A light-activated antibody catalyst

Article abstract of DOI:10.1021/ja982711r

A catalytic antibody for a multistep Norrish type II photochemical reaction was investigated. Absorption of light energy by α-ketoamide substrate 1b produced a high-energy biradical intermediate, that was then directed by the antibody microenvironment to form tetrahydropyrazine 13 with a k(cat) of 1.4 x 10^-3 min^-1 at 280 nm irradiation and an enantiomeric excess of 78%. Antibody-catalyzed reactions performed with radiolabeled substrate indicated that little self-inactivation (6.8 mol % covalent modification after four turnovers per antibody) occurred. The singular product obtained in the antibody-catalyzed reaction was not observed in the uncatalyzed reaction unless the pH was lowered below 4. Studies suggested that the interplay of conformational control and chemical catalysis were responsible for the high specificity. A change in protonation state of the antibody was correlated with the inclusion of a new reaction pathway in the antibody-catalyzed reaction, indicating that general-base catalysis was involved in the rerouting of the Norrish reaction to form 13. An X-ray crystal structure of the substrate was obtained and suggested that the antibody binds the α-ketoamide in a twisted conformation optimal for the first step of the photochemical reaction. The antibody described here is a model for the evolution of light-activated enzymes and can serve as a foundation for the development of light-dependent antibody catalysts for a range of even more complex photochemical reactions.

Full text of DOI:10.1021/ja982711r

J. Am. Chem. Soc. 1998, 120, 12783-12790
12783
A Light-Activated Antibody Catalyst
Matthew J. Taylor, Timothy Z. Hoffman, Jari T. Yli-Kauhaluoma,
Richard A. Lerner,* and Kim D. Janda*
Contribution from the The Scripps Research Institute, Department of Chemistry and The Skaggs Institute
for Chemical Biology, 10550 North Torrey Pines Road, La Jolla, California, 92037
ReceiVed July 30, 1998
Abstract: A catalytic antibody for a multistep Norrish type II photochemical reaction was investigated.
Absorption of light energy by R-ketoamide substrate 1b produced a high-energy biradical intermediate, that
was then directed by the antibody microenvironment to form tetrahydropyrazine 13 with a kcat of 1.4 × 10-3
min-1 at 280 nm irradiation and an enantiomeric excess of 78%. Antibody-catalyzed reactions performed with
radiolabeled substrate indicated that little self-inactivation (6.8 mol % covalent modification after four turnovers
per antibody) occurred. The singular product obtained in the antibody-catalyzed reaction was not observed in
the uncatalyzed reaction unless the pH was lowered below 4. Studies suggested that the interplay of
conformational control and chemical catalysis were responsible for the high specificity. A change in protonation
state of the antibody was correlated with the inclusion of a new reaction pathway in the antibody-catalyzed
reaction, indicating that general-base catalysis was involved in the rerouting of the Norrish reaction to form
13. An X-ray crystal structure of the substrate was obtained and suggested that the antibody binds the
R-ketoamide in a twisted conformation optimal for the first step of the photochemical reaction. The antibody
described here is a model for the evolution of light-activated enzymes and can serve as a foundation for the
development of light-dependent antibody catalysts for a range of even more complex photochemical reactions.
Introduction
Antibodies have been shown to catalyze a variety of reactions
that involve highly reactive intermediates.7 Notably, an antibody-
catalyzed photochemical reaction involving sensitization by a
tryptophan residue and a putative radical anion intermediate has
been reported.8 Also, several other antibodies have been shown
to process substrates by way of carbocations.9 To investigate
catalysis of the photochemical Norrish type II reaction, we chose
to utilize the R-ketoamide 1 as both the hapten and substrate
(Scheme 1). A hapten-substrate approach is rarely engaged
because, energetically, this tactic does not adhere to the usual
transition-state stabilization approach. However, here this ap-
proach is reasonable because there is an external source of
energy. Thus, we could investigate the interaction of light with
a substrate susceptible to such a transformation within the
confines of an antibody combining site where binding energy
could perturb the reaction coordinate. In so doing, we intended
to establish the fate of a biradical intermediate, determine
reaction pathways and product distributions, and assess the
effects of chemical catalysis in the antibody microenvironment.
In this regard, self-inactivation of enzymes that catalyze
reactions via reactive intermediates has been observed during
the turnover of alternate substrates.10 Furthermore, while
sensitization of the Norrish reaction by a tryptophan residue
fortuitously present at the active site would not be expected,
this did not preclude the participation by other amino acids.
Significantly, pH has been shown to affect the product outcome
Nature has evolved elegant mechanisms to utilize the energy
of light. These include the rhodopsin-activated enzymatic
cascade in vision,1 the initiation of electron-transfer events in
photosynthesis,2 and the electron-transfer induced fragmentation
of thymine dimers in DNA repair.3 A common mechanistic
feature in the enzymes that utilize light energy to catalyze
chemical reactions is that absorption of a photon by the substrate
or cofactor initiates a multistep process involving one or more
reactive intermediates. These enzymes, as well as others, have
evolved to correctly process high-energy species including
cations, anions, radical anions, and radicals.3,4 In addition to
their role in enzymatic processes, photochemical reactions have
also been of interest among physical and synthetic organic
chemists.5 A well-studied example is the photochemical Norrish
type II reaction of both ketones and R-ketoamides which
proceeds through a biradical intermediate (Figure 1).6 In the
effort to develop a catalyst for a photochemical reaction of this
type, the challenge lies in the ability to not only generate but
also control the reactive intermediates such that a single product
is formed.
* To whom correspondence should be addressed.
(1) ) Stryer, L. Cold Spring Harbor Symp. Quant. Biol. 1988, 53, 283
and references therein.
(2) Deisenhofer, J.; Michel, H.; Huber, R. Trends Biochem. Sci. 1985,
10, 243 and references therein.
(3) Begley, T. P. Acc. Chem. Res. 1994, 27, 394 and references therein.
(4) (a) Retey, J. Angew. Chem., Int. Ed. Engl. 1990, 29, 355. (b) Walsh,
C. Enzymatic Reaction Mechanisms; W. H. Freeman and Company: New
York, 1979.
(5) Turro, N. J. Modern Molecular Photochemistry; Benjamin/Cum-
mings: Menlo Park, CA, 1978.
(6) (a) Norrish, R. G. W. Trans. Faraday Soc. 1937, 33, 1521. (b)
Bamford, C. H.; Norrish, R. G. W. J. Chem. Soc. 1935, 1504. (c) Åkermark,
B.; Sjoberg, B.; Johansson, N.-G. Tetrahedron Lett. 1969, 371.
(7) (a) Schultz P. G.; Lerner, R. A. Science 1995, 269, 1835. (b) Keinan,
E.; Lerner, R. A. Isr. J. Chem. 1996, 36, 113.
(8) Jacobsen, J. R.; Cochran, A. G.; Stephans, J. C.; King, D. S.; Schultz,
P. G. J. Am. Chem. Soc. 1995, 117, 5453.
(9) (a) Li, T.; Janda, K. D.; Ashley, J. A.; Lerner, R. A. Science 1994,
264, 1289. (b) Li, T.; Janda, K. D.; Lerner, R. A. Nature 1996, 379, 326.
(10) (a) Hamilton, J. A.; Yamada, R.; Blakley, R. L.; Hogenkamp, H. P.
C.; Looney, F. D.; Winfield, M. E. Biochemistry 1971, 10, 347. (b) Kulmacz,
R. J. Arch. Biochem. Biophys. 1986, 249, 273.
10.1021/ja982711r CCC: $15.00 © 1998 American Chemical Society
Published on Web 11/21/1998

12784 J. Am. Chem. Soc., Vol. 120, No. 49, 1998
Taylor et al.
Figure 1. Mechanism of the Norrish type II reactions of ketones and R-ketoamides. Following absorption of a photon, intramolecular hydrogen
atom abstraction results in the formation of a triplet biradical which undergoes intersystem crossing (ISC) to the singlet state. In the case of the
R-ketoamide, the electronic configuration is in resonance with the zwitterionic and biradical state. This is usually followed by cyclization (in both
cases) or fragmentation (in the case of ketones).
