Efficient, scalable kinetic resolution of cis-4-benzyloxy-2,3-epoxybutanol

Article abstract of DOI:10.1016/j.tetasy.2005.09.029

A new enzyme catalysed kinetic resolution of cis-4-benzyloxy-2,3- epoxybutanol has been reported. Efficient, scalable separation of the optically active alcohol from its ester derivative has been solved with liquid-liquid and solid-liquid extraction metho

Full text of DOI:10.1016/j.tetasy.2005.09.029

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 16 (2005) 3841–3847
Efficient, scalable kinetic resolution of
cis-4-benzyloxy-2,3-epoxybutanol
a
a
a
b
Ferenc Faigl,a, Angelika Thurner, Melinda Battancs, Ferenc Farkas, Laszlo Poppe,
*
´
´
b
c
c
´
´
´
´
´
Viktoria Bodai, Ildiko Kmecz and Bela Simandi
aResearch Group at Department of Organic Chemical Technology, Hungarian Academy of Sciences, Budapest University of
Technology and Economics, H-1521 Budapest, Hungary
bInstitute for Organic Chemistry and Research Group for Alkaloid Chemistry, Budapest University of Technology and Economics,
H-1521 Budapest, Hungary
cDepartment of Chemical Engineering, Budapest University of Technology and Economics, H-1521 Budapest, Hungary
Received 14 September 2005; accepted 23 September 2005
Available online 11 November 2005
Abstract—A new enzyme catalysed kinetic resolution of cis-4-benzyloxy-2,3-epoxybutanol has been reported. Efficient, scalable sep-
aration of the optically active alcohol from its ester derivative has been solved with liquid–liquid and solid–liquid extraction meth-
ods using commercial organic solvents and supercritical carbon dioxide. cis-4-Benzyloxy-2,3-epoxy-1-butanol enantiomers were
applied for the enantioselective synthesis of (2S,3S,10S)- and (2R,3R,10R)-3-[20-(dibenzylamino)-10-hydroxyethyl]-2-phenyloxetane.
ꢀ 2005 Elsevier Ltd. All rights reserved.
1. Introduction
acylated alcohol from the unreacted enantiomer is a
key step of such protocols. The usual isolation method
is column chromatography but it can be tedious on a
multigram scale. Therefore, during development of a
new enzyme catalysed kinetic resolution process for
compound 1, different methods were also investigated
in our laboratories to find the optimum solution for
the separation of the unreacted enantiomer of 1 from
the reaction product.
Optically active oxiranes are useful intermediates in the
synthesis of biologically active compounds. Regio- and
stereoselective ring opening or rearrangement of chiral
2,3-disubstituted oxiranes are known routes to optically
active alcohols, oxetanes and cis-but-2-ene-1,4-diol
derivatives.1–3 The oxetanes, for example, can be used
as building blocks in the synthesis of oxetanocin A.4 Dif-
ferent routes are known for the preparation of nonrace-
mic chiral oxiranes. Allylic alcohols can be oxidised
enantioselectively with tert-butyl hydroperoxide, tita-
nium isopropoxide and a chiral catalyst.5 Recently we
reported the resolution of two amino group-containing
oxiranyl ethers by preparing their diastereomeric salts.6
The disadvantage of this resolution is that we have to
find special conditions for each amino derivative.
2. Results and discussion
Synthesis of racemic 1 from the commercially available
cis-2-buten-1,4-diol via monobenzylation and epoxida-
tion was performed according to the literature proce-
dure.2 The enantiomer selectivity of the enzyme
catalysed esterification of racemic 1 was screened with
a wide selection of enzymes [AK, AY, F, M, Ps and R
lipases from Amano; Novozym 435, Lipase IM 20 and
Lipase TL IM from Novozymes; porcine liver esterase
(liver acetone powder), Candida rugosa and Pseudomo-
nas fluorescens lipases from Fluka; a-chymotripsin,
papain, Candida cylindracae and porcine pancreatic
lipases from Sigma]. Vinyl acetate was used as an acyl-
ating reagent and the esterification was carried out in
hexane in these test reactions. Gas chromatographic
The kinetic resolution of cis-4-benzyloxy-2,3-epoxybuta-
nol 1 by enzyme catalysed enantiomer selective acylation
would solve this problem and make possible the prepa-
ration of a large variety of enantiomerically pure oxira-
nyl ethers. Efficient and scalable separation of the
*
Corresponding author. Tel.: +36 1463 3652; fax: +36 1463 3648;
e-mail: ffaigl@mail.bme.hu
0957-4166/$ - see front matter ꢀ 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2005.09.029

3842
F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
analyses performed on a chiral stationary phase contain-
ing capillary column helped us in monitoring the enzy-
matic reactions. Comparison of the yields of the
produced acetate 2a and the residual alcohol 1 and ee
values of 1 led us to conclude that the porcine pancreatic
lipase (PPL) was one of the most active and enantiose-
lective catalysts. Therefore all the further optimization
reactions were carried out with PPL (Scheme 1) which
accelerates esterification of the (ꢀ)-(2S,3R)-1 enantio-
mer, thus an excess of (+)-(2R,3S)-1 remained unreacted
in the reaction mixture. The absolute configuration of
optically active 1 is known.5
increased in that mixture, compared to the selectivity
observed in hexane but remained lower than that in
tetrahydrofuran (Table 1). The reaction was a bit slower
but the enantioselectivity significantly increased
(E ꢁ 23), probably due to the smaller water content of
the more concentrated reaction system and larger sub-
strate/enzyme ratio, in gram scale experiments (cf. Sec-
tion 4.2). In that case, the reaction was interrupted at
about 58% conversion to yield the unreacted (+)-1 iso-
mer in 97% ee.
