dichloro-1-propanol (E > 100) and (S)-2-chloro-1-phenyl-
ethanol (E ) 73) by means of kinetic resolution of halo-
hydrins. Here we show that azide can be used as a nucleo-
phile in the reverse reaction. Furthermore, this enzyme-
catalyzed azidolysis is both highly enantioselective and
â-regioselective toward (substituted) styrene oxides.
Using the chromogenic substrate p-nitrostyrene oxide (R)-
substrate concentrations higher than 1 mM the curve showed
first order kinetics, indicating a considerably higher K value
of the substrate. The initial activity of (S)-1a was too low to
be measured (<1 mU mg ). The epoxides 2a and 3a were
also converted with high enantioselectivity leading to opti-
cally pure epoxides (ee > 99%) and the almost optically
(98% ee) pure azido alcohols 2b and 3b. A complication is
chemical azidolysis leading mainly to the azido alcohols 2c
m
9
-
1
1a, we determined the equilibrium constant (Keq) of the
reversible reaction catalyzed by the halohydrin dehalogenase
with a variety of halides and nucleophiles. The nucleophilic
ring opening of p-nitrostyrene oxide leads to a decrease in
the extinction coefficient at 310 nm, which made it possible
to monitor the enzymatic ring opening online by the decrease
of absorbance. Using chiral HPLC, it was established that
the enzymatic ring opening is highly regioselective toward
the â-carbon atom. With chloride and bromide the equilib-
rium is predominantly on the side of the epoxide (Table 1).
and 3c. The intrinsic E-value (Table 2, ee
enzymatic resolutions based on the ee’s of a and b was higher
than 200, but the apparent E-value (conv, ee ), which takes
the unwanted chemical conversion into account, was con-
siderably lower.
s p
, ee ) of the
s
The high â-regioselectivity of the enzyme-catalyzed reac-
tion is striking and is, as mentioned, opposite to the observed
selectivity in the non-catalyzed azide ion ring opening. For
the kinetic resolution of 2a, â-attack took place to the extent
of 89% at 55% conversion compared to chemical azidolysis,
which involves only 3% reaction at the â-position. During
the enzymatic reaction the apparent selectivity percentage
is continuously lowered as a result of the chemical side
reaction. When the reaction is performed with an excess of
enzyme, the initial â-regioselectivity for all substrates was
higher than 98%, indicating an almost absolute opposite
regioselectivity compared to that of the chemical reaction.
Would the chemical azidolysis leading to product c derail
a practical application of this halohydrin dehalogenase as a
biocatalyst for large scale conversions? To determine the
optimal conditions for a large scale synthesis, the influence
of increasing azide concentrations on the conversion of 1a
to 1b was investigated by monitoring the initial activities
(concentration 1a, 250 µM) on-line at 310 nm. The apparent
Table 1. Equilibrium Constants of the Reaction of Various
Sodium Salts with (R)-1a by the Halohydrin Dehalogenase from
Agrobacterium radiobacter AD1
sodium salt
Keq (mM)a
NaBr
NaCl
NaF
480
40
no reaction
<0.03b
NaN3
a
The equilibrium constant is defined as:
[
NaX] ‚[(R)-1a]
Keq
)
[(R)-alcohol]
b
The exact equilibrium constant was too low to be determined using the
chromogenic substrate.
K
m
value for azide was determined to be 0.2 mM, from which
we may expect that above a concentration of approximately
.5 mM the enzymatic activity will become independent of
0
This limits a practical application of this reaction since a
large excess of halide would be necessary to favor the
formation of the halohydrin over the epoxide. With azide as
nucleophile, the equilibrium lies on the side of the product
of ring opening, although the exact equilibrium constant was
too small to be determined accurately using the chromogenic
substrate. This implies that a total enzymatic conversion of
the epoxide to the azido alcohol is possible with only a small
excess of sodium azide.
the azide concentration. If in a kinetic resolution, azide is
present in a higher concentration, only an increase in the
disadvantageous chemical azidolysis will be observed. To
circumvent the problem of excessive volumes owing to the
low solubility (3 mM) of 1a we added the substrate (0.47 g)
as a second solid phase to 60 mL of buffer (MOPS, pH )
7.0) containing 29 mg of purified enzyme. The amount of
substrate was 7.8 g/L, which is 17-fold the solubility. Sodium
azide (0.6 molar equiv) was slowly added over 24 h keeping
the azide concentration around 0.5-1 mM.
After isolation, the yields and ee’s of the compounds
present in the mixture were determined using chiral HPLC.
The isolated product consisted of (S)-1a in 46% yield (98%
ee) and (R)-1b in 47% yield (97% ee). Compound 1c, the
result of the nonenzymatic reaction, formed a total of 4% of
the reaction mixture, and the product of chemical hydrolysis
of the epoxide, p-nitrophenylethanediol was present in 3%
yield.
The recombinant halohydrin dehalogenase from Agrobac-
terium radiobacter AD1 catalyzed the highly enantioselective
and regioselective azidolysis of substituted styrene oxides
(Table 2). The enzyme converted p-nitrostyrene oxide 1a to
the corresponding azido alcohol 1b with high regioselectivity
and enantioselectivity. The remaining (S)-1a was obtained
with an ee > 99% and the azido alcohol (R)-1b was formed
with an ee of 96%. The corresponding E-value was calculated
to be higher than 200 either from the ee's of the epoxide
and the azido alcohol or by using conversion and ee of the
epoxide.10 The initial activity with (R)-1a at a substrate
In conclusion, we describe here for the first time a highly
enantioselective (E > 200) and â-regioselective azidolysis
of (substituted) styrene oxides using the halohydrin de-
-
1
concentration of 1 mM was 180 mU mg . Even at initial
(
9) Lutje Spelberg, J. H.; van Hylckama Vlieg, J. E. T.; Bosma, T.;
(10) Straathof, A. J. J.; Jongejan, J. A. Enzyme Microb. Technol. 1997,
21, 559.
Kellogg, R. M.; Janssen, D. B. Tetrahedron: Asymmetry 1999, 10, 2863.
42
Org. Lett., Vol. 3, No. 1, 2001