Functional Proteomics and Deep Network Interrogation Reveal a Complex Mechanism of Action of Midostaurin in Lung Cancer Cells.

Lung cancer is associated with high prevalence and mortality, and despite significant successes with targeted drugs in genomically defined subsets of lung cancer and immunotherapy, the majority of patients currently does not benefit from these therapies. Through a targeted drug screen, we found the recently approved multi-kinase inhibitor midostaurin to have potent activity in several lung cancer cells independent of its intended target, PKC, or a specific genomic marker. To determine the underlying mechanism of action we applied a layered functional proteomics approach and a new data integration method. Using chemical proteomics, we identified multiple midostaurin kinase targets in these cells. Network-based integration of these targets with quantitative tyrosine and global phosphoproteomics data using protein-protein interactions from the STRING database suggested multiple targets are relevant for the mode of action of midostaurin. Subsequent functional validation using RNA interference and selective small molecule probes showed that simultaneous inhibition of TBK1, PDPK1 and AURKA was required to elicit midostaurin's cellular effects. Immunoblot analysis of downstream signaling nodes showed that combined inhibition of these targets altered PI3K/AKT and cell cycle signaling pathways that in part converged on PLK1. Furthermore, rational combination of midostaurin with the potent PLK1 inhibitor BI2536 elicited strong synergy. Our results demonstrate that combination of complementary functional proteomics approaches and subsequent network-based data integration can reveal novel insight into the complex mode of action of multi-kinase inhibitors, actionable targets for drug discovery and cancer vulnerabilities. Finally, we illustrate how this knowledge can be used for the rational design of synergistic drug combinations with high potential for clinical translation.

Polypharmacology-based ceritinib repurposing using integrated functional proteomics.

Targeted drugs are effective when they directly inhibit strong disease drivers, but only a small fraction of diseases feature defined actionable drivers. Alternatively, network-based approaches can uncover new therapeutic opportunities. Applying an integrated phenotypic screening, chemical and phosphoproteomics strategy, here we describe the anaplastic lymphoma kinase (ALK) inhibitor ceritinib as having activity across several ALK-negative lung cancer cell lines and identify new targets and network-wide signaling effects. Combining pharmacological inhibitors and RNA interference revealed a polypharmacology mechanism involving the noncanonical targets IGF1R, FAK1, RSK1 and RSK2. Mutating the downstream signaling hub YB1 protected cells from ceritinib. Consistent with YB1 signaling being known to cause taxol resistance, combination of ceritinib with paclitaxel displayed strong synergy, particularly in cells expressing high FAK autophosphorylation, which we show to be prevalent in lung cancer. Together, we present a systems chemical biology platform for elucidating multikinase inhibitor polypharmacology mechanisms, subsequent design of synergistic drug combinations, and identification of mechanistic biomarker candidates.

APOSTL: An Interactive Galaxy Pipeline for Reproducible Analysis of Affinity Proteomics Data.

With continuously increasing scale and depth of coverage in affinity proteomics (AP-MS) data, the analysis and visualization is becoming more challenging. A number of tools have been developed to identify high-confidence interactions; however, a cohesive and intuitive pipeline for analysis and visualization is still needed. Here we present Automated Processing of SAINT Templated Layouts (APOSTL), a freely available Galaxy-integrated software suite and analysis pipeline for reproducible, interactive analysis of AP-MS data. APOSTL contains a number of tools woven together using Galaxy workflows, which are intuitive for the user to move from raw data to publication-quality figures within a single interface. APOSTL is an evolving software project with the potential to customize individual analyses with additional Galaxy tools and widgets using the R web application framework, Shiny. The source code, data, and documentation are freely available from GitHub ( https://github.com/bornea/APOSTL ) and other sources.

De novo generation of multi-target compounds using deep generative chemistry.

Polypharmacology drugs-compounds that inhibit multiple proteins-have many applications but are difficult to design. To address this challenge we have developed POLYGON, an approach to polypharmacology based on generative reinforcement learning. POLYGON embeds chemical space and iteratively samples it to generate new molecular structures; these are rewarded by the predicted ability to inhibit each of two protein targets and by drug-likeness and ease-of-synthesis. In binding data for >100,000 compounds, POLYGON correctly recognizes polypharmacology interactions with 82.5% accuracy. We subsequently generate de-novo compounds targeting ten pairs of proteins with documented co-dependency. Docking analysis indicates that top structures bind their two targets with low free energies and similar 3D orientations to canonical single-protein inhibitors. We synthesize 32 compounds targeting MEK1 and mTOR, with most yielding >50% reduction in each protein activity and in cell viability when dosed at 1-10 µM. These results support the potential of generative modeling for polypharmacology.

