Search

Basal gene expression levels have been shown to be predictive of cellular response to cytotoxic treatments. However, such analyses do not fully reveal complex genotype- phenotype relationships, which are partly encoded in highly interconnected molecular networks. Biological pathways provide a complementary way of understanding drug response variation among individuals. In this study, we integrate chemosensitivity data from a large-scale pharmacogenomics study with basal gene expression data from the CCLE project and prior knowledge of molecular networks to identify specific pathways mediating chemical response. We first develop a computational method called PACER, which ranks pathways for enrichment in a given set of genes using a novel network embedding method. It examines a molecular network that encodes known gene-gene as well as gene-pathway relationships, and determines a vector representation of each gene and pathway in the same low-dimensional vector space. The relevance of a pathway to the given gene set is then captured by the similarity between the pathway vector and gene vectors. To apply this approach to chemosensitivity data, we identify genes whose basal expression levels in a panel of cell lines are correlated with cytotoxic response to a compound, and then rank pathways for relevance to these response-correlated genes using PACER. Extensive evaluation of this approach on benchmarks constructed from databases of compound target genes and large collections of drug response signatures demonstrates its advantages in identifying compound-pathway associations compared to existing statistical methods of pathway enrichment analysis. The associations identified by PACER can serve as testable hypotheses on chemosensitivity pathways and help further study the mechanisms of action of specific cytotoxic drugs. More broadly, PACER represents a novel technique of identifying enriched properties of any gene set of interest while also taking into account networks of known gene-gene relationships and interactions., Author summary Gene expression levels have been used to study the cellular response to drug treatments. However, analysis of gene expression without considering gene interactions cannot fully reveal complex genotype-phenotype relationships. Biological pathways reveal the interactions among genes, thus providing a complementary way of understanding the drug response variation among individuals. In this paper, we aim to identify pathways that mediate the chemical response of each drug. We used the recently generated CTRP pharmacogenomics data and CCLE basal expression data to identify these pathways. We showed that using the prior knowledge encoded in molecular networks substantially improves pathway identification. In particular, we integrate genes and pathways into a large heterogeneous network in which links are protein-protein interactions and gene-pathway affiliations. We then project this heterogeneous network onto a low-dimensional space, which enables more precise similarity measurements between pathways and drug-response-correlated genes. Extensive experiments on two benchmarks show that our method substantially improved the pathway identification performance by using the molecular networks. More importantly, our method represents a novel technique of identifying enriched properties of any gene set of interest while also taking into account networks of known gene-gene relationships and interactions.

