Predicting the open conformations of protein kinases using molecular dynamics simulations

Protein kinases (PK) control phosphorylation in eukaryotic cells, and thereby regulate
metabolic pathways, cell cycle progression, apoptosis and transcription.
Consequently there is significant interest in manipulating PK activity and treat
diseases by using small-molecule drugs. All PK catalytic domains undergo large
conformational changes as a result of substrate binding and phosphorylation. The
“closed” state of a PK cataltic domain is the only state able to phosphorylate the target
substrate, which makes the two other observed states (the “open” and the
“intermediate” states) interesting drug targets. We investigate if MD simulations
starting from the closed state of the catalytic domain of protein kinase A (C-PKA) can
be used to produce realistic structures representing the intermediate and/or open
conformation of C-PKA, since this would allow for drug docking calculations and drug design using MD snapshots. We perform 36 ten-nanosecond MD simulations
starting from the closed conformation (PDB ID: 1ATP) of C-PKA in various liganded
and phosphorylated states. The results show that MD simulations are capable of
reproducing the open conformation of C-PKA with good accuracy within 1 ns of
simulation as measured by Cα RMSDs and RMSDs of atoms defining the ATPbinding
pocket. Importantly we are able to show that even without knowledge of the
structure of the open form of C-PKA, we can identify the MD snapshots resembling
the open conformation most using the open structure of a different protein kinase
displaying only 23% sequence identity to C-PKA.