Highly efficient antibody-catalyzed deuteration of carbonyl compounds

Article abstract of DOI:10.1002/1521-3765(20020104)8:1<229::AID-CHEM229>3.0.CO;2-P

Antibody 38C2 efficiently catalyzes deuterium-exchange reactions at the α position of a variety of ketones and aldehydes, including substrates that have a variety of sensitive functional groups. In addition to the regio- and chemoselectivity of these reactions, the catalytic rates (k_cat) and rate-enhancement values (k_cat/k_un) are among the highest values ever observed with catalytic antibodies. Comparison of the substrate range of the catalytic antibody with highly evolved aldolase enzymes, such as rabbit-muscle aldolase, highlights the much broader practical scope of the antibody, which accepts a wide range of substrates. The hydrogen-exchange reaction was used for calibration and mapping of the antibody active site. Isotope-exchange experiments with cycloheptanone reveal that the formation of the Schiff base species (as concluded from the ¹⁶O/¹⁸O exchange rate at the carbonyl oxygen) is much faster than the formation of the enamine intermediate (as concluded from the H/D exchange rate), and both steps are faster than the antibody-catalyzed aldol addition reaction.

Full text of DOI:10.1002/1521-3765(20020104)8:1<229::AID-CHEM229>3.0.CO;2-P

FULL PAPER
Highly Efficient Antibody-Catalyzed Deuteration of Carbonyl Compounds
Avidor Shulman,[a] Danielle Sitry,[a] Hagit Shulman,[a] and Ehud Keinan*[a, b, c]
Abstract: Antibody 38C2 efficiently
catalyzes deuterium-exchange reactions
at the a position of a variety of ketones
and aldehydes, including substrates that
have a variety of sensitive functional
groups. In addition to the regio- and
chemoselectivity of these reactions, the
catalytic rates (kcat) and rate-enhance-
ment values (kcat/kun) are among the
highest values ever observed with cata-
lytic antibodies. Comparison of the sub-
strate range of the catalytic antibody
with highly evolved aldolase enzymes,
such as rabbit-muscle aldolase, high-
lights the much broader practical scope
of the antibody, which accepts a wide
range of substrates. The hydrogen-ex-
change reaction was used for calibration
and mapping of the antibody active site.
Isotope-exchange experiments with cy-
cloheptanone reveal that the formation
of the Schiff base species (as concluded
from the 16O/18O exchange rate at the
carbonyl oxygen) is much faster than the
formation of the enamine intermediate
(as concluded from the H/D exchange
rate), and both steps are faster than the
antibody-catalyzed aldol addition reac-
tion.
Keywords: aldol reaction ¥ catalytic
antibodies ¥ deuterium ¥ isotopic
labeling ¥ reaction mechanisms
conditions required for the exchange reaction.[4] The use of
amine catalysts that form intermediate enamines could
represent a solution to this problem. This approach, however,
is applicable only in rare cases in which the enamine can be
formed in water.[5]
Introduction
Deuterium- and tritium-labeled organic compounds have
become increasingly important for the role they play in
structure determination, in mechanistic studies, in the eluci-
dation of biosynthetic pathways and in biochemical studies.
One of the most commonly used deuteration/tritiation
methods is the acid- or base-catalyzed exchange of protons
a to a carbonyl function by using D2O or tritiated water.
Unfortunately, the fairly strenuous aqueous conditions re-
quired to complete this exchange reaction are incompatible
with substrates that contain acid/base-sensitive functional
groups.[1, 2] These difficulties have inspired much activity and
innovation at the level of process engineering.[3] Deuteration
of aldehydes is a particularly difficult task because most
aldehydes are incompatible with both the basic and acidic
Biocatalysis could provide an attractive strategy for achiev-
ing these goals. Biocatalysts have already proven to be useful
tools in synthetic organic chemistry, mainly due to their high
levels of catalytic efficiency under mild reaction conditions.[6]
Catalytic antibodies, in particular, have been shown to
catalyze a remarkable range of chemical reactions with
impressive rates.[7] Of special interest are the aldolase anti-
bodies, such as 38C2 and 33F12, which have been obtained by
reactive immunization.[8] These antibodies were elicited
against a 1,3-diketone hapten to evoke covalent bonding
between the hapten and an active-site lysine residue. One of
the most remarkable features of these antibodies is their
ability to accept a broad range of substrates.[1e, 9] It has been
shown that this feature reflects termination of the immuniza-
tion process before the antibody shape complementarity has
been fully optimized by somatic mutations.[10]
[a] Prof. Dr. E. Keinan, Dr. A. Shulman, D. Sitry, Dr. H. Shulman
Department of Chemistry and
Institute of Catalysis Science and Technology
Technion–Israel Institute of Technology
Technion City, Haifa 32000 (Israel)
We reasoned that aldolase antibodies could provide an
opportunity to solve the above-mentioned problems, since the
deuterium-exchange reaction could be carried out under
neutral conditions. Here we report that antibody 38C2
efficiently catalyzes the exchange reaction of a-protons in a
variety of ketones and aldehydes. The observed catalytic rates
(kcat) and rate-enhancement values (kcat/kun) are among the
highest values ever observed with catalytic antibodies. Fur-
thermore, we show that, unlike natural aldolase enzymes, this
catalytic antibody is much more able to accept a wide range of
Fax : (‡972)4-829-3913
E-mail: keinan@tx.technion.ac.il
[b] Prof. Dr. E. Keinan
Department of Molecular Biology and
The Skaggs Institute for Chemical Biology
The Scripps Research Institute, 10550 North Torrey Pines Road
La Jolla, California 92037 (USA)
Fax : (‡1)858-784-8732
[c] Prof. Dr. E. Keinan
Incumbent of the Benno Gitter &
Ilana Ben-Ami chair of Biotechnology, Technion.
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0229 $ 17.50+.50/0
229

E. Keinan et al.
FULL PAPER
substrates, including a variety of functionalized aldehydes or
ketones. This study also highlights mechanistic details of the
catalytic machinery of antibody 38C2. For example, it shows
that the deuteration reaction proceeds by the Schiff base
mechanism, that the formation of the Schiff base is rapid in
comparison with the tautomerization to the enamine inter-
mediate, and that both steps are faster than the antibody-
catalyzed aldol addition reaction.
be twice the value of the apparent reaction velocity measured
by NMR, in accordance with the Michaelis Menten kinetic
theory. For the substrates with lower KM values (e.g. Table 1,
entries 5, 10, 12), an excellent agreement was obtained
between the NMR and MS kinetic data.
It is difficult to assess the actual rate enhancement of the
deuteration reaction because of the divergence in the
reported uncatalyzed background rates.[13] We found it
necessary to run the uncatalyzed reaction of many substrates
for periods of time up to five months in order to detect
measurable quantities of products. Thus, even taking the most
generous estimate for the uncatalyzed rates, we found that, in
some cases, the rate enhancement (kcat/kun) of the deuteration
reaction exceeded 108-fold.
