Mechanistic studies of an antibody-catalyzed elimination reaction

Article abstract of DOI:10.1021/ja9738992

Catalytic antibody 43D4-3D12, which was generated against the substituted tertiary amine 1, catalyzes the elimination of HF from β- fluoroketone 2. We have cloned and produced the antibody as a chimeric Fab and constructed a model of the active site-substrate complex. Mutagenesis studies of the active site indicate that Glu(H)50 acts as the general base and suggest that Tyr96(L) may also play a role in the elimination reaction. Antibody 43D4-3D12 also efficiently catalyzes the elimination of HBr from substrate 4 by an E2 mechanism, again involving selective abstraction of the proton β-to the nitrophenyl ring by Glu(H)50. The antibody-catalyzed reaction affords predominantly the internal olefins, whereas the major product resulting from the uncatalyzed reaction is the alcohol, which arises from the competing substitution reaction. In addition, antibody 43D43D12 catalyzes an acetal hydrolysis reaction in which Glu(H)50 likely acts as a general acid. These studies point to the Success of this particular hapten design strategy in generating an active site with a desired catalytic functional group. They also illustrate the utility of using related reactions as mechanistic probes of biological catalysis.

Full text of DOI:10.1021/ja9738992

5160
J. Am. Chem. Soc. 1998, 120, 5160-5167
Mechanistic Studies of an Antibody-Catalyzed Elimination Reaction
†
‡
Floyd E. Romesberg, Mark E. Flanagan, Tetsuo Uno, and Peter G. Schultz*
Contribution from the Howard Hughes Medical Institute and Department of Chemistry, UniVersity of
California, Berkeley, California 94720, Central Research DiVision, Pfizer Inc., Groton, Connecticut 06340,
and Symyx Technologies, 3100 Central Expressway, Santa Clara, California 95051
ReceiVed NoVember 13, 1997
Abstract: Catalytic antibody 43D4-3D12, which was generated against the substituted tertiary amine 1, catalyzes
the elimination of HF from â-fluoroketone 2. We have cloned and produced the antibody as a chimeric Fab
and constructed a model of the active site-substrate complex. Mutagenesis studies of the active site indicate
H
L
that Glu 50 acts as the general base and suggest that Tyr96 may also play a role in the elimination reaction.
Antibody 43D4-3D12 also efficiently catalyzes the elimination of HBr from substrate 4 by an E2 mechanism,
H
again involving selective abstraction of the proton â-to the nitrophenyl ring by Glu 50. The antibody-catalyzed
reaction affords predominantly the internal olefins, whereas the major product resulting from the uncatalyzed
reaction is the alcohol, which arises from the competing substitution reaction. In addition, antibody 43D4-
H
3
D12 catalyzes an acetal hydrolysis reaction in which Glu 50 likely acts as a general acid. These studies
point to the success of this particular hapten design strategy in generating an active site with a desired catalytic
functional group. They also illustrate the utility of using related reactions as mechanistic probes of biological
catalysis.
Introduction
the recombinant antibody in E. coli. A number of active site
mutants were generated based on a model of the antibody-
substrate complex, and their catalytic properties were character-
ized. In addition, we have used a number of alternative
chemical reactions as mechanistic probes of the active site
Many enzymatic transformations, including isomerization,
elimination, and condensation reactions, involve proton transfeer
1
to and from carbon. Catalytic antibodies have been shown to
catalyze many of these same reactions.2
-6
For example,
(Figure 1). These studies provide additional insights into the
antibody 43D4-3D12, which was generated against positively
charged hapten 1, was found to catalyze the elimination of HF
from â-fluoroketone 2 (Figure 1) with a rate enhancement of
requirements and scope of proton transfer in antibody- and
enzyme-catalyzed reactions.
5
approximately 10 relative to the reaction catalyzed by free
Materials and Methods
acetate in solution.2 In this case, rather than mimicking the
transition state of the reaction, hapten 1 contains an ammonium
group designed to elicit a carboxylate residue appropriately
positioned to affect catalysis. Chemical modification studies
of antibody 43D4-3D12 confirmed that either an aspartate or
glutamate residue plays a key role in catalysis, and the pH
dependence of the reaction indicated an essential residue with
a pKa of 6.3, active in the deprotonated form. Kinetic isotope
General Methods. Unless otherwise noted, materials were obtained
from commercial suppliers and were used without further purification.
Dichloromethane was dried by distillation from CaH , and tetrahydro-
2
furan was dried by distillation from Na0 and benzophenone. All
aqueous solutions were prepared from distilled, deionized water. All
oxygen or moisture-sensitive reactions were carried out in oven-dried
glassware under a positive pressure of nitrogen. Analytical and
preparative thin-layer chromatography (PTLC) were performed on
precoated silica gel plates (Merck). Flash chromatography was
performed using Keiselgel 60 (230-400 mesh) silica gel.
3
effect data are consistent with an E2 or E1cb mechanism. The
antibody has also been shown to catalyze regioselective oxime
1
H NMR spectra of compounds were recorded on a Bruker AM-
formation between 4-nitroacetophenone and hydroxylamine
4
00 (400 MHz) Fourier-transform NMR spectrometer at the University
7
(
Figure 1). This additional catalytic activity is consistent with
1
of California, Berkeley NMR facility. H Resonances are reported in
the presence of an active site carboxylate group that can function
as either a general acid or general base.
To further investigate the mechanism and structure of
antibody 43D4-3D12, we have cloned, sequenced, and produced
13
units of parts per million downfield from tetramethylsilane (TMS).
C
NMR spectra were recorded on the same instrument; all spectra are
proton-decoupled. IR spectra were recorded on a Paragon 500
spectrophotometer (Perkin-Elmer) using NaCl plates. Mass spectra
+
+
(
FAB and EI ) were recorded at the Mass Spectroscopy Laboratory
†
Pfizer Inc.
Symyx Technologies.
of the University of California, Berkeley on a VG Zab2EQ or a Kratos
MS50 (xenon beam, 7 V) mass spectrometer. Elemental analyses were
performed by the Microanalytical Laboratory operated by the College
of Chemistry at the University of California, Berkeley. Melting points
were determined using a Mel-Temp melting point apparatus and are
uncorrected.
‡
(1) Walsh, C. Enzymatic Reaction Mechanisms; W. H. Freeman: New
York, 1979.
(
2) Shokat, K. M.; Leumann, C.; Sugasawara, R.; Schultz, P. G. Nature
1
989, 383, 269.
(
3) Shokat, K.; Uno, T.; Schultz, P. G. J. Am. Chem. Soc. 1994, 116,
2
261.
4
-(4-Nitrophenyl)-2-butanone. To a stirred solution of 1-iodo-4-
(
(
4) Reymond, J. L. Angew. Chem., Int. Ed. Engl. 1995, 34, 2285.
5) Wagner, J. L.; Lerner, R. A.; Barbas, C. F., III. Science 1995, 270,
nitrobenzene (1.0 g, 4.02 mmol) and 3-butene-2-ol (0.363 g, 5.05 mmol)
in 10 mL of acetonitrile were added 9 mg (0.040 mmol) of palladium
acetate and 0.127 g (1.25 mmol) of triethylamine, and the mixture was
heated to reflux for 24 h. The reaction was diluted with ether and
1
797.
