Enzymatic dynamic kinetic resolution of epihalohydrins

Article abstract of DOI:10.1016/j.tetasy.2004.02.009

The haloalcohol dehalogenase from Agrobacterium radiobacter AD1 catalyses the reversible ring closure of vicinal haloalcohols to produce epoxides and halides. In the ring opening of epoxides, nonhalide nucleophiles such as N ₃^- are accepted. The enantioselective irreversible ring opening of an epihalohydrin by N₃^-, combined with racemisation caused by a reversible ring opening by a halide, resulted in an enzymatic dynamic kinetic resolution yielding optically active (S)-1-azido-3-halo-2-propanol. With epichlorohydrin as a substrate, the rate of ring opening by N₃^- was higher than the rate of racemisation, resulting in a mixed kinetic resolution and dynamic kinetic resolution. With epibromohydrin as the substrate, the racemisation rate was higher than the rate of ring opening, resulting in an efficient dynamic kinetic resolution. By optimising the pH of the medium and the concentrations of N ₃^- and Br^-, the product (S)-1-azido-3-bromo-2- propanol could be obtained in 84% yield and 94% ee. An (R)-enantiomer selective ring closure of this bromoalcohol, catalysed by the same enzyme, caused a simultaneously occurring kinetic resolution, yielding when the conversion progressed, an increase in enantiopurity of (S)-1-azido-3-bromo-2-propanol to >99% ee with a yield of 77%. This compound and the ring-closed product glycidyl azide can be used as chiral synthetic building blocks.

Full text of DOI:10.1016/j.tetasy.2004.02.009

Tetrahedron:
Asymmetry
Tetrahedron: Asymmetry 15 (2004) 1095–1102
Enzymatic dynamic kinetic resolution of epihalohydrins
Jeffrey H. Lutje Spelberg,a Lixia Tang,a Richard M. Kelloggb and Dick B. Janssena,*
aDepartment of Biochemistry, Groningen Biomolecular Sciences and Biotechnology Institute, University of Groningen,
Nijenborgh 4, 9747 AG, Groningen, The Netherlands
bDepartment of Organic and Molecular Inorganic Chemistry, University of Groningen, Nijenborgh 4,
9747 AG, Groningen, The Netherlands
Received 5 January 2004; accepted 9 February 2004
Abstract—The haloalcohol dehalogenase from Agrobacterium radiobacter AD1 catalyses the reversible ring closure of vicinal
haloalcohols to produce epoxides and halides. In the ring opening of epoxides, nonhalide nucleophiles such as Nꢀ3 are accepted. The
enantioselective irreversible ring opening of an epihalohydrin by Nꢀ3 , combined with racemisation caused by a reversible ring
opening by a halide, resulted in an enzymatic dynamic kinetic resolution yielding optically active (S)-1-azido-3-halo-2-propanol.
With epichlorohydrin as a substrate, the rate of ring opening by Nꢀ3 was higher than the rate of racemisation, resulting in a mixed
kinetic resolution and dynamic kinetic resolution. With epibromohydrin as the substrate, the racemisation rate was higher than the
rate of ring opening, resulting in an efficient dynamic kinetic resolution. By optimising the pH of the medium and the concentrations
of Nꢀ3 and Brꢀ, the product (S)-1-azido-3-bromo-2-propanol could be obtained in 84% yield and 94% ee. An (R)-enantiomer
selective ring closure of this bromoalcohol, catalysed by the same enzyme, caused a simultaneously occurring kinetic resolution,
yielding when the conversion progressed, an increase in enantiopurity of (S)-1-azido-3-bromo-2-propanol to >99% ee with a yield of
77%. This compound and the ring-closed product glycidyl azide can be used as chiral synthetic building blocks.
Ó 2004 Elsevier Ltd. All rights reserved.
1. Introduction
ase (HheC) catalyses the reversible enantioselective ring
closure of aliphatic and aromatic vicinal haloalcohols,
yielding optically active haloalcohols and epoxides.4
Haloalcohols are direct precursors of epoxides since ring
closure of a haloalcohol yields an epoxide with retention
of configuration. Recently it has been shown that HheC
catalyses the highly enantioselective ring opening of
para-nitrostyrene oxide by Nꢀ3 , CNꢀ and NO2ꢀ.5
Kinetic resolution of racemic mixtures of chiral com-
pounds is a useful method for obtaining enantiomer-
ically pure compounds, but suffers from the drawbacks
that the maximum yield is only 50% of the starting
material and that labourious separation of an optically
active substrate and product is required. This limitation
can be overcome via a dynamic kinetic resolution pro-
cess, in which the slower reacting enantiomer racemises
during the conversion.1 This process can theoretically
result in a single product enantiomer with 100% yield.
