ИСТИНА

A new algorithm has been developed to quantitatively measure the difference in protein motion based on molecular dynamics simulations and RMSD time-series analysis. To compare structural fluctuations at different conditions we take advantage of the fact that in most RMSD curves (including complicatedly wavy shaped ones) it is possible to determine the basic linear component – the trend in protein deviation from to the initial conformation. Representation of protein motion in terms of linear trends provides an opportunity to process in a reasonable time a vast amount of data extracted from different MD runs of proteins at various conditions. The aim of the algorithm is to reveal induced conformational changes by examining a protein structure before and after a treatment is applied. The method compares molecular motion of a protein at two conditions – hereafter referred to as A and B (treatment versus no treatment) – to detect changes in structural mobility and select a set of amino acid residues which make the most significant contribution to the observed structural rearrangement. A and B can describe an enzyme with a bound ligand versus an unbound enzyme – to study the ligand-induced conformational change; a wild type protein versus a mutant – to estimate the impact of an amino acid substitution on protein structure and function; a protein at different protonation states of ionazible residues – to study protein stability at different pH conditions, etc. It is also possible to compare different proteins which share a common structural framework (e.g., homologs within one family). Statistical comparison of RMSD curves – time-series describing the molecular displacement of protein backbone atoms from the initial reference structure in time – is used to measure relative structural mobility and to study protein movement. In order to use the method, as many as possible independent MD simulations shall be performed for a protein at each condition A and B to improve sampling and collect more data. Then RMSD curves shall be calculated for each amino acid residue in each MD simulation. These RMSD time-series are used as an input to the algorithm which has two basic steps. First, pairwise comparisons are performed between all time-series of a protein which describe the displacement of the same amino acid residue at A and B. Second, statistical analysis of the observed pairwise comparisons is performed to select residues which have shown the most significant movement as a result of a treatment. The output is a list of residues whose structural behavior before and after a treatment is significantly different from the whole protein on average, ranked by statistical significance and atomic displacement. We show that analysis of linear trends in protein dynamics is a very useful tool to identify and study particular structural regions and amino acid residues with different mobility.