Molecular basis for enantioselectivity of lipase from Chromobacterium viscosum toward the diesters of 2,3-dihydro-3-(4′-hydroxyphenyl)-1,1,3-trimethyl-1H-inden-5-ol

Article abstract of DOI:10.1021/jo005681v

2,3-Dihydro-3-(4′-hydroxyphenyl)-1,1,3-trimethyl-1H-inden-5-ol, 1, is a chiral bisphenol useful for preparation of polymers. Previous screening of commercial hydrolases identified lipase from Chromobacterium viscosum (CVL) as a highly regio- and enantioselective catalyst for hydrolysis of diesters of 1. The regioselectivity was ≥30:1 favoring the ester at the 5-position, while the enantioselectivity varied with acyl chain length, showing the highest enantioselectivity (E = 48 ± 20 S) for the dibutanoate ester. In this paper, we use a combination of nonsymmetrical diesters and computer modeling to identify that the remote ester group controls the enantioselectivity. First, we prepared nonsymmetrical diesters of (±)-1 using another regioselective, but nonenantioselective, reaction. Lipase from Candida rugosa (CRL) showed the opposite regioselectivity (> 30:1), allowing removal of the ester at the 4′-position (the remote ester in the CVL-catalyzed reaction). Regioselective hydrolysis of (±)-l-dibutanoate (150 g) gave (±)-1-5-dibutanoate (89 g, 71% yield). Acylation gave nonsymmetrical diesters that varied at the 4′-position. With no ester at the 4′-position, CVL showed no enantioselectivity, while hindered esters (3,3-dimethylbutanoate) reacted 20 times more slowly, but retained enantioselectivity (E = 22). These results indicate that the remote ester group can control the enantioselectivity. Computer modeling confirmed these results and provided molecular details. A model of a phosphonate transition state analogue fit easily in the active site of the open conformation of CVL. A large hydrophobic pocket tilts to one side above the catalytic machinery. The tilt permits the remote ester at the 4′-position of only the (S)-enantiomer to bind in this pocket. The butanoate ester fits and fills this pocket and shows high enantioselectivity. Both smaller and larger ester groups show low enantioselectivity because small ester groups cannot fill this pocket, while longer ester groups extend beyond the pocket. An improved large-scale resolution of 1-dibutanoate with CVL gave (R)-(+)-1-dibutanoate (269 g, 47% yield, 92% ee) and (S)-(-)-1-4′monobutanoate (245 g, 52% yield, 89% ee). Methanolysis yielded (R)-(+)-1 (169 g, 40% overall yield, >97% ee) and (S)-(-)-1 (122 g, 36% overall yield, >96% ee).

Full text of DOI:10.1021/jo005681v

J . Org. Chem. 2001, 66, 3041-3048
3041
Molecu la r Ba sis for En a n tioselectivity of Lip a se fr om
Ch r om oba cter iu m viscosu m tow a r d th e Diester s of
2,3-Dih yd r o-3-(4′-h yd r oxyp h en yl)-1,1,3-tr im eth yl-1H-in d en -5-ol
David G. Gascoyne,† Herman L. Finkbeiner,† Kwok P. Chan,† J anet L. Gordon,†
Kevin R. Stewart,† and Romas J . Kazlauskas*,‡
Molecular OptoElectronics Corporation, 877 25th Street, Watervliet, New York 12189, and
McGill University, Department of Chemistry, 801 Sherbrooke Street West,
Montre´al, QC H3A 2K6 Canada
romas.kazlauskas@mcgill.ca
Received October 12, 2000
2,3-Dihydro-3-(4′-hydroxyphenyl)-1,1,3-trimethyl-1H-inden-5-ol, 1, is a chiral bisphenol useful for
preparation of polymers. Previous screening of commercial hydrolases identified lipase from
Chromobacterium viscosum (CVL) as a highly regio- and enantioselective catalyst for hydrolysis of
diesters of 1. The regioselectivity was g30:1 favoring the ester at the 5-position, while the
enantioselectivity varied with acyl chain length, showing the highest enantioselectivity (E ) 48 (
20 S) for the dibutanoate ester. In this paper, we use a combination of nonsymmetrical diesters
and computer modeling to identify that the remote ester group controls the enantioselectivity. First,
we prepared nonsymmetrical diesters of (()-1 using another regioselective, but nonenantioselective,
reaction. Lipase from Candida rugosa (CRL) showed the opposite regioselectivity (>30:1), allowing
removal of the ester at the 4′-position (the remote ester in the CVL-catalyzed reaction). Regioselective
hydrolysis of (()-1-dibutanoate (150 g) gave (()-1-5-dibutanoate (89 g, 71% yield). Acylation gave
nonsymmetrical diesters that varied at the 4′-position. With no ester at the 4′-position, CVL showed
no enantioselectivity, while hindered esters (3,3-dimethylbutanoate) reacted 20 times more slowly,
but retained enantioselectivity (E ) 22). These results indicate that the remote ester group can
control the enantioselectivity. Computer modeling confirmed these results and provided molecular
details. A model of a phosphonate transition state analogue fit easily in the active site of the open
conformation of CVL. A large hydrophobic pocket tilts to one side above the catalytic machinery.
The tilt permits the remote ester at the 4′-position of only the (S)-enantiomer to bind in this pocket.
The butanoate ester fits and fills this pocket and shows high enantioselectivity. Both smaller and
larger ester groups show low enantioselectivity because small ester groups cannot fill this pocket,
while longer ester groups extend beyond the pocket. An improved large-scale resolution of
1-dibutanoate with CVL gave (R)-(+)-1-dibutanoate (269 g, 47% yield, 92% ee) and (S)-(-)-1-4′-
monobutanoate (245 g, 52% yield, 89% ee). Methanolysis yielded (R)-(+)-1 (169 g, 40% overall yield,
>97% ee) and (S)-(-)-1 (122 g, 36% overall yield, >96% ee).
In tr od u ction
fold when researchers used pure enantiomers of helicene
in the film.2 The pure enantiomers created a chiral
supramolecular array similar to a liquid crystal. In
another example, researchers used all three approaches.
They added a chiral substituent to a chromophore,
oriented it into a liquid crystal, poled it, and finally cross-
linked this orientation.3
Organic materials with nonlinear optical properties are
the recent focus of both industrial and academic research
due to the potential applications in telecommunications,
optical information processing, storage and display.1
Many nonlinear optical effects require a noncentrosym-
metric, that is, chiral, environment for the chromophore.
One way to create a chiral environment is to align the
chromophore in poled polymer films or Langmuir-
Blodgett films. A second way is to add chiral substituents
to the chromophore. A third way is to create a chiral
supramolecular array, such as a chiral liquid crystal or
a chiral polymer. Devices with the strongest nonlinear
optical effect often use several methods. For example, a
Langmuir-Blodgett film of helicene showed a small
nonlinear optical effect, but this effect increased ∼30-
To use chiral polymers as supramolecular arrays in
nonlinear optics requires polymers with good transpar-
ency and mechanical properties because they also form
the mechanical shape of the device. Chiral polymers
derived from polyacrylates or polystyrene with added
chiral side chains4 have good transparency and mechan-
ical properties, but the stereocenter lies far from the main
chain. Chiral polymers with a stereocenter in the main
chain would probably better control the polymer chain
* To whom correspondence should be addressed at McGill Univer-
sity. Phone: +1 514 398 7229, fax: +1 514 398 3797.