(6) Jackson, D. Y.; Schultz, P. G. J. Am. Chem. Soc. 1991, 113, 2319.
7) Uno, T.; Gong, B.; Schultz, P. G. J. Am. Chem. Soc. 1994, 116, 1145.
(
S0002-7863(97)03899-7 CCC: $15.00 © 1998 American Chemical Society
Published on Web 05/15/1998

Antibody-Catalyzed Elimination Reaction
J. Am. Chem. Soc., Vol. 120, No. 21, 1998 5161
2
7
Figure 1. Reactions catalyzed by the antibody 43D4-3D12 which was generated against hapten 1: (a) â-fluoride elimination; (b) oxime formation;
c) bromide elimination; and (d) acetal hydrolysis.
(
washed with saturated NH
saturated brine, then dried over MgSO
vacuo. The residue was then purified by flash chromatography (silica;
:1 hexanes/ethyl acetate) affording 0.7 g of product (92% yield): R
4
Cl (aq), saturated NaHCO
3
(aq), water, and
concentrated to dryness in vacuo. The residue was purified by PTLC
4
and concentrated to dryness in
(silica; 4:1 hexanes/ethyl acetate) affording 25 mg of product as a clear
1
oil (54% yield): R
f
) 0.53; H NMR (400 MHz) (CDCl
3
) δ TMS 1.74
2
f
(t, 3H, J ) 6.7 Hz), 2.03-2.16 (m, 2H), 2.83-2.90 (m, 1H), 2.96-
1
)
0.41, amber crystals from ethyl acetate/hexanes, mp 37-40 °C; H
3.03 (m, 1H), 4.01-4.07 (m, 1H), 7.35-7.39 (m, 2H), 8.14-8.17 (m,
NMR (400 MHz) (CDCl
3
) δ TMS 2.15 (s, 3H), 2.81 (t, 2H, J ) 7.4
2H); 13C NMR (100 MHz) (CDCl
3
) δ 26.47, 33.89, 41.92, 50.02,
123.77, 129.32, 146.56, 148.74; IR (NaCl): 2924, 2863, 1600, 1518,
Hz), 2.99 (t, 2H, J ) 7.3 Hz), 7.34 (d, 2H, J ) 8.7 Hz), 8.13 (d, 2H,
1
3
-1
+
+
J ) 8.7 Hz); C NMR (100 MHz) (CDCl
3
) δ 29.29, 30.01, 44.13,
1346, 1110, 857 cm ; mass spectrum (EI) m/z 259 (M ), 257 (M ).
Anal. Calcd for C10 12BrNO : C, 46.55; H, 4.69; N, 5.43. Found:
C, 46.20; H, 4.64; N, 5.29.
1
1
23.70, 129.22, 146.49, 148.92, 206.61; IR (NaCl) 2936, 1712, 1598,
H
2
-
1
514, 1342, 1169, 856 cm ; HRMS (FAB), calcd for C10
H
12NO
3
+
(MH ) 194.0817, found 194.0780.
cis-4-Phenyl-2-butene. To a stirred solution of benzyl bromide (1.0
g, 5.85 mmol) in 30 mL of THF was added 68 mg (0.059 mmol) of
tetrakis(triphenylphosphine)palladium followed by 11.7 mL of a 0.5
M solution of cis-propenylmagnesium bromide in THF, and the resulting
mixture was stirred at room temperature for 48 h. To the reaction was
4
-(4-Nitrophenyl)-2-butanol (5). To a stirred solution of 4-(4-
nitrophenyl)-2-butanone (39 mg, 0.202 mmol) in 5 mL of methanol
cooled to 0 °C was added 31 mg (0.807 mmol) of NaBH , and the
4
resulting reaction was stirred at 0 °C for 30 min. The mixture was
quenched with 1 N HCl (aq) and concentrated under reduced pressure.
The residue was redissolved in dichloromethane, washed with saturated
then added saturated NH
ether. The ether layer was then washed with saturated NaHCO
and saturated brine, dried over MgSO , and concentrated to dryness in
4
Cl (aq), and the mixture was extracted with
3
, water,
NaHCO
The crude product was purified by PTLC (silica; 2:1 hexanes/ethyl
acetate) affording 35 mg of product as a clear oil (89% yield): R
3
, dried over Na
2
SO
4
, and concentrated to dryness in vacuo.
4
vacuo. The residue was purified by flash chromatography (silica; 50:1
f
)
hexanes/ethyl acetate), affording 69 mg of purified product as a liquid
1
1
0
1
2
.24; H NMR (400 MHz) (CDCl
3
) δ TMS 1.24 (d, 3H, J ) 6.2 Hz),
(9% purified yield): R
f
) 0.33; H NMR (400 MHz) (CDCl
3
) δ TMS
.72-1.82 (m, 2H), 2.74-2.92 (m, 2H), 3.78-3.86 (m, 1H), 7.35 (d,
1.73-1.74 (m, 3H), 3.42 (d, 2H, J ) 5.1 Hz), 5.59-5.62 (m, 2H),
7.18-7.21 (m, 3H), 7.29 (t, 2H, J ) 8.1 Hz).
H, J ) 8.8 Hz), 8.11-8.14 (m, 2H); 1 C NMR (100 MHz) (CDCl
3
3
)
δ 23.77, 31.98, 40.10, 67.08, 123.62, 129.18, 146.27, 150.15; IR
cis-4-(4-Nitrophenyl)-2-butene (6a). To a stirred solution of cis-
-phenyl-2-butene in 1 mL of trifluoroacetic anhydride at 0 °C was
-1
(
NaCl): 3369, 2968, 1600, 1517, 1347, 1110, 856 cm ; HRMS (FAB)
4
+
calcd for C18
H14NO
3
(MH ) 196.0974, found 196.0974.
added 35 mg (0.151 mmol) of copper(II) nitrate, and the mixture stirred
at 0 °C for 30 min, then warmed to room temperature with continued
stirring for 2 h. The reaction mixture was poured onto ice, and after
the ice melted, the reaction mixture was neutralized with 1 N NaOH
followed by extraction with ether. The ether layer was washed with
2
-Bromo-4-(4-nitrophenyl)butane (4). To a stirred solution of
alcohol 5 (35 mg, 0.179 mmol) in 3 mL of acetonitrile was added 0.118
g (0.357 mmol) of carbon tetrabromide followed by 94 mg (0.357
mmol) of triphenylphosphine, and the resulting mixture was stirred at
room temperature for 3 h. Silica gel (4 g) was added and the slurry
dried to a powder under reduced pressure. The crude product was then
eluted from the silica gel with 2:1 hexanes/ethyl acetate and the eluent
saturated NaHCO
concentrated to dryness in vacuo. An aliquot was isolated by
semipreparative HPLC (C-18; 55% acetonitrile in H O, 4 mL/min;
3 4
, water, and saturated brine, dried over MgSO , and
2

5162 J. Am. Chem. Soc., Vol. 120, No. 21, 1998
Romesberg et al.
