Efficient ionic liquid-based platform for multi-enzymatic conversion of carbon dioxide to methanol

Article abstract of DOI:10.1039/c8gc02230e

Low yields commonly obtained during enzymatic conversion of CO₂ to methanol are attributed to low CO₂ solubility in water. In this study, four selected ionic liquids with high CO₂ solubility were separately added to the multi-enzyme reaction mixture and the yields were compared to the pure aqueous system (control). In an aqueous 20% [CH][Glu] system, yield increased ca. 3.5-fold compared to the control (ca. 5-fold if NADH regeneration was incorporated). Molecular dynamics simulation revealed that CO₂ remains for longer in a productive conformation in the enzyme in the presence of [CH][Glu], which explains the marked increase of yield that was also confirmed by isothermal titration calorimetry-lower energy (ΔG) binding of CO₂ to FDH. The results suggest that the accessibility of CO₂ to the enzyme active site depends on the absence/presence and nature of the ionic liquid, and that the enzyme conformation determines CO₂ retention and hence final conversion.

Full text of DOI:10.1039/c8gc02230e

View Article Online
View Journal
Green
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: Z. B. Zhang, J.
Muschiol, Y. Huang, S. B. Sigurðardóttir, N. von Solms, A. E. E. Daugaard, W. jiang, J. Luo, B. Xu, S. Zhang
and M. Pinelo, Green Chem., 2018, DOI: 10.1039/C8GC02230E.
This is an Accepted Manuscript, which has been through the
Volume 18 Number
7 7 April 2016 Pages 1821–2242
Royal Society of Chemistry peer review process and has been
accepted for publication.
Green
Chemistry
Accepted Manuscripts are published online shortly after
Cutting-edge research for a greener sustainable future
www.rsc.org/greenchem
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 1463-9262
CRITICAL REVIEW
G. Chatel et al.
Heterogeneous catalytic oxidation for lignin valorization into valuable
chemicals: what results? What limitations? What trends?
rsc.li/green-chem

Please do not adjust margins
Green Chemistry
Page 1 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
Efficient Ionic Liquid–based platform for Multi-Enzymatic
Conversion of Carbon Dioxide to Methanol
Zhibo Zhang,a, b Jan Muschiol,a Yuhong Huang,a Sigyn Björk Sigurdardóttir,a Nicolas von Solms,a
Anders E. Daugaard,a Jiang Wei,a Jianquan Luo,c Bao-Hua Xu,b Suojiang Zhang*b and Manuel
Pinelo*a
Received 00th January 20xx,
Accepted 00th January 20xx
DOI: 10.1039/x0xx00000x
www.rsc.org/
Low yields commonly obtained during enzymatic conversion of CO2 to methanol are attributed to low CO2 solubility in
water. In this study, four selected ionic liquids with high CO2 solubility were separately added to the multi-enzyme reaction
mixture and the yields were compared to the pure aqueous system (control). In an aqueous 20% [CH][Glu] system, yield
increased app. 3.5-fold compared to the control (app. 5-fold if NADH regeneration was incorporated). Molecular dynamics
simulation revealed that CO2 remains for longer in a productive conformation in the enzyme in the presence of [CH][Glu],
which explains the marked increase of yield that was also confirmed by isothermal titration calorimetry - lower energy
(ΔG) binding of CO2 to FDH. The results suggest that the accessibility of CO2 to the enzyme active site depends on the
absence/presence and nature of the ionic liquid, and that the enzyme conformation determines CO2 retention and hence
final
conversion.
Introduction
One of the greatest environmental challenges we face today is the emission of enormous amounts of carbon dioxide (CO2) into the
atmosphere each year, which contributes to global warming, ocean acidification, melting of icebergs and the energy crisis.1, 2 Ideally, CO2
ought to be converted to useful chemical and fuels (e.g. methanol) for renewable energy utilization and simultaneous alleviation of the
problem of CO2 emissions. Therefore extensive efforts have been made to bring about catalytic hydrogenation of CO2 via chemistry,
electrochemistry, photochemistry and enzymatic conversions.3-5 Due to the inherent thermodynamic stability and low reactivity of CO2,
production of methanol by enzymatic conversion has significant advantages over conventional techniques owing to the high selectivity,
high efficiency, mild experimental conditions, and environmental friendliness of enzymatic catalysis.6
Inspired by the biological metabolic pathway, sequential reduction of CO2 to formic acid, formaldehyde and methanol can be achieved
by using formate dehydrogenase (FDH), formaldehyde dehydrogenase (FaldDH), and alcohol dehydrogenase (ADH), respectively.7
However, the yield of methanol achieved in this type of system is only 43.8%.7 The low conversion is partly explained by the fact that the
reaction rate of the first reaction in the sequence (CO2 → formic acid), catalysed by FDH, is much slower than its reverse reaction (formic
acid → CO2). Indeed, Rusching et al. reported that formic acid oxidation was 30 times faster than CO2 reduction catalyzed by FDH.8 Thus
the CO2 → formic acid step is likely to contribute to the low performance reduction of CO2 to methanol. In this regard, we envisioned that
a higher yield of product, either formaldehyde or methanol, could be reached by increasing the concentration of substrate (CO2) in the
solution, which may drive the transformation of CO2 to formic acid forward. Indeed, a certain threshold concentration of formic acid is
required for the second step reaction to proceed.9 In addition, it could also be helpful to further promote the reaction rate by
incorporating a membrane in the system so that products would be immediately separated from the reaction, driving the equilibrium
towards the formic acid.
Ionic liquids (ILs) have great capacity to capture CO2 via electrostatic forces, van der Waals forces, hydrogen bonds and other physical
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 1
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 2 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
effects.10 Amine functionalized cation-tethered ILs have been used for CO2 capture, in which 0.5 mol CO2 per mol of IL could be absorbed
through a carbamate mechanism.11 Amino acid ILs could capture almost 1 mol CO2 per mol of IL by forming carbamic acid rather than
carbamate.12 Furthermore, the high potential of CO2 electronic reduction could be lowered by ILs to achieve a lower energy barrier.13, 14 To
date, ILs have been used in various enzymatic reactions, e.g. with cellulases and ADH.15-18 Recently, several biocompatible and
environmentally friendly ILs have been identified which are composed of naturally-derived materials such as sugars, amino acids, and
choline.19 Several proteins have been successfully dissolved in choline dihydrogenphosphate [CH][DHP] without denaturation.20 70% of the
initial redox activity of Cytochrome C remained more than one year after dissolving in a mixture of [CH][DHP] and water.21 Amino acid-
based ILs as benign media have been also reported in biomedical applications.22, 23 ILs could thus be promising substitutes of traditional
buffers for conducting selected enzymatic reactions.
