Preclinical-to-clinical Anti-cancer Drug Response Prediction and Biomarker Identification Using TINDL.

Prediction of the response of cancer patients to different treatments and identification of biomarkers of drug response are two major goals of individualized medicine. Here, we developed a deep learning framework called TINDL, completely trained on preclinical cancer cell lines (CCLs), to predict the response of cancer patients to different treatments. TINDL utilizes a tissue-informed normalization to account for the tissue type and cancer type of the tumors and to reduce the statistical discrepancies between CCLs and patient tumors. Moreover, by making the deep learning black box interpretable, this model identifies a small set of genes whose expression levels are predictive of drug response in the trained model, enabling identification of biomarkers of drug response. Using data from two large databases of CCLs and cancer tumors, we showed that this model can distinguish between sensitive and resistant tumors for 10 (out of 14) drugs, outperforming various other machine learning models. In addition, our small interfering RNA (siRNA) knockdown experiments on 10 genes identified by this model for one of the drugs (tamoxifen) confirmed that tamoxifen sensitivity is substantially influenced by all of these genes in MCF7 cells, and seven of these genes in T47D cells. Furthermore, genes implicated for multiple drugs pointed to shared mechanism of action among drugs and suggested several important signaling pathways. In summary, this study provides a powerful deep learning framework for prediction of drug response and identification of biomarkers of drug response in cancer. The code can be accessed at https://github.com/ddhostallero/tindl.

Anastrozole Regulates Fatty Acid Synthase in Breast Cancer.

Our previous matched case-control study of postmenopausal women with resected early-stage breast cancer revealed that only anastrozole, but not exemestane or letrozole, showed a significant association between the 6-month estrogen concentrations and risk of breast cancer. Anastrozole, but not exemestane or letrozole, is a ligand for estrogen receptor α. The mechanisms of endocrine resistance are heterogenous and with the new mechanism of anastrozole, we have found that treatment of anastrozole maintains fatty acid synthase (FASN) protein level by limiting the ubiquitin-mediated FASN degradation, leading to increased breast cancer cell growth. Mechanistically, anastrozole decreases the guided entry of tail-anchored proteins factor 4 (GET4) expression, resulting in decreased BCL2-associated athanogene cochaperone 6 (BAG6) complex activity, which in turn, prevents RNF126-mediated degradation of FASN. Increased FASN protein level can induce a negative feedback loop mediated by the MAPK pathway. High levels of FASN are associated with poor outcome only in patients with anastrozole-treated breast cancer, but not in patients treated with exemestane or letrozole. Repressing FASN causes regression of breast cancer cell growth. The anastrozole-FASN signaling pathway is eminently targetable in endocrine-resistant breast cancer.

Interaction Between SNP Genotype and Efficacy of Anastrozole and Exemestane in Early-Stage Breast Cancer.

Aromatase inhibitors (AIs) are the treatment of choice for hormone receptor-positive early breast cancer in postmenopausal women. None of the third-generation AIs are superior to the others in terms of efficacy. We attempted to identify genetic factors that could differentiate between the effectiveness of adjuvant anastrozole and exemestane by examining single-nucleotide polymorphism (SNP)-treatment interaction in 4,465 patients. A group of SNPs were found to be differentially associated between anastrozole and exemestane regarding outcomes. However, they showed no association with outcome in the combined analysis. We followed up common SNPs near LY75 and GPR160 that could differentiate anastrozole from exemestane efficacy. LY75 and GPR160 participate in epithelial-to-mesenchymal transition and growth pathways, in both cases with SNP-dependent variation in regulation. Collectively, these studies identified SNPs that differentiate the efficacy of anastrozole and exemestane and they suggest additional genetic biomarkers for possible use in selecting an AI for a given patient.

Single-nucleotide polymorphism biomarkers of adjuvant anastrozole-induced estrogen suppression in early breast cancer.

