Critical Assessment of Artificial Intelligence Methods for Prediction of hERG Channel Inhibition in the "Big Data" Era.

The rise of novel artificial intelligence (AI) methods necessitates their benchmarking against classical machine learning for a typical drug-discovery project. Inhibition of the potassium ion channel, whose alpha subunit is encoded by the human ether-à-go-go-related gene (hERG), leads to a prolonged QT interval of the cardiac action potential and is a significant safety pharmacology target for the development of new medicines. Several computational approaches have been employed to develop prediction models for the assessment of hERG liabilities of small molecules including recent work using deep learning methods. Here, we perform a comprehensive comparison of hERG effect prediction models based on classical approaches (random forests and gradient boosting) and modern AI methods [deep neural networks (DNNs) and recurrent neural networks (RNNs)]. The training set (∼9000 compounds) was compiled by integrating the hERG bioactivity data from the ChEMBL database with experimental data generated from an in-house, high-throughput thallium flux assay. We utilized different molecular descriptors including the latent descriptors, which are real-value continuous vectors derived from chemical autoencoders trained on a large chemical space (>1.5 million compounds). The models were prospectively validated on ∼840 in-house compounds screened in the same thallium flux assay. The best results were obtained with the XGBoost method and RDKit descriptors. The comparison of models based only on latent descriptors revealed that the DNNs performed significantly better than the classical methods. The RNNs that operate on SMILES provided the highest model sensitivity. The best models were merged into a consensus model that offered superior performance compared to reference models from academic and commercial domains. Furthermore, we shed light on the potential of AI methods to exploit the big data in chemistry and generate novel chemical representations useful in predictive modeling and tailoring a new chemical space.

Argonaute2 expression is post-transcriptionally coupled to microRNA abundance.

Argonaute proteins are essential components of microRNA (miRNA)- and small interfering (siRNA)-mediated post-transcriptional gene-silencing pathways. In mammals, Argonaute2 (Ago2) is the catalytic center of the RNA-induced silencing complex (RISC) that recognizes and endonucleolytically cleaves messenger RNAs of complementary sequence. Although Ago2 is essential for RISC activity, the mechanisms regulating Argonaute protein expression are largely unknown. Here we report that Ago2 expression is dependent on miRNA abundance and that unloaded Ago2 protein is unstable. We observed a low level of Ago2 protein in Dicer- or DGCR8-deficent mouse embryonic stem cells (ESCs) that could be rescued by reintroduction of the respective cDNAs or by transfection of miRNAs or siRNAs. We found expression of Ago2 protein from a transgene to be similarly regulated, further supporting a post-transcriptional control mechanism. Inhibition of Hsc70/Hsp90 led to decreased Ago2 expression consistent with the reported role of this chaperone complex in RISC assembly. We furthermore found that the degradation of Ago2 was specifically blocked by inhibition of the lysosome, but not the proteasome. Our results illuminate a novel feedback mechanism that post-transcriptionally couples Ago2 protein levels with small RNA abundance with implications for RNA-interference (RNAi) and miRNA function.

The co-chaperones Fkbp4/5 control Argonaute2 expression and facilitate RISC assembly.

Argonaute2 (Ago2) protein and associated microRNAs (miRNAs) or small interfering RNAs (siRNAs) form the RNA-induced silencing complex (RISC) for target messenger RNA cleavage and post-transcriptional gene silencing. Although Ago2 is essential for RISC activity, the mechanism of RISC assembly is not well understood, and factors controlling Ago2 protein expression are largely unknown. A role for the Hsc70/Hsp90 chaperone complex in loading small RNA duplexes into the RISC has been demonstrated in cell extracts, and unloaded Ago2 is unstable and degraded by the lysosome in mammalian cells. Here we identify the co-chaperones Fkbp4 and Fkbp5 as Ago2-associated proteins in mouse embryonic stem cells. Pharmacological inhibition of this interaction using FK506 or siRNA-mediated Fkbp4/5 depletion leads to decreased Ago2 protein levels. We find FK506 treatment inhibits, whereas Fkbp4/5 overexpression promotes, miRNA-mediated stabilization of Ago2 expression. Simultaneous treatment with a lysosome inhibitor revealed the accumulation of unloaded Ago2 complexes in FK506-treated cells. We find that, consistent with unloaded miRNAs being unstable, FK506 treatment also affects miRNA abundance, particularly nascent miRNAs. Our results support a role for Fkbp4/5 in RISC assembly.

