Enzymatic transformations. Part 55: Highly productive epoxide hydrolase catalysed resolution of an azole antifungal key synthon

Article abstract of DOI:10.1016/j.tet.2003.10.119

A highly productive bioprocess for the preparation of enantiopure azole antifungal chirons is described. These are key building blocks for the synthesis of new triazole drug derivatives known to display valuable activity against such infections as for ins

Full text of DOI:10.1016/j.tet.2003.10.119

Tetrahedron 60 (2004) 601–605
Enzymatic transformations. Part 55: Highly productive epoxide
hydrolase catalysed resolution of an azole antifungal key synthon
*
Nicolas Monfort, Alain Archelas and Roland Furstoss
´ ´ ´ ´
Groupe Biocatalyse et Chimie Fine, Faculte des Sciences de Luminy, UMR CNRS 6111, Universite de la Mediterranee, Case 901,
163 avenue de Luminy, 13288 Marseille Cedex 9, France
Received 8 August 2003; revised 9 October 2003; accepted 24 October 2003
Abstract—A highly productive bioprocess for the preparation of enantiopure azole antifungal chirons is described. These are key building
blocks for the synthesis of new triazole drug derivatives known to display valuable activity against such infections as for instance
fluconazole-resistant oro-oesophageal candidiasis. Using commercially available recombinant Aspergillus niger epoxide hydrolase under
optimised experimental conditions, the hydrolytic kinetic resolution of 1-chloro-2-(2,4-difluorophenyl)-2,3-epoxypropane was performed in
plain water, at room temperature, using a two-phase reactor. This methodology allowed the process to be run at a substrate concentration as
high as 500 g/L (i.e., 2.5 M) and afforded the (unreacted) epoxide and the corresponding vicinal diol, both in nearly enantiopure form and
quantitative yield.
q 2003 Elsevier Ltd. All rights reserved.
1. Introduction
convergent process, i.e., affording 100% ee, 100% yield
starting from a racemic substrate) are important advantages
which are surely worth considering.5
The preparation of enantio-enriched epoxides or of their
corresponding vicinal diols, two types of chiral intermedi-
ates which offer high chemical versatility, has been
considered for years as one of the most significant goals
for asymmetric synthesis. Identification of new method-
ologies for access to these targets has been an area of active
effort for several decades. Diverse conventional chemistry
approaches have been described, the best ones based on the
use of different types of transition-metal based catalysts.1,2
We (and others) have also explored, over the last decades,
the possibility offered by the use of appropriate enzymes. In
particular, we have described that microbial epoxide
hydrolases allowed efficient biocatalysed hydrolytic kinetic
resolution (BHKR) of various epoxides.3 Although this
approach surprisingly seems to be sometimes ignored, for
some unclear cultural reasons, by organic chemists, there is
no doubt that this methodology could provide interesting
complementarity4 as well as valuable practical advantages
to the conventional chemistry approach, particularly for
industrial application. For example, the ‘green chemistry’
aspect, the very high turnover numbers generally observed,
and the fact that some of these enzymes offer complemen-
tary regioselectivities (thus allowing to set up an enantio-
With the exception of lipases and baker’s yeast, nowadays
commonly accepted as ‘classical reagents’, the use of
biocatalyts is very often considered to be hampered by
several practical drawbacks. These are for instance: (a) the
fact that the appropriate enzyme is rarely commercially
available (b) the (possible) limited substrate specificity of a
given enzyme (c) the low substrate concentration usable
owing to the (generally) poor solubility of organic substrates
in water (thus leading to low volumetric productivity of the
bioreactor) or, else (d) the reaction limitation linked to
substrate and/or product inhibition. Although, depending on
the substrate, these limitations should indeed not be
neglected, they may in many cases be circumvented by
using appropriate methodologies. We describe herein our
results, focused on the BHKR of epoxide 1, where these
drawbacks could be efficiently overcome. Thus, we show
that using the commercially available recombinant Asper-
gillus niger epoxide hydrolase† it is possible to set up a
highly productive bioprocess for the preparation of
enantiopure epoxide 1 as well as of the corresponding
chloro-diol 2. Both these chirons are key building blocks for
the synthesis of new triazole drug derivatives, like for
instance D0870, known to display valuable activity against
such infections—as for instance fluconazole-resistant oro-
oesophageal candidiasis—often encountered in HIV-
infected (and other immunocompromised) patients6,7
(Scheme 1).
