Modeling protein quaternary structure of homo- and hetero-oligomers beyond binary interactions by homology.
Bertoni M, Kiefer F, Biasini M, Bordoli L, Schwede T. · 2017
Read Modeling protein quaternary structure of homo- and hetero-oligomers beyond binary interactions by homology. by Bertoni M, Kiefer F, Biasini M, Bordoli L, Schwede T. (2017) 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-pmc5585393.txt.
Introduction
Bertoni and co-authors present a modeling study in Scientific Reports. Cell work often depends on proteins joining into complexes. Lab structures for such assemblies lag far behind interaction data. Their method models whole complexes by homology, starting from amino acid sequences alone.
The approach must guess both the partner count and the overall shape. The team scores how well contact surfaces are conserved across evolution. They also cluster known complex structures to measure shape variety. Blending those scores with classic surface traits improved template choice. That gave more reliable models of protein complexes. Readers see how sequence clues can stand in when no lab structure exists.
Catalog record
| CREATOR | Bertoni M, Kiefer F, Biasini M, Bordoli L, Schwede T. |
|---|---|
| YEAR | 2017 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 69 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Modeling protein quaternary structure of homo- and hetero-oligomers beyond binary interactions by homology By Martino Bertoni, Florian Kiefer, Marco Biasini, Lorenza Bordoli, Torsten Schwede Published in: Scientific Reports DOI: 10.1038/s41598-017-09654-8 ABSTRACT Cellular processes often depend on interactions between proteins and the formation of macromolecular complexes. The impairment of such interactions can lead to deregulation of pathways resulting in disease states, and it is hence crucial to gain insights into the nature of macromolecular assemblies. Detailed structural knowledge about complexes and protein-protein interactions is growing, but experimentally determined three-dimensional multimeric assemblies are outnumbered by complexes supported by non-structural experimental evidence. Here, we aim to fill this gap by modeling multimeric structures by homology, only using amino acid sequences to infer the stoichiometry and the overall structure of the assembly. We ask which properties of proteins within a family can assist in the prediction of correct quaternary structure. Specifically, we introduce a description of protein-protein interface conservation as a function of evolutionary distance to reduce the noise in deep multiple sequence alignments. We also define a distance measure to structurally compare homologous multimeric protein complexes. This allows us to hierarchically cluster protein structures and quantify the diversity of alternative biological assemblies known today. We find that a combination of conservation scores, structural clustering, and classical interface descriptors, can improve the selection of homologous protein templates leading to reliable models of protein complexes. INTRODUCTION Macromolecular complexes are of central interest in structural biology1–3. Direct physical protein-protein interactions (PPIs), as well as indirect ones, are essential for performing and regulating cellular activities such as signal transduction, cell-cycle, morphological differentiation, cell motility, transcription and translation. A precise description of proteins’ interactions and quaternary structure (QS) is fundamental to gain a detailed molecular understanding on how these interactions are mediated and regulated. While experimental information on interacting partners obtained with high-throughput methods4–6 such as two-hybrid screening (Y2H) or affinity purification of complexes grows with an exponential trend7–10, the number of experimentally determined three-dimensional complexes and oligomeric structures is lagging behind. Shedding light on the atomic details of such interactions is challenging since the expression of protein complexes is often tightly regulated and obtaining sufficient concentrations of intact complexes for structure determination is often not trivial. Aiming to fill this gap, several computational techniques to model protein interactions have been developed, which differ in type and amount of structural information required as starting point. One of the first approaches used to model interactions de novo, when structures of the individual components are available, was macromolecular docking. The relative orientation of two proteins is sampled and scored by exploiting e.g. the components’ shape11 or physicochemical complementarity12. Recently, amino acid co-evolution analysis (see ref. 13 for a review) has been successfully applied to identify proximal residues in interfaces14 thus increasing the accuracy of the results. Docking approaches are generally more accurate when no significant conformational changes are required for interface formation, as documented by the regular CAPRI experiment (Critical Assessment of Prediction of Interactions)15. When some