In order to determine the dependence of the results on
the type of the acyl group, the esterification process
was investigated under the same reaction conditions as
described above using vinyl acetate, propionate and
butyrate, respectively (Scheme 1). The highest reaction
rate was observed with vinyl butyrate, while approxi-
mately the same rates were detected with vinyl acetate
and propionate (Table 2). The type of the acyl group
did not significantly influence the enantiomer selectivity
of the reaction. At about 62–65% conversion (calculated
to the racemic 1) 96–97% ee of the residual alcohol 1 was
detected in each case.
In order to find the optimum reaction conditions, the
influence of the solvent, the acylating reagent (vinyl ace-
tate, propionate and butyrate) and the epoxy alcohol/
PPL ratio on the result of the kinetic resolution was
investigated.
In the first run, the effect of the solvent was studied at
ambient temperature. The alcohol 1/vinyl acetate/PPL/
solvent ratios were kept constant in these experiments
(Table 1) and after 2 h of shaking the reactions were
stopped by filtration from the enzyme. The yields of
the acetate 2a and the residual (+)-1 and (ꢀ)-1 alcohols
were determined by gas chromatography.
The (ꢀ)-1 enantiomer was prepared in ethanol via PPL
catalysed alcoholysis of (ꢀ)-2a and (ꢀ)-2b, respectively
(Scheme 2, Table 3). In spite of the increased amount
of the enzyme, the reactions in ethanol were much
slower than the acylation of ( )-1 with vinyl esters
Surprisingly, the reaction was slower in tetrahydrofuran
or in hexane than in their mixture. The selectivity (E)
O
H
O
O
H
O
PPL, solvent
H
H
+
CH3
+
3
2
O
H
1
4
R
OBn
OH
OBn
O
R
OH
OBn
O
O
R = Me
R = Et
R = Pr
(-)-(2S,3R)-2a
(-)-(2S,3R)-2b
(-)-(2S,3R)-2c
( )-1
(+)-(2R,3S)-1
Scheme 1. PPL catalysed kinetic resolution of 1.
Table 1. Results of the enzymatic resolution of 1 in different solvents after 2 h reactiona
Solvent
(+)-1 (%)
(ꢀ)-1 (%)
(ꢀ)-2a (%)
2.9
ee 1 (%)
Eb
b
Chloroform
Dichloromethane
Acetone
50.9
50.4
49.1
37.6
33.6
36.2
45.4
44.6
45.8
45.7
45.3
47.0
42.4
37.1
33.5
12.8
1.9
9.1
15.2
18.9
49.2
89.3
89.0
40.3
58.4
48.9
56.8
63.2
31.8
b
8.8
b
14.0
47.5
63.1
60.1
32.6
41.4
35.9
39.3
42.2
25.8
Hexane
5.4
Vinyl acetate
THF–hexane 1:1c
Cyclohexane
Toluene
Ethyl acetate
Acetonitrile
Tetrahydrofuran
Diethyl ether
9.0
11.1
16.4
19.0
24.1
27.0
26.7
31.0
2.1
19.3
11.7
15.7
12.6
10.2
24.3
a Starting amounts of racemic 1/vinyl acetate/PPL/solvent: 20 mg/200 lL/20 mg/1800 lL, in all cases. The amounts of the formed 2a and the residual
(+)-1 and (ꢀ)-1 are given in mol %, on the basis of GC measurements.
b Degree of the enantiomer selectivity (E) was calculated from the conversion and ee of 1.8 Due to extreme sensitivity to experimental errors at small
conversions, no E values were calculated for conversion <15%.
c THF: tetrahydrofuran.

