A separation-integrated cascade reaction to overcome thermodynamic limitations in rare-sugar synthesis

Article abstract of DOI:10.1002/anie.201411279

Enzyme cascades combining epimerization and isomerization steps offer an attractive route for the generic production of rare sugars starting from accessible bulk sugars but suffer from the unfavorable position of the thermodynamic equilibrium, thus reducing the yield and requiring complex work-up procedures to separate pure product from the reaction mixture. Presented herein is the integration of a multienzyme cascade reaction with continuous chromatography, realized as simulated moving bed chromatography, to overcome the intrinsic yield limitation. Efficient production of D-psicose from sucrose in a three-step cascade reaction using invertase, D-xylose isomerase, and D-tagatose epimerase, via the intermediates D-glucose and D-fructose, is described. This set-up allowed the production of pure psicose (99.9%) with very high yields (89%) and high enzyme efficiency (300 g of D-psicose per g of enzyme).

Full text of DOI:10.1002/anie.201411279

.
Angewandte
Communications
International Edition: DOI: 10.1002/anie.201411279
German Edition: DOI: 10.1002/ange.201411279
Biocatalysis
A Separation-Integrated Cascade Reaction to Overcome
Thermodynamic Limitations in Rare-Sugar Synthesis**
Nina Wagner, Andreas Bosshart, Jurek Failmezger, Matthias Bechtold, and
Sven Panke*
Angewandte
Chemie
4182
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 4182 –4186

Angewandte
Chemie
Abstract: Enzyme cascades combining epimerization and
isomerization steps offer an attractive route for the generic
production of rare sugars starting from accessible bulk sugars
but suffer from the unfavorable position of the thermodynamic
equilibrium, thus reducing the yield and requiring complex
work-up procedures to separate pure product from the reaction
mixture. Presented herein is the integration of a multienzyme
cascade reaction with continuous chromatography, realized as
simulated moving bed chromatography, to overcome the
intrinsic yield limitation. Efficient production of d-psicose
from sucrose in a three-step cascade reaction using invertase,
d-xylose isomerase, and d-tagatose epimerase, via the inter-
mediates d-glucose and d-fructose, is described. This set-up
allowed the production of pure psicose (99.9%) with very high
yields (89%) and high enzyme efficiency (300 g of d-psicose
per g of enzyme).
of the energetically favorable final reaction by online
integration of the cascade reaction and product removal in
a continuous process (Figure 1a). However, the complexity of
the cascade reaction imposes a number of challenging
boundary conditions, in particular on the selected separation
O
ne-pot cascade reactions are considered an attractive
alternative to more conventional step-by-step synthetic
schemes. They frequently rely on enzymes as myriads of
these selective biocatalysts evolved to operate under broadly
similar reaction conditions. This combination enables com-
pact one-pot reaction schemes which avoid laborious protec-
tion/deprotection steps and intermediate isolation.[1] Among
many other benefits,[2] running concomitantly multiple reac-
tions in one vessel allows combination of critical steps, having
unfavorable position of the reaction equilibrium, with ener-
getically and highly favorable steps at the end of the cascade
to achieve high yield, and indeed this has been exploited in
a number of elegant synthetic cascade strategies.[3]
Figure 1. a) Integrated operation of enzyme cascade reaction and
product removal. b) Enzyme cascade reaction for the manufacturing of
the rare sugar d-psicose (4) from sucrose (1) with the yield limited by
the position of the equilibria of the isomerization and the epimeriza-
tion reactions. Enzymes that participate in the reaction: invertase
(INV), d-xylose isomerase (DXI), d-tagatose epimerase (DTE). Equilib-
rium constants of the iso- and epimerase reactions at 508C are
indicated.