Beyond the practical applicability of the isotopic labeling of
organic compounds, the kinetic parameters for the substrates
point to important features of the antibody active site and
suggest mechanistic details of the exchange reaction. The
remarkably broad scope of carbonyl substrates accepted by
antibody 38C2 ranges from acyclic and cyclic ketones
(Tables 1 and 2, below) to aldehydes (Table 3, below) and
polyfunctional ketones (Table 4, below).
Results and Discussion
Antibody-catalyzed deuterium-exchange reactions were stud-
ied with a broad range of ketones and aldehydes (Tables 1 4,
below). All reactions were carried out in D2O under neutral
conditions and in the presence of catalytic amounts of the
commercially available antibody 38C2.[11] The exchange
reaction was followed either by mass spectrometry (GCMS)
1
or by H NMR. The high sensitivity of the GCMS method
with respect to substrate and product concentrations allowed
for the determination of the kinetic parameters of these
reactions (kcat and KM) by using Lineweaver Burk analysis
(Figure 1).[12] In all cases that were studied by GCMS, the
Acyclic ketones: Table 1 exhibits a variety of symmetrical and
nonsymmetrical ketones that undergo the 38C2-catalyzed
deuterium-exchange reaction. The regioselectivity of ex-
change at the a versus a' positions of nonsymmetrical ketones
is of particular interest because it may provide useful
information about the exchange mechanism. For example,
with linear ketones such as butanone (entry 2), the regiose-
lectivity is known to depend on the nature of the catalyst used.
Under strong acid catalysis, exchange of the methylene
protons is generally preferred over that of the methyl protons
(kCH /kCH ˆ 2.5). A similar trend was reported for neutral
2
3
conditions (kCH /kCH ˆ 1.6).[14] Conversely, the opposite regio-
2
3
selectivity has been observed under strong base catalysis
(kCH /kCH ˆ 0.7).[14] Other ratios within this range (kCH /kCH
ˆ
2
3
2
3
1.0 2.0) have been reported for a variety of weak acid and
base catalysts, such as acetate buffer solutions.[5a, 15] We found
that the antibody-catalyzed exchange with butanone at
neutral pH is significantly more regioselective (k /k
ˆ
Figure 1. Lineweaver Burk plots of the 38C2-catalyzed deuterium ex-
change reaction with three representative substrates: heptan-2-one (Ta-
ble 1, entry 6), cycloheptanone (Table 2, entry 4), and heptanal (Table 3,
entry 3).
CH2 CH3
3.0) than the highest value reported for acid catalysis. With
larger ketone substrates, in which the two alkyl groups are
considerably different from one another such as hexan-2-one
(entry 4), the discrimination in favor of the methylene is even
higher (kCH /kCH ˆ 5.8, Figure 2). These observations show
deuterium exchange reaction followed Michaelis Menten
kinetics and was effectively inhibited by acetylacetone, which
is a mechanism-based inhibitor of antibody 38C2.[8] All NMR
measurements were carried out with a high substrate concen-
tration (100mm) in order to approach substrate saturation and
thereby determine vmax and kcat. These experiments also
provided useful regioselectivity information with nonsym-
metrical ketones (vide infra), and chemoselectivity data with
polyfunctional carbonyl compounds. The NMR approach,
however, is limited in the case of substrates with large KM
values (e.g. Table 1, entries 3, 4). In such cases the apparent
vmax measured by NMR is actually smaller than the more
accurate vmax measured by MS. For example, in cases where
KM is nearly 100mm, vmax (as measured by MS) was found to
2
3
that although the reaction is carried out under neutral
aqueous conditions, its mechanism is reminiscent of a typical
acid-catalyzed exchange reaction. This is consistent with a
protonated Schiff base intermediate that is formed from the
carbonyl moiety and the lysine residue in the antibody active
site (I in Scheme 1). We have already reported that the 38C2-
catalyzed aldol reaction has characteristics that resemble
those of an acidic mechanism.[10]
That the observed regioselectivity under antibody catalysis
is higher than that expected for simple acid catalysis could
reflect a nonsymmetrical binding mode of the substrate with
subsequent formation of a nonsymmetrical Schiff base
intermediate, I. This covalent interaction could define two
230
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0230 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 1

Deuteration of Carbonyl Compounds
229 239
Table 1. Kinetic data for the antibody 38C2-catalyzed deuteration of acyclic
ketones.
[b]
Substrate
kcat (apparent)
(NMR)[a] [minÀ1
kcat (MS)[a] KM
kcat/kun
]
[minÀ1
]
[mm]
1
2
2.6
2.6 Â 107
2.4 Â 107
2.36 (CH2)
0.8 (CH3)
4.8
3
4
10.7
9.7
79
1.1 Â 108
9.7 Â 107
3.5 (CH2)
0.6 (CH3)
7.4
105
5
6
7
8
8.8
3.5
4.4
nd
47
16
13
8.8 Â 107
6.9 Â 107
4.4 Â 107
Figure 2. a) Regioselective 38C2-catalyzed deuteration of hexan-2-one
(Table 1, entry 4) as monitored by 1H NMR. b) Time-resolved 1H NMR
monitoring of the regioselective 38C2-catalyzed deuteration of hexan-2-
one. The methylene signal (d ˆ 2.6) decreased with time while that of the
methyl group (d ˆ 2.22) remained unchanged.
9
nd
10
0.4 (CH2)
0.9 (CH3)
1.4
45
31
1.4 Â 107
binding regions in the protein that accommodate the large and
small alkyl groups of the ketone substrate.[16] Considering the
known structure of the hapten used to raise antibody 38C2, it
seems likely that the large binding region is positioned away
from the lysine group towards the solvent. Our results suggest
that a general base residue located between the lysine group
and the solvent is responsible for the regioselective deproton-
ation (Scheme 1). This general base may preferentially
deprotonate the a' position directly (as shown in the Scheme)
or the a position by a relay of water molecules. This
assumption is supported by the consistent exchange rates at
the methyl group of several methyl ketones (approximately
0.8 minÀ1, Table 1 entries 2, 4, 10, 15) in contrast to the higher
rate with acetone (2.6 minÀ1), in which one methyl group is
positioned in close proximity to the general base.
Therefore, it is interesting to compare the kinetic param-
eters of different symmetrical ketones, R2CO (Table 1, entries
1, 3, 7, 8, 11), which can have only one binding mode. This
comparison provides a rough
11
12
nd
0.1
nd
0.1
1.0 Â 106
13
14
< 0.1
nd
nd
15
1[c]
10[c]
1.0 Â 107
[a] All kinetic parameters refer to the exchange of one hydrogen atom. [b] The
consistent kun value (10À7 minÀ1), which was measured for heptan-2-one (entry
4), cyclohexanone and cycloheptanone (Table 2, entries 3, 4), was applied for all
substrates. [c] Estimated value on the basis of relative rates. nd ˆ not detected.
estimate of the size of the
smaller binding region. Thus,
while the reactivity of the
smaller substrates, with R being
a methyl, ethyl, or n-propyl
group is quite similar, symmet-
Scheme 1.
rical ketones with larger alkyl
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0231 $ 17.50+.50/0
231

E. Keinan et al.
FULL PAPER
groups (n-butyl or i-butyl) are nonsubstrates. Therefore we
conclude that the smaller binding region in the antibody can
accommodate fragments no larger than an n-propyl group.
