Enzymatic transformations 63. High-concentration two liquid-liquid phase Aspergillus niger Epoxide hydrolase-catalysed resolution: Application to trifluoromethyl-substituted aromatic epoxides

Article abstract of DOI:10.1002/adsc.200700085

The aim of this work was to perform different studies allowing us to improve the methodology we had previously described to realise the Aspergillus niger epoxide hydrolase-catalysed resolution of eight trifluoromethyl- substituted styrene oxide derivative

Full text of DOI:10.1002/adsc.200700085

FULL PAPERS
DOI: 10.1002/adsc.200700085
Enzymatic Transformations 63. High-Concentration Two Liquid-
Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed
Resolution: Application to Trifluoromethyl-Substituted Aromatic
Epoxides
Justine Deregnaucourt,a Alain Archelas,a,* Fabien Barbirato,b Jean-Marc Paris,b
and Roland Furstossa
a
Groupe Biocatalyse et Chimie Fine, FRECNRS 3005, UniversitØ de la MØditerranØe, FacultØ des Sciences de Luminy,
Case 901, 163 avenue de Luminy, 13288 Marseille Cedex 9, France
Fax : (+33)-(0)-4-9182-9145; e-mail: archelas@luminy.univ-mrs.fr
StØ RHODIA, Centre de Recherches de Lyon, 85 avenue des Fr›res Perret, 69192 Saint-Fons Cedex, France
b
Received: February 5, 2007
Abstract: The aim of this work was to perform dif- centration, i.e., 360 g/L (1.8M). The efficiencies of
ferent studies allowing us to improve the methodolo- the Aspergillus niger epoxide hydrolase toward the
gy we had previously described to realise the Asper- different epoxides studied were compared and the
gillus niger epoxide hydrolase-catalysed resolution of productivity of the practical process was evaluated.
eight trifluoromethyl-substituted styrene oxide deriv- These results amply demonstrated that, without con-
atives. A two liquid-liquid phase methodology was test, the biocatalysed hydrolytic kinetic resolution of
developed, for the first time with such an enzyme, epoxides is nowadays to be considered as a very effi-
using iso-octane as a co-solvent. Different experi- cient, mild, cheap and easy-to-use “green chemistry”
mental parameters like the global volume substrate methodology applicable for cost effective preparative
concentration, the substrate over enzyme ratio, the (i.e., industrial) scale implementation.
agitation rate and the purity of the enzyme were also
optimised. This allowed us to set up very efficient bi-
oreactors allowing performance of the hydrolytic ki- Keywords: Aspergillus niger epoxide hydrolase;
netic resolution of the different substrates by operat- enzyme catalysis; fluorinated styrene oxide deriva-
ing at room temperature (278C), within a few hours tives; high-concentration two-phase bioreactor; ki-
only and at a very high global volume substrate con- netic resolution
Introduction
strate and/or enantioselectivity, thus providing the or-
ganic chemist with a set of potential tools for per-
Epoxides – as well as their corresponding vicinal diols forming a specific epoxide resolution.[1]
– are important chiral building blocks often involved
Moreover, some of these enzymes have been
in the production of a wide range of fine chemicals cloned and overexpressed, and directed evolution
and pharmaceutical molecules. Owing to the fact that techniques have allowed, in certain cases, one to im-
the biological activity of such products is often linked prove some of their properties.[2] Thus, it was by now
to only one enantiomer (eutomer), and due to the po- amply demonstrated that such a biocatalytic approach
tential toxic side effects of the other antipode (di- can, in principle, allow one to perform the resolution
stomer), the production of these molecules in enantio- of a given racemic epoxide under very mild, cheap,
pure form is an ongoing industrial challenge.
easy-to-use and efficient experimental conditions. We,
As a consequence of several practical features, bio- for instance, have shown that, using the now commer-
catalysed hydrolytic kinetic resolution (BHKR) of a cially available Aspergillus niger (A. niger) recombi-
racemic epoxide can be a very attractive alternative nant EH, the resolutions of several types of epoxides
to potentially toxic and substrate-limited chemocata- were efficiently performed, at room temperature and
lytic methods. To date, several appropriate enzymes, by using solely demineralised water as solvent, i.e.,
i.e., epoxide hydrolases (EHs), have been discovered, following a salt-free methodology.[3] Obviously, these
showing interesting wide and complementary sub- practical features are very important as far as energy
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
1405

Justine Deregnaucourt et al.
FULL PAPERS
consumption and the overall downstream processing linked to this reaction product. We have thus demon-
costs of a (potential) industrial process are concerned. strated that, by using the recombinant A. niger EH,
Moreover, they clearly contribute to the world-wide the usable so-called “global volume substrate concen-
effort for the development of “environmentally tration”[10] could be as high as 80, 306 or even 500 g/L
gentle” chemical processes, since the biocatalyst used (respectively 0.36, 2 or 2.5M) without noticeable sub-
can very easily be obtained in unlimited amounts, is strate and/or product inhibition.[3a,11] The space-time
natural, non-toxic and easily biodegradable.
yield thus reached good to excellent values (237
One primordial key factor for such an industrial g/LhÀ1 in the last case) thus clearly opening the way
process is its volumetric productivity.[4] As a general to industrial implementation. Along this line, different
feature, a major drawback in conventional aqueous studies based on the use of water-immiscible co-sol-
biocatalysis is the low water solubility of most organic vents, including methyl tert-butyl ether,[6] dodec-
(hydrophobic) substrates (which is typically a few ane,[7,12] n-hexadecane[13] or iso-octane have also been
grams per litre) and the resulting poor space-time explored.[14] One recent result described the use of an
yield (STY) of the corresponding (theoretical) pro- isolated (overexpressed) bacterial EH, which could be
cess. In the case of epoxide hydrolases, several ap- operated in the presence of octane as a co-solvent.[15]
proaches have been developed in order to overcome However, as a general feature and although the sub-
this stringent bottleneck which severely hampers strate concentration within the organic phase was
large-scale industrial development. Different solu- claimed as being reasonable in different examples
tions, including addition of a water-miscible co-sol- (i.e., 39 g/L in the last case), the global volume sub-
vent as well as the use of two liquid-phase reactors, strate concentration (i.e., 7.8 g/L in the last case) and
have been developed. For instance, we have shown therefore the overall reactor productivity was only
that the A. niger EH could achieve the resolution of moderate. As a consequence, the space-time yield (at
para-nitrostyrene oxide at a concentration of 54 g/L the best a few g/LhÀ1) of these reactors stayed dra-
(330 mM) in the presence of 20% DMSO (a water- matically incompatible with any further (industrial)
miscible solvent) over a 6 h period, whereas no reac- implementation.
tion occurred in the absence of co-solvent.[5] Interest-
We here describe our studies on a specific trifluoro-
ingly however, the same solvent appeared to be dele- methyl-substituted epoxide family, which allowed us
terious for the whole cell EH activity of either Rho- to set up, under appropriate co-solvent and optimised
dotorula glutinis[6,7] or Rhodococcus rubber SM operational conditions, cost-effective reactors operat-
1789.[8] Subsequent studies demonstrated that, as a ing at room temperature (278C), at high global
general rule, such solvents cause severe enzyme inhib- volume substrate concentration (i.e., several hundred
ition.[9]
g/L) and over a short reaction time, thus providing an
Other types of approaches were based on the use excellent process productivity.
of a two liquid-liquid phase methodology, in which
the immiscible organic phase was either the neat ep-
oxide itself or various water-immiscible organic co- Results and Discussion
solvents. In this type of approach, the organic phase
plays the role of an internal substrate reservoir and In a preliminary work, we have recently described the
also offers the advantage both to protect the substrate A. niger EH-catalysed BHKR of the various trifluoro-
epoxide from spontaneous hydrolysis and to extract methyl-substituted styrene oxide epoxides rac-1–8
part of the formed diol from the aqueous phase, thus (Scheme 1).[16] They were synthesised with moderate
minimising the inhibition phenomena very often to excellent yield (46–99%) starting from the corre-
Scheme 1.
1406
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
sponding (commercially available) aldehyde or (this concentration was adjusted so as to afford a
ketone.[17] Our exploratory results indicated that, in 10 g/L global volume substrate concentration) was
spite of the presence of fluorine atoms, which very dispersed in a demineralised water phase, in the pres-
often disturb chemical reactivity, the BHKR of these ence of an amount of A. niger EH corresponding to a
substrates was indeed possible, at rather low substrate substrate over enzyme (S/E) ratio of 58. Very satisfac-
concentrations, however (about 2 mM). In all cases, torily, the resolution processed efficiently and a con-
the unreacted epoxide and the corresponding diol of, version ratio of nearly 50% was obtained after
respectively, (S) and (R) absolute configurations were 120 min reaction. The recovered (less reactive) epox-
obtained, as determined using a circular dichroism ide showed an ee (ees) higher than 99% and the calcu-
methodology.[16] In the present work we explored, lated Eapp value was 33.[20] Further experiments were
using the same epoxides rac-1–8, the various experi- conducted in order to explore the possibility to in-
mental options described above for increasing the crease the global volume substrate concentration.
substrate concentration in this process.