However, the scope of cascade reactions could be
substantially broadened if the thermodynamic requirement
on the last reaction could be removed. A case in point is the
production of rare sugars: here, the existence of multiple
chemically similar hydroxy groups requires demanding pro-
tection-group chemistry and the use of selective enzymes
appears particularly attractive.[4] In fact, cascade reactions
with a maximum of only three consecutive enzymatic steps
already allow the synthesis of 13 of the 20 rare standard
hexoses from either d-glucose, d-fructose, d-galactose (from
natural sources), or l-sorbose (from Reichsteinꢀs vitamin C
process[5]).[6] While this could be highly attractive for increas-
ing the flexibility and compactness of rare sugar synthesis,
these cascades consist of isomerases and epimerases with the
corresponding unfavorable equilibrium positions,[7] thus
severely reducing the yield and thus preventing such cascades
from implementation.
technique: 1) the technique needs to be sufficiently selective
to separate the product from substrate and intermediate(s);
2) the technique needs to be highly efficient to produce the
target compound in industrially relevant amounts and pro-
ductivities; and 3) the continuous operation of reaction and
separation requires both to broadly operate under the same
reaction conditions.
While these are formidable prerequisites, continuous
countercurrent chromatography, technically realized as simu-
lated moving bed (SMB) chromatography,[8] often fulfills
these criteria.[9] In fact, single enzymatic or chemical reactions
have been integrated with SMB chromatography before.[10]
Herein, we present the practical integration of a cascade
reaction with SMB chromatography to enable the highly
efficient production of the rare sugar d-psicose (4) from
sucrose (1), and it involves three biocatalytic steps, the last
two of which employ thermodynamically unfavorable isomer-
ase- and epimerase-catalyzed reactions. The sugar 4, like
many of the 20 rare sugars in the standard set of hexoses,[4] has
a high potential as either low-calorie sweetener or active
pharmaceutical ingredient.[4] The manufacture of 4 was
recently demonstrated from d-fructose (3) using the enzyme
d-tagatose epimerase (DTE).[10b] We took advantage of the
availability of sugar-converting enzymes from either natural
A promising possibility to alleviate this problem in
a potentially generic way is to compensate for the absence
[*] N. Wagner, A. Bosshart, J. Failmezger, Dr. M. Bechtold,
Prof. S. Panke
Department of Biosystems Science and Engineering, ETH Zurich
Mattenstrasse 26, 4058 Basel (Switzerland)
E-mail: sven.panke@bsse.ethz.ch
[**] We acknowledge funding from the Swiss National Science Foun-
dation.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201411279.
Angew. Chem. Int. Ed. 2015, 54, 4182 –4186
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4183

.
Angewandte
Communications
sources (invertase; INV) or our own enzyme engineering
efforts (xylose isomerase (DXI) and DTE[11]) to set up
a continuous three-step cascade consisting of the hydrolysis of
1 into d-glucose (2) and 3, the isomerization of 2 to 3, and the
epimerization of 3 to 4 (Figure 1b). This was integrated with
ligand-exchange chromatography[12] using a Ca2+-substituted
stationary phase for the separation of the sugar mixture using
water as the liquid phase.
To realize an integrated set up, we addressed first the
cascade reaction. We selected enzymes from thermophilic
organisms, as they usually exhibit high thermal and opera-
tional stability and thus can be employed at elevated temper-
atures, which are required in sugar processing to reduce
microbial contamination and the viscosity of typically highly
concentrated feed streams.[13] High temperatures also
increase catalytic activity and can shift reaction equilibria
towards a more favorable position. We recruited INV
(TnINV)[14] and DXI (TnDXI)[15] from the extreme thermo-
philic bacterium Thermotoga neapolitana, and completed the
set with a previously thermostabilized DTE from Pseudomo-
nas cichorii (PcDTE Var8).[11]
We next compared the specific activities of the three
catalysts. The isomerization step was substantially slower
(kcat = 1.56 sÀ1’) for TnDXI Mut1F1, a three-site mutant of the
wild-type enzyme[16] (see Table S3 in the Supporting Infor-
mation), than the other two steps in the cascade (kcat = 118 sÀ1
(TnINV) and kcat = 36 sÀ1 (PcDTE Var8), all at 508C). We
therefore increased its specific activity for 2 by a factor of four
by iterative saturation mutagenesis,[17] thus resulting in the
variant TnDXI-2 (see Chapter S3).
