11720 J. Am. Chem. Soc. 1996, 118, 11720-11724 Copying Nature’s Mechanism for the Decarboxylation of â-Keto Acids into Catalytic Antibodies by Reactive Immunization Robert Bjo1rnestedt, Guofu Zhong, Richard A. Lerner,* and Carlos F. Barbas III* Contribution from The Skaggs Institute for Chemical Biology and the Department of Molecular Biology, The Scripps Research Institute, 10666 North Torrey Pines Road, La Jolla, California 92037 ReceiVed June 20, 1996X Abstract: Reactive immunization was used to generate catalytic antibodies that use the enamine mechanism common to the natural class I aldolase enzymes. In order to investigate the possibility of exploiting the imine and enamine intermediates programmed into antibody catalysts by reactive immunization and the features which antibody aldolases share with naturally evolved catalysts, we have studied their ability to catalyze the decarboxylation of structurally related â-keto acids. Both aldolase antibodies were shown to efficiently catalyze the decarboxylation of two hapten- related â-keto acids with rate enhancements (kcat/kuncat) between 4959 and 14 774. Inhibition studies support the role of an essential lysine residue in the active site of the antibodies and the formation of a cyanide accessible imine intermediate in the mechanism. Investigation of the decarboxylation reaction of 2-{3′-(4′′-acetamidophenyl)propyl}- acetoacetic acid, 4, to 6-(4′-acetamidophenyl)-2-hexanone, 12, in the presence of 18O-labeled water by electrospray mass spectrometry revealed obligatory incorporation of 18O in the antibody-catalyzed reaction consistent with decarboxylation proceeding via an imine intermediate. These studies demonstrate that reactive immunization may be utilized to program in fine detail the mechanism of catalytic antibodies and our ability to exploit the programmed reaction coordinate for different catalytic tasks. Introduction nism of these aldolase antibodies was programmed using reactive immunization3 with the â-diketone hapten 1 as shown The study of amine-catalyzed decarboxylations of â-keto in Scheme 1. The â-diketone 1 is a chemical trap. Reaction acids occupies a special place in the history of bioorganic of an -amino group of lysine within the active site of the chemistry and enzymology. In a series of elegant studies antibody with a keto group of 1 results in the formation of a Westheimer and colleagues elucidated in detail the mechanism tetrahedral carbinolamine, which is dehydrated to an imine and by which the enzyme acetoacetate decarboxylase catalyzes the subsequently tautomerized to the stable vinylogous amide 3 decarboxylation of acetoacetic acid.1 These studies demon- shown in Scheme 1. The imine and enamine intermediates strated that the reaction proceeds by the formation of a Schiff developed along the reaction coordinate of the reaction between base between the -amino group of a lysine residue in the the antibody and the diketone and the aldol reaction are also enzyme and acetoacetate, followed by decarboxylation to form found along the reaction coordinates of other reactions, for an enamine which is then tautomerized to a Schiff base and example, decarboxylations, racemizations, and alkylation reac- finally hydrolyzed to release acetone and the free enzyme. The tions to name a few. Indeed, it is known that some naturally enzymatic mechanism is analogous to that originally proposed occurring class I aldolases are bifunctional. In addition to by Pederson in 1934 for simple amine-catalyzed decarboxyla- catalyzing the aldol reaction, these bifunctional catalysts also tions.2 Herein we report the programming of the chemical catalyze the decarboxylation of â-keto acids.5 In order to mechanism used by the natural enzyme acetoacetate decarboxy- investigate the possibility of exploiting common intermediates lase into catalytic antibodies by the process of reactive im- programmed into antibody catalysts by reactive immunization munization. and the features which antibody aldolases share with naturally Previously, antibodies capable of catalyzing intermolecular evolved catalysts, we have studied their ability to catalyze the aldol reactions with control of stereochemistry in both Cram decarboxylation of structurally related â-keto acids. and anti-Cram directions were produced.3a These catalytic antibodies were shown to use the enamine mechanism common Results and Discussion to the natural class I aldolase enzymes.4 The reaction mecha- X Abstract published in AdVance ACS Abstracts, November 15, 1996. Two antibody aldolases, 38C2 and 33F12, were available for (1) The following constitutes a partial listing of the most relevant study.3a,6 These antibodies catalyze the retroaldol reaction of references: (a) Hamilton, G. A.; Westheimer, F. H. J. Am. Chem. Soc. 1959, 81, 6332. (b) Fridovich, I.; Westheimer, F. H. J. Am. Chem. Soc. 1962, 84, (3) (a) Wagner, J.; Lerner, R. A.; Barbas, C. F., III Science 1995, 270, 3208. (c) Laursen, R. A.; Westheimer, F. H. J. Am. Chem. Soc. 1966, 88, 1797. (b) Wirshing, P.; Ashley, J. A.; Lo, C. -H. L.; Janda, K. D.; Lerner, 3426. (d) Tagaki, W.; Guthrie, J. P.; Westheimer, F. H. Biochemistry 1968, R. A. Science 1995, 270, 1775. 