This is further supported by the finding that di-n-propyl
ketone has a smaller KM value than diethyl ketone.
Our structural hypothesis was reinforced by computational
modeling (Figure 3) with the crystal structure of another
aldolase antibody, 33F12,[8b] which is a very close homologue
Cyclic ketones: As can be seen in Table 2, the unsubstituted
monocyclic substrates (entries 2 5) are generally more
reactive than the acyclic ketones. The lack of reactivity of
cyclobutanone (entry 1) can be attributed to the relatively
high energy required to form its enamine derivative. The
highest reactivity was observed by NMR (Figure 4) with
Table 2. Kinetic data for the antibody-38C2-catalyzed deuteration of cyclic
ketones.
[b]
Substrate
kcat (MS)[a] [minÀ1
]
KM [mm]
kcat/kun
1
2
nd
29.9
25
5
3.0 Â 108
5.9 Â 108
3
4
84.9 (194)[d]
13.1
47.4
8.8
16
23
14
1.0 Â 108
4.7 Â 108
8.8 Â 107
5
6
Figure 3. Hexan-2-one bound in the active site of antibody 33F12, after
energy minimization with the Discover module of Insight II.
of antibody 38C2. That structure reveals several active-site
residues that could function as a general base: a carboxylate
(Glu H50), an alcohol (Ser H35), and three phenols (Tyr H95,
Tyr L96, Tyr L36) all positioned within 7 ä of the active-site
lysine (Lys H93). To gain more information about this
catalytic machinery we used the Insight II software package
to carry out energy minimization of the lysine-bound hexan-2-
one within the active site of antibody 33F12. We chose to
model the reduced Schiff base intermediate so as to allow
maximum conformational flexibility during the calculation.
This model suggests that Tyr H95, which is positioned in close
proximity to the substrate a-protons, is the catalytic base. All
the other basic residues are positioned further away from the
substrate. Moreover, this calculated model suggests that the
general basicity of Tyr H95 is enhanced by hydrogen bonding
to Glu H50. The observed regioselectivity with this substrate
could be explained by the close proximity (3.3 ä) between the
C3 hydrogens and Tyr H95 (Figure 3). By contrast, the
approach of this base to the C-1 hydrogens (which are
positioned 6 7 ä away from Tyr H95) is obstructed by the
substrate itself.
7
0.07
20
7.0 Â 105
9.5 Â 106
8
9
0.95
1[c]
13.5
[a c] See notes to Table 1. [d] The number in parentheses, which refers to
saturation conditions in NMR experiment, is more accurate than the value
obtained by MS. The error limit is approximately 15%.
An interesting experimental observation that further sup-
ports our hypothesis about the location and function of the
general base is the low reactivity of branched substrates
(entries 9 14). This trend probably reflects a sterically
hindered approach of the general base towards the acidic
proton, for example, the C-3 protons in 4-methylpentan-2-one
(entry 10). In this case the steric hindrance seems to be
Figure 4. 1H NMR trace of the antibody-38C2-catalyzed deuteration of
cyclohexanone (Table 2, entry 3).
sufficiently large to cause reversal of regioselectivity (kCH
/
2
kCH ˆ 0.44).
3
232
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0232 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 1

Deuteration of Carbonyl Compounds
229 239
Table 4. Kinetic data for the antibody 38C2-catalyzed deuteration of
bifunctional ketones.
cyclohexanone, which exhibits a record value of kcat in the
order of 200 minÀ1 (kcat/kun ꢀ 6 Â 108). As was the case with the
branched acyclic ketones (vide supra), substituted cyclic
ketones (entry 6) and bicyclic substrates (entries 7 9) are
much less reactive than the simple monocyclic ketones.
Substrate
kcat (apparent)
(NMR)[a] [minÀ1
kcat (MS)[a]
KM
[mm]
]
[minÀ1
]
1
2
0.4 (CH2)
0 (CH3)
0.08
3
Aldehydes: The efficient antibody-catalyzed deuteration of
aldehydes (Table 3) is of particular synthetic significance
because aldehydes are less chemically stable than ketones.
The aldehyde substrates we used are incompatible with even
mild reaction conditions (room temperature, neutral aqueous
medium) over the prolonged periods of time required for the
formation of any detectable amounts of deuterated products
in the uncatalyzed reaction. The data in Table 3 indicate that
aldehydes are generally much better substrates than ketones,
as reflected by their relatively higher kcat and lower KM values.
Although aldehydes are known to undergo rapid self-aldol
condensations, neither uncatalyzed nor 38C2-catalyzed self
condensations were observed with valeraldehyde and higher
homologues under the reaction conditions. Thus, the only
detectable reaction with such aldehydes was the deuterium
exchange.
ꢁ 0.4 (CH)[c]
0 (CH3)
3
4
nd
< 0.1
< 0.1
5
6
1
1
5[c]
10[c]
Table 3. Kinetic data for the antibody-38C2-catalyzed deuteration of
aldehydes.
[b]
Substrate
kcat (MS)[a] [minÀ1
]
KM [mm] kcat/kun
7
246
340
[d]
1
2
3
4
5
6
19.0
0.7
[d]
27.6
45.2
127
0.6
8
9
1
10[c]
[d]
4
[d]
0.66 (CH2)
0.14 (CH3)
0.4
10
1
[d]
50[c]
50[c]
1[c]
10
1
10[c]
[d]
1[c]
[d]
7
8
5.8
0.5
11
12
0.2 (CH2)
0.2 (CH3)
nd
[d]
32.5
3
[a c] See notes to Table 1. [d] d) No background rate could be determined
due to the complete loss of starting material before any exchange product
could be detected. The error limit is approximately 15%.
13
1
5[c]
14
15
1[c]
Bifunctional ketones: Chemoselective deuteration of bifunc-
tional carbonyl compounds (Table 4) not only affords syn-
thetically useful intermediates but also provides insight into
the catalytic machinery of the antibody. Under neutral
conditions, antibody 38C2 catalyzes the deuteration of acid-/
base-sensitive substrates, such as ketoketals (entry 8), keto-
10[c]
[a and c] See notes to Table 1. The error limit is approximately 15%.
0947-6539/02/0801-0233 $ 17.50+.50/0
233
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002

E. Keinan et al.
FULL PAPER
esters (entries 9, 10), and a-haloketones (entries 11, 13). No
background reaction could be detected with these substrates.
