SKEMPI 2.0: an updated benchmark of changes in protein-protein binding energy, kinetics and thermodynamics upon mutation.

Jankauskaite J, Jiménez-García B, Dapkunas J, Fernández-Recio J, Moal IH. · 2019

Read SKEMPI 2.0: an updated benchmark of changes in protein-protein binding energy, kinetics and thermodynamics upon mutation. by Jankauskaite J, Jiménez-García B, Dapkunas J, Fernández-Recio J, Moal IH. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6361233.txt.

Introduction

Jankauskaitė and co-authors present a database article in Bioinformatics. They announce SKEMPI 2.0, a major update to a benchmark of how mutations change protein binding. The database collects measured effects for protein pairs whose structures are known.

The new version holds hand-checked data for 7,085 mutations, more than double the earlier release. It adds binding speed data for 1,844 mutations and heat-related measures for 443. It also records 440 mutations that abolish binding entirely. The authors describe their sources and checks in detail. Researchers can use the free database to test prediction methods and study how mutations cause disease.

Catalog record

CREATORJankauskaite J, Jiménez-García B, Dapkunas J, Fernández-Recio J, Moal IH.
YEAR2019
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC6361233

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

SKEMPI 2.0: an updated benchmark of changes in protein–protein binding energy, kinetics and thermodynamics upon mutation By Justina Jankauskaitė, Brian Jiménez-García, Justas Dapkūnas, Juan Fernández-Recio, Iain H Moal, Ioannis Xenarios Published in: Bioinformatics DOI: 10.1093/bioinformatics/bty635 ABSTRACT Abstract Motivation Understanding the relationship between the sequence, structure, binding energy, binding kinetics and binding thermodynamics of protein–protein interactions is crucial to understanding cellular signaling, the assembly and regulation of molecular complexes, the mechanisms through which mutations lead to disease, and protein engineering. Results We present SKEMPI 2.0, a major update to our database of binding free energy changes upon mutation for structurally resolved protein–protein interactions. This version now contains manually curated binding data for 7085 mutations, an increase of 133%, including changes in kinetics for 1844 mutations, enthalpy and entropy changes for 443 mutations, and 440 mutations, which abolish detectable binding. Availability and implementation The database is available as supplementary data and at https://life.bsc.es/pid/skempi2/. Supplementary information Supplementary data are available at Bioinformatics online. 1 INTRODUCTION Protein–protein interactions are central to almost all biological processes, from cellular signal transduction and the assembly of mesoscopic structures such as myofilaments, to viral adhesion and the immune response. Consequently the effects of changes in protein sequence on the structure, thermodynamics and kinetics of protein–protein interactions has wide implications for constraining the permissible substitutions that accrue over the course of evolution, and for understanding the molecular etiology of disease. Methods which measure, predict or optimize these changes have applications in designing de novo interactions (Fleishman et al., 2011), enhancing the specificity and affinity of biological therapeutics (e.g. Arkadash et al., 2017), designing combinatorial protein libraries (e.g. Guntas et al., 2010), uncovering the effects of pathological mutations (e.g. Tidow et al., 2006), locating druggable binding sites (e.g. Stevers et al., 2017) and binding hotspots for drug design (Guo et al., 2014), altering binding kinetics (Cohen-Khait and Schreiber, 2016; Rosenfeld et al., 2017), protein–protein docking (e.g. Epa et al., 2013), and characterizing transition states (e.g. Wu et al., 2002), binding pathways (Plattner et al., 2017), and sequence-affinity landscapes (Aizner et al., 2014). SKEMPI is a manually curated database of mutations in structurally characterized protein–protein interactions and the effect of those mutations on binding affinity and other parameters (Moal and Fernandez-Recio, 2012). The first release has been used as a basis for many further studies, including the development of energy functions (Moal and Fernandez-Recio, 2013; Moal et al., 2015b) which were subsequently implemented in the CCharPPI web server for characterizing protein–protein interactions (Moal et al., 2015a), as well as being used for ranking docked poses (Barradas-Bautista et al., 2017; Moal et al., 2013; Moal et al., 2017; Pfeiffenberger et al., 2017). SKEMPI has also been