7, 905. (e) Frey, P. A.; Kokesh, F. C.; Westheimer, F. H. Biochemistry (4) Rutter, W. J. Fed. Proc. Am. Soc. Exp. Biol. 1964, 23, 1248. Morris, 1971, 14, 7266. (f) Kokesh, F. C.; Westheimer, F. H. J. Am. Chem. Soc. A. J.; Tolan, D. R. Biochemistry 1994, 33, 12291. 1971, 93, 7270. (g) Schmidt, Jr., D. E.; Westheimer, F. H. Biochemistry (5) Kobes, R. D.; Dekker, E. E. Biochem. Biophys. Res. Commun. 1967, 1971, 10, 1249. (h) Autor, A. P.; Fridovich, I. J. Biol. Chem. 1970, 245, 27, 607. Nishihara, H.; Dekker, E. E. J. Biol. Chem. 1972, 247, 5079. 5214. (i) Westheimer, F. H. Proceedings of the Robert A. Welch Foundation Vlahos, C. J.; Dekker, E. E. J. Biol. Chem. 1986, 261, 11049. Conferences on Chemical Research; Robert A. Welch Foundation, Houston, (6) Nucleic acid sequencing of the genes coding for these antibodies 1971; Vol. 15, pp 7-50. (j) Westheimer, F. H. Tetrahedron 1995, 51,3. has revealed that they are somatic mutants of one another and differ by 18 (2) Pederson, K. J. J. Phys. Chem. 1934, 38, 559. amino acids in their variable regions. S0002-7863(96)02079-3 CCC: $12.00 © 1996 American Chemical Society Decarboxylation of â-Keto Acids J. Am. Chem. Soc., Vol. 118, No. 47, 1996 11721 Scheme 1 in the aldol reaction. Other antibodies which bound hapten 1 with high affinity albeit in a noncovalent fashion were also studied and were found not to catalyze the decarboxylation reactions. The current study differs from decarboxylations of activated model compounds catalyzed previously10 with antibodies in studies designed to elucidate medium effects in enzyme- catalyzed reactions. Our goal is to create an enzyme that uses a mechanism common to nature’s catalyst, thereby testing the ability of the experimenter to program a detailed catalytic mechanism into antibodies. To test whether this was achieved, the mechanism by which antibodies 38C2 and 33F12 operate was studied further. Inhibition of decarboxylation activity, as well as aldolase activity,3a was complete in the presence of the 4-hydroxy-6-phenyl-2-hexanone but not 2-hydroxy-6-phenyl- hapten 2 or 2,4-pentanedione when spectrophotometric titration 4-hexanone, suggesting preferential attack of the lysine at the indicated the formation of a single vinylogous amide species 2 position. This reactivity and the hapten structure suggested per active site. This is analogous to the inhibition of the enzyme 1d 4 would be a suitable â-keto acid substrate if 38C2 and 33F12 acetoacetate decarboxylase by acetopyruvate. Additionally, could tolerate substitution at the 3 position of the hapten. In incubation of antibody 38C2 and 4 in the presence of 1% v/v order to map the geometry of the active sites of these antibodies, acetone and 50 µM cyanide resulted in 88% inhibition of we studied their reactions with a series of structurally related decarboxylase activity. Activity decreases of 23% and 15% â-diketones. Both antibodies 38C2 and 33F12, shown to be were observed when the study was performed with the addition capable of reacting with the linear diketones 2 and 2,4- of either acetone or cyanide alone. The synergistic effect of pentanedione,3a also reacted with the branched diketones cyanide and acetone suggests that inhibition is due to attack of 3-methyl-2,4-pentanedione, 2-acetylcyclopentanone, and 2-ace- the acetone-antibody imine by cyanide to form a covalent tylcyclohexanone to form stable vinylogous amides which aminonitrile adduct. These data are similar to those reported for acetoacetate decarboxylase and the bifunctional enzyme exhibited intense ultraviolet absorptions, λmax, at 328, 335, and 1h,5 335 nm, respectively, in a spectral region clear of absorption 2-keto-4-hydroxyglutarate aldolase. Cyanide alone inhibits from protein. These experiments suggested an active site the antibody-catalyzed decarboxylation reaction in a concentra- geometry which tolerates considerable substitution at the 3 tion dependent manner, indicating that imine 15 is accessible position of the hapten molecule. On the basis of these to attack by cyanide. These experiments confirm that decar- experiments, the â-keto acids 4 and 5 were synthesized. boxylation and aldol reactions occur at the same site and are The lithium salt 11 of compound 4 was prepared in six steps dependent on an essential -amino group of a lysine residue in from 4-iodoaniline as shown in Scheme 2. Acetylation of the