Systematic analysis of the effect of multiple templates on the accuracy of comparative models of protein structure

Identification of novel inhibitors targeting TIRAP interactions with BTK and PKCδ in inflammation through an in silico approach

Advanced computational tools focusing on protein-protein interaction (PPI) based drug development is a powerful platform to accelerate the therapeutic development of small lead molecules and repurposed drugs. Toll/interleukin-1 receptor (TIR) domain-containing adapter protein (TIRAP) and its interactions with other proteins in macrophages signalling are crucial components of severe or persistent inflammation. TIRAP activation through Bruton's tyrosine kinase (BTK) and Protein Kinase C delta (PKCδ) is essential for downstream inflammatory signalling. We created homology-based structural models of BTK and PKCδ in MODELLER 9.24. TIRAP interactions with BTK and PKCδ in its non-phosphorylated and phosphorylated states were determined by multiple docking tools including HADDOCK 2.4, pyDockWEB and ClusPro 2.0. Food and Drug Administration (FDA)-approved drugs were virtually screened through Discovery Studio LibDock and Autodock Vina tools to target the common TIR domain residues of TIRAP, which interact with both BTK and PKC at the identified interfacial sites of the complexes. Four FDA-approved drugs were identified and found to have stable interactions over a range of 100 ns MD simulation timescales. These drugs block the interactions of both kinases with TIRAP in silico. Hence, these drugs have the potential to dampen downstream inflammatory signalling and inflammation-mediated disease.

Correlated conformational dynamics of the human GluN1-GluN2A type N-methyl-D-aspartate (NMDA) receptor

N-Methyl-D-aspartate receptors (NMDARs) are glutamate-gated ion channels found in the nerve cell membranes. As a result of overexcitation of NMDARs, neuronal death occurs and may lead to diseases such as epilepsy, stroke, Alzheimer's disease, and Parkinson's disease. In this study, human GluN1- GluN2A type NMDAR structure is modeled based on the X-ray structure of the Xenopus laevis template and missing loops are added by ab-initio loop modeling. The final structure is chosen according to two different model assessment scores. To be able to observe the structural changes upon ligand binding, glycine and glutamate molecules are docked into the corresponding binding sites of the receptor. Subsequently, molecular dynamics simulations of 1.3 μs are performed for both apo and ligand-bound structures. Structural parameters, which have been considered to show functionally important changes in previous NMDAR studies, are monitored as conformational rulers to understand the dynamics of the conformational changes. Moreover, principal component analysis (PCA) is performed for the equilibrated part of the simulations. From these analyses, the differences in between apo and ligand-bound simulations can be summarized as the following: The girdle right at the beginning of the pore loop, which connects M2 and M3 helices of the ion channel, partially opens. Ligands act like an adhesive for the ligand-binding domain (LBD) by keeping the bi-lobed structure together and consequently this is reflected to the overall dynamics of the protein as an increased correlation of the LBD with especially the amino-terminal domain (ATD) of the protein.

Comparative modelling studies of fruit bromelain using molecular dynamics simulation

Fruit bromelain is a cysteine protease accumulated in pineapple fruits. This proteolytic enzyme has received high demand for industrial and therapeutic applications. In this study, fruit bromelain sequences QIM61759, QIM61760 and QIM61761 were retrieved from the National Center for Biotechnology Information (NCBI) Genbank Database. The tertiary structure of fruit bromelain QIM61759, QIM61760 and QIM61761 was generated by using MODELLER. The result revealed that the local stereochemical quality of the generated models was improved by using multiple templates during modelling process. Moreover, by comparing with the available papain model, structural analysis provides an insight on how pro-peptide functions as a scaffold in fruit bromelain folding and contributing to inactivation of mature protein. The structural analysis also disclosed the similarities and differences between these models. Lastly, thermal stability of fruit bromelain was studied. Molecular dynamics simulation of fruit bromelain structures at several selected temperatures demonstrated how fruit bromelain responds to elevation of temperature.

