A landscape of response to drug combinations in non-small cell lung cancer

Combination of anti-cancer drugs is broadly seen as way to overcome the often-limited efficacy of single agents. The design and testing of combinations are however very challenging. Here we present a uniquely large dataset screening over 5000 targeted agent combinations across 81 non-small cell lung cancer cell lines. Our analysis reveals a profound heterogeneity of response across the tumor models. Notably, combinations very rarely result in a strong gain in efficacy over the range of response observable with single agents. Importantly, gain of activity over single agents is more often seen when co-targeting functionally proximal genes, offering a strategy for designing more efficient combinations. Because combinatorial effect is strongly context specific, tumor specificity should be achievable. The resource provided, together with an additional validation screen sheds light on major challenges and opportunities in building efficacious combinations against cancer and provides an opportunity for training computational models for synergy prediction.

Abstract LB-149: Genome-wide prediction of synthetic rescue mediators of resistance to targeted and immunotherapy

Background: Most patients with advanced cancer eventually acquire resistance to targeted therapies, spurring extensive efforts to identify molecular events mediating therapy resistance. Here we demonstrate that many of these molecular events involve synthetic rescue (SR) genetic interactions, where the reduction in cancer cell viability caused by targeted gene inactivation is rescued by an adaptive alteration of another gene (the rescuer).

Methods: Analyzing recently published large scale in vitro functional screens we identify 100,000s candidate SR interactions that show evidence of rescue events in cancer cell lines. We then analyze tumor transcriptomic, genomic, and genetic profiles together with survival and clinical characteristics of 10,000 TCGA cancer patients to identify the few thousand SR interactions that have strong evidence that they are mediators of emerging resistance in patient tumors.

Results: We identify the first rescue network, composed of SR interactions common to many cancer types. The identified SRs successfully match the resistance mediators identified in recently published clinical studies. We perform multiple in vitro analyses in head and neck and lung cancer, showing that targeting predicted rescuer genes successfully re-sensitizes resistant cancer cells, providing specific leads for targeting resistance proactively. We further demonstrate that SR-based combination therapy can improve the progression free survival in mouse xenografts models. Notably, we show that SR interactions successfully predict cancer patients' response in the TCGA compendium, showing performance superior to existing machine learning based predictive models. Finally, we show that SR analysis of melanoma patients successfully identifies known mediators of resistance to checkpoint immunotherapy (reported previously in mice) and suggests new combination therapies that counteract the resistance.

Conclusions: This work presents a new paradigm identifying and harnessing synthetic rescue interactions to counteract resistance to both targeted- and immuno-therapies. Future implementations of this approach will have two broad implications in the precision oncology era: first for determining the most effective treatment regimen based on the molecular characteristics of individual patient’s tumor; second for identifying conjunct drugs to counteract resistance to existing primary therapies, for both targeted and immune checkpoint therapies.

Citation Format: Avinash Das, Joo Sang Lee, Gao Zhang, Zhiyong Wang, Arnaud Amzallag, Genevieve Boland, Sridhar Hannenhalli, Meenhard Herlyn, Cyril Benes, J. Silvio Gutkind, Keity Flaherty, Eytan Ruppin. Genome-wide prediction of synthetic rescue mediators of resistance to targeted and immunotherapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr LB-149.

Differential Effector Engagement by Oncogenic KRAS

KRAS can bind numerous effector proteins, which activate different downstream signaling events. The best known are RAF, phosphatidylinositide (PI)-3' kinase, and RalGDS families, but many additional direct and indirect effectors have been reported. We have assessed how these effectors contribute to several major phenotypes in a quantitative way, using an arrayed combinatorial siRNA screen in which we knocked down 41 KRAS effectors nodes in 92 cell lines. We show that every cell line has a unique combination of effector dependencies, but in spite of this heterogeneity, we were able to identify two major subtypes of KRAS mutant cancers of the lung, pancreas, and large intestine, which reflect different KRAS effector engagement and opportunities for therapeutic intervention.

Quasimesenchymal phenotype predicts systemic metastasis in pancreatic ductal adenocarcinoma

