Nonspecific medium effects versus specific group positioning in the antibody and albumin catalysis of the base-promoted ring-opening reactions of benzisoxazoles

Article abstract of DOI:10.1021/ja0490727

The mechanisms by which solvents, antibodies, and albumins influence the rates of base-catalyzed reactions of benzisoxazoles have been explored theoretically. New experimental data on substituent effects and rates of reactions in several solvents, in an antibody, and in an albumin are reported. Quantum mechanical calculations were carried out for the reactions in water and acetonitrile, and docking of the transition state into a homology model of antibody 34E4 and an X-ray structure of human serum albumin was accomplished. A microenvironment made up of catalytic polar groups (glutamate in antibody 34E4 and lysine in human serum albumin) surrounded by relatively nonpolar groups is present in both catalytic proteins.

Full text of DOI:10.1021/ja0490727

Published on Web 06/15/2004
Nonspecific Medium Effects versus Specific Group
Positioning in the Antibody and Albumin Catalysis of the
Base-Promoted Ring-Opening Reactions of Benzisoxazoles
Yunfeng Hu,† K. N. Houk,*,† Kazuya Kikuchi,‡ Kinya Hotta,‡ and Donald Hilvert*,‡
Contribution from the Department of Chemistry and Biochemistry, UniVersity of California,
Los Angeles, Los Angeles, California 90095-1569 and Department of Chemistry,
Scripps Research Institute, La Jolla, California 92037, and Laboratorium fu¨r Organische
Chemie, Swiss Federal Institute of Technology, ETH-Ho¨nggerberg,
CH-8093, Zu¨rich, Switzerland
Received February 18, 2004; E-mail: houk@chem.ucla.edu; hilvert@org.chem.ethz.ch
Abstract: The mechanisms by which solvents, antibodies, and albumins influence the rates of base-
catalyzed reactions of benzisoxazoles have been explored theoretically. New experimental data on
substituent effects and rates of reactions in several solvents, in an antibody, and in an albumin are reported.
Quantum mechanical calculations were carried out for the reactions in water and acetonitrile, and docking
of the transition state into a homology model of antibody 34E4 and an X-ray structure of human serum
albumin was accomplished. A microenvironment made up of catalytic polar groups (glutamate in antibody
34E4 and lysine in human serum albumin) surrounded by relatively nonpolar groups is present in both
catalytic proteins.
Introduction
The conversions of benzisoxazoles to cyanophenoxides with
bases and various catalysts, often referred as the “Kemp
elimination,”1a-b have been used as a probe of medium effects
on the rates of reactions (Figure 1). The rates of these reactions
are very sensitive to solvent polarity. Kemp and co-workers’
classic studies of the effect of substituents and solvents
established that the acetate-promoted reaction of 5-nitro-
benzisoxazole is 108 times faster in acetonitrile than water.1c-d
This reaction has subsequently been shown to be catalyzed by
catalytic antibodies (kcat/kuncat ) 106),2 serum albumins (103),3-5
orphan antibodies (103),6 a polyamine organic host (103),7
surfactant vesicles (850),8 micelles (400),8 polyethyleneimine
“synzymes” (105 per site),9 and even coal (200).10 Because
Figure 1. Base-catalyzed decomposition of benzisoxazoles (the “Kemp
elimination”) and the hapten used to elicit antibody 34E4.
† University of California, Los Angeles.
different authors use different definitions for kcat and kuncat, those
accelerations are not always directly comparable. For example,
‡ Scripps Rsearch Institute and Swiss Federal Institute of Technology.
(1) (a) Casey, M. L.; Kemp, D. S.; Paul, K. G.; Cox, D. D. J. Org. Chem.
1973, 38, 2294-2301. (b) Kemp, D. S.; Casey, M. L. J. Am. Chem. Soc.
1973, 95, 6670-6680. (c) Kemp, D. S.; Paul, K. J. Am. Chem. Soc. 1975,
97, 7305-7312. (d) Kemp, D. S.; Cox, D. D.; Paul, K. J. Am. Chem. Soc.
1975, 97, 7312-7318.
k
uncat can be considered as a pseudo-first-order rate constant of
a water catalyzed reaction, or a second-order rate constant of
an acetate or amine base-promoted reaction in water, whereas
(2) Thorn, S. N.; Daniels, R. G.; Auditor, M.-T. M.; Hilvert, D. Nature 1995,
373, 228-230.
(3) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S. Nature 1996, 383, 60-63.
(4) Kikuchi, K.; Thorn, S. N.; Hilvert, D J. Am. Chem. Soc. 1996, 118, 8184-
8185.
kcat can be either a second-order rate constant in base-promoted
reactions in different solvents or a first-order rate constant when
saturated protein catalysts are involved. The various rate
constants in different catalytic systems and the kuncat used to
compute them are listed in Table 1. The magnitude of the rate
acceleration clearly depends on the pKa of the base responsible
for deprotonation and hence on the pH of the reaction medium.
Catalysts such as antibody 34E4 and the polyethyleneimine
“synzymes” exhibit high rate accelerations at pH values near
their pKa (ca. 6), whereas catalysts such as BSA with higher
pKa’s (ca. 10) must contend with the high background rate at
(5) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S.; Kikuchi, K.; Hilvert, D. J. Am.
Chem. Soc. 2000, 122, 1022-1029.
(6) Genre-Grandpierre, A.; Tellier, C.; Loirat, M. J.; Blanchard, D.; Hodgson,
D. R. W.; Hollfelder, F.; Kirby, A. J. Bioorg. Med. Chem. Lett. 1997, 7,
2497-2502.
(7) Kennan, A. J.; Whitlock, H. W. J. Am. Chem. Soc. 1996, 118, 3027-
3028.
(8) Perez-Juste, J.; Hollfelder, F.; Kirby, A. J.; Engberts, J. B. F. N. Org. Lett.
2000, 2, 127-130.
(9) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S. J. Org. Chem. 2001, 66, 5866-
5874.
(10) Shulman, H.; Keinan, E. Org. Lett. 2000, 2, 3747-3750.
9
10.1021/ja0490727 CCC: $27.50 © 2004 American Chemical Society
J. AM. CHEM. SOC. 2004, 126, 8197-8205
8197

A R T I C L E S
Hu et al.
Table 1. Rate Constants for the Uncatalyzed and Catalyzed
Decarboxylations of 5-Nitrobenzisoxazoles in Different Catalytic
Systems
relative importance to catalysis of polarity and positioning of
functional groups in the binding site.
