posted on 2017-01-12, 20:18authored byDonatella Callegari, Alessio Lodola, Daniele Pala, Silvia Rivara, Marco Mor, Andrea Rizzi, Anna Maria Capelli
The duration of drug efficacy in
vivo is a key aspect primarily
addressed during the lead optimization phase of drug discovery. Hence,
the availability of robust computational approaches that can predict
the residence time of a compound at its target would accelerate candidate
selection. Nowadays the theoretical prediction of this parameter is
still very challenging. Starting from methods reported in the literature,
we set up and validated a new metadynamics (META-D)-based protocol
that was used to rank the experimental residence times of 10 arylpyrazole
cyclin-dependent kinase 8 (CDK8) inhibitors for which target-bound
X-ray structures are available. The application of reported methods
based on the detection of the escape from the first free energy well
gave a poor correlation with the experimental values. Our protocol
evaluates the energetics of the whole unbinding process, accounting
for multiple intermediates and transition states. Using seven collective
variables (CVs) encoding both roto-translational and conformational
motions of the ligand, a history-dependent biasing potential is deposited
as a sum of constant-height Gaussian functions until the ligand reaches
an unbound state. The time required to achieve this state is proportional
to the integral of the deposited potential over the CV hyperspace.
Average values of this time, for replicated META-D simulations, provided
an accurate classification of CDK8 inhibitors spanning short, medium,
and long residence times.