F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
3843
Table 2. Enzymatic resolution of 1 in the presence of different acyl
group donating agentsa
from ( )-1 by
sequence.
a
combined acylation–ethanolysis
Reaction
Amount of the ester
time (min)
Due to the difference in their polarity, alcohol 1 and
esters 2a–c could be separated by column chromatogra-
phy on silica gel using hexane/ethyl acetate 2:1 (v:v)
mixture as eluent. Multigram scale separation of opti-
cally active 1 from the esters 2a,b or 2c is quite tedious
using this method. In order to find out more efficient,
scalable procedure for separation of compounds 1 and
2, alternative (liquid–liquid and liquid–solid) extractive
methods were investigated. First, selectivities of hex-
ane/aqueous methanol two-phases systems were studied
in separation of the free alcohol 1 and the acetate 2a as a
function of the methanol content of the polar liquid
phase. A mixture of 1 and 2a was distributed between
equal volumes of hexane and aqueous methanol by
shaking the mixture for 20 min. Then the phases were
separated and the distributions of 1 and 2a were moni-
tored by gas chromatography. The experimental details
and results are summarised in Table 4.
2ab with vinyl
acetate (%)
2bb with vinyl
propionate (%)
2cb with vinyl
butyrate (%)
30
60
90
120
150
180
11.8
22.0
31.8
40.3
45.5
50.2
8.6
18.5
26.8
37.1
42.8
53.6
19.1
32.4
45.2
53.5
60.5
64.9
a All reactions were accomplished using 0.29 g PPL for 1.0 g of racemic
1.
b Data are given in mol %, related to the amount of the starting alcohol
1.
PPL
O
O
O
H
H
H
H
+
R
OEt
EtOH
OBn OH
OBn
O
R
The best selectivity was achieved with a methanol/water
1:9 (v:v) mixture. According to these results, on a gram
scale, a mixture of 1 and 2a was dissolved in aqueous
methanol and the solution was extracted five times with
hexane. The hexane solutions contained a 95:5 mixture
of 2a and 1. Pure 2a could be obtained in 57% yield
when this mixture was extracted twice two times with
a methanol/water 1:9 mixture. The original aqueous
methanol solution contained chemically pure alcohol
1. It could be isolated from the solution in 75% yield
by evaporation of the solvents. However, this extraction
method requires large amounts of solvents relative to
the amounts of the separated compounds 1 and 2a.
O
(-)-2a
(-)-22b
R = Me
R = Et
(-)-(2S,3R )-1
Scheme 2. Alcoholysis of (ꢀ)-2a and (ꢀ)-2b.
Table 3. Rates of the PPL catalysed alcoholysisa of 2a and 2b
Reaction time (h)
Alcoholysis of 2a
Alcoholysis of 2b
2ab (%) (ꢀ)-1b (%) 2bb (%) (ꢀ)-1b (%)
100.0
93.2
41.8
24.4
0
1
24
49
0.0
6.8
58.2
75.6
100.0
93.8
42.2
28.8
0.0
6.2
57.8
71.2
Next, the amounts of solvents used in liquid–liquid
extraction were decreased by application of solid carri-
ers for selective adsorption of one between the polar
alcohol 1 and apolar ester 2b. Thus, the enzyme was fil-
tered off from the reaction mixture of the kinetic resolu-
tions and different carriers (Perfil, Silica gel 60, Celite
545 or Florisil) were suspended in the filtrates, respec-
tively. Then the mixture of (+)-1 and (ꢀ)-2b was
adsorbed on the carriers by evaporation of the solvents
and the solid residue was stirred with hexane and filtered
again. Ester 2b was isolated from the hexane filtrate,
a Reactions in the presence of 1.35 g of PPL for 1.0 g of substrate 2.
b Relative amounts (mol %) of 2a or 2b and (ꢀ)-1 were calculated from
GC data.
and there were no significant differences between the eth-
anolysis rates of 2a and 2b. The PPL enzyme facilitate
the reactions of the same enantiomeric form of 1 in both
directions (esterification and alcoholysis), therefore vir-
tually enantiopure (ꢀ)-1 (ee >99%) could be obtained
Table 4. Liquid–liquid extractive separation of 1 and 2a from their mixture using different hexane/aqueous methanol systems
Solvent compositions (mL)
Amounts of 2a (mg) and 1 (mg)a
In hexaneb In aqueous methanolb
Selectivityc
Methanol
Water
Hexane
2a
1
2a
1
1
3
4
5
9
9
7
6
5
1
10
10
10
10
10
9.50
6.09
1.63
0.59
0.10
0.2
6.47
6.83
6.92
7.00
7.02
7.02
50.14
42.61
39.76
26.93
4.42
0.10
0.02
0.01
0.01
9.89
14.35
15.38
15.87
a Composition of the starting mixture: 7.2 mg of 1 + 16.0 mg of 2a.
b Determined by GC using undecene as internal standard.
c Selectivity was calculated as the ratio of the hexane/aqueous methanol distribution of 2a and 1.

