Chemical Systems Biology Research Articles

Abstract IA20: General methods to conditionally regulate protein stability

Abstract There is an increasing need for experimental perturbation methods that provide scientists with a high degree of control over the proteins they study. In order to provide scientists with general techniques to conditionally regulate protein function, we have developed experimental methods to regulate protein stability rapidly and reversibly through the administration of cell-permeable small molecules to cultured cells or living animals. These perturbation strategies depend upon engineered protein domains called “destabilizing domains” that bind tightly to small molecule ligands. When a destabilizing domain is fused to a targeted protein-of-interest, investigators can tunably regulate the expression levels of the targeted fusion protein through the administration of the appropriate ligand. We have developed several experimental systems in which the addition of a ligand can be used to either stabilize or destabilize a destabilizing domain, thereby restoring or eliminating the function of the protein-of-interest. Citation Format: Thomas J. Wandless. General methods to conditionally regulate protein stability [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA20.

Abstract A35: Predicting drugs effect on transcription factor activity by regulatory network analysis of drug perturbed cellular states

Abstract Phenotypic changes during pathophysiological processes can be efficiently captured by genome-wide gene expression signatures or molecular phenotypes. Transcription factors (TFs) are the key components of the transcriptional regulatory network that integrates intracellular and extracellular stimuli and generates a response in the form of specific gene expression signatures. We have shown the existence and dissected the components of this integrative logic layer in several pathologic and physiologic processes. Moreover, perturbation of key components of this machinery can be efficiently used to induce desired phenotypic changes. However, TFs have been rather elusive targets for pharmacological intervention. Here, by using a cell-context specific network-based approach, we generated a comprehensive map between small compounds and TF activity. We leveraged the large collection of expression profiles in response to small molecule perturbagens from CMAP to first reverse-engineer transcriptionally regulatory networks in 3 different cellular context (Breast: MCF7, Prostate:PC3 and Myeloma:HL60) by the ARACNe algorithm, and then interrogate these networks with the MARINa algorithm to identify TFs whose activity is affected by the profiled compounds. We show two different evidences that validate our method. First, we correctly identified ESR1 as the very top TF whose activity is affected by the estrogen antagonists tamoxifen, raloxifen and clomifene. Second, we selected 10 compounds from the 1,400 represented in the MCF7 CMAP dataset that our method predict as STAT3 perturbagens. From these, 4 showed a significant inhibition of STAT3 activity in a cell-based STAT3 reported assay. Citation Format: Yao Shen, Mariano Alvarez, Sergey Pampou, Jorida Coku, Andrea Califano. Predicting drugs effect on transcription factor activity by regulatory network analysis of drug perturbed cellular states [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr A36.

Abstract IA13: Super-resolution fluorescence microscopy of cells and tissues

Abstract Light microscopy is an essential tool in biological research. However, the spatial resolution of light microscopy, classically limited by diffraction to a few hundred nanometers, is substantially larger than typical molecular length scales in cells. Hence many subcellular structures cannot be resolved by conventional light microscopy. We recently developed a new form of super-resolution fluorescence microscopy, stochastic optical reconstruction microscopy (STORM), that breaks the diffraction limit. STORM uses single-molecule imaging and photo-switchable fluorescent probes to temporally separate the spatially overlapping images of individual molecules. This approach allows the localization of fluorescent probes with nanometer precision and the construction of sub-diffraction-limit images. Using this method, we have achieved multicolor and three-dimensional (3-D) imaging of live cells and tissues with nanometer-scale resolution. In this talk, I will discuss the general concept, recent technological advances and biological applications of STORM. Citation Format: Xiaowei Zhuang. Super-resolution fluorescence microscopy of cells and tissues [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA13.

Abstract IA1: Cancer dependencies defined by genomic alterations and targeted by small molecules

