Expression of binding energy on an antibody reaction coordinate

Article abstract of DOI:10.1021/ja9928879

In this paper, we report the investigation of how the catalytic antibody 17E8 uses remote binding energy along the catalyzed hydrolytic reaction coordinate. With the use of alternative substrate analogues, we find that 17E8 can use free energy from binding interactions between the substrate side chain and antibody recognition pocket to equally stabilize the transition state and the Michaelis complex. In these cases, the interactions are not used to increase k(cat). We have also identified substrates for which the interactions are used to preferentially stabilize the transition state over the Michaelis complex. In these cases, the interactions are used to increase k(cat). Mechanistic studies support the idea that the differences in the substrates' kinetic activities results from differences in the expression of side-chain-pocket binding energy along the reaction coordinates. These results suggest that generating catalytic antibodies to transition-state analogues may be limited because the selective use of remote binding interactions cannot be programmed into transition-state analogues.

Full text of DOI:10.1021/ja9928879

VOLUME 121, NUMBER 51
DECEMBER 29, 1999
© Copyright 1999 by the
American Chemical Society
Expression of Binding Energy on an Antibody Reaction Coordinate
Herschel Wade and Thomas S. Scanlan*
Contribution from the Departments of Pharmaceutical Chemistry and Cellular and Molecular
Pharmacology, UniVersity of California, San Francisco, California 94143-0446
ReceiVed August 9, 1999
Abstract: In this paper, we report the investigation of how the catalytic antibody 17E8 uses remote binding
energy along the catalyzed hydrolytic reaction coordinate. With the use of alternative substrate analogues, we
find that 17E8 can use free energy from binding interactions between the substrate side chain and antibody
recognition pocket to equally stabilize the transition state and the Michaelis complex. In these cases, the
interactions are not used to increase kcat. We have also identified substrates for which the interactions are used
to preferentially stabilize the transition state over the Michaelis complex. In these cases, the interactions are
used to increase kcat. Mechanistic studies support the idea that the differences in the substrates’ kinetic activities
results from differences in the expression of side-chain-pocket binding energy along the reaction coordinates.
These results suggest that generating catalytic antibodies to transition-state analogues may be limited because
the selective use of remote binding interactions cannot be programmed into transition-state analogues.
Enzymes use binding energy to place substrates in a precise
position relative to catalytic groups in enzyme active sites. In
addition, they are able to use binding energy gained from
interactions with nonreacting portions of the substrate to stabilize
the transition state of the reaction.1-5 However, stabilization of
the transition state (obtaining a large kcat/KM) is not the only
requirement for efficient catalysis; it is also necessary to ensure
rapid turnover (obtaining a large kcat) by decreasing the free-
energy difference between transition-state complex, (E-S),‡ and
the ground-state Michaelis complex, (E-S).6-8 Enzymes achieve
* To whom correspondence should be addressed: (phone) 415-476-3620;
(fax) 415-476-0688; (e-mail) scanlan@cgl.ucsf.edu.
(1) Jencks, W. P. Cold Spring Harbor Symp. Quant. Biol. 1987, LII,
65-73.
(2) Jencks, W. P. Catalysis in Chemistry and Enzymology; Dover
Publications: Mineola, 1969.
(3) Fersht, A. R. Proc. R. Soc. London Ser. B 1974, 187, 397-407.
(4) Fersht, A. Structure and Mechanism in Protein Science: A Guide to
Enzyme Catalysis and Protein Folding; W. H. Freeman and Company: New
York, 1999.
(5) Pauling, L. Chem. Eng. News 1946, 24, 1375-1377.
(6) Avis, J. M.; Fersht, A. R. Biochemistry 1993, 32, 5321-5326.
(7) Whitty, A.; Fierke, C. A.; Jencks, W. P. Biochemistry 1995, 34,
11678-11689.
this by expressing binding energy selectively along the reaction
coordinate. By using the free energy from binding interactions
that either exclusively stabilize the (E-S)‡ or that stabilize the
(E-S)‡ more than the (E-S), enzymes achieve fast turnover
and avoid the unproductive overstabilization of ground-state
species.1-4
Structural and functional studies of catalytic antibodies have
shown that the bulk of catalytic power results directly from the
use of binding energy from interactions that are proximal to
and remote from the catalytic center.9-13 Indeed, the catalytic
antibody 17E8 uses the energy from noncovalent interactions
between the amino acid side chain of substrates and a side-
chain recognition pocket, which are both removed from the
(8) Wells, T. N. C.; Fersht, A. R. Biochemistry 1986, 25, 1881-1886.
(9) Wade, H.; Scanlan, T. S. J. Am. Chem. Soc. 1999, 121, 1434-1443.
(10) Wade, H.; Scanlan, T. S. Annu. ReV. Biophys. Biomol. Struct. 1997,
26, 461-493.
(11) Ulrich, H. D.; Mundorff, E.; Santarsiero, B. D.; Driggers, E. M.;
Stevens, R. C.; Schultz, P. G. Nature (London) 1997, 389, 271-275.
(12) Stewart, J. D.; Benkovic, S. J. Nature (London) 1995, 375, 388-
391.
(13) Benkovic, S. J. Annu. ReV. Biochem. 1992, 61, 29-54.
10.1021/ja9928879 CCC: $18.00 © 1999 American Chemical Society
Published on Web 12/14/1999

11936 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Wade and Scanlan
Table 1. Steady-State Kinetic Analysis of Substratesa.b
KM (µM)
[∆∆Gs (kcal/mol)]
kcat (s-1
)
kcat/KM (M‚s)-1
[∆∆G‡ (kcal/mol)]
[∆∆Gb(kcal/mol)]
1
2
3
4
5
6
7
8
3400 ( 400
[0.0]
1.0 ( 0.1
[0.0]
290 ( 40
[0.0]
180 ( 30
2.1 ( 0.1
[-0.4]
18 ( 1
[-1.7]
43 ( 2
[-2.1]
3.3 ( 0.5
[-0.7]
12 ( 1
[-1.4]
27 ( 2
[-1.9]
ND
12000 ( 1400
Figure 1. Hydrolytic reaction catalyzed by 17E8 and the transition
state formed by hydroxide attack.
