NEW YORK (GenomeWeb News) – Only by understanding the tissue-specific effects of genetic polymorphisms on both gene regulation and splicing will new disease associations come into focus, according to new research appearing online today in PLoS Biology.
A team of Duke University researchers used a genome-wide screen to find interactions between genetic variants, gene expression, and alternative splicing in blood and brain tissue. In doing so, they found extensive between-tissue differences in SNP effects — only about half of the polymorphisms had common effects in both tissues tested. The team is starting to catalogue the data on the effects that specific genetic variants have on gene expression and splicing in various tissues.
Similar experiments have been done in the past, senior author David Goldstein, director of Duke University’s Institute for Genome Sciences and Policy Center for Human Genome Variation, told GenomeWeb Daily News. But those previous studies have focused mainly on immortalized cells. And, the researchers argued, the influence that certain polymorphisms have over gene expression and splicing are poorly understood.
“People have been looking at gene expression and splicing in blood cells, because these are easy to obtain and work with, yet they are trying to ask what the implications might be for diseases that do not affect the tissue they are studying,” Goldstein said in a statement.
For the latest study, he and his team analyzed 93 cortical brain samples collected during autopsies and 80 peripheral blood mononucleated cell samples. They genotyped samples and did a genome-wide screen measuring exon-level expression. By averaging exons, the team was able to estimate transcript expression. They also looked exon-by-exon to determine splicing patterns.
By combining these data, they were able to find both expression quantitative trait loci, or eQTLs, and splicing quantitative trait loci, or sQTLs. On average, they looked at about 40 SNPs for each of the 22,000 or so genes evaluated. And for each gene, they assessed roughly a dozen transcripts and four exons per transcript.
The researchers found drastically different results in the brain and blood tissue. Overall, they identified 929 exon-level associations, which they pruned down by looking for high-confidence associations. Of these, almost three-quarters of the eQTLs and roughly half of the sQTLs seemed to act in only one of the two tissues tested.
In general, the results suggest that it’s not going to be good enough to look at one tissue or at immortalized cell lines and to extrapolate from there. “We really need to look at primary cells — and the right kind of primary cells,” Goldstein said.
“I was surprised because I didn’t expect the differences to be as dramatic as they were between the tissues,” lead author Erin Heinzen, a post-doctoral researcher at Duke, said in a statement. “It really shows that we need to build up a very comprehensive picture of how genetic variation influences gene expression in specific tissues.”
Goldstein, likewise, said he expected to see tissue-specific differences, but not to this magnitude. He and his co-authors noted that the results underscore the value of cataloguing tissue-specific differences in gene variants, gene expression, and gene splicing.
“These results emphasize the importance of establishing a database of polymorphisms affecting splicing and expression in primary tissue types and suggest that splicing effects may be of more phenotypic significance than overall gene expression changes,” they wrote.
For their part, Goldstein and his team have developed a free program called SNP Express for compiling data on relationships between polymorphisms and splicing and expression data across human tissues.
The researchers also looked at the expression and splicing effects associated with 84 polymorphisms identified in GWA studies. They noted that, other than non-synonymous coding SNPs, only a handful of functional mechanisms for published SNPs have been identified so far.
Based on their results, the group found evidence that splicing effects conferred by polymorphisms may be behind at least 13 of the 84 disease loci evaluated.
In the future, Goldstein said that he and his colleagues plan to extend their work to look at specific lymphocyte populations — work that will complement the team’s HIV research. They also plan to use complete whole-genome re-sequencing and cDNA sequencing to look for rare variants and their effects in lymphocytes.
“As the field transitions to the study of rare variants it will be critical to supplement these datasets with complete DNA re-sequencing data to comprehensively characterize the full spectrum of genetic regulation of expression,” the authors concluded.