An efficient method for the lipase-catalysed resolution and in-line purification of racemic flurbiprofen in a continuous-flow reactor

Article abstract of DOI:10.1016/j.molcatb.2012.02.008

The lipase-catalysed kinetic resolution of flurbiprofen was performed in a flow-chemistry reactor allowing for a significant reduction of the reaction time compared to the classical batch method. The process was implemented by adding an in-line purification step of the exiting solution, consisting in a catch and release protocol, which allows easy separation and recovery of both (S)-flurbiprofen and (R)-flurbiprofen butyl ester with an enantiomeric excess ≥90% and a chemical purity >98%.

Full text of DOI:10.1016/j.molcatb.2012.02.008

Journal of Molecular Catalysis B: Enzymatic 84 (2012) 78–82
Contents lists available at SciVerse ScienceDirect
Journal of Molecular Catalysis B: Enzymatic
journal homepage: www.elsevier.com/locate/molcatb
An efficient method for the lipase-catalysed resolution and in-line purification of
racemic flurbiprofen in a continuous-flow reactor
Lucia Tamborinia, Diego Romanob, Andrea Pintoa, Arianna Bertolania,b, Francesco Molinarib, Paola Contia,∗
a Dipartimento di Scienze Farmaceutiche ‘Pietro Pratesi’, Universitá degli Studi di Milano, via Mangiagalli 25, 20133 Milano, Italy
b Dipartimento di Scienze e Tecnologie Alimentari e Microbiologiche, Sezione Microbiologia Industriale, Universitá degli Studi di Milano, via Mangiagalli 25, 20133 Milano, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
Available online 10 March 2012
The lipase-catalysed kinetic resolution of flurbiprofen was performed in a flow-chemistry reactor allow-
ing for a significant reduction of the reaction time compared to the classical batch method. The process
was implemented by adding an in-line purification step of the exiting solution, consisting in a catch and
release protocol, which allows easy separation and recovery of both (S)-flurbiprofen and (R)-flurbiprofen
butyl ester with an enantiomeric excess ≥90% and a chemical purity >98%.
Keywords:
Enzymatic resolution
Flow reactor
Lipase
© 2012 Elsevier B.V. All rights reserved.
Flurbiprofen
Catch and release
1. Introduction
resolutions in batch, such as long reaction times, product inhibition,
and scalability.
Flurbiprofen is a non-steroidal anti-inflammatory drug (NSAID).
Cyclooxygenase (Cox) inhibiting activity, that is responsi-
ble for reducing the production of inflammation-mediating
prostaglandins, thereby providing relief from inflammation [1],
resides primarily in the (S)-enantiomer. On the other hand, (R)-
flurbiprofen (also known as Ansaid or generically as Tarenflurbil)
shows anticancer effects in vivo and in vitro [2,3]. In fact, some
recent researches demonstrate that (R)-flurbiprofen inhibits pro-
gression of prostate cancer and colon cancer in mice [4–6] and
inhibits survival of colon and prostate cancer cell lines in vitro
[7,8]. The precise mechanism by which it exerts anticancer effects
is still under investigation. At clinically relevant concentrations,
bioconversion to the Cox-inhibiting S-enantiomer is inefficient, so
(R)-flurbiprofen might not cause the gastric irritation associated
with other NSAIDs. For these reasons, (R)-flurbiprofen is currently
in clinical trials for the treatment of metastatic prostate cancer.
Although several studies concerning the biocatalysed resolution
of racemic flurbiprofen have been reported in recent years [9–13],
there is still the need to find a strategy that could lead to a fast,
automated, and easily scalable way to obtain both enantiomers of
flurbiprofen. In this contest, the use of a continuous flow reactor
could help to overcome some of the limitations of traditional kinetic
In the present study, we report the kinetic resolution of (R,S)-
flurbiprofen with an immobilised lipase B from Candida antarctica
(Novozym 435®) in a continuous flow reactor [14]. Novozym 435®
has been reported to exhibit a relatively high enantiopreference
toward (R)-flurbiprofen and it has been applied to the flurbipro-
fen resolution via either esterification or transesterification carried
out in organic media in batch [15–18]. Exploiting the advantages
of using a flow chemistry reactor, we focused our attention on
the influence of parameters such as temperature, residence time
and stoichiometry on the reactivity and selectivity of the lipase-
catalysed kinetic resolution with the aim to define the best reaction
conditions to obtain each enantiomer of flurbiprofen with a high
degree of enantiomeric purity.
