Integration of chemical-genetic and genetic interaction data links bioactive compounds to cellular target pathways

For the yeast Saccharomyces cerevisiae, each of the 6,000 potential genes characterized by the genome sequencing project has been deleted, identifying 1,000 essential genes and 5,000 viable deletion mutants¹. Testing the viable mutants for hypersensitivity to a target-specific compound identifies a chemical-genetic interaction profile, which, for highly specific compounds, consists of a set of genes that buffer the cell from defects in target activity (Fig. 1a). Because a loss-of-function mutation in a gene encoding the target of an inhibitory compound models the primary effect of the compound^{2, 3, 4}, crossing such a mutation into the set of viable mutants and scoring the resultant double mutants for reduced fitness should generate a genetic interaction profile for the target gene that resembles the chemical-genetic interaction profile of its inhibitory compound (Fig. 1b). We therefore expected that a comprehensive compendium of global genetic interaction profiles would provide a key for deciphering the pathways and targets of growth-inhibitory compounds. To establish this proof of concept, we applied a four-step strategy to link several well-characterized compounds to their target pathways and proteins.

Figure 1. Chemical-genetic interactions can be modeled by synthetic genetic interactions. (a) In a chemical-genetic interaction (at left), a deletion mutant, lacking the product of the deleted gene (represented by a black X), is hypersensitive to a normally sublethal concentration of a growth-inhibitory compound. In a synthetic lethal genetic interaction (right), two single deletions lead to viable mutants but are inviable in a double-mutant combination. Gene deletion alleles that show chemical-genetic interactions with a particular compound should also be synthetically lethal or sick with a mutation in the compound target gene. (b) Comparison of a chemical-genetic profile to a compendium of genetic interaction (synthetic lethal) profiles should identify the pathways and targets inhibited by drug treatment. In this hypothetical figure, chemical-genetic and genetic interactions are both designated by red squares. For example, deletion mutants 3, 5, 6 and 7 are hypersensitive to compound X and a mutation in query gene A leads to a fitness defect when combined with deletion alleles 1, 2, 3 and 4. Here, the chemical-genetic profile of compound X resembles the genetic profile of gene B, thereby identifying the product of gene B as a putative target of compound X. Full Figure and legend (33K)

We screened 4,700 viable yeast deletion mutants for hypersensitivity to 12 diverse inhibitory compounds (see Supplementary Tables 1 and 2 online). These compounds included benomyl, a microtubule depolymerizing agent⁶; FK506 and cyclosporin A (CsA), immunosuppressant drugs that inhibit calcineurin⁷; hydroxyurea, an inhibitor of ribonucleotide reductase⁸; camptothecin, a topoisomerase I inhibitor⁹; fluconazole, an antifungal drug that inhibits Erg11, a cytochrome P450 required for ergosterol biosynthesis^{10, 11}; cycloheximide, an inhibitor of protein synthesis; rapamycin, an inhibitor of TOR kinase signaling¹²; tunicamycin, an inhibitor of protein glycosylation¹³; wortmannin, an inhibitor of phosphatidylinositol kinase signaling¹⁴; sulfometuron methyl, an inhibitor of amino acid biosynthesis¹⁵; and caffeine, an inhibitor of cAMP phosphodiesterases¹⁶. The hypersensitive mutants were identified by arraying strains onto agar plates containing semi-inhibitory concentrations of each compound and scoring reduced colony formation. These putative chemical-genetic interactions were then confirmed by serial-dilution spot assays in a second round of analysis, such that the final data set should contain few false positives.

