Mechanistic study of proton transfer and hysteresis in catalytic antibody 16E7 by site-directed mutagenesis and homology modeling

Article abstract of DOI:10.1016/j.bmc.2004.11.041

Antibody 16E7 catalyzes the carbon protonation of enol ether 2 to hemiacetal 3, and the carbon deprotonation of benzisoxazole 7 to phenol 8. This antibody shows an extreme case of hysteresis, requiring several hours to reach full activity. Antibody 16E7 w

Full text of DOI:10.1016/j.bmc.2004.11.041

Bioorganic & Medicinal Chemistry 13 (2005) 1021–1029
Mechanistic study of proton transfer and hysteresis
in catalytic antibody 16E7 by site-directed
mutagenesis and homology modeling
Lei Zheng,a Roman Manetsch,b, Wolf-Dietrich Woggon,b
Ulrich Baumanna and Jean-Louis Reymonda,*
aDepartment of Chemistry and Biochemistry, University of Berne, Freiestrasse 3, CH-3012 Berne, Switzerland
bDepartment of Chemistry, University of Basel, St. Johanns-Ring 19, CH-4056 Basel, Switzerland
Received 11 October 2004; revised 19 November 2004; accepted 22 November 2004
Available online 19 December 2004
Abstract—Antibody 16E7 catalyzes the carbon protonation of enol ether 2 to hemiacetal 3, and the carbon deprotonation of benz-
isoxazole 7 to phenol 8. This antibody shows an extreme case of hysteresis, requiring several hours to reach full activity. Antibody
16E7 was expressed as recombinant chimeric Fab in Escherichia coli. A model for the three-dimensional structure was produced by
homology modeling and used for a docking procedure to obtain models for antibody–ligand complexes. Site-direct mutagenesis of
GluL39, identified as a possible catalytic residue by the model, to either glutamine or alanine abolished catalysis, showing that both
the protonation reaction of enol ether 2 and the deprotonation of benzisoxazole 7 are promoted by the same residue. The model
furthermore suggested that substrate access to the catalytic site might be hindered by a flexible HCDR3 loop held in closed position
by a hydrogen bond between SerH99 and GluL39, which could explain the observed hysteresis effect. In agreement with this model,
mutagenesis of SerH99 to alanine, or deletion of this residue, was found to reduce hysteresis by approximately 50%.
ꢀ 2004 Elsevier Ltd. All rights reserved.
1. Introduction
hemiacetal 3, and the base-promoted deprotonation of
benzisoxazole 7 to form cyanophenol 8 (Fig. 1). Both
types of reactions have been previously observed indi-
vidually in different catalytic antibodies.5–7 However,
no catalytic antibody has been reported previously to
catalyze both reaction types within the same binding
pocket, which renders the molecular reaction mecha-
nism of antibody 16E7 particularly interesting. In addi-
tion, catalytic antibody 16E7 exhibits an unusual case of
hysteresis for both its protonation and its deprotonation
reactions, whereby catalysis requires several hours to
reach full activity in the presence of substrate. This
retardation effect is reversible, suggesting an underlying
protein dynamics effect.
Catalytic antibodies can be obtained by immunization
against small molecule haptens that are either transition
state analogs or reactive probes corresponding to a
given reaction.1 Most catalytic antibodies accelerate
base-promoted hydrolytic processes or pericyclic reac-
tions.2 Catalytic antibodies for carbon protonation
and deprotonation reactions are of particular interest
due to the fundamental nature of this step, which is
the key to catalysis in many natural enzymes.3 We re-
cently reported the preparation of catalytic antibody
16E7, obtained by immunization against the guanidi-
nium hapten 1.4 This antibody catalyzes proton transfer
to and from carbon in two separate reactions, which are
the acid-promoted protonation of enol ethers 2 to form
Herein we report a site-directed mutagenesis study of
catalytic antibody 16E7 guided by homology modeling.
A three-dimensional model of catalytic antibody 16E7
was produced and used in a docking procedure to gen-
erate a model of the antibody complexes with hapten 1
and substrates 2 and 7. Residue GluL39 was identified
as the key catalytic residue responsible for both the pro-
tonation and the deprotonation activity of antibody
Keywords: Catalytic antibody; Proton transfer; Homology modeling;
Site-directed mutagenesis.
*
Corresponding author. Tel.: +41316314325; fax: +4131631
8057; e-mail: jean-louis.reymond@ioc.unibe.ch
Present address: The Scripps Research Institute, 10550 North Torrey
Pines Road, BCC-315, La Jolla, CA 92037, USA.
0968-0896/$ - see front matter ꢀ 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.bmc.2004.11.041

1022
L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
CAGACTCCA (the SacI restriction site is underlined);
VL anti-sense, GCAGCATCAGCCCGTTTTATTTC-
AAGCTTGG (HindIII); VH sense, AGGTCCAGC-
MeO
CF3COO
O
NH
OH
N
H
N
H
TGCTCGAGTCTGG
GAGACGGTGACCAGAGTCCCTTGG
(XhoI);
VH
anti-sense,
(BstEII).
1
The amplified fragments were cloned into appropriate
sites in the vector p4xH-M139 to give a plasmid in which
the VL and VH segments of the antibody 16E7 are fused
to human Cj and gamma1CH1regions. The chimeric
Fab 16E7 gene fragment was subcloned into
pBAD-43C9 vector10 by PCR with VL sense and CH1
anti-sense primer, CACCGCCGGTCGACTCAGTGG-
TGGTGGTGGTGGTGTGTGTGAGTTTTGTCAC
(SalI), 6XHistidine tag was added at the C-terminus of
chimeric Fd. The resulting expression plasmid pBAD-
16E7his allows one to express a 6X Histidine containing
chimeric Fab 16E7.
