Structural basis of macromolecular recognition

Generation of a high confidence set of domain-domain interface types to guide protein complex structure predictions by AlphaFold

MOTIVATION

While the release of AlphaFold (AF) represented a breakthrough for the prediction of protein complex structures, its sensitivity, especially when using full length protein sequences, still remains limited. Modeling success rates might increase if AF predictions were guided by likely interacting protein fragments. This approach requires available sets of highly confident protein-protein interface types. Computational resources, such as 3did, infer interacting globular domain types from observed contacts in protein structures. Assessing the accuracy of these predicted interface types is difficult because we lack hand-curated reference sets of verified domain-domain interface (DDI) types.

RESULTS

To improve protein complex modeling of DDIs by AF, we manually inspected 80 randomly selected DDI types from the 3did resource to generate a first reference set of DDI types. Identified cases of DDI type nonapproval (40%) primarily resulted from inaccurate Pfam domain matches, crystal contacts, and synthetic protein constructs. Using logistic regression, we predicted a subset of 2411 out of 5724 considered DDI types in 3did to be of high confidence, which we subsequently applied to 53 000 human-protein interactions to predict DDIs followed by AF modeling. We obtained highly confident AF models for 604 out of 1129 predicted DDIs. Of note, for 47% of them no confident AF structural model could be obtained using full length protein sequences.

AVAILABILITY AND IMPLEMENTATION

Code is available at https://github.com/KatjaLuckLab/DDI_manuscript.

Structural basis for the recognition of human hemoglobin by the heme-acquisition protein Shr from Streptococcus pyogenes

In Gram-positive bacteria, sophisticated machineries to acquire the heme group of hemoglobin (Hb) have evolved to extract the precious iron atom contained in it. In the human pathogen Streptococcus pyogenes, the Shr protein is a key component of this machinery. Herein we present the crystal structure of hemoglobin-interacting domain 2 (HID2) of Shr bound to Hb. HID2 interacts with both, the protein and heme portions of Hb, explaining the specificity of HID2 for the heme-bound form of Hb, but not its heme-depleted form. Further mutational analysis shows little tolerance of HID2 to interfacial mutations, suggesting that its interaction surface with Hb could be a suitable candidate to develop efficient inhibitors abrogating the binding of Shr to Hb.

Discovery and Characterization of a Bicyclic Peptide (Bicycle) Binder to Thymic Stromal Lymphopoietin

Thymic stromal lymphopoietin (TSLP) is an epithelial-derived pro-inflammatory cytokine involved in the development of asthma and other atopic diseases. We used Bicycle Therapeutics' proprietary phage display platform to identify bicyclic peptides (Bicycles) with high affinity for TSLP, a target that is difficult to drug with conventional small molecules due to the extended protein-protein interactions it forms with both receptors. The hit series was shown to bind to TSLP in a hotspot, that is also used by IL-7Rα. Guided by the first X-ray crystal structure of a small peptide binding to TSLP and the identification of key metabolites, we were able to improve the proteolytic stability of this series in lung S9 fractions without sacrificing binding affinity. This resulted in the potent Bicycle 46 with nanomolar affinity to TSLP (KD = 13 nM), low plasma clearance of 6.4 mL/min/kg, and an effective half-life of 46 min after intravenous dosing to rats.

Impact of AlphaFold on structure prediction of protein complexes: The CASP15-CAPRI experiment

We present the results for CAPRI Round 54, the 5th joint CASP-CAPRI protein assembly prediction challenge. The Round offered 37 targets, including 14 homodimers, 3 homo-trimers, 13 heterodimers including 3 antibody-antigen complexes, and 7 large assemblies. On average ~70 CASP and CAPRI predictor groups, including more than 20 automatics servers, submitted models for each target. A total of 21 941 models submitted by these groups and by 15 CAPRI scorer groups were evaluated using the CAPRI model quality measures and the DockQ score consolidating these measures. The prediction performance was quantified by a weighted score based on the number of models of acceptable quality or higher submitted by each group among their five best models. Results show substantial progress achieved across a significant fraction of the 60+ participating groups. High-quality models were produced for about 40% of the targets compared to 8% two years earlier. This remarkable improvement is due to the wide use of the AlphaFold2 and AlphaFold2-Multimer software and the confidence metrics they provide. Notably, expanded sampling of candidate solutions by manipulating these deep learning inference engines, enriching multiple sequence alignments, or integration of advanced modeling tools, enabled top performing groups to exceed the performance of a standard AlphaFold2-Multimer version used as a yard stick. This notwithstanding, performance remained poor for complexes with antibodies and nanobodies, where evolutionary relationships between the binding partners are lacking, and for complexes featuring conformational flexibility, clearly indicating that the prediction of protein complexes remains a challenging problem.

Discriminating physiological from non-physiological interfaces in structures of protein complexes: A community-wide study

Reliably scoring and ranking candidate models of protein complexes and assigning their oligomeric state from the structure of the crystal lattice represent outstanding challenges. A community-wide effort was launched to tackle these challenges. The latest resources on protein complexes and interfaces were exploited to derive a benchmark dataset consisting of 1677 homodimer protein crystal structures, including a balanced mix of physiological and non-physiological complexes. The non-physiological complexes in the benchmark were selected to bury a similar or larger interface area than their physiological counterparts, making it more difficult for scoring functions to differentiate between them. Next, 252 functions for scoring protein-protein interfaces previously developed by 13 groups were collected and evaluated for their ability to discriminate between physiological and non-physiological complexes. A simple consensus score generated using the best performing score of each of the 13 groups, and a cross-validated Random Forest (RF) classifier were created. Both approaches showed excellent performance, with an area under the Receiver Operating Characteristic (ROC) curve of 0.93 and 0.94, respectively, outperforming individual scores developed by different groups. Additionally, AlphaFold2 engines recalled the physiological dimers with significantly higher accuracy than the non-physiological set, lending support to the reliability of our benchmark dataset annotations. Optimizing the combined power of interface scoring functions and evaluating it on challenging benchmark datasets appears to be a promising strategy.

