Characterization of proton-transfer catalysis by serum albumins

Article abstract of DOI:10.1021/ja993471y

Two independent investigations of catalysis by BSA and other serum albumins are combined to provide detailed insight into the mechanism of a classical proton-transfer reaction taking place on the surface of a protein. The Kemp elimination involves the gen

Full text of DOI:10.1021/ja993471y

1022
J. Am. Chem. Soc. 2000, 122, 1022-1029
Characterization of Proton-Transfer Catalysis by Serum Albumins
†,‡
,†
,‡
§
Florian Hollfelder, Anthony J. Kirby,* Dan S. Tawfik,* Kazuya Kikuchi, and
Donald Hilvert*,
§,|
Contribution from the UniVersity Chemical Laboratory, Lensfield Road, Cambridge CB2 1EW,
United Kingdom, Centre for Protein Engineering, Hills Road, Cambridge CB2 2HQ, United Kingdom,
Departments of Chemistry and Molecular Biology, The Scripps Research Institute,
1
0550 North Torrey Pines Road, La Jolla, California 92037, and Laboratory of Organic Chemistry,
Swiss Federal Institute of Technology (ETH), UniVersit a¨ tstrasse 16, CH-8092 Z u¨ rich, Switzerland
ReceiVed September 27, 1999
Abstract: Two independent investigations of catalysis by BSA and other serum albumins are combined to
provide detailed insight into the mechanism of a classical proton-transfer reaction taking place on the surface
of a protein. The Kemp elimination involves the general base-catalyzed removal of a proton from carbon and
is known to be highly sensitive to medium effects. The serum albumins bind and catalyze the eliminative
fragmentation of 5-nitrobenzisoxazole with remarkable efficiency and “accidental specificity.” A binding site
and a likely catalytic lysine are identified, and factors contributing to the efficiency of catalysis analyzed. A
key factor appears to be a differentiated microenvironment, which allows delocalized negative charge to develop
in a stabilizing hydrophobic pocket next to a polar region where the developing ammonium cation is not
disfavored. Catalytic efficiency is discussed in terms of effective molarities for the reaction catalyzed by serum
albumins and by catalytic antibodies and compared with a selection of published EMs for enzyme-catalyzed
reactions.
5
Proton transfers are involved in almost every enzyme reaction.
calculation of an EM of the order of 10 M for the proton-
1
2
Although normally fast, they can become rate determinings
and thus need efficient catalysis. This is especially true when
the proton is transferred to or from carbon or when the reaction
is concerted with the formation or cleavage of a bond between
heavy (i.e., non-hydrogen) atoms. How enzymes stabilize
proton-transfer transition states is the subject of vigorous
transfer catalyzed by mandelate racemase, shows how enzymes
can take advantage of proximity effects to catalyze efficient
proton transfer. Results with intramolecular model systems
designed to show a large increase in hydrogen bond strength
during the course of proton transfer show that high EMs (up to
6
10 M) can be attained, given rather precise control of
1
-10
13,14
ongoing debate.
Induced intramolecularity is an obvious
geometry.
Control of geometry is of course a basic property
candidate, but effective molarities (EM; the nominal concentra-
tion of a catalytic group in solution needed to match the rate of
of enzyme active sites, and results with GlufAsp mutants show
15
how the minimal change in the position of a general acid or
1
1
16
the intramolecular reaction under identical conditions ) for
general acid/base catalysis in model systems are notoriously
lowstypically below 10 M. For nucleophilic catalysis, by
general base reduces catalytic efficiency.
Of particular interest in this context is an estimated EM of
up to 40 000 M for a reaction catalyzed by man-made antibody
34E4, generated by the bait-and-switch approach, using a
cationic hapten to elicit an active-site carboxylate general base.17
The model for much more demanding enzyme-catalyzed proton
transfers was the Kemp elimination (1 f 2), a base-catalyzed
proton transfer from carbon, in which the removal of the proton
9
13
contrast, values of up to 10 M (up to 10 when ground-state
strain is relieved during the reaction) can be achieved with
relative ease when catalytic and substrate group are brought
together on the same molecule. It has been suggested that this
low efficiency of intramolecular general acid/base catalysis may
reflect the rather loose transition-state binding of the in-flight
protonsor simply our deficient understanding of the optimal
positional requirements for a proton-transfer reaction. The recent
1
8
is concerted with N-O-cleavage.
*
Corresponding authors. Telephone: +44-1223-336370. Fax: +44-
1
223-336362. E-mail: ajk1@cam.ac.uk.
†
University Chemical Laboratory.
Centre for Protein Engineering.
The Scripps Research Institute.
Swiss Federal Institute of Technology.
‡
§
|
(
1) Gerlt, J. A.; Gassmann, P. G. Biochemistry 1993, 32, 11943-11952.
(2) Gerlt, J. A.; Gassmann, P. G. J. Am. Chem. Soc. 1993, 115, 11552-
1
1568.
(
(
(
3) Cleland, W. W.; Kreevoy, M. W. Science 1994, 264, 1887-1890.
4) Cleland, W. W.; Kreevoy, M. M. Science 1995, 269, 104.
5) Frey, P. A.; Whitt, S. A.; Tobin, J. B. Science 1994, 264, 1927-
Catalytic antibodies are the best available enzyme mimics
and have many interesting and relevant properties, but high
1
1
930.
(
6) Frey, P. A. Science 1995, 269, 104-106.
(
7) Guthrie, J. P.; Kluger, R. J. Am. Chem. Soc. 1993, 115, 11569-
(12) Bearne, S. L.; Wolfenden, R. Biochemistry 1997, 36, 1646-1656.
(13) Kirby, A. J.; Williams, N. H. J. Chem. Soc., Perkin Trans. 2 1994,
643-648.
1572.
(
(
(
8) Guthrie, J. P. Chemistry & Biology 1996, 3.
9) Shan, S. O.; Loh, S.; Herschlag, D. Science 1996, 272, 97-101.
10) Shan, S.-O.; Herschlag, D. Proc. Natl. Acad. Sci. U.S.A. 1996, 93,
(14) Kirby, A. J.; O’Carroll, F. J. Chem. Soc., Perkin Trans. 2 1994,
649-655.
1
4474-14479.
(15) Lawson, S.; Wakarchuk, W. W.; Withers, S. G. Biochemistry 1997,
36, 2257-2265.
(11) Kirby, A. J. AdV. Phys. Org. Chem. 1980, 17, 183-278.
