呃

In Situ Click Chemistry: Enzyme Inhibitors Made to Their Own Specifications Roman Manetsch,† Antoni Krasin´ski,† Zoran Radic´,‡ Jessica Raushel,† Palmer Taylor,‡ K. Barry Sharpless,† and Hartmuth C. Kolb*,† Contribution from the Department of Chemistry and the Skaggs Institute for Chemical Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, California 92037, and the Department of Pharmacology, UniVersity of California, San Diego, 9500 Gilman DriVe, La Jolla, California 92093 Received June 18, 2004; E-mail: hckolb@scripps.edu Abstract: The in situ click chemistry approach to lead discovery employs the biological target itself for assembling inhibitors from complementary building block reagents via irreversible connection chemistry. The present publication discusses the optimization of this target-guided strategy using acetylcholinesterase (AChE) as a test system. The application of liquid chromatography with mass spectroscopic detection in the selected ion mode for product identification greatly enhanced the sensitivity and reliability of this method. It enabled the testing of multicomponent mixtures, which may dramatically increase the in situ screening throughput. In addition to the previously reported in situ product syn-TZ2PA6, we discovered three new inhibitors, syn-TZ2PA5, syn-TA2PZ6, and syn-TA2PZ5, derived from tacrine and phenylphenanthridinium azides and acetylenes, in the reactions with Electrophorus electricus and mouse AChE. All in situ-generated compounds were extremely potent AChE inhibitors, because of the presence of multiple sites of interaction, which include the newly formed triazole nexus as a significant pharmacophore. Introduction The past decade has seen a paradigm shift in drug discovery from testing small numbers of “handcrafted” compounds and natural products to high-throughput screening of large combi- natorial libraries.1 These developments have gone hand in hand with dramatic improvements in methods for producing, handling, and screening large numbers of compounds.2-5 Despite these achievements, challenges related to the synthesis, purification, and diversity of compound libraries and the pharmacological properties of their members still exist,6,7 and combinatorial chemistry has not yet achieved its full potential.8,9 Since typically more than 99% of all compounds in a library are inactive in a given screen, methods for producing just the active compounds are highly desirable. Target-guided synthesis (TGS) seeks to address this challenge by using the target enzyme for assembling its own inhibitors from a collection of building block reagents. Only building blocks that adhere to the protein’s binding sites react with each other to form highly potent inhibitors that simultaneously access multiple binding pockets within the protein. These target-guided approaches avoid the classical screening of large compound libraries altogether, and hit identification can be as simple as determining whether a given combination of building blocks has resulted in a product. Follow-up tests for determining the inhibitory potency, bio- availability, toxicity, and the development of structure-activity relationships (SAR) can then be limited to a small number of target-generated compounds, which may dramatically improve the efficiency of the discovery process. The concept of target-guided synthesis was pioneered almost 20 years ago by Rideout et al., who observed a marked synergism between the cytotoxic effects of decanal and N- amino-guanidines, which was proposed to be due to the self- assembly of cytotoxic hydrazones inside cells.10,11 Since then, several approaches to target-guided synthesis have been ex- plored: (1) dynamic combinatorial chemistry,12-21 (2) stepwise target-guided synthesis,22,23 and (3) kinetically controlled target- † The Scripps Research Institute. ‡ University of California, San Diego. (1) Nicolaou, K. C.; Hanko, R.; Hartwig, W. In Handbook of Combinatorial