Several processes have been described in which the
kinetic resolution and/or in situ racemisation is cataly-
sed by an enzyme.2
Epihalohydrins are widely used as intermediates in the
preparation of various products such as resins, poly-
mers, adhesives and pharmaceuticals.6 The carbon atom
bearing the halide and especially the oxirane carbon
atoms are highly reactive. Generally, the nucleophilic
ring opening of an epihalohydrin is directed towards the
terminal position of the epoxide, yielding the corre-
sponding 3-substituted 1-halo-2-propanol. Ring closure
of this haloalcohol yields hydrogen halide and an
epoxide that can undergo another nucleophilic attack
and thus serve as a useful synthon. Enantiomerically
pure epichlorohydrin was used as a building block for
biologically active compounds such as (S)-atenolol and
(R)-carnitine.6 In the preparation of these compounds,
the epoxide ring of enantiopure epichlorohydrin is
opened by a nucleophile. An attractive nucleophile for
epoxide ring opening is an azide, which in turn affords
Recently, the biocatalytic potential of the epoxide
hydrolase and haloalcohol dehalogenase produced by
the microorganism Agrobacterium radiobacter AD1 has
been investigated. The epoxide hydrolase (EchA) was
used to obtain a variety of enantiomerically pure epox-
ides by kinetic resolution.3 The haloalcohol dehalogen-
* Corresponding author. Tel.: +31-50-3634209; fax: +31-50-3634165;
e-mail: d.b.janssen@chem.rug.nl
0957-4166/$ - see front matter Ó 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetasy.2004.02.009

1096
J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
an azidoalcohol. Optically active azidoalcohols have
been prepared using various biocatalytic methods such
as lipase-catalysed esterification7;8 and microbial reduc-
tion of azidoketones.9 The combination of a lipase-
catalysed enzymatic resolution of azidoalcohols with
ruthenium-catalysed racemisation resulted in a dynamic
kinetic resolution process in which the corresponding
azide substituted acetate esters were obtained in high
yields and high enantiomeric excesses.10 Enantiomer-
ically pure azidoalcohols can be used as precursors for
aminoalcohols and aziridines. Several chiral amino-
alcohols such as 2-amino-1-phenylethanols are biologi-
cally active and have been used as building blocks for
pharmaceutical products or product candidates.
achieve complete conversion of an epihalohydrin in
order to obtain the corresponding 1-azido-3-halo-2-
propanol in high enantiomeric purity and yield.
2. Results and discussion
2.1. Dynamic kinetic resolution with epichlorohydrin
and Nꢀ3
Epichlorohydrin was chosen as the substrate to investi-
gate the dynamic kinetic resolution process. The sepa-
rate enantiomers of this epoxide are commercially
available, making it feasible to determine several aspects
of the reaction, such as the rate of racemisation and ring
opening. The reactions that can occur starting from an
epihalohydrin and Nꢀ3 are depicted in Scheme 2. In
theory, all these reactions can be catalysed by a halo-
alcohol dehalogenase. The ring opening of epihalohy-
drin 1 by Nꢀ3 yields 1-azido-3-halo-2-propanol 2, which
can again be used as a substrate for the enzyme since a
vicinal haloalcohol moiety is present. Ring closure of 2
results in the formation of glycidyl azide 4 and a halide
(Xꢀ). Ring opening of epoxide 4 by Nꢀ3 results in the
formation of the diazido product 5.
A requirement for a dynamic kinetic resolution is rapid
in situ racemisation of the substrate. Previous work
described the enzymatic racemisation of epichloro-
hydrin, which was produced during the enantioselective
conversion of 2,3-dichloro-1-propanol.4 The b-regio-
selective cleavage of epichlorohydrin by a chloride,
yielded 1,3-dichloro-2-propanol. Ring closure of this
prochiral compound produced racemic epichlorohydrin,
causing racemisation to occur (Scheme 1).
Cl-
OH
O
O
The reaction steps shown in Scheme 2 can be influenced
by several factors such as temperature, pH, product
inhibition and the concentrations of organic substrates
and nucleophiles. The pH of the reaction medium
influences the rate of enzyme-catalysed ring-closure and
ring-opening reaction and the position of the equilib-
rium between these reactions. The rate of chemical ring
opening of epoxides by a nucleophilic compound is also
pH-dependent.14;15 The chemical hydrolysis of epichlo-
rohydrin 1a to yield 3-chloro-1,2-propanediol could
trigger the unwanted enzyme-catalysed formation of
glycidol and, in the presence of Nꢀ3 , 3-azido-1,2-pro-
panediol. To determine suitable reaction conditions, the
influence of the pH on all aspects of the dynamic kinetic
resolution process was investigated.
Cl
Cl
Cl
Cl
prochiral
(R)
Cl-
(S)
Scheme 1. Racemisation of epichlorohydrin.
