Toward Modeling Protein Multimers by Combining AlphaFold 3 Predictions with Secondary Structures from Medium-Resolution Cryo-EM Maps.

A comprehensive, unbiased inventory of synuclein forms present in Lewy bodies from patients with dementia with Lewy bodies was carried out using two-dimensional immunoblot analysis, novel sandwich enzyme-linked immunosorbent assays with modification-specific synuclein antibodies, and mass spectroscopy. The predominant modification of alpha-synuclein in Lewy bodies is a single phosphorylation at Ser-129. In addition, there is a set of characteristic modifications that are present to a lesser extent, including ubiquitination at Lys residues 12, 21, and 23 and specific truncations at Asp-115, Asp-119, Asn-122, Tyr-133, and Asp-135. No other modifications are detectable by tandem mass spectrometry mapping, except for a ubiquitous N-terminal acetylation. Small amounts of Ser-129 phosphorylated and Asp-119-truncated alpha-synuclein are present in the soluble fraction of both normal and disease brains, suggesting that these Lewy body-associated forms are produced during normal metabolism of alpha-synuclein. In contrast, ubiquitination is only detected in Lewy bodies and is primarily present on phosphorylated synuclein; it therefore likely occurs after phosphorylated synuclein has deposited into Lewy bodies. This invariant pattern of specific phosphorylation, truncation, and ubiquitination is also present in the detergent-insoluble fraction of brain from patients with familial Parkinson's disease (synuclein A53T mutation) as well as multiple system atrophy, suggesting a common pathogenic pathway for both genetic and sporadic Lewy body diseases. These observations are most consistent with a model in which preferential accumulation of normally produced Ser-129 phosphorylated alpha-synuclein is the key event responsible for the formation of Lewy bodies in various Lewy body diseases.

Although atomic structures have been determined directly from cryo-EM density maps with high resolutions, current structure determination methods for medium resolution (5 to 10 Å) cryo-EM maps are limited by the availability of structure templates. Secondary structure traces are lines detected from a cryo-EM density map for α-helices and β-strands of a protein. A topology of secondary structures defines the mapping between a set of sequence segments and a set of traces of secondary structures in three-dimensional space. In order to enhance accuracy in ranking secondary structure topologies, we explored a method that combines three sources of information: a set of sequence segments in 1D, a set of amino acid contact pairs in 2D, and a set of traces in 3D at the secondary structure level. A test of fourteen cases shows that the accuracy of predicted secondary structures is critical for deriving topologies. The use of significant long-range contact pairs is most effective at enriching the rank of the maximum-match topology for proteins with a large number of secondary structures, if the secondary structure prediction is fairly accurate. It was observed that the enrichment depends on the quality of initial topology candidates in this approach. We provide detailed analysis in various cases to show the potential and challenge when combining three sources of information.

The SEQUEST program was the first and remains one of the most widely used tools for assigning a peptide sequence within a database to a tandem mass spectrum. The cross correlation score is the primary score function implemented within SEQUEST and it is this score that makes the tool particularly sensitive. Unfortunately, this score is computationally expensive to calculate, and thus, to make the score manageable, SEQUEST uses a less sensitive but fast preliminary score and restricts the cross correlation to just the top 500 peptides returned by the preliminary score. Classically, the cross correlation score has been calculated using Fast Fourier Transforms (FFT) to generate the full correlation function. We describe an alternate method of calculating the cross correlation score that does not require FFTs and can be computed efficiently in a fraction of the time. The fast calculation allows all candidate peptides to be scored by the cross correlation function, potentially mitigating the need for the preliminary score, and enables an E-value significance calculation based on the cross correlation score distribution calculated on all candidate peptide sequences obtained from a sequence database.

Cryo-electron microscopy is a major structure determination technique for large molecular machines and membrane-associated complexes. Although atomic structures have been determined directly from cryo-EM density maps with high resolutions, current structure determination methods for medium resolution (5 to 10 Å) cryo-EM maps are limited by the availability of structure templates. Secondary structure traces are lines detected from a cryo-EM density map for α-helices and β-strands of a protein. When combined with secondary structure sequence segments predicted from a protein sequence, it is possible to generate a set of likely topologies of α-traces and β-sheet traces. A topology describes the overall folding relationship among secondary structures; it is a critical piece of information for deriving the corresponding atomic structure. We propose a method for protein structure prediction that combines three sources of information: the secondary structure traces detected from the cryo-EM density map, predicted secondary structure sequence segments, and amino acid contact pairs predicted using MULTICOM. A case study shows that using amino acid contact prediction from MULTICOM improves the ranking of the true topology. Our observations convey that using a small set of highly voted secondary structure contact pairs enhances the ranking in all experiments conducted for this case.