Concentrations of 100 and 360 g/L (0.5 and 1.8M, re-
spectively) were assayed, keeping the 10% (v/v) iso-
octane/water proportion and an unchanged S/E=58
ratio. As shown in Figure 1 (upper and lower panels),
the resolution proceeded very satisfactorily at 100 g/L
concentration but, at 360 g/L, the concentration pro-
Preliminary Experiments
Use of Water-Miscible Solvents
Following our previously described results,[5] we first
explored the possibility to enhance the substrate con-
centration in the presence of a water-miscible solvent.
Epoxide rac-2 was used as a model substrate. Thus, its
behaviour in a 20% (v/v) DMSO aqueous solution
containing a constant (weight to weight) substrate
[18,19]
over enzymatic powder S/Eratio (67)
was ex-
plored, at increasing substrate concentrations (i.e.,
0.5, 5 and 10 g/L). Unfortunately, these attempts were
unsuccessful, indicating in particular a severe phase-
transfer rate limitation of the insoluble substrate to
the aqueous/DMSO phase at a 10 g/L concentration.
Thus, in contrast to our (above described) previous
results, this limitation would obviously be worse in
the absence of any water-miscible solvent (i.e., using
our neat substrate approach).[11] We therefore ex-
plored an alternative two liquid-liquid phase method-
ology based on the use of an additional water-immis-
cible co-solvent.
Two Liquid-Liquid Phase Approach using a
Water-Immiscible Solvent
Exploratory studies: To explore this possibility, epox-
ide rac-7 was used as a model substrate. Different
water-immiscible solvents were checked, keeping in
mind that such a solvent should not have deleterious
effects on enzyme activity and should be chosen so
that partitioning of substrate and product over the
two phases facilitates downstream processing. All ex-
periments were conducted at room temperature
(278C). Among the solvents tested (diisopropyl ether,
tert-butyl methyl ether, ethyl acetate, toluene, hexane
and iso-octane) iso-octane proved to be by far the
most appropriate, leading to the fastest resolution
process. Thus, under vigorous (magnetic) stirring, a
Figure 1. Concentration profile of the A. niger EH-catalysed
resolution of rac-7 at different global volume substrate con-
centrations. Upper panel: at 10, 100 and 360 g/L using a
10% iso-octane/water medium. Lower panel: at 360 g/L
global volume substrate concentration using a 20% iso-
10% (v/v) proportion of a rac-7/iso-octane solution octane/water medium.
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1407

Justine Deregnaucourt et al.
FULL PAPERS
file clearly indicated the occurrence of a phase-trans- pares with those generally in operation with these
fer limitation. This was due to the rather surprising chemical catalysts.[22]
(and specific) physico-chemical properties of fluorine-
containing compounds.[21] Thus, increasing the sub-
strate concentration led to formation of numerous SemiPreparative Scale Experiments
substrate droplets firmly sticking to the reaction
vessel surface, thus making the substrate inaccessible On the base of these very satisfactory preliminary re-
to dispersion/emulsification in the medium. This sults it could be concluded that, at least for epoxides
could be overcome simply by increasing the iso- 4 and 7, the use of iso-octane as a water-immiscible
octane/water ratio to 20% which allowed us to avoid solvent allowed us to overcome the above discussed
this “substrate sticking” phenomenon (Figure 1, lower phase-transfer bottleneck. Moreover, this allowed a
panel). Thus, even at substrate concentrations as high two liquid-liquid phase methodology to be run at a
as 100 and 360 g/L the resolution could be achieved very high global volume substrate concentration, and
within 210 and 240 min (3.5 and 4 h) respectively.
also to noticeably minimise the amount of enzyme
Optimisation of the S/E ratio: Due to the interest to used. These results greatly encouraged us to pursue
minimise the amount of enzyme used (i.e., to opti- this study in order to explore the applicability of
mise the S/Esubstrate over enzyme ratio) which obvi-
these findings to the other epoxides. Each of these
ously would allow one to lower the overall cost of a substrates was submitted to exploratory semi-prepara-
large-scale process, we further explored this option tive trials. The experimental parameters were set so
using epoxide rac-4. This was achieved at a fixed sub- as to provide a 50 g/L global volume substrate con-
strate concentration of 250 g/L (1.22M), using a 20% centration. Thus, 1.25 g of epoxide was submitted to
iso-octane/demineralised water liquid medium and by BHKR in a 50 mL (total) volume mechanically stirred
varying the S/Eratio from 50 to 200. The concentra-
microreactor, using 25 mL liquid medium (10% v/v
tion profile of these experiments is provided in iso-octane/demineralised water) and an S/Eratio of
Figure 2. It could be observed that, even at an S/E 100. The obtained results are provided in Table 1, in-
value as high as 200, the BHKR proceeded smoothly dicating that all (but 6) epoxides were very efficiently
within 240 min. From a practical point of view, this processed under these experimental conditions. Nei-
translates into the fact that, under these experimental ther substrate nor product inhibition was observed,
conditions, the BHKR of 1 kilogram of rac-4 could be except in the case of 5, where severe product inhibi-
conducted at room temperature and within about 4 h tion was detected. The only exception was for epoxide
by using only 5 g of partly purified enzymatic powder 6, where the EH showed low activity. In this case, the
(containing about 25% pure enzyme). In this particu- reaction had to be run at 10 g/L substrate concentra-
lar case, the catalyst over substrate molecular ratio tion only, using a S/E=50 value. All resolutions were
(i.e., the E/S mol% value), a parameter which is very achieved within a few hours, the ees reaching a value
often used for transition metal catalysts, was about of at least 98% in all cases. Obviously, as in any ki-
5.7·10À4 mol%, a value which very favourably com- netic resolution, an enantiomerically pure (unreacted)
epoxide could be obtained by allowing the reaction to
proceed further. The Eapp value was determined for
each substrate using both ees and eep. It appeared to
be reasonable to excellent for the ortho- and para-
substituted derivatives, whereas it was rather low for
the meta-substituted substrates 2 and 7 which only ex-
hibited Eapp values of about 5 and 17, respectively.
This tendency is in perfect coherence with our general
(unpublished) observation about the influence of the
aromatic ring substitution pattern of styrene oxide de-
rivatives. The absolute configuration of the recovered
epoxide and of the formed diol [that is, (S) and (R),
respectively] remained unchanged as compared to our
previous results obtained in the absence of water-im-
miscible solvent.[16]
Interestingly, it is to be stressed that epoxides 7 and
8, bearing an additional gem-methyl group on the ox-
irane ring, were efficiently processed. This is a very
noteworthy observation since even the Jacobsenꢁs
salen (Co)OAc catalysts, claimed as being the best
transition metal based catalysts for achieving the
Figure 2. Influence of the S/Eratio on the resolution of rac-
~
4 at a global volume substrate concentration of 250 g/L.
:
&
^
*
S/E=50; : S/E=100; : S/E=150; : S/E=200.
1408
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
Table 1. Results obtained upon A. niger EH-catalysed hydrolytic kinetic resolution of rac-1–8, operated in the presence of
iso-octane.[a]
Substrate
1
2
3
4
5
6
7
8
Reaction time
5 h 15
2 h 05
2 h 05
1 h 10
1 h 45
-
3 h 40
5 h 10
ees
eep
Eapp
Yieldepox
Yielddiol
98
77.3
34
34
59.2
98.7
13.2
5
11
84.9
98
84.3
52
37
51.3
98.6
94.5
175
45
98.5
85
-
-
nd
-
-
98.3
59.0
17
32
64.1
99.1
88.3
85
35
58
60[b]
33
54.8
39
[a]
Standard experimental conditions: Global volume substrate concentration: 50 g/L; 10% iso-octane (v/v); S/E=100; nd=
not determined.
Experiment achieved using a 10 g/L global volume substrate concentration and S/E=50.
[b]
HKR of epoxides, are known to be inoperative on Table 2. A. niger EH-catalysed BHKR of rac-4 conducted in
a “perfectly agitated standardised reactor” at a 250 g/L
global volume substrate concentration and at different agita-
tion rates.
such trisubstituted substrates.