Improvising 5-HT7R homology model for design of high affinity ligands: model validation with docking, embrace minimization, MM-GBSA, and molecular dynamic simulations

The subtype, 5-HT₇R has been implicated in neurological disorders and presents itself as a promising target for antidepressant drugs. Design of targeted selective ligands, require a sound knowledge of 3D-receptor structure. In absence of receptor structure, structure-based design of targeted ligands relies on generation of 5-HT₇R homology model. In this study, the impact of template choice, alignment, and model building methods on the homology model of 5-HT₇R is addressed. The compactness and model quality due to the presence of cholesterol (lipidic receptor) have also been observed. The results suggest good stereochemical quality of the final model. Ramachandran Plot Analysis indicated more than 97.5% amino acid residues in the favorable region. The overall quality factor was 91.8% using ERRAT. The Z-score for backbone confirmation and packing quality were -1.248 and -1.427, respectively, using WHATCHECK. The RMS Z-score for side chain planarity was .711. Other validation results for the final model include binding site analysis in which Asp162, Val163, Phe343, Phe344, Arg350, Arg367, and Leu370 conserved residues were found in the active site, correlation coefficient of .82 in ligand-based screening and .85 in embrace minimization. Further, the model showed good correlation for agonist and antagonist in docking ([Formula: see text] ≈ .76, [Formula: see text] ≈ .82), embrace minimization ([Formula: see text] ≈ .73, [Formula: see text] ≈ .72), and MM-GBSA ([Formula: see text] ≈ .69, [Formula: see text] ≈ .75) studies. The model was subjected to Molecular Dynamics (MD) simulation of 20 ns both in ligand-free and ligand-bound receptor (agonist and antagonist) system in order to assess its stability.

An overview of comparative modelling and resources dedicated to large-scale modelling of genome sequences

Computational modelling of proteins has been a major catalyst in structural biology. Bioinformatics groups have exploited the repositories of known structures to predict high-quality structural models with high efficiency at low cost. This article provides an overview of comparative modelling, reviews recent developments and describes resources dedicated to large-scale comparative modelling of genome sequences. The value of subclustering protein domain superfamilies to guide the template-selection process is investigated. Some recent cases in which structural modelling has aided experimental work to determine very large macromolecular complexes are also cited.

Identification of a conserved 8 aa insert in the PIP5K protein in the Saccharomycetaceae family of fungi and the molecular dynamics simulations and structural analysis to investigate its potential functional role

Homologs of the phosphatidylinositol-4-phosphate-5-kinase (PIP5K), which controls a multitude of essential cellular functions, contain a 8 aa insert in a conserved region that is specific for the Saccharomycetaceae family of fungi. Using structures of human PIP4K proteins as templates, structural models were generated of the Saccharomyces cerevisiae and human PIP5K proteins. In the modeled S. cerevisiae PIP5K, the 8 aa insert forms a surface exposed loop, present on the same face of the protein as the activation loop of the kinase domain. Electrostatic potential analysis indicates that the residues from 8 aa conserved loop form a highly positively charged surface patch, which through electrostatic interaction with the anionic portions of phospholipid head groups, is expected to play a role in the membrane interaction of the yeast PIP5K. To unravel this prediction, molecular dynamics (MD) simulations were carried out to examine the binding interaction of PIP5K, either containing or lacking the conserved signature insert, with two different membrane lipid bilayers. The results from MD studies provide insights concerning the mechanistic of interaction of PIP5K with lipid bilayer, and support the contention that the identified 8 aa conserved insert in fungal PIP5K plays an important role in the binding of this protein with membrane surface. Proteins 2017; 85:1454-1467. © 2017 Wiley Periodicals, Inc.

Theoretical design of a new chimeric protein for the treatment of breast cancer

p28 and NRC peptides are two anticancer peptides with various mechanisms have shown to be effective against breast cancer. Therefore, it seems that construction of a chimeric protein containing the two peptides might cause synergistic cytotoxic effects. However, since the two peptides bear opposite charges, production of a chimeric protein in which the two moieties do not intervene each other is difficult. In this study, our goal was to find a suitable peptide linker for the new chimeric protein in a manner that none of the peptides intervene the other's function. We selected some linkers with different characteristics and lengths and created a small library of the chimeric proteins harboring these linkers. Homology modeling and molecular dynamic simulation revealed that (PA)5P and (EAAAK)3 linkers can separate the p28 and NRC peptides effectively. Thus, the chimeric protein linked with (PA)5P or (EAAAK)3 linkers might show synergistic and stronger anticancer effects than the separate peptide moieties because they could exert their cytotoxic effects freely which is not influenced by the other part.