Critical Assessment of Methods for Predicting the 3D Structure of Proteins and Protein Complexes

Advances in a scientific discipline are often measured by small, incremental steps. In this review, we report on two intertwined disciplines in the protein structure prediction field, modeling of single chains and modeling of complexes, that have over decades emulated this pattern, as monitored by the community-wide blind prediction experiments CASP and CAPRI. However, over the past few years, dramatic advances were observed for the accurate prediction of single protein chains, driven by a surge of deep learning methodologies entering the prediction field. We review the mainscientific developments that enabled these recent breakthroughs and feature the important role of blind prediction experiments in building up and nurturing the structure prediction field. We discuss how the new wave of artificial intelligence-based methods is impacting the fields of computational and experimental structural biology and highlight areas in which deep learning methods are likely to lead to future developments, provided that major challenges are overcome.

Solubility-Aware Protein Binding Peptide Design Using AlphaFold

New protein-protein interactions (PPIs) are identified, but PPIs have different physicochemical properties compared with conventional targets, making it difficult to use small molecules. Peptides offer a new modality to target PPIs, but designing appropriate peptide sequences by computation is challenging. Recently, AlphaFold and RoseTTAFold have made it possible to predict protein structures from amino acid sequences with ultra-high accuracy, enabling de novo protein design. We designed peptides likely to have PPI as the target protein using the "binder hallucination" protocol of AfDesign, a de novo protein design method using AlphaFold. However, the solubility of the peptides tended to be low. Therefore, we designed a solubility loss function using solubility indices for amino acids and developed a solubility-aware AfDesign binder hallucination protocol. The peptide solubility in sequences designed using the new protocol increased with the weight of the solubility loss function; moreover, they captured the characteristics of the solubility indices. Moreover, the new protocol sequences tended to have higher affinity than random or single residue substitution sequences when evaluated by docking binding affinity. Our approach shows that it is possible to design peptide sequences that can bind to the interface of PPI while controlling solubility.

Leveraging structural and 2D-QSAR to investigate the role of functional group substitutions, conserved surface residues and desolvation in triggering the small molecule-induced dimerization of hPD-L1

Small molecules are rising as a new generation of immune checkpoints’ inhibitors, with compounds targeting the human Programmed death-ligand 1 (hPD-L1) protein are pioneering this area of research. Promising examples include the recently disclosed compounds from Bristol-Myers-Squibb (BMS). These molecules bind specifically to hPD-L1 through a unique mode of action. They induce dimerization between two hPD-L1 monomers through the hPD-1 binding interface in each monomer, thereby inhibiting the PD-1/PD-L1 axis. While the recently reported crystal structures of such small molecules bound to hPD-L1 reveal valuable insights regarding their molecular interactions, there is still limited information about the dynamics driving this unusual complex formation. The current study provides an in-depth computational structural analysis to study the interactions of five small molecule compounds in complex with hPD-L1. By employing a combination of molecular dynamic simulations, binding energy calculations and computational solvent mapping techniques, our analyses quantified the dynamic roles of different hydrophilic and lipophilic residues at the surface of hPD-L1 in mediating these interactions. Furthermore, ligand-based analyses, including Free-Wilson 2D-QSAR was conducted to quantify the impact of R-group substitutions at different sites of the phenoxy-methyl biphenyl core. Our results emphasize the importance of a terminal phenyl ring that must be present in any hPD-L1 small molecule inhibitor. This phenyl moiety overlaps with a very unfavorable hydration site, which can explain the ability of such small molecules to trigger hPD-L1 dimerization.

Prediction of protein assemblies, the next frontier: The CASP14-CAPRI experiment

We present the results for CAPRI Round 50, the fourth joint CASP-CAPRI protein assembly prediction challenge. The Round comprised a total of twelve targets, including six dimers, three trimers, and three higher-order oligomers. Four of these were easy targets, for which good structural templates were available either for the full assembly, or for the main interfaces (of the higher-order oligomers). Eight were difficult targets for which only distantly related templates were found for the individual subunits. Twenty-five CAPRI groups including eight automatic servers submitted ~1250 models per target. Twenty groups including six servers participated in the CAPRI scoring challenge submitted ~190 models per target. The accuracy of the predicted models was evaluated using the classical CAPRI criteria. The prediction performance was measured by a weighted scoring scheme that takes into account the number of models of acceptable quality or higher submitted by each group as part of their five top-ranking models. Compared to the previous CASP-CAPRI challenge, top performing groups submitted such models for a larger fraction (70-75%) of the targets in this Round, but fewer of these models were of high accuracy. Scorer groups achieved stronger performance with more groups submitting correct models for 70-80% of the targets or achieving high accuracy predictions. Servers performed less well in general, except for the MDOCKPP and LZERD servers, who performed on par with human groups. In addition to these results, major advances in methodology are discussed, providing an informative overview of where the prediction of protein assemblies currently stands.

Current Challenges and Opportunities in Designing Protein-Protein Interaction Targeted Drugs

It has been noticed that the efficiency of drug development has been decreasing in the past few decades. To overcome the situation, protein-protein interactions (PPIs) have been identified as new drug targets as early as 2000. PPIs are more abundant in human cells than single proteins and play numerous important roles in cellular processes including diseases. However, PPIs have very different physicochemical features from the conventional drug targets, which make targeting PPIs challenging. Therefore, as of now, only a small number of PPI inhibitors have been approved or progressed to a stage of clinical trial. In this article, we first overview previous works that analyzed differences between PPIs with PPI targeting ligands and conventional drugs with their binding pockets. Then, we constructed an up-to-date list of PPI targeting drugs that have been approved or are currently under clinical trial and have bound drug-target structures available. Using the dataset, we analyzed the PPIs and their ligands using several scores of druggability. Druggability scores showed that PPI sites and their drugs targeting PPIs are less druggable than conventional binding pockets and drugs, which also indicates that PPI drugs do not follow the conventional rules for drug design, such as Lipinski's rule of five. Our analyses suggest that developing a new rule would be beneficial for guiding PPI-drug discovery.

