Acquired immunodeficiency syndrome (AIDS) is caused by human immunodeficiency virus (HIV). HIV protease, reverse transcriptase, and integrase are targets of current drugs to treat the disease. However, anti-viral drug-resistant strains have emerged quickly due to the high mutation rate of the virus, leading to the demand for the development of new drugs. One attractive target is Gag-Pol polyprotein, which plays a key role in the life cycle of HIV. Recently, we found that a combination of M50I and V151I mutations in HIV-1 integrase can suppress virus release and inhibit the initiation of Gag-Pol autoprocessing and maturation without interfering with the dimerization of Gag-Pol. Additional mutations in integrase or RNase H domain in reverse transcriptase can compensate for the defect. However, the molecular mechanism is unknown. There is no tertiary structure of the full-length HIV-1 Pol protein available for further study. Therefore, we developed a workflow to predict the tertiary structure of HIV-1 NL4.3 Pol polyprotein. The modeled structure has comparable quality compared with the recently published partial HIV-1 Pol structure (PDB ID: 7SJX). Our HIV-1 NL4.3 Pol dimer model is the first full-length Pol tertiary structure. It can provide a structural platform for studying the autoprocessing mechanism of HIV-1 Pol and for developing new potent drugs. Moreover, the workflow can be used to predict other large protein structures that cannot be resolved via conventional experimental methods.

De novo protein design enhances our understanding of the principles that govern protein folding and interactions, and has the potential to revolutionize biotechnology through the engineering of novel protein functionalities. Despite recent progress in computational design strategies, de novo design of protein structures remains challenging, given the vast size of the sequence-structure space. AlphaFold2 (AF2), a state-of-the-art neural network architecture, achieved remarkable accuracy in predicting protein structures from amino acid sequences. This raises the question whether AF2 has learned the principles of protein folding sufficiently for de novo design. Here, we sought to answer this question by inverting the AF2 network, using the prediction weight set and a loss function to bias the generated sequences to adopt a target fold. Initial design trials resulted in de novo designs with an overrepresentation of hydrophobic residues on the protein surface compared to their natural protein family, requiring additional surface optimization. In silico validation of the designs showed protein structures with the correct fold, a hydrophilic surface and a densely packed hydrophobic core. In vitro validation showed that 7 out of 39 designs were folded and stable in solution with high melting temperatures. In summary, our design workflow solely based on AF2 does not seem to fully capture basic principles of de novo protein design, as observed in the protein surface\'s hydrophobic vs. hydrophilic patterning. However, with minimal post-design intervention, these pipelines generated viable sequences as assessed experimental characterization. Thus, such pipelines show the potential to contribute to solving outstanding challenges in de novo protein design.

Deeper understanding of T-cell-mediated adaptive immune responses is important for the design of cancer immunotherapies and antiviral vaccines against pandemic outbreaks. T-cells are activated when they recognize foreign peptides that are presented on the cell surface by Major Histocompatibility Complexes (MHC), forming peptide:MHC (pMHC) complexes. 3D structures of pMHC complexes provide fundamental insight into T-cell recognition mechanism and aids immunotherapy design. High MHC and peptide diversities necessitate efficient computational modelling to enable whole proteome structural analysis. We developed PANDORA, a generic modelling pipeline for pMHC class I and II (pMHC-I and pMHC-II), and present its performance on pMHC-I here. Given a query, PANDORA searches for structural templates in its extensive database and then applies anchor restraints to the modelling process. This restrained energy minimization ensures one of the fastest pMHC modelling pipelines so far. On a set of 835 pMHC-I complexes over 78 MHC types, PANDORA generated models with a median RMSD of 0.70 Å and achieved a 93% success rate in top 10 models. PANDORA performs competitively with three pMHC-I modelling state-of-the-art approaches and outperforms AlphaFold2 in terms of accuracy while being superior to it in speed. PANDORA is a modularized and user-configurable python package with easy installation. We envision PANDORA to fuel deep learning algorithms with large-scale high-quality 3D models to tackle long-standing immunology challenges.