Metastasis following surgical resection is a leading cause of mortality in pancreatic ductal adenocarcinoma. Epithelial-mesenchymal transition is thought to play an important role in metastasis, although its clinical relevance in metastasis remains uncertain. We evaluated a panel of RNA in-situ hybridization probes for epithelial-mesenchymal transition-related genes expressed in circulating tumor cells. We assessed the predictive value of this panel for metastasis in pancreatic ductal adenocarcinoma and, to determine if the phenotype is generalizable between cancers, in colonic adenocarcinoma. One hundred fifty-eight pancreatic ductal adenocarcinomas and 205 colonic adenocarcinomas were classified as epithelial or quasimesenchymal phenotype using dual colorimetric RNA-in-situ hybridization. SMAD4 expression on pancreatic ductal adenocarcinomas was assessed by immunohistochemistry. Pancreatic ductal adenocarcinomas with quasimesenchymal phenotype had a significantly shorter disease-specific survival (P = 0.031) and metastasis-free survival (P = 0.0001) than those with an epithelial phenotype. Pancreatic ductal adenocarcinomas with SMAD4 loss also had lower disease-specific survival (P = 0.041) and metastasis-free survival (P = 0.001) than those with intact SMAD4. However, the quasimesenchymal phenotype proved a more robust predictor of metastases-area under the curve for quasimesenchymal = 0.8; SMAD4 = 0.6. The quasimesenchymal phenotype also predicted metastasis-free survival (P = 0.004) in colonic adenocarcinoma. Epithelial-mesenchymal transition defined two phenotypes with distinct metastatic capabilities-epithelial phenotype tumors with predominantly organ-confined disease and quasimesenchymal phenotype with high risk of metastatic disease in two epithelial malignancies. Collectively, this work validates the relevance of epithelial-mesenchymal transition in human gastrointestinal tumors.

Genome-wide prediction of synthetic rescue mediators of resistance to targeted and immunotherapy

Most patients with advanced cancer eventually acquire resistance to targeted therapies, spurring extensive efforts to identify molecular events mediating therapy resistance. Many of these events involve synthetic rescue (SR) interactions, where the reduction in cancer cell viability caused by targeted gene inactivation is rescued by an adaptive alteration of another gene (the rescuer). Here, we perform a genome-wide in silico prediction of SR rescuer genes by analyzing tumor transcriptomics and survival data of 10,000 TCGA cancer patients. Predicted SR interactions are validated in new experimental screens. We show that SR interactions can successfully predict cancer patients' response and emerging resistance. Inhibiting predicted rescuer genes sensitizes resistant cancer cells to therapies synergistically, providing initial leads for developing combinatorial approaches to overcome resistance proactively. Finally, we show that the SR analysis of melanoma patients successfully identifies known mediators of resistance to immunotherapy and predicts novel rescuers.

Statistical assessment and visualization of synergies for large-scale sparse drug combination datasets

BACKGROUND

Drug combinations have the potential to improve efficacy while limiting toxicity. To robustly identify synergistic combinations, high-throughput screens using full dose-response surface are desirable but require an impractical number of data points. Screening of a sparse number of doses per drug allows to screen large numbers of drug pairs, but complicates statistical assessment of synergy. Furthermore, since the number of pairwise combinations grows with the square of the number of drugs, exploration of large screens necessitates advanced visualization tools.

RESULTS

We describe a statistical and visualization framework for the analysis of large-scale drug combination screens. We developed an approach suitable for datasets with large number of drugs pairs even if small number of data points are available per drug pair. We demonstrate our approach using a systematic screen of all possible pairs among 108 cancer drugs applied to melanoma cell lines. In this dataset only two dose-response data points per drug pair and two data points per single drug test were available. We used a Bliss-based linear model, effectively borrowing data from the drug pairs to obtain robust estimations of the singlet viabilities, consequently yielding better estimates of drug synergy. Our method improves data consistency across dosing thus likely reducing the number of false positives. The approach allows to compute p values accounting for standard errors of the modeled singlets and combination viabilities. We further develop a synergy specificity score that distinguishes specific synergies from those arising with promiscuous drugs. Finally, we developed a summarized interactive visualization in a web application, providing efficient access to any of the 439,000 data points in the combination matrix ( http://www.cmtlab.org:3000/combo_app.html ). The code of the analysis and the web application is available at https://github.com/arnaudmgh/synergy-screen .

CONCLUSIONS

We show that statistical modeling of single drug response from drug combination data can help determine significance of synergy and antagonism in drug combination screens with few data point per drug pair. We provide a web application for the rapid exploration of large combinatorial drug screen. All codes are available to the community, as a resource for further analysis of published data and for analysis of other drug screens.

Harnessing synthetic lethality to predict the response to cancer treatment

While synthetic lethality (SL) holds promise in developing effective cancer therapies, SL candidates found via experimental screens often have limited translational value. Here we present a data-driven approach, ISLE (identification of clinically relevant synthetic lethality), that mines TCGA cohort to identify the most likely clinically relevant SL interactions (cSLi) from a given candidate set of lab-screened SLi. We first validate ISLE via a benchmark of large-scale drug response screens and by predicting drug efficacy in mouse xenograft models. We then experimentally test a select set of predicted cSLi via new screening experiments, validating their predicted context-specific sensitivity in hypoxic vs normoxic conditions and demonstrating cSLi's utility in predicting synergistic drug combinations. We show that cSLi can successfully predict patients' drug treatment response and provide patient stratification signatures. ISLE thus complements existing actionable mutation-based methods for precision cancer therapy, offering an opportunity to expand its scope to the whole genome.