Understanding the molecular pathways by which oncogenes drive cancerous cell growth, and how dependence on such pathways varies between tumors could be highly valuable for the design of anti-cancer treatment strategies. In this work we study how dependence upon the canonical PI3K and MAPK cascades varies across HER2+ cancers, and define biomarkers predictive of pathway dependencies. A panel of 18 HER2+ (ERBB2-amplified) cell lines representing a variety of indications was used to characterize the functional and molecular diversity within this oncogene-defined cancer. PI3K and MAPK-pathway dependencies were quantified by measuring in vitro cell growth responses to combinations of AKT (MK2206) and MEK (GSK1120212; trametinib) inhibitors, in the presence and absence of the ERBB3 ligand heregulin (NRG1). A combination of three protein measurements comprising the receptors EGFR, ERBB3 (HER3), and the cyclin-dependent kinase inhibitor p27 (CDKN1B) was found to accurately predict dependence on PI3K/AKT vs. MAPK/ERK signaling axes. Notably, this multivariate classifier outperformed the more intuitive and clinically employed metrics, such as expression of phospho-AKT and phospho-ERK, and PI3K pathway mutations (PIK3CA, PTEN, and PIK3R1). In both cell lines and primary patient samples, we observed consistent expression patterns of these biomarkers varies by cancer indication, such that ERBB3 and CDKN1B expression are relatively high in breast tumors while EGFR expression is relatively high in other indications. The predictability of the three protein biomarkers for differentiating PI3K/AKT vs. MAPK dependence in HER2+ cancers was confirmed using external datasets (Project Achilles and GDSC), again out-performing clinically used genetic markers. Measurement of this minimal set of three protein biomarkers could thus inform treatment, and predict mechanisms of drug resistance in HER2+ cancers. More generally, our results show a single oncogenic transformation can have differing effects on cell signaling and growth, contingent upon the molecular and cellular context., Author Summary Biomarkers capable of accurately predicting patient responses to alternate therapies are critical to realizing the vision of precision medicine. Identifying such biomarkers is, however, challenging due to the inherent complexity of biological networks. Here we sought to identify molecular features that predict how a genetically defined subset of cancers (HER2+) differentially depend on two oncogenic signaling pathways, the PI3K/AKT and MAPK/ERK cascades. We find that combined measurement of three non-intuitive proteins (EGFR, ERBB3, and CDKN1B) accurately predicts cellular dependence on these signaling pathways, and responsiveness to drugs targeting their constituents. Notably, this three-biomarker model outperformed both biological intuition (phosho-AKT and phospho-ERK) and current clinical practice (PIK3CA mutations). More broadly, this exemplifies how the functional consequences of a single oncogenic driver (HER2) can depend upon molecular and cellular context. Expression of these markers also varies by indication, such that breast cancers are biased toward PI3K-dependnece, while non-breast indications (lung, ovarian, and gastric) are particularly MAPK-dependent, and thus may respond differently to therapeutic strategies developed for breast cancer. Together, we believe that our results will aid the design of novel, stratified treatment strategies for HER2+ disease.

Cancer is a genetic disease that develops through a series of somatic mutations, a subset of which drive cancer progression. Although cancer genome sequencing studies are beginning to reveal the mutational patterns of genes in various cancers, identifying the small subset of “causative” mutations from the large subset of “non-causative” mutations, which accumulate as a consequence of the disease, is a challenge. In this article, we present an effective machine learning approach for identifying cancer-associated mutations in human protein kinases, a class of signaling proteins known to be frequently mutated in human cancers. We evaluate the performance of 11 well known supervised learners and show that a multiple-classifier approach, which combines the performances of individual learners, significantly improves the classification of known cancer-associated mutations. We introduce several novel features related specifically to structural and functional characteristics of protein kinases and find that the level of conservation of the mutated residue at specific evolutionary depths is an important predictor of oncogenic effect. We consolidate the novel features and the multiple-classifier approach to prioritize and experimentally test a set of rare unconfirmed mutations in the epidermal growth factor receptor tyrosine kinase (EGFR). Our studies identify T725M and L861R as rare cancer-associated mutations inasmuch as these mutations increase EGFR activity in the absence of the activating EGF ligand in cell-based assays., Author Summary Cancer progresses by accumulation of mutations in a subset of genes that confer growth advantage. The 518 protein kinase genes encoded in the human genome, collectively called the kinome, represent one of the largest families of oncogenes. Targeted sequencing studies of many different cancers have shown that the mutational landscape comprises both cancer-causing “driver” mutations and harmless “passenger” mutations. While the frequent recurrence of some driver mutations in human cancers helps distinguish them from the large number of passenger mutations, a significant challenge is to identify the rare “driver” mutations that are less frequently observed in patient samples and yet are causative. Here we combine computational and experimental approaches to identify rare cancer-associated mutations in Epidermal Growth Factor receptor kinase (EGFR), a signaling protein frequently mutated in cancers. Specifically, we evaluate a novel multiple-classifier approach and features specific to the protein kinase super-family in distinguishing known cancer-associated mutations from benign mutations. We then apply the multiple classifier to identify and test the functional impact of rare cancer-associated mutations in EGFR. We report, for the first time, that the EGFR mutations T725M and L861R, which are infrequently observed in cancers, constitutively activate EGFR in a manner analogous to the frequently observed driver mutations.