An automated decision-tree approach to predicting protein interaction hot spots

ACDMBI: A deep learning model based on community division and multi-source biological information fusion predicts essential proteins

Accurately identifying essential proteins is vital for drug research and disease diagnosis. Traditional centrality methods and machine learning approaches often face challenges in accurately discerning essential proteins, primarily relying on information derived from protein-protein interaction (PPI) networks. Despite attempts by some researchers to integrate biological data and PPI networks for predicting essential proteins, designing effective integration methods remains a challenge. In response to these challenges, this paper presents the ACDMBI model, specifically designed to overcome the aforementioned issues. ACDMBI is comprised of two key modules: feature extraction and classification. In terms of capturing relevant information, we draw insights from three distinct data sources. Initially, structural features of proteins are extracted from the PPI network through community division. Subsequently, these features are further optimized using Graph Convolutional Networks (GCN) and Graph Attention Networks (GAT). Moving forward, protein features are extracted from gene expression data utilizing Bidirectional Long Short-Term Memory networks (BiLSTM) and a multi-head self-attention mechanism. Finally, protein features are derived by mapping subcellular localization data to a one-dimensional vector and processing it through fully connected layers. In the classification phase, we integrate features extracted from three different data sources, crafting a multi-layer deep neural network (DNN) for protein classification prediction. Experimental results on brewing yeast data showcase the ACDMBI model's superior performance, with AUC reaching 0.9533 and AUPR reaching 0.9153. Ablation experiments further reveal that the effective integration of features from diverse biological information significantly boosts the model's performance.

Virtual Screening of Peptide Libraries: The Search for Peptide-Based Therapeutics Using Computational Tools

Over the last few decades, we have witnessed growing interest from both academic and industrial laboratories in peptides as possible therapeutics. Bioactive peptides have a high potential to treat various diseases with specificity and biological safety. Compared to small molecules, peptides represent better candidates as inhibitors (or general modulators) of key protein-protein interactions. In fact, undruggable proteins containing large and smooth surfaces can be more easily targeted with the conformational plasticity of peptides. The discovery of bioactive peptides, working against disease-relevant protein targets, generally requires the high-throughput screening of large libraries, and in silico approaches are highly exploited for their low-cost incidence and efficiency. The present review reports on the potential challenges linked to the employment of peptides as therapeutics and describes computational approaches, mainly structure-based virtual screening (SBVS), to support the identification of novel peptides for therapeutic implementations. Cutting-edge SBVS strategies are reviewed along with examples of applications focused on diverse classes of bioactive peptides (i.e., anticancer, antimicrobial/antiviral peptides, peptides blocking amyloid fiber formation).

Computational design of a cyclic peptide that inhibits the CTLA4 immune checkpoint

Proteins involved in immune checkpoint pathways, such as CTLA4, PD1, and PD-L1, have become important targets for cancer immunotherapy; however, development of small molecule drugs targeting these pathways has proven difficult due to the nature of their protein-protein interfaces. Here, using a hierarchy of computational techniques, we design a cyclic peptide that binds CTLA4 and follow this with experimental verification of binding and biological activity, using bio-layer interferometry, cell culture, and a mouse tumor model. Beginning from a template excised from the X-ray structure of the CTLA4:B7-2 complex, we generate several peptide sequences using flexible docking and modeling steps. These peptides are cyclized head-to-tail to improve structural and proteolytic stability and screened using molecular dynamics simulation and MM-GBSA calculation. The standard binding free energies for shortlisted peptides are then calculated in explicit-solvent simulation using a rigorous multistep technique. The most promising peptide, cyc(EIDTVLTPTGWVAKRYS), yields the standard free energy -6.6 ± 3.5 kcal mol-1, which corresponds to a dissociation constant of ∼15 μmol L-1. The binding affinity of this peptide for CTLA4 is measured experimentally (31 ± 4 μmol L-1) using bio-layer interferometry. Treatment with this peptide inhibited tumor growth in a co-culture of Lewis lung carcinoma (LLC) cells and antigen primed T cells, as well as in mice with an orthotropic Lewis lung carcinoma allograft model.

Immunoinformatics-guided design of a multi-valent vaccine against Rotavirus and Norovirus (ChRNV22)

Rotavirus (RV) and Norovirus (NV) are the main viral etiologic agents of acute gastroenteritis (AG), a serious pediatric condition associated with significant death rates and long-term complications. Anti-RV vaccination has been proved efficient in the reduction of severe AG worldwide, however, the available vaccines are all attenuated and have suboptimal efficiencies in developing countries, where AG leads to substantial disease burden. On the other hand, no NV vaccine has been licensed so far. Therefore, we used immunoinformatics tools to develop a multi-epitope vaccine (ChRNV22) to prevent severe AG by RV and NV. Epitopes were predicted against 17 prevalent genotypes of four structural proteins (NV's VP1, RV's VP4, VP6 and VP7), and then assembled in a chimeric protein, with two small adjuvant sequences (tetanus toxin P2 epitope and a conserved sequence of RV's enterotoxin, NSP4). Simulations of the immune response and interactions with immune receptors indicated the immunogenic properties of ChRNV22, including a Th1-biased response. In silico search for putative host-homologous, allergenic and toxic regions also indicated the vaccine safety. In summary, we developed a multi-epitope vaccine against different NV and RV genotypes that seems promising for the prevention of severe AG, which will be further assessed by in vivo tests.

Trimeric Architecture Ensures the Stability and Biological Activity of the Calf Purine Nucleoside Phosphorylase: In Silico and In Vitro Studies of Monomeric and Trimeric Forms of the Enzyme

Mammalian purine nucleoside phosphorylase (PNP) is biologically active as a homotrimer, in which each monomer catalyzes a reaction independently of the others. To answer the question of why the native PNP forms a trimeric structure, we constructed, in silico and in vitro, the monomeric form of the enzyme. Molecular dynamics simulations showed different geometries of the active site in the non-mutated trimeric and monomeric PNP forms, which suggested that the active site in the isolated monomer could be non-functional. To confirm this hypothesis, six amino acids located at the interface of the subunits were selected and mutated to alanines to disrupt the trimer and obtain a monomer (6Ala PNP). The effects of these mutations on the enzyme structure, stability, conformational dynamics, and activity were examined. The solution experiments confirmed that the 6Ala PNP mutant occurs mainly as a monomer, with a secondary structure almost identical to the wild type, WT PNP, and importantly, it shows no enzymatic activity. Simulations confirmed that, although the secondary structure of the 6Ala monomer is similar to the WT PNP, the positions of the amino acids building the 6Ala PNP active site significantly differ. These data suggest that a trimeric structure is necessary to stabilize the geometry of the active site of this enzyme.

Structural and functional characterization of NEMO cleavage by SARS-CoV-2 3CLpro

In addition to its essential role in viral polyprotein processing, the SARS-CoV-2 3C-like protease (3CLpro) can cleave human immune signaling proteins, like NF-κB Essential Modulator (NEMO) and deregulate the host immune response. Here, in vitro assays show that SARS-CoV-2 3CLpro cleaves NEMO with fine-tuned efficiency. Analysis of the 2.50 Å resolution crystal structure of 3CLpro C145S bound to NEMO226-234 reveals subsites that tolerate a range of viral and host substrates through main chain hydrogen bonds while also enforcing specificity using side chain hydrogen bonds and hydrophobic contacts. Machine learning- and physics-based computational methods predict that variation in key binding residues of 3CLpro-NEMO helps explain the high fitness of SARS-CoV-2 in humans. We posit that cleavage of NEMO is an important piece of information to be accounted for, in the pathology of COVID-19.

