Next generation sequencing for molecular diagnosis of neurological disorders using ataxias as a model

Abstract

Many neurological conditions are caused by immensely heterogeneous gene mutations. The diagnostic process is often long and complex with most patients undergoing multiple invasive and costly investigations without ever reaching a conclusive molecular diagnosis. The advent of massively parallel, next-generation sequencing promises to revolutionize genetic testing and shorten the 'diagnostic odyssey' for many of these patients. We performed a pilot study using heterogeneous ataxias as a model neurogenetic disorder to assess the introduction of next-generation sequencing into clinical practice. We captured 58 known human ataxia genes followed by Illumina Next-Generation Sequencing in 50 highly heterogeneous patients with ataxia who had been extensively investigated and were refractory to diagnosis. All cases had been tested for spinocerebellar ataxia 1-3, 6, 7 and Friedrich's ataxia and had multiple other biochemical, genetic and invasive tests. In those cases where we identified the genetic mutation, we determined the time to diagnosis. Pathogenicity was assessed using a bioinformatics pipeline and novel variants were validated using functional experiments. The overall detection rate in our heterogeneous cohort was 18% and varied from 8.3% in those with an adult onset progressive disorder to 40% in those with a childhood or adolescent onset progressive disorder. The highest detection rate was in those with an adolescent onset and a family history (75%). The majority of cases with detectable mutations had a childhood onset but most are now adults, reflecting the long delay in diagnosis. The delays were primarily related to lack of easily available clinical testing, but other factors included the presence of atypical phenotypes and the use of indirect testing. In the cases where we made an eventual diagnosis, the delay was 3-35 years (mean 18.1 years). Alignment and coverage metrics indicated that the capture and sequencing was highly efficient and the consumable cost was ∼£400 (€460 or US$620). Our pathogenicity interpretation pathway predicted 13 different mutations in eight different genes: PRKCG, TTBK2, SETX, SPTBN2, SACS, MRE11, KCNC3 and DARS2 of which nine were novel including one causing a newly described recessive ataxia syndrome. Genetic testing using targeted capture followed by next-generation sequencing was efficient, cost-effective, and enabled a molecular diagnosis in many refractory cases. A specific challenge of next-generation sequencing data is pathogenicity interpretation, but functional analysis confirmed the pathogenicity of novel variants showing that the pipeline was robust. Our results have broad implications for clinical neurology practice and the approach to diagnostic testing.

Keywords: ataxia; autosomal dominant cerebellar ataxia; autosomal recessive cerebellar ataxia; diagnosis; genetics.

Figure 1

Clinical features of cases. (A) Total number of cases analysis with or without pathogenic mutations categorized by age of onset, family history and progression. (B) Percentage of mutation-positive cases analysed by age of onset, family history and progression.

Figure 2

Pathogenicity analysis of H36R in PKRCG. (A) The Histidine at position 36 is located in the first cysteine-rich domain. The equivalent Histidine H101, located in the second cysteine rich-domain is associated with several known spinocerebellar ataxia 14 (SCA14) mutations (Alonso et al., 2005). (B) Comparison of the 3D structure of the cysteine-rich domains in PRKCG shows the located of H36R and other known mutations including H101Y, H101Q and H101R. (C) Comparison of HEK cells expressing wild-type (left) and H36R (right) with visible inclusions in the cells with the mutated SCA14 protein. (D) Susceptibility to aggregation in the H36R mutant compared with wild-type, reported in several SCA14 mutations (Seki et al., 2005). At least 200 cells per transfection were counted in three independent experiments. ANOVA and Bonferroni’s multiple comparison test: wild-type versus H36R P < 0.005; wild-type versus H101Q P < 0.005; wild-type versus G128D not significant; wild-type versus C150F P < 0.005.

Figure 3

Western blot of Case 15. Patient had adolescent onset ataxia with an eye movement disorder, sensorimotor neuropathy, pes cavus, amyotrophy and raised alpha fetoprotein. Lane 3 shows Case 15, with a normal-sized senataxin band. The blot was reported to be normal and the patient highly unlikely to have ataxia with oculomotor apraxia type 2, although in retrospect the band may be of reduced intensity. Two mutations in SETX were detected using the ataxia NGS panel: F1756S and 7100+2 T>C (splice mutation). Also see Supplementary Fig. 2.

Figure 4

(A) Splice predictions for MRE11 c414+4_314+7 show loss of donor site of exon 4 in four splice prediction programs [Alamut version 2.3 (Interactive Biosoftware, Rouen, France)]. (B) Retrospective western blot of Case 37 (Lane 6) showing a slight reduction of MRE11 but a clearer reduction of the interacting proteins hRAD50 and Nbs1, which is consistent with a diagnosis of ataxia-telangiectasia-like disorder caused by mutations in MRE11. Ataxia NGS panel identified two mutations, N117S and a splice mutation 314+4_314+7del. The normal size of MRE11 protein in the patient lane suggests that N117S produces a normally sized but abnormally functioning product, also reported by Stewart et al. (1999).