Evolution or Revolution: The Challenge to Today's Medicinal Chemist
Steven V. Ley
Department of Chemistry, University of Cambridge, Cambridge, United Kingdom
With the ever increasing demand for new compounds, synthetic chemists have been expected to accelerate greatly their rate of production of new chemical entities.
The preparation of biologically active and many other functional material from small, commercially available building blocks inevitably involves more that one synthetic step. For most modern drugs and other complex molecules, it is not uncommon to require at least 10 steps and sometimes many more.
In order to address these goals we believe a much better practical solution for the preparation of large chemical libraries rather than use solid phase organic synthesis would be to use solid-supported reagents in a designed sequential and multi-step fashion. In combination with advances in the use of scavenging agents and catch and release techniques even greater opportunities for organics synthesis become apparent.
This lecture will present results from our laboratories towards this goal.
The Value of Chemical Genetics in Drug Discovery
Keith Russell
Department of CNS Chemistry, AstraZeneca Pharmaceuticals, Wilmington, DE, United States of America
Bridging the knowledge gap between the data provided by the human genome and our knowledge of biological processes and systems is a requirement for the efficient and effective exploitation of this knowledge in drug discovery. We see this knowledge gap as being best bridged by a truly interdisciplinary approach and a tight integration of chemistry and biology thinking and experiment. Chemical genetics provides a framework for the systematic study of small molecules to perturb and thus understand biological systems. The adoption of chemical genetics thinking is already growing in its influence among chemists and biologists, and the fruits of this integrated approach to drug discovery promises to be an exciting and rewarding area of research for the next decade.
Changing Paradigms in Drug Discovery
Hugo Kubinyi
Universität Heidelberg, Heidelberg, Germany
The strategies of drug design changed significantly within the past few decades, Whereas chemistry, biological activity hypotheses, and animal experiments dominated drug research, especially in its “golden age”, from the sixties to the eighties of the last century, many new technologies developed over the past 20 years. A vast amount of new drugs was expected to result from combinatorial chemistry and high-throughput screening; however, the yield of new drug candidates was relatively poor. Molecular modelling, virtual screening and 3D structure-based design support the selection and rational design of high-affinity protein ligands. But high affinity to a disease-relevant target is only on important property; in addition, a drug must be orally bioavailable, it should have favourable pharmacokinetics and no unacceptable side effects or toxicity.
The presentation will discuss the following questions:
- what are the reasons for the productivity gap between R&D cost and the number of NCE’s?
- is there a “druggable genome”?
- is target focus always best?
- is poor ADME the main problem in clinical development?
- are we using the right virtual screening tools?
- what are the main problems in virtual screening and structure-based design?
- what’s wrong and could we do better?
References
D. F. Horrobin, Modern biomedical research: an internally self-consistent universe with little contact with medical reality, Nature Rev. Drug Discovery. 2, 151-154 (2003).
H. Kubinyi, Drug Research: Myths, Hype and Reality, Nature Rev. Drug Discov. 2, 665-668 (2003).
Thermodynamic-based Algorithms in Drug Design. High Affinity, Selectivity and Adaptability
Ernesto Freire
Department of Biology, Johns Hopkins University, Baltimore, United States of America
A common starting point in drug development is the identification through screening or rational design of compounds that bind or exhibit some inhibitory activity against their intended targets. Often, those initial compounds bind to their targets with micromolar and sometimes weaker affinities. To become effective drugs, the binding affinities of those compounds need to be optimized by three or more orders of magnitude. This task is not a trivial one if one considers that it needs to be done while satisfying several stringent constraints, e.g. the molecular weight cannot substantially exceed 500Da in order for the molecule to be orally bioavailable; the compound needs to exhibit appropriate target selectivity, appropriate membrane permeability and viable water solubility. Furthermore, the compound needs to exhibit an adequate pharmacokinetic profile, no toxicity, etc. These constraints emphasize the need for accurate ways of predicting the various effects of introducing diverse chemical functionalities or scaffold modifications during lead optimization. At the physical level, affinity optimization is also hampered by the ubiquitous phenomenon known as enthalpy/entropy compensation, i.e. favorable changes in binding enthalpy are compensated by opposite changes in entropy and vice versa, resulting in only marginal changes in binding affinity. The ideal optimization strategy requires the identification of enthalpic or entropic contributions that carry the lowest entropic or enthalpic penalty and therefore induce the largest change in the Gibbs energy. Since thermodynamics and microcalorimetry provide the most fundamental and rigorous description of the binding of a ligand, these approaches have provided the basis for the development of thermodynamic-based algorithms aimed at: 1) Optimization of binding affinity of lead compounds; 2) Improvement of binding selectivity between similar targets; 3) Incorporation of binding adaptability to mutations that cause drug resistance. An important advantage of thermodynamic profiling and thermodynamic-based algorithms is that they can be applied to situations in which the high resolution structure of the target is either known or unknown. In this presentation, these algorithms and their practical applications to different design situations will be discussed.
