Binding of hydrophobic hydroxamic acids enhances peroxidase's stereoselectivity in nonaqueous sulfoxidations

Article abstract of DOI:10.1021/ja012075o

Horseradish peroxidase exhibits a meager stereoselectivity (E) in the sulfoxidation of thioanisole (1a) in 99.8% (v/v) methanol. The E value, however, is greatly enhanced when the enzyme forms a complex with benzohydroxamic acid (2a). These findings are rationalized by means of molecular dynamics simulations and energy minimization which correctly explain (i) why the free enzyme is not stereoselective, (ii) why 2a inhibits peroxidase-catalyzed sulfoxidation of 1a but the enzymatic formation of one enantiomer of the sulfoxide product is inhibited much more than that of the other, thereby raising peroxidase's E, and (iii) why in the presence of 2a the enzyme favors production of the S sulfoxide of 1a. The generality of the observed ligand-induced stereoselectivity enhancement is demonstrated with other hydrophobic hydroxamic acids, as well as with additional thioether substrates.

Full text of DOI:10.1021/ja012075o

Published on Web 01/11/2002
Binding of Hydrophobic Hydroxamic Acids Enhances
Peroxidase’s Stereoselectivity in Nonaqueous Sulfoxidations
Prasanta Kumar Das, Jose M. M. Caaveiro, Susana Luque,† and
Alexander M. Klibanov*
Contribution from the Department of Chemistry, Massachusetts Institute of Technology,
Cambridge, Massachusetts 02139
Received August 30, 2001
Abstract: Horseradish peroxidase exhibits a meager stereoselectivity (E) in the sulfoxidation of thioanisole
(1a) in 99.8% (v/v) methanol. The E value, however, is greatly enhanced when the enzyme forms a complex
with benzohydroxamic acid (2a). These findings are rationalized by means of molecular dynamics simulations
and energy minimization which correctly explain (i) why the free enzyme is not stereoselective, (ii) why 2a
inhibits peroxidase-catalyzed sulfoxidation of 1a but the enzymatic formation of one enantiomer of the
sulfoxide product is inhibited much more than that of the other, thereby raising peroxidase’s E, and (iii)
why in the presence of 2a the enzyme favors production of the S sulfoxide of 1a. The generality of the
observed ligand-induced stereoselectivity enhancement is demonstrated with other hydrophobic hydroxamic
acids, as well as with additional thioether substrates.
insufficient stereoselectivity.4 In this work, we demonstrate that
complexation of horseradish peroxidase (HRP) with hydrophobic
Introduction
Improving the stereoselectivity of a given enzyme toward a
desired substrate is one of the most challenging but practically
important goals in the area of biocatalysis.1 Conducting enzyme-
catalyzed synthetic transformations in organic solvents,2 instead
of conventional aqueous reaction media, offers unprecedented
opportunities in this regard.3 Namely, enzymatic stereoselectivity
has been found to be markedly affected by the solvent and by
the history/formulation of the enzyme sample.3 In the present
study, we have proposed and validated another, independent
approach to that goal.
hydroxamic acids can greatly enhance its stereoselectivity.
Moreover, this effect can be correctly rationalized by means of
structure-based molecular modeling of the enzyme‚substrate and
enzyme‚substrate‚ligand complexes.
Results and Discussion
Peroxidase-catalyzed sulfoxidations and other asymmetric
transformations can be carried out not only in water4 but
also, and profitably so, in nearly anhydrous organic solvents.7
As the initial model reaction, herein we investigated the
oxidation of thioanisole (1a) catalyzed by lyophilized HRP in
Peroxidases are a versatile family of redox enzymes capable
of catalyzing several asymmetric processes, such as oxidation
of prochiral thioethers.4 The products of this latter process, chiral
sulfoxides, are valuable bioactives and synthons.5 For example,
the best-selling drug of all time, anti-ulcer omeprazole (Prilosec/
Losec), is such a chiral sulfoxide whose single-enantiomer
version is nearing clinical use.6 Like other enzymatic conver-
sions, peroxidase-catalyzed sulfoxidations often suffer from
99.8% (v/v) methanol.7 Peroxidase-catalyzed sulfoxidation of
this prochiral substrate with H2O2 lacked stereopreference: the
stereoselectivity value E(S/R)8 was found to be 1.1 ( 0.1 (top
line in Table 1). The E(S/R) value was only slightly higher, 1.8
( 0.2, when an alternative oxidant, tert-butyl hydroperoxide
(t-BuOOH), was used (top line in Table 1).
* Corresponding author. E-mail: klibanov@mit.edu.
† Permanent address: Department of Chemical and Environmental
Engineering, University of Oviedo, 33071 Oviedo, Spain.
(1) Faber, K.; Ottolina, G.; Riva, S. Biocatalysis 1993, 8, 91-132. Margolin,
A. L. Enzyme Microb. Technol. 1993, 15, 266-280. Roberts, S. M.; Turner,
N. J.; Willets, N. J.; Turner, M. K. Introduction to Biocatalysis Using
Enzymes and Microorganisms; Cambridge University Press: New York,
1995. Faber, K. Biotransformations in Organic Chemistry, 4th ed.; Springer-
Verlag: Berlin, 2000.
