CAG-encoded polyglutamine length polymorphism in the human genome

Abstract

Background: Expansion of polyglutamine-encoding CAG trinucleotide repeats has been identified as the pathogenic mutation in nine different genes associated with neurodegenerative disorders. The majority of individuals clinically diagnosed with spinocerebellar ataxia do not have mutations within known disease genes, and it is likely that additional ataxias or Huntington disease-like disorders will be found to be caused by this common mutational mechanism. We set out to determine the length distributions of CAG-polyglutamine tracts for the entire human genome in a set of healthy individuals in order to characterize the nature of polyglutamine repeat length variation across the human genome, to establish the background against which pathogenic repeat expansions can be detected, and to prioritize candidate genes for repeat expansion disorders.

Results: We found that repeats, including those in known disease genes, have unique distributions of glutamine tract lengths, as measured by fragment analysis of PCR-amplified repeat regions. This emphasizes the need to characterize each distribution and avoid making generalizations between loci. The best predictors of known disease genes were occurrence of a long CAG-tract uninterrupted by CAA codons in their reference genome sequence, and high glutamine tract length variance in the normal population. We used these parameters to identify eight priority candidate genes for polyglutamine expansion disorders. Twelve CAG-polyglutamine repeats were invariant and these can likely be excluded as candidates. We outline some confusion in the literature about this type of data, difficulties in comparing such data between publications, and its application to studies of disease prevalence in different populations. Analysis of Gene Ontology-based functions of CAG-polyglutamine-containing genes provided a visual framework for interpretation of these genes' functions. All nine known disease genes were involved in DNA-dependent regulation of transcription or in neurogenesis, as were all of the well-characterized priority candidate genes.

Conclusion: This publication makes freely available the normal distributions of CAG-polyglutamine repeats in the human genome. Using these background distributions, against which pathogenic expansions can be identified, we have begun screening for mutations in individuals clinically diagnosed with novel forms of spinocerebellar ataxia or Huntington disease-like disorders who do not have identified mutations within the known disease-associated genes.

Figure 1

Relationship between length of longest uninterrupted CAG-tract and Q-tract length variance. (A) All targets. HD Q-tract length variance from Andres et al. [26]. Correlation = 0.62, not including ATXN3. (B) Higher resolution view of targets with Q-tract length variance < 4.0. Dashed lines at 10 CAG and 0.79 variance represent the cutoff for identifying candidate genes for polyglutamine expansion disorders. See text for list of genes falling in this area.

Figure 2

Example distributions of normal Q-tract lengths. (A) ATXN3, ataxin 3 (B) RAI1, retinoic acid receptor 1.

Figure 3

Functional classification of CAGpolyQ genes: shared Gene Ontology term analysis. Known disease genes are marked with a 'D', candidate disease genes are marked with a 'C' and genes with invariant Q-tracts (Table 1) are marked with an 'I'. Clusters of genes are labeled with the GO terms that best described each cluster. GO terms shared by gene pairs are listed in Additional file 7 and Additional file 8. Genes not represented in a graph either had no annotation under that GO namespace or did not share a GO term with a score above the 99^th percentile. (A) Biological process. Genes not represented: ARID1B, ATXN1, ATXN2, BRD4, C9ORF43, DCP1B, HD, DENND4B, FRMPD3, MAML2, PAXIP1L, PHC1, PHLDA1, SOCS7, THAP11, TNRC15, TNRC6A, TNRC6B and TNS. (B) Molecular function. Genes not represented: ATN1, ATXN1, ATXN3, BRD4, C14ORF4, C9ORF43, CHERP, DCP1B, KCNN3, DENND4B, FRMPD3, MAML2, NUMBL, PAXIP1L, PCQAP, PHLDA1, SOCS7, ST6GALNAC5, TNRC15, TNRC6A, TNRC6B and TNS.