Designing libraries with CNS activity.

Library design is an important and difficult task. In this paper we describe one possible solution to designing a CNS-active library. CNS-actives and -inactives were selected from the CMC and the MDDR databases based on whether they were described as having some kind of CNS activity in the databases. This classification scheme results in over 15 000 actives and over 50 000 inactives. Each molecule is described by 7 1D descriptors (molecular weight, number of donors, number of acceptors, etc.) and 166 2D descriptors (presence/absence of functional groups such as NH(2)). A neural network trained using Bayesian methods can correctly predict about 75% of the actives and 65% of the inactives using the 7 1D descriptors. The performance improves to a prediction accuracy on the active set of 83% and 79% on the inactives on adding the 2D descriptors. On a database with 275 compounds where the CNS activity is known (from the literature) for each compound, we achieve 92% and 71% accuracy on the actives and inactives, respectively. The models we construct can therefore be used as a "filter" to examine any set of proposed molecules in a chemical library. As an example of the utility of our method, we describe the generation of a small library of potentially CNS-active molecules that would be amenable to combinatorial chemistry. This was done by building and analyzing a large database of a million compounds constructed from frameworks and side chains frequently found in drug molecules.

Properties of known drugs. 2. Side chains.

We continue our study of the common features present in drug molecules by looking in detail at drug side chains. Using shape description methods, we divide a database of commercially available drugs into a list of common drug side chains. On the basis of the atom pair shape descriptor (taking into account atom type, hybridization, and bond order), there are 1,246 different side chains among the 5,090 compounds analyzed. The average number of side chains per molecule is 4, and the average number of heavy atoms per side chain is 2. If we ignore the carbonyl side chain, then there are approximately 15,000 occurrences of side chains. Of these 15,000 approximately 11,000 are from the "top 20" group of side chains. This suggests that the diversity that side chains provide to drug molecules is quite low. We discuss ways that this work could be used to provide guidance for molecular design efforts.

Consensus scoring: A method for obtaining improved hit rates from docking databases of three-dimensional structures into proteins.

We present the results of an extensive computational study in which we show that combining scoring functions in an intersection-based consensus approach results in an enhancement in the ability to discriminate between active and inactive enzyme inhibitors. This is illustrated in the context of docking collections of three-dimensional structures into three different enzymes of pharmaceutical interest: p38 MAP kinase, inosine monophosphate dehydrogenase, and HIV protease. An analysis of two different docking methods and thirteen scoring functions provides insights into which functions perform well, both singly and in combination. Our data shows that consensus scoring further provides a dramatic reduction in the number of false positives identified by individual scoring functions, thus leading to a significant enhancement in hit-rates.

The SHAPES strategy: an NMR-based approach for lead generation in drug discovery.

BACKGROUND: Recently, it has been shown that nuclear magnetic resonance (NMR) may be used to identify ligands that bind to low molecular weight protein drug targets. Recognizing the utility of NMR as a very sensitive method for detecting binding, we have focused on developing alternative approaches that are applicable to larger molecular weight drug targets and do not require isotopic labeling. RESULTS: A new method for lead generation (SHAPES) is described that uses NMR to detect binding of a limited but diverse library of small molecules to a potential drug target. The compound scaffolds are derived from shapes most commonly found in known therapeutic agents. NMR detection of low (microM-mM) affinity binding is achieved using either differential line broadening or transferred NOE (nuclear Overhauser effect) NMR techniques. CONCLUSIONS: The SHAPES method for lead generation by NMR is useful for identifying potential lead classes of drugs early in a drug design program, and is easily integrated with other discovery tools such as virtual screening, high-throughput screening and combinatorial chemistry.

Recognizing molecules with drug-like properties.

A variety of successful approaches to the problem of recognizing 'drug-like' molecules have been employed. These range from simple counting schemes such as the Lipinski 'rule of five' to the analysis of the multidimensional 'chemistry space' occupied by drugs, to neural network learning systems. With this variety of tools, it now appears possible to design libraries that are enriched in compounds which have desirable or 'drug-like' properties. Verifying the robustness of these methods, and extending them, will form the basis of research in this field during the next few years.

Can we learn to distinguish between "drug-like" and "nondrug-like" molecules?