† Molecular OptoElectronics Corporation.
(2) Verbiest, T.; van Elshocht, S.; Kauranen, M.; Hellemans, L.;
Snauwaert, J .; Nuckolls, C.; Katz, T. J .; Persoons, A. Science 1998,
282, 913-915.
(3) Trollsås, M., Orrenius, C., Sahle´n, F., Gedde, U. W., Norin, T.,
Hult, A., Hermann, D., Rudquist, P., Komitov, L., Lagerwall, S. T.,
Lindstro¨m, J . J . Am. Chem. Soc. 1996, 118, 8542-8548.
‡ McGill University.
(1) Example: Nalwa, H. S.; Miyata, S., Eds. Nonlinear Optics of
Organic Molecules and Polymers; CRC: Boca Raton, FL, 1997.
10.1021/jo005681v CCC: $20.00 © 2001 American Chemical Society
Published on Web 04/12/2001

3042 J . Org. Chem., Vol. 66, No. 9, 2001
Gascoyne et al.
Ta ble 1. In flu en ce of th e Acyl Gr ou p on th e
Sch em e 1. Ca r ben iu m Ion Med ia ted
Dim er iza tion of 4-(2-P r op en yl)p h en ol Yield s
Ra cem ic 2,3-Dih yd r o-3-
En a n tioselectivity of CVL tow a r d 1-Diester s
ee of
(4′-h yd r oxyp h en yl)-1,1,3-tr im eth yl-1H-in d en -5-ol,
(()-1, a Bisp h en ol Mon om er Usefu l for th e
Syn th esis of P olym er s
con-
version
(%)a
mono-
ester
(%)b
rate
(U/mg)
ester
Ec
Straight Chain
diacetate
15
11.5
8
11
0.4
10
2.8
6.4
6.8
63d
42.3
78.5
93.2
92.4
78.4
86.4
63.8
51.1
60.1
5
dipropanoate
dibutanoate
dipentanoate
dihexanoate
diheptanoate
dioctanoate
dinonanoate
didecanoate
37
36
39
51
3.4
8.8
47
20
14 ( 3
48 ( 20e
35 ( 18e
21
12
5
5
5
orientation. The ideal monomer would be rigid, contain
a stereocenter within the main polymer chain, and form
transparent polymers with excellent mechanical proper-
ties. Unfortunately, no such examples exist. One group
recently reported binaphthol polycarbonates, but their
molecular weights were very low.5
A monomer that fits these criteria is Indane bisphenol
dimer or 2,3-dihydro-3-(4′-hydroxyphenyl)-1,1,3-trimethyl-
1H-inden-5-ol, 1. Acid-catalyzed dimerization of 4-pro-
penylphenol,6 Scheme 1, yields racemic 1 in high yield.
It also forms as an impurity in the commercial prepara-
tion of bisphenol A (2,2-bis(4-hydroxyphenyl)propane).7
Racemic 1 forms transparent polymers with good me-
chanical properties.8
We previously reported a regio- and enantioselective
hydrolysis of diesters of 19 using lipase from Chromo-
bacterium viscosum.10 Although chemists often resolve
enantiomers using hydrolase-catalyzed reactions,12 this
reaction was unusual for two reasons. First, the hydroly-
sis was highly regio- and enantioselective despite the
large distance between the stereocenter and ester car-
Branched Chain
2-methylbutan-oate
3-methylbutan-oate
3,3-dimethyl-butanoate
2
18
3.8
na
na
1.1
na
na
<1
<1
<1
<1
Monoester
25
5-butanoate
60f
9f
1.2
Nonsymmetrical Esters
5-butanoate, 4′-3,3-
dimethylbutanoate
5-butanoate, 4′benzoate
0.4
3.8
91
na
22
na
<1
<1
a
Measured by HPLC at 272 nm. Peak areas for the monoester
were divided by 1.07 to account for the difference in extinction
coefficient for diester and monoester. Data for the symmetrical,
straight chain diesters is from ref 9. na ) not applicable because
b
the substrate did not react. Hydrolysis forms the 4′-monester in
all cases. c Enantiomeric ratio as defined by Chen, C. S., Fujimoto,
Y., Girdaukas, G., Sih, C. J . J . Am. Chem. Soc. 1982, 104, 7294-
7299. 2.6% diol noted in the hydrolysis of the diacetate. e Average
d
and standard deviation for several different experiments or range
for two experiments. f Conversion refers to the amount of diol
formed; enantiomeric purity is for the diol.
bonyl at the 5-position (five bonds). Efficient resolutions
of molecules containing stereocenter as remote as this
one are rare.12 Second, the enantioselectivity varied by
approximately a factor of 10 (E ) 5-50) with changes in
the length of the acyl group.
In this paper we identify that the remote, nonreacting
acyl group caused this variation in enantioselectivity.
Further, we use molecular modeling to identify the
molecular basis of the enantioselectivity. This is the first
time that a molecular mechanism has been identified for
the enantioselectivity toward such a remote stereocenter.
(4) For example, Chen, X. H.; Herr, R. P.; Schmitt, K.; Buchecker,
R. Liq. Cryst. 1996, 20, 125-138; Firestone, M. A.; Park, J .; Minami,
N.; Ratner, M. A.; Marks, T. J .; Lin, W.; Wong, G. K. Macromolecules
1995, 28, 2247-2259.
(5) Takata, T.; Furusho, Y., Murakawa, K.; Endo, T., Matsuoka, H.;
Hirasa, T.; Matsuo, J .; Sisido, M. J . Am. Chem. Soc. 1998, 120, 4530-
4531.
(6) Chan, K. P.; Cristo, P. L.; McGrath, P. M. US Patent 5,994,596,
1999.
(7) Paryzkova, J .; Snobl, D.; Matousek, P. Chem. Prum. 1979, 29,
30-34.
(8) For example, polycarbonates: Gordon, J . L.; Chan, K. P.;
Gascoyne, D. G. Polym. Prepr. 1998, 39, 486-487; Gordon, J . L.;
Gascoyne, D. G. US Patent 5,703,197, 1997; polyethers: Saegusa, Y.;
Iwasaki, T., Nakamura, S. J . Polym. Sci., Part A: Polym. Chem. 1994,
32, 249-256; polysilixoanes: Saegusa, Y.; Kato, T.; Oshiumi, H.;
Nakamura, S. J . Polym. Sci., Part A: Polym. Chem. 1992, 30, 1401-
1406; Padmanaban, M.; Kakimoto, M.; Imai, Y. J . Polym. Sci., Part A:
Polym. Chem. 1990, 28, 2997-3005; polyesters: Wilson, J . C. J . Polym.