Benzohydroxamic acid (2a) is known to bind in the vicinity
of the active site’s heme of HRP, as well as of other
(2) Carrea, G.; Riva, S. Angew. Chem. 2000, 33, 2226-2254.
(3) Klibanov, A. M. Nature 2001, 409, 241-246.
(4) Van Deurzen, M. P. J.; Van Rantwijk, F.; Sheldon, R. A. Tetrahedron 1997,
53, 13183-13220. Colonna, S.; Gaggero, N.; Richelmi, C.; Pasta, P. Trends
Biotehchnol. 1999, 17, 163-168. Adam, W.; Lazarus, M.; Saha-Mo¨ller,
C. R.; Weichold, O.; Hoch, U.; Ha¨ring, D.; Schreier, P. AdV. Biochem.
Eng./Biotechnol. 1999, 63, 73-108.
(7) Dai, L.; Klibanov, A. M. Biotechnol. Bioeng. 2000, 70, 353-357. Xie, Y.;
Das, P. K.; Klibanov, A. M. Biotechnol. Lett. 2001, 23, 1451-1454.
(8) Defined as the ratio of the initial rate for the production of the S sulfoxide
to that of the R sulfoxide in the same reaction mixture. See: Chen, C.-S.;
Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294-
7299. Straathof, A. J. J.; Jongejan, J. A. Enzyme Microb. Technol. 1997,
21, 559-571.
(5) Carreno, M. C. Chem. ReV. 1995, 95, 1717-1760.
(6) Cannarsa, M. J. Chim. Oggi/Chem. Today 1999, 17 (9), 28-32.
9
782 VOL. 124, NO. 5, 2002 J. AM. CHEM. SOC.
10.1021/ja012075o CCC: $22.00 © 2002 American Chemical Society

Stereoselectivity of Peroxidase Sulfoxidations
A R T I C L E S
Table 1. Initial Rates and Stereoselectivities E(S/R) of the Asymmetric Sulfoxidation of Thioanisole (1a) Catalyzed by HRP in the Presence
of Varying Concentrations of Benzohydroxamic Acid (2a) in 99.8% Methanola
initial rate of product formation,b µM/min
with H O
with t-BuOOH
R sulfoxide
2
2
[2a], mM
S sulfoxide
R sulfoxide
E(S/R)8
S sulfoxide
E(S/R)8
0
4.0 ( 0.4
3.5 ( 0.2
1.1 ( 0.1
1.4 ( 0.1
1.9 ( 0.1
2.6 ( 0.1
3.1 ( 0.3
3.6 ( 0.2
4.0 ( 0.3
2.4 ( 0.3
1.3 ( 0.2
1.8 ( 0.2
2.4 ( 0.2
3.4 ( 0.4
4.6 ( 0.4c
5.0 ( 0.7
5.9 ( 0.6
0.5
1.0
2.0c
4.0
7.0
10
1.5 ( 0.1
1.1 ( 0.1
1.1 ( 0.1
0.45 ( 0.02
0.14 ( 0.02
0.10 ( 0.01
0.076 ( 0.012
0.046 ( 0.005
0.77 ( 0.06
0.49 ( 0.02
0.40 ( 0.04
0.29 ( 0.02
0.24 ( 0.01
0.41 ( 0.03
0.19 ( 0.01
0.13 ( 0.01
0.080 ( 0.004
0.060 ( 0.006
0.48 ( 0.04
0.46 ( 0.03
0.38 ( 0.05
0.27 ( 0.03
0.21 ( 0.01
<0.028
>7.1
a Lyophilized enzyme powders were suspended in methanol (0.2% water) containing 0.5 mM 1a, 1 mM peroxide, and 2a. Each reaction mixture (1
mg/mL of HRP in the case of H2O2, and 10 mg/mL in the case of t-BuOOH) was vigorously stirred at room temperature; periodically, aliquots were
withdrawn and assayed by chiral HPLC, as described in the Methods section. b All initial reaction rates were calculated by the least-squares method, with
the mean and standard error values shown in the table. c For HRP lyophilized from its aqueous solution containing 1 mM 2a (second paragraph of Results
and Discussion) and then suspended in methanol following our experimental protocol, the calculated ligand concentration in that solvent was 2 mM. However,
when 2 mM 2a was directly added to the methanol suspension of HRP lyophilized in the absence of ligand, the E(S/R) value obtained, 4.6 ( 0.4, was below
that in the former case (>7.5). This discrepancy suggests that some of the ligand co-lyophilized with HRP becomes trapped in the enzyme sample and thus
is not released into the solvent when the lyophilized powder is suspended in methanol.
oxidoreductases, and inhibit the enzyme.9-11 We decided to
explore whether the extent of this inhibition would be identical
for the enzymatic formation of S and R sulfoxides from 1a. To
this end, HRP (5 mg/mL) and 2a (0.75 mM) were dissolved in
a buffered (pH 7.0) aqueous solution, followed by lyophilization.