Abstract The ability to sequence cancer genomes coupled with advances in small-molecule science provides a new foundation for creating safe and effective cancer therapeutics – ones delivered to patients based on the specific genomic alterations present in their cancer. Several examples now exist of small-molecule cancer drugs targeting protein kinases encoded by both wild-type and oncogenic alleles, and yielding high clinical response rates in patients with specific genetic features. However, these drugs target only a narrow range of proteins, benefit <1% of patients suffering today from cancer, and provide therapeutic benefits that are not always durable. All current genetically matched drugs target the driver oncogene or its wild-type allele directly (“oncogene dependencies”). It is unknown whether similar clinical responses will result from drugs targeting non-oncogenes working in cooperation with oncogenes in patients with cancers having the relevant genetic features (“non-oncogene co-dependencies”). To accelerate the discovery of genetically matched therapies, systematic approaches are needed to identify: 1) the direct and indirect dependencies cancers acquire as a consequence of their specific mutations, translocations, and copy number alterations, and 2) small-molecule drugs that target the dependencies. Genomic cancer cell line profiling has been used to reveal patterns of small-molecule sensitivities across diverse cancer cell lines (CCLs). These efforts initially focused on relating sensitivity to the lineage of CCLs (ref NCI-60 and others), but now increasingly focus on relating sensitivity to genetic and epigenetic features of the cell lines. But these are still early days of genomic CCL profiling – there are many issues to iron out. Nevertheless, there are also signs that this approach will eventually be fruitful and impacting. For example, there are no examples of targeted cancer therapeutics today that were not predicted by genomic CCL profiling. In my lecture, I will describe an Interactive Resource to analyze and visualize a rich body of quantitative CCL sensitivity measurements with the aim of identifying novel candidate cancer dependencies targeted by small molecules. To enable the Resource, we measured the sensitivity of 243 genetically characterized CCLs to 355 small-molecule probes and drugs (the ‘Informer Set’) at eight concentrations in duplicate. The CCLs are part of a 949-member Cancer Cell Line Encyclopedia (CCLE) whose extensive genomic characterization includes the mutational status of >1,600 genes, global assessment of DNA copy number, and global gene expression analysis. These characterizations are also available freely via an interactive web portal (broadinstitute.org/ccle). The compounds included in the Informer Set are known to have selective interactions with their target, and collectively target many distinct nodes in cancer cell circuitry, including but not limited to pathways modulating apoptosis, oxidative stress, chromatin signaling, mitotic stress, hypoxic stress, proteotoxic stress, and metabolism. Using robust analytical methods including cell-line and compound filtering, we correlated the sensitivity measurements with genomic alterations in cancer cell lines to identify dependencies conferred by specific genotypes. I will describe the current content of the Resource and provide multiple illustrations of its use, including the use of additional analytical tools not included in the Resource but readily and freely accessible elsewhere. Citation Format: Stuart L. Schreiber. Cancer dependencies defined by genomic alterations and targeted by small molecules [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA1.

Abstract PR6: A cooperative molecular response to combined mTOR/HDAC inhibition revealed by transcriptional co-expression analysis

Abstract Identification of biologically relevant molecular targets responding cooperatively to drug combinations is challenging, yet potentially useful in predicting clinical outcomes. We found that combining sirolimus (rapamycin), an inhibitor of mechanistic target of rapamycin (mTOR), with entinostat (MS-275), a selective class I histone deacytlase (HDAC) inhibitor, was synergistic in controlling 90% of tested cell lines derived from B cell malignancies, including multiple myeloma, mantle cell lymphoma, and murine plasmacytoma, in vitro, and effective in limiting xenograft growth in vivo. To understand the transcriptional underpinnings of the drug synergy, weighted gene co-expression analysis (WGCNA) was adapted to analyze gene expression profile data from cells treated with sirolimus and entinostat individually and in combination. WGCNA delineated the contribution of each inhibitor to the overall gene expression change of the combination by considering not only measures of fold-change and significance testing, but also the degree of gene expression interconnectedness. A co-expression network consisting of five gene modules was defined, where each module represents a unique transcriptional effect of the drug combination. In the next phase of analysis, each gene module was individually tested for functional and clinical enrichment. The module containing genes cooperatively affected by both compounds was highly enriched (p<0.001) for genes involved in cell cycle, immune recognition, and DNA damage/repair, which we have validated. Gene set enrichment analysis of each drug-induced gene expression module demonstrated that genes of all five modules were significantly enriched when healthy donors were compared to MM patients. Finally, interrogation of the cooperative drug signature in patient GEP datasets with survival annotation found it predictive of increased survival (p<0.01), thus linking the drug combination-induced transcriptional changes to predictions for enhanced survival. Thus, systems-level genomic approaches identified markers and a gene signature associated with drug combination activity, disease specificity, and clinical potential. This proffered talk is also presented as Poster A16. Citation Format: John Kyle Simmons, Beverly A. Mock, Aleksandra Michalowski, Jyoti Patel, Bih-Rong Wei, R. Mark Simpson, W. Michael Kuehl, Shuling Zhang, Ke Zhang, Ola Landgren. Sorger. A cooperative molecular response to combined mTOR/HDAC inhibition revealed by transcriptional co-expression analysis [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR6.

