Effective fragment potentials and the enzyme active site

https://doi.org/10.1016/S0097-8485(99)00066-2Get rights and content

Abstract

Optimization of the binding conformation of a substrate in an enzyme active site using ab initio quantum chemistry methods are intractable since the active site comprises several hundred atoms. However, the active site can be decomposed into an active and spectator region where the spectator residues are represented by effective fragment potentials and reducing the number of all-electron atoms involved in the chemistry to a reasonable level. The effective fragment potentials for electrostatics and polarization are implemented in GAMESS but the repulsive and charge transfer potentials are fit to interaction energies of water with models of the residues. These repulsive/charge transfer potentials are generated for the protein residues and the EFP are then used to optimize binding of a transition state analogue to chorismate mutase (B. subtilis) and small dianions to ribonuclease A. For chorismate mutase the calculated binding conformation compares well to the comparable X-ray structure. The binding of the inhibitor to the glutamate/glutamine mutant active site is then predicted with the optimization including the glutamine residue constrained only at the Cα atom. The binding conformations suggest important roles for tyr108 and arg63, which have not been noted earlier. The electrostatic stabilization of the transition state by the active site charge distribution has to be augmented by a specific electronic activation by glu78. In ribonuclease A, the protons are found to move to provide a clustering of the charges to bind the small dianions, phosphate, thiophosphate, and sulfate.

Introduction

Structural and electronic understanding of the active site of an enzyme is a prelude to the evaluation of the catalytic reaction path. Binding of the substrate replaces the polar environment of water by the highly ionic environment of the active site. An accurate representation of the ionic hydrogen bonds requires ab initio quantum methods. Since a complete shell of residues around an active site can easily comprise several hundred atoms, all-electron methods are presently intractable. The ionic character of the active site implies that, in addition to significant electrostatic interactions, large polarization and charge transfer terms are to be expected. Effective fragment potentials (EFP) have been developed to handle solvation effects but are also appropriate for quantum calculations of the active site of a biomolecule. The active site of a large biomolecule is modeled by dividing it into an active region (A) described as an all-electron complex and a spectator region (S) that influences A through electrostatic and hydrogen-bonded interactions. All groups involved in the chemistry, which include any proton transfers, must be included in A. The groups that are included in the environment, S, can be probed by adding them to A and determining the effect on the chemistry. An iterative scheme can quickly select the essential residues that must be treated all-electron in reaction path calculations in the active site. EFP represent the protein electrostatic potential in the hamiltonian of the complete model active site. The present implementation of the EFP in GAMESS (Jensen et al., 1994, Day et al., 1996) represents non-bonded interactions between the A and S regions. The EFP replace the exchange repulsion, electrostatic potential, polarization, and a constrained charge transfer from the S region in the quantum hamiltonian of the total complex. Although some of the earliest applications of the EFP were to the calculation of reaction paths in protein (Krauss, 1995, Wladkowski et al., 1995, Krauss and Wladkowski, 1998, Kafafi and Krauss, 1999, Krauss et al., 1999), the main thrust of the applications have been to analyze the effect of water solvent on hydrogen bond interactions in water itself, formamide, and glutamic acid (Chen and Gordon, 1996, Day and Pachter, 1997, Krauss and Webb, 1997, Merrill and Gordon, 1998, Petersen and Gordon, 1999, Webb and Gordon, 1999). An accurate EFP for water has been developed and included in the GAMESS code but EFP for the protein residues are not generally available. In this note we will describe EFP for the amide backbone and the side-chains in the following residues: ala, val, leu, ser, thr, cys, arg, his, tyr, phe, glu, gln, asp, and asn. They will then be applied to binding of a transition state analogue to chorismate mutase and the small dianions, phosphate, thiophosphate, and sulfate, to ribonuclease A. More extensive studies have shown that binding, active site mutants, transition state optimizations, and spectroscopy in an EFP active site are tractable using these EFP. These will be summarized at the end of this note to indicate the general utility of the EFP in studying enzyme electronic structure, binding, reactivity, and spectroscopy at an enzyme active site.