Chemistry; Nicolaou, K. C., Hanko, R., Hartwig, W., Eds.; Wiley-VCH: Weinheim, Germany, 2002; Vol. 1, pp 3-9. (2) Kolb, H. C.; Sharpless, K. B. Drug DiscoVery Today 2003, 8, 1128-1137. (3) Terrett, N. Combinatorial Chemistry; Oxford University Press: Oxford, U.K., 1998. (4) Nicolaou, K. C.; Hanko, R.; Hartwig, W. Handbook of Combinatorial Chemistry; Wiley-VCH: Weinheim, Germany, 2002; Vol. 1. (5) Nicolaou, K. C.; Hanko, R.; Hartwig, W. Handbook of Combinatorial Chemistry; Wiley-VCH: Weinheim, Germany, 2002; Vol. 2. (6) Kassel, D. B.; Myers, P. L. Pharm. News 2002, 9, 171-177. (7) Geysen, H. M.; Schoenen, F.; Wagner, D.; Wagner, R. Nat. ReV. Drug DiscoVery 2003, 2, 222-230. (8) Fixing the drugs pipeline. The Economist, March 11, 2004; available online at http://www.economist.com/printedition/displayStory.cfm? Story_ID=2477075. (9) Kubinyi, H. Nat. ReV. Drug DiscoVery 2003, 2, 665-668. (10) Rideout, D. Science 1986, 233, 561-563. (11) Rideout, D.; Calogeropoulou, T.; Jaworski, J.; McCarthy, M. Biopolymers 1990, 29, 247-262. (12) Huc, I.; Lehn, J.-M. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 2106-2110. (13) Ramstrom, O.; Lehn, J.-M. ChemBioChem 2000, 1, 41-48. (14) Lehn, J.-M.; Eliseev, A. V. Science 2001, 291, 2331-2332. (15) Bunyapaiboonsri, T.; Ramstrom, O.; Lohmann, S.; Lehn, J.-M.; Peng, L.; Goeldner, M. ChemBioChem 2001, 2, 438-444. (16) Eliseev, A. V. Pharm. News 2002, 9, 207-215. (17) Ramstrom, O.; Lehn, J.-M. Nat. ReV. Drug DiscoVery 2002, 1, 26-36. (18) Otto, S. Curr. Opin. Drug DiscoVery DeV. 2003, 6, 509-520. (19) Erlanson, D. A.; Braisted, A. C.; Raphael, D. R.; Randal, M.; Stroud, R. M.; Gordon, E. M.; Wells, J. A. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 9367-9372. Published on Web 09/18/2004 10.1021/ja046382g CCC: $27.50 © 2004 American Chemical Society J. AM. CHEM. SOC. 2004, 126, 12809-12818 9 12809 guided synthesis.24-31 The dynamic combinatorial chemistry approach introduced by Lehn et al.12 relies on building blocks bearing complementary functional groups that react reversibly with each other to form a thermodynamically controlled mixture of products. In the presence of the enzyme, the equilibrium is skewed toward the compounds that show the highest affinity toward the enzyme. Their identification requires the equilibrium to be “frozen” (e.g., by hydride reduction or by lowering the pH) before analysis by HPLC or MS can be performed. The multistep variant of TGS makes only indirect use of the enzyme for inhibitor synthesis.22,23 In the first step, a library of building blocks is screened to identify candidates that bind to the enzyme. In the second step, the building blocks with the highest affinity are linked together using conventional combinatorial chemistry approaches. The library of “divalent” molecules is then screened for high affinity inhibitors using traditional assays. The kineti- cally controlled approach uses the enzyme target itself for the synthesis of inhibitors by equilibrium controlled sampling of various possible pairs of reactants until an irreversible reaction induced by the enzyme essentially connects the pair that best fits its binding pockets.24-31 Recently, several successful applications of the kinetically controlled approach to TGS have been reported. For example, Benkovic and Boger have developed multisubstrate adduct inhibitors (MAI) of the enzyme glycinamide ribonucleotide transformylase (GAR Tfase) by enzyme-templated alkylation of one of its substrates with a folate-derived electrophile.26-28 More recently, Huc described a similar approach, in which inhibitors of carbonic anhydrase were generated by alkylation of a thiol with R-chloroketones in the presence of the Zn(II) enzyme.29 Competition experiments revealed that the enzyme- templated reaction had produced mainly the alkylation product with the highest affinity for the target. Nicolaou and co-workers have utilized a target-accelerated combinatorial synthesis ap- proach to develop dimeric derivatives of vancomycin.30,31 