Volume 21, Number 1


  December 2004 Research Bulletin of the Supercomputing Institute

Multi-scale quantum models for RNA catalysis (pdf)

The study of biological phosphate chemistry is of key importance to understanding many cellular processes such as the hydrolysis of phosphates involved in transcription, cell signaling and respiration. Of particular interest is the study of the molecular mechanisms whereby RNA can catalyze fairly complicated reactions.  The unraveling of the details of the action of RNA enzymes, or ``ribozymes'', would allow deeper understanding of biological processes and provide valuable insight for the design of therapeutics that target viral or genetic disease and new biotechnology such as RNA chips or allosteric molecular switches in nanodevices.

Application of theoretical methods to RNA catalysis is extremely challenging. The polyanionic nature of RNA amplifies the importance of rigorous treatment of electrostatic interactions, polarization and other quantum many-body effects, interaction with solvent, monovalent and divalent ions. Moreover, RNA is considerably flexible, and consequently, simulations typically require long equilibration and sampling times relative to proteins. These factors combine to make the accurate theoretical study of RNA catalysis particularly challenging, and largely beyond the reach of many conventional techniques and models.

A promising approach is to use multi-scale quantum models to study the molecular mechanisms of phosphoryl transfer reactions, and in particular those involved in RNA catalysis (Fig. 1).  These models involve the integration of a hierarchy of theoretical levels that work together synchronously to provide detailed insight into complex biological processes that simultaneously span a broad range of spatial and temporal domains.  Professor Darrin York and his research group in the Department of Chemistry at the University of Minnesota have designed and applied new multi-scale quantum models to study the molecular mechanisms of phosphoryl transfer reactions in solution and in ribozymes.  The strategy that has been taken is to construct a large-scale database of high-level quantum calculations related to RNA catalysis.  This database is then used to design and parameterize accurate semiempirical quantum models that are 3-4 orders of magnitude faster that the high-level calculations, and that can be used in combined quantum mechanical/molecular mechanical (QM/MM) simulations or linear-scaline electronic structure calculations of biological macromolecules in solution.


Figure 1.  An example of a multi-scale quantum model: Set-up for a combined QM/MM activated dynamics simulation of hammerhead ribozyme.  An active dynamical region with particles propigated by Newtonian molecular dynamics (Active MD) is surrounded by a stochastic buffer that is propigated by Langevin molecular dynamics (SBMD).  The effects of the environment are modeled by a generalized solvent boundary potential (GSBP) that uses a variational electrostatic projection (VEP) method.  The structure is based on pioneering crystallographic work by Prof. Bill Scott at UC Santa Cruz.
The York Group has has constructed a large-scale density-functional quantum database of model phosphoryl transfer reactions (in the gas phase and with continuum solvent corrections).  The QCRNAworld's largest database of biochemical phosphoryl transfer reactions.  Recent studies resulting from the data contained in QCRNA have been published [ database has recently gone online and contains over 1,500 molecular structures and 200 chemical mechanisms, and represents the J. Am. Chem. Soc., 126, 1654 (2004); Chem. Phys. Chem., 5, 1045 (2004);  J. Biol. Inorg. Chem. 9, 807 (2004)].  Figure 2 shows one of the reaction mechanisms contained in the database: a 5'-ribose, 3'-phosphodiester model for RNA transesterification, a reaction catalyzed by the hammerhead ribozyme.
(Click on the image to view a 2.0MB MPEG movie)
Figure 2.  A 5'-ribose,3'-phosphodiester model for RNA transesterification and hydrolysis.  Movies of the reaction can be viewed as a 2.0MB MPEG file or 10MB animated GIF file.

The QCRNA database has been used to design new fast semiempirical quantum models that can be used in simulations of model RNA catalysis reactions in complex chemical environments.  Combined quantum mechanical/molecular mechanical (QM/MM) simulations have recently explored the nature of transphosphorylation thio effects (the change in reaction rate that occurs upon substitution of key phosphoryl oxygen positions with sulfur) to aid in mechanistic interpretation of experimental results [J. Am. Chem. Soc., 125, 7178 (2003); J. Am. Chem. Soc., 126, 7504 (2004)].  The free energy profiles for these simulations are illustrated in Figure 3.


Figure 3.  Free energy profiles for transphosphorylation thio effects in the 5'-ribose,3'-phosphodiester model shown in Fig. 2.  Taken from Gregersen et al., J. Am. Chem. Soc., 126, 7504 (2004).


The York Group has recently extended the QM/MM methodology to include a new smooth COSMOvariational electrostatic projection method solvation method for biological reactions, a [J. Phys. Chem. B, in press] for efficient modeling of the solvated macromolecular environment in activated dynamics simulations, and a new efficient linear-scaling Ewald technique [J. Chem. Theory Comput., in press] for long-range electrostatic interactions in collaboration with Prof. Jiali Gao.  These methods add to the arsenal of multi-cale quantum models used to study biological reactions in complex environments.

New-generation semiempirical quantum models derived from QCRNA are forthcoming, however preliminary quantum models have already emerged such as the AM1/d* method for phosphate hydrolysis reactions [Theor. Chem. Acc. 109, 149 (2003)] with Prof. Xabier Lopez, and the PM3BP method for nucleotide base pairing [J. Comput. Chem. 24, 57 (2003)] done in collaboration with the group of Prof. Christopher Cramer.  These models can be used with linear-scaling electronic structure methods, also developed by the York Group, to examine new quantum descriptors for entire solvated biological macromolecules up to tens of thousands of atoms [Proteins, 56, 724-737 (2004)].  This strategy has recently been applied to study the regioselectivity and RNA binding affinity of the HIV-1 nucleocapsid protein [J. Mol. Biol. 330, 993 (2003)] in collaboration with the group of Prof. Karin Musier-Forsyth. 


Figure 4. The local hardness map of HIV-1 nucleocapsid protein in the SL3-RNA bound conformation (based on PDB ID: 1A1T).   Taken from Khandogin et al., J. Mol. Biol. 330, 993 (2003).
Taken together, these developments, along with the QCRNA database represent major strides in the design of multi-scale quantum models to study the molecular mechanisms of RNA catalysis.  Professor York's research has been supported by the National Institutes of Health, the Donors of the Petroleum Research Fund administered by the American Chemical Society, the Army High-Performance Computing Research Center and the Minnesota Supercomputing Institute.