The resultant enzyme‚2a complex was assayed in the oxidation
of 1a with t-BuOOH in 99.8% methanol under the same
conditions as the free enzyme above. It was found that while
the complexation with the ligand lowered the initial rate of the
peroxidase-catalyzed production of the S sulfoxide 5.2-fold, that
for the R enantiomer plummeted 17-fold. Consequently, the
stereoselectivity more than trippledsfrom 1.8 ( 0.2 to 5.7 (
0.6. The E(S/R) value increased with the concentration of 2a in
the aqueous solution of the enzyme prior to lyophilization: for
example, it was merely 3.4 ( 0.2 at the 0.3 mM ligand but
exceeded 7.5 for the 1 mM ligand.
will also unequally affect the rates of S and R sulfoxidations of
1a, and hence raise stereoselectivity, in a manner similar to that
observed in the aforementioned experiments with the preformed,
lyophilized enzyme‚ligand complex. Inspection of Table 1
indicates that, whether with H2O2 or t-BuOOH as an oxidant,
2a added to a mixture of HRP and 1a in methanol significantly
inhibits the formation of both S and R sulfoxides. Furthermore,
in all cases the ligand slows down the production of the R
enantiomer far more than the S, thereby giving rise to an
enhanced stereoselectivity. At the highest 2a concentration used,
10 mM, the E(S/R) value grew from 1.1 ( 0.1 to 4.0 ( 0.3 for
H2O2 and from 1.8 ( 0.2 to >7.1 for t-BuOOH (Table 1), i.e.,
from un- or barely detectable to quite substantial.
To explain the foregoing observations mechanistically, we
employed the means of molecular modeling involving molecular
dynamics simulations and energy minimization.13 Upon reaction
with a peroxide, the ferric heme state of peroxidase is converted
to the catalytically active compound I, an FedO oxyferryl
species, which is two oxidation states above the resting state;14
in the sulfoxidation of thioanisole, the iron-bound oxygen
atom is subsequently transferred to the substrate’s sulfur
atom.15
(9) Schonbaum, G. R. J. Biol. Chem. 1973, 248, 502-511. Maltempo, M. M.;
Ohlsson, P. L.; Paul, K. G.; Petersson, L.; Ehrenberg, A. Biochemistry 1979,
18, 2935-2941. Aviram, I. Arch. Biochem. Biophys. 1981, 212, 483-490.
Kitagawa, T.; Hashimoto, S.; Teraoka, J.; Nakamura, S.; Yajima, H.;
Toichiro, H. Biochemistry 1983, 22, 2788-2792. Sakurada, J.; Takahashi,
S.; Hosoya, T. J. Biol. Chem. 1986, 261, 9657-9662. Smulevich, G.;
English, A.; Mantini, A. R.; Marzocchi, M. P. Biochemistry 1991, 30, 772-
779. LaMar, G. N.; Herna´ndez, G.; de Ropp, J. S. Biochemistry 1992, 31,
9158-9168. Veitch, N. C. Biochem. Soc. Trans. 1995, 23, 232-240.
Itakura, H.; Oda, Y.; Fukuyama, K. FEBS Lett. 1997, 412, 107-110. de
Ropp, J. S.; Mandal, P.; Brauer, S. L.; LaMar, G. N. J. Am. Chem. Soc.
1997, 119, 4732-4739. Smulevich, G.; Feis, A.; Indiani, C.; Becucci, M.;
Marzocchi, M. P. J. Biol. Inorg. Chem. 1999, 4, 39-47. Indiani, C.; Feis,
A.; Howes, B. D.; Marzocchi, M. P.; Smulevich, G. J. Am. Chem. Soc.
2000, 122, 7368-7376.
(10) Henriksen, A.; Schuller, D. J.; Meno, K.; Welinder, K. G.; Smith, A. T.;
Gajhede, M. Biochemistry 1998, 37, 8054-8060.
(11) Chang, Y.-T.; Veitch, N. C.; Loew, G. H. J. Am. Chem. Soc. 1998, 120,
5168-5178.
We found that, as in aqueous solution,9 2a is a potent inhibitor
of peroxidase’s natural reaction (phenol oxidation with H2O2)12
in 99.8% methanol. For instance, addition of 0.5 mM 2a to the
reaction mixture cut the initial rate of the enzymatic oxidation
of 0.5 mM o-methoxyphenol with 0.1 mM H2O2 some 8-fold
in that organic solvent. Therefore, we examined whether a direct
addition of 2a to lyophilized HRP suspended in 99.8% methanol
(12) Saunders: B. C.; Holmes-Siedle, A. G.; Stark, B. P. Peroxidase; Butter-
worth: Washington, DC, 1964. Everse, J.; Everse, K. E.; Grisham, M. B.,
Eds. Peroxidases in Chemistry and Biology; CRC Press: Boca Raton, FL,
1991.
(13) Ke, T.; Klibanov, A. M. J. Am. Chem. Soc. 1999, 121, 3334-3340. Shin,
J.-S.; Luque, S.; Klibanov, A. M. Biotechnol. Bioeng. 2000, 69, 577-583.
(14) Harris, D. L.; Loew, G. H. J. Am. Chem. Soc. 1996, 118, 10588-10594.