Identification of Small-Molecule Inhibitors of Fibroblast Growth Factor 23 Signaling via In Silico Hot Spot Prediction and Molecular Docking to α-Klotho

Fibroblast growth factor 23 (FGF23) is a therapeutic target for treating hereditary and acquired hypophosphatemic disorders, such as X-linked hypophosphatemic (XLH) rickets and tumor-induced osteomalacia (TIO), respectively. FGF23-induced hypophosphatemia is mediated by signaling through a ternary complex formed by FGF23, the FGF receptor (FGFR), and α-Klotho. Currently, disorders of excess FGF23 are treated with an FGF23-blocking antibody, burosumab. Small-molecule drugs that disrupt protein/protein interactions necessary for the ternary complex formation offer an alternative to disrupting FGF23 signaling. In this study, the FGF23:α-Klotho interface was targeted to identify small-molecule protein/protein interaction inhibitors since it was computationally predicted to have a large fraction of hot spots and two druggable residues on α-Klotho. We further identified Tyr433 on the KL1 domain of α-Klotho as a promising hot spot and α-Klotho as an appropriate drug-binding target at this interface. Subsequently, we performed in silico docking of ∼5.5 million compounds from the ZINC database to the interface region of α-Klotho from the ternary crystal structure. Following docking, 24 and 20 compounds were in the final list based on the lowest binding free energies to α-Klotho and the largest number of contacts with Tyr433, respectively. Five compounds were assessed experimentally by their FGF23-mediated extracellular signal-regulated kinase (ERK) activities in vitro, and two of these reduced activities significantly. Both these compounds were predicted to have favorable binding affinities to α-Klotho but not have a large number of contacts with the hot spot Tyr433. ZINC12409120 was found experimentally to disrupt FGF23:α-Klotho interaction to reduce FGF23-mediated ERK activities by 70% and have a half maximal inhibitory concentration (IC₅₀) of 5.0 ± 0.23 μM. Molecular dynamics (MD) simulations of the ZINC12409120:α-Klotho complex starting from in silico docking poses reveal that the ligand exhibits contacts with residues on the KL1 domain, the KL1-KL2 linker, and the KL2 domain of α-Klotho simultaneously, thereby possibly disrupting the regular function of α-Klotho and impeding FGF23:α-Klotho interaction. ZINC12409120 is a candidate for lead optimization.

Majority Vote Cascading: A Semi-Supervised Framework for Improving Protein Function Prediction

A method to improve protein function prediction for sparsely annotated PPI networks is introduced. The method extends the DSD majority vote algorithm introduced by Cao et al. to give confidence scores on predicted labels and to use predictions of high confidence to predict the labels of other nodes in subsequent rounds. We call this a majority vote cascade. Several cascade variants are tested in a stringent cross-validation experiment on PPI networks from S. cerevisiae and D. melanogaster, and we show that for many different settings with several alternative confidence functions, cascading improves the accuracy of the predictions. A list of the most confident new label predictions in the two networks is also reported. Code and networks for the cross-validation experiments appear at http://bcb.cs.tufts.edu/cascade.

PPI-HotspotDB: Database of Protein-Protein Interaction Hot Spots

Single-point mutations of certain residues (so-called hot spots) impair/disrupt protein-protein interactions (PPIs), leading to pathogenesis and drug resistance. Conventionally, a PPI-hot spot is identified when its replacement decreased the binding free energy significantly, generally by ≥2 kcal/mol. The relatively few mutations with such a significant binding free energy drop limited the number of distinct PPI-hot spots. By defining PPI-hot spots based on mutations that have been manually curated in UniProtKB to significantly impair/disrupt PPIs in addition to binding free energy changes, we have greatly expanded the number of distinct PPI-hot spots by an order of magnitude. These experimentally determined PPI-hot spots along with available structures have been collected in a database called PPI-Hotspot^DB. We have applied the PPI-Hotspot^DB to create a nonredundant benchmark, PPI-Hotspot+PDB^BM, for assessing methods to predict PPI-hot spots using the free structure as input. PPI-Hotspot^DB will benefit the design of mutagenesis experiments and development of PPI-hot spot prediction methods. The database and benchmark are freely available at https://ppihotspot.limlab.dnsalias.org.

Artificial intelligence based methods for hot spot prediction

Proteins interact through their interfaces to fulfill essential functions in the cell. They bind to their partners in a highly specific manner and form complexes that have a profound effect on understanding the biological pathways they are involved in. Any abnormal interactions may cause diseases. Therefore, the identification of small molecules which modulate protein interactions through their interfaces has high therapeutic potential. However, discovering such molecules is challenging. Most protein-protein binding affinity is attributed to a small set of amino acids found in protein interfaces known as hot spots. Recent studies demonstrate that drug-like small molecules specifically may bind to hot spots. Therefore, hot spot prediction is crucial. As experimental data accumulates, artificial intelligence begins to be used for computational hot spot prediction. First, we review machine learning and deep learning for computational hot spot prediction and then explain the significance of hot spots toward drug design.

A peptide derived from GAPDH enhances resistance to DNA damage in budding yeast

Social behaviors do not only exist in higher organisms but are also present in microbes that interact for the common good. Here, we report that budding yeast cells interact with their neighboring cells after exposure to DNA damage. Yeast cells irradiated with DNA-damaging ultraviolet light secrete signal peptides that can increase the survival of yeast cells exposed to DNA-damaging stress. The secreted peptide is derived from glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and it induced cell death of a fraction of yeast cells in the group. The data suggest that the GAPDH-derived peptide serves in budding yeast's social interaction in response to DNA-damaging stress. Importance Many studies have shown that microorganisms, including bacteria and yeast, display increased tolerance to stress after exposure to the same stressor. However, the mechanism remains unknown. In this manuscript, we report a striking finding that S. cerevisiae cells respond to DNA damage by secreting a peptide that facilitates resistance to DNA-damaging stress. Although it has been shown that GAPDH possesses many key functions in cells aside from its well-established role in glycolysis, this study demonstrated that GAPDH is also involved in the social behaviors response to DNA-damaging stress. The study opens the gate to an interesting research field about microbial social activity for adaptation to a harsh environment.

SARS CoV-2 Delta variant exhibits enhanced infectivity and a minor decrease in neutralization sensitivity to convalescent or post-vaccination sera

Since their identification, severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) Kappa and Delta have rapidly spread to become globally dominant. However, their infectivity and sensitivity to administered vaccines have not been documented. We monitored the neutralization potential of convalescent or BNT162b2 post-vaccination sera against Kappa and Delta SARS-CoV-2 pseudoviruses. We show that both variants were successfully neutralized by convalescent and post-vaccination sera, exhibiting a mild decrease in their neutralization sensitivity. Of the two variants, Delta presented enhanced infectivity levels compared with Kappa or wild-type SARS-CoV-2. Nevertheless, both variants were not as infectious or resistant to post-vaccination sera as the Beta variant of concern. Interestingly, the Delta plus variant (AY.1/B.1.617.2.1) exhibited high resistance to post-vaccination sera, similar to that of the Beta SARS-CoV-2. However, its infectivity levels were close to those of wild-type SARS-CoV-2. These results account for the worldwide prevalence of Delta variant of concern and confirm the efficacy of the BNT162b2 vaccine against circulating other Delta variants.