Figure 2. Constant production of 4 over 3000 min from 1 by a three-
enzyme cascade in an EMR at 508C, with concentrations of 1 (filled
triangles), 2 (empty squares), 3 (empty circles), and 4 (empty
triangles) as indicated. The solid lines indicate simulated concentra-
tions of the respective compounds taking enzyme inactivation into
consideration.
enzyme and thus to keep the conversion of substrate into
intermediates and product constant[18] over a period of at least
two days with a feed concentration of 171 gLÀ1 (500 mm) for
1 (Figure 2). An excellent agreement between the experi-
mental and simulated concentrations could be observed. As
the hydrolysis of 1 is irreversible under these reaction
conditions, we reasoned that the separation during integrated
operation can be reduced from a quaternary to a ternary
problem by providing sufficient TnINV in the EMR to run the
first reaction essentially to completion.
Subsequently, we optimized the joint operation of all
three enzymes. Activity profiles at different pH values were
recorded (see Figure S6) and a working pH of 7.0 was
assigned for cascade operation, as a trade-off of the optimum
operating pH values of the individual biocatalysts. To deter-
mine the optimal temperature for the cascade, we examined
the influence of temperature on thermal activation, unfolding,
and irreversible deactivation of each biocatalyst in an
isothermal enzyme-membrane reactor (EMR) setup as de-
scribed before.[18] Briefly, an EMR was continuously supplied
with fresh starting material, and conversion at different
temperatures was determined by HPLC from periodically
collected samples of the outlet stream.
Next, we investigated the multicomponent separation of
2, 3, and 4 by SMB using a lab-scale plant employing eight
DOWEX 50 WX4-400 (Ca2+) chromatographic columns. A
four-zone SMB, a standard technique for the continuous
separation of sugars,[20] was used, with each zone character-
ized by a designated flow-rate ratio m, representing the ratio
of net liquid phase flow to simulated stationary phase flow,
and a periodic shifting time, tsw, of the outlet ports. This
reflects the simulated countercurrent movement of the sta-
tionary phase. The sugar feed stream is provided between the
two central zones II and III (Figure 3).
The compound that is retained most strongly is trans-
ported to the extract port upstream between zone I and II,
whereas the compound that is retained more weakly is carried
in direction of the fluid flow movement (against the simulated
The resulting curves were fitted to biocatalyst inactivation
models using a previously described procedure[19] (see Fig-
ure S9), thus resulting in
inactivation rate constants
of each biocatalyst species
at 508C (TnINV: 3 ”
10À7 sÀ1, TnDXI-2: 1.4 ”
10À4 sÀ1, PcDTE Var8:
2.9 ” 10À6 sÀ1; see the Sup-
porting Information). This
study allowed us to calcu-
late the amount of
enzyme that needed to
be injected periodically
into the EMR to compen-
sate for inactivated
Figure 3. Direct-coupling of a multienzyme-cascade reaction, SMB separation, and NF recycling for the
continuous production of 4. The letters indicate interfaces at which sampling took place.
4184
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 4182 –4186

Angewandte
Chemie
solid flow; Figure 3) to the raffinate port between zones III
and IV. Zones I and IV are used to regenerate solid and
mobile phases. The desorbent stream leaving the SMB
through zone 4 is here discarded as a waste stream (open
loop SMB). Although SMB chromatography is typically used
for the separation of binary mixtures, it can be easily adapted
for the processing of multicomponent mixtures if the product
elutes either as the first or the last compound.[10a] In our case,
4 eluted after 3 and 2. We carried out SMB experiments using
a 1m (180 glÀ1) feed solution containing the three hexoses at
those concentrations which were predicted to leave the EMR
during integrated operation (464 mm 2, 409 mm 3, 127 mm 4).
Fractions were collected from the extract, raffinate, and waste
stream. We started with an operating point (No. 1 in Table 1)
previously identified suitable for the separation of 3 and 4.[21]
As expected, operating point No. 1 showed high purity of 4 in
the extract and high recovery of 3 (RCFrc) in the raffinate
stream, but only moderate recovery of 2 (RCGlc = 87%). The
balance was found in the waste stream, thus indicating that
the mobile phase was not fully regenerated. As predicted by
the triangle theory for SMB-design,[22] RCGlc could be easily
increased up to 99.5% by stepwise decreasing the mass flow
ratio in zone 4 (No. 2 to 4 in Table 1).