(15) Kobayashi, S.; Nakano, M.; Kimura, T.; Schaap, A. P. Biochemistry 1987,
26, 5019-5022.
9
J. AM. CHEM. SOC. VOL. 124, NO. 5, 2002 783

A R T I C L E S
Das et al.
Figure 1. Molecular models of the binary complex of HRP with the substrate 1a (A) and of the ternary complex of HRP with 1a and the ligand 2a (B).
The depicted oxyferryl form of the enzyme is capable of transferring its heme-bonded oxygen atom to the 1a’s sulfur atom to form a chiral sulfoxide. The
molecular models were built on the basis of the X-ray crystal structures10,16 of HRP and its complexes with 2a and with the natural substrate ferulic acid and
cyanide, as described in the Methods section. Representations: a gray ribbon, the protein backbone; black/gray/white sticks, amino acid side chains; golden
sticks, the heme; a pink ball, the oxyferryl-heme’s oxygen; blue sticks, bound 1a whose sulfur atom and its two lone pairs of electrons are denoted by a red
ball and red sticks, respectively; and green sticks, bound 2a. For clarity, (i) only the portion of the protein backbone situated within a 15-Å sphere with the
center at the substrate’s sulfur atom and (ii) the side chains situated within 7 Å from either the sulfur or oxygen atoms (red and pink balls, respectively) are
shown.
The starting structure for the modeling of the enzyme‚1a
complex was, therefore, approximated by the recently solved
X-ray crystal structure of the ternary complex of HRP with the
natural substrate ferulic acid (which we replaced by 1a) and
cyanide (replaced by oxygen).16 This starting structure was then
subjected to molecular dynamics simulations and energy
minimization, and the lowest-energy structure thus obtained was
chosen as that of the catalytically relevant oxyferryl-peroxidase‚
1a complex.
Figure 1A depicts the extended active site region of the
resultant lowest-energy structure. The protein backbone is re-
presented by a ribbon, the heme by golden sticks, the oxyferryl’s
oxygen atom by a pink ball, 1a by blue sticks, and 1a’s sulfur
atom and its two lone electron pairs by red ball-and-sticks,
respectively. In addition, all amino acid residues’ side chains
located within 7 Å from either the oxyferryl’s oxygen or 1a’s
sulfur are represented by black/white/gray sticks (note that the
impression that the benzene ring of Phe41 in the center of Figure
1A is in close proximity of the substrate’s S atom is an optical
illusion; in actuality, it is way behind the plane of the figure
with its closest atom being 3.7 Å away from the sulfur).
In the enzymatic sulfoxidation, the oxyferryl-heme’s O is
transferred to the bound 1a’s sulfur to become a co-owner of
one of its lone pairs of electrons. If the latter denoted by a red
stick pointing southwest in Figure 1A is shared by the oxygen
atom, then application of the Cahn-Ingold-Prelog convention17
yields the S enantiomer of the sulfoxide product. Conversely,
if the red stick pointing southeast is engaged, the R enantiomer
ensues. It is apparent from Figure 1A that the sulfur’s two lone
pairs are similarly accessible to the attacking oxygen atom.
Therefore, one would expect comparable concentrations of the
S and R sulfoxides, i.e., no appreciable stereoselectivity. This
was indeed observed experimentally, with the E(S/R) values
being in the 1 to 2 range depending on the oxidant used (top
line of Table 1).
On the basis of kinetic studies, 2a is generally regarded as a
competitive inhibitor of HRP.9 This view is supported by X-ray
crystal structure data: the position occupied by 2a in its binary
complex10 with the enzyme coincides with that occupied by the
natural substrate ferulic acid.16 However, our observations
indicate that at least in the case of enzymatic sulfoxidations
this cannot be the whole picture because 2a should then inhibit
the formation of S and R sulfoxides of 1a to the same extent.
As seen in Table 1, that is clearly not so, with the R
sulfoxidations being inhibited much more than the S.
A recent theoretical study11 of binding modes of 2a in the
cyanide-ligated (i.e., oxyferryl-mimicking) form of HRP re-
vealed the existence of several secondary ligand binding sites,
in addition to the main one coinciding with that for the substrate.
Therefore, we set out to build a ternary complex of the oxyferryl
form of the enzyme with 1a and 2a, whereby the former
occupies the previously identified hydrophobic substrate binding
site16 (to allow the enzymatic sulfoxidation to occur) and the
ligand occupies an optimal secondary, nonoverlapping binding
site. To this end, molecular dynamics simulations and energy
minimization were performed to derive the lowest-energy
conformer of such a ternary complex. In the resultant structure
(Figure 1B), the 2a molecule (represented by green sticks) was
found to bind to a hydrophobic pocket located at the edge of
the heme-containing cavity, a few angstroms away from the
bound 1a molecule; this binding mode is additionally stabilized
(16) Henriksen, A.; Smith, A. T.; Gajhede, M. J. Biol. Chem. 1999, 274, 35005-
35011.
(17) Carey, F. A.; Sundberg, R. J. AdVanced Organic Chemistry, 3rd ed.; Plenum
Press: New York, 1990; Part A, pp 71-73.