Allergy is becoming an intensifying disease among the world population, particularly in the developed world. Once allergy develops, sufferers are permanently trapped in a hyper-immune response that makes them sensitive to innocuous substances. The immune pathway concerned with developing allergy is the Th2 immune pathway where the IgE antibody binds to its Fc ∊ RI receptor on Mast and Basophil cells. This paper discusses a protocol that could disrupt the binding between the antibody and its receptor for a potential permanent treatment. Ten proteins were computationally designed to display a human IgE motif very close in proximity to the IgE antibody\'s Fc ∊ RI receptor\'s binding site in an effort for these proteins to be used as a vaccine against our own IgE antibody. The motif of interest was the FG loop motif and it was excised and grafted onto a Staphylococcus aureus protein (PDB ID 1YN3), then the motif + scaffold structure had its sequence re-designed around the motif to find an amino acid sequence that would fold to the designed structure correctly. These ten computationally designed proteins showed successful folding when simulated using Rosetta\'s AbinitioRelax folding simulation and the IgE epitope was clearly displayed in its native three-dimensional structure in all of them. These designed proteins have the potential to be used as a pan anti-allergy vaccine. This work employedin silicobased methods for designing the proteins and did not include any experimental verifications.

Best vitelliform macular dystrophy (BVMD) is an autosomal dominant macular degeneration. The typical central yellowish yolk-like lesion usually appears in childhood and gradually worsens. Most cases are caused by variants in the BEST1 gene which encodes bestrophin-1, an integral membrane protein found primarily in the retinal pigment epithelium.
Here we describe the spectrum of BEST1 variants identified in a cohort of 57 Italian patients analyzed by Sanger sequencing. In 13 cases, the study also included segregation analysis in affected and unaffected relatives. We used molecular mechanics to calculate two quantitative parameters related to calcium-activated chloride channel (CaCC composed of 5 BEST1 subunits) stability and calcium-dependent activation and related them to the potential pathogenicity of individual missense variants detected in the probands.
Thirty-six out of 57 probands (63% positivity) and 16 out of 18 relatives proved positive to genetic testing. Family study confirmed the variable penetrance and expressivity of the disease. Six of the 27 genetic variants discovered were novel: p.(Val9Gly), p.(Ser108Arg), p.(Asn179Asp), p.(Trp182Arg), p.(Glu292Gln) and p.(Asn296Lys). All BEST1 variants were assessed in silico for potential pathogenicity. Our computational structural biology approach based on 3D model structure of the CaCC showed that individual amino acid replacements may affect channel shape, stability, activation, gating, selectivity and throughput, and possibly also other features, depending on where the individual mutated amino acid residues are located in the tertiary structure of BEST1. Statistically significant correlations between mean logMAR best-corrected visual acuity (BCVA), age and modulus of computed BEST1 dimerization energies, which reflect variations in the in CaCC stability due to amino acid changes, permitted us to assess the pathogenicity of individual BEST1 variants.
Using this computational approach, we designed a method for estimating BCVA progression in patients with BEST1 variants.

Bioinformatics is a very resourceful tool to understand evolution of membrane proteins, such as transient receptor potential channels. Expert bioinformatics users rely on specialized scripting and programming skills. Several web servers and standalone tools are available for nonadvanced users willing to develop projects to understand their system of choice. In this case, we present a desktop-based protocol to develop evostructural hypotheses based on basic bioinformatics skills and resources, specifically for a small subgroup of TRPV channels, which can be further implemented for larger datasets.

Cryo-electron microscopy (cryo-EM) is becoming the imaging method of choice for determining protein structures. Many atomic structures have been resolved based on an exponentially growing number of published three-dimensional (3D) high resolution cryo-EM density maps. However, the resolution value claimed for the reconstructed 3D density map has been the topic of scientific debate for many years. The Fourier Shell Correlation (FSC) is the currently accepted cryo-EM resolution measure, but it can be subjective, manipulated, and has its own limitations. In this study, we first propose supervised deep learning methods to extract representative 3D features at high, medium and low resolutions from simulated protein density maps and build classification models that objectively validate resolutions of experimental 3D cryo-EM maps. Specifically, we build classification models based on dense artificial neural network (DNN) and 3D convolutional neural network (3D CNN) architectures. The trained models can classify a given 3D cryo-EM density map into one of three resolution levels: high, medium, low. The preliminary DNN and 3D CNN models achieved 92.73% accuracy and 99.75% accuracy on simulated test maps, respectively. Applying the DNN and 3D CNN models to thirty experimental cryo-EM maps achieved an agreement of 60.0% and 56.7%, respectively, with the author published resolution value of the density maps. We further augment these previous techniques and present preliminary results of a 3D U-Net model for local resolution classification. The model was trained to perform voxel-wise classification of 3D cryo-EM density maps into one of ten resolution classes, instead of a single global resolution value. The U-Net model achieved 88.3% and 94.7% accuracy when evaluated on experimental maps with local resolutions determined by MonoRes and ResMap methods, respectively. Our results suggest deep learning can potentially improve the resolution evaluation process of experimental cryo-EM maps.