A community proposal to integrate structural bioinformatics activities in ELIXIR (3D-Bioinfo Community)

Structural bioinformatics provides the scientific methods and tools to analyse, archive, validate, and present the biomolecular structure data generated by the structural biology community. It also provides an important link with the genomics community, as structural bioinformaticians also use the extensive sequence data to predict protein structures and their functional sites. A very broad and active community of structural bioinformaticians exists across Europe, and 3D-Bioinfo will establish formal platforms to address their needs and better integrate their activities and initiatives. Our mission will be to strengthen the ties with the structural biology research communities in Europe covering life sciences, as well as chemistry and physics and to bridge the gap between these researchers in order to fully realize the potential of structural bioinformatics. Our Community will also undertake dedicated educational, training and outreach efforts to facilitate this, bringing new insights and thus facilitating the development of much needed innovative applications e.g. for human health, drug and protein design. Our combined efforts will be of critical importance to keep the European research efforts competitive in this respect.

Here we highlight the major European contributions to the field of structural bioinformatics, the most pressing challenges remaining and how Europe-wide interactions, enabled by ELIXIR and its platforms, will help in addressing these challenges and in coordinating structural bioinformatics resources across Europe. In particular, we present recent activities and future plans to consolidate an ELIXIR 3D-Bioinfo Community in structural bioinformatics and propose means to develop better links across the community. These include building new consortia, organising workshops to establish data standards and seeking community agreement on benchmark data sets and strategies. We also highlight existing and planned collaborations with other ELIXIR Communities and other European infrastructures, such as the structural biology community supported by Instruct-ERIC, with whom we have synergies and overlapping common interests.

Multiple protein-DNA interfaces unravelled by evolutionary information, physico-chemical and geometrical properties

Interactions between proteins and nucleic acids are at the heart of many essential biological processes. Despite increasing structural information about how these interactions may take place, our understanding of the usage made of protein surfaces by nucleic acids is still very limited. This is in part due to the inherent complexity associated to protein surface deformability and evolution. In this work, we present a method that contributes to decipher such complexity by predicting protein-DNA interfaces and characterizing their properties. It relies on three biologically and physically meaningful descriptors, namely evolutionary conservation, physico-chemical properties and surface geometry. We carefully assessed its performance on several hundreds of protein structures and compared it to several machine-learning state-of-the-art methods. Our approach achieves a higher sensitivity compared to the other methods, with a similar precision. Importantly, we show that it is able to unravel ‘hidden’ binding sites by applying it to unbound protein structures and to proteins binding to DNA via multiple sites and in different conformations. It is also applicable to the detection of RNA-binding sites, without significant loss of performance. This confirms that DNA and RNA-binding sites share similar properties. Our method is implemented as a fully automated tool, JETDNA2, freely accessible at: http://www.lcqb.upmc.fr/JET2DNA. We also provide a new dataset of 187 protein-DNA complex structures, along with a subset of 82 associated unbound structures. The set represents the largest body of high-resolution crystallographic structures of protein-DNA complexes, use biological protein assemblies as DNA-binding units, and covers all major types of protein-DNA interactions. It is available at: http://www.lcqb.upmc.fr/PDNAbenchmarks.

Protein-DNA interactions are essential to living organisms and their impairment is associated to many diseases. For these reasons, they have become increasingly important therapeutic targets. Experimental structure determination has revealed different binding motifs and modes, associated to different functions. Yet, the available structural data gives us only a glimpse of the multiplicity and complexity of protein surface usage by DNA. In this work, we use a three-layer model to describe and predict DNA-binding sites at protein surfaces. Given a protein, we consider the way its residues are conserved through evolution, their physico-chemical properties and geometrical shapes to decrypt its surface. We are able to detect a large portion of interacting residues with good precision, even when they are ‘hidden’ by conformational changes. We highlight cases where one protein binds DNA via distinct regions to perform different functions. We are able to uncover the alternative binding sites and relate their properties with their specific roles. Our work can help guiding mutagenesis experiments and the development of new drugs specifically targeting one site while limiting possible side effects.

The nuclear Overhauser Effect (NOE) as a tool to study macromolecular confinement: Elucidation and disentangling of crowding and encapsulation effects

We propose a methodology to capture short-lived but biophysically important contacts of biomacromolecules using the biomolecule-water nuclear Overhauser effect as an indirect microscope. Thus, instead of probing the direct correlation with the foreign biomolecule, we detect its presence by the disturbance it causes in the surrounding water. In addition, this information obtained is spatially resolved and can thus be attributed to specific sites. We extend this approach to the influence of more than one change in chemical environment and show a methodological way of resolution. This is achieved by taking double differences of corresponding σ_NOE/σ_ROE ratios of the systems studied and separating specific, unspecific, and intermediate influence. While applied to crowding and encapsulation in this study, this method is generally suitable for any combination of changes in chemical environment.