9
784 J. AM. CHEM. SOC. VOL. 124, NO. 5, 2002

Stereoselectivity of Peroxidase Sulfoxidations
A R T I C L E S
Table 2. Initial Rates and Stereoselectivties E(S/R) of the
Asymmetric Sulfoxidation of Thioanisole (1a) with t-BuOOH
Catalyzed by HRP in 99.8% Methanol in the Presence of Different
Hydroxamic Acid Ligandsa
by a hydrogen bond between 2a’s carbonyl oxygen and peptide
bond’s hydrogen of Gly69.18
Comparison of Figures 1A and 1B shows that 1a in the binary
complex is oriented differently from that in the ternary complex,
i.e., that binding of 2a in the secondary site profoundly alters
the orientation of the substrate in the primary site. In particular,
while 1a’s sulfur atom in the binary HRP‚substrate complex
faces the oxyferryl’s oxygen atom (Figure 1A), in the ternary
complex it is turned away from it (Figure 1B), thus making
their reaction more problematic. This difference may be partially
responsible for the observed inhibition of the peroxidase-
catalyzed sulfoxidation of 1a in the presence of 2a (Table 1).
One can also see in Figure 1B that the two lone pairs of
electrons of the substrate’s sulfur atom are no longer equally
exposed to the attacking oxygen atom (as they are in Figure
1A). Namely, the one represented by the red stick pointing
northwest is much less accessible than its counterpart pointing
northeast. This explains why the peroxidase‚2a complex is
stereoselective, whereas the free enzyme is essentially not (Table
1). Finally, the use of the Cahn-Ingold-Prelog convention17
reveals that when the attacking oxygen engages the more
favorably oriented lone pair, the resultant sulfoxide has the S
configuration. In other words, our molecular modeling analysis
correctly predicts not only that the enzyme‚ligand complex
should be stereoselective but also the absolute configuration of
the predominant product.
initial rate of product
formation,b µM/min
ligand
[ligand], mM
S sulfoxide
R sulfoxide
E(S/R)8
none
2a
2b
2c
2d
2e
2.4 ( 0.3
1.3 ( 0.2
1.8 ( 0.2
5.9 ( 0.6
5.7 ( 0.3
5.7 ( 0.3
4.3 ( 0.6
9.7 ( 1.0
2.5 ( 0.1
4.0 ( 0.1
7.0
1.0
1.0
7.0
16
0.27 ( 0.03
0.38 ( 0.02
0.33 ( 0.02
0.12 ( 0.02
0.32 ( 0.04
0.13 ( 0.01
0.97 ( 0.02
0.046 ( 0.005
0.066 ( 0.004
0.058 ( 0.003
0.028 ( 0.003
0.033 ( 0.002
0.052 ( 0.002
0.24 ( 0.01
2f
2g
10
7.0
a Lyophilized enzyme powders were suspended (at 10 mg/mL) in
methanol (0.2% water) containing 0.5 mM thioether, 1 mM t-BuOOH,
and a ligand at the highest possible concentration that still allowed
measurement of the initial rate of the enzymatic production of the slower
(R) enantiomer, except for 2g where the ligand overlaps with the S sulfoxide
in the HPLC chromatogram. The reaction mixtures were vigorously stirred
at room temperature; periodically, aliquots were withdrawn and assayed
by chiral HPLC as described in the Methods section. b See footnote b to
Table 1.
The affinity of 2a toward the primary binding locus in the
active site of HRP is due to a combination of hydrophobic
interactions of the ligand’s benzene ring and hydrogen bonds
formed by the ligand’s hydroxamic moiety.9,10 The same holds
true for secondary binding loci (ref 11 and above). Thus other
hydrophobic hydroxamic acids should be able to play a role
akin to 2a’s. We verified this hypothesis experimentally by
examining the influence of various hydroxamic acids on the
initial rates and stereoselectivities of the HRP-catalyzed sul-
foxidation of 1a with t-BuOOH in 99.8% methanol. One can
see in Table 2 that all three ring-substituted benzohydroxamic
acids tested, 2b, 2c, and 2d, inhibit the enzymatic formation of
the R sulfoxide of 1a much stronger than that of its S enantiomer,
thus significantly increasing the stereoselectivity.8 That 2e with
its bulky adamantane moiety is also quite potent in raising the
E(S/R) value (sixth line in Table 2) suggests that the secondary
ligand binding site (Figure 1B) is spacious and that aromatic
π-π interactions play no role. However, it is important for a
stereoselectively boosting ligand to have a hydrophobic moiety;
as seen in the penultimate line in Table 2, 2f is virtually
ineffective. Finally, having the unsubstituted hydroxamic moiety
is preferable, presumably due to its greater propensity to engage
in hydrogen bonding with the enzyme, for 2g is less effective
than 2a at the same concentration (Table 2).
Figure 2. Stereoselectivities of HRP, in the absence and in the presence
of 2a, in the sulfoxidations of various aromatic thioethers in 99.8% methanol.
Note that the E(S/R)8 value of 1 (the X-axis) means no stereopreference.
