The assignment of protonation state and accurate calculation of local pKa in macromolecular structures can be an important factor in understanding and simulating biological systems. The assignment of protonation states is a difficult computational problem because of uncertainties related to conformation, solvent salts and other interactions.
On 15 October 2008 we will hold an eCheminfo Community of Practice conference session at Bryn Mawr College to be chaired by Paul Labute (President, Chemical Computing Group), to present recent methods related to macromolecular pKa prediction as well as techniques and issues related to methods validation.
A description of the session with presentation abstracts follows:
Accurate Calculation of pKas
http://echeminfo.com/COMTY_confprog08pkas
(Please follow continuation here to read abstracts. Comments can be made at the end.)
Abstracts
Modeling Ionization States and Proton Uptake/Release in Receptor-Ligand Association
Emil Alexov, Computational Biophysics and Bioinformatics, Department of Physics, Clemson University, Clemson, SC 29634, USA
Ionized groups carry net charge and thus play a major role in the electrostatic interactions between the ligand and receptor. However, their ionization states depend on such factors as the pH of the water phase, the interactions with other charges, water molecules and mobile ions. Therefore the charge states of the titratable groups have to be predicted prior to applying docking protocols. With the progress made in structural modeling, many structures are expected to be models and therefore the method of computing pKa’s should be able to tolerate structural imperfections while retaining high accuracy predictions. In addition, the formation of the receptor-ligand complex could dramatically change the electrostatic environment of the ionizable groups and cause proton uptake/release. Accounting for such phenomena can be critical for obtaining correct docking solutions. Here we report our recent investigation of applying Multi-Conformation Continuum Electrostatics (MCCE) method to calculate pKa’s of ionizable groups of more than 500 3D structural models of nucleoside mono-phosphate kinases. The proton uptake/release is modeled using 2,887 protein-protein complexes extracted at 40% sequence identity from the ProtCom database. It was shown that in the vast majority of the cases, the formation of a complex induces charge transfer from the water to the protein, resulting in pH dependence of the binding energy. Detailed analysis was performed in case of pepstatin binding to three aspartic proteases and it was demonstrated that ionization changes caused by the complexation could be distributed over the entire structures and are not located only within the binding epitope.
Benchmarking Protein pKa Predictions
Matthew Davies, Jenner Institute, UK
pKa values are a measure of the protonation of ionizable groups in proteins. Ionizable groups are involved in intra-protein, protein-solvent and protein-ligand interactions as well as solubility, protein folding and catalytic activity. The pKa shift of a group from its intrinsic value is determined by the perturbation of the residue by the environment and can be calculated from three-dimensional structural data.
The Protein pK(a) Database (PPD) v1.0 provides a compendium of protein residue-specific ionization equilibria (pK(a) values), as collated from the primary literature, in the form of a web-accessible postgreSQL relational database. Ionizable residues play key roles in the molecular mechanisms that underlie many biological phenomena, including protein folding and enzyme catalysis. The PPD serves as a general protein pK(a) archive and as a source of data that allows for the development and improvement of pK(a) prediction systems. The database is accessed through an HTML interface, which offers two fast, efficient search methods: an amino acid-based query and a Basic Local Alignment Search Tool search. Entries also give details of experimental techniques and links to other key databases, such as the National Center for Biotechnology Information and the Protein Data Bank, providing the user with considerable background information. The database can be found at the following URL: http://www.jenner.ac.uk/PPD.
We also used the PPD to analyse the performance of different pKa prediction techniques. Our work provides a benchmark of available software implementations: MCCE, MEAD, PROPKA and UHBD. Combinatorial and regression analysis was used in an attempt to find a consensus approach towards pKa prediction.
Improved pKA Prediction: Combining Empirical and Semi-microscopic Methods
Gernot Kieseritzky and Ernst-Walter Knapp, Freie Universitaet Berlin, Institut fuer Biologie/Chemie/Pharmazie, Fabeckstr. 36a, 14195 Berlin, Germany
Using three different methods we tried to compute 171 experimentally known pKA values of ionizable residues from 15 different proteins and compared the accuracies of computed pKA values in terms of the root mean square deviation from experiment. Our own semi-microscopic method (1) is based on a continuum electrostatic model of the protein including conformational flexibility (Karlsberg+) where pKA values are computed by electrostatic energy computations using a small number of optimized protein conformations derived from crystal structures. The others are empirical approaches with PROPKA (2) deploying a physically motivated energy terms with adjustable parameters and PKAcal (3) using a meaningless empirical function. PROPKA reproduced the pKA values with highest overall accuracy. Differentiating the data set into weakly and strongly shifted experimental pKA values, however, we found that PROPKA’s accuracy is better if the pKA values are weakly shifted but is on equal footing with that of KBPLUS for more strongly shifted values. On the other hand, PKAcal reproduces strongly shifted pKA values badly but weakly shifted values with the same accuracy as PROPKA. We tested different consensus approaches combining data from all three methods to find a general procedure for most accurate pKA predictions. In most of the cases we found that the consensus approach reproduced experimental data with better accuracy than any of the individual methods alone.