Keywords: Biocatalysis; Enzyme resolution; Epoxide; Kinetic resolution;
Enantioselective; Hydrolysis.
*
Corresponding author. Tel.: þ33-4-91-82-91-55; fax: þ33-4-91-82-91-
45; e-mail address: furstoss@luminy.univ-mrs.fr
This enzyme is presently commercially available from FLUKA (ref.
71832).
†
0040–4020/$ - see front matter q 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tet.2003.10.119

602
N. Monfort et al. / Tetrahedron 60 (2004) 601–605
which would translate into 15 kg/m3 in case of an industrial
application. Therefore, we have explored the possibility to
perform this resolution using plain water instead of
phosphate buffer. Indeed, we had shown previously—
using rather water soluble pyridine oxirans—that to the
contrary of many enzymes, the A. niger epoxide hydrolase
retained nearly its total activity in plain water. It remained to
be checked whether this was also applicable to the rather
water insoluble epoxide 1. The results of a comparative
study between the BHKR of rac-1, using a phosphate buffer
solution on one hand or plain water on the other hand, are
reported on Figure 1. Interestingly no significant difference
was observed between these two experiments, indicating
that a salt free process could also be used in this case.
Scheme 1.
2. Results
We have recently described our preliminary results aimed at
offering an alternative pathway to the synthesis of the
enantiopure epoxide (S)-1 and of the corresponding diol
(R)-2.8 This involved the BHKR of rac-1, performed using
the commercially available (recombinant) A. niger epoxide
hydrolase. We have shown that, when conducted at a 2 mM
(0.4 g/L) concentration of rac-1 in a 100 mM phosphate
buffer solution (pH 7) containing 10% DMSO, this could
efficiently lead to enantiopure (S)-1 and to the correspond-
ing chloro-diol (R)-2, also obtained in nearly enantiopure
form (ees.98%) (Scheme 2). The observed E value of this
resolution was higher than 200, and the a(S)/b(S) and a(R)/
b(R) regioselectivity factors, determined for each one of the
two enantiomers of 1, were calculated to be 5/95 and 1/99,
respectively, indicating that both enantiomers were essen-
tially attacked at the less substituted carbon atom.
2.1.2. Optimisation of the substrate concentration. As far
as the practical/industrial applicability of such a process is
concerned, one major bottleneck obviously could be the
usable substrate concentration. We have determined that the
aqueous solubility of 1 is about 0.7 g/L, precluding any high
concentration application using a monophasic aqueous
medium. Addition of a water miscible solvent, i.e., DMSO
as described in our preliminary results, obviously would
allow to slightly increase the aqueous concentration, but
only to a very limited extend keeping in mind that this
solvent is only tolerated by the enzyme at less than about
10% concentration (at this concentration the residual
activity is about 70%). Therefore, we decided to explore
the possibility to set up a ‘high concentration’ two-phase
reactor, the substrate playing itself the role of the organic
phase. We have already described the possibility to
efficiently operate such a two-phase BHKR reaction in the
case of para-bromo-a-methylstyrene oxide9 or para-
chlorostyrene oxide,5 for which a substrate concentration
as high as 80 and 300 g/L, respectively, could be reached. It
should be noted that these two reactions were carried out at
low temperature (4 and 0 8C respectively) in order to limit
the spontaneous chemical hydrolysis of these substrates. In
the present case, epoxide 1 appeared to be very stable
towards chemical hydrolysis (less than 0.1% per hour), thus
allowing the preparative experiments to be carried out at
nearly room temperature (27 8C).
These interesting results encouraged us to pursue this study
in order to set up an improved preparative scale process. In
particular, we wanted to address some of the drawbacks
cited above, in order to demonstrate the practical applica-
bility of such a biocatalytic procedure. To reach this goal we
explored the possibility (a) to simplify the experimental
protocol (b) to increase the substrate concentration (c) to
optimise the enzyme/substrate ratio (d) to perform this
BHKR at a several-gram scale using the thus defined
optimal experimental conditions.