From local structure to a global framework: recognition of protein folds

Protein folding has been a major area of research for many years. Nonetheless, the mechanisms leading to the formation of an active biological fold are still not fully apprehended. The huge amount of available sequence and structural information provides hints to identify the putative fold for a given sequence. Indeed, protein structures prefer a limited number of local backbone conformations, some being characterized by preferences for certain amino acids. These preferences largely depend on the local structural environment. The prediction of local backbone conformations has become an important factor to correctly identifying the global protein fold. Here, we review the developments in the field of local structure prediction and especially their implication in protein fold recognition.

Improvement of 3D protein models using multiple templates guided by single-template model quality assessment

MOTIVATION

Modelling the 3D structures of proteins can often be enhanced if more than one fold template is used during the modelling process. However, in many cases, this may also result in poorer model quality for a given target or alignment method. There is a need for modelling protocols that can both consistently and significantly improve 3D models and provide an indication of when models might not benefit from the use of multiple target-template alignments. Here, we investigate the use of both global and local model quality prediction scores produced by ModFOLDclust2, to improve the selection of target-template alignments for the construction of multiple-template models. Additionally, we evaluate clustering the resulting population of multi- and single-template models for the improvement of our IntFOLD-TS tertiary structure prediction method.

RESULTS

We find that using accurate local model quality scores to guide alignment selection is the most consistent way to significantly improve models for each of the sequence to structure alignment methods tested. In addition, using accurate global model quality for re-ranking alignments, prior to selection, further improves the majority of multi-template modelling methods tested. Furthermore, subsequent clustering of the resulting population of multiple-template models significantly improves the quality of selected models compared with the previous version of our tertiary structure prediction method, IntFOLD-TS.

AVAILABILITY AND IMPLEMENTATION

Source code and binaries can be freely downloaded from http://www.reading.ac.uk/bioinf/downloads/

Systematic assessment of accuracy of comparative model of proteins belonging to different structural fold classes

In the absence of experimental structures, comparative modeling continues to be the chosen method for retrieving structural information on target proteins. However, models lack the accuracy of experimental structures. Alignment error and structural divergence (between target and template) influence model accuracy the most. Here, we examine the potential additional impact of backbone geometry, as our previous studies have suggested that the structural class (all-α, αβ, all-β) of a protein may influence the accuracy of its model. In the twilight zone (sequence identity ≤ 30%) and at a similar level of target-template divergence, the accuracy of protein models does indeed follow the trend all-α > αβ > all-β. This is mainly because the alignment accuracy follows the same trend (all-α > αβ > all-β), with backbone geometry playing only a minor role. Differences in the diversity of sequences belonging to different structural classes leads to the observed accuracy differences, thus enabling the accuracy of alignments/models to be estimated a priori in a class-dependent manner. This study provides a systematic description of and quantifies the structural class-dependent effect in comparative modeling. The study also suggests that datasets for large-scale sequence/structure analyses should have equal representations of different structural classes to avoid class-dependent bias.

Effect of using suboptimal alignments in template-based protein structure prediction

Computational protein structure prediction remains a challenging task in protein bioinformatics. In the recent years, the importance of template-based structure prediction is increasing because of the growing number of protein structures solved by the structural genomics projects. To capitalize the significant efforts and investments paid on the structural genomics projects, it is urgent to establish effective ways to use the solved structures as templates by developing methods for exploiting remotely related proteins that cannot be simply identified by homology. In this work, we examine the effect of using suboptimal alignments in template-based protein structure prediction. We showed that suboptimal alignments are often more accurate than the optimal one, and such accurate suboptimal alignments can occur even at a very low rank of the alignment score. Suboptimal alignments contain a significant number of correct amino acid residue contacts. Moreover, suboptimal alignments can improve template-based models when used as input to Modeller. Finally, we use suboptimal alignments for handling a contact potential in a probabilistic way in a threading program, SUPRB. The probabilistic contacts strategy outperforms the partly thawed approach, which only uses the optimal alignment in defining residue contacts, and also the re-ranking strategy, which uses the contact potential in re-ranking alignments. The comparison with existing methods in the template-recognition test shows that SUPRB is very competitive and outperforms existing methods.

