Understanding the causal relationships between pathophysiology of the cell and drug effects on biomolecular targets is imperative in the development of new therapeutics. Construction of useful models of drug effects on disease pathophysiology involves (i) the collection of large and diverse data sets and information on functional connections between cellular components; and (ii) the development and application of computational learning methods to integrate the relevant information and enable prediction of cell behavior. We illustrate both the challenges presented by the development of a combined experimental and computational approach that includes functional analysis, data-driven simulation and experimental validation, and its power to identify critical cellular components in the anticancer activity of drug combinations exhibiting therapeutic synergy.

A novel class of experimental anticancer agents, the Farnesyltransferase Inhibitors (FTIs), have exhibited modest activity in the clinic as single agents, but were found to have promising activity in combination with standard chemotherapy drugs. In particular, FTIs synergize with the microtubule-stabilizing drug Taxol in several in vivo and in vitro models. Furthermore, recent phase I and phase II clinical trials have demonstrated beneficial clinical activity for this combination in chemotherapy of Taxol-resistant cancer patients. We have recently shown that the drug combination inhibits cell growth, enhances tubulin acetylation and induces apoptosis in human cancer cell lines (1). However, the cellular mechanisms underlying the synergistic effects of this promising anticancer drug combination still needs to be understood. To this end, we undertook a collaborative experimental and computational study combining (i) whole-genome transcriptome analysis (ii) measurements of drug effects on biomarkers (iii) utilization of interaction and functional databases for functional and pathway connections, (iv) computational modeling and prediction based on an in silico reverse engineering approach.

Transcriptome changes in cells exposed to the FTI Lonafarnib (LNF) (1, 5, 10 µM) and Taxol (Tx) (2, 5, 10 nM) alone or in synergistic combination (LNF 1 µM + Tx 2, 5 or 10nM) were examined using whole-genome Affymetrix HGU133 Plus 2 microarrays. The observed synergistic transcriptome changes were evaluated in the context of detailed protein-protein interaction maps and molecular networks. On this basis, they were grouped into enriched enzymatic, metabolic and signaling pathways and classified according to functional categories utilizing interaction and functional databases. These analyses were complemented by additional experimental measurements of the effects of drugs on tubulin acetylation, to serve as biomarkers for synergistic drug activity. The various data elements were combined and analyzed with a reverse engineering and forward simulation technology in order to develop an in silico model predictive of drug synergy.

We report on important insights provided by the results regarding the biology of the combined drug effects. Thus, the dose-dependent effects of Taxol or LNF treatment alone were shown to include expression of genes involved in cell death, cell cycle, DNA replication and repair and p53 signaling pathways (LNF had additional effects on genes involved in cellular assembly, organization, and motility). Genes likely to be involved in synergy belong to signaling networks specific to actin and microtubule cytoskeleton, adherens junctions, and receptor tyrosine kinase signaling, indicating the nature of the synergistic effects.

(1) Marcus, A. I., O'Brate, A. M., Buey, R. M., Zhou, J., Thomas, S., Khuri, F. R., Andreu, J. M., Diaz, F., Giannakakou, P. Farnesyltransferase inhibitors reverse taxane resistance. Cancer Research, 2006, 66 (17), 8838-8846.