Harnessing Shannon entropy-based descriptors in machine learning models to enhance the prediction accuracy of molecular properties.

Accurate prediction of molecular properties is essential in the screening and development of drug molecules and other functional materials. Traditionally, property-specific molecular descriptors are used in machine learning models. This in turn requires the identification and development of target or problem-specific descriptors. Additionally, an increase in the prediction accuracy of the model is not always feasible from the standpoint of targeted descriptor usage. We explored the accuracy and generalizability issues using a framework of Shannon entropies, based on SMILES, SMARTS and/or InChiKey strings of respective molecules. Using various public databases of molecules, we showed that the accuracy of the prediction of machine learning models could be significantly enhanced simply by using Shannon entropy-based descriptors evaluated directly from SMILES. Analogous to partial pressures and total pressure of gases in a mixture, we used atom-wise fractional Shannon entropy in combination with total Shannon entropy from respective tokens of the string representation to model the molecule efficiently. The proposed descriptor was competitive in performance with standard descriptors such as Morgan fingerprints and SHED in regression models. Additionally, we found that either a hybrid descriptor set containing the Shannon entropy-based descriptors or an optimized, ensemble architecture of multilayer perceptrons and graph neural networks using the Shannon entropies was synergistic to improve the prediction accuracy. This simple approach of coupling the Shannon entropy framework to other standard descriptors and/or using it in ensemble models could find applications in boosting the performance of molecular property predictions in chemistry and material science.

High-throughput and targeted drug screens identify pharmacological candidates against MiT-translocation renal cell carcinoma.

BACKGROUND: MiT-Renal Cell Carcinoma (RCC) is characterized by genomic translocations involving microphthalmia-associated transcription factor (MiT) family members TFE3, TFEB, or MITF. MiT-RCC represents a specific subtype of sporadic RCC that is predominantly seen in young patients and can present with heterogeneous histological features making diagnosis challenging. Moreover, the disease biology of this aggressive cancer is poorly understood and there is no accepted standard of care therapy for patients with advanced disease. Tumor-derived cell lines have been established from human TFE3-RCC providing useful models for preclinical studies. METHODS: TFE3-RCC tumor derived cell lines and their tissues of origin were characterized by IHC and gene expression analyses. An unbiased high-throughput drug screen was performed to identify novel therapeutic agents for treatment of MiT-RCC. Potential therapeutic candidates were validated in in vitro and in vivo preclinical studies. Mechanistic assays were conducted to confirm the on-target effects of drugs. RESULTS: The results of a high-throughput small molecule drug screen utilizing three TFE3-RCC tumor-derived cell lines identified five classes of agents with potential pharmacological efficacy, including inhibitors of phosphoinositide-3-kinase (PI3K) and mechanistic target of rapamycin (mTOR), and several additional agents, including the transcription inhibitor Mithramycin A. Upregulation of the cell surface marker GPNMB, a specific MiT transcriptional target, was confirmed in TFE3-RCC and evaluated as a therapeutic target using the GPNMB-targeted antibody-drug conjugate CDX-011. In vitro and in vivo preclinical studies demonstrated efficacy of the PI3K/mTOR inhibitor NVP-BGT226, Mithramycin A, and CDX-011 as potential therapeutic options for treating advanced MiT-RCC as single agents or in combination. CONCLUSIONS: The results of the high-throughput drug screen and validation studies in TFE3-RCC tumor-derived cell lines have provided in vitro and in vivo preclinical data supporting the efficacy of the PI3K/mTOR inhibitor NVP-BGT226, the transcription inhibitor Mithramycin A, and GPNMB-targeted antibody-drug conjugate CDX-011 as potential therapeutic options for treating advanced MiT-RCC. The findings presented here should provide the basis for designing future clinical trials for patients with MiT-driven RCC.

A High-Throughput Screening Platform Identifies Novel Combination Treatments for Malignant Peripheral Nerve Sheath Tumors.

