Christina
Kohlmann
a,
Susanne
Leuchs
ab,
Lasse
Greiner
*ab and
Walter
Leitner
ac
aInstitut für Technische und Makromolekulare Chemie, RWTH Aachen University, Worringer Weg 1, 52074, Aachen, Germany. E-mail: greiner@dechema.de
bDECHEMA e.V. Karl-Winnacker-Institut, Theodor-Heuss-Allee 25, 60486, Frankfurt am Main, Germany
cMax-Planck-Institut für Kohlenforschung, Mülheim a.d.R., Germany
First published on 15th February 2011
The continuous biocatalytic synthesis of (R)-2-octanol carried out in an enzyme membrane reactor showed an excellent process stability of the biocatalysts, resulting in a turnover number of 15 million for the applied alcohol dehydrogenase. Utilisation of the ionic liquid AMMOENG™ 101 as a feasible cosolvent increased substrate concentration, and improved space time yields and turnover numbers of the cofactor by factors of 3 and 6, respectively. Moreover, 80% less waste was generated, while producing the same amount of product. Integrated product separation was realized viasolid phase extraction. Extraction of the applied solid phase with supercritical carbon dioxide allowed more than 30 reuses of the solid phase.
The application of non-conventional media could contribute to overcome the limitations implied by aqueous reaction environments. For example, ionic liquids (ILs) are powerful solubilisers of various substances and beyond that in biocatalytic reactions improve activities, stabilities and selectivities of biocatalysts, as well as having positive influences on enzyme kinetics and reaction rates.3 Furthermore, supercritical fluids (scFs) could become alternative extracting agents.4 Prevalent supercritical carbon dioxide (scCO2) is used in biocatalysis, as it reaches critical conditions at moderate temperatures and pressures (>31 °C; >7.83 MPa).5 In addition, it is non-flammable, non-toxic, environmentally benign and available in reasonable quantities and at reasonable cost.6 Notably, it is very promising for the extraction of reaction media containing ILs, as it is able to dissolve in ILs, whereas ILs cannot be dissolved in scCO2.7 Unfortunately, CO2 reacts with water to form carbonic acid, resulting in a pH drop in unsuitable reaction media.8 Moreover, CO2 tends to form carbamides with the amino residues of proteins, leading to a destabilised biocatalyst.9 Thus, either the contact of scCO2 with sensitive biocatalysts needs to be avoided or different scFs should be applied.6 A potential approach to achieve separation of the synthesis reaction and scCO2 extraction is its combination with solid phase extraction (SPE). SPE can be beneficial for process engineering, as it is simple and safe in handling, and can be easily automated.10 Besides, when using an appropriate eluent, the SPE material can be recycled effectively.
Within our work, we have focused on the Lactobacillus brevisalcohol dehydrogenase (LbADH)-catalysed reduction of 2-octanone with a limiting solubility of 7.8 mmol L−1 in aqueous phosphate buffer (Fig. 1). The enantioselective reduction of 2-octanone was used as a model reaction for membrane mediated biphasic reactors11 or in biphasic systems with LbADH.12 The water miscible IL AMMOENG™ 101 (Fig. 2) was applied as a performance additive, since it has previously been described as a potential solubiliser.3a For cofactor regeneration, the glucose dehydrogenase (GDH)-catalysed oxidation of β-D-glucose was chosen, as previous findings have shown it to be superior to substrate coupled regeneration. Reactions with or without the addition of IL were carried out continuously by applying an enzyme membrane reactor (EMR). To achieve integrated product separation, a stainless steel column filled with an appropriate solid phase extraction material was integrated into the product stream. Afterwards, the SPE column was recycled viascCO2 extraction. The setup used for this approach can be seen in Fig. 3.
![]() | ||
Fig. 1 LbADH-catalysed reduction of prochiral ketones to the corresponding (R)-alcohols with GDH-catalysed cofactor regeneration. |
![]() | ||
Fig. 2 Structure of the ionic liquid AMMOENG™ 101. |
![]() | ||
Fig. 3 Flow scheme for continuous synthesis in an enzyme membrane reactor (EMR), and subsequent SPE of the loaded column downstream and extraction with carbon dioxide (P: pressure transducer, IL: ionic liquid). |
![]() | ||
Fig. 4 Activity of LbADH as a function of reactant concentrations with and without IL; ![]() |
![]() | ||
Fig. 5 Activity of GDH as function of reactant concentrations with and without IL; ![]() |
All kinetic experiments concerning LbADH show activation by AMMOENG™ 101, therefore the IL is a beneficial solubiliser for this enzyme. Unfortunately, LbADH is inhibited by both the substrate and the product, though these inhibitions are less pronounced in the presence of the IL. As expected, ADH is not affected by glucose (data not shown), while its activity is decreased by the coupled product from cofactor regeneration. The coupled product gluconic acid δ-lactone is in equilibrium with the corresponding gluconic acid; the given concentrations are based on the added amount as this resembles the reaction conditions. Notably, coproduct inhibition is reduced by the addition of AMMOENG™ 101. This may be explained by interaction of gluconic acid with tertiary ammonium salts, which are used for reactive extraction.13 The activity of GDH is not directly influenced by the addition of AMMOENG™ 101. Also, 2-octanone and 2-octanol do not affect the enzyme. However, inhibition by high concentrations of gluconic acid δ-lactone (or the corresponding gluconic acid) is reduced as well.