For in-situ removal of products (methanol and NAD+) and recycling of enzymes, a separation system platform is additionally required.
Recently, inspired by membrane fouling mechanisms, we proposed a simple approach to immobilize enzymes in membranes using “reverse
filtration” of the enzyme solution.24 In this system, enzyme immobilization was achieved by hydrogen bonding, entrapment and
hydrophobic or electrostatic adsorption.25 The activity of the immobilized enzymes could be maintained to approximately that of the free
enzymes due to the mild and fast immobilization procedure. High enzyme loading could also be maintained and the contact time for the
substrate-enzyme complex could be controlled by changing pressure. However, to produce one mole of methanol in such a cascade
reaction, three moles of reduced nicotinamide adenine dinucleotide (NADH) are stoichiometrically consumed, as the cofactor (NADH) acts
as a hydrogen and electron donor at each step of the reduction reaction.9 Converting the oxidized form of the cofactor (NAD+) to NADH is
essential for reducing cost and enhancing methanol production. Normally, NADH regeneration is accomplished by chemical,
photochemical, and electrochemical methods, but such regeneration can be also attained by adding another enzyme system which
requires NAD+ to proceed (Scheme 1).26
In this study, biocompatible ILs composed of choline and amino acids (i.e. [CH][Glu], [CH][Pro], [CH][Gly], and [CH][His]) were designed
and synthesized in order to increase CO2 solubility and stabilize FDH. These ILs were incorporated in a membrane reactor system which
enabled in situ removal of products from the reaction, as illustrated in Figure 1. Four kinds of ILs as co-solvent were evaluated in the
biocatalytic membrane reactor by passing a mixture of CO2, IL and cofactor through the enzyme-loaded membrane. To our knowledge, this
is the first report of multi-enzymatic conversion of CO2 to methanol in ILs with NADH regeneration. This integration of ILs and biocatalytic
membrane provides a promising avenue for a practical CO2-based sustainable chemistry.
Scheme 1. Multi-enzyme system for methanol synthesis from CO2 with in situ regeneration of NADH.
Experimental
Materials
Glycine (Gly), L-proline (Pro), L-histidine (His), L-glutamic acid (Glu), Choline hydroxide (aqueous solution 46 wt%), were
purchased from Sigma Aldrich (St. Louis, MO, USA). Methanol and acetonitrile were analytical grade and used without any
further purification. Double-distilled water was used in all experiments. Formate dehydrogenase (EC 1.2.1.2, homo-dimer, 76
kDa) from Candida boidinii (FDH), formaldehyde dehydrogenase (EC 1.2.1.46, homo-dimer, 150 kDa) from Pseudomonas sp.
(FaldDH), alcohol dehydrogenase (EC 1.1.1.1, homo-tetramer, 141 kDa) from Saccharomyces cerevisiae (ADH), and glucose
dehydrogenase (EC 1.1.1.118, homo-hexamer, 300 kDa) from Pseudomonas sp. (GDH) were purchased from Sigma–Aldrich (St.
Louis, MO, USA). These commercial powders or liquids are not pure enzymes and the protein content was determined by
Bradford protein assay. β-Nicotinamide adenine dinucleotide reduced form (NADH, >97 wt%), β-nicotinamide adenine
dinucleotide hydrate (NAD+), Trizma base, hydrochloric acid (37%) D-glucose, and methanol (
Sigma Aldrich (St. Louis, MO, USA). All the enzyme and substrate solutions were prepared using 0.1 M Tris
≥
99.9%) were purchased from
–HCl buffer (pH = 7.0)
unless otherwise stated. CO2 gas (>99.5%) in a cylinder was purchased from AGA A/S (Denmark). Commercial UF membranes
(PLGC, Millipore) used in this work have a regenerated cellulose skin layer on a polypropylene support, and their molecular
weight cut-off is 10 kDa.
Synthesis of [Ch][AA] ILs
ILs were prepared and purified according to the literature.27 [Ch][OH] aqueous solution (about 4 M) was added dropwise under cooling to
an amino acid aqueous solution or suspension (0.06 mol) to obtain a slight excess (about 10 mol%) of amino acid. The mixture was stirred
at about 3 oC overnight in the dark. Water was then removed under reduced pressure at 50 oC using a rotavapor. Acetonitrile/methanol (9 :
2 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 3 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
1, v/v) was then added under vigorous stirring to precipitate the excess of amino acid. The mixture was left stirring overnight and the
excess of amino acid was then filtered off. The filtrate was evaporated to remove solvents at 50 oC. The product was dried under vacuum
for 72 h at 60 oC. Characterizaiton data of ILs can be found in supporting information.
Experimental set-up and procedure
The dead-end filtrations and enzymatic reaction were performed in a stirred cell (Amicon 8050, Millipore, USA) and descriptions of
equipment and procedure can be found in previous work.24 The PLGC membranes (10 kDa) were placed on the membrane holder in
‘sandwich’ mode (with their own support layer facing the feed and an extra polypropylene support beneath the skin layer). The membrane
was first soaked in a 5% NaCl solution for 30 min and then filtered with deionized water for another 30 min at 1 bar (procedures according
to the manufacturers’ instructions). Next, the water permeability of the membranes was measured at 2 bar with buffer for 30 min. Then
each enzyme solution (30 mL) was poured into the cell with a 10 kDa membrane for enzyme immobilization. The prepared solution was
bubbled with CO2 through a syringe needle (0.6 mm × 25 mm) before entering the reactor. The flow rate of gas (measured by the speed of
bubble emission) was controlled in the same manner in all the experiments by controlling the pressure valve.
Solubility experimental apparatus and procedure
The gas solubility experimental apparatus and procedure are similar to the work of Shang et al. In the experiment CO2 of ambient pressure
was bubbled at a flow rate of about 60 mL min−1 through about 4.0 g of the IL in a glass tube with an inner diameter of 12 mm. The glass
tube was partly immersed in a water bath of desired temperature. The weight of the IL solution was determined at regular intervals by an
electronic balance (OHAUS Corp. AR2140, USA) with a resolution of 0.0001 g.