Malignant peripheral nerve sheath tumors (MPNST) are soft-tissue sarcomas that are the leading cause of mortality in patients with Neurofibromatosis type 1 (NF1). Single chemotherapeutic agents have shown response rates ranging from 18% to 44% in clinical trials, so there is still a high medical need to identify chemotherapeutic combination treatments that improve clinical prognosis and outcome. We screened a collection of compounds from the NCATS Mechanism Interrogation PlatE (MIPE) library in three MPNST cell lines, using cell viability and apoptosis assays. We then tested whether compounds that were active as single agents were synergistic when screened as pairwise combinations. Synergistic combinations in vitro were further evaluated in patient-derived orthotopic xenograft/orthoxenograft (PDOX) athymic models engrafted with primary MPNST matching with their paired primary-derived cell line where synergism was observed. The high-throughput screening identified 21 synergistic combinations, from which four exhibited potent synergies in a broad panel of MPNST cell lines. One of the combinations, MK-1775 with Doxorubicin, significantly reduced tumor growth in a sporadic PDOX model (MPNST-SP-01; sevenfold) and in an NF1-PDOX model (MPNST-NF1-09; fourfold) and presented greater effects in TP53 mutated MPNST cell lines. The other three combinations, all involving Panobinostat (combined with NVP-BGT226, Torin 2, or Carfilzomib), did not reduce the tumor volume in vivo at noncytotoxic doses. Our results support the utility of our screening platform of in vitro and in vivo models to explore new therapeutic approaches for MPNSTs and identified that combination MK-1775 with Doxorubicin could be a good pharmacologic option for the treatment of these tumors.

Brigatinib causes tumor shrinkage in both NF2-deficient meningioma and schwannoma through inhibition of multiple tyrosine kinases but not ALK.

Neurofibromatosis Type 2 (NF2) is an autosomal dominant genetic syndrome caused by mutations in the NF2 tumor suppressor gene resulting in multiple schwannomas and meningiomas. There are no FDA approved therapies for these tumors and their relentless progression results in high rates of morbidity and mortality. Through a combination of high throughput screens, preclinical in vivo modeling, and evaluation of the kinome en masse, we identified actionable drug targets and efficacious experimental therapeutics for the treatment of NF2 related schwannomas and meningiomas. These efforts identified brigatinib (ALUNBRIG®), an FDA-approved inhibitor of multiple tyrosine kinases including ALK, to be a potent inhibitor of tumor growth in established NF2 deficient xenograft meningiomas and a genetically engineered murine model of spontaneous NF2 schwannomas. Surprisingly, neither meningioma nor schwannoma cells express ALK. Instead, we demonstrate that brigatinib inhibited multiple tyrosine kinases, including EphA2, Fer and focal adhesion kinase 1 (FAK1). These data demonstrate the power of the de novo unbiased approach for drug discovery and represents a major step forward in the advancement of therapeutics for the treatment of NF2 related malignancies.

The synergy of BET inhibitors with aurora A kinase inhibitors in MYCN-amplified neuroblastoma is heightened with functional TP53.

Amplification of MYCN is a poor prognostic feature in neuroblastoma (NBL) indicating aggressive disease. We and others have shown BET bromodomain inhibitors (BETi) target MYCN indirectly by downregulating its transcription. Here we sought to identify agents that synergize with BETi and to identify biomarkers of resistance. We previously performed a viability screen of ∼1,900 oncology-focused compounds combined with BET bromodomain inhibitors against MYCN-amplified NBL cell lines. Reanalysis of our screening results prominently identified inhibitors of aurora kinase A (AURKAi) to be highly synergistic with BETi. We confirmed the anti-proliferative effects of several BETi+AURKAi combinations in MYCN-amplified NBL cell lines. Compared to single agents, these combinations cooperated to decrease levels of N-myc. We treated both TP53-wild type and mutant, MYCN-amplified cell lines with the BETi JQ1 and the AURKAi Alisertib. The combination had improved efficacy in the TP53-WT context, notably driving apoptosis in both genetic backgrounds. JQ1+Alisertib combination treatment of a MYCN-amplified, TP53-null or TP53-restored genetically engineered mouse model of NBL prolonged survival better than either single agent. This was most profound with TP53 restored, with marked tumor shrinkage and apoptosis induction in response to combination JQ1+Alisertib. BETi+AURKAi in MYCN-amplified NBL, particularly in the context of functional TP53, provided anti-tumor benefits in preclinical models. This combination should be studied more closely in a pediatric clinical trial.

