A drop of enantioselectivity in the Pseudomonas cepacia lipase-catalyzed ester hydrolysis is influenced by the chain length of the fatty acid

Article abstract of DOI:10.1016/j.tetasy.2004.06.010

A two-step molecular mechanics based computational procedure has been applied to explain the enantioselectivity observed in the hydrolysis of esters of primary alcohols, carried out in the presence of a lipase from Pseudomonas cepacia. This approach proved to be very effective in explaining an unpredictable drop in enantioselectivity, experimentally observed when the chain of the fatty acid was lengthened and to predict the chain length in correspondence of which the effect should have revealed itself.

Full text of DOI:10.1016/j.tetasy.2004.06.010

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 15 (2004) 2345–2350
A drop of enantioselectivity in the Pseudomonas cepacia
lipase-catalyzed ester hydrolysis is influenced by the chain
length of the fatty acid
Andrea Tafi,a Fabrizio Manetti,a Maurizio Botta,a,* Silvana Casatib and
Enzo Santaniellob,*
aDipartimento Farmaco Chimico Tecnologico, Universitꢀa degli Studi di Siena, Via Aldo Moro, 2 I-53100 Siena, Italy
bDipartimento di Scienze Precliniche L.I.T.A.-Vialba, Universitꢀa degli Studi di Milano, via G. B. Grassi,74 I-20157 Milano, Italy
Received 7 May 2004; accepted 10 June 2004
Available online 23 July 2004
Abstract—A two-step molecular mechanics based computational procedure has been applied to explain the enantioselectivity ob-
served in the hydrolysis of esters of primary alcohols, carried out in the presence of a lipase from Pseudomonas cepacia. This ap-
proach proved to be very effective in explaining an unpredictable drop in enantioselectivity, experimentally observed when the chain
of the fatty acid was lengthened and to predict the chain length in correspondence of which the effect should have revealed itself.
ꢀ 2004Elsevier Ltd. All rights reserved.
1. Introduction
over the course of a studied reaction, assuming that this
is a good mimic of the effectively relevant transition
state. Through such an approach,4a Tuomi and Kazl-
auskas were able to discuss the enantioselectivity of PcL
towards esters of primary alcohols in terms of compet-
ing recognition (experimentally evaluated by the
Michaelis–Menten constant) and reaction rate differ-
ences between enantiomers. These authors, however,
pointed out that their modelling could better account
for competing recognition. A pure THI model, which is
focused on the location of the lowest-energy structure,
does not favor a proper kinetic interpretation of the
enantioselectivity, based on the height of the energy
barrier in correspondence of the transition state (TS), as
it causes a rather thermodynamic perspective, based on
the depth of the tetrahedral intermediate’s energy hol-
low. The practical limitations of the pure THI approach
have been recently reported4e and consequently, a
‘pseudomolecular dynamics under constraints’ protocol
has been proposed with the aim of advancing the THI
modelling strategy.
The enantioselectivity of lipases is still mainly rational-
ized by using computational approaches exploiting
molecular mechanics (MM) related methods. Mixed
quantum mechanics/molecular mechanics (QM/MM)
methods have recently been applied with success to
rationalize the enantioselectivity of specific enzymes.1
However MM approaches are still attractive in model-
ling enzyme reactivity, despite their theoretical limita-
tions, due to their speed, flexibility, and reliability.2
Various models have been proposed over the last 10
years, with the aim of rationalizing or predicting the
behavior of several lipases towards their substrates.3 In
particular, the enantioselectivity and substrate specific-
ity of the lipase from Pseudomonas cepacia (PcL) has
been thoroughly studied and interpreted.4
The MM computational protocol, which is the most
commonly applied, consists of conformational searches
or molecular dynamics (MD) simulations to model the
lowest-energy tetrahedral intermediate (THI) formed by
each enantiomer (the fast- and the slow-reacting one)
A different kinetic approach, based on Deslongchamps’
stereoelectronic theory5 and theoretically verified by
QM calculations, was proposed and firstly applied in
1998 to rationalize the enantioselectivity of lipases.6
According to Ema’s studies, conformational require-
ments at the reacting center come into play in
* Corresponding authors. Tel.: +39-0577-234-306; fax: +39-0577-234-
333 (M.B.); tel.: +39-02-503-19693; fax: +39-02-382-10295; e-mail
addresses: botta@unisi.it; enzo.santaniello@unimi.it
