Current approaches for automated model building into cryo-EM maps using Buccaneer with CCP-EM

Deciphering the allosteric regulation of mycobacterial inosine-5'-monophosphate dehydrogenase

Allosteric regulation of inosine 5'-monophosphate dehydrogenase (IMPDH), an essential enzyme of purine metabolism, contributes to the homeostasis of adenine and guanine nucleotides. However, the precise molecular mechanism of IMPDH regulation in bacteria remains unclear. Using biochemical and cryo-EM approaches, we reveal the intricate molecular mechanism of the IMPDH allosteric regulation in mycobacteria. The enzyme is inhibited by both GTP and (p)ppGpp, which bind to the regulatory CBS domains and, via interactions with basic residues in hinge regions, lock the catalytic core domains in a compressed conformation. This results in occlusion of inosine monophosphate (IMP) substrate binding to the active site and, ultimately, inhibition of the enzyme. The GTP and (p)ppGpp allosteric effectors bind to their dedicated sites but stabilize the compressed octamer by a common mechanism. Inhibition is relieved by the competitive displacement of GTP or (p)ppGpp by ATP allowing IMP-induced enzyme expansion. The structural knowledge and mechanistic understanding presented here open up new possibilities for the development of allosteric inhibitors with antibacterial potential.

Automated model building and protein identification in cryo-EM maps

Interpreting electron cryo-microscopy (cryo-EM) maps with atomic models requires high levels of expertise and labour-intensive manual intervention in three-dimensional computer graphics programs1,2. Here we present ModelAngelo, a machine-learning approach for automated atomic model building in cryo-EM maps. By combining information from the cryo-EM map with information from protein sequence and structure in a single graph neural network, ModelAngelo builds atomic models for proteins that are of similar quality to those generated by human experts. For nucleotides, ModelAngelo builds backbones with similar accuracy to those built by humans. By using its predicted amino acid probabilities for each residue in hidden Markov model sequence searches, ModelAngelo outperforms human experts in the identification of proteins with unknown sequences. ModelAngelo will therefore remove bottlenecks and increase objectivity in cryo-EM structure determination.

DeepMainmast: integrated protocol of protein structure modeling for cryo-EM with deep learning and structure prediction

Three-dimensional structure modeling from maps is an indispensable step for studying proteins and their complexes with cryogenic electron microscopy. Although the resolution of determined cryogenic electron microscopy maps has generally improved, there are still many cases where tracing protein main chains is difficult, even in maps determined at a near-atomic resolution. Here we developed a protein structure modeling method, DeepMainmast, which employs deep learning to capture the local map features of amino acids and atoms to assist main-chain tracing. Moreover, we integrated AlphaFold2 with the de novo density tracing protocol to combine their complementary strengths and achieved even higher accuracy than each method alone. Additionally, the protocol is able to accurately assign the chain identity to the structure models of homo-multimers, which is not a trivial task for existing methods.

Dual function of OmpM as outer membrane tether and nutrient uptake channel in diderm Firmicutes

The outer membrane (OM) in diderm, or Gram-negative, bacteria must be tethered to peptidoglycan for mechanical stability and to maintain cell morphology. Most diderm phyla from the Terrabacteria group have recently been shown to lack well-characterised OM attachment systems, but instead have OmpM, which could represent an ancestral tethering system in bacteria. Here, we have determined the structure of the most abundant OmpM protein from Veillonella parvula (diderm Firmicutes) by single particle cryogenic electron microscopy. We also characterised the channel properties of the transmembrane β-barrel of OmpM and investigated the structure and PG-binding properties of its periplasmic stalk region. Our results show that OM tethering and nutrient acquisition are genetically linked in V. parvula, and probably other diderm Terrabacteria. This dual function of OmpM may have played a role in the loss of the OM in ancestral bacteria and the emergence of monoderm bacterial lineages.

Automated model building and protein identification in cryo-EM maps

Interpreting electron cryo-microscopy (cryo-EM) maps with atomic models requires high levels of expertise and labour-intensive manual intervention. We present ModelAngelo, a machine-learning approach for automated atomic model building in cryo-EM maps. By combining information from the cryo-EM map with information from protein sequence and structure in a single graph neural network, ModelAngelo builds atomic models for proteins that are of similar quality as those generated by human experts. For nucleotides, ModelAngelo builds backbones with similar accuracy as humans. By using its predicted amino acid probabilities for each residue in hidden Markov model sequence searches, ModelAngelo outperforms human experts in the identification of proteins with unknown sequences. ModelAngelo will thus remove bottlenecks and increase objectivity in cryo-EM structure determination.

