Search

A newer generation of anti-cancer drugs targeting underlying somatic genetic driver events have resulted in high single-agent or single-pathway response rates in selected patients, but few patients achieve complete responses and a sizeable fraction of patients relapse within a year. Thus, there is a pressing need for identification of combinations of targeted agents which induce more complete responses and prevent disease progression. We describe the results of a combination screen of an unprecedented scale in mammalian cells performed using a collection of targeted, clinically tractable agents across a large panel of melanoma cell lines. We find that even the most synergistic drug pairs are effective only in a discrete number of cell lines, underlying a strong context dependency for synergy, with strong, widespread synergies often corresponding to non-specific or off-target drug effects such as multidrug resistance protein 1 (MDR1) transporter inhibition. We identified drugs sensitizing cell lines that are BRAFV600E mutant but intrinsically resistant to BRAF inhibitor PLX4720, including the vascular endothelial growth factor receptor/kinase insert domain receptor (VEGFR/KDR) and platelet derived growth factor receptor (PDGFR) family inhibitor cediranib. The combination of cediranib and PLX4720 induced apoptosis in vitro and tumor regression in animal models. This synergistic interaction is likely due to engagement of multiple receptor tyrosine kinases (RTKs), demonstrating the potential of drug- rather than gene-specific combination discovery approaches. Patients with elevated biopsy KDR expression showed decreased progression free survival in trials of mitogen-activated protein kinase (MAPK) kinase pathway inhibitors. Thus, high-throughput unbiased screening of targeted drug combinations, with appropriate library selection and mechanistic follow-up, can yield clinically-actionable drug combinations.

Advances in reporters for gene expression have made it possible to document and quantify expression patterns in 2D-4D. In contrast to microarrays, which provide data for many genes but averaged and/or at low resolution, images reveal the high spatial dynamics of gene expression. Developing computational methods to compare, annotate, and model gene expression based on images is imperative, considering that available data are rapidly increasing. We have developed a sparse Bayesian factor analysis model in which the observed expression diversity of among a large set of high-dimensional images is modeled by a small number of hidden common factors. We apply this approach on embryonic expression patterns from a Drosophila RNA in situ image database, and show that the automatically inferred factors provide for a meaningful decomposition and represent common co-regulation or biological functions. The low-dimensional set of factor mixing weights is further used as features by a classifier to annotate expression patterns with functional categories. On human-curated annotations, our sparse approach reaches similar or better classification of expression patterns at different developmental stages, when compared to other automatic image annotation methods using thousands of hard-to-interpret features. Our study therefore outlines a general framework for large microscopy data sets, in which both the generative model itself, as well as its application for analysis tasks such as automated annotation, can provide insight into biological questions.