Scheme 1a
High-intensity, broad band UV irradiation (centered at 340
nm) of substrate 1b under the assay conditions (50 mM N,N-
bis(2-hydroxyethyl)glycine (Bicine) pH 8.5, 5% dimethyl sul-
foxide (DMSO)) produced primarily one diastereomer of the
â-lactam 11 and equal amounts of the oxazolidinone diastere-
omers 12, consistent with the expected product distribution from
literature precedence (Figure 2).11 Purification on the milligram
scale provided authentic standards of these products for
comparison. A panel of 22 monoclonal antibodies (mAbs)
elicited against R-ketoamide hapten 1a were purified12 and then
screened under described conditions (vide supra). Interestingly,
it was found that three of the mAbs (8C7, 2B6, and 1A3), of
which mAb 8C7 was studied in detail, produced a single product
that was not observed in the uncatalyzed reaction or in the
absence of light energy. This product was identified as the
tetrahydropyrazine 13, resulting from a rearrangement of the
biradical and/or zwitterionic intermediates 9 and 10.13 However,
it was also found that 13 could be produced in the absence of
antibody under acidic conditions (pH < 4). As an additional
control, bovine serum albumin (BSA), a protein previously
shown to bind small molecules and accelerate some reactions,14
was utilized under identical conditions and had no effect on
the photochemical reaction of the R-ketoamide substrate.
The above observations suggested that the photochemical
reaction in the antibody combining site was under strict
microenvironment control so as to produce a single, unique
product. It should be emphasized that the specificity of mAb
8C7 was very high, especially when compared to that of the
uncatalyzed reaction. This selectivity was illustrated by high-
pressure liquid chromatography (HPLC) analysis of the reaction
products (Figure 3). At pH 8.5 in the absence of antibody, no
measurable amount of 13 was formed. Even at pH 1, only 50%
of the product mixture was composed of tetrahydropyrazine 13.
Yet, virtually quantitative conversion of R-ketoamide 1b to
tetrahydropyrazine 13 was achieved by mAb 8C7. Fluorescence
quenching experiments, in which a fixed concentration of
antibody was titrated with increasing amounts of substrate,
indicated up to 2 equiv were required before quenching was
complete (data not shown). This result demonstrated the
expected presence of two combining sites per antibody molecule.
b
i
a (a) BuCl, Et3N, CH2Cl2, 0 °C (96%); (b) SeO2, pyridine, 100 °C
(85%); (c) N-acetyl-1,4-piperizine (for 1b) (40%) or tert-butyl 1-pip-
erizine-carboxylate (for 5) (98%), Et3N, BOP chloride, CHCl3; (d) i) 4
M HCl, dioxane, ii) 5-[(2,5-dioxo-1-pyrrolidinyl)oxy]-5-oxopentanoyl
chloride, diisopropylethylamine (DIPEA), CHCl3 (92%).
of the Norrish reaction11a so that general acid/base catalysis
might be observed. Our rationale suggested that the utilization
of a singular hapten-substrate could be a viable approach in
the elicitation of antibody catalysts for a multistep Norrish type
II reaction.
Results and Discussion
Hapten 1a was synthesized in five steps from 4-amino-
acetophenone (Scheme 1). The aniline substructure contained
in 2 was first acylated with isobutyryl chloride (iBuCl). Isolation
of this amide was followed by oxidation of the methyl ketone
embedded in 3 with selenium dioxide (SeO2) in pyridine to
provide the R-ketoacid 4. Coupling of 4 to tert-butyl 1-piper-
izine-carboxylate using bis(2-oxo-3-oxazolidinyl)phosphinic
(BOP) chloride afforded 5 as a white solid in 98% yield.
Installation of the linker for coupling to carrier proteins keyhole
limpet hemocyanin (KLH) and bovine serum albumin (BSA)
was achieved by deprotection of 5 with 4 M aqueous HCl and
acylation of the resulting amine with the mono-N-hydroxysuc-
cinimidyl (NHS) ester of glutaric acid chloride, providing 1a.
Substrate 1b, where the C-5 linker was replaced with an
acetamide functionality, was synthesized by coupling the
R-ketoacid 4 to N-acetyl-1,4-piperazine with BOP chloride.
(12) Kohler, G.; Milstein, C. Nature 1975, 256, 495.
(13) Spectroscopic evidence is consistent with this structural assignment.
An authentic standard of the tetrahydropyrazine 13 was synthesized by
photolysis of 1b in 5% water-acetonitrile with 1% trifluoroacetic acid and
purified by preparative reversed-phase HPLC and preparative TLC. Further
proof of the structure was obtained by derivatization through hydrogenation
of the double bond in 13 with hydrogen over palladium on carbon. This
produced a product identical to that obtained by reduction of the ketone in
1b with sodium borohydride. Control experiments indicate that 13 was not
produced from products 11 or 12 from the uncatalyzed reaction, suggesting
that these are not intermediates in the uncatalyzed formation of 13.
Analogous reactivity in other R-ketoamides has not been reported.
(14) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S. Nature 1996, 383, 60.
(11) (a) Aoyama, H.; Sakamoto, M.; Kuwabara, K.; Yoshida, K.; Omote,
Y. J. Am. Chem. Soc. 1983, 105, 1958. (b) Irradiation of solid 1b produced
a â-lactam diastereomer as the sole product, consistent with literature
precedence (see: Aoyama, H.; Hasegawa, T.; Omote, Y. J. Am. Chem. Soc.
1979, 101, 5343). A control experiment at 20 mM substrate concentration,
where precipitation was observed, revealed a product profile that was
essentially a sum of the products resulting from solid-state and solution-
state reactivity. We therefore concluded that 1b was soluble under the assay
conditions.

Light-ActiVated Antibody Catalyst
J. Am. Chem. Soc., Vol. 120, No. 49, 1998 12785
Figure 2. The products of the uncatalyzed and antibody-catalyzed reactions.
Scheme 2a
a (a) i) HOBt, EDC, DMF, ii) 1,4-piperizine, DMF (39%); (b) 4,
Et3N, BOP chloride, CHCl3 (51%); (c) H2, Pd/C, EtOH (83%); (d)
CH314COCl, Et3N, CHCl3/THF (64%).
Figure 3. HPLC analysis of uncatalyzed and antibody-catalyzed
reactions. The top trace is an uncatalyzed reaction at 10 µM 1b in 50
mM Bicine, pH 8.5, 5% DMSO. Product retention times: 8.24 min
(11), 17.17 and 18.56 min (12). The middle trace is from 10 µM 1b in
0.1 N HCl, 5% DMSO. Product retention times: 10.78 min (13), 16.84
and 18.18 min (12). The bottom trace is 10 µM antibody 8C7, 10 µM
1b in 50 mM Bicine, pH 8.5, 5% DMSO, and 9.5 µM 13 is formed.
turnover and rates (measured at Vmax) that were linear with
antibody concentration. The kinetic constants were maximal at
a wavelength of 280 nm,16 giving a Vmax ) 5.5 × 10-3 µM/
min (at 60 µM substrate, 2 µM antibody), kcat ) 1.4 × 10-3
min-1, kcat/Kd ) 1.8 × 105 M-1 min-1, and a quantum yield of
0.0012. Significantly, the uncatalyzed rate of formation of 13
was not measurable under these conditions.
In this initial assay, a high concentration of mAb and near
stoichiometric amounts of substrate were employed to qualita-
tively assess product distribution. To obtain quantitative kinetic
parameters, formation of tetrahydropyrazine 13 by the antibody
was followed using low-intensity irradiation and catalytic
amounts of antibody. Tight binding (i.e., low Ks) of substrate
1b to the antibody prevented a full Michaelis-Menten analysis,
since Km could not be accurately determined; however, equi-
librium binding constants for the substrate were obtained by
BIAcore analysis15 using 1c. Synthesis of the amine-tethered
substrate analogue 1c was performed as shown in Scheme 2.