2.1. Simplification of the experimental protocol
The possibility to increase the substrate concentration was
explored in the range of 2 mM–3.6 M concentration. The
reaction was performed using plain water at a constant
enzyme/substrate ratio of 93 U/mmol of rac-1, and was
followed by monitoring the ee of the remaining epoxide at
20, 40 and 60 min reaction time periods (Fig. 2). Interest-
ingly, it thus appeared that (a) the reaction did very nicely
proceed in such a two-phase system (b) no mass transfer
limit was apparent (c) the BHKR reaction was very efficient
2.1.1. Use of plain water instead of phosphate buffer. As a
general trend, one of the important issues for industrial
application is, whatever the type of reactant or catalyst
(chemical or enzymatic) involved, the set up of so-called
‘salt free’ processes. This obviously allows to simplify, and
therefore to lower the cost, of the subsequent downstream
processing. Our previously described procedure—which
involved in the use of a 100 mM (pH 7) phosphate buffer—
in fact implied an amount of 15 g/L of phosphate salts,
Scheme 2.

N. Monfort et al. / Tetrahedron 60 (2004) 601–605
603
Figure 3. Influence of the substrate concentration on the ee of the residual
epoxide-1 as a function of time. A %ee of (S)-1 at 0.4 g/L (2 mM); B %ee
of (S)-1 at 500 g/L (2.5 M).
therefore have explored the possibility to decrease the
enzyme/substrate ratio, and the results we have obtained are
shown in Figure 4. It appeared that the resolution process
could be carried out even when the enzyme/substrate ratio
was decreased from 100 to 25 U/mmol of rac-1, the reaction
time being then increased from 60 to 300 min for reaching
the 50% conversion ratio (ee.99%).
Figure 1. Comparative study between the BHKR of rac-1 using a
phosphate buffer solution or plain water. Full line: phosphate buffer; dashed
line: plain water; squares: ee of 1; triangles: conversion ratio c.
up to a substrate concentration as high as 2.5 M (500 g/L)
(d) above this value, the reaction rate slowed down
noticeably (e) surprisingly—at an identical reaction
time—the ee of the remaining epoxide appeared to reach
a higher value for experiments conducted at higher substrate
concentration. Thus, a comparative study, carried out in a
homogeneous system (at 2 mM, i.e., 0.4 g/L) on one hand,
and using a two-phase system (2.5 M, i.e., 500 g/L) at the
other hand, clearly showed that at high substrate concen-
tration the rate of reaction was faster by a factor of about 1.5
than at low concentration. In these cases, the ee of the
remaining epoxide reached nearly 100% after only 40 min
in the biphasic system, whereas 60 min were necessary to
reach the same value for the monophasic system (Fig. 3). A
more detailed study of this interesting effector effect, which
will be published elsewhere, showed that this activation was
due to the presence of diol 2 formed during biohydrolysis.
2.2. Gram scale reactor
Using these optimised experimental conditions, a gram
scale preparative experiment was conducted at a 2.5 M
(500 g/L) concentration in a two-liquid-phase reactor. This
was performed using 2 g of rac-1 in the presence of 275 U
of crude recombinant A. niger epoxide hydrolase (corre-
sponding to 25 mg of crude powder at 11 U/mg) i.e., an
enzyme/substrate ratio of 28, dissolved in 1.8 mL plain
water and 0.2 mL DMSO. This was stirred for 2 h at 27 8C,
and the reaction was quenched by addition of 1 mL of
acetonitrile to the reaction medium. Unreacted 1 was
extracted with pentane (3£15 mL), whereas diol 2 was
further extracted with ethyl acetate (3£15 mL). Both
products were purified by flash chromatography, then by
bulb-to-bulb distillation, which led to 0.83 g of (S)-1 [99.9%
ee; 41.5% yield; [a]2D5¼þ46.5 (c 1; THF)], and to 0.95 g of
diol (R)-2 [94.5% ee; 43.5% yield; [a]2D5¼þ3.4 (c 1; THF)].