H
Ab 16E7
MeO
MeO
HO
OMe
OH
O
OR
3 (R = H)
4 (R = Me)
2
b
a
H
H
a
O
MeO
OMe
OH
b
O
Ab 16E7
2.3. Production, purification, and characterization of
chimeric Fab fragments
OH
Me
OH
5
6
The Escherichia coli strain TOP10 (Invitrogen) was
transformed with pBAD-16E7his and its mutagenized
derivatives for expression of the wild type and mutant
antibodies. 1/40 volume of preculture was inoculated
into LB medium containing 100 lg/mL ampicillin at
20 ꢁC and overexpression was performed after induction
with 0.1% L(+)-arabinose at OD600 of 1.5. 1/10 culture
volume of Terrific Broth was added to implement
nutrition for overexpression. After 14 h induction, the
crude periplasmic lysate was prepared by sonication.
The chimeric Fab 16E7 fragment was purified by Ni2+
NTA affinity chromatography (Qiagen). Concentrations
Ab-COO
O2N
H
Ab 16E7
O2N
CN
N
O
O
7
8
Figure 1. Carbon protonation and deprotonation reaction catalyzed
by antibody 16E7 (anti-1).
of purified antibodies were determined by measuring
11
16E7. Furthermore, the modeling study provided a
mechanistic hypothesis linking the hysteresis effect to a
slow breathing motion of the HCDR3 loop, which was
tested by mutagenesis.
A280 and A260
.
The purity of the sample was con-
firmed by SDS-PAGE. The same procedure was applied
for the production and purification of mutants (see
below). WT and mutant chimeric 16E7-Fab fragments
were obtained in similar purified yields of 0.6–1.0 mg/L
culture.
2. Materials and methods
2.4. Site-directed mutagenesis
2.1. Cloning of antibody 16E7 Fab gene fragment
Site-directed mutagenesis was accomplished by a Quik-
ChangeTM site-directed mutagenesis protocol with minor
modification and including silent restriction site. All
PCRs were performed with the High Fidelity System
(Roche). Mutant codons were generally chosen to ex-
ploit the most frequently used codon for a particular
amino acid among highly expressed genes in E. coli. Oli-
gonucleotides were custom-synthesized and purified by
MicroSynth (Balgach, Switzerland) as follows:
Total RNA was isolated from the hybridoma cell line
16E7 with RNeasy kit (Qiagen) and its cDNA library
was constructed with the Omniscript kit (Qiagen). The
Fd and kappa chain gene fragments of 16E7 were ampli-
fied by PCR with the degenerated primer set as de-
scribed.8 Amplified PCR products were digested with
XhoI-SpeI for the Fd gene fragment and SacI-XbaI
for the kappa chain gene fragment, respectively, and
ligated into phage display vector pcomb3H8 to give
pcomb3H-16E7. The resulting clone was checked for
its affinity to the hapten-BSA conjugate with phage dis-
play panning methods.8 All PCRs were carried out with
Vent proofreading polymerase (New England Biolab).
D
H50A_for GGAGcTATTTACCCTGGAtccGGGAA-
TACTTACTAC (BamHI)
H50A_rev CCCggaTCCAGGGTAAATAgCTCCAA-
TCCACTCAAGTCC
L39Q_for CCTATTTAcaATGGTAtCTGCAGAAA-
CCAGGCC (KpnI deletion)
L39Q_rev GCAGaTACCATtgTAAATAGGTGTTTC-
CATTAC
L39A_for CCTATTTAGcATGGTAtCTGCAGAAA-
CCAGGCC (KpnI deletion)
D
E
2.2. Construction of chimeric Fab fragment
E
The VL and VH genes of Ab 16E7 were amplified using
pcomb3H-16E7 as the template with the following
primers: VL sense, CCAGATGTGAGCTCGTGACC-
E

L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
1023
EL39A_rev GCAGaTACCATgCTAAATAGGTGTT-
TCCATTAC
tion at 37 ꢁC. Substrate and product concentrations
were measured by RP-C18-HPLC after 12, 20, and
28 h incubation time, allowing to determine the initial
velocities vi and the final velocities vss for the different
substrate concentrations used in the measurements.
After correction of these rates for the uncatalyzed reac-
tion rates in BisTris buffer, the net rates vi and vss were
obtained. These rates were used to derive the Michaelis-
Menten constants KM and the maximum velocity Vmax
from the Lineweaver–Burk plot of 1/V versus 1/[S].
The catalytic constant kcat was obtained by dividing
Vmax by the antibody concentration. For the Kemp
elimination the obtained velocities vi and vss from
Eq. (2) were used to determine the value KM/Vmax from
the Lineweaver–Burk plot of 1/V versus 1/[S].
S
H99A_for GGTACTACGGTgcTGGCGCTGTCTCC-
TGGGGC
H99A_rev CAGCGCCAgcACCGTAGTACCCgCgg-
GCACAGAAATAGAC (SacII)
H99del_for
GGTACTACGGTGGCGCTGTCTCCTGGGGCG
H99del_rev GAGACAGCGCCACCGTAGTACCCg-
S
S
S
CggGCACAGAAATAG (SacII)
For all primers, mutagenized positions are denoted in
lowcase. The restriction sites are underlined.
