WO2007140471A2 - Ribosome and rna display of biologically active small molecules - Google Patents

Abstract

The present invention relates to Ribosome and RNA Display of Biologically Active Small Molecules ('RRDBASM'), a technology which utilizes engineered aminoacyl- tRNAs together with diverse mRNA libraries to produce small molecule ligands for biologically interesting molecules. The present invention provides for methods and compositions that may be used to produce biologically active peptidomimetic compounds with enhanced properties relative to their natural peptide counterparts.

Description

RIBOSOME AND RNA DISPLAY OF BIOLOGICALLY ACTIVE SMALL MOLECULES

SPECIFICATION CROSS REFERENCE TO RELATED APPLICATIONS The present application claims the benefit of International Patent Application

No. PCT/US2006/037905, filed September 27, 2006 (of which it is a continuation-in-part) as well as U.S. Provisional Patent Application Serial No. 60/809,766 filed May 31, 2006 the entire contents of both of which are incorporated herein by reference.

1. INTRODUCTION

The present invention relates to Ribosome and RNA Display of Biologically Active Small Molecules ("RRDBASM"), a technology which utilizes engineered aminoacyl- tRNAs together with diverse mRNA libraries to produce small molecule ligands for biologically interesting molecules. The present invention provides for methods and compositions that may be used to produce biologically active peptidomimetic compounds with enhanced properties relative to their natural peptide counterparts.

2. BACKGROUND OF THE INVENTION

Small molecule synthesis arguably is the underpinning of the modern pharmaceutical industry. Complex natural products and their derivatives are routinely synthesized both for drug development and academic research. (Rivkin et al., 2005 Angew. Chem. Int. Ed., 44: 2838-2850; Evans et al., 2003, Actualite Chimique, 35-38; Corey, 2004, J. Org. Chem. 69: 2917-2919; Nicolaou et al., 2004, Proc. Natl. Acad. Sci. U.S.A. 101: 11929-11936; Weinreb, 2001, Nature 4U : 429-431; Stork, 1980, Nature 284: 383-384; Nicolaou et al., 2000, Angew. Chem. Int. Ed. 39_i 44-122; Peterson et al., 2004, Proc. Natl. Acad. Sci. U.S.A. IOJJ. 11943-11948; Tse, et al., 2004, Chem. Biol. I k_. 1607-1617).

For the most part, natural products are synthesized in a convergent, linear fashion using organic chemistry. The Boger group reported the total synthesis of the cyclic heptadepsipeptide HUN-7293, a natural product that inhibits cell adhesion and has anti- inflammatory properties, in 1999 (Boger et al., 1999, J. Am. Chem. Soc. 121: 6197-6205). The cyclic heptadepsipeptide was synthesized in sixteen steps with a 7.8% overall yield, counting the longest linear sequence, via Mitsunobu esterification and macrocyclization of a tetrapeptide and tripeptide.

Organic synthesis is now routinely carried out on the solid phase, as well as in solution, allowing libraries of small molecules on the order of 10⁴- 10 to be assembled. (Soth et al., 1997, Curr Opin Chem Biol j_: 120-129; Bunin et al., 1992, J. Am. Chem. Soc. 114: 10997-10998). Recently there has been interest in developing alternative approaches to small molecule synthesis. These approaches include regio- and enantio-selective catalysis, DNA- templated synthesis, and engineering of the polyketide synthase (PKS) and nonribosomal peptide synthase (NRPS) biosynthetic machineries. The long-term goal of these efforts is to devise entirely new strategies for small molecule synthesis that allow complex natural products to be synthesized in just a few steps in high yield. In the area of asymmetric catalysis, Miller and co-workers have used peptide libraries to develop asymmetric catalysts that can phosphorylate inositol regio- and enantio-selectively, allowing for the synthesis of these important biological signaling molecules in just a few steps in high yield (Sculimbrene et al., 2004, J. Am. Chem. Soc. 126_i 13182-13183). MacMillan and co-workers recently reported a concise synthesis of differentially protected hexoses in high yield and stereochemical purity based on aldol coupling of three aldehyde units. (Northrup et al., 2004, Science 305: 1752-1755).

Inspired by Nature's genetic code, both Liu and Harbury are developing in vitro methods for DNA encoding of small molecule synthesis. Liu and co-workers recently reported DNA-encoded synthesis of a macrocyclic fumaramide library. (Gartner et al., 2001, J. Am. Chem. Soc. 123: 6961-6963; Gartner et al., 2004, Science 305: 1601: Kaori et al., 2005, J. Am. Chem. Soc. 127: 1660-1661). Harbury and co-workers were able to select [Leu]enkephalin peptides based on their ability to bind to the 3-E7 antibody from a 10⁶- member peptide library. (Halpin et al., 2004, PLoS Biol. 2: E173; Halpin et al., 2004, PLoS Biol 2: E174; Halpin et al., 2004, PLoS Biol. 2: E175).

Several other laboratories are working towards the engineering of modular assembly-line polyketide and non-ribosomal peptide synthetases for controlled production of novel natural product analogs. The modular nature of both the PKSs and the NRPSs has suggested that these biosynthetic pathways could be engineered for the synthesis of a wide variety of natural products. (Khosla et al, 2003, Nat. Rev. Drug. Discov. 2: 1019-1025; Clardy et al., 2004, Nature 432: 829-837; Finking et al., 2004, Annu Rev Microbiol 58: 453- 488; Sieber et al., 2005, Chem. Rev. 105: 715-738; Liu et al, 2002, Science 297_i 1170- 1173). The erythromycin biosynthetic cluster has been functionally rewired for synthesis of ring-expanded analogs of erythromycin (Jacobsen et al., 1997, Science 277: 367-369), and Walsh's group has exploited the relaxed substrate specificity of the tyrocidine thioesterase macrocyclization catalyst to generate a 168-member cyclic peptide library from peptide precursors synthesized by solid phase methods (Jacobsen et al., 1997, Science 277: 367-369).

Still in their infancy, one can only speculate at this point as to the relative merits of these different approaches to small molecule synthesis. It would likely take some time before any of these methods could rival current linear syntheses for the routine synthesis of a broad range of small molecule structures, and in the end these methods likely will be best suited to the synthesis of a particular class of small molecules (e.g. PKSs for polyketides). An analogy may be drawn to solid-phase peptide and oligonucleotide synthesis (Kaiser et al., 1984, Science 226_i 1151-1153; Gait, 1991, Curr. Opin. Biotechnol. 2: 61-68). It took decades for both to be reduced to practice. There were many technical hurdles that had to be overcome along the way. But in the end these solid-phase syntheses revolutionized the way peptides and oligonucleotides were synthesized, making both molecules readily available to the research community and hence opening up entirely new research areas (arguably molecular biology in the case of oligonucleotides).

Peptidomimetics are an important class of natural products (Ripka et al., 1998, Curr. Opin. Chem. Biol. 2j_.441-452). As evidenced by non-ribosomal peptide natural products such as bacitracin and bleomycin, relatively simple side chain substitutions and post-translational modifications can render peptides suitable even as therapeutics (Walsh, 2004, Science 303: 1805-1810). Current approaches to the synthesis of libraries of peptidomimetics largely rely on solid-phase chemical synthesis. For example, solid-phase chemical syntheses have been developed for N-alkyl glycines ("peptoids")(Simon et al.,

1992, Proc. Natl. Acad. Sci. U.S.A. 89j_.9367-9371), β-peptides (Hamuro et al., 1999, J. Am. Chem. Soc. 121: 12200-12201; Arvidsson et al., 2001, Chembiochem 2: 771-773; Porter et al., 2000 Nature 404: 565; Cheng et al, 2001, Chem. Rev. 1OJU 3219-3232), and oligocarbamates (Cho et al., 1993, Science 26Jj_. 1303-1305; Cho et al., 1998, J. Am. Chem. Soc. 120: 7706-7718). Using strategies similar to those originally developed for peptide libraries, it has been shown that peptoid ligands for GPCR (Zuckermann et al., 1994, J. Med. Chem. 37; 2678-2685) and oligocarbamate ligands for thrombin (Cho, et al., 1999, Bioorg. Med. Chem. 7, 1171-1179) can be isolated from libraries of these oligomers. Even simply using a pool of amino acids with expanded sidechain diversity has improved ligand discovery. (Soth et al, 1997, Curr Opin Chem Biol Ii 120-129; Bunin et al, 1992, J. Am. Chem. Soc. IH: 10997-10998; Sculimbrene et al., 2004, J. Am. Chem. Soc. 126: 13182- 13183; Northrup et al., 2004, Science 305: 1752-1755; Gartner et al., 2001, J. Am. Chem. Soc. 123: 6961-6963; Gartner, et al., 2004, Science 305_i 1601; Kaori et al., 2005, J. Am. Chem. Soc. 127: 1660-1661; Halpin et al., 2004, PLoS BioL 2j_. E173; Halpin et al., 2004, PLoS Biol 2ά E174; Halpin et al., 2004, PLoS Biol. 2: E175; Khosla et al., 2003, Nat. Rev. Drug. Discov. 2i 1019-1025; Clardy et al., 2004, Nature 432i 829-837; Finking et al., 2004, Annu Rev Microbiol 58i 453-488; Sieber et al., 2005, Chem. Rev. 105: 715-738; Liu et al., 2002, Science 297: 1170-1173; Jacobsen et al., 1997, Science 277: 367-369; Kaiser, 1984, Science 226: 1151-1153; Gait, 1991, Curr. Opin. Biotechnol. 2: 61-68; Ripka, et al., 1998, Curr. Opin. Chem. Biol. 2: 441-452; Walsh, 2004, Science 303: 1805-1810; Simon et a!.. 1992, Proc. Natl. Acad. Sci. USA 89: 9367-9371; Hamuro et al., 1999, J. Am. Chem. Soc. 121 : 12200-12201; Arvidsson et al., 2001, Chembiochem 2: 771-773; Porter et al., 2000, Nature 404: 565; Cheng et al., 2001, Chem. Rev. 1OJU 3219-3232; Cho et al., 1993, Science 2611_.1303-1305; Cho et al., 1998, J. Am. Chem. Soc. 120: 7706-7718; Zuckermann et al., 1994, J. Med. Chem. 37: 2678-2685; Cho et al., 1999, Bioorg. Med. Chem. 7: 1171-1179; Kirshenbaum et al., 1999, Curr. Opin. Struct. Biol. 9i 530-535).

