Predicting selective drug targets in cancer through metabolic networks

Abstract

The interest in studying metabolic alterations in cancer and their potential role as novel targets for therapy has been rejuvenated in recent years. Here, we report the development of the first genome-scale network model of cancer metabolism, validated by correctly identifying genes essential for cellular proliferation in cancer cell lines. The model predicts 52 cytostatic drug targets, of which 40% are targeted by known, approved or experimental anticancer drugs, and the rest are new. It further predicts combinations of synthetic lethal drug targets, whose synergy is validated using available drug efficacy and gene expression measurements across the NCI-60 cancer cell line collection. Finally, potential selective treatments for specific cancers that depend on cancer type-specific downregulation of gene expression and somatic mutations are compiled.

Figure 1

Cancer selectivity and pathway association of predicted growth-supporting genes. (A) Distribution of selectivity scores for the set of 199 predicted growth-supporting genes. (B) Pathway association of the highly cytostatic growth-supporting genes (cytostatic score>0.9), showing for each pathway the number of predicted genes that are known targets of current anticancer drugs, the number of known targets of drugs that are currently used for non-cancer indications and entirely novel gene targets, that is, genes without any currently known drugs that target them. For each pathway, the number of missed predictions, that is, known anticancer drug targets that are not predicted to be highly selective, is also shown.

Figure 2

The relation between drugs' efficacy and the expression of their paired target genes. (A) A schematic illustration of the expected relation between an efficacy of a drug and the expression of genes that have a synthetic lethal interaction with the drug's target. (B) The distribution of efficacy–expression correlation values for the set of predicted synthetic lethal drug-gene pairs, and for a background distribution of random pairs. The predicted synthetic lethal pairs show a significant anticorrelation compared with the random pairs (Wilcoxon P-value=0.02).

Figure 3

Cytostatic scores, synergy and pathway association of predicted target gene pairs. (A) The distribution of predicted drug target pairs across their synergy and selectivity scores. Highly synergistic and selective drug target pairs are colored red, while the remaining ones are colored blue. (B) A network of pathway associations of the predicted synergistic and highly selective drug target pairs (Supplementary Table S4). Node size represents the number of genes that participate in a pathway. Edge width represents the number of synergistic and selective pairs involving genes in the two connected pathways (its nodes).

Figure 4

Predicted synthetic lethal gene pairs in TCA cycle (A), phosphatidylserine biosynthesis (B) and pentose-phosphate pathway (C). Red arrows represent predicted synthetic lethality. In (A) SDH undergoes loss of heterozygosity resulting in a complete loss-of-function in paragangliomas and pheochromocytomas; in (B) PTDSS2, CEPT1 and PCYT2 undergo chromosomal deletions in testicular cancer (PTDSS2), pheochromocytoma (CEPT1) and renal cell carcinoma (PCYT2); in (C) RPE undergo chromosomal deletions in several cancer types, including adrenocortical carcinoma and head and neck squamous cell carcinoma. In all cases, the targeting of the corresponding synthetic lethal partner is expected to selectively affect proliferation of the cancer cells.

Figure 5

Drugs predicted to target specific cancer types based on chromosomal loss of synthetic lethal participant genes. Specific drugs may be effective in treating specific types of cancer, based on our predictions of selective synthetic lethal gene pairs. Cancer types that show a high frequency (in yellow and white) of chromosomal deletions of specific genes are susceptible to drugs inhibiting the genes' synthetic lethal complements. Experimental drugs are followed by an asterisk. In cases where multiple drugs target the same gene, only a single representative drug is shown here with the remaining drugs specified in Supplementary Table S9.