OBJECTIVES: Based on our previous findings that postmenopausal women with estrone (E1) and estradiol (E2) concentrations at or above 1.3 pg/ml and 0.5 pg/ml, respectively, after 6 months of adjuvant anastrozole therapy had a three-fold risk of recurrence, we aimed to identify a single-nucleotide polymorphism (SNP)-based model that would predict elevated E1 and E2 and then validate it in an independent dataset. PATIENTS AND METHODS: The test set consisted of 322 women from the M3 study and the validation set consisted of 152 patients from MA.27. All patients were treated with adjuvant anastrozole, had on-anastrozole E1 and E2 concentrations and genome-wide genotyping. RESULTS: SNPs were identified from the M3 genome-wide association study. The best model to predict the E1-E2 phenotype with high balanced accuracy was a support vector machine model using clinical factors plus 46 SNPs. We did not have an independent cohort that is similar to the M3 study with clinical, E1-E2 phenotypes and genotype data to test our model. Hence, we chose a nested matched case-control cohort (MA.27 study) for testing. Our E1-E2 model was not validated but we found the MA.27 validation cohort was both clinically and genomically different. CONCLUSIONS: We identified a SNP-based model that had excellent performance characteristics for predicting the phenotype of elevated E1 and E2 in women treated with anastrozole. This model was not validated in an independent dataset but that dataset was clinically and genomically substantially different. The model will need validation in a prospective study.

Pharmacogenomics of aromatase inhibitors in postmenopausal breast cancer and additional mechanisms of anastrozole action.

Aromatase inhibitors (AIs) reduce breast cancer recurrence and prolong survival, but up to 30% of patients exhibit recurrence. Using a genome-wide association study of patients entered on MA.27, a phase III randomized trial of anastrozole versus exemestane, we identified a single nucleotide polymorphism (SNP) in CUB And Sushi multiple domains 1 (CSMD1) associated with breast cancer-free interval, with the variant allele associated with fewer distant recurrences. Mechanistically, CSMD1 regulates CYP19 expression in an SNP- and drug-dependent fashion, and this regulation is different among 3 AIs: anastrozole, exemestane, and letrozole. Overexpression of CSMD1 sensitized AI-resistant cells to anastrozole but not to the other 2 AIs. The SNP in CSMD1 that was associated with increased CSMD1 and CYP19 expression levels increased anastrozole sensitivity, but not letrozole or exemestane sensitivity. Anastrozole degrades estrogen receptor α (ERα), especially in the presence of estradiol (E2). ER+ breast cancer organoids and AI- or fulvestrant-resistant breast cancer cells were more sensitive to anastrozole plus E2 than to AI alone. Our findings suggest that the CSMD1 SNP might help to predict AI response, and anastrozole plus E2 serves as a potential new therapeutic strategy for patients with AI- or fulvestrant-resistant breast cancers.

Anastrozole has an Association between Degree of Estrogen Suppression and Outcomes in Early Breast Cancer and is a Ligand for Estrogen Receptor α.

PURPOSE: To determine if the degree of estrogen suppression with aromatase inhibitors (AI: anastrozole, exemestane, letrozole) is associated with efficacy in early-stage breast cancer, and to examine for differences in the mechanism of action between the three AIs. EXPERIMENTAL DESIGN: Matched case-control studies [247 matched sets from MA.27 (anastrozole vs. exemestane) and PreFace (letrozole) trials] were undertaken to assess whether estrone (E1) or estradiol (E2) concentrations after 6 months of adjuvant therapy were associated with risk of an early breast cancer event (EBCE). Preclinical laboratory studies included luciferase activity, cell proliferation, radio-labeled ligand estrogen receptor binding, surface plasmon resonance ligand receptor binding, and nuclear magnetic resonance assays. RESULTS: Women with E1 ≥1.3 pg/mL and E2 ≥0.5 pg/mL after 6 months of AI treatment had a 2.2-fold increase in risk (P = 0.0005) of an EBCE, and in the anastrozole subgroup, the increase in risk of an EBCE was 3.0-fold (P = 0.001). Preclinical laboratory studies examined mechanisms of action in addition to aromatase inhibition and showed that only anastrozole could directly bind to estrogen receptor α (ERα), activate estrogen response element-dependent transcription, and stimulate growth of an aromatase-deficient CYP19A1-/- T47D breast cancer cell line. CONCLUSIONS: This matched case-control clinical study revealed that levels of estrone and estradiol above identified thresholds after 6 months of adjuvant anastrozole treatment were associated with increased risk of an EBCE. Preclinical laboratory studies revealed that anastrozole, but not exemestane or letrozole, is a ligand for ERα. These findings represent potential steps towards individualized anastrozole therapy.