Cell-based assays to support the profiling of small molecules with histone methyltransferase and demethylase modulatory activity.

Histone methylation is a prevalent and dynamic chromatin modification, executed by the action of histone methyltransferases (HMTs) and demethylases (HDMs). Aberrant activity of many of these enzymes is associated with human disease, hence, there is a growing interest in identifying corresponding small molecule inhibitors with therapeutic potential. To date, most of the technologies supporting the identification of these inhibitors constitute in vitro biochemical assays which, although robust and sensitive, do not study HMTs and HDMs in their native cellular state nor provide information of inhibitor's cell permeability and toxicity. The evident need for complementary cellular approaches has recently propelled the development of cell-based assays that enable screening of HMT and HDM enzymes in a more relevant environment. Here, we highlight current cellular methodologies for HMT and HDM drug discovery support. We anticipate that implementation of these cell-based assays will positively impact the discovery of pharmacologically potent HMT and HDM inhibitors.

RALDH1 Inhibition Shows Immunotherapeutic Efficacy in Hepatocellular Carcinoma.

Globally, hepatocellular carcinoma (HCC) is one of the most commonly diagnosed cancers and a leading cause of cancer-related death. We previously identified an immune evasion pathway whereby tumor cells produce retinoic acid (RA) to promote differentiation of intratumoral monocytes into protumor macrophages. Retinaldehyde dehydrogenase 1 (RALDH1), RALDH2, and RALDH3 are the three isozymes that catalyze RA biosynthesis. In this study, we have identified RALDH1 as the key driver of RA production in HCC and demonstrated the efficacy of RALDH1-selective inhibitors (Raldh1-INH) in suppressing RA production by HCC cells. Raldh1-INH restrained tumor growth in multiple mouse models of HCC by reducing the number and tumor-supporting functions of intratumoral macrophages as well as increasing T-cell infiltration and activation within tumors. Raldh1-INH also displayed favorable pharmacokinetic, pharmacodynamic, and toxicity profiles in mice thereby establishing them as promising new drug candidates for HCC immunotherapy.

A stele-enriched gene regulatory network in the Arabidopsis root.

Tightly controlled gene expression is a hallmark of multicellular development and is accomplished by transcription factors (TFs) and microRNAs (miRNAs). Although many studies have focused on identifying downstream targets of these molecules, less is known about the factors that regulate their differential expression. We used data from high spatial resolution gene expression experiments and yeast one-hybrid (Y1H) and two-hybrid (Y2H) assays to delineate a subset of interactions occurring within a gene regulatory network (GRN) that determines tissue-specific TF and miRNA expression in plants. We find that upstream TFs are expressed in more diverse cell types than their targets and that promoters that are bound by a relatively large number of TFs correspond to key developmental regulators. The regulatory consequence of many TFs for their target was experimentally determined using genetic analysis. Remarkably, molecular phenotypes were identified for 65% of the TFs, but morphological phenotypes were associated with only 16%. This indicates that the GRN is robust, and that gene expression changes may be canalized or buffered.

The interplay between transcription factors and microRNAs in genome-scale regulatory networks.

Metazoan genomes contain thousands of protein-coding and non-coding RNA genes, most of which are differentially expressed, i.e., at different locations, at different times during development, or in response to environmental signals. Differential gene expression is achieved through complex regulatory networks that are controlled in part by two types of trans-regulators: transcription factors (TFs) and microRNAs (miRNAs). TFs bind to cis-regulatory DNA elements that are often located in or near their target genes, while miRNAs hybridize to cis-regulatory RNA elements mostly located in the 3' untranslated region of their target mRNAs. Here, we describe how these trans-regulators interact with each other in the context of gene regulatory networks to coordinate gene expression at the genome-scale level, and discuss future challenges of integrating these networks with other types of functional networks.

High-Throughput Detection of Ligand-Protein Binding Using a SplitLuc Cellular Thermal Shift Assay.

The confirmation of a small molecule binding to a protein target can be challenging when switching from biochemical assays to physiologically relevant cellular models. The cellular thermal shift assay (CETSA) is an approach to validate ligand-protein binding in a cellular environment by examining a protein's melting profile which can shift to a higher or lower temperature when bound by a small molecule. Traditional CETSA uses SDS-PAGE and Western blotting to quantify protein levels, a process that is both time consuming and low-throughput when screening multiple compounds and concentrations. Herein, we outline the reagents and methods to implement split Nano Luciferase (SplitLuc) CETSA, which is a reporter-based target engagement assay designed for high-throughput screening in 384- or 1536-well plate formats.