IGF-binding proteins secreted by cancer-associated fibroblasts induce context-dependent drug sensitization of lung cancer cells.

Cancer-associated fibroblasts (CAFs) in the tumor microenvironment are often linked to drug resistance. Here, we found that coculture with CAFs or culture in CAF-conditioned medium unexpectedly induced drug sensitivity in certain lung cancer cell lines. Gene expression and secretome analyses of CAFs and normal lung-associated fibroblasts (NAFs) revealed differential abundance of insulin-like growth factors (IGFs) and IGF-binding proteins (IGFBPs), which promoted or inhibited, respectively, signaling by the receptor IGF1R and the kinase FAK. Similar drug sensitization was seen in gefitinib-resistant, EGFR-mutant PC9GR lung cancer cells treated with recombinant IGFBPs. Conversely, drug sensitivity was decreased by recombinant IGFs or conditioned medium from CAFs in which IGFBP5 or IGFBP6 was silenced. Phosphoproteomics and receptor tyrosine kinase (RTK) array analyses indicated that exposure of PC9GR cells to CAF-conditioned medium also inhibited compensatory IGF1R and FAK signaling induced by the EGFR inhibitor osimertinib. Combined small-molecule inhibition of IGF1R and FAK phenocopied the CAF-mediated effects in culture and increased the antitumor effect of osimertinib in mice. Cells that were osimertinib resistant and had MET amplification or showed epithelial-to-mesenchymal transition also displayed residual sensitivity to IGFBPs. Thus, CAFs promote or reduce drug resistance in a context-dependent manner, and deciphering the relationship between the differential content of CAF secretomes and the signaling dependencies of the tumor may reveal effective combination treatment strategies.

A census of pathway maps in cancer systems biology.

A key goal of cancer systems biology is to use big data to elucidate the molecular networks by which cancer develops. However, to date there has been no systematic evaluation of how far these efforts have progressed. In this Analysis, we survey six major systems biology approaches for mapping and modelling cancer pathways with attention to how well their resulting network maps cover and enhance current knowledge. Our sample of 2,070 systems biology maps captures all literature-curated cancer pathways with significant enrichment, although the strong tendency is for these maps to recover isolated mechanisms rather than entire integrated processes. Systems biology maps also identify previously underappreciated functions, such as a potential role for human papillomavirus-induced chromosomal alterations in ovarian tumorigenesis, and they add new genes to known cancer pathways, such as those related to metabolism, Hippo signalling and immunity. Notably, we find that many cancer networks have been provided only in journal figures and not for programmatic access, underscoring the need to deposit network maps in community databases to ensure they can be readily accessed. Finally, few of these findings have yet been clinically translated, leaving ample opportunity for future translational studies. Periodic surveys of cancer pathway maps, such as the one reported here, are critical to assess progress in the field and identify underserved areas of methodology and cancer biology.

Predicting Drug Response and Synergy Using a Deep Learning Model of Human Cancer Cells.

Most drugs entering clinical trials fail, often related to an incomplete understanding of the mechanisms governing drug response. Machine learning techniques hold immense promise for better drug response predictions, but most have not reached clinical practice due to their lack of interpretability and their focus on monotherapies. We address these challenges by developing DrugCell, an interpretable deep learning model of human cancer cells trained on the responses of 1,235 tumor cell lines to 684 drugs. Tumor genotypes induce states in cellular subsystems that are integrated with drug structure to predict response to therapy and, simultaneously, learn biological mechanisms underlying the drug response. DrugCell predictions are accurate in cell lines and also stratify clinical outcomes. Analysis of DrugCell mechanisms leads directly to the design of synergistic drug combinations, which we validate systematically by combinatorial CRISPR, drug-drug screening in vitro, and patient-derived xenografts. DrugCell provides a blueprint for constructing interpretable models for predictive medicine.

Off-target based drug repurposing opportunities for tivantinib in acute myeloid leukemia.

