Abstract
The capability of current MD simulations to be used as a tool in rational design of agonists of medically interesting enzyme RNase L was tested. Dimerization and enzymatic activity of RNase L is stimulated by 2',5'-linked oligoadenylates (pA(25)A(25)A; 2-5A).
First, it was necessary to ensure that a complex of monomeric human RNase L and 25A was stable in MD simulations. It turned out that Glu131 had to be protonated.
The non-protonated Glu131 caused dissociation of 2-5A from RNase L. Because of the atypical 2'-5' internucleotide linkages and a specific spatial arrangement of the 25A trimer, when a single molecule carries all possible conformers of the glycosidic torsion angle, several versions of the AMBER force field were tested.
One that best maintained functionally important interactions of 25A and RNase L was selected for subsequent MD simulations. Furthermore, we wonder whether powerful GPUs are able to produce MD trajectories long enough to convincingly demonstrate effects of subtle perturbations of interactions between 25A and RNase L.
Detrimental impacts of various point mutations of RNase L (R155A, F126A, W60A, K89A) on 2-5A binding were observed on a time scale of 200 ns. Finally, 2-5A analogues with a bridged 3'-O,4'-C-alkylene linkage (B) introduced into the adenosine units (A) were used to assess ability of MD simulations to distinguish on the time scale of hundreds of nanoseconds between agonists of RNase L (pA(25)A(25)B, pB(25)A(25)A, pB(25)A(25)B) and inactive analogs (pA(25)B(25)A, pA(25)B(25)B, pB(25)B(25)A, pB(25)B(25)B).
Agonists were potently bound to RNase L during 200 ns MD runs. For inactive 2-5A analogs, by contrast, significant disruptions of their interactions with RNase L already within 100 ns MD runs were found.