For example, with ethyl 5-oxohexanoate (entry 9) we failed to
detect any product of the uncatalyzed deuteration reaction
due to the faster consumption of the substrate in various side
reactions even in a neutral aqueous medium. An attempt to
catalyze this reaction with n-butylamine yielded the corre-
sponding butylamide as the only detectable product. With
38C2, however, highly chemoselective deuterium exchange
occurred at the position a to the carbonyl but not a to the
ester group (Figure 5). Furthermore, as in the case of the
Figure 6. Inhibition of the deuteration of cycloheptanone (Table 2, entry
&
^
*
fluoro-
4) by various ketones. no inhibitor;
cyclohex-2-en-1-one;
~
acetone; 3-hydroxybutanone; Â acetylacetone.
site residue by additional hydrogen bonding. In hydroxy-
ketone substrates, this additional bonding could stabilize the
Schiff base intermediate and inhibit its tautomerization to the
enamine. Yet hydroxyketones may still be active in the aldol
addition reaction because their binding mode and propensity
to form an enamine intermediate may change upon addition
of an aldol acceptor molecule.
An interesting case of high regioselectivity was exhibited by
2-methoxycyclohexanone (Table 4, entry 5). We had initially
observed in preliminary mass spectrometry experiments that
the exchange reaction does not progress beyond 35%
conversion. This observation could reflect exchange of just
one out of the three acidic protons in the molecule. Indeed,
Figure 5. 1H NMR trace of the regioselective 38C2-catalyzed deuteration
of ethyl 5-oxohexanoate (Table 4, entry 9). The C4 methylene signal (d ˆ
2.7) decreased with time, while that of the methyl group (dˆ 2.24) decreased
very slowly. All other signals were totally preserved during this reaction.
nonsymmetrical acyclic ketones, deuteration of the methylene
1
further H NMR experiments revealed that, even after long
hydrogens was highly favored (kCH /kCH ˆ 4.7). Interestingly,
2
3
reaction times, only one proton (at the C-6 methylene) was
although fluoroacetone (entry 11) is a moderate substrate,
trifluoroacetone (entry 12) is a nonsubstrate in this reaction.
Other functional groups that were found to be compatible
with the antibody-catalyzed reaction conditions are ethers
(entries 4 6), unsaturated ketones (entry 14) and 1,2-
diketones (entry 16).
1
exchanged for deuterium. Closer inspection of the H NMR
spectra revealed that only the axial proton at C-6 (td, J ˆ 13.2,
5.9 Hz) was exchanged by deuterium. Similar regioselectivity
was observed with 2-methylcyclohexanone (Table 2, entry 6).
Further studies on the use of antibody 38C2 for the kinetic
resolution of these substrates and other substituted cyclic
ketones are currently underway.
Of particular interest is the behavior of the hydroxyketones
(entries 1 3), in which the presence of a hydroxy substituent
in the molecule significantly impairs the deuterium exchange
reaction. This observation is intriguing, particularly when
considering the high reactivity of hydroxyacetone and hy-
droxybutanone in the 38C2-catalyzed aldol addition.[8] To gain
more information about the interaction of hydroxyketones
with 38C2 we carried out inhibition experiments in which the
deuteration of cycloheptanone took place in the presence of
25 mol% (with respect to the substrate) of acetylacetone, 3-
hydroxybutanone, cyclohex-2-en-1-one, and fluoroacetone
(Figure 6). As expected, acetylacetone caused almost com-
plete inhibition of the deuteration reaction. The low levels of
inhibition by cyclohex-2-en-1-one and fluoroacetone indicate
that these two ketones are competitive inhibitors, whose
binding affinities are similar to that of cycloheptanone. Yet,
the much stronger inhibition with 3-hydroxybutanone sug-
gests that the active site binds this compound more efficiently
than a simple ketone. We suggest that the hydroxyl group,
unlike simple ketones, may interact with an active
Mechanistic studies:
The 38C2-catalyzed deuteration reaction: It is interesting to
compare the mechanistic details of the antibody-catalyzed
exchange reaction with similar reactions catalyzed by natural
enzymes. Prior observations with antibody-38C2-catalyzed
aldol, retroaldol, and decarboxylation reactions[17] suggest
that all of these reactions involve the formation of a Schiff
base intermediate with an active-site lysine residue. Therefore
it is likely that the deuterium exchange reaction involves a
multistep process in which every proton is exchanged
individually in a series of equilibria, as exemplified by
Scheme 1. An important question relates to the relative rates
of the two main steps, that is, formation of the protonated
Schiff base, I, and tautomerization of the latter to enamine II.
If the first step is faster than the second, the expected kinetic
behavior of the entire process is that of consecutive pseudo-
first-order reactions with the formation and disappearance of
234
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0234 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 1

Deuteration of Carbonyl Compounds
229 239
partially deuterated products, [D1], [D2], [D3] etc. Alterna-
tively, if the second step is faster than the first one, it is to be
expected that no partially deuterated intermediates would be
observed during the overall reaction.
To check this issue, we followed the progress of the
exchange reaction of cycloheptanone, 1, by GCMS using a
low concentration of catalyst to allow detailed monitoring of
this multistep reaction (Figure 7a). It is noteworthy that the
antibody-catalyzed deuteration reaction was complete before
any background reaction could be detected. The data in
Figure 7a fit well with a kinetic model of four consecutive first
order reactions where each step is independent of the
others.[18] A single rate constant with statistical corrections
describes the seemingly complicated array of data. Thus, the
first-order rate constant (k) related to the first step (mono-
deuteration of 1 to produce [D1]1) was easily obtained from
the exponential curve, which describes the disappearance of
[D0]1. The rate constant for the next step (formation of [D2]1)
3
was determined as ³4 k. The rate constant for the third step
1
(formation of [D3]1) was ³2 k. Accordingly, the rate constant
for the fourth step (formation of [D4]1) was 1³4 k. The excellent
agreement between the experimental data and the calculated
curves in Figure 7a confirmed that the exchange reaction
occurs in a stepwise manner. This agreement justifies the
neglect of a second-order isotope effect. Thus, referring to
Scheme 1, kÀ1 is large compared with k2, and the process that
corresponds to k1/kÀ1 can be treated as a rapid pre-equilib-
rium. Similar behavior has been observed by Pratt in the
RMA-catalyzed (RMA ˆ rabbit-muscle aldolase) deuteration
of hydroxyacetone phosphate[19] and by Westheimer in the
acetoacetate decarboxylase-catalyzed de-deuteration reac-
tion with [D6]acetone.[20] This model of a rapid pre-equilib-
rium fits the kinetics of most of the ketone and aldehyde
substrates shown in Tables 1 4.
To further verify the assumption that the Schiff base
intermediate is formed in a rapid pre-equilibrium, we
examined the antibody-catalyzed 16O/18O exchange of the
carbonyl oxygen in cycloheptanone. The reaction, carried out
with cycloheptanone in H218O, was found to be very fast and to
require a low antibody concentration (0.1 mgmLÀ1) and a
short reaction time (2 min). An excellent fit with Michaelis
Menten kinetics was observed (kcat ˆ 418 minÀ1, KM ˆ 21mm).