Figure 2. The set of viable gene deletion mutants were screened for hypersensitivity to each of 12 inhibitory compounds (cyclosporin A, FK506, tunicamycin, sulfometuron methyl, wortmannin, caffeine, rapamycin, fluconazole, camptothecin, hydroxyurea, cycloheximide and benomyl). (a) Two-dimensional hierarchical cluster plot of chemical-genetic profiles. Genes are represented on the horizontal axis and compounds on the vertical axis, with chemical-genetic interactions shown in red. Both compounds and genes are clustered together based upon the similarity of their chemical-genetic interactions. (b) A section of the cluster plot (red bar, labeled 'B' on the horizontal axis of a) is enlarged to highlight gene deletions that lead to benomyl sensitivity specifically. Genes involved in tubulin folding (CIN1, CIN2, CIN4), the prefoldin actin/tubulin chaperone complex (GIM3, GIM4, GIM5, PAC2, PAC10, PFD1, YKE2), the mitotic spindle checkpoint (MAD1, MAD2, BUB3) and tubulin structure (TUB3) showed chemical-genetic interactions with benomyl. (c) A section of the cluster plot (green bar, labeled 'C' on the horizontal axis of a) is enlarged to show the overlap between the CsA and FK506 chemical-genetic interaction profiles. (d) Several different sections of the cluster plot (blue bars, labeled 'D i', 'D ii', 'D iii', 'D iv' on the horizontal axis of a) are enlarged to show the multidrug sensitivity associated with gene deletions in ERG2, ERG3, ERG4, ERG5, ERG6 and PDR5. Full Figure and legend (85K)

The hierarchical clustering revealed a number of genes associated with sensitivity to multiple compounds with diverse modes of action. Drug efflux pumps and members of the ATP-binding cassette protein family, such as Pdr5, Yor1 and Snq1, protect yeast from a range of drugs and confer multidrug resistance when overexpressed¹⁸. Membrane lipid composition also affects drug susceptibility in yeast, as deletion mutants defective for ergosterol biosynthesis have high membrane fluidity and show hypersensitivity to numerous drugs¹⁹. Indeed, we found that pdr5 cells were hypersensitive to cycloheximide, fluconazole and wortmannin and that cells deleted for genes encoding ergosterol biosynthetic enzymes such as ERG3, ERG4 and ERG6 were sensitive to wortmannin, rapamycin, camptothecin, hydroxyurea and cycloheximide (Fig. 2d). We also identified a number of new strains as multidrug sensitive (see, for example, VPH2 in Fig. 2c). To define a multidrug-resistant gene set, we constructed a chemical-genetic interaction network of the 65 genes associated with sensitivity to four or more compounds (Fig. 3). This network was enriched for genes involved in vacuolar protein sorting (VPS16, VPS25, VPS36, VPS67, VAM7, VAM6, STP22, SNF7, DID4, IES6) and genes encoding subunits of the yeast vacuolar H⁺-ATPase complex (VMA2, VMA4, VMA5, VMA6, VMA7, VMA8, VMA10, VMA13, VMA22, PPA1, VMA11, VPH2), a proton pump that maintains the low vacuolar pH (gene data obtained from the Saccharomyces Genome Database, http://www.yeastgenome.org). These findings are consistent with previous observations identifying vma and vps mutants as sensitive to staurosporine, vanadate and hygromycin B²⁰.

Figure 3. To classify multidrug resistance genes, we identified 65 deletion strains that were sensitive to at least four of ten diverse compounds: wortmannin, benomyl, tunicamycin, rapamycin, sulfometuron methyl, fluconazole, cycloheximide, FK506, caffeine and hydroxyurea. The camptothecin chemical-genetic profile was excluded from this analysis because it overlaps extensively with the hydroxyurea profile and the CsA profile was excluded because it overlaps extensively with that of FK506. In the network diagram, edges indicate a chemical-genetic interaction and nodes represent either compounds or genes, with the genes color-coded from a defined subset of GO functional attributes. Full Figure and legend (158K)

Figure 4. Overlap between the chemical-genetic profile of fluconazole and the genetic interaction profile of ERG11. Edges indicate either a chemical-genetic or a genetic interaction and nodes represent either compounds or genes, with the gene nodes color-coded from a defined subset of GO functional attributes. (a) Network of the chemical-genetic interactions with fluconazole and the genetic interactions with ERG11. (b) Venn diagram summarizing that 75 genes showed a chemical-genetic interaction with fluconazole, 27 genes showed a genetic interaction with ERG11 and 13 genes were in the overlap set. Red bracketed numbers indicate the number of genes, in each group, classified as multidrug resistant. Full Figure and legend (84K)