1
0.1021/ja993471y CCC: $19.00 © 2000 American Chemical Society
Published on Web 01/27/2000

Proton-Transfer Catalysis by Serum Albumins
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1023
efficiency is not usually one of them. So an EM as high as
Initial velocities were determined at 25 ( 1 °C with 1 (0.15-0.8 mM)
and serum albumin (15 µM) and are corrected for the rate of the
background reaction under the same conditions. Data points are means
from three measurements. A methanolic stock solution of substrate 1
was diluted with water (1:10) and added to a solution of SA in buffer
40 000 M raises important fundamental questions. As noted by
Thorn et al., the analysis in terms of catalytic efficiency is
complicated by the high sensitivity of the reaction to medium
1
9
effects, so that the catalytic package even in this simple one-
step model enzyme reaction is difficult to unravel. The
Cambridge group thought to gain insight into what constitutes
the main source of catalysis in the antibody 34E4sa medium
effect or a “perfectly poised” basesby looking for other protein
scaffolds with an appropriate combination of hydrophobic bind-
ing sites with potential general bases in close proximity. The
family of serum albumins from different species shows the
desired property and catalyses the Kemp elimination with kcat
values of the same order of magnitude. The Scripps/ETH group
observed catalysis by bovine serum albumin (BSA) indepen-
(30 mM containing 150 mM NaCl). The final methanol concentration
did not exceed 2% of the final volume of 200 µL. The rate of the
SA-catalyzed reaction was approximately halved in the presence of
1
50 mM NaCl compared to that in salt-free buffer solution. The use of
buffers bearing hydrophobic groups (e.g., cyclohexyl in 2-(N-cyclo-
hexylamino)ethanesulfonic acid (CHES) or 3-(cyclohexylamino)-1-
propanesulfonic acid (CAPS) was avoided since they inhibited the
reaction (by a factor of about 2 in 30 mM buffer, pH 9.5). Kinetic
parameters were derived by Lineweaver-Burk and Eadie-Hofstee
methods and by fitting directly to the Michaelis-Menten equation.
Calculated errors are based on standard deviations. The low solubility
of 1 in water prevented measurements at substrate concentrations above
1
7
dently, in the course of their work with catalytic antibodies.
M
K ; hence, errors greater than the indicated limits cannot be rigorously
This paper is based on our two independent preliminary re-
ports of catalysis of the Kemp elimination by serum album-
excluded. The rate constant kuncat for the uncatalyzed reaction was
obtained by extrapolation to zero buffer concentration (30-210 mM
bis-Tris buffer, [S] ) 10-37.5 mM]). Repetitive scans showed a clean
isosbestic point at 304 nm, as for the uncatalyzed reaction, ruling out
the possibility that a binding step was being monitored.
2
0,21
ins.
We present detailed information about these systems, a
discussion of the likely catalytically active residues and the glo-
bal environment of the proton-transfer active site, and compar-
ison with structural data. A comparison of serum albumins with
other available catalysts for this reaction allows some insight
into the sources of catalysis and points to a significant con-
tribution of medium effects in bringing about rate acceleration.
The Scripps/ETH group monitored reactions of 1 spectrophotometri-
cally at 380 nm as a function of time. The reactions were carried out
at 20 ( 0.2 °C in the following buffers (40 mM, containing 100 mM
NaCl): sodium acetate (pH < 6), sodium phosphate (6 < pH < 8),
and sodium carbonate (pH > 8). In all cases, addition of substrate
dissolved in 2% acetonitrile was used to initiate reaction. Stopped-
flow techniques were used for measurements at high pH (>8.5). Equal
volumes of substrate and BSA solutions were rapidly mixed, and the
time-dependent increase in absorption at 380 nm was recorded. Initial
velocities, calculated from the first 5-10% of the reaction, were
measured in triplicate and corrected for the background reaction of 1
in the absence of BSA. For saturation kinetics, the concentration of
substrate was varied from approximately 50 µM to 1 mM with BSA
concentrations around 10 µM. The catalyst concentration was increased
up to ∼50 µM for particularly slow reactions to obtain reliable rates
over background. Kinetic constants were derived by direct computer
Experimental Procedures
Materials. Bovine serum albumins were used as supplied by Sigma
(
(
St. Louis, MO) (A-7511, A-3294, A-7030) and Boehringer Mannheim
fraction V). (Preparation A-7030 (“essentially fatty acid free”) was
used for the majority of the experiments reported below, except where
otherwise indicated.) Other proteins assayed for catalytic activity toward
1
were barnase, lysozyme, trypsin, chymotrypsinogen and chymotrypsin,
and calmodulin (from Sigma; conditions: 0.2 mg/mL, 30 mM HEPES,
pH 7.4). Many of these are known to have hydrophobic binding sites,
but none catalyzed the Kemp elimination significantly.
22
Substrate Synthesis. The published preparation of 1 was improved
fit of the data to the Michaelis-Menten equation: v
+ [1]). Apparent second-order rate order constants (kcat/K
also obtained directly and independently from initial velocities obtained
at low substrate concentration (50 µM , K ). The pH-rate data were
0
) kcat[BSA][1]/
as follows. 1,2-Benzisoxazole (Aldrich, Milwaukee, WI, 4.6 g, 38.6
(K
m
m
) were
3
mM) was dissolved at 0 °C in concentrated H
2
SO
4
(19.7 cm ) and a
3
mixture of concentrated nitric acid (2.6 cm ) and concentrated H
1 cm ) was added slowly. The solution was stirred for 30 min and
poured onto an ice/water mixture (1;1, 100 cm ). The crude product
was collected as a precipitate and recrystallized from anhydrous ethanol
to yield the nitrobenzisoxazole 1 (3.1 g, 50%) as colorless needles, mp
2
SO
4
3
m
(
analyzed as described in the text. Measurements by the two groups
were generally in good agreement, but diverged above pH 9.3, where
the Scripps stopped-flow results gave significantly higher initial rates.
At these high rates (> 0.2 O. D./s) strong product inhibition had set in
before the manual measurements commenced. The stopped-flow data
3
2
2
-1
1
26-7 °C;
NO ), δ (400 MHz; CDCl
J 2, H-4), 8.50 (1 H, dd, J 2 and 9, H-6), 7.75 (1 H, d, J 9, H-7); δ
ν
max (CDCl
3
)/cm 3100 (CH), 1620 (arom); 1530, 1350
(
2
H
3
) 8.90 (1 H, d, J 0.8, H-3), 8.72 (1 H, d,
-
1
a
for BSA gave a plateau rate of 360 min and the apparent pK of 10.3
C
which we use to calculate the EM.