Comparing these results with the kinetic parameters for the
38C2-catalyzed H/D exchange with the same substrate (kcat
ˆ
13 minÀ1, KM ˆ 16mm), one may conclude that the formation
of intermediate I is indeed faster (by a factor of at least 30)
than the formation of intermediate II. Thus, we can conclude
that the rate-determining step in the deuterium exchange
reaction is the formation of the enamine intermediate, II.
Figure 7. a) Antibody-38C2-catalyzed deuteration of cycloheptanone
(Table 2, entry 4). All data points were obtained from time-resolved
GCMS measurements. The calculated curves are based on a four-step series
of consecutive first-order reactions by using a single rate constant (kobs
ˆ
0.00045 sÀ1) multiplied by the appropriate statistical factor. The back-
ground rate is undetectable under the reaction conditions. b) Antibody-
38C2-catalyzed deuteration of 1-methylpiperid-4-one (Table 4, entry 7).
All data points were obtained from time-resolved GCMS measurements.
The calculated curves are based on the kinetic model shown in Scheme 2.
The kinetic behavior of the background (uncatalyzed) reaction (c) agrees
with the simple stepwise model (calculated lines).
The 38C2-catalyzed aldol reaction: This reaction is a multistep
cascade of events in which the formation of the Schiff base (I)
and enamine (II) intermediates represent the first two steps.
The slowest step in the enzyme-catalyzed aldol addition is the
À
reaction of II with the carbonyl acceptor to form a new C C
bond. Accordingly, the antibody-catalyzed aldol addition is
expected to be a slower process in comparison with the
deuterium-exchange reaction. Indeed, comparison of the
reported 38C2-catalyzed aldol reaction rates with our ex-
change reaction rates for the same ketone substrates reveals
that the aldol reactions are generally 100 10000-fold slow-
er.[1e, 9] The aldol reaction has a much more sterically
demanding transition state than the exchange reaction. This
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0235 $ 17.50+.50/0
235

E. Keinan et al.
FULL PAPER
explains why the difference between the aldol and exchange
reaction rates for a given substrate depends strongly on the
substrate size. For example, the deuterium-exchange reaction
with acetone is 500-fold faster than the aldol reaction between
acetone and 4-acetamidodihydrocinnamaldehyde. Yet, with
more sterically demanding donors, the difference between the
corresponding reaction rates increases. For example, the
exchange reaction with cyclohexanone is 30000 fold faster
than the aldol reaction between cyclohexanone and the
above-mentioned aldehyde acceptor. Cyclooctanone, which
is an even more sterically demanding ketone, is an active
substrate for the exchange reaction (Table 2, entry 5) but is a
nonsubstrate for the aldol reaction.[9]
The H/D kinetic isotope effect in the exchange reaction
(Figure 8) was studied by comparing the rate of the trans-
formation of 1 into [D4]1 (in D2O) with the rate of the reverse
transformation ([D4]1 into 1, in H2O). No significant primary
The 38C3-catalyzed deuteration of bifunctional compounds:
While the above-described rapid pre-equilibrium results in
stepwise deuteration of simple ketones, the situation could
change with bifunctional substrates. For example, the kinetic
behavior exhibited by 1-methylpiperid-4-one (Figure 7b)
represents a different mechanistic model in which a rapid
pre-equilibrium no longer exists. While the background
deuteration reaction of 1-methylpiperid-4-one follows the
stepwise kinetics described above (see Figure 7c), the kinetic
behavior of the antibody-catalyzed reaction with this sub-
strate reflects a more complex sequence of events.
In principle, all substrates may undergo more than a single
proton exchange in every encounter with the antibody active
site. While in the case of cycloheptanone (Figure 7a) such
double and triple deuteration steps seem negligible, multiple
deuteration reactions become significant in the case of
1-methylpiperid-4-one. As can be seen from Figure 7b, the
kinetic behavior of this substrate, whose catalytic conversion
is much more efficient than that of cycloheptanone, does not
fit the simple stepwise kinetic model. In order to estimate the
rate constants in this case, we
Figure 8. Lineweaver Burk plots of the 38C2-catalyzed deuteration
reaction of cycloheptanone in D2O (kcat ˆ 13minÀ1, KM ˆ 16mm), and de-
deuteration of [D4]cycloheptanone in H2O (kcat ˆ 27minÀ1, KM ˆ 29mm).
kinetic hydrogen isotope effect was found to be associated
with the exchange reaction. The small isotope effect (kD/kH ˆ
2, kcat/KM(D)/kcat/KM(H) ˆ 1.1) can be attributed to solvent
and to equilibrium isotope effects. The lack of a large isotope
effect, which would be expected if the formation of the
enamine occurred in the rate determining step, probably
reflects a nonsymmetrical transition state. Similar observa-
tions and interpretations have already been reported for
RMA.[19] It is remarkable that a catalytic antibody and highly
evolved enzymes share the fine details of their catalytic
mechanisms.
developed a set of integrated
rate equations on the basis of
the kinetic model shown in
Scheme 2. To simplify this rath-
er complex kinetic model, we
did not take the background
reaction into account. Addi-
Comparison with enzymes: Although there are a few reported
cases of an enzyme-catalyzed exchange of protons a to
carbonyl groups under mild conditions, practically all of them
were limited to very few substrates. For example, pyruvic acid
was deuterated by using oxaloacetate decarboxylase,[21] ace-
tone and butanone were deuterated by using acetoacetate
decarboxylase under slightly acidic conditions (pH 5.9),[22]
dihydroxyacetone phosphate was tritiated by using RMA,[23]
and enantioselective deuteration of 2-carboxycyclohexanone
was achieved with acetoacetate decarboxylase.[24] Few other
examples of substrate-specific deuterium exchange reactions
have been reported.[25]
The results summarized in Tables 1 4 highlight the re-
markably broad range of substrates that are accepted by
antibody 38C2. To compare the substrate range of 38C2 with
that of the widely studied RMA, we studied the biocatalyzed
deuterium-exchange reaction with a representative selection
of substrates: heptan-2-one, heptan-4-one, cycloheptanone,
cyclooctanone, octanal, hydroxyacetone, 3-hydroxybutanone,
and ethyl-5-oxohexanoate. No RMA-catalyzed reaction was
observed with any of these substrates. Very modest catalytic
tionally, all the reverse reac-
Scheme 2.
tions and the triple deuteration
reactions (represented in
Scheme 2 by k6 and k9) were
neglected. Although this treatment should be considered as a
qualitative analysis, it yielded an interesting conclusion that
the values of the rate constants of the single and double
deuteration steps are approximately of the same magnitude.
