Studies of cell physiology and structure have identified many intriguing proteins that could be analyzed for function by using the power of yeast genetics. Unfortunately, identifying the homologous yeast gene with the two most commonly used approaches-DNA hybridization and antibody cross-reaction-is often difficult. We describe a strategy to identify yeast homologs based on protein function itself. This cloning-by-function strategy involves first identifying a yeast mutant that depends on a plasmid expressing a cloned foreign gene. The corresponding yeast gene is then cloned by complementation of the mutant defect. To detect plasmid dependence, the colony color assay of Koshland et al. [Koshland, D., Kent, J. C. & Hartwell, L. H. (1985) Cell 40,[393][394][395][396][397][398][399][400][401][402][403] is used. In this paper, we test the feasibility of this approach using the wellcharacterized system of DNA topoisomerase II in yeast. We show that (i) plasmid dependence is easily recognized; (ii) the screen efficiently isolates mutations in the desired gene; and (iii) the wild-type yeast homolog of the gene can be cloned by screening for reversal of the plasmid-dependent phenotype. We conclude that cloning by function can be used to isolate the yeast homologs of essential genes identified in other organisms.Many genes have been isolated that encode products essential for eukaryotic cells. The difficulty of genetic analysis in higher eukaryotes makes it attractive to study such genes by using the extraordinary power of yeast genetics. Unfortunately, because of the evolutionary distance between yeast and other experimental organisms, it is often difficult to identify the yeast homolog of a protein from another species by using the standard approaches of DNA hybridization and cross-species antibody reaction.Recent successes with substituting genes from other organisms for yeast genes suggest that an alternative approach will be very productive. Many examples are now available of foreign cDNAs encoding products that substitute fully for yeast proteins inactivated by mutation. For example, the human homolog of a Schizosaccharomyces pombe mitotic control gene, cdc2, was cloned by introducing a human cDNA library into a cdc2'5 strain and then screening for growth at the restrictive temperature (1). Additional examples of such conservation of function through evolution include other cell cycle proteins (2), protein transport components (3, 4), cytochrome c (5), and transcription factors (6, 7). The properties of the proteins in these examples suggest that not only monomer function, but also protein-protein interactions, have been conserved in evolution. The case of cytochrome c is particularly striking, not only because of the complex processing pathway that must be followed to produce a completely functional protein, but also because antisera raised against yeast or mouse cytochrome c do not react with the alternative cytochrome c; furthermore, DNA probes from yeast or rat cytochrome c genes do not cross-hybridize ...