Whether or not proteins can accelerate reactions merely by
altering the microenvironment around a substrate and transition
state has been the subject of an ongoing debate. Dewar attributed
catalysis of chymotrypsin and carboxypeptidase A to the ability
of these enzymes to eliminate solvent from the binding pocket
and to provide a nonpolar environment for the reaction.11 He
also emphasized the importance of a nonpolar environment for
acceleration of various reaction types.12 On the other hand,
Warshel has used thermodynamic arguments to show that
desolvation effects alone are unlikely to be the origin of enzyme
activity.13 Although a polar molecule is destabilized by being
bound into a nonpolar environment, a free energy penalty must
be paid to remove the ionic molecule from water. Warshel
proposes that electrostatic interactions between the oriented
dipoles of the preorganized protein binding pocket and the
substrate generally cause catalysis.13 In general, Warshel
embraces the effectively high polarity of protein binding sites
and attributes catalysis to the ideal electrostatic complementarity
of protein binding site and transition state.
catalyst
kbackground (s-1 j
)
kcat (s-1
)
34E4a
34E4b
BSAc
BSAd
3.1 × 10-5
0.66
0.30
2.5
0.017
0.035
0.00063
0.014
0.045
2.2 × 10-7
1.6 × 10-3
3.1 × 10-5
orphan antibody 4B2e
organic hostf
coalg
>3.1 × 10-5
1.1 × 10-7
6.8 × 10-5
synzymeh
8.0 × 10-8
Bu4NOAci
Et3Ni
1.4 × 10-4 M-1
8.2 × 10-1 M-1
a 20 °C, 40 mM phosphate buffer containing 100 mM NaCl, pH 7.4, ref
2. b 20 °C, 40 mM sodium acetate buffer containing 100 mM NaCl, pH
6.0 (the pKa of the active site carboxylic acid), ref 2 and this work. c 20
°C, 40 mM sodium carbonate buffer containing 100mM NaCl, pH 10.2
(the pKa of the active site amine), ref 4 and this work. d 20 °C, 40 mM
phosphate buffer containing 100 mM NaCl, pH 7.4, Reference 4. e 30 °C,
1% CH3CN, 40 mM phosphate buffer containing 100 mM NaCl, pH 7.1,
ref 6. The value of kbackground originally reported in ref 6 is 1.9 × 10-6 s-1
,
which is inconsistently low and is presumably a misprint. f Room temper-
ature, CDCl3, ref 7. g 4 °C, 50 mM phosphate buffer containing 2%
acetonitrile, pH 7.4, ref 10. h 25 °C, 70 mM BisTris buffer, pH 5.9, ref 9.
j
i 25° C, H2O, ref 1d. kbackground is obtained by extrapolation to zero buffer
Although some investigators have advocated pure medium
effects (“The medium is the message” in McLuhan’s words),3
others have highlighted contributions of structural effects and
specific interactions to catalysis. For example, Bruice empha-
sizes the ability of a protein to assemble substrates specifically
into a geometry suitable for reaction (NACs, near attack
conformations) regardless of the polarity of the microenviron-
ment.14
Throughout these discussions, confusion has occasionally
arisen by the use of different definitions of “medium.” The
dictionary definition emphasizes more or less homogeneous
surroundings: “Any intervening substance through which a force
acts on objects at a distance or through which impressions are
conveyed to the senses: applied, e.g., to the air, the ether, or
any substance considered with regard to its properties as a
vehicle of light or sound.”15 This definition of “medium” implies
a homogeneous environment. However, the word is often used
to indicate specific characteristics of surroundings even when
they are heterogeneous. “Microenvironment” is used here as a
term to describe the active site of a protein catalyst, especially
to convey the heterogeneity that is always a characteristic of a
protein interior.
Kinetic Studies. Kemp explored the kinetics of the reactions
of substituted benzisoxazoles (BI) with tertiary amines in water.1
The kinetics follow Brønsted linear free energy relationships
(LFERs) over a broad range of reactivity. The slope, â, of a
plot of log k vs pKa of the leaving group, pKlg, provides a
measurement of the relationship between the reaction rate and
the reaction exothermicity and yields information about the
structure and solvation of the transition state (TS).16 We have
measured the rates and âlg values for these eliminations in
several solvents and with protein catalysts. These are compared
to water as a standard. Table 2 lists the benzisoxazoles studied,
concentration.
high pH and are less efficient even though the absolute value
of kcat in the pH-independent range is higher.
Although many different interpretations have been offered
for the catalysis by these diverse materials, the binding of
substrate into a relatively nonpolar environment and particularly
the desolvation of acetate have usually been thought to be
responsible for the rate acceleration. By contrast, the amine-
catalyzed reaction is relatively insensitive to solvent polarity,
presumably because the uncharged base is not dramatically
stabilized by hydrogen bonding. Thus, the reactions involving
amine bases, such as polyamines, serum albumins, and synzymes
are not easy to explain by medium polarity effects.
Vigorous discussion has broken out about how two types of
protein catalysts, (1) tailored catalytic antibodies such as 34E4,2
and (2) “off-the-shelf” proteins such as serum albumins,3,4
accelerate this transformation. The two proteins are likely to
exploit carboxylate and amine groups as catalytic bases,
respectively. Hilvert and Kirby independently determined that
serum albumins catalyze the Kemp ring-opening with kcat similar
to that found with catalytic antibody 34E4.3,4 Hilvert identified
Lys222 as the catalytic group in bovine serum albumin
subdomain II A. There is also a set of positively charged residues
that stabilize the phenolate product formed in the reaction.4
However, Kirby proposed that the low polarity of the interiors
of the albumin and 34E4 is the factor responsible for most of
the rate acceleration in both cases.3
We have studied how these different proteins accelerate this
reaction. Kinetic measurements on substituted compounds have
been performed to show how specific interactions with nonpolar
residues and nonspecific medium effects influence the stability
and structure of the transition state. Quantum mechanical
calculations of acetate and amine catalyzed reactions in the gas
phase and in nonpolar and aqueous solutions have been used
to explore the origins of acceleration by nonpolar solvents. The
docking and binding of reactants and transition states in the
antibody and serum albumin have provided detailed pictures
of the origins of antibody and protein catalysis, specifically
(11) Dewar, M. J. S.; Storch, D. M. Proc. Natl. Acad. Sci. U.S.A. 1985, 82,
2225-2229.
(12) Dewar, M. J. S.; Healy, E. Organometallics 1982, 1, 1705-1708.
(13) Warshel, A. J. Biol. Chem. 1998, 273, 27035-29038.
(14) Bruice, T. C.; Lightstone, F. C. Acc. Chem. Res. 1999, 32, 127-136.
(15) Oxford English Dictionary, Second Edition, 1989.
(16) Hammond, G. S. J. Am. Chem. Soc. 1955, 77, 334-338; Leffler, J. E.
Science 1953, 117, 340-341.