The error bars correspond to the calculated standard errors. Experimental
conditions: lyophilized peroxidase powder was suspended (at 10 mg/mL)
in methanol containing 0.2% water (v/v), as well as the corresponding 0.5
mM thioethers (1 mM in the case of 1c), 1 mM t-BuOOH (H2O2 in the
case of 1c), and 2a in the highest concentration that still allowed accurate
measurement of the initial rates of the enzymatic production of the slower,
R product enantiomer (1.0, 0.5, and 4.0 mM for 1b, 1c, and 1d, respectively).
The reaction mixtures were vigorously stirred at room temperature;
periodically, aliquots were withdrawn and assayed by chiral HPLC. For
other conditions, see the Methods section.
three additional thioethers, 1b, 1c, and 1d, as sulfoxidation
substrates in 99.8% methanol. In all instances, the free enzyme
exhibited no appreciable stereoselectivity (E values in the 1.0-
1.5 range). And yet, for all three prochiral substrates (Figure
2), as for 1a (Table 1), in the presence of 2a the enzyme was
unequivocally stereoselective.
In closing, we discovered that complex formation of HRP
with hydrophobic hydroxamic acids strongly enhances the
stereoselectivity of enzymatic sulfoxidations. This effect was
comprehensively rationalized by means of structure-based
molecular modeling, providing new insights into the much-
researched mechanism of ligand binding to the active site of
peroxidase.
To ascertain the generality of the observed ligand-induced
enhancement of peroxidase’s stereoselectivity, we examined
(18) It is worth mentioning that when our molecular dynamics and energy
minimization approach was used to model the structures of the binary
complexes of oxyferryl-HRP with 1a (Figure 1A) and with 2a (not shown),
the substrate and the ligand were found to be located in essentially the
same site. This result is in agreement with the literature X-ray crystal-
lographic data,10,16 thereby validating our molecular modeling methodology.
The secondary binding site occupied by 2a in the ternary complex (Figure
1B) is not as hospitable energetically as the main binding site. However,
when the latter is occupied by 1a (which is required for the peroxidase-
catalyzed sulfoxidation), that secondary binding site apparently becomes
the next best option, and the resultant ternary complex (Figure 1B) was
found to have the lowest energy structure.
9
J. AM. CHEM. SOC. VOL. 124, NO. 5, 2002 785

A R T I C L E S
Das et al.
the basis of the published X-ray crystal structures10,16 of HRP and its
complexes by using a two-step procedure described below, similar to
that13 previously employed by us with hydrolases.
Materials and Methods
Materials. HRP (type II, EC 1.11.1.7) was purchased from Sigma
Chemical Co. The thioether substrates 1a-d, the ligands 2a-g, guaiacol
(o-methoxyphenol), and t-BuOOH (70% aqueous solution) were
obtained from Aldrich Chemical Co. Methanol (dried by us over 3-Å
molecular sieves prior to use) and H2O2 (30% aqueous solution) were
from Mallinckrodt, and hexane and isopropyl alcohol were from EM
Science. All other chemicals and solvents used were purchased from
Aldrich Chemical Co., were of the highest purity available, and were
used without further purification.
The 1a and 2a moieties were constructed by using standard molecular
fragments provided within the INSIGHT II software (Accelrys, Prin-
ceton, NJ). The structure of HRP in its complex with ferulic acid and
cyanide16 was obtained by retrieving the heavy atom coordinates (entry
7ATJ) from the Brookhaven Protein Data Bank.21 The ferulic acid
moiety was excised, and the heme’s iron-bound cyanide was replaced
by an oxygen atom, thus converting the enzyme to the catalytically
active compound I (an FedO oxyferryl species). Since solvent
molecules and counterions were not included in the simulations, 1a,
2a, all protein residues, and the oxyferryl-heme were modeled in their
un-ionized form. The coordinates of the added hydrogens were
generated according to idealized bond lengths and valence angles by
using the Builder module of INSIGHT II. The CVFF force field22
provided within the FDISCOVER program (Accelrys) was used for
potential and charge assignments of 1a, 2a, and all HRP’s amino acid
residues. The force field parameters and charges for the protoporphyrin
IX moiety of the heme were taken from a previous cytochrome c study23
and ab initio calculations24 on model systems of cytochrome P450,
respectively. The FedO bond distance was taken to be 1.78 Å, and a
force constant of 300 kcal/mol was used.25 The charges for the oxyferryl
moiety were obtained from an ab initio CASSCF calculation on high-
valent iron-oxo-porphyrins.26
To model the oxyferryl-peroxidase‚1a complex, the 1a molecule was
placed in the substrate binding site of HRP, which in the X-ray crystal
structure16 is occupied by ferulic acid. First, this initial structure was
energy-minimized to relieve any overlay-strained artifacts using the
steepest descent method for 500 iterations or until the RMS gradient
fell below 0.1 kcal/Å, followed by the conjugate-gradient minimization
until the maximum derivative dropped below 0.01 kcal/Å. For
subsequent calculations, the sulfur-bonded carbon of 1a’s phenyl ring
was tethered by using a harmonic potential (see below); this carbon
atom was selected because its coordinates virtually coincide with the
centers of both mass and geometry of the substrate. The purpose of
the tethering was to allow widely different conformations to be explored
while preventing the substrate from diffusing too far from the enzyme.