(
(
(
[
400 MHz; CDCl
3
) 164.4- (arom C-O), 147.1+ (arom CHdN), 144.8
Inhibition Studies. Noncovalent inhibition. Noncovalent inhibition
was measured for HSA and is expressed as the concentration of inhibitor
for which the activity was halved (IC50). Strong inhibition by product
arom C-NO
2
), 125.6+, 121.9-, 119.3+, 110.5+ (arom); m/z 164
+
+
+
100%, M ), 134 (23%, M - NO), 118 (28%, M - NO
2
) (Found:
requires M , 164.0222). (Found: C, 51.1;
O requires C, 51.2; H, 2.5; N, 17.1).
+
+
7 4 2 3
M] 164.0222 C H N O
2
was observed in all experiments (K
i
≈ 50 µM). Up to 30 turnovers
H, 2.6; N, 17.1. C12
H N
14 2
were observed before the background rate caught up with the catalyzed
rate. Dialysis of product-inhibited HSA against phosphate buffer (pH
Rate Measurements. In Cambridge initial rates were determined
spectrophotometrically by monitoring the release of product 2 at 405
nm in a Thermomax microtiter plate reader (Molecular Devices) and
analyzed using the programs SOFTmax 2.32 and Kaleidagraph 3.0.5.
7
.4) restored activity in full.
Covalent inhibition. SAs (150 µM) were incubated with pyridoxal
phosphate (0-0.9 mM) at pH 8.1 (50 mM bicine) for 2 h in the dark.
Following preincubation the solution was diluted into reaction buffer,
and initial rates were measured as described above. SAs were also
(
16) Straus, D.; Raines, R.; Kawashima, E.; Knowles, J. R.; Gilbert, W.
Proc. Natl. Acad. Sci. U.S.A. 1985, 82, 2272-2276.
17) Thorn, S. N.; Daniels, R. G.; Auditor, M.-T. M.; Hilvert, D. Nature
995, 373, 228-230.
18) Casey, M. L.; Kemp, D. S.; Paul, K. G.; Cox, D. D. J. Org. Chem.
973, 38, 2294-2301.
19) Kemp, D. S.; Cox, D. D.; Paul, K. G. J. Am. Chem. Soc. 1975, 97,
2
3
(
modified with acetylsalicylic acid following the procedure of Walker.
1
1
7
6
8
7
24
For modification with 4-nitrophenyl anthranilate, SAs (4 mg/mL) were
incubated with 4-nitrophenyl anthranilate (0-90 µM) for 24 h (100
mM phosphate, pH 7.4, 50 mM NaCl), and the release of nitrophenoxide
was followed. The sample was then dialyzed against phosphate buffer
(pH 7.4, 50 mM) and assayed as above. There was a good correlation
between the release of nitrophenoxide and the residual activity for the
(
(
312-7318.
(20) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S. Nature 1996, 383, 60-
3.
(
21) Kikuchi, K.; Thorn, S.; Hilvert, D. J. Am. Chem. Soc. 1996, 118,
184-8185.
(23) Walker, J. E. FEBS Lett. 1976, 66, 173-174.
(24) Hagag, N.; Birnbaum, E. R.; Darnall, D. W. Biochemistry 1983,
22, 2, 2420-2427.
(22) Lindemann, H.; Thiele, H. Justus Liebig’s Ann. Chem. 1926, 449,
6.

1024 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Hollfelder et al.
Table 1. Sequence Alignment of Primary Structure for Serum Albumins from Different Speciesa
IIA Site
199
211
214
215
218
219
222
223
234
238
242
257
260
261
264
290
291
292
human
bovine
equine
rat
K
R
K
K
F
R
K
K
R
F
L
V
F
W
W
W
W
R
W
W
W
W
A
S
R
R
R
R
Y
R
R
R
R
L
L
L
M
L
L
L
L
L
R
K
K
R
K
K
R
T
F
F
F
F
Y
F
F
F
F
L
L
I
L
L
L
V
S
L
L
L
L
H
H
H
N
H
H
H
N
H
R
R
R
R
M
R
R
R
R
L
L
A
A
A
A
M
L
I
I
I
I
A
A
A
A
M
A
A
S
E
E
E
E
E
E
E
E
G
S
L
I
I
A
Q
S
L
F
I
L
M
L
I
L
I
chicken
sheep
F
M
M
M
M
L
L
F
I
porcine
murine
rabbit
S
I
L
I
L
I
I
F
A
A
L
A
M
M
L
I
Y
R
Y
IIIA Site
384
387
388
391
392
395
410
411
430
433
438
449
450
453
485
489
human
bovine
equine
rat
P
L
L
L
L
V
L
L
L
L
I
I
N
N
N
N
N
N
N
N
N
C
C
C
C
C
C
C
C
C
F
F
F
Y
L
F
F
Y
Y
R
R
R
R
R
R
R
R
R
Y
Y
Y
Y
Y
Y
Y
Y
Y
L
L
L
L
M
L
L
L
L
V
V
V
V
I
L
L
L
V
C
C
C
C
C
C
C
C
C
A
T
S
E
E
E
E
E
E
E
E
E
L
L
L
L
L
L
L
L
L
R
R
R
R
R
R
R
R
R
S
S
S
S
T
S
S
S
S
H
P
P
T
P
P
P
P
V
V
V
I
V
S
chicken
sheep
T
A
V
V
porcine
murine
rabbit
I
V
V
a
Residues are grouped according to interactions with ligands in two small-molecule binding sites as described by Carter.26,35 The numbering is
56
based on the assignment residue 214 as the only Trp in HSA. The IIA and IIIA sites are identical with Sudlow’s sites I and II and Ozeki’s sites
U and R, respectively. Possible bases (or their substitutes) are printed in bold, other residues constituting the two binding sites are predominantly
hydrophobic.
5
7
Kemp elimination, consistent with one equivalent of p-nitrophenyl
anthranilate reacting. At complete modification of HSA (almost 90%
reaction, extrapolated to 60 µM p-NPOH), 70% of the Kemp elimination
activity remained. Lysine residues in BSA were modified with
fluorescein isothiocyanate (FITC) as previously described.25 Unreacted
FITC was removed by gel filtration (Sephadex G-25). The concentration
of the modified protein was determined by the method of Smith, and
-
1
-1
an extinction coefficient at 491 nm of 15,900 M cm was assumed
25
for fluorescein coupled to BSA. Residual activity was measured with
5
5 µM 1 in carbonate buffer (pH 10).