0957-4166/$ - see front matter ꢀ 2004Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2004.06.010

2346
A. Tafi et al. / Tetrahedron: Asymmetry 15 (2004) 2345–2350
correspondence to the transition state leading to the
formation of the first THI and are responsible for the
enantioselective effect, by modulation of the reaction’s
activation energy. Repulsive interactions or steric
clashes inside the active site allow7 only one enantiomer
to get to the TS’s theoretically favored conformation
(gauche with respect to the C–Oalcohol–C–Oserine dihedral
angle); as a consequence, an energy gap of a few
kcal molꢀ1 is created between the two competing TSs.6
Notably, the stereoelectronic theory allows such ther-
modynamic concepts to be transformed into geometric
constraints (reflecting the energy difference between the
diastereomeric forms of the transition states) on the
three-dimensional structure of the THIs. Accordingly,
gauche, rather than anti conformations, must be
adopted by the tetrahedral intermediates to follow a low
activation energy reaction pathway.6;8
excellent enantioselectivity of the PcL-catalyzed hydro-
lysis of esters 1b and 1c dramatically fell when the chain
of the fatty acid was lengthened to esters 1d and 1e. The
experimental results shown in the Table suggest that the
extension of the acyl chain could produce a perturbation
on the reaction pathway nullifying the enantiopreference
shown by the enzyme towards the (S)-enantiomers of
compounds 1b and 1c. With the aim of clarifying this
unpredictable effect at a molecular level, the hydrolyses
of compounds 1b–e were simulated and the results ob-
tained for 1b and 1d, taken as representative examples of
the whole set of compounds, reported here in detail.
Table 1. PcL-catalyzed hydrolysis of esters 1b–f
Substrate
Time (h)
% Conv.a
Eepb
Ec
1b
1c
1d
1e
1f
2
32
27
>98
>98
>100
>100
2
7
2416
25
1.4
1.8
A novel two-step MM based modelling procedure has
recently been developed and successfully tested to ex-
plain the enantioselectivity observed in the hydrolysis of
esters of secondary alcohols in the presence of PcL.4b
This method exploits Ema’s results to involve the ste-
reoelectronic theory concepts into a MM approach,
which simulates the two main steps (recognition and
THI formation) of the lipase catalyzed reactions. For a
given substrate, docking studies enable us to detect
potentially productive Michaelis–Menten (M–M) com-
plexes of both enantiomers, which in turn are converted
into tetrahedral intermediates. After geometric optimi-
zation of the latter, the stereoelectronic theory con-
straints are used as an acceptance test, in order to
explain or predict the course of the simulated reaction.
Hereafter, this procedure will be named the ‘Essential
Pathway Approach’ (EPA).
426
3
32
80
13
a Determined by GC.
b Evaluated by 1N NMR analysis of the corresponding MTPA esters,
according to Dale et al.12
c Determined according to Chen et al.13
EPA was first applied to acetate 1b. The docking found a
quite similar recognition geometry for both enantiomers
of this substrate into the active site of PcL (corresponding
as a whole to an alignment already described for the same
compound4a;11). In particular, the orientation in 3D space
of the hydrogen and methyl groups at the stereocenter
(showing roughly exchanged positions) appeared to be
the only difference in the recognition of the two enantio-
mers of 1b by PcL. This theoretical achievement con-
firmed again that, even though the enzyme is able to bind
in a different way with the two enantiomers of its sub-
strates, the enantioselection of PcL does not come into
play during the recognition process.4;7 In fact, no struc-
tural enantioselectivity indicators4 came out from dock-
ing to hint at any chiral discrimination. When the THI of
both enantiomers was built and optimized a critical
enantioselective differentiation between the two diaste-
reomeric structures occurred, which fully rationalized the
experimental result of the hydrolysis reaction of this
compound.4b;6;7 In the case of (S)-1b, the methyl group at
the stereocenter was easily accommodated inside a small
hydrophobic pocket of PcL, formed by residues His86,
His286, Leu287, and Ile290,4a;b;13 allowing the THI to
adopt the expected productive gauche conformation
displayed in red in Figure 2. In the case of (R)-1b (yellow
in Fig. 2), even though the overall orientation of the M–
M complex was maintained, an unproductive anti-con-
formation of the THI was adopted, that prevented the
methyl group to be directed completely outwards with
respect to the small cavity mentioned above and to be
totally solvent exposed.