An obvious requirement for a dynamic kinetic resolu-
tion process is an irreversible enantioselective ring
opening of the epoxide. Previous research has shown
that the HheC-catalysed azidolysis of substituted styrene
oxides is highly enantioselective and b-regioselective,
yielding the corresponding (R)-azidoalcohols.11 The
combination of the racemisation of epihalohydrin and
the enantioselective irreversible ring opening by Nꢀ3 ,
yielding one enantiomer of the 1-azido-3-halo-2-propa-
nol, is the concept of the dynamic kinetic resolution that
is reported herein. This 1-azido-3-halo-2-propanol and
glycidyl azide, the ring-closed product thereof, have
been used as synthetic building blocks.12;13
OH
-
X-
N3
O
O
X
N3
X
N3
1a and b
X-
4
2a and b
Four epoxides can be suitable substrates for the above
proposed principle: epifluorohydrin, epichlorohydrin,
epibromohydrin and epiiodohydrin. Alkyl iodides have
a high chemical reactivity, making the occurrence of
various unwanted side-reactions likely. HheC shows no
activity towards vicinal fluoroalcohols. As a conse-
quence, the epoxides epifluorohydrin and epiiodohydrin
were considered to be unfavourable substrates for
investigating the possibility of a dynamic kinetic reso-
lution. HheC catalyses the conversion of a wide variety
of chloroalcohols and bromoalcohols. These haloalco-
hols are generally reasonably stable in a neutral aqueous
environment. Epichlorohydrin and epibromohydrin are
the most suitable substrates for investigating the above-
proposed dynamic kinetic resolution. The goal was to
-
N3
OH
OH
N3
N3
X
X
3a and b
5
Scheme 2. Summary of possible reactions catalysed by a haloalcohol
dehalogenase starting from epihalohydrin and Nꢀ3 (a) X ¼ Cl, (b)
X ¼ Br.
No noteworthy chemical hydrolysis of epichlorohydrin
1a was observed between pH 4.5 and 7.5 (<3% hꢀ1). At
pH 8.5 (6% hꢀ1) and especially pH 9.5 (45% hꢀ1),

J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
1097
chemical hydrolysis was significant. The rate of HheC-
100
80
60
40
20
0
catalysed ring opening of epichlorohydrin (S)-1a by N3ꢀ
displayed an optimum lower than pH 6. The high rate of
the ring-opening reaction is remarkable. The maximum
activity (470 lmol minꢀ1 mgꢀ1) at pH 5 was more than
20-fold higher than the observed rate of ring closure of
1,3-dichloro-2-propanol (at pH 7.5). The rate of HheC-
mediated ring closure of 1,3-dichloro-2-propanol 3a was
highest around pH 8 (Fig. 1). The latter compound was
considered to be an optimal substrate for the HheC.16
The ring opening of epichlorohydrin was more than a
factor of 2000 higher than the earlier described ring
opening of styrene oxide.5
R
0
20
40
60
80
time (min)
500
400
300
200
100
0
25
20
15
10
5
Figure 2. Effect of pH on the chloride-mediated racemisation of (R)-
epichlorohydrin 1a catalysed by HheC. Incubations contained 10 mM
(R)-epichlorohydrin 1a, 1 mM NaCl and 0.7 lM enzyme. Symbols:
(-ꢀ-), pH 4.5; (-j-), pH 5.5; (-O-), pH 6.5; (-.-), pH 7.5; (-s-), pH 8.5;
(-d-), pH 9.5.
cally taking samples from the reaction mixture and
determining the enantiomeric excess of substrate 1a and
product 2a using chiral gas chromatography.
The ring opening of epichlorohydrin by Nꢀ3 , catalysed
by HheC, was selective towards (S)-epichlorohydrin,
yielding (S)-1-azido-3-chloro-2-propanol as the main
product. At pH 4.5 a (nondynamic) kinetic resolution
occurred, resulting in an increase in the enantiomeric
purity of the remaining (R)-1a to higher than 99% ee
(Fig. 3). The enantiomeric purity of product 2a gradu-
ally decreased during the conversion. An E-value of 21
was calculated using conversion and enantiomeric excess
of the remaining substrate 1a.17 At pH 5.5 and higher,
the enantiomeric excess of (R)-1a initially, temporarily
increased to a maximum at around 50% substrate con-
version (Fig. 3A). The maximum transient enantiomeric
purity of (R)-1a decreased from around 90% ee at pH 5.5
to less than 20% ee at pH 8.5 indicating that the kinetic
resolution became more dynamic at higher pH-values.
After the initial increase, the enantiomeric excess of (R)-
1a decreased again, indicating that the relative rate of
racemisation increased.
0
5.5
6.5
7.5
pH
8.5
9.5
Figure 1. Effect of pH on the initial rate of ring opening by Nꢀ3 (-d-)
and ring closure (-j-) reactions catalysed by HheC. Incubations con-
tained: ring opening, 20 mM (S)-epichlorohydrin 1a and 30 mM NaN3;
ring closure, 5 mM 1,3-dichloro-2-propanol 3a.
The rate of racemisation was determined both by the
conversion of epihalohydrin 1 to 1,3-dihalo-2-propanol
3 and the reverse reaction. For an efficient dynamic
kinetic resolution to take place, racemisation of the
substrate 1 must occur faster than the ring opening by
Nꢀ3 . If this is the case, the enantiomeric purity of
remaining epihalohydrin will remain low and that of the
formed 1-azido-3-halo-2-propanol 2 (eep) will be con-
stant throughout the course of the reaction. In such an
ideal dynamic kinetic resolution, the enantiomeric purity
of the desired product is determined by the E-value of
the kinetic resolution (Eq. 1).