The N-carboxybenzyl-γ-amino-n-butyric acid 6 was activated
as the 1-hydroxybenzotriazole (HOBt) ester and coupled to
piperazine. The resulting amine 7 was coupled to R-ketoacid 4
using BOP chloride to provide 8 in 51% yield. Following
removal of the carboxybenzyl (CBZ) group, the amine was
purified by preparative reversed phase HPLC to afford the
trifluoroacetate (TFA) salt 1c. A stock solution in dimethyl-
formamide (DMF) was then utilized for immobilization of the
ligand on a BIAcore chip, and BIAcore analysis was performed
according to standard protocols.15 This analysis provided a Kd
) 7.6 nM. Furthermore, mAb 8C7 was shown to exhibit
The effectiveness of any antibody catalyst for biradical
reactions could be limited, in part, by inactivation during the
course of the reaction. In studying catalytic antibodies for
cationic reactions, for instance, one antibody was observed to
undergo alkylation upon reaction with substrate.17 To observe
any covalent inactivation of mAb 8C7, the reaction was
performed in the presence of radiolabeled substrate 1d, syn-
thesized by acylation of 1c with 1-(14C)-acetyl chloride. It was
(16) The wavelength dependence of the uncatalyzed and antibody-
catalyzed reactions were examined (page S20, Supporting Information). The
uncatalyzed reaction, as judged by the rate of formation of â-lactam 11,
has a wavelength versus rate maximum at 310 nm (corrected for light
intensity), paralleling the absorption maximum of substrate 1b (λmax ) 310
nm, ꢀ310 ) 28 000 L M-1 cm-1). The mAb 8C7 catalyzed reaction, as judged
by the rate of formation of 13, has a rate maximum shifted approximately
20 nm to 290 nm. This shift in wavelength versus rate profile for the
antibody-catalyzed reaction is most consistent with a shift in absorption
maximum upon binding of substrate to antibody. The UV spectrum of the
bound complex was measured; however, the spectrum is dominated by
antibody absorbance in this region, making it difficult to judge whether a
blue-shift in UV maximum of the substrate had occurred. An alternative
model, one involving energy transfer from or sensitization by a tryptophan
indole side chain (ꢀ280 ) 5000 L M-1 cm-1) was not supported by our
data, since a large contribution from substrate absorption at 310 nm in the
wavelength versus rate profile would have been expected.
(17) Li, T.; Janda, K. D.; Hilton, S.; Lerner, R. A. J. Am. Chem. Soc.
1995, 117, 2367.
(15) Malmqvist, M. Nature 1993, 361, 186.

12786 J. Am. Chem. Soc., Vol. 120, No. 49, 1998
Taylor et al.
Figure 4. The pH-dependent product distributions for uncatalyzed and antibody-catalyzed reactions.
determined that after four turnovers only 6.8 mol % radioactivity
remained bound per antibody molecule. This indicated that little
covalent modification by the biradical intermediate occurred and
that the antibody was not subject to self-inactivation during the
photochemical Norrish type II reaction.
ring closure on either carbon or oxygen at a rate that apparently
competed favorably with proton abstraction, since formation of
13 was not observed under these conditions. However, at low
pH, protonation of 10 to form 14 competed with cyclization.
Protonation can be predicted to have two effects on the reactivity
of 10. First, it is known that cyclization on carbon is disfavored
under acidic conditions,11a which we observed as an absence
of 11 at low pH. This could be attributed to a loss of both anionic
and biradical character, as cyclization on carbon occurs from
either 9 or 10. Second, deprotonation of the resulting iminium
ion would be favored because of an increase in the acidity of
the abstracted proton, favoring the formation of 13.
Interestingly, catalysis by mAb 8C7 exhibited a pH-dependent
product distribution that strongly suggested chemical catalysis
was involved in the rerouting of the Norrish reaction. A Vmax
versus pH profile for production of tetrahydropyrazine 13 was
bell-shaped with a maximum at pH 7.5.20 Importantly, the profile
did not merely reflect a change in rate, but also the inclusion
of a new reaction pathway. It was found that at low pH the
reaction course shifted toward fragmentation, where it competed
effectively with proton abstraction, yielding multiple products,
including 15 and 16 (Figure 4).21 A model consistent with these
observations is one where a general base in the antibody
combining site is performing the proton abstraction to form 13
and that, upon protonation of this basic amino acid residue,
fragmentation becomes the favored process (Figure 5). It is also
reasonable to presume that the antibody may use binding energy
to exert conformational control over intermediates in the
multistep reaction. An X-ray crystal structure for R-ketoamide
1b was obtained and revealed a conformation where the ketone
and amide carbonyls were orthogonal (Figure 6). Therefore, we
assume that antibodies elicited to 1a will bind 1b in a twisted
conformation. This geometry is optimal for the intramolecular
Antibody-catalyzed asymmetric induction was also examined,
since a benzylic stereocenter in tetrahydropyrazine 13 was
produced from the prochiral R-ketoamide 1b during this light-
driven reaction. Stereochemical control is one of the hallmarks
of enzyme-catalyzed reactions,18 and asymmetric induction has
been achieved successfully in catalytic antibodies.19 The degree
of enantioselectivity was measured quantitatively by determining
the enantiomeric excess (ee). To this end, the antibody-catalyzed
reaction was taken to completion under stoichiometric condi-
tions, and 13 was analyzed by chiral high-pressure liquid
chromatography. Comparison to an authentic sample of racemic
13 indicated that the antibody selectively formed one enantiomer
in 78% ee. The enantioselectivity was quite high, given that no
stereochemistry was specified in the original hapten design. This
result is perhaps most simply accommodated in terms of
enantiofacial protonation of zwitterionic intermediate 10, rather
than a tantamount sequential hydrogen atom abstraction and
donation by the biradical form 9. However, the latter process
cannot be ruled out. Significantly, the presence of the radical
character at the prochiral carbon in this intermediate did not
preclude asymmetric induction.
The pH profile for the uncatalyzed reaction revealed some
important mechanistic details. In the uncatalyzed reaction at pH
8.5, there was primarily a competition between cyclization on
carbon to form 11 and on oxygen to form 12, whereas at pH 1,
cyclization to 12 and proton abstraction to form tetrahydro-
pyrazine 13 were competitive (Figure 4). This product distribu-
tion could be explained by examining the reactivity of the
biradical and/or zwitterion 9 and 10. At high pH, 10 underwent
(20) Cleland, W. W. AdV. Enzymol. 1977, 45, 273.
(21) Co-injection in the HPLC assay of the products of the antibody-
catalyzed reaction (50 mM MES, 100 mM NaCl, pH 6.0) with an authentic
standard of 15, synthesized by sodium borohydride reduction of the
R-ketoacid 4, suggested that fragmentation was occurring. Due to a lack of
UV absorbance and/or stability, 16 was not observed in the HPLC assay,
and it is important to note that two other unidentified products are also
formed.
(18) Fersht, A. Enzyme Structure and Mechanism; W. H. Freeman and
Company: New York, 1977.
(19) (a) Fujii, I.; Lerner, R. A.; Janda, K. D. J. Am. Chem. Soc. 1991,
113, 8528. (b) Nakayama, G. R.; Schultz, P. G. J. Am. Chem. Soc. 1992,
114, 780. (c) Zhong, G.; Hoffmann, T.; Lerner, R. A.; Danishefsky, S.;
Barbas, C. F., III. J. Am. Chem. Soc. 1997, 119, 8131.

Light-ActiVated Antibody Catalyst
J. Am. Chem. Soc., Vol. 120, No. 49, 1998 12787
abstracted proton by binding to the negatively charged enolate
oxygen of 10. This effect would be similar to that described
for the formation of 13 at low pH in the uncatalyzed reaction
(vide supra). It is unlikely that such a general acid also served
as a specific proton source in the reaction, since we would have
expected a much higher ee value for the formation of tetrahy-
dropyrazine 13. With regard to conformational effects, a
hydrogen bond donor might position the intermediate 10 or 14
optimally for deprotonation by the general base. Upon depro-
tonation of this amino acid side chain, the substrate is out of
alignment for efficient proton abstraction, causing a decrease
in reaction rate. Alternatively, the antibody general base might
not be properly positioned because of a protein conformational
change at high pH.
Figure 5. A model for the pH-dependent mechanism of formation of
products by mAb 8C7.
