Objectives: To determine whether single nucleotide polymorphisms (SNP) in genes related to glutathione metabolism are associated with autism risk either independently or interactively.
Methods: Glutathione genes were ranked by weighted criteria that addressed each gene’s biological relationship with glutathione, location in autism linkage regions, levels of expression in brain, and degree with which their expression correlated with expression of other glutathione genes. SNPs were then selected for the top scoring genes by selecting tag-SNPs, functional SNPs, and SNPs in conserved regions. Single-SNP odds ratios for variants in 41 glutathione genes were estimated in case-parent trios from AGRE families using the genotype- Transmission Disequilibrium Test (gTDT). Models of multi-SNP interactions and resulting odds ratios were estimated using the trio logic regression method within a conditional logistic regression framework. Logic regression is an adaptive methodology that evaluates whether Boolean combinations of covariates predict risk and selects a final model via permutation testing. We evaluated models with up to four interacting SNPs.
Results: Single-SNP associations were observed for four genes including Cystathionine Gamma-Lyase (CTH), Alcohol Dehydrogenase 5 (ADH5), Gamma-Glutamylcysteine Synthetase, Catalytic subunit (GCLC), and Glutaredoxin 3 (GLRX3). In addition, a three-SNP joint effect was observed for genotype combinations of SNPs in GLRX, GLRX3 and CTH, with an odds ratio of 3.78 (2.36, 6.04).
Conclusions: SNPs in two genes, including CTH and GCLC are directly involved in glutathione metabolism, while ADH5 and glutaredoxins are also involved in anti-oxidation within the glutathione pathway. These results suggest that variation in genes involved in counterbalancing oxidative and nitrosative stress may contribute to autism. This includes marginal effects for some genes as well as interactions between these and other genes along this pathway.