Whole genome sequencing delineates regulatory, copy number, and cryptic splice variants in early onset cardiomyopathy

Fig. 1. Yield of protein-coding and regulatory variants in 209 unrelated childhood CMP cases. a Flow-chart showing the selection process and yield of protein-coding and regulatory variants in the overall cohort and in the gene-elusive subset. Totally, 39% of all cases harbored at least one pathogenic protein-coding variant in a CMP gene; among the remaining 128 gene elusive cases, 15% harbored at least one prioritized high-risk regulatory variant in a CMP gene; and an additional 5% harbored an LoF variant in a new candidate CMP gene. b Pie diagram showing the distribution of protein-coding and regulatory variants in CMP genes and LoF variants in new CMP genes across the cohort (n = 209). WGS identified putatively pathogenic protein-coding SNVs/indels/CNVs in CMP genes in 39% of cases, high-risk variants in regulatory elements of CMP genes in an additional 15% of cases, and loss of function (LoF) variants in candidate genes in an additional 5% of cases. c Variant distribution by CMP subtypes: HCM cases had a higher yield of pathogenic protein-coding variants compared to other CMP subtypes (odds ratio 2.8, CI: 1.5–5.2, p = 7.07 × 10−4). d Variant burden by the patient: 9 cases (4.3%) had…

Fig. 2. Effect of loss of function and copy number deletions in CMP genes on myocardial gene expression. The figure shows LV myocardial gene expression using RNA sequencing in the patient harboring a loss of function or copy number deletion (red dot) compared to other cases without the variant (gray dots) (n = 35 cases). a–c Three pathogenic loss of function variants predicted to result in nonsense-mediated decay of mRNA. Scaled RPKM expression of target mRNA of variants in DSC2 (stopgain), FLNC (splice acceptor), MYBPC3 (frameshift deletion) are below the 25th percentile compared to the remaining cohort; d–f The left panels show the genomic location of three single CNV deletions in CTNNA3, JPH2, NEXN genes. The right panels show scaled RPKM expression of target mRNA below the 25th percentile compared to the remaining cohort. g Location of loss of function variant in NRAP (ENST00000359988) in the discovery cohort (orange dot). gnomAD background density maps of frameshift, splice site, and premature stop variants are shown. h Myocardial NRAP expression: RNA-seq analysis demonstrated low NRAP mRNA expression (<75th percentile) in the LV myocardium of a DCM patient harboring a homozygous…

Fig. 3. Regulatory variant burden in cases (n = 209) and controls (n = 1326). a There was a significant enrichment of high-risk regulatory variants in CMP genes in the cases (orange) compared to controls (blue) (OR 2.25, 95% CI: 1.65–3.07, p = 6.70 × 10−7). b Burden of regulatory variants genes in cases in the discovery and 100,000 Genomes Project cohorts versus controls. Top 4 genes enriched for regulatory variants compared to controls included FKTN (OR = 58.1, CI: 3.1–1083), DTNA (OR = 6.7, CI: 3.0–14.8), DSC2 (OR = 32.0, CI: 1.5–668) and DSG2 (OR = 10.6, CI: 1.4–81). Tier 1 gene and primary CMP gene classifications are denoted by plus symbols. c Replication cohort (n = 1266): scatter plot showed a positive correlation between genes enriched for high-risk regulatory variants in the CMP discovery cohort vs the 100,000 Genomes Project replication cohort (Spearman ρ2 0.555, p = 0.000936) with the top genes being similar in both CMP cohorts (FKTN, DTNA, DSC2, DSG2).

Fig. 4. Target gene and protein expression in the LV myocardium of patients harboring regulatory variants. RNA Seq, qRT-PCR, Western blot, and immunohistochemistry were performed in available LV myocardium from CMP patients (n = 35) to detect mRNA and protein expression of target genes in patients harboring regulatory variants in BRAF, DSP, FKTN, LARGE1, PRKAG2, or TGFB3. For RNA sequencing data, the target scaled RPKM gene expression was compared between the patient harboring the variant (red dot) and the remainder of the cohort (black dots) using boxplots showing median expression for the cohort, 25th and 75th percentiles, and maximum and minimum values (n = 35). For qRT-PCR, Western blot, and immunohistochemistry, target gene or protein expression in the LV myocardium of the patient harboring the variant was compared to wild-type controls including an autopsy sample from an individual without cardiac disease as well as one or more CMP patients that did not harbor any known pathogenic coding or regulatory variants. Three independent experiments were performed per sample with each experiment including three technical replicates per sample. Protein expression level of GAPDH as a house…

Fig. 5. Reporter assays in human iPSC-cardiomyocytes. a Luciferase reporter assay showing the effect of regulatory variants on transcription. The cloned promoter variants of BRAF (chr7:140624223_G/A), DTNA (chr18:32072866_A/G), FKTN (chr9:108319991_A/C, chr9:108320330_G/A), and LARGE1 (chr22:34316416_C/T) reduced luciferase activity compared to reference sequences. The promoter variant of DSP (chr6:7541776_G/A), a second promoter variant of LARGE1 (chr22:34316687_G/A), and an enhancer variant of TGFB3 (chr14:76289218_A/G) significantly increased luciferase activity compared to reference sequences. *p < 0.05 versus reference sequence. All luciferase reporter assays were performed with three biological replicates, each with three technical replicates. b Volcano plot representing the effect of 46 regulatory variants on gene expression using MPRA. Twenty-five variants had significant differences in transcriptional activity between reference and alternative allele (FDR < 0.05, represented by the horizontal black line). Gray = CMP variant activity less than reference allele; black = CMP variant activity more than reference allele. c Totally, 67% of significant variants were associated with…