Sequence Alignment and Comparison. The unique residue Trp-
2
14 of human serum albumin is taken as the anchoring point for
numbering in Table 1. Sequence data were obtained from the SwissProt
database.
pK
a
Measurements and Reactions of 1 in Solvent Mixtures. The
pK ’s of acetic acid and n-butylamine were measured under the reaction
a
conditions as the pH of a 50% free base solution using a Russell
CMAWL 757 electrode connected to a Radiometer PHM 83 pH-meter.
[
T ) 25 °C, [CH COONa] ) 50 mM, [I] ) 25 mM.] Data are displayed
3
in Figure 3. Acetate-catalyzed reactions were monitored by following
the appearance of the product chromophore at 405 nm in a Cary 3
spectrophotometer. (T ) 25 °C; [CH COONa] ) 50-5 mM; [I] ) 25
3
mM; [S] ) 0.1-0.14 mM.) Pseudo-first-order rate constants were
calculated from exponential fits of the increase in absorbance with time
and, for the reaction in 100% water, by the method of initial rates.
Pseudo-first-order rate constants were plotted against acetate concentra-
Figure 1. Second-order rate constants for the acetate-catalyzed reaction
f2 (relative to the rate constants in water) plotted against the solvent
composition in acetonitrile/water systems. Inset: Relative rates plotted
logarithmically against the measured pK values of acetic acid in water/
acetonitrile mixtures (data from Figure 3).
1
a
-
tion to yield second-order rate constants kOAc (Figure 1).
Results and Discussion
Magnitude of Rate Accelerations. All of the SAs tested
accelerate the Kemp elimination. Catalytic activity was similar
1
0 min) resulted in a loss of no more than 40% of activity,
thus ruling out labile proteins as catalytic contaminants.
Reactions were analyzed according to the Michaelis-Menten
model (Figure 2). For BSA derived kcat values (Cambridge, at
2
9
((5%) under the same conditions for different commercial
preparations, with the exception of the globulin-free cold alcohol
precipitated albumin A-7638 (80% of typical activity at pH 7.4
and 65% at pH 8.5). For example, the activities of different
preparations of BSA differed by no more than (20%. The
observed catalysis is thus independent of origin or method of
purification and constitutes a genuine catalytic effect. This
conclusion is further strengthened by the observation that
albumins from other species, with known high sequence
homology, share the catalytic power of BSA (Table 1). While
most proteins can be denatured thermally, BSA is known to be
particularly heat-resistant.26 Attempted heat inactivation (94 °C,
-1
-1
5 °C) ranged from 0.71 min at pH 6.8 to 14.6 min at pH
.0; stopped-flow measurements by the Scripps/ETH group
max
cat
show the rate finally leveling out near pH 11 (k
) 360
-1
min , measurements at 20 °C). KM values fall in the mM range,
mM at pH 8 and 0.7 mM at pH 9, although these data should
2
be taken with a degree of caution since limited substrate solubil-
ity precluded measurements above KM (Figure 2). The Scripps/
ETH data yield KM values between 0.4 and 0.7 mM for BSA
above pH 7; at lower pH values KM appears to increase making
accurate determinations difficult. The corresponding kcat for HSA
falls between 3.2 min at pH 8 and 28.8 min at pH 9.0 (KM
1
-1
-1
max
-1
.55 mM at pH 8 and 3.2 mM at pH 9, kcat 48 min ). The kcat
(
25) Taylor, R. P. J. Am. Chem. Soc. 1976, 98, 2684-2686.
(26) Carter, D. C.; Ho, J. X. AdV. Protein Chem. 1994, 45, 153-203.
values for both proteins correspond to rate accelerations

Proton-Transfer Catalysis by Serum Albumins
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1025
o
Figure 2. Plots of initial rates vs initial substrate concentration [S ]
Figure 3. Measured pK
4) vs percentage of water in acetonitrile/water mixtures. The equations
for the lines drawn are pK (HOOCCH ) ) 7.1-0.026% water and pK
) ) 9.5 + 0.014% water (correlation coefficients are >0.99).
a
values for acetic acid (b) and n-butylamine
for catalysis by BSA at pH 8.97. [Conditions: T ) 25 °C, [BSA] )
(
1
5 µM in 30 mM buffer, 150 mM NaCl].
a
3
a
+
BuNH
3
(
Table 2. Relative Activities toward 1 of Serum Albumins from
Different Species
a more or less hydrophobic active site. As the hydrophobicity
of the mixed solvent medium increases with added acetonitrile,
the pKa falls to about 9.2 (Figure 3, upper dataset).
relative activityb
speciesa
at pH 6.8
at pH 7.4
at pH 8.4
at pH 9.04
human (A-1653)
bovine (A-3249)
dog (A-3184)
22
100
40
51
100
48
51
100
35
69
100
34
The dangers of overinterpreting pH dependencies of steady-
state parameters are well-known, and our evidence for the
catalytic role of a lysine, although substantial, is necessarily
indirect. We therefore tested directly for an active-site lysine
by looking for inhibition by covalent modification. Pyridoxal-
29
horse (A-9888)
chicken (A-3014)
pig (A-2764)
45
11
24
18
222
106
105
169
78
170
95
90
92
95
95
rat (A-6272)
rabbit (A-0639)
mouse (A-3139)
70
49
49
95
34
31
5
-phosphate (PyrP) is known to bind covalently to two sites on
62
44
41
BSA, the aldehyde group forming a Schiff-base (or a gem-
a
Sigma ordering number in brackets. b Measured as initial velocities.
30
diamine) with one (or two) lysine ꢀ-amino groups. Thus cova-
[
Conditions: 25 °C, 30 mM buffer (BTBS, phosphate, bicine, AMPSO),
lent modification with 1 equiv of PyrP completely inhibits the
observed catalysis (by a lysine in its basic form with an apparent
pKa of 8.4) of the decomposition of Meisenheimer complex 3.31
We conclude below that the same lysine and the same binding
site are involved in catalysis of the Kemp elimination.
1
50 mM NaCl, [SA] ) 2 mg /ml, [S
o
] ) 0.5 mM]. The relative activity
of BSA at the respective pHs was set to 100.
3
4
over background of 10 -10 (accurate only to an order of
magnitude).27
Rate ratios for the different SAs vary with pH (Table 2),
indicating that the catalytic base has a slightly different pKa in
each case: evidently the catalytic groups occupy different envi-
ronments. In the pH-independent regions the plateau kcat values
are similar, differing by factors of no more than 5. These remark-
able rate accelerations are by no means a general property of
proteins, even those that are good substrate binders. The great
majority of antibodies screened in previous studies exhibited
no significant activity in the Kemp elimination, 1
7,28
and we
found the same to be true of calmodulin, and such enzymes as
barnase, lysozyme, trypsin, chymotrypsinogen, and chymotryp-
sin.
