Genetics of Gene Expression in Alzheimer's Disease (AD)
We postulate that a substantial proportion of the genetic risk for common diseases such AD is due to variants that affect gene expression levels. It has been suggested that polymorphisms in regulatory regions of the genome could substantially account for the genetic basis of complex traits and diseases. It is estimated that the size of the functional non-coding genome may exceed that of the coding genome. Furthermore, the number of heterozygous functional cis regulatory sites exceeds that of heterozygous coding sites. Thus, determining the genetic variation underlying gene expression can enhance our understanding of the genetic basis of complex diseases and traits.
In our preliminary studies, we utilized the genotypes from the published Mayo LOAD GWAS (Carrasquillo et al., 2009 Nature Genetics) and cerebellar expression levels of 12 LOAD candidate genes to identify variants in a strong functional candidate gene, insulin degrading enzyme (IDE), that show significant association with IDE mRNA levels in the cerebellum, as well as association with AD risk (Zou et al., 2010 Neurology). These results strongly suggest that the use of the expression endophenotypes may lead to the identification of functional disease risk variants.
We have two NIH-funded projects that are aimed at pursuing a genome-wide association study of gene expression levels obtained from the cerebellum and temporal cortex of post-mortem AD and control brains. We postulate that these studies may uncover AD risk variants that operate by influencing gene expression levels in the brain. We successfully completed cerebellar gene expression measurements of 24,526 probes on the Illumina WG-DASL arrays in 197 pathologically proven LOAD cases and 177 non-AD subjects.
We were able to detect 13,349 probes successfully in 100 percent of the subjects and >17,000 probes in >75 percent of the subjects. Using the transcriptome data from these >17,000 probes as the phenotype and the 313,530 SNP genotypes from the Mayo LOAD GWAS, we performed expression GWAS in the PLINK software. We only focused on the cis-SNPs defined as those SNPs that reside within the gene encoding the transcript analyzed ± 1cM flanking regions. We are currently in the process of analyzing this data with the following aims:
- To perform GWAS of whole transcriptome expression levels to identify significant cis-SNP/transcript associations
- To identify and validate cis-SNPs that associate with both LOAD risk and gene expression levels.
We have a parallel study aimed at obtaining whole transcriptome expression levels from the temporal cortex mRNA of 198 LOAD and 193 non-AD subjects who also have whole genome SNP genotypes. Discovery of genetic variants that influence gene expression levels in the brain has clear and significant implications, including the potential to uncover the genetic influences on diseases of the central nervous system. Our approach focused on discovery of significant eSNPs with LOAD risk association holds great promise for uncovering novel genetic variants that influence AD risk through their influence on gene expression, as well as the opportunity to test known candidate variants for their role in this important biological function. Our parallel studies in the cerebellum and temporal cortex from the same subjects will also enable important comparisons and validation across different brain regions of ADs and non-AD subjects.