Journal of Molecular Biology
Protein Interaction Energy Landscapes are Shaped by Functional and also Non-functional Partners
Graphical abstract
Introduction
Biomolecular interactions and particularly protein-protein interactions (PPIs) are central for many physiological processes and are of utmost importance for the functioning of the cell. As such, they have brought about a wealth of studies over the last decades [[1], [2], [3], [4], [5]]. The concentration of proteins in a cell has been estimated to be approximately 2–4 million proteins per cubic micron [6]. In such a highly crowded environment, proteins constantly encounter each other, and numerous nonspecific interactions are then likely to occur [[7], [8], [9], [10]]. For example, in the cytosol of S. cerevisiae, a protein can encounter up to 2000 different proteins [11]. In this complex jigsaw puzzle, each protein has evolved to bind the right piece(s) in the right way at the right moment (positive design), and to prevent misassembly and non-functional interactions (negative design) [[12], [13], [14], [15], [16]].
Positive design constrains the physicochemical properties and the evolution of protein-protein interfaces. Indeed, a strong selection pressure operates on binding sites to maintain the functional assembly, including the functional partner and the functional binding mode. For example, homologs sharing at least 30% sequence identity almost invariably interact in the same way [17]. Conversely, the negative design prevents proteins from being trapped in the numerous competing non-functional interactions inherent to the crowded environment of the cell. Many studies have described the relationship between the propensity of proteins for promiscuous interactions and their abundances or surface properties [[18], [19], [20], [21]]. Particularly, it has been shown that misinteraction avoidance shapes the evolution and physicochemical properties of abundant proteins, resulting in a slower evolution and less sticky surfaces than what is observed for less abundant ones [18,[22], [23], [24], [25], [26]]. The whole surface of abundant proteins is thus constrained, preventing them from engaging in deleterious non-specific interactions that could dramatically impact the cell at high concentrations [25]. For example, it has been shown in E. coli that the net charge, as well as the charge distribution on protein surfaces, affect the diffusion coefficients of proteins in the cytoplasm [19,27]; positively charged proteins move up to 100 times more slowly as they get caught in nonspecific interactions with ribosomes, which are negatively charged, and which, thus shape the composition of the cytoplasmic proteome [27].
All these studies show that both positive and negative designs operate on the whole protein surface. Binding sites are constrained to maintain functional assemblies (i.e., functional binding modes with functional partners) while the rest of the surface is constrained to avoid non-functional assemblies. These constraints should, therefore, shape the energy landscapes of functional but also, non-functional interactions to prevent the latter from prevailing over the former. This should have consequences on the evolution of the whole protein surface propensity to interact with its environment since non-interacting surfaces are in constant competition with functional binding sites. We can thus hypothesize that the distribution over the protein surface of its interaction propensity is constrained during evolution in order to ensure that proteins correctly bind functional partners and do not bind to other proteins to form non-functional assemblies.
In this work, we focus on protein surfaces as a “proxy” for regulating functional and non-functional PPIs. We investigate interaction energy landscapes of proteins with native and non-native partners and ask whether the distribution over the protein surface of its interaction propensity is conserved during evolution. With this aim in mind, we performed large-scale docking simulations to characterize interactions involving either native or native-related (i.e., partners of their homologs) partners or arbitrary partners. Docking simulations have the advantage of being able to simulate the interaction of a protein with arbitrary partners, which is very difficult to address with experimental approaches. Docking algorithms are now fast enough for large-scale applications and allow for the characterization of interaction energy landscapes for thousand of protein couples. Typically, a docking simulation takes from a few minutes to a couple of hours on modern processors [[28], [29], [30]], opening the way for extensive cross-docking experiments [[31], [32], [33], [34], [35]]. Here, we performed a cross-docking experiment involving 103 proteins docked with their native or native-related partners and their homologs, as well as arbitrary partners and their homologs. We represented the interaction energy landscapes resulting from each docking calculation with a two dimensional (2D) energy map in order to (i) realize the interaction propensity mapping of a protein surface in interaction with any protein and (ii) easily compare the energy maps resulting from the docking of a same protein with different sets of homologous partners. Thus, we characterize the evolution of the distribution over a protein surface of its interaction propensity for native or arbitrary partners.
Section snippets
Characterization of the interaction propensity for the whole surface of a protein
If positive and negative designs constrain the propensity of the whole surface of proteins to interact with their functional partners or random encounters, this should shape the evolution of interaction energy landscapes of functional protein pairs but also of random encounter pairs. Consequently, we expect the interaction energy landscape involving two proteins (forming functional or arbitrary partners) to be conserved for a pair of proteins that are homologous to them. Testing this hypothesis
Impact of positive and negative design on the interaction propensity of proteins
In this study, we address the impact of positive and negative design on thousands of interaction energy landscapes with a synthetic and efficient representation of the docking energy landscapes: two-dimensional energy maps. These maps reflect the spatial distribution over the whole surface of a protein (namely the receptor) of its propensity to interact with a given partner (namely, the ligand). We show that the distribution on the protein surface of all regions, including cold, intermediate,
Protein dataset
The dataset comprises 103 protein structures divided into 16 families of structural homologs (see S1 Table for a detailed list of each family). Each family is composed of a monomer (i.e., the master of the family) selected from the protein-protein docking benchmark 5.0 [69] in its unbound forms. Each master protein is a native partner of a master protein of another family (i.e., the 3D structure of the corresponding complex has been characterized experimentally). The two corresponding families
Fundings
SSM work is supported by the “Initiative d’Excellence” program from the French State (Grant “DYNAMO”, ANR-11-LABX-0011-01). HS work was supported by a French government fellowship. This work was performed using HPC resources from GENCI-[TGCC/CINES/IDRIS] (Grant 2017 - [DARI A0030707460]).