Herein, EPA is applied for the first time to explain results
obtained from the hydrolysis of racemic mixtures of
esters of 3-phenyl-2-methyl-1-propanol (compound 1a,
Fig. 1), performed in aqueous medium, in the presence of
PcL. Moreover, the original computational procedure
has been further improved by the adjustment of an energy
minimization protocol,9 which has allowed the evalua-
tion of the steric energy of enzyme–substrate complexes.
OR
1
d. R=CO(CH2)6CH3
e. R=CO(CH2)10CH3
f. R=CO(CH2)4CH3
a. R=H
b. R=COCH3
c. R=CO(CH2)2CH3
Figure 1. Structure of esters considered as substrates for PcL-catalyzed
hydrolysis reactions.
2. Results and discussion
EPA was then applied to the octanoate 1d (see Fig. 3).
Compound 1a is a primary alcohol whose stereochemi-
cal outcome in the PcL-catalyzed transesterification10
has been studied in detail from a MM point of view.4a;11
We observed (see Table 1 and the Experimental) that the
The study of the recognition of both enantiomers of this
substrate by PcL did not exhibit any relevant novelty
with respect to the corresponding enantiomers of 1b.

A. Tafi et al. / Tetrahedron: Asymmetry 15 (2004) 2345–2350
2347
covalent bond between the catalytic serine residue and
the carbonyl carbon of the substrate forces the latter to
approach PcL. Consequently, the acyl moiety must
rearrange into HA, crossing torsional energy barriers
and adopting a conformation more folded and higher in
energy, than the one displayed by the M–M complex
(three C–C torsion values were found to differ from 180ꢁ
and 60ꢁ).
In the light of the above reported computational results,
it was hypothesized that the refolding of the fatty acid
chain inside HA might produce a compulsory increase in
the activation energy of the rate-determining step (the
one leading to the first tetrahedral intermediate). This
would be effective on both enantiomers and be able to
cancel out any reaction rate difference between them,
otherwise referable to the exclusive possibility of the
(S)-enantiomer to take a gauche TS conformation.16
According to this hypothesis, any longer acyl chain may
directly agitate the system determining a new bottleneck
in the reaction pathway irrespective of the stereochem-
istry of the reacting substrate.
Figure 2. Superimposition of the calculated tetrahedral intermediates
of (S)-1b (red) and (R)-1b (yellow) located at their proper respective
3D positions inside the active site of PcL. Hydrogen atoms bound to
heteroatoms are displayed. For the sake of simplicity, only residues
useful for the discussion are shown. The surface of the pocket formed
by residues His86, His286, Leu287, and Ile290 is shown in transpar-
ency.14
With the aim of supporting such an assumption (based
on structural considerations) free energy differences
between the complexes with PcL of the (R)- and
(S)-enantiomers were roughly estimated.9;17 The values
obtained for compounds 1b–f, at both recognition and
THI formation steps, are shown in Table 2.
Table 2. Steric energy differences (kJ/mol) between PcL-enantiomer
(S and R) complexes, calculated for different substrates
a
a
Substrate
DEðR; SÞ M–Mb
DEðR; SÞ THIc
1b
1c
1d
1e
1f
6.7
5.3
5.2
6.0
4.0
)5.9
)4.0
0.9/6.0d
)1.4/0.3e
2.0
a DEðR; SÞ ¼ EðPcL-SÞ ꢀ EðPcL-RÞ.