CDC25B partners with PP2A to induce AMPK activation and tumor suppression in triple negative breast cancer.

Cell division cycle 25 (CDC25) dual specificity phosphatases positively regulate the cell cycle by activating cyclin-dependent kinase/cyclin complexes. Here, we demonstrate that in addition to its role in cell cycle regulation, CDC25B functions as a regulator of protein phosphatase 2A (PP2A), a major cellular Ser/Thr phosphatase, through its direct interaction with PP2A catalytic subunit. Importantly, CDC25B alters the regulation of AMP-activated protein kinase signaling (AMPK) by PP2A, increasing AMPK activity by inhibiting PP2A to dephosphorylate AMPK. CDC25B depletion leads to metformin resistance by inhibiting metformin-induced AMPK activation. Furthermore, dual inhibition of CDC25B and PP2A further inhibits growth of 3D organoids isolated from patient derived xenograft model of breast cancer compared to CDC25B inhibition alone. Our study identifies CDC25B as a regulator of PP2A, and uncovers a mechanism of controlling the activity of a key energy metabolism marker, AMPK.

Deep sequencing across germline genome-wide association study signals relating to breast cancer events in women receiving aromatase inhibitors for adjuvant therapy of early breast cancer.

OBJECTIVE: To identify additional genetic variants beyond those observed in a previous genome-wide association study (GWAS) in women treated on the MA.27 clinical trial in which women were randomized to 5 years of adjuvant therapy with anastrozole or exemestane. PATIENTS AND METHODS: We performed a matched case-control study in 234 women who had a recurrence of breast cancer (cases) and 649 women who had not (controls). The analysis was restricted to White women with an estrogen receptor-positive breast cancer. Multiplex PCR-based targeted deep sequencing was performed of the MIR2052HG region on chromosome 8 between positions 75.4 and 75.7, a span of 300 kb, in an attempt to identify additional functional single nucleotide polymorphisms (SNPs). RESULTS: A total of 4677 unique variants were identified that had not been identified in the previous GWAS. Clinical Annotation of Variants analysis revealed 10 variants, including eight SNPs and two insertion-deletion mutations with moderate or high impact. However, none of the common and variant regions was significant after adjustment for the most significant SNP (rs13260300) identified in our previous GWAS. We performed haplotype analysis that revealed two regions in which the haplotypes lost significance when adjusted for this prior GWAS SNP and one region with two significant haplotypes (P = 0.046 and 0.031) after adjusting for the GWAS SNP. CONCLUSION: We were unable to identify common or rare variant regions that added value to the findings from our previous GWAS. We did find two haplotypes that were significant after adjusting for our top GWAS SNP but these were considered to be of marginal value.

The lncRNA MIR2052HG regulates ERα levels and aromatase inhibitor resistance through LMTK3 by recruiting EGR1.