Sci., Polym. Chem. Ed. 1975, 13, 749-754; Imai, Y.; Tassavori, S. J .
Polym. Sci., Polym. Chem. Ed. 1984, 22, 1319-1325; Kakimoto, M.;
Harada, S.; Oishi, Y.; Imai. Y. J . Polym. Sci., Part A: Polym. Chem.
1987, 25, 2747-2753; Negi, Y. S.; Imai, Y.; Kakimoto, M. J . Polym.
Mater. 1988, 5, 67-71; Yoneyama, M.; Kakimoto, M.; Imai, Y.
Macromolecules 1989, 22, 2593-2596; polyphosphates: Imai, Y.;
Kamata, H.; Kakimoto, M. J . Polym. Sci., Polym. Chem. Ed. 1984, 22,
1259-1265.
(9) Zhang, M., Kazlauskas, R. J . Org. Chem. 1999, 64, 7498-7503;
Kazlauskas, R. J .; Zhang, X. U.S. Patent 5,959,159; 1999.
(10) Lipases from Chromobacterium viscosum and from Pseudo-
monas glumae have identical amino acid sequences and very similar
biochemical properties. Taipa, M. A.; Liebeton, K.; Costa, J . V.; Cabral,
J . M. S.; J aeger, K.-E. Biochim. Biophys. Acta 1995, 1256, 396-402;
Lang, D. Hofmann, B. Haalck, L. Hecht, H. J . Spener, F. Schmid, R.
D. Schomburg, D. J . Mol. Biol. 1996, 259, 704-717.
Resu lts
Diester s. In the previous paper,8 we examined the
effect of acyl group chain length on the enantioselectivity
of CVL-catalyzed hydrolysis of 1-diesters, Table 1. The
dibutanoate ester (E ) 48 ( 20) and dipentanoate ester
(E ) 35 ( 18) showed the highest enantioselectivity. Both
shorter and longer esters showed lower enantioselectivity
(e.g., E ) 5 for both the diacetate and didecanoate). We
also prepared several diesters of branched chain acids,
Table 1, but they were disappointing. Branched chain
diesters of 1 either did not react in CVL-catalyzed
reactions (3-methylbutanoate, 3,3-dimethylbutanoate), or
reacted slowly without enantioselectivity (2-methyl-
butanoate). Thus, the acyl groups influenced both the
rate and enantioselectivity, but we did not know which
acyl group caused these changes since we changed both
simultaneously.
(11) (a) Faber, K. Biotransformations in Organic Chemistry, 4th ed.;
Springer: Berlin, 2000. (b) Bornscheuer, U. T.; Kazlauskas, R. J .
Hydrolases in Organic Syntheses: Regio- and Stereoselective Biotrans-
formations Wiley-VCH: Weinheim, 1999. (c) Wong, C.-H.; Whitesides,
G. M. Enzymes in Synthetic Organic Chemistry, Pergamon: Oxford,
1994.
(12) Reference 11b, pp 98-102, 114-115.

Molecular Basis for Enantioselectivity of Lipase
J . Org. Chem., Vol. 66, No. 9, 2001 3043
Ta ble 2. En a n tioselectivity of CRL tow a r d Sever a l
1-Diester s
ee of
con-
mono-
rate
(U/mg)
version ester
ester
dibutanoate
(%)a
(%)b
Ec
Straight Chain
5.3
9.1
52
27
33
33
3.7-4.7d
2.2
dipentanoate
branched chain
2-methylbutanoate
3-methylbutanoate
3,3-dimethylbutanoate
9.2
1.5
<1
27
4.4
<1
27
4.5
na
1.9
1.1
na
Nonsymmetrical Esters
5-butanoate,
4′-3,3-dimethylbutanoate
5-butanoate, 4′-benzoate
3.5
35
55
na
5.4
na
<1
<1
a
Measured by HPLC at 272 nm. Peak areas for the monoester
were divided by 1.07 to account for the difference in extinction
coefficient for diester and monester. na ) not applicable because
b
the substrate did not react. Hydrolysis forms the 5-monoester.
c Enantiomeric ratio as defined by Chen, C. S., Fujimoto, Y.,
Girdaukas, G., Sih, C. J . J . Am. Chem. Soc. 1982, 104, 7294-
d
7299. Range for two experiments.
slightly changed the enantioselectivity, Table 2. As
compared to the dipentanoate (E ) 2.2), the enantio-
selectivity increased slightly for the dibutanoate (E )
3.7-4.7) and the nonsymmetrical ester 5-butanoate, 4′-
3,3-dimethylbutanoate (E ) 5.4), remained similar for
the di(2-methylbutanoate) (E ) 1.9) and decreased
slightly for di(3-methylbutanoate) (E ) 1.1). Two diesters
did not reactsthe di(3,3-dimethylbutanoate) and the
5-butanoate, 4′-benzoate. This high regioselectivity com-
bined with low enantioselectivity toward a wide range
of esters make this a useful reaction for the preparation
of 1-5-monoesters.
We prepared 1-5-monobutanoate by a CRL-catalyzed
regioselective hydrolysis of the 4′-ester group in 1-di-
butanoate, Scheme 2. We allowed the CRL-catalyzed
hydrolysis of (()-1-dibutanoate (150 g) to proceed in an
emulsion of tert-butyl methyl ether and water for 15 days
until approximately 1 equiv of base had been consumed.
This high conversion ensured that the product would be
racemic. Workup of the reaction yielded the (()-1-5-
monobutanoate as a powder in 71% yield. We also
prepared enantiomerically pure 1-5-monobutanoate using
the same reaction, but starting with enantiomerically
pure 1-dibutanoate. CRL-catalyzed hydrolysis of (R)-(+)-
1-dibutanoate (22.4 g) yielded (R)-1-5-monobutanoate in
66% yield with >99% ee.
Non sym m etr ica l Diester s. We prepared two non-
symmetrical diesters where the ester at the 5-position
(CVL-reactive) was butanoate and the ester at the 4′-
position (CVL-nonreactive) was either 3,3-dimethyl-
butanoate or benzoate, Table 1. CVL-catalyzed hydrolysis
of the nonsymmetrical ester with 3,3-dimethylbutanoate
at the 4′-position was 20 times slower than the dibutan-
oate, but the enantioselectivity decreased only by a factor
of 2 to E ) 22. With a benzoate at the 4′-position, the
butanoate at the 5-position no longer reacted. The most
revealing case was the 1-5-monobutanoate where there
is no acyl group at the 4′position. The rate of CVL-
catalyzed hydrolysis increased by a factor three, but the
enantioselectivity disappeared (E ) 1.2). Thus, structural
changes remote from the reactive group affected both the
rate and enantioselectivity of the CVL-catalyzed hydroly-
sis. We suggest therefore that variations of the remote
F igu r e 1. 1H NMR chemical shifts of the aromatic protons
in 1,1-dibutanoate, and the 1-monobutanoate isolated from a
CRL-catalyzed hydrolysis of 1-dibutanoate. The aromatic
protons in 1 resonate 0.2-0.3 ppm upfield of those in 1-di-
butanoate. In the monobutanoate, the resonances of the less-
substituted ring lie 0.21-0.36 ppm upfield from those in the
dibutanoate, while those in the more-substituted ring lie at
with 0.03 ppm of the dibutanoate. Thus, the ester group lies
on the more substituted ring and is 1-5-monobutanoate
Sch em e 2. Regioselective Hyd r olysis of th e
4′-Ester Gr ou p by CRL Yield s 1-5-Mon obu ta n oa te.
