Name: Eileen Dolan
Title: Professor of Medicine, University of Chicago
Education: 1986 — fellowship, biochemistry, the Milton S. Hershey Medical Center; 1983 — PhD, Purdue University, medicinal chemistry; 1979 — BS, University of Dayton, chemistry.
A recent study in the American Journal of Human Genetics has shown that an individual’s origin may influence the way he or she responds to drugs. In the study, researchers from the University of Chicago and Affymetrix used GeneChip Human Exon 1.0 ST arrays to examine 176 HapMap lymphoblastoid cell lines created from blood samples from healthy people from 60 nuclear families, including 30 Caucasian families from Utah and 30 Yoruban families from Nigeria.
From the 17,879 genes targeted initially, the researchers focused on the 9,156 that were reliably expressed in all the samples. Almost 5 percent of those genes were expressed differently in samples from European descent than in the Yoruban samples. On average, the differences in gene expression were approximately 1.26-fold and, based on subsequent experiments, did not seem to stem from copy number variation.
To get a sense of how these findings may impact pharmacogenomic research, as well as the dynamics behind the array-driven project, BioArray News spoke with paper co-author Eileen Dolan last week.
Why did you decide to take on this project, and how does it fit into your research?
Our area of research is pharmacogenomics and identifying genetic variants that predict a patient’s risk of toxicity due to chemotherapy. We had a conversation with John Blume [vice president of RNA development at Affymetrix]. We were working with cell lines from the International HapMap Project and we wanted to enter into collaboration with him to do expression array experiments on all 90 Caucasian and 90 Yoruban HapMap cell lines. We agreed upfront that we would make this data publicly available.
At the time the Human Exon 1.0 ST Array was just coming out. So John proposed to do exon expression array [experiments] on all 180 samples we had and release all the data to us. We prepared the samples, Tyson Clark, [senior scientist at Affymetrix,] ran the expression arrays at Affy, and sent us the data. While we did that, we realized that even though we were interested in pharmacogenomics, we were sitting on a rich source of information for many scientific questions.
One of the global questions we decided to ask was: ’Are there expression differences in cell lines derived from Yoruba people of Ibadan, Nigeria, compared to cell lines from US residents with northern and western European ancestry?’ Because we were using an array that interrogated exons, we could delve deeper and look at exon-level expression and splice variation. Since the International HapMap Project makes the genetic sequences from these individuals publicly available, we could also do a study to determine whether genetic variation is associated with expression variation.
Our laboratory had evaluated the sensitivity of these cell lines to various chemotherapeutic drugs allowing us to look at how those differences in expression play into variation in sensitivity to drugs. This allowed us to build models that consider genetic variation, expression variation, and sensitivity to drugs. It is a very integrated approach. Although this particular paper is focused solely on identifying expression differences between Caucasians and Yorubans, we have also identified genetic variants that act through their effect on expression and are associated with sensitivity to drugs.
Affy manufactures a number of different arrays for gene-expression studies. Why did you decide to use the GeneChip Human Exon 1.0 ST array?
The idea of using the exon array came from John Blume. The exon array has an advantage over the other arrays because the probes are to the exons within the gene, whereas other arrays are to the 3’ end of the gene. A recent report has shown the exon array provides a more accurate measure of the expression level of the gene. In addition, the exon array allows us to look beyond gene level to exon level and evaluate alternative splicing.
Are you using any other technologies besides arrays in your research?
We are using primarily exon array and phenotyping samples for their sensitivity to drugs. We have also performed quantitative RT-PCR on several genes and for those experiments we generally use Applied Biosystems’ TaqMan assays.
Why did you choose the populations you compared, and can you describe the variation you found between the populations?
These are part of the International HapMap samples. There was already a rich set of genetic variation data concerning these samples in the public domain. We had also produced in our lab quite a bit of data looking at sensitivity to drugs in these same cell lines. So it made sense to bring in expression data as well.
In terms of the findings, we discovered that there’s a lot of inter-individual variation in gene expression and there’s also inter-population variation as well. We also found that the inter-population differences are on average about 1.26-fold different among the populations. They are not dramatic but they are very significant in about 5 percent of genes.
Was there any prior work that influenced your research?
There was some literature evidence from a few other groups using different sample sets. Studies at the Wellcome Trust Sanger Institute have been performed under the direction of Manolis Dermitzakis looking at population expression differences. One advantage is that we used International HapMap cell lines with 90 from each population, and they are trio samples – mother, father, and child – so you can do additional genetic studies based on the trio structure.
Why do you consider this variation to be significant?
I think what is important is that there are differences in disease susceptibility and in response to drugs in different populations. Expression differences might explain why a population is more susceptible to a particular disease or why a population is more sensitive to a certain drug. One aspect of that may be due to genetic variation. In other words, if a gene is important in protecting against a particular toxicity and that is expressed to a higher extent in one population, than that population may be protected to a greater extent than another population. This paper provides researchers with information or data that gives them insight into particular genes that may be worth further investigation.
What other causes might factor in this variation? You mentioned copy number variation in the paper.
What we said in the paper was that we did not observe a higher percentage of copy number variation among the 5 percent differentially expressed genes when compared to the whole analysis set. In other words, a majority of the differential genes we identified were not within genomic regions of known copy number variation. Therefore, it is unlikely that copy number variation is a major contributor to the expression differences we observed, though the detailed contribution of copy number variation to the differential expression at the individual level is not clear.
There are a number of variables that can explain differences in expression levels. Some of it is due to differences in allele frequency for some genetic variance. We also appreciate that to some extent expression differences are due to the fact that the Caucasian lines have been around longer than the Yoruban cell lines, so that also may contribute to some of the differences.
What kinds of follow-on studies are you planning and what needs to be done to build on this research?
We are evaluating population differences in alternative splicing and associating genetic variation with expression variation. We are looking into the biological implications of differences in expression and differences in alternative splicing among populations. My main research area is to understand why certain individuals and certain populations are susceptible to toxicity associated with chemotherapy.
We are using these cell lines to develop an unbiased, whole genome approach that identifies ‘pharmacogenetic signatures’ associated with expression and susceptibility to drug toxicity that can be used to reduce a cancer patient’s chance of an adverse event. The follow-up studies to our pharmacogenetic studies include validation of our findings in clinical trials. These studies are challenging since most patients receive multi-agent regimens and you generally need a large number of patients for genotype-phenotype studies.
How can this help other researchers?
First, they could use these same cell lines to measure other cellular phenotypes – any biochemical pathway or phosphorylation of protein – and can use that to identify expression signatures correlated with susceptibility to the cellular phenotype. They can also determine whether genetic variation is associated with their phenotype through its effect on expression.
The second thing is that we have made this data publicly available in [the Gene Expression Omnibus database]. People can look at whether their favorite pathway or gene is expressed in these cell lines and if the expression is associated with genetic variation.