[-1.7]
[-2.1]
2600 ( 100
[-0.1]
7000 ( 100
[-1.8]
4000 ( 200
[+0.1]
11000 ( 700
[-2.0]
3400 ( 900
[0.0]
970 ( 300
[-0.6]
4600 ( 200
2600 ( 100
[-1.2]
[+0.1]
5000 ( 700
[+0.3]
5400 ( 800
[-1.6]
ND
ND
a The ∆∆Gb values were calculated from the equation -RT ln[(kcat
/
KM)X/(kcat/KM)1] where R is the gas constant and T is the absolute
temperature. The errors shown were obtained from calculated fits
(KaleidaGraph-Synergy Software) of the data to the Michaelis-Menton
equation. The errors in the kcat/KM values were from the propagation
of the fitted kcat and KM errors.49 b The ∆∆Gs values were calculated
from the equation -RT ln[(KM)X/(KM)1], ∆∆G‡ values were calculated
from the equation -RT ln[(kcat)X/(kcat)1], where X corresponds to a
substrate to which 1 is being compared.
Figure 2. (A) Substrates used to study side-chain-pocket interactions.
The N-formyl phenyl ester skeleton is shown above side-chains which
are designated by R. (B) The N-formyl phenyl phosphonate skeleton
of the transition-state analogues used in this study.
substrates is due to differences in the expression of binding
energy along the catalytic reaction coordinate.
reactive center, to promote ester hydrolysis (Figure 1).9,14 These
interactions are analogous to those in natural enzymes such as
proteases and tRNA synthetases that also use binding interac-
tions between amino acid side chains and recognition pockets.15-22
Experimental Section
General Methods and Reagents. Reactions requiring anhydrous
conditions were carried out in flame-dried glassware under an
atmosphere of argon. Anhydrous solvents were purchased from Aldrich.
Chromatography solvents were purchased from Fisher Corp. and were
used as received. Reagents were purchased either from Sigma or Aldrich
In a previous study, we determined that the removal of the
methylene groups from 2 to yield 1 (Figure 2) resulted in an
increase in the free energy of the 17E8-substrate transition-
state complex, indicating that 17E8 uses the side-chain-pocket
interactions for overall transition-state stabilization (Table 1).9,14
The removal of the interactions also resulted in a similar
decrease in the stability of the Michaelis complex, indicating
that 17E8 uses the side-chain-pocket interactions in a manner
that has been termed uniform binding.1-4 With uniform binding,
the interactions are used to equally stabilize both the Michaelis
and the transition-state complexes (Figure 3). Because the
additional side-chain-pocket interactions are not used to
decrease the free energy difference between the (IgG-S) and
the (IgG-S)‡, which would result in an increased kcat value,
uniform binding can be viewed as a wasteful use of binding
energy and a trait of a primitive catalyst.2,3,23
and used as received unless noted otherwise. H NMR and 13C NMR
1
were recorded on a General Electric 300 MHz instrument. The NMR
samples were prepared in 5-mm tubes, and the chemical shifts are
reported in parts per million (δ) relative to the TMS standard for
samples in CDCl3 and the TSP (3-(trimethylsilyl)-1-propanesulfonic
acid sodium salt) standard for samples in D2O. Flash chromatography
was performed with Merck silica gel 60 (230-400 mesh). Ion exchange
chromatography was performed with Diethylaminoethyl (DEAE)
Sephadex A-25 (anion exchange) and Dowex 50WX2-100 (cation
exchange). The resins were washed and prepared according to
manufacturers’ recommendations.
Synthesis of n-Formylated Amino Acid Ester Substrates, 1-8.
The racemic phenyl ester substrates were synthesized as described by
Wade and Scanlan.9 The synthesis of 5 is described by Guo et al. The
L-enantiomer of the amino acid esters (1, 2, and 4) was used in the
temperature-dependent studies. The synthesis of the enantiopure
substrates is described by Wade and Scanlan.14
We have identified alternative substrates for which the binding
energy between the amino acid side chain and recognition pocket
is used to increase the catalytic turnover number (Figure 2).
Here we show that the difference in the kinetic activity of these
n-Formyl-((S)-methyl)-cysteine phenyl ester [3]: yield 0.65 g
1
(80%) H NMR (300 MHz, CDCl3) δ 8.29 (s, 1H), 7.40 (t, J ) 7.8
Hz, 2H), 7.27 (t, J ) 7.2 Hz, 1H), 7.13 (d, J ) 8.2 Hz, 2H), 6.64 (br
d, J ) 7.5 Hz, 1H), 5.16 (dd, J ) 7.8, 4.5 Hz, 1H), 3.16 (d, J ) 5.1
Hz, 2H), 2.21 (s, 3H); 13C NMR δ 160.8, 129.6, 126.4, 121.2, 50.6,
36.3, 16.4; MS (EI) 239.0(M+), 239.1, 194.1, 160.0, 146.0, 118.0, 101.0,
(14) Wade, H.; Scanlan, T. S. J. Am. Chem. Soc. 1996, 118, 6510-
6511.
(15) Bizzozero, S. A.; Baumann, W. K.; Dutler, H. Eur. J. Biochem.
1982, 122, 251-258.
(23) For another example of context-dependent expression of binding
energy see Narlikar, G.; Herschlag, D. Biochemistry 1998, 37, 9902-9911.
In this study with a Tetrahymena group I RNA enzyme, the authors find
that the inclusion or removal of a specific binding interaction gives rise to
a uniform binding effect in the WT ribozyme and provides specific
transition-state stabilization in the context of a mutant ribozyme. These
results suggest that the optimization of a catalyst and the contributions that
binding interactions may make to catalysis depend on the context (i.e., the
starting point) and the amount of binding energy available from other
interactions. In our study, the starting point is 17E8 and the substrate that
17E8 was designed to cleave is (2).