BACKGROUND: Our previous genome-wide association study using the MA.27 aromatase inhibitors adjuvant trial identified SNPs in the long noncoding RNA MIR2052HG associated with breast cancer-free interval. MIR2052HG maintained ERα both by promoting AKT/FOXO3-mediated ESR1 transcription and by limiting ubiquitin-mediated ERα degradation. Our goal was to further elucidate MIR2052HG's mechanism of action. METHODS: RNA-binding protein immunoprecipitation assays were performed to demonstrate that the transcription factor, early growth response protein 1 (EGR1), worked together with MIR2052HG to regulate that lemur tyrosine kinase-3 (LMTK3) transcription in MCF7/AC1 and CAMA-1 cells. The location of EGR1 on the LMTK3 gene locus was mapped using chromatin immunoprecipitation assays. The co-localization of MIR2052HG RNA and the LMTK3 gene locus was determined using RNA-DNA dual fluorescent in situ hybridization. Single-nucleotide polymorphisms (SNP) effects were evaluated using a panel of human lymphoblastoid cell lines. RESULTS: MIR2052HG depletion in breast cancer cells results in a decrease in LMTK3 expression and cell growth. Mechanistically, MIR2052HG interacts with EGR1 and facilitates its recruitment to the LMTK3 promoter. LMTK3 sustains ERα levels by reducing protein kinase C (PKC) activity, resulting in increased ESR1 transcription mediated through AKT/FOXO3 and reduced ERα degradation mediated by the PKC/MEK/ERK/RSK1 pathway. MIR2052HG regulated LMTK3 in a SNP- and aromatase inhibitor-dependent fashion: the variant SNP increased EGR1 binding to LMTK3 promoter in response to androstenedione, relative to wild-type genotype, a pattern that can be reversed by aromatase inhibitor treatment. Finally, LMTK3 overexpression abolished the effect of MIR2052HG on PKC activity and ERα levels. CONCLUSIONS: Our findings support a model in which the MIR2052HG regulates LMTK3 via EGR1, and LMTK3 regulates ERα stability via the PKC/MEK/ERK/RSK1 axis. These results reveal a direct role of MIR2052HG in LMTK3 regulation and raise the possibilities of targeting MIR2052HG or LMTK3 in ERα-positive breast cancer.

Identification of pathways associated with chemosensitivity through network embedding.

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.

Anastrozole Aromatase Inhibitor Plasma Drug Concentration Genome-Wide Association Study: Functional Epistatic Interaction Between SLC38A7 and ALPPL2.

Anastrozole is a widely prescribed aromatase inhibitor for the therapy of estrogen receptor positive (ER+) breast cancer. We performed a genome-wide association study (GWAS) for plasma anastrozole concentrations in 687 postmenopausal women with ER+ breast cancer. The top single-nucleotide polymorphism (SNP) signal mapped across SLC38A7 (rs11648166, P = 2.3E-08), which we showed to encode an anastrozole influx transporter. The second most significant signal (rs28845026, P = 5.4E-08) mapped near ALPPL2 and displayed epistasis with the SLC38A7 signal. Both of these SNPs were cis expression quantitative trait loci (eQTL)s for these genes, and patients homozygous for variant genotypes for both SNPs had the highest drug concentrations, the highest SLC38A7 expression, and the lowest ALPPL2 expression. In summary, our GWAS identified a novel gene encoding an anastrozole transporter, SLC38A7, as well as epistatic interaction between SNPs in that gene and SNPs near ALPPL2 that influenced both the expression of the transporter and anastrozole plasma concentrations.

Machine Learning Helps Identify New Drug Mechanisms in Triple-Negative Breast Cancer.

This paper demonstrates the ability of mach- ine learning approaches to identify a few genes among the 23,398 genes of the human genome to experiment on in the laboratory to establish new drug mechanisms. As a case study, this paper uses MDA-MB-231 breast cancer single-cells treated with the antidiabetic drug metformin. We show that mixture-model-based unsupervised methods with validation from hierarchical clustering can identify single-cell subpopulations (clusters). These clusters are characterized by a small set of genes (1% of the genome) that have significant differential expression across the clusters and are also highly correlated with pathways with anticancer effects driven by metformin. Among the identified small set of genes associated with reduced breast cancer incidence, laboratory experiments on one of the genes, CDC42, showed that its downregulation by metformin inhibited cancer cell migration and proliferation, thus validating the ability of machine learning approaches to identify biologically relevant candidates for laboratory experiments. Given the large size of the human genome and limitations in cost and skilled resources, the broader impact of this work in identifying a small set of differentially expressed genes after drug treatment lies in augmenting the drug-disease knowledge of pharmacogenomics experts in laboratory investigations, which could help establish novel biological mechanisms associated with drug response in diseases beyond breast cancer.

