Kinome-Wide Virtual Screening by Multi-Task Deep Learning

Deep learning is a machine learning technique to model high-level abstractions in data by utilizing a graph composed of multiple processing layers that experience various linear and non-linear transformations. This technique has been shown to perform well for applications in drug discovery, utilizing structural features of small molecules to predict activity. Here, we report a large-scale study to predict the activity of small molecules across the human kinome-a major family of drug targets, particularly in anti-cancer agents. While small-molecule kinase inhibitors exhibit impressive clinical efficacy in several different diseases, resistance often arises through adaptive kinome reprogramming or subpopulation diversity. Polypharmacology and combination therapies offer potential therapeutic strategies for patients with resistant diseases. Their development would benefit from a more comprehensive and dense knowledge of small-molecule inhibition across the human kinome. Leveraging over 650,000 bioactivity annotations for more than 300,000 small molecules, we evaluated multiple machine learning methods to predict the small-molecule inhibition of 342 kinases across the human kinome. Our results demonstrated that multi-task deep neural networks outperformed classical single-task methods, offering the potential for conducting large-scale virtual screening, predicting activity profiles, and bridging the gaps in the available data.

Pan-cancer proteogenomics connects oncogenic drivers to functional states

Cancer driver events refer to key genetic aberrations that drive oncogenesis; however, their exact molecular mechanisms remain insufficiently understood. Here, our multi-omics pan-cancer analysis uncovers insights into the impacts of cancer drivers by identifying their significant cis-effects and distal trans-effects quantified at the RNA, protein, and phosphoprotein levels. Salient observations include the association of point mutations and copy-number alterations with the rewiring of protein interaction networks, and notably, most cancer genes converge toward similar molecular states denoted by sequence-based kinase activity profiles. A correlation between predicted neoantigen burden and measured T cell infiltration suggests potential vulnerabilities for immunotherapies. Patterns of cancer hallmarks vary by polygenic protein abundance ranging from uniform to heterogeneous. Overall, our work demonstrates the value of comprehensive proteogenomics in understanding the functional states of oncogenic drivers and their links to cancer development, surpassing the limitations of studying individual cancer types.

Proteogenomic analysis of chemo-refractory high-grade serous ovarian cancer

To improve the understanding of chemo-refractory high-grade serous ovarian cancers (HGSOCs), we characterized the proteogenomic landscape of 242 (refractory and sensitive) HGSOCs, representing one discovery and two validation cohorts across two biospecimen types (formalin-fixed paraffin-embedded and frozen). We identified a 64-protein signature that predicts with high specificity a subset of HGSOCs refractory to initial platinum-based therapy and is validated in two independent patient cohorts. We detected significant association between lack of Ch17 loss of heterozygosity (LOH) and chemo-refractoriness. Based on pathway protein expression, we identified 5 clusters of HGSOC, which validated across two independent patient cohorts and patient-derived xenograft (PDX) models. These clusters may represent different mechanisms of refractoriness and implicate putative therapeutic vulnerabilities.

Correction: The Genetic Landscape of Ocular Adnexa MALT Lymphoma Reveals Frequent Aberrations in NFAT and MEF2B Signaling Pathways

[This corrects the article DOI: 10.1158/2767-9764.CRC-21-0022.][This corrects the article DOI: 10.1158/2767-9764.CRC-21-0022.].

Mass spectrometry database of lacrimal gland adenoid cystic carcinoma and normal lacrimal gland tissue identifies extracellular matrix remodeling in these tumors

Adenoid cystic carcinoma of the lacrimal gland (LGACC) is a slow-growing but aggressive orbital malignancy. Due to the rarity of LGACC, it is poorly understood, which makes diagnosing, treating, and monitoring disease progression difficult. The aim is to understand the molecular drivers of LGACC further to identify potential targets for treating this cancer. Mass spectrometry was performed on LGACC and normal lacrimal gland samples to examine the differentially expressed proteins to understand this cancer's proteomic characteristics. Downstream gene ontology and pathway analysis revealed the extracellular matrix is the most upregulated process in LGACC. This data serves as a resource for further understanding LGACC and identifying potential treatment targets. This dataset is publicly available.

Spatial Transcriptomics Identifies Expression Signatures Specific to Lacrimal Gland Adenoid Cystic Carcinoma Cells

Although primary tumors of the lacrimal gland are rare, adenoid cystic carcinoma (ACC) is the most common and lethal epithelial lacrimal gland malignancy. Traditional management of lacrimal gland adenoid cystic carcinoma (LGACC) involves the removal of the eye and surrounding socket contents, followed by chemoradiation. Even with this radical treatment, the 10-year survival rate for LGACC is 20% given the propensity for recurrence and metastasis. Due to the rarity of LGACC, its pathobiology is not well-understood, leading to difficulties in diagnosis, treatment, and effective management. Here, we integrate bulk RNA sequencing (RNA-seq) and spatial transcriptomics to identify a specific LGACC gene signature that can inform novel targeted therapies. Of the 3499 differentially expressed genes identified by bulk RNA-seq, the results of our spatial transcriptomic analysis reveal 15 upregulated and 12 downregulated genes that specifically arise from LGACC cells, whereas fibroblasts, reactive fibrotic tissue, and nervous and skeletal muscle account for the remaining bulk RNA-seq signature. In light of the analysis, we identified a transitional state cell or stem cell cluster. The results of the pathway analysis identified the upregulation of PI3K-Akt signaling, IL-17 signaling, and multiple other cancer pathways. This study provides insights into the molecular and cellular landscape of LGACC, which can inform new, targeted therapies to improve patient outcomes.

