Sulin Zhang

Aberrant activation of the STING signaling pathway is involved with aging-related inflammation, yet no STING-targeting therapeutics are clinically approved. Here, we identify Omaveloxolone, an FDA-approved drug for Friedreich’s ataxia, as a potent STING degrader (DC ₅₀ = 118 nM, D _max = 98.6%) via high-throughput screening. Omaveloxolone binds non-covalently to the base of the STING ligand-binding domain (LBD) with high affinity ( K _D = 21.4 nM) and exhibits crowbar-like behavior consistent with STING conformational destabilization: it hinders LBD closure, increases hydrophobic exposure, and reduces thermal stability. This conformational perturbation is proposed to enhance recognition by the E3 ubiquitin ligase HUWE1, leading to K27-linked ubiquitination at STING lysines K236/K347/K370 and proteasomal degradation. In vivo, Omaveloxolone depletes STING, suppresses downstream inflammatory signaling, and ameliorates aging-related inflammatory phenotypes in aged male mice. Because Omaveloxolone is also an established NRF2 activator, these in vivo benefits likely reflect dual modulation of inflammatory and antioxidant pathways. This study implicates Omaveloxolone as a molecular crowbar-like STING degrader and provides a mechanistic framework for targeting STING via conformational perturbation.

Regulatory T-cell (Treg) stability is maintained by dynamic remodeling of the FOXP3 transcriptional complex, and disruption of this complex leads to Treg dysfunction and immune dysregulation. However, how specific FOXP3 mutations alter the dynamic remodeling of the FOXP3 complex and thereby contribute to pathogenic Treg reprogramming in IPEX syndrome remains unclear. Here, we demonstrate that predominant Th1 inflammation manifests in both FOXP3^V408M IPEX patients and FOXP3^V408M knock-in mice and reveal that the mutation intrinsically impairs Treg-mediated control of Th1 inflammation, revealing a distinct pathogenesis of this mutation in IPEX syndrome. Mechanistically, the V408M mutation disrupts the FOXP3–T-bet interaction, impairing the FOXP3-mediated restraint of T-bet-driven IFN-γ production and thereby contributing to increased Th1 inflammation. Using an AI-driven virtual screening approach, we identified a first-in-class small molecule, FM029, that directly binds to FOXP3 and reinforces its interaction with T-bet. FM029 strongly suppressed Treg-derived IFN-γ production and alleviated IFN-γ-driven tissue inflammation in both FOXP3^V408M mice and an acute colitis model. Collectively, these findings establish the FOXP3–T-bet interaction as a central checkpoint that governs Treg stability and IFN-γ-driven Th1 pathology, providing a proof-of-concept that pharmacologic stabilization of the FOXP3–T-bet interaction can mitigate IFN-γ-driven immune disorders.

Targeting methionine adenosyltransferase 2A (MAT2A) represents an attractive synthetic lethal approach for treating methylthioadenosine phosphorylase (MTAP)-deleted cancers. Despite promising progress in the development of MAT2A inhibitors, many reported compounds remain limited by insufficient selectivity, systemic toxicity, and the stereochemical complexities of atropisomerism. Through structure-guided optimization of our previously reported atropisomeric lead, we identified ZS97 (26), a potent MAT2A inhibitor that lacks axial chirality. The nonatropisomeric nature simplifies pharmacological development by avoiding the need for asymmetric synthesis or chiral resolution. ZS97 demonstrated potent enzymatic inhibition, effectively depleting intracellular SAM and suppressing downstream PRMT5-mediated methylation. This biochemical blockade selectively triggered the DNA damage response and cell growth inhibition in MTAP-deficient cancer cells. Furthermore, ZS97 exhibited favorable oral bioavailability and induced significant tumor inhibition in an NCI-H838 xenograft model without associated weight loss. These findings highlight ZS97 as a structurally optimized orally active candidate with therapeutic potential for MTAP-deficient cancers.