Th is Hyd r olysis Sh ow s Low En a n tioselectivity
1-5-Mon obu ta n oa te via Regioselective Hyd r olysis
1-Dibu ta n oa te. To identify which acyl group caused the
changes in enantioselectivity, we needed racemic non-
symmetrical diesters of 1. Column chromatography on
silica gel could not separate 1-4′-monobutanoate and 1-5-
monobutanoate, so we used an enzyme-catalyzed regio-
selective reaction. The ideal reaction would be highly
regioselective, but not enantioselective.
Our previous screening identified lipase from Candida
rugosa, CRL, as a highly regioselective, but poorly
enantioselective, catalyst.13 CRL favored hydrolysis of the
4′-ester group by >30:1, yielding 1-5-monobutanoate,
Scheme 2. The structure was established from the 1H
NMR chemical shifts, Figure 1. This regioselectivity is
opposite that for CVL. Changing the acyl group in CRL-
catalyzed reactions did not significantly change the
regioselectivity (data not shown), but some acyl groups
(13) Alternatively, one of the less enantioselective reactions cata-
lyzed by CVL (e.g., diacetate) would yield the other regioisomer (the
4′-monoester). This route would involve protection-deprotection steps
to prepare the 5-monoester, so we did not investigate it.

3044 J . Org. Chem., Vol. 66, No. 9, 2001
Gascoyne et al.
ester group caused most of the variations in enantio-
selectivity in the diester series.
Molecu la r Mod elin g. To rationalize why CVL favors
the (S)-enantiomer and to provide details on how the
remote ester group caused variations in enantioselectiv-
ity, we used computer modeling. First, we built a homol-
ogy model of the open conformation of CVL. Although
two groups have solved the X-ray crystal structure of
CVL, both structures show the inactive closed conforma-
tion.14 On the other hand, crystallographers have solved
the X-ray crystal structure of Pseudomonas cepacia lipase
(PCL), a closely related lipase,15 in the active open
conformation.16 These structures of CVL and PCL are
almost identical, except for the position of the lid, a
helical segment that blocks the active site in the closed
conformation. This high degree of similarity suggests that
we could build a reliable homology model of the open
conformation of CVL based on the structure for PCL. We
used Swiss-Model, an automated homology-modeling
server,17 to create this model. As expected, the quality of
this model was good with only a few high-energy orienta-
tions of side chains as identified by the program
WhatCheck.18 The minimization below relaxed these
high-energy orientations.
We built the phosphonate shown in Figure 2 into the
active site of the open conformation of CVL. This phos-
phonate mimics the transition state for hydrolysis of
1-dibutanoate at the favored butanoatesat the 5-position
of the (S)-enantiomer. The structure of 1 is relatively
rigid and fit well into the alcohol-binding site. Energy
minimization of this structure suggests a reasonable
model for the transition state. The most interesting
feature of this model is the location of the 4-butanoyl-
oxyphenyl moiety in a large hydrophobic pocket. This
pocket lies above the catalytic triad and tilts to the right.
For the (S)-enantiomer, the 4-butanoyloxyphenyl moiety
also tilts to the right and thus fits well in this pocket. In
contrast, for the (R)-enantiomer (not shown), the 4-bu-
tanoyloxyphenyl moiety points to the left and lies outside
the hydrophobic pocket. The calculated energy of the (R)-
enantiomer complex lies 3.0 kcal/mol higher in energy
than for the (S)-enantiomer complex, consistent with a
less-favorable orientation for the (R)-enantiomer. An
enantiomeric ratio of 48 corresponds to a free energy
difference of 2.3 kcal/mol. Since the modeling ignores any
F igu r e 2. Molecular modeling of a transition state analogue
in the active site of CVL rationalizes the observed preference
for the (S)-enantiomer. (a) Line drawing of the phosphonate
transition state analogue for the CVL-catalyzed hydrolysis of
favored butanoyl group: 5-position of the (S)-enantiomer. This
regioisomer and enantiomer is favored by >30:1 over other
possibilities. Energy minimization of the structure yielded the
orientation shown. The 4′-butanoyloxyphenyl group lay in the
large hydrophobic pocket, while the CMe2 group lay in the
medium pocket. The “eye” shows the direction of view for part
b. (b) Molecular model of the energy-minimized structure. The
protein is shown in space-filling representation, while the
transition state analogue is shown in stick representation.
Hydrogens are omitted for clarity. The violet spheres represent
the imidazole ring of the catalytic His 285. The medium gray
spheres represent the nine hydrophobic residues in the active
site: Leu 17, Phe 119, Val 123, Ile 139, Phe 142, Val 143, Phe
146, Leu 164, Leu 167. These residues line the large hydro-
phobic pocket, which lies above His 285 and tilts to the right.
In the transition state analogue, red represents oxygen atoms,
and an arrow points to the orange phosphorus atom. The 4′-
butanoyloxyphenyl group fits and fills the large hydrophobic
pocket. In the minimized structure for the other enantiomer
the 4′-butanoyloxyphenyl group lies to the left, outside the
large hydrophobic pocket. This difference in the orientation
of the 4′-butanoyloxyphenyl group likely accounts for the faster
reaction of the (S)-enantiomer. The starting protein structure
was a homology model based on the open conformation of PCL,
a lipase closely related to CVL.
(14) Noble, M. E. M.; Cleasby, A.; J ohnson, L. N.; Egmond, M. R.;
Frenken, L. G. J .FEBS Lett. 1993, 331, 123-128; Lang, D. Hofmann,
B. Haalck, L. Hecht, H. J . Spener, F. Schmid, R. D. Schomburg, D. J .
Mol. Biol. 1996, 259, 704-717.
(15) The amino acid sequence of CVL (319 amino acids) differs from
the sequence of PCL (320 amino acids) by fifty-one amino acids and
one deletion. Of the fifty-one amino acids, thirty are highly conservative
substitutions, twelve are conservative substitutions and only nine
amino acids are different. Homology modeling of such similar proteins
(84% identical amino acids, 97% similar amino acids) is very reliable.
A reliability check showed that homology models deviated from the
experimental structures by e2 Å deviation for the CR atoms in 79% of
the cases and by e5 Å in 95% of the cases (Guex, N.; Diemand, A.;
Peitsch, M. C. Trends Biochem. Sci. 1999, 24, 364-367.)
(16) Kim, K. K., Song, H. K., Shin, D. H., Hwang, K. Y., Suh, S. W.
