AlphaFold two years on: Validation and impact

AI-based quality assessment methods for protein structure models from cryo-EM

Cryogenic electron microscopy (cryo-EM) has revolutionized structural biology, with an increasing number of structures being determined by cryo-EM each year, many at higher resolutions. However, challenges remain in accurately interpreting cryo-EM maps. Inaccuracies can arise in regions of locally low resolution, where manual model building is more prone to errors. Validation scores for structure models have been developed to assess both the compatibility between map density and the structure, as well as the geometric and stereochemical properties of protein models. Recent advancements have introduced artificial intelligence (AI) into this field. These emerging AI-driven tools offer unique capabilities in the validation and refinement of cryo-EM-derived protein atomic models, potentially leading to more accurate protein structures and deeper insights into complex biological systems.

Structural Biology of RNA and Protein-RNA Complexes after AlphaFold3

Recent breakthroughs in AI-mediated protein structure prediction have significantly accelerated research and generated valuable hypotheses within the field of structural biology and beyond. Notably, AlphaFold2 has facilitated the determination of larger protein complexes for which only limited experimental data are available. De novo predictions can now be experimentally validated with relative ease compared to the pre-AlphaFold2 era. In May 2024, AlphaFold3 was launched with high expectations, promising the capability to accurately predict RNA structures and protein-RNA complexes - features that were absent in AlphaFold2. This review evaluates the extent to which AlphaFold3 fulfills this promise through specific examples. At present, AlphaFold3 falls short in reliably predicting RNA and protein-RNA complex structures, particularly for non-canonical interactions where training data remain scarce. As a result, users should exercise caution when using AlphaFold3 predictions as hypotheses generators for RNA and protein-RNA complex structures. In the interim, integrating AI-based predictors with data-driven docking tools is recommended to address these limitations. This approach can help bridge the gap until sufficient training data are available to enable the development of more reliable predictive algorithms.

Structure-Based Experimental Datasets for Benchmarking Protein Simulation Force Fields [Article v0.1]

This review article provides an overview of structurally oriented experimental datasets that can be used to benchmark protein force fields, focusing on data generated by nuclear magnetic resonance (NMR) spectroscopy and room temperature (RT) protein crystallography. We discuss what the observables are, what they tell us about structure and dynamics, what makes them useful for assessing force field accuracy, and how they can be connected to molecular dynamics simulations carried out using the force field one wishes to benchmark. We also touch on statistical issues that arise when comparing simulations with experiment. We hope this article will be particularly useful to computational researchers and trainees who develop, benchmark, or use protein force fields for molecular simulations.

Computational insights into the inhibitory effects of PFAS 14 on colorectal cancer targeting GSTA1 through competitive binding

This study employed computational biology approaches to investigate the interactions between per- and polyfluoroalkyl substances (PFAS) and key colorectal cancer (CRC) proteins. The results indicate that PFAS may influence CRC progression by modulating multiple proteins, particularly glutathione S-transferase A1 (GSTA1). Computational analysis revealed that PFAS 14 exhibits high binding affinity for GSTA1, occupying its glutathione-binding site. Further simulations confirmed the stable binding of PFAS 14 across different environments, forming persistent hydrogen bonds and water bridges, suggesting a potential inhibitory effect on GSTA1.GSTA1, a key member of the glutathione S-transferase family, plays a critical role in detoxification by catalyzing the conjugation of glutathione to electrophilic compounds. Dysregulation of GSTA1 has been implicated in cancer progression and chemoresistance. In CRC, altered GSTA1 expression may affect tumor metabolism and drug response, making it a potential therapeutic target.This study identifies GSTA1 as a key target of PFAS interactions, suggesting that environmental PFAS exposure may influence CRC by interfering with detoxification mechanisms. The competitive inhibition of GSTA1 by PFAS 14 may impact cancer cell survival and progression. Future research should integrate experimental validation to assess its phenotypic effects and evaluate PFAS 14 as a potential GSTA1 inhibitor.

Protein binding and folding through an evolutionary lens

Protein-protein associations are often mediated by an intrinsically disordered protein region interacting with a folded domain in a coupled binding and folding reaction. Classic physical organic chemistry approaches together with structural biology have shed light on mechanistic aspects of such reactions. Further insight into general principles may be obtained by interpreting the results through an evolutionary lens. This review attempts to provide an overview on how the analysis of binding and folding reactions can benefit from an evolutionary approach, and is aimed at protein scientists without a background in evolution. Evolution constantly reshapes existing proteins by sampling more or less fit variants. Most new variants are weeded out as generations and new species come and go over hundreds to hundreds of millions of years. The huge ongoing genome sequencing efforts have provided us with a snapshot of existing adapted fit-for-purpose protein homologs in thousands of different organisms. Comparison of present-day orthologs and paralogs highlights general principles of the evolution of coupled binding and folding reactions and demonstrate a great potential for evolution to operate on disordered regions and modulate affinity and specificity of the interactions.

Molecular Origami: Designing Functional Molecules of the Future

In the field of chemical biology, DNA origami has been actively researched. This technique, which involves folding DNA strands like origami to assemble them into desired shapes, has made it possible to create complex nanometer-sized structures, marking a major breakthrough in nanotechnology. On the other hand, controlling the folding mechanisms and folded structures of proteins or shorter peptides has been challenging. However, recent advances in techniques such as protein origami, peptide origami, and de novo design peptides have made it possible to construct various nanoscale structures and create functional molecules. These approaches suggest the emergence of new molecular design principles, which can be termed "molecular origami". In this review, we provide an overview of recent research trends in protein/peptide origami and DNA/RNA origami and explore potential future applications of molecular origami technologies in electrochemical biosensors.

Prediction of peptide structural conformations with AlphaFold2

Protein structure prediction via artificial intelligence/machine learning (AI/ML) approaches has sparked substantial research interest in structural biology and adjacent disciplines. More recently, AlphaFold2 (AF2) has been adapted for the prediction of multiple structural conformations in addition to single-state structures. This novel avenue of research has focused on proteins (typically 50 residues in length or greater), while multi-conformation prediction of shorter peptides has not yet been explored in this context. Here, we report AF2-based structural conformation prediction of a total of 557 peptides (ranging in length from 10 to 40 residues) for a benchmark dataset with corresponding nuclear magnetic resonance (NMR)-determined conformational ensembles. De novo structure predictions were accompanied by structural comparison analyses to assess prediction accuracy. We found that the prediction of conformational ensembles for peptides with AF2 varied in accuracy versus NMR data, with average root-mean-square deviation (RMSD) among structured regions under 2.5 Å and average root-mean-square fluctuation (RMSF) differences under 1.5 Å. Our results reveal notable capabilities of AF2-based structural conformation prediction for peptides but also underscore the necessity for interpretation discretion.