Section snippets

Effective fragment potentials

The EFP are one-electron terms in the quantum hamiltonian so that the calculation time for the complex is only slightly larger than that required for the all-electron part. All the EFP are represented as linear expansions of gaussian functions. An accurate representation of the electrostatic potential is achieved using a distributed multi-polar analysis (DMA) of the spectator charge distribution (Stone, 1981, Stone and Alderton, 1985) multiplied by distance-dependent screening functions to

Chorismate mutase

The rearrangement of chorismate to prephenate catalyzed by chorismate mutase is a key step in the biosynthetic pathway that ultimately yields tyrosine and phenylanaline. This pathway is now being explored for bio-commercial production of chemicals. Even though the reaction is a Claisen rearrangement, the active site of all chorismate mutase enzymes is very ionic since the substrate is a dianion and positive charges, provided by at least three arginines, are present. The catalytic reaction

Ribonuclease A

Ribonuclease A catalyzes the hydrolysis of the phosphate ester bond in single strand RNA. Although there are many X-ray structures of this enzyme, the binding of small ligands is still uncertain. As very high resolution structures (<1 Å) of this enzyme become available, the proton on a ligand such as the phosphate anion may be observed with the direct measurement of the hydrogen bonds in the active site. The significant hydrogen bonding residues in the active site have been identified for very

Conclusion

The repulsive EFP for model systems of the protein backbone and various residues have been generated. A single water molecule probes these molecular models to determine the interaction energy. An RVS decomposition of the interaction energy determines the repulsive and charge transfer components. These potentials are assumed to work for interactions of these protein models for interactions with other molecules constructed of first shell atoms. The remaining EFP components, electrostatic and

Acknowledgements

This work is supported in part by the Advanced Technology Program of the National Institute of Standards and Technology.

References (33)

  • H Basch et al.

    J. Mol. Struct.

    (1995)
  • Y.M Chook et al.

    J. Mol. Biol.

    (1994)
  • M Krauss

    Comput. Chem.

    (1995)
  • W.J Stevens et al.

    Chem. Phys. Lett.

    (1987)
  • A.J Stone

    Chem. Phys. Lett.

    (1981)
  • P.A Bartlett et al.

    J. Am. Chem. Soc.

    (1985)
  • H.A Carlson et al.

    J. Am. Chem. Soc.

    (1996)
  • W Chen et al.

    J. Chem. Phys.

    (1996)
  • S.T Cload et al.

    J. Am. Chem. Soc.

    (1996)
  • P.N Day et al.

    J. Chem. Phys.

    (1996)
  • P.N Day et al.

    J. Chem. Phys.

    (1997)
  • M.M Davidson et al.

    J. Chem. Soc. Perkin Trans.

    (1996)
  • D.R Garmer et al.

    J. Phys. Chem.

    (1989)
  • Jensen, J.H., Day, P.N., Gordon, M.S., Basch, H., Cohen, D., Garmer, D.R., Krauss, M., Stevens, W.J., 1994. In: Smith,...
  • Kafafi, S., Krauss, M., 1999. Int. J. Quant. Chem., in...
  • P Kast et al.

    J. Am. Chem. Soc.

    (1996)
  • Cited by (16)

    • Catalytic mechanism of human glyoxalase i studied by quantum-mechanical cluster calculations

      2016, Journal of Molecular Catalysis B: Enzymatic
      Citation Excerpt :

      In 2001, three groups independently proposed two different mechanisms for the catalytic reaction of GlxI [5–7]. Richter and Krauss (RK) used HF/4–31G calculations of the active site, coupled with a frozen effective fragment potential description [8,9] of eleven residues in the binding site and proposed a three-step mechanism for the reaction of the S enantiomer of the substrate [5]. Himo and Siegbahn (HS) used density functional theory (DFT) and proposed a five-step mechanism for the S substrate [6] (the step counts exclude the binding of the substrate and the dissociation of the product).

    View all citing articles on Scopus

    Workshop on New Trends in Computational Methods for Large Molecular Systems, Szklarska Poreba, Poland, 1–6 July 1999, Edited by W. Andrzej Sokalski and Morris Krauss.

    View full text