2. Materials and methods
2.1. Materials
(R,S)-Flurbiprofen, solvents, and lipase B from C. antarctica
(Novozym 435®, immobilised on acrylic resin; Sigma, L4777) were
purchased from Sigma–Aldrich.
HPLC analyses were performed with a Jasco PU-980 pump
equipped with a UV–vis detector Jasco UV-975 (wavelength:
254 nm). Column: Lux Amylose-2, 4.60 mm i.d. × 150 mm, Phe-
nomenex. Eluent: acetonitrile:water:formic acid (1:1:0.2, v/v/v).
Flow: 1.0 mL/min. The reaction solvent was evaporated and the
HPLC sample was prepared in acetonitrile (10 mM).
∗
Corresponding author. Tel.: +39 0250319329; fax: +39 0250319326.
E-mail address: paola.conti@unimi.it (P. Conti).
1381-1177/$ – see front matter © 2012 Elsevier B.V. All rights reserved.
doi:10.1016/j.molcatb.2012.02.008

L. Tamborini et al. / Journal of Molecular Catalysis B: Enzymatic 84 (2012) 78–82
79
The continuous-flow biotranformations were performed using
2.4. Catch and release procedure
a R2+/R4 flow reactor, commercially available from Vapourtec [14]
equipped with Omnifit glass column (10 mm i.d. × 100 mm length).
(a) The solution exiting from the column filled with Novozym 435®
was flowed, at the same flow rate, through a second glass col-
umn (10 mm i.d. × 100 mm length) filled with Amberlyst A21
(1.5 g). A 100 psi back-pressure regulator was applied to the
system. The solution outflowing from the second column was
evaporated recovering (R)-flurbiprofen butyl ester in a >98%
chemical purity.
(b) A solution of 5% acetic acid in toluene was flowed (flow rate:
250 ␮L/min) through the column containing Amberlyst A21 to
release the unreacted (S)-flurbiprofen, which was recovered
after evaporation of the solvent. A 100 psi back-pressure reg-
ulator was applied to the system.
2.2. Resolution of (R,S)-flurbiprofen in batch [18]
Racemic flurbiprofen 1 (10 mg, 0.04 mmol), 10.8 ␮L of n-butanol
(0.12 mmol, 3 equiv), Novozym 435® (50 mg), and molecular sieves
(50 mg) were suspended in toluene (2 mL) and stirred at 40 ◦C for
6 h (Scheme 1). The reaction was monitored by chiral HPLC, after
sample withdrawing (50 ␮L), filtration, evaporation of the solvent
and dissolution in acetonitrile (100 ␮L). The single enantiomers of
the acid 1 and the single enantiomers of the butyl ester 2 were
simultaneously separated in one HPLC run.
(R)-Flurbiprofen 1: retention time 4.3 min.
3. Results and discussion
(S)-Flurbiprofen 1: retention time 5.7 min.
(R)-Flurbiprofen butyl ester 2: retention time 29.8 min.
(S)-Flurbiprofen butyl ester 2: retention time 37.8 min.
The absolute configuration of the enantiomers of flurbiprofen
was determined by comparison with commercial standards of the
optically pure compounds; the absolute configuration of the butyl
ester was determined by comparison with the optical rotation of
authentic samples of the optically pure enantiomers obtained by
esterification of optically pure (S)-flurbiprofen and (R)-flurbiprofen
with n-butanol by conventional esterification procedures [19].
The enantiomeric ratio (E) was calculated as reported in the lit-
erature [20]. The specific reaction rate (rbatch) was calculated from
the amount of the product (nP (␮mol)), the reaction time (t (min)),
and the mass of biocatalyst employed (mE (g)) according to Eq. (1)
[21].
3.1. Resolution of (R,S)-flurbiprofen in batch
We first performed the kinetic resolution of (R,S)-flurbiprofen
with immobilised lipase B from C. antarctica (Novozym 435®),
Scheme 1 in batch, following the procedure recently reported by
Ghanem [18]. After 6 h, we evaluated the enantiomeric excess of
the substrate (ees = 52%), the enantiomeric excess of the product
(eep = 86%) and the molar conversion (c = 38%), corresponding to
E = 22 and a specific reaction rate (rbatch) of 0.84 ␮mol/min g.