Having separately identified the optimal conditions for
the operation of the cascade reactions and the SMB, we next
addressed integration of the reaction and separation
(Figure 3; for a more detailed scheme see Figure S13). In
the fully integrated mode, fresh 1 is mixed with the recycling
stream enriched in 2 and 3 and then enters the EMR, where it
is partially (ca. 11%) converted into 4. The EMR-outlet
stream enters the SMB unit for separation into product
stream (pure 4, leaving between zone I and II) and a mixed
recycling stream containing the intermediate sugars (leaving
between zone III and IV). To compensate for dilution during
SMB operation, the recycling stream is concentrated by
a nanofiltration (NF) device. The implementation of the
optimum SMB operating point exhibiting full RCFrc and RCGlc
(No. 4, Table 1) led to a strongly diluted raffinate stream, thus
requiring unfeasibly high NF concentration factors for our
lab-scale system. Thus, we adopted operating point No. 3
Table 1: SMB pre-experiments at 508C for an increased sugar recovery in
the recycling stream.
No.
SMB operating parameters
SMB performance
tsw
mI
mII
mIII
mIV
PUEx RCFrc RCGlc
[min]
[À]
[À]
[À]
[À]
[%]
[%]
[%]
(Table 1) which showed decreased recovery of 2 (RCGlc
=
1
2
3
4
5.1
5.1
5.1
5.1
1.406 0.704 1.132 0.500 99.9
1.406 0.704 1.132 0.475 99.9
1.406 0.704 1.132 0.450 99.9
1.406 0.704 1.132 0.300 99.9
99.9
99.9
99.9
99.9
87
94
97
99.5
97%), but allowed operation of the NF with a feasible
concentration factor of 1.8 to obtain again a total sugar
concentration (2 and 3) of 180 gLÀ1 in the recycling stream.
The EMR was operated at 5088C and received initially
Figure 4. Concentration profiles of the starting material 1, intermediates 2 and 3, and the product 4 during operation of the integrated process.
Samples were taken at different sampling interfaces (compare Figure 3): a) Interface A (located between NF and EMR), b) Interface B (located
between EMR and SMB), c) Interface C (located between SMB and NF) and d) at the extract port D. The time point of full integration is indicated
by a dashed line in each plot.
Angew. Chem. Int. Ed. 2015, 54, 4182 –4186
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
4185

.
Angewandte
Communications
1.5 mgmLÀ1 TnINV, 8.9 mgmLÀ1 TnDXI-2, and 0.95 mgmLÀ1
PcDTE Var8. During the operation, 1.50 mg TnINV, 0.37 mg
TnDXI-2, and 4.12 mg PcDTE Var8 were injected every
30 minutes to compensate for losses resulting from biocatalyst
inactivation. To shorten the settling time, a feed solution
containing 1, 2, and 3 at the projected steady-state target
concentrations was used during the start-up (see Chap-
ter S10). Finally, for monitoring the process during the
integrated operation, three stirred sampling reservoirs (A,
B, and C in Figure 3) were installed in the interfaces of the
unit operations.
How to cite: Angew. Chem. Int. Ed. 2015, 54, 4182–4186
Angew. Chem. 2015, 127, 4256–4260
[1] W. D. Fessner in Asymmetric Organic Synthesis with Enzymes
(Eds.: V. Gotor, I. Alfonso, E. García-Urdiales), Wiley-VCH,
Weinheim, 2008, p. 275.
[2] P. A. Santacoloma, G. Sin, K. V. Gernaey, J. M. Woodley, Org.
Process Res. Dev. 2011, 15, 203.
[3] a) W.-D Fessner, New Biotechnol. 2015, DOI: 10.1016/
j.nbt.2014.11.007; Cascade Biocatalysis—Integrating Stereoselec-
tive and Environmentally Friendy Reactions (Eds.: S. Riva, W.-D.
Fessner, S. Fessner), Wiley-VCH, Weinheim, 2014; b) M.
Rachwalski, N. Vermue, F. P. J. T. Rutjes, Chem. Soc. Rev.
2013, 42, 9268.