GSK3α has been identified as a new target in the treatment of acute myeloid leukemia (AML). However, most GSK3 inhibitors lack specificity for GSK3α over GSK3ß and other kinases. We have previously shown in lung cancer cells that GSK3α and to a lesser extent GSK3ß are inhibited by the advanced clinical candidate tivantinib (ARQ197), which was designed as a MET inhibitor. Thus, we hypothesized that tivantinib would be an effective therapy for the treatment of AML. Here, we show that tivantinib has potent anticancer activity across several AML cell lines and primary patient cells. Tivantinib strongly induced apoptosis, differentiation and G2/M cell cycle arrest and caused less undesirable stabilization of ß-catenin compared to the pan-GSK3 inhibitor LiCl. Subsequent drug combination studies identified the BCL-2 inhibitor ABT-199 to synergize with tivantinib while cytarabine combination with tivantinib was antagonistic. Interestingly, the addition of ABT-199 to tivantinib completely abrogated tivantinib induced ß-catenin stabilization. Tivantinib alone, or in combination with ABT-199, downregulated anti-apoptotic MCL-1 and BCL-XL levels, which likely contribute to the observed synergy. Importantly, tivantinib as single agent or in combination with ABT-199 significantly inhibited the colony forming capacity of primary patient AML bone marrow mononuclear cells. In summary, tivantinib is a novel GSK3α/ß inhibitor that potently kills AML cells and tivantinib single agent or combination therapy with ABT-199 may represent attractive new therapeutic opportunities for AML.

The Galaxy Platform for Reproducible Affinity Proteomic Mass Spectrometry Data Analysis.

Affinity proteomics (AP-MS) is growing in importance for characterizing protein-protein interactions (PPIs) in the form of protein complexes and signaling networks. The AP-MS approach necessitates several different software tools, integrated into reproducible and accessible workflows. However, if the scientist (e.g., a bench biologist) lacks a computational background, then managing large AP-MS datasets can be challenging, manually formatting AP-MS data for input into analysis software can be error-prone, and data visualization involving dozens of variables can be laborious. One solution to address these issues is Galaxy, an open source and web-based platform for developing and deploying user-friendly computational pipelines or workflows. Here, we describe a Galaxy-based platform enabling AP-MS analysis. This platform enables researchers with no prior computational experience to begin with data from a mass spectrometer (e.g., peaklists in mzML format) and perform peak processing, database searching, assignment of interaction confidence scores, and data visualization with a few clicks of a mouse. We provide sample data and a sample workflow with step-by-step instructions to quickly acquaint users with the process.

An immunoproteomic approach to characterize the CAR interactome and signalosome.

Adoptive transfer of T cells that express a chimeric antigen receptor (CAR) is an approved immunotherapy that may be curative for some hematological cancers. To better understand the therapeutic mechanism of action, we systematically analyzed CAR signaling in human primary T cells by mass spectrometry. When we compared the interactomes and the signaling pathways activated by distinct CAR-T cells that shared the same antigen-binding domain but differed in their intracellular domains and their in vivo antitumor efficacy, we found that only second-generation CARs induced the expression of a constitutively phosphorylated form of CD3ζ that resembled the endogenous species. This phenomenon was independent of the choice of costimulatory domains, or the hinge/transmembrane region. Rather, it was dependent on the size of the intracellular domains. Moreover, the second-generation design was also associated with stronger phosphorylation of downstream secondary messengers, as evidenced by global phosphoproteome analysis. These results suggest that second-generation CARs can activate additional sources of CD3ζ signaling, and this may contribute to more intense signaling and superior antitumor efficacy that they display compared to third-generation CARs. Moreover, our results provide a deeper understanding of how CARs interact physically and/or functionally with endogenous T cell molecules, which will inform the development of novel optimized immune receptors.

Ceritinib Enhances the Efficacy of Trametinib in BRAF/NRAS-Wild-Type Melanoma Cell Lines.