Under the same conditions the BSA-catalyzed reaction of 1
is inhibited by 2 equiv of PyrP, but only to an extent of 80%.
The pyridoxal treatment applied to HSA gave only 40%
inhibition. This suggests the possibility that both PyrP binding
sites described by Dempsey may be catalytically active to some
extent. On the other hand, 1 equiv of fluorescein isothiocyanate
Identification of Active-Site Residues and Binding Site
Mapping. The pH dependence of kcat and kcat/KM suggests the
involvement of catalytic bases of pKa in the region of 9-10
for both BSA and HSA in the low pH region.20 The stopped-
flow measurements at Scripps/ETH reveal a further rate increase
for the BSA-catalyzed reaction at higher pHs,21 with a corre-
sponding increase in the apparent pKa to 10.3. It is not
unexpected for the microenvironment of the catalytic group to
change as other ionizing groups are deprotonated. (The activity
at higher pH could also be evidence for a second catalytic site.)
A suitable candidate in each case is a lysine (pKa ≈ 11 in small
peptides) in a somewhat hydrophobic environment which desta-
bilizes the ionic form to some extent. We have mimicked the
proposed pKa depression by measuring the pKa of n-butylamine
in acetonitrile/water mixtures, a simple homogeneous model for
30
2
5,26,32
(
FITC), known to react preferentially with Lys-222 of BSA,
2
1
inhibits some 90% of the catalytic activity. This residue is
located in a binding pocket in subdomain IIA of the protein.33
The extent of modification by FITC is directly quantifiable, and
the loss of activity was found to correlate with derivatization
2
1
of a single lysine residue, providing a strong indication that
the decomposition of 1 occurs predominantly at a single site
(29) Knowles, J. R. CRC Crit. ReV. Biochem 1976, 165.
(
30) Dempsey, W. B.; Christensen, H. N. J. Biol. Chem. 1962, 237,
1
113-1120.
(31) Taylor, R. P.; Chau, V.; Bryner, C.; Berga, S. J. Am. Chem. Soc.
1
975, 97, 1934-1944.
(27) Hollfelder, F. Ph.D. Thesis, Cambridge, 1997.
(32) Andersson, L.-O.; Rehnstr o¨ m, A.; Eaker, D. Eur. J. Biochem. 1971,
(
28) Sergeeva, M. V.; Yomtova, V.; Parkinson, A.; Overgaauw, M.;
20, 371-380.
Pomp, R.; Schots, A.; Kirby, A. J.; Hilhorst, R. Isr. J. Chem. 1996, 36,
77-183.
(33) Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharmacol. 1975, 11,
824-832.
1

1026 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Hollfelder et al.
a more significant contribution of up to some 30% of overall
activity, as indicated by the results with diazepam and 4-ni-
troanthranilate.
The amino acid sequences of the two sites implicated in small
3
5
ligand binding by He and Carter for different SAs (Table 1)
show predominantly hydrophobic side chains, together with
potential general bases and some permanently cationic residues
(e.g., Arg 257, 410, and 485). The positively charged residues
could be involved in binding negatively charged portions of
the ligands 4-6 or product 2 (and perhaps the transition state
leading to it); with the hydrophobic residues binding the
aromatic rings of these compounds. This is consistent with the
binding properties of 2 and 4-6 and some pKa-lowering effect
of a hydrophobic environment on an amine base. In HSA the
IIA site, presumably with Lys-199 acting as the general base,
would be responsible for most of the activity. For BSA there is
no lysine in position 199, but a different lysine (222) that is in
turn absent in HSA is close by and available for catalysis.
Potential Sources of Rate Acceleration. We consider first
the mechanisms available for efficient catalysis of the Kemp
elimination.
(a) A catalytic base needs to be available in the free base
form at the pH of the reaction, properly positioned with respect
to the bound substrate in the catalyst active site. Its effectiveness
will depend on its positioning, and will contribute to its observed
effective molarity (EM, see above). (Assistance to leaving group
departure by protonation or by interaction with a cationic group
is also a possibility. However, Kemp observed no general acid
catalysis even for the reaction of unsubstituted benzisoxazole
Figure 4. Inhibition of HSA catalysis by small, anionic ligands.
Activities (relative to the uninhibited initial rate) are plotted against
the ratio [inhibitor]/[HSA]. Inhibitors used: (b) warfarin (4), (0)
-
ibuprofen (5), (O) diazepam (6), (∆) octanoate (CH
3
(CH
2
)
6
COO ), (2)
and triiodobenzoate. [Conditions: 25 °C, [HSA] ) 3.75-60 µM, [S
0.1 mM, in 30 mM bicine, pH 8.7].
o
]
)
aleaving group
leaving group
(
pK
≈ 6.9), so for substrate 1 (pKa
≈ 4.1)
on BSA. In contrast, alkylation of cysteines with iodoacetamide
had no effect on the BSA-catalyzed reaction, and modification
of tyrosines (and possibly lysines) with a 100-fold molar excess
of acetylimidazole reduced the specific activity by only 54%.
With HSA 4-nitrophenyl anthranilate (believed to react with
Tyr-41124 ) reduced the activity by about 30%, suggesting that
either Tyr-411 or a residue nearby in the IIIA site may contribute
to the observed catalysis.
such catalysis is even less likely, at least in a protic environ-
ment.)
(b) The work of Kemp shows that there is a large, accelerating
medium effect of aprotic solvents on the reactions of anions
with 1, a consequence of the absence of hydrogen-bond
solvation. Acetate is effectively destabilized and thus activated,
compared to the hydrogen-bonded form in water. Studies in
aprotic solvents indicate that this effect makes a major contribu-
Several noncovalent binding sites have been identified on
7
tion to the over 10 -fold rate accelerations observed for the
26
HSA for small, pharmacologically relevant molecules. Figure
shows how some of the compounds involved in the binding
decarboxylation of 7 and for catalysis of the elimination reaction
4
19
1
by acetate. Homogeneous aprotic solvents in Kemp’s studies
studies (4-6) can inhibit catalysis more or less completely. The
detailed analysis is complicated by the fact that the guests have
individual preferences for one of the two sites (e.g., warfarin 4
for IIA, ibuprofen 5 for IIIA), with relative affinities differing
by factors less than 100.26 Further conformational change may
be induced by a binding event at one of the two sites or possibly
achieved rate accelerations of at least this magnitude, regardless
of their polarity. In the case of protein catalysts the thermody-
namic cost of any desolvation of an active-site carboxylate has
been paid at the stage of protein synthesis, raising its ground-
state energy and thus increasing its intrinsic reactivity. However,
the relatively modest perturbation of the observed (kinetic) pKa
under both saturating and nonsaturating conditions argues
against substantial desolvation of the carboxylate base in the
2
6
by binding of fatty acids (e.g. octanoic acid) at a third site.