Structure 1997, 5, 173-185; Schrag, J . D., Li, Y. G., Cygler, M., Lang,
D. M., Burgdorf, T., Hecht, H. J ., Schmid, R., Schomburg, D., Rydel,
T. J ., Oliver, J . D., Strickland, L. C., Dunaway, C. M., Larson, S. B.,
Day, J ., McPherson, A. Structure 1997, 5, 187-202; Lang, D. A.,
Mannesse, M. L. M., De Haas, G. H., Verheij, H. M., Dijkstra, B. W.
Eur. J . Biochem. 1998, 254, 333-340.
(17) Peitsch, M. C. Bio/ Technology 1995, 13, 658-660; Peitsch, M.
C. Biochem. Soc. Trans. 1996, 24, 274-279; Peitsch, M. C., Guex, N.
Electrophoresis 1997, 18, 2714-2723. http://www.expasy.ch/swissmod/
SWISS-MODEL.html.
entropy contributions, the agreement between the cal-
culated 3.0 and the observed 2.3 kcal/mol is surprisingly
good. Thus, this different binding of the 4-butanoyl-
oxyphenyl moiety rationalizes the preference of CVL for
the (S)-enantiomer. Note that PCL, which has a similar
tilted large hydropobic pocket, also favored the (S)-
(18) Hooft, R. W. W.; Vriend, G.; Sander, C.; Abola, E. E. Nature
1996, 381, 272. http://www.sander.embl-heidelberg.de/whatcheck/.

Molecular Basis for Enantioselectivity of Lipase
J . Org. Chem., Vol. 66, No. 9, 2001 3045
Sch em e 3. La r ge Sca le Resolu tion of (()-1
Dibu ta n oa te Usin g th e Regio- a n d
En a n tioselective Hyd r olysis Ca ta lyzed by CVL
racemate melts significantly higher that the pure enan-
tiomer: 193 vs 149 °C. For a conglomerate the racemate
must melt at a lower temperature than the pure enan-
tiomers. The racemate also shows a higher heat of
fusion: 8.5 vs 6.8 kcal/mol. The predicted21 eutectic
composition is 82% ee with a melting point of 143 °C.
Lipase-catalyzed kinetic resolution remains the only
route to pure enantiomers of 1.
Discu ssion
In previous work,8 we identified CVL from Sigma as
the most suitable hydrolase for preparation of pure
enantiomers of 1. CVL showed both high regioselectivity
favoring the acyl group at the 5-position and high
enantioselectivity for the (S)-enantiomer. Out of a num-
ber of different esters investigated, CVL showed the
highest enantioselectivity toward the dibutanoate of 1
(E ) 48 ( 20). In this paper we reported a resolution of
1 on a larger scale and in higher yield than in previous
work. This resolution is efficient and simple enough to
prepare kilogram amounts of enantiomerically pure 1.
In this paper, we also identified the molecular mech-
anism of enantioselectivity first by identifying which
structural features are essential to enantioselectivity.
Previously we noted that the acyl chain length in diesters
of 1 affects the enantioselectivity, but did not know which
of the two esters was more important. Here we varied
the remote ester group (the ester that does not react in
CVL-catalyzed reactions) and found it was crucial to the
high enantioselectivity of CVL. We used a lipase with
the opposite regioselectivity to remove the remote ester
group. With no ester group at the 4′-position (1-5-
monobutanoate), CVL showed no enantioselectivity, while
with the corresponding diester (1-dibutanoate), CVL
showed high enantioselectivity. This result suggests that
the remote, nonreacting ester group is essential to high
enantioselectivity.22 Consistent with this suggestion, the
CVL-catalyzed hydrolysis of the nonsymmetrical diesters
1-5-butanoate, 4′-3,3-dimethylbutanoatesshowed good
enantioselectivity, only a factor of 2 lower than 1-dibu-
tanoate.
To identify which structural features in CVL recognize
this remote ester group, we used molecular modeling of
a phosphonate transition state analogue. As expected, the
chiral alcohol moiety 1 bound in the proposed alcohol-
binding crevice. This crevice contains a large and a
medium pocket. However, unlike many lipases such as
CRL which have a roughly symmetrical large pocket, the
large pocket in CVL is not symmetrical and tilts by ∼30°
to the right in Figure 2. Because of this tilt, the acyl-
oxyphenyl moiety only for the (S)-enantiomer binds in
the large pocket. Consistent with this explanation, CRL
shows low enantioselectivity because its large pocket is
roughly symmetrical, and PCL shows some enantio-
selectivity because its large pocket is also tilted. Further,
size of this large pocket also suggests a reason the
enantioselectivity of CVL varies with structural changes
enantiomer, but with lower enantioselectivity (E ∼10).8,19
On the other hand, CRL, which has a roughly sym-
metrical large hydrophobic pocket, showed no enantio-
selectivity.
(R)-(+)-1 a n d (S)-(-)-1 via La r ge-Sca le Resolu tion
of (()-1-Dibu ta n oa te. To resolve (()-1 on a large scale,
we started with 408 g of (()-1 and converted it to (()-
1-dibutanoate (568 g, 91% isolated yield). A portion of
this diester (520 g) was hydrolyzed using CVL (Sigma)
in a water and tert-butyl methyl ether emulsion to 48%
conversion in 31 h, Scheme 3. Workup and separation of
the mono-and dibutanoates by chromatography on silica
gel yielded oils: (R)-(+)-1-dibutanoate, 269 g, 92% ee,
47% yield and (S)-(-)-1-4′-monobutanoate, 245 g, 89% ee,
52% yield. The enantiomeric ratio for this reaction was
56, better than the average for the small scale experi-
ments in Table 2 above. The esters were saponified in
basic methanol, and the resulting 1 was recrystallized
from dichloromethane-methanol to give white crystals:
(R)-(+)-1, 136 g, >97% ee, 40% yield and (S)-(-)-1, 122
g, >96% ee, 36% yield from the racemic dibutanoate. This
recrystallization also raised the enantiomeric purity. The
maximum yield in a resolution is 50%. This reaction was
repeated to produce ∼500 g of enantiomerically pure 1
for polymer synthesis.
(20) J acques, J .; Collet, A.; Wilen, S. H. Enantiomers, Racemates,
and Resolutions; Wiley-Interscience: New York, 1981.
(21) Reference 20, pp 88-93.
(22) We propose that the remote ester rather than the reactive ester
is the more important for enantioselectivity. On the other hand, the
reactive ester certainly influences the rate of reaction (e.g., no reaction
for the 1-di(3,3-dimethylbutanotate, but slow reaction for the 1-5-
butanoate,4′-(3,3-dimethylbutanoate)) and may have a minor role in
enantioselectivity.
Some organic compounds (∼10%) crystallize as con-
glomerates and can be resolved by direct crystallization.20
However, 1 does not crystallize as conglomerate. The
(19) PCL was not suitable for preparative reactions because it
showed low regioselectivity (less than approximately three favoring
hydrolysis of the 5′-ester in 1-dibutanoate).

3046 J . Org. Chem., Vol. 66, No. 9, 2001
Gascoyne et al.
