MSH3 is a genetic modifier of somatic repeat instability in X-linked dystonia parkinsonism

Abstract

X-linked dystonia parkinsonism (XDP) is a progressive adult-onset neurogenerative disorder caused by the insertion of a SINE-VNTR-Alu (SVA) retrotransposon in TAF1 gene. One element of the SVA is a tandem polymorphic CCCTCT repeat tract whose length inversely correlates with the age of disease onset. Previous observations that the repeat exhibits length-dependent somatic expansion and that XDP onset is modified by variation in DNA repair gene MSH3 indicated that somatic repeat expansion is an important disease driver. Here, we sought to uncover genetic modifiers of CCCTCT instability in XDP patients and to provide a mechanistic link between somatic instability and disease. We determined quantitative metrics of both repeat expansion and repeat contraction in blood. Using genetic association analyses of exome sequencing data, as well as directed sequencing of a variant MSH3 repeat, we found that MSH3 modifies repeat expansion and contraction in blood as well as age at onset. MSH3 alleles associated with earlier disease onset were associated with more expansion and less contraction. Conversely, alleles associated with later disease onset were associated with less expansion and more contraction. Notably, MSH3 repeat alleles were also similarly associated with expansion and contraction in brain tissues. Our findings provide key evidence that MSH3's role(s) in CCCTCT repeat dynamics underlies its impact on clinical disease and indicate that therapeutic strategies to lower or inhibit MSH3 are predicted to both slow CCCTCT expansion and promote CCCTCT contraction, impacting the disease course prior to clinical onset.

Keywords: MSH3; XDP; genetic modifiers; somatic instability.

Figure 1.. Models of CCCTCT somatic instability and age at onset in XDP blood samples.

a-c) Correlations of somatic expansion (a), somatic contraction (b) and age at onset (AAO) (c) with inherited CCCTCT repeat length. d) Linear regression models of somatic instability (model 1 and 2) and AAO (model 3 and 4). CCCTCT instability and CCCTCT repeat length were measured by ABI fragment sizing. n=number of individuals. Each dot represents an XDP individual.

Figure 2.. Exome-wide association analyses for age at onset and CCCTCT somatic instability in blood.

ExWAS for residual AAO, residual somatic CCCTCT expansion and residual somatic CCCTCT contraction in blood, performed on SNVs with minor allele frequency ≥ 0.01, showing the region of the MSH3/DHFR locus in which significant associations were detected. Genomic coordinates are based on GRCh37/hg19. Signal 1 and signal 2 SNVs are indicated in pink and blue respectively and are identified as separate signals based on conditional analyses (Figure S8, Tables S4, S6, S9). Downward-pointing triangles show SNVs associated with a lower residual value (earlier AAO, less expansion, less contraction) and upward-pointing triangles show SNVs associated with a higher residual value (later AAO, more expansion, more contraction). The horizontal dotted lines indicate a genome-wide significance threshold of p-value = 5e-08.

Figure 3.. MSH3 rare coding variants and relationships to AAO, blood expansion and contraction

a) Location of the five rare variants detected (Ala58Val, Arg574Trp, Cys763Phe, Leu911Val, Asp1000Glu; MAF < 0.01) and the common variant Ile79Val specified by signal 1 top SNV rs1650697 mapped onto MSH3 motifs. (b) Three-dimensional (3D) structure of MSH3 (light blue) complexed with MSH2 (dark blue) (=MutSb dimer), modeled with PyMOL software, showing the spatial locations of the rare variants. Ala58Val, located in the unstructured N-terminal domain, is not shown. c) MSH3 rare variants plotted to show their CADD score (x-axis) and AAO residual, expansion residual and contraction residuals on the y-axes. Grey circles in the background represent all data from those individuals in which non-synonymous coding changes were identified in 15 candidate genes (Table S11). NTD, N-terminal domain; MMBD, mismatch-binding domain, CTD, C-terminal domain. Upper and lower horizonal dotted lines show the 90^th and 10^th percentile values respectively for each phenotype. CADD scores were obtained using GRCh37-v.1.6.

Figure 4.. The association of MSH3 tandem repeat variants with age at onset, blood expansion and contraction

a) Schematic of MSH3 tandem repeats detected using MiSeq analyses in 302 XDP blood samples. The repeats are upstream of the DHFR coding region; DHFR shares a promoter with MSH3 and is transcribed in the opposite direction. Each repeat variant is color-coded and the corresponding amino acids are represented. Nomenclature is according to Flower et al. 6d is a variant of 6a resulting from an Ala58Val substitution. b) Frequency of MSH3 tandem repeats detected. c) Residuals of AAO, CCCTCT expansion and CCCTCT contraction for individuals harboring 3a/3a, 3a/6a, 3a/7a, 6a/6a, 6a/7a and 7a/7a MSH3 repeat genotypes. See Table S16 for statistical analyses. Each dot represents an XDP individual.

Figure 5.. MSH3 tandem repeat variants in XDP post-mortem brain tissues.

Residuals of CCCTCT expansion (above dotted lines) and CCCTCT contraction (below dotted lines) in caudate, cerebellum, medial thalamus, temporal pole, cortex BA9 and occipital cortex for individuals harboring 3a/3a, 3a/6a, 3a/7a, 6a/6a, 6a/7a and 7a/7a MSH3 repeat genotypes. See Table S20 for statistical analyses. Each dot represents an XDP individual.

Figure 6.. MSH3 tandem repeat variants alter MSH3 expression levels in the brain

MSH3 mRNA expression levels in caudate, cerebellum, cortex BA9 and occipital cortex for individuals harboring 3a/6a, 6a/6a, 6a/7a and 7a/7a MSH3 repeat genotypes. See Table S22 for statistical analyses. Each dot represents an XDP individual.

Figure 7.. Summary of MSH3 associations

Upward arrows indicate associations with greater residual values (more expansion, more contraction, later onset and downward arrows indicate associations with lower residual values (less expansion, less contraction, earlier onset). Red is consistent with a detrimental effect and green is consistent with a beneficial effect. SNV summaries are based on meeting genome-wide significance in ExWAS data with the exception of blood contraction signal 2 where the p-value for association was 3.48e10⁻⁷ (Tables S3-S6, S8, S9). MSH3 repeat allele summaries are based on nominal significance in regression analyses (Tables S16, S17, S19, S20).