3.2. Resolution of (R,S)-flurbiprofen in flow
3.2.1. Temperature and flow rate optimisation
Firstly, flow experiments were focused on reaching a degree of
conversion similar to the one obtained in the batch process. For
this purpose, a 20 mM solution of (R,S)-flurbiprofen 1 in toluene
containing 3 equiv of n-BuOH was flowed through a glass column
(10 mm i.d. × 100 mm length) loaded with 250 mg of Novozym 435
and 250 mg of molecular sieves. Temperature and flow rate, which
sets the residence time, were varied and the results are reported in
Table 1.
nP
rbatch
=
(␮mol/ min g)
(1)
t × mE
2.3. Resolution of (R,S)-flurbiprofen in flow
Experiments reported in Table 1 were carried out as follows:
10 mg of racemic flurbiprofen 1 (0.04 mmol), 10.8 ␮L of n-butanol
(0.12 mmol, 3 equiv) were dissolved in toluene (2 mL) and the
solution was flowed through a glass column (10 mm i.d. × 100 mm
length) filled with Novozym 435® (250 mg) and molecular sieves
(250 mg). A 100 psi back-pressure regulator was applied to the sys-
tem. Temperature and flow rate were varied and their values are
reported in Table 1.
After collecting a total volume of 4 mL, the reaction outcome
was analyzed by chiral HPLC. To this aim, a sample of 100 ␮L was
withdrawn, evaporated and re-dissolved in acetonitrile (100 ␮L).
Experiments reported in Tables 2 and 3 were carried using the
same reactor set-up and following the procedure described above,
varying the concentration of reagents and the amount of catalyst
and molecular sieves, according to the data reported in the tables.
Due to the use of a large excess of molecular sieves, which are
expected to trap the produced water, this reaction can be con-
sidered quasi-irreversible; therefore, the enantiomeric ratio (E)
was calculated from two of the following three properties: the
conversion (c), the enantiomeric excess of the product (eep) and
enantiomeric excess of the remaining substrate (ees) [20].
The specific reaction rate (rflow) was calculated from the con-
centration of the product ([P] (␮mol/mL)), flow rate (f (mL/min)),
and the mass of biocatalyst employed (mE (g)) according to Eq. (2)
[21].
The data reported in Table 1 indicate that the use of the flow
reactor dramatically reduced the reaction time. In fact, running
the reaction at 40 ◦C, the degree of conversion was 37% and the
ester 2 was obtained with 88% ee (E = 26), in about 40 min (entry 2).
When temperature was increased to 60 ◦C, a further increase of the
reaction rate was achieved, while maintaining the same degree of
conversion and enantiomeric excess: the ester 2 was formed with
a 35% conversion and 90% ee (E = 30) in just 15 min (entry 9). On
the other side, using a temperature of 60 ◦C and a reaction time of
40 min (entry 4) resulted in a marked increase of the conversion
rate (64%) allowing, in this case, the obtainment of (S)-flurbiprofen
with a good enantiomeric purity (92% ee). By simply modulating
the flow rate, the present methodology allows the production of
either (S)-flurbiprofen or (R)-flurbiprofen butyl ester, with a high
degree of enantiomeric purity, according to the desired outcome of
a kinetic resolution.
3.2.2. Reaction stoichiometry optimisation
In an attempt to optimise the process, also in view of a possible
reaction scale-up, we evaluated the influence of the stoichiometry
on the reaction outcome in terms of conversion and enantiomeric
purity of the product. Operating at the optimised temperature and
flow-rate conditions previously identified (Table 1, entry 9), we
started by modifying the molar ratio flurbiprofen:n-BuOH from 1:1
to 1:5 (Table 2). No significant differences in the ee of the product
were observed in the five experiments, while the best conversion
was obtained using a molar ratio flurbiprofen:n-BuOH = 1:3. Inter-
estingly, this is the same stoichiometric ratio reported for the batch
[P] × f
rflow
=
(␮mol/min g)
(2)
mE

80
L. Tamborini et al. / Journal of Molecular Catalysis B: Enzymatic 84 (2012) 78–82
Scheme 1. Enantioselective esterification of (R,S)-flurbiprofen 1 with n-BuOH catalysed by Novozym 435 with the addition of molecular sieves, in toluene.
