Biology subjects

Nan, Y.

Publications and source records attributed to Nan, Y..

9 recordsLinked to original sources

bioRxiv2026

An AI System for Autonomous Algorithm Evolution in Drug Development

Artificial intelligence (AI) is increasingly permeating the drug development pipeline. Numerous algorithms for accelerating this multi-stage and multi-task process have been constructed, which depends heavily on expert design and labor-intensive task-specific optimization. Given that AI-driven acceleration of drug development is recognized as a cumulative, often synergistic, effect across multiple stages, the autonomous evolution of existing algorithms across the entire pipeline is demanded to achieve a holistic advancement. Here, we present DrugEvolve, a multi-role large language model system for systematic and autonomous algorithm evolution in drug development. DrugEvolve realizes a closed-loop evolution process by incorporating Researcher, Engineer, and Analyst domains, and enables an iterative design, implementation, evaluation, and refinement of algorithm by leveraging scientific knowledge and accumulated evolutionary experience. Across eleven representative tasks spanning target identification, drug discovery, preclinical study, and clinical trial, DrugEvolve autonomously evolved the corresponding task-specific algorithms and achieved substantial performance enhancement on 120 benchmark test sets. Moreover, it showed robust generalizabilities across heterogeneous data modalities (ranging from biological sequence and graph to molecular topology and textual language), and realized gains in both predictive and generative tasks. Collectively, this AI system can serve not only as an algorithmic infrastructure for drug development, but also as a transferable paradigm for broader scientific domains.

pharmacology and toxicology↗

bioRxiv2025

IRG1/itaconate/NRF2/GSH axis in tumor-associated macrophages drives therapy resistance and immune evasion in BRCA1-deficient breast cancer

Tumor-associated macrophages (TAMs) are major contributors to immunosuppression and therapeutic resistance, including resistance to PARP inhibitors (PARPi) in BRCA1-deficient breast cancer. However, the mechanisms underlying TAM-mediated PARPi resistance remain unclear. Here, we demonstrate that TAM-derived glutathione (GSH) impairs the efficacy of PARPi by protecting tumor cells from DNA damage and ferroptosis while suppressing STING-mediated immune activation. Mechanistically, STAT5-driven upregulation of the IRG1/itaconate axis in TAMs rewires mitochondrial metabolism and activates NRF2-dependent GSH biosynthesis. GSH is subsequently released into the tumor microenvironment, where it is taken up by tumor cells, protecting them from PARPi-induced cytotoxicity and dampening immune responses. Pharmacological inhibition of IRG1 reverses these effects, restoring PARPi sensitivity and enhancing anti-tumor immunity in BRCA1-deficient tumor models. Collectively, these findings uncover a TAM-specific immunometabolic program that limits PARPi efficacy and highlight the IRG1/NRF2/GSH axis as a promising therapeutic target to improve treatment outcomes in BRCA1-associated breast cancer.

cancer biology↗

bioRxiv2025

Adaptive Asymmetric Adjustment in Affective Evaluation and Reciprocity for Altruistic Behaviors from Exogenous Uncertainty to Certainty

In direct reciprocity, environmentally imposed exogenous uncertainty frequently decouples benefactors altruistic intentions from final outcomes. How do cooperative bonds remain resilient against this inevitable "noise"? Adopting a comprehensive levels-of-analysis framework across eight experiments, we integrate interpersonal paradigms, computational modeling, multivariate fMRI decoding, and third-party social evaluations to uncover a robust "asymmetric adjustment" in beneficiarys affective evaluation and reciprocity under exogenous uncertainty, systematically delineating its cognitive algorithmic rules, neural representations, and adaptive functions in sustaining human cooperation. Specifically, beneficiarys reciprocity amplifies for positive deviations from expectations while buffering against reciprocal reductions from unintended shortfalls. Modeling reveals this computational resilience is governed by an asymmetric intrinsic prosocial bias, extending beyond traditional frameworks of outcome/intention-based evaluation, prediction-error-based social learning, or loss aversion. Supported by theory-of-mind system, this "asymmetric adjustment" is perceived as morally superior to other calculations, thereby serving a critical adaptive function to sustain cooperation amidst the volatility of an uncertain world.

animal behavior and cognition↗

bioRxiv2023

Mechanism of action of Buqing Granule against Diabetic retinopathy based on network pharmacology and animal experiments