Abstract A188: Harnessing synthetic lethality to predict the response to cancer treatments

Synthetic lethality (SL) describes an interaction between a pair of genes whereby their double knockout is lethal, while their respective knockout is not. The identification of SL interactions (SLi) via large-scale genomic screens offers promising opportunities for developing selective therapies in cancer. However, our analysis of the TCGA cohort shows that many of the interactions do not carry predictive signal of patient survival or drug response. Here we present a data-driven approach termed ISLE (Identification of clinically relevant Synthetic LEthality) that mines the TCGA cohort to identify a subset of clinically relevant SL interactions (cSLi). ISLE consists of the following inference steps, analysis of tumor, cell line, and gene evolutionary data. We first create an initial pool of SL pairs identified through direct double knockout screens/isogenic cell line screens or inferred from large-scale shRNA/sgRNA single-gene knockout screens. Starting from this initial SL pool, ISLE first identifies putative SL gene pairs whose co-inactivation is under-represented in tumors, testifying that it is selected against. Second, it prioritizes candidate SL pairs whose co-inactivation is associated with improved patient’s prognosis, testifying that it may hamper tumor progression. Finally, it prioritizes SL-gene pairs with similar evolutionary phylogenetic profiles based on the notion that SL interactions are conserved across multiple species. We validate the identified SL pairs using an unseen large-scale in vitro drug response screen by showing the SL pairs marks a decent prediction accuracy (AUC~0.8). We compare ISLE’s performance to the standard supervised drug response prediction approaches in DREAM challenges, and our prediction based on generic pretreatment tumor samples (from TCGA) was within top 3 in prediction accuracy among the top predictors. ISLE-based approach also successfully distinguishes responders vs nonresponders to drug treatment (for >70% of drugs) in mouse xenografts using the activity profile of the drug target’s SL-partners. We then experimentally show the utility of SL in predicting synergistic drug combinations in patient-derived cell lines based on the notion that the two drugs whose targets have SL interactions are synergistic. Most importantly, we demonstrate for the first time that an SL network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including TCGA, where cSLi are successfully predict patients’ response for more than 70% of cancer drugs. ISLE is predictive of patients’ response for the majority of current cancer drugs without any drug-specific training. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy.

Citation Format: Joo S. Lee, Avinash Das, Livnat Jerby-Arnon, Rand Arafeh, Matthew Davidson, Arnaud Amzallag, Seung Gu Park, Kuoyuan Cheng, Welles Robinson, Dikla Atias, Chani Stossel, Ella Buzhor, Gidi Stein, Joshua J. Waterfall, Paul S. Meltzer, Talia Golan, Sridhar Hannenhalli, Eyal Gottlieb, Cyril H. Benes, Yardena Samuels, Emma Shanks, Eytan Ruppin. Harnessing synthetic lethality to predict the response to cancer treatments [abstract]. In: Proceedings of the AACR-NCI-EORTC International Conference: Molecular Targets and Cancer Therapeutics; 2017 Oct 26-30; Philadelphia, PA. Philadelphia (PA): AACR; Mol Cancer Ther 2018;17(1 Suppl):Abstract nr A188.

Abstract PR09: Harnessing synthetic lethality to predict clinical outcomes of cancer treatment

Significance: The identification of Synthetic Lethal interactions (SLi) have long been considered a foundation for the advancement of cancer treatment. The rapidly accumulating large-scale patient data now provides a golden opportunity to infer SLi directly from patient samples. Here we present a new data-driven approach termed ISLE for identifying SLi, which is then shown to be predictive of clinical outcomes of cancer treatment in an unsupervised manner, for the first time.

Methods: ISLE consists of four inference steps, analyzing tumor, cell line and gene evolutionary data: It first identifies putative SL gene pairs whose co-inactivation is underrepresented in tumors, testifying that they are selected against. Second, it further prioritizes candidate SL pairs whose co-inactivation is associated with better prognosis in patients, testifying that they may hamper tumor progression. Finally, it eliminates false positive SLi using gene essentiality screens (testifying to causal SLi relations) and prioritizing SLi paired genes with similar evolutionary phylogenetic profiles.