1
retention time, 20 min) and concentrated: H NMR (400 MHz) (CDCl
3
)
saturated NaHCO
3
, dried over Na
2 4
SO , and concentrated to dryness in
δ TMS 1.71-2.04 (m, 3H), 3.50 (d, 2H, J ) 7.3 Hz), 5.52-5.58 (m,
H), 7.37 (d, 2H, J ) 7.8 Hz), 8.13-8.16 (m, 2H); C NMR (100
vacuo. The intermediate alcohol was isolated by PTLC (silica; 2:1
hexanes/ethyl acetate), affording 21 mg of liquid product (83% yield):
1
3
2
MHz) (CDCl ) δ 12.89, 32.99, 123.67, 126.59, 126.97, 129.08, 131.41,
3
R
f
) 0.25. The alcohol was then carried on by reaction in acetonitrile
(2 mL) with 83 mg (0.249 mmol) of carbon tetrabromide and 65 mg
0.249 mmol) of triphenylphosphine for 3 h at room temperature. Silica
1
32.87.
(
trans-4-Phenyl-2-butene. To a stirred mixture of trans-1-bromo-
gel (4 g) was added and the slurry dried to a powder under reduced
pressure. The crude product was then eluted from silica gel with 2:1
hexanes/ethyl acetate and the eluent concentrated to dryness in vacuo.
1
-propene (500 mg, 4.13 mmol) in 10 mL of THF was added 239 mg
(0.207 mmol) of tetrakis(triphenylphosphine)palladium followed by 4.1
mL of a 1 M solution of benzylmagnesium chloride in ether, and the
resulting mixture was stirred at room temperature for 24 h. Saturated
The residue was purified by PTLC (silica; 4:1 hexanes/ethyl acetate),
1
affording 16 mg of product as a clear oil (49% yield): R
NMR (400 MHz) (CDCl
f
) 0.53; H
aqueous NH
separated. The organic layer was washed with saturated NaHCO
water, and saturated brine, dried over MgSO , and concentrated to
dryness in vacuo. The residue was purified by flash chromatography
silica; 50:1 hexanes/ethyl acetate), affording 314 mg of product as a
4
Cl was added followed by ether, and the layers were
) δ TMS 2.85 (1/2 ABq, 1H, J ) 13.7 Hz),
.98 (1/2 ABq, 1H, J ) 13.9 Hz), 4.02 (s, 1H), 7.33-7.38 (m, 2H),
.12-8.22 (m, 2H); 13C NMR (400 MHz) (CDCl
) δ 22.57, 33.71,
3.86, 50.01, 123.78, 129.32, 146.55, 148.75; IR (NaCl): 2950, 1600,
3
,
3
2
8
4
4
3
(
-
1
+
1
1518, 1346, 1199, 1110, 848 cm ; mass spectrum (EI) m/z 264 (M ),
clear liquid (58% yield): R
f
) 0.74; H NMR (400 MHz) (CDCl
3
) δ
H), 7.12-7.17 (m, 3H), 7.22-7.28 (m, 2H); 1 C NMR (100 MHz)
CDCl ) δ 17.85, 39.04, 125.84, 126.28, 128.31, 128.45, 130.04, 141.01;
+
2
62 (M ). Anal. Calcd for C10
7 5
H D BrNO
2
: C, 45.64; H, 4.60; N,
TMS 1.63-1.69 (m, 3H), 3.29 (d, 2H, J ) 6.5 Hz), 5.46-5.62 (m,
3
5.32. Found: C, 45.98; H, 4.63; N, 5.48.
2
(
3
4-(4-Acetamidophenyl)-2-butanone. To a stirred solution of 4-ac-
etamidoiodobenzene (1.0 g, 3.83 mmol) and 3-butene-2-ol (0.346 g,
-
1
IR (NaCl) 3027, 2916, 1494, 1452, 990, 744, 698 cm
.
4
.81 mmol) in 10 mL of acetonitrile was added 9 mg (0.038 mmol) of
trans-4-(4-Nitrophenyl)-2-butene (6b). To a stirred solution of
trans-4-phenyl-2-butene (162 mg, 1.23 mmol) in 4 mL of trifluoroacetic
anhydride at 0 °C was added 142 mg (0.623 mmol) of copper(II) nitrate.
The mixture was stirred at 0 °C for 30 min, then warmed to room
temperature with continued stirring for 2 h. The reaction mixture was
poured onto ice, and after the ice melted, the reaction mixture was
neutralized with 1 N NaOH followed by extraction with ether. The
palladium acetate and 0.121 g (1.19 mmol) of triethylamine, and the
mixture was heated to reflux for 24 h. The reaction was diluted with
ether and washed with saturated NH
water, and saturated brine, then dried over MgSO
dryness in vacuo. The residue was purified by flash chromatography
(silica; 4:1 ethyl acetate/hexanes) affording 0.42 g of product: R
0.35. White crystals were precipitated from ethyl acetate/hexanes, mp
109-111 °C: 1H NMR (400 MHz) (CDCl
) δ TMS 2.13 (s, 3H), 2.15
4
Cl (aq), saturated NaHCO
3
(aq),
4
and concentrated to
f
)
ether layer was washed with saturated NaHCO
brine, dried over MgSO , and concentrated to dryness in vacuo. An
aliquot was then purified by PTLC (silica; 50:1 hexanes/ethyl acetate)
only the tail of band at R ) 0.32 excised), affording 15 mg of pure
b as a liquid: H NMR (400 MHz) (CDCl ) δ TMS 1.70-1.72 (m,
H), 3.41 (br s, 2H), 5.55-5.58 (m, 2H), 7.33 (d, 2H, J ) 8.9 Hz),
3
, water, and saturated
4
3
(s, 3H), 2.73 (t, 2H, J ) 7.4 Hz), 2.85 (t, 2H, J ) 7.4 Hz), 7.12 (d, 2H,
J ) 8.3 Hz), 7.40 (d, 2H, J ) 8.4 Hz), 7.61 (br s, 1H); 13C NMR (100
(
f
1
6
3
8
3
3
MHz) (CDCl ) δ 24.41, 29.07, 30.07, 45.07, 120.17, 128.71, 136.02,
136.88, 168.40, 208.04; IR (NaCl) 3295, 1707, 1665, 1604, 1550, 1515,
1411, 1370, 1321, 824 cm . Anal. Calcd for C12H15NO : C, 70.22;
2
H, 7.37; N, 6.82. Found: C, 69.82; H, 7.40; N, 6.70.
-
1
-1
.14 (d, 2H, J ) 8.7; IR (NaCl) 2915, 1517, 1346 cm ; HRMS (EI),
+
calcd for C10
H
11NO
2
(M ) 177.0790, found 177.0786.
4
-(4-Nitrophenyl)-1-butene (7). To a stirred solution of trifluoro-
acetic anhydride (5 mL) at 0 °C containing 0.5 g (3.78 mmol) of
-phenyl-1-butene (Aldrich) was added 0.44 g (1.89 mmol) of copper-
II) nitrate. The resulting mixture was stirred at 0 °C for 30 min, and
4-(4-Acetamidophenyl)-2-butanol. To a stirred solution of (4-
acetamidophenyl)-2-butanone (150 mg, 0.731 mmol) in 15 mL of
methanol cooled to 0 °C was added 110 mg (2.92 mmol) of NaBH .
4
The resulting reaction was stirred at 0 °C for 30 min, then quenched
with 1 N HCl (aq) and concentrated under reduced pressure. The
residue was redissolved in dichloromethane, washed with saturated
4
(
after warming to room temperature, the reaction mixture was stirred
for an additional 2 h, at which point it was poured onto ice. The pH
was adjusted to pH 8 with 1 N NaOH (aq) and the mixture extracted
3 2 4
NaHCO , dried over Na SO , and concentrated to dryness in vacuo.