The above results indicate that at a pH lower than 8.5
the rate of epoxide racemisation was lower than the rate
of ring opening, resulting in a mixed kinetic and
dynamic kinetic resolution. At pH 8.5, the enantiomeric
purity of the epoxide 1a remained below 20%, indicating
predominantly a dynamic kinetic resolution. However,
at this pH the yield of the desired product (S)-2a was less
than 20% at complete conversion of epichlorohydrin 1a.
This was caused by the ring closure of 1-azido-3-chloro-
2-propanol 2a to glycidyl azide 4, which proceeded
faster at increasing pH values.
ðE ꢀ 1Þ
eep ¼
ꢁ 100%
ð1Þ
ðE þ 1Þ
The enzyme-catalysed racemisation of 10 mM (R)-1a
was measured in the presence of a catalytic amount
(0.1 M equiv) of Clꢀ (Fig. 2). At pH 4.5, no racemisation
occurred, since the rate of ring closure at this pH was
very low. Between pH 6.5 and 8.5 fast racemisation of
(R)-1a occurred, resulting in almost racemic epichloro-
hydrin (ee <10%) within 30 min.
The HheC-catalysed dynamic kinetic resolution, based
on the combination of the above-described ring opening
and the racemisation of 1a, was investigated at various
pH-values. The progress of the reaction between racemic
epichlorohydrin 1a and Nꢀ3 was monitored by periodi-
2.2. Dynamic kinetic resolution with epibromohydrin
and Nꢀ3
The dynamic kinetic resolution with epibromohydrin 1b
was investigated between pH 4.5 and 8.5. Identical to the

1098
J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
100
80
60
40
20
0
100
(A)
(A)
80
60
R
R
40
20
0
0
20
40
60
80
100
0
20
40
60
80
100
% conversion racemic 1a
%conversion racemic 1b
100
80
60
40
20
0
100
80
60
40
20
0
(B)
(B)
S
S
0
20
40
60
80
100
0
20
40
60
80
100
% conversion racemic 1a
%conversion racemic 1b
Figure 3. Effect of pH on enantiomeric purity of substrate (A) and
product (B) during the HheC catalysed ring opening of epichlorohy-
drin 1a by Nꢀ3 . Incubations contained: 20 mM racemic epichlorohydrin
1a and 30 mM NaN3. Symbols: (-d-), pH 4.5; (-s-), pH 5.5; (-.-)
pH 6.5; (-O-), pH 7.5; (-j-), pH 8.5.
Figure 4. Effect of pH on enantiomeric purity of substrate (A) and
product (B) during the HheC catalysed ring opening of epibromohy-
drin 1b by Nꢀ3 . Incubations contained: 20 mM racemic epibromohydrin
1b and 30 mM NaN3. Symbols: (-d-), pH 4.5; (-s-), pH 5.5; (-.-),
pH 6.5; (-O-), pH 7.5; (-j-), pH 8.5.
process described above for epichlorohydrin, the
dynamic character of the resolution improved with
increasing pH value (Fig. 4). Below pH 6.5 the reaction
showed predominantly the character of a kinetic reso-
lution resulting in an increase of the enantiomeric purity
of the remaining substrate (R)-1b and a decrease in the
enantiomeric purity of product (S)-2b during the course
of the reaction.
which also has the added advantage of a lower rate of
chemical conversion.
According to Scheme 1, the broad substrate range of the
enzyme makes the formation of various side products
possible. During the resolution of epibromohydrin 1b at
pH 6.5, 1-azido-3-bromo-2-propanol 2b, glycidyl azide 4
and 1,3-dibromo-2-propanol 3b were initially formed at
an almost equal rate (Fig. 5). The concentration of the
intermediate product 3b, causing racemisation of epi-
bromohydrin, decreased as the reaction proceeded with
only a trace being left at total substrate conversion. The
rate of formation of 5, which is the product of the
reaction of Nꢀ3 with glycidyl azide 4, accelerated towards
the end of the conversion of the substrate. This indicated
that HheC also catalyses this ring-opening reaction.
At pH 6.5 and higher, the enantiomeric excess of (R)-1b
remained below 15%, showing that an almost optimal
dynamic kinetic resolution was obtained. The initial
enantiomeric excess of (S)-2b was around 90% at all pH-
values, indicating that the kinetic resolution occurred
with an E-value of approximately 20.
At pH 6.5 and higher, the enantiomeric purity of the
product 2b remained constant but increased slightly
after 50% substrate conversion. These results indicate
that epibromohydrin 1b is a better substrate than epi-
chlorohydrin 1a for the dynamic kinetic resolution since
at pH 6.5 or higher, the rate of racemisation was higher
than the rate of ring opening. The reaction with epi-
bromohydrin can be carried out at lower pH-values,
In Table 1, the yields of the various products at several
pH-values are shown at the first data point at which the
conversion of 1b was higher than 98%. Between pH 4.5
and 7.5, the azidoalcohol 2b was formed as the major
product. Generally, with increasing pH, the yield of
epoxide 4 increased while that of haloalcohols 2b and 3b
decreased. This can be explained by the high pH opti-

J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
1099
ucts 4 and 5. Increasing the concentration of Brꢀ had no
influence on the initial rate of product 2b formation, but
resulted in an increase of product yields (Fig. 6A and
Table 2, exp. no 1, 5, 6 and 7). These increases in
product yields were accompanied by a decrease in the
formation of the side product 4. Apparently, the higher
yield of 2b at increasing Brꢀ concentrations was caused
by the inhibition of the conversion of 2b to 4.