While these methods have already been incorporated into both academic and pharmaceutical efforts for drug discovery, a limitation to these methods, particularly for the initial stages of lead discovery, remains the modest library sizes available. To deconvolute these libraries at the end of the screening process requires that a tag be introduced as each monomer unit is added to the small molecule. While several clever pooling strategies have been reported, in practice what this requires is that each member of the library be synthesized in a separate compartment, allowing libraries only of a size on the order of 10⁴- 10⁶.

The present invention provides for the preparation of peptide and peptomimetic libraries by methods which utilize the Ribosomal Biosynthetic Machinery ("RBM") and non-natural aminoacyl tRNAs. More than thirty years ago, Chapeville et al. proved the adaptor function of the tRNA molecule by demonstrating that reduction of Cys- tRNACys to Ala-tRNACys with Raney/Ni gave a hybrid tRNA that incorporated Ala in response to a Cys codon (Chapeville et al., 1962, Proc. Natl. Acad. Sci. U.S.A. 48: 1086- 1092). Studies on the specificity of EF-Tu, the protein factor that delivers the aminoacyl- tRNA to the ribosome, also suggested that the protein biosynthetic machinery could recognize substrates beyond the 20 "natural" L-α-amino acids (Chladek et al., 1985, Angew. Chem. Int. Ed. Engl. 24: 371-391; Faulhammer et al., 1987, FEBS Lett. 217: 203-211). Then Johnson, Brunner, and others showed that modified aminoacyl-tRNAs, such as Nε- acetyl Lys-tRNALys, could be used as substrates by the ribosome and associated factors (Johnson et al., 1976, Biochemistry JJK 569-575; Baldini et al., 1988, Biochemistry 27: 7951- 7959). A significant breakthrough in the field was the introduction of a chemoenzymatic approach for synthesizing aminoacyl-tRNAs that allowed a wide variety of aminoacyl-tRNAs to be synthesized and tested for ribosome catalyzed peptide and protein synthesis (Heckler et al., 1984, Biochemistry 23_i 1468- 1473; Robertson et al., 1991, J. Am. Chem. Soc. UIi 2722-2729).

Over 100 different acyl-tRNAs have been synthesized by a number of different laboratories (Cornish et al, 1995, Angew. Chem. Int. Ed. Engl. 34: 621-633; Mendel et al., 1995, Annu. Rev. Biophys. Biomol. Struct. 24: 435-462; Thorson et al., 1998, Methods. MoI. Biol. 77: 43-73; Dougherty, 2000, Curr. Opin. Chem. Biol. 4: 645- 652; Sisido et al., 2002, Curr. Opin. Chem. Biol. 6_i 809-815; Dedkova et al., 2003, J. Am. Chem. Soc. 125; 6616-6617; Gilmore et al., 1999, Topics Curr. Chem. 202; 77-99). These synthetic aminoacyl-tRNAs have been used for a variety of applications, including studies of the mechanism of translation (Hecht, 1992, Ace. Chem. Res. 25: 545-552) and incorporation of photoaffinity labels and fluorescent probes (Johnson et al., 1982, J. MoL Biol. 156: 113-140; Abrahamson et al., 1985, Biochemistry 24: 692-700; Janiak et al., 1990, Biochemistry 29: 4268-4277).

Synthetic suppressor tRNAs have been used to incorporate amino acids with altered pKa values, constrained steric conformations, and other properties for studies of enzyme mechanism and protein stability (Cornish et al., 1995, Angew. Chem. Int. Ed. Engl. 34: 621-633; Gilmore et al., 1999, Topics Curr. Chem. 202: 77-99). It has been shown that this approach can be used in Xenopus oocytes, allowing studies of membrane proteins (Nowak et al., 1995, Science 268: 439-442). A suppressor tRNA has been used to incorporate biotin-Lys in ribosome display (Li et al., 2002, J. Am. Chem. Soc. 124: 9972- 9973). Recent work has sought to extend this approach to the simultaneous incorporation of two or three different analogs using four base-pair codons, rare codons, or synthetic base pairs (Bain et al., 1992, Nature 356: 537-539; Hirao et al., 2002, Nat. Biotechnol. 2Oj_. 177- 182; Anderson et al., 2002, J. Am. Chem. Soc. 124; 9674-9675; Hohsaka,, et al., 1996, J. Am. Chem. Soc. JJ8: 9778-9779; Anderson, et al., 2002, Chem. Biol. 9: 237-244). Aminoacyl-tRNA synthetases have been engineered that can charge the suppressor tRNA in vivo, making it possible to incorporate the synthetic amino acid in vivo and generate the modified protein in high yield (Wang et al., 2001, Science 292; 498-500; Wang, et al., 2005, Angew. Chem. Int. Ed. 44; 34-66). Similarly, cells have long been fed unnatural amino acids for tRNA charging and incorporation in place of their natural amino acid counterpart (Link et al., 2003, Curr. Opin. Biotechnol. 14: 603-609).

A limited number of groups of researchers have published reports of unnatural oligomer synthesis by the RBM: (1) Fahnestock and Rich's 1971 publication where they showed in vitro synthesis of polyesters using tRNA^Phe misacylated with hydroxy-Phe (Fahnestock, et al., 1971, Science 173: 340-343); (2) Roberts' group reported translation of poly-(GUA)_n (where n = 1, 5 or 10) templates using a synthetic acyl-tRNA charged with N- methyl-Phe (Frankel et al., 2003, Chemistry & Biology 10: 1043-105) (3) Cornish's group synthesized a tripeptide containing three different side chain amino acid analogs in a row, each in response to a different natural codon (Forster et al., 2003, Proc. Natl. Acad. Sci. U.S.A. 100: 6353-6357) and (4) Hecht reported that a 5OS ribosome mutant can incorporate D-amino acids (Dedkova et al., 2003, J. Am. Chem. Soc. 125: 6616-6617).

There remains a need to produce small molecule libraries which are sufficiently diverse as to be likely to contain a compound with desired activity as well as pharmacologic properties such as stability, selectivity, etc.. One approach that researchers have taken is to utilize the ability of RNA translation to produce small molecule libraries. Prior to the present invention, the efficient incorporation of amino acid analogs using ribosome and RNA display had been problematic, so that the desired levels of diversity have been elusive.

3. SUMMARY OF THE INVENTION

The present invention relates to Ribosome and RNA Display of Biologically Active Small Molecules ("RRDBASM"), a technology which utilizes engineered aminoacyl- tRNAs together with diverse mRNA libraries to produce small molecule ligands for biologically interesting molecules. RRDBASM allows the generation of extremely large libraries (estimated >10¹⁵) but is not limited to incorporation of natural amino acids. Extending the range of monomeric building blocks that can be used by RRDBASM, according to the invention, makes it an even more powerful drug-discovery technique. The present invention provides for the use of amino acid analogs which, incorporated into a peptide or peptide mimic, confer desirable characteristics, such as enhanced activity, stability, bioavailability, immunogenicity, and/or other therapeutically beneficial property. In preferred non-limiting embodiments, the analogs of the invention undergo, either spontaneously or under particular reaction conditions, a post-translational reaction to produce an inter-residue linkage with enhanced stability.

In non-limiting embodiments, the present invention provides for methods for producing ligands, either directly or as part of a library subjected to screening, for ligand binding partners of interest, including, but not limited to, proteins with SH3 domains, XIAP, components of the Notch pathway, tubulin, and HIV and Rce 1 proteases. The present invention further provides for compositions comprising novel amino acid analogs.

In specific, non-limiting embodiments, the present invention provides for incorporation, by RRDBASM, of beta-hydroxy amino acid analogs or beta-thio-amino acid analogs into ligands of interest which may, for example, be useful in the treatment of diseases such as Alzheimer's disease, HIV infection, hypertension or malaria. RRDBASM, in systems comprising such analogs, may be used to synthesize libraries of small molecules containing β-esters as final products or as intermediates that may rearrange into β-hydroxy γ or δ peptides. Rearrangements leading to β-hydroxy-γ peptides allows the synthesis of products in the statin class.

4. BRIEF DESCRIPTION QF THE DRAWINGS

FIGURE 1. The ribosome is nature's machinery for translating mRNA templates into peptides. Only the aniinoacyl-tRNA with the cognate anticodon binds to the codon on the mRNA template and subsequently leads to the formation of the new peptide bond. According to the invention, the ribosomal machinery is co-opted for synthesis of peptidomimetic libraries by feeding synthetic acyl-tRNA substrates to a purified translation system. In the pure translation display cycle translation products binding to ligand binding partner may be selected from the size of a library having more than 10¹ ' members. In the figure, an amino acid analog is shown being incorporated into a peptide. Translation components (here, derived from E. colϊ) are initiation factors IFl, IF2, and IF3; elongation factors EF-Tu, EF-Ts, and EF-G; ribosome and natural aa-tRNAs. SD is a Shine-Delgarno ribosome binding sequence. FIGURE 2. Pure Translation Display, an extension of ribosome display. DNA is in vitro transcribed and translated using the purified translation system. From the library of ribosome complexes (peptidomimetic translation product, ribosome and associated mRNA) active peptidomimetics are selected through binding to an immobilized ligand binding partner, here, the BIR3 domain of XIAP. The mRNA of biologically active translation products is amplified by RT-PCR.