2.5. Kinetic measurements
A solution of antibody in BisTris buffer, pH 6.24 at
37 ꢁC was mixed with a substrate solution (acetoni-
trile/buffer = 1:1) providing a final solution containing
29 mM BisTris, 104 mM NaCl, 10% acetonitrile. The
final antibody concentration was 250–540 lg/mL for
the hydrolysis of the enol ether and 24–90 lg/mL for
the Kemp elimination. Product formation was followed
3. Results
3.1. Hysteresis character of chimeric Fab 16E7
Antibody 16E7 was expressed as chimeric Fab in E. coli
after joining the variable heavy and light chains regions
with the human kappa chain constant region and heavy
chain IgG1CH1region, respectively. This chimeric anti-
body catalyzes both the protonation of enol ether 2 and
the deprotonation of benzisoxazole 7 with activity
parameters comparable to those of the hybridoma de-
rived IgG antibody, including the hysteresis phenome-
non characterized by the retardation time s, the initial
velocity vi and the steady-state velocity vss (Table 1).
In the case of enol ether 2 the transition between the ini-
tial phase of catalysis (vi) and the steady-state phase of
catalysis (mss) is accompanied by an increase in substrate
binding (threefold lower KM), while the catalytic rate
constant kcat is unaffected (Table 2). Only the specificity
constant kcat/KM was determined for benzisoxazole 7
since no substrate saturation occurred within the con-
centration range accessible with this substrate, and was
found to increase by two-fold between the initial phase
and the steady-state phase.
by RP18-HPLC on
a Bischoff LiChrospher 100
(4.6 · 125 mm) column using isocratic elution at 1 mL/
min with a premixed acetonitrile/water solution with
the desired proportion. The 16E7-catalyzed elimination
was followed by measuring A380 using a Molecular De-
vices microtiter plate reader. Measurements were taken
at intervals of 60, 90, or 120 s. With all substrates, the
antibody-catalyzed reactions were carried out in parallel
with an equivalent reaction containing excess hapten 1
(20 lM) for inhibition. All reactions of WT and mutant
Fab-16E7 in the presence of hapten 1 gave background
level reaction rates, indicating that the observed activi-
ties originated from the antigen combining site of the
antibodies.
2.6. Data treatment
For the hydrolysis of the enol ether, the values for the
initial velocity vi and for the final velocity vss were graph-
ically estimated from the slopes of the progress curve.
From following equation:
3.2. Homology modeling and docking
The 3-D structure of the antibody 16E7 Fv domain was
constructed using the online antibody structure model-
ing service WAM (Web Antibody Modeling, http://
antibody.bath.ac.uk/).13 The whole molecule energy
minimization procedure resulted in 200 antibody struc-
tures for antibody 16E7, which were ordered with
ascending minimization energy. The obtained antibody
model was then used in a docking procedure for the dif-
ferent ligands using the docking program AUTODOCK
3.05.14 All docking attempts with the tight binding hap-
ten 1 or the substrates 2 and 7 failed starting with the
lowest energy model 16E7-E1. However, the docking
procedure was successful when starting from the second
lowest energy model 16E7-E2.
ðvss À viÞ
T ¼
Á s,
ð1Þ
vss
the value T of the intercept on the time axis was used to
determine the apparent rate constant s for the transition
between the two velocities vi and vss.4,12 For the Kemp
elimination the data were analyzed with Kaleidagraph
(Abelbeck Software). The three parameters vi, vss, and
s were obtained from the following equation:4,12
½Pꢀ ¼ vss Á t À ðvss À viÞ Á ð1 À eÀt=sÞ Á s:
ð2Þ
2.7. Michaelis-Menten kinetics
The hydrolysis of the enol ethers was initiated by adding
the substrate to the antibody in a capped microtiter
plate. The microtiter plate was kept in a closed plastic
box with a wet paper towel inside for humidity satura-
In the resulting model of the antibody–hapten complex,
ligand 1 appears to be buried in a very deep pocket
located at the center of Fv domain. The hapten is in
direct contact with two carboxylate residues, GluL39

1024
L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
Table 1. The kinetic parameters of antibody 16E7-catalyzed hydrolysis and deprotonation reaction
Antibody
Enol ether 2a,b
vss (lM hÀ1
Benzisoxazole 7a,c
Benzisoxazole 7a,d
vi (lM hÀ1
)
)
s (h)
vi (lM hÀ1
)
vss (lM hÀ1
)
s (h)
vi (lM hÀ1
)
vss (lM hÀ1
)
s (h)
MAb 16E7
Fab 16E7
3.42
2.79
n.a.
n.a.
n.a.
1.12
0.35
5.73
9.45
n.a.
n.a.
n.a.
9.86
5.10
12.15
12.62
n.a.
n.a.
n.a.
7.9
n.a.
n.a.
n.a.
0.25
1.4
n.a.
26.3
n.a.
n.a.
n.a.
6.11
4.79
n.a.
29.4
n.a.
n.a.
n.a.
8.15
5.29
n.a.
3.45
n.a.
n.a.
n.a.
1.52
2.51
65.8
1.95
3.12
13.39
n.a.
154.6
11.9
8.50
34.95
n.a.
GluL39Ala
GluL39Gln
AspH50Ala
SerH99Ala
0.81
0.47
n.a.
n.a.