Enzyme immobilization
Three enzymes of 100 µL liquid FDH, 1.0 mg solid FaldDH, and 1.5 mg solid ADH were immobilized in the 10 kDa regenerated cellulose
membrane. Enzyme immobilization was carried out at a pressure of 2 bar, and permeate was collected in precision cylinders for analysis.
The cylinders were replaced manually for every 4 mL. At the end of filtration, the ‘fouled’ membrane was washed with 10 mL of buffer at a
pressure of 2 bar, and then rinsed 3 times with buffer without pressure. The amount of immobilized enzyme (loading) was calculated from
ꢀꢁ
the mass balance equation, and the immobilization efficiency was expressed as enzyme loading efficiency (loading efficiency = ) where
ꢀꢂ
mi and mt are amount of immobilized and total enzyme, respectively.
Enzymatic reaction with immobilized enzymes
NADH solution (5 mM) was prepared with 0.1 M Tris–HCl buffer and ILs which had been pre-bubbled with CO2 for 30 min. 4 mL NADH
solution with saturated CO2 was added to the stirred cell equipped with 10 kDa regenerated cellulose membrane. The applied pressure was
controlled manually to ensure that 4 mL permeate was obtained in 30 min. For the enzyme reuse experiment, when 4 mL of permeate had
been obtained, the filtration was paused and another 4 mL of fresh NADH solution with saturated CO2 was added for the next cycle (each
cycle lasted about 30 min).
Enzymatic reaction with NADH regeneration
2 mg of glucose dehydrogenase (GDH) for NADH regeneration was immobilized together with the other three enzymes (i.e. FDH, FaldDH,
ADH) in the 10 kDa regenerated cellulose membrane, following the immobilization procedure given above. The 4 ml NADH solution with
saturated CO2 containing 5 mM NADH and 50 mM D-Glucose as substrate for glucose dehydrogenase was run through the membrane. The
obtained permeate was recycled as feed solution in the next reaction cycle and this operation was repeated for six times.
Analytical methods
The concentration of enzymes was measured as protein concentration using the Bradford protein assay (Perkin Elmer lambda20 UV/VIS,
o
Germany). A Hewlett Packard HP6890 gas chromatograph (GC) equipped with a FID (250 C) and a Restek XTI-5 column (30 m ×0.25 mm
i.d., film thickness 0.25 mm) was used for methanol concentration. The carrier gas was N2 with a flow rate of 0.4 mL min-1. The injector
temperature was 150 oC and the injection volume was 1 µL. Methanol GC chromatograms were calibrated with 0.01–1 mM methanol
solution in 0.1 M pH 7.0 Tris– HCl buffer. Scanning electron microscopy (SEM) was performed in an FEI Helios EBS3 dual beam electron
microscope. The skin and support samples were prepared by cutting a small square of the membrane, which was then attached to an
aluminium stub by means of double sided sticky carbon tape. The edges of the sample were mounted on the aluminium stub by means of
copper tape. After freezing the membrane sample by plunging in liquid nitrogen, cross sections of the membrane skin and the support
were cut from the with a pair of scissors. The cross sections were mounted on a slotted specimen stub and fixed with copper tape. All
samples were coated with Pt for 2 s at 80 mA in a Cressington 208HR Sputter Coater, which gave an approximate thickness of 4 nm. The
micrographs were obtained with an Everhart Thornley detector at low magnifications and with a Thru-the-Lens detector at high
magnifications, in high vacuum at 5 keV acceleration voltage and 43 pA current. 1H NMR and 13C NMR measurements. Spectra were
recorded at 298 K on NMR spectrometer (av-600 MHz, D2O, Bruker, Switzerland). Solutions were prepared by dissolving 20–30 mg of IL in
0.7 mL of D2O. Elemental analysis of C, H, and N (Elementar Vario EL, Germany) indicated that the elemental ratio of ILs agrees well with
their predicted structure.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 3
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 4 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
Isothermal titration calorimetry
The titration experiments were performed using a Nano ITC low volume titration calorimeter (TA instruments, New Castle, DE, USA).
Titrations were performed at 25 °C and consisted of enzyme (FDH) and 4.0 μL injections of ligand (NaHCO3, sodium formate, NADH, or
NAD+) at 300-second intervals. All solutions were filtered, degassed to avoid bubble formation, and equilibrated to the corresponding
temperature before each experiment. The syringe was inserted into the reaction cell, stirring (250 rpm) was initiated, and the instrument
was equilibrated at 25 °C until the base line was flat and stable. The titration data were analyzed with Nano Analyze software (TA
Instruments) using an independent model to obtain the curve fitting and thermodynamic binding data. Enthalpy of binding was determined
for three titrations of each experiment and average values were compared. The intrinsic molar enthalpy change (ΔH), the binding
stoichiometry (n) and binding constant (K) for the binding process were obtained from the best fit of the calorimetric data. Gibbs free
energy of binding and Kd were calculated from binding affinity measurements, using ΔG = - R(T) ln(1/Kd), where R is the universal gas
constant and T is temperature in Kelvin. Entropy of binding was then estimated with ΔS = (ΔH – ΔG) / T, where ΔH was the average
enthalpy of binding.