Figure 3. Superimposition between calculated recognition geometry
(violet) and tetrahedral intermediate (red) of (S)-1d, located at their
proper respective 3D positions inside the active site of PcL. The fatty
acid chain occupies, folded, the HA binding pocket in the right side
of the figure.14
b Calculated at the recognition step (M–M complexes).
c Calculated at the THI formation step.
d Dependent on the conformation of the alcoholic moiety of the THI of
(S)-1d (0.9 gauche; 6.0, anti).16
e Dependent on the conformation of the alcoholic moiety of the THI of
(S)-1e ()1.4 gauche; 0.3, anti).16
The common alcoholic moieties were found to be per-
fectly superimposable with each other [(R)-1b with (R)-
1d and (S)-1b with (S)-1d] as expected, while the achiral
fatty acid chain of both enantiomers of 1d was folded
(one C–C–C–C torsion value was approximatively equal
to 60ꢁ) to occupy the so called HA binding pocket, ‘a
hydrophobic, well defined groove that is the binding site
for the sn-3 fatty acid chain of a phosphonate triacyl-
glycerol analogue inhibiting PcL’.15 When the THI of
both enantiomers of 1d was optimized (starting from
their respective lowest energy M–M complexes), while
the binding of the alcoholic moieties followed the same
pattern of 1b (shown in Fig. 2), a remarkable confor-
mational interconversion of the fatty acid chain
occurred, involving in a non-enantioselective manner
both diastereomeric enzyme–substrate complexes. The
case of (S)-1d is shown in Fig. 3: the formation of the
Although the accuracy of the differences reported in
Table 2 could be controversial for the approximations
used in their calculation (the constraints imposed during
the minimizations), the comparison among these values
proved to be useful in estimating, in a qualitative man-
ner, the enantioselective behavior displayed by PcL in
the hydrolysis reaction of the studied substrates. In
particular, the trend of the DE values calculated in
correspondence of the THI formation (third column in
the table) correctly confirms both the advantage of (S)-
1b with respect to (R)-1b in their hydrolysis in presence
of PcL (stereoelectronic requirements) and the drop in
enantioselectivity experimentally observed going from
1b to 1d, determined by the extension of the fatty acid
chain (overflowing conformational rearrangement).18

2348
A. Tafi et al. / Tetrahedron: Asymmetry 15 (2004) 2345–2350
3. Conclusion
the subsequent calculatioꢀn1s. 3LIP was first minimized
ꢁ
A gradient value was reached,
until a 1 · 10ꢀ1 kJ molꢀ1
The robustness of the model proposed herein to ratio-
nalize the behavior of PcL towards compounds 1b–d
was tested through the prediction of the fatty acid
chain length, in correspondence of which the drop of
enantioselectivity should have revealed itself. The
simulation of the hydrolysis of the hexanoate 1f was
performed and it was predicted that PcL should have
hydrolyzed this compound with lower enantioselectivity
than 1b and 1c, as suggested by both the folding of the
acyl chain into HA (results not shown) and the DEðR; SÞ
THI values (Table 2). The experimental outcome of the
hydrolysis of 1f confirms the prediction (see Table 1),
thus demonstrating the reliability of our computational
approach.
after having fixed all the backbone atoms at their crys-
tallographic positions in 3D space. The output structure
was then subjected to a step-wise simulated annealing
protocol (applying the same set of constraints to the
backbone atoms as in the preliminary minimization and
the SHAKE algorithm) to allow both side chains and
water molecules to mutually relocate in such a way as to
optimize their mutual interactions. A 10 ps molecular
dynamics (MD) run was firstly performed at 300 K with
a 1.5 fs time step for equilibration purposes. The next
MD run lasted 150 ps with the same time step, during
which the system was slowly cooled to approximately
150 K. The third and final MD run had the system cooled
to 50 K over 20 ps with a 2.0 fs time step. Finally, the
output structure was reminimized by applying the above
described set of constraints.
4. Experimental
4.1. PcL-catalyzed hydrolysis of esters 1b–f
4.3. Flexible docking of the substrates and construction of
their tetrahedral intermediates.4b
The highly enantioselective enzyme catalyzed hydrolyses
of acetate 1b and butyrate 1c proceeded in absence of
surfactant that was required for the hydrolysis of less
soluble longer chain esters. In a separate experiment, we
observed that the addition of Triton X 100 to these
enzymatic reactions did not influence the E values of the
hydrolysis of 1b and 1c, which remained >100. For the
sake of homogeneity, we herein report the experimental
protocol for all substrates in the presence of Triton X 100.