used to study human disease (Das et al., 2014; Peng and Alexov, 2016; Petukh et al., 2015a), assessing the role of dynamics on binding (Sumbul et al., 2015), exploring the conservation of binding regions (Hu et al., 2016), evaluating experimental affinity measurement methods (Geng et al., 2016), as well serving as a data source for models which predict dissociation rate changes upon mutation (Agius et al., 2013), pathological mutations (Gossage et al., 2014), hotspot residues (e.g. Hwang et al., 2014; Liu et al., 2015; Melo et al., 2016; Simoes et al., 2017) and changes in binding energy (e.g. Barlow et al., 2018; Berliner et al., 2014; Dehouck et al., 2013; Dourado and Flores, 2014; Lai et al., 2017; Li et al., 2016; Moretti et al., 2013; Pallara et al., 2013; Pantazes et al., 2015; Petukh et al., 2015b; Pires et al., 2014; Xiong et al., 2017; Yan et al., 2017; Zhao et al., 2014). Here we present a major update to the benchmark in terms of the number of mutations in the database and the number of different systems included (Table 1). We now also include details of the experimental method for all entries, based on the categories of Geng et al., 2016, as well as mutations which abolished detectable binding or for which only an upper or lower affinity limit could be ascertained for the wild-type or mutant. [Table: Comparison with previous version] 2 MATERIALS AND METHODS 2.1 Data sources Just over two fifths of the data come from the previous version of SKEMPI (Moal and Fernandez-Recio, 2012), comprised mostly of data found in literature sources which came to the authors’ attention, in some cases during the data collection for the structural affinity benchmark (Kastritis et al., 2011) and following references therein. Some entries in SKEMPI 1.1 were found by checking the references in the ASEdb (Thorn and Bogan, 2001) and PINT (Kumar and Gromiha, 2006) databases, although not all the data passed the checks required for inclusion (see Section 2.2). Similarly, most of the new data in SKEMPI 2.0 was found by searching the literature, partly in tandem with the literature search for the more recent structural affinity benchmark (Vreven et al., 2015). During data collection three other relevant databases were published: ABbind (Sirin et al., 2016), PROXiMATE (Jemimah et al., 2017) and dbMPIKT (Liu et al., 2017). Their references were checked if they were not already included. Data from these sources comprise 4%, 3% and 6% of SKEMPI 2.0 respectively. As with ASEdb and PINT, none of the data were directly copied into SKEMPI. Moreover, the cited papers were read and data entered using the same checks and procedures as other entries. 2.2 Data collection Each entry was found in the literature and manually vetted. To ensure quality, a number of stringent checks were applied. Firstly, we ensured that the structure and the paper reporting the affinities refer to the same protein in the same species, and that structural and affinity data matched in terms of cofactors, ancillary chains and post-translational modifications. For instance, we distinguish between RasGTP and RasGppNHp, as the nucleotide modulates the affinity of Ras with its effectors. Where the full-length protein was not used, we checked to ensure that the fragment in the crystal structure matched that for which affinities are reported. Once the checks are passed, the data is collected, including the PDB file, the chains of the interacting subunits, the mutation, the wild-type and mutant affinities (KD, M), the reference, the names of the proteins, the temperature at which the experiment is performed (T, K), the experimental method used (an extension of the category scheme of Geng et al., 2016), notes on the entry and, when available, the association rate (kon, M−1s−1), dissociation rate (koff, s−1), enthalpy (ΔH, kcal.mol−1) and entropy (ΔS, cal.mol−1.K−1). For cases where multiple PDB entries are available, the higher resolution structure is chosen. Where affinities or kinetic or thermodynamic parameters are reported in different units, these are converted to the units specified above. In some cases, when not reported directly, KD, kon, koff, ΔH and ΔS were calculated using the relationships ΔG=ΔH−TΔS=RTln(KD) and KD=1/KA=koff/kon. To ensure consistency, the residue numbering in SKEMPI is the same as that reported in the PDB file. Thus, the numbering is often shifted or altered compared to that in the cited paper such that, for instance, if a crystal structure of an antibody is reported in the Kabat numbering scheme but the mutation data is not, then the mutation data is converted before entry into SKEMPI. For all

Read the complete text