Table 1
Screening of different flow rates and temperatures in the enantioselective esterification of (R,S)-flurbiprofen 1 with n-BuOH catalysed by Novozym 435 in a continuous flow
reactor.
Entry
Residence time (min)
Temp. (◦C)
Molar conversion (%)a
ees (%)b
eep (%)b
Ea
rc (␮mol/min g)
[flow-rate (␮L/min)]
1
2
3
4
5
6
7
8
9
60 [18]
40 [27]
40 [27]
40 [27]
30 [36]
30 [36]
30 [36]
15 [72]
15 [72]
40
40
50
60
40
50
60
50
60
46
37
47
64
37
46
51
22
35
66
52
74
92
52
74
64
26
48
78
88
84
52
88
88
61
90
90
15
26
25
9.7
26
34
7.8
24
30
0.7
0.8
1.0
1.4
1.1
1.3
1.5
1.3
2.0
Reaction conditions: 20 mM solution of (R,S)-flurbiprofen 1 in toluene, 3 equiv of n-BuOH, 250 mg of Novozym 435® and 250 mg of molecular sieves.
a
Calculated according to Ref. [20].
Determined by chiral HPLC.
Calculated according to Eq. (2) (Ref. [21]).
b
c
Table 2
Screening of different equivalents of n-BuOH in the enantioselective esterification of (R,S)-flurbiprofen 1 with n-BuOH catalysed by Novozym 435 in a continuous flow reactor.
Equivalents of n-BuOH
Molar conversion (%)a
ees (%)b
eep (%)b
Ea
rc (␮mol/min g)
1
2
3
4
5
33
34
35
37
40
44
46
48
52
58
90
90
90
88
86
29
29
30
26
23
1.9
2.0
2.0
2.1
2.3
Reaction conditions: 20 mM solution of (R,S)-flurbiprofen 1 in toluene, n-BuOH, 250 mg of Novozym 435® and 250 mg of molecular sieves. Flow rate: 72 ␮L/min. T = 60 ◦C.
a
Calculated according to Ref. [20].
Determined by chiral HPLC.
Calculated according to Eq. (2) (Ref. [21]).
b
c
process [18] and matches the one used in our initial experiments
(Table 1). Consequently, this parameter was kept constant in all the
following experiments.
Subsequently, we evaluated the effect of increasing substrate
concentration of the starting solution and we defined the amount
of polymer-supported enzyme necessary to convert increasing con-
centration of starting material with similar performances in terms
of enantiomeric excess and conversion (Table 3).
The proportional increase of substrate and enzyme concentra-
tions (entries 1, 3 and 4), using the same residence time, resulted
in similar results in terms of conversion and enantioselectivity;
when substrate concentration was doubled, but enzyme amount
and residence time were kept constant, the conversion decreased
from 35% to 25% (entry 1 vs. entry 2) and the enantiomeric ratio
It is important to underline that all the experiments described
so far were carried out repetitively using the same column with-
out refilling the immobilised enzyme, which maintained its activity
unchanged over the time. The reproducibility of the enzyme per-
formance over the time was assessed by periodically repeating the
best experiment (Table 1, entry 9) and verifying the matching of
the results obtained both in term of conversion and enantiomeric
purity of substrate and product.
Table 3
Screening of different concentrations of (R,S)-flurbiprofen 1 in the enantioselective esterification with n-BuOH (3 equiv) catalysed by Novozym 435 in a continuous flow
reactor at 60 ◦C.
Entry
[1] (mM)
Mass of enzyme (g) + mass
of molecular sieves (g)
Residence time (min)
[flow-rate (␮L/min)]
Molar conversion
(%)a
ees (%)b
eep (%)b
Ea
rc (␮mol/min g)
1
2
3
4
5
6
20
40
40
80
80
80
0.25 + 0.25
0.25 + 0.25
0.50 + 0.50
1.00 + 1.00
0.25 + 0.25
0.50 + 0.50
15 [72]
15 [72]
15 [146]
15 [292]
60 [18]
30 [72]
35
25
37
36
40
37
48
28
52
50
58
52
90
84
90
90
88
90
30
15
31
31
28
31
2.0
2.9
4.3
8.4
2.3
4.3
a
Calculated according to Ref. [20].