MD simulations
MD simulations were carried out to study the effect of the employed IL water mixtures on FDH. Therefore the structure of FDH from
Candida biodinii (pdb code 5DN9) in complex with NAD+ and azide was downloaded from the PDB database. In preparation for the MD
simulations the azide was replaced with a CO2 molecule in both monomers of the enzyme using the replace function of YASARA 16.9.23
(YASARA Biosciences GmbH, Vienna, Austria).28 Next, the program was used to clean the structure and optimize the hydrogen bonding
network. To run the simulation in a mixed solvent system, a cubic simulation cell extending 10 Å around all atoms was created and the
AMBER15IPQ force field was chosen.29 A cell neutralization and pKa prediction experiment at pH 6.85 was carried out to neutralize the
simulation cell and assign correct protonation states of the amino acid side-chains. For creation of the mixed solvent system, all water
molecules were deleted from the soup and the solvent density was estimated using weighted densities of the pure compounds as found in
the literature.30 The specific IL molecules were created using the Build function of YASARA to fill the simulation cell with an aqueous ionic
liquid. In the next step the cell was filled with the created molecule using the respective densities specified in Table S3. The rest of the cell
was filled with water molecules by temporarily removing the IL molecules, filling the cell with water using the specified density (Table S3),
and adding the removed IL molecules back again. To remove all bumps between solvent molecules, the protein was fixed and an energy
minimization experiment was carried out. Then all atoms were freed again and the resulting scene was saved as solvent system for the MD
simulations. The simulations themselves were done using the YASARA macro md_run with the pressure control mode “Manometer”, pH
6.85 and 298 K over a time of 18.1 ns. The resulting simulation snapshots were analysed using the YASARA macros md_analyze and
md_analyzeres. For visualization in PyMOL (The PyMOL molecular Graphics System, Version 1.1 Schrodinger, Cambridge, MA, USA), the
simulation file was converted to pdb using the YASARA macro md_convert. The movies were prepared using OBS Studio
(https://obsproject.com)
Results and discussion
Fabrication of biocatalytic membrane
The immobilization was performed in a membrane assembled in a so-called sandwich mode (polypropylene layer – skin layer –
polypropylene support) so that the membrane support layer was positioned to face the feed, while underneath the skin layer an extra
polypropylene support was placed to act as a cushion to alleviate membrane compression and peeling of the skin layer (Figure 1). Based on
the “fouling-induced immobilization” method, the three enzymes (i.e. FDH, FaldDH and ADH) were simultaneously immobilized in the
membrane. During enzyme loading in the membrane, the permeate was collected over time, as presented in Figure S1. The mechanism of
membrane fouling induced by the enzyme solution filtration was categorized into four fouling models which are presented in Table S1 and
identified as standard, intermediate, complete blocking, and cake layer models.31, 32 The cake layer model, showing a high correlation
coefficient, was found to best describe the loading mechanism. In the initial stage of the filtration process, only the cake layer model fitted
the experimental data, which indicated that most enzymes are deposited on the skin layer of the membrane. As filtration time increased,
the experimental data also correlated well with other models because the fouling layer created by the enzymes acts as an additional
membrane.24 The morphology of the enzyme-immobilized membrane was characterized by SEM, and is shown in Figure 2. The figure
shows the skin layer, which is made up of regenerated cellulose (Figure 2a), and some enzyme aggregates adsorbed on the surface of the
polypropylene support fibres (Figure 2b, c). From the mathematical modelling and characterization by SEM, the fouling-induced enzyme
immobilization was found to involve at least two mechanisms: entrapment and adsorption. In Figure 2c it can be seen that some enzymes
were bound to the support fibres by hydrophobic adsorption (additional SEM pictures can be found in Figure S3). Based on mass balance
calculations by Bradford assay, 2.66 mg of protein was immobilized in the membrane, which corresponds to enzyme loading efficiency of
76%. Accordingly, the permeability dropped to 3.04 L m-2 h-1 bar-1 after enzyme loading, which is around 100 times lower than that of virgin
membrane (330 ±6 L m-2 h-1 bar-1).
4 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 5 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
Amicon cell
Figure 1. The immobilization strategy of enzymes in membrane for multi-enzymatic cascade reaction.
The synthesis of methanol from CO2 catalyzed by the three-enzyme cascade reaction in Tris-HCl buffer was performed with the
membrane loaded with enzymes. According to the reaction route, three moles of NADH are stoichiometrically consumed to produce one
mole of methanol in the cascade reaction. Therefore the methanol yield (Ymethanol) based on NADH may be calculated using the following
equation.
ꢍꢎꢏꢂꢐꢑꢒꢓꢔ
∗ꢕ ∗ 100
ꢋ
ꢌ
ꢃꢀꢄꢅꢆꢇꢈꢉꢊ % = ꢍ
ꢖꢗꢘꢙ,ꢁꢒꢁꢂꢁꢑꢔ
Where Cmethanol is the methanol concentration (mM), and CNADH,initial is the initial NADH concentration (mM).
After 30 min reaction, a methanol yield of 24.5% was obtained for the immobilized system, whilst practically the same yield (23.5%) was
obtained for an equivalent free enzyme system (using the same amount of enzymes in free form) (Figure 3a). The similar yield obtained
confirmed that no enzyme activity was sacrificed during immobilization. The low yield of methanol obtained was explained by the kinetics
of the reaction, as reported by Luo et al.24 As explained above, the reaction rate of the forward reaction (CO2 →formic acid) is much lower
than that of the reverse reaction (formic acid →CO2). For the second enzyme, FaldDH, the reaction (formic acid →formaldehyde) was also
found to be less efficient than the reverse reaction (formaldehyde → formic acid). However, for the third enzyme, ADH, the forward
reaction (formaldehyde →methanol) was much more efficient than the reverse reaction (methanol →formaldehyde). Additionally, Luo et
al. suggested that in order to be activated, the second reaction required a threshold concentration of formic acid. Therefore the first
reaction from CO2 to formic acid probably plays a decisive role in this cascade reaction.
Figure 2. SEM images of (a) Membrane skin layer (regenerated cellulose); (b) Support layer (nonwoven polypropylene); (c) View of support layer after
enzyme
immobilization.
