NMR spectroscopy and computational analysis of interaction between Serratia marcescens chitinase B and a dipeptide derived from natural-product cyclopentapeptide chitinase inhibitor argifin

Abstract

The dipeptide N-acetyl-Arg{N(omega)-(N-methylcarbamoyl)}-N-methyl-Phe(2), which is a part of the natural-product cyclopentapeptide chitinase inhibitor argifin (1), inhibits chitinase B from Serratia marcescens (SmChiB) with a half-maximal inhibitory concentration (IC(50)) of 3.7 microM. Despite the relatively small size of 2, its inhibitory activity is comparable with that of 1 (IC(50)=6.4 microM). To elucidate the basis for this interesting phenomenon, we investigated the interaction between 2 and SmChiB using a combination of nuclear magnetic resonance spectroscopy and computational methods. The transferred nuclear Overhauser effect (TRNOE) experiment obtained structural information on the SmChiB-bound conformation of 2. The binding mode of 2 and SmChiB was modeled by the novel molecular-docking approach proposed in our laboratory, which can explicitly consider water-mediated hydrogen-bonding interactions in protein-ligand interfaces. The SmChiB-bound conformation of 2 in the resulting model satisfied all proton-proton distance constraints derived from the TRNOE experiment, indicating that our model structure of the 2-SmChiB complex is reasonable. A molecular dynamics (MD) simulation examined the stability of the resultant complex structure and suggested that 2 binds to SmChiB in a similar fashion to the binding mode observed for N(omega)-(N-methylcarbamoyl)-Arg(1) and N-methyl-Phe(2) of 1 in the crystal structure of the argifin-SmChiB complex. Finally, the binding free energies of 1 and 2 with SmChiB were estimated by the molecular mechanics Poisson-Boltzmann surface area (MM-PBSA) method using the MD trajectory. The MM-PBSA calculation suggested that both 1 and 2 bind to SmChiB with similar affinities, which is consistent with their experimental IC(50) values. Energetic analysis revealed that the van der Waals interaction of 2 with SmChiB is much less than that of 1, but is completely compensated by the more favorable contribution of solute entropy and the total electrostatic component. The improved total electrostatic component was derived from more favorable electrostatic interactions. Therefore, we conclude that dipeptide 2 was also better optimized against SmChiB than 1 in an electrostatic point of view.