SerH99 deletion
7.8
n.a.
n.a.
a All measurements were carried out with a solution containing antibody in the buffer of 29 mM BisTris, 104 mM NaCl, 10% acetonitrile, pH 6.24 at
37 ꢁC.
b The enol ether hydrolysis was performed with an antibody concentration of 0.54 mg/mL and with a substrate concentration of 300 lM.
c For the investigation of the catalytic residues all measurements were carried out with an antibody concentration of 0.27 mg/mL and a substrate
concentration of 156 lM.
d For the investigation of the hysteresis were all measurements carried out with an antibody concentration of 0.09 mg/mL and a substrate con-
centration of 633 lM.
Table 2. Kinetic parameters for antibody 16E7 catalyzed reactions
Substrates
Enol ether 2a
Steady-state phased
Benzisoxazole 7b
Steady-state phasef
Initial phasec
Initial phasee
kcat (sÀ1
)
1.00 · 10À3
7.99 · 10À7
1262
1.11 · 10À3
7.99 · 10À7
1392
n.d.
3.62 · 10À6
n.d.
n.d.
3.62 · 10À6
n.d.
kuncat (sÀ1
kcat/kuncat
)
KM (lM)
513
1.97 · 10À6
0.406
171
6.47 · 10À6
0.124
n.d.
4.97 · 10À5g
n.d.
1.13 · 10À4g
kcat/KM (lMÀ1 sÀ1
KS (lM)h
)
7.28 · 10À2
3.20 · 10À2
a All assays were performed at an active site concentration of 3.33 lM.
b All assays were performed at an active site concentration of 0.48 lM.
c The kinetic parameters were determined after 3 h incubation time.
d The kinetic parameters were determined after 15 h incubation time.
e The kinetic parameters were determined after 6 min incubation time.
f The kinetic parameters were determined after 2 h incubation time.
g The low solubility of 7 prevented measurements at substrate concentrations above 2500 lM, without evidence for substrate saturation, hence only
kcat/KM is reported.
h KS is defined as the dissociation constant of the complex between catalyst and the reactionÕs transition state and can be calculated by following
equation: KS = kuncat/(kcat/KM).
and AspH50, which are placed at 9 A distance from one
another (Fig. 2A). The hapten–antibody contact con-
sists mainly in a charge-neutralizing hydrogen bond net-
4B2, which was obtained against a guanidine-type
hapten related to 1 and which also catalyzes the depro-
tonation of substrate 7, as well as the deprotonation of a
b-c-unsaturated ketone.7
˚
work involving the negatively charged GluL39, TyrL41
,
and the nitrogen atoms of the cationic guanidinium
group of hapten 1. A similar pair of carboxylates has
been observed in the structure of catalytic antibody
Enol ether substrate 2 also docked successfully into the
Fv domain binding pocket of the 16E7-E2 model (Fig.
2B). The substrateÕs reactive enol ether function is posi-
tioned in proximity of the carboxylate group of GluL39
previously found to interact with the guanidinium group
of hapten 1, with an orientation suitable for proton
transfer. TyrL41 forms two hydrogen bonds with GluL39
and the oxygen atom of the substrateÕs enol ether func-
Ab 14D9
(anti-9)
MeO
Ar
L39
˚
tional group. The distance of 4.5 A separating Glu
NHCO(CH2)3CO2H
from the enol ether b-carbon is analogous to the situa-
tion observed in the related catalytic antibody 14D9,
an antibody raised against the quaternary ammonium
hapten 9 and which catalyzes the enantioselective pro-
tonation of enol ethers such as the tranformation of
10 to (S)-11 (Fig. 3). This suggests that a water molecule
might be involved in bridging proton transfer between
10
N
Me
O
H
H
N
Ar
OH
O
(S)-11
9
the substrate and the carboxylic acid function of GluL39
This water molecule might also trap the intermediate
.
Figure 2. Enol ether protonation reaction with antibody 14D9 raised
against hapten 9.

L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
1025
Figure 3. Views of the 16E7 Fv model combining site. The residues in the binding pocket are drawn in gray stick and the ligands are shown in green.
(A) Complex with docked hapten 1. (B) Complex with enol ether substrate 2. (C) Complex with docked benzisoxazole substrate 7. (D) HCDR3
flexibility in models 16E7 E1-E7. E1: green; E2: cyan; E3: olive; E4: orange; E5: lime; E6: magenta; E7: salmon. The GluL39 and SerH99 of model E1,
E2 are shown in green and gray sticks, respectively.
oxocarbonium cation 5 to form the acyclic hemiacetal 6
as the primary reaction product (Fig. 1). AspH50 forms a
hydrogen bond with the phenolic OH-group of substrate
2, holding this group away from the enol ether function.
This orientation might explain the fact that enol ether 2
does not undergo an intramolecular cyclization to the
corresponding acetal 4 under antibody catalysis, but
rather undergoes hydrolysis to 3 as observed in the
background reaction in water.
to either glutamine or alanine abolished catalytic activ-
ity in both, the protonation reaction with enol ether 2
and the deprotonation reaction with benzisoxazole 7,
establishing that this residue is critical for catalysis in
antibody 16E7. Residue AspH50 was mutated to alanine
to remove its hydrogen-bonding interaction with the
phenolic hydroxyl group of enol ether 2 as observed in
the docked model (Fig. 2A), with the hope that the
AspH50Ala mutant might promote an intramolecular
cyclization of enol ether 2 to form the cyclic acetal 4.