3844
F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
Table 5. Compositionsa of the hexane filtrate and the dichloromethane extract of the hexane washed carriers supplemented with mixtures of 1 and 2b
Carrier (1.0 g)
Perfil
Florisil
Celite
Silica gel
Starting composition
1 (g)
2b (g)
1 (g)
2b (g)
1 (g)
2b (g)
0.1900
0.3100
0.0662
0.2638
0.1187
0.0313
0.1900
0.3100
0.0482
0.2418
0.1057
0.0143
0.1900
0.3100
0.0687
0.2954
0.1041
—
0.1900
0.3100
—
0.2131
0.0754
0.0780
Hexane solution
Dichloromethane solution
a Determined by GC.
then alcohol 1 was recovered from the solid phase by
washing it with dichloromethane (Table 5).
mixture of 1 and 2b in 4:1 ratio while the third fraction
contained only a trace amount of these compounds. The
alcohol 1 was isolated from the residual solid by extrac-
tion with methanol (Table 6). This method afforded
good yields and sufficient purities (97% for 1, 93% for
2b) and, in addition, serves an environmentally begin
This new solid–liquid extraction method provided pure
ester 2b in 69% yield from the hexane filtrate when silica
gel was the solid carrier and pure alcohol 1 was obtained
from the dichloromethane solution in 55% yield when
Celite was used as solid support.
7
process for preparation of optically active 1 and 2.
In order to demonstrate the synthetic usefulness of
(2R,3S)-1, it was converted into a novel optically active
dibenzylamino group containing oxiranyl ether 4 via the
tosylate 3. Then compound 4 was transformed into oxe-
tane 5 by using potassium tert-butoxide activated lith-
ium diisopropylamide (LiDA–KOR, Scheme 3),
according to our previously reported rearrangement
process.3 The enantiomeric purity maintained through-
out these reactions as well as during the formation of
the new stereogenic centres in the oxetane ring of 5.
Configuration of C10 did not change during these reac-
tions. Starting from this fact, the absolute configura-
tions of the new stereogenic centres of the oxetane ring
in 5 have been determined on the basis of our earlier
NMR investigations.6 The opposite enantiomers of 4
and 5 were synthesised from (2S,3R)-1 in the same way.
It is known that supercritical carbon dioxide (scCO2)
acts as an apolar solvent like hexane. Application of
scCO2in the above mentioned solid–liquid extractive
method could further decrease the amount of the organic
solvents used. Thus, solvent free esters 2 can easily be
obtained after evaporating the carbon dioxide from
the scCO2 extract. First, this idea was tested by the
attempted separation of 1 and 2a with supercritical
fluid–solid extraction at 39 ꢁC and different pressures
(90, 160 bar) but insufficient separation was observed.
Much better separation has been achieved when the dif-
ference between the polarities of alcohol 1 and ester 2
was increased by using the propionate 2b instead of
the acetate 2a. Under these experiments, the crude reac-
tion mixture [(+)-1 and (ꢀ)-2b] obtained from the enzy-
matic reaction was mixed with silica gel and the solvents
were evaporated. The residual solid was filled into the
extractor tube and treated with scCO2 at 33 ꢁC,
100 bar. Three extract fractions were collected succes-
sively as separate samples. The ester 2b was retrieved
directly from the first fraction by decreasing the pres-
sure. The second fraction contained about 8% of the
3. Conclusion
The experimental results allowed us to conclude that
efficient kinetic resolution of 4-benzyloxy-2,3-epoxybut-
anol 1 can be accomplished by PPL catalysed acylation
Table 6. Supercritical fluid extraction of (ꢀ)-2b from its mixture with (+)-1 at 100 bar, 33 ꢁC
Starting mixturea
First fractionb
Second fractionb
Residuec
Weight
2.13 g
0.81 g, 38%
1.32 g, 62%
1.01 g
0.07 g, 7%
0.94 g, 93%
0.17 g
0.03 g, 17%
0.14 g, 83%
0.81 g
0.79 g, 97%
0.02 g, 3%
(+)-1 contentd
(ꢀ)-2b contentd
a On 4.26 g of silica gel (60 mesh).
b From the carbon dioxide extracts (18 g of scCO2 eluent was used for each fraction).
c Recovered from silica gel with methanol.
d Determined by GC.
