Search for: All records

Structure-based drug design (SBDD) leverages the three-dimensional geometry of proteins to identify potential drug candidates. Traditional approaches, rooted in physicochemical modeling and domain expertise, are often resource-intensive. Recent advancements in geometric deep learning, which effectively integrate and process 3D geometric data, alongside breakthroughs in accurate protein structure predictions from tools like AlphaFold, have significantly propelled the field forward. This article systematically reviews the state-of-the-art in geometric deep learning for SBDD. We begin by outlining foundational tasks in SBDD, discussing prevalent 3D protein representations, and highlighting representative predictive and generative models. Next, we provide an in-depth review of key tasks, including binding site prediction, binding pose generation, de novo molecule generation, linker design, protein pocket generation, and binding affinity prediction. For each task, we present formal problem definitions, key methods, datasets, evaluation metrics, and performance benchmarks. Lastly, we explore current challenges and future opportunities in SBDD. Challenges include oversimplified problem formulations, limited out-of-distribution generalization, biosecurity concerns related to the misuse of structural data, insufficient evaluation metrics and large-scale benchmarks, and the need for experimental validation and enhanced model interpretability. Opportunities lie in integrating biomedical AI agents, leveraging multimodal datasets, developing comprehensive benchmarks, establishing criteria aligned with clinical outcomes, and designing foundation models to expand the scope of design tasks. We also curatehttps://github.com/zaixizhang/Awesome-SBDD, reflecting ongoing contributions and new datasets in SBDD. 

Despite their biological significance, the study of hydropersulfides (RSSH) is often limited due to their inherent instability. Here, we introduce arylsulfonothioates as thiol activated RSSH donors and provide insight into cellular reactive sulfur species homeostasis. These precursors persulfidate physiologically relevant thiols (RSH) to form the corresponding RSSH. Real-time monitoring of hydrogen sulfide (H2S) generation via membrane inlet mass spectrometry (MIMS) was employed to follow RSSH production, revealing that electron-donating aryl substituents marginally slow RSSH release rates, whereas electron-withdrawing substituents slightly accelerate release. Furthermore, arylsulfonothioates with strong electron-withdrawing substituents offer superior protection against doxorubicin (DOX)-induced cardiotoxicity. Experiments using H9c2 cardiomyocytes affirmed the cell-permeability of arylsulfonothioates and their ability to increase intracellular RSSH levels and protein persulfidation levels. Notably, we observe the excretion of RSSH into the extracellular medium. Further investigations revealed the involvement of the cystine/glutamate antiporter SLC7A11, as cotreatment with its inhibitor, sulfasalazine, significantly reduce extracellular RSSH release. H9c2 cells exhibit tolerance to arylsulfonothioate 1g with an electronwithdrawing 4-cyano group at 1 mM; however, inhibition of the cystine antiporter results in a minor decrease in cell viability. Under oxidative stress conditions induced by DOX or hydrogen peroxide (H2O2), cotreatment with 1g diminishes the excretion of RSSH and confers cytoprotection against DOX or H2O2-mediated toxicity. Our findings show adaptive cellular responses to RSSH levels, demonstrating excretion under elevated conditions to maintain redox homeostasis and intracellular retention as a protective response during oxidative stress. 

Amorphous zeolitic imidazolate framework (aZIF) films have been recently introduced as resists for electron beam and extreme ultraviolet lithography (EBL and EUVL). aZIFs are also being considered for separation applications, including thin film membranes. However, the reported methods for aZIF deposition are currently based on highly empirical trial-and-error approaches that hinder control of film composition, thickness, and uniformity as well as scale-up and transferability to different coating geometries. This work presents a method for depositing aZIF films with controllable thickness using dilute precursors mixed immediately before encountering the substrate. Importantly, the method is amenable to quantitative analysis by computational fluid dynamics (CFD) to extract intrinsic deposition rates and limiting reactant transport diffusivities, enabling predictive physics-based modeling of the deposition process. This allows the deposition method to be adapted for spin coating on silicon wafers to prepare high-quality aZIF films with consistently controlled thickness. Using this approach, high-resolution resist performance and wafer-scale use for beyond EUV (BEUV) lithography of aZIF films is demonstrated. 

Abstract Designing protein-binding proteins is critical for drug discovery. However, artificial-intelligence-based design of such proteins is challenging due to the complexity of protein–ligand interactions, the flexibility of ligand molecules and amino acid side chains, and sequence–structure dependencies. We introduce PocketGen, a deep generative model that produces residue sequence and atomic structure of the protein regions in which ligand interactions occur. PocketGen promotes consistency between protein sequence and structure by using a graph transformer for structural encoding and a sequence refinement module based on a protein language model. The graph transformer captures interactions at multiple scales, including atom, residue and ligand levels. For sequence refinement, PocketGen integrates a structural adapter into the protein language model, ensuring that structure-based predictions align with sequence-based predictions. PocketGen can generate high-fidelity protein pockets with enhanced binding affinity and structural validity. It operates ten times faster than physics-based methods and achieves a 97% success rate, defined as the percentage of generated pockets with higher binding affinity than reference pockets. Additionally, it attains an amino acid recovery rate exceeding 63%. 

Abstract Maintaining intestinal homeostasis relies on the intricate interplay among the mucosal epithelium, immune system, and host microbiome. A key question is how innate immune cells sense and process microbes in the gut lumen, eliciting appropriate protective responses without causing tissue injury. Clearance of invading microbes and initiation of downstream inflammatory responses are central to this process and require proper function of the endolysosomal system. Dysfunction of this system can predispose the host to chronic inflammatory disorders and acute infections. Here, through forward genetic screening of N-ethyl-N-nitrosourea (ENU)-mutagenized mice and CRISPR/Cas9 validation, we identifyLrmda, encoding leucine-rich melanocyte differentiation-associated protein (LRMDA), as a key regulator of intestinal homeostasis. Using hematopoietic chimera and conditional knockouts, we show that LRMDA functions primarily in CD11c⁺cells, including mucosal dendritic cells (DCs) and macrophages, but not in non-hematopoietic cells. Proteomic, cellular, and biochemical analyses reveal that LRMDA directly and cooperatively interacts with the endolysosome-specific small GTPase Rab32 and the endosomal recycling complex Retriever. Loss of LRMDA or Retriever function increases susceptibility to dextran sodium sulfate (DSS)-induced colitis and impairs clearance ofListeria monocytogenes. Together, our findings establish the Rab32-LRMDA-Retriever complex as a critical regulator of endolysosomal trafficking in innate immune cells, essential for maintaining intestinal immune homeostasis.