Polymer-supported catalase: A green approach to the removal of hydrogen peroxide from reaction mixtures

Article abstract of DOI:10.1021/op0255074

During a programme of work directed at developing a manufacturing route for SB 214857, Lotrafiban, the need arose to find methodology for efficiently destroying hydrogen peroxide in aqueous solutions of the penultimate intermediate, SB 270051. Whilst this was initially achieved using a chemical process, a biotransformation process has now been developed that utilises a polymer-supported catalase enzyme to remove peroxide from reaction mixtures. The biocatalytic approach provides an economic and environmentally friendly solution to peroxide removal when compared to the chemical process.

Full text of DOI:10.1021/op0255074

Organic Process Research & Development 2002, 6, 505−508
Polymer-Supported Catalase: A Green Approach to the Removal of Hydrogen
Peroxide from Reaction Mixtures
Mark Alston,† Andy Willetts,† and Andy Wells*,‡,§
Zylepsis, 6 Highpoint, Ashford, Kent, TN24 8DH, England, and Chemical DeVelopment, GlaxoSmithKline
Pharmaceuticals, Old Powder Mills, near Leigh, Tonbridge, Kent TN11 9AN, England
Scheme 1. Aminocarbonylation reaction to produce SB
270051
Abstract:
During a programme of work directed at developing a manu-
facturing route for SB 214857, Lotrafiban, the need arose to
find methodology for efficiently destroying hydrogen peroxide
in aqueous solutions of the penultimate intermediate, SB 270051.
Whilst this was initially achieved using a chemical process, a
biotransformation process has now been developed that utilises
a polymer-supported catalase enzyme to remove peroxide from
reaction mixtures. The biocatalytic approach provides an
economic and environmentally friendly solution to peroxide
removal when compared to the chemical process.
Introduction
SB 214857-A is a potent GPIIb/IIIa (fibrinogen receptor)
antagonist that was being developed by GlaxoSmithKline.
Fibrinogen receptor antagonists prevent platelet aggregation;
thus, such compounds are designed for the prevention of
thrombotic events such as heart attack and stroke.1 The
manufacturing route for SB 214857 had an aminocarbony-
lation reaction as a key convergent step to assemble to
molecule.2,3 This converts SB 240093 to the penultimate
intermediate, SB 270051, Scheme 1.
Scheme 2. Hydrogenation of SB 270051 to SB 214857
When this synthetic strategy was being developed, it was
envisaged that the product, SB 270051, could be extracted
into water from the anisole solvent and directly hydrogenated
to give SB 214857, thus avoiding solids isolation and drying
(Scheme 2). When the feasibility of this approach was first
examined, the reduction of SB 270051 to SB 214845 required
large amounts (a weight equivalent!) of platinum oxide
catalyst and long reaction times (several days) to go to
completion. This could clearly never be a viable industrial
process. An extensive investigation into this reduction
showed that the presence of iodide anions acting as an
efficient catalyst poison was the cause of the long reaction
time and requirement for a high catalyst loading.4,5 In the
carbonylation reaction that produces SB 27051, 1 mol of HI
per mol of SB 270051 is generated (Scheme 1). The bulk of
the HI (∼95%+) is removed during workup as an insoluble
salt with dicyclohexylamine. The remaining iodide is present
at a level of ∼2000 ppm in the aqueous solution of SB
270051. It was found that if this level could be reduced to
less than 100 ppm, then the reduction of the aqueous SB
270051 solution to SB 214857 could be reliably achieved
with lower loading of a cheaper catalyst (palladium-on-
carbon) in hours rather than days. Thus the reaction became
feasible to consider as a manufacturing process (Scheme 2).
Removal of Iodide. Various ways of physically adsorbing
iodide were considered but were ruled out for large-scale
* To whom correspondence should be addressed. E-mail: Andrew.wells@
astrazeneca.com.
† Zylepsis.
‡ GlaxoSmithKline Pharmaceuticals.
§ Current address: Process R&D, AstraZeneca Charnwood, Bakewell Road,
Loughborough, Leicestershire LE115HT, England.