(16) Bauer, C.-A.; Thompson, R. C.; Blout, E. R. Biochemistry 1976,
15, 1291-1296.
(17) Thompson, R. C.; Blout, E. R. Biochemistry 1973, 12, 57-65.
(18) Thompson, R. C. Biochemistry 1974, 13, 5495-5501.
(19) Stroud, R. M. Am. Sci. 1974, 231, 74-88.
(20) Fersht, A. R.; Dingwall, C. Biochemistry 1979, 18, 1250-1255.
(21) Dorovska, V. N.; Varfolomeyev, S. D.; Kazanskaya, N. F.; Klyosov,
A. A.; Martinek, K. FEBS Lett. 1972, 23, 122-124.
(22) Berezin, I. V.; Kazanskaya, N. F.; Klyosov, A. A.; Martinek, K.
FEBS Lett. 1971, 15, 125-128.

Binding Energy and Antibody Reaction Coordinates
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11937
Figure 4. Correlation between log P and log KM values of the
substrates used in the 17E8 catalyzed reaction. The KM values for the
substrates that are not discussed in this paper were taken from ref 12.
The concentration units for the substrates are µM. The log P values
correspond to those of model compounds that are analogous to the side
chains of the phenyl ester substrates: pentane, log P ) 3.39, 2; cis
3-pentene, 2.83; butane, 2.89; propane, 2.36; ethane, 1.81, 1; diethyl
sulfide, 1.95, 4; diethyl ether, 0.89, 7. The values were taken from ref
24. The boxed data designate the substrates that are discussed in this
paper. The slope of the line without 4 and 7 is -0.9 ( 0.2, R ) 0.93;
with 4 and 7 included, the slope is -0.7 ( 0.2, R ) 0.90.
Figure 3. Free energy diagrams representing binding energy use and
catalysis. (A) 17E8 catalyzed reaction with 1 (black) and 2 (grey) (B)
17E8 catalyzed reaction with 1 (black) and 4 & 7 (grey). Both IgG
and S represent unbound antibody and substrate, respectively. (IgG-
S) and (IgG-S)‡ represent the Michaelis and transition-state complexes,
respectively.
94.0, 90.0, 77.0, 61.0. HRMS calcd for C12H15N1O3S1: 239.0616.
Found: 239.0616.
n-Formyl-((S)-ethyl)-cysteine phenyl ester [4]: yield 0.75 g (90%);
1H NMR (300 MHz, CDCl3) δ 8.22 (s, 1H), 7.36 (t, J ) 7.5 Hz, 2H),
2.23 (t, J ) 7.2 Hz, 1H), 7.10 (d, J ) 7.8 Hz, 2H), 6.56 (br s, 1H),
5.10 (dd, J ) 9.0, 4.8 Hz, 1H), 3.12 (d, J ) 4.5 Hz, 2H), 2.58 (dd, J
) 7.5, 6.9 Hz, 2H), 1.52 (t, J ) 7.2 Hz, 2H); 13C NMR d 169.1, 160.8,
150.3, 129.6, 126.4, 121.2; MS (EI) 253.1(M+), 208.1, 160.0, 132.0,
115.0, 104.1, 94.0, 87.0, 75.0, 65.0. HRMS Calcd for C12H15N1O3S1:
253.0773. Found: 253.0774.
n-Formyl-(O-methyl)-serine phenyl ester [6]: yield 0.73 g (88%)
1H NMR (300 MHz, CDCl3) δ 8.30 (s, 1H), 7.40 (t, J ) 7.5 Hz, 2H),
7.26 (t, J ) 7.8, 7.5 Hz, 1H), 7.10 (d, J ) 7.5 Hz, 2H), 6.54 (br d, J
) 5.4 Hz, 1H), 5.08 (dt, J ) 8.4, 2.7 Hz, 1H), 4.05 (dd, J ) 9.3, 3.0
Hz, 1H), 3.77 (dd, J ) 9.3, 3.0 Hz, 1H), 3.44 (s, 3H); 13C NMR δ
160.7, 129.5, 126.3, 121.3, 72.2, 59.7, 51.5; MS (EI) 224.0(MH+),
224.1, 130.1, 102.1, 94.0, 85.0, 74.1, 65.0. HRMS Calcd for
C11H14N1O4: 224.0923. Found: 224.0915.
Figure 5. pH-kcat profile for 17E8 catalyzed hydrolysis of 2 (inset)
and 4. The pKa values obtained from the fit to the following equation:
ko
n-Formyl-(O-ethyl)-serine phenyl ester [7]: yield 0.54 g (60%);
1H NMR (300 MHz, CDCl3) δ 8.30 (s, 1H), 7.40 (t, J ) 7.8, 7.5 Hz,
2H), 7.26 (t, J ) 6.9 Hz, 1H), 7.10 (d, J ) 7.8 Hz, 2H), 6.56 (br d, J
) 5.4 Hz, 1H), 5.10 (dd, J ) 8.4, 3.0 Hz, 1H), 4.08 (dd, J ) 9.3, 2.4
Hz, 1H), 3.79 (dd, J ) 9.3, 2.4 Hz, 1H), 1.52 (q, J ) 6.9 Hz, 2H) 1.22
(t, J ) 6.9 Hz, 3H); 13C NMR δ 160.6, 129.5, 126.2, 121.3, 70.1, 67.1,
51.5, 15.0; MS (EI) 238.1(MH+), 238.1, 144.0, 133.0, 116.0, 94.0, 82.9,
70.0, 60.0. HRMS Calcd for C12H15N1O4: 237.1001. Found: 237.0995.
n-Formyl-(O-methyl)-homoserine phenyl ester [8]: yield 0.57 g
kcat(obs)
)
10pK -pH + 1 + 10pH-pK
′
a1
a2
where kcat(obs) is the observed kcat value and ko is the pH independent
kcat value. The equation was derived from a scheme based on two
ionizable antibody residues (for a model see ref 27). The pKa1 and pKa2
values obtained for the substrates are 9.2 ( 0.2 and 9.7 ( 0.2, R )
0.993 (4), and 9.0 ( 0.1 and 10.0 ( 0.1, R ) 0.994 (2).