9
8198 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Ring-Opening Reactions of Benzisoxazoles
A R T I C L E S
Table 2. Kinetic Parameters for the Decomposition of Substituted Benzisoxazoles
pK
phenol
kAcO-,H O
kAcO-,CH CN
kBuNH ,H O
kBuNH ,CH CN
(kcat/Km)34E4
(M-1 s-1
(kcat/Km)BSA
(M-1 s-1
)
a
2
3
2
2
2
3
benzisoxazole
(M-1 s-1
)
(M-1 s-1
)
(M-1 s-1
)
(M-1 s-1
)
)
5,7-(NO2)2
5,6-(NO2)2
5-NO2, 6-Cl
5-NO2
5-CN
6-NO2
5,6-Cl2
6-Cl
5-Cl
5-F
unsubstituted
0.6
2.5
3.6
4.1
4.7
5.2
5.6
6.1
6.4
6.8
6.9
2.76 × 10-3
1.88 × 10-4
2.73 × 10-5
1.33 × 10-5
4.52 × 10-6
2.47 × 10-6
1.52 × 10-6
759 000
57 800
4510
650 000
3840
22 600
5450
4530
457
147
23.9
1.81
1.52
0.116
0.949
0.261
0.0603
0.0247
0.0205
1000
0.0326
0.0218
0.0112
0.00854
144
0.177
9.64
0.635
3.3
104
1.78
19.4
0.255
0.548
0.0218
1.73
0.735
0.659
0.155
a Experimental conditions: The second-order rate constants for the acetate and butylamine dependent reactions were determined in water or acetonitrile
at 20 °C under pseudo-first order conditions as previously described. The reactions with 34E4 and BSA were performed in 40 mM phosphate, 100 mM
NaCl, pH 7.4 and 20 °C.
Table 3. Relative Second Order Rate Constants and Brønsted
Coefficients for the Decomposition of Substituted Benzisoxazoles
no. of
points
base/catalyst, medium
acetate, H2O
k
âlg
r2
a
rel
1
-0.67 ( 0.01 0.998
8
8
5
5
8
acetate, CH3CN
butylamine, H2O
butylamine, CH3CN
7.5 × 107 -0.95 ( 0.05 0.983
2.5 × 103 -0.53 ( 0.07 0.946
2.0 × 104 -0.86 ( 0.09 0.967
1.8 × 106 -0.82 ( 0.07 0.946 10
antibody 34E4, H2O (pH 7.5) 4.1 × 108 -1.48 ( 0.13b 0.956
BSA, H2O (pH 7.5)
a The second-order rate constant (kbase or kcat/Km) for the 5-NO2 derivative.
b Excluding 6-nitro, 5,6-dinitro, and 5,7-dinitrobenzisoxazoles, which deviate
systematically from the correlation due to comparatively large Km values
(see Figure 2).
along with the pKa values of the leaving group of the
corresponding phenols and the rate constants measured in water
and acetonitrile (MeCN) with different bases and catalysts. Table
3 lists the âlg values derived from the plots shown in Figure 2.
Excellent LFERs are found for the eliminations promoted by
butylamine or acetate as base, in both water and acetonitrile
solvents, and with the catalytic antibody 34E4 and bovine serum
albumin (BSA) as catalysts.
The âlg values for these correlations vary from -0.53 to
-1.48. In each case, the reaction rate increases with decreasing
salicylonitrile pKa, indicating substantial transfer of negative
charge to the ether oxygen of the benzisoxazole in the transition
state. The reactions involving acetate and butylamine in water
have the smallest âlg values; they are least sensitive to the
stability of the leaving group. The similarity of the values
suggests that the structures and strengths of the base have little
influence on the transition state structure and its solvation in
water. The change to acetonitrile as solvent causes a 108
acceleration with acetate as base, but only a 10-fold effect with
the amine base. Despite dramatically different effects on the
rates with different bases, the Brønsted âlg values are -0.9 to
-1.0 for both acetate and butylamine in acetonitrile. The values
are close to -1, so that the effect of leaving group stability is
strongly felt in the transition state. However, the aqueous pKa
values are used for this correlation, while one would expect a
larger spread of pKa values for a related nonpolar solvent. In
the lower dielectric environment, there is greater sensitivity to
charge development at oxygen.
Figure 2. (A) Brønsted plot for the aqueous reaction of substituted
benzisoxazoles with BuNH2 (O) or with BSA (b). The apparent second-
order rate constant (kcat/Km)BSA obtained at pH 7.4 was extrapolated to its
maximum value using the known pH-dependence of the BSA-catalyzed
reaction. (B) Brønsted plot for the aqueous reactions of BI with acetate
(0) or 34E4 (9). The reactions of 34E4 with 5,7-dinitro-, 5,6-dinitro-, and
6-nitrobenzisoxazole ([) deviate systematically from the correlation relating
the other eight substrates.
value correlating these reaction rates is similar to that obtained
with butylamine in acetonitrile. This suggests that the protein
binding pocket that contains Lys222, the likely catalytic base,9,17
presents a microenvironment to the developing phenoxide that
is similar to that of acetonitrile.
Antibody 34E4 likewise accepts a wide range of substituted
benzisoxazoles as substrates, but as shown in Figure 2, the
reaction rate is unusually sensitive to leaving group effects. Eight
of the eleven substrates tested are correlated with âlg ) -1.5,
which is much larger than any of the other Brønsted coefficients.
In this environment, the elimination rates respond more to
substituent effects than aqueous pK values do. This mostly likely
These results can be compared and contrasted with those
obtained with the protein catalysts BSA and antibody 34E4.
BSA accepts a wide variety of mono- and di-substituted
benzisoxazoles as substrates (Table 2), and the Brønsted âlg
(17) Carter, D. C.; Ho, J. X. AdV. Prot. Chem. 1994, 45, 153-203.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8199

A R T I C L E S
Hu et al.
Figure 3. Computed transition states for reaction of acetate and butylamine
with 4-nitrobenzisoxazole. Relevant bond lengths and charges are shown.
arises from the enhanced basicity of the phenoxide leaving group
in a relatively nonpolar environment, which leads to a larger
spread in phenoxide pK values than in water or acetonitrile and
a â value greater than -1.
Figure 4. Free energy (∆G) diagrams for reactions of acetate with
4-nitrobenzisoxazole in gas phase, acetonitrile, water, and antibody 34E4.
Numbers in bold face are experimental values and those in italics are the
averaged calculated values based on PCM and CPCM models.
