Forecasting roles of combinatorial chemistry in the age of genomically derived drug discovery targets.

Genomics has caused an explosion in the number of potential therapeutic targets with varying degrees of validated pathophysiology. Among the first applications of combinatorial chemistry in genomics-driven drug discovery is the search for surrogate ligands or substrates. In the event that no surrogate is found for molecular assays, more exotic functional screens in whole cells or model organisms are used. Protein-protein interaction mapping by yeast and mammalian two-hybrid systems dominates empirical functional genomics, and this will lead to a bias for screening projects targeting this type of interaction. Drug discovery for protein-protein interactions has a poor track record, and this will challenge prevailing views on the design of combinatorial libraries. Genomics based on structural homology will yield many putative kinases, receptors, enzymes, transporter proteins, ion channels and GPCRs. Most of these projects will require new surrogate agonists, ligands or substrates, and then pharmaceutically useful agonists or antagonists will need to be found. Again, combinatorial chemistry might be essential to these studies. Given the need to screen hundreds of targets at great risk of irrelevance to pathophysiology, combined with the challenge of finding surrogate or natural ligands for these new targets, there is an urgent need for efficiency. Different groups are addressing these concerns by developing biologically-driven combinatorial libraries in order to achieve a higher density of bioactivity. Early efforts in this regard will be described.

Combinatorial chemistry defines general properties of linkers for the optimal display of peptide ligands for binding soluble protein targets to TentaGel microscopic beads.

Affinity chromatography and the binding of soluble target proteins to novel or known ligands attached to solid supports are important phenomena to basic and applied research. Satisfactory display of a ligand for the acceptor protein is critical for successful binding to occur. Here we describe the application of combinatorial chemistry to systematically explore the properties of linkers used to present peptide ligands to various protein targets. Our main interest is in drug discovery, and our results probably explain, in large part, the disappointing efficiency of an early drug discovery method known as the "Selectide Process" (Lam, K. S., et al. (1991) Nature 358, 82-84). Interestingly, for all seven protein targets studied, a cationic feature was found to be a common theme for optimal linkers displaying peptide ligands on TentaGel beads, and this is not likely to be caused by ionic exchange mechanisms.

Efficient discovery of inhibitory ligands for diverse targets from a small combinatorial chemical library of chimeric molecules.

Living systems are mainly composed and regulated by compounds in four biochemical classes and their polymers-nucleotides, carbohydrates, lipids, and amino acids. Early combinatorial chemistry libraries consisted of peptides. The present report describes the general bioactivity and biophysical properties of a combinatorial chemical library that used glyco, nucleotidyl, and lipid building blocks. The resulting chimeric combinatorial library of 361 compounds had a confirmed cumulative hit rate of 0.16%, which is 8-fold higher than a commonly claimed industrial benchmark of 0. 02%. It produced 7 structurally confirmed hits for a third of 12 proprietary drug discovery projects, and these comprised a variety of molecular targets. Diversity analyses demonstrated that despite the small number of compounds, a wider range of diversity space was covered by this library of biochemical chimeras than by a branched tripeptide library of the same size and similar generic formula.

Combinatorial chemistry reveals a new motif that binds the platelet fibrinogen receptor, gpIIbIIIa.

Among cell adhesion molecules, the classic Arg-Gly-Asp (RGD) motif is the best studied. We used combinatorial chemical and affinity immunochemical methods to find a novel motif of unnatural peptide ligands for the fibrinogen receptor of platelets, gpIIbIIIa (alphaIIbbeta3). The new d-amino acid motif, p(f/y)l, is unique among the ligands that bind the RGD pocket: It lacks the carboxylic acid group that is believed to coordinate with calcium in the MIDAS motif of the receptor. With an IC50 of 14 microM for the most potent compound, these linear p(f/y)l peptides had affinities similar to those of linear peptides containing RGD, and reversed sequences failed to compete with binding up to 1 mM. As the new motif was so different, molecular modeling was employed to suggest a model for molecular recognition. A reversed binding mechanism common for d-amino acid mimics of natural l-amino acid peptides offers an attractive hypothesis that suggests three points of contact similar to those made by the RGD-mimicking monoclonal antibody, OPG2. Interestingly, the model proposes that pi-electrons in the new motif may substitute for the carboxylate group present in all other RGD-types of ligands. Although modeling linear peptides is subjective, the pi-bonding model provides intriguing possibilities for medicinal chemistry after appropriate confirmatory studies.

Discovery of a novel, potent, and specific family of factor Xa inhibitors via combinatorial chemistry.

A series of low molecular weight peptide inhibitors of factor Xa, unrelated to any previously described, was identified by screening a combinatorial peptide library composed of L-amino acids. The minimal inhibitory sequence is a tripeptide, L-tyrosinyl-L-isoleucyl-L-arginyl, which competitively inhibits the hydrolysis of small chromogenic substrates by factor Xa but binds in an orientation which prevents a productive nucleophilic attack by serine 195 of the catalytic triad on the carbonyl carbon of the carboxyterminal arginine. The initial leads identified in an octamer combinatorial peptide library ranged in potency from 4 to 15 microM. These peptides were modified into peptide mimetics with a greater than 1000-fold increase in potency while retaining unusual selectivity for factor Xa over the related serine proteases thrombin, factor VIIa/tissue factor, plasmin, activated protein C, kallikrein, and trypsin. One of the most potent analogues, SEL 2711, with a Ki of 0.003 microM for factor Xa and 40 microM for thrombin, is active in in vitro and ex vivo coagulation assays, suggesting the potential application of these inhibitors in anticoagulant therapy.

