Switched enantiopreference of Humicola lipase for 2-phenoxyalkanoic acid ester homologs can be rationalized by different substrate binding modes

Article abstract of DOI:10.1016/S0957-4166(99)00438-3

Humicola lanuginosa lipase was used for enantioselective hydrolyses of a series of homologous 2-phenoxyalkanoic acid ethyl esters. The enantioselectivity (E-value) of the enzyme changed from an (R)-enantiomer preference for the smallest substrate, 2-pheno

Full text of DOI:10.1016/S0957-4166(99)00438-3

Pergamon
Tetrahedron: Asymmetry 10 (1999) 4191–4202
Switched enantiopreference of Humicola lipase for
2-phenoxyalkanoic acid ester homologs can be rationalized by
different substrate binding modes
Per Berglund,a,∗ Imre Vallikivi,b Linda Fransson,a Heinz Dannacher,c Mats Holmquist,a
Mats Martinelle,a Fredrik Björkling,c Omar Parve b and Karl Hult a
aDepartment of Biotechnology, Royal Institute of Technology, SE-100 44 Stockholm, Sweden
bDepartment of Bioorganic Chemistry, Institute of Chemistry at Tallinn Technical University, Akadeemia tee 15, Tallinn 12618,
Estonia
cLeo Pharmaceutical Products, Industriparken 55, DK-2750 Ballerup, Denmark
Received 22 September 1999; accepted 4 October 1999
Abstract
Humicola lanuginosa lipase was used for enantioselective hydrolyses of a series of homologous 2-
phenoxyalkanoic acid ethyl esters. The enantioselectivity (E-value) of the enzyme changed from an (R)-enantiomer
preference for the smallest substrate, 2-phenoxypropanoic acid ester, to an (S)-enantiomer preference for the
homologous esters with longer acyl moieties. The E-values span the range from E=13 (R) to E=56 (S). A
molecular modeling study identified two different substrate-binding modes for each enantiomer. We found that the
enantiomers favored different modes. This discovery provided a model that offered a rational explanation for the
observed switch in enantioselectivity. © 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction
Lipases are frequently used in various chemical applications today and more than 30 lipases are
commercially available.1–4 Due to their generally excellent ability in enantiomer discrimination they
have a widely recognized potential for the production of enantiomerically pure compounds.5 They are
frequently used both in kinetic resolutions of racemates and in asymmetrizations of prochiral and meso
compounds.6,7 However, identifying the best lipase for a given transformation often leads to a necessary
but expensive and time-consuming screening procedure. One approach to overcome this limitation would
∗
Corresponding author. Tel: +46 8 790 75 12; fax: +46 8 22 46 01; e-mail: per.berglund@biochem.kth.se
0957-4166/99/$ - see front matter © 1999 Elsevier Science Ltd. All rights reserved.
PII: S0957-4166(99)00438-3

