The high failure rate of drugs in the development phase requires a strategy to reduce risks by generating lead candidates from different chemical classes. We describe a new three-dimensional computational approach for lead evolution, based on multiple pharmacophore hypotheses. Using full conformational models for both active and inactive compounds, a large number of pharmacophore hypotheses are analyzed to select the set or "ensemble" of hypotheses that, when combined, is most able to discriminate between active and inactive molecules. The ensemble hypothesis is then used to search virtual chemical libraries to identify compounds for synthesis. This method is very rapid, allowing very large virtual libraries on the order of a million compounds to be filtered efficiently. In applying this method to alpha(1)-adrenergic receptor ligands, we have demonstrated lead evolution from heterocyclic alpha(1)-adrenergic receptor ligands to highly dissimilar active N-substituted glycine compounds. Our results also show that the active N-substituted glycines are part of our smaller filtered library and thus could have been identified by synthesizing only a portion of the N-substituted glycine library.

Combinatorial libraries for drug discovery can be produced with several different goals: maximum diversity, "drug like" properties, bias toward particular receptor families, lead optimization, or structure based drug design. The most suitable approach to designing a library depends on its specific purpose. Never the less, some general tools such as the creation of a property space, statistical experimental design, and stratified sampling from candidate bins can be useful in a wide variety of library design problems.This paper will compare the designs of 4 different libraries, intended for 4 different purposes: a pure diversity design, a design biased toward the physicochemical properties of orally available drugs, a library biased by docking into the active site of thrombin, and a lead optimization library designed to increase the potency and decrease the molecular weight of an initial hit that bound to Urokinase Plasminogen Activation Receptor (UPAR).

Distance geometry methods have been used extensively to build models of molecules of various sizes, including small molecules, peptides, and proteins. These methods are often overlooked as tools for conformational analysis, even though they often perform as well as other conformational sampling methods. We have implemented two new distance geometry approaches in the DGEOM95 package. In the first new method, the traditional embedding algorithm is replaced with a procedure that generates random 4D coordinates for each atom, followed by refinement of these coordinates into 3D using the distance geometry error function. The conformational sampling produced by this method is comparable to that obtained with partial metrization, and superior to that obtained with the original embedding procedure. In the second method, a molecular dynamics step is included in the refinement stage. Although this method can be applied to any embedding algorithm, substantial improvements in sampling are seen primarily with the original embedding algorithm.

A new field of research, "molecular diversity", represents a paradigm shift for drug discovery in the '90s. In the identification of drug candidates, the automated, permutational, and combinatorial use of chemical building blocks now allows the generation and screening of unprecedented numbers of compounds. Drug discovery-better, faster, cheaper? Note that diversity-based discovery programs are currently producing many, many potential leads every year. Indeed, more compounds have been made and screened in the '90s than in the previous century of pharmaceutical research. Of course, diversity is much more than a game of numbers. Key elements of variety, complexity, spatial features, and multiple physicochemical parameters contribute to diversity. And, combinatorial. synthesis begs for new assay schemes to be established, including affinity selection techniques, tagging methodologies, deconvolution tactics, and other brute-force strategies. However, the resulting demands placed on informatics can be staggering. Massive data sets are now amassed in a period of weeks to months, and innovative tools for data handling and analysis are critical. Lead discovery is often followed by lead optimization, although diversity libraries based on well-known medicinal pharmacophores have the potential to produce development candidates directly. Where optimization is needed, diversity can also play a role in speeding and enhancing the research process, yielding better drug candidates, more quickly. Discovery and development should be contrasted in relation to diversity libraries. One might predict that any highly diverse set of building blocks might yield potent and selective leads. In contrast, the need for orally available therapies can restrict the viable chemistries to those producing low molecular weight compounds, with stability in appropriate biological milieus, suitable safety characteristics, etc. Furthermore, there are noteworthy ramifications of the diversity game beyond targeted pharmaceutical research, for example, in novel ways of representing multi-dimensional physicochemical parameter space in our limited three-to five-dimensional world (3-dimensions, x, y, and z, plus color and time), in biophysics and spectroscopy applied to assays, for example, mass spectrometric identification of leads and fluorescence-activated cell sorting techniques, and in other disciplines too. Thus, an integrated approach to exploiting molecular diversity, taking into account both discovery and development considerations, plus multiple scientific disciplines, is ultimately required for optimal success.

