Tracing binding modes in hit-to-lead optimization: chameleon-like poses of aspartic protease inhibitors.

Successful lead optimization in structure-based drug discovery depends on the correct deduction and interpretation of the underlying structure-activity relationships (SAR) to facilitate efficient decision-making on the next candidates to be synthesized. Consequently, the question arises, how frequently a binding mode (re)-validation is required, to ensure not to be misled by invalid assumptions on the binding geometry. We present an example in which minor chemical modifications within one inhibitor series lead to surprisingly different binding modes. X-ray structure determination of eight inhibitors derived from one core scaffold resulted in four different binding modes in the aspartic protease endothiapepsin, a well-established surrogate for e.g. renin and ß-secretase. In addition, we suggest an empirical metrics that might serve as an indicator during lead optimization to qualify compounds as candidates for structural revalidation.

Structure-based design of inhibitors of the aspartic protease endothiapepsin by exploiting dynamic combinatorial chemistry.

Structure-based design (SBD) can be used for the design and/or optimization of new inhibitors for a biological target. Whereas de novo SBD is rarely used, most reports on SBD are dealing with the optimization of an initial hit. Dynamic combinatorial chemistry (DCC) has emerged as a powerful strategy to identify bioactive ligands given that it enables the target to direct the synthesis of its strongest binder. We have designed a library of potential inhibitors (acylhydrazones) generated from five aldehydes and five hydrazides and used DCC to identify the best binder(s). After addition of the aspartic protease endothiapepsin, we characterized the protein-bound library member(s) by saturation-transfer difference NMR spectroscopy. Cocrystallization experiments validated the predicted binding mode of the two most potent inhibitors, thus demonstrating that the combination of de novo SBD and DCC constitutes an efficient starting point for hit identification and optimization.

Total chemical synthesis of human T-cell leukemia virus type 1 protease via native chemical ligation.

Human T-cell leukemia virus 1 (HTLV-1) protease, a member of the aspartic acid protease family, plays critical roles in the pathogenesis of the virus and is an attractive viral target for therapeutic intervention. HTLV-1 protease consists of 125 amino acid residues and functions as a homodimer stabilized in part by a four-stranded beta-sheet comprising the N- and C-termini. Compared with many other viral proteases such as HIV-1 protease, HTLV-1 protease is elongated by an extra 10 amino acid residue "tail" at the C-terminus. The structural and functional role of the extra C-terminal residues in the catalysis of HTLV-1 protease has been a subject of debate for years. Using the native chemical ligation technique pioneered by Kent and coworkers, we chemically synthesized a full-length HTLV protease and a C-terminally truncated form encompassing residues 1-116. Enzyme kinetic analysis using three different peptide substrates indicated that truncation of the C-terminal tail lowered the turnover number of the viral enzyme by a factor of 2 and its catalytic efficiency by roughly 10-fold. Our findings differ from the two extreme views that the C-terminal tail of HTLV-1 protease is either fully dispensable or totally required for enzyme dimerization and/or catalysis.

Active-site specificity of digestive aspartic peptidases from the four species of Plasmodium that infect humans using chromogenic combinatorial peptide libraries.

Two targeted chromogenic octapeptide combinatorial libraries, comprised of 38 pools each containing 361 different peptides, were used to analyze the enzyme/substrate interactions of five plasmepsins. The first library (P1 library) was based on a good synthetic aspartic peptidase substrate [Westling, J., Cipullo, P., Hung, S. H., Saft, H., Dame, J. B., and Dunn, B. M. (1999) Protein Sci. 8, 2001-2009; Scarborough, P. E., and Dunn, B. M. (1994) Protein Eng. 7, 495-502] and had the sequence Lys-Pro-(Xaa)-Glu-P1*Nph-(Xaa)-Leu. The second library (P1' library) incorporated results with the plasmepsins from the first library and had the sequence Lys-Pro-Ile-(Xaa)-Nph*P1'-Gln-(Xaa). In both cases, P1 and P1' were fixed residues for a given peptide pool, where Nph was a para-nitrophenylalanine chromogenic reporter and Xaa was a mixture of 19 different amino acids. Kinetic assays monitoring the rates of cleavage of these libraries revealed the optimal P1 and P1' residues for the five plasmepsins as hydrophobic substitutions. Extended specificity preferences were obtained utilizing liquid chromatography-mass spectrometry (LC-MS) to analyze the cleavage products produced by enzyme-catalyzed digestion of the best pools of each peptide library. LC-MS analysis of the P1-Phe and P1'-Phe pools revealed the favored amino acids at the P3, P2, P2', and P3' positions. These analyses have provided new insights on the binding preferences of malarial digestive enzymes that were used to design specific methyleneamino peptidomimetic inhibitors of the plasmepsins. Some of these compounds were potent inhibitors of the five plasmepsins, and their possible binding modes were analyzed by computational methods.

Microwave-enhanced medicinal chemistry: a high-speed opportunity for convenient preparation of protease inhibitors.

The unique properties of microwave in situ heating offer unparalleled opportunities for medicinal chemists to accelerate lead optimization processes in early drug discovery. The technology is ideal for palladium-catalyzed alteration chemistry as it allows for complete control over reactions, with the use of non-inert conditions, providing high chemoselectivity and rapid feedback. To illustrate the advantages of this methodology, we describe our applications and approaches for the rapid synthesis of novel aspartyl protease inhibitors using dedicated microwave equipment. Biological results from chemical studies of the different side-chain positions of HIV-1 and malarial plasmepsin I and II protease inhibitors are summarized.