EPCO-14. ISOSCELES: AN INTEGRATIVE FRAMEWORK FOR THE PREDICTION OF TREATMENT RESISTANT GLIOBLASTOMA CELLS

Glioblastoma (GBM) remains the most common and lethal adult primary brain cancer. Two of the most significant issues preventing the development of effective GBM treatments are inter- and intra-tumor heterogeneity. To address these issues, we developed a novel platform termed ISOSCELES (Inferred cell Sensitivity Operating on the integration of Single-Cell Expression and L1000 Expression Signatures). ISOSCELES integrates single-cell gene expression data in individual GBM tumors with perturbation-response data derived from the NIH Library of Integrated Network-Based Cellular Signatures (LINCS) L1000 dataset to predict sensitive and resistant tumor cell populations. Importantly, we analyzed the predictive power of ISOSCELES in an in vivo xenograft model and demonstrated that ISOSCELES reveals the GBM cell identities primed for lineage expansion during treatment with the aurora kinase inhibitor alisertib. These studies suggest that ISOSCELES can be used to identify sensitive and resistant cell populations to targeted therapies in GBM, which can inform treatment decisions in ongoing and future clinical trials.

Connecting omics signatures and revealing biological mechanisms with iLINCS

There are only a few platforms that integrate multiple omics data types, bioinformatics tools, and interfaces for integrative analyses and visualization that do not require programming skills. Here we present iLINCS ( http://ilincs.org ), an integrative web-based platform for analysis of omics data and signatures of cellular perturbations. The platform facilitates mining and re-analysis of the large collection of omics datasets (>34,000), pre-computed signatures (>200,000), and their connections, as well as the analysis of user-submitted omics signatures of diseases and cellular perturbations. iLINCS analysis workflows integrate vast omics data resources and a range of analytics and interactive visualization tools into a comprehensive platform for analysis of omics signatures. iLINCS user-friendly interfaces enable execution of sophisticated analyses of omics signatures, mechanism of action analysis, and signature-driven drug repositioning. We illustrate the utility of iLINCS with three use cases involving analysis of cancer proteogenomic signatures, COVID 19 transcriptomic signatures and mTOR signaling.

A Multiparametric Pharmacogenomic Strategy for Drug Repositioning predicts Therapeutic Efficacy for Glioblastoma Cell Lines

Background

Poor prognosis of glioblastoma patients and the extensive heterogeneity of glioblastoma at both the molecular and cellular level necessitates developing novel individualized treatment modalities via genomics-driven approaches.

Methods

This study leverages numerous pharmacogenomic and tissue databases to examine drug repositioning for glioblastoma. RNAseq of glioblastoma tumor samples from The Cancer Genome Atlas (TCGA, n=117) were compared to “normal” frontal lobe samples from Genotype-Tissue Expression Portal (GTEX, n=120) to find differentially expressed genes (DEGs). Using compound-gene expression data and drug activity data from the Library of Integrated Network-Based Cellular Signatures (LINCS, n=66,512 compounds) CCLE (71 glioma cell lines), and Chemical European Molecular Biology Laboratory (ChEMBL) platforms, we employed a summarized reversal gene expression metric (sRGES) to “reverse” the resultant disease signature for GBM and its subtypes. A multi-parametric strategy was employed to stratify compounds capable of blood brain barrier penetrance with a favorable pharmacokinetic profile (CNS-MPO).

Results

Significant correlations were identified between sRGES and drug efficacy in GBM cell lines in both ChEMBL(r=0.37,p<.001) and Cancer Therapeutic Response Portal (CTRP) databases (r=0.35, p<0.001). Our multiparametric algorithm identified two classes of drugs with highest sRGES and CNS-MPO: HDAC inhibitors (vorinostat and entinostat) and topoisomerase inhibitors suitable for drug repurposing.

Conclusions

Our studies suggest that reversal of glioblastoma disease signature correlates with drug potency for various GBM subtypes. This multiparametric approach may set the foundation for an early-phase personalized -omics clinical trial for glioblastoma by effectively identifying drugs that are capable of reversing the disease signature and have favorable pharmacokinetic and safety profiles.

EPCO-18. A MESENCHYMAL CELL POPULATION MEDIATES RESISTANCE TO AURORA KINASE INHIBITORS IN GLIOBLASTOMA

Glioblastoma (GBM) remains the most common adult brain cancer, with a dismal average patient survival of less than two years. No new treatments have been approved for GBM since the introduction of the alkylating agent temozolomide in 2005. Even then, temozolomide treatment only increases the average survival of GBM patients by a few months. Thus, novel therapeutic options are direly needed. The aurora kinases A and B are targetable and overexpressed in GBM, and their expression is highly correlated with patient survival outcomes. Our lab has found that small molecule aurora kinase inhibition reduces GBM tumor growth in vitro and in vivo, however, eventually tumors still grow. Computational analysis integrating compound transcriptional response signatures from the LINCS L1000 dataset with the single-cell RNA-sequencing data of patient GBM tumors resected at the University of Miami predicts that aurora inhibition targets a subset of cells present within any GBM tumor. Results of in vivo single-cell perturbation experiments with the aurora kinase inhibitor alisertib coincide with our predictions and reveal a cellular transcriptional phenotype resistant to aurora kinase inhibition, characterized by a mesenchymal expression program. We find that small molecules that are predicted to target different cell populations from alisertib, including this resistant mesenchymal population, synergize with alisertib to kill GBM cells. As a whole, we have identified the cellular population resistant to aurora kinase inhibition and have developed an analytical framework that identifies synergistic small molecule combinations by identifying compounds that target transcriptionally distinct cellular populations within GBM tumors.