Conclusion
Of the three light-activated enzymes known,26,27 perhaps the
best studied is DNA photolyase.27 In this enzyme, one cofactor
functions as “antenna” (a folate) and another as electron-transfer
component (a flavin) in the fragmentation of thymine photo-
dimers in duplex DNA. The wavelength range (300-500 nm)
for activity is defined by the absorbance of the cofactors in the
enzyme. Even so, upon removal of the cofactors the enzyme-
catalyzed reaction still proceeds, albeit poorly, with a tryptophan
indole side chain now acting as the electron-transfer compo-
nent.28 Consequently, this form of DNA photolyase utilizes light
energy that now corresponds to tryptophan absorbance (250-
300 nm). These observations suggest that two challenges existed
in the evolution of light-activated enzymes. First, nature evolved
a chemical mechanism intrinsic to the enzyme protein frame-
work. In DNA photolyase, this included binding of the thymine
dimer, inclusion of a proximal indole side chain for the electron-
transfer step, and processing of a radical anion intermediate.
Second, the enzyme then evolved the capacity to utilize
wavelengths of light energy available in the environment by
installing an array of cofactors. In the antibody catalyst described
here, the mechanism involved binding of the R-ketoamide and
the routing of a biradical intermediate to form a single, unique
product through the interplay of conformational control and
combining site chemical catalysis. Hence, in essence, the
antibody is representative of the early evolution of a light-
activated enzyme in that the chemistry is intrinsic to the protein
itself without any significant tuning of the optimal wavelength
range by extrinsic cofactors. The installation of additional
chemical machinery, through hapten design or protein engineer-
ing, could afford light-dependent catalytic antibodies adapted
to utilize specific wavelengths of light for a wide range of
photochemical reactions.
Figure 6. An ORTEP view of the X-ray crystal structure of substrate
1b.
hydrogen atom abstraction in the photochemical reaction,22 and
rationalizes the observed successful reaction in the antibody
combining site. Furthermore, calculations suggest that the ideal
geometries for resonance forms 9 and 10 are nonplanar and
planar, respectively.23 Hence, the zwitterionic form should be
destabilized by the antibody. It is known that 11 is preferentially
formed from biradical 9 and this transformation was not
catalyzed by the mAb. The observed reactivity is best explained
by rapid, enantioselective protonation of an intermediate twisted
structure with zwitterionic character to form the nonplanar
intermediate 14.24 This could occur through antibody-mediated
conformational destabilization of the zwitterion 10, effectively
lowering the transition-state energy for protonation to form 14.
Upon formation of 14, deprotonation and fragmentation compete
effectively with cyclization on oxygen, as formation of 12 was
not observed. The behavior at high pH, where the sole product
was 13 and the reaction rate decreased, was more complicated
to interpret. We could speculate that either substrate and/or
antibody conformation or chemical catalysis was affected by
deprotonation of a general acid involved in the mechanism for
formation of 6. A hydrogen bond, possibly recruited through a
“bait-and-switch”25 in the immune response by the tertiary amide
carbonyl in the hapten, could increase the acidity of the
Experimental Section
(22) The optimal geometries for the Norrish type II reactions of ketones
have been determined (see: Gudmundsdottir A. D.; Lewis, T. J.; Randall,
L. H.; Scheffer, J. R.; Rettig, S. J.; Trotter, J.; Wu, C.-H. J. Am. Chem.
Soc. 1996, 118, 6167). Both the atomic distances and bond angles obtained
from the coordinates of the R-ketoamide 1b crystal structure compare very
favorably with those optimal for intramolecular hydrogen atom abstraction
in the Norrish reaction of ketones.
(23) Chesta, C. A.; Whitten, D. G. J. Am. Chem. Soc., 1992, 114, 2188.
(24) A reviewer suggested the possibility of amino acid side chains (e.g.,
cysteine) in the antibody combining site participating in radical chemistry
with the bound substrate. We find this intriguing because the transformation
shown in Figure 2 is formally equivalent to a hydrogen atom abstraction/
donation process. However, we have no evidence for this mechanism and
therefore chose to discuss the simpler model, exemplified in Figure 5, for
the mechanism of the antibody-catalyzed reaction involving a protonation/
deprotonation mechanism. X-ray crystallographic analysis, cloning, and site-
directed mutagenesis would provide data bearing on this question, but since
these experiments are time-consuming and large in scope, they will be the
subject of future reports.
All reactions were performed in an atmosphere of nitrogen with a
magnetic stirrer unless otherwise noted. Analytical HPLC was per-
formed on a Hitachi L7000 series HPLC using a Vydac 201TP54
reversed phase column. Preparative HPLC was performed on a Rainin
HPLC system using a Vydac 201HS1022 reversed phase column.
Methylene chloride and chloroform were distilled from calcium hydride.
Tetrahydrofuran (THF) was distilled from sodium/benzophenone.
Methanol was distilled from magnesium. Scintillation counting was
performed on a Beckman L3801 scintillation counter using ScintiVerse
(25) Janda, K. D.; Weinhouse, M. I.; Danon, T.; Pacelli, K. A.; Schloeder,
D. M. J. Am. Chem. Soc. 1991, 113, 5427.
(26) (a) Beer, N. S.; Griffiths, T. W. Biochem. J. 1981, 195, 83. (b)
Heelis, P. F.; Liu, S. J. Am. Chem. Soc. 1997, 119, 2936.
(27) Sancar, A. Biochemistry 1994, 33, 2.
(28) Kim, S.-T.; Li, Y. F.; Sancar, A. Proc. Natl. Acad. Sci. U.S.A. 1992,
89, 900.

12788 J. Am. Chem. Soc., Vol. 120, No. 49, 1998
Taylor et al.
II (Fischer) as scintillant. High-intensity photolysis was performed in
a Rayonet photochemical reactor with RPR-3400 bulbs. Low-intensity
photolysis was performed in a SPF-500 spectrofluorometer from SLM-
Amico Instruments. Beam intensity was determined using both ferri-
oxalate actinometry (performed according to standard protocols29 using
a Hewlett-Packard Vectra series 4 UV-visible spectrometer) and a
model 70260 radiant power energy meter from Oriel Instruments.
Dialysis was performed using slide-a-lyzer cassettes purchased from
Pierce. Sep-pak cartridges were purchased from Waters. 1-(14C)-acetyl
chloride was purchased from American Radiolabeled Chemical at a
specific activity of 50 mCi/mmol.
N-(4-Acetylphenyl)-2-methyl-propanamide (3). 4-aminoacetophe-
none (2.65 g, 19.6 mmol) and triethylamine (2.73 mL, 19.6 mmol) in
70 mL of dry CH2Cl2 were cooled in an ice bath. Isobutyryl chloride
(2.9 mL, 19.6 mmol) was added dropwise via a syringe. The reaction
was stirred for 1 h at room temperature and was judged complete by
thin-layer chromatography (TLC) (CH2Cl2). Then 100 mL of CH2Cl2
was added and the organic layer washed with 50 mL of 1 N HCl, 50
mL of saturated NaHCO3, and 50 mL of brine. The organic layer was
then dried over anhydrous Na2SO4 and filtered, and the solvent was
removed in vacuo to provide 3.9 g 3 as a pale yellow solid (96%): 1H
NMR (CDCl3, 500 MHz) δ 1.27 (d, 6H, J ) 6.6 Hz), 2.55 (sept, 1H,
J ) 7 Hz), 2.57 (s, 3H), 7.47 (bs, 1H), 7.65 (d, 2H, J ) 8.8 Hz), 7.93
(d, 2H, 8.5 Hz); 13C NMR (CDCl3, 125 MHz) δ 19.4, 26.4, 36.5, 119.0,
129.6, 132.4, 142.8, 176.1, 197.3; HRMS for C12H15NO2 (M+) calcd
205.1103, found 205.1180.
4-[(2-Methyl-1-oxopropyl)amino]-R-oxo-benzeneacetic Acid (4).
Compound 3 (2.52 g, 12.3 mmol) and selenium dioxide (2.04 g, 18.4
mmol) in 50 mL of pyridine were heated to 100 °C for 14 h. The
selenium was filtered and the orange solution concentrated in vacuo to
approximately 5 mL in volume. The residue was dissolved in 100 mL
of 5% NaOH and extracted with two 30-mL portions of diethyl ether.
