Improved enzyme activity and enantioselectivity in organic solvents by methyl-β-cyclodextrin

Article abstract of DOI:10.1021/ja990515u

The use of enzymes in organic solvents to introduce chirality to a number of relevant organic compounds has been well documented. However, there are still major drawbacks in such applications, in particular the frequently much lower enzyme activity under nonaqueous conditions. In addition, the reaction outcome (substrate enantioselectivity and reaction rates) cannot be accurately predicted. To overcome these limitations, herein we introduce methyl-β-cyclodextrin (MβCD) as a new macrocyclic additive to simultaneously enhance the activity and enantioselectivity of dehydrated subtilisin Carlsberg suspended in neat organic solvents. MβCD was efficient in dramatically increasing the activity and significantly improving the enantioselectivity of subtilisin in co-lyophilizates when compared to the powder lyophilized from buffer alone. The initial rate determined for the transesterification between sec-phenethyl alcohol and vinyl butyrate increased by up to 164-fold and the enantioselectivity could be doubled. In addition, marked solvent effects were noted. To investigate the possible relationship between enzyme structure and these kinetic data, the secondary structure of subtilisin was investigated by Fourier transform infrared (FTIR) spectroscopy under all relevant conditions. Using the α-helix content determined from the amide I vibrational band as the main quantitative parameter, we found that MβCD is partially efficient in ameliorating dehydration-induced structural perturbations. Suspension of the subtilisin - MβCD co-lyophilizate in the various solvents revealed solvent-induced structural perturbations in some of them (e.g., acetonitrile), while no such changes were observed in others (e.g., THF and 1,4-dioxane). For the first time the results demonstrated that enantioselectivity and structural intactness in the various solvents were clearly related. Increase in the enzyme activity in contrast is mainly caused by increased structural mobility of subtilisin in the solvents by MβCD. We conclude that it is important to carefully select the additive and the solvent system to achieve high enantioselectivity and activity in such applications. Simultaneous improvement of both enzyme properties requires careful optimization of the enzyme formulation and proper selection of a suitable solvent. FTIR spectroscopy has proven to be a very valuable methodology to structurally guide such an optimization procedure.

Full text of DOI:10.1021/ja990515u

J. Am. Chem. Soc. 1999, 121, 8157-8163
8157
Improved Enzyme Activity and Enantioselectivity in Organic Solvents
by Methyl-â-cyclodextrin
†
‡
†
‡
Kai Griebenow, Yanira D ´ı az Laureano, Ang e´ lica M. Santos, Ileana Monta n˜ ez Clemente,
‡
†
,‡
Luiz Rodr ´ı guez, Michael W. Vidal, and Gabriel Barletta*
Contribution from the Departments of Chemistry, UniVersity of Puerto Rico, College at Humacao, CUH
Station, Humacao, Puerto Rico 00791, and UniVersity of Puerto Rico, R ´ı o Piedras Campus, P.O. Box
2
3346, San Juan, Puerto Rico 00931-3346
ReceiVed February 17, 1999
Abstract: The use of enzymes in organic solvents to introduce chirality to a number of relevant organic
compounds has been well documented. However, there are still major drawbacks in such applications, in
particular the frequently much lower enzyme activity under nonaqueous conditions. In addition, the reaction
outcome (substrate enantioselectivity and reaction rates) cannot be accurately predicted. To overcome these
limitations, herein we introduce methyl-â-cyclodextrin (MâCD) as a new macrocyclic additive to simultaneously
enhance the activity and enantioselectivity of dehydrated subtilisin Carlsberg suspended in neat organic solvents.
MâCD was efficient in dramatically increasing the activity and significantly improving the enantioselectivity
of subtilisin in co-lyophilizates when compared to the powder lyophilized from buffer alone. The initial rate
determined for the transesterification between sec-phenethyl alcohol and vinyl butyrate increased by up to
1
64-fold and the enantioselectivity could be doubled. In addition, marked solvent effects were noted. To
investigate the possible relationship between enzyme structure and these kinetic data, the secondary structure
of subtilisin was investigated by Fourier transform infrared (FTIR) spectroscopy under all relevant conditions.
Using the R-helix content determined from the amide I vibrational band as the main quantitative parameter,
we found that MâCD is partially efficient in ameliorating dehydration-induced structural perturbations.
Suspension of the subtilisin-MâCD co-lyophilizate in the various solvents revealed solvent-induced structural
perturbations in some of them (e.g., acetonitrile), while no such changes were observed in others (e.g., THF
and 1,4-dioxane). For the first time the results demonstrated that enantioselectivity and structural intactness in
the various solvents were clearly related. Increase in the enzyme activity in contrast is mainly caused by
increased structural mobility of subtilisin in the solvents by MâCD. We conclude that it is important to carefully
select the additive and the solvent system to achieve high enantioselectivity and activity in such applications.
Simultaneous improvement of both enzyme properties requires careful optimization of the enzyme formulation
and proper selection of a suitable solvent. FTIR spectroscopy has proven to be a very valuable methodology
to structurally guide such an optimization procedure.
Introduction
employing suspended, dehydrated enzyme powders in such
applications. Furthermore, possible solvent-induced denaturation
Enantiomeric purity in synthetic organic reactions has become
very important, in particular in the synthesis of pharmaceuticals.