Blind prediction of homo- and hetero-protein complexes: The CASP13-CAPRI experiment

We present the results for CAPRI Round 46, the third joint CASP-CAPRI protein assembly prediction challenge. The Round comprised a total of 20 targets including 14 homo-oligomers and 6 heterocomplexes. Eight of the homo-oligomer targets and one heterodimer comprised proteins that could be readily modeled using templates from the Protein Data Bank, often available for the full assembly. The remaining 11 targets comprised 5 homodimers, 3 heterodimers, and two higher-order assemblies. These were more difficult to model, as their prediction mainly involved "ab-initio" docking of subunit models derived from distantly related templates. A total of ~30 CAPRI groups, including 9 automatic servers, submitted on average ~2000 models per target. About 17 groups participated in the CAPRI scoring rounds, offered for most targets, submitting ~170 models per target. The prediction performance, measured by the fraction of models of acceptable quality or higher submitted across all predictors groups, was very good to excellent for the nine easy targets. Poorer performance was achieved by predictors for the 11 difficult targets, with medium and high quality models submitted for only 3 of these targets. A similar performance "gap" was displayed by scorer groups, highlighting yet again the unmet challenge of modeling the conformational changes of the protein components that occur upon binding or that must be accounted for in template-based modeling. Our analysis also indicates that residues in binding interfaces were less well predicted in this set of targets than in previous Rounds, providing useful insights for directions of future improvements.

Polyhedra, spindles, phage nucleus and pyramids: Structural biology of viral superstructures

Viral infection causes comprehensive rearrangements of the cell that reflect as much host defense mechanisms as virus-induced structures assembled to facilitate infection. Regardless of their pro- or antiviral role, large intracellular structures are readily detectable by microscopy and often provide a signature characteristic of a specific viral infection. The structural features and localization of these assemblies have thus been commonly used for the diagnostic and classification of viruses since the early days of virology. More recently, characterization of viral superstructures using molecular and structural approaches have revealed very diverse organizations and roles, ranging from dynamic viral factories behaving like liquid organelles to ultra-stable crystals embedding and protecting virions. This chapter reviews the structures, functions and biotechnological applications of virus-induced superstructures with a focus on assemblies that have a regular organization, for which detailed structural descriptions are available. Examples span viruses infecting all domains of life including the assembly of virions into crystalline arrays in eukaryotic and bacterial viruses, nucleus-like compartments involved in the replication of large bacteriophages, and pyramid-like structures mediating the egress of archaeal viruses. Among these superstructures, high-resolution structures are available for crystalline objects produced by insect viruses: viral polyhedra which function as the infectious form of occluded viruses, and spindles which are potent virulence factors of entomopoxviruses. In turn, some of these highly symmetrical objects have been used to develop and validate advanced structural approaches, pushing the boundary of structural biology.

Configurational Entropy Components and Their Contribution to Biomolecular Complex Formation

Configurational entropy change is a central constituent of the free energy change in noncovalent interactions between biomolecules. Due to both experimental and computational limitations, however, the impact of individual contributions to configurational entropy change remains underexplored. Here, we develop a novel, fully analytical framework to dissect the configurational entropy change of binding into contributions coming from molecular internal and external degrees of freedom. Importantly, this framework accounts for all coupled and uncoupled contributions in the absence of an external field. We employ our parallel implementation of the maximum information spanning tree algorithm to provide a comprehensive numerical analysis of the importance of the individual contributions to configurational entropy change on an extensive set of molecular dynamics simulations of protein binding processes. Contrary to commonly accepted assumptions, we show that different coupling terms contribute significantly to the overall configurational entropy change. Finally, while the magnitude of individual terms may be largely unpredictable a priori, the total configurational entropy change can be well approximated by rescaling the sum of uncoupled contributions from internal degrees of freedom only, providing support for NMR-based approaches for configurational entropy change estimation.

Infinite Assembly of Folded Proteins in Evolution, Disease, and Engineering

Mutations and changes in a protein's environment are well known for their potential to induce misfolding and aggregation, including amyloid formation. Alternatively, such perturbations can trigger new interactions that lead to the polymerization of folded proteins. In contrast to aggregation, this process does not require misfolding and, to highlight this difference, we refer to it as agglomeration. This term encompasses the amorphous assembly of folded proteins as well as the polymerization in one, two, or three dimensions. We stress the remarkable potential of symmetric homo-oligomers to agglomerate even by single surface point mutations, and we review the double-edged nature of this potential: how aberrant assemblies resulting from agglomeration can lead to disease, but also how agglomeration can serve in cellular adaptation and be exploited for the rational design of novel biomaterials.

Towards capturing cellular complexity: combining encapsulation and macromolecular crowding in a reverse micelle

This paper studies the orientational structure and dynamics of multi-protein systems under confinement and discusses the implications on biological cells.

Self-Consistent Framework Connecting Experimental Proxies of Protein Dynamics with Configurational Entropy

The recently developed NMR techniques enable estimation of protein configurational entropy change from the change in the average methyl order parameters. This experimental observable, however, does not directly measure the contribution of intramolecular couplings, protein main-chain motions, or angular dynamics. Here, we carry out a self-consistent computational analysis of the impact of these missing contributions on an extensive set of molecular dynamics simulations of different proteins undergoing binding. Specifically, we compare the configurational entropy change in protein complex formation as obtained by the maximum information spanning tree approximation (MIST), which treats the above entropy contributions directly, and the change in the average NMR methyl and NH order parameters. Our parallel implementation of MIST allows us to treat hard angular degrees of freedom as well as couplings up to full pairwise order explicitly, while still involving a high degree of sampling and tackling molecules of biologically relevant sizes. First, we demonstrate a remarkably strong linear relationship between the total configurational entropy change and the average change in both methyl and backbone-NH order parameters. Second, in contrast to canonical assumptions, we show that the main-chain and angular terms contribute significantly to the overall configurational entropy change and also scale linearly with it. Consequently, linear models starting from the average methyl order parameters are able to capture the contribution of main-chain and angular terms well. After applying the quantum-mechanical harmonic oscillator entropy formalism, we establish a similarly strong linear relationship for X-ray crystallographic B-factors. Finally, we demonstrate that the observed linear relationships remain robust against drastic undersampling and argue that they reflect an intrinsic property of compact proteins. Despite their remarkable strength, however, the above linear relationships yield estimates of configurational entropy change whose accuracy appears to be sufficient for qualitative applications only.

