A sufficiently stable noncovalent association complex
between a
covalent inhibitor and its protein target is regarded as a prerequisite
for the formation of a covalent complex. As this transient form can
hardly be assessed experimentally, computational modeling is required
to probe the suitability of a given ligand at this particular stage.
To investigate which criteria should be fulfilled by suitable candidates
in a molecular dynamics (MD) assessment, a systematic study was conducted
with 20 complexes of cathepsin K, a papain-like cysteine protease
of pharmaceutical relevance. The covalent inhibitors in these complexes
were converted to their pre-reaction states, and the resulting noncovalent
complexes were subjected to MD simulations. The critical distance
between the electrophilic and nucleophilic reaction partners was monitored
as a potential parameter to assess the suitability for covalent bond
formation. Across various warhead types, a distance between 3.6 and
4.0 Å, comparable to the sum of the generalized Born radii of
carbon and sulfur, was found to be stably maintained under appropriate
conditions. The protonation state of the catalytic dyad and the resulting
solvation pattern dramatically affected the noncovalent binding mode
and the distance of the warhead to the active site. For several complexes,
fluctuations in the orientation of the warhead were observed due to
torsional rotations in adjacent bonds. This observation helped to
explain the gradual transitions from noncovalent to covalent complexes
observed in the crystal structures of three closely related nitrile-based
inhibitors. According to comparative simulations conducted for a set
of 13 cathepsin S complexes, the overall findings of the study appear
to be transferable to related cysteine proteases as targets of covalent
inhibitors.