Appropriately functionalized monomeric vancomycin derivatives were subjected to olefin metathesis or disulfide formation in the presence of vancomycin’s target, Ac-D-Ala-D-Ala or Ac2- L-Lys-D-Ala-D-Ala, resulting in the formation of highly potent dimers. The scope of most TGS methods is limited because of their use of highly reactive reagents (strong electrophiles or nucleo- philes, metathesis catalysts etc.), which can react in many “unproductive” pathways, including ones that destroy the enzyme target. In contrast, the recently developed in situ click chemistry approach to kinetically controlled TGS24 uses bio- orthogonal reactions and reagents, for example, the [1,3]-dipolar cycloaddition reaction32 between azides and acetylenes. This system is especially well-suited for TGS, since (a) the reaction is extremely slow at room temperature, despite the very high driving force that makes it irreversible, (b) it does not involve components that might disturb the binding sites (external reagents, catalysts, byproducts), and (c) the reactants are inert to biological molecules. Mock et al. had previously provided proof-of-concept by demonstrating that the azide/acetylene [1,3]- dipolar cycloaddition is accelerated by 4 to 5 orders of magnitude by the synthetic receptor cucurbituril to give exclusively the anti-triazole regioisomer.33-35 The biological target for the initial in situ click chemistry study, acetylcholinesterase (AChE), catalyzes the hydrolysis of the neurotransmitter acetylcholine and thus plays a key role in the central and peripheral nervous system.36 Its inhibitors have been employed for over a century in various therapeutic regimens and to investigate the role of acetylcholine in neurotransmission.37,38 The catalytic site of the enzyme is located at the bottom of a 20 Å deep narrow gorge. A second, peripheral binding site is positioned at the other end of this gorge, near the protein surface.39,40 A building block library of azides and acetylenes based on the known site-specific inhibitors tacrine (active site ligand) and phenylphenanthridinium (peripheral site ligand) was developed to probe whether the enzyme would combine selected pairs of complementary reagents to synthesize its “divalent” inhibitors (cf. Scheme 1).24 A series of 49 binary mixtures of these reagents was incubated with Electrophorus electricus AChE (electric eel AChE) at room temperature for 6 days, potentially giving rise to 98 products. Analysis of the crude reaction mixtures by desorption/ionization on silicon mass spectrometry41 (DIOS-MS) revealed only one product, TZ2PA6, which was shown by HPLC to be only the 1,5-disubstituted triazole (“syn-triazole”) (Scheme 1). This compound, formed by the enzyme, turned out to be the most potent noncovalent AChE inhibitor known to date, with Kd values between 77 fM (Torpedo californica) and 410 fM (murine AChE). In contrast, the anti-TZ2PA6 isomer, not formed by the enzyme, is less active by 2 orders of magnitude. Recent X-ray structures of both the syn- and anti-TZ2PA6 mouse AChE complexes confirmed the multivalent nature of the protein ligand interactions, with the tacrine moiety accessing the active center of the enzyme and the phenylphenanthridinium group the peripheral site (Figure 1).25 Interestingly, these studies (20) Erlanson, D. A.; Lam, J. W.; Wiesmann, C.; Luong, T. N.; Simmons, R. L.; DeLano, W. L.; Choong, I. C.; Burdett, M. T.; Flanagan, W. M.; Lee, D.; Gordon, E. M.; O’Brien, T. Nat. Biotechnol. 2003, 21, 308-314. (21) Ramstro¨m, O.; Lohmann, S.; Bunyapaiboonsri, T.; Lehn, J.-M. Chem.- Eur. J. 2004, 10, 1711-1715. (22) Maly, D. J.; Choong, I. C.; Ellman, J. A. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 2419-2424. (23) Kehoe, J. W.; Maly, D. J.; Verdugo, D. E.; Armstrong, J. I.; Cook, B. N.; Ouyang, Y.