3-(S-methyl)-propyl (corresponding substrate is 5) (44%): 1H NMR
(300 MHz, D2O) δ 8.16 (s, 1H), 7.41 (t, J ) 6.8 Hz, 2H), 7.22 (t, J )
7.6 Hz, 1H), 7.17 (d, J ) 8.4 Hz, 2H), 4.45 (ddd, J ) 14.8, 11.2, 3.2
Hz, 1H), 2.69 (ddd, J ) 13.6, 8.4, 4.8 Hz, 1H), 2.80 (dt, J ) 13.2, 8.4
Hz, 1H), 2.17 (m, 1H), 2.11 (s, 3H), 1.96 (s, 1H) LSIMS-MS(-) 288.4
m/z (MH--HCl).
Steady-State Kinetics of the Phenyl Esters. Michaelis-Menton
parameters for the substrates were determined by continuous measure-
ment at 270 nm (phenol release ꢀ ) 1400 M-1‚cm-1) using a Uvikon
930 (Kontron Instrument) UV-vis spectrophotometer. All assays were
performed with cuvette holders thermostated at 24.5 ( 0.5 °C with a
Lauda RM6 temperature control unit. Cells of 1-cm path length (0.5
mL) were used in each experiment. The buffer used in all kinetic
experiments was 50 mM borate-150 mM NaCl, pH 8.7. The antibody
concentrations used in the experiments ranged from 0.2 to 1.4 µM. All
substrates were soluble at these substrate concentrations. The substrate
concentrations used were: 1, 800 µM to 30 mM; 2, 30 µM to 1 mM;
3, 650 µM to 22 mM; 4, 200 µM to 25 mM. The reactions were initiated
by adding 20 mL of the substrate stock in DMSO to a solution of 13-
1
(63%) H NMR (300 MHz, CDCl3) δ 8.26 (s, 1H), 7.39 (t, J ) 7.8
Hz, 2H), 7.25 (t, J ) 7.2 Hz, 1H), 7.10 (d, J ) 7.8 Hz, 2H), 6.85 (br
d, J ) 5.4 Hz, 1H), 5.10 (dt, J ) 7.2, 5.4 Hz, 1H), 3.58 (t, J ) 5.7
2H), 3.35 (s, 3H), (q, J ) 5.4 Hz, 2H) 1.22 (t, J ) 6.9 Hz, 3H); 13C
NMR δ 170.3, 160.8, 150.5, 129.4, 126.0, 121.2, 68.9, 58.9, 49.7, 15.0;
MS (EI) 238.1(MH+), 179.1, 158.0, 116.1, 94.0, 82.9, 70.0, 65.0, 56.0.
HRMS Calcd for C12H16N1O4: 238.1079. Found: 238.1087.
Synthesis of n-Formylated Phenyl Phosphonates (see Figure 6a,
6b, and text for phosphonates used). The phosphonates were prepared
by the route similar to that described by Guo et al. The use of aldehydes
with different side chains yielded the series.
Phenyl [1-(1-N-Formylamino)alkyl]phosphonates. 2-((S)-ethyl)-
ethyl (corresponding substrate is 4 in Figure 2): yield 0.20 g (44%);
1H NMR (300 MHz, D2O) δ 8.22 (s, 1H), 7.45 (t, J ) 8 Hz, 2H), 7.23
(t, J ) 8.0 Hz, 1H), 7.19 (d, J ) 9 Hz, 2H), 4.44 (ddd, J ) 15.2, 12,
3.2 Hz, 1H), 3.22 (ddd, J ) 14.4, 4.8, 3.2 Hz, 1H), 2.80 (ddd, J )
14.4, 12, 6 Hz, 1H), 2.61 (dq, J ) 7.2, 2.4 Hz, 2H), 1.20 (t, J ) 7.2
Hz, 3H) LSIMS-MS(-) 288.4 m/z (MH--HCl).

11938 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Wade and Scanlan
Figure 6. Transition-state versus substrate complementarity. (A) ln
KI vs ln kcat/KM (slope ) -1.08 ( 0.11 R ) 0.973). (B) ln KI vs ln KM
(slope ) 1.64 ( 0.05 R ) 0.999). The data used in the plots were
obtained using various phenyl ester substrates and phosphonates. The
substrate and phosphonate skeletons are shown in Figure 2. The side
chains are shown next to the data points. The data were fit with the
KaleidaGraph curve-fitting program (data taken from refs 9 and 14
and Table 1). The boxed data designate the substrates that are discussed
in this paper.
25 mL of 17E8 (in PBS) and 455-467 mL of pH 8.7 borate buffer. In
the background reactions, the IgG was replaced with PBS. All catalyzed
assays were performed in triplicate. The background reactions were
performed in duplicate. The catalyzed rate was obtained by subtracting
the average of the background reaction rate from the rate of the
catalyzed reaction. The data (V vs [S]) from the experiments were fit
with the KaleidaGraph (Synergy Software) curve-fitting program using
the Michaelis-Menton equation.
pH-Rate Dependence of 17E8 Catalysis. The steady-state kinetic
parameters of 17E8 catalysis of 4 at different pH values were obtained
in the same manner as described above for the experiments performed
at pH 8.7. The following buffer systems were used in this analysis:
50 mM NaHPO4, 150 mM NaCl, pH 7.2; 50 mM Tris, 150 mM NaCl,
pH 7.8; 50 mM borate, 150 mM NaCl, pH 8.4, pH 8.7; 50 mM CHES,
150 mM NaCl, pH 9.0, pH 9.5, pH 9.8, and pH 10.0. The buffer system
used for 2 is listed in ref 27. The parameters for the pH values >10
were not obtained due to a high background reaction. The kcat values
were plotted as a function of pH to the equation described in the legend
of Figure 5 using the KaleidaGraph plotting program.