Stable operation of the integrated process was accom-
plished for more than 4000 minutes (Figure 4), with
3800 minutes of it in a fully integrated manner. The process
immediately delivered 12.5 glÀ1 of 4 in very high purity
(PUEx ꢀ 99.9%). Taking into account the incomplete rejec-
tion of the NF membrane for 2 and 3, the theoretical process
yield was 92% based on 1, and we found an experimental
yield of 89% by comparing the molar mass flows of 1 entering
the process and 4 leaving the system through the extract port
(see Chapter S10). This deviation resulted from the small
amount of 2 leaving the system in the SMB waste stream,
which could be easily circumvented on lager scale by further
reducing the setpoint for mIV and employing a NF unit which
can cope with the higher dilution of the raffinate flow. The
experimental yield compares well with a process yield of
approximately 14%, which could be maximally obtained by
running the same cascade reaction in batch mode. A total of
146 mg enzyme was provided over the investigated run time
of 4200 minutes, thus translating into a total turnover of 0.3 kg
of 4 per gram of total enzyme and a remarkably high EMR
space time yield of 1.6 kglÀ1 dayÀ1. We produced preparative
amounts of pure 4 (14 gdayÀ1) in the course of this work and it
could be isolated from the aqueous phase in high yields (ꢀ
85%) according to a previously described protocol.[21]
In summary, we demonstrate an efficient and, given the
achieved yield and productivity, eminently suitable strategy
for the synthesis of the rare sugar 4 by employing epimerases
and isomerases. Given the diversity of aqueous-media-
compatible stationary phases and the wide range of hexose-
interconverting iso- and epimerases,[4] we argue that this
integrated synthetic strategy enables general access to
a variety of rare hexoses.
[4] K. Beerens, T. Desmet, W. Soetaert, J. Ind. Microbiol. Biotech-
nol. 2012, 39, 823.
[5] C. Bremus, U. Herrmann, S. Bringer-Meyer, H. Sahm, J.
Biotechnol. 2006, 124, 196.
[6] K. Izumori, J. Biotechnol. 2006, 124, 717.
[7] R. N. Goldberg, Y. B. Tewari, T. N. Bhat, Bioinformatics 2004,
20, 2874.
[8] M. Juza, M. Mazzotti, M. Morbidelli, Trends Biotechnol. 2000,
18, 108.
[9] H. Lorenz, A. Seidel-Morgenstern, Angew. Chem. Int. Ed. 2014,
53, 1218; Angew. Chem. 2014, 126, 1240.
[10] a) A. G. OꢀBrien, Z. Horvath, F. Levesque, J. W. Lee, A. Seidel-
Morgenstern, P. H. Seeberger, Angew. Chem. Int. Ed. 2012, 51,
7028; Angew. Chem. 2012, 124, 7134; b) N. Wagner, M. Fuereder,
A. Bosshart, S. Panke, M. Bechtold, Org. Process Res. Dev. 2012,
16, 323.
[11] A. Bosshart, S. Panke, M. Bechtold, Angew. Chem. Int. Ed. 2013,
52, 9673; Angew. Chem. 2013, 125, 9855.
[12] M. Stefansson, D. Westerlund, J. Chromatogr. A 1996, 720, 127.
[13] S. H. Bhosale, M. B. Rao, V. V. Deshpande, Microbiol. Rev. 1996,
60, 280.
[14] L. Dipasquale, A. Gambacorta, R. A. Siciliano, M. F. Mazzeo, L.
Lama, Extremophiles 2009, 13, 345.
[15] C. Vieille, J. M. Hess, R. M. Kelly, J. G. Zeikus, Appl. Environ.
Microbiol. 1995, 61, 1867.
[16] D. Sriprapundh, C. Vieille, J. G. Zeikus, Protein Eng. 2003, 16,
683.
[17] M. T. Reetz, J. D. Carballeira, Nat. Protoc. 2007, 2, 891.
[18] Unpublished results.
[19] M. Bechtold, S. Panke, Chem. Eng. Sci. 2012, 80, 435.
[20] D. E. Rearick, M. Kearney, D. D. Costesso, Chemtech. 1997, 27,
36.
[21] Unpublished results.
[22] C. Migliorini, M. Mazzotti, M. Morbidelli, J. Chromatogr. A
1998, 827, 161.
Keywords: biocatalysis · carbohydrates · enzymes · isomerases ·
synthetic methods
Received: November 20, 2014
Published online: February 16, 2015
4186
www.angewandte.org
ꢀ 2015 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2015, 54, 4182 –4186