Three-dimensional structures of compounds (R)-1b, (R)-
1c, (R)-1d, (R)-1e, (R)-1f, (S)-1b, (S)-1c, (S)-1d, (S)-1e,
and (S)-1f were constructed and optimized in Macro-
Model/BatchMin. The BatchMin Monte Carlo Multiple
Minimum methodology (MCMM command) was then
chosen to carry out simulations involving 10,000 steps,
in which each substrate (guest compound) was subjected
to a flexible docking into PcL (host compound).
BatchMin command MOLS permitted us to perform
rotations and translations of the guest, with a maximum
value of the random rotational angle equal to 180ꢁ and a
To a suspension of the ester (0.8 mmol) in a phosphate
buffer (0.05 M at pH 7, 2 mL) under magnetic stirring,
Triton X 100 (50 mg) and PcL (Amano, 30 U/mg,
16 mg) were sequentially added. The reaction was
continuously monitored by a pH meter with the pH
kept at 7 by adding 0.1 M NaOH. When the required
conversion was reached, 1 M HCl and NaCl were ad-
ded, the reaction mixture extracted with dichlorome-
thane (3 · 5 mL) and dried over Na2SO4. Alcohol 1a
was purified by silica gel column chromatography
(petroleum ether/acetate, 8:2).
ꢁ
maximum allowed translation movement of 6.0 A at
each step. At the same time, the compound was sub-
jected to a statistical conformational search inside the
active site of PcL by random changes in the torsion
angles of all its rotatable bonds (4in the simplest case of
1b, up to 12 in the case of 1e) as given by the TORS
command. Values of 30ꢁ and 180ꢁ, respectively, were
chosen as the minimum and maximum angular incre-
ment to be added or subtracted at each Monte Carlo
step. Each docking simulation was repeated a few times
starting from different arbitrary geometries to produce a
complete sampling of the whole potential energy
hypersurface of the selected PcL-substrate system. En-
ergy minimizations of the output structures that satisfied
three distance constraints (distance between the car-
bonyl carbon of the guest and the oxygen atom (Oc) of
4.2. Preparation of the enzyme starting structure
The coordinates of Pseudomonas cepacia lipase (PcL)
utilized in this study were collected from the structure
with entry number 3LIP in the Brookhaven Protein Data
Bank.19 The ‘enzyme starting structure’ to be used in all
computations was prepared as follows. The coordinates
of PcL were red into the software package MacroModel/
BatchMin (version 5.5)20 equipped with the Amber*
united atoms force field.21 Accordingly to this choice,
hydrogen atoms were added to heteroatoms only. Bound
water molecules were maintained throughout the calcu-
lations with the exception of four located at the active
site cleft, which were removed to prevent hindrance to
the docking of substrates. The generalized Born solvent-
accessible surface area (GB/SA) continuum solvation
method of BatchMin was applied to simulate the effects
of water.22 Amber* and GB/SA water were used for all of
ꢁ
Ser87 were shorter than 4.5 A; distances between the
carbonyl oxygen of the guest and the backbone NHs of
ꢁ
Leu17 and Gln88 were shorter than 4.5 A) were carried
out in order to obtain eligible Michaelis–Menten (M–M)
complexes. BatchMin least square superimposition
routine (COMP command) was selected to eliminate
duplicate minima and the global chirality checking
command (CHIG) was used to reject any structure
whose chirality had been changed by the minimization.
Due to the large number of atoms in the whole system, a
region of PcL surrounding the active site cleft was
selected, which comprised of 71 residues plus 10 crystal
water molecules for a total of 692 atoms. All calcula-

A. Tafi et al. / Tetrahedron: Asymmetry 15 (2004) 2345–2350
2349
tions were performed on this internal subset, while the
external residues and water molecules were not included
in the calculations.
of PcL was maintained by fixing all the host atoms at
their starting positions, with the exception of the side
chains of Ser87, Asp264, His286, Glu289, and W510,
which were relaxed together with each substrate. In this
way unwanted steric energy fluctuations were com-
pletely avoided.