Results: We applied ISLE to analyze the TCGA tumor collection and generated the first clinically-derived pan-cancer SL-network, composed of SLi common across many cancer types. We validated that these SLi match the known, experimentally identified SLi (AUC=0.87), and show that the SL-network is predictive of patient survival in an independent breast cancer dataset (METABRIC). Based on the predicted SLi, we predicted drug response in a wide variety of in vitro, mouse xenograft and patient data, altogether encompassing >700 single drugs and >5,000 drug combinations in >1,000 cell lines, 375 xenograft models and >5,000 patient samples. Importantly, these predictions were performed in an unsupervised manner, reducing the known risk of over-fitting the data commonly associated with supervised prediction methods. SL-derived predictions are based on computing an SL-score that estimates the efficacy of a given drug in a given tumor based on the latter's omics data. The SL-score counts the number of inactive SL-partners of a given drug target(s) in the given tumor, reflecting the notion that a drug is likely to be more effective in tumors where many of its targets' SL-partners are inactive. The predicted SL-scores show significant correlations (R > 0.4) with large-scale in vitro and in vivo drug response screens for the majority of drugs tested. Based on the conjecture that synergism between drugs may be mediated by underlying SLi between their targets, we additionally provide accurate predictions of drug synergism for both in vitro and in vivo drug combination screens (AUC~0.8). Most importantly, we demonstrate for the first time that an SL-network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including the TCGA, where SLis successfully predict patients' response for 75% of cancer drugs.

Conclusions: ISLE is predictive of the patients' response for the majority of current cancer drugs. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy. The predictive performance of ISLE is likely to further improve with the expected rapid accumulation of additional cancer omics and clinical phenotypic data.

Citation Format: Joo Sang Lee, Avinash Das, Livnat Jerby-Arnon, Dikla Atias, Arnaud Amzallag, Cyril H. Benes, Talia Golan, Eytan Ruppin. Harnessing synthetic lethality to predict clinical outcomes of cancer treatment [abstract]. In: Proceedings of the AACR Precision Medicine Series: Opportunities and Challenges of Exploiting Synthetic Lethality in Cancer; Jan 4-7, 2017; San Diego, CA. Philadelphia (PA): AACR; Mol Cancer Ther 2017;16(10 Suppl):Abstract nr PR09.

Detection of dysregulated protein-association networks by high-throughput proteomics predicts cancer vulnerabilities

The formation of protein complexes and the co-regulation of the cellular concentrations of proteins are essential mechanisms for cellular signaling and for maintaining homeostasis. Here we use isobaric-labeling multiplexed proteomics to analyze protein co-regulation and show that this allows the identification of protein-protein associations with high accuracy. We apply this 'interactome mapping by high-throughput quantitative proteome analysis' (IMAHP) method to a panel of 41 breast cancer cell lines and show that deviations of the observed protein co-regulations in specific cell lines from the consensus network affects cellular fitness. Furthermore, these aberrant interactions serve as biomarkers that predict the drug sensitivity of cell lines in screens across 195 drugs. We expect that IMAHP can be broadly used to gain insight into how changing landscapes of protein-protein associations affect the phenotype of biological systems.

Abstract 543: Harnessing synthetic lethality to predict clinical outcomes of cancer treatment

Significance: The identification of Synthetic Lethal interactions (SLi) has long been considered a foundation for the advancement of cancer treatment. The rapidly accumulating large-scale patient data now provides a golden opportunity to infer SLi directly from patient samples. Here we present a new data-driven approach termed ISLE for identifying SLi, which is then shown to be predictive of clinical outcomes of cancer treatment in an unsupervised manner, for the first time.

Methods: ISLE consists of four inference steps, analyzing tumor, cell line and gene evolutionary data: It first identifies putative SL gene pairs whose co-inactivation is underrepresented in tumors, testifying that they are selected against. Second, it further prioritizes candidate SL pairs whose co-inactivation is associated with better prognosis in patients, testifying that they may hamper tumor progression. Finally, it eliminates false positive SLi using gene essentiality screens (testifying to causal SLi relations) and prioritizing SLi paired genes with similar evolutionary phylogenetic profiles.

Results: We applied ISLE to analyze the TCGA tumor collection and generated the first clinically-derived pan-cancer SL-network, composed of SLi common across many cancer types. We validated that these SLi match the known, experimentally identified SLi (AUC=0.87), and show that the SL-network is predictive of patient survival in an independent breast cancer dataset (METABRIC). Based on the predicted SLi, we predicted drug response of single agents and drug combinations in a wide variety of in vitro, mouse xenograft and patient data, altogether encompassing >700 single drugs and >5,000 drug combinations in >1,000 cell lines, 375 xenograft models and >5,000 patient samples. Of note, these predictions were performed in an unsupervised manner, reducing the known risk of over-fitting the data commonly associated with supervised prediction methods. Our prediction is based on the notion that a drug is likely to be more effective in tumors where many of its targets’ SL-partners are inactive, and drug synergism may be mediated by underlying SLi between their targets. Most importantly, we demonstrate for the first time that an SL-network can successfully predict the treatment outcome in cancer patients in multiple large-scale patient datasets including the TCGA, where SLis successfully predict patients’ response for 75% of cancer drugs.