Structural and functional characterization of NEMO cleavage by SARS-CoV-2 3CLpro.. [PMID: 34816264 PMCID: PMC8609902 DOI: 10.1101/2021.11.11.468228] [Citation(s) in RCA: 0] [Impact Index Per Article: 0]

In addition to its essential role in viral polyprotein processing, the SARS-CoV-2 3C-like (3CLpro) protease can cleave human immune signaling proteins, like NF-κB Essential Modulator (NEMO) and deregulate the host immune response. Here, in vitro assays show that SARS-CoV-2 3CLpro cleaves NEMO with fine-tuned efficiency. Analysis of the 2.14 Å resolution crystal structure of 3CLpro C145S bound to NEMO_226–235 reveals subsites that tolerate a range of viral and host substrates through main chain hydrogen bonds while also enforcing specificity using side chain hydrogen bonds and hydrophobic contacts. Machine learning- and physics-based computational methods predict that variation in key binding residues of 3CLpro-NEMO helps explain the high fitness of SARS-CoV-2 in humans. We posit that cleavage of NEMO is an important piece of information to be accounted for in the pathology of COVID-19.

Designing an epitope vaccine against Dermatophagoides pteronyssinus: An in silico study

The house dust mite, Dermatophagoides pteronyssinus, is a major source of the inhaled allergen Der p 1, which causes immunoglobulin E (IgE)-mediated hypersensitivity reactions manifesting in allergic diseases. To date, no drugs or vaccines effectively treat or prevent Der p 1 sensitization. We applied in silico immunoinformatics to design T-cell and B-cell epitopes that were specified and developed from the allergen Der p 1 of D. pteronyssinus. We identified the conserved epitope areas by predicting the accessibility and flexibility of B-cell epitopes, and the percentage of human leukocyte antigen representing T cells. Molecular docking using HADDOCK software indicated three optimal clusters: cluster 6 (z-score: -2.1), cluster 1 (z-score: -1.2), and cluster 3 (z-score: -0.6). The most negative Z-score was found in cluster 6, which represented three epitopes. The interaction between A chain proteins (IgE protein residues) and B chains (Der p 1 protein residues) exhibited a knowledge-based FADE and contact value >1, suggesting the best protein interactions occurred in the conserved area. Molecular dynamic simulation further predicted the stable nature of Der p 1 protein. The IQRDNGYQP region is the best candidate to be utilized as a D. pteronyssinus epitope vaccine, which could be used in the development of allergen-specific immunotherapy.

Docking-based generation of antibodies mimicking Cry1A/1B protein binding sites as potential insecticidal agents against diamondback moth (Plutella xylostella)

BACKGROUND

Broad use of insecticidal Cry proteins from Bacillus thuringiensis in biopesticides and transgenic crops has resulted in cases of practical field resistance, highlighting the need for novel approaches to insect control. Previously we described an anti-Cry1Ab idiotypic-antibody (B12-scFv) displaying toxicity against rice leafroller (Cnaphalocrocis medinalis) larvae, supporting the potential of antibodies for pest control. The goal of the present study was to generate insecticidal antibodies against diamondback moth (Plutella xylostella) larvae.

RESULTS

Four genetically engineered antibodies (GEAbs) were designed in silico from B12-scFv using three-dimensional (3D) structure and docking predictions to alkaline phosphatase (ALP) as a Cry1Ac receptor in P. xylostella. Among these GEAbs, the GEAb-dV_L antibody consisting of two light chains had overlapping binding sites with Cry1A and Cry1B proteins and displayed high binding affinity to P. xylostella midgut brush border membrane (BBM) proteins. Proteins in BBM identified by pull-down assays as binding to GEAb-dV_L included an ABC transporter and V-ATPase subunit A protein. Despite lacking the α-helical structures in Cry1A that are responsible for pore formation, ingestion of GEAb-dV_L disrupted the P. xylostella larval midgut epithelium and resulted in toxicity. Apoptotic genes were activated in gut cells upon treatment with GEAb-dV_L .

CONCLUSION

This study describes the first insecticidal GEAb targeting P. xylostella by mimicking Cry proteins. Data support that GEAb-dV_L toxicity is associated to activation of intracellular cell death pathways, in contrast to pore-formation associated toxicity of Cry proteins. This work provides a foundation for the design of novel insecticidal antibodies for insect control. © 2021 Society of Chemical Industry.

The dual role of SrbA from Paracoccidioides lutzii: a hypoxic regulator

The fungus Paracoccidioides lutzii is one of the species of the Paracoccidioides genus, responsible for a neglected human mycosis, endemic in Latin America, the paracoccidioidomycosis (PCM). In order to survive in the host, the fungus overcomes a hostile environment under low levels of oxygen (hypoxia) during the infectious process. The hypoxia adaptation mechanisms are variable among human pathogenic fungi and worthy to be investigated in Paracoccidoides spp. Previous proteomic results identified that P. lutzii responds to hypoxia and it has a functional homolog of the SrbA transcription factor, a well-described hypoxic regulator. However, the direct regulation of genes by SrbA and the biological processes it governs while performing protein interactions have not been revealed yet. The goal of this study was to demonstrate the potential of SrbA targets genes in P. lutzii. In addition, to show the SrbA three-dimensional aspects as well as a protein interaction map and important regions of interaction with predicted targets. The results show that SrbA-regulated genes were involved with several biological categories, such as metabolism, energy, basal processes for cell maintenance, fungal morphogenesis, defense, virulence, and signal transduction. Moreover, in order to investigate the SrbA's role as a protein, we performed a 3D simulation and also a protein-protein network linked to this hypoxic regulator. These in silico analyses revealed relevant aspects regarding the biology of this pathogen facing hypoxia and highlight the potential of SrbA as an antifungal target in the future.

Hotspot Coevolution Is a Key Identifier of Near-Native Protein Complexes

Protein-protein interactions play a key role in mediating numerous biological functions, with more than half the proteins in living organisms existing as either homo- or hetero-oligomeric assemblies. Protein subunits that form oligomers minimize the free energy of the complex, but exhaustive computational search-based docking methods have not comprehensively addressed the challenge of distinguishing a natively bound complex from non-native forms. Current protein docking approaches address this problem by sampling multiple binding modes in proteins and scoring each mode, with the lowest-energy (or highest scoring) binding mode being regarded as a near-native complex. However, high-scoring modes often match poorly with the true bound form, suggesting a need for improvement of the scoring function. In this study, we propose a scoring function, KFC-E, that accounts for both conservation and coevolution of putative binding hotspot residues at protein-protein interfaces. We tested KFC-E on four benchmark sets of unbound examples and two benchmark sets of bound examples, with the results demonstrating a clear improvement over scores that examine conservation and coevolution across the entire interface.