Combinatorial Biosynthesis of Nonribosomal Lipopeptides
Jason Micklefield
Department of Chemistry, University of Manchester Institute of Science and Technology, Manchester, United Kingdom
Nonribosomal peptides are among the most structurally diverse and widespread secondary metabolites in nature. They comprise proteinogeneic as well as unusual amino acid residues and often contain carbohydrate, fatty acid or polyketide moieties. Many of these complex peptides are widely used therapeutic agents including the immunosuppressant cyclosporin, the antitumour agent bleomycin and the antibiotic of last resort vancomycin. Consequently there is a great demand to develop new methods that will allow for the construction of novel variants, particularly new antimicrobial agents owing to the continued emergence of pathogens with resistance to current antibiotic treatments. Given their structural complexity it is extremely unlikely that total synthesis will be able to supply this demand. On the other hand the reprogrammed, combinatorial biosynthesis of nonribosomal peptide variants is much more realistic.
The calcium-dependent antibiotic (CDA) from Streptomyces coelicolor A3(2) is a nonribosomal lipopeptide. CDA shares a similar structure and probably a related mode of action to daptomycin, which recently became the first new class of natural antibiotic to reach the clinic in many years. As a result there is now significant interest in the biosynthesis, and particularly the development of new methods for engineering, lipopeptide antibiotics. In this presentation the isolation and structural determination of several new wild-type CDA lipopeptides, containing unusual C-terminal Z-dehydrotryptophan residues, will be described.1 The organisation and putative function of genes required for the biosynthesis of CDA will be discussed and an hypothesis for the biosynthesis of CDA will be presented. Several aspects of this hypothesis have now been confirmed by appropriate biochemical experiments and several new enzymes have been characterised.
In addition we have developed several methods for biosynthetically engineering new lipopeptide antibiotics, which will be the major focus of this presentation. For example we deleted a putative 4-hydroxymandelic acid synthase encoding gene (hmaS), involved in the biosynthesis of HPG. The resulting mutant DhmaS was incapable of producing CDA. By feeding synthetic analogues of HPG and its biosynthetic precursors to DhmaS, we were able to affect the mutasynthesis of novel lipopeptides (CDA2fa, 2fb, 2d) with modified 4–fluorophenylglycine and phenylglycine residues.1 In addition to this site directed mutagenesis of nonribosomal peptide synthetase (NRPS) adenylation (A) domains was investigated as a means to engineer new CDAs.2 Single- and double-point mutants of the CDA NRPS module 7, A-domain were generated, which were predicted to alter the specificity of this domain from Asp to Asn. The double-point mutant produced a new peptide CDA2a-7N containing Asn at position 7 as expected. However in both the single- and double-point mutants significant hydrolysis of the CDA-6mer intermediate was evident. One explanation for this is that the mutant module 7 A-domain activates Asn instead of Asp, however the Asn–thioester intermediate is only weakly recognised by the upstream C-domain acceptor site allowing a water molecule to intercept the hexapeptidyl intermediate in the donor site. 2
Finally we have been investigating module swapping strategies and natural evolution events leading to new NRPSs. Notably we characterised two CDA-non-producing S. coelicolor mutants. One produces a ring-expanded dodecapeptide with an additional Asp residue CDA4a+D, and another produces a ring-contracted CDA decapeptide missing an Asp residue CDA4a–D. Subsequent genetic analysis revealed that these mutants had undergone a rare natural genetic recombination event resulting in NRPS genes where a whole Asp module (4 or 5) had either been deleted or amplified in-frame. This finding not only provides evidence of how modular NRPS have evolved to create structurally diverse peptide products, but also provides invaluable insight into the ways in which we may direct such evolutionary events in the laboratory. Specifically the design of module swapping experiments, using homologous recombination techniques to generate new in-frame NRPS encoding genes and peptide products, will be greatly facilitated by this knowledge.