Temperature Dependence of Catalysis. All reactions were per-
formed in 50 mM borate, 150 mM NaCl (pH 8.0). The steady-state
kinetic parameters of 17E8 catalysis of 1, 2, and 4 at the different
temperatures were obtained in the same manner as described above
for the experiments performed at 24.5 °C. The temperature range
covered for each of the substrates was from 5 to 40 °C. The kcat, KM,
and kcat/KM values for the different temperatures were plotted as a
function of 1/T (K) to the equation described in the legend of Figure
7 using the KaleidaGraph plotting program.
Figure 7. Temperature dependence of KM, kcat, and kcat/KM for 1 (A),
2 (B), and 4 (C). The KM data for 1 and 4 were fit to the equation: ln
KM ) (∆Ho/RT) - (∆So/R). The KM data for 2 was fit to the equation
ln KM ) (∆HoT(0)/RT) - ((∆ST‡(0) - ∆Cpo)/R) - (∆Cpo ln T/R), where
∆HoT(0) and ∆STo(0) are the enthalpy and entropy change at 0 K. The kcat
data for 1 and 2 were fit to the equation: ln(kcat/T) ) ln(kb/h) + (∆S‡/
R) - (∆H‡/RT). The kcat data for 4 was fit to the equation ln kcat
ln(kb/h) - (∆HT‡(0)/RT) + (1 + (∆Cp‡/R)) ln T + ((∆S‡T(0)
)
-
∆Cp‡)/R), where ∆H‡T(0) and ∆ST‡(0) are the enthalpy and entropy of
activation at 0 K, respectively. The kcat/KM data for 1 was fit to the
equation: ln(kcat/KMT) ) ln(kb/h) + (∆S‡/R) - (∆H‡/RT). The kcat/KM
data for 2 and 4 were fit to the equation: ln(kcat/KM) ) ln(kb/h) -
(∆H‡T(0)/RT) + (1 + (∆Cp‡/R)) ln T + ((∆S‡T(0) - ∆Cp‡)/R).
Results and Discussion
used to decrease the free-energy difference between the (IgG-
S) and the (IgG-S)‡ and increase kcat. We decided to exploit
the correlation between the stability of the (IgG-S), log KM,
and the hydrophobicity parameter, log P, for the substrate side
chains (Figure 4). The relationship suggests that the Michaelis
complex can be destabilized by using substrates with side chains
To make 17E8 behave in a more catalytically optimized
manner than that demonstrated with the substrates 1 and 2, we
attempted to circumvent the uniform use of the side-chain-
pocket binding energy by destabilizing the (IgG-S). This
destabilization would leave some of the binding energy to be

Binding Energy and Antibody Reaction Coordinates
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11939
that are less hydrophobic than 2. There is also a modest
correlation between log(kcat/KM) and log P.9 It is unclear if either
correlation is due to the removal of side-chain-pocket contacts
or to side-chain hydrophobicity. Unfortunately, these factors are
inextricably linked, so we decided to change the hydrophobicity
of the side chain while maintaining the maximum number of
heavy-atom-pocket contacts to reduce destabilization of the
(IgG-S)‡ complex. By replacing methylene groups in the side
chain with sulfur and oxygen atoms, the hydrophobicity of the
side chain is substantially altered as indicated by the changes
in the log P values (Figure 4) which are directly correlated to
the free energy of transfer from octanol to water for model
compounds.24
inhibitor of the 17E8 reaction and has an inhibition and binding
constant similar to that of the phosphonate corresponding to 2
(data not shown). The ln KI and ln(kcat/KM) relationship for the
substrates and their corresponding transition-state analogues is
also maintained with the inclusion of substrates 3 and 4,
supporting the similarity in transition-state structure for 2, 3,
and 4 hydrolysis (Figure 6).28-31 These results support the
kinetic evidence that points to differential binding as the cause
for the increase in turnover rate. The tight binding of the
transition-state analogue also discounts the possibility that the
increase in turnover rate is solely due to an increase in the
intrinsic reactivity of the ester with the replacement of the
heteroatom.32
To obtain further insight into the mechanistic differences in
the catalyzed hydrolysis of substrates 1, 2, and 4, the temperature
dependence of 17E8 catalysis with each of the substrates was
studied. The temperature dependence of the steady-state pa-
rameters KM, kcat, and kcat/KM yielded the thermodynamic
quantities shown in Table 2.33 The ∆HKM values obtained for
substrate binding are -7.0, -2.9, and -2.8 kcal/mol for 1, 2,
and 4, respectively. The ∆SKM values are 6.3, 1.4, -0.5 cal/
(mol‚K).34 The van’t Hoff plot for substrate 2 results in a curved
data fit which suggests that the ∆HKM and ∆SKM values are not
constant with temperature and that a ∆CpKM must be included
in the van’t Hoff equation (Figure 7B).35 The ∆CpKM value
obtained from the fit is 700 cal/(mol‚K).36 There is no observed
curvature in the ln KM-1/T plot with substrates 1 and 4 (Figure
7A,C).37 The data for the kcat dependence on temperature yields
∆H‡kcat values of 10.1, 5.6, and 8.6 kcal/mol and ∆S‡kcat values
of -32, -25, and -32 cal/(mol‚K) for 1, 2, and 4, respectively.