20
15
10
5
100
(A)
0
80
0
10
20
30
40
time (min)
60
Figure 5. Progress curves of the dynamic kinetic resolution of epi-
bromohydrin 1b and the formation of products catalysed by HheC.
The incubation contained 20 mM epibromohydrin 1b, 30 mM NaN3
and 1.8 lM enzyme at pH 6.5. Symbols: (-d-), epibromohydrin 1b;
(-O-), 1-azido-3-bromo-2-propanol 2b; (-.-), 1,3-dibromo-2-propanol
3b; (-s-), glycidyl azide 4; (-j-), 1,3-diazido-2-propanol 5.
S
40
20
0
0
50
100
150
200
250
time (min)
Table 1. Effect of pH on the formation of the products during the
dynamic kinetic resolution of epibromohydrin 1ba
100
80
60
40
20
0
pH
% Yield
2b
% Ee
(S)-2b
% Yield
4
% Yield
3b
% Yield
5
(B)
4.5
5.5
6.5
7.5
8.5
67
81
72
65
14
65
87
94
95
97
8
6
18
8
7
6
20
19
72
2
6
0
16
12
S
0
a Incubations contained 20 mM epibromohydrin 1b and 30 mM NaN3.
Reaction mixtures were analysed at the first data point at which
>98% conversion of epibromohydrin 1b was observed.
0
20
40
60
80
mum of the ring-closure reaction (around pH 8.0). At
pH 8.5 the formation of 2b was slower than the
sequential ring closure to glycidyl azide 4 resulting in the
formation of the latter as the major product.
time (min)
Figure 6. Effect of NaBr (A) and NaN3 (B) concentration on the
formation of product (S)-2b at pH 6.5. The incubations contained: (A)
20 mM epibromohydrin 1b, 22.5 mM NaN3 and various concentra-
tions NaBr; (B) 20 mM epibromohydrin 1b and various concentrations
NaN3. Symbols: (A) (-d-), 0 mM NaBr; (-s-), 10 mM NaBr; (-.-),
20 mM NaBr; (-O-), 50 mM NaBr and (B) (-d-), 22.5 mM NaN3;
(-s-), 30 mM NaN3; (-.-), 45 mM NaN3; (-O-), 90 mM NaN3. Rect-
angles around data points correspond to entries in Table 2.
2.3. Optimisation of reactant concentrations
The results described above showed that the dynamic
kinetic resolution of epibromohydrin with Nꢀ3 as a
nucleophile is feasible. At pH 6.5 the azidoalcohol (S)-2b
was obtained in 72% yield at a moderate enantiomeric
excess of 94%. The experiments were performedꢀwith
20 mM epibromohydrin 1b and 30 mM Nꢀ3 . No B r was
added since the catalytic amount of Brꢀ, which was
formed by the conversion of 2b to 4, was sufficient
enough to cause racemisation. To test if variation of the
initial concentrations of Brꢀ and N3ꢀ would have any
effect on enantiomeric excess and yield of the desired
product (S)-2b, the concentrations of these anions were
varied. These experiments were performed at pH 6.5
since a pH lower than 6.5 resulted in a less dynamic
character of the kinetic resolution and a pH higher than
6.5 resulted in an increase in the amount of side prod-
The rate of substrate conversion increased with a higher
concentration of Nꢀ3 resulting in an increased transient
yield of (S)-2b (Fig. 6B). Complete conversion of 1b
(20 mL, 20 mM) was achieved in 60 min with 22.5 mM
Nꢀ3 and in less than 20 min with 90 mM Nꢀ3 , using
1.8 lM enzyme. However, after the initial rapid accu-
mulation of 2b, its concentration slowly decreased. The
combination of both a higher Nꢀ3 and higher Brꢀ con-
centration did not yield an increase in yield of 2b (Table
2, exp. no 8 and 9). Varying the Brꢀ or N3ꢀ concentra-
tions did not have an effect on the enantioselectivity of
the ring opening since all conversions yielded (S)-2b in
94% to 95% ee at total conversion of the substrate.

1100
J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
Table 2. Effect of NaBr and NaN3 concentration on product formation during the dynamic kinetic resolution of epibromohydrin 1ba
Exp. no
Concd NaBr (mM) Concd NaN3 (mM)
% Yield 2b
% Ee (S)-2b
% Yield 4
% Yield 3b
% Yield 5
1
2
3
4
5
6
7
8
9
0
0
22.5
30
65
72
74
78
72
75
84
84
84
94
94
95
94
95
95
95
94
94
26
20
17
12
19
17
10
9
3
2
1
1
2
3
3
3
3
4
6
8
9
5
5
5
4
5
0
45
90
0
10
20
50
50
50
22.5
22.5
22.5
30
45
8
a Incubations contained 20 mM epibromohydrin 1b and 1.8 lM enzyme. Reaction mixtures were analysed at the first datapoint at which >98%
conversion of epibromohydrin 1b was observed.