Abstract IA11: Functionalization the cancer genome

Abstract As we begin to implement next-generation sequencing into patient care, a central challenge will involve determining which aberrations in a given tumor represent “drivers” that determine tumor behavior versus “passengers” resultant from the inherent instability of cancer genomes. Critically even within cancer genes, many aberrations could still be passengers. Indeed, given the broad landscape of aberrations in human tumors, it is possible that genomic aberrations in a cancer gene could be hypomorphs (decreasing activity of a tumor suppressor gene), hypermorphs (increasing activity of an oncogene) or neomorphs (changing function of the protein). Neomorphs such as those engendered by mutations in IDH1 in glioma could be particularly challenging as their therapeutics liabilities are not likely to be predictable from known functions of the wild-type protein. Targeting driver aberrations should improve outcomes for patients, while targeting passenger aberrations would be without benefit and may instead result in unnecessary toxicity and delay implementation of effective therapies. It will also be critical to select the drug or drugs most likely to benefit each patient based on underlying aberrations in their own tumor. Unfortunately, there is a complete lack of practical high-throughput platforms able to identify and select optimal therapies targeting driver aberrations. We have implemented a high-throughput pipeline that consists of generation of patient tumor mutations in open reading frames (ORF) and determining the effect of shRNA, wild-type ORFs and mutant ORFs in the BAF3 IL3 dependent sensor line. Aberrations that increase cell viability and proliferation are established as stable lines. Stable lines are then probed with a chemical genomic library of tool compounds targeting critical pathways as well as through a high-throughput reverse phase protein array (RPPA) functional proteomics platform. The intersection of functional effects in the platforms identifies potential drivers and elicits their mechanistic activity and potential therapeutic liabilities. This pipeline has the potential to elicit tumor drivers and identify individualized therapeutic approaches. Citation Format: Gordon B. Mills, Yiling Lu, Wai Ting Cheung, Kenneth Scott, Han Liang. Functionalization the cancer genome [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA11.

Abstract PR9: Investigating combinatorial ligand addiction provides insights into rational drug combinations in cancer therapy

Abstract Cancer, the second most common cause of death in the United States, is a collection of diseases caused by uncontrolled cell growth and metastasis. The main treatment for cancer is chemotherapy, which generally kills fast growing cells nonspecifically and has many side effects. A different type of cancer treatment, called targeted therapy, aims to avoid general toxicity by using drugs that block the activity of specific gene products, usually encoded by oncogenes, which have been shown to drive tumor growth. To date, targeted therapies, alone or in combination with chemotherapies, have mainly been successful in rare subsets of patients with tumors addicted to single oncogenes. This has created a rationale to mainly treat patients with an oncogene-addiction (such as those carrying mutated or overexpressed kinases) with targeted therapies like erlotinib and trastuzumab, which inhibit human epidermal growth factor receptor (EGFR) and human epidermal growth factor receptor 2 (HER2/ErbB2), respectively. Here, evidence is provided that targeted therapies are also effective in tumors that are dependent on multiple growth factors – a phenomenon that is called combinatorial ligand addiction. Specifically, it is shown that ligands that bind the EGFR family and the hepatocyte growth factor receptor (HGFR/MET) can activate protein kinase B (PKB/AKT) across a broad set of cancer cell lines, suggesting that ligand signaling is redundant and widespread. It is also shown that ErbB ligands have distinct signaling dynamics and strengths, which provides a rationale for investigating each component of the ErbB signaling network. Using a systematic approach, we found that ErbB3 is an important therapeutic target even though it is not overexpressed and lacks kinase activity. Furthermore, it is shown that cell lines with and without known oncogene-addiction express autocrine ligands and have improved growth inhibition with drug combinations that include autocrine ligand-blocking antibodies. This research demonstrates that combinatorial ligand addiction creates a new rationale for therapeutic combinations to improve efficacy and prevent resistance in cancer cells that are treated with current targeted drugs. This proffered talk is also presented as Poster A22. Citation Format: Emily A. Pace, Ulrik B. Nielsen, Birgit Schoeberl, Diana H. Chai, Anand Parikh, Ashish Kalra, Shinji Oyama, Bryan Johnson, Gege Tan, Aaron Fulgham, Raghida Bukhalid. Investigating combinatorial ligand addiction provides insights into rational drug combinations in cancer therapy [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR9.