The steady-state kinetic parameters for the substrates are
shown in Table 1. As predicted by the log P-log KM correlation
for the homologous aliphatic side-chain series, the heteroatom
replacement results in destabilization of the (IgG-S) as indicated
by the large increases in KM and the unfavorable ∆∆Gs values
which are as large as 2 kcal/mol (relative to 2). The KM values
for the substrates with the oxygen replacements (6-8) are larger
than those for those with the sulfur replacements (3-5) further
substantiating the log P-log KM trend. The inclusion of these
substrates on the log KM-log P plot maintains the linear
relationship between the two parameters (Figure 4). The
heteroatom replacements do not result in significant changes
in the kcat/KM values (compare 4, 5, and 7 to 1) which is in
contrast to the result of deleting methylene groups (compare 1
and 2). The heteroatom replacements do result in increased kcat
values for the 17E8 catalyzed reactions. The kcat increase is as
high as 20-fold (8) and is dependent on the heteroatom position
in the side chain. The γ-replacements (3, 4 and 6, 7) had much
larger increases than the δ-replacements (5 & 8).
The kinetic behavior of these alternative substrates suggests
that the side-chain-pocket binding energy is being used
differentially between the (IgG-S) and the (IgG-S)‡ (see Figure
3).1,2,4 When the additional binding groups are added to 1 to
yield 3-7, the (IgG-S) complex is not further stabilized, in
contrast to the 1 to 2 side-chain change. However, the additional
groups do significantly stabilize the (IgG-S)‡ complex as shown
by the negative ∆∆Gb values that are more favorable by as much
as 2 kcal/mol. These increases in kcat values for substrates 3-7
suggest that the additional interactions are used to decrease the
free-energy difference between the (IgG-S) and (IgG-S)‡
complexes. The ∆∆G‡ value (relative to 1) is decreased by as
much as 1.7 kcal/mol.
The plot of ln(kcat/T) vs 1/T for 4 suggests that ∆H‡
and
kcat
‡
∆S kcat values are not constant with temperature for the reaction
with 4 (Figure 7C).38 The ∆Cp‡ kcat value obtained from the fit
is 140 cal/(mol‚K). No curvature is seen in the ln(kcat/T) vs 1/T
plots for 1 and 2 (Figure 7A,B). The kcat/KM temperature
dependence data yields ∆H‡TS values of 2.6, 2.5, and 5.8 kcal/
(28) Mader, M. M.; Bartlett, P. A. Chem. ReV. 1997, 97, 1281-1301.
(29) Bartlett, P. A.; Marlowe, C. K. Biochemistry 1983, 22, 4618-4624.
(30) Wolfenden, R. Annu. ReV. Biophys. Bioeng. 1976, 5, 271-306.
(31) Gandour, R. D.; Schowen, R. L. Transition States of Biochemical
Processes; Plenum Press: New York, 1978.
(32) The larger kcat value for 4 (compared with that of 7) also suggests
that an increase in the intrinsic reactivity of the carbonyl due to the
heteroatom replacement is not a large factor as the oxygen replacement
should render the carbonyl group more reactive than the sulfur atom
replacement.
(33) It should be pointed out that because there is a pH rate dependence
for the catalyzed reaction, it is therefore possible that the parameters obtained
from the temperature dependence (∆H‡, ∆HKM, ∆S‡, and ∆SKM) include
contributions from the protonation and deprotonation of active-site residues.
For the study presented, we emphasize the differences in the parameters
obtained for the substrates. The contributions from the protonation and
deprotonation of the active-site residues are most likely the same for the
substrates, as they are cleaved by the same mechanism, with the participation
of the same active-site residues, and have similar transition-state structures.
(34) Although the ∆SKM may seem anomalous for a binding event, similar
values have been noted for binding of substrates in other enzymatic systems.
Hinz, H.; Weber, K.; Flossdorf, J.; Kula, M. Eur. J. Biochem. 1976, 71,
437-442; Sturtevant, J. M. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 2236-
2240; Fukuda, M.; Kunugi, S. Eur. J. Biochem. 1984, 71, 565-570; and
Ross, P. D.; Subramanian, S. Biochemistry 1981, 20, 3096-3102. These
positive and near-zero values for the association of the substrate with 17E8
are consistent with the release of bound water for the 17E8 active site and
from the substrate’s hydrophobic moieties (phenyl group and side chain).
(35) In this study, we assume that KM is approximately equal to KS. We
believe this assumption to be reasonable for several reasons. The turnover
To ensure that the changes in the kinetic parameters were
due to differential binding and not to a change in the catalytic
mechanism, we performed a pH-rate study for substrate 4, which
is the substrate that has the highest increase in turnover number
(Figure 5). The shape of the plot and the two pKa values obtained
for 4 are similar to those for 2, suggesting that the same residues
participate in the rate-determining step for the catalyzed
hydrolysis of 2 and 4.25-27
The phosphonate transition-state analogue corresponding to
substrate 4 was also synthesized and tested for its binding and
inhibition activity related to the 17E8 catalyzed reaction (see
Figure 2 for structure). The phosphonate is a competitive
(24) Hansch, C.; Leo, A.; Hoekman, D. Exploring QSAR: Hydrophobic,
Electronic, and Steric Constants; American Chemical Society: Washington,
D. C., 1995.
rate for 17E8 is not fast enough to warrant the suggestion that koff > kcat
,
(25) Zhou, G. W.; Guo, J.; Huang, W.; Fletterick, R. J.; Scanlan, T. S.
Science (Washington, D.C.) 1994, 265, 1059-1064.
(26) Guo, J.; Huang, W.; Zhou, W.; Fletterick, R. J.; Scanlan, T. S. Proc.
Natl. Acad. Sci. U.S.A. 1995, 92, 1694-1698.
(27) Guo, J.; Huang, W.; Scanlan, T. S. J. Am. Chem. Soc. 1994, 116,
6062-6069.
thus leaving term KM devoid of chemical steps. The existence of additional
unimolecular nonchemical steps (i.e., protein conformational change) which
would precede the rate-determining chemical step is unlikely, due to the
simple catalytic mechanisms elicited by phosphonate haptens. The rate-
determining nature of the hydrolysis of the phenyl ester has been determined
in other mechanistic studies (see ref 27).