FIGURE 3. N-alkyl peptide synthesis using thia-Pro analogs.

FIGURE 4. De nuovo designed genetic code, where compounds 1-8 are α- hydroxy amino acid analogs (referred to herein as compounds (1-8)HA, respectively.) FIGURE 5. Synthetic scheme for the synthesis of aminoacyl-tRNA. Reagents include (a) i)6-Nitroveratrylcarbonyl chloride (NvoCl), Na₂CO₃, dioxan/water 1:1 (Robertson et al, 1991, J. Am. Chem. Soc. 113:2722-2729) ii)NvoCL, triethylamine, THF; iii) Chloroacetonitril, triethylamine; (b) pdCpA, DMF, ii) tRNA^"CA, T$ RNA ligase; iii) hv (Ellman et al, 1991, Methods Enzymol. 202:301-336). FIGURE 6A-B. Close-up of EF-Tu residues interacting with aa-tRNA. (A)

(Left panel): EF-Tu from T. aquaticus with yeast Phe-tRNAPhe and the GTP analog GDPNP (Nissen et al., 1995, Science 270: 1464-1472). EF-Tu is rendered in a blue ribbon, and the aminoacylated tRNA is in CPK colors. The rectangular black box encompasses the aminoacylated acceptor stem. (B) (Right panel): enlargement of the boxed region, after rotation of approximately 90 degrees toward the viewer. The view is centered on the phenylalanine residue carried by the tRNA. The aminoacyl tRNA is in orange, His 67 of EF- Tu is magenta, and other residues that contact the acceptor stem are in blue. Labeled residues are targets for the mutagenic screen summarized in the text.

FIGURE 7. N-Alkyl pure ribosome display library design. FIGURE 8. A new genetic code.

FIGURE 9. Central biochemical events involved in transducing and modulating Notch signals. Key steps represented in the figure are discussed in the text.

FIGURE 10. Incorporation of α-hydroxy acids.

FIGURE 11. Mechanism of intramolecular rearrangement. After acylation of the α-hydroxy function the nucleophilic attach of the free amino function onto the ester carbonyl is facilitated by formation of a five-membered ring. The posttranslational rearrangement leads to the desired α-hydroxy-β-peptide backbone.

FIGURE 12. The aspartyl protease catalyzed protein hydrolysis of substrate (structure 11) . Analogs of the proposed transition state T (structure 12) of the HIV protease catalyzed hydrolysis of the peptide bond are potential protease inhibitors (Rl and R2 = aromatic or hydrophobic).

FIGURE 13A-E. The compounds A-D contain an α-hydroxy-β-amino acids substructure that can act as analogs of the transition state T of the aspartyl protease catalyzed protein hydrolysis (FIGURE 12) (De Clercq, 1995, J. Med. Chem. 38: 2491; Boehme et al., 1995, Ann. Rep. Med. Chem. 30: 139; Thaisrivongs, 1994, Annu. Rep. Med. Chem. 29: 133; Huff, 1991, J. Med. Chem. 34: 2305; Kempf, 1996, Curr. Pharm. Des. 2: 225). E. An example of an α-hydroxy-β-amino acid-containing peptidomimetic compound that may be synthesized by RRDBASM, with acyl-tRNAs carrying components of the structure shown at the bottom of the figure, prior to joining the analogs at the ribosome. FIGURE 14A-J. A is a general structure for 34 related α-hydroxy-β-amino acids. Compared to (2S, 3S)-allophenylnorstatine C (FIGURE 13) the α-hydroxy-β-amino acids B-D have bigger aromatic substituents. In the α-hydroxy-β-amino acids E, F and J the aromatic structure is connected by a longer spacer to the α-hydroxy-β-amino acid core and the α-hydroxy-β-amino acids G-J have an additional hydroxyl function. FIGURE 15. α-hydroxy compounds of the invention.

FIGURE 16. α-thio compounds of the invention.

FIGURE 17A- C. A. General structures of the β-hydroxy-γ-amino acids or β- thio-γ-amino acids (17), the β-hydroxy-δ-amino or β-thio-δ-amino acids (18) and the β- hydroxy acids (19), where X can be OH or SH. B. A non-limiting example of a β-hydroxy-γ- amino acid of interest (20). C. Examples of β-hydroxy-amino acids that are found in bioactive small molecules (20, 36-38).

FIGURE 18A-B. A. The natural product pepstatin A (39) is a statin that contains two β-hydroxy-γ-amino acid (20) structures (see legend FIGURE 17B-C). B. The general structure of the two bis-statins (34 and 35), the core of which contains the β-hydroxy- γ-amino core of structures 20 and 36 (see legend FIGURE 17C). FIGURE 19. Peptidomimetic 40, which contains the β-hydroxy-γ-amino 37 (see legend FIGURE 17C) as the active core.

FIGURE 20. The β-hydroxy acid 38 is found as a substructure of a metabolite 41 of FK228.

5. DETAILED DESCRIPTION OF THE INVENTION

For clarity of description and not by way of limitation, the detailed description of the invention is divided into the following subsections:

(i) amino acid analogs; (ii) artificial aminoacyl tRNAs;

(iii) cell-free translation systems;

(iv) preparing compounds using RRDBASM; and

(v) ligand binding partner targets of special interest.

5.1 AMINO ACID ANALOGS

An "amino acid analog" as that term is used herein is a molecule having general formula 1 or 6:

which is not one of the naturally occurring amino acids. An amino acid analog may or may not share other structural features with one of the naturally occurring nucleic acids. In formula 1 or 6, R may be a non-naturally occurring side chain with or without reactive functional groups and biological activity. Reactive functional groups may be used for post- translational modifications. In non-limiting embodiments, X in formula 6 may be either OH (α-hydroxy acids) or SH (α-thio acids).

In various embodiments, the present invention provides for the use of particular classes of amino acid analogs, including N-alkyl analogs, proline analogs (including aza-, oxo- and thia-proline analogs) and α-hydroxy analogs (including α-hydroxy- β-amino analogs).

N-Alkyl analogs include, but are not limited to, the following, referred to as compounds (21-27)N-Alk, respectively:

Proline analogs according to the invention may be represented by the following general formulas 2 and 3:

In formulas 2 and 3, X may be N (for aza-proline analogs), O (for oxo-proline analogs) or S (for thia-proline analogs). In formulas 2 and 3, each of Ri - R₈ may be, for example and not by way of limitation, H, methyl, dimethyl, (C₁ - C₄)alkyl, aryl (e.g., phenyl or substituted phenyl), hydroxy(Ci - C₄)alkyl, carboxyl(Cj - C₄)alkyl, or ammo(C_! - C₄)alkyl. In specific, non-limiting embodiments of the invention, Rj and R₂ (for formula 2) or R₇ and R₈ (for formula 3) may be the same or different and may be selected from the group consisting of H, methyl, dimethyl, (Ci - C₄)alkyl, aryl (e.g., phenyl or substituted phenyl), hydroxy(Cj - C₄)alkyl, carboxyl(Ci - C₄)alkyl, or amino(Ci - C₄)alkyl, where in a non-limiting subset of embodiments at least one of the pair Ri and R₂ (for formula 2) or the pair R₇ and R₈ (for formula 3) is H. A general scheme for N-alkyl peptide synthesis using thia-proline analogs is shown in FIGURE 3.

Thia-proline analogs include, but are not limited to, the following specific compounds, referred to as compounds (28-33)Thia-Pro, respectively:

In additional non-limiting embodiments, the present invention provides for analogs of (28-33) Thia-Pro in which the S in the ring is substituted with either N, to produce analogous compounds (28-33)Aza-Pro, or O, to produce analogous compounds (28-33)Oxo-Pro.

Of note, post-translational reduction of a thiaproline having formula 2 may yield formula 4 or 5 (in the context of a peptide or peptoid), where Rj - R₈ may be as indicated for formulas 2 and 3:

Thia-proline analogs, incorporated into a peptide or peptoid, may be postranslationally reduced, for example, using Raney/Ni hydrogenation. Advantageously, post-translational rearrangement of α-hydroxy-β-amino acids may produce a α-hydroxy-β-peptide backbone, having enhanced stability, as shown in FIGURE 11.

In further non-limiting embodiments of the invention, the amino acid analog comprises an α-hydroxy acid moiety, e.g., as in formula 6:

where X is either OH (α-hydroxy acids) or SH (α-thio acids). Ester or thio ester hydrolysis leads to site specific backbone cleavage where the α-hydroxy acid or α-thio acid is incorporated. In certain non-limiting embodiments, the R group of formula 6 may be varied so that the resulting analog is a α-hydroxy-β-amino acid, an α-hydroxy-γ-amino acid, an α- thio-β-amino acid, or an α-thio-γ-amino acid. In non-limiting embodiments, where the analog is an α-hydroxy-β-amino acid, R may be aryl(Ci - C₄)alkylamino, hydroxy(Cj - C₄)alkylamino, carboxy(Ci - C₄)alkylamino, (C₁ - C₄)alkoxy(Cj - C₄)alkylamino, amino(Ci - C₄)alkoxy(C] - C₄)alkylamino, aryl(C] - C₄)alkoxy(Ci - C₄)alkylamino, phenyl(Ci - C₄)alkoxy(Ci - C₄)alkylamino, or naphthyl(Ci - C₄)alkoxy(Ci - C₄)alkylamino.

