Superfund Research Program
Using Computational Approaches to Investigate Ligand-Receptor Interactions
A ligand is simply a molecule that binds specifically to a protein - yet ligand-receptor interactions are critical in virtually all biological processes. Ligand-receptor binding is involved in cell functions including gene regulation, transport across membranes, immune responses and enzymes catalysis. Two key features of ligand-receptor interactions are that they are highly specific and that the interaction brings about an event, or a cascade of events, which results in an observable response. An increased understanding of the molecular nature of ligand-receptor interactions and the series of events that they set in motion could provide valuable information on the mechanisms involved in normal physiological processes and responses to environmental contaminants.
Dr. Sandor Vajda leads the Bioinformatics and Molecular Modeling Core at the Boston University Superfund Basic Research Program. His team supports research projects investigating environmental pollutants that bind to specific protein receptors or enzymes, affecting pathways involved in reproduction and development. Dr. Vajda's lab provides hardware and software tools, and expertise in bioinformatics, molecular modeling and visualization, prediction of protein structures by homology modeling, and docking and the analysis of receptor-ligand interactions.
Dr. Vajda's team has developed a number of computational tools that are very useful to the investigators in the BU SBRP:
- Computational solvent mapping - determines binding sites computationally rather than experimentally. The method places molecular probes (small molecules or functional groups) on a protein surface in order to find the most favorable binding positions. Since the small molecules prefer binding in the active site, computational mapping can be used for the identification and characterization of functional sites of the enzyme. The method has been applied to cytochrome P450s, and revealed interesting differences between bacterial and mammalian molecules. In bacterial P450s, the mapping shows well-defined binding sites for both bound and unbound structures, with very little difference between the two. In contrast, mammalian P450s exhibit a largely open binding channel and a well-defined pocket is formed only due to substrate binding. This plasticity of the binding site explains why mammalian P450s can oxidize a structurally very diverse range of substrates. The mapping also shows that the binding of the first ligand in mammals leaves a large fraction of the binding channel unoccupied, resulting in a pocket just above the heme iron that can bind a second ligand molecule with higher affinity, thereby demonstrating the importance of P450s in drug-drug interactions.
- Docking to nuclear receptors - discrimination of agonists from partial agonists by performing calculations for a panel of different structures. Although x-ray structures are available for many nuclear receptors, docking of small molecules to these proteins is far from easy, because the binding of an agonist, partial agonist, or antagonist leads to specific and different conformational changes that affect cofactor binding and gene activation. Due to such conformational changes, one cannot dock a strong agonist to a protein structure that has been crystallized without a ligand or in complex with a partial agonist. Similarly, a partial agonist will not fit well into the agonist-bound form of a nuclear protein. This difficulty of cross-docking was used to classify potential ligands by docking them to a panel of receptor structures that have been generated from structures co-crystallized either with an agonist, partial agonist, or antagonist. The type of receptor structure for which a given ligand yields the lowest free energy determines if the compound is an agonist, partial agonist, antagonist, or a non-binder.
- Consensus - an algorithm that consistently provides a high quality alignment of the core regions for comparative modeling. The method is currently freely available to the scientific community as a server at http://structure.bu.edu.
The Bioinformatics and Molecular Modeling Core provides direct support to several projects at the BU SBRP, providing data on xenobiotic ligands such as dioxins, PCBs, phthalates in bacteria, fish, mouse and human models.
In support of Dr. John Stegeman's research, which is focused to determine the function and regulation of CYP P450 as potentially associated with developmental effects of chemicals in fish, Dr. Vajda's team used available sequence and structural information on P450s to predict the enzyme structure. They employed homology modeling, using a number of known structures of related sequences to predict the enzyme shape, then applied robust docking methods, computationally simulating the fit of a ligand into the active site of the predicted enzyme structure. They found that the ligand 3,3',4,4'-tetrachlorobiphenyl (TCB) binds significantly closer to the heme group in human CYP1A1s than in fish, a result which is good agreement with experimental evidence. This information suggests that different species may exhibit differences in oxidative metabolism by P450s and that non-human models may not be appropriate surrogates to assess the risks that particular compounds pose to human health.
In collaboration with Dr. David Waxman, Dr. Vajda investigated the binding of phthalates in peroxisome proliferators activated receptors (PPARs).
Phthalates are industrial plasticizers, known to interact with PPARs, and are established reproductive toxicants. The researchers first docked 15 phthalates with known values of PPARg activation, and showed that the calculated binding free energies are in good agreement with experimental data. The same docking and scoring calculations have been then applied to 70 phthalate monoesters with no activation data. Some of the compounds with the lowest calculated binding free energies were known to be weak activators of PPARg, and some others will be evaluated experimentally for PPARg activation.
Dr. Vajda's work is a good example of interdisciplinary collaboration, with new and innovative engineering techniques shedding light on biology, and the biology acting as a driver for technical development. Such combinations of experimental and computational methods to help identify ligands of proteins and enzymes will facilitate the integration of computational approaches into more traditional xenochemical toxicity screening programs. These studies will help identify environmental chemicals that warrant closer scrutiny as potential ligands for human receptors.
For More Information Contact:
To learn more about this research, please refer to the following sources:
- Kortvelyesi T, Dennis S, Silberstein S, Brown L, Vajda S. 2003. Algorithms for computational solvent mapping of proteins. Proteins: Structure, Function, and Genetics 51(3):340-351. PMID:12696046
- Dennis S, Kortvelyesi T, Vajda S. 2002. Computational mapping identifies the binding sites of organic solvents on proteins. Proc Natl Acad Sci U S A 99(7):4290-4295. PMID:11904374
To receive monthly mailings of the Research Briefs, send your email address to firstname.lastname@example.org.