2.4. Increased enantiomeric excess by a sequential kinetic
resolution
conversions of 2a or 2b were performed in the presence
of an equimolar amount of Nꢀ3 the (R)-enantiomers
were converted preferentially resulting in kinetic reso-
lutions with E-values of 15 and 8, respectively. Due to
this sequential kinetic resolution, (S)-1-azido-3-bromo-
2-propanol 2b can be obtained enantiomerically pure if
the reaction is allowed to proceed after the dynamic
kinetic resolution of epibromohydrin is completed
(Table 3). With the addition of an excess of Nꢀ3 and no
Brꢀ (exp. no 3 and 4), a prolonged incubation resulted in
a disadvantageous decrease in the yield of (S)-1-azido-3-
bromo-2-propanol 2b. This was due to the HheC-ca-
talysed ring opening of glycidyl azide 4 by Nꢀ3 yielding
1,3-diazido-2-propanol 5. This shifted the equilibrium
between 2b and 4 in the direction of 4, and increased the
degree of conversion of 2b.
During the conversion of epibromohydrin 1b, the
enantiomeric purity of (S)-2b remained almost constant
during the first 50% of the conversion and steadily in-
creased during the second 50%. In a normal dynamic
kinetic resolution process, the enantiomeric purity of the
product remained constant throughout the course of the
reaction. The increase in enantiomeric purity of (S)-2b
can be explained by the occurrence of a sequential
kinetic resolution, the (R)-selective ring closure of 1-
azido-3-bromo-2-propanol 2b yielding glycidyl azide 4.
This reaction caused a decrease in the yield of 2b, but an
increase in enantiomeric excess. However, when racemic
2b was subjected to HheC in a separate experiment, only
20% conversion occurred. This can be explained by the
position of the equilibrium between 2b and 4, which is
apparently positioned towards 2b. During a dynamic
kinetic resolution, HheC also catalyses the ring opening
of glycidyl azide 4 by N3ꢀ and thereby draws the con-
version of (R)-2b to completion. When the enzymatic
The optimal reaction conditions (pH 6.5, 50 mM Brꢀ
and 30 mM Nꢀ3 ), combined with a prolonged incubation
after complete conversion of substrate 1b, resulted in
(S)-1-azido-3-bromo-2-propanol 2b in higher than 99%
ee with a yield of 77%. The overall reaction scheme of
Table 3. Increase of enantiomeric purity of (S)-2b due to the sequential kinetic resolutiona
Exp. no
Concd NaBr (mM) Concd NaN3(mM)
% Yield of 2b at 94.5%
( 0.5%) ee of (S)-2b
% Yield of 2b at 97%
( 0.5%) ee of (S)-2b
% Yield of 2b at
99% ee of (S)-2b
3
4
8
9
0
0
45
90
30
45
74
78
84
84
66
68
80
77
58
51
77
71
50
50
a Incubations contained 20 mM epibromohydrin 1b and 1.8 lM enzyme.
slow
OH
O
O
fast
Br
N3
N3
Br
-
-
-
Br
Br
Br
-
N3
N3
(S)-1b
(S)-4
(S)-2b
OH
OH
Kinetic
resolution
Kinetic
resolution
N3
N3
Br
Br
racemisation
5
3b
-
N3
-
OH
O
fast
O
slow
Br
N3
Br
N3
-
-
N3
Br
(R)-1b
(R)-4
(R)-2b
Scheme 3. Overall reaction scheme of the combined dynamic kinetic resolution process of epibromohydrin 1b and subsequent kinetic resolution of
the formed product 1-azido-3-bromo-2-propanol. All 12 reaction steps are catalysed by the haloalcohol dehalogenase HheC.

J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
1101
the dynamic kinetic resolution with the sequential
kinetic resolution is shown in Scheme 3.
Brꢀ and Nꢀ3 but also CNꢀ and NOꢀ2 .5 Any compound
that contains an epihalohydrin as a substructure, forms
a prochiral intermediate after ring opening by a halide
and can be accepted by the enzyme, is expected to be a
substrate for the described dynamic kinetic resolution
process.
3. Discussion and conclusion
An efficient dynamic kinetic resolution was developed
for the conversion of racemic epibromohydrin to enan-
tiomerically pure (S)-1-azido-3-bromo-2-propanol 2b.
The reactions depicted in Scheme 3 occurred simulta-
neously and were all catalysed by the haloalcohol
dehalogenase HheC. With epibromohydrin 1b at pH 6.5
only a slight transient enantiomeric excess of the
remaining substrate was observed during the reaction,
indicating that the rate of racemisation was higher than
the rate of ring opening. Optimisation of the reaction
conditions resulted in 1-azido-3-bromo-2-propanol 2b in
84% yield and 94% ee. Due to an in situ sequential
kinetic resolution by ring closure of 2b catalysed by the
same enzyme, the product can be obtained enantio-
merically pure (ee >99%).
4. Experimental
4.1. General
The enantiomeric excesses (% eeÕs) and the yields of all
the aliphatic compounds were determined with a Hew-
lett–Packard 5890 gas chromatograph equipped with a
FID-detector, using a Chiraldex G-TA capillary column
(col I, 50 m, Astec) or a Chiraldex A-TA capillary col-
umn (col II, 25 m, Astec) all of 0.25 mm inside diameter.