Abstract IA4: Designing biological systems to report on disease

Abstract Biology presents us with an array of design principles. From studies of both simple and more complex systems, we understand some of the fundamentals of how nature works and what goes wrong in the disease state. We are interested in using the foundations of biology to engineer cells in a logical and predictable way to perform certain functions with relevance to therapeutics and diagnostics. By necessity, the predictable engineering of biology requires knowledge of quantitative behavior of individual cells and communities and the ability to construct reliable models. By building and analyzing synthetic systems, we learn more about the fundamentals of biological design as well as engineer useful living devices with myriad applications. For example, we are interested in building cells that can perform specific tasks, such as remembering past events and thus acting as a biological computer. In particular, we have built genetic circuits whereby cells can ‘remember’ stimuli for many generations including DNA damage and hypoxia. By characterizing these ‘memory’ cells, we can identify key genes that may act as therapeutic targets and explanations for persistence of a subset of cells in tumors. Citation Format: Pamela A. Silver. Designing biological systems to report on disease [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA4.

Abstract PR5: The receptor tyrosine kinase layer of breast cancer cell lines is predictive of the response to therapeutic drugs

Abstract Currently, clinicians use the levels of the hormone receptors for estrogen (ER) and progesterone (PR) and amplifications of the ErbB2 receptor tyrosine kinase (RTK) to distinguish three groups of breast cancers and guide the choice of therapeutic treatments. In this work we show that a limited number of measurements characterizing breast cancer cell lines at the level of RTKs can be used to predict cell line sensitivity to therapeutic inhibitors. We chose a panel of approximately 40 breast cancer cell lines, treated them with 22 different growth factors and cytokines at a saturating and subsaturating dose for 10, 30, and 90 min and measured the phosphorylation of key kinases downstream of the receptors. In addition, we measured the expression and phosphorylation levels of approximately 20 RTKs under serum starvation conditions. We used the measurements to predict the GI50 values of the cell lines in response to over 70 therapeutic inhibitors by partial least square regression. We found that our measurements are sufficient to predict the response to about half of the drugs studied. The majority of drugs we are able to predict target RTKs directly or kinases immediately downstream of the receptors. In some cases we are able to predict the response to drugs that target processes seemingly unrelated to our measurements. We are currently exploring the data to uncover what drives sensitivity and resistance to drugs, gain mechanistic insight in the action of cancer therapeutics, and find predictors of drug response that may be useful in a clinical setting. Our results suggest that a carefully selected set of measurements might be sufficient to classify responsiveness of breast cancers to a number of cancer drugs currently used in the clinic. This is particularly relevant in the case of cancers like triple-negative breast cancers for which choosing effective treatment has proven difficult. This proffered talk is also presented as Poster A17.This proffered talk is also presented as Poster A17. Citation Format: Mario Niepel, Marc Hafner, Emily Pace, Birgit Schoeberl, Peter K. Sorger. The receptor tyrosine kinase layer of breast cancer cell lines is predictive of the response to therapeutic drugs [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR5.

Abstract A15: High-Content screen for inhibitors of cell migration in cancer metastasis using adenoviral knock-down

Abstract Enhanced cell migration is a hallmark of metastatic cancer cells. The propensity of cancer cells to close an open wound in a cell monolayer is thought to predict this ability. Using our adenoviral shRNA knockdown library we have established a high-throughput wound healing assay to identify novel genes involved in cell migration. A 96-pin tool was designed to apply a constant mechanical scratch-wound in the cellular monolayer. We used transmitted light imaging for segmentation and quantification of the scratch wound that remained open. Genes whose knockdown inhibit cell migration can be identified by their effect on the open wound. We demonstrated that two knockdown constructs targeting a known player in motility, CXCR4, inhibit wound healing, thereby validating our approach. Using this wound healing assay we have identified a number of novel genes associated with cancer cell motility. These targets have consecutively been validated for their role in 3-D invasion using Boyden chambers. As our adenoviral knockdown libraries focus on druggable targets, these validated targets can quickly be employed to generate small-molecule compounds or antibody therapeutics targeting cancer metastasis. Citation Format: Remko de Pril, Annemarie Lekkerkerker, Desire van Steenhoven, Ilhem Maghrani, Tim Perera, Janine Arts, David Fischer, Richard Janssen. High-Content screen for inhibitors of cell migration in cancer metastasis using adenoviral knock-down [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr A15.