1
(()-1-Di(3-m eth ylbu ta n oa te): H NMR δ 1.02 (m, 15H),
1.31 (s, 3H), 1.66 (s, 3H), 2.20 (m, 3H), 2.38 (m, 5H), 6.81 (d,
1H, 6), 6.97 (m, 3H), 7.17 (m, 3H). 13C NMR δ 22.0, 22.1, 25.5,
30.2, 30.3, 30.4, 42.2, 42.9, 43.0, 50.1, 59.2, 117.6, 120.4, 120.7,
122.9, 127.2, 147.5, 148.3, 149.0, 149.5, 171.0, 171.1. HRMS
(EI) Calcd for C28H36O4 (M+) 436.26136, found 436.26134.
(()-1-Di(2,2-d im eth ylp r op a n oa te): 1H NMR δ 1.05 (s,
3H), 1.32 (s, 12H), 1.34 (s, 9H), 1.67 (s, 3H), 2.21 (d, 1H, J )
13), 2.39 (d, 1H, J ) 13), 6.78 (d, 1H, J ) 2), 6.95 (m, 3H),
7.18 (m, 3H). 13C NMR δ 26.9, 27.0, 30.3, 30.5, 30.6, 38.8, 42.4,
50.2, 59.4, 117.6, 120.4, 120.7, 123.0, 127.4, 147.6, 148.7, 149.0,
149.7, 150.0, 176.8, 176.9. HRMS (EI) Calcd for C28H36O4 (M+)
436.26136, found 436.26134.
in the remote ester group. With no acyl group or a small
acyl group at the 4′-position, binding of the phenoxy or
acyloxyphenyl group to the large hydrophobic pocket is
weak. The (S)-enantiomer gains little advantage by
having the correct tilt for the large pocket and the
resulting enantioselectivity is low. Similary, with a large
acyl group (>C6), only part of the acyl chain fits in the
large pocket; the rest binds to the surface of the lipase
for either enantiomer. Again difference in binding for the
two enantiomers in low and the enantioselectivity is low.
Only intermediate acyl groups (butanoate, pentanoate)
maximize the difference in binding of the two enanti-
omers because the acyl group fits and fills the large
pocket. Enantioselectivity is high for these intermediate
acyl groups. This explanation is the first detailed proposal
for how lipases recognize enantiomers with remote ste-
reocenters.
The ready access to enantiomerically pure 1 will allow
evaluation of enantiomerically pure polymers as matrixes
for nonlinear optical devices. We have prepared polycar-
bonates and other polymers from 123 and are measuring
the properties of nonlinear optical devices and polarizing
filters made from these chiral polymers.
Although the main focus of this work was the prepara-
tion of enantiomerically pure monomers to make chiral
polymers for nonlinear optical applications, chiral poly-
mers derived from 1 may also be useful as catalysts or
catalyst supports for asymmetric synthesis.24
(()-1-5-Mon ob u t a n oa t e via CR L-Ca t a lyzed R egio-
select ive H yd r olysis. A solution of lipase from Candida
rugosa (CRL, 9.0 mg solid, Sigma L 8525 dissolved in phos-
phate buffer, 0.1 M, 56 mL) was added to a solution of
1-dibutanoate (150 g, 367 mmol, distilled under vacuum) in
tert-butyl methyl ether (500 mL). The mixture was stirred to
form an emulsion, and the pH was maintained at 7.2 by a pH
stat, which controlled the addition of 5.0 M NaOH. After 13
d, 270 mmol of base was consumed and additional CRL (7.2
mg solid) was added. After 15 d, a total of 350 mmol of base
was consumed and the stirring was stopped. The emulsion was
allowed to settle and the water layer was discarded. The
organic phase was washed with distilled water (2 × 250 mL)
and concentrated to a gum by rotary evaporation followed by
vacuum (800 µm Hg) at 120 °C. The gum was was washed
with hexane (200 mL), leaving a powder, 89 g, 71%. mp 123-
126 °C, 1H NMR δ 1.00 (m, 6H), 1.28 (s, 3H), 1.59 (s, 3H), 1.76
(sextet, 2H), 2.12 (d, 1H, J ) 16), 2.36 (d, 1H, J ) 16), 2.51 (t,
2H, J ) 8), 6.36 (s, 1H, OH), 6.59 (d, 2H, J ) 9), 6.76 (d, 1H,
J ) 2), 6.96 (m, 3H), 7.14 (broadened d, 1H, J ) 8). 13C NMR
δ 13.5, 18.3, 30.3, 30.5, 30.6, 36.2, 42.3, 49.9, 59.3, 114.7, 117.6,
120.3, 123.1, 127.5, 142.1, 149.5, 150.6, 153.6, 173.2. HRMS
(EI) Calcd for C22H26O3 (M+) 338.18819, found 338.18818.
(R)-(+)-1-5-Mon obu ta n oa te via CRL-Ca ta lyzed Regio-
selective Hyd r olysis. A similar procedure starting with (R)-
(+)-1-dibutanoate (22.4 g, 54.8 mmol) yielded white powder,
12.3 g, 66%, >99% ee by HPLC on a Chiralcel AD column, mp
98-101 °C. The 1H and 13C NMR were identical to those for
the racemate.
Exp er im en ta l Section
Gen er a l. NMR spectra were recorded at 270 or 300 MHz
in deuteriochloroform. Coupling constants are given in hertz.
Mass spectra were obtained either by direct inlet electron
ionization or by fast atom bombardment (6 kV Xe) in a glycerol
matrix. THF was dried by distillation from sodium benzo-
phenone ketyl under nitrogen. CVL was purchased from
Sigma. Melting points and enthalpies of fusion for (()-1 and
(-)-1 were measured by differential scanning calorimetry. The
eutectic composition was calculated as the intersection of the
lines from the Schro¨der-van Laars equation, which predicts
the part of the phase diagram lying between the pure enan-
tiomer and the eutectic, and the Prigogine-Defay equation
which predicts the phase diagram between the two eutectics.20
(()-2,3-Dih yd r o-3-(4′-h yd r oxyp h en yl)-1,1,3-tr im eth yl-
1H-in d en -5-ol, (()-1, was prepared by acid-catalyzed dimer-
ization of 4-(2-propenyl)phenol.5
Non sym m etr ica l Diester s of (()-1. Nonsymmetrical di-
esters were prepared by acylation of (()-1-5-monobutanoate.
