Biomedical Discovery Review with DNA Arrays
Transcriptional programs reveal how global gene expression is remodeled during changes in cell growth, physiology, or environment. For example, the transcripRichard A. Young* Whitehead Institute for Biomedical Research Cambridge, Massachusetts 02142 tional program of the yeast cell cycle has been described Department of Biology in detail (Cho et al., 1998; Spellman et al., 1998). Similarly, Massachusetts Institute of Technology a portion of the transcriptional program has been estabCambridge, Massachusetts 02139 lished for Drosophila development during metamorphosis (White et al., 1999). These expression profiles provide a rich source of information for identifying genes that DNA microarray technologies permit systematic ap- may be coregulated and for modeling regulatory mechaproaches to biological discovery that have begun to nisms and networks. have a profound impact on biological research, pharma- DNA array studies have also been designed to identify cology, and medicine. The ability to obtain quantitative genes whose expression depends on a cell state or the functions of specific components of signaling or information about the complete transcription profile of cells promises to be an exceptionally powerful means transcription apparatuses. In many cases, the nature of these genes has led to new insights into biological to explore basic biology, diagnose disease, facilitate drug development, tailor therapeutics to specific pathol- processes. For example, studies designed to identify ogies, and generate databases with information about genes whose expression depends on key components living processes. The new discipline of expression profil- of the transcription initiation machinery in yeast led to the insight that the general transcription apparatus is ing will link many fields of biology and medicine with a physically remodeled when cells encounter limiting nushared knowledge base and a common mathematical trients (Holstege et al., 1998). The set of genes regulated language. However, major conceptual and technical by genome ploidy in yeast revealed how ploidy can challenges lie ahead in this rapidly evolving arena. This influence cell cycle progression and explained why review highlights recent discoveries that come from the greater cell size is associated with higher ploidies (Galituse of microarray technologies and considers the revoski et al., 1999). Specific genes associated with agelutionary changes and challenges that lie ahead in related phenotypes and diseases in humans have been biology. identified, implicating mitotic misregulation in aging (Ly et al., 2000).
DNA Microarray Technologies
The value of such expression profiles has yet to be Several different DNA microarray technologies are curfully realized, even by the investigators who produced rently in use, some examples of which are shown in and published them. This limitation is due to the paucity Figure 1. DNA arrays can be synthesized in situ with of computational tools necessary to analyze large dataphotochemical techniques or with ink jet technology sets, inadequate experience with modeling biological (Marton et al., 1998; Lipshutz et al., 1999). Alternatively, problems and systems using large amounts of data, prefabricated DNA molecules are printed in arrays on and the lack of methods for systematic experimental glass slides or nylon membranes using robots (DeRisi examination of such models. Nonetheless, it is clear that et al., 1997). Several of these technical platforms have expression profiles are remarkably robust signatures of been put to the test with the complete set of yeast genes biological conditions. These signatures are a precise and have produced expression data of sufficient quality molecular phenotype of the cell in a specific state. The for biological discovery. Microarrays produced by each importance and utility of these signatures is illustrated of these technologies will likely accommodate the entire by recent microarray-based studies into gene function population of genes carried within the human genome, in yeast and classification of diseases in humans. but improvements in microarray hardware, experimental design, and computational methods of analysis will need Gene Function and Drug Discovery to be made before the full potential of expression profil- Two issues of interest to biologists who wish to exploit ing is obtained in mammalian cells. These issues will be new genome sequence information have been examined discussed in more detail below. in a paper in this issue of Cell (Hughes et al., 2000). The first issue addressed by this work is whether expression profiles of mutant cells can be used to accurately clasTranscriptional Programs sify the functions of previously uncharacterized genes. DNA microarrays have been used to estimate the levels The second issue explored by these investigators is of RNA species throughout the RNA population of living whether the expression profile of cells treated with a cells and to explore transcriptional programs. The popupharmacological agent can be used to identify the target lation of RNA species in a cell, the transcriptome, has of that drug. The evidence indicates that the function of been estimated for yeast under certain growth condia gene can be accurately predicted from the expression tions and, at least in part, for more complex eukaryotes profile of a cell with a mutation in that gene. The results (Holstege et al., 1998; Lockhart and Winzeler, 2000). also provide an important proof-of-principle for new approaches to pharmacological research and develop-