The Genetic Landscape of Ocular Adnexa MALT Lymphoma Reveals Frequent Aberrations in NFAT and MEF2B Signaling Pathways

A comprehensive constellation of somatic non-silent mutations and copy number (CN) variations in ocular adnexa marginal zone lymphoma (OAMZL) is unknown. By utilizing whole-exome sequencing in 69 tumors we define the genetic landscape of OAMZL. Mutations and CN changes in CABIN1 (30%), RHOA (26%), TBL1XR1 (22%), and CREBBP (17%) and inactivation of TNFAIP3 (26%) were among the most common aberrations. Candidate cancer driver genes cluster in the B-cell receptor (BCR), NFkB, NOTCH and NFAT signaling pathways. One of the most commonly altered genes is CABIN1, a calcineurin inhibitor acting as a negative regulator of the NFAT and MEF2B transcriptional activity. CABIN1 deletions enhance BCR-stimulated NFAT and MEF2B transcriptional activity, while CABIN1 mutations enhance only MEF2B transcriptional activity by impairing binding of mSin3a to CABIN1. Our data provide an unbiased identification of genetically altered genes that may play a role in the molecular pathogenesis of OAMZL and serve as therapeutic targets.

A proteogenomic portrait of lung squamous cell carcinoma

Lung squamous cell carcinoma (LSCC) remains a leading cause of cancer death with few therapeutic options. We characterized the proteogenomic landscape of LSCC, providing a deeper exposition of LSCC biology with potential therapeutic implications. We identify NSD3 as an alternative driver in FGFR1-amplified tumors and low-p63 tumors overexpressing the therapeutic target survivin. SOX2 is considered undruggable, but our analyses provide rationale for exploring chromatin modifiers such as LSD1 and EZH2 to target SOX2-overexpressing tumors. Our data support complex regulation of metabolic pathways by crosstalk between post-translational modifications including ubiquitylation. Numerous immune-related proteogenomic observations suggest directions for further investigation. Proteogenomic dissection of CDKN2A mutations argue for more nuanced assessment of RB1 protein expression and phosphorylation before declaring CDK4/6 inhibition unsuccessful. Finally, triangulation between LSCC, LUAD, and HNSCC identified both unique and common therapeutic vulnerabilities. These observations and proteogenomics data resources may guide research into the biology and treatment of LSCC.

Abstract 2147: Discovering novel targets for adenoid cystic carcinoma of the lacrimal gland

Lacrimal gland adenoid cystic carcinoma is a rare but highly lethal indolent cancer originating in the cells of secretory glands, primarily of the head and neck. With the 10-year survival rate of 20%, the only life extending technique is to remove the eye and surrounding socket contents entirely. Despite radical treatment, ACC is a very slow growing tumor and most patients will present with metastatic disease to the brain, lung, liver, or bones several years out from initial diagnosis and before symptoms are noticeable. Due to the rarity of LGACC, it is not well understood, but extensive genome analysis of ACC of the head and neck has previously uncovered a high mutation rate of MYB and NOTCH in patient samples, though unfortunately neither can be directly targeted by drugs. We have RNA sequencing data for 7 primary LGACC tumors, mass spectrometry data for 12 primary tumors and RNA sequencing for 15 cell lines derived from tumors to discover possible targetable biomarkers in LGACC and to further understand the biology of these tumors. Our analysis has revealed the signatures and drivers of LGACC, most significantly involving extracellular matrix mechanisms including HAPLN1, FABP7, and VCAN. We compared the genes of proteins differentially expressed in the mass spectrometry and RNA sequencing data and discovered 194 genes that were differentially expressed in both analysis types. With this extensive understanding of the molecular signatures, we used iLINCS integrated web-based platform to determine compounds that would target the differentially expressed genes. We have selected compounds based on a rubric we generated and will test them individually and in combination on LGACC cell lines to determine the best specific treatment for primary tumors and metastatic disease.

Citation Format: Acadia H. Moeyersoms, Vasileios Stathias, Michelle Zhang, Ryan A. Gallo, David T. Tse, Daniel Pelaez. Discovering novel targets for adenoid cystic carcinoma of the lacrimal gland [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2147.

Abstract 2170: Proteogenomic and metabolomic characterization of human glioblastoma

Glioblastoma (GBM) is the most aggressive nervous system cancer, with median survival under 2 years. Understanding its molecular pathogenesis is crucial for improving diagnosis and treatment. We performed an integrated analysis of genomic, proteomic, post-translational modification and metabolomic data on 99 treatment-naive GBMs. We identified key phosphorylation events (e.g., phosphorylated PTPN11 and PLCG1) as potential switches mediating oncogenic pathway activation as well as potential targets for EGFR-, TP53- and RB1-altered tumors. We detected immune subtypes, driven by the presence of distinct immune cell populations using bulk omics, validated by snRNA-seq, and they were correlated with specific expression and histone acetylation patterns. Acetylation of histone H2B in classical-like and immune-low GBM was driven largely by BRDs, CREBBP, and EP300. Integrated metabolomic and proteomic data identified specific lipid distributions across subtypes and distinct global metabolic changes in IDH mutated tumors. By comparing the adult GBM proteomics to the adolescent and young adult (AYA) GBM cohort from the HOPE study, we found downregulated IDH1 expression and up-regulated expression of genes in the NADH dehydrogenase family, including NDUFB1, and NDUFB3 among others, which may be related to high IDH1 mutation frequency in AYA. This work highlights biological relationships which could potentially aid GBM patient stratifications for more effective treatments.

Citation Format: Liang-Bo Wang, Alla Karpova, Marina A. Gritsenko, Jennifer E. Kyle, Song Cao, Yize Li, Dmitry Rykunov, Antonio Colaprico, Joseph Rothstein, Runyu Hong, Vasileios Stathias, MacIntosh Cornwell, Francesca Petralia, Richard D. Smith, Antonio Iavarone, Milan G. Chheda, Jill S. Barnholtz-Sloan, Karin D. Rodland, Tao Liu, Li Ding, Clinical Proteomic Tumor Analysis Consortium. Proteogenomic and metabolomic characterization of human glioblastoma [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2170.