Conclusions: ISLE is predictive of the patients’ response for the majority of current cancer drugs. Of paramount importance, the predictions of ISLE are based on SLi between (potentially) all genes in the cancer genome, thus prioritizing treatments for patients whose tumors do not bear specific actionable mutations in cancer driver genes, offering a novel approach to precision-based cancer therapy. The predictive performance of ISLE is likely to further improve with the expected rapid accumulation of additional patient data.

Citation Format: Joo Sang Lee, Avinash Das, Livnat Jerby-Arnon, Seung Gu Park, Matthew Davidson, Dikla Atias, Arnaud Amzallag, Chani Stossel, Ella Buzhor, Welles Robinson, Kuoyuan Cheng, Joshua J. Waterfall, Paul S. Meltzer, Sridhar Hannenhalli, Cyril H. Benes, Talia Golan, Emma Shanks, Eytan Ruppin. Harnessing synthetic lethality to predict clinical outcomes of cancer treatment [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 543. doi:10.1158/1538-7445.AM2017-543

Abstract B19: Targeting dependencies within apoptotic pathways through inhibition of BET bromodomains

Cancer is driven in large part by dysregulated transcriptional programs that allow for the acquisition of the many ‘Hallmarks of Cancer.' Multiple regulatory factors are essential for the establishment and maintenance of these cancer-specific transcriptional programs, and this dependence creates vulnerabilities that can be therapeutically targeted. Here we describe a selective dependence on the bromodomain and extraterminal (BET) family of proteins for the reprogramming of apoptotic signaling networks and demonstrate how this dependence can be predicted prior to therapeutic intervention. We demonstrate in phenotypically sensitive cell lines that BET inhibition results in a rapid and robust transcriptional response among regulators of apoptosis, and that this transcriptional response is correlated with changes in the apoptotic threshold of target cells and subsequent apoptosis. We show that the robustness of the apoptotic response, and not that of the cytostatic response, predicts phenotypic sensitivity to BETi. Consistent with this, we observed that acquired BETi tolerance in two disparate cellular models is driven by dysregulated expression of anti-apoptotic BCL2 family proteins, and that genetic or pharmacological manipulation of apoptotic signaling can modify the phenotypic response to BETi. We further demonstrate that the basal expression levels of a set of apoptotic factors significantly predict preclinical response to BETi, and in particular note that BETi preferentially targets those cells that are dependent on BCL2 for survival. Our findings suggest that tumor cells have acquired a dependence on BET bromodomain function to evade apoptosis, and highlight opportunities to exploit this dependence in the clinic through rational patient selection and drug combination strategies.

Citation Format: Andrew R. Conery, Richard C. Centore, Kerry L. Spillane, Nicole E. Follmer, Archana Bommi-Reddy, Charlie Hatton, Barbara M. Bryant, Patricia Greninger, Arnaud Amzallag, Cyril H. Benes, Jennifer A. Mertz, Robert J. Sims, III. Targeting dependencies within apoptotic pathways through inhibition of BET bromodomains. [abstract]. In: Proceedings of the AACR Precision Medicine Series: Targeting the Vulnerabilities of Cancer; May 16-19, 2016; Miami, FL. Philadelphia (PA): AACR; Clin Cancer Res 2017;23(1_Suppl):Abstract nr B19.

Preclinical Anticancer Efficacy of BET Bromodomain Inhibitors Is Determined by the Apoptotic Response

Small-molecule inhibitors of the bromodomain and extraterminal (BET) family of proteins are being tested in clinical trials for a variety of cancers, but patient selection strategies remain limited. This challenge is partly attributed to the heterogeneous responses elicited by BET inhibition (BETi), including cellular differentiation, senescence, and death. In this study, we performed phenotypic and gene-expression analyses of treatment-naive and engineered tolerant cell lines representing human melanoma and leukemia to elucidate the dominant features defining response to BETi. We found that de novo and acquired tolerance to BETi is driven by the robustness of the apoptotic response, and that genetic or pharmacologic manipulation of the apoptotic signaling network can modify the phenotypic response to BETi. We further reveal that the expression signatures of the apoptotic genes BCL2, BCL2L1, and BAD significantly predict response to BETi. Taken together, our findings highlight the apoptotic program as a determinant of response to BETi, and provide a molecular basis for patient stratification and combination therapy development.

Landscape of Targeted Anti-Cancer Drug Synergies in Melanoma Identifies a Novel BRAF-VEGFR/PDGFR Combination Treatment

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 BRAF^V600E 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.

