A Network of Conserved Synthetic Lethal Interactions for Exploration of Precision Cancer Therapy

Abstract

An emerging therapeutic strategy for cancer is to induce selective lethality in a tumor by exploiting interactions between its driving mutations and specific drug targets. Here we use a multi-species approach to develop a resource of synthetic lethal interactions relevant to cancer therapy. First, we screen in yeast ∼169,000 potential interactions among orthologs of human tumor suppressor genes (TSG) and genes encoding drug targets across multiple genotoxic environments. Guided by the strongest signal, we evaluate thousands of TSG-drug combinations in HeLa cells, resulting in networks of conserved synthetic lethal interactions. Analysis of these networks reveals that interaction stability across environments and shared gene function increase the likelihood of observing an interaction in human cancer cells. Using these rules, we prioritize ∼10(5) human TSG-drug combinations for future follow-up. We validate interactions based on cell and/or patient survival, including topoisomerases with RAD17 and checkpoint kinases with BLM.

Figure 1. Study design, quantitative genetic interaction mapping in S. cerevisiae

A Scheme illustrating selection of tumor suppressor genes (TSG) and druggable targets (DT) in S. cerevisiae. B Percent of patients in the TCGA harboring either a somatic mutation (n = 6911) or homozygous deletion (n = 7462) in any of the TSG chosen for screening. Incidence of both somatic mutation and homozygous deletion is higher for the TSG with yeast orthologs included in this study relative to a random set of genes (Inset). P-value was calculated via 1000 random samples; error bars indicate +/− 1 SD. C Deletions of yeast TSG orthologs cause defects in cellular functions and phenotypes associated with human cancer. Significance was assessed using a Fisher’s exact test. DDC, DNA Damage Checkpoint, taken from Gene Ontology (Ashburner et al., 2000). GCR Supp, Gross Chromosomal Rearrangement Suppression, lists (1) and (2) both taken from (Putnam et al., 2012). Mutator supp, Mutator suppression, taken from (Huang et al., 2003). Short lived, taken from (Fabrizio et al., 2010). D For each TSG (x-axis), the plot shows the fraction of druggable genes screened for synthetic lethal interactions in prior studies in yeast (Ryan et al., 2012) (y-axis). For approximately 50% of TSG, fewer than half of relevant interactions had been tested prior to this study (dotted lines). E Number of synthetic lethal (SL) hits per gene for both DT and TSG. See also Figure S1; Tables S1 and S2

Figure 2. Chemo-genetic interaction mapping in a human cancer cell line

A Design of human screen based on the yeast network. B Representative dose response curve for the drug vorinostat. Such a curve was created for each drug to establish IC₂₀ and IC₄₀ doses for screening. Error bars represent +/− SD. C Heat map of chemical-gene interactions, blue represents synthetic-sick/lethal (negative) interaction, yellow represents epistatic (positive) interaction. Interactions highlighted in red are discussed in greater detail in the text. D Cumulative number of interactions identified as a function of the interaction score threshold, highlighting numbers of interactions at 3 and 5 standard deviations (z) below the mean. Recovery of gold-standard interactions of olaparib with BRCA1 and BRCA2 is also shown. See also Figure S2 and S4; Tables S3 and S4.

Figure 3. Conservation between human and yeast

A Evidence of synthetic lethality in yeast, as well as context stability, increases the likelihood of observing a human synthetic-sick/lethal interaction. Gene pairs are ranked (x-axis) by each type of evidence (colored curves); Likelihood score (y-axis) is computed using synthetic-lethal gene pairs identified in the human chemo-genetic screen as a gold standard. B Venn diagram showing number of interactions in CoCaNet (at two stringencies) relative to the number of interactions tested in both species. C Network diagram of top 10% strongest synthetic-sick/lethal interactions (CoCaNet10); square nodes on outside ring represent DT, circular nodes represent TSG. S. cerevisiae gene names are below human gene names in parentheses. Red edges represent interactions previously reported in literature, grey edges are first reported in this study. D Network diagram of top 2% strongest synthetic-sick/lethal interactions (CoCaNet2) organized by gene function. Thickness of edge represents strength of interaction conservation score; arrows indicate direction of edge (DT to TSG). E Likelihood score (for top 10% of yeast gene pairs) is shown for various lines of evidence. See also Table S5.

Figure 4. Validation of cross-species interaction networks for RAD17 and XRCC3

A Network map of all conserved synthetic-sick/lethal interactions in CoCaNet10 for the TSG RAD17. Square nodes represent druggable genes; oval nodes represent drugs used to inhibit these genes. Green edges indicate validation by clonogenic assay. B Sample plate images from clonogenic assay. C Clonogenic assay with TOP1 inhibitor irinotecan in HeLa cells with either stable knockdown of RAD17 or non-targeting (SCR) control. Error bars represent +/− SD, * denotes t-test p < 0.05 at that dose. D Similar clonogenic assay with TOP2 inhibitor etoposide in HeLa cells. E Network map of all conserved synthetic-sick/lethal interactions in CoCaNet10 for the TSG XRCC3 with annotations as in A. F Clonogenic assay with HDAC inhibitor entinostat in LN428 cells, with either stable knockdown of XRCC3 or non-targeting (SCR) control. G Similar clonogenic assay with HDAC inhibitor vorinostat in LN428 cells. H Synthetic genetic array in S. cerevisiae for rpd3Δ, rad57Δ and rpd3Δrad57Δ, p-values as indicated. See also Figure S3.

Figure 5. Clinical potential of deeply conserved interactions

A Kaplan-Meier plot of overall survival, selecting the highest 10% (ISL) or lowest 10% (Non-ISL) of patients in METABRIC ranked by CoCaNet score. B Upper quartile survival for METABRIC cohort stratified by the indicated genetic interaction networks. C Histogram of CoCaNet interactions, binned by the number of patients the ISL group in A whose tumors under-express both of the genes involved in the interaction. D For those TSG interacting with the target of an FDA-approved drug, the number of mutations or deletions seen per patient in TCGA cohort is shown. See also Table S6