Proteogenomic and metabolomic characterization of human glioblastoma

Glioblastoma (GBM) is the most aggressive nervous system cancer. Understanding its molecular pathogenesis is crucial to improving diagnosis and treatment. Integrated analysis of genomic, proteomic, post-translational modification and metabolomic data on 99 treatment-naive GBMs provides insights to GBM biology. We identify key phosphorylation events (e.g., phosphorylated PTPN11 and PLCG1) as potential switches mediating oncogenic pathway activation, as well as potential targets for EGFR-, TP53-, and RB1-altered tumors. Immune subtypes with distinct immune cell types are discovered using bulk omics methodologies, validated by snRNA-seq, and correlated with specific expression and histone acetylation patterns. Histone H2B acetylation in classical-like and immune-low GBM is driven largely by BRDs, CREBBP, and EP300. Integrated metabolomic and proteomic data identify specific lipid distributions across subtypes and distinct global metabolic changes in IDH-mutated tumors. This work highlights biological relationships that could contribute to stratification of GBM patients for more effective treatment.

Machine and deep learning approaches for cancer drug repurposing

Knowledge of the underpinnings of cancer initiation, progression and metastasis has increased exponentially in recent years. Advanced "omics" coupled with machine learning and artificial intelligence (deep learning) methods have helped elucidate targets and pathways critical to those processes that may be amenable to pharmacologic modulation. However, the current anti-cancer therapeutic armamentarium continues to lag behind. As the cost of developing a new drug remains prohibitively expensive, repurposing of existing approved and investigational drugs is sought after given known safety profiles and reduction in the cost barrier. Notably, successes in oncologic drug repurposing have been infrequent. Computational in-silico strategies have been developed to aid in modeling biological processes to find new disease-relevant targets and discovering novel drug-target and drug-phenotype associations. Machine and deep learning methods have especially enabled leaps in those successes. This review will discuss these methods as they pertain to cancer biology as well as immunomodulation for drug repurposing opportunities in oncologic diseases.

Integrated Proteogenomic Characterization across Major Histological Types of Pediatric Brain Cancer

We report a comprehensive proteogenomics analysis, including whole-genome sequencing, RNA sequencing, and proteomics and phosphoproteomics profiling, of 218 tumors across 7 histological types of childhood brain cancer: low-grade glioma (n = 93), ependymoma (32), high-grade glioma (25), medulloblastoma (22), ganglioglioma (18), craniopharyngioma (16), and atypical teratoid rhabdoid tumor (12). Proteomics data identify common biological themes that span histological boundaries, suggesting that treatments used for one histological type may be applied effectively to other tumors sharing similar proteomics features. Immune landscape characterization reveals diverse tumor microenvironments across and within diagnoses. Proteomics data further reveal functional effects of somatic mutations and copy number variations (CNVs) not evident in transcriptomics data. Kinase-substrate association and co-expression network analysis identify important biological mechanisms of tumorigenesis. This is the first large-scale proteogenomics analysis across traditional histological boundaries to uncover foundational pediatric brain tumor biology and inform rational treatment selection.

The Clinical Kinase Index: A Method to Prioritize Understudied Kinases as Drug Targets for the Treatment of Cancer

The approval of the first kinase inhibitor, Gleevec, ushered in a paradigm shift for oncological treatment-the use of genomic data for targeted, efficacious therapies. Since then, over 48 additional small-molecule kinase inhibitors have been approved, solidifying the case for kinases as a highly druggable and attractive target class. Despite the role deregulated kinase activity plays in cancer, only 8% of the kinome has been effectively "drugged." Moreover, 24% of the 634 human kinases are understudied. We have developed a comprehensive scoring system that utilizes differential gene expression, pathological parameters, overall survival, and mutational hotspot analysis to rank and prioritize clinically relevant kinases across 17 solid tumor cancers from The Cancer Genome Atlas. We have developed the clinical kinase index (CKI) app (http://cki.ccs.miami.edu) to facilitate interactive analysis of all kinases in each cancer. Collectively, we report that understudied kinases have potential clinical value as biomarkers or drug targets that warrant further study.

LINCS Data Portal 2.0: next generation access point for perturbation-response signatures

The Library of Integrated Network-Based Cellular Signatures (LINCS) is an NIH Common Fund program with the goal of generating a large-scale and comprehensive catalogue of perturbation-response signatures by utilizing a diverse collection of perturbations across many model systems and assay types. The LINCS Data Portal (LDP) has been the primary access point for the compendium of LINCS data and has been widely utilized. Here, we report the first major update of LDP (http://lincsportal.ccs.miami.edu/signatures) with substantial changes in the data architecture and APIs, a completely redesigned user interface, and enhanced curated metadata annotations to support more advanced, intuitive and deeper querying, exploration and analysis capabilities. The cornerstone of this update has been the decision to reprocess all high-level LINCS datasets and make them accessible at the data point level enabling users to directly access and download any subset of signatures across the entire library independent from the originating source, project or assay. Access to the individual signatures also enables the newly implemented signature search functionality, which utilizes the iLINCS platform to identify conditions that mimic or reverse gene set queries. A newly designed query interface enables global metadata search with autosuggest across all annotations associated with perturbations, model systems, and signatures.