The challenge of modeling protein assemblies: the CASP12-CAPRI experiment

We present the quality assessment of 5613 models submitted by predictor groups from both CAPRI and CASP for the total of 15 most tractable targets from the second joint CASP-CAPRI protein assembly prediction experiment. These targets comprised 12 homo-oligomers and 3 hetero-complexes. The bulk of the analysis focuses on 10 targets (of CAPRI Round 37), which included all 3 hetero-complexes, and whose protein chains or the full assembly could be readily modeled from structural templates in the PDB. On average, 28 CAPRI groups and 10 CASP groups (including automatic servers), submitted models for each of these 10 targets. Additionally, about 16 groups participated in the CAPRI scoring experiments. A range of acceptable to high quality models were obtained for 6 of the 10 Round 37 targets, for which templates were available for the full assembly. Poorer results were achieved for the remaining targets due to the lower quality of the templates available for the full complex or the individual protein chains, highlighting the unmet challenge of modeling the structural adjustments of the protein components that occur upon binding or which must be accounted for in template-based modeling. On the other hand, our analysis indicated that residues in binding interfaces were correctly predicted in a sizable fraction of otherwise poorly modeled assemblies and this with higher accuracy than published methods that do not use information on the binding partner. Lastly, the strengths and weaknesses of the assessment methods are evaluated and improvements suggested.

Structural basis for clonal diversity of the human T-cell response to a dominant influenza virus epitope

Influenza A virus (IAV) causes an acute infection in humans that is normally eliminated by CD8⁺ cytotoxic T lymphocytes. Individuals expressing the MHC class I molecule HLA-A2 produce cytotoxic T lymphocytes bearing T-cell receptors (TCRs) that recognize the immunodominant IAV epitope GILGFVFTL (GIL). Most GIL-specific TCRs utilize α/β chain pairs encoded by the TRAV27/TRBV19 gene combination to recognize this relatively featureless peptide epitope (canonical TCRs). However, ∼40% of GIL-specific TCRs express a wide variety of other TRAV/TRBV combinations (non-canonical TCRs). To investigate the structural underpinnings of this remarkable diversity, we determined the crystal structure of a non-canonical GIL-specific TCR (F50) expressing the TRAV13-1/TRBV27 gene combination bound to GIL-HLA-A2 to 1.7 Å resolution. Comparison of the F50-GIL-HLA-A2 complex with the previously published complex formed by a canonical TCR (JM22) revealed that F50 and JM22 engage GIL-HLA-A2 in markedly different orientations. These orientations are distinguished by crossing angles of TCR to peptide-MHC of 29° for F50 versus 69° for JM22 and by a focus by F50 on the C terminus rather than the center of the MHC α1 helix for JM22. In addition, F50, unlike JM22, uses a tryptophan instead of an arginine to fill a critical notch between GIL and the HLA-A2 α2 helix. The F50-GIL-HLA-A2 complex shows that there are multiple structurally distinct solutions to recognizing an identical peptide-MHC ligand with sufficient affinity to elicit a broad anti-IAV response that protects against viral escape and T-cell clonal loss.

Proteins evolve on the edge of supramolecular self-assembly

The self-association of proteins into symmetric complexes is ubiquitous in all kingdoms of life. Symmetric complexes possess unique geometric and functional properties, but their internal symmetry can pose a risk. In sickle-cell disease, the symmetry of haemoglobin exacerbates the effect of a mutation, triggering assembly into harmful fibrils. Here we examine the universality of this mechanism and its relation to protein structure geometry. We introduced point mutations solely designed to increase surface hydrophobicity among 12 distinct symmetric complexes from Escherichia coli. Notably, all responded by forming supramolecular assemblies in vitro, as well as in vivo upon heterologous expression in Saccharomyces cerevisiae. Remarkably, in four cases, micrometre-long fibrils formed in vivo in response to a single point mutation. Biophysical measurements and electron microscopy revealed that mutants self-assembled in their folded states and so were not amyloid-like. Structural examination of 73 mutants identified supramolecular assembly hot spots predictable by geometry. A subsequent structural analysis of 7,471 symmetric complexes showed that geometric hot spots were buffered chemically by hydrophilic residues, suggesting a mechanism preventing mis-assembly of these regions. Thus, point mutations can frequently trigger folded proteins to self-assemble into higher-order structures. This potential is counterbalanced by negative selection and can be exploited to design nanomaterials in living cells.

Entropy in molecular recognition by proteins

Molecular recognition by proteins is fundamental to molecular biology. Dissection of the thermodynamic energy terms governing protein-ligand interactions has proven difficult, with determination of entropic contributions being particularly elusive. NMR relaxation measurements have suggested that changes in protein conformational entropy can be quantitatively obtained through a dynamical proxy, but the generality of this relationship has not been shown. Twenty-eight protein-ligand complexes are used to show a quantitative relationship between measures of fast side-chain motion and the underlying conformational entropy. We find that the contribution of conformational entropy can range from favorable to unfavorable, which demonstrates the potential of this thermodynamic variable to modulate protein-ligand interactions. For about one-quarter of these complexes, the absence of conformational entropy would render the resulting affinity biologically meaningless. The dynamical proxy for conformational entropy or "entropy meter" also allows for refinement of the contributions of solvent entropy and the loss in rotational-translational entropy accompanying formation of high-affinity complexes. Furthermore, structure-based application of the approach can also provide insight into long-lived specific water-protein interactions that escape the generic treatments of solvent entropy based simply on changes in accessible surface area. These results provide a comprehensive and unified view of the general role of entropy in high-affinity molecular recognition by proteins.