-B.; Moore, K. L.; Ellman, J. A.; Bertozzi, C. R. Bioorg. Med. Chem. Lett. 2002, 12, 329-332. (24) Lewis, W. G.; Green, L. G.; Grynszpan, F.; Radic, Z.; Carlier, P. R.; Taylor, P.; Finn, M. G.; Sharpless, K. B. Angew. Chem., Int. Ed. 2002, 41, 1053- 1057. (25) Bourne, Y.; Kolb, H. C.; Radicü , Z.; Sharpless, K. B.; Taylor, P.; Marchot, P. Proc. Natl. Acad. Sci. U.S.A. 2004, 101, 1449-1454. (26) Inglese, J.; Benkovic, S. J. Tetrahedron 1991, 47, 2351-2364. (27) Boger, D. L.; Haynes, N.-E.; Kitos, P. A.; Warren, M. S.; Ramcharan, J.; Marolewski, A. E.; Benkovic, S. J. Bioorgan. Med. Chem. 1997, 5, 1817- 1830. (28) Greasley, S. E.; Marsilje, T. H.; Cai, H.; Baker, S.; Benkovic, S. J.; Boger, D. L.; Wilson, I. A. Biochemistry 2001, 40, 13538-13547. (29) Nguyen, R.; Huc, I. Angew. Chem., Int. Ed. 2001, 40, 1774-1776. (30) Nicolaou, K. C.; Hughes, R.; Cho, S. Y.; Winssinger, N.; Smethurst, C.; Labischinski, H.; Endermann, R. Angew. Chem., Int. Ed. 2000, 39, 3823- 3828. (31) Nicolaou, K. C.; Hughes, R.; Cho, S. Y.; Winssinger, N.; Labischinski, H.; Endermann, R. Chem.-Eur. J. 2001, 7, 3824-3843. (32) Huisgen, R. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A., Ed.; Wiley: New York, 1984; Vol. 1, pp 1-176. (33) Mock, W. L.; Irra, T. A.; Wepsiec, J. P.; Manimaran, T. L. J. Org. Chem. 1983, 48, 3619-3620. (34) Mock, W. L.; Irra, T. A.; Wepsiec, J. P.; Adhya, M. J. Org. Chem. 1989, 54, 5302-5308. (35) Mock, W. L. Top. Curr. Chem. 1995, 175, 1-24. (36) Taylor, P.; Radic, Z. Annu. ReV. Pharmacol. Toxicol. 1994, 34, 281-320. (37) Argyl-Robertson, D. Edinburgh Med. J. 1863, 8, 815-820. (38) Dale, H. H. J. Pharmacol. Exp. Ther. 1914, 6, 147-190. (39) Sussman, J. L.; Harel, M.; Frolow, F.; Oefner, C.; Goldman, A.; Toker, L.; Silman, I. Science 1991, 253, 872-879. (40) Harel, M.; Schalk, I.; Ehret-Sabatier, L.; Bouet, F.; Goeldner, M.; Hirth, C.; Axelsen, P. H.; Silman, I.; Sussman, J. L. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 9031-9035. (41) Wei, J.; Buriak, J. M.; Siuzdak, G. Nature 1999, 399, 243-246. A R T I C L E S Manetsch et al. 12810 J. AM. CHEM. SOC. 9 VOL. 126, NO. 40, 2004 revealed that the triazole unit, created by the azide/alkyne cycloaddition, engages in hydrogen bonding and stacking interactions with amino acid residues in the wall of the gorge. Several important conclusions can be drawn from this observa- tion. First, triazoles are not just passive linkers, but rather active pharmacophores that may contribute significantly to protein binding, as in the case of the in situ-generated product, syn- TZ2PA6. Second, the tremendous rate acceleration by AChE42 is due not only to entropic effects, but also to an enthalpic stabilization of the triazole-like transition state, leading to the observed product. In a more general sense, it appears likely that an “in situ hit” is a good binder, because the same entropic and enthalpic factors that cause the observed rate acceleration may also stabilize the newly formed triazole in the complex and thus add to the overall binding interactions, which also involve the two residues that are held together by the triazole linker. The higher potency of syn-TZ2PA6 compared to the anti- isomer manifests itself in a strikingly different binding mode at the peripheral site. The phenylphenanthridinium moiety of the tightly bound syn-product inserts itself between tryptophan- 286 and tyrosine-72 residues near the gorge rim (Figure 1), causing the enzyme to adopt a minor abundance conformation in which the tryptophan residue swings out into the solvent to make room for the ligand. This conformation has never been seen before in AChE X-ray structures of the free enzyme or its inhibitor complexes. Thus, the in situ click chemistry approach allows one to identify conformations that associate with high affinity inhibitors and that would not be detected by conventional structural methods. These findings have interesting implications for drug discovery, as it is possible to trap a flexible enzyme in a minor abundance conformation by an inhibitor, which is formed inside its binding pockets through the irreversible reaction of complementary building