Determined by chiral HPLC.
Calculated according to Eq. (2) (Ref. [21]).
b
c

L. Tamborini et al. / Journal of Molecular Catalysis B: Enzymatic 84 (2012) 78–82
81
Scheme 2. Outline of the overall resolution and purification process.
(E) was halved. This set of experiments shows that it is possible to
convert increasing concentration of flurbiprofen in equal time and
with the same degree of enantiomer selectivity by proportionally
augmenting the amount of enzyme filled in the column. In all the
experiments the amount of molecular sieves was proportionally
increased in order to keep constant the ratio enzyme/molecular
sieves.
We then evaluated the effect of raising the concentration of
racemic flurbiprofen (from 20 to 80 mM), keeping constant the
amount of enzyme, but simultaneously increasing the residence
time. Interestingly, we noticed that we could convert a solution of
substrate four times more concentrated (80 mM), increasing four
times the residence time and using only 0.25 g of enzyme (Table 3,
entry 1 vs. 5). In summary, increased concentrations of flurbiprofen
could be efficiently converted provided that the amount of enzyme
or the residence time were proportionally augmented. For instance
an 80 mM solution of substrate can be converted with a compara-
ble outcome either using 1.00 g of enzyme and setting the residence
time to 15 min, or using 1/4 of the enzyme (0.25 g) and a residence
time 4 times longer (60 min). Worth noting, the same outcome can
also be obtained balancing the amount of enzyme and the resi-
dence time, i.e. an 80 mM solution can be converted using 0.50 g of
enzyme and a residence time of 30 min (Table 3, entry 6).
containing the polymer supported base Amberlyst A21 was added
down-stream in order to catch the unreacted carboxylic acid
(Scheme 2, step 1). In this way the exiting solution only contains
(R)-flurbiprofen butyl ester 2 with excellent chemical purity (>98%)
and good ee (90%). Subsequently the acid can be released from
Amberlyst A21 using a 5% solution of acetic acid in toluene at room
temperature (Scheme 2, step 2). By applying the suitable reaction
conditions the same set-up can be exploited to prepare chemically
and enantiomerically pure (S)-flurbiprofen: in fact, if the conditions
reported in Table 1, entry 4 are applied, (S)-flurbiprofen is obtained
in a chemical purity >98% and an enantiomeric excess of 92%.
4. Conclusions
The results reported in the present work have highlighted
the advantages of performing a lipase-catalysed kinetic resolu-
tion of flurbiprofen in a flow-chemistry reactor compared to the
traditional batch process. In particular, we observed a significant
reduction of the reaction time from 6 h (in batch) to 15–60 min
in flow, depending on the desired degree of conversion. To better
compare the results obtained for the batch and the flow reactions,
we calculated the specific reaction rate of the two processes at a
similar degree of conversion. Notably, for the batch reaction the
specific reaction rate was 0.84 ␮mol/min g, while, in a flow mode,
using the conditions reported in Table 3, entry 4, it was possible to
reach a significant increase of the productivity of about 10 times
(8.40 ␮mol/min g).
We also evaluated the scalability of the process up to 80 mM
concentration of the starting solution and we have assessed the
reproducibility of the enzyme performances upon several hours
of continuous work. Finally, the process was implemented by
adding an in-line purification step of the exiting solution, con-
sisting in a separation of products from substrates thanks to a
catch and release protocol. Worth noting, after the initial set-up of
the system, the obtainment of the pure products can be achieved
avoiding tedious extraction/separation techniques. In conclusion,
the present protocol allows the obtainment and easy recovery
of both (S)-flurbiprofen and (R)-flurbiprofen butyl ester with an
enantiomeric excess ≥90% and a chemical purity >98%, by sim-
ply modulating the reaction conditions (temperature and residence
time).