The Brønsted plot for kcat similarly yields âlg ) -1.5 (r2 )
0.97) as a consequence of the similar Km values (100-200 µM)
exhibited by these substrates. Benzisoxazoles containing a nitro
substituent at the 6- or 7-position deviate negatively from the
correlation in 34E4 shown in Figure 2 and appear to lie along
a line that is roughly parallel to that through the data for the
other substrates. These substrates have substantially larger Km
values than the others, suggesting that they adopt a different
binding mode at the active site. However, the large kinetic
isotope effect (V/KD ) 5.7) measured for the most reactive
compound, 5,7-dinitrobenzisoxazole (kcat ) 660 s-1, Km ) 1.0
mM), shows that CH bond-breaking is still rate limiting for these
substrates. The large â value for 34E4 indicates that the
developing negative charge is relatively destabilized by the
environment so that the substituent effects are very large, and
reflect the very large role of substituent, rather than micro-
environment, on charge stabilization. The results suggest that
the local environment of 34E4 around the leaving group is
effectively less polar than acetonitrile.
Therefore, the E2 type geometry of the Kemp ring-opening does
not change significantly with different bases.
The calculated activation barrier in the gas phase is -7.8
kcal/mol for the acetate catalyzed reaction, because there is an
ion-molecule complex between the ionic reactant, acetate, and
benzisoxazole, which is lower in energy than the reactants and
the transition state for elimination. The conversion of this
complex to the transition state has a normal positive activation
barrier in solutions.
Solvation Energy Calculations. The calculated solvation free
energies of acetate are all within 3 kcal/mol of the experimental
aqueous solvation free energy of acetate (-77 kcal/mol).23 The
free energies of reactants and transition states in solvents are
sums of the single point solvation free energies based on the
(18) Gaussian 98, Revision A.9, Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.;
Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Zakrzewski, V. G.;
Montgomery, J. A.; Stratmann, Jr., R. E.; Burant, J. C.; Dapprich, S.;
Millam, J. M.; Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.;
Tomasi, J.; Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.;
Adamo, C.; Clifford, S.; Ochterski, J.; Petersson, G. A.; Ayala, P. Y.; Cui,
Q.; Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J. B.; Cioslowski, J.; Ortiz, J. V.; Baboul, A. G.; Stefanov, B.
B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin,
R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara,
A.; Challacombe, M.; Gill, P. M. W.; Johnson, B.; Chen, W.; Wong, M.
W.; Andres, J. L.; Gonzalez, C.; Head-Gordon, M.; Replogle, E. S.; Pople,
J. A. Gaussian, Inc., Pittsburgh, PA, 1998.
Quantum Mechanical Studies of the Base-Promoted Benz-
isoxazole Ring-Opening. Quantum mechanical calculations18
and solvation energy estimations19-22 were performed as
described in the Experimental Section. The optimized geometries
of the two transition states with acetate and butylamine as bases
in the gas phase are shown in Figure 3. The proton is more
completely transferred in the amine reaction, but the Cs(H)s
O and Cs(H)sN distances are similar, at 2.62 and 2.71 Å,
respectively. The lengths of the breaking NsO bonds in the
five-membered ring are 1.81 and 1.99 Å in the two transition
states, respectively. The charges on nitrogen (-0.33) and oxygen
(-0.44) of the cleaving benzisoxazole ring are almost identical.
(19) Barone, V.; Cossi, M.; Tomasi, J. J. Comput. Chem. 1998, 19, 404-417.
(20) Miertus, S.; Scrocco, E.; Tomasi, J. Chem. Phys. 1981, 55, 117-129.
(21) JAGUAR (Schro¨dinger, Inc., Portland, OR) Version 4.0, 2000.
(22) “AMSOL 6.5”, Hawkins, G. D.; Giesen, D. J.; Lynch, G. C.; Chambers,
C. C.; Rossi, I.; Storer, J. W.; Rinaldi, D.; Liotard, D. A.; Cramer, C. J.;
Truhlar, D. G.
(23) Pearson, R. G. J. Am. Chem. Soc. 1986, 108, 6109-6114.
9
8200 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Ring-Opening Reactions of Benzisoxazoles
A R T I C L E S
Figure 5. Free energy (∆G) diagrams for reactions of butylamine with
4-nitrobenzisoxazole in the gas phase, acetonitrile, water and BSA. Numbers
in bold face are experimental values and those in italics are the averaged
calculated values based on PCM and CPCM models.
Figure 6. Docked binding mode of TS minus AcO- in the 34E4 binding
site. (a) Side view (b) Top view.
gas-phase geometries, and the relevant free energies in the gas
phase. All of the methods overestimate the activation barriers
in both water and MeCN. The CPCM model shows a change
in the activation energy (7 kcal/mol) from water to acetonitrile
which is closes to the experimental value (11 kcal/mol).
Figure 4 summarizes the energetics of acetate- and 34E4-
promoted eliminations. CPCM calculations predict that acetate
and benzisoxazole are stabilized by a total of 63.6 kcal/mol in
acetonitrile, while the transition state 1 is solvated by only 33.6
kcal/mol. Because the reactants are stabilized much more than
the transition state, the activation barrier changes from -7.8
kcal/mol in the gas phase to 22.2 kcal/mol in acetonitrile. The
reactants and transition state are stabilized further by 15.8 and
9.4 kcal/mol, respectively, when changing the solvent from
acetonitrile to water. The new activation barrier is calculated
to be 28.6 kcal/mol, which is 6.4 kcal/mol higher than in
acetonitrile, experiment shows an even larger activation barrier
increase of 10.7 kcal/mol. Larger basis sets are likely to improve
agreement with experiment. For the acetate-promoted process,
the gas phase activation barrier (and consequently all the others)
is decreased by 2 kcal/mol.
For the amine-catalyzed reaction, summarized in Figure 5,
the activation barrier in the gas phase is 31.1 kcal/mol due to
the difficulty of charge separation in the gas phase. The reactants
are barely solvated by acetonitrile (0.7 kcal/mol), while the
transition state is stabilized by 12.3 kcal/mol due to its high
polarity. This leads to a smaller activation barrier (19.5 kcal/
mol) in acetonitrile than in the gas phase. The additional changes
of solvation free energies upon transfer from acetonitrile to water
for reactants and transition state 2 are 8.9 and 6.7 kcal/mol,
respectively. The resulting activation barrier in water (21.7 kcal/
mol) is slightly higher than in acetonitrile, because the reactants
are stabilized more than transition state. These calculations show
good agreement with experiment for both acetonitrile and water.
Acetate has an experimental pKa value of 4.8 in water and
22.3 in acetonitrile,24 and this large change is reflected in the
large rate acceleration of the elimination in acetonitrile.
However, experiment and calculation show that there is only a
2∼3 kcal/mol barrier change for the amine-catalyzed reaction.
