One of the major projects in my lab is investigating the basic principles of genetic interaction networks. This work stems from the finding that the majority, about 5,000, or about 80 percent, of yeast's approximate 6,000 genes are not required for viability, suggesting that multiple compensatory pathways regulate essential processes. One way to test for and identify these compensatory connections is to make double mutants and examine their phenotypes systematically.
In particular, synthetic lethality defines a relationship in which two different single mutations lead to viable single mutants, whereas combining the same mutations generates a lethal double-mutant phenotype. Identifying synthetically lethal double-mutant combinations highlights genes whose products buffer one another and impinge on the same essential process—defining a functional relationship between the genes and their corresponding pathways. Thus, large-scale mapping of synthetic lethal networks should generate a global map of functional relationships.
My group has established the methodology and infrastructure to generate this global genetic interaction map for yeast cells. We apply an automated approach to yeast genetics, referred to as synthetic genetic array (SGA) analysis, to enable systematic isolation of yeast double mutants and quantitative analysis of their phenotypes. The single mutants we cross to one another were derived from a set of about 5,000 viable yeast deletion mutants and a set of about 1,000 conditional temperature-sensitive alleles of essential genes, which we are constructing.
We developed a quantitative scoring system that enables us to estimate both single- and double-mutant fitness from images of high-density yeast arrays. These measurements allow us to identify both negative genetic interactions, in which the double mutant grows more slowly than expected for the combination of single mutants—synthetic lethality being the most extreme case—and positive genetic interactions, in which the double mutant grows better than expected, such that one mutation often suppresses the phenotype of the other.
We have tested about 5 million gene pairs (roughly 30 percent of all pairs) for synthetic genetic interactions. In doing so, we have identified about 170,000 interactions, with twice as many negative as positive interactions. Whereas some genes show only a few interactions and others, such as essential genes, appear to be highly connected hubs on the network, the average gene shows about 30 negative genetic interactions. The genetic network is highly complex, but the interactions are not random; genes with similar roles tend to interact with one another. Moreover, genes with a highly similar function often share a similar pattern of genetic interactions. We exploited this property to construct a global network, grouping together genes with similar genetic interaction patterns: Nodes in the resultant network represent genes, and edges connect gene pairs that share common sets of genetic interactions or similar genetic interaction profiles (see figure).
This network highlights genetic relationships between diverse biological processes and inherent functional organization of the cell. Genes with tightly correlated profiles form discernible clusters, corresponding to distinct bioprocesses. More detailed analysis of the clusters and the specific genetic interactions that connect the genes enables us to assign genes to specific pathways and to predict the precise function of the gene product.
Genetic Interactions and Genotype to Phenotype
Because genetic interactions are prevalent among deletion alleles in the inbred yeast model system, we anticipate that genetic interactions will also occur among the different alleles of genes carried by a given individual in an outbred population, such that genetic interactions may underlie a significant component of the inherited phenotypes for an individual. Thus, when studying inherited diseases, a major challenge is identifying the potential for a combination of natural genetic variants to modulate the activity of specific disease-associated pathways and assessing the extent to which genetic interactions contribute to the genotype-to-phenotype relationship of an individual.
Because a loss-of-function mutation models the physiological effects of an inhibitory compound that blocks the action of the corresponding deleted gene product, yeast genetic interaction maps are related to yeast chemical-genetic interaction maps, which can be created by scoring the sensitivities of single mutants to bioactive compounds. We have shown that compounds whose chemical-genetic profile resembles the synthetic lethal genetic interaction profile of a specific gene may very well target the product of the identified gene or its corresponding pathway. We are comparing our synthetic lethal and chemical-genetic interaction networks computationally and predicting the target pathways of specific compounds.
Last updated September 26, 2012