Abstract A21: High-throughput small molecule screening of genetically diverse glioblastoma stem cells

Abstract Survival statistics for glioblastoma multiforme (GBM) have remained static despite years of research. Mounting evidence suggests that glioblastoma stem cells (GSCs) are a primary cause of GBM growth, spread, vascularity, and chemotherapeutic resistance. Increasing survival, therefore, demands a therapeutic strategy with efficacy against GSCs. To address this, we have conducted a high-throughput screen of over 3,000 small molecule compounds for activity against three independently derived GSC lines that vary in their expression of TCGA subtype-related genes. Our fluorescence-based assay is sensitive, reliable, and largely free of row- and column-based systematic errors. We identify novel compounds that impair the viability of all three of these genetically diverse GSC lines but have comparably limited toxicity against a normal human glial cell line. Many of these compounds are small and lipophilic, as well, making them ideal candidates for follow-up in vivo evaluation either through intravenous delivery or through encapsulation in polymeric nanoparticles. Citation Format: Garth W. Strohbehn, Jiangbing Zhou, Michael Fu, Toral R. Patel, Joseph M. Piepmeier, W. Mark Saltzman. High-throughput small molecule screening of genetically diverse glioblastoma stem cells [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr A21.

Abstract PR2: A systems biology approach to elucidating the consequences of complex ternary interactions of heparin, FGF ligands, and FGF receptor on downstream signaling in NSCLC cells

Abstract Autocrine/paracrine FGF signaling has been implicated in non-small cell lung cancer (NSCLC) development and progression. In multiple NSCLC cell lines, a switch from FGFR2-IIIb expressed in lung epithelial cells to FGFR1-IIIc and FGFR2-IIIc, results in autocrine response to FGF2 and FGF9 ligands also expressed by these cell lines. FGF2 and FGFR1 expression have also been detected in patient samples and shown to correlate with short survival times. The FGF signaling complex involves a ternary interaction of FGF ligand, receptor, and heparin sulfate glycosaminoglycans (HSGAGs). HSGAGs bind directly to both ligand and receptor, and can serve both to stabilize the receptor-ligand complex and to sequester autocrine/paracrine ligands, increasing local concentrations. A systems approach combining experiment and modeling was used to understand the complex interplay of these components and their implications on cell signaling and behavior. We focused on the signaling behavior of NSCLC cell line NCI-H1703, which predominantly expressed FGFR1. We interrogated pathway dynamics with both FGF2 and FGF9 ligands and found varying Erk1/2 activation kinetics. Using a model encompassing the ternary complex formation in the cellular membrane, ligand-mediated receptor dimerization and phosphorylation, and subsequent kinase cascade to activate Erk1/2, we were able to identify potential mechanisms driving these differences. Additional perturbations of extracellular interactions and intracellular signaling were able to support these hypotheses. These results highlight the role of the extracellular interactions between ligand, receptor, and HSGAGs in controlling downstream behavior. This proffered talk is also presented as Poster A5. Citation Format: Diana H. Chai, Jaeyeon Kim, Jannik Vollmer, Andrijana Radivojevic, Jitendra Kanodia, Marco Muda, Birgit Schoeberl. A systems biology approach to elucidating the consequences of complex ternary interactions of heparin, FGF ligands, and FGF receptor on downstream signaling in NSCLC cells [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR2.

Abstract IA8: Genetic approaches to cancer

Abstract Breast cancer is a collection of diseases with distinct clinical behaviors and underlying genetic causes. We have searched genetically for genes that have cancer relevant phenotypes including genes that alter cellular proliferation, cellular transformation, cell survival, cellular senescence, and genes that are essential for the proliferation of cancer cells. We have approached this by the generation of libraries of shRNAs for loss of function experiments and libraries of ORFs for gain of function experiments. We will discuss these technologies and how they can be applied to the functional dissection of genes important for breast cancer. Citation Format: Nicole L. Solimini, Andrea L. Bredemeyer, Mamie Z. Li, Rameen Beroukhim, Matthew Meyerson, Stephen J. Elledge, Natalya N. Pavlova, Qikai Xu, Craig H. Mermel, Anthony C. Liang, Michael R. Schlabach, Ji Luo, Anna E. Burrows, Anthony N. Anselmo. Genetic approaches to cancer [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA8.