A new clustering of antibody CDR loop conformations

Previous analyses of the complementarity-determining regions (CDRs) of antibodies have focused on a small number of "canonical" conformations for each loop. This is primarily the result of the work of Chothia and coworkers, most recently in 1997. Because of the widespread utility of antibodies, we have revisited the clustering of conformations of the six CDR loops with the much larger amount of structural information currently available. In this work, we were careful to use a high-quality data set by eliminating low-resolution structures and CDRs with high B-factors or high conformational energies. We used a distance function based on directional statistics and an effective clustering algorithm with affinity propagation. With this data set of over 300 nonredundant antibody structures, we were able to cover 28 CDR-length combinations (e.g., L1 length 11, or "L1-11" in our CDR-length nomenclature) for L1, L2, L3, H1, and H2. The Chothia analysis covered only 20 CDR-lengths. Only four of these had more than one conformational cluster, of which two could easily be distinguished by gene source (mouse/human; κ/λ) and one could easily be distinguished purely by the presence and the positions of Pro residues (L3-9). Thus, using the Chothia analysis does not require the complicated set of "structure-determining residues" that is often assumed. Of our 28 CDR-lengths, 15 have multiple conformational clusters, including 10 for which the Chothia analysis had only one canonical class. We have a total of 72 clusters for non-H3 CDRs; approximately 85% of the non-H3 sequences can be assigned to a conformational cluster based on gene source and/or sequence. We found that earlier predictions of "bulged" versus "nonbulged" conformations based on the presence or the absence of anchor residues Arg/Lys94 and Asp101 of H3 have not held up, since all four combinations lead to a majority of conformations that are bulged. Thus, the earlier analyses have been significantly enhanced by the increased data. We believe that the new classification will lead to improved methods for antibody structure prediction and design.

Identification of subfamily-specific sites based on active sites modeling and clustering

MOTIVATION

Current computational approaches to function prediction are mostly based on protein sequence classification and transfer of annotation from known proteins to their closest homologous sequences relying on the orthology concept of function conservation. This approach suffers a major weakness: annotation reliability depends on global sequence similarity to known proteins and is poorly efficient for enzyme superfamilies that catalyze different reactions. Structural biology offers a different strategy to overcome the problem of annotation by adding information about protein 3D structures. This information can be used to identify amino acids located in active sites, focusing on detection of functional polymorphisms residues in an enzyme superfamily. Structural genomics programs are providing more and more novel protein structures at a high-throughput rate. However, there is still a huge gap between the number of sequences and available structures. Computational methods, such as homology modeling provides reliable approaches to bridge this gap and could be a new precise tool to annotate protein functions.

RESULTS

Here, we present Active Sites Modeling and Clustering (ASMC) method, a novel unsupervised method to classify sequences using structural information of protein pockets. ASMC combines homology modeling of family members, structural alignment of modeled active sites and a subsequent hierarchical conceptual classification. Comparison of profiles obtained from computed clusters allows the identification of residues correlated to subfamily function divergence, called specificity determining positions. ASMC method has been validated on a benchmark of 42 Pfam families for which previous resolved holo-structures were available. ASMC was also applied to several families containing known protein structures and comprehensive functional annotations. We will discuss how ASMC improves annotation and understanding of protein families functions by giving some specific illustrative examples on nucleotidyl cyclases, protein kinases and serine proteases.

AVAILABILITY

http://www.genoscope.fr/ASMC/.

Low-homology protein threading

Motivation: The challenge of template-based modeling lies in the recognition of correct templates and generation of accurate sequence-template alignments. Homologous information has proved to be very powerful in detecting remote homologs, as demonstrated by the state-of-the-art profile-based method HHpred. However, HHpred does not fare well when proteins under consideration are low-homology. A protein is low-homology if we cannot obtain sufficient amount of homologous information for it from existing protein sequence databases.