Matrix Drug Screen Identifies Synergistic Drug Combinations to Augment SMAC Mimetic Activity in Ovarian Cancer.

Inhibitor of apoptosis (IAP) proteins are frequently upregulated in ovarian cancer, resulting in the evasion of apoptosis and enhanced cellular survival. Birinapant, a synthetic second mitochondrial activator of caspases (SMAC) mimetic, suppresses the functions of IAP proteins in order to enhance apoptotic pathways and facilitate tumor death. Despite on-target activity, however, pre-clinical trials of single-agent birinapant have exhibited minimal activity in the recurrent ovarian cancer setting. To augment the therapeutic potential of birinapant, we utilized a high-throughput screening matrix to identify synergistic drug combinations. Of those combinations identified, birinapant plus docetaxel was selected for further evaluation, given its remarkable synergy both in vitro and in vivo. We showed that this synergy results from multiple convergent pathways to include increased caspase activation, docetaxel-mediated TNF-α upregulation, alternative NF-kB signaling, and birinapant-induced microtubule stabilization. These findings provide a rationale for the integration of birinapant and docetaxel in a phase 2 clinical trial for recurrent ovarian cancer where treatment options are often limited and minimally effective.

Moving targets in drug discovery.

Drug Discovery is a lengthy and costly process and has faced a period of declining productivity within the last two decades resulting in increasing importance of integrative data-driven approaches. In this paper, data mining and integration is leveraged to inspect target innovation trends in drug discovery. The study highlights protein families and classes that have received more attention and those that have just emerged in the scientific literature, thus highlighting novel opportunities for drug intervention. In order to delineate the evolution of target-driven research interest from a biological perspective, trends in biological process annotations from Gene Ontology and disease annotations from DisGeNET are captured. The analysis reveals an increasing interest in targets related to immune system processes, and a recurrent trend for targets involved in circulatory system processes. At the level of diseases, targets associated with cancer-related pathologies, intellectual disability, and schizophrenia are increasingly investigated in recent years. The methodology enables researchers to capture trends in research attention in target space at an early stage during the drug discovery process. Workflows, scripts, and data used in this study are publicly available from https://github.com/BZdrazil/Moving_Targets . An interactive web application allows the customized exploration of target, biological process, and disease trends (available at https://rguha.shinyapps.io/MovingTargets/ ).

Drugs Targeting Tumor-Initiating Cells Prolong Survival in a Post-Surgery, Post-Chemotherapy Ovarian Cancer Relapse Model.

Disease recurrence is the major cause of morbidity and mortality of ovarian cancer (OC). In terms of maintenance therapies after platinum-based chemotherapy, PARP inhibitors significantly improve the overall survival of patients with BRCA mutations but is of little benefit to patients without homologous recombination deficiency (HRD). The stem-like tumor-initiating cell (TIC) population within OC tumors are thought to contribute to disease recurrence and chemoresistance. Therefore, there is a need to identify drugs that target TICs to prevent relapse in OC without HRD. RNA sequencing analysis of OC cells grown in TIC conditions revealed a strong enrichment of genes involved in drug metabolism, oxidative phosphorylation and reactive oxygen species (ROS) pathways. Concurrently, a high-throughput drug screen identified drugs that showed efficacy against OC cells grown as TICs compared to adherent cells. Four drugs were chosen that affected drug metabolism and ROS response: disulfiram, bardoxolone methyl, elesclomol and salinomycin. The drugs were tested in vitro for effects on viability, sphere formation and markers of stemness CD133 and ALDH in TICs compared to adherent cells. The compounds promoted ROS accumulation and oxidative stress and disulfiram, elesclomol and salinomycin increased cell death following carboplatin treatment compared to carboplatin alone. Disulfiram and salinomycin were effective in a post-surgery, post-chemotherapy OC relapse model in vivo, demonstrating that enhancing oxidative stress in TICs can prevent OC recurrence.

Novel renal medullary carcinoma cell lines, UOK353 and UOK360, provide preclinical tools to identify new therapeutic treatments.

