Microarrays Have Arrived: Gene Expression Tool Matures

The world of oncology research is slowly discarding the gene and the protein. In their place are genomics and proteomics—the vast collections of genes and proteins in an organism. As evidence of this changing of the guard, the January Oncogenomics Conference in Tuscon, Ariz., expanded its usual focus on microarrays to include genomics and protein research. The bulk of the meeting, however, was a testimonial to the robustness of microarray technology. Microarrays, small glass chips embedded with ordered rows of DNA, allow researchers to compare thousands of genes expressed in one biological sample to those in a second sample. Commercial arrays contain as many as 4,000 genes, and within a year or so that number may grow to 16,000, or about half of the entire human genome. (See sidebar, next page.) Some investigators are using the technology to study basic biologic pathways, while those with more of a clinical bent are looking for genes that might be used for early detection or treatment. One group is trying to identify the cellular targets of the BRCA1 protein by comparing normal cells to cells lacking the BRCA1 gene; another is using arrays to differentiate the function of p53 from its related family members, p63 and p73. More clinically oriented groups are looking for genes expressed in early- and late-stage ovarian tumors or genes that are expressed in prostate tumors and absent in normal tissue. “Everyone is using arrays now, whereas a year ago, there weren’t that many people getting good data,” said Lewis A. Chodosh, M.D., Ph.D., from the University of Pennsylvania School of Medicine, Philadelphia. “Over the past year, microarray efforts went from being very focused on proving that the technology works, to producing volumes of reliable data, to developing higher order analytical methods to make sense of the data. The pace of this transition is extremely impressive, if not overwhelming.” And the results are fascinating. Chodosh and co-workers have been able to identify genes differentially expressed during various development stages of the mouse mammary gland. By looking at the changes in expression of 5,500 genes, his group was surprised to discover that genetic pathways involving fatty acid metabolism, angiogenesis, and extracellular matrix synthesis were involved in mammary developmental regulation. Several clinical studies are beginning to bear fruit. Reproducible patterns of gene expression in tumors make it possible to identify subtypes of tumors as well as identify the cells that contribute to the tumor’s molecular profile. In work from the laboratory of David Botstein, Ph.D., at Stanford University School of Medicine in Stanford, Calif., in collaboration with the Norwegian Radium Hospital in Oslo, microarray analysis of 8,102 genes from 65 primary breast tumor samples showed that the tumors could be classified into four subtypes. The researchers also identified expressed genes commonly seen in normal endothelial cells, B lymphocytes, T lymphocytes, macrophages, and stromal cells. The results appear in the Aug. 17, 2000, issue of Nature. “All the cell types in the primary carcinoma are not there by accident,” said Botstein, who is the chairman of the Department of Genetics. “They are not noise that needs to be eliminated or cut away with a laser. On the contrary, they are part of the picture just as much as many different cell types are part of the picture of a lymph node or a lung or a brain or liver.” The scientists also found that the molecular program of a primary tumor is retained in its metastases. Perhaps more importantly, they found that estrogen receptor-negative breast tumors may consist of at least two biologically distinct subtypes of tumors that may need to be treated as distinct diseases. In a similar analysis with lung cancer tumors, microarray expression identified four distinct types of lung cancer, two of which were adenocarcinomas that appear to correlate with different survival times. Both Chodosh and Christopher Lee, Ph.D., from the Chemistry and Biochemistry and Bioinformatics Interdepartmental Program at the University of California at Los Angeles, pointed out that researchers are still learning how to query the chips in an unbiased way. Ultimately, they think it’s important to be able to look at all the genes in a systematic way to find, as Lee said, “the right, relevant new thing out of the infinite possibilities.” “That will require someone who is very mathematically, statistically savvy who is driven by the data,” said Lee. “Biological systems have high information content and that means you have to compute everything. Groups that have a good interdisciplinary collaboration on very focused problems will probably make the most progress.” While researchers sort through hundreds of expressed genes from microarrays and the 3.1 million bits of data from the genome project, analyzing the proteins in a cell is an even more daunting challenge. In the first place, the 30,000 or so expressed genes seem paltry compared with the 1 to 10 million proteins estimated to be present in human cells. In addition, the current protein technology, 2-D gels and mass spectrometry for isolating and identifying proteins, is technically challenging. Right now, of the 1,000 proteins visible on a typical 2-D gel, about 100 can be identified by mass spectrometry. The other 900 remain a mystery. One of the protein researchers at the conference, Cheryl Arrowsmith, Ph.D., associate professor at the Ontario Cancer Institute in Toronto, works in the area of structural proteomics, the determination of the 3-D protein structures on a genome-wide scale. “The technology in the microarray area is much more mature than in proteins. There’s really no protein equivalent of the cDNA chip,” she said. “But, the technologies will get better, and the results from the microarrays will help us chose which proteins in cancer pathways to focus on.” While it seems clear that cDNA and protein microarrays have the potential to revolutionize the world of oncology, they have also added fresh momentum and energy to basic biology. The possibility of understanding the myriad of molecular connections that take place in a cell seems remotely possible. Chodosh believes that computational modeling will soon take scientists to a point where they can view the cell as a three-dimensional network instead of seeing signaling pathways as two-dimensional—where one thing is perturbed and selected downstream molecules are measured. Researchers will be able to introduce a perturbation into any complex network of molecules and accurately measure the impact on the expression of essentially every gene in the genome. Ruth A. Van Bogelen, Ph.D., who oversees both RNA and protein profiling groups at Pfizer Global Research and Development in Ann Arbor, Mich., is sympathetic to Chodosh’s more global perspective. “What you need to do is pull several fields together—physiology, biochemistry, RNA profiling, protein structure,” she said. “That way we can get much further in understanding what biology is all about.” Some companies have already created computer simulations of biological systems, such as “disease maps” for asthma and obesity; others have modeled viruses and red blood cells. Although these integrative systems appear to be an ultimate goal, even without them the spectacular successes of the sequencing and microarray technologies have created a palpable sense of excitement in the research community. As one participant said, “There’s a lot of enthusiasm that we’re going somewhere. The momentum is incredibly high.” Dr. Lewis A. Chodosh

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.