Targeted therapy options are currently lacking for the heterogeneous population of patients whose melanomas lack BRAF or NRAS mutations (∼35% of cases). We undertook a chemical biology screen to identify potential novel drug targets for this understudied group of tumors. Screening a panel of 8 BRAF/NRAS-WT melanoma cell lines against 240 targeted drugs identified ceritinib and trametinib as potential hits with single-agent activity. Ceritinib enhanced the efficacy of trametinib across the majority of the BRAF/NRAS-WT cell lines, and the combination showed increased cytotoxicity in both three-dimensional spheroid culture and long-term colony formation experiments. Coadministration of ceritinib and trametinib led to robust inhibition of tumor growth in an in vivo xenograft BRAF/NRAS-WT melanoma model; this was not due to ALK inhibition by ceritinib. Mechanistic studies showed the ceritinib-trametinib combination to increase suppression of MAPK and TORC1 signaling. Similar results were seen when BRAF/NRAS-WT melanoma cells were treated with a combination of trametinib and the TORC1/2 inhibitor INK128. We next used mass spectrometry-based chemical proteomics and identified known and new ceritinib targets, such as IGF1R and ACK1, respectively. Validation studies suggested that ceritinib could suppress mTORC1 signaling in the presence of trametinib through inhibition of IGF1R and/or ACK1 in a cell line-dependent manner. Together, our studies demonstrated that combining a specific inhibitor (trametinib) with a more broadly targeted agent (ceritinib) has efficacy against tumors with heterogeneous mutational profiles. Mol Cancer Ther; 17(1); 73-83. ©2017 AACR.

Targeting the BRD4-HOXB13 Coregulated Transcriptional Networks with Bromodomain-Kinase Inhibitors to Suppress Metastatic Castration-Resistant Prostate Cancer.

Resistance to androgen receptor (AR) antagonists is a significant problem in the treatment of castration-resistant prostate cancers (CRPC). Identification of the mechanisms by which CRPCs evade androgen deprivation therapies (ADT) is critical to develop novel therapeutics. We uncovered that CRPCs rely on BRD4-HOXB13 epigenetic reprogramming for androgen-independent cell proliferation. Mechanistically, BRD4, a member of the BET bromodomain family, epigenetically promotes HOXB13 expression. Consistently, genetic disruption of HOXB13 or pharmacological suppression of its mRNA and protein expression by the novel dual-activity BET bromodomain-kinase inhibitors directly correlates with rapid induction of apoptosis, potent inhibition of tumor cell proliferation and cell migration, and suppression of CRPC growth. Integrative analysis revealed that the BRD4-HOXB13 transcriptome comprises a proliferative gene network implicated in cell-cycle progression, nucleotide metabolism, and chromatin assembly. Notably, although the core HOXB13 target genes responsive to BET inhibitors (HOTBIN10) are overexpressed in metastatic cases, in ADT-treated CRPC cell lines and patient-derived circulating tumor cells (CTC) they are insensitive to AR depletion or blockade. Among the HOTBIN10 genes, AURKB and MELK expression correlates with HOXB13 expression in CTCs of mCRPC patients who did not respond to abiraterone (ABR), suggesting that AURKB inhibitors could be used additionally against high-risk HOXB13-positive metastatic prostate cancers. Combined, our study demonstrates that BRD4-HOXB13-HOTBIN10 regulatory circuit maintains the malignant state of CRPCs and identifies a core proproliferative network driving ADT resistance that is targetable with potent dual-activity bromodomain-kinase inhibitors.

Escape Excel: A tool for preventing gene symbol and accession conversion errors.

BACKGROUND: Microsoft Excel automatically converts certain gene symbols, database accessions, and other alphanumeric text into dates, scientific notation, and other numerical representations. These conversions lead to subsequent, irreversible, corruption of the imported text. A recent survey of popular genomic literature estimates that one-fifth of all papers with supplementary gene lists suffer from this issue. RESULTS: Here, we present an open-source tool, Escape Excel, which prevents these erroneous conversions by generating an escaped text file that can be safely imported into Excel. Escape Excel is implemented in a variety of formats (http://www.github.com/pstew/escape_excel), including a command line based Perl script, a Windows-only Excel Add-In, an OS X drag-and-drop application, a simple web-server, and as a Galaxy web environment interface. Test server implementations are accessible as a Galaxy interface (http://apostl.moffitt.org) and simple non-Galaxy web server (http://apostl.moffitt.org:8000/). CONCLUSIONS: Escape Excel detects and escapes a wide variety of problematic text strings so that they are not erroneously converted into other representations upon importation into Excel. Examples of problematic strings include date-like strings, time-like strings, leading zeroes in front of numbers, and long numeric and alphanumeric identifiers that should not be automatically converted into scientific notation. It is hoped that greater awareness of these potential data corruption issues, together with diligent escaping of text files prior to importation into Excel, will help to reduce the amount of Excel-corrupted data in scientific analyses and publications.

Sustained activation of the AKT/mTOR and MAP kinase pathways mediate resistance to the Src inhibitor, dasatinib, in thyroid cancer.