AptaNet as a deep learning approach for aptamer-protein interaction prediction

Aptamers are short oligonucleotides (DNA/RNA) or peptide molecules that can selectively bind to their specific targets with high specificity and affinity. As a powerful new class of amino acid ligands, aptamers have high potentials in biosensing, therapeutic, and diagnostic fields. Here, we present AptaNet-a new deep neural network-to predict the aptamer-protein interaction pairs by integrating features derived from both aptamers and the target proteins. Aptamers were encoded by using two different strategies, including k-mer and reverse complement k-mer frequency. Amino acid composition (AAC) and pseudo amino acid composition (PseAAC) were applied to represent target information using 24 physicochemical and conformational properties of the proteins. To handle the imbalance problem in the data, we applied a neighborhood cleaning algorithm. The predictor was constructed based on a deep neural network, and optimal features were selected using the random forest algorithm. As a result, 99.79% accuracy was achieved for the training dataset, and 91.38% accuracy was obtained for the testing dataset. AptaNet achieved high performance on our constructed aptamer-protein benchmark dataset. The results indicate that AptaNet can help identify novel aptamer-protein interacting pairs and build more-efficient insights into the relationship between aptamers and proteins. Our benchmark dataset and the source codes for AptaNet are available in: https://github.com/nedaemami/AptaNet .

Computationally-obtained structural insights into the molecular interactions between Pidilizumab and binding partners DLL1 and PD-1

Pidilizumab is a monoclonal antibody tested against several types of malignancies, such as lymphoma and metastatic melanoma, showing promising results. In 2016, the FDA put Pidilizumab's clinical studies on partial hold due to emerging evidence pointing to the antibody target uncertainty. Although initial studies indicated an interaction with the PD-1 checkpoint receptor, recent updates assert that Pidilizumab binds primarily to Notch ligand DLL1. However, a detailed description of which interactions coordinate antibody-antigen complex formation is lacking. Therefore, this study uses computational tools to identify molecular interactions between Pidilizumab and its reported targets PD-1 and DLL1. A docking methodology was validated and applied to determine the binding modes between modeled Pidilizumab scFvs and the two antigens. We used Molecular Dynamics (MD) simulations to verify the complexes' stability and submitted the resulting trajectory files to MM/PBSA and Principal Component Analysis. A set of different prediction tools determined scFv interface hot-spots. Whereas docking and MD simulations revealed that the antibody fragments do not interact straightforwardly with PD-1, ten scFv hot-spots, including Met93 and Leu112, mediated the interaction with the DLL1 C2 domain. The interaction triggered a conformational selection-like effect on DLL1, allowing new hydrogen bonds on the β3-β4 interface loop. The unprecedented structural data on Pidilizumab's interactions provided novel evidence that its legitimate target is the DLL1 protein and offered structural insight on how these molecules interact, shedding light on the pathways that could be affected by the use of this essential immunobiological.Communicated by Ramaswamy H. Sarma.

Robust principal component analysis-based prediction of protein-protein interaction hot spots

Proteins often exert their function by binding to other cellular partners. The hot spots are key residues for protein-protein binding. Their identification may shed light on the impact of disease associated mutations on protein complexes and help design protein-protein interaction inhibitors for therapy. Unfortunately, current machine learning methods to predict hot spots, suffer from limitations caused by gross errors in the data matrices. Here, we present a novel data pre-processing pipeline that overcomes this problem by recovering a low rank matrix with reduced noise using Robust Principal Component Analysis. Application to existing databases shows the predictive power of the method.

Decoding Protein-protein Interactions: An Overview

Drug discovery has focused on the paradigm "one drug, one target" for a long time. However, small molecules can act at multiple macromolecular targets, which serves as the basis for drug repurposing. In an effort to expand the target space, and given advances in X-ray crystallography, protein-protein interactions have become an emerging focus area of drug discovery enterprises. Proteins interact with other biomolecules and it is this intricate network of interactions that determines the behavior of the system and its biological processes. In this review, we briefly discuss networks in disease, followed by computational methods for protein-protein complex prediction. Computational methodologies and techniques employed towards objectives such as protein-protein docking, protein-protein interactions, and interface predictions are described extensively. Docking aims at producing a complex between proteins, while interface predictions identify a subset of residues on one protein that could interact with a partner, and protein-protein interaction sites address whether two proteins interact. In addition, approaches to predict hot spots and binding sites are presented along with a representative example of our internal project on the chemokine CXC receptor 3 B-isoform and predictive modeling with IP10 and PF4.

Computational Hot-Spot Analysis of the SARS-CoV-2 Receptor Binding Domain/ACE2 Complex*

Infection and replication of SARS CoV-2 (the virus that causes COVID-19) requires entry to the interior of host cells. In humans, a protein-protein interaction (PPI) between the SARS CoV-2 receptor-binding domain (RBD) and the extracellular peptidase domain of ACE2 on the surface of cells in the lower respiratory tract is an initial step in the entry pathway. Inhibition of the SARS CoV-2 RBD/ACE2 PPI is currently being evaluated as a target for therapeutic and/or prophylactic intervention. However, relatively little is known about the molecular underpinnings of this complex. Employing multiple computational platforms, we predicted "hot-spot" residues in a positive-control PPI (PMI/MDM2) and the CoV-2 RBD/ACE2 complex. Computational alanine scanning mutagenesis was performed to predict changes in Gibbs' free energy that are associated with mutating residues at the positive control (PMI/MDM2) or SARS RBD/ACE2 binding interface to alanine. Additionally, we used the Adaptive Poisson-Boltzmann Solver to calculate macromolecular electrostatic surfaces at the interface of the positive-control PPI and SARS CoV-2/ACE2 PPI. Finally, a comparative analysis of hot-spot residues for SARS-CoV and SARS-CoV-2, in complex with ACE2, is provided. Collectively, this study illuminates predicted hot-spot residues, and clusters, at the SARS CoV-2 RBD/ACE2 binding interface, potentially guiding the development of reagents capable of disrupting this complex and halting COVID-19.

A Sequence-segment Neighbor Encoding Schema for Protein Hotspot Residue Prediction

Background:

Hotspots are those residues that contribute major free energy of binding in protein-protein interactions. Protein functions are frequently dependent on hotspot residues. At present, hotspot residues are always identified by Alanine scanning mutagenesis technology, which is costly, time-consuming and laborious.

Objective:

Therefore, more accurate and efficient methods have to be developed to identify protein hotspot residues.

Methods:

This paper proposed a novel encoding schema of sequence-segment neighbors and constructed a random forest-based model to identify hotspots in protein interaction interfaces. Firstly, 10 amino acid physicochemical properties, 16 features related to the PI and DI, and 25 features related to ASA were extracted. Different from the previous residue encoding schemas, such as auto correlation descriptor or triplet combination information, this paper employed the influence of amino acids neighbors to hotspot residues and amino acids with a certain distance in sequence to the hotspot.

Results:

Moreover, the proposed model was compared with other hotspot prediction methods, including APIS, Robetta, FOLDEF, KFC, MINERVA models, etc.

Conclusion:

The experimental results showed that the proposed model can improve the prediction ability of protein hotspot residues on the same test set.