We characterize the in vitro and in vivo pharmacology of CHIR 2279, an N-substituted glycine peptoid previously identified from a combinatorial library as a novel ligand to alpha 1-adrenoceptors. Competitive receptor-binding assays with [3H]prazosin showed that CHIR 2279 was similar to prazosin in binding to alpha 1A (rat submaxillary), alpha 1a, alpha 1b, and alpha 1 d (cDNA expressed in LTK- cells) with high and approximately equipotent affinity. Ki values for CHIR 2279 ranged from 0.7 to 3 nM, and were 10-fold weaker than with prazosin. Functional assays for postsynaptic alpha 1-adrenoceptors showed CHIR 2279 was approximately equipotent in antagonizing agonist-induced contractile responses with rat was deferens (alpha 1A), canine prostate (alpha 1A), rat spleen (alpha 1B) and rat aorta (alpha 1D). The pA2 for CHIR 2279 averaged 7.07 in these assays, indicating a 10- to 100-fold lower in vitro potency than prazosin. In dogs, CHIR 2279 antagonized the epinephrine-induced increase in intraurethal pressure (pseudo pA2, 6.86) and in rats antagonized the phenylephrine-induced increase in mean arterial blood pressure. In rats and guinea pigs, CHIR 2279 induced a dose-dependent decrease in mean arterial blood pressure without eliciting the tachycardia commonly observed with other alpha 1-blockers. Pharmacokinetic/pharmacodynamic modeling showed the i.v. system clearance rate of CHIR 2279 was 60 and 104 ml/min/kg in rats and guinea pigs, respectively, and the in vivo potency for mean arterial blood pressure reduction was twice as great in guinea pigs (EC50, 520 ng/ml) than rats (EC50, 1170 ng/ml).

We present the derivation of a new molecular mechanical force field for simulating the structures, conformational energies, and interaction energies of proteins, nucleic acids, and many related organic molecules in condensed phases. This effective two-body force field is the successor to the Weiner et al. force field and was developed with some of the same philosophies, such as the use of a simple diagonal potential function and electrostatic potential fit atom centered charges. The need for a 10-12 function for representing hydrogen bonds is no longer necessary due to the improved performance of the new charge model and new van der Waals parameters. These new charges are determined using a 6-31G* basis set and restrained electrostatic potential (RESP) fitting and have been shown to reproduce interaction energies, free energies of solvation, and conformational energies of simple small molecules to a good degree of accuracy. Furthermore, the new RESP charges exhibit less variability as a function of the molecular conformation used in the charge determination. The new van der Waals parameters have been derived from liquid simulations and include hydrogen parameters which take into account the effects of any geminal electronegative atoms. The bonded parameters developed by Weiner et al. were modified as necessary to reproduce experimental vibrational frequencies and structures. Most of the simple dihedral parameters have been retained from Weiner et al., but a complex set of 4 and yj parameters which do a good job of reproducing the energies of the low-energy conformations of glycyl and alanyl dipeptides has been developed for the peptide backbone.

Screening synthetic combinatorial libraries, such as mixtures of oligo(N-substituted)glycines, facilitates rapid drug lead discovery and optimization by vastly increasing the number of candidate molecules made and tested. Discovery efficiency and productivity can be further improved by using experimental design to maximize molecular diversity for a given library size or to bias the library with key features for a specific receptor. We describe new methods to quantify molecular diversity using descriptors that characterize lipophilicity, shape and branching, chemical functionality, and specific binding features. Experimental design methods select sets of side chains that are diverse in these properties, and "flower plots" allow the diversity to be graphically compared. We also quantify the overall diversity accessible to different families of combinatorial chemistry.

Peptoids, oligomers of N-substituted glycines, are described as a motif for the generation of chemically diverse libraries of novel molecules. Ramachandran-type plots were calculated and indicate a greater diversity of conformational states available for peptoids than for peptides. The monomers incorporate t-butyl-based side-chain and 9-fluorenylmethoxy-carbonyl alpha-amine protection. The controlled oligomerization of the peptoid monomers was performed manually and robotically with in situ activation by either benzotriazol-1-yloxytris(pyrrolidino)phosphonium hexafluorophosphate or bromotris(pyrrolidino)phosphonium hexaflurophosphate. Other steps were identical to peptide synthesis using alpha-(9-fluorenylmethoxycarbonyl)amino acids. A total of 15 monomers and 10 oligomers (peptoids) are described. Preliminary data are presented on the stability of a representative oligopeptoid to enzymatic hydrolysis. Peptoid versions of peptide ligands of three biological systems (bovine pancreatic alpha-amylase, hepatitis A virus 3C proteinase, and human immunodeficiency virus transactivator-responsive element RNA) were found with affinities comparable to those of the corresponding peptides. The potential use of libraries of these compounds in receptor- or enzyme-based assays is discussed.