Electron cryo-microscopy (cryo-EM) as a cutting edge technology has carved a niche for itself in the study of large-scale protein complex. Although the protein backbone of complexes cannot be derived directly from the medium resolution (5-10 A) of amino acids from three-dimensional (3D) density images, secondary structure elements (SSEs) such as alpha-helices and beta-sheets can still be detected. The accuracy of SSE detection from the volumetric protein density images is critical for ab initio backbone structure derivation in cryo-EM. So far it is challenging to detect the SSEs automatically and accurately from the density images at these resolutions. This dissertation presents four computational methods - SSEtracer, SSElearner, StrandTwister and StrandRoller for solving this critical problem. An effective approach, SSEtracer, is presented to automatically identify helices and β- sheets from the cryo-EM three-dimensional maps at medium resolutions. A simple mathematical model is introduced to represent the β-sheet density. The mathematical model can be used for β-strand detection from medium resolution density maps. A machine learning approach, SSElearner, has also been developed to automatically identify helices and β-sheets by using the knowledge from existing volumetric maps in the Electron Microscopy Data Bank (EMDB). The approach has been tested using simulated density maps and experimental cryo-EM maps of EMDB. The results of SSElearner suggest that it is effective to use one cryo-EM map for learning in order to detect the SSE in another cryo-EM map of similar quality. Major secondary structure elements such as a-helices and β-sheets can be computationally detected from cryo-EM density maps with medium resolutions of 5-10A. However, a critical piece of information for modeling atomic structures is missing, since there are no tools to detect β-strands from cryo-EM maps at medium resolutions. A new method, StrandTwister, has been proposed to detect the traces of β-strands through the analysis of twist, an intrinsic nature of β-sheet. StrandTwister has been tested using 100 β-sheets simulated at 10A resolution and 39 β-sheets computationally detected from cryoEM density maps at 4.4-7.4A resolutions. StrandTwister appears to detect the traces of β-strands on major β-sheets quite accurately, particularly at the central area of a β-sheet. β-barrel is a structure feature that is formed by multiple β-strands in a barrel shape. There is no existing method to derive the β-strands from the 3D image of β-barrel. A new method, StrandRoller, has been proposed to generate small sets of possible β-traces from the density images at medium resolutions of 5-10A. The results of StrandRoller suggest that it is possible to derive a small set of possible β-traces from the β-barrel cryo-EM image at medium resolutions even when it is not possible to visualize the separation of β-strands.

Building an accurate atomic structure model of a protein into a cryo-electron microscopy (cryo-EM) map at worse than 3 Å resolution is difficult. To facilitate this task, we devised a method for assigning the amino acid residue sequence to the backbone fragments traced in an input cryo-EM map (EMSequenceFinder). EMSequenceFinder relies on a Bayesian scoring function for ranking 20 standard amino acid residue types at a given backbone position, based on the fit to a density map, map resolution, and secondary structure propensity. The fit to a density is quantified by a convolutional neural network that was trained on ~5.56 million amino acid residue densities extracted from cryo-EM maps at 3-10 Å resolution and corresponding atomic structure models deposited in the Electron Microscopy Data Bank (EMDB). We benchmarked EMSequenceFinder by predicting the sequences of 58,044 distinct α-helix and β-strand fragments, given the fragment backbone coordinates fitted in their density maps. EMSequenceFinder identifies the correct sequence as the best-scoring sequence in 77.8% of these cases. We also assessed EMSequenceFinder on separate datasets of cryo-EM maps at resolutions from 4 to 6. The accuracy of EMSequenceFinder (63.5%) was better than that of three tested state-of-the-art methods, including findMysequence (45%), ModelAngelo (27%), and sequence_from_map in Phenix (12.9%). We further illustrate EMSequenceFinder by threading the SARS-CoV-2 NSP2 sequence into eight cryo-EM maps at resolutions from 3.7 to 7.0 Å. EMSequenceFinder is implemented in our open-source Integrative Modeling Platform (IMP) program. Thus, it is expected to be helpful for integrative structure modeling based on a cryo-EM map and other information, such as models of protein complex components and chemical crosslinks between them. EMSequenceFinder is available as part of our open source IMP distribution at https://integrativemodeling.org/.