This finding suggests that kÀ1 (Scheme 1) is no longer much
larger than k2. This, in turn, results in an increased residence
time of this substrate in the active site of 38C2. The increased
reactivity of this substrate could also result from the close
proximity between the tertiary amine, which is probably
protonated under neutral pH, and the Schiff base. This
proximity could facilitate the conversion of the protonated
Schiff base intermediate (I in Scheme 1) to the enamine
intermediate II.
236
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0236 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 1

Deuteration of Carbonyl Compounds
229 239
Catalysts: Catalytic antibody 38C2 was purchased from Aldrich in the form
of a lyophilized powder. For all deuteration experiments, the antibody was
dissolved in a solution of D2O containing 0.1m NaCl. The antibody final
concentration was established by its absorbance at 280 nm. The same
results were obtained with an antibody solution that was obtained from
dialysis of a concentrated antibody sample in phosphate-buffered saline
(PBS H2O, pH 7.4) against a solution of 0.1m NaCl in D2O. In the reactions
in which a deuterium atom was exchanged for hydrogen, the antibody was
dissolved in H2O (HPLC grade) containing 0.1m NaCl. For the oxygen
exchange experiments, the antibody (1 mg) was dissolved in 1 mL of H218O
(97%, purchased from Rotem Industries Ltd., Israel) containing 0.1m
NaCl.
activity (significantly lower than that of 38C2) was observed
with heptanal and cyclohexane carboxaldehyde. RMA ex-
hibited high reactivity only with its specific substrate,
dihydroxyacetone phosphate.
Conclusion
We have shown here that antibody 38C2 catalyzes deuterium
exchange reactions with a broad variety of ketones and
aldehydes. The catalytic rates and rate-enhancement values
are among the highest values ever observed with catalytic
antibodies. Since all reactions were carried out under neutral
conditions the method was compatible with aldehydes and
other sensitive functional groups. The usefulness of this
catalytic reaction is evident from the ability to carry out the
reaction with a broad scope of substrates, from the regiose-
lectivity with nonsymmetrical ketones, and from the chemo-
selectivity observed with polyfunctional substrates. A relative
reactivity study with many substrates represents a useful
method of mapping and calibrating the active site. The
observed regioselectivity suggests the existence of two bind-
ing regions in the antibody that accommodate the small and
large alkyl groups of unsymmetrical ketones. Furthermore,
qualitative computational modeling suggests that either Tyr
H95 or Glu H50 or a combination of both residues could
function as the general base in the catalytic machinery.
Determination of the background reactions: The background deuteration
reaction was monitored along with the catalytic reaction. In most cases,
during the time period suitable for obtaining the kinetic parameters of the
catalyzed reaction, no background products could be detected. For entries
6 in Table 1; 3, 4 in Table 2; 1, 3, 4 in Table 3; and 9 in Table 4, the
background reaction was measured over a period of 5 months. Five
different samples were prepared for each substrate (0.5mm in D2O
containing 10% acetonitrile and 0.1m NaCl). These samples were main-
tained at room temperature for a period of 5 months and then analyzed by
GCMS-CI. The average rate of these measurements was calculated as the
uncatalyzed rate of the deuteration reaction.
NMR kinetics: All deuteration reactions were conducted by using 0.1m of
substrate and 1 mgmLÀ1 of antibody in D2O containing 0.1m NaCl. The
control samples were measured in D2O containing 0.1m NaCl without
antibody. The reaction mixtures contained 10% v/v of deuterated
acetonitrile, except in the case of acetone, butanone, acetol, fluoroacetone,
and 1,1,1, trifluoroacetone.
With most substrates analyzed by NMR, a peak resulting from nonex-
changeable protons was used as an internal standard to calibrate the
integration. This allowed monitoring of the decrease in peak area of the a-
hydrogens. With substrates containing only exchangeable protons, the
reaction mixture contained 0.1m of acetonitrile, and the acetonitrile signal
was used as an internal standard.
Isotope-exchange experiments with cycloheptanone reveal
that the formation of the Schiff base species (as concluded
from the 16O/18O exchange rate at the carbonyl oxygen) is
much faster than the formation of the enamine intermediate
(as concluded from the H/D exchange rate), and both steps
are faster than the antibody-catalyzed aldol addition. These
findings reinforce the hypothesis that the slowest step in the
The initial rate of hydrogen exchange was monitored during the first 24 h at
time intervals of 2 3 h. Each sample was analyzed by 24 scans by using a
relaxation delay of 10 s and file size of 16 K. In the case of cyclohexanone,
due to the extremely high exchange rate, the reaction was monitored only
during the first hour at time intervals of 2 min by using 16 scans with a
relaxation delay of 5 s.
À
catalytic aldol addition reaction is the formation of a new C C
MS-CI kinetics: In all cases, a preliminary investigation of the reaction rate
was conducted by monitoring the deuteration reaction with 2mm substrate
and 0.2 mgmLÀ1 antibody. This allowed for the determination of the time
needed to measure the initial rates (approximately 5% conversion) with a
good signal to noise ratio. Samples of the catalyzed and uncatalyzed
reactions (70 mL each) were withdrawn at different time intervals during a
two day period. These samples were quenched with dichloromethane
(1 mL) and shaken vigorously for 1 min to allow efficient extraction of the
substrate and products to the organic solvent and to destroy all antibody
activity. The organic phase was separated and analyzed by GCMS-CI.
bond. Time-resolved studies show that polydeuteration of
simple ketones and aldehydes occurs by a stepwise mecha-
nism in which each deuteration step is independent. By
contrast, with bifunctional substrates, such as 1-methylpiper-
id-4-one, multiple exchanges may occur before substrate
dissociation. Comparison of the substrate range of the
catalytic antibody with highly evolved aldolase enzymes, such
as rabbit-muscle aldolase, highlights the much broader
practical scope of the antibody. Further applications of these
results are currently being developed, particularly with
respect to enantioselectivity and preparative-scale reactions.
The kinetic measurements for each substrate were carried out by
monitoring
a set of 15 reaction mixtures with a constant antibody
concentration (0.1 to 1 mgmLÀ1 for different substrates) and various
substrate concentrations (0.1mm to 10mm). All kinetic measurements were
conducted in D2O containing 0.1m NaCl and 10% acetonitrile. The
reactions were allowed to reach 5% conversion, and then quenched with
dichloromethane as described for the preliminary experiment. Analysis of
the product distribution was carried out by MS. The rate of the uncatalyzed
reaction was subtracted when necessary. The kinetic parameters kcat and KM
were elucidated from Lineweaver Burk plots.
Experimental Section
General methods: All kinetic studies that were based on mass spectrometry
analysis were carried out with a Finnigan TSQ-70 GCMS-CIlinked to a
Varian GC equipped with a DB-5ms capillary column. 1H NMR kinetic
measurements were recorded on a Bruker AM200 operating at 200 MHz
or a Bruker AM400 operating at 400 MHz. UV/Vis measurements were
recorded on a Shimadzu UV-1601 spectrophotometer with a UV/Vis
microcuvette (100 mL). Long-period reactions were maintained at 258C by
using a Friocell incubator. All substrates (analytical grade) and deuterated
solvents were purchased from Aldrich and were used without further
purification. All solvents (HPLC grade) were purchased from either
Aldrich or Merck.