Abstract 4690: Next-generation screen for integrative subtyping and target discovery for KRAS-mutant cancer

Mutations in the small GTPase, KRAS, are found in ∼140,000 new cases of cancer every year in the United States. This heterogeneous class of cancers manifests primarily as adenocarcinomas of the lung, colon and pancreas. These cancers display a wide spectrum of KRAS-dependency and differentially activate downstream effector signaling. The tumors further diverge in their array of co-occurring secondary mutations, expression signatures and KRAS mutant allele. Ultimately, the sole trait these cancers share in common is an obstinate resistance to chemo- and targeted-therapies, making identification of effective treatments an urgent need. To identify treatments for such a heterogeneous class of cancers, we developed a strategy to stratify KRAS-mutant cell lines into subtypes by integrating next-generation RNAi screening and “Omics” database mining. Each subtype is characterized by unique biomarkers and distinct patterns of effector dependency, both of which represent potential targets for personalized therapeutic strategies.

Our RNAi screen systematically evaluates sensitivity to siRNA-mediated knockdown of 40 KRAS effector nodes in a panel of 135 lung, colorectal and pancreatic cancer cell lines. Data is analyzed on the single cell level, through the simultaneous measurement of 5 functional parameters. This single-cell, multi-dimensional approach allows for a comprehensive assessment of cellular homeostasis, with unprecedented depth and dynamic range that allows robust classification of cell lines by similarity. We identify subtypes of KRAS-mutant cell lines that rely on particular effector pathways such as the RAL, RSK, MTOR and autophagy pathways, which are not engaged by all KRAS-mutant cell lines, and thus may represent targets for personalized treatment. We further identify widely shared dependencies such as on the RAF, glycolytic and cell cycle pathways. Through integrative data mining of exome, transcriptome and drug/siRNA sensitivity databases for each KRAS-mutant subtype, we can identify unique biomarkers that will serve to stratify patients in the clinic and recommend personalized treatment strategies.

Citation Format: Tina L. Yuan, Rachel Bagni, Ming Yi, Arnaud Amzallag, Shervin Afghani, Katie Beam, William Burgan, Nicole Fer, Leslie Garvey, Brian Smith, Andrew Waters, Robert Stephens, Cyril Benes, Frank McCormick. Next-generation screen for integrative subtyping and target discovery for KRAS-mutant cancer. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 4690. doi:10.1158/1538-7445.AM2015-4690

Abstract LB-247: Querying the RAS genomic network with siRNAs and and flow cytometry: Automatic, multidimensional phenotyping of 135 cancer cell lines by Gaussian mixture fitting and expectation maximization

To discover novel therapeutic modalities and genomic predictors of response, large screen utilizing small molecules or sh/siRNA are performed on increasingly large collections of cancer cell lines. However these screens suffer from two main limitations: 1) the off-target effects of the probes 2) the coarse measurement of the cellular response that cannot distinguish between different outcomes such as proliferation block and apoptosis.

Here we profile 50 lung cancer cell lines using highly specific combinations of siRNAs against effector nodes of KRAS, and measured several characteristics of phenotypic response by flow cytometry including viability, level of reactive oxygen species and cell membrane integrity.

For each assay [node-cell line], typically 25,000 events were measured. We often observed multi modal distributions following node silencing. We used the expectation maximization algorithm to fit the different cell populations induced by the gene silencing. This allows us to automatically extract the proportion of dying cells, and also provides estimates of the cell growth impairment. Assessment of replicates shows that our results are highly reproducible, provide an accurate estimate of the proportion of dying cells, and reveal the complexity of multi-modal response of KRAS cancer cell lines to perturbation of key signaling nodes. Using the genomic profiling and pharmaceutical screen of the same cell lines, we reveal the association of specific node silencing with 1) the basal genomic state of the cell lines 2) the activity of a panel of drugs with the goal of identifying new targeting modalities and patient selection strategies for KRAS mutant cancers.

Citation Format: Arnaud Amzallag, Tina L. Yuan, Rachel Bagni, Ming Yi, Robert Stephens, Sridhar Ramawamy, Frank McCormick, Cyril H. Benes. Querying the RAS genomic network with siRNAs and and flow cytometry: Automatic, multidimensional phenotyping of 135 cancer cell lines by Gaussian mixture fitting and expectation maximization. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr LB-247. doi:10.1158/1538-7445.AM2015-LB-247