Renal medullary carcinoma (RMC) is a rare, aggressive disease that predominantly afflicts individuals of African or Mediterranean descent with sickle cell trait. RMC comprises 1% of all renal cell carcinoma diagnoses with a median overall survival of 13 months. Patients are typically young (median age-22) and male (male:female ratio of 2:1) and tumors are characterized by complete loss of expression of the SMARCB1 tumor suppressor protein. Due to the low incidence of RMC and the disease's aggressiveness, treatment decisions are often based on case reports. Thus, it is critical to develop preclinical models of RMC to better understand the pathogenesis of this disease and to identify effective forms of therapy. Two novel cell line models, UOK353 and UOK360, were derived from primary RMCs that both demonstrated the characteristic SMARCB1 loss. Both cell lines overexpressed EZH2 and other members of the polycomb repressive complex and EZH2 inhibition in RMC tumor spheroids resulted in decreased viability. High throughput drug screening of both cell lines revealed several additional candidate compounds, including bortezomib that had both in vitro and in vivo antitumor activity. The activity of bortezomib was shown to be partially dependent on increased oxidative stress as addition of the N-acetyl cysteine antioxidant reduced the effect on cell proliferation. Combining bortezomib and cisplatin further decreased cell viability both in vitro and in vivo that single agent bortezomib treatment. The UOK353 and UOK360 cell lines represent novel preclinical models for the development of effective forms of therapy for RMC patients.

Cytotoxic Profiling of Annotated and Diverse Chemical Libraries Using Quantitative High-Throughput Screening.

Cell-based phenotypic screening is a commonly used approach to discover biological pathways, novel drug targets, chemical probes, and high-quality hit-to-lead molecules. Many hits identified from high-throughput screening campaigns are ruled out through a series of follow-up potency, selectivity/specificity, and cytotoxicity assays. Prioritization of molecules with little or no cytotoxicity for downstream evaluation can influence the future direction of projects, so cytotoxicity profiling of screening libraries at an early stage is essential for increasing the likelihood of candidate success. In this study, we assessed the cell-based cytotoxicity of nearly 10,000 compounds in the National Institutes of Health, National Center for Advancing Translational Sciences annotated libraries and more than 100,000 compounds in a diversity library against four normal cell lines (HEK 293, NIH 3T3, CRL-7250, and HaCat) and one cancer cell line (KB 3-1, a HeLa subline). This large-scale library profiling was analyzed for overall screening outcomes, hit rates, pan-activity, and selectivity. For the annotated library, we also examined the primary targets and mechanistic pathways regularly associated with cell death. To our knowledge, this is the first study to use high-throughput screening to profile a large screening collection (>100,000 compounds) for cytotoxicity in both normal and cancer cell lines. The results generated here constitute a valuable resource for the scientific community and provide insight into the extent of cytotoxic compounds in screening libraries, allowing for the identification and avoidance of compounds with cytotoxicity during high-throughput screening campaigns.

Therapeutic strategies for diffuse midline glioma from high-throughput combination drug screening.

Diffuse midline gliomas (DMGs) are universally lethal malignancies occurring chiefly during childhood and involving midline structures of the central nervous system, including thalamus, pons, and spinal cord. These molecularly related cancers are characterized by high prevalence of the histone H3K27M mutation. In search of effective therapeutic options, we examined multiple DMG cultures in sequential quantitative high-throughput screens (HTS) of 2706 approved and investigational drugs. This effort generated 19,936 single-agent dose responses that inspired a series of HTS-enabled drug combination assessments encompassing 9195 drug-drug examinations. Top combinations were validated across patient-derived cell cultures representing the major DMG genotypes. In vivo testing in patient-derived xenograft models validated the combination of the multi-histone deacetylase (HDAC) inhibitor panobinostat and the proteasome inhibitor marizomib as a promising therapeutic approach. Transcriptional and metabolomic surveys revealed substantial alterations to key metabolic processes and the cellular unfolded protein response after treatment with panobinostat and marizomib. Mitigation of drug-induced cytotoxicity and basal mitochondrial respiration with exogenous application of nicotinamide mononucleotide (NMN) or exacerbation of these phenotypes when blocking nicotinamide adenine dinucleotide (NAD+) production via nicotinamide phosphoribosyltransferase (NAMPT) inhibition demonstrated that metabolic catastrophe drives the combination-induced cytotoxicity. This study provides a comprehensive single-agent and combinatorial drug screen for DMG and identifies concomitant HDAC and proteasome inhibition as a promising therapeutic strategy that underscores underrecognized metabolic vulnerabilities in DMG.