Time resolved MS experiments: The exchange reaction was followed as a
function of time. Each reaction was carried out as described above for the
preliminary kinetic experiments, by using 0.1 mgmLÀ1 of antibody and a
substrate concentration of 0.2mm. Samples were withdrawn at various time
intervals over a period of 24 h and analyzed by MS.
Kinetic models for the polydeuteration reactions: The time-resolved MS
technique allowed for the determination of each of the partially deuterated
species in terms of molar ratio. This analysis reflected the product
distribution as a function of time. The following integrated rate equations
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0237 $ 17.50+.50/0
237

E. Keinan et al.
FULL PAPER
were developed[18] in accordance with the consecutive pseudo first order
reactions model ([D0] ! [D1] ! [D2] ! [D3] ! [D4]) in which d00 is the
initial concentration of the molecule with no deuterium atoms and dn refers
to the concentration of molecules with n deuterium atoms:
antibody 33F12 were acquired from the Brookhaven Protein Data Bank
(entry 1axt). All water molecules and all residues except those amino acids
that construct the Fv region of the protein were deleted from the model.
The geometry of the Fv region (Light 1 107, Heavy 5 110) was optimized
with no constraints. A three-carbon unit simulating acetone was then
bound to the e-nitrogen of lysine H-93 by using the Builder module. The
resultant amine was protonated to simulate the positive charge of the
protonated Schiff base intermediate in the antibody-catalyzed reaction.
The geometry of the structure was optimized until the root-mean-square
(rms) energy gradient was less than 0.01 KcalmolÀ1. The larger substrate
molecules were built stepwise onto the resultant scaffold. The geometry of
the structures was repeatedly optimized after each modification. The
dielectric constant was set to 1 throughout the calculations. Glutamate H50
was deprotonated by setting the pH to 7.0, which was the pH of all
antibody-catalyzed reactions. In all calculations with the modified protein,
the bound substrate as well as all six CDR regions of the Fv (H31 37,
H50 65, H90 102; L-24 36, L-50 56, L-89 98) were not constrained.
d0 ˆ d00eÀk t
0
d1 ˆ 4d00(eÀk t À eÀk t
)
1
0
d2 ˆ 6d00(eÀk t À 2eÀk t ‡ eÀk t
)
0
1
2
d3 ˆ 4d00(ÀeÀk t ‡ 3eÀk t À 3eÀk t ‡ eÀk t
)
0
1
2
3
d4 ˆ d00(1 ‡ eÀk t À 4eÀk t ‡ 6eÀk t À 4eÀk t
)
0
1
2
3
These equations were used to produce the lines shown in Figure 7a,
representing the best fit between the calculated model and the exper-
imental results. For the case of 1-methylpiperid-4-one (Figure 7b), another
set of integrated rate equations, developed on the basis of the kinetic model
shown in Scheme 2, was used.
Preparation of [D4]cycloheptanone: Cycloheptanone (50 mmol, 5.9 mL)
was dissolved in anhydrous ether (30 mL) and D2O (20 mL) under argon.
Aqueous NaOD (8 g, 40% in D2O, 80 mmol) was added, and the mixture
was stirred vigorously for 5 days. D2O and NaCl were added, and the
mixture was extracted twice with diethyl ether. The combined organic
phase was dried over magnesium sulfate and filtered, and the solvent was
removed under reduced pressure to yield 5 g of partially deuterated
Acknowledgement
We thank Dr. Noam Adir for assistance with the computational methods.
We are grateful to Jacob Katzir and Alex Etinger of the Israel Center for
Mass spectrometry, Technion-Israel Institute of Technology for their
dedicated assistance with MS measurements. We thank the Israel Science
Foundation and the Skaggs Institute for Chemical Biology for financial
support. A.S. thanks the Clore Foundation for a graduate fellowship.
1
cycloheptanone (by H NMR analysis). This procedure was repeated two
more times to yield 3.9 g [D4]cycloheptanone (isotopic purity above 98%).
1H NMR (CDCl3): d ˆ 1.62 (brs).
Kinetic H/D isotope effect: Two kinetic experiments were carried out
according to the general procedure described above by using GCMS. In the
first experiment, nondeuterated cycloheptanone was treated with antibody
38C2 (0.1 mgmLÀ1) in saline solution (D2O containing 0.1m NaCl). The
second experiment was carried out in the same way with [D4]cyclo-
heptanone in H2O containing 0.1m NaCl. Both experiments were carried
out with substrate concentrations between 0.4 and 4.5mm. The reaction
mixtures were quenched after 40 min at room temperature.
[1] a) P. G. Gassman, D. H. Aue, D. S. Patton, J. Am. Chem. Soc. 1968, 90,
7271; b) T. H. Kinstle, O. L. Chapman, M. Sung, J. Am. Chem. Soc.
1968, 90, 1227.
[2] a) J. Seibl, T. G‰umann, Helv. Chim. Acta 1963, 46, 2857 b) D. S.
Weinberg, C. Djerassi, J. Org. Chem. 1966, 31, 115.
Oxygen-exchange experiments: The rate of Schiff base formation was
measured by following the 16O/18O exchange rate of the carbonyl oxygen of
cycloheptanone. A set of nine reaction mixtures, all containing antibody
38C2 (0.1 mgmLÀ1) and cycloheptanone (0.4 4.0mm) in H218O (containing
0.1m NaCl), was kept at room temperature. All reactions were quenched
after 2 min as described above and analyzed by MS. The background
reaction in the absence of 38C2 was allowed to stand for 40 minutes at
room temperature before being quenched. A Lineweaver Burk plot of the
¬
[3] a) G. Frejaville, J. Jullien, French Pat. 1,424,496, 1964; Chem. Abst.
1966, 65, 10169; b) K. Mislow, M. A. W. Glass, H. B. Hopps, E. Simon,
G. H. Wahl, Jr., J. Am. Chem. Soc. 1964, 86, 1710; c) M. Senn, W. J.
Richter, A. L. Burlingame, J. Am. Chem. Soc. 1965, 87, 680; d) D. K.
Albert, S. Meyerson, Analyt. Chem. 1967, 39, 1904.
[4] Even with weak bases, such as tertiary amines, the exchange reaction
with aldehydes results in a loss of starting material, requires a lot of
deuterium oxide, and gives only an asymptotic approach to isotopic
purity: W. Kirmse, H.-D. von Scholz, H. Arold, Justus Liebigs Ann.
Chem. 1968, 22, 711.
[5] a) J. Hine, J. Mulders, J. G. Houston, J. P. Idoux, J. Org. Chem. 1967, 32,
2205; b) J. Hine, B. C. Menon, J. Mulders, J. P. Idoux, J. Org. Chem.