Analogues of aspartic proteases synthesized by densely covering silica gel with carboxyl groups.

Aspartic protease analogues synthesized by covering the surface of silica gel with carboxyl groups effectively hydrolyzed hemoglobin and gamma-globulin. It is proposed that the carboxyl group is involved in both complexation of the protein substrate and the catalytic cleavage of the peptide bonds of the complexed proteins.

Design, synthesis and biological activity of novel non-peptidyl endothelin converting enzyme inhibitors, 1-phenyl-tetrazole-formazan analogues.

A novel non-peptidyl endothelin converting enzyme inhibitor was obtained through a pharmacophore analysis of known inhibitors and three-dimensional structure database search. Analogues of the new inhibitor were designed using the structure-activity relationship of known inhibitors and synthesized. In anesthetized rats, intraperitoneal administration of the analogues suppressed the pressor responses induced by big endothelin-1.

Screening aspartyl proteases with combinatorial libraries.

Large numbers of pharmaceutically relevant low-molecular weight compounds can now be synthesized using combinatorial methods. Screening these large libraries of compounds requires high throughput assays. These methods are utilized to search for inhibitors of the aspartyl proteases, plasmepsin II and cathepsin D. Plasmepsin II, a protease found in the malaria parasite, hydrolyzes human hemoglobin, the nutrient source for the parasite and is a new target for anti-malaria therapy. Cathepsin D may be involved in many biological processes and inhibitors would help to clarify the role of cathepsin D in these processes. Plasmepsin II and cathepsin D are approximately 35% identical in amino acid sequence. Therefore, a comparison of the screening results of these two enzymes will be very useful in determining each enzyme's specificity and demonstrating the power of utilizing encoded combinatorial libraries.

Comparative properties of feline immunodeficiency virus (FIV) and human immunodeficiency virus type 1 (HIV-1) proteinases prepared by total chemical synthesis.

The aspartyl proteinase (PR) encoded by the feline immunodeficiency virus (FIV) was prepared by total chemical synthesis. The 116-amino-acid polypeptide chain was assembled in a stepwise fashion using a Boc chemistry solid-phase peptide synthesis approach and subsequently folded into the biologically active dimeric proteinase. The synthetic enzyme showed proteolytic activity against a variety of different peptide substrates corresponding to putative cleavage sites of the Gag and Gag-Pol polyproteins of FIV. A comparative study with the proteinase of human immunodeficiency virus type 1 (HIV-1) showed that the FIV and HIV-1 enzymes have related but distinct substrate specificities. In particular, HIV-1 PR and FIV PR each show a strong preference for their own MA/CA substrates, despite identical amino acid residues at four of seven positions from P3-P4' of the substrate including an identical MA/CA cleavage site (between Tyr approximately Pro residues). FIV PR also showed a requirement for a longer peptide substrate than HIV-1 PR. Defining the similarities and the differences in the properties of these two retroviral enzymes will have a significant impact on structure-based drug design.

Application of the multiple antigenic peptides (MAP) strategy to the production of prohormone convertases antibodies: synthesis, characterization and use of 8-branched immunogenic peptides.

Antiserum against an N-terminal sequence of murine prohormone convertase-1 (mPC1) incorporating the sequence immediately following the junction between the putative pro-region and the active enzyme was obtained. This was accomplished using the multiple antigenic peptide (MAP) approach whereupon an 8-branched polylysine core to which are grafted multiple copies of a 16 amino acid peptide representing the N-terminal sequence of mPC1 (positions 84-99) was synthesized by solid-phase Fmoc chemistry. The ensuing peptide was purified and fully characterized by RP-HPLC, 1H-NMR, amino acid composition, peptide sequencing and ion-spray mass spectrometry. The immunological properties of the resulting antibodies in detecting recombinant PC1 in both crude and purified preparations were compared with antibodies raised against a similar N-terminal segment of PC1 but using the conventional method of peptide-carrier protein conjugation and also developed against a C-terminal fusion protein of PC1. Our data indicate that the MAP antibody was as efficient as both the amino and carboxy-terminal antibodies in qualitative as well as quantitative analysis of PC1 encoded protein by radioimmunoassay. Following an identical approach, antibodies against other prohormone convertases like furin, PC5/6 and PACE4 were also developed and subsequently applied to a number of biochemical and immunological studies. In each case, the case of preparation and high immunogenicity of the MAP approach were confirmed and reside in the simplicity and rapidity with which a potent and useful antiserum is obtained.

Solid phase synthesis of the proteinase of bovine leukemia virus. Comparison of its specificity to that of HIV-2 proteinase.

The 126-residue proteinase (PR) of bovine leukemia virus (BLV) was synthesized by solid-phase peptide synthesis and its activity was shown using various oligopeptide substrates representing cleavage sites in BLV, human T-cell leukemia virus type 1 (HTLV-1), murine leukemia virus (MuLV) and human immunodeficiency virus type 1 (HIV-1). The specificity of the BLV PR was also compared to that of chemically synthesized human immunodeficiency virus type 2 (HIV-2) PR. Many of the peptides were cleaved at the expected site, however, 6 out of 15 were hydrolyzed only by one of the PRs. Furthermore, one BLV peptide was processed differently by the two enzymes. These results, together with the relative activities and the lack of inhibition of BLV PR by two HIV-1 PR inhibitors, suggest that the BLV PR specificity is substantially different from that of HIV PRs.