In various non-limiting embodiments, the present invention provides for compounds having the structural formulas 8 (α-hydroxy-β-amino acids), 9 (α-hydroxy-γ- amino acids) or 10 (respectively) as follows:

where X = HO or HS. PG is a protective group that needs to be present as long as the α- hydroxy(or -thio)-γ-amino acids 9 is on the tRNA. Here R₁-R₆ can be part of cyclic or acyclic structures. Each OfR₁ - R₆ may be, for example and not by way of limitation, H, methyl, dimethyl, (C₁ - C₄)alkyl, aryl (e.g., phenyl or substituted phenyl), hydroxy(Ci - C₄)alkyl, carboxyl(Ci - C₄)alkyl, or amino(Ci - C₄)alkyl.

In specific non-limiting embodiments, the invention provides for the use of hydroxy acids numbered 1 -8 in FIGURE 4, referred to hereafter, respectively, as compounds (1-8)HA, as well as hydroxy acids depicted in FIGURES 13A-D and 14A-J.

In one specific, non-limiting embodiment, the invention provides for the use of alpha hydroxy O-methyl serine. Other specific, non-limiting examples of α-hydroxy and α-thio compounds which may be used according to the invention are depicted, respectively, in FIGURES 15 and l6. In further non-limiting embodiments, the invention provides for the use of β- hydroxy- and β-thio amino acid analogs. General structures are shown in FIGURE 17A of the β-hydroxy-γ-amino acids or β-thio-γ-amino acids (17), the β-hydroxy-δ-amino or β-thio- δ-amino acids (18) and the β-hydroxy-acids (19), where X can be OH or SH. R₁ - Rg of these general formulas may be any of the natural amino acid side chains or may be unnatural side chains with or without reactive functional groups. Reactive functional groups, if present, may be used for post-translational modifications. A non-limiting example of a β-hydroxy-γ- amino acid of interest (20) is shown in FIGURE 17B.

In particular, non-limiting embodiments, such β-hydroxy or β-thio amino acid analogs may be used, in RRDBASM, to prepare molecules that simulate natural structures. Examples of β-hydroxy-amino acids that are found in bioactive small molecules (20, 36-38) are shown in FIGURE 17C. As one specific non-limiting example, RRDBASM operating on such analogs may be used to simulate molecules of the statin class, which contain a β- hydroxy-γ-amino structure. FIGURE 18A depicts the natural product pepstatin A, a statin which contains two β-hydroxy-γ-amino structures. FIGURE 18B shows two bis-statins (structures 34 and 35) which contain the β-hydroxy-γ-amino core of structures 20 and 36 (see FIGURE 17C). Structure 34 is currently in commercial development, and acts as a human β- secretase (BACE-I) inhibitor. Modification of this compound (structure 35) resulted in a decrease of the IC ₅₀ for BACE-I inhibition from 0.5 μM for structure 34 to 21 nM for structure 35. RRDBASM, using the β-hydroxy or β-thio amino acid analogs of the invention, may be used to provide candidates which may be screened for even greater potencies. Of note, structure 34 would be well-suited for RRDBASM initiating translation with AcMet- tRNAi. In a second subset of particular non-limiting embodiments, β-hydroxy- and/or β-thio amino acid analogs may be used to produce analogs of other compounds of biological interest. For example, the peptidomimetic compound 40, shown in FIGURE 19, is a potent inhibitor of plasmepsin 1 having a Ki of 0.5 nM, and therefore a candidate agent for treating malaria. RJRDBASM using β-hydroxy and/or β-thio amino acid analogs may be used to produce analogs of compound 40 with improved activity.

In a third subset of particular non-limiting embodiments, β-hydroxy and/or β- thio amino acid analogs may be used in RRDBASM to produce analogs of a metabolite of the HDAC inhibitor FK228 (FIGURE 20) that is reported to be an antitumor agent by Chan and Xiao, International Patent Application Publication No. WO 2007040522.

In a fourth subset of particular non-limiting embodiments, β-hydroxy and/or β- thio amino acid analogs may be used in RRDBASM to produce modulators of an Asp protease.

In a fifth subset of particular non-limiting embodiments, β-hydroxy and/or β- thio amino acid analogs may be modified by introducing a side chain using N-alkylation (for example, operating on R₆ or Rg of structures 17 or 18 of FIGURE 17A. used in RRDBASM to produce analogs of a metabolite of the HDAC inhibitor FK228 (FIGURE 20).

In a sixth subsert of particular non-limiting embodiments, the present invention provides for the use of β-hydroxy analogs of the α-hydroxy acids shown FIGURES 13A-D, 14A-J, 15 or 16 where, for each compound a carbon atom, substituted or unsub- stituted, is inserted between the carbon bearing the hydroxyl group and the carbonyl group.

5.2 ARTIFICIAL AMINOACYL tRNAs

Artificial aminoacyl tRNAs may be prepared by linking an amino acid analog, as described in the preceding section, to a tRNA using any method known in the art.

In a non-limiting embodiment of the invention, an artificial tRNA may be prepared as follows, using a chemoenzymatic method as set forth in Heckler, et al., 1984, Biochemistry 23:1468-1473 and later modified by Noren et al., 1989, Science 244, 182-188. This method takes advantage of the fact that all tRNAs end in an invariant CCA-3". A transcription and then purified by precipitation or gel electrophoresis. An aminoacyl-pdCpA dinucleotide may be synthesized and then ligated to the tRNA-CA using T4 RNA ligase. The aminoacyl-pdCpA dinucleotide may then be made by acylating pdCpA with the cyanomethyl active ester of the amino acid. The dinucleotide may be synthesized using standard nucleotide chemistry. Briefly, 6-N, 6-N, 2"-O, 3"-O-tetrabenzoyl adenosine may be prepared by transiently protecting the 5" hydroxyl of adenosine with dimethoxytrityl. The protected adenosine may then be coupled to the 2"-deoxycytidinylphosphoramidite and oxidized under standard conditions. A phosphoramidite may then be added to the 5" hydroxyl group of deoxycytidine and subsequently oxidized. Finally, the benzoyl and cyanoethyl protecting groups are removed under basic conditions, and the pdCpA may be purified by HPLC and "activated" as the tetrabutylammonium salt. A photolabile protecting group may be used for the amino acid because it can be removed after coupling of the aa- pdCpA to the tRNA-CA. The amino acid then may be prepared as the N- nitroveratryloxycarbonyl cyanomethyl active ester. The NVOC protecting group may be installed under standard conditions, and then the cyanomethyl group may be introduced using chloroacetonitrile and triethylamine as the base. The cyanomethyl active ester may then be used to selectively acylate pdCpA at the 2 "/3" hydroxyl group (the two rapidly interconvert at room temperature) using the tetrabutylammonium salt of pdCpA. The aa-pdCpA may be purified using HPLC and then coupled to tRNA-CA enzymatically and photodeprotected. In specific non-limiting embodiments, synthesis of 5mg of an aa-pdCpA may be sufficient for at least 100 translation reactions.

In specific non-limiting embodiments, the present invention provides for compositions comprising an aminoacyl tRNA, said aminoacyl tRNA further comprising, (and, in a translational sense, "charged with") one of the amino acid analogs described herein. One specific, non-limiting embodiment of a synthetic scheme for the synthesis of amino-acyl tRNA is shown in FIGURE 5.

Further methods for preparing aminoacyl tRNAs include those published in Murakam et al., 2006, Nat Methods 3_i 357-9; Seebeck and Szostak, 2006, J Am Chem Soc 128: 7150-1 ; Hartman et al., 2006, Proc Natl Acad Sci U. S. A. 103_i 4356-61 ; and Josephson et al., 2005, J Am Chem Soc 127: 11727-35. 5.3 CELL-FREE TRANSLATION SYSTEMS

FIGURE 1 schematically depicts a cell-free translation system. Any suitable cell-free translation system known in the art may be used. It is desirable to control the populations of various tRNAs so that incorporation of a synthetic aminoacyl tRNA carrying an amino acid analog is not substantially inhibited by the presence of acyl-tRNAs competing for the same codon.

In preferred embodiments, the cell-free translation system of Forster et al., 2003, Proc. Natl. Acad. Sci. U.S.A. 100:6353 may be used, as follows. This system utilizes ribosomes purified exhaustively to remove measurable contaminating RS charging activities, recombinant translation factors (Forster et al., 2001, Anal. Biochem. 297:60—70), in vitro- synthesized mRNAs, in vzYro-charged native tRNA isoacceptors, and chemoenzymatically synthesized aatRNAs. Briefly, mRNAs and translation mixes may be prepared as described in Forster et al., 2001, Anal. Biochem. 297:60-70 , except that (i) neither polyethylene glycol nor His-tagged EF-Ts (Hwang, et al., 1997, Arch. Biochem. Biophys. 348: 157-162) are added to the translation components, (ii) initiation factor (IF)2 may be further purified by gelfiltration chromatography, and (iii) the ribosomes may be subjected to additional washing, in that an additional high-speed spin of 1 min may precede the final pelleting of the four- times-washed ribosomes to remove residual insoluble material. Ribosomes and factors desirably are not contaminated with RSs or proteases, as measured by charging of total tRNA (Sigma) with 15 14C-labeled amino acids (New England Nuclear) and by stability of peptides. Macromolecular concentrations in translations may be adjusted slightly to give 0.5 μM each of IFl, IF2, IF3, EF-G, and EF-Ts, 2.5 μM EF-Tu, four-times-washed ribosomes at 0.029 A260 unit/μl [27 nM estimated to be active], 1 μM mRNA, 0.2 μM , and 0.5 μM for each elongator aa-tRNA. Translations may be performed at 37°C for 30 min without preincubation. Translations may then be analyzed, for example by cation-exchange

(treatment with alkali, acidification, then minichromatography to separate anionic formylated peptides from unformylated amino acids).