Multi-enzymatic reaction in the ILs with co-immobilization of enzymes
The four selected ILs have a high capacity to absorb CO2, where one mole of IL can chemically absorb half a mole of CO2. The adsorption of
half a mole of CO2 by each mole of IL was proposed by Han et al, and similar mechanisms of CO2 adsorption by ILs were also reported by
other authors.33, 34 Han et al also demonstrated that [CH][AA] can be repeatedly recycled for CO2 adsorption, since CO2 can be desorbed
from the IL by bubbling with N2. Therefore, the process of CO2 adsorption is reversible, and ILs can provide FDH with a high enough CO2
concentration with a slow-releasing system, as required by FDH. As illustrated in Figure 3a (red points), the measured molar ratio of CO2 to
the ILs could slightly exceed 0.5. The minor excess in the CO2: IL molar ratio suggests that physical adsorption also contributes to the
uptake of CO2. Based on the CO2 solubility in the pure ILs, the CO2 concentration in the four kinds of 20% ILs was calculated to be in the
range from 467 to 714 mM, which is far higher than the CO2 solubility in water (33 mM).35 Furthermore, the CO2 adsorption rate in an
aqueous IL solution is faster than in the pure IL. Indeed, CO2 adsorption equilibrium is reached after 20 minutes in aqueous 5 wt% [CH][AA],
whilst the saturation of CO2 in pure [CH][AA] will take at least 4 h.36 Therefore aqueous ILs may be ideal media for enzymatic reactions.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 5
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 6 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
5x103
100
(a)
1.0
0.8
0.6
0.4
0.2
0.0
(b)
80
4x103
80
60
40
60
3x103
40
2x103
Free
system
20
1x103
20
0
0
10%
20%
40%
60%
0
[CH][Glu] concentration (V /V
)
Tris-HCl Tris-HCl [CH][Glu] [CH][Pro] [CH][Gly] [CH][His]
IL buffer
Figure 3. (a) Yield of methanol in free enzyme Tris-HCl (pH=7.0) and in four kinds of 20% ILs (left); solubility of CO2 in the absolute ILs with molar ratio of CO2
to the IL (MCO2/MIL) (left, red circle); CO2 concentration in ILs (left, blue triangle). All the enzymatic reactions were conducted in the membrane except “Free
system”. (b) Methanol production at different concentrations of [CH][Glu]. An enzymatic membrane reactor equipped with a 10 kDa regenerated cellulose
membrane (skin layer facing feed) was used. Recycling of immobilized enzymes was conducted using fresh NADH solution in Tris-HCl buffer and containing
20% [CH][Glu] buffer (NADH = 5 mM).
CO2 reduction to methanol with the enzyme-loaded membrane (i.e. FDH, FaldDH, and ADH) was performed in the presence of 20% ILs
(VIL/Vbuffer = 20%) (Figure 3a). The maximum methanol yield was achieved using [CH][Glu] and was ~3.5 times higher than the results of
multi-enzymatic reactions in Tris-HCl buffer. The same reaction was similarly examined for the other three ILs, but conversion of CO2 was
not significantly improved compared to the reactions in Tris-HCl buffer. In previous work, Liu et al. increased the pressure during operation
as an additional strategy to achieve high CO2 concentration in solution. They found that the reaction rate increased from (1.20 ± 0.09) ×
10−3 to (2.17 ± 0.07) × 10−3 µmol/min when CO2 pressure was increased from 0.2 MPa to 0.5 MPa, but almost no change was detected
when pressure was further increased to 1.0 MPa.37 The stable yield detected when pressure was increased above 0.5 MPa was attributed
to the fact that NADH concentration probably became the limiting factor. In our case, methanol yield could be improved almost three fold
by increasing the CO2 concentration 15 times in the [CH][Glu] system compared to reactions in Tris-HCl buffer.
To further improve the efficiency of CO2 conversion, the [CH][Glu] concentration was studied over the range from 10% to 60%. The yield of
methanol was clearly increased with increasing [CH][Glu] from 10% to 20%. CO2 captured in the solution was noticeably increased by
increasing [CH][Glu] concentration, thus shifting the reaction equilibrium towards the production of methanol. However, the yield of
methanol declined with further increase of the [CH][Glu] concentration from 20% to 60%. Such decrease can be explained by the negative
role that large amounts of [CH][Glu] exert on the enzyme activity, which has been attributed to the change of electrostatic balance
between charges in proteins when subjected to high salt (IL) concentrations. Additionally, stability, crystallization and aggregation
tendency of proteins have been also reported to change dramatically with increasing concentration of ILs in aqueous media.38-40 As a
matter of fact, it has been reported that enzymes lose all activity when in pure ILs.41, 42 Therefore, a balance between increasing CO2
solubility and maintaining enzyme performance has to be found. Water-based solvents with small amounts of salts (water-rich IL mixtures)
have been indeed reported to be the best media for proteins, which is consistent with our results.41
pH effect on enzyme activity
Liu et al. reported that the optimal pH for reduction of CO2 to formic acid by FDH was 6.0.42 However, the highest yield of formaldehyde
catalysed by FaldDH was produced at pH 7.0. ADH, the third enzyme in the sequence, was reported to have an optimal activity at pH 8.1 by
Shrabon.43 In our case, the optimum pH value when the reaction took place in buffer (without ionic liquids) was 6.5, which is similar to the
optimum pH of FDH (Figure 4a); this result suggests that the first reaction (CO2 → Formic acid) plays a decisive role in the performance of
the whole reaction. The relative activity was also found to significantly decrease when the pH value of the buffer was either below 6.5 or
above 7.5, which suggests a reduced enzyme activity under either acid or alkali conditions. Previous studies suggested that the activity and
structure of the enzymes might be affected by strongly acidic or alkaline media.44 Our results showed that the pH of the 20% [CH][Glu]
mixture (pH 6.85) was lower than the pH of the other ILs screened, which could also have influenced the higher conversion of the former
compared to the ILs evaluated. Unfortunately, these ILs were synthesized by acid-base neutralization of choline and amino acids.
Therefore lowering the pH of the other ILs was not be possible because a change of pH would have resulted in decomposition and amino
acid precipitation from the solution; thus no direct comparison among the four ILs at the same pH could be performed.
6 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 7 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
FDH+FaldDH+ADH
20%[CH][His]
20%[CH][Gly]
20%[CH][Pro]
20%[CH][Glu]
8.39
100
80
60
40
20
7.73
7.79
6.85
(a)
0
4
（b）
5
6
7
8
9
10
0
1
2
3
4
5
pH
6
7
8
9
pH
Figure 4. (a) Activity of the three enzymes at different pH from 4.5 to 9.5; (b) pH values in different kinds of 20% ILs (VIL/VBuffer) after 30 min bubbling with
CO2.
Recycle of the biocatalytic membrane
To evaluate the recyclability of the biocatalytic membrane, FDH, FaldDH and ADH were co-immobilized in the 10 kDa regenerated cellulose
membrane and fresh substrate was fed to the reactor after each of the six 30-minutes reaction cycles. The yield of methanol in the Tris-HCl
buffer was maintained at 20% - 40% for six runs (Figure 5), which confirmed that enzyme leakage was low and that enzyme activity was not
lost during immobilization. In the presence of [CH][Glu], the yield of methanol increased and was between 60% to 75% during the six runs.