We have used a Bayesian neural network to distinguish between drugs and nondrugs. For this purpose, the CMC acts as a surrogate for drug-like molecules while the ACD is a surrogate for nondrug-like molecules. This task is performed by using two different set of 1D and 2D parameters. The 1D parameters contain information about the entire molecule like the molecular weight and the the 2D parameters contain information about specific functional groups within the molecule. Our best results predict correctly on over 90% of the compounds in the CMC while classifying about 10% of the molecules in the ACD as drug-like. Excellent generalization ability is shown by the models in that roughly 80% of the molecules in the MDDR are classified as drug-like. We propose to use the models to design combinatorial libraries. In a computer experiment on generating a drug-like library of size 100 from a set of 10 000 molecules we obtain at least a 3 or 4 order of magnitude improvement over random methods. The neighborhoods defined by our models are not similar to the ones generated by standard Tanimoto similarity calculations. Therefore, new and different information is being generated by our models, and so it can supplement standard diversity approaches to library design.

Hepatitis C virus NS3 RNA helicase domain with a bound oligonucleotide: the crystal structure provides insights into the mode of unwinding.

BACKGROUND: Hepatitis C virus (HCV) represents a major health concern as it is responsible for a significant number of hepatitis cases worldwide. Much research has focused on the replicative enzymes of HCV as possible targets for more effective therapeutic agents. HCV NS3 helicase may provide one such suitable target. Helicases are enzymes which can unwind double-stranded regions of DNA or RNA in an ATP-dependent reaction. The structures of several helicases have been published but the structural details as to how ATP binding and hydrolysis are coupled to RNA unwinding are unknown. RESULTS: The structure of the HCV NS3 RNA helicase domain complexed with a single-stranded DNA oligonucleotide has been solved to 2.2 A resolution. The protein consists of three structural domains with the oligonucleotide lying in a groove between the first two domains and the third. The first two domains have an adenylate kinase like fold, including a phosphate-binding loop in the first domain. CONCLUSIONS: HCV NS3 helicase is a member of a superfamily of helicases, termed superfamily II. Residues of NS3 helicase which are conserved among superfamily II helicases line an interdomain cleft between the first two domains. The oligonucleotide binds in an orthogonal binding site and contacts relatively few conserved residues. There are no strong sequence-specific interactions with the oligonucleotide bases.

Bovine viral diarrhea virus NS3 serine proteinase: polyprotein cleavage sites, cofactor requirements, and molecular model of an enzyme essential for pestivirus replication.

Members of the Flaviviridae encode a serine proteinase termed NS3 that is responsible for processing at several sites in the viral polyproteins. In this report, we show that the NS3 proteinase of the pestivirus bovine viral diarrhea virus (BVDV) (NADL strain) is required for processing at nonstructural (NS) protein sites 3/4A, 4A/4B, 4B/5A, and 5A/5B but not for cleavage at the junction between NS2 and NS3. Cleavage sites of the proteinase were determined by amino-terminal sequence analysis of the NS4A, NS4B, NS5A, and NS5B proteins. A conserved leucine residue is found at the P1 position of all four cleavage sites, followed by either serine (3/4A, 4B/5A, and 5A/5B sites) or alanine (4A/4B site) at the P1' position. Consistent with this cleavage site preference, a structural model of the pestivirus NS3 proteinase predicts a highly hydrophobic P1 specificity pocket. trans-Processing experiments implicate the 64-residue NS4A protein as an NS3 proteinase cofactor required for cleavage at the 4B/5A and 5A/5B sites. Finally, using a full-length functional BVDV cDNA clone, we demonstrate that a catalytically active NS3 serine proteinase is essential for pestivirus replication.

The structural basis for the specificity of pyridinylimidazole inhibitors of p38 MAP kinase.

BACKGROUND: The p38 mitogen-activated protein (MAP) kinase regulates signal transduction in response to environmental stress. Pyridinylimidazole compounds are specific inhibitors of p38 MAP kinase that block the production of the cytokines interleukin-1beta and tumor necrosis factor alpha, and they are effective in animal models of arthritis, bone resorption and endotoxin shock. These compounds have been useful probes for studying the physiological functions of the p38-mediated MAP kinase pathway. RESULTS: We report the crystal structure of a novel pyridinylimidazole compound complexed with p38 MAP kinase, and we demonstrate that this compound binds to the same site on the kinase as does ATP. Mutagenesis showed that a single residue difference between p38 MAP kinase and other MAP kinases is sufficient to confer selectivity among pyridinylimidazole compounds. CONCLUSIONS: Our results reveal how pyridinylimidazole compounds are potent and selective inhibitors of p38 MAP kinase but not other MAP kinases. It should now be possible to design other specific inhibitors of activated p38 MAP kinase using the structure of the nonphosphorylated enzyme.

