Johannes Kirchmair studied pharmacy at the Leopold-Franzens University of Innsbruck, Austria, from 1999 to 2004 and received his Ph.D. degree under the guidance of Professor Thierry Langer in 2007. For his Ph.D., he specialized in structure-based virtual high-throughput screening and parallel screening techniques. His career began at Inte:Ligand GmbH in Vienna, Austria, working on computer-guided drug development using 3D virtual high-throughput screening techniques for the identification and optimization of novel anticancer agents. His research interests cover medicinal chemistry, computational chemistry, and drug design as well as QSAR and 3D-QSAR molecular modeling techniques. Since April 2008, Dr. Kirchmair has been a researcher at the University of Innsbruck and teaches computational chemistry.
Gerhard Wolber received his Ph.D. in pharmaceutical chemistry at the University of Innsbruck in 2003 after his studies of computer science and pharmacy at the University of Innsbruck and University of Vienna, Austria, respectively. As one of the founders of the drug design company Inte:Ligand, he has been working as head of cheminformatics and research since 2003, where he has been developing the two programs ilib diverse and LigandScout. In 2008, he took a position as a lecturer in pharmaceutical chemistry at the Institute of Pharmacy at the University of Innsbruck, where he now heads his own research group and teaches computational and medicinal chemistry. His research interests include structure- and ligand-based drug design, efficient algorithms for virtual screening, 2D and 3D visualization techniques, and 3D pharmacophore modeling.
Information about the Computer-aided Molecular Design group at the University of Innsbruck is available at www.uibk.ac.at/pharmazie/
Johannes Kirchmair, Ph.D., and Gerhard Wolber, Ph.D.
Drs. Kirchmair and Wolber are coauthors of the recently published The Protein Data Bank (PDB), Its Related Services and Software Tools as Key Components for In Silico Guided Drug Discovery.
Johannes Kirchmair, Patrick Markt, Simona Distinto, Daniela Schuster, Gudrun M. Spitzer, Klaus R. Liedl, Thierry Langer and Gerhard Wolber (2008) J. Med. Chem. 51: 7021–7040
University of Innsbruck
Q: You recently published an extensive paper describing the PDB archive, its history, and related resources. What surprised you when you were preparing this manuscript?
A: We were overwhelmed by the number and diversity of tools provided by the PDB portals and related websites. Before this work, we regularly used only a small part of PDB-related tools, simply because we were accustomed to them. After beginning to thoroughly investigate services and software available for structure-based PDB-related drug development, we immediately started using many of these tools and applications for research as well as for teaching. Another positively surprising fact was that many of these approaches take care of small organic ligands, while providing a high level of cross-linking; i.e., it becomes possible to solve a specific problem by jumping from one service to another one without losing intermediate results. The best thing is that most of these services are free for non-commercial use despite their high quality. Naturally, this is great for teaching students, since we have access to a similar level of information as an industrial environment. The PDB bridges the two worlds of biology (macromolecules) and of medicinal chemistry (small molecules); it also provides a large quantity of easy-to-use tools for scientists that may not have been too much involved in computational chemistry or modeling so far. We see a strong trend for chemists to use PDB data to derive new ideas for synthesis and SAR within a short time without the need for installing any software; everything’s on the web–free for academics!
Q: What do you think your online category of PDB-related tools (www.uibk.ac.at/pharmazie/phchem
/camd/pdbtools.html) will look like in 10 years?
A: Looking at the development in the past few years, we hope–and are confident–that ligand chemistry will become more important to the PDB. A tighter integration with initiatives like PubChem certainly bears great potential, such as being able to correlate ligand similarity with binding pocket similarity, which could lead to integrating virtual screening tools into the web interface of the PDB. 3D pharmacophores could be a good way to formulate the interaction of a ligand with its surrounding protein. Another possibility is that more software could be developed to further analyze the binding site. Protein-ligand-docking is also an interesting but currently controversial topic. If docking were to be regarded less commercially, eventually the PDB could offer a freely parameterizable docking toolbox that could help solving the scoring problem by large-scale statistics. We also hope that there will be more membrane proteins crystallized in the next 10 years, which would trigger the creation of a plethora of new tools that deal with membrane-drug interactions and homology modeling.
Q: How do you use the PDB when training pharmacy students?
A: The RCSB PDB is an invaluable resource for teaching: the web application has improved so much in the past few years that many aspects of computational chemistry teaching can be directly covered using the standard RCSB PDB interface. Visualization of the proteins and binding pockets are only one; the ability to perform sequence similarity searches and the EC-classification to identify similar proteins with and without bound ligands are others. We also use the PDB as input to our own tools, such as the 3D pharmacophore generator LigandScout for developing structure-based 3D pharmacophores. For teaching medicinal chemistry, the clarity of the RCSB PDB web interface allows for demonstrating essential structure-activity relationships (e.g., Which geometric chemical features are essential for ligand binding? If there is a reactive group on the ligand, why is an irreversible inhibitor bad? The PDB structure complex can show that the ligand is covalently bound to its co-factor).
Q: What are some challenges facing structure-based drug discovery today?
A: The ligand affinity problem: protein-ligand docking has frequently addressed, but never solved this challenge. The practical approach that most scientists choose is to define rule-based scoring functions for their problems, and for that the PDB can help to better understand a problem by providing experimental data. However, there are still several issues with how the PDB stores small organic molecules. In some cases, it is still impossible to get the correct chemistry of a ligand from the PDB in an automated way. Sometimes, crystallographers do not pay much attention to the ligand or crystal waters and ions. Hence, it could be useful to store initial, un-refined electron densities without model bias only for the ligand to allow for re-interpretation. Other challenges are the lack of crystal structures for important protein classes, such as membrane proteins, and protein flexibility, especially conformational flexibility at the binding site which could be analyzed by multiple X-ray structures of one and the same target interaction site with multiple ligands.
Q: What are some of the new exciting opportunities in drug discovery? What role would the PDB play in these?
A: The large collection of useful tools shows that the PDB provides extremely useful data for drug discovery–also for regarding small molecules, which probably has never been the primary focus of the PDB. Ligand Expo shows that ligands are becoming important, and we see a huge potential in paying more attention to ligand chemistry. Getting correct ligands directly from the PDB bears the potential of providing a lot of new cross-linking applications. Structure-based parallel screening and polypharmacology approaches are exciting topics that seem to be tailored for a database like the PDB.