Alternatively, cell-free translations as described in United States Patent No. 6,977,150 by Forster orUnited States Patent Application Publication No. 2002/0123101 by Inoue, et al. may be used.

In particular non-limiting embodiments, the present invention provides for cell-free translation systems in which the amount or type of EF-Tu is modified relatively to conventionally used systems. Without being bound to any particular theory, it is hypothesized that linear iV-alkyl amino acids are excluded from peptides/peptoids synthesized in purified cell-free translation systems largely due to EF-Tu preferentially binding to natural amino acids. A number of approaches may be used to compensate for any such binding disadvantage manifested by an amino acid analog. First, the relative concentrations of EF-Tu and aminoacyl-tRNA substrates may be modulated. A simple competitive inhibitor model for analog-tRNA binding to EF-Tu in the presence of a pool of natural aa-tRNA substrates ("competitive inhibitors") shows that even modest decreases in EF-Tu affinity would significantly affect EF-Tu occupancy. For example, where EF-Tu and the analog-tRNA and aa-tRNA substrates are all present at ca. 1 μM, because the KDs of the natural aa-tRNA substrates for EF-Tu»GTP are ca. InM, for an analog-tRNA impaired only two orders of magnitude in its KD for EF-TuOTP, the effective KD is:

K^e _D ^ff = KD • • (1 + il^Ml) = 10OnM • (1 + 10³) = lOOμM

where the natural aa-tRNAs are the competitive inhibitors I with KI = KD for EF-Tu binding. With an effective KD well above both the concentrations of both EF-Tu and the analog- tRNA, <1% of the analog-tRNA would be bound by EF-Tu. This simple competitive inhibition model predicts that analogs impaired only one or two orders of magnitude in EF- Tu binding can be "fixed" simply by using EF-Tu in excess or optimizing the relative concentrations of EF-Tu and the natural and analog-tRNA substrates in the translation mixture based on their relative KDs. In non-limiting embodiments, the concentration of EF- Tu may be increased by a factor of at least 10, at least 50, at least 100, or at least 1000.

In alternative approaches, EF-Tu may be genetically engineered to promote binding. Both rational design and screening may be used to redesign the substrate specificity of EF-Tu. There are two high-resolution structures of EF-Tu»GTP^#aa-tRNA ternary complexes.106, 108 These structures paint a consistent picture of the residues that line the amino acid binding pocket (Fig. 13, Note that the residue numbers are from T. aquaticus, and the corresponding residues from E. coli Tu are: H66 for H67, E215 for E226, F218 for F229, F262 in place of H273, N273 for N285, and V274 for V286. The N285 residue is conserved in greater than 98% of EF-Tu sequences, emphasizing its importance). Cassette mutagenesis may be used to randomize each active-site residue independently. Both for library generation and subsequent characterization, EF-Tu may be purified and activated. Then, pools of EF-Tu mutants may be screened, for example in a 96-well plate, using a fluorescent GTP analog (mant-GTP) to score ribosome-dependent GTPase activation of EF- Tu as a change in fluorescence.115 The fluorescence GTPase assay is more readily adapted to a 96-well plate format and eliminates EF-Tu mutants with improved binding, but defective on the ribosome. In specific non-limiting embodiments, the present invention provides for a variant of E. coli EF-Tu which is at least 90 percent homologous to NCBI Accession No. AAC76954, comprises a mutation at one or more {e.g., 2, 3, 4, 5 or 6) of the following residues: H66, E215, F218, F262, N273, or V274, and is a functional elongation factor.

As a further alternative, EF-Tu from another strain, species, or genus may be used. As particular examples, an EF-Tu from a halophilic or thermophilic bacteria may be used, including, but not limited to, Thermus thermophilus, Bacillus stearothermophilus, and Thermus thermophilus HB8.

5.4 PREPARING COMPOUNDS USING RRDBASM

For preparation of a library of small molecules that comprise amino acid residues, which may be naturally occurring amino acids or amino acid analogs, ribosome and RNA display technology may be used. Such techniques are described in, for example, Forster et al, 2003, Proc. Natl. Acad. Sci. U.S.A. 100:6353-6357; Tan et al., 2004, J. Am. Chem. Soc. 126:12752-12753; and United States Patent No. 6,977,150. Additional references may be found in the list of references provided below. Schematic diagrams of ribosome and RNA display technologies are presented in FIGURES 1 and 2.

In certain embodiments, the present invention may be used to synthesize a particular compound of interest. The compound may be a peptide or a peptidomimetic. A peptide, as defined herein, is a molecule comprised of natural amino acids where the backbone or main chain of the molecule consists of units joined by peptide bonds. A peptidomimetic as defined herein is a molecule comprised of amino acids and/or amino acid analogs, where an amino acid with an unnatural side chain is incorporated and/or the backbone or main chain of the molecule comprises at least one bond that is not a peptide bond or where the peptide bond is N-substituted. For example, but not by way of limitation, such a bond may be an ester bond. In non-limiting embodiments of the invention, the peptidomimetic may be a polyester (see below). In non-limiting embodiments of the invention, the peptide or peptidomimetic may be between about 2 and 100, or between about 2 and 80, or between about 2 and 60, or between about 2 and 40, or about 2 and 20, or about 2 and 10, or about 2 and 5, residues in length.

The present invention may be used to generate libraries of small molecule ligands that are amenable to ribosome synthesis. "Ligand," as that term is used herein, refers to a compound that is able to bind to a biologically active molecule, its "ligand binding partner," where the relationship between ligand and ligand binding partner is such that there is a binding affinity between them; that is to say, the relationship may be enzyme/substrate; hormone/receptor; antigen/antibody; etc.

Therefore, for example, a ligand of interest, which may be a naturally occurring or a synthetic molecule, may be used as a model for which small molecule analogs may be developed according to the invention. Non-limiting examples of ligands for which such analogs may be developed include kynostatins, tubulin polymerization inhibitors such as hemiasterlin, talbotulins (e.g., HTI-286), belamide A, or dolastatins.

Alternatively, the ligand binding partner may be used as the basis for identifying members of a small molecule library that are suitable modulators (for example, by screening a pure translational display system using ligand binding partner bound to a solid phase (e.g., a substrate, matrix, plate or bead) to select suitable candidates which are later amplified by PCR (directed evolution)). Suitable ligand binding partner targets include, but are not limited to, Human Immunodeficiency Virus (HIV) protease, Rce 1 protease, the Crk SH3 Domain, and IAP proteins.

The present invention provides for methods of producing diverse libraries of small molecule ligands for a ligand binding partner of interest. One source of diversity may be the mRNA directing the translational synthesis of the small molecules, such that the coding sequence of said RNA may contain one or more residue which is varied relative to a sequence encoding a ligand of interest. Methods of the invention encompass the use of a single RNA species or a plurality of RNA species (a library), where the latter can increase diversity by orders of magnitude.

A second source of diversity may be the use of multiple amino acid analogs. Such multiple amino acid analogs may be linked to a single species of tRNA or to a plurality of species of tRNA. In specific non-limiting embodiments of the invention, the redundancy of the genetic code (and the availability of more than 20 possible codons) may be utilized to expand the number of possible amino acids which may be incorporated into a translation product.

For illustration, as a specific, non-limiting example, aminoacyl tRNAs bound to the natural amino acids and to various analogs may be utilized in a de nuovo genetic code, as shown in FIGURE 4.

The assignment of a particular analog to be carried by different tRNAs may result in different levels of incorporation of said analog, thereby determining its representation in any peptide or peptidomimetic library produced. Furthermore, different concentrations of competitor RNAs corresponding to the same codon but charged with different analogs can further affect library composition.

FIGURE 2 schematically depicts a rationale by which peptide or peptidomimetic ligands generated according to the invention may be allowed to bind to their ligand binding partner while still associated with the ribosome and encoding mRNA, thereby allowing for the selection of RNAs which encode bindable ligands; said selected RNAs may be amplified by PCR and then allowed to pass through one or more "round(s)" of selection.

5.5 LIGAND BINDING PARTNER TARGETS OF SPECIAL INTEREST

5.5.1. SH3 DOMAINS AS LIGAND BINDING PARTNERS

SH3 domains typically bind peptide ligands that adopt a polyproline type II helical conformation, with proline residues strongly favored at the P-I and P2 positions (FIGURE 7). The peptide recognition surface has two pockets at the P-I and P2 positions that accommodate the structural property of proline residues that is unique among the 20 natural amino acids: N-substitution (Nguyen et al., 1998, Science 282: 2088-2092). Lim and colleagues exploited this feature to identify three Crk ligands with higher affinity than the natural peptide sequence among a mere 22 -sequence peptide/peptoid library with N- substituted residues in either the P-I or P2 site in the peptide sequence (Nguyen et al., 2000, Chem. Biol. Jl 463-473 )

The N-substitution preference of SH3 domains may be exploited in the design of a library by using building blocks enriched in N-alkyl amino acids and Pro analogs. The library may contain 16 AsnB tRNAs, with GNN anticodons assigned to 16 different building blocks (see FIGURE 8 for an enlarged genetic code), to read a modified genetic code constructed from the 16 NNC codons. The mRNAs in the library may encode the peptidomimetic sequence MPxPxxPRxx (FIGURE 7), where sites of variation in the library are defined by the x and include the P-I and P2 positions. Note (FIGURE 8) that the 16-letter alphabet may be enriched in N-alkyl amino acids, but not restricted to them, in order to maximize the structural complexity of the library. The alphabet may also be designed to include a mix of hydrophobic, polar, and charged amino acid analogs to ensure peptide solubility. Peptides/peptoids from the library may then be selected by ribosome display (FIGURE 2) for binding to the Crk SH3 domain, immobilized as a GST fusion protein on glutathione-agarose beads (or, alternatively, a biotin-modified Crk SH3 domain-ref 84).137 After the binding step, the mRNA associated with the bound peptides may be recovered by addition of EDTA (which dissociates the ribosomal subunits and releases the mRNA), then amplified by RT- PCR. This cycle of translation, selection, recovery and amplification may be repeated for additional cycles until the library converges, with the identity of a set number (e.g., a dozen, 20, etc.) clones determined after each round by DNA sequencing.