Scale-Up Experiments
Agitation rate (rpm)
500
800
1000
1200
ees
eep
Eapp
98.6
95.7
226
98.7
96.3
267
99.3
95.4
250
99.0
96.0
259
In order to explore the possibility to scale-up this two
liquid-liquid phase BHKR methodology, we decided
to further optimise two important parameters, i.e., (a)
the agitation rate, which partly governs the intimate
nature of the emulsion created and therefore the
phase-transfer rate of the substrate and (b) the purity the Eapp value was determined on the base of ees and
of the enzyme to be used, which is another key factor eep, the later being measured after completion of the
as far as the overall cost of a large-scale application is reaction and extraction and purification of the formed
concerned. These parameters were studied using ep- diol. The obtained results are provided on Table 2
oxides rac-4 and 8, respectively. Epoxide 4 was and their kinetic profile is illustrated on Figure 3.
chosen due to its excellent Eapp value (Eapp =175 at
It could be observed that, except for the experi-
50 g/L) and to the high activity of the A. niger EH ment conducted at 500 rpm, the reaction rate as well
against this substrate. The choice of rac-8 (which as the calculated Eapp value were very comparable
showed an Eapp =85 at 50 g/L) was motivated by the one to each other and stayed nearly unchanged from
fact that the best conventional chemical catalysts one experiment to the other. To the contrary, both
known to day are inefficient on this type of gem-dis- these parameters were lower for the experiment con-
ubstituted substrate. All experiments were conducted
in a standardised stirred-tank reactor (STR) equipped
with a rotatory Rushton turbine. This type of reactor
is commonly used in ours laboratories as a “scale-up
tool” allowing accurate determination of most reactor
key parameters, thus facilitating direct extrapolation
to large-scale application.
Agitation Rate Optimisation
Using such a 100-mL reactor, 12.5 g of rac-4 were
submitted to BHKR at room temperature (278C),
using a 250 g/L global volume substrate concentra-
tion, a 20% (v/v) iso-octane/total liquid volume and
an S/Eratio of 200. Four experiments were per-
formed, each of them at a different rotation rate, i.e.,
at respectively 500, 800, 1000 and 1200 rpm. Aliquots
were withdrawn each 15 min and analysed for ees
Figure 3. Reaction profile of the BHKR of rac-4 at different
~
&
*
^
agitation rates. : 500 rpm; : 800 rpm; : 1000 rpm;
:
using chiral GC chromatography. In each experiment, 1200 rpm.
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1409

Justine Deregnaucourt et al.
FULL PAPERS
Table 3. Influence of the enzyme purity on the A. niger-catalysed BHKR of rac-4. Experiments conducted at a global
volume substrate concentration of 100 g/L and 10% (v/v) iso-octane/water.
Enzyme
Amount of sub- Reaction
Agitation rate Reaction time ees eep Eapp Yield epox- Yield diol
(amount U)
strate (g)
volume (mL)
(rpm)
(min)
ide [%]
[%]
purified
5
50
800
220
99.2 84.3 63 40
49
(1100)
crude (1100)
crude (5600)
5
25
50
250
800
600
290
150
99.8 78.8 56 41
99.7 78.5 52 42
49
49
ducted at 500 rpm. This result clearly indicated that room temperature (278C) using a 100 g/L global
no substrate phase-transfer limitation did occur pro- volume substrate concentration, 50 mL of a 10%
vided the agitation rate was maintained at least at (v/v) iso-octane/water liquid medium, a 800 rpm agita-
800 rpm. This is a valuable information since minimis- tion rate and 1100 U of enzyme.[18] The stirred-tank
ing energy cost is another important goal for running reactor used was the same as the one mentioned
an industrial process. Hence, it is also interesting to above for the BHKR of rac-4. The reactions were fol-
stress that the optimum Eapp value (about 250) was lowed by controlling the ees using chiral GC and the
noticeably higher in these cases as compared to the Eapp values were calculated on the base of the ees and
value of 175 observed during the semi-preparative eep after extraction and purification of the reaction
scale experiment described above, conducted in a product. Interestingly enough, no difference between
non-optimised reactor. These results clearly indicate the two experiments did appear, the kinetic profiles
that intimate dispersion of the substrate in the of the two reactions being nearly superimposable. The
medium – achieved by using an optimised reactor set Eapp values appeared to be identical as well (Table 3).
and an adequate agitation rate – is another key factor From these experiments it could be concluded that
of this resolution. Using these optimised experimental the purity of the enzyme used influenced neither the
conditions, a total amount of 50 g of rac-4 was proc- reaction rate nor the product formation of the reac-
essed in four (identical) experiments of 12.5 g each, tion. In particular, no conflicting side reactions, which
which proved to be highly reproducible. These were could have been catalysed by other enzymes present
performed using a 250 g/L global volume substrate in the crude extract, did occur.
concentration, an agitation rate of 800 rpm and an S/
Eratio of 200 (i.e., a total amount of 312.5 mg enzy-
As a validation of this observation, a 25-g prepara-
tive-scale experiment using the crude enzyme was op-
matic powder). Each of these experiments only neces- erated under the same experimental conditions. Inter-
sitated about 210 min to reach an ees >99%. Extrac- estingly, due to the improved geometrical characteris-
tion and purification of the remaining epoxide and of tics of the 500-mL standardised STR we used, a lower
the formed diol provided as a pool 27.0 g of (S)-4 (ee (600 rpm) agitation rate was sufficient and a shorter
99%, 43% yield) and 29.1 g of the corresponding reaction time was observed.
(R)-diol (ee 95%, 44% yield).
Resolution of rac-1-8 using Jacobsenꢀs Catalysts
Use of a Crude Enzymatic Preparation
As far as the resolution of racemic epoxides is con-
An additional important parameter is the cost of the cerned, different conventional chemistry-based meth-
enzyme production itself. As described previously, the odologies have been described over the recent years,
A. niger EH is produced by culturing the recombinant the best known implying the use of transition metal-
fungal strain, which only necessitates a simple and based catalysts. In order to compare the efficiency of
cheap culture medium. In a first step, a straightfor- our enzymatic methodology versus these chemical ap-
ward isolation procedure affords the crude enzyme in proaches, we have explored the possibility to perform
aqueous solution. In a second step, this can be further the resolution of rac-1–8 using the so-called sa-
purified to obtain the partly purified enzyme in a len(Co)OAc catalyst, using the best appropriate ex-
lyophilised form (which is in fact the material which perimental conditions.[22] Thus, 1 mmol of each epox-
was used up to now throughout this study). We now ide was submitted to a HKR using (R,R)-(salen)Co-
checked whether the crude material could be used to
A
perform the BHKR of rac-8 as a model substrate and, values were determined after extraction of the reac-
for the sake of comparison, the same experiment was tion medium and chiral GC analysis. When appropri-
conducted using the partly purified EH. For each one ate, the ee of the formed diol (eep) was similarly deter-
of these experiments, 5 g of rac-8 were processed at mined after chemical derivatisation. The respective
1410
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
Table 4. Analytical scale HKR of rac-1–8 using (R,R)-(salen)Co
A
strate at 278C, using 0.007 mmol (4.76 mg) of catalyst and 0.55 mmol (9.9 mL) of water. Reaction time: 48 h.
Substrate
1
2
3
4
5
6
7
8
c
ees
eep
E
E
77.7
23.3
nd
~1
34
61
nd
nd
nd
25.8
0.2
nd
~1
nd
2
19.6
0.2
nd
~1
85
100
nd
>30
5
97.8
79.3
38
98.6
89.0
84
99.6
89.4
110
60
0.1
nd
~1
17
app chemical
app enzymatic (at 50 g/L)
52
175
Eapp values were calculated by using either the c and the time necessary for this transformation.[4] The third
ees values or the ees and eep values. The results ob- value (aTOF) provides a good image of the average
tained are summarised in Table 4. For a comparison efficiency of the catalyst over time. The fourth (STY)
purpose, the Eapp values obtained by our A. niger EH parameter describes the overall efficiency of a given
enzymatic resolution under biphasic conditions (i.e., reactor by indicating the productivity of the process
10% v/v iso-octane) at a global volume substrate con- (generally expressed in grams per litre per hour). Ob-
centration of 50 g/L are also recalled in Table 4. Our viously, the higher these TOF, TON, aTOF and STY
results indicated that the outcome of the reaction was values, the better the catalyst and the overall process
clearly depending upon the substitution pattern of the will be. An additional difficulty to the calculation of
aromatic ring, as well as of the epoxide moiety. Thus, the correct TON and aTOF values lies in the fact that
in the case of the monosubstituted epoxides (a) the it is very difficult to accurately tune the minimum
ortho-substituted substrate 1 was not resolved by the amount of (bio)catalyst to be used to just reach this
salen(Co) catalyst, whereas an Eapp value of about 34 ees and to, simultaneously, exhaust the (bio)catalyst.
was measured using the biocatalytic approach, (b) the Therefore, the amount of catalyst used will be higher
Eapp value of the meta-substituted substrate 2 was than strictly necessary, and the experimentally evalu-
higher with the salen(Co) catalyst, (c) in the case of ated TON and aTOF will be lower than the “real”
the para-substituted epoxides 3, 4 and 5, two out of values. However, provided the amount of catalyst
three substrates showed a better Eapp value under en- used is only slightly higher than the minimum re-
zymatic catalysis, (d) hence, the chemical approach quired, the calculated value will – although underesti-
was inoperative for all gem-disubstituted epoxides 6, mated – provide a good image of the efficiency of the
7 and 8.