The small activation barrier change in the amine-catalyzed
reaction is related to the smaller pK fluctuation from 10.8 in
water to 18.3 in acetonitrile.23
Transfer of Substrates and Transition States from Water
to Protein. The binding energies for reactants, estimated from
measured Km values, are 5.2 and 4.1 kcal/mol for 34E4 and
BSA, respectively. The calculated free energy corresponding
to kcat/Km is 12.1 kcal/mol for the reaction catalyzed by 34E4
and 12.0 kcal/mol for BSA. These results are also shown in
Figures 4 and 5. According to the thermodynamic cycle, 34E4
and BSA stabilize the transition states by 11.7 and 7.1 kcal/
mol relative to water, respectively. In acetate- and amine-
catalyzed reactions, both the reactants and transition states are
stabilized upon proceeding along the series from vacuum to
acetonitrile, water, and protein. The protein catalysis occurs by
lowering the free energy of the transition state in the protein
complex more than the substrate in the protein complex. A
similar trend was also found in our study of another solvent
sensitive reaction, the antibody-catalyzed Kemp decarboxyl-
ation.25
Docking Studies with 34E4. To explore in detail the
mechanism of antibody catalysis, the substrate and transition
state were docked into the binding sites of 34E4. In the absence
of a crystal structure for 34E4, we used a homology-based model
(24) Kolthoff, I. M.; Chantoon, M. K.; Bhowmik, S. J. Am. Chem. Soc. 1968,
90, 23-28.
(25) Ujaque, G.; Tantillo, D.; Hu, Y.; Houk, K.; Hotta, K.; Hilvert, D. J. Comput.
Chem. 2003, 24, 98-110.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8201

A R T I C L E S
Hu et al.
Figure 7. 34E4 surface including residues within 5 Å (top) and 8 Å
(bottom) of the docked transition state.
Figure 8. Human serum albumin. (a) Binding pockets and cavities
calculated by CAST (b) The docked transition state and the myristic acid
inhibitor.
based on the sequence of 34E4 obtained here (see Experimental
Section). Figure 6 shows the best transition state docking mode
for the acetate-catalyzed reaction in 34E4 in both side and top
views. It has an estimated free energy of binding of -7 kcal/
mol, according to the AUTODOCK free energy scoring func-
tion. In the vicinity of the benzisoxazole moiety, there are mostly
hydrophobic residues. The nitro group is buried in the bottom
of the 34E4 binding pocket, and forms two hydrogen bonds
with Thr L34. The tight steric fit of the substrate within the
pocket also suggests why substituents at the 6-position influence
substrate binding (Km) to a considerable extent in kinetic
experiments (Table 2). Near the entrance to the binding site,
the benzisoxazole transition moiety is surrounded by Tyr L32,
Trp L91, Tyr H102, Tyr H103, and Glu H50, all of which are
within 5 Å of the transition state. A potential base, Glu H50 is
placed perfectly to deprotonate the benzisoxazole and initiate
ring-opening. The distance of the nearest carboxylate oxygen
from the transferring proton is 2.5 Å. The proposed role in
catalysis is consistent with preliminary mutagenesis results
showing that activity is abolished when GluH50 is substituted
with glutamine.27
deeply inside the complementary binding pocket, which has a
very narrow entrance formed by several hydrophobic residues.
The second view shows that the transition state in 3D repre-
sentation is in relatively tight contact with the surroundings.
The region of negative electrostatic potential (red) is concen-
trated on the Glu H50 side, near where the proton of the
benzisoxazole is located.
Antibody 34E4 provides a preorganized microenvironment
made of hydrophobic and polar residues around the reaction
center, with the basic site oriented for deprotonation of the
benzisoxazole. The oxygen of the benzisoxazole ring is situated
so as to be solvated by external water as it is transformed into
cyanophenoxide.
Serum Albumin Catalysis. Serum albumins are ubiquitous
proteins in the blood stream; albumins bind various small
molecules in multiple binding sites. Hilvert and Kirby’s experi-
ment show that BSA and human serum albumin (HSA) have
similar effects on the benzisoxazole elimination, and there is a
binding site with lysine group present in each case.5 Because
the BSA crystal structure is unavailable, we used the HSA
crystal structure for the docking study.
Two binding pockets with the highest CAST score (see
Experimental Section) are the most likely binding sites (Figure
8). Subdomains IIA and IIIA, shown as light yellow and light
blue, are also the two sites that have been proposed to bind
small heterocyclic substrates such as warfarin experimen-
Figure 7 shows the solvent accessible surfaces of the 34E4
binding pocket including residues within 5 Å (top) or 8 Å
(bottom) of the transition state. The transition state is trapped
(26) 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.
(27) Seebeck, F., unpublished results.
9
8202 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Ring-Opening Reactions of Benzisoxazoles
A R T I C L E S
Figure 10. Surface representation of residues within 8 Å of the transition
state.
next to Lys 199. The protonation states of these two lysine
groups have been studied both experimentally and theoretically.
The combination of a neutral Lys 199 and protonated Lys 195
is the most favored one since they can form a hydrogen bond
with water.31 Experiments show that Lys 199 has a low pKa
(∼8).32 Therefore, we postulate that Lys 199 is the catalytic
group in human serum albumin, whereas Lys222 is the catalytic
group in bovine serum albumin. Experiments also indicate that
HSA loses most of its catalytic ability upon Lys199 modifica-
tion,5 confirming the predictions from docking experiments.
Figure 9. Transition state moiety in HSA subdomain IIA. (a) Residues
within 5 Å of the transition state (b) Surface representation of residues
within 5 Å of the transition state.
Conclusion
The mechanisms of catalysis of the Kemp elimination have
been elucidated for reactions occurring in two proteins, and the
rates have been computed in aqueous and nonpolar solvents.
Antibody 34E4 catalyzes this reaction by the alignment of
catalytic base Glu H50 and substrate benzisoxazole within a
relatively nonpolar pocket. Similarly, HSA has a deep nonpolar
pocket at subdomain IIA with Lys 199 that can deprotonate
5-nitrobenzisoxazole.
tally.29,30 The five myristic acids determined from the same
crystal structure are shown in the second view (Figure 8b).
Although there are lysine residues on the HSA surface, none
of them can be considered as effective catalytic bases because
of their high pKa values on the protein surface. There is one
lysine in subdomain IIA and no lysine residues inside the
subdomain IIIA. Consequently, we focused our attention on the
IIA subdomain.
An analogy between enzymes and organic solvents is often
made, but the heterogeneous microenvironment of a protein
pocket is clearly different from free solvent molecules in that
its fixed structure can be tuned to interact with and stabilize
disparate components of a reacting molecule. The efficiency of
these biological catalysts derives from having a catalytic base
geometrically positioned in a generally nonpolar active site, a
consequence of successful hapten design in 34E4, but seren-
dipitous in albumin. From the evolutionary perspective, there
is no fundamental difference between our specific residue
positioning argument and Kirby’s more recent proposal of a
specific medium effect.9,33 The specific medium effect involves
dispersion and electrostatic interactions with the transition state,
which is a necessary outcome of the specific group positioning.