ObjectiveFor this study, network pharmacology and animal experiments were used together to get a better idea of how BQKL works at the molecular level to treat DR. MethodsIn this study, we obtained the relevant action target information of BQKL from the TCMSP and also combined the GeneCards, OMIM, TTD, and PharmGKB databases with the GEO database to obtain the relevant target information of DR. The intersection of these targets was determined using a Venn diagram to identify the target of action for the treatment of DR with BQKL. The target proteins of BQKL for DR were then uploaded to the String database. The resultant data were imported into Cytoscape 3.9.0 to construct PPI networks and identify key targets of action. The DAVID database was used to do a GO and KEGG pathway enrichment analysis of target genes for treating DR with BQKL. Molecular docking was performed to validate the core action targets with the core compounds of BQKL. In addition, we induced DR production in rats by a high-fat, high-sugar diet and intraperitoneal injection of STZ and validated the results obtained from the network pharmacological analysis by changes in body weight and blood glucose, serum levels of biochemical markers, HE staining, immunohistochemistry, qRT-PCR, and Western blot experiments in rats. ResultsIn this study, quercetin, kaempferol, {beta}-sitosterol, lignanserin, and stigmasterol were identified as the key components, TP53, AKT1, JUN, CASPASE3, MAPK3, and MAPK1 as the key targets, and PI3K-Akt, AGE-RAGE, and MAPK signaling pathways as the main pathways involved. The results of animal experiments showed that BQKL could not only effectively reduce the degree of blood glucose, blood lipids, and oxidative damage in diabetic rats but also slow down the development process of DR. At the same time, it can significantly up-regulate the expression of AKT1, MAPK1, and MAPK3 and down-regulate the expression of CASPASE3, c-JUN, and TP53 in retinal tissue. ConclusionBQKL ameliorates oxidative stress, apoptosis, and inflammation due to hyperglycemia-related stress by regulating key targets of CASPASE3, AKT1, c-JUN, TP53, MAPK1, and MAPK3, thereby delaying the onset and progression of DR.

molecular biology↗

bioRxiv2023

m6A Demethylase FTO Stabilizes LINK-A to Exert Oncogenic Roles via MCM3-Mediated Cell Cycle Progression and HIF-1α Activation

RNA N6-methyladenosine (m6A) modification, balanced by methyltransferases and demethylases, has recently been shown to play critical roles in multiple cancers. However, the mechanism by which m6A modification regulates long noncoding RNA (lncRNA) stability and function during cancer progression remains unclear. Here, we show that m6A demethylase fat mass and obesity-associated protein (FTO) removes the m6A modification on long intergenic noncoding RNA for kinase activation (LINK-A) and stabilizes it to promote cell proliferation and cytotoxic chemotherapy resistance in esophageal squamous cell carcinoma (ESCC). Mechanistically, LINK-A enhances the interaction between minichromosome maintenance complex component 3 (MCM3) and cyclin-dependent kinase 1 (CDK1) to promote MCM3 phosphorylation by CDK1. MCM3 is a subunit of the hexameric protein complex and its phosphorylation facilitates loading of the MCM complex onto chromatin, which promotes cell cycle progression and subsequent cell proliferation. Meanwhile, LINK-A prevents the interaction of MCM3 and hypoxia-inducible factor 1 (HIF-1), abrogates MCM3-mediated transcriptional repression of HIF-1, and promotes glycolysis and chemoresistance of cancer cells. These results elucidate a mechanism whereby FTO-stabilized LINK-A plays oncogenic roles and present the FTO/LINK-A/MCM3/HIF-1 axis as a promising therapeutic target for ESCC.

cancer biology↗

bioRxiv2023

Prion-like characteristics of Hepatitis E virus ORF3 protein are associated with virus release and pathogenesis

Hepatitis E virus (HEV) is a common causative agent of acute hepatitis. Due to the shortage of efficient in vitro model, the viral assembly and release processes are still poorly understood. In this study, we found that ORF3, an HEV structural protein showed prion-like properties. The prion domain (PrD) of yeast prion Sup35 could be functionally replaced by HEV-ORF3 and its candidate PrD (cPrD). A single amino acid substitution in the cPrD reduced the aggregation propensity of ORF3 and blocked the function of ORF3 in enhancing the stability of microtubules in HEV-infected cells, thus blocking virion release from the infected cells and resulting in reduced HEV pathogenicity in Mongolian gerbils. These data suggest that HEV-ORF3 is a novel functional prion-like protein which assembles into a filamentous structure for virion release, supporting the hypothesis that the self-propagating properties of prion proteins are widely exploited as epigenetic information carriers in nature.

microbiology↗

bioRxiv2022

HCP5 prevents ubiquitination-mediated UTP3 degradation to inhibit apoptosis by activating c-Myc transcriptional activity