Principled multi-omic analysis reveals gene regulatory mechanisms of phenotype variation.

Recent studies have analyzed large-scale data sets of gene expression to identify genes associated with interindividual variation in phenotypes ranging from cancer subtypes to drug sensitivity, promising new avenues of research in personalized medicine. However, gene expression data alone is limited in its ability to reveal cis-regulatory mechanisms underlying phenotypic differences. In this study, we develop a new probabilistic model, called pGENMi, that integrates multi-omic data to investigate the transcriptional regulatory mechanisms underlying interindividual variation of a specific phenotype-that of cell line response to cytotoxic treatment. In particular, pGENMi simultaneously analyzes genotype, DNA methylation, gene expression, and transcription factor (TF)-DNA binding data, along with phenotypic measurements, to identify TFs regulating the phenotype. It does so by combining statistical information about expression quantitative trait loci (eQTLs) and expression-correlated methylation marks (eQTMs) located within TF binding sites, as well as observed correlations between gene expression and phenotype variation. Application of pGENMi to data from a panel of lymphoblastoid cell lines treated with 24 drugs, in conjunction with ENCODE TF ChIP data, yielded a number of known as well as novel (TF, Drug) associations. Experimental validations by TF knockdown confirmed 41% of the predicted and tested associations, compared to a 12% confirmation rate of tested nonassociations (controls). An extensive literature survey also corroborated 62% of the predicted associations above a stringent threshold. Moreover, associations predicted only when combining eQTL and eQTM data showed higher precision compared to an eQTL-only or eQTM-only analysis using pGENMi, further demonstrating the value of multi-omic integrative analysis.

Germline genome-wide association studies in women receiving neoadjuvant chemotherapy with or without bevacizumab.

Neoadjuvant chemotherapy (NAC) for breast cancer is widely utilized, and we performed genome-wide association studies (GWAS) to determine whether germ-line genetic variability was associated with benefit in terms of pathological complete response (pCR), disease-free survival, and overall survival in patients entered on the NSABP B-40 NAC trial, wherein patients were randomized to receive, or not, bevacizumab in addition to chemotherapy. Patient DNA samples were genotyped with the Illumina OmniExpress BeadChip. Replication was attempted with genotyping data from 1398 HER2-negative patients entered on the GeparQuinto NAC study in which patients were also randomized to receive, or not, bevacizumab in addition to chemotherapy. A total of 920 women from B-40 were analyzed, and 237 patients achieved a pCR. GWAS with three phenotypes (pCR, disease-free survival, overall survival) revealed no single nucleotide polymorphisms (SNPs) that were genome-wide significant (i.e. P≤5E-08) signals; P values for top SNPs were 2.04E-07, 5.61E-08, and 5.63E-08, respectively, and these SNPs were not significant in the GeparQuinto data. An ad-hoc GWAS was performed in the patients randomized to bevacizumab (457 patients with 128 pCR) who showed signals on chromosome 6, located within a gene, CDKAL1, that approached, but did not reach, genome-wide significance (top SNP rs7453577, P=2.97E-07). However, this finding was significant when tested in the GeparQuinto data set (P=0.04). In conclusion, we identified no SNPs significantly associated with NAC. The observation, in a hypothesis-generating GWAS, of an SNP in CDKAL1 associated with pCR in the bevacizumab arm of both B-40 and GeparQuinto requires further validation and study.

Differential roles of ERRFI1 in EGFR and AKT pathway regulation affect cancer proliferation.