4192
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
be the rational tailoring of enzyme enantioselectivity by protein engineering methods. In order to achieve
this, a more detailed knowledge of lipase catalysis on a molecular level is needed.
Chiral carboxylic acids are important building blocks for the synthesis of many pharmaceuticals,5
pesticides,8 and natural compounds such as pheromones,9 where often substances of very high enantio-
meric excess (>99.5% ee) are needed. Lipase-catalyzed resolution of racemic carboxylic acids is kinet-
ically more complex than the resolution of racemic alcohols since, in the former case, two diastereomeric
acyl enzymes are formed.10 Hence, both the formation and breakdown of the acyl enzymes pass through
transition states that might be involved in defining the enantioselectivity of the resolution of a chiral
acid.11
Switched enantiomer preference within a homologous series of racemic chiral acyl donors has
previously been observed for instance with the serine hydrolase subtilisin.12 In this case, the existence
of two different modes of binding the substrate to the enzyme active site explaining the change in
enantiopreference has recently been confirmed by X-ray crystallography using enantiomeric boronic
acid inhibitors.13 In Candida rugosa lipase, methyl-branched chiral fatty acids have been used as model
compounds and modeling studies have predicted the existence of two different modes of binding of
the enantiomers of such a chiral acyl donor to the active site.14 The fast reacting enantiomer orients its
hydrophobic acyl chain into the active-site tunnel, which is a unique feature of this kind of lipase,15
whereas the slow reacting enantiomer leaves the tunnel empty. The existence of these two substrate-
binding modes has recently been experimentally confirmed.16 Similarly, for Candida antarctica lipase
B, it has recently been shown that 4-methyl-branched acids can be resolved, and that the enantiomers
orient their acyl chain differently in the active site of the enzyme.17 Two different modes of binding
substrate enantiomers have also been described for homologous secondary alcohols with C. antarctica
lipase B.18 In this case, longer alcohols have an inverted orientation in the enzyme active site.
Humicola lanuginosa lipase (HLL) has a molecular weight of 30 kDa. The X-ray crystal structure of
HLL is known.19 A helix, a lid, covers the superficial active site of the lipase in its closed conformation.
HLL is produced recombinantly in extensive amounts (Lipolase®, Novo Nordisk A/S) and can readily be
obtained in large quantities and high purity. According to the crystal structure of an HLL–inhibitor com-
plex, the lipase has a hydrophobic crevice, which extends from the active-site serine and accommodates
the first seven to eight carbons of the acyl chain.19 Also, tryptophane 89 (Trp89) in the lid of the lipase
has been shown to be important for binding the acyl chain of natural substrates.20 However, Trp89 has a
negative impact on the rate and enantioselectivity in hydrolysis of non-natural substrates such as racemic
2-alkyl substituted alkanoates.21
In this work, we are studying lipase catalysis and the lipase enantiomer discrimination processes in
detail using rational substrate engineering in combination with molecular modeling. This provides new
knowledge of substrate/enzyme interactions and of the mechanisms controlling enzyme enantioselec-
tivity, and will guide future protein engineering attempts for controlling and maximizing selectivity of
synthetically useful enzymes.
2. Results
Racemic ethyl 2-phenoxyalkanoates 1–7, the related methyl ester 8, and the phenyl-substituted analog
9 were synthesized from the corresponding 2-bromoalkanoyl esters and phenol in good yields.

P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
4193
Humicola lanuginosa lipase-catalyzed hydrolyses of 1–9 were performed in a pH-stat instrument. The
initial rates and enantioselectivity (E) displayed by HLL as well as the sign of the optical rotation for the
remaining unhydrolyzed substrate are shown in Table 1.
Table 1
Results of the HLL-catalyzed hydrolyses of compounds 1–9
The first seven substrates with increasing acyl chain lengths, 1–7, displayed a gradual change in
enantioselectivity with (R)-enantiomer preference for 1 switching to (S) preference for the substrates
with longer acyl chains. Substrate 2 is close to the switch and displayed very low enantioselectivity with
(S)-enantiomer preference. Changing the alcohol moiety from ethyl in 1 to methyl in 8 had little effect
on both the reaction rate and enantioselectivity. The phenyl-substituted 9 was slow and displayed a low
enantioselectivity.
The highest E-value and a high initial rate were seen with substrate 5. When a mutant of HLL was used
where the tryptophan residue in the lid of the lipase had been replaced with a glycine (HLL-W89G), the
initial rate towards 5 increased twofold but the E-value remained virtually the same (Table 2).
In order to rationalize the fact that the lipase changed its enantiopreference from (R) to (S) with
an increasing alkyl chain length on the stereocenter (substrates 1–7), a molecular modeling study was
undertaken. Initially, two substrates, 1 and 5, were selected for modeling. These substrates displayed the
largest difference in enantioselectivity within the homologous series and represented the extremes of E-
values with opposite enantiopreference. Substrate 6, displaying the second highest E-value, was included
at a later stage to give a larger data set.