Modeling protein-protein and protein-peptide complexes: CAPRI 6th edition

We present the sixth report evaluating the performance of methods for predicting the atomic resolution structures of protein complexes offered as targets to the community-wide initiative on the Critical Assessment of Predicted Interactions (CAPRI). The evaluation is based on a total of 20,670 predicted models for 8 protein-peptide complexes, a novel category of targets in CAPRI, and 12 protein-protein targets in CAPRI prediction Rounds held during the years 2013-2016. For two of the protein-protein targets, the focus was on the prediction of side-chain conformation and positions of interfacial water molecules. Seven of the protein-protein targets were particularly challenging owing to their multicomponent nature, to conformational changes at the binding site, or to a combination of both. Encouragingly, the very large multiprotein complex with the nucleosome was correctly predicted, and correct models were submitted for the protein-peptide targets, but not for some of the challenging protein-protein targets. Models of acceptable quality or better were obtained for 14 of the 20 targets, including medium quality models for 13 targets and high quality models for 8 targets, indicating tangible progress of present-day computational methods in modeling protein complexes with increased accuracy. Our evaluation suggests that the progress stems from better integration of different modeling tools with docking procedures, as well as the use of more sophisticated evolutionary information to score models. Nonetheless, adequate modeling of conformational flexibility in interacting proteins remains an important area with a crucial need for improvement. Proteins 2017; 85:359-377. © 2016 Wiley Periodicals, Inc.

Linking in domain-swapped protein dimers

The presence of knots has been observed in a small fraction of single-domain proteins and related to their thermodynamic and kinetic properties. The exchanging of identical structural elements, typical of domain-swapped proteins, makes such dimers suitable candidates to validate the possibility that mutual entanglement between chains may play a similar role for protein complexes. We suggest that such entanglement is captured by the linking number. This represents, for two closed curves, the number of times that each curve winds around the other. We show that closing the curves is not necessary, as a novel parameter G', termed Gaussian entanglement, is strongly correlated with the linking number. Based on 110 non redundant domain-swapped dimers, our analysis evidences a high fraction of chains with a significant intertwining, that is with |G'| > 1. We report that Nature promotes configurations with negative mutual entanglement and surprisingly, it seems to suppress intertwining in long protein dimers. Supported by numerical simulations of dimer dissociation, our results provide a novel topology-based classification of protein-swapped dimers together with some preliminary evidence of its impact on their physical and biological properties.

Prediction of homoprotein and heteroprotein complexes by protein docking and template-based modeling: A CASP-CAPRI experiment

We present the results for CAPRI Round 30, the first joint CASP-CAPRI experiment, which brought together experts from the protein structure prediction and protein-protein docking communities. The Round comprised 25 targets from amongst those submitted for the CASP11 prediction experiment of 2014. The targets included mostly homodimers, a few homotetramers, and two heterodimers, and comprised protein chains that could readily be modeled using templates from the Protein Data Bank. On average 24 CAPRI groups and 7 CASP groups submitted docking predictions for each target, and 12 CAPRI groups per target participated in the CAPRI scoring experiment. In total more than 9500 models were assessed against the 3D structures of the corresponding target complexes. Results show that the prediction of homodimer assemblies by homology modeling techniques and docking calculations is quite successful for targets featuring large enough subunit interfaces to represent stable associations. Targets with ambiguous or inaccurate oligomeric state assignments, often featuring crystal contact-sized interfaces, represented a confounding factor. For those, a much poorer prediction performance was achieved, while nonetheless often providing helpful clues on the correct oligomeric state of the protein. The prediction performance was very poor for genuine tetrameric targets, where the inaccuracy of the homology-built subunit models and the smaller pair-wise interfaces severely limited the ability to derive the correct assembly mode. Our analysis also shows that docking procedures tend to perform better than standard homology modeling techniques and that highly accurate models of the protein components are not always required to identify their association modes with acceptable accuracy. Proteins 2016; 84(Suppl 1):323-348. © 2016 Wiley Periodicals, Inc.

Specific Electrostatic Molecular Recognition in Water

The identification of pairs of small peptides that recognize each other in water exclusively through electrostatic interactions is reported. The target peptide and a structure-biased combinatorial ligand library consisting of ≈78 125 compounds were synthesized on different sized beads. Peptide-peptide interactions could conveniently be observed by clustering of the small, fluorescently labeled target beads on the surface of larger ligand-containing beads. Sequences of isolated hits were determined by MS/MS. The interactions of the complex showing the highest affinity were investigated by a novel single-bead binding assay and by 2D NMR spectroscopy. Molecular dynamics (MD) studies revealed a putative mode of interaction for this unusual electrostatic binding event. High binding specificity occurred through a combination of topological matching and electrostatic and hydrogen-bond complementarities. From MD simulations binding also seemed to involve three tightly bound water molecules in the interface between the binding partners. Binding constants in the submicromolar range, useful for biomolecular adhesion and in nanostructure design, were measured.

Molecular basis of in vitro affinity maturation and functional evolution of a neutralizing anti-human GM-CSF antibody

X-ray structure analysis of 4 antibody Fab fragments, each in complex with human granulocyte macrophage colony stimulating factor (GM-CSF), was performed to investigate the changes at the protein-protein binding interface during the course of in vitro affinity maturation by phage display selection. The parental antibody MOR03929 was compared to its derivatives MOR04252 (CDR-H2 optimized), MOR04302 (CDR-L3 optimized) and MOR04357 (CDR-H2 and CDR-L3 optimized). All antibodies bind to a conformational epitope that can be divided into 3 sub-epitopes. Specifically, MOR04357 binds to a region close to the GM-CSF N-terminus (residues 11-24), a short second sub-epitope (residues 83-89) and a third at the C-terminus (residues 112-123). Modifications introduced during affinity maturation in CDR-H2 and CDR-L3 led to the establishment of additional hydrogen bonds and van der Waals contacts, respectively, providing a rationale for the observed improvement in binding affinity and neutralization potency. Once GM-CSF is complexed to the antibodies, modeling predicts a sterical clash with GM-CSF binding to GM-CSF receptor α and β chain. This predicted mutually exclusive binding was confirmed by a GM-CSF receptor α chain ligand binding inhibition assay. Finally, high throughput sequencing of clones obtained after affinity maturation phage display pannings revealed highly selected consensus sequences for CDR-H2 as well for CDR-L3, which are in accordance with the sequence of the highest affinity antibody MOR04357. The resolved crystal structures highlight the criticality of these strongly selected residues for high affinity interaction with GM-CSF.