The ultimate biological catalyst must utilize a perfect match
Autodock produced a binding mode that is shown in Figures
9 and 10. The average estimated free energy of binding is -6
kcal/mol. Figure 9 shows stick and space-filling renderings of
the protein with the transition state surrounded by residues in
the cleft of subdomain IIA within 5 Å. The pocket has mainly
nonpolar residues including Trp 214, Ala 219, Phe 211, Ser
202, His 242, and the putative base, Lys 199. The latter is 5.4
Å away from the transferring hydrogen. The only tryptophan
in albumin, Trp214, has π-π interactions with benzisoxazole
aromatic ring. There is barely enough room for the transition
state to stay in such a steep and narrow pocket. This can be
seen from the comparison of 5 Å (Figure 9b) and 8 Å (Figure
10) surfaces. The transition state resides in the concave part of
the binding pocket. Such a binding mode is not easily accessible
from outside without conformational changes of the albumin
upon binding the ligand. Lys 195 is located in subdomain IIA
(31) (a) Diaz, N.; Suarez, D.; Sordo, T. L.; Merz, Jr., K. M. J. Med. Chem.
2001, 44, 250-260. (b) Bhattacharya, A. A.; Petitpas, I.; Twine, S.; East,
M.; Curry, S. J. Biol. Chem. 2001, 276, 22 804-809.
(32) (a) Gerig, J. T.; Reiheimer, J. D. J. Am. Chem. Soc. 1975, 97, 168-173.
(b) Gerig, J. T.; Katz, K. E.; Reinheimer, J. D. Biochim. Biophys. Acta
1978, 534, 196-209.
(28) (a) Liang, J.; Edelsbrunner, H.; Woodward, C. Protein Science 1998, 7,
1884-1897. (b) Liang, J.; Edelsbrunner, H.; Fu, P.; Sudhakar, P. V.;
Subramaniam, S. Proteins 1998, 33, 1-17.
(29) Bhattacharya, A. A.; Grune, T.; Curry, S. J. Mol. Biol. 2000, 303, 721-
732.
(33) Hollfelder, F.; Kirby, A. J.; Tawfik, D. S. J. Am. Chem. Soc. 1997, 119,
9578-9579.
(30) Zaton, A. M. L.; Villamor, J. P. Chem.-Biol. Interact 2000, 124, 1-11.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8203

A R T I C L E S
Hu et al.
dinitro (380 nm), 5-nitro-6-chloro (380 nm), 5-nitro (380 nm), 5-cyano
(324 nm), 6-nitro (404 nm), 5,6-dichloro (342 nm), 6-chloro (329 nm),
5-chloro (339 nm), 5-fluoro (338 nm), and unsubstituted (325 nm).
Extinction coefficients were measured for each compound under the
reaction conditions. The second-order rate constants for the acetate and
butylamine-dependent reactions were determined in water or MeCN
at 20 °C. Pseudo-first-order rate constants were plotted against acetate
or butylamine concentration, and the slope of these plots gave the
second-order rate constant for the base-catalyzed reaction. The reactions
with 34E4 and BSA were performed in 40 mM phosphate, 100 mM
NaCl, pH 7.4 and 20 °C. Initial velocities were calculated by standard
linear-regression analysis using the initial linear portion of absorbance
vs time plots and were corrected for background activity. The kcat and
between the catalytic group and the transition state plus an
elaborate active site made of polar and nonpolar groups, which
provides favorable interactions with the transition state including
electrostatics, solvation and other factors.34 The most important
feature for catalysis is the assembly of catalytic functionality,
here the basic site, with appropriate orientation and reactivity.
Of course catalysts that accidentally have an appropriate
environment do exist, such as albumins in this case.
Catalytic antibodies represent a successful effort to biological
catalyst design for nonbiological reactions. The problems in such
design and execution are known5 to be difficult but not
insoluble.36,37 A remaining challenge is how to use the program-
mable nature of the antibody pocket along with medium effects
to act in concert with other active site parameters. Careful design
by increasing the strength of the catalytic base through desol-
vation and by providing specific stabilizing interactions with
the leaving group such as forming hydrogen bonds with
phenoxide, should lead to better catalysts for such a solvent
sensitive reaction.38
Km values were calculated by fitting the kinetic data to the Michaelis-
Menten equation using the program KaleidaGraph. Two percent
acetonitrile was used to dissolve the substrate for all reactions in aqueous
medium. Crystallized BSA, treated to remove bound lipids, was
purchased from Sigma and used without further purification; 34E4 was
isolated and purified as previously described.
Cloning of the Antibody Genes. Poly (A)+ mRNA was isolated
from the hybridoma producing antibody 34E4.2 A cDNA library was
constructed with the Great Lengths cDNA synthesis kit (Clontech).
The VL and VH genes were cloned using the polymerase chain reaction
with the following primers: VL sense, GTACATTTGCTCTTCGGT-
TCACAGGCTGTTGTGACTCAGGAA (the Sac I restriction site is
in italics); VL anti-sense, ATGAGTTTTTGTTCTGCGGCCGCCT-
TGGGCTGACCTAGGACAGT (Not I); VH sense AGGTCCAGCT-
GCTCGAGTCTGG (Xho I); VH anti-sense GTTCTGACTAGT-
GGGCACTCTGGGCTC (Spe I). The amplified fragments were puri-
fied, cloned into appropriate sites in the vector p4xH-M1342 and
sequenced. The resulting expression plasmid, p4xH-34E4, allows
production of 34E4 as a chimeric murine-human Fab fragment in which
the VL and VH segments of the catalytic antibody are fused to human
Cκ and gamma CH1 regions, respectively.42
Experimental Section
Synthesis of Substituted Benzisoxazoles. Unsubstituted benz-
isoxazole and the 5-nitro-, 6-nitro-, 5,7-dinitro-, 5-chloro-, and 6-chloro-
substituted derivatives were prepared according to the literature.1a,39
4-Cyano- and 4-fluorosalicylaldehyde were prepared from the corre-
sponding phenols by a Duff reaction;40 4,5-dichlorosalicylaldehyde was
prepared by a Reimer-Tiemann reaction.41 The substituted salicyl-
aldehydes were converted to benzisoxazoles by a standard literature
procedure.39b
5-Cyanobenzisoxazole. 1H NMR (300 MHz, CDCl3) δ 7.76 (q, 1H),
7.84 (q, 1H), 8.16 (d, 1H), 8.83 (d, 1 H). MS FABM+ 145 (M++H+).
mp 245 °C (decomp.).
5,6-Dichlorobenzisoxazole. 1H NMR (300 MHz, CDCl3) δ 7.80 (d,
1H), 7.85 (s, 1H), 8.68 (d, 1H). MS FABM+ 187/189 (M++H+), 210/
212 (M++Na+). mp 92 °C (sub.).
