The Subtle Trade-Off between Evolutionary and Energetic Constraints in Protein-Protein Interactions

Design and Test of Molecules that Interfere with the Recognition Mechanisms between the SARS-CoV-2 Spike Protein and Its Host Cell Receptors

The disruptive impact of the COVID-19 pandemic has led the scientific community to undertake an unprecedented effort to characterize viral infection mechanisms. Among these, interactions between the viral glycosylated Spike and the human receptors ACE2 and TMPRSS2 are key to allowing virus invasion. Here, we report and test a fully rational methodology to design molecules that are capable of perturbing the interactions between these critical players in SARS-CoV-2 pathogenicity. To this end, we computationally identify substructures on the fully glycosylated Spike protein that are not intramolecularly optimized and are thus prone to being stabilized by forming complexes with ACE2 and TMPRSS2. With the aim of competing with the Spike-mediated cell entry mechanisms, we have engineered the predicted putative interaction regions in the form of peptide mimics that could compete with Spike for interaction with ACE2 and/or TMPRSS2. Experimental models of viral entry demonstrate that the designed molecules are able to interfere with viral entry into ACE2/TMPRSS2 expressing cells, while they have no effects on the entry of control viral particles that do not harbor the Spike protein or on the entry of Spike-presenting viral particles into cells that do not display its receptors on their surface.

Conformational Dynamics, Energetics, and the Divergent Evolution of Allosteric Regulation: The Case of the Yeast MAPK Family

Allosteric mechanisms provide finely-tuned control over signalling proteins. Proteins of the same family may share high sequence identity and structural similarity but show distinct traits of allosteric control and evolutionary divergent regulation. Revealing the determinants of such properties may be important to understand the molecular bases of different regulatory pathways. Herein, we investigate whether and how evolutionarily-divergent traits of allosteric regulation in homologous proteins can be decoded in terms of internal dynamics and interaction networks that support functionally oriented conformations. In this framework, we start from the comparative analysis of the dynamics and energetics of the yeast MAP Kinases (MAPKs) Fus3 and Kss1 in their native basins. Importantly, distinctive dynamic and energetic stabilization features emerge, which can be related to the two proteins' differential ability to be phosphorylated and engage with the allosteric activator Ste5. We then expanded our study to other evolutionarily-related MAPKs. We show that the dynamical and energetical traits defining the distinct regulatory profiles of Fus3 and Kss1 can be traced along their evolutionary tree. Overall, our approach is able to reconnect (latent) allostery with the principal elements of protein structural stabilization and dynamics, showing how allosteric regulation was encrypted in MAPKs structure well before Ste5 appearance.

NLRP3 monomer functional dynamics: From the effects of allosteric binding to implications for drug design

The protein NLRP3 and its complexes are associated with an array of inflammatory pathologies, among which neurodegenerative, autoimmune, and metabolic diseases. Targeting the NLRP3 inflammasome represents a promising strategy for easing the symptoms of pathologic neuroinflammation. When the inflammasome is activated, NLRP3 undergoes a conformational change triggering the production of pro-inflammatory cytokines IL-1β and IL-18, as well as cell death by pyroptosis. NLRP3 nucleotide-binding and oligomerization (NACHT) domain plays a crucial role in this function by binding and hydrolysing ATP and is primarily responsible, together with conformational transitions involving the PYD domain, for the complex-assembly process. Allosteric ligands proved able to induce NLRP3 inhibition. Herein, we examine the origins of allosteric inhibition of NLRP3. Through the use of molecular dynamics (MD) simulations and advanced analysis methods, we provide molecular-level insights into how allosteric binding affects protein structure and dynamics, remodelling of the conformational ensembles populated by the protein, with key reverberations on how NLRP3 is preorganized for assembly and ultimately function. The data are used to develop a Machine Learning model to define the protein as Active or Inactive, only based on the analysis of its internal dynamics. We propose this model as a novel tool to select allosteric ligands.