The aqueous layer was acidified with concentrated HCl and extracted
with three 50-mL portions of ethyl acetate. The combined organic layers
were washed with 30 mL of brine, dried over anhydrous Na2SO4 and
filtered, and the solvent was removed in vacuo. Flash chromatography
(ethyl acetate (EtOAc)/hexane/acetic acid (HOAc) 15/1/1) provided 2.46
g (85%) of 4 as a yellow solid: 1H NMR (d6-DMSO, 500 MHz) δ
1.10 (dd, 6H, J1 ) 7 Hz, J2 ) 3.3 Hz), 2.64 (dsept, 1H, J1 ) 7 Hz, J2
) 3.3 Hz), 7.82 (dd, 2H, J1 ) 8.8 Hz, J2 ) 3.3 Hz), 7.88 (dd, 2H, J1
) 8.8 Hz, J2 ) 3.3 Hz), 10.34 (s, 1H); 13C NMR (d6-DMSO, 125 MHz)
δ 18.9, 34.7, 39.0, 118.3, 125.8, 130.5, 145.1, 166.0, 175.7, 186.9;
HRMS for C12H13NO4 (M+) calcd 235.0845, found 235.0919.
N-[4-[(4-Acetyl-1-piperazinyl)oxoacetyl]phenyl]-2-methyl-pro-
panamide (1b). Compound 4 (888 mg, 3.78 mmol), N-acetylpiperazine
(480 mg, 3.74 mmol), and triethylamine (0.5 mL, 3.59 mmol) were
dissolved in 9 mL of dry CHCl3 and cooled on ice. Bis(2-oxo-3-
oxazolidinyl)phosphinic chloride (BOP chloride) (960 mg, 3.77 mmol)
was added as a solid in portions. The reaction mixture was stirred for
1 h on ice, then warmed to room temperature, and stirred overnight.
Then 150 mL of EtOAc was added and the organic solution extracted
with 40 mL of 1 N HCl, 40 mL of saturated NaHCO3, and 40 mL of
brine. The organic layer was dried over anhydrous Na2SO4 and filtered,
and the solvent was removed in vacuo. Flash chromatography (EtOAc)
provided 522 mg (40%) of 1b as a white solid: 1H NMR (CDCl3, 600
MHz) δ 1.27 (d, 6H, J ) 7 Hz), 2.10 and 2.16 (s, 3H), 2.55 (sept, 1H,
J ) 7 Hz), 3.33-3.35 (m, 1H), 3.38-3.40 (m, 1H), 3.46-3.48 (m,
1H), 3.58-3.63 (m, 2H), 3.75 (s, 2H), 3.78-3.80 (m, 1H), 7.41 (bs,
1H), 7.70 (dd, 2H, J1 ) 8.8 Hz, J2 ) 2.2 Hz), 7.93 (dd, 2H, J1 ) 8.8
Hz, J2 ) 3 Hz); 13C NMR (CDCl3, 125 MHz) δ 19.3, 21.3, 36.9, 40.9,
41.1, 41.2, 41.4, 45.6, 45.8, 45.9, 46.2, 119.1, 119.2, 128.0, 131.1, 144.4,
165.6, 165.8, 169.1, 175.8, 189.5; HRMS for C18H23N3O4 (M+) calcd
345.1688, found 345.1773.
phosphinic chloride (BOP chloride) (0.44 g, 1.74 mmol) was added to
the stirred solution and the reaction mixture brought to room temper-
ature and stirred for 2 h. The mixture was diluted with EtOAc (150
mL), and the organic layer was washed successively with solutions of
0.1 M HCl (50 mL), saturated NaHCO3 (30 mL), and brine (2 × 50
mL). The organic layer was dried with anhydrous MgSO4 and filtered,
and the solvent was removed in vacuo. The residue was crystallized
twice from EtOAc/hexanes to give 0.64 g (91%) of 5 as a white solid:
1H NMR δ (d6-DMSO, 300 MHz) δ 1.09 (d, 6H, J ) 9.4 Hz), 1.40 (s,
9H), 2.63 (sept, 1H, J ) 9.4 Hz), 3.21 (m, 2H), 3.28 (m, 2H), 3.46 (m,
2H), 3.59 (m, 2H), 7.88 (m, 4H), 10.35 (br s, 1H); 13C NMR (CDCl3,
150 MHz) δ 19.3, 28.2, 36.3, 41.1, 45.7, 80.6, 119.3, 127.7, 130.8,
144.8, 154.3, 165.9, 176.3, 189.8; HRMS for C21H29N3O5 (M+) calcd
403.2107, found 403.2117.
N-[4-[[4-[5-[(2,5-Dioxo-1-pyrrolidinyl)oxy]-1,5-dioxopentyl]-1-
piperazinyl]oxoacetyl]phenyl]-2-methyl-propanamide (1a). A mix-
ture of tert-Boc piperazinyl amide 5 (0.64 g, 1.59 mmol) and 4 M
hydrogen chloride in 1,4-dioxane (8.0 mL, 32 mmol) was stirred at
room temperature for 30 min. 1,4-Dioxane was removed in vacuo, and
the crystalline residue was lyophilized to give the amine hydrochloride
(0.53 g, 98%). The crude product was used without further purification
(LRMS for C16H21N3O3 calcd 303 (M+), found 303). A mixture of the
amine hydrochloride (0.53 g, 1.56 mmol), 5-[(2,5-dioxo-1-pyrrolidinyl)-
oxy]-5-oxopentanoyl chloride (0.42 g, 1.72 mmol) and N,N-diisoprop-
ylethylamine (0.57 mL, 3.28 mmol) in CHCl3 (10 mL) was stirred at
room temperature for 13 h. The reaction mixture was diluted with CH2-
Cl2 (100 mL) and washed with solutions of 0.5 M HCl (20 mL),
saturated NaHCO3 (30 mL), and brine (2 × 30 mL). The organic layer
was dried with anhydrous MgSO4 and filtered, and the solvent was
removed in vacuo to give crude 1a as an off-white solid. Recrystalli-
zation (EtOAc/hexanes) gave 0.76 g (92%) of 1a as white crystals:
1H NMR (d6-DMSO, 300 MHz) δ 1.11 (d, 6H, J ) 9.4 Hz), 1.72-
1.90 (m, 4H), 1.95 (t, 2H, J ) 8.4 Hz), 2.64 (m, 1H), 2.82 (bs, 4H),
3.22 (m, 2H), 3.28 (m, 2H), 3.41 (m, 2H), 3.61 (m, 2H), 7.79-7.92
(m, 4H), 10.38 (br s, 1H); 13C NMR (CDCl3, 150 MHz) δ 19.5, 20.2,
25.6, 29.7, 30.2, 30.9, 36.9, 41.3, 41.4, 41.7, 45.0, 45.5, 45.7, 45.9,
119.2, 128.3, 131.3, 144.2, 165.6, 165.8, 168.4, 169.1, 170.4, 175.6,
189.6; HRMS for C25H31N4O8 (M+) calcd 515.2141, found 515.2146.
[4-Oxo-4-(piperazinyl)butyl]carbamic Acid, Phenylmethyl Ester
(7). To N-CBZ-γ-amino-n-butyric acid 6 (200 mg, 0.84 mmol) and
1-hydroxybenzotriazole (HOBt) (114 mg, 0.84 mmol) in 2 mL of DMF
was added 1-[3-(dimethylamino)propyl]-3-ethylcarbodiimide hydro-
chloride (EDC) (165 mg, 0.86 mmol). After 30 min, the reaction was
judged complete by TLC (CH2Cl2). The resulting activated ester was
added dropwise to 1,4-piperazine (360 mg, 8.4 mmol) in 1 mL of DMF.
The reaction mixture was then added to 30 mL of EtOAc and the
organic solution extracted with two 30-mL portions 1 N HCl. The
combined aqueous layer was adjusted to pH 10 with 1 N NaOH and
saturated NaHCO3. The aqueous layer was extracted with four 100-
mL portions EtOAc. The organic layer was then dried over anhydrous
Na2SO4 and filtered, and the solvent was removed in vacuo to provide
a yellow oil. Flash chromatography (10-25% CH3OH/CH2Cl2) pro-
vided 100 mg (39%) of 7 as a pale yellow oily solid: 1H NMR (d6-
DMSO, 600 MHz) δ 1.60 (quintet, 2H, J ) 7.2 Hz), 2.26 (t, 2H, J )
7.4 Hz), 2.60 (dt, 4H, J1 ) 11 Hz, J2 ) 4.8 Hz), 2.99 (dt, 2H, J1 ) 5.9
Hz, J2 ) 7 Hz), 3.33 (dt, 4H, J2 ) 4.8 Hz), 4.99 (s, 2H), 7.25-7.34
(m, 5H); 13C NMR (CD3OD, 150 MHz) δ 27.5, 31.9, 42.1, 44.2, 47.1,
47.5, 48.3, 68.2, 129.7, 129.8, 130.3, 139.4, 159.7, 174.2; HRMS for
C16H23N3O3 (M+) calcd 305.1739, found 305.1809.
