AlphaFold2 for Protein Structure Prediction: Best Practices and Critical Analyses.

AlphaFold2 (AF2) has emerged in recent years as a groundbreaking innovation that has revolutionized several scientific fields, in particular structural biology, drug design, and the elucidation of disease mechanisms. Many scientists now use AF2 on a daily basis, including non-specialist users. This chapter is aimed at the latter. Tips and tricks for getting the most out of AF2 to produce a high-quality biological model are discussed here. We suggest to non-specialist users how to maintain a critical perspective when working with AF2 models and provide guidelines on how to properly evaluate them. After showing how to perform our own structure prediction using ColabFold, we list several ways to improve AF2 models by adding information that is missing from the original AF2 model. By using software such as AlphaFill to add cofactors and ligands to the models, or MODELLER to add disulfide bridges between cysteines, we guide users to build a high-quality biological model suitable for applications such as drug design, protein interaction, or molecular dynamics studies.

Molecular dynamics of the human RhD and RhAG blood group proteins.

Introduction: Blood group antigens of the RH system (formerly known as "Rhesus") play an important role in transfusion medicine because of the severe haemolytic consequences of antibodies to these antigens. No crystal structure is available for RhD proteins with its partner RhAG, and the precise stoichiometry of the trimer complex remains unknown. Methods: To analyse their structural properties, the trimers formed by RhD and/or RhAG subunits were generated by protein modelling and molecular dynamics simulations were performed. Results: No major differences in structural behaviour were found between trimers of different compositions. The conformation of the subunits is relatively constant during molecular dynamics simulations, except for three large disordered loops. Discussion: This work makes it possible to propose a reasonable stoichiometry and demonstrates the potential of studying the structural behaviour of these proteins to investigate the hundreds of genetic variants relevant to transfusion medicine.

HumanLectome, an update of UniLectin for the annotation and prediction of human lectins.

The UniLectin portal (https://unilectin.unige.ch/) was designed in 2019 with the goal of centralising curated and predicted data on carbohydrate-binding proteins known as lectins. UniLectin is also intended as a support for the study of lectomes (full lectin set) of organisms or tissues. The present update describes the inclusion of several new modules and details the latest (https://unilectin.unige.ch/humanLectome/), covering our knowledge of the human lectome and comprising 215 unevenly characterised lectins, particularly in terms of structural information. Each HumanLectome entry is protein-centric and compiles evidence of carbohydrate recognition domain(s), specificity, 3D-structure, tissue-based expression and related genomic data. Other recent improvements regarding interoperability and accessibility are outlined.

Evaluation of the Potential Impact of In Silico Humanization on VHH Dynamics.

Camelids have the peculiarity of having classical antibodies composed of heavy and light chains as well as single-chain antibodies. They have lost their light chains and one heavy-chain domain. This evolutionary feature means that their terminal heavy-chain domain, VH, called VHH here, has no partner and forms an independent domain. The VHH is small and easy to express alone; it retains thermodynamic and interaction properties. Consequently, VHHs have garnered significant interest from both biotechnological and pharmaceutical perspectives. However, due to their origin in camelids, they cannot be used directly on humans. A humanization step is needed before a possible use. However, changes, even in the constant parts of the antibodies, can lead to a loss of quality. A dedicated tool, Llamanade, has recently been made available to the scientific community. In a previous paper, we already showed the different types of VHH dynamics. Here, we have selected a representative VHH and tested two humanization hypotheses to accurately assess the potential impact of these changes. This example shows that despite the non-negligible change (1/10th of residues) brought about by humanization, the effect is not drastic, and the humanized VHH retains conformational properties quite similar to those of the camelid VHH.

Structural Space of the Duffy Antigen/Receptor for Chemokines' Intrinsically Disordered Ectodomain 1 Explored by Temperature Replica-Exchange Molecular Dynamics Simulations.