2
× 25 mL with ether. The ether extracts were combined, washed
with saturated NaHCO , water, and saturated brine, then dried over
MgSO and concentrated to dryness in vacuo. The crude product was
purified by flash chromatography (silica; 20:1 hexanes/ethyl acetate),
affording 0.54 g of product as an amber liquid (81% yield): R ) 0.81.
An aliquot was then further purified by PTLC (silica; 50:1 hexanes/
The crude product was purified by recrystallization from ethyl acetate/
hexanes, affording 127 mg of product as a white crystalline solid (84%
3
1
yield), mp 132-135 °C; H NMR (400 MHz) (DMSO-d
6
) δ HOD 1.02
4
(d, 3H, J ) 6.2 Hz), 1.49-1.58 (m, 2H), 1.98 (s, 3H), 2.42-2.59 (m,
2H), 3.50-3.56 (m, 1H), 4.40 (d, 1H, J ) 4.8 Hz), 7.05 (d, 2H, 8.4
Hz), 7.41 (d, 2H, J ) 8.4 Hz), 9.79 (s, 1H); 13C NMR (100 MHz)
f
1
ethyl acetate): R
MHz) (CDCl ) δ TMS 2.37-2.44 (m, 2H), 2.82 (t, 2H, J ) 8.0 Hz),
.98-5.06 (m, 2H), 5.78-5.86 (m, 1H), 7.32-7.36 (m, 2H), 8.13-
f
) 0.21 (only top of band excised); H NMR (400
6
(DMSO-d ) δ 23.26, 23.55, 30.62, 40.57, 64.83, 118.68, 127.96, 136.59,
136.63, 167.62; IR (NaCl) 3292, 2926, 1665, 1604, 1545, 1514, 1412,
3
-1
4
8
1
1
1371, 1321, 824 cm . Anal. Calcd for C12
N, 6.76. Found: C, 69.42; H, 8.34; N, 6.70.
-(4-Acetamidophenyl)-2-bromobutane (10). To a stirred solution
2
H17NO : C, 69.53; H, 8.26;
1
3
.16 (m, 2H); C NMR (100 MHz) (CDCl
23.47, 129.19, 131.94, 136.77, 146.26; IR (NaCl): 3079, 2936, 2858,
3
) δ 34.72, 35.06, 115.76,
4
-
1
642, 1606, 1519, 1347, 1110, 996, 916, 859, 744, 698 cm ; HRMS
of 4-(4-acetamidophenyl)-2-butanol (30 mg, 0.145 mmol) in 2 mL of
acetonitrile was added 96 mg (0.289 mmol) of carbon tetrabromide
followed by 76 mg (0.289 mmol) of triphenylphosphine. The resulting
mixture was stirred at room temperature for 3.5 h, silica gel (4 g) was
then added, and the slurry was dried to a powder under reduced pressure.
The crude product was then eluted from the silica gel with 2:1 ethyl
acetate/hexanes and the eluent concentrated to dryness in vacuo. The
+
(FAB), calcd for C10
H12NO
2
(MH ) 178.0868, found 178.0826.
4
5
-(4-Nitrophenyl)-2-butanone-1,1,1,3,3-d . To a stirred solution
of 4-(4-nitrophenyl)-2-butanone (40 mg, 0.207 mmol) dissolved in 3
mL of methyl alcohol-d (Aldrich) was added 10 µL of 40% NaOD in
D O and the resulting mixture stirred at room temperature for 8 h. The
2
reaction mixture was then concentrated under reduced pressure and
the residue purified by PTLC (silica; 2:1 hexanes/ethyl acetate),
residue was purified by PTLC (silica; 3:1 ethyl acetate/hexanes),
1
affording 25 mg of product as an amber oil (61% yield): R
f
) 0.40;
) δ TMS: 3.03 (s, 2H), 7.38 (d, 2H, J )
.6 Hz), 8.16 (d, 2H, J ) 8.7 Hz).
-Bromo-4-(4-nitrophenyl)butane-1,1,1,3,3-d
tion of 4-(4-nitrophenyl)-2-butanone (25 mg, 0.126 mmol) in 3 mL of
methanol-d was cooled to 0 °C, and 19 mg (0.504 mmol) of NaBH
affording 12 mg of product (32% yield): R
MHz) (CDCl
2.16 (s, 3H), 2.67-2.86 (m, 2H), 4.02-4.08 (m, 1H), 7.14 (d, 2H, J )
.4 Hz), 7.35 (br s, 1H), 7.41 (d, 2H, J ) 8.4 Hz); 13C NMR (100
MHz) (CDCl ) δ 24.53, 26.49, 33.33, 42.61, 50.75, 120.16, 129.01,
35.93, 136.95, 169.14; IR (NaCl): 3284, 3120, 2922, 1661, 1601,
f
) 0.57; H NMR (400
1
H NMR (400 MHz) (CDCl
3
3
) δ TMS 1.72 (d, 3H, J ) 6.6 Hz), 1.96-2.14 (m, 2H),
8
8
2
5
(8). A stirred solu-
3
1
1
4
-
1
514, 1409, 1320, 838, 763 cm
O-Methanesulfonyl-4-(4-nitrophenyl)-2-butanol (9). To a stirred
solution of alcohol 5 (40 mg, 0.205 mmol) in 10 mL of dichloromethane
.
was added. The resulting reaction was stirred at 0 °C for 30 min, then
was concentrated under reduced pressure. The residue was triturated
with dichloromethane, and the dichloromethane slurry was washed with

Antibody-Catalyzed Elimination Reaction
J. Am. Chem. Soc., Vol. 120, No. 21, 1998 5163
cooled to 0 °C was added 70 mg (0.615 mmol) of methanesulfonyl
chloride followed by 104 mg (1.02 mmol) of triethylamine, and the
resulting mixture was stirred at 0 °C for 30 min, then at room
temperature for 2 h. The reaction was washed with water, dried over
Kinetics and Product Analysis. UV kinetic measurements of the
fluoride elimination reaction were carried out as described previously.
2
High-pressure liquid chromatography (HPLC) kinetic measurements
were determined using a Rainin SD-200 binary HPLC system equipped
with a Rainin UV-1 UV detector and a Rainin C-18 Microsorb analytical
column (4.6 mm × 25 cm). Initial velocities were determined by
measuring integrated product peak differences as a function of time
(corrected for background) and comparing to authentic standard
calibration curves. Bromide elimination reactions were carried out
using 10 µM Fab as determined by OD280 nm (1.37 absorbance units
for 1.0 mg/mL of 43D4-3D12 and a molecular weight of 50 000) in
50 mM sodium phosphate monobasic buffer (adjusted to pH 8.0 with
2 N NaOH) containing 10% acetonitrile (37 °C) at substrate concentra-
tions ranging from 50 to 250 µM. Products were analyzed by reverse
phase HPLC (C-18, 55% acetonitrile in 50 mM sodium acetate, pH
5.0, 1 mL/min, 254 nm detection) over 30 min. Retention times for
the cis (6a), trans (6b), and terminal olefin (7) products were 21.5,
22.7, and 19.9 min, respectively. The retention time for the alcohol
product (5) was 4.7 min. Acetal hydrolyses were carried out using 10
µM Fab in 50 mM sodium acetate, 50 mM NaCl (pH 5.0), and 5%
acetonitrile (37 °C) at substrate concentrations ranging from 20 µM to
4 mM. Products were analyzed by reverse phase HPLC (C-18, 60%
acetonitrile in 50 mM sodium acetate, pH 5.0, 254-nm detection) over
15 min. Kinetic analyses of uncatalyzed reactions were carried out in
the absence of antibody, but under otherwise identical conditions.