Chemical crosslinking can identify the neighborhood relationships between specific amino-acid residues in proteins. The interpretation of crosslinking data is typically performed using single, static atomic structures. However, proteins are dynamic, undergoing motions spanning from local fluctuations of individual residues to global motions of protein assemblies. Here we demonstrate that failure to explicitly accommodate dynamics when interpreting crosslinks structurally can lead to considerable errors. We present a method and associated software, DynamXL, which is able to account directly for flexibility in the context of crosslinking modeling. Our benchmarking on a large dataset of model structures demonstrates significantly improved rationalization of experimental crosslinking data, and enhanced performance in a protein-protein docking protocol. These advances will provide a considerable increase in the structural insights attainable using chemical crosslinking coupled to mass spectrometry.

Despite the recent success of newly developed direct-acting antivirals against hepatitis C, the disease continues to be a global health threat due to the lack of diagnosis of most carriers and the high cost of treatment. The heterodimer formed by glycoproteins E1 and E2 within the hepatitis C virus (HCV) lipid envelope is a potential vaccine candidate and antiviral target. While the structure of E1/E2 has not yet been resolved, partial crystal structures of the E1 and E2 ectodomains have been determined. The unresolved parts of the structure are within the realm of what can be modeled with current computational modeling tools. Furthermore, a variety of additional experimental data is available to support computational predictions of E1/E2 structure, such as data from antibody binding studies, cryo-electron microscopy (cryo-EM), mutational analyses, peptide binding analysis, linker-scanning mutagenesis, and nuclear magnetic resonance (NMR) studies. In accordance with these rich experimental data, we have built an in silico model of the full-length E1/E2 heterodimer. Our model supports that E1/E2 assembles into a trimer, which was previously suggested from a study by Falson and coworkers (P. Falson, B. Bartosch, K. Alsaleh, B. A. Tews, A. Loquet, Y. Ciczora, L. Riva, C. Montigny, C. Montpellier, G. Duverlie, E. I. Pecheur, M. le Maire, F. L. Cosset, J. Dubuisson, and F. Penin, J. Virol. 89:10333-10346, 2015, https://doi.org/10.1128/JVI.00991-15). Size exclusion chromatography and Western blotting data obtained by using purified recombinant E1/E2 support our hypothesis. Our model suggests that during virus assembly, the trimer of E1/E2 may be further assembled into a pentamer, with 12 pentamers comprising a single HCV virion. We anticipate that this new model will provide a useful framework for HCV envelope structure and the development of antiviral strategies.IMPORTANCE One hundred fifty million people have been estimated to be infected with hepatitis C virus, and many more are at risk for infection. A better understanding of the structure of the HCV envelope, which is responsible for attachment and fusion, could aid in the development of a vaccine and/or new treatments for this disease. We draw upon computational techniques to predict a full-length model of the E1/E2 heterodimer based on the partial crystal structures of the envelope glycoproteins E1 and E2. E1/E2 has been widely studied experimentally, and this provides valuable data, which has assisted us in our modeling. Our proposed structure is used to suggest the organization of the HCV envelope. We also present new experimental data from size exclusion chromatography that support our computational prediction of a trimeric oligomeric state of E1/E2.

Research projects in structural biology increasingly rely on combinations of heterogeneous sources of information, e.g. evolutionary information from multiple sequence alignments, experimental evidence in the form of density maps and proximity constraints from proteomics experiments. The OpenStructure software framework, which allows the seamless integration of information of different origin, has previously been introduced. The software consists of C++ libraries which are fully accessible from the Python programming language. Additionally, the framework provides a sophisticated graphics module that interactively displays molecular structures and density maps in three dimensions. In this work, the latest developments in the OpenStructure framework are outlined. The extensive capabilities of the framework will be illustrated using short code examples that show how information from molecular-structure coordinates can be combined with sequence data and/or density maps. The framework has been released under the LGPL version 3 license and is available for download from http://www.openstructure.org.