Abstract PR7: Visualizing the dynamic and heterogeneous responses of single cells to inhibitors of the growth factor signaling network

Abstract The growth factor signaling network, including the Akt, ERK, mTOR and other pathways, is deregulated in many diseases and is a prominent target for therapeutic inhibitors. The effectiveness of these inhibitors is limited by the existence of homeostatic feedback loops that act to restore signaling and attenuate the effects of inhibition. To obtain the single-cell data necessary to fully characterize these rapid and heterogeneous responses, we developed a library of mammary epithelial cells stably expressing over 90 fluorescent protein-based reporters. This library comprises reporters for: cellular processes including cell death, cell cycle, protein translation, and metabolism; kinases including ERK, Akt and AMPK; and microRNAs involved in the regulation of cellular differentiation and stress response. Using high-throughput live-cell microscopy and computational image analysis, we monitored the responses of these pathways to inhibitors of EGFR, ErbB2, PI3K, mTOR, MEK, Raf, and Bcl-2. This analysis revealed unexpectedly dynamic fluctuations in key pathways including ERK, Akt, and glycolysis. In one example, inhibition of EGFR transformed constitutive ERK activity into intermittent bursts of activity ranging in duration from 20-90 minutes with high cell-to-cell variability. Characterization of ERK activity pulses revealed that their frequency depended on the level of EGF receptor activity and that their duration was essential for controlling the commitment to cell proliferation. In sharp contrast, inhibition of MEK altered the amplitude, rather than the frequency, of ERK activity. Using these live-cell measurements in combination with high content immunofluorescence, we constructed a computational model of feedback loops controlling ERK activity pulses. Quantitative mapping of ERK output to proliferation rate revealed a distinct and exquisitely sensitive threshold of ERK output required for proliferation. Potential strategies for reducing ERK output below this threshold and effectively blocking proliferation include dual inhibition of upstream regulators or intermittent high-dose inhibition. Single-cell quantitative analysis of inhibitor responses is now being extended to other pathways to identify signaling dynamics and information transfer properties that enable effective therapeutic targeting of the network. This proffered talk is also presented as Poster A19. Citation Format: John Albeck, Yin P. Hung, Gregory Leya, Gary Yellen, Gordon Mills, Joan Brugge. Visualizing the dynamic and heterogeneous responses of single cells to inhibitors of the growth factor signaling network [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR7.

Abstract IA15: Predicting drug sensitivity from cancer cell lines

Abstract High-throughput molecular characterization technologies have enabled detailed genetic and genomic characterization of large panels of cancer cell lines, including profiling of gene expression, copy number variation (CNV), SNP, mutation, and RNA-seq. The same cell lines have also been profiled for sensitivity to potential anti-cancer compounds, allowing development of computational models used to predict genotype-specific drug treatments linked to tumor molecular subtypes. Inferring such models is challenging because the model inputs contain far more features than observations, known as the p>>n problem, precluding the use of classical statistical models such as least squares (a.k.a. multiple linear regression). Recently, generalized linear models or penalized linear models have been proposed and extended for feature selection and overcoming the p>>n. Bayesian extension of such methods have also been developed, using algorithms such as Metropolis-Hasting and Markov Chain Monte Carlo, and demonstrated to improve prediction accuracy at the expense of increased computing cost. A comprehensive evaluation of predictive modeling techniques is essential to leverage cell line studies to infer the most accurate genetic predictors of drug sensitivity, which can then be used to inform patient selection strategies in clinical trials and to identify functional genetic determinants of drug sensitivity or resistance. In this study, we evaluate state-of-the-art predictive models which utilize dimension reduction methods (e.g., partial least square, support vector machine, principal component regression), feature selection (e.g., LASSO, RIDGE, and elastic-net), and Bayesian feature selection (e.g., Bayesian LASSO and Bayesian RIDGE). We assess each method based on 1) the accuracy of sensitivity predictions in a cross validation setting and in an independent dataset, and 2) the biological coherence of inferred predictive features by comparison to publicly available pathway databases. Citation Format: Adam Arne Margolin, In Sock Jang, Stephen Friend. Predicting drug sensitivity from cancer cell lines. [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA15.