Thus diazepam, which was located in the crystal structure in
the IIIA site, inhibited the reaction by up to 30%. Since only a
single association constant was measured34 this may be the only
inhibitor that is specific for a single site (Kd ) 2 µM), although
it is possible that a second association constant was missed.
This is possible also for warfarin: up to 3 molar equiv gave
17
active site of 34E4.
For catalysis by amines this effect will be in the opposite
direction. The pKa of an ammonium ion is depressed in less
polar media (Figure 3) because the cation is less effectively
solvated, and the same will be true of (this part of) the transition
state in reactions where an amine acts as a general base. The
modest rate increases observed by Kemp for the amine-catalyzed
elimination reaction of 1 must therefore result from a different
effect or combination of effects.
54% of inhibition, but at higher molar excesses further inhibition
could be observed.
Taken together, the inhibition data suggest that catalysis by
BSA occurs predominantly at the IIA binding site where an
amine base, presumably a lysine, serves as the catalytically
active residue. The well-defined IIIA binding pocket, implicated
in various hydrolytic reactions, appears to be less important: if
it contributes to the observed catalysis, it is to a much lesser
extent. On the other hand the IIIA site in HSA appears to make
(c) In addition to medium effects involving hydrogen bonding
we can identify specific medium effects on the free-energy of
activation. These include fine-tuning of the effective dielectric
constant of the active site, through-space electrostatic interac-
tions, and dispersion interactions, in particular those involving
(34) M u¨ ller, W. E.; Wollert, U. Mol. Pharmacol. 1973, 11, 52-60.
(35) He, X. M.; Carter, D. C. Nature 1992, 358, 209-215.

Proton-Transfer Catalysis by Serum Albumins
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1027
the delocalized π-system of the benzisoxazole ring in the transi-
difference in basicity between acetate and the active-site carbox-
ylate, is 5000 M. An EM of this magnitude for a proton-transfer
process is even more remarkable, and raises in more acute form
the question of how far the figure results from precise position-
ing of the active-site carboxylate, and how far it reflects the
special microenvironment of the active site. It is not possible
to dissect the observed catalysis in these systems neatly in terms
of specific and nonspecific medium effects vs positioning of
the general base. However, the relatively small shifts of the pKa’s
from normal solution values suggest that hydrogen-bonding
desolvation of the general base in the active site is not the major
factor and is worth no more than perhaps an order of magnitude
in rate for 34E4. For the serum albumins this factor must be is
smaller still, since it is in the opposite direction, and unlikely
to be worth more than 5. The most convincing candidate for
the major factor, certainly in the case of BSA and HSA and
probably in the case of 34E4, is specific solvation of the
transition state (source (c) above). This still leaves scopesbut
does not formally requiresa contribution to the EM of the order
of 10 M from positioning, which falls in the range we might
expect for general base catalysis in such a system.
tion state. Large, polarizable anions are better solvated in apro-
1
9,36
tic solvents,
although data which allow the effect to be is-
olated and quantified are sparse.37 It is possible to make limit-
ing, order-of-magnitude estimates of the importance of these
effects.
(i) By comparing the effects of different aprotic solvents tested
by Kemp. In the absence of H-bonding solvation rates vary by
up to 100-fold, consistent with a specific medium effect of at
least this magnitude.
(ii) By measuring transfer parameters for the species of int-
erest. In a rare example, the enthalpy of transfer of the Mei-
senheimer adduct of 2,4,6-trinitroanisole and methoxide from
methanol to DMSO has been measured as at least 6 kcal
1
9
-
1 38
mol . This would correspond in isolation to a rate effect of
4
greater than 10 . Evidently large effects on the solvation of
delocalized nitroaromatic anions are possible.
Calculation of Effective Molarities. The EM is the ratio of
the first-order rate constant for an intramolecular reaction and
the second-order rate constant for the corresponding bimolecular
reaction, measured under the same conditions.11 The EM is
interpreted in model systems in terms of the (fairly standard)
entropic benefit of proximity plus any enthalpic advantage
derived from the specific positioning of substrate and catalytic
groups with respect to each other. For an enzyme an EM can
be obtained by dividing kcat by the second order rate constant
for the intermolecular reaction. This ratio includes not only the
effect of induced intramolecularity, but any effect (medium
effects, desolvation, etc.) involved in stabilizing the transition
state in question and destabilizing the corresponding ground-
Implications and Conclusions
The rationale behind using SAs was simple: to “partition”
the substrate from solution into a hydrophobic binding site of
a protein known to bind a number of small, hydrophobic mol-
26
ecules. To achieve catalysis in such an environment a catalytic
base must be part of the binding site. Given the relative sim-
plicity of this working hypothesis for the search of a suitable
protein, it is remarkable that the search was successful: especi-
ally in view of the statistically low success rate for designed
catalytic antibody catalysts. This success encouraged us to try
to explore the factors involved in catalysis in this simple system,
perhaps to shed light on more efficientsand more complex
catalytic systems. One trivial reason for our success in observing
catalysis is undoubtedly that the single transition-state reaction
enz
state Michaelis complex. Thus, an EM is an oVerall measure
of efficiency, providing only an upper limit for the effect due
to positioning.
For any EM it is important to compare second- and first-or-
der rate constants as far as possible under identical conditions.
For BSA and HSA the Cambridge group measured second-order
rate constants in both water and acetonitrile for catalysis of the
Kemp elimination of 5-nitrobenzisoxazole 1 by 2-methoxyethyl-
4
1
1 f 2 has modest energetic requirements for catalysis, com-
42
pared with more demanding proton transfers in nature. Another
is likely to be the convenient geometry of the process.43 Our
analysis suggests a linear transition state with at one end a delo-
calized anion developing in a hydrophobic, stabilizing environ-
ment. This is generated by the removal of a proton by an
adjacent catalytic general base at the other end which remains
largely solvated and thus can take advantage of hydrogen-
bonding solvation as the ammonium cation develops.
-
1
-1
amine. The rate constant in water, 1.02 M min , corrected,
using Kemp’s Bronsted â of 0.7239 to the pKa of the catalytic
base gives an EM of 150 M for the lysine NH2 group of BSA
-
1
when compared with the plateau rate of 360 min (apparent
pKa ) 10.3) reached near pH 11. Using the second-order rate
constant for catalysis by n-butylamine21 gives a corrected EM
of 360 M.
