Perturbational Gene-Expression Signatures for Combinatorial Drug Discovery

Area of Science:

• Oncology
• Genomics
• Pharmacology

Background:

• Cancer survival depends on both mutated (oncogene) and non-mutated (non-oncogene) genes, termed addictions.
• Overcoming drug resistance in cancer requires effective combination therapies, but identifying synergistic drug pairs is difficult.
• Current methods often focus on known drug-protein targets, limiting the discovery of novel combinations.

Purpose of the Study:

• To develop a scalable, gene-expression-based approach for predicting synergistic drug combinations in cancer.
• To identify novel therapeutic strategies targeting both oncogene and non-oncogene addictions.
• To validate the efficacy of predicted drug combinations experimentally.

Main Methods:

• Inferred recurrent perturbation-transcript regulatory relationships from a large compendium of chemical and genetic perturbation experiments.
• Utilized gene expression data across multiple cell lines to identify these relationships.
• Predicted synergistic drug pairs based on identified transcript-level recurrences.
• Experimentally validated predicted drug combinations in vitro.

Main Results:

• Identified transcript-level recurrences distinct from known compound-protein targets.
• These recurrences were reproducible in external datasets and correlated with small-molecule sensitivity.
• Two unexpected synergistic drug combinations were experimentally confirmed in vitro.
• The gene-expression approach successfully predicted effective combination therapies.

Conclusions:

• A gene-expression-based strategy can effectively identify synergistic drug pairs for cancer therapy.
• This approach facilitates combinatorial drug screening and aids in targeting non-mutated genes.
• The findings support a new paradigm for developing combination therapies to overcome cancer drug resistance.