AKT signaling is modulated by a complex network of regulatory proteins and is commonly deregulated in cancer. Here, we present a dual mechanism of AKT regulation by the ERBB receptor feedback inhibitor 1 (ERRFI1). We show that in cells expressing high levels of EGFR, ERRF1 inhibits growth and enhances responses to chemotherapy. This is mediated in part through the negative regulation of AKT signaling by direct ERRFI1-dependent inhibition of EGFR In cells expressing low levels of EGFR, ERRFI1 positively modulates AKT signaling by interfering with the interaction of the inactivating phosphatase PHLPP with AKT, thereby promoting cell growth and chemotherapy desensitization. These observations broaden our understanding of chemotherapy response and have important implications for the selection of targeted therapies in a cell context-dependent manner. EGFR inhibition can only sensitize EGFR-high cells for chemotherapy, while AKT inhibition increases chemosensitivity in EGFR-low cells. By understanding these mechanisms, we can take advantage of the cellular context to individualize antineoplastic therapy. Finally, our data also suggest targeting of EFFRI1 in EGFR-low cancer as a promising therapeutic approach.

Therapy response testing of breast cancer in a 3D high-throughput perfused microfluidic platform.

BACKGROUND: Breast cancer is the most common invasive cancer among women. Currently, there are only a few models used for therapy selection, and they are often poor predictors of therapeutic response or take months to set up and assay. In this report, we introduce a microfluidic OrganoPlate® platform for extracellular matrix (ECM) embedded tumor culture under perfusion as an initial study designed to investigate the feasibility of adapting this technology for therapy selection. METHODS: The triple negative breast cancer cell lines MDA-MB-453, MDA-MB-231 and HCC1937 were selected based on their different BRCA1 and P53 status, and were seeded in the platform. We evaluate seeding densities, ECM composition (Matrigel®, BME2rgf, collagen I) and biomechanical (perfusion vs static) conditions. We then exposed the cells to a series of anti-cancer drugs (paclitaxel, olaparib, cisplatin) and compared their responses to those in 2D cultures. Finally, we generated cisplatin dose responses in 3D cultures of breast cancer cells derived from 2 PDX models. RESULTS: The microfluidic platform allows the simultaneous culture of 96 perfused micro tissues, using limited amounts of material, enabling drug screening of patient-derived material. 3D cell culture viability is improved by constant perfusion of the medium. Furthermore, the drug response of these triple negative breast cancer cells was attenuated by culture in 3D and differed from that observed in 2D substrates. CONCLUSIONS: We have investigated the use of a high-throughput organ-on-a-chip platform to select therapies. Our results have raised the possibility to use this technology in personalized medicine to support selection of appropriate drugs and to predict response to therapy in a real time fashion.

SNPs near the cysteine proteinase cathepsin O gene (CTSO) determine tamoxifen sensitivity in ERα-positive breast cancer through regulation of BRCA1.

Tamoxifen is one of the most commonly employed endocrine therapies for patients with estrogen receptor α (ERα)-positive breast cancer. Unfortunately the clinical benefit is limited due to intrinsic and acquired drug resistance. We previously reported a genome-wide association study that identified common SNPs near the CTSO gene and in ZNF423 associated with development of breast cancer during tamoxifen therapy in the NSABP P-1 and P-2 breast cancer prevention trials. Here, we have investigated their roles in ERα-positive breast cancer growth and tamoxifen response, focusing on the mechanism of CTSO. We performed in vitro studies including luciferase assays, cell proliferation, and mass spectrometry-based assays using ERα-positive breast cancer cells and a panel of genomic data-rich lymphoblastoid cell lines. We report that CTSO reduces the protein levels of BRCA1 and ZNF423 through cysteine proteinase-mediated degradation. We also have identified a series of transcription factors of BRCA1 that are regulated by CTSO at the protein level. Importantly, the variant CTSO SNP genotypes are associated with increased CTSO and decreased BRCA1 protein levels that confer resistance to tamoxifen. Characterization of the effect of both CTSO SNPs and ZNF423 SNPs on tamoxifen response revealed that cells with different combinations of CTSO and ZNF423 genotypes respond differently to Tamoxifen, PARP inhibitors or the combination of the two drugs due to SNP dependent differential regulation of BRCA1 levels. Therefore, these genotypes might be biomarkers for selection of individual drug to achieve the best efficacy.

Knowledge-guided gene prioritization reveals new insights into the mechanisms of chemoresistance.