Rapid prediction of crucial hotspot interactions for icosahedral viral capsid self-assembly by energy landscape atlasing validated by mutagenesis

Icosahedral viruses are under a micrometer in diameter, their infectious genome encapsulated by a shell assembled by a multiscale process, starting from an integer multiple of 60 viral capsid or coat protein (VP) monomers. We predict and validate inter-atomic hotspot interactions between VP monomers that are important for the assembly of 3 types of icosahedral viral capsids: Adeno Associated Virus serotype 2 (AAV2) and Minute Virus of Mice (MVM), both T = 1 single stranded DNA viruses, and Bromo Mosaic Virus (BMV), a T = 3 single stranded RNA virus. Experimental validation is by in-vitro, site-directed mutagenesis data found in literature. We combine ab-initio predictions at two scales: at the interface-scale, we predict the importance (cruciality) of an interaction for successful subassembly across each interface between symmetry-related VP monomers; and at the capsid-scale, we predict the cruciality of an interface for successful capsid assembly. At the interface-scale, we measure cruciality by changes in the capsid free-energy landscape partition function when an interaction is removed. The partition function computation uses atlases of interface subassembly landscapes, rapidly generated by a novel geometric method and curated opensource software EASAL (efficient atlasing and search of assembly landscapes). At the capsid-scale, cruciality of an interface for successful assembly of the capsid is based on combinatorial entropy. Our study goes all the way from resource-light, multiscale computational predictions of crucial hotspot inter-atomic interactions to validation using data on site-directed mutagenesis' effect on capsid assembly. By reliably and rapidly narrowing down target interactions, (no more than 1.5 hours per interface on a laptop with Intel Core i5-2500K @ 3.2 Ghz CPU and 8GB of RAM) our predictions can inform and reduce time-consuming in-vitro and in-vivo experiments, or more computationally intensive in-silico analyses.

Molecular dynamics analysis of the binding of human interleukin-6 with interleukin-6 α-receptor

Human interleukin-6 (hIL-6) is a multifunctional cytokine that regulates immune and inflammatory responses in addition to metabolic and regenerative processes and cancer. hIL-6 binding to the IL-6 receptor (IL-6Rα) induces homodimerization and recruitment of the glycoprotein (gp130) to form a hexameric signaling complex. Anti-IL-6 and IL-6R antibodies are clinically approved inhibitors of IL-6 signaling pathway for treating rheumatoid arthritis and Castleman's disease, respectively. There is a potential to develop novel small molecule IL-6 antagonists derived from understanding the structural basis for IL-6/IL-6Rα interactions. Here, we combine homology modeling with extensive molecular dynamics (MD) simulations to examine the association of hIL-6 with IL-6Rα. A comparison with MD of apo hIL-6 reveals that the binding of hIL-6 to IL-6Rα induces structural and dynamic rearrangements in the AB loop region of hIL-6, disrupting intraprotein contacts and increasing the flexibility of residues 48 to 58 of the AB loop. In contrast, due to the involvement of residues 59 to 78 in forming contacts with the receptor, these residues of the AB loop are observed to rigidify in the presence of the receptor. The binary complex is primarily stabilized by two pairs of salt bridges, Arg181 (hIL-6)- Glu182 (IL-6Rα) and Arg184 (hIL-6)- Glu183 (IL-6Rα) as well as hydrophobic and aromatic stacking interactions mediated essentially by Phe residues in both proteins. An interplay of electrostatic, hydrophobic, hydrogen bonding, and aromatic stacking interactions facilitates the formation of the hIL-6/IL-6Rα complex.

Efficient Classification of Hot Spots and Hub Protein Interfaces by Recursive Feature Elimination and Gradient Boosting

Proteins are not isolated biological molecules, which have the specific three-dimensional structures and interact with other proteins to perform functions. A small number of residues (hot spots) in protein-protein interactions (PPIs) play the vital role in bioinformatics to influence and control of biological processes. This paper uses the boosting algorithm and gradient boosting algorithm based on two feature selection strategies to classify hot spots with three common datasets and two hub protein datasets. First, the correlation-based feature selection is used to remove the highly related features for improving accuracy of prediction. Then, the recursive feature elimination based on support vector machine (SVM-RFE) is adopted to select the optimal feature subset to improve the training performance. Finally, boosting and gradient boosting (G-boosting) methods are invoked to generate classification results. Gradient boosting is capable of obtaining an excellent model by reducing the loss function in the gradient direction to avoid overfitting. Five datasets from different protein databases are used to verify our models in the experiments. Experimental results show that our proposed classification models have the competitive performance compared with existing classification methods.

Computational Hot-Spot Analysis of the SARS-CoV-2 Receptor Binding Domain / ACE2 Complex

Infection and replication of SARS CoV-2 (the virus that causes COVID-19) requires entry to the interior of host cells. In humans, a Protein-Protein Interaction (PPI) between the SARS CoV-2 Receptor-Binding Domain (RBD) and the extracellular peptidase domain of ACE2, on the surface of cells in the lower respiratory tract, is an initial step in the entry pathway. Inhibition of the SARS CoV-2 RBD / ACE2 PPI is currently being evaluated as a target for therapeutic and/or prophylactic intervention. However, relatively little is known about the molecular underpinnings of this complex. Employing multiple computational platforms, we predicted hot-spot residues in a positive control PPI (PMI / MDM2) and the CoV-2 RBD/ACE2 complex. Computational alanine scanning mutagenesis was performed to predict changes in Gibbs free energy that are associated with mutating residues at the positive control (PMI/MDM2) or SARS RBD/ACE2 binding interface to alanine. Additionally, we used the Adaptive Poisson-Boltzmann Solver to calculate macromolecular electrostatic surfaces at the interface of the positive control PPI and SARS CoV-2 / ACE2 PPI. Collectively, this study illuminates predicted hot-spot residues, and clusters, at the SARS CoV-2 RBD / ACE2 binding interface, potentially guiding the development of reagents capable of disrupting this complex and halting COVID-19.

iPNHOT: a knowledge-based approach for identifying protein-nucleic acid interaction hot spots

Background

The interaction between proteins and nucleic acids plays pivotal roles in various biological processes such as transcription, translation, and gene regulation. Hot spots are a small set of residues that contribute most to the binding affinity of a protein-nucleic acid interaction. Compared to the extensive studies of the hot spots on protein-protein interfaces, the hot spot residues within protein-nucleic acids interfaces remain less well-studied, in part because mutagenesis data for protein-nucleic acids interaction are not as abundant as that for protein-protein interactions.

Results

In this study, we built a new computational model, iPNHOT, to effectively predict hot spot residues on protein-nucleic acids interfaces. One training data set and an independent test set were collected from dbAMEPNI and some recent literature, respectively. To build our model, we generated 97 different sequential and structural features and used a two-step strategy to select the relevant features. The final model was built based only on 7 features using a support vector machine (SVM). The features include two unique features such as ∆SASsa^1/2 and esp3, which are newly proposed in this study. Based on the cross validation results, our model gave F1 score and AUROC as 0.725 and 0.807 on the subset collected from ProNIT, respectively, compared to 0.407 and 0.670 of mCSM-NA, a state-of-the art model to predict the thermodynamic effects of protein-nucleic acid interaction. The iPNHOT model was further tested on the independent test set, which showed that our model outperformed other methods.

Conclusion

In this study, by collecting data from a recently published database dbAMEPNI, we proposed a new model, iPNHOT, to predict hotspots on both protein-DNA and protein-RNA interfaces. The results show that our model outperforms the existing state-of-art models. Our model is available for users through a webserver: http://zhulab.ahu.edu.cn/iPNHOT/.