A slight decrease in the yield (albeit not statistically significant) was observed after the second run, but no further decrease was observed
during the remaining four runs. The decrease of yield observed after the second cycle could be anyway ascribed to the decrease of
available NADH due to adsorption in the fouling layer.24
100
Tris-HCl buffer
20% [CH][Glu]
80
60
40
20
0
l
o
n
a
h
t
e
m
Y
1
2
3
4
5
6
Number of recycles
Figure 5. Stability of biocatalytic membrane in Tris-HCl buffer and 20% [CH][Glu]. One-way ANOVA statistical analysis showed that Ymethanol did not change
during the three hours of reaction.
Cofactor (NADH) regeneration with glucose dehydrogenase
NADH acts as a terminal electron donor and hydrogen donor in the cascade enzymatic reaction and is consumed stoichiometrically at each
step. As a result, three molar equivalents of NADH are consumed to transform one molar equivalent of CO2 in methanol. Efficient
regeneration of NADH is crucial for such a cascade enzymatic reaction since the cost of NADH is very high and in-situ generated NAD+ will
in turn inhibit the reduction of CO2 and promote the reverse oxidative reaction. In the current study, glucose dehydrogenase (GDH) was
used for NADH regeneration and the enzymatic reaction was coupled to the main cascade enzymatic reaction. Therefore GDH was
immobilized together with the three enzymes (FDH, FaldDH, ADH) in the 10 kDa regenerated cellulose membrane. D-glucose (50 mM) as
the GDH substrate was mixed with NADH solution (4 mL) containing 20% [CH][Glu], which was pre-bubbled with CO2 for 30 minutes. As
seen in Figure 6, the yield of methanol increased from 73% to 100% after 30 to 120 min reaction and then reached equilibrium after 120
min. The result indicates that NADH was efficiently regenerated with high activity of GDH. A number of researchers have also reported that
the yield of methanol increased with increasing NADH concentration.45-47 Total turnover number (TTN) of up to 10000 has been obtained
by using GDH for the regeneration of NAD+, as reported by Obon.48 The reduction rate from NAD+ to NADH catalysed by GDH is faster than
the oxidation of NADH to NAD+ by ADH, as reported by Fauziah et al.26 The reaction rate for the reaction from NAD+ to NADH catalysed by
GDH was 6.3 µmol/mg·min, while the reaction rate for the reaction from NADH to NAD+ catalysed by ADH was 4.7 µmol/mg·min.
Furthermore, converting NADH to NAD+ catalysed by ADH was much more efficient than with the other two enzymes (FDH and FaldDH).24
The reaction rate for the cascade reaction is limited by the slowest reaction. Therefore the reaction rate for converting NAD+ to NADH is far
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 7
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 8 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
higher than the oxidation of NADH by the three enzymes. Lastly, after two hours there was no further improvement in yield of methanol.
Probably, due to product inhibition by high methanol concentration in the solution, the reversible enzymatic reaction would not progress
any further.
120
NADH regeneration with GDH
100
80
60
40
30
60
90
120
150
180
Reaction time (min)
Figure 6. Methanol production as a function of reaction time with coupled GDH for cofactor (NADH) regeneration.
Molecular Dynamics (MD) Simulations
MD simulations on FDH with different solvents (i.e. water, 20%[CH][Gly], 20%[CH][Glu], 20%[CH][His] and 20% [CH][Pro]) were performed
in order to evaluate the effect of the aqueous IL on the enzyme. As depicted in Figure 7a, all simulations using aqueous ILs showed a similar
trend with respect to the root mean square deviation (RMSD) value of the Cα backbone over the MD simulation time, with the exception of
20% [CH][Glu] which approached a lower maximum value. The average RMSD values for the simulation in water and using the aqueous ILs
(except [CH][Glu]) ranged from 1.7 to 1.9 Å (Figure S2). In contrast, the average RMSD values for the 20% [CH][Glu] simulation of 1.3 Å
were markedly lower compared to the other solvents (Figure S2). These results might also indicate a stabilizing effect of the aqueous 20%
[CH][Glu] on protein structural integrity and therefore be another explanation for the good performance of the 20% [CH][Glu] as solvent
for the reaction system.
15
(a)
(b)
2
1
0
10
5
Water
[Ch][Glu]
[Ch][Gly]
[Ch][Pro]
[Ch][His]
0
]
]
]
]
s
i
H
u
o
2
y
l
G
r
P
l
G
[
r
[
[
]
[
]
h
C
]
h
]
e
t
h
h
C
a
W
0
5
10
t [ns]
15
20
C
[
C
[
[
[
%
%
%
0
2
%
0
2
0
0
2
Figure 7. (a) RMSD time course during the molecular dynamics simulations using H2O (black line), 20% [Ch][Gly] (blue line), 20% [Ch][Glu] (red line), 20%
[Ch][His] (green line) and 20% [Ch][Pro] (pink line) in water as solvents. (b) Distance of the Tyr73 OH-group and Phe 285 Cζ over the MD simulation time.
8 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 9 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
Figure 8. Surface representation of formate dehydrogenase (PDB 5DN9) at the beginning (A) and end (B-F) of MD simulation in different solvent systems:
H2O (B), 20% [Ch][Glu] (C), 20% [Ch][Gly] (D), 20% [Ch][His] (E), 20% [Ch][Pro] (F). For clarity the bulk solvent is not shown. The NADH cofactor is shown as
yellow sticks, the catalytically important Arg is shown as cyan sticks, Tyr73 and Phe285 are highlighted as greens ticks and green surface, CO2 is shown as
cyan
spheres.
The
solvent
accessible
active
sites
are
highlighted
with
the
black
arrows.
Closer inspection of the MD simulations revealed that in all solvent systems (except 20% [CH][Glu]) the active site, which binds the CO2
molecule, quickly became solvent accessible (Figure 8, Movies S1-5). This was especially pronounced in the simulation with water as
solvent where one of the CO2 molecules has left the active site after 12.5 ns (Figure 8b, Movie S1). As a measure of this surface opening,
the distance of the amino acid side chain OH-group of Tyr73 and the Cζ of the side-chain of Phe285 was recorded over the simulation time.
