Artificial Intelligence in Cryo-Electron Microscopy

Single particle cryo-EM map and model validation: It's not crystal clear

The application of single particle cryogenic electron microscopy (cryo-EM) to structure determination continues to have a transformative impact on our understanding on biological systems. While there has been a great deal of algorithmic development focused on improving attainable resolutions and streamlining atomic model building, there has not been commensurate development of validation metrics to ensure the accuracy of our cryo-EM maps and models. This review emphasizes the persistent issues that currently complicate single particle cryo-EM structure validation, and highlights the metrics that are gaining broad acceptance by the community. This article aims to underscore the need for further development of validation criteria and the potential role of machine learning methodologies in confidently assessing the quality of cryo-EM structures.

How structural biology has changed our understanding of icosahedral viruses

Cryo-electron microscopy and tomography have allowed us to unveil the remarkable structure of icosahedral viruses. However, in the past few years, the idea that these viruses must have perfectly symmetric virions, but in some cases, it might not be true. This has opened the door to challenging paradigms in structural virology and raised new questions about the biological implications of "unusual" or "defective" symmetries and structures. Also, the continual improvement of these technologies, coupled with more rigorous sample purification protocols, improvements in data processing, and the use of artificial intelligence, has allowed solving the structure of sub-viral particles in highly heterogeneous samples and finding novel symmetries or structural defects. In this review, I initially analyzed the case of the symmetry and composition of hepatitis B virus-produced spherical sub-viral particles. Then, I focused on Alphaviruses as an example of "imperfect" icosahedrons and analyzed how structural biology has changed our understanding of the Alphavirus assembly and some biological implications arising from these discoveries.

Allostery Illuminated: Harnessing AI and Machine Learning for Drug Discovery

In the past several years there has been rapid adoption of artificial intelligence (AI) and machine learning (ML) tools for drug discovery. In this Microperspective, we comment on recent AI/ML applications to the discovery of allosteric modulators, focusing on breakthroughs with AlphaFold, structure-based drug discovery (SBDD), and medicinal chemistry applications. We discuss how these technologies are facilitating drug discovery and the remaining challenges to identify allosteric binding sites and ligands.

Refinement of cryo-EM 3D maps with a self-supervised denoising model: crefDenoiser

Cryogenic electron microscopy (cryo-EM) is a pivotal technique for imaging macromolecular structures. However, despite extensive processing of large image sets collected in cryo-EM experiments to amplify the signal-to-noise ratio, the reconstructed 3D protein-density maps are often limited in quality due to residual noise, which in turn affects the accuracy of the macromolecular representation. Here, crefDenoiser is introduced, a denoising neural network model designed to enhance the signal in 3D cryo-EM maps produced with standard processing pipelines. The crefDenoiser model is trained without the need for `clean' ground-truth target maps. Instead, a custom dataset is employed, composed of real noisy protein half-maps sourced from the Electron Microscopy Data Bank repository. Competing with the current state-of-the-art, crefDenoiser is designed to optimize for the theoretical noise-free map during self-supervised training. We demonstrate that our model successfully amplifies the signal across a wide variety of protein maps, outperforming a classic map denoiser and following a network-based sharpening model. Without biasing the map, the proposed denoising method leads to improved visibility of protein structural features, including protein domains, secondary structure elements and modest high-resolution feature restoration.

Machine learning approaches to cryoEM density modification differentially affect biomacromolecule and ligand density quality

The application of machine learning to cryogenic electron microscopy (cryoEM) data analysis has added a valuable set of tools to the cryoEM data processing pipeline. As these tools become more accessible and widely available, the implications of their use should be assessed. We noticed that machine learning map modification tools can have differential effects on cryoEM densities. In this perspective, we evaluate these effects to show that machine learning tools generally improve densities for biomacromolecules while generating unpredictable results for ligands. This unpredictable behavior manifests both in quantitative metrics of map quality and in qualitative investigations of modified maps. The results presented here highlight the power and potential of machine learning tools in cryoEM, while also illustrating some of the risks of their unexamined use.

Exploring advances in single particle CryoEM with apoferritin: From blobs to true atomic resolution

Deciphering the three-dimensional structures of macromolecules is of paramount importance for gaining insights into their functions and roles in human health and disease. Single particle cryoEM has emerged as a powerful technique that enables direct visualization of macromolecules and their complexes, and through subsequent averaging, achieve near atomic-level resolution. A major breakthrough was recently achieved with the determination of the apoferritin structure at true atomic resolution. In this review, we discuss the latest technological innovations across the entire single-particle workflow, which have been instrumental in driving the resolution revolution and in transforming cryoEM as a mainstream technique in structural biology. We illustrate these advancements using apoferritin as an example that has served as an excellent benchmark sample for assessing emerging technologies. We further explore whether the existing technology can routinely generate atomic structures of dynamic macromolecules that more accurately represent real-world samples, the limitations in the workflow, and the current approaches employed to overcome them.

