Modeling synthetic lethality

Abstract

Background: Synthetic lethality defines a genetic interaction where the combination of mutations in two or more genes leads to cell death. The implications of synthetic lethal screens have been discussed in the context of drug development as synthetic lethal pairs could be used to selectively kill cancer cells, but leave normal cells relatively unharmed. A challenge is to assess genome-wide experimental data and integrate the results to better understand the underlying biological processes. We propose statistical and computational tools that can be used to find relationships between synthetic lethality and cellular organizational units.

Results: In Saccharomyces cerevisiae, we identified multi-protein complexes and pairs of multi-protein complexes that share an unusually high number of synthetic genetic interactions. As previously predicted, we found that synthetic lethality can arise from subunits of an essential multi-protein complex or between pairs of multi-protein complexes. Finally, using multi-protein complexes allowed us to take into account the pleiotropic nature of the gene products.

Conclusions: Modeling synthetic lethality using current estimates of the yeast interactome is an efficient approach to disentangle some of the complex molecular interactions that drive a cell. Our model in conjunction with applied statistical methods and computational methods provides new tools to better characterize synthetic genetic interactions.

Figure 1

Mechanisms for synthetic genetic interactions proposed by Kaelin [1]. Each node or circle is a multi-protein complex. Synthetic genetic interactions between members of the multi-protein complexes are presented by red lines. Synthetic lethality in loss-of-function alleles can arise from at least four different mechanisms. (a) The cellular organizational units might be uniquely redundant with respect to an essential function; multi-protein complexes A and A' might share paralogues. (b) They might be two sub-units of an essential multi-protein complex: a-c form a sub-complex and synthetically interact with e-h. (c) They might be two interconnected components in an essential linear pathway: each mutation decreases the flow through the pathway. (d) They might participate in parallel pathways that are together essential: one pathway might be needed to compensate for the damage caused by mutations in the other pathway.

Figure 2

Synthetic genetic interactions are not randomly distributed in the interactome. The figure represents the distribution of the interactions (positive edges) observed within multi-protein complexes or between pairs of multi-protein complexes in Tong et al. [10] and in the two different permutation models. The observed data are represented by the green curve and the data derived from the permutation models are shown by the blue and pink curves. The center of the distribution for the observed data has a greater density value than those for the two simulations, meaning that synthetic genetic interactions are not randomly distributed in the interactome but rather cluster within or between pairs of multi-protein complexes.

Figure 3

Within multi-protein complex versus between multi-protein complex synthetic lethal interactions. The kinetochore complexes are composed of sub-complexes that present many synthetic lethal interactions, especially between them. In bold are proteins that are part of either the kinetochore complex [GO:0000776] or the condensed nuclear chromosome kinetochore [GO:0000778]. In italics are the proteins that are not currently listed as part of those kinetochore complexes. The essential genes are labeled in red. The observed synthetic genetic interactions are indicated by a red line. The number associated with a red line indicates the number of synthetic genetic interactions within or between the sub-complexes. Most of the synthetic genetic interactions are between sub-complexes that contain few or no essential genes. Systematic names for these genes are available in Additional data file 1.