Enzymes are extremely valuable catalysts in this context because
of their properties, which include high specificity and enantio-
and prochiral selectivity. These natural catalysts have also
been shown to be very useful for the introduction of chirality
under nonaqueous conditions. However, enzymes are fre-
quently poorly active in organic media when compared to their
respective activities in their natural aqueous environment. It is
well documented that lyophilization (the most frequent method
of enzyme preparation) causes pronounced structural perturba-
tions for most proteins, including the model enzyme in this
study, subtilisin Carlsberg. Such lyophilization-induced struc-
tural perturbations could contribute to the observed loss of
enzymatic activity and in particular enantioselectivity when
1
(2) (a) Zacks, A.; Klibanov, A. M. J. Biol. Chem. 1988, 263, 3194-
201. (b) Haraldsson, G. G., Ed.The application of lipases in organic
3
synthesis. In Suppl. B: Chem. Acid DeriV. 1992, 2, 1395-1473. (c) Lalonde,
J. J.; Govardhan, C.; Khalaf, N.; Martinez, A. G.; Visuri, K.; Margolin, A.
L. J. Am. Chem. Soc. 1995, 117, 6845-6852. (d) Lin, C. H.; Sugai, T.;
Halcomb, R. L.; Ichikawa, Y.; Wong, C. H. J. Am. Chem. Soc. 1992, 114,
10138-10145. (e) Terradas, F.; Teston-Henry, M.; Fitzpatrick, P. A.;
Klibanov, A. M. J. Am. Chem. Soc. 1993, 115, 5, 390-396. (f) Wescott,
C. R.; Klibanov, A. M. J. Am. Chem. Soc. 1993, 115, 1629-1631. (g)
Fitzpatrick, P. A.; Klibanov, A. M. J. Am. Chem. Soc. 1991, 113, 3166-
2
3
3171. (h) Schoffers, E.; Golebiowski, A.; Johnson, C. R. Tetrahedron 1996,
52, 3769-3826. (i) Faber, K.; Ottolina, G.; Riva, S. Biocatalysis 1993, 8,
91-132. (j) Klibanov, A. M. Acc. Chem. Res. 1990, 23, 114-120.
(3) (a) Schmitke, J. L.; Wescott, C. R.; Klibanov, A. M. J. Am. Chem.
4
Soc. 1996, 118, 3360-3365. (b) Klibanov, A. M. Trends Biotechnol. 1997,
1
5, 97-101.
(
4) (a) Desai, U. R.; Klibanov, A. M. J. Am. Chem. Soc. 1995, 117,
3
5
940-3945. (b) Griebenow, K.; Klibanov, A. M. Biotechnol. Bioeng. 1996,
3, 351-362. (c) Griebenow, K.; Klibanov, A. M. Proc. Natl. Acad. Sci.
*
Corresponding author.
R ´ı o Piedras Campus.
College at Humacao.
U.S.A. 1995, 92, 10969-10976. (d) Dong, A.; Meyer, J. D.; Kendrick, B.
S.; Manning, M. C.; Carpenter, J. F. Arch. Biochem. Biophys. 1996, 334,
406-414. (e) Griebenow, K.; Santos, A. M.; Carrasquillo, K. G. Internet
J. Vibr. Spectrosc. 1999, 3 (1), [http://www.ijvs.com]. (f) Prestrelski, S. J.;
Tedischi, N.; Arakawa, T.; Carpenter, J. F. Biophys. J. 1993, 65, 661-
671.
†
‡
(
1) (a) Caddick, S.; Jenkins, K. Chem. Soc. ReV. 1996, 447-456. (b)
Stinson, S. C. Chiral drug market shows signs of maturity. In Chem. Engn.
News 1997, October 20.
1
0.1021/ja990515u CCC: $18.00 © 1999 American Chemical Society
Published on Web 08/31/1999

8158 J. Am. Chem. Soc., Vol. 121, No. 36, 1999
Griebenow et al.
Scheme 1
of the suspended enzyme catalysts might also contribute to the
loss in activity and enantioselectivity.
inclusion complexes with a number of guest molecules. As a
result, they are useful in a number of applications ranging from
chiral separations to drug delivery and as enzyme mimics.8
These diverse applications of cyclodextrins motivated us to study
their effect on the activity of subtilisin in organic solvents. It
has been shown recently that cyclodextrins when added to a
buffer solution containing a protein reduce its mean unfolding
To overcome these problems and make enzymes as asym-
metric catalysts more appealing to the organic chemist, many
strategies increasing their potency in organic solvents have been
5
a
explored. These include the mode of enzyme preparation, the
5
b
5c
control of the pH value, co-lyophilization with lyoprotectants
and salts,5 addition of water-mimicking agents, imprinting
with substrates and substrate analogues,
d,e
5f
temperature. The author concluded that cyclodextrins form
9
5g,h
5i,j
5r
immobilization,
inclusion complexes with the side chains of buried hydrophobic
residues of the protein, shifting the equilibrium in favor of the
unfolded polypeptide. The potential binding of cyclodextrins
to amino acid residues is very intriguing in light of the enzyme
5k-p
5q
solubilization,
mutagenesis, formation of substrate salts,
and cross-link crystallization.2 One of the most successful
c
groups of activating additives identified thus far are crown
6
6
ethers. While the mechanism of activation is still somewhat
activation by crown ethers in organic solvents. Therefore, we
speculative, co-lyophilization of various enzymes with the crown
ether 18-crown-6 in particular resulted in highly active and
enantioselective formulations in organic solvents. It is reasonable
to assume that such crown ethers bind to surface amino acid
residues (in particular the ꢀ-amino group of lysine) and that
this binding alters the enzyme properties beneficially. For
example, due to the amphiphilic character of the crown ethers
they could increase the flexibility of the enzymes in the solvents
as a result of such binding. Such increased structural mobility
has been demonstrated by Broos et al. (1996) as a possible
mechanism of how some solvents enhance the enzyme catalytic
decided to investigate their effect on the serine protease subtilisin
suspended in organic solvents, and in particular, on the reaction
rates and the enzyme enantioselectivity. After finding the
additive methyl-â-cyclodextrin (MâCD) very efficient in en-
hancing the enzyme performance, we also addressed simulta-
neously the influence of this additive on enzyme structure and
rigidity under various conditions leading to new mechanistic
insights.