NMR-spectra were recorded in CDCl3 or DMSO-d6.
Epichlorohydrin 1a, (R)-1a, (S)-1a and epibromohydrin
1b were purchased from Aldrich. The haloalcohol
dehalogenase from A. radiobacter AD1 (HheC) was
overexpressed and purified as described before.3;4
This sequential kinetic resolution did not take place if
the product, glycidyl azide 4 was not converted to the
diazido product 5. The HheC catalysed conversion of 4
to 5 removes the product and draws the equilibrium
between 2b and 4 towards the latter. This explains that
the optimal concentration of Nꢀ3 is between 1.2- and 1.5-
fold higher than the concentration of epibromohydrin.
If an equimolar amount of azide were to be used, no N3ꢀ
would be left to react with glycidyl azide 4. However if a
too large excess of Nꢀ3 was used (Table 3, exp. no 3 and
4) the rate of conversion of 2b became very fast,
resulting in a lower yield of 2b and the formation of 5 as
the major side product. A considerable loss in yield of 2b
was observed since the E-value of this second kinetic
resolution is only 8.
4.2. General procedure for biocatalytic conversions
Stock solutions of NaCl, NaBr and NaN3 were freshly
prepared. The substrate was dissolved in 20 mL buffer
and incubated at 22 °C. The proper volumes of stock
solutions were added to the substrate solution. The
reaction was started by the addition of the enzyme. The
reaction was monitored by periodically taking 0.5 mL
samples from the closed reaction vessel. The samples
were extracted with 2 mL diethyl ether containing 1-
chlorohexane as an internal standard. Prior to analysis
by chiral GC, the samples were dried by passing them
through a small column containing MgSO4. The fol-
lowing initial substrate concentrations were used: ep-
oxides 1a, (S)-1a and 1b, 20 mM; (R)-1a and 2a, 10 mM;
1,3-dichloro-2-propanol 3a, 5 mM. The effect of pH on
the conversions was determined using the following
buffers: pH 4.5, 100 mM sodium acetate; pH 5.5,
100 mM sodium citrate; pH 6.5, 100 mM sodium phos-
phate; pH 7.5 and pH 8.5, 100 mM Tris-sulfate; pH 9.5,
100 mM glycine–NaOH.
Jacobsen and Schaus have reported a similar dynamic
kinetic resolution process. In this reaction, the enantio-
selective ring opening (krel >100) of epichlorohydrin by
TMSN3 was catalyzed by a (Salen)Cr(III)N3 complex
resulting in almost enantiomerically pure 3-azido-1-
chloro-2-trimethylsiloxypropane (ee >97%) in a 76%
yield.13 A sequential kinetic resolution was not de-
scribed. Similar to the results described above for epi-
chlorohydrin, the limiting factor in this process was the
slow racemisation of epoxide (R)-1a. This problem
could be circumvented by a controlled addition of azide,
thereby increasing the relative rate of racemisation.
4.3. Synthesis of racemic reference compounds
Compounds 2a and 2b were prepared from the corre-
sponding epoxides, 1a and 1b.15 The epoxide (2.5 mmol)
was dissolved in an aqueous solution of sodium azide
(13.0 mmol in 4.0 mL), 2.3 mL acetic acid then added
and the solution stirred for 5 h at 30 °C. The solution
was extracted with diethyl ether (3 ꢂ 4 mL). The
combined organic phase was washed five times with
5 mL portions of sodium phosphate buffer (50 mM,
pH 6.5). The organic phase was dried and the diethyl
ether removed on a rotary evaporator. NMR data:
This is the first description of a dynamic kinetic reso-
lution process using a haloalcohol dehalogenase. Al-
though the overall reaction scheme was somewhat
complicated, since various reactions occurred simulta-
neously (Scheme 3), choosing the optimal conditions
allowed an efficient conversion of racemic epi-
bromohydrin to enantiomerically pure (S)-1-azido-3-
bromo-2-propanol 2b. In this research, the substrate was
limited to one type of epoxide and one nucleophile (Nꢀ3 ).
The substrate range of this reaction is not limited to
epihalohydrins and Nꢀ3 . A recent evaluation of nucleo-
philes showed that the enzyme accepts as well as Clꢀ,
1
Compound 2a (DMSO-d6): H NMR d 3.32 (m, 2H),
3.60 (m, 2H), 3.85 (m, 1H), 5.7 (br s, 1H); 13C NMR d
43.9 (C-1), 50.6 (C-3), 67.1 (C-2). Compound 2b

1102
J. H. Lutje Spelberg et al. / Tetrahedron: Asymmetry 15 (2004) 1095–1102
(DMSO-d6): 1H NMR d 3.34 (m, 2H), 3.47 (m, 2H) 3.86
(m, 1H), 5.8 (d, 1H); 13C NMR d 33.5 (C-1), 51.3 (C-3),
66.8 (C-2). Compound 4 was prepared from racemic 2b.
