Binding Affinities for a Series of Selective Inhibitors of Gelatinase-A Using Molecular Dynamics with a Linear Interaction Energy Approach

Abstract

The binding of a series of hydroxamate inhibitors with gelatinase-A is examined to evaluate the viability of calculating free energies of binding, ∆G b , utilizing molecular dynamics (MD) simulations with a linear interaction energy approach. In our simulations, a bonded model was used to represent the potentials of the catalytic zinc center. The electrostatic distribution of this model was derived using a two-stage electrostatic potential fitting calculations. The resulting bonded model was then used to generate the MD trajectories. Coulombic, van der Waals, and coordinate bond energy components determined from MD simulations of the bound and unbound inhibitors solvated in water were correlated with the free energies of binding for the 15 hydroxamate inhibitors. In the correlation process, several linear models consisted of different energy components were tested. We found that besides the usually used Coulombic and van der Waals energy terms, the introduction of a constant term could significantly improve the correlation. The best model yields an average error of 0.6 kcal/mol for the 15 binding affinities, which cover an observed range of 7.2 kcal/mol. The predictive ability of the best model was revealed by the high value of q 2 (0.854) from the leave-one-out cross-validation. To this series of inhibitors, the constant term can be treated as effective adjustment to the entropy contribution in the binding free energies. The MD simulations predicted the binding mode of the gelatinase-A with the studied inhibitors, and also provided insights into the interactions occurring in the active site and the origins of variations in ∆G b . The P 1 ′ groups of inhibitors make extensive van der Waals and hydrophobic contacts with the nonpolar side chains of four residues in the S1′ subsite, including Leu 197, Val 198, Leu 218, and Tyr 223, which directly influence the ligand binding. Hydrogen bonds between hydroxamates and gelatinase-A are very important to stabilize the inhibitors in the active site. The hydrogen bonds between the P 3 ′ group and gelatinase-A can produce more favorable electrostatic interactions.

Predictions of Binding of a Diverse Set of Ligands to Gelatinase-A by a Combination of Molecular Dynamics and Continuum Solvent Models

“…In the current work, the potentials of the catalytic zinc center were represented by a nonbonded model, whereas in our previous work, the bonded model was used. 55 Using those two different potential models, we found that the structures of the catalytic zinc bear some differences, but the obtained LIE models are of similar statistical significance. The bonded or nonbonded models may not be completely correct.…”

“…In our previous work, when adopting the bonded zinc model, the coordinate form of the zinc atom was well-preserved. 55 In the current work, when adopting the nonbonded model, the catalytic zinc is also pentacoordinated, but the coordinate form is different from that in the initial model. Nevertheless, the coordinate form from our MD simulations should be further validated by experiments.…”

“…41 Second, in previous work on the prediction of the free energy of bindings between 15 hydroxamates and MMP-2, a bonded zinc center was used. 55 Therefore, in this paper, the use of the nonboned model can be directly compared with the use of the bonded model.…”

“…The new fit resulted in substantial improvement (r 2 ) 0.88, q 2 ) 0.57). In previous work on the binding free-energy calculations of 15 hydroxamate inhibitors with gelatinase-A, 55 several linear models consisting of different energy components were tested. We found that besides the Coulombic and van der Waals energy terms that are normally used the introduction of a constant term could significantly improve the correlation.…”

Simulation‐Based Predictions of Binding Affinities of Matrix Metalloproteinase Inhibitors

“…The timeaveraged geometries were used to calculate pertinent ensemble averages after 5, 10, 25, 50, 100, and 200 ps. The respective conformations for the most active ligand (20 , Table 1) after different simulation times are illustrated in Figure 2. In addition to using SASA in Eq.…”