This Title All WIREs
How to cite this WIREs title:
WIREs Comput Mol Sci
Impact Factor: 25.113

Structural prediction of protein interactions and docking using conservation and coevolution

Full article on Wiley Online Library:   HTML PDF

Can't access this content? Tell your librarian.

Abstract Knowledge of the detailed structure of macromolecular interactions is key to a better understanding and modulation of essential cellular functions and pathological situations. Great efforts are invested in the development of improved computational prediction methods, including binding site prediction and protein–protein docking. These tools should benefit from the inclusion of evolutionary information, since the pressure to maintain functional interactions leads to conservation signals on protein surfaces at interacting sites and coevolution between contacting positions. However, unveiling such constraints and finding the best way to integrate them into computational pipelines remains a challenging area of research. Here, we first introduce evolutionary properties of interface structures, focusing on recent work exploring evolutionary mechanisms at play in protein interfaces and characterizing the complexity of evolutionary signals, such as interface deep mutational scans. Then, we review binding site predictors and interface structure modeling methods that integrate conservation and coevolution as important ingredients to improve predictive capacity, ending with a section dedicated to the prediction of binding modes between a globular protein domain and a short motif located within an intrinsically disordered or flexible region. Throughout the review, we discuss practical applications and recent promising developments, in particular regarding the use of coevolutionary information for interface structural prediction and the integration of these coevolution signals with machine learning and deep learning algorithms. This article is categorized under: Structure and Mechanism > Molecular Structures Structure and Mechanism > Computational Biochemistry and Biophysics Molecular and Statistical Mechanics > Molecular Interactions
(a) Schematic representation of the interaction networks between the yMLH1–yMLH2 heterodimer and yMER3 in yeast Saccharomyces cerevisiae and between their mouse orthologs, the mMLH1–mPMS1 heterodimer and mHFM1. yMER3 and mHFM1 are composed of five globular domains represented by squares surrounded by disordered N‐terminal and C‐terminal extensions (indicated by “N” and “C” labels). The width of the links between each pair of proteins is indicative of the experimentally observed relative interaction strength. (b) Compared architecture of the mixed lineage leukemia (MLL) complexes involving either MLL1 (left, reference PDB structure: 6KIU) or MLL3 (right, reference PDB structure: 6KIW). WDR5 subunit is colored purple, ASH2L is orange, MLL1 and MLL3 are two different shades of dark green, histone octamer is cyan, RBBP5 is yellow, ubiquitin is pink and DNA is black. Top views of the two complexes (with the nucleosome at the bottom) are provided where the nucleosomes and the RBBP5 subunits are exactly in the same orientation. Due to differences between MLL1 and MLL3, the relative positions of WDR5 and even more ASH2L are quite different between the two complexes even though the same overall architecture is maintained, providing a likely explanation for the large difference in binding affinity for RBBP5‐ASH2L between MLL1 and MLL3. These differences in the details of the assembly reflect a different functional role for MLL1 compared to MLL3 and other MLLs
[ Normal View | Magnified View ]
Graphical summary of a selection of methods presented in the review for the structural prediction of protein interactions. Those methods are available as web servers, except InterPep and InterPep2 (see Table for links and references). Methods for predicting interactions between globular domains are presented on the left in oval shapes, methods for predicting protein–peptide interactions mediated by short motifs are presented on the right in rectangles and methods suitable for both are in the middle in rounded rectangles. The background color code denotes how structural modeling makes use of evolutionary information: through sequence conservation (light orange), sequence coevolution (dark orange) or structural homology (blue), or indirectly through information provided by upstream methods that predict binding sites, motifs or contacts (green). Template‐based modeling, which relies on homologous complexes of known structure, bypasses the use of upstream methods compared to docking‐based predictions
[ Normal View | Magnified View ]
(a) Prediction of the CAPRI target T95 nucleosome‐ubiquitin ligase. The structure of the best model found by InterEvDock2 is shown in blue (ranked fourth and of acceptable quality) and compared to the experimental reference structure in gray (PDB: 4R8P). The nucleosome is shown in white. (b) Prediction of CAPRI target T134 between the dynein domain and a MAG3 disordered region. In this case, the PSSM profile of the dynein binding domain motif (top) could be used to identify the binding region of dynein on MAG3. Close and distant homologous dynein‐peptide complexes were used to generate the PSSM profile centered on the anchor residue in these interactions (marked with a star and most frequently a glutamine in position 0 of the PSSM). The binding region was identified as corresponding to the frame with the only positive score (highlighted in yellow and centered on leucine 613, which is marked by a star) when gliding the PSSM along the MAG3 peptide sequence (middle). Three groups managed to return a high quality model in first position, one for which the structure is represented above in magenta (bottom) and compared to the experimental reference structure in gray (PDB: 6GZJ)
[ Normal View | Magnified View ]
Flowchart of the protocols and tools described in the review to carry out structural modeling of protein interactions taking into account evolutionary information. When starting from the sequences of interacting proteins, structural modeling of their assembly can follow two strategies, both relying on evolutionary relationships. The first one (1), generally more accurate but restricted to a limited number of cases, uses homology relationships and template‐based docking methods to generate structures of assemblies, which are reviewed in two subsections of this review for globular and disordered regions, respectively. The second strategy (2) relies on a combination of approaches involving structural modeling of the partners when possible, evolutionary analysis of the disordered regions and use of evolutionary information to identify binding patches at the surface of globular domains (3, 6). Combined with coevolution analyses, free docking methods can incorporate all these levels of information to produce models of assemblies (4, 5, and 7). These methods are reviewed for both globular and disordered systems
[ Normal View | Magnified View ]
Interface residue prediction of the RBBP5 protein using different programs: (a) Rate4Site, (b) SPPIDER, and (c) ISPRED4. Interface residue predictions are displayed on the surface of RBBP5 and color‐coded from white (predicted as noninterface) to yellow to red (highest predicted probability to be involved in an interface). The RBBP5 subunit is involved in interfaces with six different partners (five proteins and one DNA) in the MLL1 complex associated with the nucleosome (interfaces 1, 2, 3, 4, and 6 in reference PDB structure: 6KIU) and in one intramolecular interaction (interface 5 in reference PDB structure: 6KM7). These interfaces are mediated either by its globular beta‐propeller domain (interfaces 1, 4, 5, and 6) or by its N‐terminal intrinsically disordered region (interfaces 1, 2, and 3). ISPRED4 prediction exhibits remarkable sensitivity in the detection of interface residues in RBBP5 for all seven interfaces with almost no false positives. As in Figure b, WDR5 subunit is colored purple (1), ASH2L is orange (2), MLL1 is dark green (3), histones octamer is cyan (4), RBBP5 is yellow and lime (5), ubiquitin is pink (6) and DNA is black
[ Normal View | Magnified View ]

Browse by Topic

Molecular and Statistical Mechanics > Molecular Interactions
Structure and Mechanism > Computational Biochemistry and Biophysics
Structure and Mechanism > Molecular Structures

Access to this WIREs title is by subscription only.

Recommend to Your
Librarian Now!

The latest WIREs articles in your inbox

Sign Up for Article Alerts