blocks. The goal of this study was to optimize the in situ approach to drug discovery and to investigate its scope. We started by optimizing the mass spectroscopy-based analysis method, since a highly sensitive and reliable method was deemed crucial for success. We then revisited the AChE system to search for additional in situ hits from binary azide/acetylene mixtures and from multireagent mixtures (combinatorial screening), to study the species dependence of product formation, and to determine syn/anti ratios and binding affinities for all products. Results and Discussion Optimization of the Analysis Method. Poor and variable levels of purity of the acetylcholinesterase enzyme make great demands on the analytical techniques used to detect the (42) The enzyme-free reaction under these conditions ([TZ2] ) 4.6 íM.; [PA6] ) 24 íM) is extremely slow, taking several thousand years to reach 50% conversion at 18 °C (second-order rate constant at 18 °C in 1-BuOH, K ) 1.9 � 10-5 M-1 min-1 24). Apart from the expected entropic stabilization, the transition state may also experience stabilization through hydrogen bonding and stacking interactions with the protein.25 Scheme 1. In Situ Click Chemistry Screeninga a Fifty-two binary mixtures of azide and alkyne building blocks were incubated with eel AChE, as indicated by the double arrows, potentially giving rise to 104 products. Previous work was done without PZ5, potentially giving rise to 98 products from 49 reagent combinations, and providing syn-TZ2PA6 as the sole product of the in situ reaction with the enzyme.24 Enzyme Inhibitors Made to Their Own Specifications A R T I C L E S J. AM. CHEM. SOC. 9 VOL. 126, NO. 40, 2004 12811 formation of templated inhibitors, especially when they are tightly bound. In previous experiments, the DIOS-MS method was found to be capable of detecting small quantities of new molecules in the presence of large amounts of protein, but signal- to-noise ratios were still very low. This issue has now been resolved by purifying the samples before MS analysis using standard LC/MS techniques with selected ion monitoring to increase sensitivity even further. The analysis is extremely easy to perform, allowing crude reaction mixtures to be screened and products to be unambiguously identified by their molecular weights and retention times. The new analytical method was validated on the known in situ hit TZ2PA6. After incubating the building blocks TZ2 and PA6 with eel AChE for 6 h, analysis by LC/MS-SIM gave a distinct product signal with a characteristic molecular weight and retention time (Figure 2). Thus, the high sensitivity of this analysis method allowed us to reduce the incubation time from 6 days to as little as 6 h, thereby significantly enhancing the efficiency of lead discovery by in situ click chemistry.43 Control experiments, in which mixtures of the same building blocks were incubated in the presence of bovine serum albumin (BSA) instead of AChE, or in the absence of any protein, failed to give detectable amounts of triazole. In Situ Lead Discovery. Encouraged by these results, we decided to revisit the AChE system using a library of tacrine and phenylphenanthridinium building blocks (“T-P library”), which contained one additional member, PZ5, compared to previous work, and to screen for additional in situ hits with the more sensitive LC/MS-SIM method. In the “in situ screening mode”, potential hits are identified by looking for significant differences between the chromatograms of the enzyme reactions and the control reactions (BSA in place of AChE, absence of any protein). The potential hits are then validated by additional control experiments (e.g., performing the enzyme reaction in the presence of a known active site inhibitor) and eventually by comparing retention times with synthetic samples (cf. Table 3). This screening procedure led to the identification of three new hit compoundssTZ2PA5, TA2PZ6, and TA2PZ5sin addition to the known hit, TZ