Na
2 4
SO , and concentrated to dryness in vacuo, affording 54 mg of
1
product (96% yield): H NMR (400 MHz) (CDCl
3
) δ TMS 1.47 (d,
3
3
2
H, J ) 6.3 Hz), 1.65-2.11 (m, 2H), 2.79-2.91 (m, 2H), 3.04 (s,
H), 4.84-4.89 (m, 1H), 7.37 (d, 2H, J ) 8.8 Hz), 8.14-8.18 (m,
H).
4
-((Methoxymethoxy)methyl)nitrobenzene (11). To a stirred
solution of dichloromethane (10 mL) containing 100 mg (0.653 mmol)
of 4-nitrobenzyl alcohol was added 79 mg (0.980 mmol) of chlorom-
ethyl methyl ether followed by 253 mg (1.96 mmol) of N,N-
diisopropylethylamine, and the resulting mixture was stirred at room
temperature for 18 h. The reaction was diluted with dichloromethane,
washed with saturated NaHCO
to dryness in vacuo. The residue was then purified by preparative TLC
silica; 1:1 ethyl acetate/hexanes), affording 37 mg of product as a
3 2 4
, dried over Na SO , and concentrated
(
1
yellow oil (29% yield): R
TMS 3.41 (s, 3H), 4.69 (s, 2H), 4.74 (s, 2H), 7.52 (d, 2H, J ) 8.9 Hz),
8
9
1
f
) 0.4; H NMR (100 MHz) (CDCl
3
) δ
.20 (d, 2H, J ) 8.7 Hz); 13C NMR (400 MHz) (CDCl
6.12, 123.60, 127.81, 145.61, 147.35. IR (NaCl): 2942, 2889, 1723,
607, 1522, 1348, 1151, 1107, 1050, 847, 739 cm ; HRMS (FAB)
3
) δ 55.55, 67.94,
-
1
+
calcd for C
9
H12NO
4
(MH ) 198.0766, found 198.0741.
O-Tetrahydropyranyl-4-nitrobenzyl Alcohol (13). To a stirred
solution of 4-nitrobenzyl alcohol (500 mg, 3.26 mmol) in 30 mL of
dichloromethane was added 411 mg (4.89 mmol) of 3,4-dihydro-2H-
pyran followed by 31 mg (0.163 mmol) p-toluenesulfonic acid
monohydrate. The resulting mixture was stirred at room temperature
for 2 h, then diluted with dichloromethane, washed with saturated
Homology Model of Fab 43D4-3D12. The generation of a model
of the Fab 43D4-3D12 variable region is possible due to the existence
of conserved antibody framework regions, an experimental crystal-
lographic database of canonical loop structures whose conformations
13
depend on well-defined residues, and experimental constraints. An
NaHCO
affording 743 mg of product as a yellow liquid (96% yield): H NMR
400 MHz) (CDCl
3 2 4
, dried over Na SO , and concentrated to dryness in vacuo
initial model of Fab 43D4-3D12 was generated using the AbM
1
14
program. Briefly, coordinates corresponding to the most homologous
(
3
)
8
3
1
7
2
3
) δ TMS 1.55-1.93 (m, 6H), 3.54-3.59 (m, 1H),
.86-3.92 (m, 1H), 4.61 (1/2 ABq, 1H, J ) 13.5 Hz), 4.74 (t, 1H, J
framework regions (1baf and 1bbj, respectively), as well as for canonical
loops L2, H1, and H2 (L2, class 1; H1, class 1; H2, class 2), were
selected from an antibody data base. Side chains were then changed
to those corresponding to 43D4-3D12 using maximum overlap. f
Coordinates for the noncanonical loops were selected by combined
conformational and CR database searching. The hypervariable loops
were preminimized with several hundred iterations using the steepest-
3.5 Hz), 4.89 (1/2 ABq, 1H, J ) 13.5 Hz), 7.52-7.55 (m, 2H),
13
.19-8.22 (m, 2H); C NMR (400 MHz) (CDCl ) δ 19.18, 25.26,
3
0.36, 62.18, 67.52, 98.21, 123.46, 123.46, 127.66, 127.66, 146.07,
47.16; IR (NaCl) 2944, 2870, 1607, 1520, 1345, 1126, 1036, 973,
-
1
+
39 cm ; HRMS (EI) calcd C12
4
H16NO (MH ) 238.1079, found
38.1072.
15
descents algorithm within the program Discover. Substrates 2, 4, and
Antibody (Fab, 43D4-3D12) Expression, Purification, and Mu-
11 were docked manually, employing the following experimental
constraints. First, it was assumed that the p-nitrophenyl group would
be oriented into the binding pocket. During immunization, hapten 1
was conjugated to keyhole limpet hemocyanin (KLH) via a linker, and
this orientation for hapten binding is the only one allowing the antibody
to approach hapten. We assume that the substrate binds analogously.
tagenesis. Total RNA was isolated from the hybridoma cell line by
8
the method of Chomczynski and Sacchi. Total RNA was enriched
for messenger RNA encoding 43D4-3D12 variable regions by affinity
chromatography with oligo(dT) cellulose (Pharmacia). Constant region
3
′ primers were used to reverse transcribe cDNA. PCR amplification
with the 3′ constant region primer and the 5′ primers described by Huse
This is supported by several crystal structures of p-nitrophenyl
9
et al. yielded sufficient DNA for cloning and sequencing. After
16,17
complexes.
Second, the acidic protons were positioned proximate
κ
H
determination of the J and J regions used by 43D4-3D12, 3′ primers
to the catalytic residue. Several exploratory steepest-descents mini-
mizations of the hypervariable loops were started from various initial
conformations. The binding pocket of the lowest energy structure was
then minimized by a combination of steepest-descent and conjugate-
gradient methods to a root mean square derivative less than 0.001 kcal/
mol‚Å). Multiple substrate orientations were examined, including
those corresponding to syn and anti deprotonation, as well as those
leading to cis and trans products.
corresponding to the appropriate J regions and containing restriction
sites were synthesized and used in conjunction with the 5′ primers to
PCR amplify DNA corresponding to the variable regions of 43D4-
3
D12. The V
construct fuses the murine V
constant regions, respectively. Mutagenesis was performed by the
H
and V
L
genes were cloned into plasmid p4X.10 This
and V genes to human C 1 and C 1
H
L
H
κ
(
1
1
method of Kunkel. Sequence determination was performed by the
1
2
method of Sanger. The Fab fragments were isolated from E. coli
5F2 transformed with the appropriate plasmids, according to the
2
1
0
Results and Discussion
published protocol. Fab fragments were further purified by protein
G affinity chromatography (Sepharose CL-6B, Pierce). The Fab
fragments were concentrated and stored in either kinetics buffer or
Cloning and Production of Fab 43D4-3D12. The sequences
of the 43D4-3D12 VL and VH chains are shown in Figure 2.