(bio)catalyst.
Catalytic Efficiency
Experimental Determination of the Efficiency
Parameters (h2)
In addition to the Eapp value, which obviously is of
fundamental importance in the context of a prepara- According to the above described definitions, the dif-
tive-scale resolution process, some other important ferent efficiency parameters related to the resolution
parameters – which unfortunately are very often not experiments of each epoxide 1–8 were determined
addressed – are (a) the so-called turnover frequency and are summarised in Table 5.
(TOF), (b) the total turnover number (TON), (c) the
As commented above, the experimental conditions
average turnover frequency (aTOF) (i.e., the TON under which the different parameters of Table 5 were
value divided by the time necessary to reach comple- recorded were not optimised for each one of the ex-
tion of the reaction), and the space-time yield (STY) periments. Therefore, as a general rule, all these
of the process.[4,23] The first parameter relates to the values surely could be further improved. Neverthe-
efficiency of a given catalyst by indicating the number less, it can be observed that most of them are already
of molecules of substrate transformed – in a given very satisfactory. Thus, all the E/S mol% values range
time period – by one molecule of catalyst. This TOF at values of about 10À3, which very favourably com-
value corresponds in fact to the kcat value used in en- pare with those generally used for transition metal-
zymology and is generally measured at the beginning based catalysts.[22] For example, this value is 1.1·10À3
of the reaction.[24] The second parameter (TON) re- mol% for rac-4 at S/E=100 and 50 g/L, i.e., under
lates to the overall efficiency and stability of the cata- non-optimised conditions (see Table 1). However, we
lyst in the given experimental conditions (catalyst have demonstrated (see above) that, for this same
productivity), by indicating the total number of sub- substrate, an S/Eratio of 200 could be used to achieve
strate molecules transformed by one molecule of cata- the nearly optimised resolution at a 250 g/L global
lyst until this catalyst is exhausted, independently of volume substrate concentration. In this case, the cal-
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1411

Justine Deregnaucourt et al.
FULL PAPERS
Table 5. Process efficiency parameters calculated upon resolution of epoxides 1–8 catalysed by the A. niger EH.
Epoxide
Eapp
S/E
[w/w]
E/S[a]
mol %
TON[a]
[mol sub/mol enz]
aTOF[a]
[mol sub/mol enz/h]
STYdiol
STYepoxide
[g/LhÀ1]
[g/LhÀ1]
1
2
3
4
4
5
6
7
8
8
34
5
100[b]
100[b]
100b]
100[b]
200[c]
50[b]
10À3
5.7·104
8.1·104
4.8·104
4.8·104
7.8·104
2·104
11·103
39.5·103
23.4·103
44.1·103
22·103
14·103
nc
5.7
3.3
2.7
9
20.5
30.7
2.3
-
4.7
3.4
16.8
10À3
20.7
12.5
29.9
34
2.9
nc
9.4
5.7
21.5
52
175
267
60
nd
17
85
52
10À3
1.1·10À3
5.7·10À4
2.4·10À3
1.1·10À3
1.1·10À3
1.1·10À3
2.3·10À3
100[b]
100[b]
100[b]
nc
5.6·104
2.6·104
2.2·104
16.6·103
5.1·103
8.7·103
[d]
-
[a]
Calculated assuming a 25% content of EH in the enzymatic extract and a molecular weight of 45 kDa of the protein. nc:
not calculated.
Standard experimental conditions described in Table 1.
Experimental conditions corresponding to the preparative scale biohydrolysis of rac-4 at 250 g/L in an STR.
Experimental conditions corresponding to the preparative scale biohydrolysis of 25 g of rac-8 at 100 g/L in a STR using
crude enzymatic extract as biocatalyst.
[b]
[c]
[d]
culated E/S mol% value was as low as 5.7·10À4 %. tion and combination of all the different parameters
Similarly, under the non-optimised experimental con- of a specific process have, as a general rule, to be con-
ditions, the recorded TON value of this resolution sidered for correct evaluation.
was 4.8·104, whereas under the optimised conditions, a
noticeably higher TON value of about 7.8·104 was cal-
culated.
More generally, the different parameters recorded
Conclusions
in Table 5 afford interesting information about the ef- In the course of this work, we have developed differ-
ficiency of the resolution of rac-1–8 by the A. niger ent studies aimed at improving the methodology we
EH. It can be seen that, for all epoxides (but rac-6) had previously described to perform the Aspergillus
the TON values, which lie in the range 2·104 to 8·104, niger epoxide hydrolase-catalysed resolution of eight
very favourably compare with those generally accept- trifluoromethyl-substituted styrene oxide derivatives
ed for conventional chemical catalysis, which ought to (rac-1–8). Our present results indicated that, by using
be >1,000 for high-value products and >50,000 for a 10 to 20% (v/v) proportion of iso-octane as a co-
large-scale or less-expensive products.[23] Similarly, solvent, we were able to overcome the severe global
due to the catalytic efficiency of the enzyme (short re- volume substrate concentration bottleneck observed
action time), the aTOF values, which lie in the range in our preliminary studies. Thus, very efficient and
5·103 to 4.4·104 mol sub/mol enz/h, again are excellent cost-effective resolution processes could be run. For
as compared to those of chemical processes were example, rac-7 could be resolved at a global volume
aTOF values of 500 mol sub/mol cat/h for small and> substrate concentration as high as 360 g/L (1.8M)
10,000 mol sub/mol cat/h for large scale products are using a 20% iso-octane/demineralised water liquid
considered as being good to excellent.[23]
medium in a two liquid-liquid phase reactor.
Finally, another important parameter – i.e., the
For the sake of cost effectiveness, other key param-
STY value – was calculated for the formed diol as eters, i.e., (a) the substrate over biocatalyst S/Eratio,
well as for the recovered epoxide. The obtained (b) the agitation rate and (c) the purity of the enzyme
values (Table 5) in particular highlight the fact that, used, were also optimised. Thus, resolution of epoxide
as for instance for rac-2, excellent TON, aTOF and rac-4 could be performed at 250 g/L global volume
STYdiol values do not necessarily reflect an interesting substrate concentration (1.22M), using an S/Eratio of
process. This is due to the poor Evalue ( E=5) of this 200. This in fact represents a 5.7·10À4 mol% ratio of
resolution. To the contrary, excellent STYdiol and catalyst, a value which very favourably compares with
À1
STYepox values of over 30 g/Lh
could be observed the ones used with the best transition metal-based
for rac-4, for which an Evalue of 267 was estimated
at 250 g/L global substrate concentration.
catalysts. From a practical point of view, this result
translates into the fact that, under these experimental
These different results and considerations clearly il- conditions, the BHKR of 1 kilogram of rac-4 could be
lustrate the fact that, due to the multiparametric conducted, within a 4 h period, by using only 5 g of
nature of such a resolution process, careful examina- (commercially available) enzymatic powder. As far as
1412
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
(buffer A). The suspension was homogenised using a me-
chanical grinder and the cells were disrupted by pressure
shock (27 kpsi, Cell Disrupter from Constant System Ltd,
Daventry Northants, United Kingdom) to liberate the intra-
cellular EH activity. Unbroken cells and cellular debris were
precipitated by centrifugation (13,300”g, 90 min) at 48C
and discarded. The supernatant, containing the EH was
treated by microfiltration using a membrane with a cut-off
of 0.1 mm (Inceltech, PSFH, France) to obtain a clear ex-
tract. The microfiltration process was repeated 3 times after
dilution of the retentate with buffer A (2 L). The resulting
microfiltrates, were concentrated to 200 mL by tangential
flow ultrafiltration through a membrane with a cut-off of
40 kDa (Inceltech, 40UFIB/1/S2, France). The concentrated
enzyme solution was finally desalted by gel filtration (Se-
pharose G-25) using water as eluant. The active fractions
were pooled then freeze-dried to furnish the enzymatic
powder (1.5 g) stored at 48C. All procedures were carried
out at 48C.
the agitation rate is concerned, its optimum value was
determined. Depending on the specific reactor vessel
geometry, moderate values of 600 or 800 rpm were
shown to be sufficient. Finally, we have demonstrated
that the (very easily prepared) crude enzyme could be
used in place of the purified one. Indeed, its purity
did not influence the reaction rate and no conflicting
side reactions, which could have been catalysed by
other enzymes present in the crude extract, did occur.
