Predicting tumor cell line response to drug pairs with deep learning.

BACKGROUND: The National Cancer Institute drug pair screening effort against 60 well-characterized human tumor cell lines (NCI-60) presents an unprecedented resource for modeling combinational drug activity. RESULTS: We present a computational model for predicting cell line response to a subset of drug pairs in the NCI-ALMANAC database. Based on residual neural networks for encoding features as well as predicting tumor growth, our model explains 94% of the response variance. While our best result is achieved with a combination of molecular feature types (gene expression, microRNA and proteome), we show that most of the predictive power comes from drug descriptors. To further demonstrate value in detecting anticancer therapy, we rank the drug pairs for each cell line based on model predicted combination effect and recover 80% of the top pairs with enhanced activity. CONCLUSIONS: We present promising results in applying deep learning to predicting combinational drug response. Our feature analysis indicates screening data involving more cell lines are needed for the models to make better use of molecular features.

The National Cancer Institute ALMANAC: A Comprehensive Screening Resource for the Detection of Anticancer Drug Pairs with Enhanced Therapeutic Activity.

To date, over 100 small-molecule oncology drugs have been approved by the FDA. Because of the inherent heterogeneity of tumors, these small molecules are often administered in combination to prevent emergence of resistant cell subpopulations. Therefore, new combination strategies to overcome drug resistance in patients with advanced cancer are needed. In this study, we performed a systematic evaluation of the therapeutic activity of over 5,000 pairs of FDA-approved cancer drugs against a panel of 60 well-characterized human tumor cell lines (NCI-60) to uncover combinations with greater than additive growth-inhibitory activity. Screening results were compiled into a database, termed the NCI-ALMANAC (A Large Matrix of Anti-Neoplastic Agent Combinations), publicly available at https://dtp.cancer.gov/ncialmanac Subsequent in vivo experiments in mouse xenograft models of human cancer confirmed combinations with greater than single-agent efficacy. Concomitant detection of mechanistic biomarkers for these combinations in vivo supported the initiation of two phase I clinical trials at the NCI to evaluate clofarabine with bortezomib and nilotinib with paclitaxel in patients with advanced cancer. Consequently, the hypothesis-generating NCI-ALMANAC web-based resource has demonstrated value in identifying promising combinations of approved drugs with potent anticancer activity for further mechanistic study and translation to clinical trials. Cancer Res; 77(13); 3564-76. ©2017 AACR.

Phosphorylated fraction of H2AX as a measurement for DNA damage in cancer cells and potential applications of a novel assay.

Phosphorylated H2AX (γ-H2AX) is a sensitive marker for DNA double-strand breaks (DSBs), but the variability of H2AX expression in different cell and tissue types makes it difficult to interpret the meaning of the γ-H2AX level. Furthermore, the assays commonly used for γ-H2AX detection utilize laborious and low-throughput microscopy-based methods. We describe here an ELISA assay that measures both phosphorylated H2AX and total H2AX absolute amounts to determine the percentage of γ-H2AX, providing a normalized value representative of the amount of DNA damage. We demonstrate the utility of the assay to measure DSBs introduced by either ionizing radiation or DNA-damaging agents in cultured cells and in xenograft models. Furthermore, utilizing the NCI-60 cancer cell line panel, we show a correlation between the basal fraction of γ-H2AX and cellular mutation levels. This additional application highlights the ability of the assay to measure γ-H2AX levels in many extracts at once, making it possible to correlate findings with other cellular characteristics. Overall, the γ-H2AX ELISA represents a novel approach to quantifying DNA damage, which may lead to a better understanding of mutagenic pathways in cancer and provide a useful biomarker for monitoring the effectiveness of DNA-damaging anticancer agents.

Gene expression profiling of 49 human tumor xenografts from in vitro culture through multiple in vivo passages--strategies for data mining in support of therapeutic studies.