VL possesses the highest homology to the Vκ21 group as
2
+
2+
phosphate-buffered saline (Ca , Mg free, deionized water).
(
(
8) Chomczynski, P. and Sacchi, N. Anal. Biochem.1987, 162, 156.
9) Huse, W. D.; Sastry, L.; Kang, S. A.; Alting-Mees, M.; Burton, D.
(13) Chothia, C.; Lesk, A. M. J. Mol. Biol. 1987, 196, 901.
(14) Oxford Molecular, 1994.
(15) Biosym Technologies, 1992.
(16) Patten, P. A.; Gray, N. S.; Yang, P. L.; Marks, C. B., Wedemayer,
G. L.; Boniface, J. J.; Stevens, R. C.; Schultz, P. G. Science 1996, 271,
1086.
(17) Hsieh-Wilson, L. C.; Schultz, P. G.; Stevens, R. C. Proc. Natl. Acad.
Sci. U.S.A. 1996, 93, 5363.
R.; Benkovic, S. J.; Lerner, R. A. Science 1989, 246, 1275.
10) Ulrich, H. D.; Patten, P. A.; Yang, P. L., Romesberg, F. E.; Schultz,
P. G. Proc. Natl. Acad. Sci. U.S.A. 1995, 88, 8784.
11) Kunkel, T. A.; Roberts, J. D.; Zakour, R. A. Methods in Enzy-
mol.1987, 154, 367.
12) Sanger, W.; Nicklen, S.; Coulson, A. R. Proc. Natl. Acad. Sci. U.S.A.
977, 74, 5463.
(
(
(
1

5164 J. Am. Chem. Soc., Vol. 120, No. 21, 1998
Romesberg et al.
Mutagenesis and Modeling Studies. Previous affinity
labeling experiments indicated that the likely active site general
base is either Glu46 or Glu50 of the heavy chain.3 To
unequivocally assign the catalytic residue, both glutamate
residues were mutated to glutamine by site-directed mutagenesis.
The mutants were then assayed for their ability to catalyze
H
fluoride elimination from substrate 2. The Glu46 Gln mutant
-
1
retained full catalytic activity, kcat ) 0.30 min and Km ) 214
H
µM at pH 6.5, while the Glu50 Gln mutant had no measurable
activity above the background reaction. We therefore assign
H
Glu50 as the catalytic carboxylate residue which acts as an
H
efficient general base catalyst. Glu50 was also mutated to a
histidine to probe the stereoelectronic requirements for proton
H
transfer. The Glu50 His mutant also possessed no catalytic
activity above background at pH 6.5, possibly arising from poor
alignment of the basic imidazole nitrogen with the CR protons.
Figure 2. Amino acid sequences of 43D4-3D12 light and heavy chain
variable regions (V and V ). Hypervariable regions are underlined.
In an attempt to gain insight into the structural basis for the
catalytic properties of 43D4-3D12 a computational model of
the Fab was generated. The model is based on the high
structural homology between antibody framework regions and
a crystallographic database of antibodies sharing canonical loop
L
H
delineated by Potter et al.18,19 and is joined in frame to Jκ2.
The nucleotide sequence of the 43D4-3D12 VH shows that the
2
0,21
antibody uses a germline gene from the VH1 (J558) family
1
3
structures with Fab 43D4-3D12. Additionally, experimental
constraints allow for the positioning of the substrate in the
antibody combining site. Specifically, the active site base
along with JH4. While the 43D4-3D12 VH chain appears to
contain a significant number of somatic mutations, it is
interesting to note that the catalytic Glu50H (see below) is
H
Glu50 is positioned so that it is in close proximity to the acidic
present in the majority of the germline genes and therefore
probably does not represent a somatic mutation.
CH group of substrate 2 and the nitroaryl ring is oriented into
the binding pocket in a manner consistent with the geometry
of the hapten-KLH conjugate used to elicit antibody 43D4-3D12.
The recombinant 43D4-3D12 was produced in E. coli as
chimeric Fab possessing murine variable regions fused to human
constant regions CH1 and Cκ1. The wild-type (wt) protein
expression levels were found to be low, relative to a variety of
other catalytic antibodies expressed similarly10 (approximately
The modeled Michaelis complex between substrate 2 and Fab
43D4-3D12 is shown in Figure 3. The model predicts that the
substrate interacts with residues in the active site of 43D4-3D12
via both hydrogen bonding and hydrophobic packing interac-
tions. Residues from L3, H1, H2, and H3 all participate in
forming a deep, occluded binding pocket for the substrate. At
the bottom of the pocket the p-nitro group of 2 is hydrogen
1
4
00 µg/L). The catalytic constants of the recombinant Fab
-
1
3D4-3D12, kcat ) 0.30 min and KM ) 214 µM at pH 6.5,
2
are indistinguishable from those of the fully murine antibody.
In the course of mutagenesis studies of 43D4-3D12 (see
below), several active site mutants were generated by site
directed mutagenesis, e.g., each residue of the CDR3 loop
H
bonded to the indole ring of Trp47 and the carboxamide group
L
L
H
of Asn89 . The side chains of Tyr96 and Cys95 pack against
the phenyl moiety of the substrate. Both enantiomers position
the fluoride leaving group to form a hydrogen bond to the
(residues 92-96) was mutagenized to alanine, histidine, and
lysine. Surprisingly, 11 mutants corresponding to either alanine,
histidine, or lysine at positions 92-94 and either histidine or
lysine (not alanine or phenylalanine) at position 96 resulted in
a greater than 1 order of magnitude increase in protein
expression levels relative to wt protein. Given that two of three
wt codons were present in the DNA encoding each mutant, low
expression of the recombinant wt protein is not likely a
consequence of codon usage. Additionally, control experiments
were performed by generating silent mutations in the same
residues of the wt antibody. In all cases low protein expression
levels were found. It is possible that the increased expression
levels of the mutant proteins result from the removal of
deleterious interactions involving the light chain CDR3 loop in
either the folded protein (light chain CDR3 residues supply
approximately 20% of the light chain contacts with the heavy
chain) or in some intermediate folded state.22
L
hydroxyl group of Tyr96 , which could facilitate loss of the
fluoride by stabilizing developing negative charge in the
transition state. Minimization of both complexed enantiomers
with substrate geometries corresponding to either a syn- or anti-
fluoride elimination reaction resulted in isoenergetic minima,
with all complexes maintaining the hydrogen bond between the
L
fluoride group and Tyr96 . The only stable conformation of
the incipient olefin was found to be trans, consistent with the
2
H
experimental result. Aromatic residues, predominately Trp33 ,
H
H
Tyr97 , and Tyr100c are predicted to limit access of water to
the binding pocket. With the exception of Asn94 , all of the
L
side chains in close proximity to the catalytic Glu50H are
hydrophobic, consistent with the elevated pKa of Glu50H
observed experimentally. The hydrophobic, tightly packed
substrate environment is also consistent with the unsuccessful
attempts to catalyze the exchange of deuterium for hydrogen
in p-nitrobenzyl acetone when bound to antibody 43D4-3D12.
(
18) Potter, M.; Newell, J. B.; Rudikoff, S.; Haber, E. Mol. Immunol.