New targeted therapies are needed for advanced thyroid cancer. Our lab has shown that Src is a key mediator of tumorigenic processes in thyroid cancer. However, single-agent Src inhibitors have had limited efficacy in solid tumors. In order to more effectively target Src in the clinic, our lab has previously generated four thyroid cancer cell lines that are resistant to dasatinib through gradual dose escalation. We further tested two additional Src inhibitors and shown the dasatinib-resistant (DasRes) cells exhibit cross-resistance to saracatinib, but are sensitive to bosutinib, suggesting that unique off-targets of bosutinib play an important role in mediating sensitivity to bosutinib. To identify the kinases targeted by dasatinib and bosutinib, we utilized an unbiased compound centric chemical proteomics screen. We identified 33 kinases that were enriched in the bosutinib pull down. Using the STRING database to map protein-protein interactions of the unique bosutinib targets, we identified a signaling axis which included mTOR, FAK, and MEK. Inhibition of the mTOR, MEK, and Src/FAK nodes simultaneously was the most effective at reducing cell growth and survival. Overall, these studies have identified key mediators of Src inhibitor resistance, and show that targeting these signaling nodes are necessary for anti-tumor efficacy.

Proteome-wide Profiling of Clinical PARP Inhibitors Reveals Compound-Specific Secondary Targets.

Poly(ADP-ribose) polymerase (PARP) inhibitors (PARPi) are a promising class of targeted cancer drugs, but their individual target profiles beyond the PARP family, which could result in differential clinical use or toxicity, are unknown. Using an unbiased, mass spectrometry-based chemical proteomics approach, we generated a comparative proteome-wide target map of the four clinical PARPi, olaparib, veliparib, niraparib, and rucaparib. PARPi as a class displayed high target selectivity. However, in addition to the canonical targets PARP1, PARP2, and several of their binding partners, we also identified hexose-6-phosphate dehydrogenase (H6PD) and deoxycytidine kinase (DCK) as previously unrecognized targets of rucaparib and niraparib, respectively. Subsequent functional validation suggested that inhibition of DCK by niraparib could have detrimental effects when combined with nucleoside analog pro-drugs. H6PD silencing can cause apoptosis and further sensitize cells to PARPi, suggesting that H6PD may be, in addition to its established role in metabolic disorders, a new anticancer target.

Chemoproteomics Reveals Novel Protein and Lipid Kinase Targets of Clinical CDK4/6 Inhibitors in Lung Cancer.

Several selective CDK4/6 inhibitors are in clinical trials for non-small cell lung cancer (NSCLC). Palbociclib (PD0332991) is included in the phase II/III Lung-MAP trial for squamous cell lung carcinoma (LUSQ). We noted differential cellular activity between palbociclib and the structurally related ribociclib (LEE011) in LUSQ cells. Applying an unbiased mass spectrometry-based chemoproteomics approach in H157 cells and primary tumor samples, we here report distinct proteome-wide target profiles of these two drug candidates in LUSQ, which encompass novel protein and, for palbociclib only, lipid kinases. In addition to CDK4 and 6, we observed CDK9 as a potent target of both drugs. Palbociclib interacted with several kinases not targeted by ribociclib, such as casein kinase 2 and PIK3R4, which regulate autophagy. Furthermore, palbociclib engaged several lipid kinases, most notably, PIK3CD and PIP4K2A/B/C. Accordingly, we observed modulation of autophagy and inhibition of AKT signaling by palbociclib but not ribociclib.

GSK3 alpha and beta are new functionally relevant targets of tivantinib in lung cancer cells.

Tivantinib has been described as a potent and highly selective inhibitor of the receptor tyrosine kinase c-MET and is currently in advanced clinical development for several cancers including non-small cell lung cancer (NSCLC). However, recent studies suggest that tivantinib's anticancer properties are unrelated to c-MET inhibition. Consistently, in determining tivantinib's activity profile in a broad panel of NSCLC cell lines, we found that, in contrast to several more potent c-MET inhibitors, tivantinib reduces cell viability across most of these cell lines. Applying an unbiased, mass-spectrometry-based, chemical proteomics approach, we identified glycogen synthase kinase 3 (GSK3) alpha and beta as novel tivantinib targets. Subsequent validation showed that tivantinib displayed higher potency for GSK3α than for GSK3ß and that pharmacological inhibition or simultaneous siRNA-mediated loss of GSK3α and GSK3ß caused apoptosis. In summary, GSK3α and GSK3ß are new kinase targets of tivantinib that play an important role in its cellular mechanism-of-action in NSCLC.