BACKGROUND: Development of cancer therapeutics partially depends upon selection of appropriate animal models. Therefore, improvements to model selection are beneficial. RESULTS: Forty-nine human tumor xenografts at in vivo passages 1, 4 and 10 were subjected to cDNA microarray analysis yielding a dataset of 823 Affymetrix HG-U133 Plus 2.0 arrays. To illustrate mining strategies supporting therapeutic studies, transcript expression was determined: 1) relative to other models, 2) with successive in vivo passage, and 3) during the in vitro to in vivo transition. Ranking models according to relative transcript expression in vivo has the potential to improve initial model selection. For example, combining p53 tumor expression data with mutational status could guide selection of tumors for therapeutic studies of agents where p53 status purportedly affects efficacy (e.g., MK-1775). The utility of monitoring changes in gene expression with extended in vivo tumor passages was illustrated by focused studies of drug resistance mediators and receptor tyrosine kinases. Noteworthy observations included a significant decline in HCT-15 colon xenograft ABCB1 transporter expression and increased expression of the kinase KIT in A549 with serial passage. These trends predict sensitivity to agents such as paclitaxel (ABCB1 substrate) and imatinib (c-KIT inhibitor) would be altered with extended passage. Given that gene expression results indicated some models undergo profound changes with in vivo passage, a general metric of stability was generated so models could be ranked accordingly. Lastly, changes occurring during transition from in vitro to in vivo growth may have important consequences for therapeutic studies since targets identified in vitro could be over- or under-represented when tumor cells adapt to in vivo growth. A comprehensive list of mouse transcripts capable of cross-hybridizing with human probe sets on the HG-U133 Plus 2.0 array was generated. Removal of the murine artifacts followed by pairwise analysis of in vitro cells with respective passage 1 xenografts and GO analysis illustrates the complex interplay that each model has with the host microenvironment. CONCLUSIONS: This study provides strategies to aid selection of xenograft models for therapeutic studies. These data highlight the dynamic nature of xenograft models and emphasize the importance of maintaining passage consistency throughout experiments.

The exomes of the NCI-60 panel: a genomic resource for cancer biology and systems pharmacology.

The NCI-60 cell lines are the most frequently studied human tumor cell lines in cancer research. This panel has generated the most extensive cancer pharmacology database worldwide. In addition, these cell lines have been intensely investigated, providing a unique platform for hypothesis-driven research focused on enhancing our understanding of tumor biology. Here, we report a comprehensive analysis of coding variants in the NCI-60 panel of cell lines identified by whole exome sequencing, providing a list of possible cancer specific variants for the community. Furthermore, we identify pharmacogenomic correlations between specific variants in genes such as TP53, BRAF, ERBBs, and ATAD5 and anticancer agents such as nutlin, vemurafenib, erlotinib, and bleomycin showing one of many ways the data could be used to validate and generate novel hypotheses for further investigation. As new cancer genes are identified through large-scale sequencing studies, the data presented here for the NCI-60 will be an invaluable resource for identifying cell lines with mutations in such genes for hypothesis-driven research. To enhance the utility of the data for the greater research community, the genomic variants are freely available in different formats and from multiple sources including the CellMiner and Ingenuity websites.

Downregulation of Pdcd4 by mir-21 facilitates glioblastoma proliferation in vivo.

MicroRNAs (miRNAs) are small, noncoding RNAs that play a critical role in developmental and physiological processes and are implicated in the pathogenesis of several human diseases, including cancer. They function by regulating target gene expression post-transcriptionally. In this study, we examined the role of oncogenic mir-21 in the pathogenesis of glioblastoma, the most aggressive form of primary brain tumor. We have previously reported that mir-21 is expressed at higher levels in primary glioblastoma-tissue and glioblastoma-derived cell lines than in normal brain tissue. We demonstrate that downregulation of mir-21 in glioblastoma-derived cell lines results in increased expression of its target, programmed cell death 4 (Pdcd4), a known tumor-suppressor gene. In addition, our data indicate that either downregulation of mir-21 or overexpression of its target, Pdcd4, in glioblastoma-derived cell lines leads to decreased proliferation, increased apoptosis, and decreased colony formation in soft agar. Using a glioblastoma xenograft model in immune-deficient nude mice, we observe that glioblastoma-derived cell lines in which mir-21 levels are downregulated or Pdcd4 is over-expressed exhibit decreased tumor formation and growth. Significantly, tumors grow when the glioblastoma-derived cell lines are transfected with anti-mir-21 and siRNA to Pdcd4, confirming that the tumor growth is specifically regulated by Pdcd4. These critical in vivo findings demonstrate an important functional linkage between mir-21 and Pdcd4 and further elucidate the molecular mechanisms by which the known high level of mir-21 expression in glioblastoma can attribute to tumorigenesis--namely, inhibition of Pdcd4 and its tumor-suppressive functions.