[4-[4-[[4-[(2-Methyl-1-oxopropyl)amino]phenyl]oxoacetyl]-1-pip-
erazinyl]-4-oxobutyl]-carbamic acid, Phenylmethyl Ester (8). Com-
pounds 4 (63 mg, 0.2 mmol), 7 (49 mg, 0.2 mmol), and triethylamine
(28 µL, 0.2 mmol) were combined in 0.5 mL of CHCl3 and cooled on
ice. BOP chloride (51 mg, 0.2 mmol) was added and the reaction
mixture stirred first on ice for 30 min and then at room temperature
for 3 h. Then 25 mL of EtOAc was added and the organic layer
extracted with 4 mL of 1 N HCl, 4 mL of saturated NaHCO3, and 4
mL of brine. The organic layer was then dried over anhydrous Na2SO4
and filtered, and the solvent was removed in vacuo. Flash chromatog-
raphy (EtOAc) followed by preparative TLC (two 0.5-mm plates,
4-[[4-[(2-Methyl-1-oxopropyl)amino]phenyl]oxoacetyl]-1-pipera-
zineacetic acid, 1,1-Dimethylethyl Ester (5). A mixture of 4 (0.41 g,
1.74 mmol), tert-butyl 1-piperizine-carboxylate (0.32 g, 1.74 mmol)
and N,N-diisopropylethylamine (0.61 mL, 3.49 mmol) in anhydrous
CHCl3 (10 mL) was cooled to 0 °C. Bis (2-oxo-3-oxazolidinyl)-
(29) Murov, S. L.; Carmichael, I.; Hug, G. L. Handbook of Photochem-
istry; Marcel Dekker: New York, 1993; pp 299-305.

Light-ActiVated Antibody Catalyst
J. Am. Chem. Soc., Vol. 120, No. 49, 1998 12789
EtOAc, eluted with 20% CH3OH/CH2Cl2) provided 54 mg (51%) of 8
as a white foam: 1H NMR (CDCl3, 600 MHz) δ 1.27 (d, 6H, J ) 7
Hz), 1.85-1.88 (m, 2H), 2.33-2.41 (m, 2H), 2.55 (sept, 1H, J ) 7
Hz), 3.24-3.27 (m, 2H), 3.32 (bs, 2H), 3.40 (bs, 1H), 3.54 (bs, 1H),
3.60 (bs, 1H), 3.72-3.74 (m, 3H), 5.00 (bs, 1H), 5.06-5.09 (m, 2H),
7.32-7.35 (m, 5H), 7.43 (bs, 1H), 7.70 (d, 2H, J ) 8.8 Hz), 7.91-
7.93 (m, 2H); 13C NMR (CDCl3, 150 MHz) δ 19.3, 25.0, 30.2, 36.4,
40.5, 41.2, 41.5, 45.3, 45.5, 45.7, 66.5, 93.0, 119.4, 127.8, 127.9, 128.0,
128.4, 131.0, 136.5, 144.8, 156.6, 165.8, 171.0, 176.2, 189.6; HRMS
for C28H34N4O6 (M+) calcd 522.2478, found 522.2454.
N-[4-[[4-(4-Amino-1-oxobutyl)-1-piperazinyl]oxoacetyl]phenyl]-
2-methyl-propanamide, Trifluoroacetate Salt (1c). Compound 8 (50
mg, mmol) was dissolved in 2 mL of EtOH and 5 mg of 10% Pd/C
added. The mixture was then stirred under 1 atm H2 for 4 h and filtered
through Celite, and the solvent was removed in vacuo. Preparative
HPLC (20:80 CH3CN/H2O w/0.1% TFA) provided 40 mg (83%) of 1c
as a clear oil: 1H NMR (CD3OD, 600 MHz) δ 1.20 (d, 6H, J ) 7 Hz),
1.90-1.95 (m, 2H), 2.54 (t, 1H, J ) 7 Hz), 2.61 (t, 1H, J ) 7 Hz),
2.66 (sept, 1H, J ) 7 Hz), 2.94-2.99 (m, 2H), 3.37-3.39 (m, 1H),
3.41-3.43 (m, 1H), 3.52-3.54 (m, 1H), 3.59-3.61 (m, 1H), 3.66-
3.68 (m, 1H), 3.74 (s, 2H), 3.78-3.80 (m, 1H), 7.81 (d, 2H, J ) 8.8
Hz), 7.90-7.92 (m, 2H); 13C NMR (CD3OD, 150 MHz) δ 19.8, 23.7,
23.8, 30.9, 37.2, 40.4, 42.1, 42.3, 42.9, 45.9, 46.5, 46.7, 46.9, 120.6,
129.2, 132.1, 146.9, 167.7, 167.8, 172.7, 179.1, 191.4; HRMS for
C20H28N4O4 (M+) calcd 388.2111, found 388.2178.
N-[4-[[4-[4-(14C-Acetylamino)-1-oxobutyl]-1-piperazinyl]oxoacetyl]-
phenyl]-2-methyl-propanamide (1d). Compound 1c (8.4 mg, 16.7
µmol) was transferred to a small vial with dry methanol and the solvent
removed in vacuo. A small stirbar, triethylamine (100 µL, 50.2 µmol)
in 100 µL of dry CHCl3, and 100 µL of dry THF were added, and the
reaction mixture was cooled on ice. An ampuole of 1-(14C)-acetyl
chloride was cooled on ice and 1 mL of dry CHCl3 containing acetyl
chloride (1.6 µL, 22 µmol) added to the reservoir above the reagent
chamber. The top of the chamber was then broken, and the resulting
solution of radiolabeled acetyl chloride in was CHCl3 drawn into a
syringe and added to the reaction mixture dropwise. The transfer was
completed with 1 mL of CHCl3. The reaction mixture was stirred for
1 h at room temperature, followed by drying in a warm bath (37 °C)
with a stream of nitrogen. The dried reaction mixture was then loaded
onto a 0.5-mm preparative TLC plate with CH2Cl2/EtOAc/CH3OH and
eluted with 20% CH3OH/CH2Cl2, and the solvent was removed in vacuo
to provide 4.6 mg (64%) of 1d as a white foam. Comparison to a
nonradioactive standard by TLC confirmed the identity and purity of
1d. The specific activity of 1d, using the relationship 1 Ci ) 2.22 ×
1012 cpm, was determined by scintillation counting to be 12.7 mCi/
mmol. Spectral data for nonradioactive 1d: 1H NMR (CD3OD, 600
MHz) δ 1.20 (d, 6H, J ) 7 Hz), 1.75-1.80 (m, 2H), 1.90 and 1.93 (s,
3H), 2.40 (t, 1H, J ) 7.4 Hz), 2.47 (t, 1H, J ) 7.4 Hz), 2.66 (sept, 1H,
J ) 7 Hz), 3.16-3.22 (m, 2H), 3.36-3.38 (m, 1H), 3.40-3.42 (m,
1H), 3.52-3.54 (m, 1H), 3.57-3.59 (m, 1H), 3.67-3.69 (m, 1H), 3.73
(bs, 2H), 3.78-3.80 (m, 1H), 7.81 (d, 2H, J ) 8.8 Hz), 7.92 (dd, 2H,
J1 ) 8.8 Hz, J2 ) 1.8 Hz); 13C NMR (CD3OD, 150 MHz) δ 20.3,
23.0, 26.5, 31.7, 37.7, 40.4, 42.6, 42.7, 42.9, 43.3, 46.6, 47.1, 47.2,
47.5, 121.0, 129.7, 132.6, 147.3, 168.2, 168.3, 173.9, 174.0, 179.5,
191.9; HRMS for C22H30N4O5 (M+) calcd 430.2216, found 430.2127.