The antibody sequences for the variable segments of 34E4 are given
in Chart 1.
Construction of a Homology Model. A search of the structure
database identified antibodies HC19 (1gig.pdb)43 and J539 (2fbj.pdb)44
as having the highest sequence similarity to the 34E4 VL and VH
domains, respectively. Because HC19 and 34E4 have CDR H3 loops
that are the same length and that possess reasonable sequence similarity
(28.6%), modeling of the VL-VH interface45 is facilitated. A least-
squares fit of the VH domains of HC19 and J539, minus the CDR H3
loop, allowed overlay of the two molecules with a rmsd. of 1.40 Å for
all CR atoms except those of the CDR L1 (residues L24 to L34) and
H3 (residues H95 to H102). In this overlay, the VH domains of the
two antibodies were superimposed with rmsd. of 0.88 Å for all non-
CDR H3 CR atoms, whereas the VL domains were overlaid with rmsd.
of 1.37 Å for all non-CDR L1 CR atoms. In particular, the residues
near the VL/VH interface from the two molecules, namely residues L42-
L46, L94-L100, H43-H50, H58-H61, and H103-H105 (27 residues in
total), coincided with each other with rmsd. of 0.87 Å for all CR atoms.
Accordingly, visual inspection of the interface between the HC19 VL
and J539 VH domains revealed that no significant steric interference
was introduced by the way the two domains were put together. After
5-Fluorobenzisoxazole. 1H NMR (300 MHz, CDCl3) δ 7.32 (q, 1H),
7.38 (q, 1H), 7.59 (q, 1H), 8.7 (d, 1 H). MS FABM+ 139 (M++H+).
mp 68 °C (sub.).
5,6-Dinitrobenzisoxazole. 6-Nitrobenzisoxazole (200 mg) was dis-
solved in ice-cooled cH2SO4 (1.6 mL). Fuming nitric acid (200 µl)
was added dropwise, and the reaction mixture was heated at 80 °C for
30 min. The starting material was completely consumed. The reaction
mixture was poured onto ice and then extracted immediately with
methylene chloride. The organic layer was dried and purified by silica
gel chromatography using methylenechloride as eluent to give 142 mg
1
of the desired product in 56% yield. H NMR (300 MHz, CDCl3) δ
8.17 (q, 1H), 8.52 (d, 1H), 9.15 (d, 1H). MS FABM+ 210 (M++H+).
mp 112 °C.
5-Nitro-6-chlorobenzisoxazole. 6-Chlorobenzisoxazole was nitrated
at the 5-position by the same method used to prepare 5,6-dinitrobenz-
1
isoxazole in 62% yield. H NMR (300 MHz, CDCl3) δ 7.88 (d, 1H),
8.36 (d, 1H), 8.86 (d, 1H). MS FABM+ 198/200 (M++H+). mp 106
°C.
Kinetic Analyses. All kinetic experiments were performed at 20
°C. Fast reactions were measured by stopped flow techniques. The
reactions were initiated by adding the benzisoxazole substrate to the
reaction mixture and product formation was monitored spectrophoto-
metrically at the following wavelengths: 5,7-dinitro (352 nm), 5,6-
(39) (a) Kemp, D. S.; Woodward R. B. Tetrahedron 1965, 21, 3019-3035. (b)
Casey, M. L.; Kemp, D. S.; Paul, K. G.; Cox, D. D. J. Org. Chem. 1973,
38, 2294-2301.
(40) Suzuki, Y.; Takahashi, H. Chem. Pharm. Bull. 1983, 31, 1751-1753.
(41) Postmus, C.; Kaye, I. A.; Craig, C. A.; Matthews, R. S. J. Org. Chem.
1964, 29, 2693-2698.
(42) Ulrich, H. D., Patten, P. A., Yang, P. L., Romesberg, F. E., & Schultz, P.
G. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 11 907-11 911.
(43) Bizebard, T.; Daniels, R.; Kahn, R.; Golinellipimpaneau, B.; Skehel, J. J.;
Knossow, M. Acta Crystallog. 1994, D50, 768-777.
(44) Suh, S. W.; Bhat, T. N.; Navia, M. A.; Cohen, G. H.; Rao, D. N.; Rudikoff,
S.; Davies, D. R. Proteins: Struct. Funct. Genet. 1986, 1, 74-80.
(45) Stanfield, R. L.; Takimoto-Kamimura, M., Rini, J. M.; Profy, A. T.; Wilson,
I. A. Structure 1993, 1, 83-93.
(34) Houk, K. N.; Leach. A. G.; Kim, S. P.; Zhang, X. Angew Chem., Int. Ed.
2003, 42, 4872-4897.
(35) Zhang, X.; Houk, K. N., submitted for publication.
(36) Tantillo, D. J.; Houk, K. N. J. Org. Chem. 1999, 64, 3066-3076.
(37) Barbany, M.; Gutierrez-de-Teran, H.; Sanz, F.; Villa-Freixa, J.; Warshel,
A. ChemBioChem 2003, 4, 277-285.
(38) Parker, A. J. Chem. ReV. 1969, 69, 1-32.
9
8204 J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004

Ring-Opening Reactions of Benzisoxazoles
A R T I C L E S
Table 4. Comparison of Experimental and Calculated Solvation
Free Energies and Nitrobenzisoxazole Elimination Activation
Energies for Acetate-Catalysis with Different Theoretical Models
Chart 1
34E4 light chain variable domain
1
2
3
val
4
5
6
7
8
9
10 11 12
- - - leu thr
solvation free energy
solvation free energy
gln
ala
val thr gln
16 17 18
gly glu thr
glu ser ala
in H O (kcal/mol)
in CH CN (kcal/mol)
2
3
13
14
15
pro
19 20 21
val thr leu
22 23 24
thr cys arg
‡
‡
AcO-
NBI
TS
∆∆G
AcO-
NBI
TS
∆∆G
thr
ser
25
ser
26
ser
27
ser
27A 27B 27C 28 29 30
31 32 33
asn tyr ala
CPCMa
PCMa
-74.1
-75.3
-77.4
-5.2 -43.0 28.6 -63.0 -0.6 -33.6 22.2
-6.7 -43.5 30.7 -63.4 -0.6 -33.6 22.6
-5.7 -46.4 28.9 -75.1 -12.4 -54.4 25.3
gly ala val
37 38 39
gln glu lys
49 50 51
gly gly thr
61 62 63
arg phe ser
73 74 75
leu thr ile
thr thr ser
PBb
34
thr
35
trp
36
val
40 41 42
pro asp his
43 44 45
leu phe thr
SM5.2Rc
-77.3 -11.4 -51.6 29.3 -75.6 -15.7 -56.3 27.2
experiment -77.0d
23.8 -62.7e
13.1
46
gly
47
leu
48
ile
52 53 54
asn lys arg
55 56 57
ala pro gly
a B3LYP/6-31+G* with CPCM or PCM solvation models in Gaussian
98. b B3LYP/6-31+G* with Poisson-Boltzmann solvation model in Jaguar
4.0.21 c AM1 with SM5.2R solvation model in AMSOL 6.5.22 d Ref 23.
e Pliego, Jr., J. R.; Riveros, J. M. Phys. Chem. Chem. Phys. 2002, 4, 1622.
