Harnessing protein folding neural networks for peptide–protein docking

Highly accurate protein structure predictions by deep neural networks such as AlphaFold2 and RoseTTAFold have tremendous impact on structural biology and beyond. Here, we show that, although these deep learning approaches have originally been developed for the in silico folding of protein monomers, AlphaFold2 also enables quick and accurate modeling of peptide–protein interactions. Our simple implementation of AlphaFold2 generates peptide–protein complex models without requiring multiple sequence alignment information for the peptide partner, and can handle binding-induced conformational changes of the receptor. We explore what AlphaFold2 has memorized and learned, and describe specific examples that highlight differences compared to state-of-the-art peptide docking protocol PIPER-FlexPepDock. These results show that AlphaFold2 holds great promise for providing structural insight into a wide range of peptide–protein complexes, serving as a starting point for the detailed characterization and manipulation of these interactions.

Holotoxin disassembly by protein disulfide isomerase is less efficient for Escherichia coli heat-labile enterotoxin than cholera toxin

Cholera toxin (CT) and Escherichia coli heat-labile enterotoxin (LT) are structurally similar AB5-type protein toxins. They move from the cell surface to the endoplasmic reticulum where the A1 catalytic subunit is separated from its holotoxin by protein disulfide isomerase (PDI), thus allowing the dissociated A1 subunit to enter the cytosol for a toxic effect. Despite similar mechanisms of toxicity, CT is more potent than LT. The difference has been attributed to a more stable domain assembly for CT as compared to LT, but this explanation has not been directly tested and is arguable as toxin disassembly is an indispensable step in the cellular action of these toxins. We show here that PDI disassembles CT more efficiently than LT, which provides a possible explanation for the greater potency of the former toxin. Furthermore, direct examination of CT and LT domain assemblies found no difference in toxin stability. Using novel analytic geometry approaches, we provide a detailed characterization of the positioning of the A subunit with respect to the B pentamer and demonstrate significant differences in the interdomain architecture of CT and LT. Protein docking analysis further suggests that these global structural differences result in distinct modes of PDI-toxin interactions. Our results highlight previously overlooked structural differences between CT and LT that provide a new model for the PDI-assisted disassembly and differential potency of these toxins.

Explanation of Structure and Function of kv1.3 Potent Blocker From Mesobuthus eupeus Venom Gland: A New Promise in Drug Development

Background: Scorpions and other venomous animals are sought with great concern because venom is a source of novel peptides with exciting features. Some toxins of scorpion venom are effectors of potassium channels. Previous studies strongly support the importance of potassium channel toxins for use as pharmacological tools or potential drugs. Objectives: Here, a three-dimensional (3-D) structure and function of a potent acidic blocker of the human voltage-gated potassium ion channel, Kv1.3, previously identified in the scorpion Mesobuthus eupeus venom gland, were interpreted. Methods: The 3-D structure of meuK2-2 was generated using homology modeling. The interaction of meuK2-2 with the Kv1.3 channel was evaluated using a computational protocol employing peptide-protein docking experiments, pose clustering, and 100 ns molecular dynamic simulations to make the 3-D models of the meuK2-2/Kv1.3 complex trustworthy. Results: A CSα/β (cysteine-stabilized α-helical and β-sheet) fold was found for the 3-D structure of meuK2-2. In a different mechanism from what was identified so far, meuK2-2 binds to both turret and pore loop of Kv1.3 through two key residues (Ala28 and Ser11) and H-bonds. The binding of meuK2-2 induces some conformational changes to Kv1.3. Eventually, the side chain of a positively charged amino acid (His9) occupies the channel's pore. All together blocks the ion permeation pathway. Conclusions: MeuK2-2 could block Kv1.3 by a new mechanism. So, it could be a unique target for further investigations to develop a pharmacological tool and potential drug.

Analyzing the interaction of human ACE2 and RBD of spike protein of SARS-CoV-2 in perspective of Omicron variant

The newly identified Omicron (B.1.1.529) variant of Severe Acute Respiratory Syndrome Voronavirus 2 (SARS-CoV-2) has steered concerns across the world due to the possession of large number of mutations leading to high infectivity and vaccine escape potential. The Omicron variant houses 32 mutations in S protein alone. The viral infectivity is determined mainly by the ability of spike (S) protein receptor binding domain (RBD) to bind to the human Angiotensin I Converting Enzyme 2 (hACE2) receptor. In this paper, the interaction of the RBDs of SARS-CoV-2 variants with hACE2 was analyzed by using protein-protein docking and compared with the novel Omicron variant. Our findings reveal that the Omicron RBD interacts strongly with hACE2 receptor via unique amino acid residues as compared to the Wuhan and many other variants. However, the interacting residues of RBD are found to be the same in Lamda (C.37) variant. These unique binding of Omicron RBD with hACE2 suggests an increased potential of infectivity and vaccine evasion potential of the new variant. The evolutionary drive of the SARS-CoV-2 may not be exclusively driven by RBD variants but surely provides for the platform for emergence of new variants.