Results and Discussion
Subtilisin, our model enzyme, was lyophilized from an
aqueous buffer solution (pH 7.8) with or without various
cyclodextrins. The preparations were suspended in various
typical organic solvents (THF, acetonitrile, 1,4-dioxane, toluene,
dichloromethane, and octane). The kinetics of the product
formation in the well-characterized and well-understood trans-
esterification between sec-phenethyl alcohol and vinyl butyrate
was followed by gas chromatography (Scheme 1).2
7
rate. Alternatively, crown ethers could also activate enzymes
by facilitating the removal of water molecules from the active
site upon substrate binding.6b
In the current study we present data and mechanistic insights
on enzyme activation by another class of macrocyclic excipients,
cyclodextrins. They are versatile cyclic compounds that can form
g,10
(5) (a) Noritomi, H.; Almarsson, O¨ .; Barletta, G. L.; Klibanov, A. M.
Biotechnol. Bioeng. 1996. (b) Yang, Z.; Zacherl, D.; Russell, A. J. J. Am.
Chem. Soc. 1993, 115, 12251-12257. (c) Dabulis, K.; Klibanov, A. M.
Biotechnol. Bioeng. 1993, 41, 566-571. (d) Khmelnitsky, Y. L.; Welch,
S. H.; Clark, D. E.; Dordick, J. S. J. Am. Chem. Soc. 1994, 116, 2647-
No enhancement effect on the enzyme enantioselectivity or
the reaction rate was observed when subtilisin was prepared in
the presence of R- and γ-cyclodextrin. However, a marginal
increase in the enzyme enantioselectivity and catalytic rate was
observed when the enzyme was prepared in the presence of
â-cyclodextrin (âCD), at a 1:1 weight ratio (data not shown).
The effect resembles those found with crown ethers where the
size of the cavity determines the efficiency of the additive in
2
648. (e) Ru, M. T.; Dordick, J. S.; Reimer, J. A.; Clark, D. S. Biotechnol.
Bioeng. 1999, 63, 233-241. (f) Kitaguchi, H.; Itoh, I.; Ono, M. Chem.
Lett. 1990, 1203-1206. (g) Rich, J. O.; Dordick, J. S. J. Am. Chem. Soc.
1
1
997, 119, 3245-3252. (h) Russell, A. J.; Klibanov, A. M. J. Biol. Chem.
988, 263, 11624-11626. (i) Petro, M.; Svec, F.; Fr e´ chet, J. M. J.
Biotechnol. Bioeng. 1996, 49, 355-363. (j) Orsat, B.; Drtina, G. J.;
Williams, M. G.; Klibanov, A. M. Biotechnol. Bioeng. 1994, 44, 1265-
6
enzyme activation. For example, R-chymotrypsin activity is
1269. (k) Okahata, Y.; Hatano, A.; Ijiro, K. Tetrahedron: Asymmetry. 1995,
6, 1311-1322. (l) Okahata, Y.; Fujinoto, Y.; Ijiro, K. J. Org. Chem. 1995,
60, 2244-2250. (k) Xu, K.; Klibanov, A. M. J. Am. Chem. Soc. 1996,
118, 9815-9819. (m) Paradkar, V. M.; Dordick, J. S. J. Am. Chem. Soc.
1994, 116, 5009-5010. (n) Wangikar, P. P.; Michels, P.; Clark, D. S.;
enhanced more by 18-crown-6 than by 15-crown-5 or 12-crown-
6
4
. Regarding the chemical similarity of R-, â-, and γ-CD, the
difference in activation suggests specific macrocycle-enzyme
Dordick, J. S. J. Am. Chem. Soc. 1997, 119, 70-76. (o) Meyer, J. D.;
Kendrick, B. S.; Matsuura, J. E.; Ruth, J. A.; Bryan, P. N.; Manning, M.
Int. J. Pept. Protein Res. 1996, 47, 177-181. (p) Xu, K.; Griebenow, K.;
Klibanov, A. M. Biotechnol. Bioeng. 1997, 56, 485-491. (q) Chen, K.;
Arnold, F. H. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 5618-5622. (r) Ke,
T.; Klibanov, A. M. J. Am. Chem. Soc. 1999, 121, 3486-3493.
interactions. The cavity sizes of R-, â-, and γ-CD are ca. 0.47-
8g
0
.53, 0.6-0.65, and 0.75-0.83 nm, respectively.
Since it has been reported in the literature that the magnitude
of the enzyme enhancement in organic solvents by macrocyclic
(
6) (a) Reinhoudt, D. N.; Eendebak, A. M.; Nijenhuis, W. F.; Verboom,
W.; Kloosterman, M.; Schoemaker, H. E. J. Chem. Soc., Chem. Commun.
989, 399-400. (b) Broos, J.; Sakodinskaya, I. K.; Engbersen, J. F. J.;
Verboom, W.; Reinhoudt, D. N. J. Chem. Soc., Chem. Commun. 1995, 255-
56. (c) Itoh, T.; Takagi, Y.; Murakami, T.; Hiyama, Y.; Tsukube, H. J.
(8) (a) Albers, E.; Muller, B. W. Crit. ReV. Ther. Drug Carrier Syst.
1995, 12, 311-37. (b) Stella, V. J.; Rajewski, R. A. Pharm. Res. 1997, 14,
556-67. (c) Breslow, R.; Huang, Y.; Zhang, X.; Yang, J. Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 11156-11158. (d) Luong, J. H.; Brown, R. S.; Male
K. B.; Cattaneo M. V.; Zhao S. Trends Biotechnol. 1995, NoV. 13, 457-
463. (e) Karuppiah, N.; Sharma, A. Biochem. Biophys. Res. Commun. 1995,
211, 60-66. (f) Fanali S. J. Chromatogr. A 1997, 792, 227-267. (g) Li,
S.; Purdy, W. C. Chem. ReV. 1992, 92, 1457-1470.
1
2
Org. Chem. 1996, 61, 2158-2163. (d) van Unen, D.-J.; Engbersen, J. F.
J.; Reinhoudt, D. N. Biotechnol. Bioeng. 1998, 59, 553-556. (e) Engbersen,
J. F. J.; Broos, J.; Verboom, W.; Reinhoudt, D. N. Pure Appl. Chem. 1996,
6
8, 2171-2178.