1967, 32, 3850.
results (9 points) afforded a good linear fit with kcat ˆ 418minÀ1 and KM
ˆ
21mm (R2 ˆ 0.997).
Enzymatic reactions: Rabbit-muscle aldolase (EC 4.1.2.13), which was
purchased from Sigma in the form of a lyophilized powder, was dissolved in
D2O containing 0.1m NaCl. The kinetic measurements with this enzyme
were based on 1H NMR and GCMS analyses as described above for
antibody 38C2.
[6] C.-H. Wong, G. M. Whitesides, Enzymesin Synthetic Organic
Chemistry, Pergamon, Oxford, 1994.
[7] a) P. G. Schultz, R. A. Lerner, Science 1995, 269, 1835; b) E. Keinan,
R. A. Lerner, Isr. J. Chem. 1996, 36, 113; c) R. A. Lerner, S. J.
Benkovic, P. G. Schultz, Science 1991, 252, 659; d) N. R. Thomas, Natl.
Prod. Rep. 1996, 479.
Inhibition experiments: Preliminary inhibition experiments were conduct-
ed as described above for the preliminary kinetic experiments with the
following substrates: heptan-2-one, heptanal, and cycloheptanone. The
inhibitors used were: acetylacetone, cyclohex-2-en-1-one, 3-hydroxybuta-
none and fluoroacetone. In all cases a substrate concentration of 2mm and
an inhibitor concentration of 0.5mm were used. The reactions were
monitored over a period of 48 h.
[8] a) J. Wagner, R. A. Lerner, C. F. Barbas III, Science 1995, 270, 1797;
b) C. F. Barbas III, A. Heine, G. Zhong, T. Hoffmann, S. Gramatiko-
va, R. Bjˆrnestedt, B. List, J. Anderson, E. A. Stura, E. A. Wilson,
R. A. Lerner, Science 1997, 278, 2085; c) T. Hoffmann, G. Zhong, B.
List, D. Shabat, J. Anderson, S. Gramatikova, R. A. Lerner, C. F.
Barbas III, J. Am. Chem. Soc. 1998, 120, 2768; d) B. List, D. Shabat,
C. F. Barbas III, R. A. Lerner, Chem. Eur. J. 1998, 4, 881; e) G. Zhong,
D. Shabat, B. List, J. Anderson, S. C. Sinha, R. A. Lerner, C. F.
Barbas III, Angew. Chem. 1998, 110, 2609; Angew. Chem. Int. Ed.
1998, 37, 2481; f) G. Zhong, R. A. Lerner, C. F. Barbas III, Angew.
Chem. 1999, 111, 3957; Angew. Chem. Int. Ed. 1999, 38, 3738; g) S. C.
Sinha, J. Sun, G. Miller, C. F. Barbas III, R. A. Lerner, Org. Lett. 1999,
1, 1623; h) J. M. Turner, T. Bui, R. A. Lerner, C. F. Barbas III, B. List,
Chem. Eur. J. 2000, 6, 2772.
Stereoselective deuteration of 2-methoxycyclohexanone: The deuterium
exchange reaction was monitored by 1H NMR (400 MHz) according to the
general procedure described above. Both the antibody-catalyzed reaction
and the corresponding uncatalyzed reaction were carried out for 42 days at
room temperature. 1H NMR (D2O): d ˆ 4.15 (dd, J ˆ 12.0, 6.2 Hz, 1H;
H2ax), 3.44 (s, 1H; H7), 2.54 (td, J ˆ 13.2, 5.9 Hz, 1H; H6ax), 2.44 (m, 2H;
H6eq, H3eq), 2.11 (dm, 1H; H5eq), 1.93 (dm, 1H; H4eq), 1.78 (qt, J ˆ 13.0,
3.59 Hz, 1H; H4ax), 1.62, (qt, J ˆ 13.0, 4.0 Hz, 1H; H5ax), 1.55 (qd, J ˆ 12.4,
4.0 Hz, 1H; H3ax).
Molecular mechanics calculations: The molecular mechanics studies were
accomplished by using the Discover module with CVFF force field within
the Insight-II molecular simulation package. The calculations were carried
out on a Silicon Graphics Indigo work station. The initial coordinates for
[9] T. Hoffmann, G. Zhong, B. List, D. Shabat, J. Anderson, S.
Gramatikova, R. A. Lerner, C. F. Barbas III, J. Am. Chem. Soc.
1998, 120, 2768.
238
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0238 $ 17.50+.50/0
Chem. Eur. J. 2002, 8, No. 1

Deuteration of Carbonyl Compounds
229 239
[10] H. Shulman, C. Makarov, A. K. Ogawa, F. Romesberg, E. Keinan, J.
Am. Chem. Soc. 2000, 122, 10743.
[18] S. Benson, The Foundationsof Chemical Kinetics , McGraw Hill, New
York, NY, 1960, p. 39.
[11] Aldrich catalog 47, 995 0.
[19] R. S. Pratt, Biochem. 1977, 16, 3988.
[12] Due to limited substrate solubility, the kinetic measurements could
not be carried out with higher substrate concentrations.
[13] a) J. Warkentin, R. A. Cox. J. Org. Chem. 1968, 33, 1301; b) J. Hine,
K. G. Hampton, B. C. Menon, J. Am. Chem. Soc. 1967, 89, 2664; c) J.
Jullien, N. Thoi-Lai, Bull. Soc. Chim. Fr. 1970, 11, 3948.
[14] C. Rappe, Acta Chem. Scand. 1966, 20, 2236.
[15] J. A. Warkentin, O. S. Tee, J. Am. Chem. Soc. 1966, 88, 5540.
[16] E. Keinan, E. K. Hafeli, K. K. Seth, R. Lamed, J. Am. Chem. Soc.
1986, 108, 162.
[20] G. H. Hammons, F. H. Westheimer, K. Nakaokas, R. Kluger, J. Am.
Chem. Soc. 1975, 97, 1568.
[21] G. W. Kosicki, Biochemistry 1968, 7, 4310.
[22] a) W. Tagaki, F. H. Westheimer, Biochemistry 1968, 7, 901;
b) Ref. [20].
[23] a) I. A. Rose, S. V. Rieder, J. Biol. Chem. 1958, 233, 315; b) Ref. [19].
[24] S. A. Benner, J. D. Rozzell, Jr., J. Am. Chem. Soc. 1981, 103, 993.
[25] S. R. Grady, E. E. Dekker, Biochim. Biophys. Acta 1979, 568, 234, and
references therein.
[17] R. Bjˆrnestedt, G. Zhong, R. A. Lerner, C. F. Barbas III, J. Am. Chem.
Soc. 1996, 118, 11720.
Received: May 25, 2001 [F3282]
Chem. Eur. J. 2002, 8, No. 1
¹ WILEY-VCH Verlag GmbH, 69451 Weinheim, Germany, 2002
0947-6539/02/0801-0239 $ 17.50+.50/0
239