Acknowledgments
We thank F. Fraternali, R. Guerois, E. Laine, and M. Montes for their constructive comments on the manuscript.
References (78)
- et al.
Macromolecule diffusion and confinement in prokaryotic cells
Curr. Opin. Biotechnol.
(2011) Macromolecular crowding: an important but neglected aspect of the intracellular environment
Curr. Opin. Struct. Biol.
(2001)- et al.
High-resolution mapping of protein concentration reveals principles of proteome architecture and adaptation
Cell Rep.
(2014) - et al.
A de novo protein binding pair by computational design and directed evolution
Mol. Cell
(2011) - et al.
The relationship between sequence and interaction divergence in proteins
J. Mol. Biol.
(2003) - et al.
Mistranslation-induced protein misfolding as a dominant constraint on coding-sequence evolution
Cell
(2008) - et al.
Identification of protein–protein interaction sites from docking energy landscapes
J. Mol. Biol.
(2004) - et al.
A simple method for displaying the hydropathic character of a protein
J. Mol. Biol.
(1982) A new method for mapping macromolecular topography
J. Mol. Graph. Model.
(2003)- et al.
Evolution of specificity in protein-protein interactions
Biophys. J.
(2014)
Modeling protein association mechanisms and kinetics
Curr. Opin. Struct. Biol.
Proteins feel more than they see: fine-tuning of binding affinity by properties of the non-interacting surface
J. Mol. Biol.
Protein–protein interactions: hot spots and structurally conserved residues often locate in complemented pockets that pre-organized in the unbound states: implications for docking
J. Mol. Biol.
Hot regions in protein–protein interactions: the organization and contribution of structurally conserved hot spot residues
J. Mol. Biol.
Identification of protein interaction partners and protein–protein interaction sites
J. Mol. Biol.
Protein abundance biases the amino acid composition of disordered regions to minimize non-functional interactions
J. Mol. Biol.
Protein binding specificity versus promiscuity
Curr. Opin. Struct. Biol.
Understanding protein–protein interactions using local structural features
J. Mol. Biol.
Surface map comparison: studying function diversity of homologous proteins
J. Mol. Biol.
Updates to the integrated protein–protein interaction benchmarks: docking benchmark version 5 and affinity benchmark version 2
J. Mol. Biol.
A Dirichlet process mixture model for brain MRI tissue classification
Med. Image Anal.
A computational interactome and functional annotation for the human proteome
ELife
Protein–protein interaction and quaternary structure
Q. Rev. Biophys.
Protein promiscuity and its implications for biotechnology
Nat. Biotechnol.
New embo MEMBER’S review: diversity of protein-protein interactions
EMBO J.
The molecular sociology of the cell
Nature
What is the total number of protein molecules per cell volume? A call to rethink some published values: insights & Perspectives
Bioessays
Diffusion, crowding & protein stability in a dynamic molecular model of the bacterial cytoplasm
PLoS Comput. Biol.
Biomolecular interactions modulate macromolecular structure and dynamics in atomistic model of a bacterial cytoplasm
ELife
Natural -sheet proteins use negative design to avoid edge-to-edge aggregation
Proc. Natl. Acad. Sci.
Physicochemical principles that regulate the competition between functional and dysfunctional association of proteins
Proc. Natl. Acad. Sci.
Robust protein protein interactions in crowded cellular environments
Proc. Natl. Acad. Sci.
Proteins evolve on the edge of supramolecular self-assembly
Nature
Topology of protein interaction network shapes protein abundances and strengths of their functional and nonspecific interactions
Proc. Natl. Acad. Sci.
Exploring weak, transient protein–protein interactions in crowded in vivo environments by in-cell nuclear magnetic resonance spectroscopy
Biochemistry
Quinary interactions weaken the electric field generated by protein side-chain charges in the cell-like environment
J. Am. Chem. Soc.
Physicochemical code for quinary protein interactions in Escherichia coli
Proc. Natl. Acad. Sci. U.S.A.
Highly expressed genes in yeast evolve slowly
Genetics
Constraints imposed by non-functional protein–protein interactions on gene expression and proteome size
Mol. Syst. Biol.
Cited by (7)
Kinematic Vibrational Entropy Assessment and Analysis of SARS CoV‑2 Main Protease
2022, Journal of Chemical Information and ModelingSURFMAP: A Software for Mapping in Two Dimensions Protein Surface Features
2022, Journal of Chemical Information and ModelingFrom complete cross-docking to partners identification and binding sites predictions
2022, PLoS Computational BiologyPrediction of Protein–Protein Binding Affinities from Unbound Protein Structures
2022, Methods in Molecular BiologyMoving pictures: Reassessing docking experiments with a dynamic view of protein interfaces
2021, Proteins: Structure, Function and Bioinformatics