Efficiency of Antibody Catalysis. Antibodies can catalyze
a wide variety of reactions and are indisputably the best enzyme
mimics currently available. They can match natural enzymes
These are relatively high EMs for proton-transfer reactions
and clearly refer in the case of BSA to the same catalytic lysine,
presumably in somewhat different microenvironments. (The
active site of HSA appears to be closely similar.) In intramo-
lecular reactions in water, general acid/base catalysis is almost
always very inefficient and correspondingly insensitive to
geometry. On the other hand, the small number of highly
efficient intramolecular systems now known are typically rigid
4
4
in selectivity and stereospecificity, but rate accelerations kcat/
6
kuncat greater than 10 are rare exceptions (Table 3). Thus,
calculations of EMs for reactions catalyzed by catalytic antibod-
ies (Table 3) show that these are generally low, with very few
4
5
45,46
exceptions. The highest EMs (as high as 10 -10 M
)
typically come at the price of severe product inhibition.46
A
40
large hapten gives the antibody extensive recognition possibili-
ties which allow it to bind the substrate well, but since most
recognition interactions do not change during the course of the
reaction, turnover is prevented.
and highly sensitive to geometry. It seems unlikely that such
geometrical precision would be attained in several different
protein binding sites with proximate lysine side chains.
enz
EM for the same reaction catalyzed by the carboxylate
17
group in the active site of antibody 34E4, corrected for the
(41) Menger, F. M.; Ladika, M. J. Am. Chem. Soc. 1987, 109, 3145-
146.
3
(
(
(
36) Coetzee, J. F. Prog. Phys. Org. Chem. 1967, 4, 45.
37) Buncel, E.; Wilson, H. Acc. Chem. Res. 1979, 12, 42-48.
38) Larsen, J. W.; Amin, K.; Fendler, J. H. J. Am. Chem. Soc. 1971,
(42) Bearne, S. L.; Wolfenden, R. J. Am. Chem. Soc. 1995, 117, 9588-
9589.
(43) Na, J.; Houk, K. N.; Hilvert, D. J. Am. Chem. Soc. 1996, 118, 6462-
6471.
(44) Kirby, A. J. Acta Chem. Scand. 1996, 50, 203-210.
(45) Wirsching, P.; Ashley, J. A.; Benkovic, S. J.; Janda, K. D.; Lerner,
R. A. Science 1991, 252, 680-685.
9
3, 3, 2910-2913.
39) Kemp, D. S.; Casey, M. L. J. Am. Chem. Soc. 1973, 95, 6670-
680.
40) Kirby, A. J. Acc. Chem. Res. 1997, 30, 290-296.
(
6
(

1028 J. Am. Chem. Soc., Vol. 122, No. 6, 2000
Hollfelder et al.
Table 3: EMenz Values for Antibody-Catalyzed Reactions
Such figures do not look very different from those for
catalytic antibodies (Table 3). However, calculation of the EM
in the same way as for the antibodies (i.e., by comparison with
the corresponding intermolecular reaction in solution, EM )
kcat/k2) reveals the true extent of catalysis by enzymes. For NDP
kinase the comparison with the second-order reaction of
no. of
system
4E4
reaction type
substrates EMenz (M)a
ref
17
4
3
3
4
proton transfer
1
1
1
4.1 × 10
b,c
(
5000)
4
5F10
proton transfer
proton transfer
2.2 × 10
17
58
b,c
(
6100)
3D4-3D3
40.5
15
52
imidazole with ATP gives a figure of the order of 10 M,
and a figure of the same order of magnitude is likely for
(3.3)
d
MOPC 315
PCP21H3
PCP21H3
ester hydrolysis
transesterification
transesterification
transesterification
transesterification
acyl azide aminolysis
ester aminolysis
ester aminolysis
ester aminolysis
ester aminolysis
imine formation
Diels-Alder
1
2
2
2
2
2
2
2
2
2
2
2
2
2
0.05
59, 60
45
5
4 × 10
12
5
4
3
imidazole catalysis of the racemisation of mandelate anion.
4.4 × 10
5.9 × 10
2.5 × 10
0.158
45
1
1
9
1
2
1
1
1
7
2
1
8R.136.1
8R.136.1
B5.1
46
“Overall” EMs for enzyme reactions include contributions from
a range of catalytic effects and can be substantially higher than
those for the positioning of any particular group. Although the
number of specific examples is limited, we can conclude that
high EMs are commonplace in enzymes but a rare exception in
reactions catalyzed by catalytic antibodies.
46
61
e
7G8
10.3
62
4B11
63G
7.86
63
3
10
64
63G
194
64
7C5-11C2
D4
0.21
4.75
18.3
1000
65
66
2C8
E9
Diels-Alder
66
The failure of antibodies to attain the levels of efficiency of
natural enzymes also stems from a combination of factors: (a)
transition-state analogues are at best poor mimics of true
transition states; (b) many reactions involve more than one
transition state that requires stabilization; (c) the majority of
catalytic antibody reactions do not actually involve a functional
group acting as a catalyst, but simply stabilize the transition
state for the reaction; (d) even if functional group(s) are involved
in catalysis the probability of generating arrays capable of
interacting synergistically, even given “good TSAs”, is infini-
tesimally small. Even bait and switch haptens (exemplified by
the entry for 43D4-3D3), although designed to tap the catalytic
potential of conveniently located functional groups, appear to
give EMs no higher than do TSA haptens. More generally, (e)
only a tiny fraction of the immune response is typically sampled;
and (f) hapten binding rather than catalysis drives selection in
most cases (removing the possibility of optimizing catalysis by
Darwinian evolution in the immune system). It is also conceiv-
able that the immunoglobulin fold places intrinsic limitations
Diels-Alder
67, 68
a
EMenz are derived from kcat/k
2
, and thus include both the effects of
b
positioning and other catalytic effects. Corrected for the difference
corr
_pKa
. c Using k in water. d Modified
in pK
antibody. The k
the fraction of deprotonated amine (pK
a
values of using k
2
) k
measured at pH 8 for butylamine was corrected for
11) at this pH.
2
/10
2
e
2
a
How do these data compare with EMs for natural enzymes?