Abstract A33: Integrated analysis of molecular and metabolomic data to identify drug resistance mechanisms in lung cancers

Abstract Consistent bioinformatics analysis approaches of chemically perturbed biological systems are needed in order to achieve meaningful integration of results from multiple experiments in support of cancer systems biology research. Our initial efforts have focused on identifying and contrasting molecular mechanisms that reveal biochemical events affecting non-small cell lung cancers in the context of combination drug therapies as a model for defining these approaches. While the link between growth factor receptor mutations, target inhibitor drugs and non-small cell lung cancer is well established, the basis of variable response in different patient cohorts is still poorly understood and is the subject of ongoing investigations. Data for our efforts were provided by the NCI's Center for Advanced Preclinical Research. Mice genetically engineered to develop erlotinib-sensitive lung adenocarcinomas were subjected to drug compounds that are currently in clinical trials, as well as erlotinib, individually and in combination, to study early response and development of late resistance. Metabolite data, gene expression data, and histological data were collected and used in defining the initial analysis approach. While evaluation of analysis approaches continues, preliminary results already indicate these data can be effective in assessing drug efficacy on reducing tumor burden. Further efforts are underway to develop analysis techniques to explore the underlying dynamic processes. Citation Format: Anand S. Merchant, Natalie Fedorova Abrams, Eric A. Stahlberg, Zoe Weaver. Integrated analysis of molecular and metabolomic data to identify drug resistance mechanisms in lung cancers [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr A33.

Abstract A23: Novel ursolic acid derivatives with potent antitumor activity

Abstract Cancer is the fourth leading cause of death worldwide, within pancreatic cancer is the fourth leading cause of cancer death and without effective medical therapy.1 Natural products provide a promising source for developing effective anticancer agents.2 Ursolic acid is a pentacyclic triterpenoid present in plants, vegetables, and fruits that has been found to have several biological activities, including antitumor activity.3 In order to improve the preexisting antitumor activity of ursolic acid, a panel of semisynthetic derivatives were prepared, characterized, and evaluated for the antiproliferative activities in several cancer cell lines. Ursolic acid was chemically modified by the introduction of heterocyclic rings in several points of the backbone structure and was found that modifications at C2 had improved activities against proliferation of pancreatic cancer cell lines, inducing apoptosis with upregulation of p53 and NOXA levels, and downregulation of XIAP levels. Fluor is a highly desirable atom, in key positions of active molecules can improve metabolic and chemical stability, membrane permeability and binding affinity.4 We have introduced fluor into the ring C of ursolic acid and a series of ursane-type fluor-derivatives were synthesized. The most potent derivative was 19-fold more active than ursolic acid to inhibit Aspc-1 cell growth and to arrest cell cycle at G1 phase with upregulation of p21waf1, and apoptosis at higher concentrations with upregulation of NOXA and downregulation of c-FLIP. These new semisynthetic derivatives of ursolic acid may have potential for drug discovery to treat cancer. Citation Format: Ana S. Leal, Jorge A.R. Salvador, Jing Yongkui. Novel ursolic acid derivatives with potent antitumor activity [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr A23.

Abstract PR11: Dissecting signaling transduction network to infer master regulators of non small cell lung cancer