7) Broos, J.; Visser, A. J. W. G.; Engbersen, F. J.; Verboom, W.; van
Hoek, A.; Reinhoud, D. N. J. Am. Chem. Soc. 1995, 117, 12657-12663.
(9) Cooper, A. J. Am. Chem. Soc. 1992, 114, 9208-9209
(
(10) Colombo, G.; Toba, S.; Merz, K. M., Jr. J. Am. Chem. Soc. 1999,
121, 3486-3493.

ImproVed Enzyme ActiVity with Methyl-â-cyclodextrin
J. Am. Chem. Soc., Vol. 121, No. 36, 1999 8159
Table 1. Enantioselectivity, Initial Rate, and R-Helix Content of Subtilisin Carlsberg in Different Organic Solvents
solvent
THF
lyophilizeda,b
MâCD co-lyophilizatea,c
enhancement
R-helix (%)d
32 ( 1
e
V
V
E
V
V
E
V
V
E
V
V
E
V
V
E
V
V
S
0.014 ( 0.004
0.00043 ( 0.00015
32 ( 2
0.034 ( 0.021
0.0011 ( 0.0005
30.7 ( 8.9
0.0014
0.00015
2.3 ( 0.6
164
91
1.8
112
77
1.5
19
17
1.1
17
13
1.2
11
18
R
f
0.039 ( 0.009
59 ( 4
1
,4-dioxane
S
3.8 ( 1.5
30 ( 1
26 ( 1
22 ( 1
26 ( 1
29 ( 0
R
0.085 ( 0.03
45 ( 1.5
CH
2
Cl
2
S
0.027 ( 0.002
0.0026 ( 0.0002
10.2 ( 0.03
0.085 ( 0.006
0.016 ( 0.003
5.4 ( 0.8
R
9.5
acetonitrile
toluene
S
0.0050 ( 0.0026
0.0012 ( 0.0007
4.6 ( 0.6
0.0035 ( 0.0005
0.00035 ( 0.00005
11.8 ( 0.6
0.015 ( 0.005
0.0015 ( 0.0005
R
S
0.04 ( 0.027
0.0062 ( 0.0023
7.4 ( 1.4
0.058 ( 0.033
0.010 ( 0.003
R
0.63
4
6
octane
S
R
a
Enzyme concentration 1 mg/mL in all experiments. b Lyophilized from aqueous phosphate buffer, pH 7.8. c Lyophilized from aqueous phosphate
buffer containing MâCD at a 1:6 weight ratio of subtilisin to excipient. d R-helix content determined for the suspended subtilisin-MâCD co-
-
1
lyophilized by Gaussian curve fitting of the amide I IR spectra after resolution enhancement by Fourier-self-deconvolution (FWHM 24 cm , k )
e
-1
-1 f
2
M R M S
.4). Initial rates in µmol mg min . [kcat/K ] /[kcat/K ] (enzyme enantioselectivity).
compounds depends on the ratio of additive to enzyme,6b we
increased the concentration of the additive. It was necessary to
employ methyl-â-cyclodextrin (MâCD) in this instance, due to
the limited solubility of âCD in water. MâCD is highly soluble
in water, and yet it shares similar chemical properties with its
counterpart âCD. When MâCD was employed at a 1:6 weight
ratio of enzyme to excipient, a significant activation of subtilisin
occurred when compared with the data for the enzyme lyoph-
ilized from buffer alone (Table 1). The initial rates were larger
for the MâCD formulation under all solvent conditions tested.
The activation of the enzyme was in particular pronounced in
the solvents THF and 1,4-dioxane. The initial rates obtained
for the “S” enantiomer using this new enzyme preparation were
hydrophobicity,11 dipole moment,2g and dielectric constant.2g
However, in each of these examples specific reasons were
hypothesized to account for the observed effects, in particular
differences in water replacement from the active site by the
1
2
different enantiomers. In the case of the substrate chosen in
this work, no correlation was found with any of the preparations
(powder lyophilized from buffer alone or MâCD-enzyme
formulation), between the enzyme enantioselectivity and any
of the solvents physicochemical properties mentioned above.2g
To further investigate the somewhat puzzling observation that
reaction rates improved under all conditions while enantiose-
lectivity increased or decreased depending on the solvent, FTIR
studies were conducted to investigate possible structural con-
tributions. We determined the effect of the additive MâCD on
the secondary structure of subtilisin in the lyophilized state and
suspended in organic solvents. It is well established that
lyophilization causes significant structural changes in subtilisin.4
Significant spectral changes occurred in subtilisin upon lyo-
philization from aqueous buffer solution (Figure 1). The
amplitude of all spectral components increased relative to the
1
64- and 112-fold larger in THF and 1,4-dioxane, respectively,
than for the enzyme lyophilized without the additive. In addition,
enzyme enantioselectivity of this new preparation was also
significantly higher in THF (2 times) and in 1,4-dioxane (1.5
times) than for the enzyme lyophilized without MâCD. In other
solvents enantioselectivity was not significantly influenced by
the additive (in CH2Cl2 and acetonitrile) or even dropped (in
toluene and octane). We excluded nonenzymatic cyclodextrin
mediated transesterification in control experiments. No reaction
was observed in the absence of the enzyme. In an additional
control experiment, subtilisin was lyophilized from buffer and
suspended in THF, and MâCD was added next to the reaction
mixture at the 1:6 weight ratio as before. No effect of the
additive on the initial rates and enantioselectivity was noted in
these experiments. Two conclusions can be derived from this.
First, MâCD must indeed interact in a specific way with the
enzyme, probably by binding to surface residues. This behavior
must be the result of the pretreatment of the enzyme (lyophiliza-
tion in the presence of the additive). Therefore, the situation
resembles those found with crown ethers where highly active
preparations can also only be obtained when the enzymes are
co-lyophilized with the additive.6b Second, substrate-MâCD
inclusion complexes formed in the solvent can be excluded to
contribute to the improvement in enantioselectivity. In principle,
preferential binding of one enantiomer over the other by MâCD
in the solvent could also have caused changes in the enantio-
selectivity.