The energy-minimized structure was then subjected to 1000 steps of
molecular dynamics simulation at 900 K using the Verlet leapfrog
integrator with a step size of 1 fs and a template harmonic force of 2
kcal/Å. After each simulated ps, the resulting structure was energy-
minimized as outlined above, except that the maximum derivative for
the conjugate gradient method was set to be less than 0.005 kcal/Å
and a template forcing constant of 0.2 kcal/Å was used. The coordinates
of the minimized structure were then saved, and its energy was
calculated. This cycle was repeated until 100 energy-minimized
structures were obtained; four lowest energy ones of them were chosen
for further analysis.
Enzyme Preparation. HRP was dissolved (5 mg/mL) in a 50 mM
aqueous phosphate buffer (pH 7.0); then the solution was frozen with
liquid N2 and lyophilized for 48 h in a Labconco freeze-drier (-50
°C, 40-60 µm Hg). When the enzyme was lyophilized in the presence
of 2a, the latter was added to the buffered solution at a desired
concentration, and the pH was adjusted to 7.0 before and after the
addition of enzyme. The lyophilized enzyme powder was suspended
in 99.8% (v/v) methanol containing a thioether, a peroxide, and a ligand
(if any), and briefly sonicated.
Kinetic Measurements. The initial rates of the enzymatic sulfoxi-
dation of 1a and other thioethers with peroxides were followed by
isocratic high-performance liquid chromatography (HPLC) using a
Chiralcel OD-H column (4.6 mm i.d. x 250 mm). Chiral HPLC
conditions (the mobile phase and flow rate) for the separation of the
enantiomers of the sulfoxides produced from 1a-d and the retention
times for the R and S enantiomers were respectively as follows: (1a)
90:10 (henceforth, all such ratios are v/v) hexane/isopropyl alcohol,
0.5 mL/min, 23.0 and 30.0 min; (1b) 90:10 hexane/isopropyl alcohol,
0.7 mL/min, 30.2 and 38.1 min; (1c) 92:8:0.1 hexane/ethanol/trifluo-
roacetic acid, 0.5 mL/min, 45.0 and 52.0 min; and (1d) 92:8 hexane/
isopropyl alcohol, 0.6 mL/min, 28.5 and 34.0 min. In the presence of
certain hydroxamic acid ligands, HPLC conditions/retention times were
distinct from those listed above for methyl phenyl sulfoxide enanti-
omers: (2a, 2d-2f) 90:10 hexane/isopropyl alcohol, 0.5 mL/min, 23.0
and 30.0 min; (2b, 2c) 95:5 hexane/isopropyl alcohol, 0.8 mL/min,
24.2 and 32.8 min; and (2g) 96:4 hexane/isopropyl alcohol, 1.1 mL/
min, 22.1 and 31.0 min. The elution order for the R and S enantiomers
of all sulfoxides was assumed to be that previously established7 and
confirmed19 for their homologue methyl phenyl sulfoxide. The absor-
bance of the effluent was monitored at 242 nm for methyl phenyl
sulfoxide and 254 nm for all other sulfoxides.
The initial sulfoxidation rates were typically measured as follows:
1 mL of 99.8% methanol containing 0.5 mM thioether (1 mM in the
case of 1c), 1 mM peroxide (2 µL of 0.5 M in water), and a desired
concentration of hydroxamic acids was added to lyophilized HRP (1
mg/mL in the case of H2O2 and 10 mg/mL in the case of t-BuOOH;
note that “mg” refers to the enzymic protein, i.e., after the calculated
weight of the buffer salt and the ligand has been subtracted from the
overall weight of the powder). The mixture was vigorously stirred at
room temperature; periodically, 25-µL aliquots were withdrawn and
centrifuged, 5 µL of the supernatant was analyzed by chiral HPLC,
and the sulfoxide product concentration was plotted as a function of
the enzymatic reaction time.
Second, each of the four lowest energy conformers thus identified
was used as an initial model for a new set of dynamics/minimization
calculations. The procedure was the same as described in the preceding
paragraph, except that the initial optimization was omitted. The cycle
was repeated until 125 structures of each lowest energy conformer were
obtained; the resultant total of 500 energy-minimized structures was
Peroxidase-catalyzed oxidation of guaiacol with H2O2 in 99.8%
methanol was monitored spectrophotometrically at 436 nm as previously
described.20 The enzyme (0.05 mg/mL) was suspended directly in a
spectrophotometric cuvette containing 3 mL of a reaction mixture
consisting of 0.5 mM guaiacol, 0.1 mM H2O2, and a desired concentra-
tion of 2a at room temperature.
(21) Berman, H. M.; Westbrook, J.; Feng, Z.; Gilliand, G.; Bhat, T. N.; Weissig,
H.; Shindyalov, I. N.; Bourne, P. E. Nucleic Acid Res. 2000, 28, 235-
242.