Refolding parameters for the allosteric homodimeric guanylyl cyclase catalytic core from the atrial natriuretic peptide receptor.

Protein folding continues to be an important biophysical topic in molecular biology. We report the parameters for successfully refolding the guanylyl cyclase core of the ANP receptor, an allosteric homodimeric enzyme. Urea was a better chaotropic solvent than guanidine HCl, and physiological salt concentrations and pH were needed for optimal recovery of enzymatic activity. Renaturation was more sensitive to alkaline compared to acidic deviations in solvent conditions. The time course of refolding was sigmoidal producing an enzyme with a specific activity of 16,000 pmol cGMP/min/mg using 60 microM concentration of substrate. Additional factors are described in this unusual case of renaturing an allosteric homodimeric enzyme in vitro.

The guanylyl cyclase core of an atrial natriuretic peptide receptor: enzymatic properties and basis for cooperativity.

The enzymatic properties of a cloned atrial natriuretic peptide receptor are described. The renatured catalytic core had maximal activity with Mn2+, and all nucleoside triphosphates inhibited the enzyme competitively. The catalytic specificity of the enzyme was tested directly. The cyclase reaction was specific for guanine, producing cGMP and cyclic deoxyGMP. Surprisingly, deoxyguanylyl cyclase kinetics were classical, unlike the positive cooperativity seen for guanylyl cyclase activity, suggesting that the 2' hydroxyl group of GTP is necessary for the allosteric mechanism.

Overexpression of dimeric guanylyl cyclase cores of an atrial natriuretic peptide receptor.

Using a bacterial expression system, large amounts of the catalytic core of an atrial natriuretic peptide receptor guanylyl cyclase were produced and purified. After refolding the protein from a buffer containing urea, the enzyme had positively cooperative kinetics with a Hill coefficient, nH = 1.42 +/- 0.08. Size exclusion chromatography and denaturing polyacrylamide gel electrophoresis demonstrated that the enzyme is composed of homodimers with interacting catalytic sites.

The carboxyl region contains the catalytic domain of the membrane form of guanylate cyclase.

A polypeptide containing the catalytic domain of an atrial natriuretic peptide receptor guanylate cyclase has been produced using a bacterial expression system. A carboxyl fragment of the membrane form of guanylate cyclase from rat brain, which contains a region homologous to soluble guanylate and adenylate cyclases, was expressed in Escherichia coli with a double plasmid system that encodes T7 RNA polymerase (Tabor, S., and Richardson, C.C. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 1074-1078). Application of this expression system permitted exclusive radiolabeling of the cloned gene product, thereby providing a means to evaluate the level of expression and stability of encoded proteins. Fusion proteins were formed with the T7 bacteriophage gene 10 product and the 293 carboxyl-terminal residues of guanylate cyclase and two deletional mutants encoding 105 and 69 residues. Extracts prepared from bacteria expressing the carboxyl region, but not those expressing further deletions in this region, had substantial guanylate cyclase activity. There was no associated adenylate cyclase activity, suggesting that the catalytic domain retained its enzymatic specificity. These results provide direct evidence that the carboxyl portion of the membrane form of guanylate cyclase contains a catalytic domain. Homologous regions of the soluble form of guanylate cyclase and adenylate cyclase are likely to have enzymatic properties.

The membrane form of guanylate cyclase. Homology with a subunit of the cytoplasmic form of the enzyme.

A cDNA clone for the membrane form of guanylate cyclase has been isolated from the testis of the sea urchin Strongylocentrotus purpuratus. An open reading frame predicts a protein of 1125 amino acids including an apparent signal peptide of 21 residues; a single transmembrane domain of 25 amino acids divided the mature protein into an amino-terminal, extracellular domain of 485 amino acids and a carboxyl domain of 594 intracellular amino acids. Three potential Asn-linked glycosylation sites were present in the proposed extracellular domain. The deduced protein sequence was homologous to the protein kinase family and contained limited but significant regions of identity with a low molecular weight atrial natriuretic peptide receptor. The carboxyl region (202 amino acids) was 42% identical with a subunit of the cytoplasmic form of guanylate cyclase recently cloned from bovine lung (Koesling, D., Herz, J., Gausepohl, H., Niroomand, F., Hinsch, K.-D., Mulsch, A., Bohme, E., Schultz, G., and Frank, R. (1988) FEBS Lett. 239, 29-34). Therefore, the membrane form of guanylate cyclase is a member of an apparently large family of proteins that includes the low molecular weight atrial natriuretic peptide receptor, the soluble form of guanylate cyclase and protein kinases.

Membrane guanylate cyclase is a cell-surface receptor with homology to protein kinases.

Guanylate cyclase has been strongly implicated as a cell-surface receptor on spermatozoa for a chemotactic peptide, and on various other cells as a receptor for atrial natriuretic peptides. Resact (Cys-Val-Thr-Gly-Ala-Pro-Gly-Cys-Val-Gly-Gly-Gly-Arg-Leu-NH2), the chemotactic peptide released by sea urchin Arbacia punctulata eggs, is specifically crosslinked to A. punctulata spermatozoan guanylate cyclase. After the binding of the peptide the state of guanylate cyclase phosphorylation modulates enzyme activity. We report here that the deduced amino-acid sequence of the spermatozoan membrane form of guanylate cyclase predicts an intrinsic membrane protein of 986 amino acids with an amino-terminal signal sequence. A single transmembrane domain separates the protein into putative extracellular and cytoplasmic-catalytic domains. The cytoplasmic carboxyl-terminal 95 amino acids contain 20% serine, the likely regulatory sites for phosphorylation. Unexpectedly, the enzyme is homologous to the protein kinase family.