Study: 'Homemade' Gene Expression Technology Unreliable

Otherwise, scientists using DNA microarrays, also known as gene chips, risk having their research results called into question, said Peter Spencer, Ph.D., professor of neurology in the OHSU School of Medicine.

Spencer, director of the OHSU Center for Research on Occupational and Environmental Toxicology, co-authored with several OHSU colleagues one of three articles about microarrays appearing this month in the journal Nature Methods. They show that geographically separated multi-investigator teams adopting common commercial, rather than homemade, microarray platforms and common sets of procedures are able to generate comparable data.

"The important point of the three papers is that with contemporary microarray platforms, we have a relatively reliable method with which to assess gene expression, we can do so reproducibly within an individual laboratory, and we can be confident that a similar result would be obtained if the experiment is repeated elsewhere," Spencer said.

Gene chips contain tens of thousands of tiny droplets containing a cell's whole gene sequences that are laid out on a single microscope slide by fast-moving robotic machines. Scientists determine how the expression of individual genes is turned up or down by placing copies of DNA or RNA molecules labeled with fluorescent dyes on the slide, and examining whether the molecules that bind to a particular gene light up when viewed with a special scanner.

By interrogating thousands of genes at once, scientists can quickly pinpoint genes affected by drugs being tested to treat heart disease, mental illness, infectious diseases and cancer. In the past, researchers were only able to analyze a few genes at once, and they were often uncertain whether these were of greatest importance.

Spencer's group is using the technology in two National Institute of Environmental Health Sciences (NIEHS) centers led by CROET � one with Oregon State University and the Battelle-run Pacific Northwest National Laboratory focuses on mechanisms underlying Superfund chemicals with neurotoxic properties; the other with the OHSU School of Medicine's Department of Pediatrics focuses on neurotoxicogenomics and child health.

Srinivasa Nagalla, M.D., associate professor of pediatrics, and cell and developmental biology, OHSU School of Medicine, led the initial bioinformatics research that supported the Nature Methods publication.

"The gene chip is a revolution in technology that is hoped rapidly to advance understanding of biological mechanisms and methods to assess the actions and effects of drugs and chemicals," said Spencer, who estimates there are "hundreds, if not thousands" of laboratories around the country using DNA microarrays.

Microarray platforms came into widespread use about five years ago. Pioneers in the field constructed their own gene chips, but these have now been replaced by much more reliable commercial platforms that generate highly reproducible data. One of these is used by CROET and another by OHSU's West Campus microarray resource facility.

"It's possible to buy a robot which will take some copies of RNA on a small pin and then write that on a glass slide repetitively until one can build up hundreds of thousands of different spots on this glass array," Spencer said.

But as commercial platforms improved over time, many laboratory or "home-built" platforms have not, he said. "The fruits of the research done on some of the early homemade and early commercial platforms have entered the literature, but the platforms were not reliable because they did not produce reproducible results."

Spencer co-authored the Nature Methods study, titled "Standardizing Global Gene Expression Analysis Between Laboratories and Across Platforms," as part of the Toxicogenomics Research Consortium, a group led by the NIEHS whose members study how the genome is involved in responses to environmental stressors and toxicants.

The three-year, NIEHS-funded study compared a lab-built spotted long oligonucleotide microarray, and a commercially produced long oligonucleotide microarray. The two types were represented among 12 microarray platforms used by seven consortium laboratories, all of which generated data from two standard RNA expression samples, one derived from mouse livers and the other taken from tissues of several mouse organs.

According to the study, reproducibility between platforms and across laboratories was generally poor, but reproducibility between laboratories dramatically increased to acceptable levels when a commercial microarray was used with standardized protocols for labeling the RNA, processing the microarrays, acquiring data and other elements.

"The bottom line is that if you use commercial platforms, you get very good interlaboratory consistency and correlation of data," Spencer said. "We've now entered a new era in which we can move forward confident that we have reliable platforms."

And this is particularly important as scientists delve deeper into the genomes of humans and other animal species in their quest to find cures for a variety of diseases.

"Human genome mapping has made the production of these microarrays possible," Spencer said. "Our interest is in how environmental factors -- drugs, pollutants, workplace substances, natural toxins, food chemicals, fragrance raw materials, other factors -- impact or interact with the genome to produce disease. That's the big task before us."