Biomolecular Structure, Dynamics and Function


Rationale. It is currently not possible to unravel the mechanism of enzyme regulation, catalysis and specificity at a detailed atomic level using any one single technique. Thus, our efforts in this area are designed to (i) assess the accuracy of molecular dynamics (MD) simulations versus that of experimental techniques for characterizing protein structure and dynamics; (ii) supplement existing experimental information, and aid in the mechanistic interpretation of experimental results; and (iii) correlate the structural and dynamical features of certain residues and regions with the biological function of the protein in question.

Specific Aims. Our preliminary goals were to

  1. assess the accuracy of MD simulations vs. NMR and X-ray spectroscopy for characterizing protein structure in solution;
  2. assess the accuracy and precision of MD simulations vs. NMR relaxation experiments for characterizing protein backbone dynamics; and
  3. supplement existing NMR information on the interatomic interactions at the phosphotyrosine (pTyr) binding region of an SH2-domain-phosphopeptide complex and aid in the interpretation of chemical shift information.

Approach. To achieve each of the above specific aims in turn,

  1. a range of NMR data (inter-proton NOE's, dihedral and hydrogen-bond restraints, dihedral angle order parameters), pertaining to the solution structure of E. Coli Ribonuclease H (RNase HI) represented by an ensemble of 15 structures, were correlated with two X-ray structures and a 2-nanosecond MD simulation of the protein in water (Philippopoulos & Lim, 1995; 1998; 1999);
  2. 3 independent data sets of backbone amide generalized order parameters originating from 15N NMR relaxation experiments, and 2 independent data sets derived from simulations of RNase HI were compared (Philippopoulos et al., 1997);
  3. 3 MD simulations of the free, phosphate-ion-bound and phosphopeptide-bound C-terminal SH2 domain of phospholipase C-gamma1 (PLCC .pY) were performed.

Key Results and Their Significance

  1. Our analysis revealed the capacity of the simulation to bring about structural transition from the crystal to the solution state (Philippopoulos & Lim, 1999). The agreement of the conformations between the NMR and MD ensembles for a number of dihedral angles indicates the existence of "hidden" picosecond-timescale dynamic information in the family of NMR structures. This strengthens the validity of the corresponding conformations sampled during the simulation. This "harvesting" of explicit dynamic information in an NMR ensemble of structures, coupled with NMR relaxation data, is an attractive alternative to the exclusive use of NMR relaxation experiments for the characterization of protein internal mobility. NMR relaxation techniques alone provide no insight into the actual protein conformations sampled and their populations, which are crucial in linking protein mobility and biological function. Our comparison of the MD simulation of RNase HI with the NMR ensemble and the crystal structures have suggested that it is possible to reliably distinguish the presence of dynamic processes from sheer scarcity of NMR constraints as the two main causes of conformationally variable regions of NMR structures. This result opens the way for the incorporation of (mainly side-chain) mobility into current protein-docking and structure-prediction approaches on the basis of well-refined and unbiased NMR ensembles of structures.
  2. The data set comparisons of RNase HI suggested that the NMR and MD backbone N-H order parameters are of comparable accuracy for residues exhibiting motions on a fast time scale (<100 ps) (Philippopoulos et al., 1997). In addition, MD order parameters for motions on a fast (<100 ps) timescale were overall found to be more precisely determined than their NMR counterparts, thereby permitting the more detailed dynamic characterization of some biologically important residues by the MD simulation than was possible with the NMR relaxation experiments.
  3. The average structure of the PLCCc.pY complex, determined by an MD simulation starting from the average NMR structure, revealed a more complete picture of inter-atomic interactions in the pTyr-binding pocket than was possible with chemical shift and NOE data alone. This allowed the more unequivocal identification of the key pTyr-recognition residues (Feng et al., 1996). The simulations of the PLCC SH2 domain in its three forms offered several alternative interpretations of the chemical shift data to those suggested in the analogous NMR investigation. This insight is valuable, as the experimentally observed chemical shifts could have resulted from a number of possible pictures of interactions. Our study showed that the combination of MD simulations and ab initio chemical shift calculations can enhance the hydrogen-bonding, amino-aromatic and aliphatic-aromatic information content of NOE- and chemical-shift-based protein structures and serve as a complementary tool for interpreting chemical shift data at the atomic level.

Publications

Feng, M.-H., Philippopoulos, M., A. D. MacKerell Jr. & Lim, C. (1996). Structural Characterization of the Phosphotyrosine Binding Region of A High Affinity SH2 Domain-Phosphopeptide Complex by Molecular Dynamics Simulation and Chemical Shift Calculations. J. Am. Chem. Soc. 118:11265-11277.

Philippopoulos, M., Xiang, Y.-F., & Lim, C. (1995).Identifying the Mechanism of Protein Loop Closure: A Molecular Dynamics Simulation of the Bacillus Stearothermophilus LDH Loop in Solution. Protein Engineering 8:563-571.

Philippopoulos, M. & Lim, C. (1995). Molecular Dynamics Simulation of E. coli Ribonuclease HI in Solution. Correlation with NMR and X-Ray Data and Insights into Biological Function. J. Mol. Biol. 254:771-792.

Philippopoulos, M., Mandel, A. M., Palmer III, A. G. & Lim, C. (1997). Accuracy and Precision of NMR Relaxation Experiments and Molecular Dynamics Simulations for Characterizing Protein Dynamics. Proteins: Struct. Funct. Genetics 28:481-493.

Philippopoulos, M. & Lim, C. (1998).Protein Dynamics: Correlation Between MD Simulation, NMR Spectroscopy and X-ray Crystallography. Advances in Computational Life Sciences: Human to Proteins. M. Michalewicz (Editor), CSIRO Mathematical and Information Sciences 2:167-182.

Philippopoulos, M. & Lim, C. (1999). Exploring the Dynamic Information Content of Protein NMR Structures: Comparison of a MD Simulation with the NMR and X-Ray Structures of E. Coli RNase HI. Proteins: Struct. Func. Genetics 36:87-110.

BACK     HOME