We next examined calcineurin, a highly conserved protein phosphatase required in yeast under a number of environmental conditions such as ionic stress, high pH and exposure to prolonged mating pheromone²⁶. Calcineurin can be inactivated genetically, by deletion of CNB1, which encodes its regulatory subunit²⁷, or chemically, by treatment with the immunosuppressant drugs FK506 or CsA⁷. From SGA analysis with a cnb1 query mutation, we identified 3 of the 4 nonessential genes known to be synthetically lethal with CNB1 (refs. 17,28; FKS1, CUP5, VPH2) and identified 35 new synthetic genetic interactions (see Supplementary Table 4 online). Comparison of the CNB1 synthetic genetic interactions with the FK506 and CsA chemical-genetic interactions showed that they largely overlap (Fig. 5a). Of the 38 genes that interacted genetically with CNB1, 23 were hypersensitive to both FK506 and CsA and a further 5 were hypersensitive to FK506 alone.

Figure 5. Overlap between the chemical-genetic profiles of FK506 and CsA and the genetic interaction profile of CNB1. Edges indicate either a chemical-genetic or a genetic interaction and nodes represent either compounds or genes, with the gene nodes color-coded from a defined subset of GO functional attributes. (a) Network of the chemical-genetic interactions of FK506 and CsA and the genetic interactions with CNB1. (b) Venn diagram summarizing that 35 genes showed a chemical-genetic interaction with FK506, 28 genes showed a chemical genetic interaction with CsA and 38 genes showed a genetic interaction with CNB1, with 24 interactions common to all three screens. Red bracketed numbers indicate the number of genes, in each group, classified as multidrug resistant. Full Figure and legend (83K)

For all three drugs, the overlap between the chemical-genetic and genetic interaction profiles was found to be highly significant when compared to random data sets. For example, it is highly unlikely that one would obtain the observed overlap between the genetic profile of ERG11 and the chemical profile of fluconazole by chance (P = 3.8 10^-56). Discrepancy between the chemical-genetic and genetic profiles of compounds and their targets could be a result of incomplete inactivation of the target protein function by the drugs or may reflect the inherent differences in genetic versus chemical mechanisms of target inhibition; unlike a gene deletion that eliminates the target protein from the cell, chemical inhibitors form a complex with the target proteins and the inactive complex remains physically present. In particular, genes that show a chemical-genetic interaction with a drug but not a genetic interaction with the drug target could be involved in cellular import or export of compounds or may provide insight into drug off-target effects.

Twenty-nine genes found in the fluconazole profile were members of the multidrug-resistant gene set (Fig. 4b, red bracketed numbers) and removal of these interactions increased the overlap with the ERG11 genetic interaction profile. In the case of the FK506 and CsA profiles, 15 genes were members of multidrug-resistant gene set; however, 12 of these also showed a genetic interaction with CNB1 (Fig. 5b), including nine vacuolar H⁺-ATPase (vma) mutants, which reflects an established synthetic lethal relationship^{17, 28}. Because vma mutants do not seem to be identified commonly in SGA screens (ref 5 and A.H.Y. Tong, G. Lesage, G. Bader, H.D., H. Xu, X. Xin et al., unpublished data), this may be a rare example where the chemical-genetic interactions observed for genes within the multidrug-resistant gene set actually reflect a target gene specificity. Nevertheless, even with the 12 multidrug-resistant genes removed from the FK506 and CsA profiles, the overlaps with the CNB1 genetic interaction profile remain highly significant (P = 4.4 10^-53 and P = 2.7 10^-60, respectively).