11940 J. Am. Chem. Soc., Vol. 121, No. 51, 1999
Table 2. The Thermodynamic Quantities Obtained from the
Wade and Scanlan
From these thermodynamic quantities, a model of the use of
side-chain-pocket interactions along 17E8’s hydrolytic reaction
coordinate can be postulated. The more favorable ∆HKM for 2
suggests that more interactions are formed in the (2-17E8)
complex than in the (S-17E8) complexes with 1 and 4. The
∆∆HKM of binding between 1 and 3 is essentially zero,
suggesting that a similar number of interactions are formed in
the (1-17E8) and (4-17E8) complexes. The large ∆CpKM value
suggests that the interactions formed in the (2-17E8) complex
are hydrophobic in nature which is expected due to the structure
of the n-butyl moiety and the residues that surround the side-
chain in the transition-state-17E8 complex.39-43 The lack of
an observable ∆CpKM and the enthalpic similarity between (1-
17E8) and (4-17E8) suggests that the side chain does not make
substantial contact with the recognition pocket in the Michaelis
complex of 17E8 and 4.
a
Temperature Dependence of kCat, KM, and kCat/KM
a
1/KM
b
∆HKM
∆SKM
cal/(mol‚K)
∆CpKM
cal/(mol‚K)
substrate
substrate
substrate
kcal/mol
-2.8 ( 0.2
-7.0 ( 0.9
-2.9 ( 0.2
0.5 ( 0.6
6.4 ( 0.7
700 ( 80
-1.4 ( 0.6
kcat
c
∆H‡
∆S‡
∆Cp‡
cal/(mol‚K)
kcat
kcat
kcat
kcal/mol
cal/(mol‚K)
-32 ( 2
8.6 ( 0.5
10.1 ( 0.4
5.6 ( 0.7
-26 ( 1
-34 ( 5
The ∆H‡
is most favorable for 4, suggesting that more
kcat
binding interactions are used in the transformation of (IgG-4)
to (IgG-4)‡ than with 1 and 2. The curvature in the ln(kcat/
T)-1/T plot suggests that the hydrophobic effect is playing a
role in the turnover of the (IgG-4) complex. The ∆Cp‡ (4)
value is smaller than the ∆CpKM (1) value. This is expected as
the group contribution of -S- to ∆Cp is smaller than that of
a -CH2- group.42,44 The large negative ∆S‡kcat for the substrates
suggests that the transition-state complexes are substantially
more ordered than the substrates’ respective Michaelis com-
plexes. This can be explained by the fact that more contacts
are made in the transition-state complex (especially with the
oxyanion hole) and the fact that a water molecule must become
fixed to participate in the transition state of the reaction.
The postulated models for 17E8’s use of the side-chain-
pocket interactions are shown in Figure 8. The side-chain
contacts are used extensively with 2 in the Michaelis complex
and much less so with 4 as suggest by the ∆HKM values and
the ∆CpKM for 2. The side-chain contacts are used in converting
the (IgG-4) complex to its transition-state complex as suggested
-140 ( 50
kcat/KM
∆H‡
∆S‡
∆Cp‡
cal/(mol‚K)
TS
TS
TS
kcal/mol
cal/(mol‚K)
-32 ( 2
5.8 ( 0.6
2.0 ( 0.3
2.1 ( 0.5
-32 ( 1
-33 ( 4
-620 ( 110
-170 ( 50
a The values of 1/KM are shown so that the values represent those
for formation of the Michaelis complex as KM formally represents a
dissociation constant. b The ∆HKM and ∆SKM values at 298 K for 2
were calculated from the fits in Figure 7 using the equations:
∆H2o98K ) ∆HoT(0) + T∆Cpo and ∆So298K ) ∆SoT(0) + ∆Cpo ln T
c The ∆H‡ and ∆S‡ values at 298 K for 2 were calculated from
kcat
the fits in kFcaitgure 7 using the equations:
∆H2‡98K ) ∆H‡T(0) + T∆Cp‡ and ∆S‡298K ) ∆S‡T(0) + ∆Cp‡ ln T
by the ∆H‡
and ∆Cp‡
values. Substrate 1 cannot make
kcat
kcat
many contacts in either complex and thus has been used as a
control substrate to determine if the curvature in the temperature-
dependent plots indeed results from the extended side-chain-
pocket contacts. The similar ∆H‡TS and ∆S‡TS values for 2 and
4 suggest that their transition-state complexes are highly similar.
Thus, it seems that the side chains for 2 and 4 are being used
differently in the 17E8-catalyzed reactions.
One reason that enzymes are able to achieve differential
binding of specific groups on substrates is because their active
sites are often more complementary to the transition state than
to the Michaelis complex.2,4,28,31,45 The existence of a linear
correlation between ln(kcat/KM) and ln KI, a lack of a correlation
between ln KM and ln KI for alternative substrates, and transition-
state analogues with several enzymes is proof of this comple-
mol and ∆S‡ values of -32, -33, and -32 cal/(mol‚K) for
TS
1, 2, and 4, respectively. The curved ln(kcat/KM) vs 1/T plot yields
∆Cp‡ values of 620 and 170 cal/(mol‚K) for 2 and 4,
TS
respectively.
(36) We believe this ∆Cp value to be an overestimate. One reason is
the inherent difficulty in obtaining accurate heat-capacity changes from van’t
Hoff plots. Another reason it that the heat capacity of binding of the
phosphonates that correspond to 1, 2, and 4 obtained by calorimetric
experiments are much smaller (>2-fold) than the ∆CPKM and ∆Cp‡ obtained
by the temperature dependence of the kinetic constants. The heat capacity
changes for phosphonate binding are probably underestimates as charged
moieties contribute to heat capacity with opposite magnitude as do
hydrophobic moieties. The fit obtained from Figure 8 is not very sensitive
to the ∆Cp variable. Changing the ∆Cp from 700 to 300 cal/(mol‚K) had
a very small change on ∆HKM (7.0 and 6.8 kcal/mol) and ∆SKM (-6.4 and
-6.6 cal/(mol‚K)), respectively. The correlation coefficients are 0.995 and
0.971, respectively.