(1) Samanen, J. M.; Ali, F. A.; Barton, L. S.; Bondinell, W. E.; Burgree, J. L.;
Callahan, J. F.; Calvo, P. R.; Chen, W.; Chen, L.; Erhard, K.; Feuerstein,
G.; Heys, R.; Hwang, S.-M.; Jakas, D. R.; Keenan, R. M.; Ku, T. W.;
Kwon, C.; Lee, C.-P.; Miller, W. H.; Newlander, K. A.; Nichols, A.; Parker,
M.; Peishoff, C. E.; Rhodes, G.; Ross, S.; Shu, A.; Simpson, R.; Takata,
D.; Yellin, T.-O.; Uzsinskas, I.; Veslavsky, J. W.; Yuan, C.-K.; Huffman,
W. F. J. Med. Chem. 1996, 39, 4867.
(2) Etridge, S. K.; Hayes, J. F.; Walsgrove, T. C.; Wells, A. S. Org. Process
Res. DeV. 1999, 3, 60.
(3) Andrews, I.; Atkins, R.; Bellingham, R. F.; Breen, G.; Carey, J. S.; Etridge,
S. K.; Hayes, J.; Hussain, N.; Morgan, D. O.; Share, A.; Smith, S.;
Walsgrove, T. C.; Wells, A. S. Tetrahedron Lett. 2001, 42, 4915.
(4) Baltzly, R. J. Org. Chem. 1976, 41, 928.
(5) Greenfield, H. J. Org. Chem. 1963, 28, 2434.
10.1021/op0255074 CCC: $22.00 © 2002 American Chemical Society
Published on Web 06/11/2002
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
505

use. The use of metals (e.g., silver nitrate) to form insoluble
iodides leads to the precipitation of highly insoluble SB
270051 carboxylate salts. The most efficient process for
large-scale use was to oxidise the iodide and then to extract
out the resulting iodine. Small-scale studies identified a
number of suitable inorganic and organic oxidants although
some oxidants were also capable of oxidising SB 270051.
The most efficient way to remove iodide, with no loss of
SB 270051, was found to be treatment with excess hydrogen
peroxide at pH 5 (10 equiv to the amount of iodide known
to be present in solution). Under these operating conditions
the level of iodide could easily be reduced to 10 ppm or
less.
The need to use an excess of hydrogen peroxide was
driven by the kinetics of the iodide oxidation. At pH 5 the
rate of oxidation is slow, requiring an excess of peroxide to
drive the reaction to completion in a reasonable time (1-2
h). The use of stoichiometric hydrogen peroxide requires low
pH (∼1) or the use of metal catalysts such as molybdenum
or vanadium to get an acceptable reaction rate. SB 270051
is unstable in water below pH 5, and the removal of
molybdenum or vanadium salts was not considered an
attractive prospect. After removal of iodine, the aqueous
solution of SB 270051 had a hydrogen peroxide concentra-
tion of ∼2700 mg/L.
are widespread and used in many bulk industries and
healthcare products, for example, food stuffs, treating
peroxide waste after paper bleaching, and contact lens
cleaning. In recent years, a number of instances of the use
of catalase enzymes to remove hydrogen peroxide (formed
as the reduction product of oxygen) from reactions producing
pharmaceutical intermediates have been reported.7-9
Thus, catalase enzymes presented themselves as poten-
tially useful biocatalysts to remove hydrogen peroxide from
aqueous solutions of SB 270051 after iodide removal. A
range of commercially available catalases was tested, with
mixed results. The use of bovine catalase was ruled out for
obvious reasons. In small-scale screening tests, some catalase
preparations did destroy peroxide in crude reaction mixtures,
but large amounts had to be added. The most promising
catalases identified were from the organisms Aspergillus
niger and Corynebacterium glutamicum. These are both
available on a large scale, in high purity, and are from
nonmammalian sources. The catalase was simply added
directly to the aqueous SB 270051 solution and stirred at 25
°C until the residual peroxide level had dropped to below
10 ppm. The solution could then be directly hydrogenated.