As a validation of this observation, a 25 g prepara-
tive-scale experiment using this crude enzyme was op-
erated and led to excellent results. Owing to the im-
portance of the efficiency parameters for any industri-
al process, the total turnover number (TON), the
average turnover frequency (aTOF) and the space-
time yield (STY) were determined. Although not op-
timised, they were shown to be good to excellent and
to very favourably compare with those generally ac-
cepted for industrial applications of transition metal-
based catalysis. According to these results, the biocat-
alysed hydrolytic kinetic resolution of epoxides
proved to be a very mild, cheap, easy-to-use, efficient
and cost effective methodology. Therefore, it clearly
does contribute to the world-wide effort for develop-
ment of “environmentally gentle” (bio)chemical pro-
cesses, since the biocatalyst used can easily be ob-
Crude enzymatic preparation: The experimental protocol
was the same as described above except that, after centrifu-
gation of the cell debris, the supernatant was just concen-
trated by microfiltration in order to obtain a final volume of
200 mL. This crude enzymatic solution was frozen in liquid
nitrogen then stored at À808C.
Synthesis of epoxides
tained in unlimited amounts, is natural, non-toxic and Epoxides 1–8 were synthesized from the corresponding alde-
hyde or ketone precursor by reaction with either
(CH3)3S+ IÀ (method I) or (CH3)3SO+ IÀ (method II) in the
presence of NaH.
biodegradable. Work is going on in our laboratory in
order to further explore the applicability of this novel
two liquid-liquid phase process to the resolution of
other interesting epoxides.
General procedure: (CH3)3S+IÀ [or (CH3)3SO+IÀ]
(20 mmol) was dissolved in DMSO (15 mL). To this solution
NaH (20 mmol, 60% in oil) was added under nitrogen at
room temperature. After stirring for 20 min, the correspond-
ing aldehyde or ketone (17 mmol) dissolved in DMSO
(20 mL) was added dropwise within 20 min. After stirring
for 1–2 h at room temperature, the reaction mixture was
poured into cold water (80 mL) and the epoxide was ex-
tracted with ethyl acetate (3”100 mL). The collected organ-
ic phase fractions were washed 3 times with water (3”
100 mL), dried over MgSO4 and concentrated under
vacuum. The crude epoxide was purified by bulb-to-bulb
distillation under reduced pressure or by flash chromatogra-
phy.
Experimental Section
General Remarks
GC analyses were performed with a Shimadzu GC14 A ap-
paratus equipped with an FID detector and helium as carri-
er gas. Determination of the enantiomeric excesses was per-
formed using 2 different chiral columns (0.25 mm, 25 m),
i.e., Chirasil-Dex CB (column I, Varian) and Lipodex E
(column II, Macherey–Nagel). 1H NMR, 13C NMR and
19F NMR spectra were recorded on a Bruker AC250 instru-
ment at 250, 62.9 and 235 MHz, respectively. All measure-
ments were carried out at room temperature in CDCl3, ace-
tone-d6 or DMSO-d6.
(2-Trifluoromethylphenyl)-oxirane (1): This epoxide was
obtained (method I) as a yellow liquid; yield: 78%; bp 508C
(0.08 mbar); 1H NMR: d=2.75 (dd, 1H, J=5.5 Hz, J=
2.5 Hz), 3.21 (dd, 1H, J=5.5 Hz, J=4.25 Hz), 4.15 (m, 1H),
7.44 (d, 1H, J=7.75 Hz), 7.54 (t, J=7.5 Hz), 7.63 (t, 1H, J=
7.5 Hz), 7.75 (d, 1H, J=7.75 Hz); 13C NMR (DMSO-d6): d=
48.5 (q, JCF =2.7 Hz, CH), 50.2 (CH2), 124.3 (q, JCF =273.5,
C), 125.5 (q, JCF =5.4 Hz, CH), 125.5 (m, CH), 126.9 (q,
Enzyme Preparation
Partly purified enzyme in a lyophilised form: The recombi-
nant A. niger Gbcf 79 containing the EH gene was cultivat-
ed in a simple medium with only glucose (10 g/L) and corn
steep liquor (20 g/L) as nutrients in 10 L fermentors at 278C
as previously described.[5] After 40 h of culture incubation,
the mycelium was recovered by vacuum filtration and re-
suspended in 1.4 L of 10 mM phosphate buffer, pH 7.1, con-
taining 1 mM EDTA, 1 mM cysteine and 0.3 mM phenylme-
thanesulfonyl fluoride to prevent enzyme inactivation
J
CF =30.7 Hz, C) 128.2 (CH), 132.9 (m, CH), 136.2 (m, C);
19F NMR (CDCl3): d=À59.62; exact mass calcd. for
C9H7F3O: 187.0371; found: 187.0374. Chiral GC analysis
[column I, T=1008C: (R)=3.8 min; (S)=4.8 min].
(3-Trifluoromethylphenyl)-oxirane (2): This epoxide was
obtained (method II) as a yellow-orange liquid; yield: 52%;
1
bp 788C (1.2 mbar); H NMR (CDCl3): d= 2.79 (dd, 1H,
J=2.5 Hz, J=5.3 Hz), 3.19 (dd, 1H, J=4.0 Hz, J=5.3 Hz),
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1413

Justine Deregnaucourt et al.
FULL PAPERS
3.92 (dd, 1H, J=2.5 Hz, J=4.0 Hz), 7.38–7.65 (m, 4H);
13C NMR: d=51.4 (CH2), 51.7 (CH), 122.3 (q, JCF =3.8 Hz,
CH), 124.0 (q, JCF =272.0 Hz, C), 125.0 (q, JCF =3.7 Hz,
CH), 128.7 (m, CH), 129.8 (CH), 131.0 (q, JCF =32.4 Hz, C),
138.8 (C); 19F NMR (CDCl3): d=À62.34; exact mass calcd.
for C9H7F3O: 187.0371; found: 187.1324. Chiral GC analysis
[column II, T=908C: (S)=13.2 min; (R)=13.7 min].
(4-Trifluoromethylphenyl)-oxirane (3): This epoxide was
obtained (method I) as a yellow liquid; yield: 70%; bp 708C
(0.08 mbar); 1H NMR (CDCl3): d=2.77 (dd, 1H, J=
5.75 Hz, J=2.75 Hz), 3.18 (dd, 1H, J=5.75 Hz, J=4.0 Hz),
3.91 (dd, 1H, J=4 Hz, J=2.75 Hz), 7.39 (d, 1H, J=8 Hz),
7.60 (d, 2H, J=8 Hz); 13C NMR (CDCl3): d=51.4 (CH2),
201.0527; found: 201.0533. Chiral GC analysis [column I,
T=1008C: (R)=8.5 min; (S)=9.1 min].
2-Methyl-2-(4-trifluoromethylphenyl)-oxirane (8): This
epoxide was obtained (method II) as a colourless liquid;
yield: 99%; bp 408C (0.07 mbar); 1H NMR (CDCl3): d=
1.73 (s, 3H), 2.76 (d, 1H, J=5.4 Hz), 3.00 (d, 1H, J=
5.4 Hz), 7.48 (d, 2H, J=8.1 Hz), 7.59 (d, 2H, J=8.1 Hz);
13C NMR (CDCl3): d=22.2 (CH3), 57.1 (CH2), 57.8 (C),
124.9 (q, JCF =270.3 Hz, C), 126.1 (q, JCF =3.8 Hz, CH),
126.5 (CH), 130.5 (q, JCF =31.9 Hz, C), 146.1 (C); 19F NMR
(CDCl3): d=À62.13; exact mass calcd. for C10H9F3O:
201.0527; found: 201.0536. Chiral GC analysis [column I,
T=1208C: (R)=6.4 min; (S)=7.1 min].
51.7 (CH), 124.0 (q, JCF =271.8 Hz, C), 125.5 (q, JCF
=
4.0 Hz, CH), 125.7 (CH), 130.4 (q, JCF =32.2 Hz, C), 141
(C); 19F NMR (CDCl3): d=À62.28; exact mass calcd. for
C9H7F3O: 187.0371; found: 187.0374. Chiral GC analysis
[column I, T=1208C: (R)=9.7 min; (S)=10.7 min].