Crystal structure of the hepatitis C virus NS3 protease domain complexed with a synthetic NS4A cofactor peptide.

An estimated 1% of the global human population is infected by hepatitis C viruses (HCVs), and there are no broadly effective treatments for the debilitating progression of chronic hepatitis C. A serine protease located within the HCV NS3 protein processes the viral polyprotein at four specific sites and is considered essential for replication. Thus, it emerges as an attractive target for drug design. We report here the 2.5 angstrom resolution X-ray crystal structure of the NS3 protease domain complexed with a synthetic NS4A activator peptide. The protease has a chymotrypsin-like fold and features a tetrahedrally coordinated metal ion distal to the active site. The NS4A peptide intercalates within a beta sheet of the enzyme core.

Free energy perturbation studies on binding of A-74704 and its diester analog to HIV-1 protease.

Free energy simulations have been employed to rationalize the binding differences between A-74704, a pseudo C2-symmetric inhibitor of HIV-1 protease and its diester analog. The diester analog inhibitor, which misses two hydrogen bonds with the enzyme active site, is surprisingly only 10-fold weaker. The calculated free energy difference of 1.7 +/- 0.6 kcal/mol is in agreement with the experimental result. Further, the simulations show that such a small difference in binding free energies is due to (1) weaker hydrogen bond interactions between the two (P1 and P1') NH groups of A-74704 with Gly27/Gly27' carbonyls of the enzyme and (2) the higher desolvation free energy of A-74704 compared with its ester analog. The results of these calculations and their implications for design of HIV-1 protease inhibitors are discussed.

The properties of known drugs. 1. Molecular frameworks.

In order to better understand the common features present in drug molecules, we use shape description methods to analyze a database of commercially available drugs and prepare a list of common drug shapes. A useful way of organizing this structural data is to group the atoms of each drug molecule into ring, linker, framework, and side chain atoms. On the basis of the two-dimensional molecular structures (without regard to atom type, hybridization, and bond order), there are 1179 different frameworks among the 5120 compounds analyzed. However, the shapes of half of the drugs in the database are described by the 32 most frequently occurring frameworks. This suggests that the diversity of shapes in the set of known drugs is extremely low. In our second method of analysis, in which atom type, hybridization, and bond order are considered, more diversity is seen; there are 2506 different frameworks among the 5120 compounds in the database, and the most frequently occurring 42 frameworks account for only one-fourth of the drugs. We discuss the possible interpretations of these findings and the way they may be used to guide future drug discovery research.

Structure and mechanism of inosine monophosphate dehydrogenase in complex with the immunosuppressant mycophenolic acid.

The structure of inosine-5'-monophosphate dehydrogenase (IMPDH) in complex with IMP and mycophenolic acid (MPA) has been determined by X-ray diffraction. IMPDH plays a central role in B and T lymphocyte replication. MPA is a potent IMPDH inhibitor and the active metabolite of an immunosuppressive drug recently approved for the treatment of allograft rejection. IMPDH comprises two domains: a core domain, which is an alpha/beta barrel and contains the active site, and a flanking domain. The complex, in combination with mutagenesis and kinetic data, provides a structural basis for understanding the mechanism of IMPDH activity and indicates that MPA inhibits IMPDH by acting as a replacement for the nicotinamide portion of the nicotinamide adenine dinucleotide cofactor and a catalytic water molecule.

CONCERTS: dynamic connection of fragments as an approach to de novo ligand design.

We have implemented and tested a new approach to de novo ligand design, CONCERTS (creation of novel compounds by evaluation of residues at target sites). In this method, each member of a user-defined set of fragments is allowed to move independently about a target active site during a molecular dynamics simulation. This allows the fragments to sample various low-energy orientations. When the geometry between proximal fragments is appropriate, bonds can be formed between the fragments. In this fashion, larger molecules can be built. The bonding arrangement can subsequently be changed-breaking bonds between chosen fragment pairs and forming them between other pairs-if the overall process creates lower energy molecules. We have tested this method with various mixes of fragments against the active sites of the FK506 binding protein (FKBP-12) and HIV-1 aspartyl protease. In several cases, CONCERTS suggests ligands which are in surprisingly good agreement with known inhibitors of these proteins.

Calculation of solvation and binding free energy differences between VX-478 and its analogs by free energy perturbation and AMSOL methods.