Conformational frustration in calmodulin-target recognition

Calmodulin (CaM) is a primary calcium (Ca(2+) )-signaling protein that specifically recognizes and activates highly diverse target proteins. We explored the molecular basis of target recognition of CaM with peptides representing the CaM-binding domains from two Ca(2+) -CaM-dependent kinases, CaMKI and CaMKII, by employing experimentally constrained molecular simulations. Detailed binding route analysis revealed that the two CaM target peptides, although similar in length and net charge, follow distinct routes that lead to a higher binding frustration in the CaM-CaMKII complex than in the CaM-CaMKI complex. We discovered that the molecular origin of the binding frustration is caused by intermolecular contacts formed with the C-domain of CaM that need to be broken before the formation of intermolecular contacts with the N-domain of CaM. We argue that the binding frustration is important for determining the kinetics of the recognition process of proteins involving large structural fluctuations.

Structural basis of protein-protein interactions

Regulated interactions between proteins govern signaling pathways within and between cells. Structural studies on protein complexes formed reversibly and/or transiently illustrate the remarkable diversity of interactions, both in terms of interfacial size and nature. In recent years, "domain-peptide" interactions have gained much greater recognition and may be viewed as both pre-translational and posttranslational-dependent functional switches. Our understanding of the multistep regulation of auto-inhibited multidomain proteins has also grown. Their activity may be understood as the "combinatorial" output of multiple input signals, including phosphorylation, location, and mechanical force. The prospects for bridging the gap between the new "systems biology" data and the traditional "reductionist" data are also discussed.

Dynamic and thermodynamic response of the Ras protein Cdc42Hs upon association with the effector domain of PAK3

Human cell division cycle protein 42 (Cdc42Hs) is a small, Rho-type guanosine triphosphatase involved in multiple cellular processes through its interactions with downstream effectors. The binding domain of one such effector, the actin cytoskeleton-regulating p21-activated kinase 3, is known as PBD46. Nitrogen-15 backbone and carbon-13 methyl NMR relaxation was measured to investigate the dynamical changes in activated GMPPCP·Cdc42Hs upon PBD46 binding. Changes in internal motion of the Cdc42Hs, as revealed by methyl axis order parameters, were observed not only near the Cdc42Hs-PBD46 interface but also in remote sites on the Cdc42Hs molecule. The binding-induced changes in side-chain dynamics propagate along the long axis of Cdc42Hs away from the site of PBD46 binding with sharp distance dependence. Overall, the binding of the PBD46 effector domain on the dynamics of methyl-bearing side chains of Cdc42Hs results in a modest rigidification, which is estimated to correspond to an unfavorable change in conformational entropy of approximately -10kcalmol(-1) at 298K. A cluster of methyl probes closest to the nucleotide-binding pocket of Cdc42Hs becomes more rigid upon binding of PBD46 and is proposed to slow the catalytic hydrolysis of the γ phosphate moiety. An additional cluster of methyl probes surrounding the guanine ring becomes more flexible on binding of PBD46, presumably facilitating nucleotide exchange mediated by a guanosine exchange factor. In addition, the Rho insert helix, which is located at a site remote from the PBD46 binding interface, shows a significant dynamic response to PBD46 binding.

Blind prediction of interfacial water positions in CAPRI

We report the first assessment of blind predictions of water positions at protein-protein interfaces, performed as part of the critical assessment of predicted interactions (CAPRI) community-wide experiment. Groups submitting docking predictions for the complex of the DNase domain of colicin E2 and Im2 immunity protein (CAPRI Target 47), were invited to predict the positions of interfacial water molecules using the method of their choice. The predictions-20 groups submitted a total of 195 models-were assessed by measuring the recall fraction of water-mediated protein contacts. Of the 176 high- or medium-quality docking models-a very good docking performance per se-only 44% had a recall fraction above 0.3, and a mere 6% above 0.5. The actual water positions were in general predicted to an accuracy level no better than 1.5 Å, and even in good models about half of the contacts represented false positives. This notwithstanding, three hotspot interface water positions were quite well predicted, and so was one of the water positions that is believed to stabilize the loop that confers specificity in these complexes. Overall the best interface water predictions was achieved by groups that also produced high-quality docking models, indicating that accurate modelling of the protein portion is a determinant factor. The use of established molecular mechanics force fields, coupled to sampling and optimization procedures also seemed to confer an advantage. Insights gained from this analysis should help improve the prediction of protein-water interactions and their role in stabilizing protein complexes.

Conserved electrostatic fields at the Ras–effector interface measured through vibrational Stark effect spectroscopy explain the difference in tilt angle in the Ras binding domains of Raf and RalGDS

Vibrational Stark effect (VSE) spectroscopy was used to measure the electrostatic fields present at the interface of the human guanosine triphosphatase (GTPase) Ras docked with the Ras binding domain (RBD) of the protein kinase Raf.

Structural templates for modeling homodimers

Oligomeric proteins are more abundant in nature than monomeric proteins, and involved in all biological processes. In the absence of an experimental structure, their subunits can be modeled from their sequence like monomeric proteins, but reliable procedures to build the oligomeric assembly are scarce. Template-based methods, which start from known protein structures, are commonly applied to model subunits. We present a method to model homodimers that relies on a structural alignment of the subunits, and test it on a set of 511 target structures recently released by the Protein Data Bank, taking as templates the earlier released structures of 3108 homodimeric proteins (H-set), and 2691 monomeric proteins that form dimer-like assemblies in crystals (M-set). The structural alignment identifies a H-set template for 97% of the targets, and in half of the cases, it yields a correct model of the dimer geometry and residue-residue contacts in the target. It also identifies a M-set template for most of the targets, and some of the crystal dimers are very similar to the target homodimers. The procedure efficiently detects homology at low levels of sequence identities, and points to erroneous quaternary structures in the Protein Data Bank. The high coverage of the target set suggests that the content of the Protein Data Bank already approaches the structural diversity of protein assemblies in nature, and that template-based methods should become the choice method for modeling oligomeric as well as monomeric proteins.