BACKGROUND: Identification of genes whose basal mRNA expression predicts the sensitivity of tumor cells to cytotoxic treatments can play an important role in individualized cancer medicine. It enables detailed characterization of the mechanism of action of drugs. Furthermore, screening the expression of these genes in the tumor tissue may suggest the best course of chemotherapy or a combination of drugs to overcome drug resistance. RESULTS: We developed a computational method called ProGENI to identify genes most associated with the variation of drug response across different individuals, based on gene expression data. In contrast to existing methods, ProGENI also utilizes prior knowledge of protein-protein and genetic interactions, using random walk techniques. Analysis of two relatively new and large datasets including gene expression data on hundreds of cell lines and their cytotoxic responses to a large compendium of drugs reveals a significant improvement in prediction of drug sensitivity using genes identified by ProGENI compared to other methods. Our siRNA knockdown experiments on ProGENI-identified genes confirmed the role of many new genes in sensitivity to three chemotherapy drugs: cisplatin, docetaxel, and doxorubicin. Based on such experiments and extensive literature survey, we demonstrate that about 73% of our top predicted genes modulate drug response in selected cancer cell lines. In addition, global analysis of genes associated with groups of drugs uncovered pathways of cytotoxic response shared by each group. CONCLUSIONS: Our results suggest that knowledge-guided prioritization of genes using ProGENI gives new insight into mechanisms of drug resistance and identifies genes that may be targeted to overcome this phenomenon.

Model-based unsupervised learning informs metformin-induced cell-migration inhibition through an AMPK-independent mechanism in breast cancer.

We demonstrate that model-based unsupervised learning can uniquely discriminate single-cell subpopulations by their gene expression distributions, which in turn allow us to identify specific genes for focused functional studies. This method was applied to MDA-MB-231 breast cancer cells treated with the antidiabetic drug metformin, which is being repurposed for treatment of triple-negative breast cancer. Unsupervised learning identified a cluster of metformin-treated cells characterized by a significant suppression of 230 genes (p-value < 2E-16). This analysis corroborates known studies of metformin action: a) pathway analysis indicated known mechanisms related to metformin action, including the citric acid (TCA) cycle, oxidative phosphorylation, and mitochondrial dysfunction (p-value < 1E-9); b) 70% of these 230 genes were functionally implicated in metformin response; c) among remaining lesser functionally-studied genes for metformin-response was CDC42, down-regulated in breast cancer treated with metformin. However, CDC42's mechanisms in metformin response remained unclear. Our functional studies showed that CDC42 was involved in metformin-induced inhibition of cell proliferation and cell migration mediated through an AMPK-independent mechanism. Our results points to 230 genes that might serve as metformin response signatures, which needs to be tested in patients treated with metformin and, further investigation of CDC42 and AMPK-independence's role in metformin's anticancer mechanisms.

A network-based phenotype mapping approach to identify genes that modulate drug response phenotypes.

To better address the problem of drug resistance during cancer chemotherapy and explore the possibility of manipulating drug response phenotypes, we developed a network-based phenotype mapping approach (P-Map) to identify gene candidates that upon perturbed can alter sensitivity to drugs. We used basal transcriptomics data from a panel of human lymphoblastoid cell lines (LCL) to infer drug response networks (DRNs) that are responsible for conferring response phenotypes for anthracycline and taxane, two common anticancer agents use in clinics. We further tested selected gene candidates that interact with phenotypic differentially expressed genes (PDEGs), which are up-regulated genes in LCL for a given class of drug response phenotype in triple-negative breast cancer (TNBC) cells. Our results indicate that it is possible to manipulate a drug response phenotype, from resistant to sensitive or vice versa, by perturbing gene candidates in DRNs and suggest plausible mechanisms regulating directionality of drug response sensitivity. More important, the current work highlights a new way to formulate systems-based therapeutic design: supplementing therapeutics that aim to target disease culprits with phenotypic modulators capable of altering DRN properties with the goal to re-sensitize resistant phenotypes.