1O
O
O
O
H
H
H
H
H
H
Bn2NH
2 LiDA-KOR
THF, -75oC
TsCl
H
H
3
2
2'
Bn2N
pyridine,
-10˚C
1'
KI, DMF,
0 oC
Ph
OH
,3S,1'
NBn2 OBn
OH
OBn
OTs
(2
OBn
(2
R
,3S
)-1
R
,3S)-3
(2 ,3 )-4
R
S
(2S
S)-5
Scheme 3. Synthesis of (2R,3S)-4 and (2S,3S,10S)-5.

F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
3845
of ( )-1 with vinyl esters in tetrahydrofuran–hexane
mixture. The enantioselectivity did not significantly
depend on the quality of the acyl moiety but the ease of
separation of the formed new esters (ꢀ)-2a, (ꢀ)-2b and
(ꢀ)-2c from the residual alcohol (+)-1 depended on
the polarity difference between the ester 2 and the alco-
hol 1. Thus novel, solid–liquid extractive methods were
developed for the separation of 1 and 2b using different
solid carriers and eluents as hexane or supercritical car-
bon dioxide. The PPL catalysed acylation of ( )-1 with
vinyl propionate and the PPL catalysed alcoholysis of
(ꢀ)-2b combined with the above mentioned solid–liquid
extractive separation method provide a novel, efficient
and scalable process for the preparation of both enanti-
omers of 1 in 97–99% ee. Starting from (+)-1 and (ꢀ)-1,
new, optically active amino group containing oxirane 4
and oxetane 5 derivatives could also be synthesised.
to GC) the enzyme was filtered off and washed with ace-
tone. The filtrate was concentrated in vacuum and the
residual oil was separated by column chromatography
on silica gel (hexane/ethyl acetate 2:1 v:v) to yield opti-
cally active alcohol 1 and ester 2a,b or 2c.
4.2.1. (+)-(2R,3S)-4-Benzyloxy-2,3-epoxybutanol, (+)-
1. Oil (40% of the starting racemate2), 1H NMR
(250 MHz): d 2.30 (1H, br s, OH), 3.3 (2H, m, oxirane
CH), 3.7 (4H, m, CH2O), 4.46 (1H, d, J 11.8,
CHaHbPh), 4.61 (1H, d, J 11.8, CHaHbPh), 7.3 (5H,
m, Ph); [a]D +23.0 (c 2.4, CHCl3), ee 97%.
4.2.2. (ꢀ)-(2S,3R)-1-Benzyloxy-2,3-epoxybutyl acetate,
(ꢀ)-2a. Oil, 1H NMR (300 MHz): d 2.10 (3H, s,
COCH3), 3.3 (2H, m, oxirane CH), 3.59 (1H, dd, J
6.0, 11.4, CHaHbOBn), 3.71 (1H, dd, J 4.2, 11.4, CHaHb-
OBn), 4.03 (1H, dd, J 6.9, 12.3, CHaHbO), 4.32 (1H, dd,
J 3.9, 12.3, CHaHbO), 4.53 (1H, d, J 12.0, CHaHbPh),
4.62 (1H, d, J 12.0, CHaHbPh), 7.4 (5H, m, Ph); 13C
NMR (300 MHz): d 170.93, 137.75, 128.69, 128.10,
128.01, 73.59, 68.01, 62.80, 54.91, 53.20, 20.95; GC–
MS (EI, m/z): 237 ([M+H]+), 173, 107, 91; HRMS
(FAB): [M+H]+ found 237.11202, C13H17O4 requires
237.11268; [a]D = ꢀ12.4 (c 2.2, CHCl3), ee 71% (at
58% conversion of the starting racemate).
4. Experimental
4.1. Materials and methods
NMR spectra were recorded on a Bruker WM 250 or
Bruker DRX 300 spectrometer using deuteriochloro-
form as solvent. Chemical shifts are given relative to
the signal of tetramethylsilane (dTMS = 0.00 ppm). Cou-
pling constants are given in hertz. IR spectra were taken
on a Perkin–Elmer 1600 FT spectrometer. Gas chroma-
tography was carried out on Agilent 4890D instrument
equipped with FID detector (nitrogen as carrier gas,
injector 250 ꢁC, detector 250 ꢁC, head pressure: 15 psi,
at 1:100 split ratio). GC–MS spectra were recorded on
a Finnigan Mat/Automass II GC/MS. Optical rotations
were determined on Perkin–Elmer 241 polarimeter. TLC
was carried out on Kieselgel 60 F254 (Merck) sheets.