58
val
59
pro
60
ala
64 65 66
gly ser leu
67 68 69
ile gly asp
70
arg
71
ala
72
ala
76 77 78
thr gly ala
79 80 81
gln thr glu
82
asp
83
glu
84
ala
85
ile
86
87
88 89 90
cys ala leu
91 92 93
trp asn ser
charges were obtained at B3LYP/6-31+G(d) level. Several solvation
models including CPCM,19 a polarizable conductor-like solvation model,
PCM,20 a polarizable continuum model, a Poisson-Boltzmann (PB)
model21 and the Cramer-Truhlar SM5.2R solvation model22 were
compared for solvation calculations on acetate-catalyzed reaction (Table
4). All solvation energies from continuum models (PCM and CPCM)
are single point energies computed based on the geometries in the gas
phase.
Docking Calculations. Autodock 3.026 was employed for docking
purposes. Autodock uses a hybrid Lamarckian genetic algorithm (LGA)
for docking and predicts the best binding modes according to an
empirical scoring function that predicts the binding energy. Acetate
was left out purposely from the gas-phase transition state docking
experiments, since a carboxylate of an aspartate or glutamate in the
binding site is considered to be the base. CHELPG charges were taken
from the computed transition state structure in the gas phase. The
receptor is fixed in all the dockings and represented by grid maps. The
transition states do not have any torsional degrees of freedoms because
of their structural rigidity. A total of 100 ga-run were conducted in
each case and resulting conformations were clustered with 1 Å root-
mean-square deviations (RMSD). The representative docked conforma-
tion of the best cluster in terms of binding free energy and population
was chosen as the final binding mode.
tyr phe
94
asn
95
his
96
leu
97
98 99
100 101 102
gly gly thr
103 104 105
lys leu thr
val phe gly
106
val
106A 107
leu
108 109 110
gln pro (lys)*
gly
34E4 heavy chain variable domain
1
2
3
4
5
6
7
8
9
10 11 12
gly leu ala
(glu) val* (lys)* leu leu glu
ser gly gly
13
gln
14
pro
15
gly
16
17
18
19 20 21
lys leu ser
22 23 24
cys ala ala
gly ser leu
25
ser
26
gly
27
phe
28 29 30
asp phe arg
40 41 42
ala pro gly
52 52A 53
asn pro asp
63 64 65
leu lys asp
75 76 77
lys asn ser
84 85 86
ser glu asp
96 97 98
31 32 33
arg tyr trp
34 35 36
met thr trp
37
val
38
arg
39
gln
43 44 45
lys gly leu
46 47 48
glu trp ile
49
gly
50
glu
51
ile
54 55 56
ser arg thr
57 58 59
ile asn tyr
60
met
61
pro
62
ser
66 67 68
lys phe ile
69 70 71
ile ser arg
72
asp
73
asn
74
ala
78 79 80
leu tyr leu
81 82 82A
gln leu ser
82B 82C 83
arg
87 88 89
ser ala leu
90 91 92
tyr tyr cys
leu
arg
Location of Binding Sites in HSA. The CASTp program was used
to locate possible binding cavities and pockets for albumin using a
solvent probe (sphere of 1.4 Å) analytically.28 Multiple binding pockets
and cavities for HSA (PDB ID: 1BJ5) have been found as shown in
Figure 8 (top). This crystal structure includes five bound myristic acids.
These were removed for the CASTp analysis. The spheres represent
the surfaces of the key C, N, and O atoms that define the boundary
between bulk solvent and protein pockets or cavities.
93
val
94
arg
95
leu
99 100 100A
val tyr asn
B
C
D
asp phe asp
his tyr tyr
E
val
F
leu
101
asp
102 103 104
tyr trp gly
105 106 107
gln gly thr
108 109 110
ser val thr
111
val
112
ser
113
ser
the superimposition of the HC19 VL and J539 VH domains, residues
H90 to H104 (the CDR H3-containing loop) of the J539 VH domain
were replaced by the corresponding loop residues from HC19 to form
a HC19 VL-J539 VH chimera with the H3 loop from HC19. All residues
were then manually modified, or mutated, to reflect the 34E4 sequence.
For the “mutated” residues, the side-chain rotamer that gave the best
fit to the local environment as judged by visual inspection was chosen.
Other known structures were also consulted in remodeling the local
sequence environment around the mutated residues when the change
required a significant structural alteration. For example, replacing MetH82
with a branched leucine in the J539 VH domain caused a steric clash
with the surrounding residues. However, structural information from
antibody CHA255 (1ind.pdb),46 which has a leucine residue at position
H82, helped resolve the problem.
Acknowledgment. We thank Jim Na (Pfizer, La Jolla),
Jeehiun K. Lee (Rutgers), and Qiaolin Deng (Merck) for
preliminary results and extensive discussions. We are grateful
to the National Institute of General Medical Sciences, National
Institutes of Health (D.H. and K.N.H.) and National Science
Foundation (K.N.H.) for financial support of this research and
to the Naito Foundation (Tokyo, Japan) for a postdoctoral
fellowship to K.K. We owe special thanks to UCLA Academic
Technology Services (Y.H. and K.N.H.) and ETH and Novartis
(D.H.).
Quantum Mechanical Methods. Both transition state geometries
were optimized in the gas phase with Density Functional Theory
(DFT)47 at B3LYP/6-31+G(d) level with GAUSSIAN 98.18 CHELPG48
JA0490727
(47) (a) Becke, A. D. J. Chem. Phys. 1993, 98, 5648-5652. (b) Lee, C. T.;
Yang, W. T.; Parr, R. G. Phys. ReV. B 1988, 37, 785-789.
(48) (a) Chirlian, L. E.; Francl, M. M. J. Comput. Chem. 1987, 8, 894-905.
(b) Breneman, C. M.; Wiberg, K. B. J. Comput. Chem. 1990, 11, 361-
373.
(46) Love, R. A.; Villafranca, J. E.; Aust, R. M.; Nakamura, K. K.; Jue, R. A.;
Major, Jr., J. G.; Radhakrishnan, R.; Butler, W. F. Biochemistry 1993, 32,
10 950-10 959.
9
J. AM. CHEM. SOC. VOL. 126, NO. 26, 2004 8205