Analytical Scale Biohydrolysis of rac-2 using DMSO
as Co-solvent
20 mL of an A. niger solution (0.37–7.5 mgmLÀ1 in deminer-
alised water, S/Eratio of 67) were added to rac-2 (0.5–
10 mg, final epoxide concentration of 0.5–10 gLÀ1) previous-
ly mixed into a mixture of DMSO (250 mL) containing 3-
(trifluoromethyl)-acetophenone (0.3–0.6 mL) as an internal
standard and demineralised water (780 mL). After mixing
for 20 min at 278C the enzymatic resolution was stopped by
adding acetonitrile (500 mL) into the medium. The remain-
ing epoxide was extracted with iso-octane (1 mL) and its en-
antiomeric excess and the conversion ratio were determined
by chiral GC analysis.
(4-Trifluoromethoxyphenyl)-oxirane (4): This epoxide
was obtained (method I) as a colourless liquid; yield: 73%;
1
bp 1008C (0.07 mbar); H NMR (CDCl3): d=2.79 (dd, 1H,
J=2.5 Hz, J=5.3 Hz), 3.19 (dd, 1H, J=4.0 Hz, J=5.3 Hz),
3.92 (dd, 1H, J=2.5 Hz, J=4.0 Hz), 7.38–7.65 (m, 4H);
13C NMR (CDCl3): d=51.3 (CH2), 51.7 (CH), 120.4 (q, JCF
=
265.5 Hz, C), 121.1 (CH), 126.9 (CH), 136.4 (C), 149.1 (q,
J=1.9 Hz, C); 19F NMR (CDCl3): d=À57.35; exact mass
calcd. for C9H7F3O2: 203.0320; found: 203.0311; anal. calcd.
for C9H7F3O2: C 52.95, H 3.46, F 27.92; found: C 52.60, H
3.38, F 29.04. Chiral GC analysis [column I, T=1208C:
(R)=4.4 min; (S)=4.8 min].
Analytical Scale Biohydrolysis of rac-7 using Iso-
octane as Co-solvent
(4-Trifluoromethylthiophenyl)-oxirane (5): This epoxide
(method I) was purified by flash chromatography (pentane/
ethyl acetate, 90:10) then distilled (bp 1008C, 0.1 mbar) to
50 mL of an A. niger solution (3.44–124 mgmLÀ1 in deminer-
alised water to obtain a constant S/Eratio of 58) were
added to rac-7 (final epoxide concentration of 10–360 gLÀ1)
previously mixed into a mixture of iso-octane (100 mL or
200 mL) containing 3-(trifluoromethyl)-acetophenone (20
mL) as an internal standard and demineralised water (390–
840mL, final total volume 1 mL). The kinetic resolution
(278C, under vigorous magnetic stirring) was followed by
taking aliquots from the reaction medium (1–10 mL) imme-
diately mixed with ethyl acetate (500 mL) to stop the enzy-
matic reaction and to extract the remaining epoxide and the
formed diol. The enantiomeric excess of the remaining ep-
oxide and the conversion ratio were determined by chiral
GC analysis.
1
provide a colourless liquid; yield: 75%; H NMR (CDCl3):
d=2.78 (dd, 1H, J=2.5 Hz, J=5.5 Hz), 3.18 (dd, 1H, J=
4.0 Hz, J=55 Hz), 3.89 (dd, 1H, J=2.5 Hz, J=40 Hz), 7.37
(d, 2H, J=8.2 Hz), 7.64 (d, 2H, J=8.2 Hz); 13C NMR
(DMSO-d6): d=66.9 (CH2), 73.1 (CH), 124.0 (q, J=1.9 Hz,
C), 127.9 (CH), 129.6 (q, JCF =307.6 Hz, C), 135.7 (CH),
147.4 (C); 19F NMR (CDCl3): d=À43.27; exact mass calcd.
for C9H7F3OS: 219.0091; found: 219.0100; anal. calcd. for
C9H7F3OS: C 49.09, H 3.20, F 25.88, S 14.56; found: C 49.41,
H 3.22, F 27.11, S 13.38. Chiral GC analysis [column I, T=
1308C: (R)=7.2 min; (S)=7.6 min].
2-Methyl-2-(2-trifluoromethylphenyl)-oxirane (6): This
epoxide (method I) was purified by flash chromatography
(pentane/ethyl acetate, 95:5) to provide a colourless liquid;
General Procedure for Semi-preparative
Biohydrolysis of 1–8 at 50g/L
yield: 75%; 13C NMR (CDCl3): d=24.6 (CH3), 55.3 (q, JCF
=
3.6 Hz, C), 57.6 (CH2), 125.1 (q, JCF =271.2, C), 126.8 (q,
Typically, 1 mL of an A. niger EH solution (12.5 mgmLÀ1 in
demineralised water) was added to rac-1–8 (1.25 g) mixed
into an emulsion of iso-octane (2.5 mL) and water
(21.5 mL). The reaction was performed at 278C. Aliquots (1
mL) were withdrawn at regular time intervals and immedi-
ately added to ethyl acetate (40 mL). After extraction the
enantiomeric excess of residual epoxide was determined by
chiral GC chromatography until ee >97%. Then the enzy-
matic reaction was stopped and the remaining epoxide was
extracted by addition of ethyl acetate (30 mL). After de-
cantation the water phase was extracted two times using
ethyl acetate (2”50 mL) then the collected organic fractions
were washed with brine (30 mL) before drying over MgSO4.
J
CF =5.3 Hz, CH), 127.8 (q, JCF =30.9 Hz, C), 128.6 (CH),
129.9 (CH), 132.8 (CH), 140.6 (m, C); 19F NMR (CDCl3):
d=À62.24. Chiral GC analysis [column I, T=908C: (R)=
10.4 min; (S)=10.9 min].
2-Methyl-2-(3-trifluoromethylphenyl)-oxirane (7): This
epoxide was obtained (method II) as a colourless liquid;
yield: 99%; bp 1178C (39 mbar); H NMR (CDCl3): d=1.55
1
(s, 3H), 2.75 (d, 1H, JHH =5.4 Hz), 2.98 (d, 1H, J=5.4 Hz),
7.40–7.59 (m, 4H); 13C NMR (CDCl3): d=21.5 (CH3), 56.2
(CH2), 57.0 (C), 122.2 (q, JCF =3.8 Hz, CH), 124.1 (q, JCF
=
272.5 Hz, C), 124.3 (q, JCF =3.8 Hz, CH), 128.7 (CH), 128.9
(CH), 130.8 (q, JCF =32.1 Hz, C), 142.4 (C); 19F NMR
(CDCl3): d=À62.27; exact mass calcd. for C10H9F3O:
1414
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
The ethyl acetate was removed under atmospheric pressure
and the residue was subjected to flash chromatography
(silica gel: hexane/ethyl acetate, gradient). The isolated re-
maining epoxide and formed diol were further purified by
bulb-to-bulb distillation. The H NMR spectra of the enan-
tiopure epoxides 1–8 were identical to those of the race-
mates.
120.8 (C), 130.0 (q, JCF =318.4 Hz, C), 127.9 (CH), 135.8
(CH), 147.4 (C); 19F NMR (CDCl3): d=À42.22; exact mass
calcd. for C9H9F3O2S: 237.0197; found: 237.0190; anal. calcd.
for C9H9F3O2S: C 48.35, H 3.81, F 23.92, S 13.46; found: C
46.04, H 3.68, F 24.62, S 13.69. Chiral GC analysis after deri-
vatisation to the corresponding acetonide [column I, T=
1408C: (R)=10.9 min; (S)=12.7 min].
1
(S)-7: (ee 98.3%); [a]2D2: +8.3 (c 1, CHCl3).
All diols were isolated as white solids at room tempera-
ture.
(R)-7d: (ee 59%); [a]D22: À5.9 (c 1.25, CHCl3); mp 488C;
1H NMR (CDCl3): d=1.52 (s, 3H), 2.37 (m, 1H), 2.99 (m,
1H), 3.69 (m, 2H), 7.43–7.73 (m, 4H); 13C NMR (CDCl3):
d=26.0 (CH3), 70.7 (CH2), 74.7 (C), 122.1 (q, JCF =3.8 Hz,
CH), 124.0 (q, JCF =3.8 Hz, CH), 124.2 (q, JCF =272.4 Hz,
C), 128.6 (CH), 128.8 (CH), 130.7 (q, JCF =32.1 Hz, C),
142.4 (C); 19F NMR (CDCl3): d=À61.98; exact mass calcd.
for C10H11F3O2: 219.0633; found: 219.0630; anal. calcd. for
C10H11F3O2: C 54.55, H 5.04, F 25.88; found: C 53.57, H
5.03, F 25.68. Chiral GC analysis after derivatisation to the
corresponding acetonide [column I, T=1208C: (S)=
9.0 min; (R)=9.6 min].
(S)-1: (ee 97.9%); [a]2D2: +62.4 (c 0.89, CHCl3).