Abstract The dissection of signal transduction networks is lagging significantly with respect to their transcriptional and protein-complex counterparts. Innovative use of multiplex and microarray-based approaches, where multiple antibodies can be used to probe an ensemble of phosphoproteins, is starting to become sufficiently mature to allow characterization of small pathways. Yet, this is still far from providing an unbiased, genome-wide view of signal-transduction processes. Using data from the first genome-wide profiling of phosphopeptide abundance in normal and tumor-related lung tissue, we have designed a novel algorithm, pARACNe, to generate an unbiased, genome-wide network representing tyrosine kinase signal transduction pathways. Specifically, pARACNe was used to infer tyrosine kinase substrates in non-small cell lung cancer (NSCLC) from tandem mass-spec abundance profiling of phosphotyrosine-enriched peptides from ˜250 samples derived from primary tumors, cell lines, and normal lung tissue. SILAC validation of EGFR, MET and ALK substrates, confirmed >80% of the pARACNe-inferred interactions. Further analysis of large-scale NSCLC microarray expression profile data, using the MINDy algorithm, showed significant convergence between the signal transduction pathways inferred by proteomics and gene expression-based methods. The availability of an accurate NSCLC Tyrosine Kinase Interaction Network (TKIN) has significant implications for our ability to elucidate master regulators of oncogenesis and tumor progression and for the identification and prioritization of druggable targets. We used the MARINa algorithm and the TKIN to infer a complete repertoire of master regulator (MRs) kinases, both individual as well as synergistic, in 46 cell lines. Initial validation shows that the method could recapitulate both cancer-specific MRs, such as EGFR, as well as cell line-specific ones, such as ALK and PTK2 in H2228 cells. Follow-up experimental assays have shown that the PTK2, which is active in tumor-related but not in normal cells, is required for H2228 cell viability and may constitute a novel oncogene addiction in NSCLC and a candidate therapeutic target. We also validated synergistic addiction of EGFR and MET in several cell lines by showing no effect on colony formation after treating cell lines with inhibitors of individual genes but synergistic decrease in colony formation with dual treatment. The TKI suggests that current pathway models are incomplete and sets the stage for quantitative, network-based methods to identify key regulators of physiologic and pathologic cell phenotypes. This proffered talk is also presented as Poster A14. Citation Format: Mukesh Bansal, Michael Peyton, Manjunath Kustagi, Luc Girard, Klarisa Rikova, Michael Comb, Michael White, John Minna, Andrea Califano. Dissecting signaling transduction network to infer master regulators of non small cell lung cancer [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr PR11.

Abstract IA19: Small-molecule modulators of the Hedgehog signaling pathway

Abstract The Hedgehog (Hh) signaling pathway plays critical roles in embryonic patterning and tissue homeostasis, and its inappropriate activation can lead to basal cell carcinoma, medulloblastoma, and other cancers. Small molecules that modulate Hh signaling proteins can therefore be valuable probes of Hh pathway-dependent physiology and targeted therapies for Hh pathway-related diseases. While high-throughput screening campaigns in academia and industry have yielded several Hh pathway modulators that potently and selectively target the transmembrane protein Smoothened (Smo), there is a growing need for compounds that inhibit Hh target gene expression through other mechanisms. Such small molecules could provide new insights into the cellular processes that regulate Gli transcription factor function and possibly prevent or overcome the emergence of tumors that are resistant to Smo antagonists. Toward this goal, we have completed large-scale screens for Hh pathway inhibitors that act downstream of Smo, revealing several compounds that inhibit Gli-dependent transcription through distinct mechanisms of action. We have successfully determined the cellular target of one inhibitor from our screens, and mechanistic studies of other Gli antagonists are currently underway. Citation Format: James K. Chen. Small-molecule modulators of the Hedgehog signaling pathway [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA19.

Abstract IA5: Designing synthetic regulatory RNAs: New tools for temporal and spatial control in biological systems

Abstract Advances in synthetic biology are transforming our ability to design and build synthetic biological systems. While progress has been made in the design of complex genetic circuits, capabilities for constructing large genetic systems currently surpass our ability to design such systems. This growing “design gap” has highlighted the need to develop methods that support the generation of new functional biological components and scalable design strategies for complex genetic circuits that will lay the foundation for integrated biological devices and systems. As examples of functional RNA molecules playing key roles in the behavior of natural biological systems have grown over the past decade, there has been growing interest in the design and implementation of synthetic counterparts. Researchers are taking advantage of the relative ease with which RNA molecules can be modeled and designed to engineer functional RNA molecules that act as diverse components including sensors, regulators, controllers (ligand-responsive RNA regulators), and scaffolds. These synthetic regulatory RNAs are providing new tools for temporal and spatial control in biological systems. I will describe recent work in the design of RNA controllers and advances in addressing challenges faced in their broad implementation as user-programmed control systems in living cells. In particular, I will describe the development of high-throughput cell-based screens for rapidly generating synthetic regulatory RNAs with specified quantitative properties. I will also discuss how the application of synthetic regulatory RNAs as controllers in different biological pathways are leading to the elucidation of integrated systems design strategies and new capabilities for programming genetic systems. Citation Format: Christina D. Smolke. Designing synthetic regulatory RNAs: New tools for temporal and spatial control in biological systems [abstract]. In: Proceedings of the AACR Special Conference on Chemical Systems Biology: Assembling and Interrogating Computational Models of the Cancer Cell by Chemical Perturbations; 2012 Jun 27-30; Boston, MA. Philadelphia (PA): AACR; Cancer Res 2012;72(13 Suppl):Abstract nr IA5.