N-[4-(7-Acetylhexahydro-3-oxo-5H-oxazolo[3,2-a]pyrazin-2-yl)-
phenyl]-2-methyl-propanamide (12) and N-[4-(4-Acetyl-7-hydroxy-
8-oxo-1,4-diazabicyclo[4.2.0]oct-7-yl)phenyl]-2-methyl-propana-
mide (11). Compound 1b (20 mg, 58 µmol) was dissolved in 1 mL of
1:1 CH3CN/H2O in a glass vial. The solution was photolyzed at 340
nm for 1 h, then approximately 10 mL of CH3CN was added, and the
solvent was removed in vacuo. The oxazolidinones 12 (7 mg) were
isolated as a mixture by preparative TLC (1-mm plate, developed three
times with 10% CH3OH/CH2Cl2) as well as crude 11 (ca. 3 mg). The
oxazolidinones were then separated by preparative TLC (0.25-mm plate,
developed four times with 10% CH3OH/CH2Cl2) to provide 1 mg each
(5%) of the diastereomers. For high Rf diastereomer 12: 1H NMR
(CDCl3, 600 MHz) δ 1.24 (d, 6H, J ) 6.6 Hz), 2.16 and 2.17 (s, 3H),
2.45-2.51 (m, 1.5H), 2.57 (t, 0.5H, J ) 12.7 Hz), 2.99-3.02 (m, 1.5H),
3.12 (t, 0.5H, J ) 11.8 Hz), 3.82 (d, 0.5H, J ) 12.7 Hz), 4.11-4.17
(m, 1.5H), 4.71 (d, 0.5H, J ) 10.5 Hz), 5.06 (d, 0.5H, J ) 11 Hz),
5.18 (d, 0.5H, J ) 8.3 Hz), 5.24 (d, 0.5H, J ) 9.2 Hz), 5.30 (s, 1H),
7.22-7.24 (m, 1H), 7.36 (d, 2H, J ) 8.3 Hz), 7.35-7.56 (m, 2H); 13
C
NMR (CDCl3, heteronuclear multidimensional quantum correlation
(HMQC) at 150 MHz) δ 19.2, 21.5, 36.7, 39.6, 40.7, 45.3, 46.5, 52.4,
79.2, 84.2, 84.4, 119.2, 126.8; HRMS for C18H23N3O4 (M+) calcd
345.1688, found 345.1732. For low Rf diastereomer 12: 1H NMR
(CDCl3, 600 MHz) δ 1.25 (d, 6H, J ) 6.6 Hz), 2.17 and 2.18 (s, 3H),
2.47-2.57 (m, 2H), 2.98-3.03 (m, 2H), 3.86 (d, 0.5H, J ) 11.8 Hz),
4.08 (dd, 0.5H, J1 ) 13.2, J2 ) 3.5 Hz), 4.11 (dd, 0.5H, J1 ) 13.2 Hz,
J2 ) 3.5 Hz), 4.22 (d, 0.5H, J ) 13.2 Hz), 4.74 (d, 0.5H, J ) 10.1
Hz), 5.06 (dd, 0.5H, J1 ) 10 Hz, J2 ) 2.2 Hz), 5.12-5.14 (m, 1H),
5.28 (d, 1H, J ) 9.7 Hz), 7.25 (bs, 1H), 7.37-7.39 (m, 2H), 7.56 (d,
2H, J ) 9.7 Hz);13C NMR (CDCl3, HMQC at 150 MHz) δ 19.5, 21.2,
36.5, 39.5, 40.5, 45.3, 47.8, 52.8, 79.2, 83.3, 86.6, 119.4, 127.2; HRMS
for C18H23N3O4 (M+) calcd 345.1688, found 345.1595. For 11, the crude
material was purified by preparative HPLC to afford 1 mg (5%). For
â-lactam 11: 1H NMR (CD3OD, 600 MHz) δ 1.19 (d, 6H, J ) 6.6
Hz), 1.93-1.97 (m, 0.5H), 2.06 and 2.10 (s, 3H), 2.46-2.51 (m, 0.5H),
2.57-2.58 (m, 0.5H), 2.62 (sept, 1H, J ) 7 Hz), 2.98-3.14 (m, 1.5H),
3.65-3.68 (m, 0.5H), 3.78-3.89 (m, 2H), 4.41-4.50 (m, 1H), 7.44
(dd, 2H, J1 ) 15.4 Hz, J2 ) 8.8 Hz), 7.62-7.64 (m, 2H); 13C NMR
(CD3OD, HMQC at 150 MHz) δ 20.0, 21.3, 37.0, 39.4, 39.6, 45.2,
46.8, 47.0, 49.3, 49.2, 49.9, 62.0, 62.4, 121.3, 128.8; HRMS for
C18H23N3O4 (M+) calcd 345.1688, found 345.1586.
N-[4-[2-(4-Acetyl-3,4-dihydro-1(2H)-pyrazinyl)-1-hydroxy-2-oxo-
ethyl]phenyl]-2-methyl-propanamide (13). Compound 1b (29 mg, 84
µmol) was dissolved in 3 mL of CH3CN and aliquoted into six glass
vials. Then 150 µL of water with 1% trifluoroacetic acid was added to
each vial, and the reactions were photolyzed for 30 min. The product
mixture was then dried in vacuo. Two consecutive purifications by
preparative TLC (1-mm plate, developed three times with 10% CH3-
OH/CH2Cl2, then 0.5-mm plate developed five times with 10% CH3-
OH/EtOAc) provided 5.9 mg (20%) of 13: 1H NMR (d6-DMSO, 600
MHz) δ 1.07 (d, 6H, J ) 7 Hz), 2.02, 2.04, 2.06, and 2.07 (s, 3H),
2.56 (sept, 1H, J ) 7 Hz), 3.36-3.82 (m, 4H), 5.40-5.49 (m, 1H),
5.75-5.82 (m, 1H), 6.14 (d, 0.3H, J ) 6.6 Hz), 6.31 (d, 0.3H, J ) 7
Hz), 6.38-6.41 (m, 0.4H), 6.44-6.47 (m, 0.4H), 6.57 (d, 0.1H, J ) 7
Hz), 6.64 (d, 0.1H, J ) 7.5 Hz), 7.24-7.28 (m, 2H), 7.55-7.59 (m,
2H), 9.80 (bs, 1H); 13C NMR (CDCl3, 150 MHz) δ 19.6, 20.9, 21.0,
21.3, 36.7, 38.3, 38.8, 39.7, 39.9, 42.3, 43.4, 106.6, 107.5, 107.6, 109.9,
111.0, 112.2, 120.2, 120.3, 120.3, 128.1, 134.1, 138.5, 167.0, 169.0,
169.3, 175.3; HRMS for C18H23N3O4 (M+) calcd 345.1688, found
345.1760.
R-Hydroxy-4-[(2-methyl-1-oxopropyl)amino]-benzeneacetic acid
(15). Compound 4 (20 mg, 85 µmol) was dissolved in 1 mL of EtOH.
Sodium borohydride (16 mg, 227 µmol) was added and the solution
stirred for 1 h. The product mixture was then dried in vacuo and
redissolved in 0.5 mL of water. The aqueous solution was extracted
with three 1-mL portions of EtOAc, the combined organic layers were
dried over anhydrous Na2SO4 and filtered, and the solvent was removed
in vacuo. The residue was dissolved in 5 mL of water and loaded onto
a 35-g Sep-pak cartridge, washed with 20 mL of water, and eluted
with 10 mL of CH3CN. The solvent was removed in vacuo to provide
15 mg (75%) of 15 as a white solid: 1H NMR (CD3OD, 500 MHz) δ
1.18 (dd, 6H, J1 ) 7 Hz, J2 ) 2.2 Hz), 2.61 (dsept, 1H, J1 ) 7 Hz, J2
) 2.2 Hz), 5.06 (s, 1H), 7.40 (d, 2H, J ) 6.6 Hz), 7.55 (d, 2H, J ) 6.6
Hz); 13C NMR (CD3OD, 100 MHz) δ 20.8, 37.9, 75.0, 122.0, 129.3,
137.6, 140.8, 179.5, 191.4; HRMS for C12H15NO4 (M+) calcd 237.0899,
found 237.0892.