The 20% [CH][Glu] exhibited significantly different behaviour compared to the other solvent systems and had a mean distance of 4.7 Å
while that of the other solvent systems ranged from 6.9 – 8.5 Å (Figure 7b). Furthermore, detailed analysis of distance between residues
and ligands involved in formation of the enzyme transition state (TS, Figure 9A), as reported by Castillo, et al.49, revealed clearly that
formation of the TS would be favoured far more in the 20% [CH][Glu] system than in all the other systems (Table S4). This effect was also
reflected in the root mean square fluctuation (RMSF) values of the residues involved in positioning the CO2 molecule inside the active site
(
Figure 9B). The amino acid side chains of Asn119, Arg258 and His311 in particular showed a significantly lower fluctuation over the MD
simulation time in 20% [CH][Glu] compared to the other solvent systems. As an explanation for the high yield of methanol in 20% [CH][Glu]
we hypothesize that the increased rigidity of the FDH on the one hand also increases protein stability itself and on the other hand leads to
an increased residence time of CO2 in the active site in a more TS-like conformation than in the other studied solvent systems. This
prolonged TS-like residence time of CO2 in the active site increases the probability of a productive positioning of the reactants for the
formation of formic acid, which thus could result in a shift of the reaction equilibrium towards the less favoured reduction of CO2 to formic
acid.
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 9
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 10 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
1.6
(b)
1.2
0.8
0.4
0.0
V93 N119 T256 R258 D282 Q287 H311 S313
Figure 9. (A) Transition state of the FDH catalyzed reaction as proposed by Castillo et al. The NADH cofactor is highlighted in orange and the
substrate CO2 in pink. (B) RMSF values for the amino acid residues involved in substrate and co-factor binding as well as TS formation over
MD simulation time for the different solvent systems: H2O (olive bars), 20% [CH][Glu] (Cyan bars), 20% [CH][Pro] (Blue bars), 20% [CH][Gly]
(Orange bars), 20% [CH][Pro] (Purple bars).
For further validation of the MD simulation, isothermal titration calorimetry (ITC) was carried out to elucidate the interaction of FDH
with CO2 in the different solvents (i.e. water and 20% [CH][Glu]). Adsorption heat due to interaction between FDH with CO2 in water and
[CH][Glu] was fitted with the Nano-Analyze software (TA Instruments) using an independent model to obtain the Gibbs free energy (∆G).
As shown in Table S1, the △
G value of CO2 interactions with FDH in water was - 17.12 kJ mol-1, which is higher than that of CO2 and FDH
interaction in the presence of [CH][Glu] (-18.75 kJ mol-1). This indicates the binding of CO2 due to lower enzyme flexibility in 20% [CH][Glu].
Conclusions
A straightforward enzyme loading approach inspired by membrane fouling mechanisms provided a novel platform for conducting a
sequential enzyme reaction in the presence of ILs. The biocatalytic system had a promising capacity for enzyme loading and showed high
stability after several experimental cycles. Previous studies have shown that the first reaction of the sequence, conversion of CO2 to formic
acid, plays a decisive role in conversion of CO2 to methanol and represents the “bottleneck” that determines the progress of the whole
reaction. A strategy to enhance the conversion of this reaction step was attempted by increasing the concentration of CO2 in the reaction
system through adding ionic liquids as a co-solvent. The system that offered the best conversion results was an aqueous mixture 20%
[CH][Glu] in which CO2 concentration was around 15 times higher than in Tris-HCl buffer as the control.
Though CO2 concentration achieved was similar in the other ionic liquids selected, the reaction yield significantly increased only when
[CH][Glu] was used. The reason for this difference was further investigated. We found that pH might play a significant role in the reaction,
because slightly acidic pH seems to favour the conversion. The most convincing explanation, however, was provided by molecular
simulation dynamics. While CO2 easily diffuses out of the active site in the other ILs tested, and especially in water, the conformation of
FDH in the presence of [CH][Glu] is such that CO2 stays for a longer time in the vicinity of the active site of the enzyme. Such longer
retention times may therefore result in higher conversion of CO2.
Further systematic research on this topic must provide more specific information about the mechanisms and specific functional
groups of ionic liquids, which are responsible for the kinds of conformational enzyme changes that can support higher conversions.
Conflicts of interest
The authors declare no conflict of interest.
Acknowledgements
The authors are thankful for support from National Science Foundation of China (U1704251) and (U1662133), CAS 100-Talent Program
(2014).
References
1.
X. Z. Lim, Nature, 2015, 526, 628-630.
10 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 11 of 12
View Article Online
DOI: 10.1039/C8GC02230E
Journal Name
ARTICLE
2.
Y. M. Al-Saleh, G. Vidican, L. Natarajan and V. V. Theeyattuparampil, Futures, 2012,
44, 105-115.
3.
4.
X. D. Xu and J. A. Moulijn, Energ Fuel, 1996, 10, 305-325.
W. B. Hou, W. H. Hung, P. Pavaskar, A. Goeppert, M. Aykol and S. B. Cronin, Acs Catal
,
2011, 1, 929-936.
5.
6.
H. Li, P. H. Opgenorth, D. G. Wernick, S. Rogers, T. Y. Wu, W. Higashide, P. Malati, Y.
X. Huo, K. M. Cho and J. C. Liao, Science, 2012, 335, 1596-1596.
J. F. Shi, Y. J. Jiang, Z. Y. Jiang, X. Y. Wang, X. L. Wang, S. H. Zhang, P. P. Han
and C. Yang, Chem Soc Rev, 2015, 44, 5981-6000.
7.
8.
R. Obert and B. C. Dave, J Am Chem Soc, 1999, 121, 12192-12193.
U. M. Ulrich RUSCHING, Peter WILLNOW, Thomas HÖPNER, European Journal of Biochemistry
,
1976, 70, 325-330.
9.
X. L. Wang, Z. Li, J. F. Shi, H. Wu, Z. Y. Jiang, W. Y. Zhang, X. K. Song and Q. H. Ai,
Acs Catal, 2014, 4, 962-972.
10.
11.
12.
G. K. Cui, J. J. Wang and S. J. Zhang, Chem Soc Rev, 2016, 45, 4307-4339.
E. D. Bates, R. D. Mayton, I. Ntai and J. H. Davis, J Am Chem Soc, 2002, 124, 926-927.
B. E. Gurkan, J. C. de la Fuente, E. M. Mindrup, L. E. Ficke, B. F. Goodrich, E. A.
Price, W. F. Schneider and J. F. Brennecke, J Am Chem Soc, 2010, 132, 2116-+.
13.
B. A. Rosen, W. Zhu, G. Kaul, A. Salehi-Khojin and R. I. Masel, J Electrochem Soc
,
2013, 160, H138-H141.
