Probing binding hot spots at protein–RNA recognition sites
| Autor: | Sunandan Mukherjee, Ranjit Prasad Bahadur, Amita Barik, Chandran Nithin, Naga Bhushana Rao Karampudi |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2015 |
| Předmět: |
0301 basic medicine
Models Molecular Protein Conformation Structural alignment Molecular Sequence Data RNA-binding protein Plasma protein binding Biology Conserved sequence Evolution Molecular 03 medical and health sciences Protein structure Genetics Humans Amino Acid Sequence Binding site Databases Protein Peptide sequence Conserved Sequence Binding Sites Models Statistical RNA RNA-Binding Proteins Water 030104 developmental biology Biochemistry Biophysics Methods Online Nucleic Acid Conformation Thermodynamics Protein Binding |
| Zdroj: | Nucleic Acids Research |
| ISSN: | 1362-4962 0305-1048 |
| Popis: | We use evolutionary conservation derived from structure alignment of polypeptide sequences along with structural and physicochemical attributes of protein-RNA interfaces to probe the binding hot spots at protein-RNA recognition sites. We find that the degree of conservation varies across the RNA binding proteins; some evolve rapidly compared to others. Additionally, irrespective of the structural class of the complexes, residues at the RNA binding sites are evolutionary better conserved than those at the solvent exposed surfaces. For recognitions involving duplex RNA, residues interacting with the major groove are better conserved than those interacting with the minor groove. We identify multi-interface residues participating simultaneously in protein-protein and protein-RNA interfaces in complexes where more than one polypeptide is involved in RNA recognition, and show that they are better conserved compared to any other RNA binding residues. We find that the residues at water preservation site are better conserved than those at hydrated or at dehydrated sites. Finally, we develop a Random Forests model using structural and physicochemical attributes for predicting binding hot spots. The model accurately predicts 80% of the instances of experimental ΔΔG values in a particular class, and provides a stepping-stone towards the engineering of protein-RNA recognition sites with desired affinity. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|