Structure of NKp65 bound to its keratinocyte ligand reveals basis for genetically linked recognition in natural killer gene complex

The natural killer (NK) gene complex (NKC) encodes numerous C-type lectin-like receptors that govern the activity of NK cells. Although some of these receptors (Ly49s, NKG2D, CD94/NKG2A) recognize MHC or MHC-like molecules, others (Nkrp1, NKRP1A, NKp80, NKp65) instead bind C-type lectin-like ligands to which they are genetically linked in the NKC. To understand the basis for this recognition, we determined the structure of human NKp65, an activating receptor implicated in the immunosurveillance of skin, bound to its NKC-encoded ligand keratinocyte-associated C-type lectin (KACL). Whereas KACL forms a homodimer resembling other C-type lectin-like dimers, NKp65 is monomeric. The binding mode in the NKp65-KACL complex, in which a monomeric receptor engages a dimeric ligand, is completely distinct from those used by Ly49s, NKG2D, or CD94/NKG2A. The structure explains the exceptionally high affinity of the NKp65-KACL interaction compared with other cell-cell interaction pairs (KD = 6.7 × 10(-10) M), which may compensate for the monomeric nature of NKp65 to achieve cell activation. This previously unreported structure of an NKC-encoded receptor-ligand complex, coupled with mutational analysis of the interface, establishes a docking template that is directly applicable to other genetically linked pairs in the NKC, including Nkrp1-Clr, NKRP1A-LLT1, and NKp80-AICL.

The dynamical response of hen egg white lysozyme to the binding of a carbohydrate ligand

It has become clear that the binding of small and large ligands to proteins can invoke significant changes in side chain and main chain motion in the fast picosecond to nanosecond timescale. Recently, the use of a "dynamical proxy" has indicated that changes in these motions often reflect significant changes in conformational entropy. These entropic contributions are sometimes of the same order as the total entropy of binding. Thus, it is important to understand the connections amongst motion between the manifold of states accessible to the native state of proteins, the corresponding entropy, and how this impacts the overall energetics of protein function. The interaction of proteins with carbohydrate ligands is central to a range of biological functions. Here, we examine a classic carbohydrate interaction with an enzyme: the binding of wild-type hen egg white lysozyme (HEWL) to the natural, competitive inhibitor chitotriose. Using NMR relaxation experiments, backbone amide and side chain methyl axial order parameters were obtained across apo and chitotriose-bound HEWL. Upon binding, changes in the apparent amplitude of picosecond to nanosecond main chain and side chain motions are seen across the protein. Indeed, binding of chitotriose renders a large contiguous fraction of HEWL effectively completely rigid. Changes in methyl flexibility are most pronounced closest to the binding site, but average to only a small overall change in the dynamics across the protein. The corresponding change in conformational entropy is unfavorable and estimated to be a significant fraction of the total binding entropy.

3DSwap: curated knowledgebase of proteins involved in 3D domain swapping

Three-dimensional domain swapping is a unique protein structural phenomenon where two or more protein chains in a protein oligomer share a common structural segment between individual chains. This phenomenon is observed in an array of protein structures in oligomeric conformation. Protein structures in swapped conformations perform diverse functional roles and are also associated with deposition diseases in humans. We have performed in-depth literature curation and structural bioinformatics analyses to develop an integrated knowledgebase of proteins involved in 3D domain swapping. The hallmark of 3D domain swapping is the presence of distinct structural segments such as the hinge and swapped regions. We have curated the literature to delineate the boundaries of these regions. In addition, we have defined several new concepts like ‘secondary major interface’ to represent the interface properties arising as a result of 3D domain swapping, and a new quantitative measure for the ‘extent of swapping’ in structures. The catalog of proteins reported in 3DSwap knowledgebase has been generated using an integrated structural bioinformatics workflow of database searches, literature curation, by structure visualization and sequence–structure–function analyses. The current version of the 3DSwap knowledgebase reports 293 protein structures, the analysis of such a compendium of protein structures will further the understanding molecular factors driving 3D domain swapping.

Database URL:http://caps.ncbs.res.in/3dswap

Secreted dengue virus nonstructural protein NS1 is an atypical barrel-shaped high-density lipoprotein

Dengue virus (DENV) causes the major arboviral disease of the tropics, characterized in its severe forms by signs of hemorrhage and plasma leakage. DENV encodes a nonstructural glycoprotein, NS1, that associates with intracellular membranes and the cell surface. NS1 is eventually secreted as a soluble hexamer from DENV-infected cells and circulates in the bloodstream of infected patients. Extracellular NS1 has been shown to modulate the complement system and to enhance DENV infection, yet its structure and function remain essentially unknown. By combining cryoelectron microscopy analysis with a characterization of NS1 amphipathic properties, we show that the secreted NS1 hexamer forms a lipoprotein particle with an open-barrel protein shell and a prominent central channel rich in lipids. Biochemical and NMR analyses of the NS1 lipid cargo reveal the presence of triglycerides, bound at an equimolar ratio to the NS1 protomer, as well as cholesteryl esters and phospholipids, a composition evocative of the plasma lipoproteins involved in vascular homeostasis. This study suggests that DENV NS1, by mimicking or hijacking lipid metabolic pathways, contributes to endothelium dysfunction, a key feature of severe dengue disease.