14.
15.
D. T. Whipple and P. J. A. Kenis, J Phys Chem Lett, 2010, 1, 3451-3458.
W. Hussain, D. J. Pollard, M. Truppo and G. J. Lye, J Mol Catal B-Enzym, 2008, 55, 19-
29.
16. S. Datta, B. Holmes, J. I. Park, Z. W. Chen, D. C. Dibble, M. Hadi, H. W. Blanch, B. A.
Simmons and R. Sapra, Green Chem, 2010, 12, 338-345.
17.
18.
19.
M. Eckstein, M. Villela, A. Liese and U. Kragl, Chem Commun
,
2004, DOI:
10.1039/b401065e, 1084-1085.
R. M. Lau, F. van Rantwijk, K. R. Seddon and R. A. Sheldon, Org Lett, 2000, 2, 4189-
4191.
K. D. Weaver, H. J. Kim, J. Z. Sun, D. R. MacFarlane and G. D. Elliott, Green Chem
,
2010, 12, 507-513.
20.
21.
M. D. Fujita K, Forsyth M., Chem Commun, 2005, 38, 4804-4806.
F. M. Fujita K, MacFarlane DR, Reid RW, Elliott GD., Biotechnol Bioeng, 2006, 94,
1209-1213.
22. F. A. E. Silva, F. Siopa, B. F. H. T. Figueiredo, A. M. M. Goncalves, J. L. Pereira, F.
Goncalves, J. A. P. Coutinho, C. A. M. Afonso and S. P. M. Ventura, Ecoto' Environ
Safe, 2014, 108, 302-310.
23.
M. Petkovic, J. L. Ferguson, H. Q. N. Gunaratne, R. Ferreira, M. C. Leitao, K. R.
Seddon, L. P. N. Rebelo and C. S. Pereira, Green Chem, 2010, 12, 643-649.
J. Q. Luo, A. S. Meyer, R. V. Mateiu and M. Pinelo, New Biotechnol, 2015, 32, 319-327.
J. Q. Luo, F. Marpani, R. Brites, L. Frederiksen, A. S. Meyer, G. Jonsson and M.
Pinelo, J Membrane Sci, 2014, 459, 1-11.
24.
25.
26.
F. Marpani, Z. Sarossy, M. Pinelo and A. S. Meyer, Biotechnol Bioeng, 2017, 114, 2762-
2770.
27. S. De Santis, G. Masci, F. Casciotta, R. Caminiti, E. Scarpellini, M. Campetella and L.
Gontrani, Phys Chem Chem Phys, 2015, 17, 20687-20698.
28.
E. Krieger and G. Vriend, Bioinformatics, 2014, 30, 2981-2982.
29. K. T. Debiec, D. S. Cerutti, L. R. Baker, A. M. Gronenborn, D. A. Case and L. T. Chong,
This journal is © The Royal Society of Chemistry 20xx
J. Name., 2013, 00, 1-3 | 11
Please do not adjust margins

Please do not adjust margins
Green Chemistry
Page 12 of 12
View Article Online
DOI: 10.1039/C8GC02230E
ARTICLE
Journal Name
J Chem Theory Comput, 2016, 12, 3926-3947.
30.
L. del Olmo, I. Lage-Estebanez, R. Lopez and J. M. G. de la Vega, J Phys Chem B, 2016,
120, 10327-10335.
31.
32.
33.
A. Affandy, E. Keshavarz-Moore and H. K. Versteeg, J Membrane Sci, 2013, 437, 150-159.
L. Palacio, C. C. Ho and A. L. Zydney, Biotechnol Bioeng, 2002, 79, 260-270.
X. Y. Li, M. Q. Hou, Z. F. Zhang, B. X. Han, G. Y. Yang, X. L. Wang and L. Z. Zou,
Green Chem, 2008, 10, 879-884.
34.
S. Z. Jianmin Zhang, Kun Dong, Yanqiang Zhang, Youqing Shen, Xingmei Lv, Chem-Eur J
,
2006, 12, 4021=4026.
35.
36.
J. A. Dean, Lange's Handbook of Chemistry.
Y. C. Shengjuan Yuan, Xiaoyan Ji, Zhuhong Yang, Xiaohua Lu Fluid Phase Equilibr, 2017,
445, 14-24.
37.
38.
Y. Z. Wang, M. F. Li, Z. P. Zhao and W. F. Liu, J Mol Catal B-Enzym, 2015, 116, 89-94.
C. C. Hermann Weingärtner, Christian Herrmannc Phys Chem Chem Phys, 2012, 14, 415-426.
39. M. S. P. Marc L. Pusey, Megan B. Turner, and Robin D. Rogers, Crystal Growth ' Design
,
2007, 7, 787-793.
40.
C. J. B. Natalie Debeljuh, Luke Hendersonb, Nolene Byrne, Chem Commun, 2011, 47, 6371-
6373.
41.
42.
H. O. Yuki Kohno, Chem Commun, 2012, 48, 7119-7130.
W. F. Liu, Y. H. Hou, B. X. Hou and Z. P. Zhao, Chinese J Chem Eng, 2014, 22, 1328-
1332.
43.
44.
S. Sarcar, T. K. Jain and A. Maitra, Biotechnol Bioeng, 1992, 39, 474-478.
F. S. Baskaya, X. Y. Zhao, M. C. Flickinger and P. Wang, Appl Biochem Biotech, 2010,
162, 391-398.
45.
Y. J. Qianyun Sun, Zhongyi Jiang, Lei Zhang, Xiaohui Sun, Jian Li, Ind Eng Chem Res
,
2009, 48, 4210-4215.
46.
47.
D. D. Bilal El-Zahab, Ping wang, Biotechnol Bioeng, 2007, 99, 508-514.
J. D. Rémi Cazelles, François Fajula, Ovidiu Ersen, Simona Moldovanb, Anne Galarneau,
New J Chem, 2013, 37, 3721-3730.
48.
49.
M. A. Obón JM, Iborra JL., Biotechnol Bioeng, 1998, 57, 510-517.
M. O. R. Castillo, S. Martí, V. Moliner, J Phys Chem B, 2008, 32, 10012-10022.
12 | J. Name., 2012, 00, 1-3
This journal is © The Royal Society of Chemistry 20xx
Please do not adjust margins