DoubleHelix: nucleic acid sequence identification, assignment and validation tool for cryo-EM and crystal structure models

Sequence assignment is a key step of the model building process in both cryogenic electron microscopy (cryo-EM) and macromolecular crystallography (MX). If the assignment fails, it can result in difficult to identify errors affecting the interpretation of a model. There are many model validation strategies that help experimentalists in this step of protein model building, but they are virtually non-existent for nucleic acids. Here, I present doubleHelix-a comprehensive method for assignment, identification, and validation of nucleic acid sequences in structures determined using cryo-EM and MX. The method combines a neural network classifier of nucleobase identities and a sequence-independent secondary structure assignment approach. I show that the presented method can successfully assist sequence-assignment step in nucleic-acid model building at lower resolutions, where visual map interpretation is very difficult. Moreover, I present examples of sequence assignment errors detected using doubleHelix in cryo-EM and MX structures of ribosomes deposited in the Protein Data Bank, which escaped the scrutiny of available model-validation approaches. The doubleHelix program source code is available under BSD-3 license at https://gitlab.com/gchojnowski/doublehelix.

Large language models generate functional protein sequences across diverse families

Deep-learning language models have shown promise in various biotechnological applications, including protein design and engineering. Here we describe ProGen, a language model that can generate protein sequences with a predictable function across large protein families, akin to generating grammatically and semantically correct natural language sentences on diverse topics. The model was trained on 280 million protein sequences from >19,000 families and is augmented with control tags specifying protein properties. ProGen can be further fine-tuned to curated sequences and tags to improve controllable generation performance of proteins from families with sufficient homologous samples. Artificial proteins fine-tuned to five distinct lysozyme families showed similar catalytic efficiencies as natural lysozymes, with sequence identity to natural proteins as low as 31.4%. ProGen is readily adapted to diverse protein families, as we demonstrate with chorismate mutase and malate dehydrogenase.

Outer membrane utilisomes mediate glycan uptake in gut Bacteroidetes

Bacteroidetes are abundant members of the human microbiota, utilizing a myriad of diet- and host-derived glycans in the distal gut¹. Glycan uptake across the bacterial outer membrane of these bacteria is mediated by SusCD protein complexes, comprising a membrane-embedded barrel and a lipoprotein lid, which is thought to open and close to facilitate substrate binding and transport. However, surface-exposed glycan-binding proteins and glycoside hydrolases also play critical roles in the capture, processing and transport of large glycan chains. The interactions between these components in the outer membrane are poorly understood, despite being crucial for nutrient acquisition by our colonic microbiota. Here we show that for both the levan and dextran utilization systems of Bacteroides thetaiotaomicron, the additional outer membrane components assemble on the core SusCD transporter, forming stable glycan-utilizing machines that we term utilisomes. Single-particle cryogenic electron microscopy structures in the absence and presence of substrate reveal concerted conformational changes that demonstrate the mechanism of substrate capture, and rationalize the role of each component in the utilisome.

Mechanism of protein-primed template-independent DNA synthesis by Abi polymerases

Abortive infection (Abi) is a bacterial antiphage defense strategy involving suicide of the infected cell. Some Abi pathways involve polymerases that are related to reverse transcriptases. They are unique in the way they combine the ability to synthesize DNA in a template-independent manner with protein priming. Here, we report crystal and cryo-electron microscopy structures of two Abi polymerases: AbiK and Abi-P2. Both proteins adopt a bilobal structure with an RT-like domain that comprises palm and fingers subdomains and a unique helical domain. AbiK and Abi-P2 adopt a hexameric and trimeric configuration, respectively, which is unprecedented for reverse transcriptases. Biochemical experiments showed that the formation of these oligomers is required for the DNA polymerization activity. The structure of the AbiK–DNA covalent adduct visualized interactions between the 3′ end of DNA and the active site and covalent attachment of the 5′ end of DNA to a tyrosine residue used for protein priming. Our data reveal a structural basis of the mechanism of highly unusual template-independent protein-priming polymerases.