3.3. Process scale up feasibility evaluation
After having assessed the efficiency of the system upon expo-
sure to higher substrate concentrations, we finally evaluated its
stability upon longer time of continuous work. While all the exper-
iments described so far were carried out on a 2 mL scale, we now
scaled up the process using 20 mL of a 80 mM solution of racemic
flurbiprofen. Operating under the conditions reported in Table 3,
entry 6, the reactor was let to work continuously for 6 h. The out-
come of the reaction (in terms of degree of conversion, ees and eep)
was perfectly in line with the data reported in Table 3, entry 6.
3.4. Catch and release protocol
As a further implementation of this method we exploited a
catch and release strategy to separate the product of the biotrans-
formation from the unreacted substrate. To this aim a column

82
L. Tamborini et al. / Journal of Molecular Catalysis B: Enzymatic 84 (2012) 78–82
References
[10] G.-H. Shin, K.-W. Lee, Y.-H. Lee, J. Mol. Catal. B: Enzyme 37 (2005) 109–111.
[11] H.-A. Bae, K.-W. Lee, Y.-H. Lee, J. Mol. Catal. B: Enzyme 40 (2006) 24–29.
[12] E.G. Lee, H.S. Won, H.-S. Ro, Y.-W. Ryu, B.H. Chung, J. Mol. Catal. B: Enzyme 26
(2003) 149–156.
[13] P. Spizzo, A. Basso, C. Ebert, L. Gardossi, V. Ferrario, D. Romano, F. Molinari,
Tetrahedron 63 (2007) 11005–11010.
[14] Vapourtec© R2+/R4 system was used; website: www.vapourtec.co.uk.
[15] G. Morrone, G. Nicolosi, A. Patti, M. Piattelli, Tetrahedron: Asymmetry 6 (1995)
1773–1778.
[16] H.Y. Zhang, X. Wang, C.B. Ching, J.C. Wu, Biotechnol. Appl. Biochem. 42 (2005)
67–71.
[17] G. Duan, C.B. Ching, Biochem. Eng. J. 2 (1998) 237–245.
[18] A. Ghanem, Chirality 22 (2010) 597–603.
[19] H. Yang, E. Henke, U.T. Bornscheuer, J. Org. Chem. 64 (1999) 1709–1712.
[20] C.-S. Chen, Y. Fujimoto, G. Girdaukas, C.J. Sih, J. Am. Chem. Soc. 104 (1982)
7294–7299.
[1] J.L. Masferrer, B.S. Zweifel, S.M. Colburn, et al., Am. J. Ther. 2 (1995) 607–610.
[2] S. Wynne, D. Djakiew, Mol. Cancer Res. 8 (2010) 1656–1664.
[3] H. Jin, Z. Wang, L. Liu, L. Gao, L. Sun, X. Li, H. Zhao, Y. Pan, H. Shi, N.
Liu, L. Hong, J. Liang, Q. Wu, Z. Yang, K. Wu, D. Fan, Mol. Pharm. 7 (2010)
156–168.
[4] W.J. Wechter, D. Kantoci, E.D. Murray, D.D. Quiggle, D.D. Leipold, K.M. Gibson,
J.D. McCracken, Cancer Res. 57 (1997) 4316–4324.
[5] W.J. Wechter, D.D. Leipold, E.D. Murray, D. Quiggle, J.D. McCracken, R.S. Barrios,
N.M. Greenberg, Cancer Res. 60 (2000) 2203–2208.
[6] W.J. Wechter, E.D. Murray, D. Kantoci, D.D. Quiggle, D.D. Leipold, K.M. Gibson,
J.D. McCracken, Life Sci. 66 (2000) 745–753.
[7] S. Grösch, I. Tegeder, K. Schilling, T. Maier, E. Niederberger, G. Geisslinger, FASEB
J. 17 (2003) 1316–1318.
[8] E.J. Quann, F. Khwaja, K.H. Zavitz, D. Djakiew, Cancer Res. 67 (2007) 3254–3262.
[9] G.-H. Shin, K.-W. Lee, T.-K. Kim, H.-D. Shin, Y.-H. Lee, J. Mol. Catal. B: Enzyme
33 (2005) 93–98.
[21] C. Csajági, G. Szatzker, E.R. To˝ ke, L. Ürge, F. Darvas, L. Poppe, Tetrahedron:
Asymmetry 19 (2008) 237–246.