5.5.2. NOTCH AS LIGAND BINDING PARTNER

Drosophila Notch and its homologs in humans and other multicellular animals define a unique class of highly conserved transmembrane receptors that normally regulate cell growth, differentiation, and death in a variety of tissue types. In general, the consequences of Notch signaling vary as a function of dose and context. Activation of Notch can favor choice of one cell fate over another, promote cell proliferation or cell cycle arrest, cause differentiation or self-renewal, and enhance survival or apoptosis (Artavanis-Tsakonas et al., 1999, Science 284; 770-6; Weng et al, 2004, Curr Opin Genet Dev J_4j_.48-54). Notch 1 , one of four Notch homologues in mammals, is normally required at several stages in the development and maturation of T-cells. Evidence implicating Notch 1 in the choice of T cell lineage commitment comes from both gain and loss-of function experiments in mice (Pear et al., 2003, Semin Immunol JJK 69-79). Expression of constitutively active hNl in hematopoietic stem cells inhibits normal marrow B cell development and induces the development of CD4+CD8+ double positive (DP) immature T- cells in the bone marrow (Pui et al., 1999, Immunity Hi 299-308). Conversely, inducible notchl knockout mice fail to develop mature T cells due to a requirement for Notchl during early stages of intrathymic T cell development (Radtke et al, 1999, Immunity Kk 547-58).

On the other hand, constitutive, unrestrained Notchl signaling is associated with T cell acute lymphocytic leukemia/lymphoma (T-ALL), both in primary human tumors (Ellisen et al., 1991, Cell 66: 649-61) and in mouse models (Pear et al., 1996, J. Exp. Med. 183: 2283-91; Aster et al., 2000, MoI Cell Biol 20: 7505-15). Activation or inactivation of Notch signaling has also been linked to a variety of other cancers, including skin, breast, lung, pancreas, and CNS tumors (Axelson, 2004, Semin Cancer Biol Ui 317-9). Because of the broad importance of Notch in differentiation and proliferation, interventions that prevent Notch receptors signaling may not only lead to new forms of treatment for T-ALL, but manipulation of Notchl activity may also be of general value in management of other cancers.

See FIGURE 9, a schematic diagram of Notch-associated cell signals. Normally, a Notch signal is activated when ligand binding induces proteolytic cleavages that release the intracellular portion of Notch (ICN) from the membrane, permitting it to translocate to the nucleus, where it turns on transcription of target genes (Lai, 2004, Development 131 : 965-73; Hansson et al., 2004, Semin Cancer Biol Hi 320-8). Once unleashed from the membrane, Notch orchestrates transcriptional activation of target genes by forming a nuclear complex that includes the transcription factor CSL, ICN, and a co- activator protein of the mastermind-like family (MAML). Among the transcriptional targets of Notch in T-cells are the bHLH protein HES-I, the Notch modifier Deltex-1, and the pre-T cell receptor alpha subunit, but there is considerable debate about what other genes are direct transcriptional targets of Notchl in T-cells. Delivery of a dominant-negative form of MAML-I arrests T-ALL tumor cell lines dependent on Notchl for growth, indicating that the activity of nuclear Notchl complexes is not only important for T-cell development, but also important for tumor cell proliferation.

According to the invention, a desirable ligand binding partner to use for selection is the CSL transcription factor, which is the only known effector of activated Notch proteins. Several reasons justify this choice. First, a 13-residue peptide derived from the intracellular part of Notch (RRQHGQL WFPEGF (SEQ ID NO: I)) suffices to bind the CSL transcription factor with high affinity (Kovall et al., 2004, Embo J 23_i 3441-51), indicating that selection of a library-encoded peptidomimetic that competes with Notch for CSL binding should be possible. Second, the selected ligand (or a suitable related one) may be a valuable reagent for probing the CSL-dependent consequences of Notch activation. Third, activating mutations of human Notch 1 are found in more than 50 % of acute T-cell lymphocytic leukemias (T-ALL), and T-ALL cell lines with such activating mutations undergo growth arrest when Notch activity is blocked, indicating that Notch activation plays a central role in the molecular pathogenesis in T-ALL.

For selection of CSL ligands, complexes between CSL and DNA duplexes that contain a 5"-biotinylated nucleotide on one strand may be captured onto avidin-coated dishes. As one specific non-limiting example, for selection of a CSL ligand, a 10-residue library constructed from the 16 NNC codons may be used, with cycles of translation, selection, recovery and amplification repeated for 10 cycles or until the library converges, with the identity of 20 clones determined after each round by DNA sequencing. The ability of the selected ligands to bind to the CSL-DNA complex may then be evaluated by fluorescence polarization or by titration calorimetry.

5.5.3. XIAP AS LIGAND BINDING PARTNER

XIAP is the most potent member of the class of inhibitors of apoptosis proteins. Through binding to caspases, XIAP prevents apoptosis but the complete function of XIAP and IAPs in general is not certain. Peptidomimetics binding to IAPs can be used for the investigation of apoptosis and the elucidation of the role of IAPs. XIAP inhibitors may provide leads for the discovery of anti-cancer drugs.

X-ray analysis has shown that SMAC N-terminus binds to the BIR3 domain of XIAP with only the first four amino acid residues AVPI (Liu et al, 2000, Nature 408:1004- 1008; Wu et al., 2000, Nature 408:1008-1012; Chai et al., 2000, Nature 406:855-862) but the N-terminal pentapeptide AVPFY of the functional Smac analog HID binds four times stronger (K_d = 60 nM) to the BIR3 domain of XIAP than the corresponding Smac derived peptide. Oost and co workers found that replacement of the proline decreased the binding affinity to the XIAP BIR3 domain (Oost et al., 2004, J. Med. Chem. 47:4417-4426).

According to one particular non-limiting embodiment of the invention, a peptidomimetic library may be generated to bind to the BIR3 domain. For example, the two positions on both sides of the proline residue may be randomized and only the four amino acids A, V, F and Y may be used for these positions. This would result in a library size of 256 pentamers. A limited alphabetic code assigning the well-established AsnB tRNAs (AsnBoAu, ASΠBGGU, AsnBouu, AsnBGcu, AsnBAAc) to the amino acids A, V, F₅Y and P reading the codons AUC, ACC, AAC, AGC, and GTT may be used. The initial library may be created using PCR. The 5" primer may encompass a fixed upstream non-coding region encoding a Shine-Delgarno sequence followed by a start codon, the library insert consisting of (ANC)₂GTT(ANC)₂ (GTT is coding for P) and an 18 nt sequence complementary to the spacer poly(V/T). The 3" primer may consist of an oligonucleotide complementary to the DNA sequence that follows the cloned poly(V/T) insert. mRNA templates may be prepared by runoff transcription, and the libraries may be translated using the purified system as described in Forster et al., 2004, Analyt. Biochem. 333:358-364. The stalled ribosome- mRNA peptide ternary complexes may be as described in Hanes et al., 2000, Meth. Enzymol. 328:404-430.

For the selection step the residues 241-356 of the human XIAP BIR3 protein may be immobilized as a GST fusion protein on glutathione-agarose beads (Nguyen et al.,

1998, Science 282:2088-2092) or a biotin-modified (Hanes et al., 1997, Proc. Natl. Acad. Sci. U.S.A. 94:4937-4942) XIAP BIR3 domain. After the binding step, the mRNA may be recovered by addition of EDTA and amplified by RT-PCR. This cycle of transcription, translation, selection and recovery may be repeated until DNA sequencing that may be carried out after each round indicates that the library size is converging to a predetermined number (e.g., 20 clones). IfAVPFY is not recovered in the selection, whether or not the hits are binding to the XIAP BIR3 domain may be tested, and then whether AVPYF can be isolated, when mixed with much weaker binding AGPYF,may be tested at different molar ratios (e.g., ranging from 1/100 to 1,000,000). If a stalled ternary complex is encountered, it may be desirable to switch to a mRNA display method as set forth in Roberts et al. 1997,

Proc. Natl. Acad. Sci. U.S.A. 94:12297-12302 and Keefe and Szostak, 2001, Nature 410:715- 718.

Members of the library determined to bind to XIAP BIR3 may be synthesized in quantity by solid phase peptide synthesis. A C-terminal QSEK sequence may be attached and the lysine residue may be linked to fluorescein-5-6-carboxyamidocaproid acid N- succimidyl ester (FAM). The K_d of each fluorescently labeled peptidomimetic for binding to the human XIAP BIR3 domain may be determined using a fluorescence polarization based competitive binding assay (Nikolovska-Coleska et al., 2004, J. Med. Chem. 47:2430-2440). Activity of said peptides may be tested in a caspase-9 activation assay, an analysis of apoptosis and a cell growth inhibition assay (Nikolovska-Coleska et al., 2004, J. Med. Chem. 47:2430-2440).