(R)-1d: (ee 77.3%); [a]2D2: À47.9 (c 0.98, CHCl3); mp
1
518C; H NMR (CDCl3): d=2.45 (s, 1H), 3.00 (s, 1H), 3.67
(m, 2H), 5.23 (m, 1H) 7.44–7.65 (m, 4H); 13C NMR
(CDCl3): d=67.8 (CH2), 74.4 (CH), 125.6 (q, JCF =5.8 Hz,
CH), 124.2 (q, JCF =273.7 Hz, C), 127.3 (q, JCF =30.3 Hz, C),
128.0 (CH), 128.2 (m, CH), 132.2 (m, CH), 139.2 (C);
19F NMR (CDCl3): d=À57.93; exact mass calcd. for
C9H9F3O2: 205.0476; found: 205.0468. Chiral GC analysis
after derivatisation to the corresponding dimethoxy ether
[column I, T=808C: (S)=22.0 min; (R)=22.6 min].
(S)-2: (ee 98.7%); [a]2D2: +9.1 (c 0.92, CHCl3), Lit.[25] [a]2D2:
+ 2.7 (c 0.92, CHCl3).
(S)-8: (ee 99.1%); [a]2D2: +16.7 (c 0.89, CHCl3).
(R)-8d (ee 88.3 %); [a]D22: À9.4 (c 1.03, CHCl3); mp 598C;
1H NMR (CDCl3): d=1.55 (s, 3H), 1.84 (dd, 1H, J=5 Hz,
J=7.25 Hz), 2.68 (s, 1H), 3.74 (ddd, 2H, J=11 Hz, J=5 Hz,
J=7.25 Hz), 7.58 (d, 2H, J=8.75 Hz), 7.63 (d, 2H, J=
8.75 Hz); 13C NMR (acetone-d6): d=26.4 (CH3), 71.5 (CH2),
750 (C), 125.4 (q, JCF =3.8 Hz, CH), 125.6 (q, JCF =269.6 Hz,
C), 127.1 (CH), 128.3 (q, JCF =32.1 Hz, C), 152.6 (C);
19F NMR (CDCl3): d=À61.98; exact mass calcd. for
C10H11F3O2: 219.0633; found: 219.0614. Chiral GC analysis
after derivatisation to the corresponding acetonide [column
I, T=1308C: (S)=8.1 min; (R)=8.6 min].
1
(R)-2d: (ee 13.2 %); [a]22: À5.7 (c 0.98, CHCl3); H NMR
(CDCl3): d=2.26 (m, 1H)D, 2.87 (m, 1H), 3.57 (dd, 1H, J=
11 Hz, J=8.25 Hz), 3.73 (dd, 1H, J=11 Hz, J=3.25 Hz),
4.81 (dd, 1H, J=3.25 Hz, J=8.25 Hz) 7.44–7.65 (m, 4H);
13C NMR (CDCl3): d=67.9 (CH2), 74.0 (CH), 122.8 (q, JCF
3.8 Hz, CH), 124 (q, JCF =3.7 Hz, CH), 124.8 (q, JCF
272.4 Hz, C), 129.0 (CH), 129.4 (CH), 130.9 (q, JCF
=
=
=
32.3 Hz, C), 139.4 (C). Chiral GC analysis after derivatisa-
tion to the corresponding acetonide [column I, T=1408C:
(S)=5.3 min; (R)=5.8 min).
(S)-3: (ee 97.9%); [a]2D2: +18 (c 1.13, CHCl3).
(R)-3d: (ee 84.3%); [a]2D2: À39.3 (c 1.03, CHCl3), Lit.[26]
Preparative Scale Biohydrolysis of rac-4 at 250 gLÀ1
in a Standardised STR
1
[a]2D2: À57 (c 1, CHCl3); mp 1018C; H NMR (CDCl3): d=
2.07 (m, 1H), 2.70 (m, 1H), 3.88 (m, 2H), 4.89 (m, 1H),
7.50 (d, 2H, J=8.5 Hz), 7.63 (d, 2H, J=8.5 Hz); 13C NMR
In a standardised STR (100 mL total volume) equipped with
a rotatory Rushton turbine and thermostated at 278C, rac-4
(12.5 g, 61.3 mmol) was added in iso-octane (10 mL) and
demineralised water (20 mL). An emulsion was formed by
agitation at 800 rpm before careful addition of the enzyme
(62.5 mg, containing 688 U,[19] dissolved in 7.5 mL of demin-
eralised water). The kinetic resolution was followed by
taking aliquots from the reaction medium (100 mL) which
were immediately mixed with a mixture of acetonitrile (300
mL) and iso-octane (1 mL) to stop the enzymatic reaction
and to extract the remaining epoxide. After vigorous vortex-
ing of samples (30 s) then centrifugation (13.000 rpm, 2 min)
the organic phase was analysed using a chiral GC column to
determine the enantiomeric excess of the remaining epox-
ide. The reaction was stopped when the ee of the residual
epoxide reached 99% (210 min) by adding AcOEt (30 mL)
directly into the reactor. After decantation the aqueous
phase was extracted twice with ethyl acetate (2”30 mL).
The collected organic fractions were washed with brine then
dried over MgSO4 and concentrated in vacuum. Using these
experimental conditions a total amount of 50 g of rac-4 was
processed in four experiments of 12.5 g each leading to a
total amount of 64.9 g of crude extract. Addition of hexane
to this crude extract allowed us to obtain the crystallization
of a part of the formed diol-4d. After filtration 24.4 g of
diol-4d were isolated as white crystals. The filtrate was con-
(acetone-d6): d=68.6 (CH2), 74.7 (CH), 125.5 (q, JCF
=
269.4 Hz, C), 125.6 (q, JCF =3.8 Hz, CH), 127.8 (CH), 129.5
(q, JCF =31.9 Hz, C), 148.5 (C); 19F NMR (acetone-d6): d=
À62.21; exact mass calcd. for C9H9F3O2: 205.0476; found:
205.0457. Chiral GC analysis after derivatisation to the cor-
responding acetonide [column I, T=1408C: (S)=5.5 min;
(R)=6.1 min].
(S)-4: (ee 98.6%); [a]2D2: + 13.7 (c 1.58, CHCl3).
(R)-4d: (ee 94.5%); [a]2D2: À41.5 (c 1.04, CHCl3); mp
648C; 1H NMR (CDCl3): d=2.13 (m, 1H), 2.69 (m, 1H),
3.70 (m, 2H), 4.84 (m, 1H), 7.21 (d, 2H, J=8.75 Hz), 7.41
(d, 2H, J=8.5 Hz); 13C NMR (acetone-d6): d=68.7 (CH2),
74.5 (CH), 121.4 (CH), 121.5 (q, JCF =253.3 Hz, C), 128.8
(CH), 143.1 (C), 148.9 (C); 19F NMR (CDCl3): d=À57.61;
exact mass calcd. for C9H9F3O3: 221.0426; found: 205.0457;
anal. calcd. for C9H9F3O3: C 48.66, H 4.08, F 25.65; found: C
48.79, H 4.08, F 26.41. Chiral GC analysis after derivatisa-
tion to the corresponding acetonide [column I, T=1408C:
(S)=5.5 min; (R)=6.1 min].
(S)-5: (ee 98.5%); [a]2D2: + 21.1 (c 1.2, CHCl3).
(R)-5d: (ee 85%): [a]D22: À5 (c 1.18, CHCl3); mp 728C;
1H NMR (DMSO-d6): d=3.46 (m, 2H), 4.61 (dd, 1H, J=
4.25 Hz, J=5.75 Hz), 4.83 (dd, 1H, J=5.75 Hz, J=5.75 Hz),
5.47 (d, 1H, J=4.25 Hz), 7.51 (d, 2H, J=8 Hz), 7.67 (d, 2H,
J=8 Hz); 13C NMR (DMSO-d6): d=67.9 (CH2), 74.7 (CH),
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1415

Justine Deregnaucourt et al.
FULL PAPERS
centrated in vacuum then a part of the remaining epoxide 4 References
(22.1 g) was isolated by distillation (758C, 10 mbar). Addi-
tion of hexane to the residue of distillation led to the crys-
tallization of a second part of diol-4d (4.7 g). After filtration
the filtrate was concentrated and the residue was finally sub-
jected to flash chromatography (silica gel: pentane/ethyl
acetate, 90/10) which gave 4.9 g of 4. Altogether, extraction
and purification of 4 and 4d provided as a pool 27.0 g of (S)-
4 (ee 99%, 43% yield) and 29.1 g of (R)-4d (ee 95%, 44%
yield).
[1] a) A. Archelas, R. Furstoss, Cur. Opin. Chem. Biol.