Analysis of Food and Drug Administration-approved anticancer agents in the NCI60 panel of human tumor cell lines.

Since the early 1990s the Developmental Therapeutics Program of the National Cancer Institute (NCI) has utilized a panel of 60 human tumor cell lines (NCI60) representing 9 tissue types to screen for potential new anticancer agents. To date, about 100,000 compounds and 50,000 natural product extracts have been screened. Early in this program it was discovered that the pattern of growth inhibition in these cell lines was similar for compounds of similar mechanism. The development of the COMPARE algorithm provided a means by which investigators, starting with a compound of interest, could identify other compounds whose pattern of growth inhibition was similar. With extensive molecular characterization of these cell lines, COMPARE and other user-defined algorithms have been used to link patterns of molecular expression and drug sensitivity. We describe here the results of screening current Food and Drug Administration (FDA)-approved anticancer agents in the NCI60 screen, with an emphasis on those agents that target signal transduction. We analyzed results from agents with mechanisms of action presumed to be similar; we also carried out a hierarchical clustering of all of these agents. The addition of data from recently approved anticancer agents will increase the utility of the NCI60 databases to the cancer research community. These data are freely accessible to the public on the DTP website (http://dtp.cancer.gov/). The FDA-approved anticancer agents are themselves available from the NCI as a plated set of compounds for research use.

Bisindenoisoquinoline bis-1,3-{(5,6-dihydro-5,11-diketo-11H-indeno[1,2-c]isoquinoline)-6-propylamino}propane bis(trifluoroacetate) (NSC 727357), a DNA intercalator and topoisomerase inhibitor with antitumor activity.

Indenoisoquinolines are topoisomerase (Top) I inhibitors developed to overcome some of the limitations of camptothecins and expand their anticancer spectrum. Bis-1,3-{(5,6-dihydro-5,11-diketo-11H-indeno[1,2-c]isoquinoline)-6-propylamino}-propane bis(trifluoroacetate) (NSC 727357) is a novel dimeric indenoisoquinoline derivative with potent antiproliferative activity in the NCI-60 cell line panel, promising hollow fiber activity (score of 32) and activity against xenografts. Submicromolar concentrations of the bisindenoisoquinoline NSC 727357 induce Top1 cleavage complexes at specific sites in biochemical assays. At higher concentrations, inhibition of Top1 catalytic activity and DNA intercalation is observed. NSC 727357 also induces a limited number of Top2-DNA cleavage complexes. In contrast to the effect of other Top1 inhibitors, cells treated with the bisindenoisoquinoline NSC 727357 show an arrest of cell cycle progression in G(1) with no significant inhibition of DNA synthesis after a short exposure to the drug. Moreover, unlike camptothecin and the indenoisoquinoline MJ-III-65 (NSC 706744, 6-[3-(2-hydroxyethyl)aminopropyl]-5,6-dihydro-5,11-diketo-2,3-dimethoxy-(methylenedioxy)-11H-indeno[1,2-c]isoquinoline hydrochloride), the cytotoxicity of bisindenoisoquinoline NSC 727357 is only partially dependent on Top1 and p53, indicating that this drug has additional targets besides Top1 and Top2.

Ligand-directed surface profiling of human cancer cells with combinatorial peptide libraries.

