Next generation massively parallel sequencing of targeted exomes to identify genetic mutations in primary ciliary dyskinesia: implications for application to clinical testing

Abstract

Purpose: Advances in genetic sequencing technology have the potential to enhance testing for genes associated with genetically heterogeneous clinical syndromes, such as primary ciliary dyskinesia. The objective of this study was to investigate the performance characteristics of exon-capture technology coupled with massively parallel sequencing for clinical diagnostic evaluation.

Methods: We performed a pilot study of four individuals with a variety of previously identified primary ciliary dyskinesia mutations. We designed a custom array (NimbleGen) to capture 2089 exons from 79 genes associated with primary ciliary dyskinesia or ciliary function and sequenced the enriched material using the GS FLX Titanium (Roche 454) platform. Bioinformatics analysis was performed in a blinded fashion in an attempt to detect the previously identified mutations and validate the process.

Results: Three of three substitution mutations and one of three small insertion/deletion mutations were readily identified using this methodology. One small insertion mutation was clearly observed after adjusting the bioinformatics handling of previously described SNPs. This process failed to detect two known mutations: one single-nucleotide insertion and a whole-exon deletion. Additional retrospective bioinformatics analysis revealed strong sequence-based evidence for the insertion but failed to detect the whole-exon deletion. Numerous other variants were also detected, which may represent potential genetic modifiers of the primary ciliary dyskinesia phenotype.

Conclusions: We conclude that massively parallel sequencing has considerable potential for both research and clinical diagnostics, but further development is required before widespread adoption in a clinical setting.

Figure 1. Family segregation analysis supports the possibility of digenic inheritance of DNAH11 and DNAH2 mutations in patient 998

A patient with a phenotype consistent with PCD (II-1) was previously found to have a single mutation in DNAH11, which encodes a component of the outer dynein arm. Targeted exome sequencing found no additional possible disease-causing mutations in DNAH11 but identified a heterozygous missense mutation of a highly conserved residue of DNAH2, which encodes a component of the inner dynein arm. Family segregation analysis revealed that the DNAH11 mutation was inherited from the patient’s father (I-1), while the DNAH2 mutation was inherited from the patient’s mother (I-2). Furthermore, the three unaffected siblings were heterozygous for only the DNAH2 allele (brother II-2), heterozygous for only the DNAH11 allele (sister II-4), or wild-type for both alleles (brother II-3).

Figure 2. Sequence analysis favors the existence of a heterozygous insertion in patient 1205

A. Local alignment of selected sequence reads from patient 1205 in the region containing a single nucleotide insertion (hg18; chr5:13,754,404 – 13,754,447). The top line of the alignment represents the NCBI 36 reference sequence at this location. Gaps introduced into the alignment are represented by a “-“. Yellow highlights the stretch of 7 adenine (7A) nucleotides where the insertion should result in 8 adenine (8A) nucleotides. B. The proportion of reads containing 5, 6, 7, 8, 9, or 10 adenine nucleotides reveal a clear difference in patient 1205. Patients 475, 998, and 1072 act as “controls” since they are homozygous for reference 7A alleles. The highest proportions of reads in these samples were 7A, with Gaussian distributions (R² > 0.998) having peaks at approximately 6.75 (6.607, 6.778, and 6.842), consistent with a homozygous 7A genotype at this position. In contrast, the number of adenine residues in the sample from patient 1205 ranged from 5 to 10, with a Gaussian distribution (R² = 0.964) that was broader than the first three patients and had a peak at 7.54. C. When compared to simulated genotypes, the distribution from patient 1205 most closely resembles the heterozygous distribution. The distributions of the three “control” samples were averaged to obtain a simulated homozygous 7A distribution (SIM-7A). This distribution was shifted to the right to generate a simulated homozygous 8A distribution (SIM-8A). The SIM-7A and SIM-8A distributions were averaged to simulate the expected distribution in an individual heterozygous for 7A and 8A (SIM-7A/8A Het).

Figure 3. Small insertion/deletion variants are observed in distinct proportions of sequence reads

A. Variants were separated by class and whether they were novel or previously reported in dbSNP, and plotted according to the percentage of sequence reads in which the variant was observed. Dashed lines represent 25% and 75% of sequence reads. Substitution variants largely clustered near 50% or 100% of sequence reads as would be expected for heterozygous or homozygous variants, respectively. Known insertion/deletion variants (Ins/Del) also tended to cluster near 50% or 100% of sequence reads. However, novel small insertion/deletion variants had a much wider distribution with a significant proportion of variants detected in fewer than 25% of the sequence reads. B. The depth of coverage, defined by the number of sequence reads in the alignment for a given variant, is shown for the different types of sequence variants. No obvious differences were observed between the classes of variants in terms of the total depth of coverage. C-F. Known substitutions, known insertions/deletions, novel substitutions, and novel insertions/deletions were plotted according to the total read depth and the percentage of reads in which the variant was observed. The novel insertion/deletion variants clearly follow a distinct pattern, with a large proportion of the variants detected in fewer than 25% of sequence reads but no clear relationship between depth of coverage and the percentage of reads in which the variant was observed.