2001, 5, 112–119; b) E. J. de Vries, D. B. Janssen, Curr.
Opin. Biotechnol. 2003, 14, 414–420; c) W. C. Choi,
C. Y. Choi, Biotechnol. Bioprocess Eng. 2005, 10, 167–
179.
[2] a) D. Belder, M. Ludwig, L.-W. Wang, M. T. Reetz,
Angew. Chem. Int. Ed. 2006, 45, 2463–2466; b) R.
Rink, J. H. Lutje Spelberg, R. J. Pieters, J. Kingma, M.
Nardini, R. M. Kellogg, B. W. Dijkstra, D. B. Janssen, J.
Am. Chem. Soc. 1999, 121, 7417–7418; c) M. T. Reetz,
C. Torre, A. Eipper, R. Lohmer, M. Hermes, B. Brun-
ner, A. Maichele, M. Bocola, M. Arand, A. Cronin, Y.
Genzel, A. Archelas, R. Furstoss, Org. Lett. 2004, 6,
177–180.
[3] a) Fluka, ref. 71832; b) M. Cleij, A. Archelas, R. Fur-
stoss, Tetrahedron: Asymmetry 1998, 9, 1839; c) Y.
Genzel, A. Archelas, Q. B. Broxterman, B. Schulze, R.
Furstoss, J. Org. Chem. 2001, 66, 538–543.
[4] A. Liese, K. Seelbach, C. Wandrey, Industrial Biotrans-
formation, Wiley-VCH, Weinheim, 2000, 57–64.
[5] C. Morisseau, H. Nellaiah, A. Archelas, R. Furstoss,
J. C. Baratti. , Enzyme Microb. Technol. 1997, 20, 446–
452.
[6] A. Goswami, M. J. Totleben, A. K. Singh, R. N. Patel,
Tetrahedron: Asymmetry 1999, 10, 3167–3175.
[7] W. J. Choi, C. Y. Choi, J. A. M. de Bont, G. A. G. M.
Weijers, Appl. Microbiol. Biotechnol 1999, 53, 7–11.
[8] H. Hellstrçm, A. Steinreiber, S. F. Mayer, K. Faber,
Biotechnol. Lett. 2001, 23, 169–173.
[9] a) H. Nellaiah, C. Morisseau, A. Archelas, R. Furstoss,
J. Baratti, Biotechnol. Bioeng. 1996, 49, 70–77; b) C. Li,
Q. Liu, X. Song, D. Ding, A. Ji, Y. Qu, Biotechnol.
Lett. 2003, 25, 2113–2116; c) J. Lotter, A. L. Botes,
M. S. van Dyk, J. C. Breytenbach, Biotechnol. Lett.
2004, 26, 1191–1195.
[10] The term “global volume substrate concentration” re-
lates to the amount of substrate present in the total re-
actor volume and therefore includes both the dissolved
and non dissolved substrate, see: B. Doum›che, A. Ar-
chelas, R. Furstoss, Adv. Synth. Catal. 2006, 348, 1948–
1957.
[11] a) K. M. Manoj, A. Archelas, J. Baratti, R. Furstoss,
Tetrahedron 2001, 57, 695–701; b) N. Monfort, A. Ar-
chelas, R. Furstoss, Tetrahedron 2004, 60, 601–604.
[12] W. J. Choi, C. Y. Choi, J. A. M. de Bont, G. A. G. M.
Weijers, Appl. Microbiol. Biotechnol. 2000, 54, 641–
646.
Preparative Scale Biohydrolysis of rac-8 at 100 gLÀ1
in a Standardised STR
In a standardised STR (500mL total volume) equipped with
a rotatory Rushton turbine and thermostated at 278C, rac-8
(25g, 123.8mmol) was added in iso-octane (25mL) and
demineralised water (112.5mL). An emulsion was formed
by agitation at 600 rpm before addition of a crude enzymatic
solution (112.5mL containing 5600 U[19]). The reaction was
stopped when the ee of the residual epoxide reached 99 %
(150min) by adding AcOEt (60mL) directly into the reac-
tor. After decantation the aqueous phase was extracted with
ethyl acetate (2”60mL). The collected organic fractions
were washed with brine then dried over MgSO4 and concen-
trated in vacuum. Separation by flash-chromatography (pen-
tane/Et2O, 80/20 then 0/100) afforded the (S)-8 epoxide
(10.5g, 42% yield, ee 99.7%) and the (R)-8d diol (13.41g,
49% yield, ee 78.5%).
General Procedure for Resolution of rac-1–8 using
Jacobsenꢀs Catalyst
The (R,R)-(salen)Co
N
N
the (R,R)-(salen)Co(II) precatalyst commercially available
from Aldrich. The best results were obtained when a mix-
ture of (R,R)-(salen)Co(II) (202 mg, 0.332 mmol), CH2Cl2
(1.66 mL), and acetic acid (66 mL, 1.33 mmol) was stirred
while open to the air for 10 min at room temperature.[22]
The solvent was removed by rotary evaporation, and the
brown residue was dried under vacuum. In a mini-vial
(1 mL) rac-1–8 (1 mmol) were added to Co(III) complex
C
(4.76 mg, 0.007 mmol) previously prepared. Water (9.9 mL,
0.55 mmol) was added and the reaction mixture was stirred
at room temperature for 48 h. The remaining epoxide was
extracted with iso-octane (10 mL) containing 3-(trifluoro-
methyl)-acetophenone (0.5 gLÀ1) as a standard and its enan-
tiomeric excess and the conversion ratio were determined
by chiral GC analysis.
[13] C. C. C. R. de Carvalho, F. van Keulen, M. R. da Fonse-
ca, Biocatal. Biotransform. 2000, 18, 223–235.
[14] a) P.-F. Gong, J.-H. Xu, Enzyme Microb. Technol. 2005,
36, 252–257; b) Y. Simeo, K. Faber, Tetrahedron:
Asymmetry 2006, 17, 402–409.
[15] a) H. Baldascini, K. J. Ganzeveld, D. Janssen,
A. A. C. M. Beenackers, Biotechnol. Bioeng. 2001, 73,
44–54; b) H. Baldascini, D. Janssen, Enzyme Microb.
Technol. 2005, 36, 285–293.
Acknowledgements
This work is part of the Ph.D. thesis of J. Deregnaucourt.
The CNRS and the Rhodia Research are greatly acknowl-
edged for their financial support. The authors thank Pascal
Pitiot (Rhodia Research) for his grateful contribution to the
agitation rate optimisation part of the study.
[16] J. Deregnaucourt, A. Archelas, F. Barbirato, J.-M.
Paris, R. Furstoss, Adv. Synth. Catal. 2006, 348, 1165–
1169.
1416
asc.wiley-vch.de
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Adv. Synth. Catal. 2007, 349, 1405 – 1417

Liquid-Liquid Phase Aspergillus niger Epoxide Hydrolase-Catalysed Resolution
FULL PAPERS
[17] E. J. Corey, M. Chaykovsky, J. Am. Chem. Soc. 1965,
87, 1353–1364.
[22] a) E. N. Jacobsen, M. Tokunaga, J. F. Larrow, US
Patent 6,262,278 B1, 2001; b) J. F. Larrow, K. E. Hem-
berger, S. Jasmin, H. Kabir, P. Morel Tetrahedron:
Asymmetry 2003, 14, 3589–3592; c) J. F. Larrow, E. N.
Jacobsen Topics Organomet. Chem. 2004, 6, 123–152.
[23] H.-U. Blaser, Chem. Commun. 2003, 293–296.
[24] D. L Nelson, M. M. Cox, Lehningerꢀs Principles of Bio-
chemistry, 3rd edn., Worth Publishers, New York, 2000,
p 263.
[18] These experiments were all conducted using a partly
purified epoxide hydrolase containing about 25% pure
enzyme, and showing an activity of 11 units against sty-
rene oxide, as measured following the UV spectroscop-
ic test published previously: See ref.[19]
[19] C. Mateo, A. Archelas, R. Furstoss, Anal. Biochem.
2003, 314, 135–141.
[20] S. Chen, Y. Fujimoto, G. Girdaukas, C. J. Sih, J. Am.
Chem. Soc. 1982, 104, 7294–7299.
[21] a) B. E. Smart, J. Fluorine Chem. 2001, 109, 3–11;
[25] M. J. Ferris, European Patent EP 40,000, 1981.
[26] K. Hirose, K. Ogasahara, K. Nishioka, Y. Tobe, K. J.
Naemura, J. Chem. Soc., Perkin Trans. 2 2000, 1984–
1993.
b) J. D. Dunitz, ChemBioChem. 2004, 5, 614–621.
Adv. Synth. Catal. 2007, 349, 1405 – 1417
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
asc.wiley-vch.de
1417