5.5.4. RCE-I AS LIGAND BINDING PARTNER

Rce 1 protease may be a ligand binding partner for which small molecules are developed using the invention. For the generation of libraries resembling the C-terminal CAAX-bix which is the substrate of the RCE-I protease that is involved in the maturation of the RAS protein, the C-term that is typically linked to the RNA would need to be accessed. After post-translational cyclization by any of the foregoing methods, followed by cleavage of a backbone ester, a pseudo C-term resembling the CAAX box may be generated.

5.5.5 HIV PROTEASE AS LIGAND BINDING PARTNER The HIV protease is essential for the virus life cycle. It is responsible for the site specific cleavage of the Gag-polyprotein and the Gag/Pol-polyprotein and thereby generates the functional viral proteins and enzymes that are required for the viral maturation (Beaulieu et al., 1997, J. Med. Chem. 40: 2164; Darke and Huff, 1994, Adv. Pharmacol. 25: 399; Darke et al., 1988, Biochem. Biophys. Res. Commun. 156j_.297). The HIV protease is a target for antiviral agents in the treatment of AIDS. The HIV protease is an aspartyl protease that preferentially cleaves between aromatic amino acids and proline or between pairs of hydrophobic and aromatic amino acids (FIGURE 12) (Badelassi et al., 2002, HeIv. Chim. Acta 85: 3090; Tomasselli and Henrikson, 1994, Methods Enzymol. 2Ah 279; Beck et al., 2000, Virology 274: 391. The potent HIV-I protease inhibitor kynostatins (KNI)-227 A and (KNI)-272 B contain the same α-hydroxy-/£-amino substructure as (2S, 35)- allophenylnorstatine C (FIGURE 13). Their protease inhibition potency is based on the mimickery of the transition state T of the peptide hydrolysis (FIGURE 12)( Bunnage et al., 1994, Tetrahedron 50i 3975). The (2S, 35)-allophenylnorstatine C is a much stronger HIV protease inhibitor than its or-epimer. The α-epimer (2R, 3S)-phenylnorstatine D though is found in a variety of renin inhibitors. Screening of a library of 34 compounds (general structure shown in figure 6) by Huang and co workers showed that substitution of the (2S, 35)-allophenylnorstatine C by its naphtyl analog (FIGURE 14B) results in increased HIV protease inhibition (Huang et al, 2002, J. Med. Chem. 45; 333). A good substrate for the HIV protease is the internally quenched FRET substrate (D ABCYL)-SQNYPIVQ-(ED ANS) that contains a HIV protease cleavage site between tyrosine and proline. This octapeptide has been used for inhibition studies of HIV protease inhibitors (Matayoshi et al., 1989, Science 254: 954). Substitution of the tyrosine in the HIV protease substrate SQNYPIVQ by any of the α-hydroxy-/?-amino acids FIGURE 13C or FIGURE 14B-J might lead to peptidomimetics with selective HIV protease inhibition properties. FIGURE 13E shows a scheme for RRDBASM synthesis of an analog of an HIV protease inhibitor.

Additional non-limiting examples of the invention are set forth below.

6. EXAMPLE 1: SYNTHESIS QF POLYESTERS

The incorporation efficiency of hydroxy acids was assayed using a mRNA template coding for MVE. The incorporation yield is determined based on the Dowex assay in which the amount of ³H-labeled glutamic acid that is incorporated into the depsipeptides fM-aHa-E is compared to the formation of the tripeptide fM-V-E. For the positive control natural amino acids on fully modified tRNAs are used while the α-hydoxy acids are loaded onto the tRNAo_AC ^AsnB- This Asn-based tRNA adaptor was engineered to read the VaI codon GUU. The yields for the hydroxy acids shown in Figure 10 are 98% for aHA, 50-55% for aHOMS and 65% for aHF. The polyester fM-aHA-aHOMS-aHF-E was formed with a yield of 50-60% using the mRNA for MNTVE and the acyl-tRNAs pairings aHOMS- tRNA^AsnBGGU, aHA-tRNA^AsnBGUU and aHF-tRNA^AsnBGAC. α-hydroxy acid monomers showing high efficiency in single-site incorporation experiments were chosen. In our purified translation system (PTS) aHF and aHA were previously incorporated with 65% and 98% yield, respectively. To enhance the solubility of the final oligomer product aHOMS was selected as a third α-hydroxy acid unit.

The incorporation efficiency of hydroxy acids is assayed using a mRNA template coding for MVE. The incorporation yield is determined based on the Dowex assay in which the amount of ³H-labeled glutamic acid that is incorporated into the depsipeptides fM-aHa-E is compared to the formation of the tripeptide fM-V-E. For the positive control natural amino acids on fully modified tRNAs are used while the α-hydoxy acids are loaded onto the previously used tRNA_GAc^AsnB (PNAS/JACS/MetO5). This Asn-based tRNA adaptor was engineered to read the VaI codon GUU. Racemic aHOMS was synthesized form methyl 2-bromo-3- methoxypropionate. The ONvoc protected derivatives of aHOMS were synthesized and ligated onto tRNAs^~CA. The resulting aHOMS-tRNA_GAc^AsnB was tested for single-site incorporation into the depsipeptide fM-aHOMS-E. Based on the Dowex assay the incorporation efficiency of aHOMS is 53%.

The <7-Nvoc protected derivatives of the α-hydroxy acids were synthesized and ligated onto tRNAs.^"CA Racemic aHOMS was synthesized form methyl 2-bromo-3- methoxypropionate (SynQuest laboratories, Alachua FL). The carboxylic acid of 4- nitrobenzoic acid was activated with cesium carbonate in DMF for the nucleophilic attack onto the bromide. Saponification with 1 M aqueous lithium hydroxide gave the deprotected aHOMS. The 0-Nvoc protection of the hydroxy function was carried out in dry THF using triethylamine as base. The carboxylic acid was activated as cyanomethyl ester using chloroacetonitrile and triethylamine. Under dry conditions five equivalents of the active ester were reacted with one equivalent of pdCpA in DMF as previously reported. The reaction was accelerated by addition of tertbutylammonium acetate.

A system was developed by constructing a mRNA with the three adjacent test codons AAC, ACC, and GUU and Asn-based tRNA adaptors engineered to read these codons. This genetic code was reassigned to the three α-hydroxy acids aHA, aHOMS and aHF. The chemoenzymatic synthesis of aHOMS-tRNA^AsnBGGU, aHA-tRNA^AsnBGUU and aHF-tRNA^AsnBGAC and the mRNA was carried out using methods known in the art. Based on the Dowex assay, the yield for the ribosomal synthesis of the polyester 1 fM-aHA- aHOMS-aHF-E was 57% compared to the formation of the tripeptide fM-V-E.

Synthetic genes were cloned to enable in vitro synthesis of tRNA^"CA species for ligation to aHa-pdCpA (aHa is used as abreviaiton for a nonspecific α-hydroxy acid). The tRNA sequences contained substitutions at their 5' and 3' termini to maintain the secondary structure of the aminoacyl stems while enabling efficient transcription initiation at the first nucleotide with GMP by T7 RNA polymerase. The <9-Nvoc-aHa-pdCpA derivatives of aHOMS, aHA and aHF were prepared and ligated to tRNA^"CA species by using general methods. Natural aa-tRNAs were prepared from pure isoacceptors or with pure recombinant RSs. The specific activity for the H-labeled glutamic acid was 8,400 dpm/pmol. mRNAs and translation mixes were prepared using published methods. Typically translations were typically performed with 1 pmol of limiting input of fMet-tRNA _fMet i (5 μL) for Dowex analysis, 4 pmol (20 μL) for HPLC analysis and at a 10 pmol (50 μL) for mass spectrometric analysis. The concentrations in translations were 0.6 μM of IFl, 0.5 μM each of IF2, IF3, EF-G, and EF-Ts, 3.6 μM EF-Tu, four-times-washed ribosomes at 0.029,4260 unit/μl [27 nM estimated to be active], 1 μM mRNA, 0.2 μM fMet-tRNAfMet _is 0.5 μM ³H-labeled E-tRNAi^GIu, and 1 μM for elongator tRNAs (photodeprotected aHa- tRNAs or aa-tRNAs, Val-tRNA^Val) or 1 μM tRNA_GAC ^AsnB"CA. Translations were performed without preincubation at 37°C for 30 min.

Product yields were determined from translations (5 μL) using the Dowex assay. Translation products were released with IM NaOH (1 μL) at 37°C for 30 min. The translation mixtures were acidified with 0.5 M HCl (0.5 niL) and loaded onto Dowex cation- exchange columns. Formylated peptides were eluted into scintillation liquid while unacylated ³H-labeled glutamic acid remains on the column. The radioactivity eluting form negative controls using truncated tRNAoAC^{AsnB" A} was defined as background and the formation of the tripeptide fM-V-E using Val-tRNA^Val as 100%.

As shown in FIGURE 10, single incorporation of the hydroxy acid (aHOMS = alpha hydroxy O-methyl serine) was 50 - 60% compared to natural VaI (defined as 100% on the MVE mRNA) and the two previously reported hydroxy acids (Lactic acid 98% and Phenyl lactic acid 65%).

Further, as shown in FIGURE 10, the polyester formation MTVE with N = lactic acid, T = aHOMS and V = Phenyl lactic acid (and reading the codon M = formyl Met and codon E = GIu) was also formed with about 50-60% yield. This yield is in the same range as the formation of a side chain modified oligomer previously reported in Forster et al., 2003, Proc. Natl. Acad. Sci. U.S.A. 100:6353-6357.