4194
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
Table 2
Hydrolysis of substrate 5 catalyzed by HLL and the mutant HLL-W89G
The basis for lipase enantioselectivity is the difference in transition-state energy between the reacting
enantiomers.11 In serine hydrolases the tetrahedral intermediate formed by the substrate after nucleophilic
attack by the serine has been used as a model for the transition state of the reaction. An ester substrate
with a chiral acyl part forms two diastereomeric acyl enzymes. In the subsequent step a water molecule
cleaves the acyl enzyme. In principle, both these steps can contribute to the overall enantioselectivity
in the reaction.23 However, formation of the acyl enzyme can be considered irreversible in a hydrolytic
reaction. We thus focused the modeling studies on the acylation step intermediate.
It is important to note that in molecular modeling only one point along a reaction coordinate will be
picked from the complete reaction path. The modeler manufactures a starting structure, and then the
simulation gives an ensemble of structures for that single point. This means that it is possible to give
the system an unrealistic starting structure which is not on the reaction path at all, i.e. starting from an
enzyme–substrate complex that is not reachable for a true process. A minimum requirement is thus that
effort is put into finding an unstrained starting structure for the modeling. If the achieved ensemble of
structures are diverging from the initial structure although the starting complex was tension free, then we
are likely to having simulated a non-reachable complex.
Not all of the possible tetrahedral intermediates were considered reactive. In order to be catalytically
active, the tetrahedral intermediate was required to form five hydrogen bonds that are necessary for sta-
bilization of the transition state18 (Fig. 1). Furthermore, it was assumed that an additional hydrogen bond
involving the hydrogen on the catalytic histidine would severely influence the catalysis by restricting the
transfer of the hydrogen to the leaving alcohol.
Figure 1. Tetrahedral intermediate of the acylation step used as the transition-state mimic in the molecular modeling studies.
The five hydrogen bonds required for catalysis are shown in dashed lines. An additional requirement was imposed in that no
additional hydrogen bond (crossed-out) should be formed with the hydrogen on the histidine
To rationalize the observed reversed enantiopreference two questions need to be answered: (i) how do
the substrates bind in the active site? and (ii) which enantiomer is the fastest reacting one?
2.1. How do the substrates bind in the active site?
To explore how the tetrahedral intermediates bind in HLL, they were first manually built into the
active site. The position of the central part of the substrates was well defined due to the requirement of
the formation of catalytic hydrogen bonds. The alcohol part (ethyl) of the substrates was small and had

P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
4195
plenty of space in the alcohol pocket. For the acyl parts of the substrates more considerations had to be
made. The acyl parts of the substrates contained the stereocenter with three different groups bound: an
alkyl chain of varying lengths, a phenoxy group, and a hydrogen. The acyl-binding site of the HLL active
site consists of a large crevice along the surface. From crystal structures of lipase–inhibitor complexes it
is known that a crevice is used to accommodate one group from the acyl part of the substrate.19 Since the
acyl part had three different substituents on the stereocenter, there were three possibilities, three modes,
to position the substrate. In HLL, however, only two modes were possible. The crevice is too narrow
to accept a larger group than a hydrogen pointing into its sidewall. This means that the modes with
the hydrogen positioned in the crevice would be impossible. Thus, there were two modes, one with the
phenoxy group accommodating the crevice, the down mode (with the alkyl chain pointing upwards, out
from the enzyme), and the other mode with the phenoxy group pointing upwards, the up mode (the alkyl
chain positioned in the crevice) (Fig. 2).
Figure 2. Four combinations of substrate configuration and orientation exemplified for substrate 5. Index up refers to the
phenoxy substituent pointing outwards from the enzyme with the acyl chain in the crevice, while down means the opposite
To further explore the two modes of substrate binding, dynamics simulations were undertaken for both
binding modes of each enantiomer of the three substrates 1, 5, and 6, a total of 12 simulations. When the
simulations were examined, it was noticed that the down mode for the (S)-enantiomers of all substrates
(Sdown) never found a stable conformation. One explanation could be that the conformation forced onto
the system by introducing this particular combination of configuration and orientation never would have
occurred in reality. This led us to reject the Sdown mode. Further, the up mode for all (R)-enantiomers (as
well as Sdown for 1) developed a catalytically unfavorable hydrogen bond between the catalytic histidine
and the oxygen on the phenoxy group of the substrate. Therefore, the Rup binding mode was rejected as
well.
The remaining modes were Sup and Rdown, and it was then concluded that the (S)-enantiomers of the
substrates reacted having their phenoxy substituent pointing upwards, while the (R)-enantiomers had the
phenoxy group in the crevice (Table 3).
2.2. Which enantiomer is the fastest reacting one?
We had a hypothesis where we assumed that each substrate utilizes the binding energy provided by the
crevice as much as possible, in accordance with the binding of the inhibitor in the inhibited enzyme.19
Therefore, it was assumed that for substrate 1, with a short alkyl chain (methyl), the enantiomer preferring
the phenoxy group in the crevice (the down mode) would be favored. For substrates 5 and 6, however,
the binding energy would be best utilized with the more hydrophobic alkyl chains (pentyl and hexyl)
accommodated by the crevice and the phenoxy group pointing upwards, the up mode. This assumption
predicts a preference of the lipase for the (R)-enantiomer for substrate 1 and the (S)-enantiomer for
substrates 5 and 6, which is in accordance with the experimental results (Table 1). To support these