This experiment allowed us to determine (a) the turnover
frequency (TOF) of the enzyme against 1, which was
2.1.3. Optimisation of the enzyme over substrate ratio.
One other important aspect of the cost effectiveness of such
a bioprocess is obviously the amount of enzyme necessary
to run the reaction, and therefore, the enzyme/substrate ratio
is one of the important factors to be optimised. This is
particularly true in the case of epoxide resolution, racemic
epoxides being generally relatively cheap compounds. We
Figure 4. Influence of the enzyme/substrate ratio on the ee of the residual
epoxide-1 as a function of time at an initial concentration of 500 g/L. The
enzymatic powder used throughout this study showed an activity of 9.6 U/
mg. Enzyme/substrate ratio (U/mmol): X, 100; W, 50; B, 37.5; A, 25.
Figure 2. Substrate concentration increase. The enzyme/substrate ratio was
kept constant at 93 U/mmol of rac-1 throughout the scale up. Reaction
time: white box: 20 min; grey box: 40 min; black box: 60 min.

604
N. Monfort et al. / Tetrahedron 60 (2004) 601–605
estimated to be about 1000 h21; (b) the total turn over
number (TON) which was about 21,500 and (c) the space
time yield of the reactor which was about 2850 g/L/day. As
can be seen, these values are very satisfactory and indicate
that the industrial implementation of such a process could be
interestingly envisaged (Scheme 3).
linked to a resolution process. This is an additional
illustration to the fact that this novel green chemistry type
of process should offer very interesting opportunities for
industrial preparation of both enantio-enriched epoxides and
of their corresponding vicinal diols. Further work is going
on in our laboratory in order to scale up this type of process
to a multigram scale and to explore other possible
applications of this two phase methodology.
As a complement to this process it is to be stressed that, as
we have described previously,8 it is possible to chemically
cyclise the obtained chloro-diol (R)-2 into epoxy-alcohol
(R)-3, which can itself be transformed back into epoxide
(S)-1 via treatment with triphenylphosphine in carbon
tetrachloride (Scheme 3). Both these reactions were
shown to occur without loss of stereochemical integrity.
Thus, by pooling the 41.5% of the unreacted and the thus
obtained (S)-1 epoxide, it is possible to render this BHKR
strategy enantioconvergent (i.e., to overcome the 50% yield
limitation linked to a classical resolution process), the
epoxide (S)-1 of 100% ee being easily obtained by
performing a second-round BHKR of the thus obtained
highly enantiomerically enriched substrate.
4. Experimental
4.1. General
Gas chromatography (GC) analysis were performed using
the chiral column Lipodex G (25 m£0.25 mm, Macherey–
Nagel, 1 kg/cm2 helium). NMR spectra were recorded in
CDCl3 using a Bruker 250 MHz instrument. The crude
powders of the recombinant A. niger epoxide used
throughout this study respectively showed an activity of 7,
8 and 11 U per mg against styrene oxide (Units are
expressed in mmol/min/mg of powder), as measured by
our new (sodium periodate based) spectrophotometric
assay.10
3. Conclusion
The aim of this work was to optimise the experimental
conditions of the BHKR of epoxide rac-1. This epoxide is a
key building block for the synthesis of D0870, a triazole
drug derivatives known to display efficient activity against
infections—as for instance fluconazole-resistant oro-oeso-
phageal candidiasis—often encountered in HIV-infected
(and other immunocompromised) patients. By using our
commercially available (recombinant) A. niger epoxide
hydrolase as biocatalyst, we have shown that (a) this
enzyme was very active against this particular substrate, (b)
this BHKR could be performed at nearly room temperature,
in plain water (and 5% DMSO) i.e., using a salt free
methodology (c) a two-phase reactor could be conducted at
a concentration as high as 2.5 M (i.e., 500 g/L) without
apparent substrate and/or product inhibition (d) both the
unreacted epoxide (S)-1 and the formed chloro-diol (R)-2
could be obtained in nearly enantiopure form and nearly
quantitative yield. Interestingly, both the TOF and the TON,
as well as the space time yield of this reaction, were
excellent.