Figure 6. Two-dimensional hierarchical clustering analysis of chemical-genetic and genetic interaction profiles. (a) Six chemical-genetic profiles (FK506, CsA, fluconazole, benomyl, hydroxyurea and camptothecin) were clustered with 57 genetic interaction profiles. Chemical-genetic or genetic interactions are represented in red. Interacting genes are plotted on the horizontal axis, with the gene cluster tree above. Compounds and query genes are plotted on the vertical axis, with the cluster tree on the outermost left side of the plot. Compounds that cluster with their target genes and pathways are colored similarly. The vertical red bar indicates a set of query genes involved in microtubule-based functions. The vertical orange bar indicates a set of query genes involved in DNA synthesis and repair. Sources of genetic interaction data include this study (CNB1, ERG11-DHFR, TUB2-403); ref. 5 (ARC40, ARP2, BNI1, BIM1, RAD27, SGS1); and (A.H.Y. Tong, G. Lesage, G. Bader, H.D., H. Xu, X. Xin et al., unpublished data) (RVS161, RVS167, SMY1, CLA4, CHS3, CHS7, CHS6, YUR1, PHO85, FKS1, GAS1, RIC1, YPT6, ALG6, DIE2, ALG8, ROT2, BIK1, CIN4, RBL2, TUB3, CIN1, CIN2, MAD2, DYN1, PAC1, NUM1, KAR3, KAR9, GIM3, CTF4, CTF8, POL32, DBF4, CDC2-1, CDC45-1, ELG1, CDC7-1, MMS4, MUS81, RTT107, RAD50, RAD52, RAD24, HOG1, RAS2, SKN1, ERV29). All genetic interaction data are available at http://biodata.mshri.on.ca/grid. (b) Overlap of the chemical-genetic screens of camptothecin and hydroxyurea and genetic interaction screens for the set of query genes involved in DNA synthesis and repair. A section of the cluster plot (orange box in a) is enlarged. VID21 and YBR094W are marked with an asterisk to indicate that these previously uncharacterized genes were identified by both chemical-genetic and genetic screens. Full Figure and legend (94K)

	Top

	Top

Yeast deletion strains derived from BY4741 (MATa his31 leu20 met150 ura30) and generated by the S. cerevisiae deletion consortium¹ were maintained in an ordered array on agar plates at a density of 768 strains (384 unique strains arrayed in duplicate) per plate and manipulated robotically with a colony arrayer (Bio-Rad)⁵.

For the chemical-genetic screens, drugs were added from concentrated stocks to autoclaved rich (YPD; 2% peptone, 1% yeast extract, 2% glucose and 2% agar) or synthetic complete (SC; 0.67% yeast nitrogen base without amino acids, 0.2% amino acid add back, 2% glucose and 2% agar) medium cooled to 50 °C. Most drugs were screened and confirmed by spot assays at the same concentration; some were confirmed at a lower concentration, as noted: FK506 (2 g/ml in SC; gift of M. Cyert and purchased from AG Scientific), CsA (100 g/ml in SC; AG Scientific), hydroxyurea (100 mM in YPD; Sigma), camptothecin (15 g/ml in YPD; gift of S. Brill and purchased from AG Scientific), fluconazole (15 g/ml in YPD; gift of J. Anderson), tunicamycin (5 g/ml and 0.5 g/ml in YPD; AG Scientific), wortmannin (1.3 g/ml in YPD; AG Scientific), caffeine (0.15% in SC; Sigma), rapamycin (0.015 g/ml and 0.01 g/ml in YPD; AG Scientific), benomyl (15 g/ml in YPD; Sigma); sulfometuron methyl (3 g/ml in SC; Sigma) and cycloheximide (0.1 g/ml in YPD; Sigma).

Genome-wide chemical-genetic profiles were generated by robotically pinning arrayed yeast strains onto solid medium containing drug at a semi-inhibitory concentration (a concentration where wild-type yeast growth is slightly compromised as compared to the no-drug control). All plates were incubated at 30 °C. The sensitivity of each of the 4,700 haploid deletion strains to a particular drug was assayed by comparing colony size on drug plates versus a no-drug control. Each chemical was screened against the complete array at least three times; each interaction that was identified at least twice by manual scoring or by computer-based scoring was confirmed using serial-dilution spot assays. For the spot assays, deletion strains were grown overnight to saturation at 30 °C in 130 l of YPD or SC liquid in a 96-well plate. Using a multichannel pipettor, each culture was diluted by five serial 100-fold dilutions and 2 l of each dilution was spotted onto medium containing drug and a no-drug control medium. Plates were incubated at 30 °C for 2−4 d and scored for drug sensitivity. Mutants inviable at all dilutions on drug plates as compared to a no-drug control were designated as strongly sensitive and given a score of 3, mutants viable at only the first one or two dilutions were designated as moderately sensitive and given a score of 2, and mutants that grew at all dilutions, but more slowly, were designated as mildly sensitive and given a score of 1. Approximately 50−80% of interactions identified as putatively sensitive from the array screens were subsequently confirmed with spot assays. For the FK506, CsA, fluconazole and benomyl screens, all deletion mutants identified as synthetic lethal or synthetic sick with the appropriate drug target were also tested for chemical sensitivity by spot assay. All deletion mutants sensitive to camptothecin were also tested for sensitivity to hydroxyurea and all deletion mutants sensitive to hydroxyurea were also tested for sensitivity to camptothecin.