4196
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
Table 3
Evaluation of the 12 simulations
arguments, the average potential energies for the last 30 ps of the dynamics simulations were calculated
(Table 4). The energies predicted the right enantiomer as well.
Table 4
Average potential energies (without crystal waters) calculated for the last 30 ps of the dynamics
simulations for the reactive modes Rdown and Sup
3. Discussion
The suggested explanation for the switched enantiopreference can be summarized in three statements:
(i) there are two different modes of positioning the substrate in the lipase active site — either the phenoxy
group is placed in the crevice (the down mode), or the alkyl chain (the up mode); (ii) the two modes are
preferred by different enantiomers — (R)-enantiomers react in the down mode and (S)-enantiomers in
the up mode; and (iii) the enantiopreference of a certain substrate is controlled by optimal utilization of
binding energy of the acyl group crevice.
The modeling was performed on the substrates 1, 5, and 6, but the general conclusions could be
tested on other substrates as well: the homologous series 1–7 was correctly predicted, although no
prediction could be made regarding the point where the switch in enantiopreference occurred. Compound
9 was of interest from this point of view. The phenoxy and phenyl groups are very similar. According
to the above reasoning, the crevice would probably accommodate the phenoxy group better than

P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
4197
the phenyl, which predicts a down mode from which an (R) preference follows. Since the absolute
configuration of (−)-9 is unknown, we could not confirm the predicted enzyme enantiomer preference
but the low enantioselectivity (E=2) observed was in accordance with the small difference between the
two substituents.
Another aspect of the switch in enantiopreference was seen if the data set was arranged to show the
free energy difference of the diastereomeric transition state–lipase complexes (∆∆G=) as a function of
the length of the acyl chain for substrates 1–7 (Fig. 3). Increasing the chain length increased the free
energy difference between the transition states, but only to a certain degree. Further elongation did not
affect the free energy difference markedly.
Figure 3. (a) Enantioselectivity E as a function of the number of carbon atoms n; (b) enantioselectivity expressed as the free
energy difference in the transition state (∆∆G==RTlnE). The free energy difference increased with increasing chain length but
remained fairly constant after n=4 (substrate 5)
It has been demonstrated that Trp89 in the lid plays a role in the hydrolysis of substrates.20,21 The
activity displayed by the Trp89Gly mutant of HLL for substrate 5 was twofold higher than wild-type HLL
(Table 2), but the E-value did not change significantly. This can be explained by unfavorable interactions
between the side chain of Trp89, covering the active site, and the part of the acyl chain that is directed
outwards from the crevice. The unfavorable interaction would be similar for both the (R)-enantiomer and
the (S)-enantiomer of substrate 5, resulting in similar E-values both for wild-type and mutant HLL, but a
higher overall activity in the latter case.
4. Conclusion
We have shown that the enantioselectivity of HLL is reversed within a homologous series of substrates.
Molecular modeling suggests that the (R)-enantiomers of all substrates investigated orient their phenoxy
substituent in the hydrophobic crevice. In contrast, the (S)-enantiomers orient their phenoxy group out
from the active site and accommodate their alkyl chain in the crevice. The results from molecular
modeling support the hypothesis that the catalytically relevant substrate binding mode giving the optimal
interactions with the acyl binding crevice is the favorable one, and is the fast reacting substrate binding
mode. Therefore, the switch in enantiomer preference of the enzyme is a consequence of a change of the
energetically favorable substrate-binding mode.
Furthermore, we have shown that for H. lanuginosa lipase and 2-phenoxycarboxylic acid esters, the
highest enantioselectivities were obtained with (S) preference for 5 and with (R) preference for 1 and 8.
It is notable from the series of homologous substrates that a small change in substrate structure causes
a reversed enantiomer preference of the enzyme. These findings will assist future protein engineering
work, and they facilitate rational substrate engineering in order to control and maximize enzyme
enantioselectivity in synthetic applications.