A collection of 60 cell lines derived from human tumors (NCI-60) has been widely explored as a tool for anticancer drug discovery. Here, we profiled the cell surface of the NCI-60 by high-throughput screening of a phage-displayed random peptide library and classified the cell lines according to the binding selectivity of 26,031 recovered tripeptide motifs. By analyzing selected cell-homing peptide motifs and their NCI-60 recognition patterns, we established that some of these motifs (a) are similar to domains of human proteins known as ligands for tumor cell receptors and (b) segregate among the NCI-60 in a pattern correlating with expression profiles of the corresponding receptors. We biochemically validated some of the motifs as mimic peptides of native ligands for the epidermal growth factor receptor. Our results indicate that ligand-directed profiling of tumor cell lines can select functional peptides from combinatorial libraries based on the expression of tumor cell surface molecules, which in turn could be exploited as "druggable" receptors in specific types of cancer.

Correlation of PIK3Ca mutations with gene expression and drug sensitivity in NCI-60 cell lines.

The gene that encodes the alpha-isoform of phosphatidylinositol 3-kinase (PIK3Ca) is frequently mutated in human cancers. We profiled the mutation status of the PIK3Ca gene in the National Cancer Institute (NCI)-60 panel of human cancer cell lines maintained by the Developmental Therapeutics Program of the NCI. Mutation hotspots on the gene were PCR amplified and sequenced, and the trace data were analyzed with software designed to detect mutations. Seven of the cell lines tested have PIK3Ca mutations: two lines derived from breast cancer, two from colon cancer, two from ovarian cancer, and one from lung cancer. BRAF and EGFR genes were normal in the PIK3Ca mutant lines. Two of the cell lines with mutant PIK3Ca also have a mutant version of the KRAS gene. The mutation status was correlated with array-based gene expression that is publicly available for the NCI-60 cell lines. We found increased expression levels for estrogen receptor (ER) and ERBB2 in PIK3Ca mutant lines. The PIK3Ca mutation status was also correlated with compound screening data for the cell lines. PIK3Ca-mutant cell lines were relatively more sensitive than PIK3Ca-normal cell lines to the ER inhibitor tamoxifen and the AKT inhibitor triciribine, among other compounds. The results provide insights into the role of mutant PIK3Ca in oncogenic signaling and allow preliminary identification of novel targets for therapeutic intervention in cancers harboring PIK3Ca mutations.

HLA class I and II genotype of the NCI-60 cell lines.

Sixty cancer cell lines have been extensively characterized and used by the National Cancer Institute's Developmental Therapeutics Program (NCI-60) since the early 90's as screening tools for anti-cancer drug development. An extensive database has been accumulated that could be used to select individual cells lines for specific experimental designs based on their global genetic and biological profile. However, information on the human leukocyte antigen (HLA) genotype of these cell lines is scant and mostly antiquated since it was derived from serological typing. We, therefore, re-typed the NCI-60 panel of cell lines by high-resolution sequence-based typing. This information may be used to: 1) identify and verify the identity of the same cell lines at various institutions; 2) check for possible contaminant cell lines in culture; 3) adopt individual cell lines for experiments in which knowledge of HLA molecule expression is relevant. Since genome-based typing does not guarantee actual surface protein expression, further characterization of relevant cell lines should be entertained to verify surface expression in experiments requiring correct antigen presentation.

Transcription profiling of gene expression in drug discovery and development: the NCI experience.

Transcript profiling, using microarray or other analogous technologies, to query on a large-scale the expression of genes in tumours or their derivative cell lines has numerous potential uses in oncology drug discovery and development. Characterisation of genes expressed in tumours may allow tumours to be separated into subsets defining subtypes that have a distinctive pathway utilisation. The molecular entities comprising the pathways which distinguish one disease subset from another then become potential candidate drug targets. Alternatively, gene expression patterns may be correlated with the degree of antiproliferative effect of candidate drug leads. This can reveal aspects of the drug's action that could serve to provide a further basis for benchmarking the generation of analogues or provide important information about pathways potentially modulated by the drug in achieving cytotoxicity. New information is emerging that the expression of drug transport-related molecules is a major variable that can be usefully explored using gene expression data, and the features promoting successful drug handling by the tumour cell may be an additional variable which can be illuminated by gene expression studies.