Page and Jencks estimated an efficiency gain simply from
8
47,48
intramolecularity of the order of 10 -fold,
and cyclization
reactions of model systems in which the reacting groups are
integrated into the same molecule show rate accelerations of
up to this magnitude.11 Intramolecular general acid-base
catalysis is typically much less efficient, but EMs as high as
5
40
1
0 M can be attained with tight control of geometry, and
comparable values have been derived for enzymes. Thus
Wolfenden estimated EMs for general species catalysis by
mandelate racemase by comparing (extrapolated) uncatalyzed
rates with those for the native and mutant enzymes.12 For general
5
4
acid catalysis by Glu-137 and general base catalysis by Lys-
on what can be accomplished catalytically. As a result,
reactions of unactivated substrates that require a complex
assembly of catalytic groups acting at different loci along the
time axis are not promising targets at this time.
5
1
66 EMs were 3 × 10 and for 622 M, respectively. Similarly
the mutation GlufAsp, a change primarily in positioning,
reduces efficiency by 2-3 orders of magnitude but does not
eliminate catalysis entirely in such enzymes as triose phosphate
isomerase and certain glycosidases, and catalytic groups
built-in to the substrate can compete with or replace
More practical objectives are reactions with modest thermo-
dynamic requirements, perhaps accelerated by medium effects,
although recent experiments that exploit reactive mechanism-
based inhibitors such as haptens suggest that some more
complex reaction manifolds can be successfully addressed. The
generation of highly versatile aldolase antibodies that exploit
the enamine-mechanism of natural aldolase enzymes is notable
1
6
15
49
50,51
active-
site general acids and bases in suitable mutants, substituting
simple intramolecular for specific active-site positioning rela-
tively effectively when the active site is otherwise intact. Thus,
for general acid-base catalysis EMs due to positioning in
enzyme active sites appear to fall within the extended range
(53) Wang, Q.; Graham, R. W.; Trimbur, D.; Warren, R. A. J.; Withers,
4
0
established by recent measurements with model systems.
S. G. J. Am. Chem. Soc., 1994, 116, 11594-5.
Interestingly, EMs for nucleophilic catalysis as low as 100
M have recently been obtained, by comparing wild type kcat
with reactions of mutants where a deleted active-site catalytic
group is replaced by a corresponding exogenous nucleophile,
but the active site is otherwise intact. Examples include the use
of imidazole to replace His-122 of NDP kinase,52 and formate
for Glu-358 of an Agrobacterium â-glucosidase (acting on an
activated substrate)53
(54) Padlan, E. A. Mol. Immunol. 1994, 31, 169-217.
(55) Barbas, C. F., III; Heine, A.; Zhong, G. F.; Hoffmann, T.;
Gramatikova, S.; Bj o¨ rnestedt, R.; Anderson, J.; Stura, E. A.; Wilson, I. A.;
Lerner, R. A. Science 1997, 278, 2085-2092.
(56) Sudlow, G.; Birkett, D. J.; Wade, D. N. Mol. Pharmacol. 1976, 12,
1
052-1061.
(57) Ozeki, Y.; Kurono, T.; Yotsuyanagi, T.; Ikeda, K. Chem. Pharm.
Bull. 1980, 28, 535.
(58) Shokat, K. M.; Leumann, C. J.; Sugasawara, R.; Schultz, P. G.
Nature 1989, 338, 269-271.
(59) Pollack, S. J.; Nakayama, G.; Schultz, P. G. Science 1988, 242,
(
46) Jacobsen, J. R.; Prudent, J. R.; Kochersperger, L.; Yonkovich, S.;
Schultz, P. G. Science 1992, 256, 365-367.
47) Page, M. I.; Jencks, W. P. Proc. Natl. Acad. Sci. U.S.A. 1971, 68,
678-1683.
48) Jencks, W. P. AdV. Enzymol. Relat. Areas Mol. Biol. 1975, 43, 219-
10.
49) Tobias, P.; Heidema, J. H.; Lo, K. W.; Kaiser, E. T.; Kezdy, F. J.
1038-1040.
(60) Pollack, S. J.; Schultz, P. G. J. Am. Chem. Soc. 1989, 111, 1929-
(
1931.
1
4
(61) Jacobsen, J. R.; Schultz, P. G. Proc. Natl. Acad. Sci. U.S.A. 1994,
91, 5888-5892.
(
(62) Janda, K. D.; Lerner, R. A.; Tramontano, A. J. Am. Chem. Soc.
1988, 110, 4835-4837.
(
J. Am. Chem. Soc. 1969, 91, 202-3.
(63) Benkovic, S. J.; Napper, A. D.; Lerner, R. A. Proc. Natl. Acad.
Sci. U.S.A. 1988, 85, 5355-5358.
(
(
(
50) Carter, P.; Wells, J. A. Science 1987, 237, 394-399.
51) Wang, Q.; Withers, S. G. J. Am. Chem. Soc. 1995, 117, 10137-8.
52) Admiraal, S.; Schneider, B.; Meyer, P.; Janin, J.; Veron, M.;
(64) Hirschmann, R.; Smith, A. B.; Taylor, C. M.; Benkovic, P. A.;
Taylor, S. D.; Yager, K. M.; Sprengeler, P. A.; Benkovic, S. J. Science
1994, 265, 234-237.
DevilleBonne, D.; Herschlag, D. Biochemistry 1999, 38, 4701-4711.

Proton-Transfer Catalysis by Serum Albumins
J. Am. Chem. Soc., Vol. 122, No. 6, 2000 1029
in this regard.55 The impressive and reliable chiral selectivity
routinely available in their reactions makes further development
of these tailor-made catalysts interesting for synthetic as well
as potential medical applications.
Acknowledgment. We thank Professor Dan Kemp for
helpful comments and Dr. Marina Resmini for help with curve
fitting. D.H. acknowledges support from NIH (Grant No.
GM38273). D.S.T. is grateful to FEBS for a Fellowship and to
Alan Fersht for support. F.H. was supported by a Marie Curie
Fellowship of the European Commission and is a Walter Grant
Scott Research Fellow in Bioorganic Chemistry at Trinity Hall,
Cambridge.
(
65) Cochran, A. G.; Pham, T.; Sugasawara, R.; Schultz, P. G. J. Am.
Chem. Soc. 1991, 113, 6670-6672.
66) Gouverneur, V. E.; Houk, K. N.; Depascualteresa, B.; Beno, B.;
Janda, K. D.; Lerner, R. A. Science 1993, 262, 204-208.
67) Hilvert, D.; Hill, K. W.; Nared, K. D.; Auditor, M.-T. M. J. Am.
Chem. Soc. 1989, 111, 9261-9262.
68) Xu, J.; Deng, Q.; Chen, J.; Houk, K. N.; Bartek, J.; Hilvert, D.;
Wilson, I. A., Science, 1999, 286, 2345-2348.
(
(
(
JA993471Y