Analysis of SARS-CoV-2 Spike Protein as The Key Target in the Development of Antiviral Candidates for COVID-19 through Computational Study

The recent public health crisis is threatening the world with the emergence of the spread of the new coronavirus 2019 (2019-nCoV) or severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This virus originates from bats and is transmitted to humans through unknown intermediate animals in Wuhan, China in December 2019. Advances in technology have opened opportunities to find candidates for natural compounds capable of preventing and controlling COVID-19 infection through inhibition of spike proteins of SARS-CoV-2. This research aims to identify, evaluate, and explore the structure of spike protein macromolecules from three coronaviruses (SARS-CoV, MERS-CoV, and SARS-CoV-2) and their effects on Angiotensin-Converting Enzyme 2 (ACE-2) using computational studies. Based on the identification of the three spike protein macromolecules, it was found that there was a similarity between the active binding sites of ACE-2. These observations were then confirmed using a protein-docking simulation to observe the interaction of the protein spike to the active site of ACE-2. SARS-COV-2 spike protein has the strongest bond to ACE-2, with an ACE score of −1341.85 kJ/mol. Therefore, some of this information from the results of this research can be used as a reference in the development of competitive inhibitor candidates for SARS-CoV-2 spike proteins for the treatment of COVID-19 infectious diseases.

Molecular Docking and Interaction Studies of Identified Abscisic Acid Receptors in Oryza sativa: An In-Silico Perspective on Comprehending Stress Tolerance Mechanisms

Background: Abiotic stresses affect plants in several ways and as such, phytohormones such as abscisic acid (ABA) play an important role in conferring tolerance towards these stresses. Hence, to comprehend the role of ABA and its interaction with receptors of the plants, a thorough investigation is essential. Aim: The current study aimed to identify the ABA receptors in Oryza sativa, to find the receptor that binds best with ABA and to examine the mutations present to help predict better binding of the receptors with ABA Methods: Protein sequences of twelve PYL (Pyrabactin resistance 1) and seven PP2C (type 2C protein phosphatase) receptors were retrieved from Rice Annotation Project database and their 3D structures were predicted using RaptorX. Protein-ligand molecular docking studies between PYL and ABA was performed using AutoDock 1.5.6, followed by 100ns molecular dynamic simulation studies using Desmond to determine the acceptable conformational changes after docking via root mean square deviation RMSD plot analysis. Protein-protein docking was then carried out in three sets: PYL-PP2Cs, PYL-ABA-PP2C and PYL(mut)-ABA-PP2C to scrutinize changes in structural conformations and binding energies between complexes. The amino acids of interest were mapped at its respective genomic coordinates using SNP-seek database to ascertain if there were any naturally occurring single nucleotide polymorphisms (SNPs) responsible for triggering rice PYLs mutations Results: Initial protein-ligand docking studies revealed good binding between the complexes, wherein PYL6-ABA complex showed the best energy of -8.15 kcal/mol. The 100ns simulation studies revealed changes in the RMSD values after docking, indicating acceptable conformational changes. Furthermore, mutagenesis study performed at specific PYL-ABA interacting residues followed by downstream PYL(mut)-ABA-PP2C protein-protein docking results after induction of mutations demonstrated a binding energy of -8.17 kcal/mol for PP2C79-PYL11-ABA complex. No naturally occurring SNPs that were responsible for triggering rice PYL mutations were identified when specific amino acid coordinates were mapped at respective genomic coordinates. Conclusion: Thus, the present study provides valuable insights on the interactions of ABA receptors in rice and induced mutations in PYL11 that can enhance the downstream interaction with PP2C

Molecular Modeling, Interacting Residues and other Structural Analyses for Human SOCS3, Gp130 and JAK Proteins: A Detailed Computational Approach for Proteins Involved in Feedback Inhibition

