A genetic association study of glutamine-encoding DNA sequence structures, somatic CAG expansion, and DNA repair gene variants, with Huntington disease clinical outcomes

Abstract

Background: Huntington disease (HD) is caused by an unstable CAG/CAA repeat expansion encoding a toxic polyglutamine tract. Here, we tested the hypotheses that HD outcomes are impacted by somatic expansion of, and polymorphisms within, the HTT CAG/CAA glutamine-encoding repeat, and DNA repair genes.

Methods: The sequence of the glutamine-encoding repeat and the proportion of somatic CAG expansions in blood DNA from participants inheriting 40 to 50 CAG repeats within the TRACK-HD and Enroll-HD cohorts were determined using high-throughput ultra-deep-sequencing. Candidate gene polymorphisms were genotyped using kompetitive allele-specific PCR (KASP). Genotypic associations were assessed using time-to-event and regression analyses.

Findings: Using data from 203 TRACK-HD and 531 Enroll-HD participants, we show that individuals with higher blood DNA somatic CAG repeat expansion scores have worse HD outcomes: a one-unit increase in somatic expansion score was associated with a Cox hazard ratio for motor onset of 3·05 (95% CI = 1·94 to 4·80, p = 1·3 × 10^-6). We also show that individual-specific somatic expansion scores are associated with variants in FAN1 (pFDR = 4·8 × 10^-6), MLH3 (pFDR = 8·0 × 10^-4), MLH1 (pFDR = 0·004) and MSH3 (pFDR = 0·009). We also show that HD outcomes are best predicted by the number of pure CAGs rather than total encoded-glutamines.

Interpretation: These data establish pure CAG length, rather than encoded-glutamine, as the key inherited determinant of downstream pathophysiology. These findings have implications for HD diagnostics, and support somatic expansion as a mechanistic link for genetic modifiers of clinical outcomes, a driver of disease, and potential therapeutic target in HD and related repeat expansion disorders.

Funding: CHDI Foundation.

Keywords: DNA repair; Genetic association study; Huntington disease; Somatic expansion.

Fig. 1

Study design and sample exclusions.

Fig. 2

Allelic variation at the HTT exon one repeat locus. The schematic diagram shows the typical reference allele structure of the CAG/CCG repeat region in exon one of HTT (top). We here define the number of pure CAGs as Q¹ and the number of additional downstream glutamine-encoding CAA/CAG repeats as Q². Thus, total encoded-glutamine Q^T = Q¹ + Q². We here define the number of copies of the proline codons before the pure CCG repeat as P¹, the number of pure CCGs as P², and the number of copies of the CCT proline codons as P³. Thus, total encoded-proline P^T = P¹ + P² + P³. The figure shows schematic representations of the atypical alleles observed. Repeat codons are depicted: CAG glutamine codons as red boxes; CAA glutamine codons as green boxes; CCG proline codons as blue boxes; CCA proline codons as yellow boxes; and CCT proline codons as pink boxes. Disease-associated alleles with either zero, three or four downstream glutamine codons (Q² = 0, 3 or 4), are indicated with a red upward triangle, green diamond, or green downward triangle, respectively. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Somatic expansion in Huntington disease. A) The ratio of somatic expansions of the HTT exon one CAG repeat is allele length and age-dependent. The graph shows the ratio of somatic expansions $\left(\frac{\mathit{number}\mathit{of}\mathit{somatic}\mathit{expansion}\mathit{products}}{\mathit{number}\mathit{of}\mathit{progenitor}\mathit{allele}\mathit{products}}\right)$ in blood DNA plotted against the age at sampling for 746 participants in the TRACK-HD and Enroll-HD cohorts. Each point is colour coded with respect to the length of the inherited progenitor allele length (Q¹). The scatterplot also shows the sex- and cohort-adjusted fitted regression lines for each pure CAG repeat length (Q¹) derived from model SEQ¹ which includes an interaction between age at sampling and pure CAG repeat length (Q¹) (Table S2, appendix). B) Somatic expansion scores in blood DNA predict age at onset in Huntington disease. The graph shows adjusted time to event curves from a Cox proportional hazard model of HD motor onset according to the sign of the pure CAG somatic expansion score (determined using model SEQ¹, Table S2, appendix). The solid lines represent the age-dependent probability of HD motor onset: orange for participants with positive expansion scores; and green for participants with negative expansion scores. The light shaded regions represent the 95% CI for the onset probability curves. A vertical tick mark on the curves indicates that a participant was censored at this time. Vertical dashed lines indicate median survival for each somatic expansion score category: individuals with a positive blood DNA somatic expansion score had a median age at onset = 46 years (95% CI 45 to 48); whilst individuals with a negative blood DNA somatic expansion score had a median age at onset = 49 years (95% CI 47 to 50). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Associations between HTT exon one CAG repeat structures and HD clinical outcomes and biases associated with the prediction of HD clinical outcomes based on fragment length estimates of CAG. The adjusted probability curves of age at HD motor onset (A/B) show the probability of motor onset predicted on the basis of the pure CAG length (Q¹, A) or the fragment length estimate of CAG (Q^FL, B) for each number of additional glutamine codons genotype (Q² = 0, red; Q² = 2, black; Q² = 4, green). The light shaded regions represent the 95% CI for the onset probability curves. A vertical tick mark on the curves indicate that a participant was censored at this time. Vertical dashed lines indicate median survival for each Q² genotype. For the pure CAG (Q¹) estimates (A): individuals Q² = 0 had a median age at onset = 38 years (95% CI = 34 to 45); individuals Q² = 2 had a median age at onset = 48 years (95% CI = 46 to 49); and individuals Q² = 4 had a median age at onset = 48 years (95% CI = 45 to 54). For the CAG fragment length (Q^FL) estimates (B): individuals Q² = 0 had a median age at onset = 32 years (95% CI = 29 to 39); individuals Q² = 2 had a median age at onset = 48 years (95% CI = 46 to 49); and individuals Q² = 4 had a median age at onset = 51 years (95% CI = 48 to 59). The scatterplots (C/E) and the whisker plots (D/F) show the ranked TRACK-HD progression score predicted on the basis of the fragment length estimate of CAG (Q^FL, C/D) or pure CAG (Q¹, E/F) dependent on the number of additional glutamine codons (Q²). The progression scores are normally distributed around zero, with zero representing the population average disease progression. Negative values represent individuals with slower than average disease progression and positive values represent individuals with faster than average disease progression. The whisker plots show the mean (diamond) and 95% CI (whiskers) on the basis of the fragment length estimate of CAG (Q^FL, E) or pure CAG length (Q¹, F) dependent on the number of additional glutamine codons (Q² = 0, red; Q² = 2, black; Q² = 4, green). The whisker plots (G/H/I/J) show the least-square means for the rate of change of TMS (TMS rate, G) and TFC (TFC rate, H) predicted on the basis of the pure CAG length (Q¹), and TMS predicted on the basis of the pure CAG (Q¹, G) or fragment length estimate of CAG (Q^FL, H) for each number of additional glutamine codons (Q²) genotype (i.e. the means associated with each number of additional glutamine codons (Q²) genotype adjusted for all other variables in the linear regression).