Genome-wide synthetic lethal screens were performed using synthetic genetic array (SGA) analysis as described previously⁵. The MAT starting strains, cnb1::URA3 mfa1::MFA1pr-HIS3 can1 (Y3404), erg11 DHFR-ts::natR can1::MFA1pr-HIS3-MF1pr-LEU2 (Y4830) and tub2-403::URA3 can1::MFA1pr-HIS3-MF1pr-LEU2 (Y4499), were derived from wild-type strain BY4742 (ref. 38). For SGA analysis, first the ordered array of 4,700 MATa XXX::kanR strains, where XXX::kanR designates one of the 4,700 deletion alleles, was crossed to a starting strain; second, the resultant diploids were selected; third, the diploids were pinned onto sporulation medium to induce meiosis; fourth, MATa meiotic progeny were germinated specifically; fifth, MATa XXX::kanR meiotic progeny were selected; and sixth, MATa double mutants were selected. Diploids were selected on SC -Ura-Lys plates or YPD plates supplemented with G418 and clonNAT for 1 d at 30 °C and then sporulated for 5 d at room temperature. In particular, to select for MATa can1::MFA1pr-HIS3::MF1pr-LEU2 meiotic progeny, spores were germinated on SC -His-Arg medium supplemented with L-canavanine for 2 d at 30 °C and then transferred onto a fresh plate of the same medium for another day of growth at 30 °C. The MATa meiotic progeny were then transferred to SC/MSG -His-Arg supplemented with L-canavanine and G418 to select for MATa XXX::kanR meiotic progeny. Finally, cells were transferred to either SC/MSG -His-Arg-Ura supplemented with L-canavanine and G418 or SC/MSG -His-Arg supplemented with L-canavanine, G418 and clonNAT, to select for MATa double-mutant meiotic progeny. MATa double-mutant progeny with growth defects were identified by scoring colony area as compared to that of progeny derived from wild-type control screens. Synthetic sick or synthetic lethal interactions were confirmed by tetrad dissection or random spore analysis.

For random spore analysis, spores were inoculated in 3 ml of liquid haploid selection medium (SC medium lacking histidine and arginine but containing canavanine; SC -His -Arg +canavanine) and incubated at 30 °C for 2 d. The germinated MATa spore progeny were diluted in sterile distilled, deionized water and plated out on medium that selects for the query-gene mutation (SC/MSG -His-Arg +canavanine/clonNAT), the deletion mutant array (DMA) mutation (SC/MSG -His-Arg +canavanine/G418) or both the query-gene and DMA mutations (SC/MSG -His-Arg +canavanine/clonNAT/G418), then incubated at 30 °C for 2 d. Colony growth under the three conditions was compared and the double mutants were scored as synthetic sick (SS), synthetic lethal (SL) or no interaction (No).

Alternatively, spores were resuspended in sterile distilled, deionized water and plated out on the haploid selection medium (SC -His-Arg +canavanine) and medium selecting for the query-gene mutation, the DMA mutation and both the query-gene and DMA mutations, then incubated at 30 °C for 2 d. Colony growth under the four conditions was compared and double mutants were scored as synthetic sick (SS), synthetic lethal (SL) or no interaction (No).