4.1.1. Synthesis of 1-chloro-2-(2,4-difluorophenyl)-2,3-
epoxypropane 1. This was synthesised following the
procedure described by Murakami and Mochizuki.11 This
epoxide was obtained as a low-melting solid. Mp¼42 8C. 1H
NMR d: 2.9 (d, J¼5 Hz, 1H, OCH2), 3.2 (d, J¼5 Hz, 1H,
OCH2), 3.7 (d, J¼12 Hz, 1H, CH2Cl), 4.1 (d, J¼12 Hz, 1H,
CH2Cl), 6.9 (m, 2H, ArH), 7.4 (m, 1H, ArH); GC-condition:
100 8C; (R)-1¼28.7 min; (S)-1¼29.7 min.
4.2. Comparative study between the BHKR of rac-1
conducted in phosphate buffer solution or in plain water
8 mg of chloro-epoxide 1 dissolved in 2 mL of DMSO were
mixed to 18 mL of plain water (or pH 7 phosphate buffer,
0.1 M). To this solution, 0.5 mg of A. niger EH powder
(7 U/mg solid) dissolved in 100 mL of water were added and
the medium was stirred at 27 8C. Aliquots of 400 mL were
withdrawn at different time intervals; 600 mL of acetonitrile
were added to stop the reaction. The epoxide was extracted
with 600 mL of isooctane containing 4-bromo-acetophe-
none (2 mM) as internal standard and analysed by chiral GC
for conversion ratio and ee.
Moreover we have previously shown that (R)-2 could be
easily transformed chemically into (S)-1 without loss of
stereochemical integrity. Thus, in this particular case, it is
possible to set up an enantioconvergent process, i.e., to
overcome the theoretical 50% yield limitation theoretically
4.3. Optimisation of the substrate concentration
General procedure. These enzymatic reactions were carried
Scheme 3. Optimal biocatalysed hydrolytic kinetic resolution of rac-1, using a two-phase process at a substrate concentration of 500 g/L (i.e., 2.5 M).

N. Monfort et al. / Tetrahedron 60 (2004) 601–605
605
1
out at 27 8C in microvials containing an adapted final
volume of 1000–50 mL as a function of the substrate
concentration (0.4–750 g/L). To different substrate emul-
sions in plain water containing 10% of DMSO a calculated
enzyme extract quantity (7 U/mg solid) was added in order
to keep constant the enzyme/substrate ratio at 93 U/mmol of
rac-1. The determination of the ee of the residual epoxide
after 20, 40 and 60 min of reaction time was carried out as
described above using one microvial for each value.
(S)-3¼19 min (small peak); (R)-3¼20.2 min. H NMR d:
2.2 (s, 1H, OH), 2.85 (d, J¼5 Hz, 1H, OCH2), 3.3 (d,
J¼5 Hz, 1H, OCH2), 4.0 (m, 2H, CH2OH), 6.7–7.0 (m, 2H,
ArH), 7.3–7.5 (m, 1H, ArH).
4.4.3. Synthesis of (S)-1-chloro-2-(2,4-difluorophenyl)-
2,3-epoxypropane 1 from (R)-3. 20 mg (0.107 mmol) of
(R)-3 obtained previously and 32 mg (0.122 mmol) of
triphenylphosphine were dissolved in 1 mL carbon tetra-
chloride, and heated for 7 h under reflux.11 After cooling
and addition of water (2 mL), the mixture was extracted
with methylene chloride (2 mL). The organic phase was
washed with brine, dried over magnesium sulfate and
evaporated. Injection on chiral GC at 100 8C led to identical
retention time (29.7 min) as one recovered after
biohydrolysis.
4.4. Optimisation of the enzyme over substrate ratio
This study was carried out at 27 8C in microvials containing
a final volume of 50 mL. In each vial 25 mg of chloro-
epoxide 1 were mixed at 27 8C to 25 mL of plain water
containing 10% of DMSO and different quantity of enzyme
powder (7 U/mg solid; enzyme/substrate ratio: 25–100 U/
mmol). Very small aliquots were withdrawn at different
time intervals and analysed as described above.