VX-478 belongs to a novel class of HIV-1 protease inhibitors that are based on N,N-disubstituted benzene sulfonamides. Force field parameters for the N,N-dialkyl benzene sulfonamide moiety have been assembled from the literature and from our own ab initio calculations. These parameters were employed to calculate solvation and binding free energy differences between VX-478 and two analogs. The free energy perturbation method has been used to determine these differences using two approaches. In the first approach, intergroup interaction terms only were included in the calculation of free energies (as in most reports of free energy calculations using AMBER). In the second approach, both the inter- and intragroup interaction terms were included. The results obtained with the two approaches are in excellent agreement with each other and are also in close agreement with the experimental results. The solvation free energies of N,N-dimethyl benzene sulfonamide derivatives (truncated models of the inhibitors), calculated using continuum solvation (AMSOL) methods, are found to be in qualitative agreement with the experimental and free energy perturbation results. The binding and solvation free energy results are discussed in the context of structure-based drug design to show how physicochemical properties (for example aqueous solubilities and bioavailabilities) of these HIV-I protease inhibitors were improved, while maintaining their inhibitory potency.

FK506-binding protein mutational analysis: defining the active-site residue contributions to catalysis and the stability of ligand complexes.

The 12 kDa FK506-binding protein FKBP12 is a cis-trans peptidyl-prolyl isomerase that binds the macrolides FK506 and rapamycin. We have examined the role of the binding pocket residues of FKBP12 in protein-ligand interactions by making conservative substitutions of 12 of these residues by site-directed mutagenesis. For each mutant FKBP12, we measured the affinity for FK506 and rapamycin and the catalytic efficiency in the cis-frans peptidyl-prolyl isomerase reaction. The mutation of Trp59 or Phe99 generates an FKBP12 with a significantly lower affinity for FK506 than wild-type protein. Tyr26 and Tyr82 mutants are enzymatically active, demonstrating that hydrogen bonding by these residues is not required for catalysis of the cis-trans peptidyl-prolyl isomerase reaction, although these mutations alter the substrate specificity of the enzyme. We conclude that hydrophobic interactions in the active site dominate in the stabilization of FKBP12 binding to macrolide ligands and to the twisted-amide peptidyl-prolyl substrate intermediate.

Comparative X-ray structures of the major binding protein for the immunosuppressant FK506 (tacrolimus) in unliganded form and in complex with FK506 and rapamycin.

FK506 (tacrolimus) is a natural product now approved in the US and Japan for organ transplantation. FK506, in complex with its 12 kDa cytosolic receptor (FKBP12), is a potent agonist of immunosuppression through the inhibition of the phosphatase activity of calcineurin. Rapamycin (sirolimus), which is itself an immunosuppressant by a different mechanism, completes with FK506 for binding to FKBP12 and thereby acts as an antagonist of calcineurin inhibition. We have solved the X-ray structure of unliganded FKBP12 and of FKBP12 in complex with FK506 and with rapamycin; these structures show localized differences in conformation and mobility in those regions of the protein that are known, by site-directed mutagenesis, to be involved in calcineurin inhibition. A comparison of 16 additional X-ray structures of FKBP12 in complex with FKBP12-binding ligands, where those structures were determined from different crystal forms with distinct packing arrangements, lends significance to the observed structural variability and suggests that it represents an intrinsic functional characteristic of the protein. Similar differences have been observed for FKBP12 before, but were considered artifacts of crystal-packing interactions. We suggest that immunosuppressive ligands express their differential effects in part by modulating the conformation of FKBP12, in agreement with mutagenesis experiments on the protein, and not simply through differences in the ligand structures themselves.

Design, synthesis and structure of non-macrocyclic inhibitors of FKBP12, the major binding protein for the immunosuppressant FK506.

We have synthesized a series of non-macrocyclic ligands to FKBP12 that are comparable in binding potency and peptidyl prolyl isomerase (PPIase) inhibition to FK506 itself. We have also solved the structure of one of these ligands in complex with FKBP12, and have compared that structure to the FK506-FKBP12 complex. Consistent with the observed inhibitory equipotency of these compounds, we observe a strong similarity in the conformation of the two ligands in the region of the protein that mediates PPIase activity. Our compounds, however, are not immunosuppressive. In the FKBP12-FK506 complex, a significant portion of the FK506 ligand, its 'effector domain', projects beyond the envelope of the binding protein in a manner that is suggestive of a potential interaction with a second protein, the calcium-dependent phosphatase, calcineurin, whose inhibition by the FKBP 12-FK506 complex interrupts the T-cell activation events leading to immunosuppression. In contrast, our compounds bind within the surface envelope of FKBP12, and induce significant changes in the structure of the FKBP12 protein which may also affect calcineurin binding indirectly.