When this was scaled up in the pilot plant, several deficien-
cies became apparent. After an initial reaction, gas evolution
would cease before all the peroxide had been consumed. The
reaction could be restarted by the addition of fresh enzyme,
which indicates a problem with the stability or inhibition of
the enzyme in this reaction mixture. If too much enzyme is
added, however, this led to a retardation of the rate of the
subsequent reduction step. This is an inherent problem with
the application of a soluble protein that can interfere with
the metal catalyst. The aqueous SB 270051 solution would
need to be ultrafiltered to remove any protein prior to
hydrogenation, but this would add an extra process step to
the synthesis. All of these problems arise from the use of
soluble enzymes in such a processes.
A number of immobilised catalase preparations have been
reported, but most could be ruled out for various reasons.
The most attractive carrier for the protein that would be
compatible with large-scale batch processing was a Eupergit
type (macroporous acryate beads).10 The catalase enzymes
of choice could be easily attached, by covalent bonding, to
Eupergit C250L, a support containing oxirane functionalities
that react with various groups on the protein.11 Two other
supports, Eupergit C and cyanogen bromide-activated
Sepharose were also used to support the catalase but were
not progressed on the basis of cost, activity, and stability
comparisons. Mixing a solution of catalase with the support
at ambient temperature gave at least a 98% attachment of
protein. Although binding was effective in a few hours, the
protein was mixed with the resin for 72 h, resulting in a
Since the aqueous feedstock of SB 270051 was subse-
quently charged to a hydrogenation reaction, the excess
hydrogen peroxide first had to be reduced to less than 1 mg/L
for safety reasons.
Removal of Hydrogen Peroxide. Our initial strategy was
to add inorganic reducing agents such as sodium metabisul-
phite or sodium sulphite. These quickly destroyed peroxide,
but sulphur residues made the resulting solutions very
difficult to hydrogenate.
The use of low-valent metal salts such as stannous
chloride was also deemed unattractive. The process that was
first scaled-up used palladium-on-carbon as a catalyst to
decompose the excess peroxide to oxygen and water. The
amount of catalyst and temperature (65-70 °C) were
adjusted to give a smooth gas evolution that could be diluted
with nitrogen and safely vented.
One peculiar aspect of aqueous SB 270051 is that
solutions at ca. neutral pH have the tendency to crystallise
when heated. If this occurred during the reaction, the resulting
solid coated the catalyst, and the reaction ceased. To ensure
that SB 270051 remained in solution during the peroxide
destruction, the aqueous solution had to be diluted with
2-propanol (IPA) prior to addition of the Pd catalyst. Whilst
this procedure worked and was scaled-up, it was not ideal
having an organic solvent mixture heated with an inhomo-
geneous Pd catalyst in the presence of oxygen. Hence, we
decided to look for a safer, more environmentally friendly
way to remove excess peroxide.
(7) Leise, A.; Seelbach, K.; Wnndrey, C. Industrial Biotransformations; Wiley-
VCH: New York, 2000.
(8) Waldemar, A.; Lazarus, M.; Boss, B.; Saha-Moller, C. R.; Humpf, H.-U.;
Schreier, P. J. Org. Chem. 1997, 62, 7841.
(9) Cabri, W.; Verga, R.; Cambiaghi, S.; Bernasconi, E. Chim. Ind. (Milan)
1999, 81, 461.
(10) Seip, J. E.; Fager, S. K.; Gavagan, J. E.; Anton, D. L.; Cosimo, R. D.
Bioinorg. Med. Chem. 1994, 2, 371.
Enzymic Destruction of Hydrogen Peroxide. Catalase
enzymes (EC 1.11.1.6) are used by living systems to break
down hydrogen peroxide to water and oxygen.6 Catalases
(6) Boyer, P. D., Ed. The Enzymes XIII, Part C; Academic Press: New York,
(11) Katchalski-Katzir, E.; Kraemer, D. M. J. Mol. Catal. B: Enzym. 2000, 10,
157.
1976.