-
1
R-helix band at ca. 1658 cm . Quantitative data revealed a
significant decrease in the R-helix content from 34% to 25%
and an increase in the â-sheet content from 15% to 32% in
4
b
agreement with previous data. Co-lyophilization of subtilisin
with MâCD prevented these structural changes to some extent.
-1
In particular, the bands at ca. 1646 and 1631 cm did not show
a significant intensity increase when compared to the aqueous
spectrum (Figure 1). However, significant amide I spectral
-
1
changes still occurred at wavenumbers above 1658 cm .
Quantitative data agree with the reduced extent of lyophilization-
induced structural perturbations in the MâCD co-lyophilizate.
The decrease in the R-helix content (30 ( 1%) was ap-
proximately halved compared to that in the case of no additive.
Next, we suspended the MâCD-subtilisin co-lyophilizate in
the organic solvents used in the kinetic experiments. It has been
established previously that suspension of subtilisin lyophilized
from a phosphate buffer in nonprotein dissolving organic
4b
solvents caused very small spectral changes. In the aforemen-
(11) (a) Tawaki, S.; Klibanov, A. M. J. Am. Chem. Soc. 1992, 114, 1882-
884. (b) Sakurai, T.; Margolin, A. L.; Russell, A. J.; Klibanov, A. M. J.
1
Next, we tried to relate the determined enantioselectivity
values with various solvent parameters. In some previous works,
enantioselectivity has been shown to be related to solvent
Am. Chem. Soc. 1988, 110, 7236-7237.
12) Wescott, C. R.; Klibanov, A. M. Biochim. Biophys. Acta 1994, 1206,
1-9.
(

8160 J. Am. Chem. Soc., Vol. 121, No. 36, 1999
Griebenow et al.
Figure 2. Fourier self-devonvoluted FTIR spectra and Gaussian curve-
fitting of subtilisin Carlsberg co-lyophilized with MâCD and suspended
in various solvents in the amide I spectral region. The solid lines
represent superimposed FSD spectra and the results of the curve-fitting;
individual Gaussian bands are shown with a dashed line.
Figure 1. Amide I FTIR spectra of subtilisin under various condi-
tions: (left) after second derivatization and (right) Fourier self-
devonvoluted FTIR spectra and Gaussian curve-fitting. The solid lines
represent the superimposed FSD spectra and the results of the curve
fitting; individual Gaussian bands are shown with a dashed line.
preservation of enzyme structure is essential in obtaining highly
active enzyme preparations in organic solvents. However, the
lack of correlation between the enzyme structure and initial rates
indicates that there must be other factors which also contribute
tioned work, the R-helix content for the lyophilized protein was
determined at 26%. For suspensions of this powder in the
following solvents the R-helix contents were determined as
follows in octane (28%), toluene (26%), THF (27%), dioxane
3
significantly to the reaction rates. Such factors include solvent-
dependent variations in enzyme flexibility and substrate des-
(26%), and acetonitrile (27%). In contrast, we found significant
3
olvation. The latter observation is particularly evident when
solvent-dependent structural changes in the MâCD co-lyo-
philizate. While suspension of the co-lyophilizate in THF,
dioxane, and octane (Figure 2) caused very small spectral
changes when compared to the spectrum of the co-lyophilizate
in the dry state (Figure 1), suspension of it in toluene,
dichloromethane, and in particular acetonitrile caused significant
solvent-induced spectral and thus structural alterations. Quan-
titative analysis of the FTIR spectra obtained supports this
observation (Table 1). The R-helix content determined for the
subtilisin-MâCD co-lyophilizate was the same in suspensions
in THF, dioxane, and octane as prior to suspension. A minor
increase in the â-sheet content was noted (data not shown). In
contrast, significant decreases were diagnosed for the R-helix
content when the co-lyophilizate was suspended in toluene,
dichloromethane, and acetonitrile. Next, we investigated whether
these solvent-induced protein structural perturbations and activity
could be related. No correlation is evident when plotting the
initial rate for the “S” enantiomer (VS) versus the R-helix content
as indicator for the overall structure of the enzyme (Table 1).
However, in the two cases where significant enzyme activation
occurred (in THF and dioxane), no solvent-induced denaturation
occurred and enzyme structure was the same as for the
co-lyophilizate with MâCD. This might be an indication that
analyzing the situation in acetonitrile. Even though the FTIR
spectra (Figures 1 and 2) and R-helix content (Table 1) indicate
significant solvent-induced structural perturbations in the MâCD
co-lyophilizate, activity is still higher than for the powder
lyophilized from phosphate buffer. These results are in agree-
4b
ment with the data reported by Griebenow and Klibanov and
4
d
Dong et al. where no correlation between enzyme structure
and initial rates was found.
The larger susceptibility of the subtilisin-MâCD co-lyo-
philizate to solvent-induced structural alterations, when com-
pared to the enzyme without the additive, indicates a reduction
in the conformational stability (“rigidity”) of subtilisin caused
by the additive. Similar conclusions have been derived when
6
employing crown ethers as additives and also when the
molecular lubricant water was added to the organic solvent.13
To further investigate differences in the stability of subtilisin
in the presence and absence of MâCD we conducted thermal
inactivation experiments. In the first set of experiments the
lyophilized subtilisin samples were incubated at 45 °C in the
organic solvent THF (which was chosen because the subtilisin
(13) Griebenow, K.; Klibanov, A. M. J. Am. Chem Soc. 1996, 118,
11695-11700.

ImproVed Enzyme ActiVity with Methyl-â-cyclodextrin
J. Am. Chem. Soc., Vol. 121, No. 36, 1999 8161
Figure 4. Dependence of the enzyme enantioselectivity on the R-helix
content for the subtilisin-MâCD formulation. Solvents: (a) THF, (b)
Figure 3. Thermal denaturation curves for subtilisin: (4) in D
2
O, pD
1
,4-dioxane, (c) CH
2
Cl
2
, (d) acetonitrile, and (e) toluene.