982, 19, 1.
19) Heinrich, G.; Traunecker, A.; Tonegawa, S., J. Exp. Med. 1984,
59, 417.
1
(
The above model suggests that residues 92-96 of the light
1
H
chain CDR3 loop are proximal to Glu50 . In an effort to
(
(
(
20) Dildrop, R. Immunol. Today 1984, 5, 85.
21) Brodeur, P. H.; Riblet, R. Eur. J. Immunol. 1984, 14, 922.
22) (a) Dul, J. L.; Argon, Y. Proc. Natl. Acad. Sci. U.S.A. 1990, 87,
modulate the activity of 43D4-3D12, each of these residues was
mutagenized to alanine, histidine, and lysine, and the catalytic
activity of the mutant Fab was assayed using â-fluoroketone 2
as a substrate. The activity of the antibody was remarkably
invariant to substitutions at residues 92-94. All three of the
8
2
135. (b) Dul, J. L.; Burrone, O. R.; Argon, Y. J. Immunol. 1992, 127,
609. (c) Wu, G. E.; Hozumi, N.; Murialdo, H. Cell 33, 77. (d) Martin, T.
M.; Kowalczyk, C.; Stevens, S.; Wiens, G. D.; Stenzel-Poore, M. P.;
Rittenberg, M. B. J. Immunol. 1996, 157, 4341.

Antibody-Catalyzed Elimination Reaction
J. Am. Chem. Soc., Vol. 120, No. 21, 1998 5165
Figure 3. Modeled Michaelis complex of 43D4-3D12 and (A) substrate 2 in anti-trans configuration; (B) substrate 4 in syn-cis configuration; and
C) substrate 11. The antibody heavy and light chains are colored purple and orange, respectively. Critical antibody side chains are labeled with
(
their single-letter notation. The substrates are colored according to atom: carbon is green, nitrogen is blue, oxygen is red, fluoride is grey, and
bromide is brown.
alanine mutants, the histidine mutants at position 92 and 94, as
well as the lysine mutant at position 92, had approximately 75%
of wt specific activity at pH 6.5. Histidine and lysine mutations
at position 93 resulted in catalysts with approximately 30% of
wt antibody specific activity. The mutant with lysine at position
94 was approximately 20% more active than wt antibody
between pH 5.8 and 6.5. Thus, it is likely that these residues
do not significantly perturb the active site structure or reactivity
H
of Glu50 .
To evaluate the potential contributions of π-stacking and
hydrogen bond donation to the substrate leaving group, which
are both predicted by the modeled Michaelis complex, we
L
mutagenized Tyr96 to alanine, histidine, lysine, and phenyla-
L
lanine. At pH 6.5, the Tyr96 Phe mutant catalyzed the fluoride
elimination from substrate 2 at a rate within a factor of 2 of
Figure 4. Kinetics and deuterium isotope effects for cis and trans
eliminations of 4 and 8: (open diamonds) trans elimination of protio
substrate; (closed squares) trans elimination of deutero substrate; (closed
diamonds) cis elimination of protio substrate; (open squares) cis
elimination of deutero substrate. All reactions were carried out in 50
mM phosphate buffer and 10% acetonitrile, pH 8.0 (37 °C), for 24 h
in the presence of 10 µM Fab (43D4-3D12) and are background
corrected.
-
1
that for wild-type 43D4-3D12: kcat ) 0.17 min and KM )
20 µM. It appears that the predicted hydrogen bond does not
1
contribute significantly to catalysis, or alternatively, the predicted
L
interaction of Tyr96 with the substrate may be incorrect.
L
However, mutation of Tyr96 to Ala, His, or Lys led to the
complete loss of catalytic activity. Thus, the residue may make
important van der Waals packing interactions with the substrate,
but does not appear to stabilize the transition state by hydrogen
bonding interactions.
group (bromide) relative to the nitrophenyl ring is shifted in
comparison to that of the fluoride group in 2. Consequently, if
the nitrophenyl group orients the substrate in a similar way in
Characterization of the Antibody-Catalyzed Bromide
Elimination Reaction. It has been shown that an antibody can
catalyze a number of different reactions if the substrates share
a common recognition element (e.g., nitrophenyl ring) and the
reactions have overlapping mechanistic requirements (e.g., a
H
the antibody active site, Glu50 should be positioned to catalyze
the elimination reaction to afford the unconjugated olefin
products 6a,b.
4
,7
Initial rates for the antibody-catalyzed reaction were deter-
mined at concentrations of substrate 4 ranging from 50 to 250
µM in the presence of 10 µM Fab (50 mM phosphate buffer
and 10% acetonitrile, pH 8.0 for 24 h at 37 °C). Analysis of
the reactions rates (V) for formation of 6a and 6b over back-
ground as a function of substrate concentration afforded the
apparent second-order rate constants (kcat/KM) of 1.6 and 3.5
general base). Indeed, analysis of the catalytic activity of an
antibody using different reactions can provide important mecha-
nistic and structural insights. We therefore assayed the ability
of Fab 43D4-3D12 to catalyze the dehydrobromination reaction
of substrate 4 (Figure 1), which lacks acidic protons R to the
leaving group. Consequently, this elimination reaction is not
likely to involve an E1cb mechanism, but rather an E2 or E1
mechanism. There is also a competing solvolysis pathway in
the corresponding acetate-catalyzed reaction which yields, in
addition to the cis and trans internal olefins 6a,b and the
terminal olefin 7, the substitution product, alcohol 5, in a product
ratio of 6% (6a), 14% (6b), 5% (7), and 75% (5), respectively.
In the dehydrobromination reaction, the position of the leaving
-
1
-1
M
min for the cis and trans elimination reactions, respec-
tively (Figure 4). The limited solubility of 4 under the reaction
conditions precluded independent determination of kcat and Km.
These reaction rates can be compared to the corresponding
second-order rate constants for the acetate-catalyzed reactions
-
6
-6
-1
under the same conditions of 7.1 × 10 and 7.4 × 10
M

5166 J. Am. Chem. Soc., Vol. 120, No. 21, 1998
Romesberg et al.
Figure 5. Mechanistic probes of 43D4-3D12 (see text).
min- for the cis and trans elimination reactions, respectively.
1
This results in a ratio of second-order rate constants ((kcat/Km)/
Figure 6. Catalysis of the hydrolysis of acetal 11 by 43D4-3D12.
Reactions were carried out in 50 mM NaOAc, 50 mM NaCl, and 5%
acetonitrile (pH 5.0) at 37 °C for 24 h in the presence of 10 µM Fab
5
k-OAc) of 2.2 × 10 for the cis elimination reaction and 4.8 ×
5
1
0 for the trans elimination reaction, showing the antibody to
4
3D4-3D12.
be an efficient catalyst for these reactions. Furthermore,
antibody 43D4-3D12 selectively catalyzes the formation of
internal olefins 6a (18%) and 6b (72%) relative to terminal
olefin 7 (1%) and alcohol 5 (9%) (corrected for the uncatalyzed
reaction). In contrast, the alcohol resulting from the competing
substitution reaction is the major product of the uncatalyzed
reaction. Thus, the antibody-catalyzed reaction is again selective
for abstraction of the proton â- to the aryl ring, whose position
corresponds to that of the ammonium group in hapten 1.
