Genetic variation is what makes a person unique. It also underlies susceptibility to a multitude of human disorders, including Alzheimer disease (AD). To what extent does genetic variation control human gene expression? Thanks to recent advances in whole-genome and transcriptome analysis, scientists are beginning to come to grips with this question. In today’s Nature Genetics, a team of researchers led by Amanda Myers at the University of Miami, Florida, reports the first-ever attempt to correlate genomic variation with transcriptome changes in the brain. “This type of analysis could eventually help identify novel biological pathways that are involved in AD,” said Myers, in an interview with ARF.

The project was made possible in part by brain banks funded through the National Alzheimer Coordinating Center (ADCs). Myers, together with colleagues at the NIA, Bethesda, Maryland; the Translational Genome Research Institute in Phoenix, Arizona; and University College London, England, obtained human cortex samples from some 20 ADC brain banks, and carried out whole-genome genotyping and expression analysis on 193 normal control tissues. The samples come from adults aged 65 to 102, who had been age- and sex-matched to AD patients. Whole-genome analysis was carried out on the cortex samples using a gene array chip (Affymetrix GeneChip Human Mapping 500K Array Set) that detects about 500,000 single nucleotide polymorphisms. Gene expression was also quantitated in the same series of samples using a chip-based assay (Illumina HumanRefseq-8 Expression BeadChip), one that quantifies approximately 24,000 different RNA transcripts. The researchers correlated the genotype data with the expression of individual transcripts to link allele dosage with gene expression.

“Since we are the only people doing this type of study in human brain, we were first interested to see what percentage of the transcriptome is expressed in the brain,” said Myers. The researchers found that about 58 percent of the human genome is expressed in the cortex and that the transcription profile of about 20 percent of those varied with genotype. The authors found that some of the genotype-transcript correlations were in cis, meaning that the SNP that correlated with expression was within 1 Mb of the 5’ or 3’ end of the gene. But the vast majority was in trans, or farther away from the gene ends, suggesting those SNPs have indirect effects, such as altering transcription factor levels. The authors discounted interactions that could not be determined to have a linear dosage effect; in other words, out of three possible alleles, for example, AA, AG, and GG, the heterozygote should have an intermediary expression level and the homozygotes should be high and low. This helps to eliminate false positives or spurious correlations.

“The nice thing about our study is that there were two positive controls for proof of principle,” said Myers. Because she had previously carried out genotyping/expression studies on the tau gene, MAPT, the researchers were able to use those findings to validate the data (see Myers et al., 2007; Myers et al., 2005). They found that the whole-genome genotyping/expression correlation matched with haplotypes that lead to higher tau expression. Their data also fit with previous genotype/expression correlates described for human lymphoblasts. SNPs that correlated with increased expression of the ribosomal protein RPS26 in the brain lay very close to SNPs that increase lymphoblast expression of the same gene (see Cheung et al., 2005).

This type of analysis could prove immensely valuable in deciphering the pathology behind common neurologic or psychiatric disorders. For example, when genetic associations are found, this data could be used to predict their effect on mRNA levels. Myers said she is mainly interested in function and in trying to elucidate biological pathways that may be important in Alzheimer pathology. “Just because there is a genetic variation that is enriched in an Alzheimer’s cohort doesn’t necessarily mean that DNA variation is actually causing a change, because it could be genetically inherited with a different variation that is responsible for the change,” said Myers. With the transcriptome analysis, you get more information to help you narrow down what is really going on, she added. The study also begins to test, at a genomic level, the “hypothesis of mass action” that Myers and NIA colleagues Andrew Singleton and John Hardy put forward for neurodegenerative disorders (Hardy is now at University College London). Their premise is that many sporadic neurodegenerative diseases, including amyloidoses, tauopathies, synucleinopathies, and prion-based diseases, are triggered by elevated production of normal proteins (see Singleton et al., 2004).

Myers stressed that this is just a pilot study. The researchers plan to use different DNA chips to search for additional variations, and also to increase the sample size so that fewer common variants might be detected. The researchers have already started working on samples from patients who had AD. That analysis, of 176 samples, is ongoing.—Tom Fagan

Comments

No Available Comments

Make a Comment

To make a comment you must login or register.

References

Paper Citations

  1. . The MAPT H1c risk haplotype is associated with increased expression of tau and especially of 4 repeat containing transcripts. Neurobiol Dis. 2007 Mar;25(3):561-70. PubMed.
  2. . The H1c haplotype at the MAPT locus is associated with Alzheimer's disease. Hum Mol Genet. 2005 Aug 15;14(16):2399-404. PubMed.
  3. . Mapping determinants of human gene expression by regional and genome-wide association. Nature. 2005 Oct 27;437(7063):1365-9. PubMed.
  4. . The law of mass action applied to neurodegenerative disease: a hypothesis concerning the etiology and pathogenesis of complex diseases. Hum Mol Genet. 2004 Apr 1;13 Spec No 1:R123-6. PubMed.

Further Reading

No Available Further Reading

Primary Papers

  1. . A survey of genetic human cortical gene expression. Nat Genet. 2007 Dec;39(12):1494-9. Epub 2007 Nov 4 PubMed.