Building an accurate atomic structure model of a protein into a cryo-electron microscopy (cryo-EM) map at worse than 3 Å resolution is difficult. To facilitate this task, we devised a method for assigning the amino acid residue sequence to the backbone fragments traced in an input cryo-EM map (EMSequenceFinder). EMSequenceFinder relies on a Bayesian scoring function for ranking 20 standard amino acid residue types at a given backbone position, based on the fit to a density map, map resolution, and secondary structure propensity. The fit to a density is quantified by a convolutional neural network that was trained on ~5.56 million amino acid residue densities extracted from cryo-EM maps at 3-10 Å resolution and corresponding atomic structure models deposited in the Electron Microscopy Data Bank (EMDB). We benchmarked EMSequenceFinder by predicting the sequences of 58,044 distinct ɑ-helix and β-strand fragments, given the fragment backbone coordinates fitted in their density maps. EMSequenceFinder identifies the correct sequence as the best-scoring sequence in 77.8% of these cases. We also assessed EMSequenceFinder on separate datasets of cryo-EM maps at resolutions from 4 to 6 Å. The accuracy of EMSequenceFinder (58%) was better than that of three tested state-of-the-art methods, including findMysequence (45%), ModelAngelo (27%), and sequence_from_map in Phenix (12.9%). We further illustrate EMSequenceFinder by threading the Severe Acute Respiratory Syndrome Coronavirus 2 Non-Structural Protein 2 sequence into eight cryo-EM maps at resolutions from 3.7 to 7.0 Å. EMSequenceFinder is implemented in our open-source Integrative Modeling Platform (IMP) program. Thus, it is expected to be helpful for integrative structure modeling based on a cryo-EM map and other information, such as models of protein complex components and chemical crosslinks between them. EMSequenceFinder is available as part of our open-source IMP distribution at https://integrativemodeling.org/.

The accuracy of the secondary structure element (SSE) identification from volumetric protein density maps is critical for de-novo backbone structure derivation in electron cryo-microscopy (cryoEM). It is still challenging to detect the SSE automatically and accurately from the density maps at medium resolutions (∼5-10 Å). We present a machine learning approach, SSELearner, to automatically identify helices and β-sheets by using the knowledge from existing volumetric maps in the Electron Microscopy Data Bank. We tested our approach using 10 simulated density maps. The averaged specificity and sensitivity for the helix detection are 94.9% and 95.8%, respectively, and those for the β-sheet detection are 86.7% and 96.4%, respectively. We have developed a secondary structure annotator, SSID, to predict the helices and β-strands from the backbone Cα trace. With the help of SSID, we tested our SSELearner using 13 experimentally derived cryo-EM density maps. The machine learning approach shows the specificity and sensitivity of 91.8% and 74.5%, respectively, for the helix detection and 85.2% and 86.5% respectively for the β-sheet detection in cryoEM maps of Electron Microscopy Data Bank. The reduced detection accuracy reveals the challenges in SSE detection when the cryoEM maps are used instead of the simulated maps. Our results suggest that it is effective to use one cryoEM map for learning to detect the SSE in another cryoEM map of similar quality.

Although cryo-electron microscopy (cryo-EM) has been successfully used to derive atomic structures for many proteins, it is still challenging to derive atomic structures when the resolution of cryo-EM density maps is in the medium resolution range, such as 5–10 Å. Detection of protein secondary structures, such as helices and β-sheets, from cryo-EM density maps provides constraints for deriving atomic structures from such maps. As more deep learning methodologies are being developed for solving various molecular problems, effective tools are needed for users to access them. We have developed an effective software bundle, DeepSSETracer, for the detection of protein secondary structure from cryo-EM component maps in medium resolution. The bundle contains the network architecture and a U-Net model trained with a curriculum and gradient of episodic memory (GEM). The bundle integrates the deep neural network with the visualization capacity provided in ChimeraX. Using a Linux server that is remotely accessed by Windows users, it takes about 6 s on one CPU and one GPU for the trained deep neural network to detect secondary structures in a cryo-EM component map containing 446 amino acids. A test using 28 chain components of cryo-EM maps shows overall residue-level F1 scores of 0.72 and 0.65 to detect helices and β-sheets, respectively. Although deep learning applications are built on software frameworks, such as PyTorch and Tensorflow, our pioneer work here shows that integration of deep learning applications with ChimeraX is a promising and effective approach. Our experiments show that the F1 score measured at the residue level is an effective evaluation of secondary structure detection for individual classes. The test using 28 cryo-EM component maps shows that DeepSSETracer detects β-sheets more accurately than Emap2sec+, with a weighted average residue-level F1 score of 0.65 and 0.42, respectively. It also shows that Emap2sec+ detects helices more accurately than DeepSSETracer with a weighted average residue-level F1 score of 0.77 and 0.72 respectively.