ModelCraft: an advanced automated model-building pipeline using Buccaneer

Interactive model building can be a difficult and time-consuming step in the structure-solution process. Automated model-building programs such as Buccaneer often make it quicker and easier by completing most of the model in advance. However, they may fail to do so with low-resolution data or a poor initial model or map. The Buccaneer pipeline is a relatively simple program that iterates Buccaneer with REFMAC to refine the model and update the map. A new pipeline called ModelCraft has been developed that expands on this to include shift-field refinement, machine-learned pruning of incorrect residues, classical density modification, addition of water and dummy atoms, building of nucleic acids and final rebuilding of side chains. Testing was performed on 1180 structures solved by experimental phasing, 1338 structures solved by molecular replacement using homologues and 2030 structures solved by molecular replacement using predicted AlphaFold models. Compared with the previous Buccaneer pipeline, ModelCraft increased the mean completeness of the protein models in the experimental phasing cases from 91% to 95%, the molecular-replacement cases from 50% to 78% and the AlphaFold cases from 82% to 91%.

Better, Faster, Cheaper: Recent Advances in Cryo-Electron Microscopy

Cryo-electron microscopy (cryo-EM) continues its remarkable growth as a method for visualizing biological objects, which has been driven by advances across the entire pipeline. Developments in both single-particle analysis and in situ tomography have enabled more structures to be imaged and determined to better resolutions, at faster speeds, and with more scientists having improved access. This review highlights recent advances at each stageof the cryo-EM pipeline and provides examples of how these techniques have been used to investigate real-world problems, including antibody development against the SARS-CoV-2 spike during the recent COVID-19 pandemic.

Cryo-EM structures of pentameric autoinducer-2 exporter from Escherichia coli reveal its transport mechanism

Bacteria utilize small extracellular molecules to communicate in order to collectively coordinate their behaviors in response to the population density. Autoinducer-2 (AI-2), a universal molecule for both intra- and inter-species communication, is involved in the regulation of biofilm formation, virulence, motility, chemotaxis, and antibiotic resistance. While many studies have been devoted to understanding the biosynthesis and sensing of AI-2, very little information is available on its export. The protein TqsA from Escherichia coli, which belongs to the AI-2 exporter superfamily, has been shown to export AI-2. Here, we report the cryogenic electron microscopic structures of two AI-2 exporters (TqsA and YdiK) from E. coli at 3.35 Å and 2.80 Å resolutions, respectively. Our structures suggest that the AI-2 exporter exists as a homo-pentameric complex. In silico molecular docking and native mass spectrometry experiments were employed to demonstrate the interaction between AI-2 and TqsA, and the results highlight the functional importance of two helical hairpins in substrate binding. We propose that each monomer works as an independent functional unit utilizing an elevator-type transport mechanism.

The structure of EXTL3 helps to explain the different roles of bi-domain exostosins in heparan sulfate synthesis

Heparan sulfate is a highly modified O-linked glycan that performs diverse physiological roles in animal tissues. Though quickly modified, it is initially synthesised as a polysaccharide of alternating β-D-glucuronosyl and N-acetyl-α-D-glucosaminyl residues by exostosins. These enzymes generally possess two glycosyltransferase domains (GT47 and GT64)-each thought to add one type of monosaccharide unit to the backbone. Although previous structures of murine exostosin-like 2 (EXTL2) provide insight into the GT64 domain, the rest of the bi-domain architecture is yet to be characterised; hence, how the two domains co-operate is unknown. Here, we report the structure of human exostosin-like 3 (EXTL3) in apo and UDP-bound forms. We explain the ineffectiveness of EXTL3's GT47 domain to transfer β-D-glucuronosyl units, and we observe that, in general, the bi-domain architecture would preclude a processive mechanism of backbone extension. We therefore propose that heparan sulfate backbone polymerisation occurs by a simple dissociative mechanism.

Sequence-assignment validation in cryo-EM models with checkMySequence

A new method, checkMySequence, for the fast and automated detection of register errors in protein models built into cryo-EM reconstructions is presented.