Antibody Assays. Antibody stock concentrations were determined
by UV-vis spectroscopy using 1 OD ) 1.25 mg/mL at 280 nm and 1
mg/mL ) 6.7 µM. For initial screening, 100 µM substrate 1b and 20
µM antibody were dissolved in 4-mL glass vials in 50 mM Bicine, pH
8.5 with 5% DMSO. High-intensity photolysis was performed for 5-
and 20-min intervals, and an aliquot of each reaction was quenched in
a one-to-one ratio with an external standard (4′-fluoroacetanilide in 15%
CH3CN/85% H2O). Analysis by analytical HPLC was then performed.
For determination of initial rates from which kinetic parameters were
derived, samples were placed in 1-mL quartz cuvettes in a SPF-500C
spectrofluorimeter. Irradiation was performed at specified wavelengths
with a band-pass of 2.5 nm at a beam intensity of 4.2 × 10-9 einsteins/

12790 J. Am. Chem. Soc., Vol. 120, No. 49, 1998
Taylor et al.
min (for 280 nm). Light intensity was attenuated with neutral density
filters, and it was found that both the uncatalyzed (for formation of
11) and the antibody-catalyzed (for formation of 13) reaction rates were
linearly dependent on light intensity between 1.1 × 10-9 and 4.2 ×
10-9 einsteins/min. No drop in rate was observed upon decreasing the
substrate concentration; therefore the antibody-catalyzed reaction was
assumed to follow saturation kinetics, and rates were taken to be at
formula, C18H23N3O4; formula weight, 345.39; crystal system, ortho-
rhombic; space group, P2111; color, colorless; unit cell, a ) 6.051 Å,
b ) 15.928 Å, c ) 19.031 Å; Z ) 4; final R value ) 0.0733; goodness
of fit ) 1.063.
Radiolabeling Experiment. To a 2.5 µM solution (200 µL, PBS
pH 7.4) of mAb 8C7 was added 10 µM (1 µL of a 2 mM stock) of 1d.
The reaction mixture was photolyzed for 20 min, followed by the
addition of another aliquot (1 µL) of 1d and further photolysis for 20
min; this procedure was repeated 3×. Control experiments containing
antibody and substrate alone, as well as antibody and substrate without
photolysis, were performed in parallel. All reactions were diluted 1:1
with 6 M guanidinium-HCl and heated to 55 °C for 2 h, followed by
exhaustive dialysis against 100 mM ammonium acetate, pH 5.5. The
dialyzed samples were then diluted into 10 mL of scintillation fluid,
and the remaining activity was measured by scintillation counting. It
was determined that 0.016 nmol 14C (for a specific activity of 12.7
mCi/mmol) was bound per 0.5 nmol antibody combining site.
Vmax. The number of turnovers was found to be limited by both substrate
depletion and product inhibition by 12 and 13. The maximum number
of turnovers was determined by pulsing the reaction (2.5 µM antibody)
with aliquots (10 µM substrate each) of substrate stock solution and
then driving the reaction to completion with high-intensity irradiation.
After five pulses of substrate, 11 µM of 13 was found, indicating that
the reaction proceeded to two turnovers per antibody combining site
before product inhibition by 12 and 13 terminated the reaction.
Light intensity from an SPF-500 spectrofluorometer was measured
by both ferrioxalate actinometry and a radiant power meter from Oriel
Instruments and was determined to be 3.0 × 10-5 W at 280 nm. Using
the conversion factor of 4.3 × 105 J mol photons-1 at 280 nm and a
Enantioselectivity of mAb 8C7. Enantioselectivity was determined
as follows: A 0.4-mL volume reaction (PBS, pH 7.4, 5% DMSO)
containing 10 µM substrate and 10 µM antibody 8C7 was irradiated
until no substrate remained as judged by HPLC analysis. The aqueous
solution was extracted (3 × 1 mL, EtOAc), the organic phase was dried
with anhydrous sodium sulfate, and then the solvent was removed in
vacuo. The aqueous phase was analyzed by HPLC to verify extraction
of the product (>90%), and then 13 was analyzed (injected in 35 µL
of dry chloroform) on a Chiralpak HPLC column (Daicel). Condi-
tions: 18% 2-propanol/hexane with 0.1% trifluoroacetic acid, 40 °C;
retention times for enantiomers from analysis of authentic racemic 13,
21 min, 28 min. The ratio of the difference in peak areas over the sum
of the peak areas provided the ee value. It was also verified at this
time that authentic standards of 11 and 12 were racemic by chiral HPLC
analysis.
V
max value for conversion of 1b to 13 by mAb 8C7 of 5.5 × 10-3 µM
min-1 in a 1-mL irradiated volume, a quantum yield of 0.0012 was
obtained for the reaction. For the uncatalyzed reaction, a disappearance
quantum yield of 0.00076 at 280 nm was calculated using an observed
rate of disappearance of 1b at 60 µM concentration of 3.2 × 10-3 µM
min-1
.
The pH study versus rate study was performed using 60 µM
substrate, 2 µM mAb 8C7, and 5% DMSO in 50 mM 2,2-bis-
(hydroxymethyl)-2,2′,2′′-nitrilotriethanol (Bis-Tris) and 4-morpho-
lineethanesulfonic acid (MES) (100 mM NaCl) pH 6.0, 50 mM
phosphate buffer saline (PBS) pH 6.8, 50 mM PBS and 4-morpho-
linepropanesulfonic acid (MOPS) (100 mM NaCl) pH 7.4, 50 mM
Bicine and N-(tris(hydroxymethyl)methyl)glycine (Tricine) (100 mM
NaCl) pH 8.5, 50 mM Bicine and 2-(cyclohexylamino)ethanesulfonic
acid (CHES) (100 mM NaCl) pH 9.2, and 50 mM CHES (100 mM
NaCl) pH 10.0.
Antibody Fluorescence Quenching. Monoclonal antibody 8C7 (50
nM) in PBS pH 7.4, 5% DMSO was titrated with substrate 1b and
antibody fluorescence (excitation at 280 nm, emission at 340 nm)
followed. It was found that antibody fluorescence was quenched only
up to the addition of 2 equiv (100 nM) of 1b. A control experiment
using tryptophan in place of antibody demonstrated that no quenching
of tryptophan fluorescence occurred upon adding 1b, indicating that
the quenching of antibody fluorescence was not due to a filter effect.
X-ray Crystal Structure Determination for 1b. The substrate 1b
was crystallized over a period of one week by vapor diffusion from
acetonitrile/water to produce crystals suitable for X-ray analysis. A
crystal was mounted, and data were collected with a Rigaku AFC6R
diffractometer equipped with a copper rotating anode and a highly
oriented graphite monochromator. Unit cell dimensions and standard
deviations were obtained by least-squares fit to 25 reflections (50 <
2θ < 80°). The data were corrected for Lorentz and polarization effects.
The structure was solved by direct methods using SHELXS86. All
calculations were done on a Silicon Graphics Personal Iris 4D/35 and
an IBM-compatible PC using programs TEXSAN (data reduction),
SHELXL-93 (refinement), and SHELXTL-PC (plotting). Crystal data:
Acknowledgment. This work was supported by the National
Institutes of Health (GM 43858) and The Skaggs Institute for
Chemical Biology, and also in part by TEKES, Technology
Development Centre, Helsinki, Finland (J.T.Y-K.) We thank
Shenlan Mao for BIAcore analysis, Raj K. Chadha for the X-ray
crystal structure determination, Professor Peter Wirsching for
a critical reading of the manuscript, as well as Professor David
Millar and Professor Albert Eschenmoser for helpful discussions.
Supporting Information Available: 1H, 13C, and HMQC
(1H-13C coupled) NMR spectra for 13, 1H and HMQC NMR
spectra for 1b, 11, and 12, X-ray structural information for 1b,
the pH versus Vmax plot for formation of 13 by mAb 8C7, and
plots of wavelength versus rate data for the uncatalyzed and
antibody-catalyzed reactions as well as plots of absorption
spectra of 1b and 1b bound to mAb 8C7 (19 pages, print/PDF).
See any current masthead page for ordering information and
Web access instructions.
JA982711R