4198
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
5. Experimental
5.1. General
MOPS and gum arabic were purchased from Sigma (St. Louis, MO); ACN from Fisher Scientific
(Leicestershire, UK); 1-PrOH, 2-PrOH from Riedel-de Häen (Germany); NaH2PO4·H2O from J. T.
Baker Chemicals B.V. (Deventer, Holland); Na2HPO4·12H2O, HCl, LiOH, NaOH, and CaCl2 from
Merck (Darmstadt, Germany); Na-citrate from Kebo (Stockholm, Sweden). All chemicals used were of
analytical or HPLC grade. Silica gel 60 F254 pre-coated TLC aluminum sheets from Merck (Darmstadt,
1
Germany) were used. H and 13C NMR spectra were recorded on a Bruker ARX300 spectrometer at
300 and 75 MHz, respectively, using CDCl3 as solvent and tetramethylsilane (TMS) as internal standard.
Mass spectra were recorded at 70 eV using a Micromass, VG Autospec, spectrometer.
5.2. α-Substituted carboxylic acids, method A
An equimolar amount of 2-bromoalkanoyl ester, phenol and an excess of K2CO3 was heated in acetone
for about 70 h. The mixture was poured into water and extracted with Et2O, and then washed (NaCl, sat.
aq.) and dried.
5.3. α-Substituted carboxylic acids, method B
An equimolar amount of 2-bromoalkanoyl ester and potassium phenolate was stirred in DMF at 25°C
for 24 h. Workup as above (method A).
5.4. α-Substituted carboxylic acids, method C
An equimolar amount of 2-bromoalkanoyl ester, 2-hydroxybenzaldehyde and an excess of K2CO3 was
stirred in DMF at 25°C for 24 h. Workup as above (method A).
5.5. Ethyl 2-phenoxypropanoate 124
1
Prepared according to method A in 31% isolated yield. H NMR δ: 1.22 (3H, t), 1.61 (3H, d), 4.21
(2H, q), 4.74 (1H, q), 6.82 (2H, d), 6.96 (2H, t), 7.27 (1H, t).
5.6. Ethyl 2-phenoxybutanoate 225
1
Prepared according to method B in 75% isolated yield. H NMR δ: 1.08 (3H, t), 1.24 (3H, t), 1.99
(2H, dt), 4.21 (2H, q), 4.55 (1H, t), 6.88 (2H, d), 6.97 (2H, t), 7.26 (1H, t). LRMS m/z: 208 (M+·), 135,
131, 107, 94, 77.
5.7. Ethyl 2-phenoxypentanoate 325
Prepared according to method A in 39% isolated yield. 1H NMR δ: 0.97 (3H, t), 1.24 (3H, t), 1.57 (2H,
m), 1.92 (2H, m), 4.20 (2H, q), 4.60 (1H, dd), 6.86 (2H, d), 6.95 (2H, t), 7.27 (1H, t). HRMS calculated
for C13H18O3 (M+·): 222.1256; found: 222.1259. LRMS m/z: 222 (M+·), 149, 129, 107, 94.