Unfortunately, the AspH50Ala mutant was completely
inactive for the protonation reaction. Nevertheless, this
mutant retained 25% of the WT activity for the deproto-
nation reaction with benzisoxazole 7. The selective loss
of catalysis for the enol ether reaction suggests that this
reaction is more sensitive to perturbations than the
KempÕs elimination reaction of benzisoxazole 7, in
agreement with the fact that the later reaction is very
sensitive to non-specific catalysis effects.16 The loss of
activity in the case of enol ether 2 might be due to per-
turbed substrate binding.
Benzisoxazole 7 also docked into 16E7-E2 model to pro-
duce a stable complex (Fig. 2C). In this case, the critical
acidic residue GluL39 forms two hydrogen bonds, one
with TyrL41, and another with the nitrogen atom in
the substrateÕs benzisoxazole ring. The substrateÕs
˚
deprotonation site is located at a distance of 3 A from
the carboxylate of GluL39 in an arrangement suitable
for direct proton transfer.
3.3. Mutagenesis of catalytic residues
Mutagenesis experiments suggested by the modeling and
docking study were carried out to probe the reaction
mechanism of antibody 16E7 (Table 1). The role of
GluL39, which appeared as the best candidate for a cat-
alytic residue, was investigated first. Mutation of GluL39
3.4. Analysis of loop movements
The fact that docking hapten 1 or the substrates onto
the lowest energy 16E7-E1 model did not produce a

1026
L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
Table 3. The conformation characters of HCDR3 with increasing
energy
Energy minimization
conformation
DG
(kcal/mol)
Loop flexibility
(deg)
E10
E2
E3
E4
E5
E6
E7
0
1.30
2.07
4.22
4.48
4.65
4.82
67.64
33.47
54.21
47.68
42.03
23.15
E1–E7 are the generated conformations during whole molecular min-
imization and ordered by increasing energies. DG is the energy differ-
ence between every conformation and the minimized conformation.
Loop angles were measured with the program Swisspdbviewer. The Ca
atom of GlycineH98 in the HCDR3 conformation of E1was chosen as
reference point and defined as an angle of 0ꢁ. Angles are defined for the
motion of the Ca atom of GlyH98 in relation to the Ca atom of ArgH95
as fixed point.
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
10
20
30
40
50
0
Figure 4. Comparsion of 16E7 models molecular surface. Top: the
16E7-E1 model, bottom: the 16E7-E2 model.
Model number
0.4
0.3
0.2
0.1
0
complex suggested that the ligands could not enter the
binding pocket in this model. Indeed AUTODOCK 3.05
employs a molecule ÔwalkingÕ procedure on the surface
of the molecule to search the potential binding pocket.
The structural difference between the 16E7-E1 model
and the 16E7-E2 model used to produce docked
complexes was identified as a large conformational
change in the HCDR3 loop. In the 16E7-E1 model, this
loop closes the binding pocket unusually deeply, pre-
venting ligand access. In the 16E7-E2 model, by con-
trast, the loop is placed such as to leave the pocket
open (Fig. 4).
-0.1
-0.2
-0.3
-0.4
0
10
20
30
40
5
0
Model number
Figure 5. Scalar representation of CDR3 loops flexibility of 16E7 Fv
models E1–E50. Two atoms on the tip of the CDR3 loops of heavy
and light chain were used to calculate the movement of their loops.
Top: Ca atom of the GlyH98, bottom: Ca atom of ValL99. Their
positions in the E1model are set to zero and compared to the
corresponding atoms in the conformations of the other models.
The five next lower energy models for 16E7 were exam-
ined. These were found to represent intermediate states
of a loop movement between the 16E7-E1 and the
16E7-E2 conformation (Fig. 2D). The HCDR3
loop movement involved seven contiguous residues
3.5. Mutagenesis of Serine H99
(TyrH96-GlyH97-GlyH98-SerH99-AlaH100-ValH101-SerH102
)
The movement of the HCDR3 observed between the
16E7-E1 model and the 16E7-E2 model might explain
the hysteresis behavior of antibody 16E7. Indeed, the
substrates might meet a similar entry problem in the real
antibody to that observed in the docking procedure with
the 16E7-E1 model. Thus, the 16E7-E1 model might
represent a conformation existing in the catalytic anti-
body during the initial kinetic phase (vi), while the
16E7-E2 model might represent the antibody conforma-
tion during the activated state of the kinetic phase (vss).
and featured a rotation spanning 67ꢁ as measured for
the Ca atom of GlyH98 in relation to the Ca atom of
ArgH95 as fixed point. The calculated energy difference
between E1and the least stable of these seven confor-
mers amounted to 4.8 kcal/mol, suggesting that the loop
rearrangement required a significant activation energy
(Table 3). Further examination of the 50 lowest energy
conformers showed that the HCDR3 loop was more
flexible compared to its LCDR3 counterpart (Fig. 5).

L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
1027
Since SerH99 played a crucial role in locking the closed
conformation of the 16E7-E1 model by hydrogen-bond-
ing interactions with GluL39, we reasoned that mutating
this residue to an alanine might destabilize the inactive
state of antibody 16E7 and therefore reduce the activa-
tion time s for reaching full activity. Substituting alanine
for serine at position H99 produced a catalytically active
mutant for both the protonation of enol ether 2 and the
deprotonation of benzisoxazole 7, showing approxi-
mately 25% of WT activity in both cases. More impor-
tantly, the hysteresis time s in the SerH99Ala mutant
was reduced to 44% of WT for the reaction of benzisox-
azole 7, and to 62% of WT for the protonation of enol
ether 2 (Table 1, Fig. 6). Similar reductions in hysteresis
were observed when SerH99 was deleted. These effects are
consistent with the mechanistic hypothesis assigning the
hysteresis behavior to a movement of the HCDR3 loop
between a more stable, catalytically less active con-
former (model 16E7-E1) and a slightly less stable, cata-
lytically more active conformer (model 16E7-E2), taking
place in a slow transition requiring a significant activa-
tion energy.