Background: When STAT3 is activated only by the IL6 family of proteins, then gp130 (having a phosphopeptide motif) interacts with human SOCS3 which further binds to JAK and inhibits its protein kinase activity. Interaction of gp130 with SOCS3 targets only the IL-6 signaling cascade. The interaction occurs when SOCS3 binds to a particular motif on gp130 (centered upon pTyr759) after its phosphorylation. Previously, wet laboratory studies were done but computational exploration for the participating residues remained unexplored. Methodology: The 3D structure of human SOCS3 protein was modeled and its stereo-chemical parameters were satisfied. Crystallographic structures of gp130-phosphopeptide and JAK were studied. After protein docking, the complex underwent minimization and molecular dynamics simulation. Different stability parameters and binding patterns with residues were evaluated Results, Discussion and Conclusion: The best modeled structure of SOCS3 protein was selected and found that it had three helices and seven sheets interspersed with coils. Arg133, Tyr137 and Tyr98 from SOCS3 formed manifold binding patterns with gp130 (mainly with pTyr759 and Glu758). Lys62, Lys63 and Arg65 from SOCS3 were also found to interact with Val762 of gp130. Interactions with JAK were also studied. Residue 53, 62-65, 98, 133, 136 and 137 formed the predominant binding pockets in SOCS3. They can serve as important target sites as well. Altogether, it created elctrostatically charged pockets to accommodate the partner proteins for each other. Gp130 phosphopeptide was observed to be tightly accommodated in the electrostatically positive zones on SOCS3 surface. Net area for solvent accessibility was also found to get drastically reduced implying high participation of residues. Earlier studies documented that the interaction of these three proteins occurs with affinity and have satisfactory association with each other. Here in this study, free energy of binding for the triple protein interaction through the ΔG values helped to infer that SOCS3 interacted spontaneously (in thermodynamic sense). Many helical conformations formed coiled-coils providing high flexibility to interact spontaneously. Most of the interactions were through the responsible SH2 domain (46-127 residue length) of SOCS3. Residues 53, 62-64 and 98 formed coils while the residue number 137adopted sheet conformation from coils. Future Scope: This study shall instigate to block the gp130-binding sites of SOCS3 through targeting of drugs, thereby preventing SOCS3-gp130 interaction. This would allow JAK-STAT signaling cascade which is paramount for several biological functions

Molecular Docking and Validation of Methicillin Resistant Staphylococcus aureus Targets against Geninthiocin

Antibiotic resistance has been a serious public health concern in recent years. Methicillin resistant “Staphylococcus aureus” (MRSA) is a superbug that causes life threatening infections of Humanity which is difficult to treat. Geninthiocin is a macrocyclic thiopeptide with a 35-membered core moiety, which was isolated from marine streptomyces sp. ICN19, which has proven potent activity against MRSA. Five target proteins PDB ID: 4YMX, 3ZDS, 3QLB, 4IEN and 1DXL were identified from MRSA for their presumptive action for Geninthiocin. In this study, we used molecular docking and molecular dynamic simulation, in order to validate Geninthiocin’s potential target protein. Target proteins were subjected to ligand-protein docking studies. Based on their docking scores and Hydrogen bonding interactions, two possible proteins 4YMX and 3ZDS were further subjected to simulation strategies to validate the protein-drug interaction. Out of which, homogentisate1,2 dioxygenase turned out to be a possible drug target for Geninthiocin. The compound Geninthiocin could be developed as a potential inhibitor against the target protein homogentisate1,2-dioxygenase for exhibiting an effective antimicrobial activity.

InterPepScore: A Deep Learning Score for Improving the FlexPepDock Refinement Protocol

Motivation: Interactions between peptide fragments and protein receptors are vital to cell function yet difficult to experimentally determine the structural details of. As such, many computational methods have been developed to aid in peptide-protein docking or structure prediction. One such method is Rosetta FlexPepDock which consistently refines coarse peptide-protein models into sub-Ångström precision using Monte-Carlo simulations and statistical potentials. Deep learning has recently seen increased use in protein structure prediction, with graph neural network seeing use in protein model quality assessment. Results: Here, we introduce a graph neural network, InterPepScore, as an additional scoring term to complement and improve the Rosetta FlexPepDock refinement protocol. InterPepScore is trained on simulation trajectories from FlexPepDock refinement starting from thousands of peptide-protein complexes generated by a wide variety of docking schemes. The addition of InterPepScore into the refinement protocol consistently improves the quality of models created, and on an independent benchmark on 109 peptide-protein complexes its inclusion results in an increase in the number of complexes for which the top-scoring model had a DockQ-score of 0.49 (Medium quality) or better from 14.8% to 26.1%.