Expression of quasi-equivalence and capsid dimorphism in the Hepadnaviridae

Hepatitis B virus (HBV) is a leading cause of liver disease. The capsid is an essential component of the virion and it is therefore of interest how it assembles and disassembles. The capsid protein is unusual both for its rare fold and that it polymerizes according to two different icosahedral symmetries, causing the polypeptide chain to exist in seven quasi-equivalent environments: A, B, and C in AB and CC dimers in T = 3 capsids, and A, B, C, and D in AB and CD dimers in T = 4 capsids. We have compared the two capsids by cryo-EM at 3.5 Å resolution. To ensure a valid comparison, the two capsids were prepared and imaged under identical conditions. We find that the chains have different conformations and potential energies, with the T = 3 C chain having the lowest. Three of the four quasi-equivalent dimers are asymmetric with respect to conformation and potential energy; however, the T = 3 CC dimer is symmetrical and has the lowest potential energy although its intra-dimer interface has the least free energy of formation. Of all the inter-dimer interfaces, the CB interface has the least area and free energy, in both capsids. From the calculated energies of higher-order groupings of dimers discernible in the lattices we predict early assembly intermediates, and indeed we observe such structures by negative stain EM of in vitro assembly reactions. By sequence analysis and computational alanine scanning we identify key residues and motifs involved in capsid assembly. Our results explain several previously reported observations on capsid assembly, disassembly, and dimorphism.

Hepatitis B virus has infected approximately one third of the human population and causes almost 1 million deaths from liver disease annually. The capsid is a defining feature of a virus, distinct from host components, and therefore a target for intervention. Unusually for a virus, Hepatitis B assembles two capsids, with different geometries, from the same dimeric protein. Geometric principles dictate that the subunits in this system occupy seven different environments. From comparing the two capsids by cryo-electron microscopy at high resolution under the exact same conditions we find that the polypeptide chains adopt seven different conformations. We use these structures to calculate potential energies (analogous to elastic deformation or strain) for the individual chains, dimers, and several higher-order groupings discernible in the two lattices. We also calculate the binding energies between chains. We find that some groupings have substantially lower energy and are therefore potentially more stable, allowing us to predict likely intermediates on the two assembly pathways. We also observe such intermediates by electron microscopy of in vitro capsid assembly reactions. This is the first structural characterization of the early assembly intermediates of this important human pathogen.

MiR193a Modulation and Podocyte Phenotype

Apolipoprotein L1 (APOL1)-miR193a axis has been reported to play a role in the maintenance of podocyte homeostasis. In the present study, we analyzed transcription factors relevant to miR193a in human podocytes and their effects on podocytes’ molecular phenotype. The motif scan of the miR193a gene provided information about transcription factors, including YY1, WT1, Sox2, and VDR-RXR heterodimer, which could potentially bind to the miR193a promoter region to regulate miR193a expression. All structure models of these transcription factors and the tertiary structures of the miR193a promoter region were generated and refined using computational tools. The DNA-protein complexes of the miR193a promoter region and transcription factors were created using a docking approach. To determine the modulatory role of miR193a on APOL1 mRNA, the structural components of APOL1 3’ UTR and miR193a-5p were studied. Molecular Dynamic (MD) simulations validated interactions between miR193a and YY1/WT1/Sox2/VDR/APOL1 3′ UTR region. Undifferentiated podocytes (UPDs) displayed enhanced miR193a, YY1, and Sox2 but attenuated WT1, VDR, and APOL1 expressions, whereas differentiated podocytes (DPDs) exhibited attenuated miR193a, YY1, and Sox2 but increased WT1, VDR, APOL1 expressions. Inhibition of miR193a in UPDs enhanced the expression of APOL1 as well as of podocyte molecular markers; on the other hand, DPD-transfected with miR193a plasmid showed downing of APOL1 as well as podocyte molecular markers suggesting a causal relationship between miR193a and podocyte molecular markers. Silencing of YY1 and Sox2 in UPDs decreased the expression of miR193a but increased the expression of VDR, and CD2AP (a marker of DPDs); in contrast, silencing of WT1 and VDR in DPDs enhanced the expression of miR193a, YY1, and Sox2. Since miR193a-downing by Vitamin D receptor (VDR) agonist not only enhanced the mRNA expression of APOL1 but also of podocyte differentiating markers, suggest that down-regulation of miR193a could be used to enhance the expression of podocyte differentiating markers as a therapeutic strategy.

The In Silico Prediction of Hotspot Residues that Contribute to the Structural Stability of Subunit Interfaces of a Picornavirus Capsid

The assembly of picornavirus capsids proceeds through the stepwise oligomerization of capsid protein subunits and depends on interactions between critical residues known as hotspots. Few studies have described the identification of hotspot residues at the protein subunit interfaces of the picornavirus capsid, some of which could represent novel drug targets. Using a combination of accessible web servers for hotspot prediction, we performed a comprehensive bioinformatic analysis of the hotspot residues at the intraprotomer, interprotomer and interpentamer interfaces of the Theiler’s murine encephalomyelitis virus (TMEV) capsid. Significantly, many of the predicted hotspot residues were found to be conserved in representative viruses from different genera, suggesting that the molecular determinants of capsid assembly are conserved across the family. The analysis presented here can be applied to any icosahedral structure and provides a platform for in vitro mutagenesis studies to further investigate the significance of these hotspots in critical stages of the virus life cycle with a view to identify potential targets for antiviral drug design.

Identification of a Cry1Fa binding site of cadherin in Plutella xylostella through fragment exchanging and molecular docking methods

Binding to the cadherin in target pests is the primary step in the action mechanism of Cry toxins, but little is known regarding the interaction of Cry1Fa with cadherin. Our previous study suggested that a Plutella xylostella cadherin fragment (PxCad-TBR) can bind to Cry1Fa, while its homologous fragment (HaCad-TBR) in Helicoverpa armigera cannot. In this study, we expressed two cadherin fragments that combine parts of PxCad-TBR and HaCad-TBR in Escherichia coli and tested their binding to the Cry1Fa. The results showed that the fragment containing amino acids T1202-A1341 of P. xylostella cadherin showed binding ability to Cry1Fa. Furthermore, two regions (V1219-E1233 and D1326-F1337) were predicted as hot spot regions that are involved in the interaction of Cry1Fa and PxCad-TBR with computer-aided molecular docking. We then constructed two PxCad-TBR mutations by fragment exchanging based on the molecular docking results and verified the mutations' binding abilities to the Cry1Fa. The results showed that the region that contains amino acids D1326-F1337 was one important binding site to Cry1Fa in P. xylostella cadherin. These results suggested that a combination of computer-aided molecular docking and fragment exchanging is an effective way to locate the key binding sites of Bt toxins in receptors.

Hot Spots and Their Contribution to the Self-Assembly of the Viral Capsid: In Silico Prediction and Analysis

The viral capsid is a macromolecular complex formed by a defined number of self-assembled proteins, which, in many cases, are biopolymers with an identical amino acid sequence. Specific protein-protein interactions (PPI) drive the capsid self-assembly process, leading to several distinct protein interfaces. Following the PPI hot spot hypothesis, we present a conservation-based methodology to identify those interface residues hypothesized to be crucial elements on the self-assembly and thermodynamic stability of the capsid. We validate the predictions through a rigorous physical framework which integrates molecular dynamics simulations and free energy calculations by Umbrella sampling and the potential of mean force using an all-atom molecular representation of the capsid proteins of an icosahedral virus in an explicit solvent. Our results show that a single mutation in any of the structure-conserved hot spots significantly perturbs the quaternary protein-protein interaction, decreasing the absolute value of the binding free energy, without altering the protein's secondary nor tertiary structure. Our conservation-based hot spot prediction methodology can lead to strategies to rationally modulate the capsid's thermodynamic properties.