GENE-36. INTEGRATING TRANSCRIPTOMICS AND KINOMICS IDENTIFIES SYNERGISTIC DRUG COMBINATIONS FOR GLIOBLASTOMA TREATMENT

Glioblastoma (GBM) is the most common and malignant adult brain tumor. Despite years of research, few advancements have been made in its management. One promising avenue of research has been treatment with BRD4 inhibitors, which decrease oncogene expression in GBM cells. However, resistance to these inhibitors is rapidly acquired. Kinome reprogramming is thought to underlie this resistance, suggesting a need for combination therapy with kinase inhibitors. The goal of this study is to determine whether transcriptomic and kinomic profiling of GBM tumors will identify synergistic drug pairs for GBM treatment. We profiled the active kinome on a set of three newly-diagnosed GBM patient-derived xenograft (PDX) tumors and three recurrent tumors using quantitative SILAC mass spectrometry. Additionally, kinome reprogramming following BET inhibition was profiled in vitro using the BET inhibitor JQ1. Kinome activity was uploaded into our novel computational platform, SynergySeq, to assess the synergistic potential of kinase inhibitors with JQ1. Additionally, single cell RNA-sequencing of a GBM tumor was used to determine the cell populations affected by each inhibitor. To quantify synergy in response to combination therapy, cells were treated with a combination matrix of JQ1 and a kinase inhibitor in variable concentrations and cell death was quantified via ATP levels. Our results showed that newly-diagnosed tumors were predicted to be sensitive to combined BET inhibition with Bcr/Abl, EGFR, and FGFR kinase inhibitors. Recurrent tumors were sensitive to combined Bcr/Abl and FGFR inhibitors but were not sensitive to EGFR inhibitors. We screened inhibitors in vitro and found a synergistic effect for the combination of JQ1 and TAS120, a pan-FGFR inhibitor. These data suggest that clinically a brain penetrant BET inhibitor should be effective in combination with a brain penetrant FGFR inhibitor. Importantly, our computational platform is a novel informatics-based approach for targeted therapy in a patient-specific and disease-specific manner.

COMP-16. COMPREHENSIVE TRANSCRIPTOMIC ANALYSIS OF SINGLE CELLS FROM RECURRENT AND PRIMARY GLIOBLASTOMA TO PREDICT CELL-TYPE SPECIFIC THERAPEUTICS

Glioblastoma (GBM) remains the most common adult brain tumor, with poor survival expectations, and no new therapeutic modalities approved in the last decade. Our laboratories have recently demonstrated that the integration of a transcriptional disease signature obtained from The Cancer Genome Atlas’ GBM dataset with transcriptional cell drug-response signatures in the LINCS L1000 dataset yields possible combinatorial therapeutics. Considering the extreme intra-tumor heterogeneity associated with the disease, we hypothesize that the utilization of single-cell RNA-sequencing (scRNA-seq) of patient tumors will further strengthen our predictive model by providing insight on the unique transcriptomes of the cellular niches present within these tumors, and into the transcriptional dynamics of these same cellular niches. By sequencing single-cell transcriptomes from recurrent GBM tumors resected from patients at the University of Miami, and integrating our datasets with previously published scRNA-seq data from primary GBM tumors, we are able to gain additional insight into the differences between these clinical distinctions. We have analyzed the differential expression of kinases both across and within distinct cell populations of primary and recurrent GBM tumors. This transcriptional map of kinase expression represents the heterogeneity of potential targets within individual tumors and between recurrent and primary GBM. Additionally, by generating disease signatures unique to each cellular population, and integrating these with transcriptional drug-response signatures from LINCS, we are able to predict compounds to target specific cell populations within GMB tumors. Additional computational techniques such as RNA velocity analysis and cell cycle scoring elucidate temporal insights to further prioritize these cell-type specific therapeutics, and reveal the intra-cellular dynamics present within these tumors. Collectively, our studies suggest that we have developed a novel omics pipeline based on the single cell RNA-sequencing of individual GBM cells that addresses intra-tumor heterogeneity, and may lead to novel therapeutic combinations for the treatment of this incurable disease.

Data Portal for the Library of Integrated Network-based Cellular Signatures (LINCS) program: integrated access to diverse large-scale cellular perturbation response data

The Library of Integrated Network-based Cellular Signatures (LINCS) program is a national consortium funded by the NIH to generate a diverse and extensive reference library of cell-based perturbation-response signatures, along with novel data analytics tools to improve our understanding of human diseases at the systems level. In contrast to other large-scale data generation efforts, LINCS Data and Signature Generation Centers (DSGCs) employ a wide range of assay technologies cataloging diverse cellular responses. Integration of, and unified access to LINCS data has therefore been particularly challenging. The Big Data to Knowledge (BD2K) LINCS Data Coordination and Integration Center (DCIC) has developed data standards specifications, data processing pipelines, and a suite of end-user software tools to integrate and annotate LINCS-generated data, to make LINCS signatures searchable and usable for different types of users. Here, we describe the LINCS Data Portal (LDP) (http://lincsportal.ccs.miami.edu/), a unified web interface to access datasets generated by the LINCS DSGCs, and its underlying database, LINCS Data Registry (LDR). LINCS data served on the LDP contains extensive metadata and curated annotations. We highlight the features of the LDP user interface that is designed to enable search, browsing, exploration, download and analysis of LINCS data and related curated content.