Cryo-electron microscopy-based drug design

Structure-based drug design (SBDD) has gained popularity owing to its ability to develop more potent drugs compared to conventional drug-discovery methods. The success of SBDD relies heavily on obtaining the three-dimensional structures of drug targets. X-ray crystallography is the primary method used for solving structures and aiding the SBDD workflow; however, it is not suitable for all targets. With the resolution revolution, enabling routine high-resolution reconstruction of structures, cryogenic electron microscopy (cryo-EM) has emerged as a promising alternative and has attracted increasing attention in SBDD. Cryo-EM offers various advantages over X-ray crystallography and can potentially replace X-ray crystallography in SBDD. To fully utilize cryo-EM in drug discovery, understanding the strengths and weaknesses of this technique and noting the key advancements in the field are crucial. This review provides an overview of the general workflow of cryo-EM in SBDD and highlights technical innovations that enable its application in drug design. Furthermore, the most recent achievements in the cryo-EM methodology for drug discovery are discussed, demonstrating the potential of this technique for advancing drug development. By understanding the capabilities and advancements of cryo-EM, researchers can leverage the benefits of designing more effective drugs. This review concludes with a discussion of the future perspectives of cryo-EM-based SBDD, emphasizing the role of this technique in driving innovations in drug discovery and development. The integration of cryo-EM into the drug design process holds great promise for accelerating the discovery of new and improved therapeutic agents to combat various diseases.

Structural Analysis of Protein Complexes by Cryo-Electron Microscopy

Structural studies of bio-complexes using single particle cryo-Electron Microscopy (cryo-EM) is nowadays a well-established technique in structural biology and has become competitive with X-ray crystallography. Development of digital registration systems for electron microscopy images and algorithms for the fast and efficient processing of the recorded images and their following analysis has facilitated the determination of structures at near-atomic resolution. The latest advances in EM have enabled the determination of protein complex structures at 1.4-3 Å resolution for an extremely broad range of sizes (from ~100 kDa up to hundreds of MDa (Bartesaghi et al., Science 348(6239):1147-1151, 2015; Herzik et al., Nat Commun 10:1032, 2019; Wu et al., J Struct Biol X 4:100020, 2020; Zhang et al., Nat Commun 10:5511, 2019; Zhang et al., Cell Res 30(12):1136-1139, 2020; Yip et al., Nature 587(7832):157-161, 2020; https://www.ebi.ac.uk/emdb/statistics/emdb_resolution_year )). In 2022, nearly 1200 structures deposited to the EMDB database were at a resolution of better than 3 Å ( https://www.ebi.ac.uk/emdb/statistics/emdb_resolution_year ).To date, the highest resolutions have been achieved for apoferritin, which comprises a homo-oligomer of high point group symmetry (O432) and has rigid organization together with high stability (Zhang et al., Cell Res 30(12):1136-1139, 2020; Yip et al., Nature 587(7832):157-161, 2020). It has been used as a test object for the assessments of modern cryo-microscopes and processing methods during the last 5 years. In contrast to apoferritin bacterial secretion systems are typical examples of multi protein complexes exhibiting high flexibility owing to their functions relating to the transportation of small molecules, proteins, and DNA into the extracellular space or target cells. This makes their structural characterization extremely challenging (Barlow, Methods Mol Biol 532:397-411, 2009; Costa et al., Nat Rev Microbiol 13:343-359, 2015). The most feasible approach to reveal their spatial organization and functional modification is cryo-electron microscopy (EM). During the last decade, structural cryo-EM has become broadly used for the analysis of the bio-complexes that comprise multiple components and are not amenable to crystallization (Lyumkis, J Biol Chem 294:5181-5197, 2019; Orlova and Saibil, Methods Enzymol 482:321-341, 2010; Orlova and Saibil, Chem Rev 111(12):7710-7748, 2011).In this review, we will describe the basics of sample preparation for cryo-EM, the principles of digital data collection, and the logistics of image analysis focusing on the common steps required for reconstructions of both small and large biological complexes together with refinement of their structures to nearly atomic resolution. The workflow of processing will be illustrated by examples of EM analysis of Type IV Secretion System.