The WTX Tumor Suppressor Interacts with the Transcriptional Corepressor TRIM28

WTX encodes a tumor suppressor implicated in the pediatric kidney cancer Wilms tumor and in mesenchymal differentiation with potentially distinct functions in the cytoplasm, at the plasma membrane, and in the nucleus. Although modulating components of the WNT signaling pathway is a proposed function for cytoplasmic and membrane-bound WTX, its nuclear properties are not well understood. Here we report that the transcriptional corepressor TRIM28 is the major binding partner for nuclear WTX. WTX interacted with the coiled coil domain of TRIM28 required for its binding to Krüppel-associated box domains of transcription factors and for its chromatin recruitment through its own coiled coil and proline-rich domains. Knockdown of endogenous WTX reduced the recruitment of TRIM28 to a chromatinized reporter sequence and its ability to repress a target transcript. In mouse embryonic stem cells where TRIM28 plays a major role in repressing endogenous retroviruses and long interspersed elements, knockdown of either TRIM28 or WTX combined with single molecule RNA sequencing revealed a highly significant shared set of differentially regulated transcripts, including derepression of non-coding repetitive sequences and their neighboring protein encoding genes (p < 1e-20). In mesenchymal precursor cells, depletion of WTX and TRIM28 resulted in analogous β-catenin-independent defects in adipogenic and osteogenic differentiation, and knockdown of WTX reduced TRIM28 binding to Pparγ promoter. Together, the physical and functional interaction between WTX and TRIM28 suggests that the nuclear fraction of WTX plays a role in epigenetic silencing, an effect that may contribute to its function as a regulator of cellular differentiation and tumorigenesis.

Patient-derived models of acquired resistance can identify effective drug combinations for cancer

Targeted cancer therapies have produced substantial clinical responses, but most tumors develop resistance to these drugs. Here, we describe a pharmacogenomic platform that facilitates rapid discovery of drug combinations that can overcome resistance. We established cell culture models derived from biopsy samples of lung cancer patients whose disease had progressed while on treatment with epidermal growth factor receptor (EGFR) or anaplastic lymphoma kinase (ALK) tyrosine kinase inhibitors and then subjected these cells to genetic analyses and a pharmacological screen. Multiple effective drug combinations were identified. For example, the combination of ALK and MAPK kinase (MEK) inhibitors was active in an ALK-positive resistant tumor that had developed a MAP2K1 activating mutation, and the combination of EGFR and fibroblast growth factor receptor (FGFR) inhibitors was active in an EGFR mutant resistant cancer with a mutation in FGFR3. Combined ALK and SRC (pp60c-src) inhibition was effective in several ALK-driven patient-derived models, a result not predicted by genetic analysis alone. With further refinements, this strategy could help direct therapeutic choices for individual patients.

dREAM co-operates with insulator-binding proteins and regulates expression at divergently paired genes

dREAM complexes represent the predominant form of E2F/RBF repressor complexes in Drosophila. dREAM associates with thousands of sites in the fly genome but its mechanism of action is unknown. To understand the genomic context in which dREAM acts we examined the distribution and localization of Drosophila E2F and dREAM proteins. Here we report a striking and unexpected overlap between dE2F2/dREAM sites and binding sites for the insulator-binding proteins CP190 and Beaf-32. Genetic assays show that these components functionally co-operate and chromatin immunoprecipitation experiments on mutant animals demonstrate that dE2F2 is important for association of CP190 with chromatin. dE2F2/dREAM binding sites are enriched at divergently transcribed genes, and the majority of genes upregulated by dE2F2 depletion represent the repressed half of a differentially expressed, divergently transcribed pair of genes. Analysis of mutant animals confirms that dREAM and CP190 are similarly required for transcriptional integrity at these gene pairs and suggest that dREAM functions in concert with CP190 to establish boundaries between repressed/activated genes. Consistent with the idea that dREAM co-operates with insulator-binding proteins, genomic regions bound by dREAM possess enhancer-blocking activity that depends on multiple dREAM components. These findings suggest that dREAM functions in the organization of transcriptional domains.

Distinct, strict requirements for Gfi-1b in adult bone marrow red cell and platelet generation

Strict, lineage-intrinsic requirement for continuous adult Gfi-1b expression at two distinct critical stages of erythropoiesis and megakaryopoiesis.

The zinc finger transcriptional repressor Gfi-1b is essential for erythroid and megakaryocytic development in the embryo. Its roles in the maintenance of bone marrow erythropoiesis and thrombopoiesis have not been defined. We investigated Gfi-1b’s adult functions using a loxP-flanked Gfi-1b allele in combination with a novel doxycycline-inducible Cre transgene that efficiently mediates recombination in the bone marrow. We reveal strict, lineage-intrinsic requirements for continuous adult Gfi-1b expression at two distinct critical stages of erythropoiesis and megakaryopoiesis. Induced disruption of Gfi-1b was lethal within 3 wk with severely reduced hemoglobin levels and platelet counts. The erythroid lineage was arrested early in bipotential progenitors, which did not give rise to mature erythroid cells in vitro or in vivo. Yet Gfi-1b^−/− progenitors had initiated the erythroid program as they expressed many lineage-restricted genes, including Klf1/Eklf and Erythropoietin receptor. In contrast, the megakaryocytic lineage developed beyond the progenitor stage in Gfi-1b’s absence and was arrested at the promegakaryocyte stage, after nuclear polyploidization, but before cytoplasmic maturation. Genome-wide analyses revealed that Gfi-1b directly regulates a wide spectrum of megakaryocytic and erythroid genes, predominantly repressing their expression. Together our study establishes Gfi-1b as a master transcriptional repressor of adult erythropoiesis and thrombopoiesis.