Abstract 5103: The dark cancer kinome - untapped opportunities for the development of novel drugs

Kinases are firmly established drug targets in cancer. There are currently 44 FDA approved kinase drug and hundreds of compounds are in clinical development. However, less than 10% of the Kinome is currently targeted and a large proportion is considered understudied by the NIH Illuminating the Druggable Genome Program (https://druggablegenome.net/). No small molecule inhibitors are known for these “dark” proteins, yet many may be opportune novel cancer targets.We developed a computational pipeline to identify and prioritize understudied kinases as cancer drug targets. We analyzed the complete set of tumors in The Cancer Genome Atlas (TCGA). For 33 different cancers we performed differential expression analysis and identified 39 dark kinases that exhibit significant upregulation in at least four types. Using co-expression analysis we built functional networks prioritizing drug targets. To identify small molecules that reverse their expression levels, we leveraged transcriptional response signatures obtained from dozens of human cancer cell lines exposed to tens of thousands of small molecules from the Library of Integrated Network-based Cellular Signatures (LINCS). To identify small molecules that directly bind to and inhibit dark kinases, we have have combined an advanced AI (artificial intelligence) model trained on activity data from across the Kinome with structure-based simulations.Using the computational pipeline, we identified the dark Ca2+/Calmodulin dependent kinase PNCK as the most differentially overexpressed kinase in kidney cancer patients. Our analyses have demonstrated statistically significant correlation between PNCK mRNA levels and various clinical and pathological outcomes, including histologic grade, clinical staging and overall survival. We have confirmed high levels of PNCK expression in 5 renal cell carcinoma cell lines (Caki-1, ACHN, 786-O, A704 and A498). Knockdown and overexpression studies have suggested PNCK and the CaMK pathway may contribute to cellular proliferation and cell cycle progression. We have applied our AI-based screening pipeline to a library of >20 million commercially available compounds and confirmed three PNCK inhibiting chemotypes. In summary, using a novel computational pipeline, we have identified and experimentally validated PNCK as a prospective novel drug target in an understudied pathway that is highly upregulated in kidney cancer. We identified first in class small molecules that target this previously dark kinase as prospective starting points for optimization into a clinical candidate.

Citation Format: Derek J. Essegian, Rimpi Khurana, Vasileios Stathias, Valery Chavez, Jaime R. Merchan, Stephan Schürer. The dark cancer kinome - untapped opportunities for the development of novel drugs [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 5103.

DrugCentral 2018: an update

DrugCentral is a drug information resource (http://drugcentral.org) open to the public since 2016 and previously described in the 2017 Nucleic Acids Research Database issue. Since the 2016 release, 103 new approved drugs were updated. The following new data sources have been included: Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS), FDA Orange Book information, L1000 gene perturbation profile distance/similarity matrices and estimated protonation constants. New and existing entries have been updated with the latest information from scientific literature, drug labels and external databases. The web interface has been updated to display and query new data. The full database dump and data files are available for download from the DrugCentral website.

CARM1 Is Essential for Myeloid Leukemogenesis but Dispensable for Normal Hematopoiesis

Chromatin-modifying enzymes, and specifically the protein arginine methyltransferases (PRMTs), have emerged as important targets in cancer. Here, we investigated the role of CARM1 in normal and malignant hematopoiesis. Using conditional knockout mice, we show that loss of CARM1 has little effect on normal hematopoiesis. Strikingly, knockout of Carm1 abrogates both the initiation and maintenance of acute myeloid leukemia (AML) driven by oncogenic transcription factors. We show that CARM1 knockdown impairs cell-cycle progression, promotes myeloid differentiation, and ultimately induces apoptosis. Finally, we utilize a selective, small-molecule inhibitor of CARM1 to validate the efficacy of CARM1 inhibition in leukemia cells in vitro and in vivo. Collectively, this work suggests that targeting CARM1 may be an effective therapeutic strategy for AML.

The Library of Integrated Network-Based Cellular Signatures NIH Program: System-Level Cataloging of Human Cells Response to Perturbations

The Library of Integrated Network-Based Cellular Signatures (LINCS) is an NIH Common Fund program that catalogs how human cells globally respond to chemical, genetic, and disease perturbations. Resources generated by LINCS include experimental and computational methods, visualization tools, molecular and imaging data, and signatures. By assembling an integrated picture of the range of responses of human cells exposed to many perturbations, the LINCS program aims to better understand human disease and to advance the development of new therapies. Perturbations under study include drugs, genetic perturbations, tissue micro-environments, antibodies, and disease-causing mutations. Responses to perturbations are measured by transcript profiling, mass spectrometry, cell imaging, and biochemical methods, among other assays. The LINCS program focuses on cellular physiology shared among tissues and cell types relevant to an array of diseases, including cancer, heart disease, and neurodegenerative disorders. This Perspective describes LINCS technologies, datasets, tools, and approaches to data accessibility and reusability.

Ontological representation, integration, and analysis of LINCS cell line cells and their cellular responses

Background

Aiming to understand cellular responses to different perturbations, the NIH Common Fund Library of Integrated Network-based Cellular Signatures (LINCS) program involves many institutes and laboratories working on over a thousand cell lines. The community-based Cell Line Ontology (CLO) is selected as the default ontology for LINCS cell line representation and integration.

Results

CLO has consistently represented all 1097 LINCS cell lines and included information extracted from the LINCS Data Portal and ChEMBL. Using MCF 10A cell line cells as an example, we demonstrated how to ontologically model LINCS cellular signatures such as their non-tumorigenic epithelial cell type, three-dimensional growth, latrunculin-A-induced actin depolymerization and apoptosis, and cell line transfection. A CLO subset view of LINCS cell lines, named LINCS-CLOview, was generated to support systematic LINCS cell line analysis and queries. In summary, LINCS cell lines are currently associated with 43 cell types, 131 tissues and organs, and 121 cancer types. The LINCS-CLO view information can be queried using SPARQL scripts.

Conclusions

CLO was used to support ontological representation, integration, and analysis of over a thousand LINCS cell line cells and their cellular responses.

Drug target ontology to classify and integrate drug discovery data

Background

One of the most successful approaches to develop new small molecule therapeutics has been to start from a validated druggable protein target. However, only a small subset of potentially druggable targets has attracted significant research and development resources. The Illuminating the Druggable Genome (IDG) project develops resources to catalyze the development of likely targetable, yet currently understudied prospective drug targets. A central component of the IDG program is a comprehensive knowledge resource of the druggable genome.