(22) Dauber-Osguthorpe, P.; Roberts, V. A.; Osguthorpe, D. J.; Wolff, J.; Genest,
M.; Hagler, A. T. Proteins 1988, 4, 31-47.
(23) Laberge, M.; Vanderkooi, J. M.; Sharp, K. A. J. Phys. Chem. 1996, 100,
10793-10801.
Molecular Modeling. Molecular models of the binary HRP‚1a and
HRP‚2a complexes and the ternary HRP‚1a‚2a complex were built on
(24) de Groot, M. J.; Havenith, R. W. A.; Vinkers, H. M.; Zwaans, R.;
Vermeulen, N. P. E.; van Lenthe, J. H. J. Comput.-Aided Mol. Des. 1998,
12, 183-193.
(19) Colonna, S.; Gaggero, N.; Carrea, G.; Pasta, P. J. Chem. Soc., Chem.
(25) Paulsen, M. D.; Ornstein, R. L. J. Comput.-Aided Mol. Des. 1992, 6, 449-
Commun. 1992, 357-358.
460.
(20) Dai, L.; Klibanov, A. M. Proc. Natl. Acad. Sci. U.S.A. 1999, 96, 9475-
(26) Yamamoto, S.; Teraoka, J.; Kashiwagi, H. J. Chem. Phys. 1988, 88, 303-
312.
9478.
9
786 J. AM. CHEM. SOC. VOL. 124, NO. 5, 2002

Stereoselectivity of Peroxidase Sulfoxidations
A R T I C L E S
collected, and their energies were calculated. Finally, the single lowest
energy conformer obtained overall was chosen as that representing the
catalytically relevant oxyferryl-HRP‚1a complex (Figure 1A).
During all simulations, nonbonded interactions were evaluated with
a group-based switching function between 23 and 25 Å, and the
nonbonded pair list was updated every 10 time steps. Only the atoms
of the substrate, of the oxyferryl-heme, and of the amino acid residues
within HRP’s heme cavity (Arg38, Phe41, His42, Phe68, Gly69, Asn70,
Ala71, Asn72, Ser73, Leu138, Pro139, Ala140, Pro141, Phe142,
Phe143, and Phe179) were allowed to move. Neither cross-terms nor
Morse functions were used, and the dielectric constant of the medium
was set to 4.0.
manually docked into both aforementioned secondary binding sites;
the coordinates of the two structures thus generated were saved and
used as the input for the subsequent simulations. The two-step procedure
used here was analogous to that described for the binary enzyme‚1a
complex, except that 2a and the amino acid residues of the enzyme
that surround both binding sites were also allowed to move during the
simulations. The lowest energy conformer for each secondary binding
site was identified, and it was found that the conformation of the ternary
complex with 2a neighboring the main binding site was 2.0 kcal/mol
more stable than the other.11 Therefore, the structure in which 2a is
bound to the enzyme in the vicinity of 1a was chosen as the catalytically
relevant conformation of the ternary HRP‚1a‚2a complex (Figure 1B).
To model the ternary HRP‚1a‚2a complex, we had to identify where
2a is bound within the enzyme when the main binding site is occupied
by 1a. A previous theoretical study of binding modes of 2a in HRP11
has shown that, in addition to the main substrate binding site,10,16 2a
can also be accommodated in several other loci regarded as secondary
binding sites. The first such putative secondary binding site11 is located
at the proximal side of the heme more than 10 Å from the main binding
site. In this location, 2a is surrounded by Phe68, Pro141, Phe142,
Arg178, Phe179, Ile180, Met181, Asp182, Phe187, Ser188, Ile244, and
Gln245. The other secondary binding site was identified during our
analysis of the molecular dynamics trajectories of the binary HRP‚1a
complex, whereby it was observed that 1a is transiently accommodated
in a pocket adjacent to the main binding site. When 2a is docked in
that secondary site, it is surrounded by 1a and Asp66, Ala67, Phe68,
Gly69, Asn70, Ala71, Asn72, Ser73, Pro141, Gln176, Arg178, and
Phe179 residues of the enzyme.
Finally, to verify our computational techniques, we applied them to
model the structure of the previously solved10 HRP‚2a complex. To
that end, 2a was placed in three different locations within the free
enzyme: the main substrate binding site, and the two secondary binding
sites used for the modeling of the ternary complex. The three initial
structures generated were saved and employed as the input for
subsequent simulations by using the same procedure as that described
for modeling the ternary complex. When the lowest energy conformers
of each simulation were selected, they revealed that the energetically
preferred binding site for 2a in HRP is the main binding site, which is
indeed the one experimentally identified by the X-ray crystallographic
analysis.10
Acknowledgment. This study was financially supported by
grant no. BES-9712497 from the National Science Foundation.
J.M.M.C. is a recipient of a postdoctoral fellowship from the
Spanish Ministry of Education, Culture, and Sports.
The coordinates of the lowest energy conformer obtained from the
simulation of the binary HRP‚1a complex were used as the basis for
the modeling of the ternary HRP‚1a‚2a complex. The ligand was
JA012075O
9
J. AM. CHEM. SOC. VOL. 124, NO. 5, 2002 787