Results: We present a profile-entropy dependent scoring function for low-homology protein threading. This method will model correlation among various protein features and determine their relative importance according to the amount of homologous information available. When proteins under consideration are low-homology, our method will rely more on structure information; otherwise, homologous information. Experimental results indicate that our threading method greatly outperforms the best profile-based method HHpred and all the top CASP8 servers on low-homology proteins. Tested on the CASP8 hard targets, our threading method is also better than all the top CASP8 servers but slightly worse than Zhang-Server. This is significant considering that Zhang-Server and other top CASP8 servers use a combination of multiple structure-prediction techniques including consensus method, multiple-template modeling, template-free modeling and model refinement while our method is a classical single-template-based threading method without any post-threading refinement.

Contact:jinboxu@gmail.com

Modern homology modeling of G-protein coupled receptors: which structural template to use?

The recent availability in the literature of new crystal structures of inactive G-protein coupled receptors (GPCRs) prompted us to study the extent to which these crystal structures constitute an advantage over the former prototypic rhodopsin template for homology modeling of the transmembrane (TM) region of human class A GPCRs. Our results suggest that better templates than those currently available are required by the majority of these GPCRs to generate homology models that are accurate enough for simple virtual screening aimed at computer-aided drug discovery. Thus, we investigated: (1) which class A GPCRs would have the highest impact as potential templates for homology modeling of other GPCRs, if their structures were solved, and (2) the extent to which multiple-template homology modeling (using all currently available GPCR crystal structures) provides an improvement over single-template homology modeling, as evaluated by the accuracy of rigid protein-flexible ligand docking on these models.

Alignment of multiple protein structures based on sequence and structure features

Comparing the structures of proteins is crucial to gaining insight into protein evolution and function. Here, we align the sequences of multiple protein structures by a dynamic programming optimization of a scoring function that is a sum of an affine gap penalty and terms dependent on various sequence and structure features (SALIGN). The features include amino acid residue type, residue position, residue accessible surface area, residue secondary structure state and the conformation of a short segment centered on the residue. The multiple alignment is built by following the 'guide' tree constructed from the matrix of all pairwise protein alignment scores. Importantly, the method does not depend on the exact values of various parameters, such as feature weights and gap penalties, because the optimal alignment across a range of parameter values is found. Using multiple structure alignments in the HOMSTRAD database, SALIGN was benchmarked against MUSTANG for multiple alignments as well as against TM-align and CE for pairwise alignments. On the average, SALIGN produces a 15% improvement in structural overlap over HOMSTRAD and 14% over MUSTANG, and yields more equivalent structural positions than TM-align and CE in 90% and 95% of cases, respectively. The utility of accurate multiple structure alignment is illustrated by its application to comparative protein structure modeling.

Protein structure prediction and model quality assessment

Protein structures have proven to be a crucial piece of information for biomedical research. Of the millions of currently sequenced proteins only a small fraction is experimentally solved for structure and the only feasible way to bridge the gap between sequence and structure data is computational modeling. Half a century has passed since it was shown that the amino acid sequence of a protein determines its shape, but a method to translate the sequence code reliably into the 3D structure still remains to be developed. This review summarizes modern protein structure prediction techniques with the emphasis on comparative modeling, and describes the recent advances in methods for theoretical model quality assessment.

New tips for structure prediction by comparative modeling

Comparative modelling is utilized to predict the 3-dimensional conformation of a given protein (target) based on its sequence alignment to experimentally determined protein structure (template). The use of such technique is already rewarding and increasingly widespread in biological research and drug development. The accuracy of the predictions as commonly accepted depends on the score of sequence identity of the target protein to the template. To assess the relationship between sequence identity and model quality, we carried out an analysis of a set of 4753 sequence and structure alignments. Throughout this research, the model accuracy was measured by root mean square deviations of Calpha atoms of the target-template structures. Surprisingly, the results show that sequence identity of the target protein to the template is not a good descriptor to predict the accuracy of the 3-D structure model. However, in a large number of cases, comparative modelling with lower sequence identity of target to template proteins led to more accurate 3-D structure model. As a consequence of this study, we suggest new tips for improving the quality of omparative models, particularly for models whose target-template sequence identity is below 50%.