1) Z. Hojati, et al., C. P. Smith and J. Micklefield “Structure, Biosynthetic Origin, and Engineered Biosynthesis of Calcium-Dependent Antibiotics from Streptomyces coelicolor” Chem. Biol. 2002, 9, 1175-1187. (lead article featured on the front cover “Chemobiosynthesis leads to new lipopeptide antibiotics” and reviewed Chem. Biol. 2002, 9, 1163-1164.)
2) G. C Uguru, C. Milne, M. Borg, F. Flett, C.P Smith and J. Micklefield, “Active site modification of adenylation domains leads to hydrolysis of upstream peptidyl thioester intermediates” J. Am. Chem. Soc. 2004, 126, 5032-5033
Chemical Complementation: A Genetic Assay for Protein Evolution and Proteomics
Virginia Cornish
Department of Chemistry, Columbia University, New York, United States of America
A high-throughput assay for enzyme activity has been developed that is reaction independent. In this assay, a small molecule yeast three-hybrid system is used to link enzyme catalysis to transcription of a reporter gene in vivo. Here, we demonstrate the feasibility of this approach using a well-studied enzyme-catalysed reaction, cephalosporin hydrolysis by the Enterobacter cloacae P99 cephalosporinase. We show that the three-hybrid system can be used to read-out cephalosprinase activity in vivo as a change in the level of transcription of a lacZ reporter gene and that the wild-type cephalosprinase can be isolated from a pool of inactive mutants using a lacZ screen. The assay has been designed so that it can be applied to different chemical reactions without changing the components of the three-hybrid system. A reaction-independent, high-throughput assay for protein function should be a powerful tool for protein engineering and enzymology, drug discovery, and proteomics.
Using Structural Similarity in the Search for Bioactivity
Peter Willett
Department of Information Studies, University of Sheffield, Sheffield, United Kingdom
The measurement of inter-molecular structural similarity lies at the heart of many approaches to virtual screening. This paper will discuss recent work in Sheffield that has investigated two ways of further enhancing the effectiveness of similarity measures when they are used to scan a database for molecules with some specific biological activity. The first approach involves a new machine learning technique, binary kernel discrimination (BKD), that has been applied to the prediction of bioactivity in a large corporate agrochemical database. BKD involves the calculation of similarities between a test molecule and the active and inactive members of a training-set of compounds for which bioactivity data are already available. BKD is shown to be noticeably more effective than existing substructural analysis approaches that use such data. Alternatively, if only some small number of active molecules is available, then methods based on extensions of conventional similarity searching (which normally requires just a single active target structure) can be used for virtual screening. We have tested several ways of combining the results from multiple similarity searches, and demonstrate that an approximate form of the BKD approach can be used that is, again, superior in effectiveness to existing approaches.