The availability of new artificial intelligence-based protein-structure-prediction tools has radically changed the way that cryo-EM maps are interpreted, but it has not eliminated the challenges of map interpretation faced by a microscopist. Models will continue to be locally rebuilt and refined using interactive tools. This inevitably results in occasional errors, among which register shifts remain one of the most difficult to identify and correct. Here, checkMySequence, a fast, fully automated and parameter-free method for detecting register shifts in protein models built into cryo-EM maps, is introduced. It is shown that the method can assist model building in cases where poorer map resolution hinders visual interpretation. It is also shown that checkMySequence could have helped to avoid a widely discussed sequence-register error in a model of SARS-CoV-2 RNA-dependent RNA polymerase that was originally detected thanks to a visual residue-by-residue inspection by members of the structural biology community. The software is freely available at https://gitlab.com/gchojnowski/checkmysequence.

Structure of the type V-C CRISPR-Cas effector enzyme

RNA-guided CRISPR-Cas nucleases are widely used as versatile genome-engineering tools. Recent studies identified functionally divergent type V Cas12 family enzymes. Among them, Cas12c2 binds a CRISPR RNA (crRNA) and a trans-activating crRNA (tracrRNA) and recognizes double-stranded DNA targets with a short TN PAM. Here, we report the cryo-electron microscopy structures of the Cas12c2-guide RNA binary complex and the Cas12c2-guide RNA-target DNA ternary complex. The structures revealed that the crRNA and tracrRNA form an unexpected X-junction architecture, and that Cas12c2 recognizes a single T nucleotide in the PAM through specific hydrogen-bonding interactions with two arginine residues. Furthermore, our biochemical analyses indicated that Cas12c2 processes its precursor crRNA to a mature crRNA using the RuvC catalytic site through a unique mechanism. Collectively, our findings improve the mechanistic understanding of diverse type V CRISPR-Cas effectors.

The human SKI complex regulates channeling of ribosome-bound RNA to the exosome via an intrinsic gatekeeping mechanism

The superkiller (SKI) complex is the cytoplasmic co-factor and regulator of the RNA-degrading exosome. In human cells, the SKI complex functions mainly in co-translational surveillance-decay pathways, and its malfunction is linked to a severe congenital disorder, the trichohepatoenteric syndrome. To obtain insights into the molecular mechanisms regulating the human SKI (hSKI) complex, we structurally characterized several of its functional states in the context of 80S ribosomes and substrate RNA. In a prehydrolytic ATP form, the hSKI complex exhibits a closed conformation with an inherent gating system that effectively traps the 80S-bound RNA into the hSKI2 helicase subunit. When active, hSKI switches to an open conformation in which the gating is released and the RNA 3′ end exits the helicase. The emerging picture is that the gatekeeping mechanism and architectural remodeling of hSKI underpin a regulated RNA channeling system that is mechanistically conserved among the cytoplasmic and nuclear helicase-exosome complexes.

•

hSKI has closed and open states connected to different helicase conformations

•

The intrinsic closed state traps the RNA 3′ end and blocks the RNA exit path

•

ATP induces the open state of hSKI, allowing 80S ribosome-bound RNA extraction

•

The hSKI open state primes hSKI2 for channeling RNA to the cytosolic exosome

CryoEM analysis of small plant biocatalysts at sub-2 Å resolution

Enzyme catalysis has emerged as a key technology for developing efficient, sustainable processes in the chemical, biotechnological and pharmaceutical industries. Plants provide large and diverse pools of biosynthetic enzymes that facilitate complex reactions, such as the formation of intricate terpene carbon skeletons, with exquisite specificity. High-resolution structural analysis of these enzymes is crucial in order to understand their mechanisms and modulate their properties by targeted engineering. Although cryo-electron microscopy (cryoEM) has revolutionized structural biology, its applicability to high-resolution structural analysis of comparatively small enzymes has so far been largely unexplored. Here, it is shown that cryoEM can reveal the structures of plant borneol dehydrogenases of ∼120 kDa at or below 2 Å resolution, paving the way for the rapid development of new biocatalysts that can provide access to bioactive terpenes and terpenoids.

findMySequence: a neural-network-based approach for identification of unknown proteins in X-ray crystallography and cryo-EM

Although experimental protein-structure determination usually targets known proteins, chains of unknown sequence are often encountered. They can be purified from natural sources, appear as an unexpected fragment of a well characterized protein or appear as a contaminant. Regardless of the source of the problem, the unknown protein always requires characterization. Here, an automated pipeline is presented for the identification of protein sequences from cryo-EM reconstructions and crystallographic data. The method's application to characterize the crystal structure of an unknown protein purified from a snake venom is presented. It is also shown that the approach can be successfully applied to the identification of protein sequences and validation of sequence assignments in cryo-EM protein structures.