4. Discussion
4.1. Homology modeling and docking
Homology modeling has been used broadly for antibod-
ies17 and the validity of the models produced has been
confirmed in several cases by crystallographic studies.18
Computational simulations have also been used exten-
sively to analyze protein dynamics,19 and the results
confirmed by fluorescence,20 CD spectroscopy and
NMR studies.21 While the overall structure of antibod-
ies is largely conserved, modeling concentrates on pre-
dicting the conformation of the CDR in the active site.
In this study we used the online antibody structure mod-
eling service WAM (Web Antibody Modeling, http://
antibody.bath.ac.uk/),13 which uses a combination of se-
quence comparison and conformation searches for pre-
dicting the three-dimensional structure from sequence
data. In order to probe the reliability of this service we
submitted the sequence of catalytic antibody 14D9 from
our group prior to its publication.15 In this case we ob-
tained a modeled structure, which was very close to the
actual structure as observed in the crystal structure,
showing that the model is capable of accurate predic-
tions for antibodies raised against small molecule hap-
tens. Docking of ligands into the 16E7 model was
accomplished using the docking program AUTODOCK
3.0514 This program has been described to be generally
suitable for elucidating interactions between antibodies
and ligands.22 In the present study modeling and dock-
ing were used as a guide for selecting mutagenesis exper-
iments to probe the reaction mechanism and hysteresis
effect.
200
(A)
150
100
50
0
0
510
1520
time / (h)
25
160
120
80
40
0
(B)
4.2. Catalytic mechanism of antibody 16E7
The mechanism of proton transfer to and from carbon
in enzyme catalysis is of particular interest since this step
is generally kinetically slow and may be rate limiting in a
catalytic cycle. The data above for antibody 16E7 clearly
point to GluL39 as the key catalytic residue responsible
for both the protonation catalysis with enol ether 2
and the deprotonation catalysis with benzisoxazole 7.
The direct carbon deprotonation reaction of substrate
7 by GluL39 proposed here for antibody 16E7 is similar
to the general base mechanism proposed for catalytic
antibody 2B4 on the basis of its crystal structure.7 On
0
4
8
12
16
20
24
time / (h)
Figure 6. Hysteresis effect in catalytic antibody 16E7. (A) Hydrolysis
of the enol ether 2. Hysteresis observed for WT Fab-16E7 (h) and for
the mutated Fab-16E7(SerH99Ala) (j). The values for the initial
velocity mi and final velocity mss are fitted according to the equation
described in the literature.4,12 Assays were performed at an initial
substrate concentration of 300 lM and an antibody concentration of
0.54 mg/mL (29 mM BisTris, 104 mM NaCl, 10% acetonitrile,
pH 6.24, 37 ꢁC). (B) Observed hysteresis of the chimeric WT Fab
16E7-catalyzed (h) and of the Fab-16E7 (SerH99Ala)-catalyzed (j)
elimination of the Kemp substrate 7. Assays were performed at an
initial substrate concentration of 475 lM and an antibody concentra-
tion of 0.09 mg/mL (29 mM BisTris, 104 mM NaCl, 10% acetonitrile,
pH 6.24, 37 ꢁC).
˚
the other hand, we observed a distance of 4.5 A between
the catalytic carboxyl group of GluL39 and the b-carbon
of enol ether 2 in the docked model of 16E7. This situ-
ation is analogous to that observed in the crystal struc-
ture of catalytic antibody 14D9, which also catalyzes an
enol ether protonation reaction, suggesting a similar
mechanism with an intervening water molecule.15 The
data with antibody 16E7 thus bridges both situations
and suggests a common mechanism for proton transfer
to and from carbon in catalytic antibodies. The situation
is also similar in several enzymes where glutamate

1028
L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
residues have been found to catalyze protonation or
deprotonation reactions.7,23
consistent with the slow hysteresis if attributed to the
modeled conformational rearrangement between E1
and E2. However, the energy barrier of 4.8 kcal/mol
given by the model is not sufficient to quantitatively
explain the very long activation times observed. The
modeled conformational rearrangement is also consis-
tent with the threefold tighter binding of enol ether 2
in the steady state (open conformer E2) versus initial
phase (closed conformed E1). Although the KM values
are not accessible, the twofold higher activity with benz-
isoxazole 7 in the steady state versus initial phase might
also be caused by tighter substrate binding.
The hydrogen-bonding interaction between the phenolic
hydroxyl group of substrate 2 and AspH50 observed in
the docked model of antibody 16E7 with the enol ether,
which places this phenolic hydroxyl group away from
the enol ether, provides a structural model to explain
the exclusive trapping of the intermediate carbocation
by water, which takes place in the antibody catalysis
despite of the desolvation occurring upon substrate
binding. Desolvation should indeed rather favor an
intramolecular cyclization pathway to acetal 4, as
observed upon treatment of enol ether 2 with strong
acid under non-aqueous conditions.4 The binding
conformation of substrate 2 in the docked model is
also in agreement with the observation that antibody
16E7 is not inhibited by product 3 and its cyclic acetal
analog 4, an observation which alone suggests that the
binding pocket is not suited for forming the cyclized
product.
