Identifying drugs that reverse disease-associated transcriptomic features has been widely explored for drug repurposing, but its potential for de novo drug discovery remains underexplored. Here, we present gene expression profile predictor on chemical structures (GPS), a deep-learning-based drug discovery platform, guided by transcriptomic features, that screens large compound libraries and optimizes lead molecules. We first develop a model that captures transcriptomic perturbation signatures solely from chemical structures and deploy it to library compounds. We refine scoring methods and employ a tree-search method for optimization. By incorporating structure-gene-activity relationships, we uncover drug mechanisms from transcriptomic data. We evaluate GPS across multiple diseases and conduct extensive validation in two cases. In hepatocellular carcinoma, we discover two unique compound series with favorable cellular selectivity and in vivo efficacy. In idiopathic pulmonary fibrosis, we identify one repurposing candidate and one novel anti-fibrotic compound by reversing gene expression of multiple distinct cell types derived from single-cell transcriptomics.
Copyright © 2026 The Authors. Published by Elsevier Inc. All rights reserved.

Triple-negative breast cancer (TNBC), a highly aggressive subtype, currently lacks potent targeted therapies. ARID1B, a key SWI/SNF chromatin remodeling complex subunit, is linked to high-grade malignancies and poor prognosis, making it a potential biomarker and therapeutic target. However, its function and regulation remain unclear. Here, it is found that uncontrolled accumulation of ARID1B and its dysregulated nuclear import promoted oncogenesis and drug resistance. ARID1B negatively regulates ARID1A, impairing SWI/SNF-mediated tumor suppression and enhancing tumor survival. Using protein complex purification and mass spectrometry, the KPNA2-KPNB1-RANBP2 protein cascade is identified as critical for facilitating ARID1B nuclear import. Replacing R1518, H1519, and D1522 residues on ARID1B with T1518, G1519, and G1522 attenuates the ARID1B-KPNA2/KPNB1 interaction, preventing recruitment of ARID1B to the nuclear pore complex (NPC). Pharmacologically inhibiting KPNB1 suppressed ARID1B translocation, limiting its nuclear levels. In TNBC mouse models, ARID1B knockout (KO) significantly reduces tumor growth and enhances PARP inhibitor efficacy. Collectively, these findings uncover an undocumented mechanism for ARID1B nuclear translocation and reveal that blockade of ARID1B nuclear translocation can be a new therapeutic strategy for TNBC.
© 2025 The Author(s). Advanced Science published by Wiley‐VCH GmbH.

Dysregulated amino acid metabolism creates cancer-specific vulnerabilities. Neuroblastoma tumors have dysregulated arginine metabolism that renders them sensitive to systemic arginine deprivation. Arginase therapy has been proposed as a therapeutic approach for neuroblastoma treatment and has a favorable safety profile in pediatric cancer patients, however optimal therapeutic combinations remain unexplored.
The anti-tumor effects of BCT-100, a pegylated human arginase, were studied in neuroblastoma cell models by metabolite profiling, proteomics, and viability, clonogenicity, and protein translation assays. BCT-100 efficacy was assessed in the Th-MYCN transgenic neuroblastoma mouse model and in neuroblastoma cell line and patient-derived xenograft models.
In vitro, depletion of arginine by BCT-100 arrested protein translation and cellular proliferation, with effects on clonogenicity enhanced in combination with standard-of-care chemotherapeutics SN-38/temozolomide and mafosfamide/topotecan. In vivo, BCT-100 treatment spared liver arginine while significantly depleting plasma and tumor arginine in Th-MYCN mice, and extended tumor latency (> 100 vs. 45.5 days) in mice pre-emptively treated at weaning. In mice with established tumors, BCT-100 prolonged tumor progression delay when combined with standard-of-care chemo- (> 90 vs. 25 days) or chemo-immuno-therapy (49.5 vs. 35.5 days). Tumor progression delay was also observed in cell line and patient-derived xenografts with BCT-100 treatment, including relapsed/refractory disease models. No increased toxicity was observed with the addition of BCT-100 to established therapies.
The arginase BCT-100 profoundly disrupts neuroblastoma growth in vitro and in vivo, an effect enhanced in combination with standard-of-care chemo-immuno-therapy. Our data supports further assessment of arginine-depleting combination therapies as a new treatment strategy for neuroblastoma.
© 2025. The Author(s).

The nucleotide-binding pockets (NBPs) in virus-specific proteins have proven to be the most successful antiviral targets for several viral diseases. Functionally important NBPs are found in various structural and non-structural proteins of SARS-CoV-2. In this study, the first successful multi-targeting attempt to identify effective antivirals has been made against NBPs in nsp12, nsp13, nsp14, nsp15, nsp16, and nucleocapsid (N) proteins of SARS-CoV-2. A structure-based drug repurposing in silico screening approach with ADME analysis identified small molecules targeting NBPs in SARS-CoV-2 proteins. Further, isothermal titration calorimetry (ITC) experiments validated the binding of top hit molecules to the purified N-protein. Importantly, cell-based antiviral assays revealed antiviral potency for INCB28060, darglitazone, and columbianadin with EC50 values 15.71 μM, 5.36 μM, and 22.52 μM, respectively. These effective antivirals targeting multiple proteins are envisioned to direct the development of antiviral therapy against SARS-CoV-2 and its emerging variants.
Copyright © 2022 Elsevier Inc. All rights reserved.

For the treatment of autoimmune diseases, depletion of B cells specific for auto-antigens is important because they will be a source of plasmablasts/plasma cells to produce autoantibodies. However, because some types of B cells like naïve B cells and memory B cells are at quiescent phase, they are insensitive to anticancer drugs which exert cytotoxicity by arresting the cell cycle. Here we show that B cell receptor (BCR) stimulation increases the sensitivity of anticancer drugs by promoting the proliferation of quiescent B cells. The BCR stimulation to primary naïve B cells enhanced sensitivity to several anticancer drugs which arrest the cell cycle through different mechanisms. The present results indicated that combination of the BCR stimulation and anticancer drugs is a promising strategy for the antigen-specific depletion of pathogenic quiescent B cells.