P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
4199
5.8. Ethyl 2-phenoxyhexanoate 425
1
Prepared according to method B in 71% isolated yield. H NMR δ: 0.92 (3H, t), 1.26 (3H, t), 1.40
(2H, q), 1.50 (2H, m), 1.95 (2H, m), 4.21 (2H, q), 4.60 (1H, dd), 6.87 (2H, d), 6.96 (2H, t), 7.27 (1H, t).
5.9. Ethyl 2-phenoxyheptanoate 525
1
Prepared according to method B in 92% isolated yield. H NMR δ: 0.89 (3H, t), 1.24 (3H, t), 1.33
(4H, m), 1.55 (2H, m), 1.93 (2H, m), 4.21 (2H, q), 4.59 (1H, dd), 6.87 (2H, d), 6.96 (2H, t), 7.26 (1H, t).
5.10. Ethyl 2-phenoxyoctanoate 6
1
Prepared according to method B in 64% isolated yield. H NMR δ: 0.88 (3H, t), 1.28 (3H, t), 1.33
(6H, m), 1.52 (2H, m), 1.93 (2H, m), 4.21 (2H, q), 4.58 (1H, dd), 6.87 (2H, d), 6.96 (2H, t), 7.26 (1H, t);
13C NMR δ: 172.0, 158.0, 129.5, 121.5, 115.1, 76.8, 61.1, 32.9, 31.6, 28.9, 25.2, 22.5, 14.2, 14.0. HRMS
calculated for C16H24O3 (M+·): 264.1725; found: 264.1721. LRMS m/z 264 (M+), 191, 171, 97, 94, 55.
5.11. Ethyl 2-phenoxydecanoate 7
1
Prepared according to method B in 66% isolated yield. H NMR δ: 0.87 (3H, t), 1.27 (3H, t), 1.29
(10H, m), 1.50 (2H, m), 1.93 (2H, m), 4.21 (2H, q), 4.58 (1H, dd), 6.86 (2H, dd), 6.96 (2H, t), 7.26 (1H,
t); 13C NMR δ: 172.0, 158.0, 129.5, 121.5, 115.1, 76.8, 61.1, 32.9, 31.8, 29.4, 29.2 (2C), 25.2, 22.7,
14.2, 14.1. HRMS calculated for C18H28O3 (M+·): 292.2038; found: 292.2040. LRMS m/z: 292 (M+·),
219, 199, 94, 69, 55.
5.12. Methyl 2-phenoxypropanoate 825
1
Prepared according to method A in 29% isolated yield. H NMR δ: 1.61 (3H, d), 3.74 (3H, s), 4.76
(1H, q), 6.85 (2H, d), 6.96 (2H, t), 7.27 (1H, t).
5.13. Ethyl 2-phenoxy-2-phenylethanoate 926
1
Prepared according to method A in 98% isolated yield. H NMR δ: 1.18 (3H, t), 4.17 (2H, m), 5.62
(1H, s), 6.95 (3H, m), 7.26 (2H, t), 7.37 (3H, m), 7.58 (2H, d).
5.14. Enzymes
Highly purified samples of H. lanuginosa lipase and of the Trp89Gly mutant of HLL were from Novo
Nordisk A/S, Denmark. The enzyme preparations were pure according to SDS-PAGE (sodium dodecyl
sulfate-polyacrylamide gel electrophoresis) with Commassie Blue staining. The enzyme was dissolved
in 10 mM MOPS buffer with pH 7.5. Enzyme concentration was determined spectrophotometrically at
280 nm (HLL ε=4.3×104 M−1 cm−1, HLL-W89G ε=3.6×104 M−1 cm−1).21

4200
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
5.15. Enzymatic hydrolysis of esters
The hydrolyses were run with a pH-stat instrument equipped with an ABU91 Autoburette (1 ml)
connected to a VIT90 Video Titrator (Radiometer, Copenhagen). Hydrolysis procedure: The substrate
emulsion (1.5 ml, 0.2 M substrate, 5% w/v gum arabicum, 0.2 M CaCl2, pH 7.5) was first emulsified by
sonication over 1 min. The reaction was then started by the addition of an enzyme solution to the stirred
thermostated substrate emulsion (5–20 µl of an enzyme solution of 0.5–1.0 mg/ml in 10 mM MOPS
buffer, pH 7.5). Titration was performed with NaOH (10 or 100 mM) at pH 7.5 and 25°C. The reaction
times were 30 min to 6 h. For HPLC analyses, samples (10–30 µl) were withdrawn at different stages of
conversion and were added to a water solution of pH 1 in order to stop the reactions.
5.16. Determination of enantiomeric excess
Analyses of the enantiomeric excess were performed on a HPLC system equipped with a LC pump
(Shimadzu LC6A), a CHIRAL-AGP column 100.4 (ChromTech, Sweden), a UV–vis spectrophotometric
detector (Shimadzu SPD-6AV), and an integrator (Perkin–Elmer LCI-100). Also, an HPLC system from
Waters was used equipped with a 486 tuneable detector, 600S system controller and a Waters 616 pump.
The product/substrate mixtures were extracted from the acidic water phase with diethyl ether. The ether
extract was then evaporated and the mobile phase (0.1 M Na-citrate, pH 3.5, or 0.1 M phosphate, pH
7, containing 1–25% of either ACN, 1-PrOH or 2-PrOH) was added to dissolve the product/substrate
mixture. The detection was carried out at 225 or 265 nm, depending on the buffer, and the flow rate was
0.9 ml/min. The ees and eep values were calculated from the base-line separated peak areas. The observed
change of enantiopreference was supported by the chiral HPLC analysis. The reacting enantiomer was
eluted first for substrate 3 and second for substrates 1 and 8.
5.17. Determination of the optical rotation
The substrate was hydrolyzed to ∼40% conversion using the pH-stat, and the remaining substrate and
product were extracted from the water phase using diethyl ether. The ether phase was then removed and
the extracted substance was flash-chromatographed on silica gel (Merck 60) with n-hexane:ethyl acetate
(92.5:7.5) as eluent. The purities of the fractions containing the substrate ester were determined by TLC.
The signs of the optical rotations of the esters were measured in n-hexane at 546 nm using a Perkin–Elmer
Polarimeter 241 instrument.
5.18. Calculation of E-values
The E-values were determined by using the enantiomeric excess (eep and ees) values and conversion
values from pH-stat runs. A curve fit of numeric data (eep, ees and/or conversion) was performed by
using the program E&K calculator (Anthonsen, H. W. E&K Calculator, version 2.03 for Macintosh. For
information about the program: http://bendik.mnfak.unit.no), based on the equations by Chen et al.27,28
For every E-value, at least four data points were used for the curve fit.
5.19. Molecular modeling
The molecular modeling was based on a crystal structure of HLL19 with a bound inhibitor showing the
lipase in an open conformation. Hydrogens were added to the heavy atoms, and the structure was allowed