	Top

1. Winzeler, E.A. et al. Functional characterization of the S. cerevisiae genome by gene deletion and parallel analysis. Science 285, 901–906 (1999).
2. Hughes, T.R. et al. Functional discovery via a compendium of expression profiles. Cell 102, 109–926 (2000).
3. Marton, M.J. et al. Drug target validation and identification of secondary drug target effects using DNA microarrays. Nat. Med. 4, 1293–1301 (1998).
4. Hartwell, L.H., Szankasi, P., Roberts, C.J., Murray, A.W. & Friend, S.H. Integrating genetic approaches into the discovery of anticancer drugs. Science 278, 1064–1068 (1997).
5. Tong, A.H. et al. Systematic genetic analysis with ordered arrays of yeast deletion mutants. Science 294, 2364–2368 (2001).
6. Thomas, J.H., Neff, N.F. & Botstein, D. Isolation and characterization of mutations in the -tubulin gene of Saccharomyces cerevisiae. Genetics 111, 715–734 (1985).
7. Liu, J. et al. Calcineurin is a common target of cyclophilin-cyclosporin A and FKBP-FK506 complexes. Cell 66, 807–815 (1991).
8. Rittberg, D.A. & Wright, J.A. Relationships between sensitivity to hydroxyurea and 4-methyl-5-amino-1-formylisoquinoline thiosemicarbazone (MAIO) and ribonucleotide reductase RNR2 mRNA levels in strains of Saccharomyces cerevisiae. Biochem. Cell. Biol. 67, 352–357 (1989).
9. Hsiang, Y.H., Lihou, M.G. & Liu, L.F. Arrest of replication forks by drug-stabilized topoisomerase I–DNA cleavable complexes as a mechanism of cell killing by camptothecin. Cancer Res. 49, 5077–5082 (1989).
10. Turi, T.G. & Loper, J.C. Multiple regulatory elements control expression of the gene encoding the Saccharomyces cerevisiae cytochrome P450, lanosterol 14 -demethylase (ERG11). J. Biol. Chem. 267, 2046–2056 (1992).
11. Truan, G., Epinat, J.C., Rougeulle, C., Cullin, C. & Pompon, D. Cloning and characterization of a yeast cytochrome b5-encoding gene which suppresses ketoconazole hypersensitivity in a NADPH-P-450 reductase-deficient strain. Gene 142, 123–127 (1994).
12. Zheng, X.F., Florentino, D., Chen, J., Crabtree, G.R. & Schreiber, S.L. TOR kinase domains are required for two distinct functions, only one of which is inhibited by rapamycin. Cell 82, 121–130 (1995).
13. Kuo, S.C. & Lampen, J.O. Tunicamycin—an inhibitor of yeast glycoprotein synthesis. Biochem. Biophys. Res. Commun. 58, 287–295 (1974).
14. Cutler, N.S., Heitman, J. & Cardenas, M.E. STT4 is an essential phosphatidylinositol 4-kinase that is a target of wortmannin in Saccharomyces cerevisiae. J. Biol. Chem. 272, 27671–27677 (1997).
15. Falco, S.C. & Dumas, K.S. Genetic analysis of mutants of Saccharomyces cerevisiae resistant to the herbicide sulfometuron methyl. Genetics 109, 21–35 (1985).
16. Parsons, W.J., Ramkumar, V. & Stiles, G.L. Isobutylmethylxanthine stimulates adenylate cyclase by blocking the inhibitory regulatory protein, Gi. Mol. Pharmacol. 34, 37–41 (1988).
17. Garrett-Engele, P., Moilanen, B. & Cyert, M.S. Calcineurin, the Ca²⁺/calmodulin-dependent protein phosphatase, is essential in yeast mutants with cell integrity defects and in mutants that lack a functional vacuolar H⁺-ATPase. Mol. Cell. Biol. 15, 4103–4114 (1995).
18. Bauer, B.E., Wolfger, H. & Kuchler, K. Inventory and function of yeast ABC proteins: about sex, stress, pleiotropic drug and heavy metal resistance. Biochim. Biophys. Acta 1461, 217–236 (1999).
19. Mukhopadhyay, K., Kohli, A. & Prasad, R. Drug susceptibilities of yeast cells are affected by membrane lipid composition. Antimicrob. Agents Chemother. 46, 3695–3705 (2002).
20. Yoshida, S. & Anraku, Y. Characterization of staurosporine-sensitive mutants of Saccharomyces cerevisiae: vacuolar functions affect staurosporine sensitivity. Mol. Gen. Genet. 263, 877–888 (2000).
