39A8) exhibit preferential expression in striated muscle tissues and mutations in TTN, BAG3, FLNC and FHOD3 are known to cause familial cardiomyopathy. We also investigated a panel of 48 known cardiomyopathy genes. Collectively, rare (n = 228, P = 0.0033) or common (n = 36, P = 0.019) variants with elevated in silico severity scores were associated with DCM, indicating that the spectrum of genes contributing to sporadic DCM extends beyond those identified here.

Conclusion: We identified eight loci independently associated with sporadic DCM. The functions of the best candidate genes at these loci suggest that proteostasis regulation might play a role in DCM pathophysiology.

Fig 1. Manhattan Plot of association P-values.

95,499 variants were investigated for association with DCM by logistic regression analysis. Associations are summarized in a Manhattan plot (R/qqman package) which displays the eleven SNVs significantly associated with DCM (Q-values < 0.01) as green dots. Note that the applied logistic model assumed an additive mode of inheritance. For variants on chromosome 15 in the ALPK3 region, a dominant mode of inheritance was better supported by the data (see Table 1 for corresponding P-values)

Fig 2. Forest Plots of odd-ratios (log) in the different populations.

The results show that the associations were largely homogeneous across populations (See also heterogeneity column in Table 2)

Fig 3. Regional plots of association at 8 loci.

P-values, obtained from logistic regression analysis of genotyped and imputed variants in genomic regions demonstrating significant association with DCM are depicted. In each region, the genotyped lead-SNV is identified by its position and the other variants are colored to reflect their LD with the lead-SNV.

Fig 4. BAG3 interacts with HspB7.

(A) GST Pull-Down showing interaction of GFP-BAG3 expressed in HEK293 cells and recombinant GST-HSPB7. GFP-BAG3 was expressed in HEK293 cells (cell extract panel) and and GST-HSPB7 was produced in a bacterial expression system. GST-HSPB7 co-sediment with GFP-BAG3 but not with GFP alone indicating specific BAG3/GST-HSPB7 interaction (Pull-Down panel). (B) Co-immunoprecipitation experiment showing interaction of Flag-HSPB7 and GFP-BAG3 in HEK293 cells. GFP alone or GFP-BAG3 were co-expressed together with Flag-HSPB7 in HEK293 cells (cell extract panel) and subjected to immunoprecipitation with an antibody against GFP. Only GFP-BAG3 immunoprecipitated with FLAG-HSPB7 (IP:GFP panel). Western blottings in (A) and (B) used HSPB7 (for GST-HSPB7), GFP (for GFP and GFP-BAG3), and α-tubulin specific antibodies.