506
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development

more stable form of the immobilised enzyme; this presum-
ably arises from multiple points of attachment of the protein
to the support. A range of protein loadings from 10 to 200
mg/g did not appear to have a significant affect on protein
activity (data not shown).
technology would have produced a robust and “green”
manufacturing process. It also provides a good example of
a “nonchiral” use of a biotransformation in process chemistry,
an area in which we feel more thought could be directed.
Experimental Section
The supported catalase enzymes proved to be much more
reliable. A 0.5-g portion of dry resin containing 0.7 mg of
enzyme could treat a 200-mL portion of SB 270051 solution,
containing around 20 g of SB 270051, in less than 1 h at 25
°C. The resin could be filtered off at the end of the reaction,
and the SB 270051 in aqueous solution could be hydroge-
nated directly to give the same yield and quality of product
as obtained in the palladium-treated route. Typically, a 20
wt % charge (based on SB 270051) of 10% Pd on carbon
catalyst and 50 psi hydrogen pressure would reduce solutions
of SB 270051 obtained from either peroxide destruction
process in ∼4 h. The yields of SB 214857 were typically
70% from SB 240093 and the products indistinguishable by
analysis (weight-based assay, impurity profile, chiral assay).
The recovered enzyme resin could be reused in subse-
quent, sequential reactions with identical results. Nine
sequential recycles of the supported enzyme were followed
by soaking the resin in the reaction mixture solution for 24
h with no loss of activity. Similarly, sequential challenge of
the enzyme with hydrogen peroxide solution did not cause
any reduction in the activity level or dissociation of the
protein from the support.
Thus, the catalase resins provided an effective, economi-
cal, safe, and environmentally friendly solution to the
problem of removing peroxide in the processing of SB
270051 solutions.
(1) No palladium-on-carbon is used for peroxide destruc-
tion.
(2) The reactions are run in water alonesno IPA cosolvent
is required. This removes the need to dispose of IPA waste
and also makes the reaction more volume-efficient.
(3) No heating or cooling is required, thus saving energy.
The chemical process runs at 70 °Csthe enzyme reaction
proceeds at ambient temperature, releasing oxygen in a safe
and controlled manner.
Corynebacterium glutamicum catalase was obtained from
Boehringer Mannheim as a freeze-dried preparation and
stored at 4 °C. Eupergit C 250L was obtained from Ro¨hm
GMBH as macroporous, spherical “beads” and stored at
<-15 °C.
Method for the Production of Eupergit C 250L-
Immobilised Catalase. The catalase was reconstituted in
KPO4 buffer (1 M, pH 7.5) at a concentration of 1.67 g of
powder/L of buffer, and mixed gently. After approximately
30 min the enzyme was dissolved, and to the solution was
added Eupergit C 250L at 167 g/L with mixing. The mixture
was allowed to stand at ambient temperature for 72-96 h,
after which the immobilised enzyme preparation was washed
with 3 vols of KPO4 buffer (1 M, pH 7.5) followed by 3
vols of deionised water.
Chemistry. Details of the large-scale aminocarbonylation
of SB240093 to SB 270051 are to be published elsewhere.3
Aqueous solutions of SB 270051. To avoid isolation and
solids handling, SB 270051 was not isolated, but formed
from SB 240093 and carried through in solution.
The aminocarbonylation reaction was run in 8 vols of
anisole. When complete the reaction was quenched with
toluene (2 vols) and water (8 vols). The solids (dicyclohexyl-
amine HI salt and some palladium black) were removed by
filtration and washed with water (2 vols). The aqueous phase
was basified with 2.0 M sodium hydroxide to pH 12 and
washed with ethyl acetate to remove residual dicyclohexyl-
amine. This gave an aqueous solution of SB 270051 at ∼10
wt %/v containing ∼2000 ppm iodide (as determined by ion
chromatography analysis). The pH was then adjusted to 5.0
with 30 wt %/wt phosphoric acid solution.
Typical treatments for 200 mL of this solution are as
follows.
Palladium-on-Carbon. Celite (1.5 g) was added. During
the peroxide treatment some tars are precipitated. Adsorption
onto Celite was found to be the most effective way to remove
the tar without encountering downstream filtration problems.
Hydrogen peroxide solution (36 mL, 3 wt %/wt) was added.