Small-Molecule Lead Generation in the 21st Century
Konrad H. Bleicher
Pharma Research Basel Discovery Chemistry, F. Hoffmann-La Roche Ltd., Basel, Switzerland
In the 1980’s, computational approaches and bio-structure based technologies were supposed to revolutionize pharmaceutical research. Classification of the binding site by X-ray crystallography and docking of potential ligands were the ultimate and most economic strategy to essentially fully exploit the ‘lock & key principle’ introduced decades earlier. Although various applications indeed delivered valuable hits and leads using structure based design it became clear that, in spite of the fact that many biological targets cannot be addressed in such a simplistic manner, bio-structure based research needed to be complemented to increase the speed and productivity of early phase drug discovery. In the early 1990’s, miniaturization and robotics paved the way for high-throughput screening. Suddenly the pendulum swung in the opposite direction to where bio-structural information seemed to become rather irrelevant due to the large number of compounds that could be screened in a relatively short period of time. Therefore a strategic move from information driven single compound design towards brute force library screening was favored. Combinatorial chemistry was supposed to address this issue in the mid 90’s. Synthesis strategies ranging from ‘single-bead technologies’ to solution phase ‘multi-component reactions’ were established allowing the generation of combinatorial arrays up to a multimillion compound range to be fed into the HTS-machinery. The logical consequence to couple high-throughput screening with high-throughput chemistry was nevertheless not delivering the promised results due to the inherent lack of diversity, the often built in overload of complexity and the resulting undesirable physicochemical properties of such combinatorial libraries. In addition, there is certainly a price to be paid for the speed on both ends: the biochemical testing as well as the chemistry itself. The rather high noise observed within high-throughput screening and the low compound quality, in particular compound purity of combinatorial arrays, are limiting the outcome of such high potential technologies. It is rather the integration of library design, parallel chemistry and biochemical as well as physicochemical testing that is required to successfully deliver high content lead series with the potential for further development into clinical candidates and ultimately new medicines. Multidimensional optimization, meaning the validation of chemical compounds not only in terms of their binding affinity but rather their pharmacological and pharmacokinetic profile as early as possible, is setting the stage for contemporary lead generation activities. This presentation will cover critical aspects of modern lead generation strategies at F. Hofmann-La Roche implemented for the rapid identification of high content small molecule modulators of protein function. An overview will be given for a range of established and new methods to generate lead molecules with a focus on G-protein coupled receptors (GPCRs). This will in addition to technological aspects also encompass, the application of ‘privileged structure’ based library design as well as virtual screening. The impact of a chemogenomics platform to the hit & lead generation process for GPCRs will also be addressed.
Chemical Approaches in Cellular Microbiology
Nicholas J. Westwood
School of Chemistry and Centre for Biomolecular Sciences, University of St. Andrews, St. Andrews, United Kingdom
The Genome programmes inspire and facilitate a large range of research goals in both biology and biochemistry. Arguably, these programmes should have just as profound an effect on the synthetic chemistry community. There is a fantastic opportunity for researchers with skills in solid phase organic synthesis to help determine the biological function of the many new proteins that are being identified.
A key approach to this goal (termed chemical genetics) requires the synthesis of large collections of small molecules. High-throughput assay systems are then used to mine these compound collections for small molecules that elicit the required biological response (inhibition of enzyme activity or a defined phenotype). Once small molecule-protein pairs have been identified, it is possible to use the small molecule to study the role of the protein in vivo as a function of time and location.
This seminar will provide recent examples of these types of studies with a particular emphasis on application in cellular microbiology.
References
“Communication between microtubules and the cytokinetic ring dissected with a novel myosin II inhibitor.” Aaron F. Straight, Amy Cheung, John Limouze, Irene Chen, Nick J. Westwood, James R. Sellers, and Timothy J. Mitchison. Science, 12003; 299:1743-1747.
“Using small molecules to ask big questions in cellular microbiology.” G.E. Ward, K.L. Carey and N.J. Westwood. Cellular Microbiology, 2002, 4(8):471-482.
“A small molecule approach to study host-pathogen interaction: identification of modulators of host cell invasion by Toxoplasma gondii.” N.J. Westwood, K.L. Cary, T.J. Mitchison, G.E. Ward, PNAS.
Multidimensional Exploration into Biochemical Pathways
Johann Gasteiger1,2, Martin Reitz 1, Oliver Sacher 2
1Computer-Chemie-Centrum, Universität Erlangen-Nürnberg, Erlangen, Germany
2 Molecular Networks GmbH, Erlangen, Germany
Biochemical processes in living organisms are often represented by complicated two-dimensional networks. Finding the desired information, and, in particular, perceiving relationships between individual reactions in such networks can be quite difficult. We have stored the contents of the poster "Biochemical Pathways" originally distributed by Boehringer Mannheim (now Roche) in a reaction database and have enriched it with additional information. The database contains 1,500 structures and 2,200 reactions. Small as this database is, it nevertheless stores information on the most important reactions, those that keep us alive.