Spots were visualised by UV and treatment with iodine
or with an aqueous solution of (NH4)6Mo7O24,
Ce(SO4)2 and sulfuric acid, followed by heating of the
dried plates. Solvents were freshly distilled. Commercial
starting materials and PPL enzyme were purchased from
FLUKA AG and Sigma, respectively, and were used
without further purification. Butyllithium in hexane
solution was supplied by Chemetall AG. Racemic cis-
4.2.3. (ꢀ)-(2S,3R)-1-Benzyloxy-2,3-epoxybutyl propio-
1
nate, (ꢀ)-2b. Oil, H NMR (300 MHz): d 1.15 (3H, t,
J 7.5, CH3), 2.37 (2H, q, J 7.5, COCH2), 3.3 (2H, m, oxi-
rane CH), 3.59 (1H, dd, J 6.5, 11.1, CHaHbOBn), 3.72
(1H, dd, J 4.0, 11.0, CHaHbOBn), 4.04 (1H, dd, J 7.0,
12.5, CHaHbO), 4.33 (1H, dd, J 3.5, 12.5, CHaHbO),
4.54 (1H, d, J 12.0, CHaHbPh), 4.62 (1H, d, J 12.0,
CHaHbPh), 7.3 (5H, m, Ph); 13C NMR (300 MHz): d
174.39, 137.77, 128.69, 128.10, 128.01, 73.59, 68.06,
62.65, 54.94, 53.26, 27.55, 9.21; GC–MS (EI, m/z): 251
([M+H]+), 175, 107, 91; HRMS (FAB): [M+H]+ found
251.12937, C14H19O4 requires 251.12833; [a]D = ꢀ10.8
(c 2.0, CHCl3), ee 52% (at 65% conversion of the starting
racemate).
4.2.4. (ꢀ)-(2S,3R)-1-Benzyloxy-2,3-epoxybutyl butyrate,
1
(ꢀ)-2c. Oil, H NMR (300 MHz): d 0.95 (3H, t, J 7.5,
4-benzyloxy-2,3-epoxybutan-1-ol
1
was prepared
CH3), 1.7 (2H, m, CH2), 2.33 (2H, t, J 7.5, COCH2), 3.3
(2H, m, oxirane CH), 3.61 (1H, dd, J 6.0, 11.1,
CHaHbOBn), 3.73 (1H, dd, J 4.2, 11.4, CHaHbOBn),
4.04 (1H, dd, J 6.9, 12.5, CHaHbO), 4.32 (1H, dd, J
3.9, 12.5, CHaHbO), 4.54 (1H, d, J 12.0, CHaHbPh),
4.62 (1H, d, J 12.0, CHaHbPh), 7.4 (5H, m, Ph); 13C
NMR (300 MHz): d 173.58, 137.77, 128.70, 128.11,
128.02, 73.60, 68.07, 62.55, 54.95, 53.28, 36.13, 18.55,
13.85; GC–MS (EI, m/z): 265 ([M+H]+), 107, 91;
HRMS (FAB): [M+] found 264.13699. C15H20O4
requires 264.13616; [a]D = ꢀ12.3 (c 2.0, CHCl3), ee
54% (at 65% conversion of the starting racemate).
according to the literature procedure.2
4.2. Kinetic resolution of cis-4-benzyloxy-2,3-epoxybuta-
nol 1, general procedure
The PPL enzyme (0.6 g) was added into a solution of
racemic cis-4-benzyloxy-2,3-epoxybutan-1-ol 1 (10.7
mmol, 2.07 g) made in tetrahydrofuran (45 mL), hexane
(45 mL) and vinyl ester (acetate, propionate or butyrate,
respectively; 20 mL) mixture. The resulting suspension
was stirred at room temperature and analysed by gas
chromatography [HP-Chiral B233 capillary column
(30 m · 0.25 mm, 0.25 lm film), 5 min at 180 ꢁC, 4 ꢁC/
min to 200 ꢁC, then kept at 200 ꢁC; retention times:
(+)-(2R,3S)-1: 18.49 min, (ꢀ)-(2S,3R)-1: 18.59 min; 2a
(sum of enantiomers): 21.12 min; 2b (sum of enantio-
mers): 26.31 min; 2c (sum of enantiomers): 33.98 min,
respectively]. At about 58–65% conversion (according
4.2.5. Alcoholysis of (ꢀ)-2, general procedure. The PPL
enzyme (3.2 g) was added to a solution of (ꢀ)-2a or (ꢀ)-
2b (10 mmol, 2.36 g or 2.50 g, ee 71%) in ethanol
(60 mL) and the resulting suspension was stirred at
room temperature for 48 h and analysed by GC (HP-
Chiral B233 capillary column, for conditions, see

3846
F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
Section 4.2). The enzyme was filtered off, washed with
acetone and the filtrate was concentrated in vacuum.
The residue was purified by column chromatography
on silica gel (hexane/ethyl acetate 2:1 v:v) to yield (ꢀ)-
1 (1.23 g, 63%); [a]D = ꢀ23.8 (c 2.2, CHCl3); ee >99%.
