Y. Ren, et al.
MolecularCatalysis468(2019)52–56
Fig. 1. Structures of lamivudine (1a), ent-lamivudine (1b)
and emtricitabine (1c).
catalyzed hydrolysis reactions [44,45], but also due to the benzoyl
group in the primary position. However, selective formation of primary
alcohols can be observed with cyclic substrates, including carbohy-
To potentially achieve higher yields in the second step, we returned
to enantioenriched isomer 4a, obtained in the STS-catalyzed DKR
process when the reaction was carried out in the presence of phenyl
acetate and triethylamine at 4 °C. Compound 4 was in this case obtained
as a mixture, consisting of two pairs of diastereomers, in which the
major isomer 4a exhibited 50% ee (Scheme 3). As shown in Fig. S2, the
ratio of the trans- and cis-isomers was 84:16, compared to a ratio of
approximately 2:1 in the reaction not catalyzed by enzyme [34]. The
enzyme preparation thus further favored formation of the trans isomers.
CAL B was subsequently used to catalyze the hydrolysis of mixture 4 in
a PBS/THF (1:1) system. The results were conspicuous, and selective
hydrolysis of the acetate group near the C2 position was observed, re-
sulting in pure primary alcohol 7 (Scheme 3, Fig. S3). According to 1H
NMR spectroscopy and chiral HPLC analysis in form of acetylated
compound 4a, an excellent diastereomeric ratio (dr) (33:1) and an ex-
cellent ee (> 99%) were determined for compound 7. Combined with
NOE-NMR spectroscopy and chiral HPLC studies [34], the absolute
configuration of compound 7 could be elucidated as 2R,5R. The results
indicate that CAL B catalyzes the regio- and stereoselective hydrolysis
of the C-2(R)-ester as well as the C-5(S)-ester, resulting in the exclusive
formation of isomer 7.
With access to the enantiopure isomer, different parameters were
next examined for optimization of the STS-CAL B-mediated enzymatic
protocol. In the first step, when toluene was used as solvent, both the ee
and the yield decreased significantly compared with THF. Moreover,
prolonging the reaction time from 2 d to 4 d could slightly improve the
yield, however at the cost of reducing the ee from 57% to 42%. In ad-
dition, increasing the STS loading did not influence the reaction sig-
nificantly (Table 1). In the cascade hydrolysis step, THF was proven to
be the optimal solvent compared with TBME and toluene [34].
From these results, THF was considered the optimal solvent for both
processes, establishing the cascade enzyme-mediated synthesis of en-
antiopure 1,3-oxathiolane derivatives. Compounds 2 and 3 were first
allowed to react at 4 °C in the presence of STS for 2 d, at which time the
solid was removed by filtration. PBS buffer and CAL B were subse-
quently added to the filtrate, and the reaction proceeded at ambient
condition overnight. Following this procedure, compound 7 could be
obtained in 50% yield and > 99% ee after purification by column
chromatography. This DKR-KR protocol for the formation of en-
antiomerically pure (2R,5R)-1,3-oxathiolane derivatives thus display
advantages in, for example, reducing the number of synthetic steps and
increasing the final yields in comparison to traditional KR processes.
Enzyme reusability is an important aspect when considering prac-
tical applications, not only for lowering the costs, but also for avoiding
unnecessary environmental impact. In order to examine the reusability
of the STS preparation, experiments were performed under the same
reaction conditions as used for the results in Table 1 (entry 2). After
each cycle, the enzyme was separated by filtration and subsequently
added to the new reaction vial for the next cycle. At least five con-
secutive cycles were carried out, and STS proved highly reusable in the
addition-cyclization-acetylation reaction without any significant loss of
Scheme 1. Enzyme-catalyzed asymmetric formation of (2R,5R)-5-acetoxy-1,3-
oxathiolan-2-yl)methyl acetate (4a), (2S,5S)-5-acetoxy-1,3-oxathiolan-2-yl)
methyl acetate (4b).
enantioenriched (2R,5R)-1,3-oxathiolane (4a) catalyzed by STS,
whereas the (2S,5S)-isomer (4b) was the main product in the CAL B-
mediated process [17,34]. Furthermore, replacing compound 2 with the
benzoyl-protected aldehyde 5 led to a higher enantiomeric excess of
(2R)-1,3-oxathiolane derivative 5a (82% ee) in the STS-catalyzed DKR
process.
For further use in, e.g., NRTI synthesis, the enantiopure (2R)-isomer
is desired. For this, we envisaged that applying a resolution protocol to
the enantioenriched isomer 5a would result in the enantiomerically
pure compound. Since KR processes focus on resolving racemic mix-
tures involving compounds with single stereocenters, thereby yielding a
maximum of 50% yield, expanding the scope to resolve a racemate
containing two chiral centers would at most provide 25% overall yield.
Therefore, the present resolution design for the enantioenriched isomer
was expected to result in better yields compared to the conventional KR
process for racemic mixtures. As the most frequently used catalysts for
selective hydrolysis of esters [35–38], lipases were chosen in current
case.
First, racemic mixture 5 was applied to hydrolysis in a biphasic
phosphate-buffered saline (PBS)/THF system in the presence of lipases,
and the reaction was followed by chiral HPLC chromatography. Lipase
preparations from Candida rugosa (CRL), Burkholderia cepacia (PS-C I)
and Candida antarctica (CAL B) were evaluated, of which CAL B resulted
in formation of isomer 5a, indicating that selective hydrolysis of either
the benzoyl- or the acetyl-groups occurred. Furthermore, the en-
antioenriched isomer 5a (C-5-configuration unknown) was prepared
[17], and subjected to hydrolysis in the presence of CAL B in a PBS/THF
mixture. Chiral HPLC analysis showed that the enantiomer of 5a was
completely eliminated, giving rise to an excellent ee of > 99%. How-
ever, a diastereomeric ratio (5a/5b) of 4:1 was recorded (Fig. S1). To
identify the configuration of minor isomer 5b, the 5a/5b mixture was
converted to the nucleoside by conventional Vorbrüggen coupling and
deprotection (cf. Scheme 2). By comparing the chiral HPLC results of
the product with lamivudine (reference), it was observed that the
product consisted of lamivudine 1a and ent-lamivudine 1b in a ratio of
4:1. This indicates that isomer 5b possessed the 2S configuration,
thereby not being suitable for lamivudine synthesis. According to the
results in Scheme 1, the configuration of isomer 5a, prepared using the
STS-system, could be deduced to be the (2R,5R)-isomer, in which case
diastereomer 5b would have a (2S,5R) configuration. Therefore, rather
than acting upon the primary benzoyl ester in the 1,3-oxathiolane, CAL
B appeared to selectively hydrolyze the secondary acetyl ester group in
the 5S-position [39–43]. This was expected, in part since secondary
alcohols have been deemed more favorable products in most enzyme-
53