Plasmodium vivax malaria affects 14 million people each year. Its invasion requires interactions between the parasitic Duffy-binding protein (PvDBP) and the N-terminal extracellular domain (ECD1) of the host's Duffy antigen/receptor for chemokines (DARC). ECD1 is highly flexible and intrinsically disordered, therefore it can adopt different conformations. We computationally modeled the challenging ECD1 local structure. With T-REMD simulations, we sampled its dynamic behavior and collected its most representative conformations. Our results suggest that most of the DARC ECD1 domain remains in a disordered state during the simulated time. Globular local conformations are found in the analyzed local free-energy minima. These globular conformations share an α-helix spanning residues Ser18 to Ser29 and in many cases they comprise an antiparallel ß-sheet, whose ß-strands are formed around residues Leu10 and Ala49. The formation of a parallel ß-sheet is almost negligible. So far, progress in understanding the mechanisms forming the basis of the P. vivax malaria infection of reticulocytes has been hampered by experimental difficulties, along with a lack of DARC structural information. Our collection of the most probable ECD1 structural conformations will help to advance modeling of the DARC structure and to explore DARC-ECD1 interactions with a range of physiological and pathological ligands.

Special Issue: New Advances in Bioinformatics and Biomedical Engineering Using Machine Learning Techniques, IWBBIO-2022.

Bioinformatics is revolutionizing Biomedicine in the way we treat and diagnose pathologies related to biological manifestations resulting from variations or mutations of our DNA [...].

A perspective on the sharing of docking data.

Computational approaches are nowadays largely applied in drug discovery projects. Among these, molecular docking is the most used for hit identification against a drug target protein. However, many scientists in the field shed light on the lack of availability and reproducibility of the data obtained from such studies to the whole community. Consequently, sustaining and developing the efforts toward a large and fully transparent sharing of those data could be beneficial for all researchers in drug discovery. The purpose of this article is first to propose guidelines and recommendations on the appropriate way to conduct virtual screening experiments and second to depict the current state of sharing molecular docking data. In conclusion, we have explored and proposed several prospects to enhance data sharing from docking experiment that could be developed in the foreseeable future.

Structural and Dynamic Differences between Calreticulin Mutants Associated with Essential Thrombocythemia.

Essential thrombocythemia (ET) is a blood cancer. ET is characterized by an overproduction of platelets that can lead to thrombosis formation. Platelet overproduction occurs in megakaryocytes through a signaling pathway that could involve JAK2, MPL, or CALR proteins. CALR mutations are associated with 25-30% of ET patients; CALR variants must be dimerized to induce ET. We classified these variants into five classes named A to E; classes A and B are the most frequent classes in patients with ET. The dynamic properties of these five classes using structural models of CALR's C-domain were analyzed using molecular dynamics simulations. Classes A, B, and C are associated with frameshifts in the C-domain. Their dimers can be stable only if a disulfide bond is formed; otherwise, the two monomers repulse each other. Classes D and E cannot be stable as dimers due to the absence of disulfide bonds. Class E and wild-type CALR have similar dynamic properties. These results suggest that the disulfide bond newly formed in classes A, B, and C may be essential for the pathogenicity of these variants. They also underline that class E cannot be directly related to ET but corresponds to human polymorphisms.

General Trends of the Camelidae Antibody VHHs Domain Dynamics.

Conformational flexibility plays an essential role in antibodies' functional and structural stability. They facilitate and determine the strength of antigen-antibody interactions. Camelidae express an interesting subtype of single-chain antibody, named Heavy Chain only Antibody. They have only one N-terminal Variable domain (VHH) per chain, composed of Frameworks (FRs) and Complementarity Determining regions (CDRs) like their VH and VL counterparts in IgG. Even when expressed independently, VHH domains display excellent solubility and (thermo)stability, which helps them to retain their impressive interaction capabilities. Sequence and structural features of VHH domains contributing to these abilities have already been studied compared to classical antibodies. To have the broadest view and understand the changes in dynamics of these macromolecules, large-scale molecular dynamics simulations for a large number of non-redundant VHH structures have been performed for the first time. This analysis reveals the most prevalent movements in these domains. It reveals the four main classes of VHHs dynamics. Diverse local changes were observed in CDRs with various intensities. Similarly, different types of constraints were observed in CDRs, while FRs close to CDRs were sometimes primarily impacted. This study sheds light on the changes in flexibility in different regions of VHH that may impact their in silico design.