Integrating Protein Interaction Surface Prediction with a Fragment-Based Drug Design: Automatic Design of New Leads with Fragments on Energy Surfaces

Protein-protein interactions (PPIs) have emerged in the past years as significant pharmacological targets in the development of new therapeutics due to their key roles in determining pathological pathways. Herein, we present fragments on energy surfaces, a simple and general design strategy that integrates the analysis of the dynamic and energetic signatures of proteins to unveil the substructures involved in PPIs, with docking, selection, and combination of drug-like fragments to generate new PPI inhibitor candidates. Specifically, structural representatives of the target protein are used as inputs for the blind physics-based prediction of potential protein interaction surfaces using the matrix of low coupling energy decomposition method. The predicted interaction surfaces are subdivided into overlapping windows that are used as templates to direct the docking and combination of fragments representative of moieties typically found in active drugs. This protocol is then applied and validated using structurally diverse, important PPI targets as test systems. We demonstrate that our approach facilitates the exploration of the molecular diversity space of potential ligands, with no requirement of prior information on the location and properties of interaction surfaces or on the structures of potential lead compounds. Importantly, the hit molecules that emerge from our ab initio design share high chemical similarity with experimentally tested active PPI inhibitors. We propose that the protocol we describe here represents a valuable means of generating initial leads against difficult targets for further development and refinement.

Computational Epitope Prediction and Design for Antibody Development and Detection

The design of optimized protein antigens is a fundamental step in the development of new vaccine candidates and in the detection of therapeutic antibodies. A fundamental prerequisite is the identification of antigenic regions that are most prone to interact with antibodies, namely, B-cell epitopes. Here, we describe an efficient structure-based computational method for epitope prediction, called MLCE. In this approach, all that is required is the 3D structure of the antigen of interest. MLCE can be applied to glycosylated proteins, facilitating the identification of immunoreactive versus immune-shielding carbohydrates.

Evaluation of docking procedures reliability in affitins-partners interactions

Affitins constitute a class of small proteins belonging to Sul7d family, which, in microorganisms such as Sulfolobus acidocaldarius, bind DNA preventing its denaturation. Thanks to their stability and small size (60–66 residues in length) they have been considered as ideal candidates for engineering and have been used for more than 10 years now, for different applications. The individuation of a mutant able to recognize a specific target does not imply the knowledge of the binding geometry between the two proteins. However, its identification is of undoubted importance but not always experimentally accessible. For this reason, computational approaches such as protein-protein docking can be helpful for an initial structural characterization of the complex. This method, which produces tens of putative binding geometries ordered according to a binding score, needs to be followed by a further reranking procedure for finding the most plausible one. In the present paper, we use the server ClusPro for generating docking models of affitins with different protein partners whose experimental structures are available in the Protein Data Bank. Then, we apply two protocols for reranking the docking models. The first one investigates their stability by means of Molecular Dynamics simulations; the second one, instead, compares the docking models with the interacting residues predicted by the Matrix of Local Coupling Energies method. Results show that the more efficient way to deal with the reranking problem is to consider the information given by the two protocols together, i.e. employing a consensus approach.

SARS-CoV-2 Spike Protein Mutations and Escape from Antibodies: A Computational Model of Epitope Loss in Variants of Concern

The SARS-CoV-2 spike (S) protein is exposed on the viral surface and is the first point of contact between the virus and the host. For these reasons it represents the prime target for Covid-19 vaccines. In recent months, variants of this protein have started to emerge. Their ability to reduce or evade recognition by S-targeting antibodies poses a threat to immunological treatments and raises concerns for their consequences on vaccine efficacy. To develop a model able to predict the potential impact of S-protein mutations on antibody binding sites, we performed unbiased multi-microsecond molecular dynamics of several glycosylated S-protein variants and applied a straightforward structure-dynamics-energy based strategy to predict potential changes in immunogenic regions on each variant. We recover known epitopes on the reference D614G sequence. By comparing our results, obtained on isolated S-proteins in solution, to recently published data on antibody binding and reactivity in new S variants, we directly show that modifications in the S-protein consistently translate into the loss of potentially immunoreactive regions. Our findings can thus be qualitatively reconnected to the experimentally characterized decreased ability of some of the Abs elicited against the dominant S-sequence to recognize variants. While based on the study of SARS-CoV-2 spike variants, our computational epitope-prediction strategy is portable and could be applied to study immunoreactivity in mutants of proteins of interest whose structures have been characterized, helping the development/selection of vaccines and antibodies able to control emerging variants.

Protein-protein interactions at a glance: Protocols for the visualization of biomolecular interactions

Protein-protein interactions (PPIs) play a key role in many biological processes and are intriguing targets for drug discovery campaigns. Advancements in experimental and computational techniques are leading to a growth of data accessibility, and, with it, an increased need for the analysis of PPIs. In this respect, visualization tools are essential instruments to represent and analyze biomolecular interactions. In this chapter, we reviewed some of the available tools, highlighting their features, and describing their functions with practical information on their usage.