Acknowledgements
4.4.1. Preparative scale resolution of rac-1 at 0.4 g/L.
400 mg of chloro-epoxide 1 dissolved in 100 mL of DMSO
were mixed to 900 mL of phosphate buffer (0.1 M, pH 7).
The biohydrolysis was started by addition of 13 mg of A.
niger powder (8 U/mg). The medium was stirred and
maintained at 27 8C. When the ee of 1 reached about 98%
(4 h, 45 min) the reaction was stopped by adding 250 mL of
acetonitrile. The medium was saturated with NaCl and then
extracted with AcOEt (4£200 mL). After drying over
MgSO4 and concentration in vacuum (S)-1 [163 mg, 41%
yield, [a]2D6¼þ45 (c 1.2; THF)] and (R)-2 [163 mg, 38%
This work was partly funded by the European Community
Contract ‘Engineering integrated biocatalysts for the
production of chiral epoxides and other pharmaceutical
intermediates’ (No. QLK3-CT2000-00426).
References and notes
1. Ito, Y. N.; Katsuki, T. Bull. Chem. Soc. Jpn 1999, 72,
603–619.
1
yield, [a]2D6¼þ3.6 (c 1; THF); H NMR d: 2.2 (dd, J¼5.4,
2. Schaus, S. E.; Brandes, B. D.; Larrow, J. F.; Tokunaga, M.;
Hansen, K. B.; Gould, A. E.; Furrow, M. E.; Jacobsen, E. N.
J. Am. Chem. Soc. 2002, 124, 1307–1315.
3. Archelas, A.; Furstoss, R. Curr. Opin. Chem. Biol. 2001, 5,
112–119.
7.7 Hz, 1H, OH), 3.3 (s, 1H, OH), 3.8–4.1 (m, 4H, CH2Cl
and CH2O), 6.7–7.0 (m, 2H, ArH), 7.6–7.75 (m, 1H, ArH)]
were separated and purified by flash chromatography then
bulb-to-bulb distillation (130 8C, 5 mbar for 1 and 200 8C,
0.1 mbar for 2). The diol 2 was obtained as a viscous oil.
4. Genzel, Y.; Archelas, A.; Broxterman, Q. B.; Schulze, B.;
Furstoss, R. J. Org. Chem. 2001, 66, 538–543.
5. Manoj, K. M.; Archelas, A.; Baratti, J.; Furstoss, R.
Tetrahedron 2001, 57, 695–701.
4.4.2. Synthesis of (R)-1-hydroxy-2-(2,4-difluorophenyl)-
2,3-epoxypropane 3 from (R)-2. To a stirred solution of
73 mg (0.33 mmol) of (R)-2 in dry THF (4 mL) cooled on an
ice bath, were added 20 mg (0.5 mmol) of sodium hydride
(60% mineral oil dispersion). The resulting suspension was
then agitated at 0 8C for 2 h. After addition of water (2 mL),
the mixture was extracted twice with ether (5 mL). The
combined extracts were washed twice with a saturated
NH4Cl solution (2 mL) then twice with brine (2 mL) and
dried over magnesium sulfate. After concentration in
vacuum and purification by bulb-to-bulb distillation
(0.1 mbar, 200 8C) 59 mg of (R)-3 (96% yield) were
isolated. [a]2D6¼þ44.6 (c 1; THF); lit.11 for (S)-3:
[a]D¼242 (c 1; THF); GC-condition: 120 8C,
6. Molina, J.; Brener, Z.; Romanha, J. A.; Urbina, J. A.
J. Antimicrob. Chemother. 2000, 46, 137–140.
7. Davies, E.; Mallion, K.; Pittam, J. D.; Taylor, N. World Patent
WO 96/04256, 1994.
8. Monfort, N.; Archelas, A.; Furstoss, R. Tetrahedron: Asym-
metry 2002, 13, 2399–2401.
9. Cleij, M.; Archelas, A.; Furstoss, R. Tetrahedron: Asymmetry
1998, 9, 1839–1842.
10. Mateo, C.; Archelas, A.; Furstoss, R. Anal. Biochem. J. 2003,
314, 135–141.
11. Murakami, K.; Mochizuki, H. European Patent EP
0472392A2, 1992.