RNA-interference-based functional genomics in mammalian cells: reverse genetics coming of age Jose Silva 1 , Kenneth Chang 1 , Gregory J Hannon* 1 and Fabiola V Rivas 1 1 Cold Spring Harbor Laboratory, Watson School of Biological Sciences, 1 Bungtown Road, Cold Spring Harbor, NY 11724, USA Sequencing of complete genomes has provided researchers with a wealth of information to study genome organiza- tion, genetic instability, and polymorphisms, as well as a knowledge of all potentially expressed genes. The identification of all genes encoded in the human genome opens the door for large-scale systematic gene silencing using small interfering RNAs (siRNAs) and short hairpin RNAs (shRNAs). With the recent development of siRNA and shRNA expression libraries, the application of RNAi technology to assign function to cancer genes and to delineate molecular pathways in which these genes affect in normal and transformed cells, will contribute signifi- cantly to the knowledge necessary to develop new and also improve existing cancer therapy. Oncogene (2004) 23, 8401–8409. doi:10.1038/sj.onc.1208176 Keywords: RNAi; high throughput screening; func- tional genomics; cancer; apoptosis; synthetic lethality Introduction RNA interference (RNAi) is a conserved biological response discovered in the nematode Caenorhabditis elegans, as a response to double-stranded RNA (dsRNA). Initially demonstrated by Mello and co- workers, who showed that injection of long dsRNA into C. elegans led to sequence-specific degradation of the corresponding mRNAs, this silencing response has been subsequently found in other eukaryotes from yeast (Neurospora crassa and Schizosaccharomyces pombe) to mammals (Fire et al., 1998; Hannon, 2002; Montgom- ery, 2004). Although knowledge of the biological mechanism of RNAi has grown exponentially over the last few years, application of RNAi at the genome-wide level had to await the development of optimal techni- ques of delivery. These were pioneered in model organisms; for example, RNAi can be triggered by soaking C. elegans (Tabara et al., 1998) and Drosophila cells (Clemens et al., 2000) in a solution of dsRNA, or by feeding worms with Escherichia coli expressing gene- specific dsRNAs (Timmons and Fire, 1998). In mam- malian cells, however, long dsRNAs (>30 nucleotides) elicit an antiviral interferon response (Minks etal., 1979; Manche etal., 1992). Thus, RNAi technology could not be applied to mammals until the discovery that short dsRNA duplexes, processed from long dsRNA into 21– 28 cleavage fragments termed small interfering RNAs (siRNAs), were sufficient to trigger gene-specific silen- cing upon transfection into mammalian cells (Elbashir et al., 2001; Harborth et al., 2001). However, the silencing response to transfected siRNAs is transient, lasting from 3 to 7 days depending upon the rate of cell division making this approach unsuitable for analysis of long-term effects of silencing. The search for a more sustained silencing response has resulted in the devel- opment of an additional class of triggers, short hairpin RNAs (shRNAs), that can establish stable gene silen- cing by continuously supplying the RNAi trigger (see for review Paddison and Hannon, 2002). Researchers are now using this technology to understand biological mechanisms in both normal cells and in malignant ones, one of the major goals being to unravel the mysteries of transformation and to improve current cancer therapy. Viewing cancer as a global epidemic of discrete afflictions is a confounding oversimplification. Each cancer is a unique disease arising from multiple genetic alterations, and the particular combination of genes mutated in any given patient probably determines the degree of malignancy and potential therapeutic vulner- abilities of that individual’s cancer. Improved preven- tion, diagnosis, and treatment of cancer in patients, will require a detailed understanding of the specific mole- cular mechanisms that go away in specific cancers. This understanding must be derived both from an examination of the cancerous cell itself and from an investigation of the interactions between the cancer and its host. One of the ways in which these insights can be obtained is through functional genetic approaches in mammals. Traditionally, functional genetic studies are divided into forward or reverse screens. In a typical forward genetic study, genes are mutated at random. The resulting changes in the phenotype of a cell or organism are then attributed to the mutated genes and, by inference, to their protein products. After identification of an abnormal phenotype, the mutations must be mapped, a process which is usually time-consuming and not easily applicable to mammalian systems. Conver- sely, reverse genetic approaches involve the disruption of a gene of interest, so as to determine its function and/ *Correspondence: GJ Hannon, Cold Spring Harbor Laboratory, Watson School of Biological Sciences, 1 Bungtown Road, Cold Spring Harbor, NY 11724, USA; E-mail: hannon@cshl.edu Oncogene (2004) 23, 8401–8409 & 2004 Nature Publishing Group All rights reserved 0950-9232/04 $30.00 www.nature.com/onc