Predicting small molecule–protein interactions across nonhomologous proteins remains challenging because shared ligand recognition is often not evident from sequence, fold, or pocket similarity. Here, we introduce pocket hopping, a machine-learning framework that learns residue-level interaction patterns from coligand binding pockets and infers compatibility between nonhomologous pockets for similar chemotypes. Using shared ligands as supervision rather than explicit geometric alignment, pocket hopping identifies pocket relationships that are not readily captured by conventional chemical-, sequence-, or structure-based comparisons. In two case studies, pocket hopping demonstrates broad utility in drug discovery by enabling de novo hit identification and mechanistic interpretation, identifying fedratinib and its analogues as helicase WRN inhibitors. The model also identified the clinical-stage HDAC inhibitor abexinostat as a direct ENPP1 binder and inhibitor, and cellular assays showed enhanced cGAMP–STING signaling under cGAMP stimulation. Together, these results indicate that pocket-level compatibility can complement existing approaches for target identification, hit discovery, and polypharmacology analysis.

Accelerating molecular probe discovery and lead optimization requires accurate and efficient binding affinity prediction. Here we present PBCNet2.0, a Cartesian tensor-based Siamese neural network for protein-ligand relative binding affinity prediction. Trained on 8.6 million protein-ligand complex pairs, PBCNet2.0 achieves zero-shot accuracy similar to computationally intensive physics-based simulations while remaining highly efficient. Retrospective prioritization experiments show that PBCNet2.0 improves optimization efficiency by 7.18-fold and reduces resource use by 41%. Mechanistic analyses indicate that the model captures intermolecular interactions and encodes spatial geometric constraints, enabling sensitivity to subtle effects such as fluorine orthogonal multipolar interactions. Notably, although not trained on mutation data, PBCNet2.0 exhibits an emergent capability to predict affinity changes induced by binding pocket residue variations, supporting resistance analysis. We prospectively validated these capabilities on ENPP1 and ALDH1B1, accurately resolving affinity shifts from minor interaction and conformational differences and identifying critical binding residues with a hit rate of five out of six selected residues.

Despite recent advances in immunotherapy for microsatellite instability-high (MSI-H) tumors, challenges of resistance persist, necessitating alternative therapeutic approaches. The synthetic lethal interaction between WRN helicase inhibition and MSI status presents a promising therapeutic strategy. Guided by the key structural features of the WRN-HRO761 complex, we rationally designed two classes of WRN inhibitors─spirocyclic compounds and benzo-fused heterocyclic analogs─by targeting the solvent-exposed region. Among these, Q15 exhibited potent in vitro activity and excellent cellular selectivity. Compared with HRO761 in vivo, Q15 demonstrated more favorable oral pharmacokinetic properties and superior antitumor efficacy at 10 mg/kg, comparable efficacy at 20 mg/kg, and achieved complete tumor regression at 40 mg/kg. Preliminary safety evaluation further indicated that Q15 possesses a favorable safety profile. These findings support Q15 as a promising WRN inhibitor candidate for the treatment of MSI-H tumors.

Precision oncology faces critical challenges in interpreting complex cellular signals and predicting drug responses across heterogeneous cancer environments. Here, we present BioGDR, a multimodal interpretable deep learning framework that integrates structure-based predicted biological features, including differential gene expression and kinase inhibition profiles, eliminating the need for experimental measurements. By modeling tumor transcriptomic states through pathway-informed graph neural networks and employing a drug-guided attention strategy, BioGDR enables mechanistic insights into drug sensitivity across compound and cellular contexts. Comprehensive evaluations demonstrate that BioGDR outperforms existing methods in compound screening relevant to early-stage drug discovery and in predicting cell line sensitivity across heterogeneous cellular states characteristic of precision oncology, while analyses on clinical patient cohorts further confirm its practical utility and generalization capability. Experimental validation with a novel ALDH1B1 inhibitor confirms its ability to identify sensitive cell populations and reveal underlying mechanisms. This work establishes a robust, biologically informed framework that bridges preclinical drug development and clinical applications, advancing precision oncology through integrative, multimodal learning and interpretable mechanism analysis.