Two further restraints were imposed in order to main-
tain the 3D structure of the enzyme largely unmodified
during the docking simulations:
(a) the side chains of 18 residues inside the subset
(namely Leu17, Thr18, Tyr23, Tyr29, His86, Ser87,
Ser117, Ph119, Val123, Leu164, Leu167, Asp264,
Val266, Val267, His286, Leu287, Glu289, Ile290)
and four water molecules (W465, W501, W510,
W532), either localized on the walls of the active site
cleft or connected to residues playing a role in the
catalytic reaction, were fully minimized together
with the guest in order to guarantee the balance be-
tween the surfaces of the two partners;
Acknowledgements
We thank the ‘Centro Universitario per l’Informatica e
la Telematica’ of the University of Siena and the Uni-
versity of Milan for financial support (to E.S. and S.C.).
Support from the Research Training Network (HPRN-
CT-2000-00018) is gratefully acknowledged.
(b) all the other atoms of the internal subset were fixed
in 3D space even if their nonbonded interactions
with all the relaxing atoms were calculated.
References and notes
1. Colombo, G.; Carrea, G. J. Biotech. 2002, 96, 23–33, and
references cited therein.
The distance between the hydroxyl hydrogen atom of
the Ser87 and the adjacent Ne atom of His286, as well as
the distance between the hydrogen of imidazole NH of
His286 and the carboxylate ion of Asp264, were moni-
tored during the docking simulations to ensure that the
relative orientations of the residues of the catalytic triad
were compatible with the mechanism of the hydrolytic
reaction.
2. Kazlauskas, R. J. Curr. Opin. Chem. Biol. 2000, 4, 81–88,
and references cited therein.
3. Bornscheuer, U. T. Curr. Opin. Biotech. 2002, 13, 543–547,
and references cited therein.
4. (a) Tuomi, W. V.; Kazlauskas, R. J. J. Org. Chem. 1999,
64, 2638–2647; (b) Tafi, A.; van Almsick, A.; Corelli, F.;
Crusco, M.; Laumen, K. E.; Schneider, M. P.; Botta, M. J.
Org. Chem. 2000, 65, 3659–3665; (c) Gentner, C.; Schmid,
R. D.; Pleiss J. Colloid Surf. B 2002, 26, 57–66; (d) Tomic’,
S.; Kojic’-Prodic’, B. J. Mol. Graph. Model 2002, 21, 241–
252; (e) Guieysse, D.; Salagnad, C.; Monsan, P.; Remaud-
Simeon, M.; Tran, V. Tetrahedron: Asymmetry 2003, 14,
1807–1817.
5. Deslongchamps, P. Stereoelectronic Effects in Organic
Chemistry; Pergamon: Oxford, 1983.
6. Ema, T.; Kobayashi, J.; Maeno, S.; Sakai, T.; Utaka, T.
Bull. Chem. Soc. Jpn. 1998, 71, 443–453.
7. Ema, T.; Jittani, M.; Furuie, K.; Utaka, M.; Sakai, T. J.
Org. Chem. 2002, 67, 2144–2151.
8. Dugas, H. Bioorganic Chemistry. A Chemical Approach to
Enzyme Action, 3rd ed.; Springer: New York, 1996;
Chapter 4.
9. A relaxed structure of PcL, the so called ‘enzyme starting
structure (see Experimental), has been fully included and
kept invariably steady during these minimizations by
fixing all the enzyme atoms at the positions they occupied
in the input structure, with the exception of the side chains
of Ser87, Asp264, His286, Glu289, and W510, which were
relaxed together with each substrate. In this way the
uncontrollable and unwanted steric energy fluctuations,
previously detected by other authors in the same kind of
calculations,2;4a have been completely avoided furnishing
an original solution to the challenging task of the energy
evaluation of enzyme–substrate complexes.
10. Ferraboschi, P.; Casati, S.; De Grandi, S.; Grisenti, P.;
Santaniello, E. Biocatalysis 1994, 10, 279–288.
11. Tomic’, S.; Dobovicnik, V.; Sunjic’, V.; Kojic’-Prodic’, B.
Croat. Chem. Acta 2001, 74, 343–357.
12. Dale, J. A.; Dull, D. L.; Mosher, H. S. J. Org. Chem. 1969,
34, 2543–2549.