The mixture was stirred for 30 min and then treated with
dichloromethane (40 mL).
The mixture was stirred for 3 min and then filtered; the
organic layer was discarded. The aqueous layer was re-
extracted with dichloromethane (2 × 40 mL).
(4) The catalase resin can be recovered and reused many
times. The actual number of possible reuses has yet to be
determined, but the results from the initial small-scale tests
suggest the catalase on the resin is very stable, for example,
at least 10 cycles possible with no inhibition by the reaction
mixture and no detectable protein leaching. Eupergit supports
also seem to be very robust under processing conditions. It
should be noted that Pen G acylase on Eupergit resin can be
reused ∼700 times.11
The pH was adjusted to 7 with 2 M sodium hydroxide
and IPA added (60 mL) followed by palladium-on-carbon
catalyst (2 g wet catalyst, 10% dry weight Pd, Johnson
Matthey type 87L). The mixture was heated to 60° C for
2-3 h after which the peroxide level was less than 1 mg/L
as judged by analysis with Quantofix peroxide test strips,
ensuring appropriate dilutions were made at higher peroxide
concentrations to keep the assay with the linear range.
The mixture was cooled and filtered. The IPA-water
solution was concentrated under reduced pressure to adjust
On a manufacturing scale, if 30 t of SB 214857 were to
be produced per year this would require ∼3 t of palladium-
on-carbon catalyst just to remove peroxide. If the catalase
resin could be reused just 50 times, only 0.015 t would be
required! Apart from the green aspects, at this level the cost
of enzyme and resin become negligible.
Unfortunately, the project was terminated before the
supported catalase resin step could be scaled-up and tested.
The laboratory work would indicate however that this
Vol. 6, No. 4, 2002 / Organic Process Research & Development
•
507

the concentration SB 270051 to ∼10 wt %/v. The pH was
adjusted to 5 with phosphoric acid solution and the solution
was then hydrogenated using the following conditions, 20wt
%/wt based on SB 270051 10% Pd on carbon catalyst/50
°C/50 psi H2 pressure. After 4 h, HPLC analysis showed
less than 1% residual SB 270051. SB 214857 could then be
isolated in ∼70% overall yield from SB 240093.
Multiple Reuse of Catalase Resins. A 20 mL sample of
SB 270051/peroxide reaction mixture was treated with 0.45
g of wet resin (∼0.2 g dry weight). All the peroxide was
destroyed after 3 min at 25 °C. The resin was removed by
filtration, washed with water, and reused. In nine consecutive
reactions the same batch of catalase resin destroyed the
peroxide in less than 3 min. The catalase resin was then
soaked in reaction mixture for 24 h and reused. On the 10th
reuse no loss of activity at all was detected, confirming that
the solid-supported catalase was not inhibited by the reaction
mixture. Identical results were found for both the Eupergit
C-250L and cyanogen bromide-activated Sepharose resins.
Catalase Resin (Corynebacterium glutamicum on Eu-
pergit C250-L). After dichloromethane extraction of the
liberated iodine, the pH was adjusted to 7 with 2 M sodium
hydroxide solution. A portion of the Eupergit C250-L
catalase resin was added (0.95 g wet resin,∼0.46 g dry
weight containing ∼0.7 mg catalase). The reaction was
stirred at 25° C.
Acknowledgment
We would like to thank Tim Walsgrove, Jerome Hayes,
Richard Atkins and Analytical Sciences at Tonbridge for their
support and help in the course of this work, and Dr John
Sime and Dr Lawson Powell of Zylepsis for support and
advice.
After 1 h all the peroxide had been destroyed (less than
1 mg/L). The resin was simply filtered off and the aqueous
solution of SB 270051 hydrogenated after adjustment to pH
5.The recovered resin was reused in a second reaction with
identical results. Both hydrogenations proceeded at rates
identical to those in palladium-on-carbon treated solutions
(4 h to reduce SB 270051 to less than 1%) and the product,
SB 214857, was isolated in identical yield and quality.
Received for review January 23, 2002.
OP0255074
508
•
Vol. 6, No. 4, 2002 / Organic Process Research & Development