Results

As part of that effort, we have developed a framework to integrate, navigate, and analyze drug discovery data based on formalized and standardized classifications and annotations of druggable protein targets, the Drug Target Ontology (DTO). DTO was constructed by extensive curation and consolidation of various resources. DTO classifies the four major drug target protein families, GPCRs, kinases, ion channels and nuclear receptors, based on phylogenecity, function, target development level, disease association, tissue expression, chemical ligand and substrate characteristics, and target-family specific characteristics. The formal ontology was built using a new software tool to auto-generate most axioms from a database while supporting manual knowledge acquisition. A modular, hierarchical implementation facilitate ontology development and maintenance and makes use of various external ontologies, thus integrating the DTO into the ecosystem of biomedical ontologies. As a formal OWL-DL ontology, DTO contains asserted and inferred axioms. Modeling data from the Library of Integrated Network-based Cellular Signatures (LINCS) program illustrates the potential of DTO for contextual data integration and nuanced definition of important drug target characteristics. DTO has been implemented in the IDG user interface Portal, Pharos and the TIN-X explorer of protein target disease relationships.

Conclusions

DTO was built based on the need for a formal semantic model for druggable targets including various related information such as protein, gene, protein domain, protein structure, binding site, small molecule drug, mechanism of action, protein tissue localization, disease association, and many other types of information. DTO will further facilitate the otherwise challenging integration and formal linking to biological assays, phenotypes, disease models, drug poly-pharmacology, binding kinetics and many other processes, functions and qualities that are at the core of drug discovery. The first version of DTO is publically available via the website http://drugtargetontology.org/, Github (http://github.com/DrugTargetOntology/DTO), and the NCBO Bioportal (http://bioportal.bioontology.org/ontologies/DTO). The long-term goal of DTO is to provide such an integrative framework and to populate the ontology with this information as a community resource.

Electronic supplementary material

The online version of this article (10.1186/s13326-017-0161-x) contains supplementary material, which is available to authorized users.

Abstract 416: Identification of therapeutic combinations in glioblastoma using personalized gene expression networks

The goal of our study was to identify patient-specific gene expression networks from Glioblastoma Patient-Derived Xenografts (PDXs) and determine novel therapeutic compound combinations using those networks.

Glioblastoma is the most common malignant primary adult brain tumor with a standard of care consisting of maximal surgical resection followed by radiotherapy and adjuvant temozolomide (TMZ) chemotherapy. However, despite medical advances in the field, recurrence is almost universal, suggesting the need for more personalized and targeted therapeutic approaches.

For this, we obtained, transcriptional data from Glioblastoma PDXs and used them to identify their respective differentially expressed genes. Patient-specific gene expression networks were then created and their biological relevance was supplemented by integrating them with TCGA Glioblastoma transcriptional data. In order to identify compound combinations specific for those networks, we used the extensive chemical perturbation signatures from the Library of Network-based Cellular Signatures (LINCS). From the large number of L1000 transcriptional data we extracted gene expression signatures that were indicative of specific LINCS compounds. We then compared those signatures to the patient-specific networks in order to prioritize compound combinations that were inducing discordant transcriptional changes in distinct sub-networks of the PDXs transcriptome. The most orthogonal compound combinations were then chosen and used in in-vitro cell viability assays of Glioblastoma PDXs in order to evaluate their effectiveness. The above process can be used to prioritize compound combinations with potential therapeutic effect in a patient-specific manner.

Citation Format: Vasileios Stathias, Michele Forlin, Bryce Allen, Stephan Schürer, Nagi G. Ayad. Identification of therapeutic combinations in glioblastoma using personalized gene expression networks [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 416. doi:10.1158/1538-7445.AM2017-416

Abstract 775: Combinatorial compound stratification based on integration of LINCS, TCGA and PubChem data

The purpose of this study is to leverage the large number of perturbation signatures from the NIH Library of Integrative Network-based Cellular Signatures (LINCS) and integrate them with patient data from The Cancer Genome Atlas (TCGA) and PubChem bioactivity data in an effort to prioritize compounds based on their synergistic effect in cancer treatment.

From the large number of LINCS L1000 transcriptional data capturing cellular responses after chemical or genetic perturbations, we extracted gene expression signatures that were indicative of specific LINCS compounds. Moreover, we compared the L1000 transcriptional profiles with ones from TCGA in order to identify characteristic signatures of major cancer types and prioritized small molecule compounds with discordant expression profiles to those cancer types. We then linked the above compounds to protein target annotations and therefore produced a compound-specific protein target profile. For this, we utilized biochemical data produced by the LINCS KinomeScan and KiNative assays and also biochemical data obtained through PubChem. Using the above information, we obtained pairs of compounds that would inhibit unlinked gene sub-networks that were produced through processing of the TCGA transcriptional data.

The above process can be used as a means to suggest compound combinations towards specific cancer types and to prioritize the development of compounds with targeted polypharmacology.

Citation Format: Vasileios Stathias, Bryce Allen, Jennifer Clarke, Nagi G. Ayad, Stephan Schürer. Combinatorial compound stratification based on integration of LINCS, TCGA and PubChem data. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 775.

Epigenetic pathways and glioblastoma treatment: insights from signaling cascades

There is an urgent need to identify novel therapies for glioblastoma (GBM) as most therapies are ineffective. A first step in this process is to identify and validate targets for therapeutic intervention. Epigenetic modulators have emerged as attractive drug targets in several cancers including GBM. These epigenetic regulators affect gene expression without changing the DNA sequence. Recent studies suggest that epigenetic regulators interact with drivers of GBM cell and stem-like cell proliferation. These drivers include components of the Notch, Hedgehog, and Wingless (WNT) pathways. We highlight recent studies connecting epigenetic and signaling pathways in GBM. We also review systems and big data approaches for identifying patient specific therapies in GBM. Collectively, these studies will identify drug combinations that may be effective in GBM and other cancers.