A novel high-prevalence antigen in the Lutheran system, LUGA (LU24), and an updated, full-length 3D BCAM model.

BACKGROUND: The basal cell adhesion molecule (BCAM) carries the antigens of the Lutheran (LU, ISBT005) system. We report a novel Lutheran antigen and propose an updated, full-length 3D model of BCAM. STUDY DESIGN AND METHODS: Red blood cell testing, antibody identification, and BCAM genomic DNA sequencing were done by standard methods. Multi-template homology modeling of BCAM used structural templates selected for coverage, highest sequence identity, and protein domain family. All variants causing the loss or gain of a Lutheran antigen were analyzed for residue accessibility and intraprotein interactions. RESULTS: An antibody to a high-prevalence antigen in the plasma of a pregnant woman was determined to be directed at a novel Lutheran antigen. Sequencing of BCAM found three homozygous changes: c.212G > A (p.Arg71His) and two silent, c.711C > T and c.714C > T. The model was built from the first two immunoglobulin crystallized domains of BCAM (D1, D2), three other templates (for D3, D4 and D5 with a higher sequence identity with the target than those used for the model proposed by Burton and Brady in 2008, and for the transmembrane region) and RaptorX (for the intracellular domain). All residues associated with a Lutheran antigen were found to be exposed in wild-type or variant proteins, except p.447 associated with loss of Lu13 expression. CONCLUSION: The c.212G > A change results in the loss of LUGA (LU24) antigen. Whole genome sequencing continues to reveal polymorphisms with uncertain immunogenicity. This model and demonstration that nearly all residues associated with the expression of a Lutheran antigen are exposed will help evaluate the significance of new polymorphisms.

Quality assessment of VHH models.

Heavy Chain Only Antibodies are specific to Camelid species. Despite the lack of the light chain variable domain, their heavy chain variable domain (VH) domain, named VHH or nanobody, has promising potential applications in research and therapeutic fields. The structural study of VHH is therefore of great interest. Unfortunately, considering the huge amount of sequences that might be produced, only about one thousand of VHH experimental structures are publicly available in the Protein Data Bank, implying that structural model prediction of VHH is a necessary alternative to obtaining 3D information besides its sequence. The present study aims to assess and compare the quality of predictions from different modelling methodologies. Established comparative & homology modelling approaches to recent Deep Learning-based modelling strategies were applied, i.e. Modeller using single or multiple structural templates, ModWeb, SwissModel (with two evaluation schema), RoseTTAfold, AlphaFold 2 and NanoNet. The prediction accuracy was evaluated using RMSD, TM-score, GDT-TS, GDT-HA and Protein Blocks distance metrics. Besides the global structure assessment, we performed specific analyses of Frameworks and CDRs structures. We observed that AlphaFold 2 and especially NanoNet performed better than the other evaluated softwares. Importantly, we performed molecular dynamics simulations of an experimental structure and a NanoNet predicted model of a VHH in order to compare the global structural flexibility and local conformations using Protein Blocks. Despite rather similar structures, substantial differences in dynamical properties were observed, which underlies the complexity of the task of model evaluation.Communicated by Ramaswamy H. Sarma.

Conformational and binding mode assessment of the human IL-3 recognition by its alpha receptor.