Moreover the competing solvolysis reaction is not significantly
catalyzed by the antibody, indicating that the antibody selec-
tively stabilizes the transition state for the elimination reaction
relative to the substitution reaction, even under aqueous condi-
tions. This suggests that water is restricted from the 43D4-
H
Tyr100c sequester the substrate from water. This is consistent
with the antibody’s demonstrated ability to selectively catalyze
bromide elimination relative to solvolysis. The bromide leaving
group is not predicted to hydrogen bond to an active site residue.
Docking of both substrate enantiomers in geometries corre-
sponding to either the syn- or anti-bromide elimination reaction
followed by minimization led to distinct but nearly isoenergetic
minima. Additionally, both cis and trans conformations of the
incipient olefin were found to be roughly equal in energy,
consistent with the above described experiments.
Antibody-Catalyzed Acetal Hydrolysis Reaction. We also
investigated the ability of antibody 43D4-3D12 to catalyze the
hydrolysis of acetal 11 (Figure 1). Again, this substrate shares
a nitrophenyl recognition element with hapten 1. However, in
3
D12 active site, consistent with the model described above in
H
H
H
which the side chains of residues Trp33 , Tyr97 , and Tyr100c
sequester the substrate from solvent.
H
this case, it was expected that Glu50 might act as a general
acid to protonate the benzylic oxygen, facilitating expulsion of
the benzyl alcohol leaving group. The resulting carboxylate
anion could stabilize the positively-charged oxocarbonium ion
intermediate. Indeed, it has been shown previously that haptens
containing positively-charged ammonium ions can lead to the
generation of antibodies capable of catalyzing the hydrolysis
Isotope effect experiments were carried out with the penta-
deutero substrate 8. The ratios of the second-order rate constants
(kcat/Km)H/(kcat/Km)D were found to be 2.9 and 4.1 for the cis
and trans elimination reactions, respectively. These kinetic
isotope effects indicate a significant contribution of an E2-like
transition state to the elimination reaction. In the case of
substrate 2, delineating between an E2 or an E1cb mechanism
was not possible due to the acidity of the protons R to the
carbonyl group. With substrate 4, however, the pKa’s of the
homobenzylic protons are sufficiently high to make an E1cb
mechanism for the bromide elimination unlikely. As was the
case in the fluoride elimination reaction, no catalysis was
23
of acetal bonds.
At pH 5 acetal 11 has a slow but measurable background
rate in 50 mM NaOAc, 50 mM NaCl, and 5% acetonitrile buffer
at 37 °C. Initial reaction rates for the antibody-catalyzed
reaction were determined under the same conditions at concen-
trations of 11 ranging from 125 µM to 4 mM in the presence
of 10 µM Fab. Analysis of the reactions rates (V) of the
antibody-catalyzed formation of alcohol 12 (Figure 1) as a
function of substrate concentration afforded an apparent second-
H
observed for substrate 4 in the case of the Glu50 Gln mutant,
H
demonstrating the importance of the Glu50 carboxylate group
in the E2 elimination of bromide from substrate 4.
-2
-1
-1
order rate constant (kcat/KM) of 1 × 10
M
min . The
Mesylate 9 and bromide 10 (Figure 5) were also examined
as potential substrates for antibody 43D4-3D12. Both uncata-
lyzed elimination reactions proceeded with measurable rates in
aqueous buffer to yield olefin products. However, neither
reaction was accelerated over background in the presence of
antibody 43D4-3D12, demonstrating high substrate specificity
of the antibody toward the hapten programmed recognition
elements.
limited solubility of 11 under the reaction conditions precluded
determination of kcat and KM. This reaction rate can be
-6
-1
compared to the second-order rate constant of 1.1 × 10
M
-1
min for the reaction catalyzed by acetic acid. The ratio (kcat/
3
KM)/kHOAc is 8.8 × 10 , showing antibody 43D4-3D12 to be an
effective catalyst. Antibody 43D4-3D12 was also assayed with
THP acetal substrate 13 (Figure 5). The antibody did not
catalyze the hydrolysis, again demonstrating substrate specificity.
We also modeled substrate 4 into the combining site of
antibody 43D4-3D12 (Figure 3). As with the 43D4-3D12
substrate 2 complex, substrate 4 is predicted to bind in a deep
hydrophobic pocket with the p-nitro group hydrogen bonded
(
23) (a) Yu, J.; Hsieh, L.; Kochersperger, L.; Yonkovich, S.; Stephens,
J. C.; Gallop, M. A.; Schultz, P. G. Angew. Chem., Int. Ed. Engl. 1994, 33,
39. (b) Reymond, J. L.; Janda, K. D.; Lerner, R. A. Angew. Chem., Int.
3
Ed. Engl. 1991, 30, 1771. (c) Yu, J. and Schultz, P. G. J. Chem. Soc., Chem.
Commun., in press. (d) Janda, K. D.; Lo, L.-C.; Shin, M.-M.; Wang, R.;
Wong, C.-H.; Lerner, R. A. Science 1997, 275, 945. (e) Yu, J.; Choi, S.
Y.; Moon, K.-D.; Chung, H.-H.; Youn, H. J.; Jeong, S.; Park, H.; Schultz,
P. G. Proc. Natl. Acad. Sci. U.S.A., in press.
H
to the indole ring of Trp47 and the carboxamide group of
L
L
H
Asn89 . The side chains of Tyr96 and Cys95 pack against
H
H
the phenyl moiety of the substrate, and Trp33 , Tyr97 and

Antibody-Catalyzed Elimination Reaction
J. Am. Chem. Soc., Vol. 120, No. 21, 1998 5167
The modeled Michaelis complex between 43D4-3D12 and
substrate 11 is shown in Figure 3. Again the substrate is
predicted to bind in a deep hydrophobic pocket with the p-nitro
catalyze the elimination reaction of substrate 4 over the
competing substitution reaction under aqueous reaction condi-
tions again points to the degree to which the antibody active
site can control chemical reactivity.
L
group hydrogen bonded to the carboxamide group of Asn89
and the side chains of Tyr96 and Cys95H packing on the
L
H
substrate phenyl ring. The protonated Glu50 is predicted to
Acknowledgment. This work was supported by the Director,
Office of Energy Research, Office of Basic Energy Sciences,
Division of Material Sciences, and also by the Division of
Energy Biosciences of the U.S. Department of Energy under
Contract No. DE-AC03-76SF00098. M.E.F. acknowledges a
fellowship from the National Institutes of Health (Grant No.
1F32AI09093-01A1). P.G.S. is a Howard Hughes Medical
Institute Investigator and a W. M. Keck Foundation Investigator.
We gratefully acknowledge Jodie Chin for assistance with Figure
3.
hydrogen bond to the benzylic oxygen of substrate 11 consistent
with the catalytic mechanism described above. The δ-oxygen
L
is predicted to accept a hydrogen bond from Asn94 which is
the only other hydrophilic residue positioned in the 43D4-3D12
combining site.
Conclusion. These studies underscore the ability to introduce
a catalytic residue into an antibody combining site that can
function as a general acid or a general base by using an
appropriately designed immunogen. In the case of antibody
H
4
3D4-3D12, the side chain of Glu50 catalyzes both elimination
and acetal hydrolysis reactions. The ability of the antibody to
JA9738992