Computational Analysis of Non-covalent Interactions in Phycocyanin Subunit Interfaces

Protein-protein interactions are an important phenomenon in biological processes and functions. We used the manually curated non-redundant dataset of 118 phycocyanin interfaces to gain additional insight into this phenomenon using a robust inter-atomic non-covalent interaction analyzing tool PPCheck. Our observations indicate that there is a relatively high composition of hydrophobic residues at the interfaces. Most of the interface residues are clustered at the middle of the range which we call "standard-size" interfaces. Furthermore, the multiple interaction patterns founded in the present study indicate that more than half of the residues involved in these interactions participate in multiple and water-bridged hydrogen bonds. Thus, hydrogen bonds contribute maximally towards the stability of protein-protein complexes. The analysis shows that hydrogen bond energies contribute to about 88 % to the total energy and it also increases with interface size. Van der Waals (vdW) energy contributes to 9.3 %±1.7 % on average in these complexes. Moreover, there is about 1.9 %±1.5 % contribution by electrostatic energy. Nevertheless, the role by vdW and electrostatic energy could not be ignored in interface binding. Results show that the total binding energy is more for large phycocyanin interfaces. The normalized energy per residue was less than -16 kJ mol^-1 , while most of them have energy in the range from -6 to -14 kJ mol^-1 . The non-covalent interacting residues in these proteins were found to be highly conserved. Obtained results might contribute to the understanding of structural stability of this class of evolutionary essential proteins with increased practical application and future designs of novel protein-bioactive compound interactions.

Antibody Fc engineering for enhanced neonatal Fc receptor binding and prolonged circulation half-life

The neonatal Fc receptor (FcRn) promotes antibody recycling through rescue from normal lysosomal degradation. The binding interaction is pH-dependent with high affinity at low pH, but not under physiological pH conditions. Here, we combined rational design and saturation mutagenesis to generate novel antibody variants with prolonged half-life and acceptable development profiles. First, a panel of saturation point mutations was created at 11 key FcRn-interacting sites on the Fc region of an antibody. Multiple variants with slower FcRn dissociation kinetics than the wildtype (WT) antibody at pH 6.0 were successfully identified. The mutations were further combined and characterized for pH-dependent FcRn binding properties, thermal stability and the FcγRIIIa and rheumatoid factor binding. The most promising variants, YD (M252Y/T256D), DQ (T256D/T307Q) and DW (T256D/T307W), exhibited significantly improved binding to FcRn at pH 6.0 and retained similar binding properties as WT at pH 7.4. The pharmacokinetics in human FcRn transgenic mice and cynomolgus monkeys demonstrated that these properties translated to significantly prolonged plasma elimination half-life compared to the WT control. The novel variants exhibited thermal stability and binding to FcγRIIIa in the range comparable to clinically validated YTE and LS variants, and showed no enhanced binding to rheumatoid factor compared to the WT control. These engineered Fc mutants are promising new variants that are widely applicable to therapeutic antibodies, to extend their circulation half-life with obvious benefits of increased efficacy, and reduced dose and administration frequency.

Extraction of molecular features for the drug discovery targeting protein-protein interaction of Helicobacter pylori CagA and tumor suppressor protein ASSP2

Half of the world population is infected by the Gram-negative bacterium Helicobacter pylori (H. pylori). It colonizes in the stomach and is associated with severe gastric pathologies including gastric cancer and peptic ulceration. The most virulent factor of H. pylori is the cytotoxin-associated gene A (CagA) that is injected into the host cell. CagA interacts with several host proteins and alters their function, thereby causing several diseases. The most well-known target of CagA is the tumor suppressor protein ASPP2. The subdomain I at the N-terminus of CagA interacts with the proline-rich motif of ASPP2. Here, in this study, we carried out alanine scanning mutagenesis and an extensive molecular dynamics simulation summing up to 3.8 μs to find out hot spot residues and discovered some new protein-protein interaction (PPI)-modulating molecules. Our findings are in line with previous biochemical studies and further suggested new residues that are crucial for binding. The alanine scanning showed that mutation of Y207 and T211 residues to alanine decreased the binding affinity. Likewise, dynamics simulation and molecular mechanics with generalized Born surface area (MMGBSA) analysis also showed the importance of these two residues at the interface. A four-feature pharmacophore model was developed based on these two residues, and top 10 molecules were filtered from ZINC, NCI, and ChEMBL databases. The good binding affinity of the CHEMBL17319 and CHEMBL1183979 molecules shows the reliability of our adopted protocol for binding hot spot residues. We believe that our study provides a new insight for using CagA as the therapeutic target for gastric cancer treatment and provides a platform for a future experimental study.

Modeling of Protein⁻Protein Interactions in Cytokinin Signal Transduction

The signaling of cytokinins (CKs), classical plant hormones, is based on the interaction of proteins that constitute the multistep phosphorelay system (MSP): catalytic receptors-sensor histidine kinases (HKs), phosphotransmitters (HPts), and transcription factors-response regulators (RRs). Any CK receptor was shown to interact in vivo with any of the studied HPts and vice versa. In addition, both of these proteins tend to form a homodimer or a heterodimeric complex with protein-paralog. Our study was aimed at explaining by molecular modeling the observed features of in planta protein-protein interactions, accompanying CK signaling. For this purpose, models of CK-signaling proteins' structure from Arabidopsis and potato were built. The modeled interaction interfaces were formed by rather conserved areas of protein surfaces, complementary in hydrophobicity and electrostatic potential. Hot spots amino acids, determining specificity and strength of the interaction, were identified. Virtual phosphorylation of conserved Asp or His residues affected this complementation, increasing (Asp-P in HK) or decreasing (His-P in HPt) the affinity of interacting proteins. The HK-HPt and HPt-HPt interfaces overlapped, sharing some of the hot spots. MSP proteins from Arabidopsis and potato exhibited similar properties. The structural features of the modeled protein complexes were consistent with the experimental data.

Computational analysis of hot spots and binding mechanism in the PD-1/PD-L1 interaction

Programmed cell death protein-1 (PD-1) is an important immunological checkpoint and plays a vital role in maintaining the peripheral tolerance of the human body by interacting with its ligand PD-L1. The overexpression of PD-L1 in tumor cells induces local immune suppression and helps the tumor cells to evade the endogenous anti-tumor immunity. Developing monoclonal antibodies against the PD-1/PD-L1 protein–protein interaction to block the PD-1/PD-L1 signaling pathway has demonstrated superior anti-tumor efficacy in a variety of solid tumors and has made a profound impact on the field of cancer immunotherapy in recent years. Although the X-ray crystal structure of the PD-1/PD-L1 complex has been solved, the detailed binding mechanism of the PD-1/PD-L1 interaction is not fully understood from a theoretical point of view. In this study, we performed computational alanine scanning on the PD-1/PD-L1 complex to quantitatively identify the hot spots in the PD-1/PD-L1 interaction and characterize its binding mechanisms at the atomic level. To the best of our knowledge, this is the first time that theoretical calculations have been used to systematically and quantitatively predict the hot spots in the PD-1/PD-L1 interaction. We hope that the predicted hot spots and the energy profile of the PD-1/PD-L1 interaction presented in this work can provide guidance for the design of peptide and small molecule drugs targeting PD-1 or PD-L1.

The hot spots quantitatively predicted by the recently developed MM/GBSA/IE method reveal a hydrophobic core in the PD-1/PD-L1 interaction.