Bending modes of DNA directly addressed by cryo-electron microscopy of DNA minicircles

We use cryo-electron microscopy (cryo-EM) to study the 3D shapes of 94-bp-long DNA minicircles and address the question of whether cyclization of such short DNA molecules necessitates the formation of sharp, localized kinks in DNA or whether the necessary bending can be redistributed and accomplished within the limits of the elastic, standard model of DNA flexibility. By comparing the shapes of covalently closed, nicked and gapped DNA minicircles, we conclude that 94-bp-long covalently closed and nicked DNA minicircles do not show sharp kinks while gapped DNA molecules, containing very flexible single-stranded regions, do show sharp kinks. We corroborate the results of cryo-EM studies by using Bal31 nuclease to probe for the existence of kinks in 94-bp-long minicircles.

Probabilistic base calling of Solexa sequencing data

Background

Solexa/Illumina short-read ultra-high throughput DNA sequencing technology produces millions of short tags (up to 36 bases) by parallel sequencing-by-synthesis of DNA colonies. The processing and statistical analysis of such high-throughput data poses new challenges; currently a fair proportion of the tags are routinely discarded due to an inability to match them to a reference sequence, thereby reducing the effective throughput of the technology.

Results

We propose a novel base calling algorithm using model-based clustering and probability theory to identify ambiguous bases and code them with IUPAC symbols. We also select optimal sub-tags using a score based on information content to remove uncertain bases towards the ends of the reads.

Conclusion

We show that the method improves genome coverage and number of usable tags as compared with Solexa's data processing pipeline by an average of 15%. An R package is provided which allows fast and accurate base calling of Solexa's fluorescence intensity files and the production of informative diagnostic plots.

3D reconstruction and comparison of shapes of DNA minicircles observed by cryo-electron microscopy

We use cryo-electron microscopy to compare 3D shapes of 158 bp long DNA minicircles that differ only in the sequence within an 18 bp block containing either a TATA box or a catabolite activator protein binding site. We present a sorting algorithm that correlates the reconstructed shapes and groups them into distinct categories. We conclude that the presence of the TATA box sequence, which is believed to be easily bent, does not significantly affect the observed shapes.

[Coronary occlusion immediately following a successful coronary angioplasty. Treatment by repeat angioplasty]

The authors report the case of a 52 year old patient with a significant stenosis of the medial portion of the left anterior descending artery (LAD) with excellent left ventricular function. Transluminal coronary angioplasty (TCA) was indicated following a positive exercise stress test. This was initially performed successfully. Fifteen minutes after the end of the procedure, a total obstruction occurred at the site of dilatation immediately eliciting significant precordial chest pain and massive elevation of the ST segment. Isosorbide dinitrate (ISDN) at a dose of 2 mg was injected into the artery 3 times without success as was an attempt to pass through the obstruction with a guide wire. Another TCA was then attempted without administration of the thrombolytic agent. The dilating catheter passed easily by the obstruction permitting several dilatations which restored rapid coronary artery flow, relieved completely the chest pain, and normalized electrocardiographic abnormalities. This procedure represents a new therapeutic approach to obstruction, an often unpredictable and serious complication of coronary angioplasty in the absence of collateral circulation, thereby preventing the development of a myocardial infarction and an emergency aortocoronary bypass operation.

Genetic fine structure of the leucine operon of Escherichia coli K-12

The order of mutational sites in 10 independently isolated leucine auxotrophys of Escherichia coli K-12 was determined by three-point reciprocal transductions. The sites of mutation mapped in linear sequence in a cluster; all leucine auxotrophic mutations were cotransducible with mutations in the arabinose operon. The mutations were assigned to four complementation groups by abortive transduction tests, designated D, C, B, and A, reading in a clockwise direction from the arabinose operon. Enzyme analyses showed that strains with a mutation in gene A lacked alpha-isopropylmalate synthetase activity (EC 4.1.3), and those with a mutation in gene B lacked beta-isopropylmalate dehydrogenase activity (EC 1.1.1). It is concluded that the gross structure of the leucine operon in E. coli is closely similar to, if not identical with, the gross structure of the leucine operon in Salmonella typhimurium.