References
[1] G. Kieseritzky, E.W. Knapp, Proteins: Structure, Function, and Bioinformatics. 2008, 71 (3), 1335-1348.
[2] H. Li, A. D. Robertson and J. H. Jensen, Proteins: Structure, Function, and Bioinformatics. 2005, 61 (4), 704-721.
[3] Y. He, J. Xu and X. M. Pan, Proteins: Structure, Function, and Bioinformatics. 2007, 69 (1), 75-82.
Calculations of pH-dependent Binding of Peptides to Biological Membranes
Maja Mihajlovic and Themis Lazaridis, Department of Chemistry, City College of New York/ CUNY, New York, NY 10031, USA
Biological processes, such as cell signaling, immune response and membrane trafficking, often involve pH-dependent binding of proteins to membranes. We recently developed a computational method for predicting absolute, pH-dependent membrane binding free energies. The binding free energy is calculated as a sum of a) the free energy cost of ionization state changes, b) the effective energy of transfer from solvent to the membrane, c) the translational/rotational entropy cost of binding, and d) an ideal entropy term which depends on the relative volume of the bound and free state and hence depends on lipid concentration. pKa values of acidic residues in solvent and on the membrane are determined using a combinatorial method. All energies required by the method are obtained from molecular dynamics trajectories on an implicit membrane (IMM1-GC). The method is tested on the helical peptide VEEKS, derived from the membrane-binding domain of phosphocholine cytidylyltransferase.
Improving the Accuracy of Calculated Protein pKa Values using NMR Spectroscopy
Jens Erik Nielsen, University College Dublin, Ireland
The ability to calculate protein residue pKa values is of significant use for understanding, manipulating and inhibiting enzymatic activity and protein-ligand interactions. Current pKa calculation methods are typically accurate to within 0.75 pKa unit when comparing to experimentally measured pKa values, and there is thus much room for improvement in the performance of protein pKa calculation methods.
pKa values are typically calculated by combining three types of energies: desolvation energies, background interaction energies and charge-charge interaction energies. Recently it has been shown that NMR can be used to measure charge-charge interaction energies in proteins, and it is therefore now possible to validate a single energy component of calculated pKa values. This prospect should lead to a dramatic improvement in our understanding of electrostatic forces in proteins, and result in improved pKa calculation methods.
It is well-known that the loss of a proton at a titratable site can be monitored in the NMR chemical shift for many nuclei in a protein. If one is able to identify the titratable group responsible for such a ‘ghost titration’ in the chemical shift of the backbone amide atoms, then it is possible to calculate the strength of the electric field due to the titratable group at the amide nucleus in question. However, since most acidic residues in a protein titrate around pH 4, it is often quite hard to determine which titratable group is responsible for a specific ghost titration. We carried out NMR titration experiments for the wild type and 9 mutants of Hen Egg White Lysozyme (HEWL) to ascertain the source of all acidic ghost titrations in the molecule. The results give us detailed information on how efficiently the electric field propagates through a protein structure, and allows us to construct a 3-dimensional map of the dielectric of the HEWL structure.
Theoretical analysis and analyses of 3-dimensional dielectric maps give information on the determinants of the permittivity of proteins, and allows us to construct more accurate protein pKa calculation methods. Future developments include the use of 3D dielectric maps in drug docking software and molecular dynamics simulations to give a more accurate description of the electric energies in proteins.
Barry Hardy
eCheminfo Community of Practice
Structural Biology eCheminfo cheminformatics chemoinformatics bioinformatics Medicinal Chemistry Computational Chemistry Virtual Screening pKa Protonation Molecular Modelling Molecular Modeling pharmaceutical pharma meeting workshop training Oxford Critical Path toxicology Bursary Life Sciences Pharma Drug Discovery Research and Development Drug Development Healthcare Innovation Knowledge Management events
Comments