Gene Explorers

For this article, Steve Bunk interviewed David Botstein, chairman, department of genetics, Stanford University School of Medicine, and Michael B. Eisen, scientist, Lawrence Berkeley National Laboratory, and assistant adjunct professor of genetics and development, University of California, Berkeley. Data from the Web of Science (ISI, Philadelphia) show that Hot Papers are cited 50 to 100 times more often than the average paper of the same type and age. M.B. Eisen, P.T. Spellman, P.O. Brown, D. Botstein, "Cluster analysis and display of genome-wide expression patterns," Proceedings of the National Academy of Sciences of the United States of America, 95:14863-8, December, 1998. (Cited in more than 195 papers) V.R. Iyer, M.B. Eisen, D.T. Ross, G. Schuler, T. Moore, J.C. Lee, J.M. Trent, L.M. Staudt, J. Hudson Jr., M.S. Boguski, D. Lashkari, D. Shalon, D. Botstein, P.O. Brown, "The transcriptional program in the response of human fibroblasts to serum," Science, 283:83-7, Jan. 1, 1999. (Cited in more than 235 papers) Any doubters that the scientist as explorer survives at the molecular level of discovery should witness those who use microarray technology for genomewide studies. In taking the next step beyond sequencing, such researchers are beginning to make sense of all that data, and to present it in graphical ways that other biologists can intuitively grasp. These two Hot Papers represent early examples of how that daunting task is being accomplished, and of how it requires the open-mindedness of pure exploration. "In my view, science really has two phases," comments David Botstein, genetics department chairman at Stanford University School of Medicine. "One is discovery, in which you form hypotheses, and the other is testing of hypotheses in order to validate them." He adds that technology developed primarily in the lab of coauthor Patrick O. Brown, a Stanford associate professor of biochemistry and a Howard Hughes Medical Institute investigator, has enabled his research team to be discoverers. Botstein's former departmental colleague, Michael B. Eisen, now a scientist at Lawrence Berkeley National Laboratory and assistant adjunct professor of genetics and development at the University of California, Berkeley, observes that some initial reaction to these papers was critical of their absence of explicit hypotheses. "They were evaluating the exploratory research as if it were hypothesis-testing research." David Botstein Since then, researchers have increasingly used the microarray hybridization method developed by the team. The process begins with the robotic printing on glass slides of arrays of thousands of discrete DNA sequences. Two cell samples to be studied are labeled with flourescent dyes of different colors, say, red and green. They are then mixed and hybridized with the arrayed DNA spots and measurements are taken of the different colors in each spot. Eisen developed algorithms that can sort through this data to find pairs of genes that behave the most similarly in each experiment, progressively adding others to form clusters of apparently co-regulated genes. (His popular software programs for this process, called Cluster and Treeview, are free to other researchers and can be downloaded at either rana.lbl.gov or www.microarrays.org.) Color tables are created from the expression data that enable biologists to visually discern the patterns. Using this technique, the Stanford group collaborated with researchers from Cold Spring Harbor Laboratory in Cold Spring Harbor, N.Y., on another frequently cited paper that appeared concurrently with these Hot Papers.1 In that study, about 800 genes of the yeast Saccharomyces cerevisiae that are regulated at five different phases of the cell cycle were identified. That is almost twice the number of cell cycle-regulated genes previously discovered. "We thought a great way of introducing this comprehensive technology was to look at all the genes of a class," Botstein explains. In the first of these two Hot Papers, the team used the clustering algorithms to create a genomewide grouping of yeast genes of similar functions. The paper shows that such genes express at the same time in the experiments, providing a way to analyze and present information across the genome. There is strong evolutionary logic behind the idea that genes expressed together will perform similar tasks, says Eisen. "Prior to this work, this had never been demonstrated on a genomewide scale. By identifying and visualizing coexpressed genes, we could see that such genes shared a common cellular role, which imparted biological meaning to coherent features of the data." Although algorithms and graphical displays had been used previously to show that genes of similar function express together under various conditions, these papers represented a leap in scale and scope. Their exploration was systematic and comprehensive, giving order and meaning to voluminous data. In the second Hot Paper, human fibroblast cells were starved of serum for 48 hours, to induce them into quiescence. They then were re-fed serum, and the DNA microarray technology was used to measure changes in messenger RNA levels of about 8,600 genes at 12 different times, from 15 minutes to 24 hours after refeeding. "We thought we were telling these cells to grow by giving them serum, but the cells were interpreting it in a more complex way, thinking it was a wound," Eisen says. Serum, the soluble fraction of clotted blood, is normally encountered by cells in the context of a wound. The team discovered that not only could genes of similar function be clustered according to their temporal patterns of expression, but many genes appeared that are known to participate in wound healing, and numerous other genes were identified that had no known role in either wound repair or fibroblast proliferation. All this suggests a much larger role for fibroblasts in wound healing than previously suspected. The original goal of the experiment was merely to investigate the cell's control of its transition from quiescence to proliferation. But Eisen notes that genomewide microarray studies always seem to take researchers into new territory. "The experiment doesn't let you have a narrow view of what's going on." The Stanford group and Eisen continue to collaborate on another application of microarray hybridization; that of classifying gene expression patterns in human tumors. Botstein cites two studies that appeared in Nature during 2000--one concerning breast tumors2, and the other on B-cell lymphomas3--as good examples of the team's explorations in this genetic terra incognita. Steve Bunk (sbunk@uswest.net) is a contributing editor for The Scientist. References 1. P.T. Spellman et al., "Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization," Molecular Biology of the Cell, 9:3273-97, 1998. (Cited in more than 175 papers) 2. C.M. Perou et al., "Molecular portraits of human breast tumors," Nature, 406:747-52, Aug. 17, 2000. 3. A.A. Alizadeh et al., "Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling," Nature, 403:503-11, Feb. 3, 2000.

Gene Explorers

Interested in reading more?

Become a Member of