(×b1()-1-5-Bu ta n oa te,4′-ben zoa te: Benzoyl chloride (3.51
g, 33 mmol) was added to a solution of (()-1-5-monobutanoate
(7.1 g, 21 mmol) and triethylamine (2.55 g, 25 mmol) in tert-
butyl methyl ether (20 mL), and the mixture was stirred for 2
h. The reaction mixture was washed with water (3 × 20 mL),
dried over MgSO4, and concentrated by rotary evaporation to
1
an orange gum: 8.3 g, 90%; H NMR δ 1.10 (m, 6H), 1.39 (s,
3H), 1.74 (s, 3H), 1.82 (m, 2H), 2.27 (d, 1H, J ) 13), 2.48 (d,
1H, J ) 13), 2.57 (t, 2H, J ) 8), 6.86 (d, 1H, J ) 2), 7.03 (dd,
1H, J ) 2), 7.06 (dd, 1H, J ) 2), 7.14 (d, 2H, J ) 9), 7.22 (d,
2H, J ) 8), 7.28 (d, 2H, J ) 9), 7.53 (dd, 2H, J ) 8), 7.63 (dd,
1H, J ) 7), 8.23 (d, 2H, J ) 8).13C NMR δ 13.6, 18.4, 30.4,
30.6, 36.2, 42.5, 50.4, 59.4, 117.8, 120.6, 121.0, 123.2, 127.6,
128.5, 129.5, 130.0, 133.4, 148.0, 148.7, 149.3, 149.8, 149.9,
165.1, 172.2. HRMS (EI) Calcd for C29H30O4 (M+) 442.21441,
found 442.21439.
(()-1-5-Bu tan oate,4′-(2,2-dim eth ylpr opan oate): The same
procedure as above, but using 2,2-dimethylpropanoyl chloride.
1H NMR δ 0.99 (m, 6H), 1.30 (m, 12H), 1.68 (m, 5H), 2.20 (d,
1H, J ) 8), 2.43 (m, 3H), 6.84 (d, 1H, J ) 2), 6.96 (m, 3H),
7.17 (m, 3H). 13C NMR δ 13.2, 17.9, 26.6, 30.0, 30.2, 35.6, 38.4,
42.0, 49.9, 59.0, 117.5, 120.3, 120.5, 122.8, 127.0, 147.2, 148.5,
148.7, 149.3, 149.5, 171.4, 176.1. HRMS (EI) Calcd for C27H34O4
(M+) 422.24571, found 422.42569.
La r ge Sca le P r ep a r a tion of (()-1-Dibu ta n oa te. A 5-L,
three-necked, round-bottom flask fitted with an overhead
stirrer, a 250-mL addition funnel, and a water-cooled con-
denser was charged with a solution of racemic 1 (408 g, 1.52
mol) and triethylamine (338 g, 3.34 mol) in tert-butyl methyl
ether (2.8 L). The flask was cooled in an ice bath while
butanoyl chloride (356 g, 3.34 mol) was added over 1 h. After
Sym m etr ica l (()-1-Diester s. Acid chloride (8.9 mmol, 2.4
eq) in dry THF (25 mL) was added over 10 min to a solution
of 1 (1.0 g, 3.7 mmol) and triethylamine (0.90 g, 8.9 mmol, 2.4
eq) in dry THF (25 mL). TLC showed complete consumption
of 1 after stirring overnight at room temperature. HCl (50 mL,
1 M) was added and the mixture was extracted with ethyl
acetate (3 × 20 mL). The combined extracts were washed with
NaHCO3 (10%, 3 × 20 mL) and water (2 × 20 mL) and dried
over magnesium sulfate. Chromatography on silica gel eluted
with dichloromethane yielded colorless oils in 90-95% yield.
(()-1-Di(2-m et h ylb u t a n oa t e): 1H NMR δ 1.00 (m, 9H),
1.25 (m, 6H), 1.33 (s, 3H), 1.60 (m, 2H), 1.67 (s, 3H), 1.80 (m,
2H), 2.21 (d, 1H, J ) 13), 2.38 (d, 1H, J ) 13), 2.42 (d, 1H, J
) 13), 2.59 (m, 2H), 6.79 (d, 1H, J ) 6), 6.97 (m, 3H), 7.18 (m,
3H). 13C NMR δ 11.3, 11.4, 16.4, 26.6, 30.3, 30.5, 40.8, 40.9,
42.4, 50.2, 59.3, 117.7, 120.4, 120.8, 123.0, 127.4, 147.6, 148.5,
149.1, 149.7, 174.9, 175.0. HRMS (EI) Calcd for C28H36O4 (M+)
436.26136, found 436.26134.
(23) Gordon, J . L.; Stewart, K. R.; Chan, K. P. US Patent 5,777,063,
1998.
(24) Review: Pu, L. Tetrahedron: Asymmetry 1998, 9, 1457-1477;
recent example: Yu, H.-B.; Hu, Q.-S.; Pu, L. J . Am. Chem. Soc. 2000,
122, 6500-6501.

Molecular Basis for Enantioselectivity of Lipase
J . Org. Chem., Vol. 66, No. 9, 2001 3047
Ta ble 3. Su m m a r y of Cr ysta lliza tion of P u r e
En a n tiom er s of 1a
stirring overnight at room temprature, HPLC analysis showed
>99% 1-dibutanoate. Water (1 L) was added to dissolve the
suspended triethylamine hydrochloride. The mixture was
transferred to a 6-L separatory funnel. The water layer was
discarded, and the organic phase was washed with distilled
water (2 × 500 mL), dried over MgSO4, filtered, and concen-
trated to a yellow gum by rotary evaporation. Vacuum distil-
lation (800 µm Hg) yielded a fraction boiling between 220 and
245 °C: oil, 568 g, 91%. HPLC and NMR showed >99% pure
(()-1-dibutanoate. 1H NMR δ 1.06 (m, 10 H), 1.32 (s, 3H), 1.65
(s, 3H), 1.78 (m, 4H), 2.21 (d, 1H), 2.40 (d, 1H), 2.52 (dt, 4H),
6.78 (d, 1H, J ) 2), 6.95 (d, 2H, J ) 8), 6.98 (dd, 1H, J ∼ 2, J
∼ 8),, 7.17 (two overlapping d, 3H, J ∼ 8); MS (FAB, M+H+)
409.
crop
amount (g)
ee (%)
(R)-(+)-1
1
2
3
4
88.6
20.8
40.4
7.3b
99
28
97
97
total
136.3 g with >97% ee
(S)-(-)-1
1
2
3
4
79.8
40.4
42.5
c
96
24
97
77
520-g Sca le Resolu tion of (()-1-Dibu ta n oa te. A 2-L,
three-neck, round-bottom flask fitted with an overhead stirrer
and a pH electrode connected to a pH controller was charged
with (()-1-dibutanoate (520 g, 1.27 mol), tert-butyl methyl
ether (400 mL), and phosphate buffer (0.1 M, pH 7.5, 150 mL).
After stirring to form an emulsion, a solution of CVL (40.8
mg, Sigma catalog no. L 0763, 120 000 units) in phosphate
buffer (0.1 M, pH 7.5, 50 mL) was added. The pH controller
maintained a pH of 7.2 by controlling the addition of 5.0 M
NaOH. Aliquots (100 µL) for HPLC analysis (Chiralpak AD)
were diluted in hexanes-ethanol (12: 1, 10 mL), dried over
MgSO4, and filtered through a 2 µm filter. After 31 h, 0.58
mol of base had been added and the solution contained 48.4
mol % (+)-1-dibutanoate, 50.4 mol % (-)-1-4′-monobutanoate
(89% ee), and 1.2 mol % of diol 1. The amount of base added
(0.58 mol) is slightly less than expected from the HPLC
analysis (0.67 mol).