7
1
1
.8; (b) dried from aqueous buffer solution, pH 7.8, and exposed to
,4-dioxane; and (0) co-dried from aqueous buffer with MâCD at a
:6 weight ratio of protein-additive in 1,4-dioxane. The spectral
7
(
“flexibility”) and the reaction rates. Therein, it has been
reported that for lyophilized subtilisin suspended in different
organic solvents, kinetic data and fluorescence anisotropy data
correlation coefficient (r) was calculated using the second derivative
amide I FTIR spectra of the samples at 30 °C and at elevated
temperatures. A decrease in the correlation coefficient to values <1
indicates temperature-induced protein structural perturbations. The
dotted curve is the inverted first derivative of the thermal denaturation
7
could be related. High reaction rates and high enzyme enan-
tioselectivity were extrapolated to higher flexibility of the
enzyme catalyst. According to Bross et al., the more flexible
the enzyme is, the greater the probability that the enzyme
achieves a more active and enantioselective conformation. This
concept is very attractive, in particular, in light of recent
of subtilisin in D
maximum, as indicated.
2
O (4), and the T
m
can be obtained from its
2
5
activation was most pronounced in it, Table 1). After various
incubation times the activity and enantioselectivity were deter-
mined. We found that the initial rates for the lyophilized powder
decreased exponentially by factors of 6 (S enantiomer) and 5
1
6
investigations involving thermostable enzymes. It has been
shown that at optimum activity level, the mesophilic and
thermophilic isoenzymes 3-isopropylmalate dehydrogenase adopt
a very similar conformational flexibility. At room temperature
the thermophilic form is nearly inactive and conformational
mobility largely restricted. All results on the conformational
stability of subtilisin reported in this work indicate decreased
rigidity of subtilisin in the organic solvents in the presence of
MâCD. Therefore, we conclude that activation of subtilisin by
MâCD is likely primarily due to increased flexibility of the
enzyme in the solvents. It can be excluded that enzyme structural
alterations in typical applications largely contribute to differ-
ences in enzyme activity. It is exciting to note that the same
delicate balance between stability and flexibility seems to play
an important role for enzyme function in organic solvents as
well. To achieve high reaction rates at moderate temperatures
it seems to be necessary to adjust the structural mobility.
However, further investigations similar to those conducted in
(R enantiomer) within the time period of 100 h. For the
subtilisin-MâCD co-lyophilizate initial rates decreased by a
factor of more than 20 for both enantiomers within less than
5
0 h. This highlights the increased susceptibility of subtilisin
in THF toward thermal inactivation in the presence of MâCD.
Similarily, we observed a drop in activity for lyophilized
subtilisin in 1,4-dioxane by factors of 4 (S enantiomer) and 1.6
(
°
R enantiomer) when increasing the temperature from 45 to 85
C. Under the same experimental conditions, activity of the
14
subtilisin-MâCD co-lyophilizate decreased by a factor of ca.
50 and 131, respectively. Again, these experiments suggest
4
that the additive MâCD caused a significant decrease in the
conformational stability of the enzyme. Finally, thermal dena-
turation experiments were conducted in conjunction with FTIR
spectroscopy. For these experiments subtilisin was prepared as
a thin dry film on a CaF2 window in the absence and presence
of MâCD. Measurements were performed exposing these films
16
aqueous solution measuring the molecular mobility and activity
at various temperatures are certainly necessary to validate this
point of view.
14
to 1,4-dioxane. From plots of the spectral correlation coef-
ficient r4e,f as an indicator for overall structural changes versus
Having identified possible rationales to explain the subtilisin
activation by MâCD, we then focused on the possible relation-
ship between enantioselectivity and protein structural integrity.
When plotting the enantioselectivity E versus the R-helix
content, a clear correlation between enzyme structure and
function emerged (Figure 4). The higher the R-helix content
the temperature, it is evident that the thermal stability of
subtilisin is significantly decreased by MâCD (Figure 3). This
indicates that the conformational flexibility of dehydrated
subtilisin is enhanced by MâCD in 1,4-dioxane. However,
thermal stability still was much higher than for subtilisin in
aqueous solution where the Tm determined was 73.2 °C at
pD 7.8.
It is intriguing that the only relationship between structural
and dynamic properties of enzymes in organic solvents and their
activity identified thus far is that between the structural mobility
(and thus, the less perturbed the enzyme structure), the higher
the enzyme enantioselectivity. The only exception is for the data
point obtained in octane. However, in this solvent the limited
(15) (a) Carrasquillo, K. G.; Cordero, R. A.; Ho, S.; Franquiz, J. M.;
Griebenow, K. Pharm. Pharmacoll. Commun. 1998, 4, 563-571. (b)
Carrasquillo, K. G.; Costantino, H. R.; Cordero, R. A.; Hsu, C. C.;
Griebenow, K. J. Pharm. Sci. 1999, 88, 166-173. (c) Costantino, H. R.;
Carrasquillo, K. G.; Cordero, R. A.; Mummenthaler, M.; Hsu, C. C.;
Griebenow, K. J. Pharm. Sci. 1998, 87, 1412-1420.
(14) 1,4-Dioxane was chosen as the solvent in the thermal denaturation
experiments because of its relatively high boiling point (101.3 °C). The
enhancements in catalytic rate and enantioselectivity in this solvent when
employing the MâCD-subtilisin co-lyophilizate were not as excellent as
in THF, but still significant.
(16) Z a´ vodszky, P.; Kardos, J.; Svingor, A.; Petsko, G. A. Proc. Natl.
Acad. Sci. U.S.A. 1998, 95, 7406-7411.

8162 J. Am. Chem. Soc., Vol. 121, No. 36, 1999
Griebenow et al.
solubility of MâCD is a likely explanation. Due to insignificant
solubility the additive might partially block the active site of
the enzyme preventing the substrate from “correct” binding.