Searches can now be performed for names, full structures and substructures, reaction partners, enzymes and coenzymes, organisms, reaction centers, etc. By using a standard structure format, other chemical databases and computer programs can be connected to this database. Furthermore, connection to bioinformatics databases can be made through enzyme names and enzyme codes.
As an application, we have investigated the geometric and electronic requirements of enzyme reactions. Three-dimensional models were automatically built by the 3D structure generator CORINA for all molecules involved in biochemical pathways. This then allowed to test the transition state hypothesis, stating that the role of an enzyme is primarily to stabilize the transition state of a reaction. This hypothesis was tested with inhibitors of some enzyme reactions establishing the geometric requirements of those reactions.
In order to investigate the electronic requirements of enzyme reactions, various physicochemical effects such as charge distribution as well as inductive, resonance, and polarizability effects were calculated for the atoms and bonds of the reaction center, i.e. those directly participating in the reaction. These values were then used to train a self-organizing (Kohonen) neural network, clustering these reactions. These clusters by and large correspond to the classification of enzymes by the EC code. However, sometimes differences are observed indicating deficiencies of the EC classification and pointing out that the physicochemical descriptors show finer details of enzyme reactions.
Thus, this database provides deeper insights into the mechanisms of biochemical pathways and can also be used for making inferences on the metabolism of compounds.
In-silico Protein-Ligand Dynamics: New Paradigms
Chandra Verma
Department of Computational Biology, Bioinformatics Institute, Singapore
Computational methods enable experiments of biological molecules that are unique in that non-physical probes can be carried and they often reveal rich insights. Two studies, one of ligand entry into proteins and the other of the actual binding of the ligand to the protein, both lead to surprising insights. Both lead to suggestions that new paradigms need to be invoked that might not only be physically more intuitive, but could also pave the way for better strategies in principles of design.
Molecular Information Theory: From Clinical Applications to Molecular Machine Efficiency
Thomas D. Schneider
Laboratory of Experimental and Computational Biology, National Cancer Institute, Frederick, MD, United States of America
Information theory was introduced by Claude Shannon in 1948 to precisely characterize data flows in communications systems. The same mathematics can also be fruitfully applied to molecular biology problems. We start with the problem of understanding how proteins interact with DNA at specific sequences called binding sites. Information theory allows us to make an average picture of the binding sites and this can be shown with a computer graphic called a
(http://www.lecb.ncifcrf.gov/~toms/glossary.html#sequence_logo).
Sequence logos show how strongly parts of a binding site are conserved, on a scale in bits of information. They have been used to study a variety of genetic control systems. More recently the same mathematics has been used to look at individual binding sites using another computer graphic called a sequence walker. (http://www.lecb.ncifcrf.gov/~toms/glossary.html#sequence_walker).
Sequence walkers are being used to predict whether changes in human genes cause mutations or are neutral polymorphisms. It may soon be possible to predict the degree of colon cancer by this method.
Information theory can also be used to understand the relationship between the binding energy dissipated when two molecules stick together and the amount of sequence conservation of the molecules measured in bits. Using the Second Law of Thermodynamics, this relationship can be expressed as the efficiency of the molecular interaction. Surprisingly, many molecular systems including genetic systems, visual pigments and motility proteins have efficiencies near 70%. A purely geometrical explanation of this result shows that although biological systems are selected to have the highest efficiency, it is restricted to 70% because having precisely distinguishable molecular states is more important.
Handling Far-From-Equilibrium Processes in Metabolic Models
Athel Cornish-Bowden
CNRS - Bioenergétique et Ingénierie des Proteines, Marseille, France
Most models of metabolic systems include some processes with equilibrium constants so large that they are almost irreversible. They raise a number of questions about how such reactions can be handled without invalidating the model. In practice, during about 40 years of metabolic modelling it was almost always assumed that if it was legitimate to ignore the reverse reaction (the negative term in the numerator of a rate equation) it was equally legitimate to ignore all effects of products on the rate (typically positive terms in the denominator). However, this improperly links two different assumptions that are logically separate and that need to be kept separate if the model is to give valid results. In other words, it is perfectly possible, and many examples exist, for a reaction to be significantly inhibited by its product even though the equilibrium constant so strongly favours the forward reaction that it is effectively irreversible. In algebraic terms this just means that the positive term in product concentration in the denominator of the rate equation can be kinetically important even if the negative term in the numerator is utterly negligible.