Induced-fit upon ligand binding and pre-equilibrium
have been considered as driving forces for conforma-
tional changes in several hysteresis enzymes and anti-
bodies.27,30,32 In the present case of antibody 16E7, the
closed, catalytically inactive E1conformation could be
favored in the absence of substrate. The open E2 confor-
mation would represent the tighter binding conforma-
tion, and would exist mainly with bound ligands. The
substrate–antibody interaction in this model may be de-
scribed either as induced-fit or as pre-equilibrium
depending on whether the conformational rearrange-
ment occurs with or without bound substrate. In any
event this two-state model is consistent with the fact that
the constant s decreases upon increasing substrate con-
centration, and with the fact that the inactive state is
regenerated upon repurification of the antibody by pro-
tein G affinity chromatography.
4.3. Hysteresis effect
Hysteresis effects in catalysis have been observed and
characterized in several enzymes and linked to protein
motions such as loop movements and quaternary con-
formational rearrangements. These phenomena are re-
ferred to as Ôprotein breathingÕ when their time scale is
on the order of several hours.24,25 Antibodies have also
been shown to display protein dynamics effects ranging
from adjustment of a single amino acid side chain26
and loop rearrangements27 to entire domain move-
ments.28 By contrast, catalytic antibodies have been gen-
erally regarded as possessing a relatively rigid binding
pocket, and numerous crystal structures of catalytic
antibodies show no significant differences between apo-
and hapten-bound structures.29 A study by Tawfik and
co-workers suggested that a protein conformational
change might be responsible for a relatively short hyster-
esis effect, observed as a delay of a few tens of seconds,
in the activation of a catalytic antibody upon substrate
binding.30 The protein dynamical aspects of antibody
binding have been recently demonstrated by the same
authors by the crystallographic observation of two isola-
ble distinct conformations of a single antibody binding
to two different haptens.31
5. Conclusion
The mechanistic model proposed on the basis of site-
directed mutagenesis and molecular modeling provides
a consistent picture of the mechanism of proton transfer
to and from carbon in catalytic antibody 16E7. Muta-
genesis of the key residues GluL39, AspH50, and
SerH99 shows that these residues control the catalysis
and hysteresis phenomenon of the antibody. Proton
transfer depends on GluL39 and proceeds analogously
to proton transfer steps observed in related systems,
which supports a common mechanism for proton trans-
fer in catalytic antibodies. The unusual hysteresis of
antibody 16E7 might be explained by a slow conforma-
tional movement of the HCDR3 loop.
The homology modeling suggests that the unusual hys-
teresis in antibody 16E7 catalysis, which is observed
with both substrates, originates in a slow conforma-
tional movement of the HCDR3 loop switching from
a closed conformation blocking substrate access to the
catalytic site (16E7-E1 model) to a catalytically active
open conformation (16E7-E2 model). This mechanism
is supported by the 38–56% reduction in hysteresis time
observed in the SerH99Ala and the SerH99 deletion
mutants (Table 1). This effect can be interpreted in terms
of residue SerH99 locking the closed conformation by
hydrogen bonding to GluL39 as suggested by the
model. Interestingly models E3 to E7 displaying inter-
mediate states of the loop movement have a significantly
higher calculated energy relative to E1and E2, which is
Acknowledgements
This work was supported by the Swiss National Science
Foundation, the University of Basel, and the University
of Berne.
References and notes
1. Schultz, P. G.; Yin, J.; Lerner, R. A. Angew. Chem., Int.
Ed. 2002, 41, 4427–4437.
2. Stevenson, J. D.; Thomas, N. R. Nat. Prod. Rep. 2000, 17,
535–537.
3. Kemp, D. S. Nature 1995, 373, 196–197.

L. Zheng et al. / Bioorg. Med. Chem. 13 (2005) 1021–1029
1029
4. Manetsch, R.; Zheng, L.; Reymond, M. T.; Woggon,
W.-D.; Reymond, J.-L. Chem. Eur. J. 2004, 10, 2487–
2506.
5. Reymond, J.-L.; Jahanghiri, G. K.; Stoudt, C.; Lerner, R.
A. J. Am. Chem. Soc. 1993, 115, 3909–3917.
Roberts, V. A., III; Liu, J.; Neagu, I.; Iwama, S.; Phillips,
B. W.; Smith, A. B., III; Hirschmann, R. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 1953–1958; (d) Kolesnikov, A. V.;
Kozyr, A. V.; Alexandrova, E. S.; Koralewski, F.; Demin,
A. V.; Titov, M. I.; Avalle, B.; Tramontano, A.; Paul, S.;
Thomas, D.; Gabibov, A. G.; Friboulet, A. Proc. Natl.
Acad. Sci. U.S.A. 2000, 97, 13526–13531.
6. Thorn, S. N.; Daniels, R. G.; Auditor, M.-T. M.; Hilvert,
D. Nature 1995, 373, 228–230.
7. Golinelli-Pimpaneau, B.; Goncalves, O.; Dintinger, T.;
Blanchard, D.; Knossow, M.; Tellier, C. Proc. Natl. Acad.
Sci. U.S.A. 2000, 97, 9892–9895.
8. Barbas, C. F.; Kang, A. S.; Lerner, R. A.; Benkovic, S. J.
Proc. Natl. Acad. Sci. U.S.A. 1991, 88, 7978–7982.
9. Ulrich, H. D.; Patten, P. A.; Yang, P. L.; Romesberg,
F. E.; Schultz, P. G. Proc. Natl. Acad. Sci. U.S.A. 1995,
92, 11907–11911.