The Answer Lies in the Energy: How Simple Atomistic Molecular Dynamics Simulations May Hold the Key to Epitope Prediction on the Fully Glycosylated SARS-CoV-2 Spike Protein

SARS-CoV-2 is a health threat with dire socioeconomical consequences. As the crucial mediator of infection, the viral glycosylated spike protein (S) has attracted the most attention and is at the center of efforts to develop therapeutics and diagnostics. Herein, we use an original decomposition approach to identify energetically uncoupled substructures as antibody binding sites on the fully glycosylated S. Crucially, all that is required are unbiased MD simulations; no prior knowledge of binding properties or ad hoc parameter combinations is needed. Our results are validated by experimentally confirmed structures of S in complex with anti- or nanobodies. We identify poorly coupled subdomains that are poised to host (several) epitopes and potentially involved in large functional conformational transitions. Moreover, we detect two distinct behaviors for glycans: those with stronger energetic coupling are structurally relevant and protect underlying peptidic epitopes, and those with weaker coupling could themselves be prone to antibody recognition.

Simple Model of Protein Energetics To Identify Ab Initio Folding Transitions from All-Atom MD Simulations of Proteins

A fundamental requirement to predict the native conformation, address questions of sequence design and optimization, and gain insights into the folding mechanisms of proteins lies in the definition of an unbiased reaction coordinate that reports on the folding state without the need to compare it to reference values, which might be unavailable for new (designed) sequences. Here, we introduce such a reaction coordinate, which does not depend on previous structural knowledge of the native state but relies solely on the energy partition within the protein: the spectral gap of the pair nonbonded energy matrix (ENergy Gap, ENG). This quantity can be simply calculated along unbiased MD trajectories. We show that upon folding the gap increases significantly, while its fluctuations are reduced to a minimum. This is consistently observed for a diverse set of systems and trajectories. Our approach allows one to promptly identify residues that belong to the folding core as well as residues involved in non-native contacts that need to be disrupted to guide polypeptides to the folded state. The energy gap and fluctuations criteria are then used to develop an automatic detection system which allows us to extract and analyze folding transitions from a generic MD trajectory. We speculate that our method can be used to detect conformational ensembles in dynamic and intrinsically disordered proteins, revealing potential preorganization for binding.

Chemical Perturbation of Oncogenic Protein Folding: from the Prediction of Locally Unstable Structures to the Design of Disruptors of Hsp90-Client Interactions

Protein folding quality control in cells requires the activity of a class of proteins known as molecular chaperones. Heat shock protein-90 (Hsp90), a multidomain ATP driven molecular machine, is a prime representative of this family of proteins. Interactions between Hsp90, its co-chaperones, and client proteins have been shown to be important in facilitating the correct folding and activation of clients. Hsp90 levels and functions are elevated in tumor cells. Here, we computationally predict the regions on the native structures of clients c-Abl, c-Src, Cdk4, B-Raf and Glucocorticoid Receptor, that have the highest probability of undergoing local unfolding, despite being ordered in their native structures. Such regions represent potential ideal interaction points with the Hsp90-system. We synthesize mimics spanning these regions and confirm their interaction with partners of the Hsp90 complex (Hsp90, Cdc37 and Aha1) by Nuclear Magnetic Resonance (NMR). Designed mimics selectively disrupt the association of their respective clients with the Hsp90 machinery, leaving unrelated clients unperturbed and causing apoptosis in cancer cells. Overall, selective targeting of Hsp90 protein-protein interactions is achieved without causing indiscriminate degradation of all clients, setting the stage for the development of therapeutics based on specific chaperone:client perturbation.