Cryo-EM has revealed the structures of many challenging yet exciting macromolecular assemblies at near-atomic resolution (3–4.5Å), providing biological phenomena with molecular descriptions. However, at these resolutions, accurately positioning individual atoms remains challenging and error-prone. Manually refining thousands of amino acids – typical in a macromolecular assembly – is tedious and time-consuming. We present an automated method that can improve the atomic details in models that are manually built in near-atomic-resolution cryo-EM maps. Applying the method to three systems recently solved by cryo-EM, we are able to improve model geometry while maintaining the fit-to-density. Backbone placement errors are automatically detected and corrected, and the refinement shows a large radius of convergence. The results demonstrate that the method is amenable to structures with symmetry, of very large size, and containing RNA as well as covalently bound ligands. The method should streamline the cryo-EM structure determination process, providing accurate and unbiased atomic structure interpretation of such maps.DOI: http://dx.doi.org/10.7554/eLife.17219.001

Maturation of dsDNA bacteriophages involves assembling the virus prohead from a limited set of structural components followed by rearrangements required for the stability that is necessary for infecting a host under challenging environmental conditions. Here, we determine the mature capsid structure of T7 at 1 nm resolution by cryo-electron microscopy and compare it with the prohead to reveal the molecular basis of T7 shell maturation. The mature capsid presents an expanded and thinner shell, with a drastic rearrangement of the major protein monomers that increases in their interacting surfaces, in turn resulting in a new bonding lattice. The rearrangements include tilting, in-plane rotation, and radial expansion of the subunits, as well as a relative bending of the A- and P-domains of each subunit. The unique features of this shell transformation, which does not employ the accessory proteins, inserted domains, or molecular interactions observed in other phages, suggest a simple capsid assembling strategy that may have appeared early in the evolution of these viruses.

Comparison of three similarity scores for bullet LEA matching

Deep learning-based local quality estimation for protein structure models from cryo-EM maps

Rapid advancements in cryogenic electron microscopy (cryo-EM) have revolutionized the field of structural biology by enabling the determination of complex macromolecular structures at unprecedented resolutions. When cryo-EM density maps have a resolution around 3Å, the atomic structure can be modeled manually. However, as the resolution decreases, analyzing these density maps becomes increasingly challenging. For modeling structures in lower resolution maps, deep learning can be used to identify structural features in the maps to assist in structure modeling.Here, we present a suite of deep learning-based tools developed by our lab that enable structural biologists to work with cryo-EM maps of a wide range of resolutions. For cryo-EM maps at near-atomic resolution (5Å or better), DeepMainmast automatically models all-atom structures by tracing the main chain from local map features of amino acids and atoms detected by deep learning; DAQ score quantifies map-model fit and indicates potential misassignments in protein models. In intermediate resolution maps (5-10Å), Emap2sec and Emap2sec+ can accurately detect protein secondary structures and nucleic acids. These tools and more are available at our web server: https://em.kiharalab.org/ .

Advances in microscopy instruments and image processing algorithms have led to an increasing number of Cryo-electron microscopy (cryo-EM) maps. However, building accurate models for the EM maps at 3-5 Å resolution remains a challenging and time-consuming process. With the rapid growth of deposited EM maps, there is an increasing gap between the maps and reconstructed/modeled three-dimensional (3D) structures. Therefore, automatic reconstruction of atomic-accuracy full-atom structures from EM maps is pressingly needed. We present a semi-automatic de novo structure determination method using a deep learning-based framework, named as DeepMM, which builds atomic-accuracy all-atom models from cryo-EM maps at near-atomic resolution. In our method, the main-chain and Cα positions as well as their amino acid and secondary structure types are predicted in the EM map using Densely Connected Convolutional Networks. DeepMM was extensively validated on 40 simulated maps at 5 Å resolution and 30 experimental maps at 2.6-4.8 Å resolution as well as an Electron Microscopy Data Bank-wide dataset of 2931 experimental maps at 2.6-4.9 Å resolution, and compared with state-of-the-art algorithms including RosettaES, MAINMAST and Phenix. Overall, our DeepMM algorithm obtained a significant improvement over existing methods in terms of both accuracy and coverage in building full-length protein structures on all test sets, demonstrating the efficacy and general applicability of DeepMM. http://huanglab.phys.hust.edu.cn/DeepMM. Supplementary data are available at Bioinformatics online.