Redeployment of automated MrBUMP search-model identification for map fitting in cryo-EM

In crystallography, the phase problem can often be addressed by the careful preparation of molecular-replacement search models. This has led to the development of pipelines such as MrBUMP that can automatically identify homologous proteins from an input sequence and edit them to focus on the areas that are most conserved. Many of these approaches can be applied directly to cryo-EM to help discover, prepare and correctly place models (here called cryo-EM search models) into electrostatic potential maps. This can significantly reduce the amount of manual model building that is required for structure determination. Here, MrBUMP is repurposed to fit automatically obtained PDB-derived chains and domains into cryo-EM maps. MrBUMP was successfully able to identify and place cryo-EM search models across a range of resolutions. Methods such as map segmentation are also explored as potential routes to improved performance. Map segmentation was also found to improve the effectiveness of the pipeline for higher resolution (<8 Å) data sets.

Cryo-EM single-particle structure refinement and map calculation using Servalcat

In 2020, cryo-EM single-particle analysis achieved true atomic resolution thanks to technological developments in hardware and software. The number of high-resolution reconstructions continues to grow, increasing the importance of the accurate determination of atomic coordinates. Here, a new Python package and program called Servalcat is presented that is designed to facilitate atomic model refinement. Servalcat implements a refinement pipeline using the program REFMAC5 from the CCP4 package. After the refinement, Servalcat calculates a weighted Fo - Fc difference map, which is derived from Bayesian statistics. This map helps manual and automatic model building in real space, as is common practice in crystallography. The Fo - Fc map helps in the visualization of weak features including hydrogen densities. Although hydrogen densities are weak, they are stronger than in the electron-density maps produced by X-ray crystallography, and some H atoms are even visible at ∼1.8 Å resolution. Servalcat also facilitates atomic model refinement under symmetry constraints. If point-group symmetry has been applied to the map during reconstruction, the asymmetric unit model is refined with the appropriate symmetry constraints.

Cryo-EM Map-Based Model Validation Using the False Discovery Rate Approach

Significant technological developments and increasing scientific interest in cryogenic electron microscopy (cryo-EM) has resulted in a rapid increase in the amount of data generated by these experiments and the derived atomic models. Robust measures for the validation of 3D reconstructions and atomic models are essential for appropriate interpretation of the data. The resolution of data and availability of software tools that work across a range of resolutions often limit the quality of derived models. Hence, the final atomic model is often incomplete or contains regions where atomic positions are less reliable or incorrectly built. Extensive manual pruning and local adjustments or rebuilding are usually required to address these issues. The presented research introduces a software tool for the validation of the backbone trace of atomic models built in the cryo-EM density maps. In this study, we use the false discovery rate analysis, which can be used to segregate molecular signals from the background. Each atomic position in the model can be associated with an FDR backbone validation score, which can be used to identify potential mistraced residues. We demonstrate that the proposed validation score is complementary to existing validation metrics and is useful especially in cases where the model is built in the maps having varying local resolution. We also discuss the application of the score for automated pruning of atomic models built ab-initio during the iterative model building process in Buccaneer. We have implemented this score in the CCP-EM software suite.

Cryo-EM model validation recommendations based on outcomes of the 2019 EMDataResource challenge

This paper describes outcomes of the 2019 Cryo-EM Model Challenge. The goals were to (1) assess the quality of models that can be produced from cryogenic electron microscopy (cryo-EM) maps using current modeling software, (2) evaluate reproducibility of modeling results from different software developers and users and (3) compare performance of current metrics used for model evaluation, particularly Fit-to-Map metrics, with focus on near-atomic resolution. Our findings demonstrate the relatively high accuracy and reproducibility of cryo-EM models derived by 13 participating teams from four benchmark maps, including three forming a resolution series (1.8 to 3.1 Å). The results permit specific recommendations to be made about validating near-atomic cryo-EM structures both in the context of individual experiments and structure data archives such as the Protein Data Bank. We recommend the adoption of multiple scoring parameters to provide full and objective annotation and assessment of the model, reflective of the observed cryo-EM map density.