A Genome-Edited ERα-HiBiT Fusion Reporter Cell Line for the Identification of ERα Modulators Via High-Throughput Screening and CETSA.

The estrogen receptor α (ERα) is a target of intense pharmacological intervention and toxicological biomonitoring. Current methods to directly quantify cellular levels of ERα involve antibody-based assays, which are labor-intensive and of limited throughput. In this study, we generated a post-translational reporter cell line, referred to as MCF7-ERα-HiBiT, by fusing a small pro-luminescent nanoluciferase (NLuc) tag (HiBiT) to the C-terminus of endogenous ERα in MCF7 cells. The tag allows the luminescent detection and quantification of endogenous ERα protein by addition of the complementary NLuc enzyme fragment. This MCF7-ERα-HiBiT cell line was optimized for quantitative high-throughput screening (qHTS) to identify compounds that reduce ERα levels. In addition, the same cell line was optimized for a qHTS cellular thermal shift assay to identify compounds that bind and thermally stabilize ERα. Here, we interrogated the MCF7-ERα-HiBiT assay against the NCATS Pharmacological Collection (NPC) of 2,678 approved drugs and identified compounds that potently reduce and thermally stabilize ERα. Our novel post-translational reporter cell line provides a unique opportunity for profiling large pharmacological and toxicological compound libraries for their effect on ERα levels as well as for assessing direct compound binding to the receptor, thus facilitating mechanistic studies by which compounds exert their biological effects on ERα.

Genome-Edited Coincidence and PMP22-HiBiT Fusion Reporter Cell Lines Enable an Artifact-Suppressive Quantitative High-Throughput Screening Strategy for PMP22 Gene-Dosage Disorder Drug Discovery.

Charcot-Marie-Tooth 1A (CMT1A) is the most common form of hereditary peripheral neuropathies, characterized by genetic duplication of the critical myelin gene Peripheral Myelin Protein 22 (PMP22). PMP22 overexpression results in abnormal Schwann cell differentiation, leading to axonal loss and muscle wasting. Since regulation of PMP22 expression is a major target of therapeutic discovery for CMT1A, we sought to establish unbiased approaches that allow the identification of therapeutic agents for this disease. Using genome editing, we generated a coincidence reporter assay that accurately monitors Pmp22 transcript levels in the S16 rat Schwann cell line, while reducing reporter-based false positives. A quantitative high-throughput screen (qHTS) of 42 577 compounds using this assay revealed diverse novel chemical classes that reduce endogenous Pmp22 transcript levels. Moreover, some of these classes show pharmacological specificity in reducing Pmp22 over another major myelin-associated gene, Mpz (Myelin protein zero). Finally, to investigate whether compound-mediated reduction of Pmp22 transcripts translates to reduced PMP22 protein levels, we edited the S16 genome to generate a reporter assay that expresses a PMP22-HiBiT fusion protein using CRISPR/Cas9. Overall, we present a screening platform that combines genome edited cell lines encoding reporters that monitor transcriptional and post-translational regulation of PMP22 with titration-based screening (e.g., qHTS), which could be efficiently incorporated into drug discovery campaigns for CMT1A.

Discovery and Optimization of 2H-1λ2-Pyridin-2-one Inhibitors of Mutant Isocitrate Dehydrogenase 1 for the Treatment of Cancer.

Neomorphic mutations in isocitrate dehydrogenase 1 (IDH1) are oncogenic for a number of malignancies, primarily low-grade gliomas and acute myeloid leukemia. We report a medicinal chemistry campaign around a 7,7-dimethyl-7,8-dihydro-2H-1λ2-quinoline-2,5(6H)-dione screening hit against the R132H and R132C mutant forms of isocitrate dehydrogenase (IDH1). Systematic SAR efforts produced a series of potent pyrid-2-one mIDH1 inhibitors, including the atropisomer (+)-119 (NCATS-SM5637, NSC 791985). In an engineered mIDH1-U87-xenograft mouse model, after a single oral dose of 30 mg/kg, 16 h post dose, between 16 and 48 h, (+)-119 showed higher tumoral concentrations that corresponded to lower 2-HG concentrations, when compared with the approved drug AG-120 (ivosidenib).