Consequently, substrate binding would be less specific and
therefore enantioselectivity lower than expected from the
structural analysis. The sluggish increase in activity in octane
also indicates the partial blockage of the active site by MâCD
in this solvent (Table 1). The MâCD additive is soluble in the
other solvents studied.17
Table 2. Enantioselectivity and Initial Rates of Candida rugosa
Lipase Suspended in THF
lyophilized (buffer only)a
lyophilized with MâCDa
b
V
V
E
S
0.00082
0.0019
2.3
0.0051
0.032
6.3
R
c
a
b
Enzyme concentration 1 mg/mL in all experiments. Initial rates
-1
-1 c
in µmol mg min
M R M S
[kcat/K ] /[kcat/K ] (enzyme enantioselectivity)
That enantioselectivity and structure are related is not
unexpected because of the need to preserve the three-
dimensional structure of the enzymes’ active site to achieve high
enantioselectivity. Differences in enantioselectivity arise from
co-lyophilized with MâCD, a 16.8-fold increase in the initial
rate of the “R” enantiomer and a 2.7-fold increase in the enzyme
enantioselectivity was observed (Table 2). Therefore, enzyme
enhancement by MâCD seems to be of a general nature.
1
2
differences in the binding of the S and R enantiomer. Similar
conclusions were derived recently when comparing lyophilized
subtilisin with subtilisin cross-linked enzyme crystals (CLECs).
Experimental Section
18
Materials. The solvents were purchased in the anhydrous form (in
Aldrich Sure/Seal bottles, water content below 0.005%). All solvents
used for this study were predried prior to their use as recommended
As mentioned earlier, several factors that contribute to the
solvent dependence of enzyme enantioselectivity have been
identified. These factors include the desolvation of the substrate
tetrahedral intermediate, exclusion of water from the active site,
the flexibility of the catalyst, and the structural integrity of the
suspended catalyst. Our data provide evidence that structural
preservation is a criterion of utmost importance when trying to
improve enantioselectivity by solvent engineering. If the catalyst
undergoes solvent-induced structural changes, then the latter
influence the enantioselectivity strongly. Consequently, solvent
effects on enantioselectivity cannot be studied under such
conditions. In this light, it is expected that in this work we would
not find a correlation between any of the solvent parameters
and the enantioselectivity. This is indeed the case: enantiose-
lectivity does not correlate with any of the solvent parameters
tested (hydrophobicity, dielectric constant, dipole moment, and
polarity). Our results and those reported by others imply that
CLECs should be primarily employed in mechanistic investiga-
tions of the effect of the solvent on enantioselectivity and other
selectivity because solvent effects on enzyme structure are
3
(THF and dioxane were dried by distilling them from Na, and CH CN
from CaH), and they were transferred to the reaction vial under N2.
âCD and MâCD were purchased from Aldrich. The enzymes,
subtilisin Carlsberg (EC 3.4.21.62) and the lipase from Candida rugosa
(EC 3.1.1.3), were purchased as lyophilized powders from Sigma
Chemicals, Inc.
Methods. (a) Enzyme Preparation. The enzyme, subtilisin Carls-
berg, was prepared as follows: The lyophilized enzyme powder was
dissolved (5 mg/mL) in a 20 mM phosphate buffer (at pH 7.8, the
enzyme’s optimum pH). This solution was then rapidly frozen (in liquid
2
N , to avoid destruction of the enzyme due to self-hydrolysis) and
lyophilized for 48 h. The lipase was prepared as described for subtilisin,
but it was lyophilized from a buffer at pH 7.5. Co-lyophilization of
the enzymes with MâCD was performed in the same manner, except
that MâCD (at a 1:6 weight ratio of enzyme to MâCD) was added to
the buffer solution prior to lyophilization.
(
b) Cyclodextrin Solubility Experiments. The cyclodextrin (MbCD
or âCD), 50 mg, was placed in a 1.0 mL vial; then the solvent (THF,
,4-dioxane, acetonitrile, toluene, CH Cl , or octane) was added (5 mL
portions) until the cyclodextrin was dissolved.
c) Kinetic Measurements. Product formation was followed by gas
chromatography (GC). The GC instruments (a Varian 3350 and a HP
890, with Chirasil CB columns, FID detectors, He as carrier gas) were
1
2
2
18-20
prohibited, or at least minimized, due to structural constraints.
(
However, while there may be advantages in such mechanistic
investigations to employ CLECs, there are also significant
drawbacks. For example, subtilisin CLECs do not have a much
6
2g
calibrated with the chiral esters, synthesized as reported. The enzyme
powder (about 5 mg) was placed in a 2-mL screw-cap scintillation vial
fitted with a mininert cap. The organic solvent (1.0 mL) and the
substrates (alcohol and vinyl butyrate substrates) were then added to
initiate the reaction. The vial was sealed and subjected to careful
sonication using a sonication bath to homogenize the suspension. The
mixture was then placed in a controlled-temperature shaker and agitated
vigorously (at 45 °C, at 300 rpm). Periodically 0.5 µL of the reacting
solution was withdrawn and analyzed by chiral-GC. Under all conditions
kinetic experiments were terminated before 10% of the product had
been formed. The enzyme enantioselectivity was determined by
measuring the initial rates of enzymatic reactions (from plots of product
formation vs time) of both enantiomers. The enzyme enantioselectivity
3
higher activity in organic solvents than the lyophilized powder.
Currently no efficient strategies for the activation of such crystals
2
1
are available.
Last, we addressed the question whether the improvements
in enzyme activity and enantioselectivity afforded by MâCD
are also found with a different enzyme. We studied the kinetics
of the model transesterification reaction between sec-phenethyl
alcohol and vinyl butyrate catalyzed by Candida rugosa lipase
in THF. Our results show that the effect of MâCD on the lipase
is similar to that observed with subtilisin. When the lipase was
(17) Solubility of MâCD in 1,4-dioxane, THF, CH2Cl2, and acetoni-
for either substrate is equal to the ratio: [kcat/K
]
M R
/[kcat/K
]
M S
) V
R
[S]/
trile: more than 1 g of MâCD/mL of solvent; in toluene: 0.07 g of MâCD/
mL of toluene; in octane: not soluble within detection limits. Note that the
additive âCD is not soluble in 1,4-dioxane, THF, and acetonitrile and that
only a marginal increase in the enzyme enantioselectivity and in the reaction
rates was observed with this additive.