P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
4201
to relax in the force field: all enzyme hydrogens were allowed to move during a dynamics simulation of
1000 fs length, and the energy for the hydrogens in the last structure was minimized. The above steps
were repeated for the water hydrogens. Then the energy for all hydrogens was minimized and finally the
energy of all atoms was minimized. The inhibitor was then removed and replaced with a built substrate.
The substrates were assigned empirical charges29 and atom types consistent with the force field. The
substrate was allowed to relax in the enzyme through repeated dynamics simulations and minimizations
on subgroups of the substrates: first the hydrocarbon chain in the acyl part of the substrate was allowed
to move for 1 ps and the last structure was minimized. Then the phenoxy group, the alcohol group, and
finally the whole substrate one at a time were allowed to relax in the same way, whereafter, the energy of
the whole system was minimized. The same procedure was repeated for two different conformations of
the two enantiomers for three substrates, giving a total of 12 structures. These 12 minimized structures
were the starting points for dynamics simulations, lasting for 100 ps each. In the dynamics simulations
all atoms were allowed to move. All bond lengths were set to constant values using SHAKE,30 which
made it possible to use a time step of 2 fs. The structure was heated with 50 K per 10 ps up to 300 K,
giving a total heating time of 50 ps. All modeling was performed with the software package SYBYL6.5
(Sybyl6.5, Tripos Inc., 1699 South Hanley Rd., St Louis, Missouri, 63144, USA) on an SGI Octane
computer. Tripos’ implementation of the Amber force field31,32 considering all atoms was used.
Acknowledgements
Financial support from the Swedish Council for Forestry and Agricultural Research (SJFR) and
from the Swedish National Board for Industrial and Technical Development (NUTEK) is gratefully
acknowledged.
References
1. Wong, C.-H.; Whitesides, G. M. Enzymes in Synthetic Organic Chemistry; Pergamon Press: Oxford, 1994.
2. Faber, K. Biotransformations in Organic Chemistry, 3rd ed.; Springer-Verlag: Heidelberg, 1997.
3. Kazlauskas, R. J.; Bornscheuer, U. T. In Biotechnology, 2nd ed., Biotransformations I; Kelly, D. R., Ed. Biotransformations
with lipases. VCH: Weinheim, 1998; Vol. 8a, pp. 37–191.
4. Schmid, R. D.; Verger, R. Angew. Chem., Int. Ed. Engl. 1998, 37, 1609–1633.
5. Zaks, A.; Dodds, D. R. Drug Discovery Today 1997, 2, 513–531.
6. Johnson, C. R. Acc. Chem. Res. 1998, 31, 333–341.
7. Berglund, P.; Hult, K. In Stereoselective Biocatalysis; Patel, R., Ed. Biocatalytic synthesis of enantiopure compounds using
lipases. Marcel Dekker: New York, 1999; pp. 633–657.
8. Stereoselectivity of Pesticides: Biological and Chemical Problems; Ariëns, E. J.; van Rensen, J. J. S.; Welling, W., Eds.;
Elsevier: Amsterdam, 1988.
9. Högberg, H.-E.; Berglund, P.; Edlund, H.; Fägerhag, J.; Hedenström, E.; Lundh, M.; Nordin, O.; Servi, S.; Vörde. C. Catal.
Today 1994, 22, 591–606.
10. Hult, K.; Holmquist, M. Methods Enzymol. 1997, 286, 386–405.
11. Sih, C. J.; Wu, S.-H. In Topics in Stereochemistry; Eliel, E. L.; Wilen, S. H., Eds.; Wiley: New York, 1989; Vol. 19, pp.
63–125.
12. Martichonok, V.; Jones, J. B. J. Am. Chem. Soc. 1996, 118, 950–958.
13. Stoll, V. S.; Eger, B. T.; Hynes, R. C.; Martichonok, V.; Jones, J. B.; Pai, E. F. Biochemistry 1998, 37, 451–462.
14. Holmquist, M.; Hæffner, F.; Norin, T.; Hult, K. Protein Sci. 1996, 5, 83–88.
15. Grochulski, P.; Bouthillier, F.; Kazlauskas, R. J.; Serreqi, A. N.; Schrag, J. D.; Ziomek, E.; Cygler, M. Biochemistry 1994,
33, 3490–3500.
16. Berglund, P.; Holmquist, P.; Hult, K. J. Mol. Catal. B: Enzym. 1998, 5, 283–287.