7. EXAMPLE 2

Incorporation of analogs having formula 6, where X=OH, and R=CH₃, CH₂C₆H₅ or CH₂OCH, respectively, are incorporated in yields of 100%, 65% and 53%, respectively. The data also indicates that multiple adjacent incorporation of all three analogs in a row forming the polyester depsipeptide (formula 7) in good yield (57%).

8. EXAMPLE 3

Good incorporation yields (80%) were likewise observed for formula 10, also referred to as abF 10.

AbF 10 was Nvoc protected and activated as cynomethyl active ester for acylation of pdCpA. The Resulting Nvoc-protected pdCpA was ligated with T4 RNA ligase onto the truncated (- CA) Asn-based tRNA with the anticodon GAC. Prior to translation of the Nvoc protective group was photolytically removed. The yield for single incorporation of abF was determined using the MVE mRNA. The incorporation of 3H-labeled glutamic acid for abF was compared to the incorporation of natural valine on fully modified tRNA with the same anticodon GAC. 70-80% single incorporation yield has been observed for an α-hydroxy-β- amino acid having a benzyl substituent at R₃ in formula 8 (below), with all other R groups being H. This is an active substructure of HIV protease inhibitors. Therefore, peptidomimetics according to the invention comprising this analog may be used as HIV protease inhibitors.

SM-319777 is Structure 13, SM-319777, is an example of an HIV-I protease K (Kj = 35 pM) inhibitor containing the α-hydroxy-β-amino acids abF 10 and the thiaproline analog with the general structure of formula 2 where Ri = R₂ = H and R₃ = R₄ =CH₃. SM-319777 is structurally closely related to a family of natural products called kynostatins.

13 SM-319777 9. EXAMPLE 4

The thia-Pro analog with the general structure 2 (above) with X = S, Rl = CH3 and R2-4 = H was incorporated with a yield in the range of 30-40% (without any optimization).

Combining the properties of natural amino acids and unnatural analogs such as the ones mentioned above with post-translational enzymatic and/or chemical modifications may be used to introduce cyclic substructures by cross-linking of the side chain residues or the N-terminal amino, thio- or hydroxy functions. Additional cyclic substructures could lead to more diverse libraries and rigid structures with enhanced properties.

Enzymatic removal of the N-terminal fM allows generation of a free NH₃ or NH-alkyl moiety at the N terminus which may either be of importance for biological activity or used for a post translational cyclization or chemical modification. The removal of the N- terminal fM may also be achieved by chemical methods. Treatment with cyanogen bromide cleaves at methionine or cystein residue natural amino acids and treatment with I₂ cleaves at the unnatural amino acid allylglycine. As shown below, removal of the N-terminal dipeptide fM-allylglycine results in the free N-terminus (in the ellipse) resembling belamide A (structure 15, infra).

Cyclization may be achieved using either enzymes such as thioesterases or

chemical methods. After removal of the N-terminal fM, the resulting free terminus may be reacted with other functional groups on the side chains. For example:

- macrolactonization may be achieved by enzymatic or chemical reaction of a side chain carboxylic acid, ester, or thioester with the amino function of a free N-terminus or of a side chain;

- alkenes may be used to induce cyclization between two side chain alkenes;

- a side chain carbonyl and an amine may be used to form an imine and reduction of the imine may be used to make a cyclic amine by reductive animation;

- a side chain carbonyl and a hydroxylamine may be used to form an oxime; - a halide may be reacted with a thiol, hydroxy, or amino function; - an amide may be reacted with an acetylene to form a 1,2,3-triazole and a macrocucle; and/or

- oxidative coupling of a histidine and a tryptophane residue may be used to form a macrocycle as found in the family of celogentins.

10. EXAMPLE 5

The present invention may be used to produce, either directly or by selection from a library of variants, peptide or peptidomimetic compounds resembling the following tubulin inhibitors:

14 HTI-286

15 Belarnide A.

 Various publications are cited herein, the contents of which are hereby incorporated in their entireties.

Claims

WHAT IS CLAIMED IS:

1. A peptidomimetic comprising α-hydroxy-O-methyl serine.

2. A peptidomimetic comprising AbF 10.

3. A peptidomimetic comprising a proline analog selected from the group consisting of compounds (28-33)Thia-Pro; (28-33)Aza-Pro and (28-33)Oxo-Pro.

4. A peptidomimetic comprising a compound selected from compounds depicted in FIGURE 14B-J.

5. A peptidomimetic comprising a compound depicted in FIGURE 15.

6. A peptidomimetic comprising a compound depicted in FIGURE 16.

7. An aminoacyltRNA carrying an amino acid analog which is α-hydroxy-O-methyl serine.

8. An aminoacyltRNA carrying an amino acid analog which is AbF 10.

9. An aminoacyltRNA carrying an amino acid analog which is a thiaproline analog.

10. An aminoacyltRNA carrying an amino acid analog which is an oxoproline analog.

11. An aminoacyltRNA carrying an amino acid analog which is an azaproline analog.

12. An aminoacyltRNA carrying an amino acid analog which is selected from the group consisting of compounds (28-33)Thia-Pro; (28-33)Aza-Pro and (28-33)Oxo-Pro.

13. An aminoacyltRNA carrying an amino acid analog which is selected from compounds depicted in FIGURE 14B-J.

14. An aminoacyltRNA carrying an amino acid analog which is a compound depicted in FIGURE 15.

15. An aminoacyl tRNA carrying an amino acid analog which is a compound depicted in FIGURE 16.

16. A method of identifying a ligand for a ligand binding partner of interest, comprising; (i) generating a mRNA library encoding potential ligands; (ii) contacting the mRNA library with a cell free translation system, said cell free translation system comprising an aminoacyl tRNA according to any one of claims 7-15, to form ribosome/RNA/translation product complexes; and (iii) selecting mRNA encoding a ligand for the ligand binding partner by affinity binding of translation product in the complexes to a binding site of the ligand binding partner.

17. The method of claim 16, further comprising a step of amplifying the mRNA selected in step (iii) and then repeating steps (ii) and (iii).

18. The method of claim 16, where the ligand binding partner of interest is an SH3 domain.

19. The method of claim 17, where the ligand binding partner of interest is an SH3 domain.

20. The method of claim 16, where the ligand binding partner of interest is a component of a Notch signaling pathway.

21. The method of claim 17, where the ligand binding partner of interest is a component of a Notch signaling pathway.

22. The method of claim 16, where the ligand binding partner of interest is a XIAP protein.

23. The method of claim 17, where the ligand binding partner of interest is a XIAP protein.

24. The method of claim 16, where the ligand binding partner of interest is a Rce-1 Protease.

25. The method of claim 17, where the ligand binding partner of interest is a Rce-1 Protease.

26. The method of claim 16, where the ligand binding partner of interest is a Human Immunodeficiency Virus Protease.

27. The method of claim 17, where the ligand binding partner of interest is a Human Immunodeficiency Virus Protease.

28. The method of claim 16, where the ligand binding partner of interest is a tubulin.

29. The method of claim 17, where the ligand binding partner of interest is a tubulin.

30. The method of claim 16, where the cell free translation system comprises an excess of EF-Tu.

31. The method of claim 17, where the cell free translation system comprises an excess of EF-Tu.

32. A method of preparing a peptidomimetic compound, comprising adding an aminoacyl tRNA according to any one of claims 7-15 to a cell free translation system under conditions in which synthesis of the peptidomimetic occurs.

33. The method of claim 32, where the cell free translation system comprises an excess of EF-Tu.

34. An aminoacyl-tRNA carrying an amino acid analog which is β-hydroxy amino acid analog.

35. The aminoacyl-tRNA of claim 34, where the carried analog is a β-hydroxy γ-amino amino acid analog.

36. The aminoacyl-tRNA of claim 34, where the carried analog is a β-hydroxy δ-amino amino acid analog.

37. An aminoacyl-tRNA carrying an amino acid analog which is β-thio amino acid analog.

38. The aminoacyl-tRNA of claim 37 where the carried analog is a β-thio γ-amino amino acid analog.

39. The aminoacyl-tRNA of claim 37, where the carried analog is a β-thio δ-amino amino acid analog.

40. A method of identifying a ligand for a ligand binding partner of interest, comprising; (i) generating a mRNA library encoding potential ligands; (ii) contacting the mRNA library with a cell free translation system, said cell free translation system comprising an aminoacyl tRNA according to any one of claims 34-39, to form ribosome/RNA/translation product complexes; and

(iii) selecting mRNA encoding a ligand for the ligand binding partner by affinity binding of translation product in the complexes to a binding site of the ligand binding partner.

41. The method of claim 40, further comprising a step of amplifying the mRNA selected in step (iii) and then repeating steps (ii) and (iii).

42. The method of claim 40, where the ligand binding partner of interest is BACE-I .

43. The method of claim 40, where the ligand binding partner of interest is plasmepsin II.

44. The method of claim 40, where the ligand binding partner of interest is HDAC.

45. The method of claim 40, where the ligand binding partner of interest is an aspartic acid protease.

46. The method of claim 40, where the cell free translation system comprises an excess of EF-Tu.

47. The method of claim 41 , where the cell free translation system comprises an excess of EF-Tu.

48. A method of preparing a peptidomimetic compound, comprising adding an aminoacyl tRNA according to any one of claims 34-39 to a cell free translation system under conditions in which synthesis of the peptidomimetic occurs.

49. The method of claim 48, where the cell free translation system comprises an excess of EF-Tu.