The reaction mixture was transferred to a 3-L separatory
funnel. tert-Butyl methyl ether (500 mL) and distilled water
(500 mL) were added, the mixture was shaken, and the phases
were allowed to separate (∼15 min). The aqueous phase was
discarded. The organic phase was washed with distilled water
(2 × 500 mL) and concentrated first by rotary evaporation and
then by heating under vacuum at 120 °C to give an orange
gum, 479 g. This gum was dissolved in methylene chloride (250
mL) and chromatographed in six portions on silica gel (1.5 kg,
60-200 mesh) eluted first with dichloromethane. When the
column effluent no longer contained (+)-1-dibutanoate (moni-
tored by TLC), the solvent was switched to methanol to elute
(-)-1-4′-monobutanoate. The column was repacked with virgin
silica gel after each run. Removal of solvent from column
fractions gave (+)-1-dibutanoate (yellow gum, 269 g) and (-)-
1-4′-monobutanoate: orange gum, 245 g, 1H and 13C NMR
match those reported previously.8
(R)-(+)-1 a n d (S)-(-)-1. Solid KOH (1.2 mol for each molar
equivalent of ester) was added to a 2-L, one-neck, round-
bottomed flask containing (+)-1-dibutanoate (269 g) dissolved
in methanol (500 mL). A strong exotherm heated the solution
to reflux. The solution was stirred for 1 h at room temperature
and then neutralized with an equivalent amount of concen-
trated HCl. Distilled water (500 mL) and tert-butyl methyl
ether (500 mL) were added, and the mixture was transferred
to a separatory funnel and shaken gently, and the phases
allowed to separate. The aqueous layer was removed and
saved. The organic phase was washed twice with distilled
water (500 mL). All of the aqueous washes were combined and
washed twice with tert-butyl methyl ether (500 mL). The
combined organic phases were concentrated by rotary evapora-
tion yielding (+)-1 as a light brown solid, 194 g, 89% ee. A
similar procedure yielded (-)-1, 224 g, 87% ee.
Solid (+)-1 was recystallized by dissolved in refluxing CHCl3
(500 mL) and MeOH (50 mL). Cooling in the freezer overnight
yield the first crop of crystals which were recovered by
filtration, Table 3. The filtrate was concentrated to half its
volume by rotary evaporation and cooled to room temperature.
A fine powdery precipitate formed after 1 h and was recovered
by filtration (crop 2). The remaining solution was cooled in
the freezer overnight. A third crop of crystals was recovered
by filtration. The remaining solution was concentrated by
rotary evaporation, and the resulting solid was sublimed under
vacuum to obtain a fourth crop of crystals: total of 136.3 g of
total
122.3 g with >96% ee
From chloroform-methanol (10:1). By sublimation. c Not
isolated.
a
b
(+)-1 with >97% ee, 39.8% yield, [R]20 +170 ( 20 (c ) 3.9,
D
MeOH), mp 148-150 °C (racemic mp 203-206 °C). Solid (-)-1
was recrystallized in the same manner: total 122.3 g of (-)-1
with >96% ee, 35.7% yield, [R]20D -150 ( 20 (c ) 4.1, MeOH),
mp 145-147 °C. The maximum yield in a resolution is 50%.
En a n tiom er ic P u r ity. Enantiomers of 1 were separated
on a Chiralcel OD HPLC column as reported previously.9
Enantiomers of 1, the four isomers of 1-monobutanoates and
1-dibutanoate were also separated on two Chiralpak AD HPLC
columns connected in series (Chiral Technologies Inc. Exton,
PA) eluted with hexanes-ethanol (92/8, 1 mL/min): (()-1-
dibutanoate: k′ ) 0.158; 1-5-monobutanoate: R ) 1.09, k′S
)
0.563, k′R ) 0.611; 1-4′-monobutanoate: R ) 1.11, k′R ) 0.755,
k′S ) 0.838; 1: R ) 1.11, k′S ) 2.25, k′R ) 2.50. Enantiomers
of the diesters did not separate on either the OD or the AD
column. To determine the enantiomeric purity of 1-diesters,
they were cleaved with methanol/sodium methoxide, and the
enantiomeric purity of the resulting 1 was determined as
above.
Hom ology Mod el of th e Op en Con fom a tion of CVL. We
submitted the amino acid sequence of CVL (Swiss Prot code
Q05489) and the open structure of PCL (Brookhaven protein
databank code 1oil) to Swiss-Model,16 an automated protein
modeling server, where the structure was modeled using
ProMod 2.0. A calcium ion was added manually to the calcium
ion site between Asp241 and Asp 287. The structure did not
have any water molecules. A check of this model using the
program WhatCheck17 revealed a number of side chain atoms
in high-energy orientations, but the minimizations below
corrected these errors.
Mod elin g of Tr a n sition Sta te An a logu es in CVL. All
modeling was done with Discover, version 2.9.7 (Biosym/MSI,
San Diego, CA) using the Amber force field with a distance
dependent dielectric constant of 4.0 and the 1-4 van der Waals
interactions scaled by 50%. The distance dependent dielectric
constant damps long-range electrostatic interactions to com-
pensate for the lack of explicit solvation. Results were dis-
played using Insight II version 95.0 (Biosym/MSI). Protein
structures in Figure 2 were created using RasMac v2.6.25 Using
the Biopolymer module of Insight II, hydrogen atoms were
added to correspond to pH 7.0. Histidines were uncharged,
aspartates and glutamates were negatively charged and
arginines and lysines were positively charged. The catalytic
histidine (His286) was protonated. The phosphonate transition
state analogue was built manually and covalently linked to
Ser 87. Energy minimization proceeded in four stages. First,
200 of iterations steepest descent algorithm, all protein atoms
constrained with a force constant of 10 kcal mol-1 Å-2; second,
200 iterations of conjugate gradients algorithm with the same
constraints; and third, 500 iterations of conjugate gradients
algorithm with only the backbone constrained by a 10 kcal
mol-1 Å-2 force constant. For the fourth stage, minimization
was continued using conjugate gradients algorithm without
(25) Sayle, R. A.; Milner-White, E. J . Trends Biochem. Sci. 1995,
20, 374-376. http://www.umass.edu/microbio/rasmol/.

3048 J . Org. Chem., Vol. 66, No. 9, 2001
Gascoyne et al.
any constraints until the rms deviation reached less than 0.005
Å mol-1. The five expected hydrogen bonds between the
transition state analogue and the catalytic amino acid residues
were similar for both the (R)- and (S)-enantiomers, but two of
these hydrogen bonds were significantly longer than the
expected 3.0 ( 0.2 Å. The N-O distances between the main
chain amide of Gln88 and the oxyanion was 3.4-3.5 Å, and
this distance between the imidazole and the O-aryl was 3.8
Å. We attribute these long bonds to either the lack of water
molecules in the structure or inaccuracies in the homology
model.
Ack n ow led gm en t. We thank the US Air Force for
funding and NSERC (Canada) for the funds to purchase
the modeling computer and software. We thank
Michael (Xiaoming) Zhang for initial experiments with
CRL.
J O005681V