Scaling up cryo-EM for biology and chemistry: The journey from niche technology to mainstream method

Cryoelectron microscopy (cryo-EM) methods have made meaningful contributions in a wide variety of scientific research fields. In structural biology, cryo-EM routinely elucidates molecular structure from isolated biological macromolecular complexes or in a cellular context by harnessing the high-resolution power of the electron in order to image samples in a frozen, hydrated environment. For structural chemistry, the cryo-EM method popularly known as microcrystal electron diffraction (MicroED) has facilitated atomic structure generation of peptides and small molecules from their three-dimensional crystal forms. As cryo-EM has grown from an emerging technology, it has undergone modernization to enable multimodal transmission electron microscopy (TEM) techniques becoming more routine, reproducible, and accessible to accelerate research across multiple disciplines. We review recent advances in modern cryo-EM and assess how they are contributing to the future of the field with an eye to the past.

Quantitative analysis of protein dynamics using a deep learning technique combined with experimental cryo-EM density data and MD simulations

Protein functions associated with biological activity are precisely regulated by both tertiary structure and dynamic behavior. Thus, elucidating the high-resolution structures and quantitative information on in-solution dynamics is essential for understanding the molecular mechanisms. The main experimental approaches for determining tertiary structures include nuclear magnetic resonance (NMR), X-ray crystallography, and cryogenic electron microscopy (cryo-EM). Among these procedures, recent remarkable advances in the hardware and analytical techniques of cryo-EM have increasingly determined novel atomic structures of macromolecules, especially those with large molecular weights and complex assemblies. In addition to these experimental approaches, deep learning techniques, such as AlphaFold 2, accurately predict structures from amino acid sequences, accelerating structural biology research. Meanwhile, the quantitative analyses of the protein dynamics are conducted using experimental approaches, such as NMR and hydrogen-deuterium mass spectrometry, and computational approaches, such as molecular dynamics (MD) simulations. Although these procedures can quantitatively explore dynamic behavior at high resolution, the fundamental difficulties, such as signal crowding and high computational cost, greatly hinder their application to large and complex biological macromolecules. In recent years, machine learning techniques, especially deep learning techniques, have been actively applied to structural data to identify features that are difficult for humans to recognize from big data. Here, we review our approach to accurately estimate dynamic properties associated with local fluctuations from three-dimensional cryo-EM density data using a deep learning technique combined with MD simulations.

From Deep Mutational Mapping of Allosteric Protein Landscapes to Deep Learning of Allostery and Hidden Allosteric Sites: Zooming in on "Allosteric Intersection" of Biochemical and Big Data Approaches

The recent advances in artificial intelligence (AI) and machine learning have driven the design of new expert systems and automated workflows that are able to model complex chemical and biological phenomena. In recent years, machine learning approaches have been developed and actively deployed to facilitate computational and experimental studies of protein dynamics and allosteric mechanisms. In this review, we discuss in detail new developments along two major directions of allosteric research through the lens of data-intensive biochemical approaches and AI-based computational methods. Despite considerable progress in applications of AI methods for protein structure and dynamics studies, the intersection between allosteric regulation, the emerging structural biology technologies and AI approaches remains largely unexplored, calling for the development of AI-augmented integrative structural biology. In this review, we focus on the latest remarkable progress in deep high-throughput mining and comprehensive mapping of allosteric protein landscapes and allosteric regulatory mechanisms as well as on the new developments in AI methods for prediction and characterization of allosteric binding sites on the proteome level. We also discuss new AI-augmented structural biology approaches that expand our knowledge of the universe of protein dynamics and allostery. We conclude with an outlook and highlight the importance of developing an open science infrastructure for machine learning studies of allosteric regulation and validation of computational approaches using integrative studies of allosteric mechanisms. The development of community-accessible tools that uniquely leverage the existing experimental and simulation knowledgebase to enable interrogation of the allosteric functions can provide a much-needed boost to further innovation and integration of experimental and computational technologies empowered by booming AI field.

A review of the approaches used to solve sub-100 kDa membrane proteins by cryo-electron microscopy

Membrane proteins (MPs) are essential components of all biological membranes, contributing to key cellular functions that include signalling, molecular transport and energy metabolism. Consequently, MPs are important biomedical targets for therapeutics discovery. Despite hardware and software developments in cryo-electron microscopy, as well as MP sample preparation, MPs smaller than 100 kDa remain difficult to study structurally. Significant investment is required to overcome low levels of naturally abundant protein, MP hydrophobicity as well as conformational and compositional instability. Here we have reviewed the sample preparation approaches that have been taken to successfully express, purify and prepare small MPs for analysis by cryo-EM (those with a total solved molecular weight of under 100 kDa), as well as examining the differing approaches towards data processing and ultimately obtaining a structural solution. We highlight common challenges at each stage in the process as well as strategies that have been developed to overcome these issues. Finally, we discuss future directions and opportunities for the study of sub-100 kDa membrane proteins by cryo-EM.