Protein-protein interactions (PPIs) are attractive targets as they are critical in a variety of biological processes and pathologies. As an illustration, the interleukin 3 (IL-3) and its α subunit receptor (IL-3Rα) are two proteins belonging to the cytokine or receptor ßc family and are involved in several disorders like inflammatory diseases or hematological malignancies. This PPI exhibits a low binding affinity and a complex formed by a mutant form of IL-3 (superkine) and IL-3Rα have emerged from the literature, with an increase of the affinity. Therefore, in this study, we performed molecular dynamics simulations and binding energy calculation in order to evaluate protein dynamics and to characterize the main interactions between IL-3 and IL-3Rα, considering both wild-type and mutant. First, in the case of IL-3Rα/IL-3 wild-type complex, IL-3Rα can adopt three different conformations essentially driven by NTD domain, including the open and closed conformations, previously observed in crystal structures. Additionally, our results reveal a third conformation that has a distinct interaction profile that the others. Interestingly, these conformational changes are attenuated in IL-3Rα/IL-3 mutant complex. Then, we highlighted the contribution of different residues which interact principally with IL-3 or IL-3Rα conserved region. As for the mutated residue at position 135 of IL-3, other residues such as IL-3 E138, IL-3 D40, IL-3Rα Y279, IL-3Rα K235, or IL-3Rα R277 seem important for a low or a high binding affinity. Altogether these findings yield new information that could be exploited in a drug discovery process.

An agnostic analysis of the human AlphaFold2 proteome using local protein conformations.

Knowledge of the 3D structure of proteins is a valuable asset for understanding their precise biological mechanisms. However, the cost of production of 3D structures and experimental difficulties limit their obtaining. The proposal of 3D structural models is consequently an appealing alternative. The release of the AlphaFold Deep Learning approach has revolutionized the field. The recent near-complete human proteome proposal makes it possible to analyse large amounts of data and evaluate the results of the approach in greater depth. The 3D human proteome was thus analysed in light of the classic secondary structures, and many less-used protein local conformations (PolyProline II helices, type of γ-turns, of ß-turns and of ß-bulges, curvature of the helices, and a structural alphabet). Without questioning the global quality of the approach, this analysis highlights certain local conformations, which maybe poorly predicted and they could therefore be better addressed.

Two new Scianna variants causing loss of high prevalence antigens: ERMAP model and 3D analysis of the antigens.

BACKGROUND: Scianna (Sc) antigens, seven high and two of low prevalence, are expressed on erythrocyte membrane-associated protein (ERMAP). We investigated SC (ERMAP) in individuals who made antibodies to high prevalence Scianna antigens, and propose a 3D model for ERMAP to precisely localize the residues associated with the known antigens. METHODS: Serological testing and DNA sequencing was performed by standard methods. A 3D structural model was built using a multi-template homology approach. Protein structures representing missense variants associated with the loss or gain of an antigen were generated. Residue accessibility and intraprotein interactions were compared with the wild-type protein. RESULTS: Two new SC alleles, one with c.349C > T (p.Arg117Cys) in a woman from South India with anti-Sc3 in her plasma, and a c.217_219delinsTGT (p.Arg73Cys) in an African-American woman with an antibody to a new high prevalence antigen, termed SCAC, were identified. Six structural templates were used to model ERMAP. 3D analysis showed that residues key for Scianna antigen expression were all exposed at the surface of the extracellular domain. The p.Arg117Cys change was predicted to abolish interactions between residues 93 and 117, with no compensating interactions. CONCLUSION: We confirm the extracellular location of Scianna residues responsible for antigen expression which predicts direct accessibility to antibodies. Loss of intraprotein interactions appear to be responsible for a Sc null and production of anti-Sc3 with p.117Cys, SC*01 N.03, and for loss of a high prevalence antigen with p.73Cys, termed SCAC for Sc Arg to Cys. Comparative modeling aids our understanding of new alleles and Scianna antigen expression.

A Perspective on the (Rise and Fall of) Protein ß-Turns.