Identifying Driver Interfaces Enriched for Somatic Missense Mutations in Tumors

Human cancers often harbor large numbers of somatic mutations. However, only a small proportion of these mutations are expected to contribute to tumor growth and progression. Therefore, determining causal driver mutations and the genes they target is becoming an important challenge in cancer genomics. Here we describe an approach for mapping somatic mutations onto 3D structures of human proteins in complex to identify "driver interfaces." Our strategy relies on identifying protein-interaction interfaces that are unexpectedly biased toward nonsynonymous mutations, which suggests that these interfaces are subject to positive selection during tumorigenesis, implicating the interacting proteins as candidate drivers.

Machine Learning Approaches for Protein⁻Protein Interaction Hot Spot Prediction: Progress and Comparative Assessment

Hot spots are the subset of interface residues that account for most of the binding free energy, and they play essential roles in the stability of protein binding. Effectively identifying which specific interface residues of protein–protein complexes form the hot spots is critical for understanding the principles of protein interactions, and it has broad application prospects in protein design and drug development. Experimental methods like alanine scanning mutagenesis are labor-intensive and time-consuming. At present, the experimentally measured hot spots are very limited. Hence, the use of computational approaches to predicting hot spots is becoming increasingly important. Here, we describe the basic concepts and recent advances of machine learning applications in inferring the protein–protein interaction hot spots, and assess the performance of widely used features, machine learning algorithms, and existing state-of-the-art approaches. We also discuss the challenges and future directions in the prediction of hot spots.

Enhanced Prediction of Hot Spots at Protein-Protein Interfaces Using Extreme Gradient Boosting

Identification of hot spots, a small portion of protein-protein interface residues that contribute the majority of the binding free energy, can provide crucial information for understanding the function of proteins and studying their interactions. Based on our previous method (PredHS), we propose a new computational approach, PredHS2, that can further improve the accuracy of predicting hot spots at protein-protein interfaces. Firstly we build a new training dataset of 313 alanine-mutated interface residues extracted from 34 protein complexes. Then we generate a wide variety of 600 sequence, structure, exposure and energy features, together with Euclidean and Voronoi neighborhood properties. To remove redundant and irrelevant information, we select a set of 26 optimal features utilizing a two-step feature selection method, which consist of a minimum Redundancy Maximum Relevance (mRMR) procedure and a sequential forward selection process. Based on the selected 26 features, we use Extreme Gradient Boosting (XGBoost) to build our prediction model. Performance of our PredHS2 approach outperforms other machine learning algorithms and other state-of-the-art hot spot prediction methods on the training dataset and the independent test set (BID) respectively. Several novel features, such as solvent exposure characteristics, second structure features and disorder scores, are found to be more effective in discriminating hot spots. Moreover, the update of the training dataset and the new feature selection and classification algorithms play a vital role in improving the prediction quality.

Structures of Hepatitis B Virus Core- and e-Antigen Immune Complexes Suggest Multi-point Inhibition

Hepatitis B virus (HBV) is the leading cause of liver disease worldwide. While an adequate vaccine is available, current treatment options are limited, not highly effective, and associated with adverse effects, encouraging the development of alternative therapeutics. The HBV core gene encodes two different proteins: core, which forms the viral nucleocapsid, and pre-core, which serves as an immune modulator with multiple points of action. The two proteins mostly have the same sequence, although they differ at their N and C termini and in their dimeric arrangements. Previously, we engineered two human-framework antibody fragments (Fab/scFv) with nano- to picomolar affinities for both proteins. Here, by means of X-ray crystallography, analytical ultracentrifugation, and electron microscopy, we demonstrate that the antibodies have non-overlapping epitopes and effectively block biologically important assemblies of both proteins. These properties, together with the anticipated high tolerability and long half-lives of the antibodies, make them promising therapeutics.

Evolution of In Silico Strategies for Protein-Protein Interaction Drug Discovery

The advent of advanced molecular modeling software, big data analytics, and high-speed processing units has led to the exponential evolution of modern drug discovery and better insights into complex biological processes and disease networks. This has progressively steered current research interests to understanding protein-protein interaction (PPI) systems that are related to a number of relevant diseases, such as cancer, neurological illnesses, metabolic disorders, etc. However, targeting PPIs are challenging due to their "undruggable" binding interfaces. In this review, we focus on the current obstacles that impede PPI drug discovery, and how recent discoveries and advances in in silico approaches can alleviate these barriers to expedite the search for potential leads, as shown in several exemplary studies. We will also discuss about currently available information on PPI compounds and systems, along with their usefulness in molecular modeling. Finally, we conclude by presenting the limits of in silico application in drug discovery and offer a perspective in the field of computer-aided PPI drug discovery.

Protein-protein interface hot spots prediction based on a hybrid feature selection strategy

Background

Hot spots are interface residues that contribute most binding affinity to protein-protein interaction. A compact and relevant feature subset is important for building machine learning methods to predict hot spots on protein-protein interfaces. Although different methods have been used to detect the relevant feature subset from a variety of features related to interface residues, it is still a challenge to detect the optimal feature subset for building the final model.

Results

In this study, three different feature selection methods were compared to propose a new hybrid feature selection strategy. This new strategy was proved to effectively reduce the feature space when we were building the prediction models for identifying hotspot residues. It was tested on eighty-two features, both conventional and newly proposed. According to the strategy, combining the feature subsets selected by decision tree and mRMR (maximum Relevance Minimum Redundancy) individually, we were able to build a model with 6 features by using a PSFS (Pseudo Sequential Forward Selection) process. Compared with other state-of-art methods for the independent test set, our model had shown better or comparable predictive performances (with F-measure 0.622 and recall 0.821). Analysis of the 6 features confirmed that our newly proposed feature CNSV_REL1 was important for our model. The analysis also showed that the complementarity between features should be considered as an important aspect when conducting the feature selection.

Conclusion

In this study, most important of all, a new strategy for feature selection was proposed and proved to be effective in selecting the optimal feature subset for building prediction models, which can be used to predict hot spot residues on protein-protein interfaces. Moreover, two aspects, the generalization of the single feature and the complementarity between features, were proved to be of great importance and should be considered in feature selection methods. Finally, our newly proposed feature CNSV_REL1 had been proved an alternative and effective feature in predicting hot spots by our study. Our model is available for users through a webserver: http://zhulab.ahu.edu.cn/iPPHOT/.

Electronic supplementary material

The online version of this article (10.1186/s12859-018-2009-5) contains supplementary material, which is available to authorized users.

Prediction of Protein Hotspots from Whole Protein Sequences by a Random Projection Ensemble System

Hotspot residues are important in the determination of protein-protein interactions, and they always perform specific functions in biological processes. The determination of hotspot residues is by the commonly-used method of alanine scanning mutagenesis experiments, which is always costly and time consuming. To address this issue, computational methods have been developed. Most of them are structure based, i.e., using the information of solved protein structures. However, the number of solved protein structures is extremely less than that of sequences. Moreover, almost all of the predictors identified hotspots from the interfaces of protein complexes, seldom from the whole protein sequences. Therefore, determining hotspots from whole protein sequences by sequence information alone is urgent. To address the issue of hotspot predictions from the whole sequences of proteins, we proposed an ensemble system with random projections using statistical physicochemical properties of amino acids. First, an encoding scheme involving sequence profiles of residues and physicochemical properties from the AAindex1 dataset is developed. Then, the random projection technique was adopted to project the encoding instances into a reduced space. Then, several better random projections were obtained by training an IBk classifier based on the training dataset, which were thus applied to the test dataset. The ensemble of random projection classifiers is therefore obtained. Experimental results showed that although the performance of our method is not good enough for real applications of hotspots, it is very promising in the determination of hotspot residues from whole sequences.