The haloalcohol was dissolved in a diluted aqueous
NaOH solution and stirred for 0.2 h at 22 °C. The
solution was extracted with diethyl ether. The organic
phase was dried and the diethyl ether removed on a
rotary evaporator yielding compound 4 as an oil. NMR
(QLK3-2000-00426), by STW (GCH.3877) and by
Innovation Oriented Research Program (IOP) on
Catalysis (No 94007a) of the Dutch Ministry of Eco-
nomic Affairs.
1
References and notes
data: Compound 4 (CDCl3): H NMR d 2.68 (dd, 1H),
2.81 (dd, 2H), 3.16 (m, 1H), 3.26 (dd, 1H), 3.53 (dd, 1H);
13C NMR d 42.2, 47.7, 49.7. Compound 5 was prepared
by the ring opening of 4 with an azide, according to the
procedure described above for 2a and 2b. Since the
diazido compound 5 is an unstable and explosive sub-
stance, it was not isolated in a pure form, but a solution
containing 5 was analysed using gas chromatography.
1. (a) Ebbers, E. J.; Ariaans, G. J. A.; Houbiers, J. P. M.;
Bruggink, A.; Zwanenburg, B. Tetrahedron 1997, 28, 9417;
(b) Noyori, R.; Tokunaga, M.; Kitamura, M. Bull. Chem.
Soc. Jpn. 1995, 68, 36.
2. Stecher, H.; Faber, K. Synthesis 1997, 29, 1.
3. Lutje Spelberg, J. H.; Rink, R.; Kellogg, R. M.; Janssen,
D. B. Tetrahedron: Asymmetry 1998, 9, 459.
4. Lutje Spelberg, J. H.; van Hylckama Vlieg, J. E. T.;
Bosma, T.; Kellogg, R. M.; Janssen, D. B. Tetrahedron:
Asymmetry 1999, 10, 2863.
4.4. Absolute configuration and chiral analysis
5. Lutje Spelberg, J. H.; Tang, L.; van Gelder, M.; Kellogg,
R. M.; Janssen, D. B. Tetrahedron: Asymmetry 2002, 13,
1083.
6. Kasai, N.; Suzuki, T.; Furukawa, Y. J. Mol. Catal. B
1998, 4, 237.
7. Foelsche, E.; Hickel, A.; Honig, H.; Seufer-Wassertal, P.
J. Org. Chem. 1990, 55, 1749.
8. Pchelka, B. K.; Loupy, A.; Plenkiewicz, J.; Blanco, L.
Tetrahedron: Asymmetry 2000, 11, 2719.
The absolute configurations of 1-azido-3-chloro-2-pro-
panol 2a and glycidyl azide 4 were determined by
co-injection with the enantiomerically pure compounds.
Enantiomerically pure (S)-2a was prepared from (S)-1a
as described above. Ring closure of (S)-2a under basic
conditions yielded (S)-4. During the dynamic kinetic
resolution of epibromohydrin 1a and epibromohydrin
2a, (S)-4 was formed as major product. From this result
we concluded that the enantiopreference towards epi-
bromohydrin must be identical to that of epichlorohy-
drin. The retention times of the analysed compounds are
as follows: Col I: temperature program: 5 min at 80 °C,
10 °C/min to 170 °C, 1 min at 170 °C. (S)-1a, 3.1 min;
(R)-1a, 3.3 min; (S)-1b, 4.4 min; (R)-1b, 4.9 min; (R)-2a,
10.8 min; (S)-2a, 11.0 min. Col I: temperature program:
100 °C; (R)-2b, 20.0 min; (S)-2b, 20.9 min. Col II: tem-
perature program: 65 °C; (S)-4, 22.3 min; (R)-4,
22.6 min.
9. Besse, P.; Veschambre, H.; Dickman, M.; Chenevert, R.
J. Org. Chem. 1994, 59, 8288.
ꢀ
€
10. Pamies, O.; Backvall, J. E. J. Org. Chem. 2001, 66, 4022.
11. Lutje Spelberg, J. H.; Tang, L.; van Hylckama Vlieg, J. E.
T.; Kellogg, R. M.; Janssen, D. B. Org. Lett. 2001, 3, 41.
12. Brittelli, D. R. US Patent no 4,877,892, 1987.
13. Schaus, S. E.; Jacobsen, E. N. Tetrahedron Lett. 1996, 37,
7937.
14. Blumenstein, J. J.; Ukachukwu, V. C.; Mohan, R. S.;
Whalen, D. L. J. Org. Chem. 1993, 58, 924.
15. Fringuelli, F.; Piermatti, O.; Pizzo, F.; Vaccaro, L. J. Org.
Chem. 1999, 64, 6094.
16. van Hylckama Vlieg, J. E. T.; Tang, L.; Lutje Spelberg,
J. H.; Smilda, T.; Poelarends, G. J.; Bosma, T.; van
Merode, A.; Fraaije, M. W.; Janssen, D. B. J. Bacteriol.
2001, 183, 5058.
Acknowledgements
This research was financially supported by the Quality
of Life Programme of the European Commission
17. Chen, C. S.; Fujimoto, Y.; Girdaukas, G.; Sih, C. J. J. Am.
Chem. Soc. 1982, 104, 7294.