(37) The curvature of the ln KM-1/T plot could also arise because of a
shift in equilibrium between several (17E8-substrate) complexes or the
involvement of chemical rate constants in the Michaelis constant. The
involvement of several Michaelis complex forms seems unlikely due to
the absence of curvature seen in the ln KM-1/T plots with substrates 1 and
4. If there was an obligatory isomerization of the Michaelis complex for
17E8 catalysis, one would expect to see the same with other substrates that
are processed by the catalyst. The linear nature of the ln(kcat/T)-1/T plot is
also consistent with the nonexistence of several Michaelis-like complexes.
The “well-behaved” nature of the plot also suggests that there is no
equilibrium shift in the ground state for 17E8 catalysis.
(39) Ross, P. D.; Subramanian, S. Biochemistry 1981, 20, 3096-3102.
(40) Baldwin, R. L. Proc. Natl. Acad. Sci. U.S.A. 1986, 83, 8069-8072.
(41) Sturtevant, J. M. Proc. Natl. Acad. Sci. U.S.A. 1977, 74, 2236-
2240.
(42) Edsall, J. T. J. Am. Chem. Soc. 1935, 57, 1506-1507.
(43) In addition to the hydrophobic effect, it has been noted that the
existence of a heat capacity change may also result from contributions from
intramolecular vibrations, electrostatic charges, hydrogen bonds, confor-
mational entropy, and possibly changes in equilibria. The contribution from
electrostatic charges and hydrogen bonds is thought to be small due to the
small net change in electrostatic interactions and hydrogen bonds upon
binding. No correction has been made for conformational entropy and
internal vibrational contributions to the heat-capacity change. The likelihood
of possible changes in equilibria has been discussed (see footnote 37 and
ref 41).
(38) The data also can also be fitted to the standard Eyring equation (no
∆Cp‡ included). This yields ∆H‡ and ∆S‡ values of 6.3 kcal/mol and -32
cal/(mol‚K), respectively. The R value for the fit to the standard equation
is 0.993. The R value for the fit that includes a ∆Cp‡ contribution is 0.998.
(44) Creighton, T. E. Proteins: Structure and Molecular Properties, 2nd
ed.; W. H. Freeman and Company: New York, 1993.

Binding Energy and Antibody Reaction Coordinates
J. Am. Chem. Soc., Vol. 121, No. 51, 1999 11941
Figure 8. Postulated model of the side-chain-pocket interactions that result in uniform binding (substrate 2) and differential binding (substrate 4).
In uniform binding, all the side-chain-pocket interactions are made in the (IgG-S) complex and no additional interactions are formed in the
(IgG-S)‡. In differential binding, the side-chain-pocket interactions are formed exclusively in the (IgG-S)‡.
mentarity.28,29,46-49 We have shown in our system that the ln-
(kcat/KM) values of an extensive panel of substrates correlates
well with the binding of the corresponding phosphonates,
indicating that the side-chain-pocket interactions are similarly
presented by both the catalytic transition state and the phos-
phonates.9 The correlation between lnKM and lnKI in Figure 6
indicates that the side-chain-pocket interactions are similar in
the Michaelis complex and in the transition state. This correla-
tion suggests that part of 17E8’s active site is complementary
to both the ground state and the transition state. This precludes
the possibility that the side-chain-pocket can be used differ-
entially between the two species. The substrates 3 and 4 deviate
from the KM-KI correlation which is consistent with the side
chains binding differently in the ground and the transition states.
In addition to engineering catalytic efficiency in the form of
high kcat/KM values, an important concern in catalyst design is
high turnover (large kcat values). The properties of enzymes
indicate that high turnover can be obtained by binding nonre-
active portions of the substrate differentially between the ground-
and transition-state complexes. This points out a potentially
significant limitation of generating catalytic antibodies with
hapten molecules designed to mimic a transition state. There
may often be a subtle structural difference between the transition
state and the Michaelis complex for a chemical reaction that is
not effectively exploited with a synthetic transition-state mimic.
As a result, the antibodies generated against the transition-state
analogue also bind corresponding nonreactive portions of the
substrate with high affinity. Thus, haptens programs binding
interactions needed to stabilize the transition state, but do not
enforce the differential use of binding energy of nonreactive
portions of the substrate. This inevitably results in the unproduc-
tive stabilization of antibody-bound ground state species. Thus,
the development of new general strategies that can program or
select for differential binding is undoubtedly an essential step
for achieving truly efficient catalytic antibodies.
Acknowledgment. This study was supported by a grant from
the National Institutes of Health (GM 50672). T.S.S. is an Alfred
P. Sloan Research Fellow. H.W. is supported by a NSF
Predoctoral and an UNCF-Merck Initiative Dissertation Fel-
lowship. Mass spectral analyses of synthetic compounds was
provided by the UCSF Mass Spectrometry Facility (A.L.
Burlingame, Director) supported by the Biomedical Research
Technology Program of the National Center for Research
Resources, NIH, NCRR, BRTP 01614, and NIH NIEHS
ES04705. We gratefully acknowledge D.A. Agard and S.M.
Miller for critical review of the manuscript and useful discussions.
(45) Wolfenden, R. Molecular and Cellular Biochemistry 1974, 3, 207-
211.
(46) Bartlett, P.; Otake, A. J. Org. Chem. 1995, 60, 3107-3111.
(47) Bartlett, P. A.; Giangiordano, M. A. J. Org. Chem. 1996, 61, 3433-
3438.
(48) Phillips, M. A.; Kaplan, A. P.; Rutter, W. J.; Bartlett, P. A.
Biochemistry 1992, 31, 959-963.
(49) Bevington, P. R.; Robinson, D. K. Data Reduction and Error
Analysis for the Physical Sciences; 2nd ed.; McGraw-Hill: New York, 1992;
p 328.
JA9928879