Scaffold-Based Analytics: Enabling Hit-to-Lead Decisions by Visualizing Chemical Series Linked across Large Datasets.

We present a method for visualizing and navigating large screening datasets while also taking into account their activities and properties. Our approach is to annotate the data with all possible scaffolds contained within each molecule. We have developed a Spotfire visualization, coupled to a fuzzy clustering approach based on the scaffold decomposition of the screening deck, used to drive the hit triage process. Progression decisions can be made using aggregate scaffold parameters and data from multiple datasets merged at the scaffold level. This visualization reveals overlaps that help prioritize hits, highlight tractable series, and posit ways to combine aspects of multiple hits. The structure-activity relationship of a large and complex hit is automatically mapped onto all constituent scaffolds making it possible to navigate, via any shared scaffold, to all related hits. This scaffold "walking" helps address bias toward a handful of potent and ligand-efficient molecules at the expense of coverage of chemical space. We consider two scaffold generation methods and explored their similarities and differences both qualitatively and quantitatively. The workflow of a Spotfire visualization used in combination with fuzzy clustering and structure annotation provides an intuitive view of large and diverse screening datasets. This allows teams to effortlessly navigate between structurally related molecules and enriches the population of leads considered and progressed in a manner complementary to established approaches.

A High-Throughput Screen of a Library of Therapeutics Identifies Cytotoxic Substrates of P-glycoprotein.

The ATP-binding cassette transporter P-glycoprotein (P-gp) is known to limit both brain penetration and oral bioavailability of many chemotherapy drugs. Although US Food and Drug Administration guidelines require that potential interactions of investigational drugs with P-gp be explored, often this information does not enter the literature. In response, we developed a high-throughput screen to identify substrates of P-gp from a series of chemical libraries, testing a total of 10,804 compounds, most of which have known mechanisms of action. We used the CellTiter-Glo viability assay to test library compounds against parental KB-3-1 human cervical adenocarcinoma cells and the colchicine-selected subline KB-8-5-11 that overexpresses P-gp. KB-8-5-11 cells were also tested in the presence of a P-gp inhibitor (tariquidar) to assess reversibility of transporter-mediated resistance. Of the tested compounds, a total of 90 P-gp substrates were identified, including 55 newly identified compounds. Substrates were confirmed using an orthogonal killing assay against human embryonic kidney-293 cells overexpressing P-gp. We confirmed that AT7159 (cyclin-dependent kinase inhibitor), AT9283, (Janus kinase 2/3 inhibitor), ispinesib (kinesin spindle protein inhibitor), gedatolisib (PKI-587, phosphoinositide 3-kinase/mammalian target of rampamycin inhibitor), GSK-690693 (AKT inhibitor), and KW-2478 (heat-shock protein 90 inhibitor) were substrates. In addition, we assessed direct ATPase stimulation. ABCG2 was also found to confer high levels of resistance to AT9283, GSK-690693, and gedatolisib, whereas ispinesib, AT7519, and KW-2478 were weaker substrates. Combinations of P-gp substrates and inhibitors were assessed to demonstrate on-target synergistic cell killing. These data identified compounds whose oral bioavailability or brain penetration may be affected by P-gp. SIGNIFICANCE STATEMENT: The ATP-binding cassette transporter P-glycoprotein (P-gp) is known to be expressed at barrier sites, where it acts to limit oral bioavailability and brain penetration of substrates. In order to identify novel compounds that are transported by P-gp, we developed a high-throughput screen using the KB-3-1 cancer cell line and its colchicine-selected subline KB-8-5-11. We screened the Mechanism Interrogation Plate (MIPE) library, the National Center for Advancing Translational Science (NCATS) pharmaceutical collection (NPC), the NCATS Pharmacologically Active Chemical Toolbox (NPACT), and a kinase inhibitor library comprising 977 compounds, for a total of 10,804 compounds. Of the 10,804 compounds screened, a total of 90 substrates were identified of which 55 were novel. P-gp expression may adversely affect the oral bioavailability or brain penetration of these compounds.