Families of the lowest energy output docking geometries
were transformed into the tetrahedral intermediates: one
long covalent bond was drawn between the Ser87 Oc
oxygen atom and the carbonyl carbon atom of the dif-
ferent compounds while transferring one H atom from
Ser87 to His286 (deletion and drawing of one bond
without modifying the position of the atom). Finally the
carbonyl group of the substrates and the imidazole ring
of His286 were changed into charged species (C–Oꢀ and
imidazolium ion, respectively). Direct energy minimi-
zations were then performed with MacroModel/Batch-
Min in order to optimize the new structures, while
maintaining the same set of restraints imposed during
the docking.
4.4. Evaluation of differences in free energy of activation
A suitable energy minimization protocol was set up, at
both recognition (M–M complexes) and tetrahedral
intermediate (THI) formation steps, in which the whole
structure of PcL was included, with the aim of esti-
mating (in a qualitative manner) the differences between
the free energy of activation of the enantiomers of
selected derivatives. Each input complex (enzyme and
bound substrate) was put together after having super-
imposed the so called ‘enzyme starting structure’ (see
above) to the output structure of PcL obtained from the
docking, by deletion of the latter (so as to locate
the substrate at its proper position into the active site of
the former). Geometry optimizations of the complexes
were then performed, during which the input structure
13. Chen, C.-S.; Fujiimoto, Y.; Girdaukas, G.; Sih, C. J. J.
Am. Chem. Soc. 1982, 104, 7294–7299.
14. The software package InsightII of Accelrys, San Diego,
CA, USA was used to perform the graphics manipula-
tions.

2350
A. Tafi et al. / Tetrahedron: Asymmetry 15 (2004) 2345–2350
15. Lang, D. A.; Mannesse, M. L. M.; De Haas, G. H.; Verheij,
H. M.; Dijkstra, W. Eur. J. Biochem. 1998, 254, 333–340.
16. On this subject it has to be noted that, when the THIs
deriving from the global minimum recognition geometry
of (S)-1d and (R)-1d were optimized, the gauche-anti
enantioselective differentiation of the alcoholic moiety,
already described for 1b, was observed again. Some M–M
complexes of (S)-1d were found however (at energy values
very close to the global minimum), whose acyl chains were
occupying the HA binding pocket with slightly different
folding patterns, in accordance with the observation that
HA snugly accommodates fatty acid chains.15 When a few
of these geometries were transformed into the correspond-
ing THI’s and relaxed, an anti-conformation of the
alcoholic moiety was detected, probably due to propaga-
tion of the perturbation caused by the rearrangement of
the acyl chain into HA. Such a gauche-anti conformational
interconversion, if real, would be even more effective in
slowing down the reaction speed of the fast-reacting
enantiomer in light of the stereoelectronic theory require-
ments (see the third column of Table 2). A similar overall
picture––not discussed in the text––was obtained too,
when the hydrolysis of 1e was simulated.
17. Haeffner, F.; Norin, T.; Hult, K. Biophys. J. 1998, 74,
1251–1262.
18. The DE values reported in the second column of Table 2
suggest a steady preference of PcL towards the (R)-
enantiomers of the primary alcohols during the recogni-
tion step. The trend of these data––related to the exper-
imentally observed drop of enantioselectivity––could
be interpreted as a further sign that lipase enantioselec-
tivity does not come into play during the recognition
process.
19. Schrag, J. D.; Li, Y.; Cygler, M.; Lang, D.; Burgdorf, T.;
Hecht, H.-J.; Schmid, R.; Schomburg, D.; Rydel, T. J.;
Oliver, J. D.; Strickland, L. C.; Dunaway, M. C.; Larson,
S. B.; Day, J.; McPherson, A. Structure 1997, 5,
187.
20. Mohamadi, F.; Richards, N. G. J.; Guida, W. C.;
Liskamp, R.; Lipton, M.; Caufield, C.; Chang, G.;
Hendrickson, T.; Still, W. C. J. Comp. Chem. 1990, 11,
440.
21. McDonald, D. Q.; Still, W. C. Tetrahedron Lett. 1992, 33,
7743.
22. Still, W. C.; Tempczyk, A.; Hawley, R. C.; Hendrickson,
T. J. Am. Chem. Soc. 1990, 112, 6127.