experimental details of the interaction are available (e.g. EM density maps, crosslinking, SAXS or NMR data, co-evolution analysis, etc.), different “hybrid-modeling” tools can be used (e.g. the Integrative Modeling Platform (IMP)16, the Rosetta Suite17, or HADDOCK18) to apply experimental constrains when modeling sizable assemblies. The number of ways proteins interact in nature is probably limited19, 20, and it has been observed that similar binding modes can be identified for almost all known protein-protein interactions21. Furthermore, Honig’s group noted that the location of the interface in structural homologs is often conserved22. These observations paved the way for homology modeling (aka comparative or template-based modeling) of protein complexes, where uncharacterized interactions are modeled using experimental structures of homologous interacting protomers (interologs) as templates. Approaches based on homology are scalable to full genomes and successfully reduced the gap between known interactions and those that are structurally characterized for several practical applications21, 23–25. While some in silico docking techniques exploit information about the stoichiometry or the symmetry of the complex26–29 to predict multimeric assemblies, the majority of docking and homology based approaches are focused on dimeric interactions, bypassing higher-order quaternary structures. The importance of prediction of complex assemblies has been highlighted by the introduction of quaternary structure prediction assessment in the recent CASP XII (Critical Assessment of protein Structure Prediction)30, 31 and the CAMEO (Continuous Automated Model Evaluation)32 experiments. In this study, we propose an approach to identify the stoichiometry and overall structure of protein complexes using amino acid sequences as starting point. We focus on efficiently using the information on quaternary structures available in the PDB repository and encoded in multiple sequence alignments for extending the scope and automating homology modeling to appropriately address protein assemblies. Overall, throughout a given protein family quaternary structure is less conserved than tertiary structure, i.e. while the fold of a polypeptide chain remains structurally similar the number of subunits forming the biologically relevant quaternary structure can vary significantly33, 34. However, if a specific interaction between two protein chains plays a structural or functional role, it is reasonable to expect that residues at the corresponding interface are less free to vary hence increasing evolutionary conservation in these regions35, 36. Here, we introduce a refined analysis of interface conservation which captures how interface conservation varies as a function of evolutionary distance within a protein family. We employ this analysis (which we refer to as Protein-Protein Interaction (PPI) fingerprints) for two critical tasks: first, the discrimination of crystal artifacts from biological contacts, which is a crucial step in determining the correct quaternary state of crystal structures to be used as templates in homology modeling; and second, the evaluation of interface quality in models to assess the confidence in the predicted quaternary structure. In parallel to these evolutionary considerations we also analyze the geometry of oligomers. Even at high sequence identity, proteins are often represented in multiple different conformations and quaternary structures in the PDB. Hence, selecting correct templates for homology modeling is essential. We define a distance measure (QS-score) that quantifies the similarity between interfaces as a function of shared interfacial contacts. QS-score thereby discriminates between alternative quaternary structures and binding modes. We use this distance measure to evaluate the diversity of quaternary conformations represented in experimental structures and for measuring the accuracy of models. Using a supervised machine learning approach, Support Vector Machines (SVM), we combine interface conservation, structural clustering and other template features to rank and automatically select templates that maximize the predicted interface quality for a specific protein of interest. Based on this approach we were able to assign the correct quaternary structure for the majority of proteins of our data set. Finally, the application of our approach is illustrated by the prediction of fructose bisphosphate aldolase (FBA) from Haloferax volcanii, which exemplifies the modeling challenges faced when homologs in closely related organisms assume a variety of oligomeric conformations. RESULTS AND DISCUSSION Interface conservation: PPI fingerprints Proteins acquire oligomeric organization for a variety of functional and biophysical advantages: modular elements are less prone to coding errors, oligomeric regulation add an additional level of control, large structures are more stable and can perform their function cooperatively37. These and other processes are influencing the evolution of proteins’ interface formation34, 38. During evolution,