4202
P. Berglund et al. / Tetrahedron: Asymmetry 10 (1999) 4191–4202
17. Heinsman, N. W. J. T.; Orrenius, S. C.; Marcelis, C. L. M.; De Sousa Teixeira, A.; Franssen, M. C. R.; Van Der Padt, A.;
Jongejan, J.; De Groot, A. Biocatal. Biotransform. 1998, 16, 145–162.
18. Orrenius, C.; Hæffner, F.; Rotticci, D.; Öhrner, N.; Norin, T.; Hult, K. Biocatal. Biotransform. 1998, 16, 1–15.
19. Lawson, D. M.; Brzozowiski, A. M.; Rety, S.; Verma, C.; Dodson, G. G. Protein Eng. 1994, 7, 543–550.
20. Holmquist, M.; Martinelle, M.; Clausen, I. G.; Patkar, S.; Svendsen, A.; Hult, K. Lipids 1994, 29, 599–603.
21. Martinelle, M.; Holmquist, M.; Clausen, I. G.; Patkar, S.; Svendsen, A.; Hult, K. Protein Eng. 1996, 9, 519–524.
22. Jacques, J.; Gros, C.; Bourcier, S. In Stereochemistry. Fundamentals and Methods; Kagan, H. B., Ed. Absolute
configurations of 6000 selected compounds with one asymmetric carbon. Georg Thieme: Stuttgart, 1977; Vol. 4.
23. Hult, K. In Micobial Reagents in Organic Synthesis, Servi, S. Ed. A kinetic interpretation of acids and alcohols influence on
the enantioselectivity in enzyme catalyzed resolutions. Kluwer Academic Publishers: the Netherlands, 1992; pp. 289–298.
24. Hill, C. M.; Schofield, H. I.; Spriggs, A. S.; Hill, E. J. Am. Chem. Soc. 1951, 73, 1660–1662.
25. Bomben, A.; Marques, C. A.; Selva, M.; Tundo, P. Tetrahedron 1995, 51, 11573–11580.
26. Guss, O. J. Am. Chem. Soc. 1949, 71, 3460–3462.
27. Chen, C.-S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1982, 104, 7294–7299.
28. Chen, C.-S.; Wu, S.-H.; Girdaukas, G.; Sih, C. J. J. Am. Chem. Soc. 1987, 109, 2812–2817.
29. Berthod, H.; Pullman, A. J. Chem. Phys. 1965, 62, 942–946.
30. Ryckaert, J. P.; Ciccoti, G.; Berendsen, H. J. C. J. Comp. Phys. 1977, 23, 327–341.
31. Weiner, S. J.; Kollman P. A.; Case, D. A.; Singh, U. C.; Ghio, C.; Alagona, G.; Profeta, S.; Weiner, P. J. Am. Chem. Soc.
1984, 106, 765–784.
32. Weiner, S. J.; Kollman, P. A.; Nguyen, D. T.; Case, D. A. J. Comp. Phys. 1986, 7, 230–252.