A Comparative Study of Target Engagement Assays for HDAC1 Inhibitor Profiling.

Histone deacetylases (HDACs) are epigenetic modulators linked to diseases including cancer and neurodegeneration. Given their therapeutic potential, highly sensitive biochemical and cell-based profiling technologies have been developed to discover small-molecule HDAC inhibitors. Ultimately, the therapeutic action of these inhibitors is dependent on a physical engagement with their intended targets in cellular and tissue environments. Confirming target engagement in the cellular environment is particularly relevant for HDACs since they function as part of cell type-specific multiprotein complexes. Here we implemented two recently developed high-throughput target engagement technologies, NanoBRET and SplitLuc CETSA, to profile 349 compounds in the Epigenetic-Focused collection for HDAC1 binding. We found that the two HDAC1 target engagement assays correlated well with each other and with orthogonal activity-based assays, in particular those carried out in cellular environments rather than with isolated HDAC proteins. The assays detected a majority of the previously described HDAC1 inhibitors in the collection and, importantly, triaged HDAC inhibitors known to target other HDACs.

A high-throughput screening platform for Polycystic Kidney Disease (PKD) drug repurposing utilizing murine and human ADPKD cells.

Autosomal dominant polycystic kidney disease (ADPKD) is one of the most common inherited monogenic disorders, characterized by a progressive decline in kidney function due in part to the formation of fluid-filled cysts. While there is one FDA-approved therapy, it is associated with potential adverse effects, and all other clinical interventions are largely supportive. Insights into the cellular pathways underlying ADPKD have revealed striking similarities to cancer. Moreover, several drugs originally developed for cancer have shown to ameliorate cyst formation and disease progression in animal models of ADPKD. These observations prompted us to develop a high-throughput screening platform of cancer drugs in a quest to repurpose them for ADPKD. We screened ~8,000 compounds, including compounds with oncological annotations, as well as FDA-approved drugs, and identified 155 that reduced the viability of Pkd1-null mouse kidney cells with minimal effects on wild-type cells. We found that 109 of these compounds also reduced in vitro cyst growth of Pkd1-null cells cultured in a 3D matrix. Moreover, the result of the cyst assay identified therapeutically relevant compounds, including agents that interfere with tubulin dynamics and reduced cyst growth without affecting cell viability. Because it is known that several ADPKD therapies with promising outcomes in animal models failed to be translated to human disease, our platform also incorporated the evaluation of compounds in a panel of primary ADPKD and normal human kidney (NHK) epithelial cells. Although we observed differences in compound response amongst ADPKD and NHK cell preparation, we identified 18 compounds that preferentially affected the viability of most ADPKD cells with minimal effects on NHK cells. Our study identifies attractive candidates for future efficacy studies in advanced pre-clinical models of ADPKD.

Identification of a G-Protein-Independent Activator of GIRK Channels.

G-protein-gated inwardly rectifying K+ (GIRK) channels are essential effectors of inhibitory neurotransmission in the brain. GIRK channels have been implicated in diseases with abnormal neuronal excitability, including epilepsy and addiction. GIRK channels are tetramers composed of either the same subunit (e.g., homotetramers) or different subunits (e.g., heterotetramers). Compounds that specifically target subsets of GIRK channels in vivo are lacking. Previous studies have shown that alcohol directly activates GIRK channels through a hydrophobic pocket located in the cytoplasmic domain of the channel. Here, we report the identification and functional characterization of a GIRK1-selective activator, termed GiGA1, that targets the alcohol pocket. GiGA1 activates GIRK1/GIRK2 both in vitro and in vivo and, in turn, mitigates the effects of a convulsant in an acute epilepsy mouse model. These results shed light on the structure-based development of subunit-specific GIRK modulators that could provide potential treatments for brain disorders.

Small-molecule activation of lysosomal TRP channels ameliorates Duchenne muscular dystrophy in mouse models.

Duchenne muscular dystrophy (DMD) is a devastating disease caused by mutations in dystrophin that compromise sarcolemma integrity. Currently, there is no treatment for DMD. Mutations in transient receptor potential mucolipin 1 (ML1), a lysosomal Ca2+ channel required for lysosomal exocytosis, produce a DMD-like phenotype. Here, we show that transgenic overexpression or pharmacological activation of ML1 in vivo facilitates sarcolemma repair and alleviates the dystrophic phenotypes in both skeletal and cardiac muscles of mdx mice (a mouse model of DMD). Hallmark dystrophic features of DMD, including myofiber necrosis, central nucleation, fibrosis, elevated serum creatine kinase levels, reduced muscle force, impaired motor ability, and dilated cardiomyopathies, were all ameliorated by increasing ML1 activity. ML1-dependent activation of transcription factor EB (TFEB) corrects lysosomal insufficiency to diminish muscle damage. Hence, targeting lysosomal Ca2+ channels may represent a promising approach to treat DMD and related muscle diseases.