21. Simon, S., Roy, D. & Schindler, M. Intracellular pH and the control of multidrug resistance. Proc. Natl. Acad. Sci. USA 91, 1128–1132 (1994).
22. Ma, L. & Center, M.S. The gene encoding vacuolar H⁺-ATPase subunit C is overexpressed in multidrug-resistant HL60 cells. Biochem. Biophys. Res. Commun. 182, 675–681 (1992).
23. Drose, S. et al. Inhibitory effect of modified bafilomycins and concanamycins on P- and V-type adenosine triphosphatases. Biochemistry 32, 3902–3906 (1993).
24. Ouar, Z. et al. Inhibitors of vacuolar H⁺-ATPase impair the preferential accumulation of daunomycin in lysosomes and reverse the resistance to anthracyclines in drug-resistant renal epithelial cells. Biochem. J. 370, 185–193 (2003).
25. Dohmen, R.J., Wu, P. & Varshavsky, A. Heat-inducible degron: a method for constructing temperature-sensitive mutants. Science 263, 1273–1276 (1994).
26. Cyert, M.S. Genetic analysis of calmodulin and its targets in Saccharomyces cerevisiae. Annu. Rev. Genet. 35, 647–672 (2001).
27. Cyert, M.S. & Thorner, J. Regulatory subunit (CNB1 gene product) of yeast Ca²⁺/calmodulin-dependent phosphoprotein phosphatases is required for adaptation to pheromone. Mol. Cell. Biol. 12, 3460–3469 (1992).
28. Tanida, I., Hasegawa, A., Iida, H., Ohya, Y. & Anraku, Y. Cooperation of calcineurin and vacuolar H⁺-ATPase in intracellular Ca²⁺ homeostasis of yeast cells. J. Biol. Chem. 270, 10113–10119 (1995).
29. Aasland, R., Stewart, A.F. & Gibson, T. The SANT domain: a putative DNA-binding domain in the SWI-SNF and ADA complexes, the transcriptional co-repressor N-CoR and TFIIIB. Trends Biochem. Sci. 21, 87–88 (1996).
30. Boyer, L.A. et al. Essential role for the SANT domain in the functioning of multiple chromatin remodeling enzymes. Mol. Cell. 10, 935–942 (2002).
31. Bellaoui, M. et al. Elg1 forms an alternative RFC complex important for DNA replication and genome integrity. EMBO J. 22, 4304–13 (2003).
32. Ben-Aroya, S., Koren, A., Liefshitz, B., Steinlauf, R. & Kupiec, M. ELG1, a yeast gene required for genome stability, forms a complex related to replication factor C. Proc. Natl. Acad. Sci. USA 100, 9906–9911 (2003).
33. Ersfeld, K. et al. Characterization of the tubulin-tyrosine ligase. J. Cell. Biol. 120, 725–732 (1993).
34. Shoemaker, D.D., Lashkari, D.A., Morris, D., Mittmann, M. & Davis, R.W. Quantitative phenotypic analysis of yeast deletion mutants using a highly parallel molecular bar-coding strategy. Nat. Genet. 14, 450–456 (1996).
35. Giaever, G. et al. Genomic profiling of drug sensitivities via induced haploinsufficiency. Nat. Genet. 21, 278–283 (1999).
36. Barstead, R. Genome-wide RNAi. Curr. Opin. Chem. Biol. 5, 63–66 (2001).
37. Shi, Y. Mammalian RNAi for the masses. Trends Genet. 19, 9–12 (2003).
38. Brachmann, C.B. et al. Designer deletion strains derived from Saccharomyces cerevisiae S288C: a useful set of strains and plasmids for PCR-mediated gene disruption and other applications. Yeast 14, 115–132 (1998).
39. Breitkreutz, B.J., Stark, C. & Tyers, M. Osprey: a network visualization system. Genome Biol. 4, R22 (2003).
40. Ashburner, M. et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat. Genet. 25, 25–29 (2000).
41. Mewes, H.W., Albermann, K., Heumann, K., Liebl, S. & Pfeiffer, F. MIPS: a database for protein sequences, homology data and yeast genome information. Nucleic Acids Res. 25, 28–30 (1997).
42. Breitkreutz, B.J., Stark, C. & Tyers, M. The GRID: the General Repository for Interaction Datasets. Genome Biol. 4, R23 (2003).

	Top

Competing interests statement:  The authors declare that they have no competing financial interests.