The silica gel was filtered off then the methanol solution
was concentrated in vacuum to yield 90% of the starting
alcohol 1 (0.81 g, 97% 1 content, GC).
4.6. Conversion of (+)-(2R,3S)-1 and (ꢀ)-(2S,3R)-1 into
optically active cis-3-[20-(dibenzylamino)-10-hydroxyethyl]-
2-phenyloxetane2 (+)-5 and (ꢀ)-5
4.3. Separation of 1 and 2 by liquid–liquid extraction, a
typical procedure
4.6.1. Optically active cis-4-benzyloxy-2,3-epoxybutyl
tosylates, (2R,3S)-3 and (2S,3R)-3. Compounds
(2R,3S)-3 and (2S,3R)-3 were synthesised from (+)-
(2R,3S)-1 and (ꢀ)-(2S,3R)-1 according to the literature
procedure for racemic 3.2 Tosylates 3 were used without
further purification.
The mixture of alcohol 1 (0.1022 g) and ester 2a
(0.1020 g) was dissolved in a mixture of water (45 mL)
and methanol (5 mL), then extracted with hexane
(5 · 20 mL). The hexane extracts were collected and
concentrated to 25 mL volume in vacuum, then washed
two times with a mixture of water (3.6 mL) and metha-
nol (0.4 mL), dried and evaporated to give pure 2
(0.058 g, 57%, purity 99%, GC). After removal of meth-
anol from the first aqueous methanol solution in vac-
uum, the aqueous phase was extracted with
dichloromethane (25 mL). The dichloromethane solu-
tion was dried on sodium sulfate and concentrated in
vacuum to yield 1 (0.077 g, 75%, purity 99%, GC).
Compound (2R,3S)-3: Oil (98%);1H NMR (250 MHz): d
2.44 (3H, s, Me–Ph), 3.24 (2H, m, oxirane CH), 3.53
(1H, dd, J 5.4, 11.4, CHaHbO), 3.58 (1H, d, J 3.6,
11.4, CHaHbO), 4.05 (1H, dd, J 6.6, 11.5, CHaHbOSO2),
4.24 (1H, dd, J 3.8, 11.5, CHaHbOSO2), 4.48 (1H, d, J
11.8, CHaHbPh), 4.54 (1H, d, J 11.8, CHaHbPh), 7.3
(7H, m, Ph), 7.78 (2H, d, J 8.3, Ph).
1
4.4. Separation of (+)-1 and (ꢀ)-2b by solid–liquid extra-
Compound (2S,3R)-3: H NMR spectrum of (2S,3R)-3
ction using different carriers
was indistinguishable from that of (2R,3S)-3.
A mixture of (+)-1 (0.98 mmol, 0.19 g) and (ꢀ)-2b
(1.24 mmol, 0.31 g) was dissolved in dichloromethane
(5 mL), then a solid carrier (Perfil, Silica gel 60, Celite
545 or Florisil, 1.0 g) was added and the suspension
was evaporated in vacuum. The dry solid was stirred
with hexane (15 mL) for an hour, then filtered off and
washed with hexane (2 · 5 mL).
4.6.2. Optically active cis-4-benzyloxy-1-dibenzylamino-
2,3-epoxybutanes,2 (ꢀ)-4 and (+)-4. Compounds (ꢀ)-4
and (+)-4 were synthesised from (2R,3S)-3 and
(2S,3R)-3 by replacement of the tosyl group with diben-
zylamine according to the literature procedure.2
Compound (ꢀ)-(2R,3S)-4 (from (+)-(2R,3S)-1): Oil,
(74%); 1H NMR (250 MHz): d 2.44 (1H, dd, J 4.0,
13.0, CHaHbN), 2.76 (1H, dd, J 4.0, 13.0, CHaHbN),
3.2 (2H, m, oxirane CH), 3.4 (1H, m, CHaHbO), 3.50
(2H, d, J 13.6, NCH2Ph), 3.6 (1H, m, CHaHbO), 3.83
(2H, d, J 13.6, NCH2Ph), 4.47 (1H, d, J 11.2, CHaHbPh),
4.59 (1H, d, J 11.2, CHaHbPh), 7.3 (15H, m, Ph).
The filtrate was concentrated in vacuum. Using silica gel
as carrier, pure propionate (ꢀ)-2b (0.22 g, 69% of the
starting ester, [a]D = ꢀ11.5 (c 2.0, CHCl3), ee 58%)
was obtained from the hexane solution. The solid phase
was suspended in dichloromethane (10 mL) and stirred
for 30 min. The suspension was filtered off, then washed
with dichloromethane (2 · 5 mL). The filtrate was con-
centrated in vacuum. Using Celite as carrier, pure alco-
hol (+)-1 (0.10 g, 55% of the starting alcohol (+)-1,
[a]D = + 23.4 (c 2.1, CHCl3), ee 98%) was obtained from
the dichloromethane solution.