Refinement of pairwise potentials via logistic regression to score protein-protein interactions

Protein-protein interactions (PPIs) are ubiquitous and functionally of great importance in biological systems. Hence, the accurate prediction of PPIs by protein-protein docking and scoring tools is highly desirable in order to characterize their structure and biological function. Ab initio docking protocols are divided into the sampling of docking poses to produce at least one near-native structure, and then to evaluate the vast candidate structures by scoring. Concurrent development in both sampling and scoring is crucial for the deployment of protein-protein docking software. In the present work, we apply a machine learning model on pairwise potentials to refine the task of protein quaternary structure native structure detection among decoys. A decoy set was featurized using the Knowledge and Empirical Combined Scoring Algorithm 2 (KECSA2) pairwise potential. The highly unbalanced decoy set was then balanced using a comparison concept between native and decoy structures. The resultant comparison descriptors were used to train a logistic regression (LR) classifier. The LR model yielded the optimal performance for native detection among decoys compared with conventional scoring functions, while exhibiting lesser performance for the detection of low root mean square deviation decoy structures. Its deployment on an independent benchmark set confirms that the scoring function performs competitively relative to other scoring functions. The scripts used are available at https://github.com/TanemuraKiyoto/PPI-native-detection-via-LR.

Coevolutionary data-based interaction networks approach highlighting key residues across protein families: The case of the G-protein coupled receptors

We present an approach that, by integrating structural data with Direct Coupling Analysis, is able to pinpoint most of the interaction hotspots (i.e. key residues for the biological activity) across very sparse protein families in a single run. An application to the Class A G-protein coupled receptors (GPCRs), both in their active and inactive states, demonstrates the predictive power of our approach. The latter can be easily extended to any other kind of protein family, where it is expected to highlight most key sites involved in their functional activity.

Small protein-protein interfaces rich in electrostatic are often linked to regulatory function

Protein-protein interaction (PPI) is critical for several biological functions in living cells through the formation of an interface. Therefore, it is of interest to characterize protein-protein interfaces using an updated non-redundant structural dataset of 2557 homo (identical subunits) and 393 hetero (different subunits) dimer protein complexes determined by X-ray crystallography. We analyzed the interfaces using van der Waals (vdW), hydrogen bonding and electrostatic energies. Results show that on average homo and hetero interfaces are similar. Hence, we further grouped the 2950 interfaces based on percentage vdW to total energies into dominant (≥60%) and sub-dominant (<60%) vdW interfaces. Majority (92%) of interfaces have dominant vdW energy with large interface size (146 ± 87 (homo) and 137 ± 76 (hetero) residues) and interface area (1622 ± 1135 Ų (homo) and 1579 ± 1060 Ų (hetero)). However, a proportion (8%) of interfaces have sub-dominant vdW energy with small interface size (85 ± 46 (homo) and 88 ± 36 (hetero) residues) and interface area (823 ± 538 Ų (homo) and 881 ± 377 Ų (hetero)). It is found that large interfaces have two-fold more interface area and interface size than small interfaces with increasing hydrogen bonding energy to interface size. However, small interfaces have three-fold more electrostatics energy than large interfaces with increasing electrostatics to interface size. Thus, 8% of complexes having small interfaces with limited interface area and sub-dominant vdW energy are rich in electrostatics. It is interesting to observe that complexes having small interfaces are often associated with regulatory function. Hence, the observed structural features with known molecular function provide insights for the better understanding of PPI.Communicated by Ramaswamy H. Sarma.

Interfaces Between Alpha-helical Integral Membrane Proteins: Characterization, Prediction, and Docking

Protein-protein interaction (PPI) is an essential mechanism by which proteins perform their biological functions. For globular proteins, the molecular characteristics of such interactions have been well analyzed, and many computational tools are available for predicting PPI sites and constructing structural models of the complex. In contrast, little is known about the molecular features of the interaction between integral membrane proteins (IMPs) and few methods exist for constructing structural models of their complexes. Here, we analyze the interfaces from a non-redundant set of complexes of α-helical IMPs whose structures have been determined to a high resolution. We find that the interface is not significantly different from the rest of the surface in terms of average hydrophobicity. However, the interface is significantly better conserved and, on average, inter-subunit contacting residue pairs correlate more strongly than non-contacting pairs, especially in obligate complexes. We also develop a neural network-based method, with an area under the receiver operating characteristic curve of 0.75 and a Pearson correlation coefficient of 0.70, for predicting interface residues and their weighted contact numbers (WCNs). We further show that predicted interface residues and their WCNs can be used as restraints to reconstruct the structure α-helical IMP dimers through docking for fourteen out of a benchmark set of sixteen complexes. The RMSD100 values of the best-docked ligand subunit to its native structure are <2.5 Å for these fourteen cases. The structural analysis conducted in this work provides molecular details about the interface between α-helical IMPs and the WCN restraints represent an efficient means to score α-helical IMP docking candidates.