5r
V
S
[R]. Note that this relationship is valid only when a racemic mixture
of the substrates is being studied (as in our experiments), so that both
chiral substrates are competing for the binding site simultaneously.
(d) FTIR Spectroscopy. FTIR studies were conducted with a Nicolet
(
18) Ke, T.; Wescott, C. R.; Klibanov, A. M. J. Am. Chem. Soc. 1996,
18, 3366-3374.
19) (a) Fitzpatrick P. A.; Ringe D.; Klibanov A. M. Biochem. Biophys.
4b,c,13,15
Magna-IR System 560 optical bench as described.
A total of
56 scans at 2 cm resolution using Happ-Ganzel apodization were
averaged to obtain each spectrum. Lyophilized protein powders were
1
-
1
2
(
Res. Commun. 1994, 198, 675-681. (b) Fitzpatrick P. A.; Steinmetz A.
C.; Ringe D.; Klibanov A. M. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 8653-
4
b,c
measured as KBr pellets (1 mg of protein per 200 mg of KBr).
Enzymes were suspended in organic solvents, sonicated for 2 min, and
measured in a FTIR cell equipped with CaF windows and 50-µm thick
spacer. Each protein sample was measured at least five times. When
necessary, spectra were corrected for the background and water vapor
contributions in an interactive manner using the Nicolet OMNIC 3.1
software to obtain the protein’s vibration spectra.4
8
657. (c) Schmitke, J. L.; Stern, L. J.; Klibanov, A. M. Proc. Natl. Acad.
Sci. U.S.A. 1997, 94, 4250-4255.
20) Wescott, C. R.; Noritomi, H.; Klibanov, A. M. J. Am. Chem. Soc.
996, 118, 10365-10370.
21) Thus far, other than adjusting the water activity to its optimum
2
(
1
2
2a
(
2
2b
and controlling the pH
no method has been established to achieve
b,c,13
activation of CLECs in organic solvents.

ImproVed Enzyme ActiVity with Methyl-â-cyclodextrin
e) FTIR Data Aanalysis. All spectra were analyzed by second
J. Am. Chem. Soc., Vol. 121, No. 36, 1999 8163
(
the R-helix content is used as the structural parameter describing the
enzyme integrity, only the area contribution of the band at ca. 1657
cm assigned to R-helices is given.
(f) Thermal Denaturation Experiments. Thin films were produced
by drying 100 µL of a solution of 10 mg/mL of subtilisin in 10 mM
phosphate buffer (pH 7.8) under a stream of dry nitrogen gas for at
derivatization in the amide I region for their component compo-
4
b,c,15,23,24
-1
sition.
Second derivative spectra were smoothed with an 11-
-1
point smoothing function (10.6 cm ). Fourier self-deconvolution (FSD)
was applied to the background- and water vapor-corrected spectra to
enable quantification of the secondary structure in the amide I region
by Gaussian curve-fitting using the program OMNIC 3.1.
did not observe any over-deconvolution with the parameters chosen
4
b,c,15,23,24
We
least 30 min on a CaF
MâCD at a 1:6 weight ratio of enzyme to additive. The CaF
2
window. Films were produced with or without
window
2
-
1
(
24 cm for the full-width-at-half-maximum and k ) 2.4 for the
was then mounted in a model HT-32 liquid cell using a 50 µm thick
enhancement factor). Note that FSD alters the band shapes, but
preserves the integrated band intensities when over-deconvolution is
spacer (Spectra Tech) and the cell filled with 1,4-dioxane. Spectra of
subtilisin in D O were measured at 50 mg/mL of protein concentration
2
2
3b,24a
avoided.
region after band narrowing of the protein vibrational spectra by FSD
Gaussian curve fitting was performed in the amide I
using 50 µm thick spacers and the above liquid cell. The temperature
was controlled with a microprocessor programmable controller (Spectra
Tech). Overall structural changes were quantified by calculating the
1
3,15,24
as described.
those in the literature.
solution and subtilisin lyophilized from buffer are within experimental
The band assignment in the amide I region followed
1
3,15,24
-1
Data obtained in this work for the aqueous
correlation coefficients from the amide I (1700-1600 cm ) second
derivative spectrum of the sample at 30 °C and those at elevated
4
b
4e,f
error of those reported previously. However, since in this work only
temperatures. Melting temperatures could not be determined from
the thermal denaturation curves in 1,4-dioxane because the experiment
could only be performed up to 95 °C and the protein structural transition
was not finished.
(
22) (a) Partridge, J.; Hutcheon, G. A.; Moore, B. D.; Halling, P. J. Am.
Chem. Soc. 1996, 118, 12873-12877. (b) Xu, K.; Klibanov, A. M. J. Am.
Chem. Soc. 1996, 118, 9815-9819.
(
23) (a) Susi, H.; Byler, D. M. Methods Enzymol. 1986, 130, 290-311.
(
b) Mantsch, H. H.; Casal, H. L.; Jones, R. N. Spectroscopy of Biological
Acknowledgment. The Puerto Rico NSF-EPSCoR program
Systems; Wiley: New York, 1986; pp 1-46.
(
(OSR-9452893), the University of Puerto Rico (FIPI program),
24) (a) Byler, D. M.; Susi, H. Biopolymers 1986, 25, 469-487. (b)
the NIH-AREA program (R-15-GM55896-01), and the NIH-
MBRS program (GM-08216) supported this work.
G o¨ rne-Tschelnokow, U.; Naumann, D.; Weise, C.; Hucho, F. Eur. J.
Biochem. 1993, 213, 1235-1242.
(25) Lozano, P.; De Diego, T.; Iborra, J. L. Eur. J. Biochem. 1997, 248,
8
0-85.
JA990515U