The ß-turn is the third defined secondary structure after the α-helix and the ß-sheet. The ß-turns were described more than 50 years ago and account for more than 20% of protein residues. Nonetheless, they are often overlooked or even misunderstood. This poor knowledge of these local protein conformations is due to various factors, causes that I discuss here. For example, confusion still exists about the assignment of these local protein structures, their overlaps with other structures, the potential absence of a stabilizing hydrogen bond, the numerous types of ß-turns and the software's difficulty in assigning or visualizing them. I also propose some ideas to potentially/partially remedy this and present why ß-turns can still be helpful, even in the AlphaFold 2 era.

SWORD2: hierarchical analysis of protein 3D structures.

Understanding the functions and origins of proteins requires splitting these macromolecules into fragments that could be independent in terms of folding, activity, or evolution. For that purpose, structural domains are the typical level of analysis, but shorter segments, such as subdomains and supersecondary structures, are insightful as well. Here, we propose SWORD2, a web server for exploring how an input protein structure may be decomposed into 'Protein Units' that can be hierarchically assembled to delimit structural domains. For each partitioning solution, the relevance of the identified substructures is estimated through different measures. This multilevel analysis is achieved by integrating our previous work on domain delineation, 'protein peeling' and model quality assessment. We hope that SWORD2 will be useful to biologists searching for key regions in their proteins of interest and to bioinformaticians building datasets of protein structures. The web server is freely available online: https://www.dsimb.inserm.fr/SWORD2.

VHH Structural Modelling Approaches: A Critical Review.

VHH, i.e., VH domains of camelid single-chain antibodies, are very promising therapeutic agents due to their significant physicochemical advantages compared to classical mammalian antibodies. The number of experimentally solved VHH structures has significantly improved recently, which is of great help, because it offers the ability to directly work on 3D structures to humanise or improve them. Unfortunately, most VHHs do not have 3D structures. Thus, it is essential to find alternative ways to get structural information. The methods of structure prediction from the primary amino acid sequence appear essential to bypass this limitation. This review presents the most extensive overview of structure prediction methods applied for the 3D modelling of a given VHH sequence (a total of 21). Besides the historical overview, it aims at showing how model software programs have been shaping the structural predictions of VHHs. A brief explanation of each methodology is supplied, and pertinent examples of their usage are provided. Finally, we present a structure prediction case study of a recently solved VHH structure. According to some recent studies and the present analysis, AlphaFold 2 and NanoNet appear to be the best tools to predict a structural model of VHH from its sequence.

Prediction of structural alphabet protein blocks using data mining.

3D protein structures determine proteins' biological functions. The 3D structure of the protein backbone can be approximated using the prototypes of local protein conformations. Sets of these prototypes are called structural alphabets (SAs). Amongst several approaches to the prediction of 3D structures from amino acid sequences, one approach is based on the prediction of SA prototypes for a given amino acid sequence. Protein Blocks (PBs) is the most known SA, and it is composed of 16 prototypes of five consecutive amino acids which were identified as optimal prototypes considering the ability to correctly approximate the local structure and the prediction accuracy of prototypes from an amino acid sequence. We developed models for PBs prediction from sequence information using different data mining approaches and machine learning algorithms. Besides the amino acid sequences, the results of the following tools were used to train the models: the Spider3 predictor of protein structure properties, several predictors of the protein's intrinsically disordered regions, and a tool for finding repeats in amino acid sequences. The highest accuracy of the constructed models is 80%, which is a significant improvement compared to the previous best available prediction, whose accuracy was 61%. Analyzing the models constructed by applying different algorithms, it was noticed that the significance of input attributes differs among the models constructed by algorithms. Using the information about amino acids belonging to intrinsically disordered regions and repeats improves the precision of prediction for some PBs using the CART classification algorithm, while this is not the case with the C5.0 classification algorithm. Improved prediction approaches can have interesting applications in protein structural model approaches or computational protein design.