Inducing cancer cell apoptosis through cytotoxic reagents is the main therapeutic strategy for diverse cancer types. However, several antiapoptotic factors impede curative cancer therapy by driving cancer cells to resist cytotoxic agent-induced apoptosis, thus leading to refractoriness and relapse. To define critical antiapoptotic factors that contribute to chemoresistance in esophageal squamous cell carcinoma (ESCC), we generated two pairs of parental and apoptosis-resistant cell models through cisplatin (DDP) induction and then performed whole-transcriptome sequencing. We identified the long noncoding RNA (lncRNA) histocompatibility leukocyte antigen complex P5 (HCP5) as the chief culprit for chemoresistance. Mechanistically, HCP5 interacts with UTP3 small subunit processome component (UTP3) and prevents UTP3 degradation from E3 ligase tripartite motif containing 29 (TRIM29)-mediated ubiquitination. UTP3 then recruits c-Myc to activate vesicle-associated membrane protein 3 (VAMP3) expression. Activated VAMP3 suppresses caspase-dependent apoptosis and eventually leads to chemoresistance. Accordingly, the expression level of the HCP5/UTP3/c-Myc/VAMP3 axis in chemoresistant patients is significantly higher than that in chemosensitive patients. Thus, our study demonstrated that the HCP5/UTP3/c-Myc/VAMP3 axis plays an important role in the inhibition of cancer cell apoptosis and that HCP5 can be a promising chemosensitive target for cancer treatment.

cancer biology↗

bioRxiv2022

DLGAP1-AS2-Mediated Phosphatidic Acid Synthesis Confers Chemoresistance via Activation of YAP Signaling

Squamous cell carcinomas (SCCs) constitute a group of human malignancies that originate from the squamous epithelium. Most SCC patients experience treatment failure and relapse and have a poor prognosis due to de novo and acquired resistance to first-line chemotherapeutic agents. To identify chemoresistance mechanisms and explore novel chemosensitizer targets, we performed whole-transcriptome sequencing of paired resistant/parental SCC cells. We identified DLGAP1 antisense RNA 2 (D-AS2) as a crucial noncoding RNA that contributes to chemoresistance in SCC. Mechanistically, D-AS2 associates with histones to regulate the distal elements of FAM3 metabolism regulating signaling molecule D (FAM3D) and reduces extracellular FAM3D protein secretion. FAM3D interacts with Gi-coupled G protein-coupled receptor (GPCR) formyl peptide receptor (FPR) 1 and FPR2 to suppress phospholipase D (PLD) activity; thus, reduced FAM3D activates PLD signaling. Moreover, activated PLD promotes phosphatidic acid (PA) production and subsequent yes-associated protein (YAP) nuclear translocation. Accordingly, in vivo administration of a D-AS2-targeting antisense oligonucleotide sensitizes SCC to cisplatin treatment. In summary, our study reveals that D-AS2/FAM3D-mediated PLD/PA lipid signaling is essential in SCC chemoresistance and that D-AS2 can be targeted to sensitize SCC to cytotoxic chemotherapeutic agents. SignificanceThis study identifies D-AS2 as a targetable lipid-related lncRNA that activates YAP signaling via PLD/PA axis to trigger chemoresistance in SCC.

cancer biology↗

bioRxiv2021

Combining Multi-Dimensional Molecular Fingerprints to Predict hERG Cardiotoxicity of Compounds

At present, drug toxicity has become a critical problem with heavy medical and economic burdens. acLQTS (acquired Long QT Syndrome) is acquired cardiac ion channel disease caused by drugs blocking the hERG channel. Therefore, it is necessary to avoid cardiotoxicity in the drug design and computer models have been widely used to fix this plight. In this study, we present a molecular fingerprint based on the molecular dynamic simulation and uses it combined with other molecular fingerprints (multi-dimensional molecular fingerprints) to predict hERG cardiotoxicity of compounds. 203 compounds with hERG inhibitory activity (pIC50) were retrieved from a previous study and predicting models were established using four machine learning algorithms based on the single and multi-dimensional molecular fingerprints. Results showed that MDFP has the potential to be an alternative to traditional molecular fingerprints and the combination of MDFP and traditional molecular fingerprints can achieve higher prediction accuracy. Meanwhile, the accuracy of the best model, which was generated by consensus of four algorithms with multi-dimensional molecular fingerprints, was 0.694 (RMSE) in the test dataset. Besides, the number of hydrogen bonds from MDFP has been determined as a critical factor in the predicting models, followed by rgyr and sasa. Our findings provide a new sight of MDFP and multi-dimensional molecular fingerprints in building models of hERG cardiotoxicity prediction.

pharmacology and toxicology↗