10. Yin, J.; Mundorff, E. C.; Yang, P. L.; Wendt, K. U.;
Hanway, D.; Stevens, R. C.; Schultz, P. G. Biochemistry
2001, 40, 10764–10773.
11. Harlow, E.; Lane, D. Antibodies: A Laboratory Manual;
Cold Spring Harbor Lab. Press: Plainview, NY, 1988.
12. Neet, K. E.; Ainslie, R. Methods Enzymol. 1980, 64, 192–
226.
19. Hansson, T.; Oostenbrink, C.; Van Gunsteren, W. Curr.
Opin. Struct. Biol. 2002, 12, 190–196.
20. Chattopadhyay, K.; Saffarian, S.; Elson, E. L.; Frieden, C.
Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 14171–14176.
21. Kroon, G.; Martinez-Yamout, M.; Krebs, J. F.; Chung, J.;
Dyson, H. J.; Wright, P. E. J. Biomol. NMR 1999, 15, 83.
22. (a) Sotriffer, C. A.; Flader, W.; Winger, R. H.; Rode, B.
M.; Liedl, K. R.; Varga, J. M. Methods: Companion
Methods Enzymol. 2000, 20, 280–291; (b) Heine, A.; Stura,
E. A.; Yli-Kauhaluoma, J. T.; Gao, C.; Deng, Q.; Beno, B.
R.; Houk, K. N.; Janda, K. D.; Wilson, I. A. Science 1998,
279, 1934–1940.
23. Mitra, B.; Kallarakal, A. T.; Kozarich, J. W.; Gerlt, J. A.;
Clifton, J. G.; Petsko, G. A.; Kenyon, G. L. Biochemistry
1995, 34, 2777–2787.
13. Whitelegg, N. R. J.; Rees, A. R. Protein Eng. 2000, 13,
819–824.
24. Cannon, W. R.; Benkovic, S. J. J. Biol. Chem. 1998, 273,
26257–26260.
14. Morris, G. M.; Goodsell, D. S.; Halliday, R. S.; Huey, R.;
Hart, W. E.; Belew, R. K.; Olson, A. J. J. Comput. Chem.
1998, 19, 1639–1662.
25. Frieden, C. Annu. Rev. Biochem. 1979, 48, 471–489.
26. Stanfield, R. L.; Fieser, T. M.; Lerner, R. A.; Wilson, I. A.
Science 1990, 248, 712–719.
15. Zheng, L.; Baumann, U.; Reymond, J.-L. Proc. Natl.
Acad. Sci. U.S.A. 2004, 101, 3387–3392.
27. Rini, J. M.; Schulze-Gahmen, U.; Wilson, I. A. Science
1992, 255, 959–965.
16. Kirby, A. J.; Hollfelder, F.; Tawfik, D. S. Appl. Biochem.
Biotechnol. 2000, 83, 173–180.
28. Guddat, L. W.; Shan, L.; Broomell, C.; Ramsland, P. A.;
Fan, Z.; Anchin, J. M.; Linthicum, D. S.; Edmundson, A.
B. J. Mol. Biol. 2000, 302, 853–872.
17. (a) Roberts, V. A.; Stewart, J. D.; Benkovic, S. J.; Getzoff,
E. D. J. Mol. Biol. 1994, 235, 1098–1116; (b) Miyashita,
H.; Hara, T.; Tanimura, R.; Fukuyama, S.; Cagnon, C.;
Kohara, A.; Fujii, I. J. Mol. Biol. 1997, 267, 1247–1257;
(c) Stewart, J. D.; Roberts, V. A.; Thomas, N. R.; Getzoff,
E. D.; Benkovic, S. J. Biochemistry 1994, 33, 1994–2003;
(d) Fujii, I.; Tanaka, F.; Miyashita, H.; Tanimura, R.;
Kinoshita, K. J. Am. Chem. Soc. 1995, 117, 6199–6209.
18. (a) Thayer, M. M.; Olender, E. H.; Arvai, A. S.; Koike, C.
K.; Canestrelli, I. L.; Stewart, J. D.; Benkovic, S. J.;
Getzoff, E. D.; Roberts, V. A. J. Mol. Biol. 1999, 291, 329–
345; (b) Kristensen, O.; Vassylyev, D. G.; Tanaka, F.;
Morikawa, K.; Fujii, I. J. Mol. Biol. 1998, 281, 501–511;
(c) Smithrud, D. B.; Benkovic, P. A.; Benkovic, S. J.;
29. (a) Hugot, M.; Bensel, N.; Vogel, M.; Reymond, M. T.;
Stadler, B.; Reymond, J.-L.; Baumann, U. Proc. Natl.
Acad. Sci. U.S.A. 2002, 99, 9674–9678; (b) Wedemayer, G.
J.; Patten, P. A.; Wang, L. H.; Schultz, P. G.; Stevens, R.
C. Science 1997, 276, 1665–1669; (c) Gigant, B.; Char-
bonnier, J. B.; Eshhar, Z.; Green, B. S.; Knossow, M.
Proc. Natl. Acad. Sci. U.S.A. 1999, 94, 7857–7861.
30. Lindner, A. B.; Eshhar, Z.; Tawfik, D. S. J. Mol. Biol.
1999, 285, 421–430.
31. James, L. C.; Roversi, P.; Tawfik, D. S. Science 2003, 299,
1362–1367.
32. Jaaskelainen, S.; Verma, C. S.; Hubbard, R. E.; Linko, P.;
Caves, L. S. Protein Sci. 1998, 7, 1359–1367.