Correction: A High-Content Assay Enables the Automated Screening and Identification of Small Molecules with Specific ALDH1A1-Inhibitory Activity.

[This corrects the article DOI: 10.1371/journal.pone.0170937.].

Quantitative high-throughput phenotypic screening of pediatric cancer cell lines identifies multiple opportunities for drug repurposing.

Drug repurposing approaches have the potential advantage of facilitating rapid and cost-effective development of new therapies. Particularly, the repurposing of drugs with known safety profiles in children could bypass or streamline toxicity studies. We employed a phenotypic screening paradigm on a panel of well-characterized cell lines derived from pediatric solid tumors against a collection of ∼3,800 compounds spanning approved drugs and investigational agents. Specifically, we employed titration-based screening where compounds were tested at multiple concentrations for their effect on cell viability. Molecular and cellular target enrichment analysis indicated that numerous agents across different therapeutic categories and modes of action had an antiproliferative effect, notably antiparasitic/protozoal drugs with non-classic antineoplastic activity. Focusing on active compounds with dosing and safety information in children according to the Children's Pharmacy Collaborative database, we identified compounds with therapeutic potential through further validation using 3D tumor spheroid models. Moreover, we show that antiparasitic agents induce cell death via apoptosis induction. This study demonstrates that our screening platform enables the identification of chemical agents with cytotoxic activity in pediatric cancer cell lines of which many have known safety/toxicity profiles in children. These agents constitute attractive candidates for efficacy studies in pre-clinical models of pediatric solid tumors.

A widely-applicable high-throughput cellular thermal shift assay (CETSA) using split Nano Luciferase.

Assessment of the interactions between a drug and its protein target in a physiologically relevant cellular environment constitutes a major challenge in the pre-clinical drug discovery space. The Cellular Thermal Shift Assay (CETSA) enables such an assessment by quantifying the changes in the thermal stability of proteins upon ligand binding in intact cells. Here, we present the development and validation of a homogeneous, standardized, target-independent, and high-throughput (384- and 1536-well formats) CETSA platform that uses a split Nano Luciferase approach (SplitLuc CETSA). The broad applicability of the assay was demonstrated for diverse targets, and its performance was compared with independent biochemical and cell-based readouts using a set of well-characterized inhibitors. Moreover, we investigated the utility of the platform as a primary assay for high-throughput screening. The SplitLuc CETSA presented here enables target engagement studies for medium and high-throughput applications. Additionally, it provides a rapid assay development and screening platform for targets where phenotypic or other cell-based assays are not readily available.

Discovery of Orally Bioavailable, Quinoline-Based Aldehyde Dehydrogenase 1A1 (ALDH1A1) Inhibitors with Potent Cellular Activity.

Aldehyde dehydrogenases (ALDHs) are responsible for the metabolism of aldehydes (exogenous and endogenous) and possess vital physiological and toxicological functions in areas such as CNS, inflammation, metabolic disorders, and cancers. Overexpression of certain ALDHs (e.g., ALDH1A1) is an important biomarker in cancers and cancer stem cells (CSCs) indicating the potential need for the identification and development of small molecule ALDH inhibitors. Herein, a newly designed series of quinoline-based analogs of ALDH1A1 inhibitors is described. Extensive medicinal chemistry optimization and biological characterization led to the identification of analogs with significantly improved enzymatic and cellular ALDH inhibition. Selected analogs, e.g., 86 (NCT-505) and 91 (NCT-506), demonstrated target engagement in a cellular thermal shift assay (CETSA), inhibited the formation of 3D spheroid cultures of OV-90 cancer cells, and potentiated the cytotoxicity of paclitaxel in SKOV-3-TR, a paclitaxel resistant ovarian cancer cell line. Lead compounds also exhibit high specificity over other ALDH isozymes and unrelated dehydrogenases. The in vitro ADME profiles and pharmacokinetic evaluation of selected analogs are also highlighted.