Abstract 65: An integrated bioinformatics approach for identifying patient-specific gene networks and novel therapeutic targets in glioblastoma

The purpose of our approach is to identify highly relevant and patient specific gene networks from next generation sequencing data and use the resulting gene combinations to identify potential novel therapeutic targets.

Glioblastoma multiforme (GBM) is the most common and aggressive malignant brain tumor. Despite medical advances in the field, the median survival time is still below 2 years since recurrence is nearly universal. Thus, the discovery of novel and specific molecular targets is needed.

As with all cancers, GBM arises due to complex alterations in a patient's genome. Identifying patient-specific differentially expressed genes (DEGs) though, can impose many difficulties given the discordance of different analysis methods when the sample size is low. We used the data on the large number of GBM patients within TCGA (The Cancer Genome Atlas) as filter for identifying variations in expression. With the help of our algorithm, we produced enriched single patient RNAseq DEG lists and we calculated a hypergeometric probability and a correlation coefficient for every gene pair on these lists. By using the most significant of these pairs, we generated gene association networks. To further validate the biological relevance of the gene pairs, we looked into the presence of these networks in other types of cancer and searched for published experimental data supporting the connections of our networks. In order to assess the therapeutic potential of these genes, we used complimentary experiments involving GBM cell lines recorded through the LINCS (Library of Integrated Network-based Cell Signatures) database.

This study identifies a putative workflow for uncovering differentially expressed patient specific genes and gene networks for GBM, and capitalizes on the LINCS database to assess the potential of novel therapeutic targets.

Citation Format: Vasileios Stathias, Chiara Pastori, Ricardo Komotar, Ming Zhang, Stephan Schürer, Jennifer Clarke, Nagi G. Ayad. An integrated bioinformatics approach for identifying patient-specific gene networks and novel therapeutic targets in glioblastoma. [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 65. doi:10.1158/1538-7445.AM2015-65

Identifying glioblastoma gene networks based on hypergeometric test analysis

Patient specific therapy is emerging as an important possibility for many cancer patients. However, to identify such therapies it is essential to determine the genomic and transcriptional alterations present in one tumor relative to control samples. This presents a challenge since use of a single sample precludes many standard statistical analysis techniques. We reasoned that one means of addressing this issue is by comparing transcriptional changes in one tumor with those observed in a large cohort of patients analyzed by The Cancer Genome Atlas (TCGA). To test this directly, we devised a bioinformatics pipeline to identify differentially expressed genes in tumors resected from patients suffering from the most common malignant adult brain tumor, glioblastoma (GBM). We performed RNA sequencing on tumors from individual GBM patients and filtered the results through the TCGA database in order to identify possible gene networks that are overrepresented in GBM samples relative to controls. Importantly, we demonstrate that hypergeometric-based analysis of gene pairs identifies gene networks that validate experimentally. These studies identify a putative workflow for uncovering differentially expressed patient specific genes and gene networks for GBM and other cancers.

Support of the histaminergic hypothesis in Tourette syndrome: association of the histamine decarboxylase gene in a large sample of families

BACKGROUND

Gilles de la Tourette Syndrome is a neurodevelopmental disorder that is caused by the interaction of environment with a complex genetic background. The genetic etiology of the disorder remains, so far, elusive, although multiple promising leads have been recently reported. The recent implication of the histamine decarboxylase (HDC) gene, the key enzyme in histamine production, raises the intriguing hypothesis of a possible role of histaminergic dysfunction leading to TS onset.

METHODS

Following up on the finding of a nonsense mutation in a single family with TS, we investigated variation across the HDC gene for association with TS. As a result of a collaborative international effort, we studied a large sample of 520 nuclear families originating from seven European populations (Greek, Hungarian, Italian, Polish, German, Albanian, Spanish) as well as a sample collected in Canada.

RESULTS AND CONCLUSIONS

Interrogating 12 tagging SNPs (tSNP) across the HDC region, we find strong over-transmission of alleles at two SNPs (rs854150 and rs1894236) in the complete sample, as well as a statistically significant associated haplotypes. Analysis of individual populations also reveals signals of association in the Canadian, German and Italian samples. Our results provide strong support for the histaminergic hypothesis in TS etiology and point to a possible role of histamine pathways in neuronal development.

Exploring genomic structure differences and similarities between the Greek and European HapMap populations: implications for association studies

Studies of the genomic structure of the Greek population and Southeastern Europe are limited, despite the central position of the area as a gateway for human migrations into Europe. HapMap has provided a unique tool for the analysis of human genetic variation. Europe is represented by the CEU (Northwestern Europe) and the TSI populations (Tuscan Italians from Southern Europe), which serve as reference for the design of genetic association studies. Furthermore, genetic association findings are often transferred to unstudied populations. Although initial studies support the fact that the CEU can, in general, be used as reference for the selection of tagging SNPs in European populations, this has not been extensively studied across Europe. We set out to explore the genomic structure of the Greek population (56 individuals) and compare it to the HapMap TSI and CEU populations. We studied 1112 SNPs (27 regions, 13 chromosomes). Although the HapMap European populations are, in general, a good reference for the Greek population, regions of population differentiation do exist and results should not be light-heartedly generalized. We conclude that, perhaps due to the individual evolutionary history of each genomic region, geographic proximity is not always a perfect guide for selecting a reference population for an unstudied population.