Another point that needs to be understood is the extent to which it is possible to measure valid thermodynamic parameters in biological systems if they are studied over a narrow range of temperature. The accessible temperature range is typically from about 5 to about 40C, and may be much less, which means that there is not enough information available in the data to provide meaningful estimates of entropies of activation. Finally, we consider the extent to which enzymes can evolve to be much more effective catalysts in one direction of reaction than in the other without violating thermodynamic principles.
Simulating Self-Organization in Biomolecular Systems
Alan E. Mark
Department of Biophysical Chemistry, University of Groningen, Groningen, The Netherlands
Molecular Dynamics simulation techniques are increasingly used to understand and predict interactions within (bio)molecular systems. The direct simulation of processes such as peptide folding and the spontaneous self-assembly of peptides or lipids is now possible. In addition, simulations are being used to predict the binding properties of potential ligands and to determine the reaction mechanism of enzymes. The question is how reliable are such calculations? In particular, as it is not yet possible to directly observe the time resolved energetic and dynamic properties of biomolecular systems at atomic resolution using experimental methods, how can we ensure that the results based on such simulations are realistic. This question will be addressed in relation to recent simulations of self-organization in (bio)molecular systems.
Catalytic Strategies for the Activation of Sulfate
Thomas S. Leyh
Department of Biochemistry, The Albert Einstein College of Medicine, Bronx, NY, United States of America
Sulfate, by most norms, is a stable, non-reactive compound; yet, its chemistry is readily manipulated by the cell to produce the hundreds of sulfur-containing compounds needed for its metabolism. To position sulfate to enter in a facile, energetically favorable way into its metabolic activities, it is chemically activated, by forming the phosphoric-sulfuric acid anhydride bond that is the chemical hallmark of activated sulfate (APS or adenosine 5-phosphosulfate). This remarkably high-energy bond is synthesized by the enzyme ATP sulfurylase in a reaction in which the adenylyl-moiety (~AMP) of ATP is transferred to sulfate. The synthesis of this ancient high energy donor is sufficiently unfavorable (Keq for APS synthesis is ~ 1 x 10-8) that it presents both a mass-action and a kinetic barrier to the acquisition of sulfur. Nature has invented several mechanisms to satisfy the selective demands created by these barriers. Among them is the recent discovery that enzymes in the cysteine biosynthetic pathway self organize and that from their interactions emerge “new” catalytic function – ATP hydrolysis. Furthermore, the hydrolysis is stoichiometrically and energetically linked to the synthesis of APS. The complex embodies a hierarchy of linked catalytic functions that can be viewed as a metabolic energy pump, each stroke of which (each ATP that is hydrolyzed) delivers one molecule of APS into the cysteine biosynthetic pathway.
Navigation in Chemical Space - Correlation-Vector Methods for Similarity Searching
Gisbert Schneider
Institut für Organische Chemie und Chemische Biologie, Universität Frankfurt, Frankfurt/Main, Germany
The rational design of compound libraries that are enriched in certain “biological activities” requires a meaningful representation of molecular structures, an appropriate similarity metric or fitness function, and a method for systematic searching for candidate structures in large virtual compound spaces. The usefulness of individual molecular descriptors and chemical space metrics is context-dependent. Our recent research activities have concentrated on the investigation of the usefulness of correlation-vector representations of molecular structures and properties in combination with various similarity metrics. These alignment-free descriptors seem to be particularly applicable when a course-grain filtering of data sets is required in combination with a high execution speed. Significant enrichment of actives was obtained by retrospective analysis. Different descriptors retrieved only weakly overlapping sets of active molecules among the top-ranking compounds. Generally, none of the different descriptors tested in this study clearly outperformed the others. For ligand-based similarity searching it is thus recommended to exploit several descriptors in parallel. Prospective application of similarity searching resulted in novel receptor ligands for several receptors. From these studies we concluded that it might be advantageous to employ several molecular descriptors and similarity metrics in parallel and benefit from a unification of the various definitions of “chemical similarity”.
