A chemical genomics approach to understanding drug action Guri Giaever Stanford Genome Technology Center, 855 California Avenue, Palo Alto, CA 94304, USA The complete collection of yeast deletion strains rep- resents a unique, living biological computer for under- standing gene function. The molecular ‘barcodes’ present in each of the deletion strains allow a quantitat- ive ranking of the importance of any gene under any experimental condition of choice. In this article, some of the recent results generated from experiments that exploit the yeast deletion collection to understand mechanisms of drug action are discussed. Identifying the precise mechanism of action of a compound is a challenging and essential task in biochemistry and drug discovery. Although numerous target-directed bio- chemical screens have resulted in the identification of an abundance of compounds for which the target – compound interaction is well characterized, cell-based assays are required to: (i) assess the efficacy of compounds within cells; and (ii) guide compound optimization in drug development. It is often very difficult to correlate global cellular phenotypes with molecular mechanisms. An assay that could align biochemical and molecular data would be useful for both drug discovery and mechanism of action studies. One such tool is the yeast deletion collection, which consists of , 21 000 haploid strains (including both MATa and MATa mating types) and heterozygous and homozygous diploid strains of Saccharomyces cerevisiae that each possess a precise deletion (from the start to the stop codon) in one of , 6000 yeast genes [1,2]. Thus, in any experimental screen this collection of ‘6000 genomes’ can be assayed and all genes required for growth can be identified. The ease of molecular manipulation of yeast, combined with its level of gene conservation with humans (40 – 50%), makes S. cerevisiae a powerful model organism to identify genes involved in drug responses. In this article, the findings of several groups who have used this collection to understand the consequences of perturbing a cell with a variety of agents [3–9] are discussed. Looking for what we know: screening compounds that target DNA Many anticancer therapies act by damaging DNA. Under- standing the molecular changes that increase sensitivity to chemotherapeutic agents will aid the development of new therapies and assist the determination of the therapy of choice for distinct tumors. Several groups have employed the yeast deletion collection to study the effects of DNA damage [4–8]. Bennett et al. used a conventional approach whereby the homozygous deletion strains are screened individually for their sensitivity to ionizing radiation by pinning the strains in a 96-well format onto agarose plates and exposing them to treatment. The strains are then analyzed for sensitivity by visual inspection. Although informative, such a methodology suffers from a lack of sensitivity to subtle changes because the screens are performed on solid media rather than in liquid culture. Confirmation of sensitivity was performed by replica plating of serial dilutions of cells followed by irradiation. Bennett et al. identified 130 genes that when deleted increased sensi- tivity to ionizing radiation; 107 of these genes were previously unknown. A somewhat surprising result in these studies was the finding that when 20 of the most sensitive diploid strains were examined for radiation sensitivity in the haploid parent, only four strains exhibited sensitivity. Furthermore, the MATa parents were more resistant than the MATa parents. Further experiments are required to determine if this difference is due to the difference in mating type [10] or the higher level of aneuploidy (the gain or loss of individual chromosomes) in diploids [11]. Bennett and colleagues characterized these strains further by examining their sensitivities to ultraviolet (UV) radiation and the DNA-damaging agents bleomycin, methyl methanesulfonate (MMS), hydroxyurea and camp- tothecin, and surveyed the effects of gene deletions on recombination, replication and checkpoint functions. This study uncovered the usual suspects of DNA repair-related genes, but also revealed sensitive genes that reside in functional categories such as mitochondrial activity and cell wall maintenance. It will be fascinating to understand the relative importance of the unpredicted role of such gene products in the DNA damage response. Two other groups [4,6,7] took advantage of the ‘molecular barcodes’ (20-nucleotide sequences that uniquely identify each deletion mutant) incorporated into each deletion strain and assayed the strains in parallel. In these studies, the collection was pooled into one culture, subjected to treatment and allowed to recover. Cells were collected over time, genomic DNA prepared and the molecular barcodes amplified using polymerase chain reaction (PCR). The abundance of each strain in the pool was then assessed by hybridization of the molecular barcodes to an oligonucleotide array carrying the comp- lement of the barcodes (Figure 1; [1,2]). Strains depleted from the pool after treatment were designated sensitive relative to a control treatment. When two of the datasets Corresponding author: Guri Giaever (ggiaever@stanford.edu). Update TRENDS in Pharmacological Sciences Vol.24 No.9 September 2003 444 http://tips.trends.com