In vitro evaluation of dimethane sulfonate analogues with potential alkylating activity and selective renal cell carcinoma cytotoxicity.

We identified five structurally related dimethane sulfonates with putative selective cytotoxicity in renal cancer cell lines. These compounds have a hydrophobic moiety linked to a predicted alkylating group. A COMPARE analysis with the National Cancer Institute Anticancer Drug Screen standard agent database found significant correlations between the IC50 of the test compounds and the IC50 of alkylating agents (e.g., r = 0.68, P < 0.00001 for chlorambucil). In this report, we examined whether these compounds had activities similar to those of conventional alkylating agents. In cytotoxicity studies, chlorambucil-resistant Walker rat carcinoma cells were 4- to 11-fold cross-resistant to the test compounds compared with 14-fold resistant to chlorambucil. To determine effects on cell cycle progression, renal cell carcinoma (RCC) line 109 was labeled with bromodeoxyuridine prior to drug treatment. Complete cell cycle arrest occurred in cells treated with an IC90 dose of NSC 268965. p53 protein levels increased as much as 5.7-fold in RCC line 109 and as much as 20.4-fold in breast cancer line MCF-7 following an 18-hour drug exposure. Finally, DNA-protein cross-links were found following a 6-hour pretreatment with all compounds. Thus, the dimethane sulfonate analogues have properties expected of some alkylating agents but, unlike conventional alkylating agents, appear to possess activity against RCC.

Two CD95 tumor classes with different sensitivities to antitumor drugs.

CD95 type I and II cells differ in their dependence on mitochondria to execute apoptosis, because antiapoptotic members of the Bcl-2 family render only type II cells resistant to death receptor-induced apoptosis. They can also be distinguished by a more efficient formation of the death-inducing signaling complex in type I cells. We have identified a soluble form of CD95 ligand (S2) that is cytotoxic to type II cells but does not kill type I cells. By testing 58 tumor cell lines of the National Cancer Institute's anticancer drug-screening panel for apoptosis sensitivity to S2 and performing death-inducing signaling complex analyses, we determined that half of the CD95-sensitive cells are type I and half are type II. Most of the type I cell lines fall into a distinct class of tumor cells expressing mesenchymal-like genes, whereas the type II cell lines preferentially express epithelium-like markers. This suggests that type I and II tumor cells represent different stages of carcinogenesis that resemble the epithelial-mesenchymal transition. We then screened the National Cancer Institute database of >42,000 compounds for reagents with patterns of growth inhibition that correlated with either type I or type II cell lines and found that actin-binding compounds selectively inhibited growth of type I cells, whereas tubulin-interacting compounds inhibited growth of type II cells. Our analysis reveals fundamental differences in programs of gene expression between type I and type II cells and could impact the way actin- and microtubule-disrupting antitumor agents are used in tumor therapy.

COMPARE: a web accessible tool for investigating mechanisms of cell growth inhibition.

For more than 10 years the National Cancer Institute (NCI) has tested compounds for their ability to inhibit the growth of human tumor cell lines in culture (NCI screen). Work of Ken Paull [J. Natl. Cancer Inst. 81 (1989) 1088] demonstrated that compounds with similar mechanism of cell growth inhibition show similar patterns of activity in the NCI screen. This observation was developed into an algorithm called COMPARE and has been successfully used to predict mechanisms for a wide variety of compounds. More recently, this method has been extended to associate patterns of cell growth inhibition by compounds with measurements of molecular entities (such as gene expression) in the cell lines in the NCI screen. The COMPARE method and associated data are freely available on the Developmental Therapeutics Program (DTP) web site (http://dtp.nci.nih.gov/). Examples of the use of COMPARE on these web pages will be explained and demonstrated. Published by Elsevier Science Inc.