A further concept employs 3D pharmacophore models in combination with correlation-vector based database screening. A quantitative pharmacophore-based approach for the compilation of focused screening libraries was developed (SQUID). The pharmacophore model is represented by a number of spheres of Gaussian-distributed feature densities. Different degrees of “fuzziness” can be introduced to influence the resolution of a model. Appropriately weighted fuzzy pharmacophore models performed better in retrospective screening than similarity searching using only a single query molecule.
The correlation-vector approach was also used for structure-based virtual screening. “Virtual ligands” were constructed inside known binding pockets, representing a blueprint of potential receptor-ligand interaction sites. The virtual ligand was then used as a template for rapid database screening. Several descriptions of interaction energies or pharmacophore sites were tested for their suitability. High enrichment factors were obtained in retrospective screening studies. This method could be used to complement automated molecular docking techniques.
Evolutionary Perspectives on Protein Folding, Structure, and Dynamics
Richard A. Goldstein
Department of Mathematical Biology, National Institute for Medical Research, London, United Kingdom
Proteins fold into their native-state conformations in milliseconds, to seconds, ignoring theoretical estimates that this process should take many times the age of the universe. Much work is directed to understanding how proteins are so much smarter than theorists, who cannot even reliably predict what the final folded states will be. Proteins have one major advantage over theorists – proteins have been working on this problem for billions of years.
We can consider different ways in which proteins may have evolved to solve the protein-folding problem. Using simple theoretical models, we can show how neutral evolution and population dynamics combined with the need to fold can explain many of the observed properties of proteins, including the way proteins fold, the distribution of observed protein structures, the marginal stability of proteins, and how the evolutionary robustness of protein structures co-exists with sequence plasticity.
Modelling and Simulation of Pharmacokinetic and Pharmacodynamic Systems-Approaches in Drug Discovery
Alex J. MacDonald
Department of Non-Clinical Drug Safety, F. Hoffmann-La Roche Ltd., Basel, Switzerland
Pharmacokinetics and pharmacodynamics in drug discovery are often viewed simple data generation processes. Candidate compounds are screened for various ADME and physico-chemical properties together with their potency in vitro and their effectiveness in in vivo models. In many cases simple summary parameters are used for comparison between drug candidates and for project decision making. However, weighing the relevance and importance of such data in isolation or in a qualitative manner is not a simple task. Modelling and simulation provides a framework for integrating these data, providing outputs that contain more information than can be elucidated from the data in isolation. The use of biologically realistic models allows for the separation of the biological and compound-specific components of the pharmacokinetic and pharmacodynamic systems. One can then begin to develop a generic approach that is applicable to the drug discovery process. An outline of the approach being developed by Roche together with practical considerations and examples will be presented.
Physiologically-based pharmacokinetic (PBPK) modelling is integral to Roche’s approach. PBPK models, by design, are capable of integrating information about various pharmacokinetic processes, including absorption, metabolism and distribution. They can be used not only to estimate summary in vivo pharmacokinetic parameters, but also to predict the complete drug concentration time-course in both plasma and tissues. However, a commonly held view is that PBPK models are complex and data-intensive, and, therefore, not applicable to the early phases of drug development. Many of the biochemical and physico-chemical parameters are generated routinely in vitro in the lead generation and optimisation phases. Quantitative structure-activity relationships and mechanism-based in silico models exist for estimating other drug properties, most critically, the partitioning of drugs into different body tissues. In vitro – in vivo scaling methods are becoming routine for the estimation of hepatic metabolism. The necessary physiological and anatomical data (e.g. tissue volumes and blood flows) are available in the scientific literature for many commonly used laboratory animals and humans. Therefore, most of required data are produced already.
By combining PBPK models with simple pharmacodynamic models, for example based on in vitro or in vivo efficacy data, the link between basic compound properties and effect in vivo is made. This allows the project teams to compare compounds with the target profile and with each other over a range of simulated doses and regimens. Crucially, an attempt at predicting drug effect in vivo in the target species, human, can be made long before the drug reaches the clinic.