[a]D = ꢀ19.7 (c 2.3, CHCl3), ee 97%; (+)-(2S,3R)-4
1
(from (ꢀ)-(2S,3R)-1): oil, (74%); The H NMR spec-
trum was indistinguishable from that of (ꢀ)-(2R,3S)-4;
[a]D = + 20.0 (c 2.2, CHCl3), ee >99%.
4.6.3. Optically active cis-3-[20-(dibenzylamino)-10-
hydroxyethyl]-2-phenyloxetanes,2 (+)-5 and (ꢀ)-5.
Enantiomers of 5 were prepared via rearrangement of
optically active 4, induced by the superbasic mixture
of potassium tert-butoxide and lithium diisopropyla-
mide, according to the literature procedure.2,6
4.5. Separation of 1 and 2b by supercritical fluid extraction
A mixture of 1 (4.48 mmol, 0.87 g) and 2b (5.03 mmol,
1.26 g) was dissolved in dichloromethane (10 mL), then
silica gel (4.26 g) was added and the solvent was
removed from the suspension by vacuum rotary evapora-
tor. The solid residue was filled into an extractor tube
and treated with supercritical carbon dioxide (3 · 18 g)
at 100 bar, 33 ꢁC. (A more detailed description of the
equipment and the extraction is given elsewhere.8) From
the first fraction 75% of the starting 2b (1.01 g, 93% 2b
content, GC) was recovered. The second fraction con-
tained a mixture of 1 and 2b in approximately 1:5 ratio
(0.17 g, Table 6) and the third fraction contained only
trace amounts of 1 and 2b. The residual solid was
removed from the extractor, suspended in methanol
(25 mL) and stirred at room temperature for 30 min.
Compound (+)-(2S,3S,10S)-5 (from (ꢀ)-4): oil (47%);
1H NMR (250 MHz): d 1.60 (1H, br s, OH), 2.3 (2H,
m, NCH2), 2.8 (1H, m, oxetane CH), 3.34 (2H, d, J
13.2, NCH2Ph), 3.89 (2H, d, J 13.2, NCH2Ph), 4.1
(1H, m, CHOH), 4.61 (2H, d, J 8.0, oxetane CH2O),
5.56 (1H, d, J 6.2, oxetane CHPh), 7.3 (15H, m, Ph);
[a]D = ꢀ35.6 (c 2.3, CHCl3), ee 97%.
Compound (ꢀ)-(2R,3R,10R)-5 (from (+)-4): oil (47%); 1H
NMR spectrum was undistinguishable from that of (+)-
(2S,3S,10S)-5; [a]D = + 37.7 (c 2.1, CHCl3), ee > 99%.

F. Faigl et al. / Tetrahedron: Asymmetry 16 (2005) 3841–3847
3847
}
2. Thurner, A.; Faigl, F.; Toke, L.; Mordini, A.; Vallachi, M.;
Reginato, G.; Czira, G. Tetrahedron 2001, 57, 8173–
8180.
Acknowledgements
The project was supported by the Hungarian Scientific
Research Foundation (T 042805, T 048854 and T
049362). The authors are grateful to Dr. Gabor Tarka-
nyi, for discussions on the NMR spectra, Dr. Laszlo
Vida for recording GC–MS spectra, and Dr. Alessandro
Mordini, for cooperation and discussions on the base
promoted rearrangement reactions.
3. Thurner, A.; Faigl, F.; Mordini, A.; Bigi, A.; Reginato, G.;
}
Toke, L. Tetrahedron 1998, 54, 11597–11602.
´
´
´
´
´
4. Ichikawa, E.; Kato, K. Synthesis 2002, 1–28.
5. Gao, Y.; Hanson, R. M.; Klunder, J. M.; Ko, S. Y.;
Masamune, H.; Sharpless, K. B. J. Am. Chem. Soc. 1987,
109, 5765–5780.
´
´
´
6. Faigl, F.; Thurner, A.; Tarkanyi, G.; Kovari, J.; Mordini,
A. Tetrahedron: Asymmetry 2002, 13, 59–68.
7. Matsuda, T.; Harada, T.; Nakamura, K.; Ikariya, T.
Tetrahedron: Asymmetry 2005, 16, 909–915.
References
´
´
8. Simandi, B.; Keszei, S.; Fogassy, E.; Kemeny, S.; Sawinsky,
1. Mordini, A.; Bindi, S.; Pecchi, S.; DeglÕInnocenti, A.;
Reginato, G.; Serci, A. J. Org. Chem. 1996, 61, 4374–4378.
J. J. Supercrit. Fluids 1998, 13, 331–336.