With regard to kinetics in buffer, high concentrations of all the reactants decrease biocatalyst activity, whereas in the presence of the IL, not only increased substrate concentrations are feasible, but also all types of inhibitions are reduced or, as for gluconic acid δ-lactone, become negligible. To be able to carry out effective syntheses, continuous reactions applying an EMR were chosen as a promising strategy, as only negligible substrate concentrations are present when performed at high conversions.
As the kinetic investigations advised continuous reactions, where the long term performance of the biocatalyst is crucial, the storage stabilities of both enzymes were also determined. The experimental values for ADH stability were perfectly in line with previously published results;3a in pure buffer, LbADH showed a half-life of 41 h, whereas with the addition of 50 g L−1AMMOENG™ 101, improvement by a factor 1.5 to 61 h was possible. Unfortunately, the GDH had a half-life of 7.9 h in buffer that dropped to merely 2.5 h in the presence of the IL. Nevertheless, as storage stabilities are not directly comparable to process stabilities and might show similar trends, these findings do not preclude continuous reactions.
A wide range of SPE materials for different purposes are commercially available. Therefore, we selected different materials from manufacturer instruction guides and tested their ability to extract 2-octanone and 2-octanol from aqueous potassium phosphate buffer.
As emphasised by the maximal loadings of the different materials shown in Fig. 6, especially the polystyrenedivinylbenzene copolymer-based materials, HR-X and HR-P are very effective in binding 2-octanol and 2-octanone. With these materials, loadings of up to 0.70 g per gram of material was possible, while with the different silica based material, at best 0.35 g per gram of material could be extracted. For cost effectiveness, we chose HR-P over HR-X for integrated product separation.
![]() | ||
Fig. 6 Maximum loading of adsorption materials (light grey: 2-octanone, dark grey: 2-octanol). |
Reaction medium | Pure buffer | 50 g IL L−1 |
---|---|---|
a Calculated for the time with more than 90% conversion. | ||
c 2-octanone /mmol L−1 | 6.0 | 25.0 |
m ADH /mg | 0.6 | 1.2 |
m GDH /mg | 10.0 | 20.0 |
τ/h | 2.4 | 8.0 |
STY/mmol L−1d−1 a | 56.1 | 72.0 |
ee (%) | >99.9 | >99.9 |
TONADHa | 11.0 × 106 | 15.2 × 106 |
TONGDHa | 1.5 × 105 | 2.1 × 105 |
TONNADP+a | 58 | 245 |
E factora | 1420 | 380 |
In comparison to the continuous reduction to (S)-2-octanol in a membrane-mediated biphasic system carried out by Liese et al., 11 an 18-fold higher cofactor utilisation could be achieved, whereas the STY is about 2-fold lower. Eckstein et al.12 reported turnovers for the cofactor higher than 1000 in biphasic batches by recycling of the aqueous reactive phase, but at a lower STY of 20 mmol L−1d−1.
Surprisingly, the reaction progress of the continuous reaction runs demonstrate an excellent process stability of the biocatalyst system. While, in the beginning, full conversion was obtained for both reaction media, in the buffer, the conversion did not drop below 90% for 100 residence times (10 days, Fig. 7). Meanwhile, with the IL buffer mixture, 90% conversion was reached after 72 residence times (24 days, Fig. 8). Apparently, storage and process stability are not comparable. Also the TONs of both biocatalysts showed similar results in both reaction media. The selectivity of LbADH does not change in the presence of 50 g L−1AMMOENG™ 101, resulting in an enantiomeric excess of more than 99.9% in both cases. The addition of the IL allowed for an increase in substrate concentration. This gave a number of improvements that were mirrored in the key performance indicators (Table 1). Firstly, the STY was 1.3-fold increased. Secondly, the TONNADP+ was improved by a factor of 4.2. Finally, the E-factor as an indicator for produced waste per unit of product could be reduced 3.7 times, denoting that 70% less waste is generated.
![]() | ||
Fig. 7 Conversion X of 2-octanonevs. residence time or time, respectively. |
![]() | ||
Fig. 8 Conversion X of 2-octanonevs. residence time or time, respectively with the addition of 50 g L−1AMMOENG™ 101. |
![]() | ||
Fig. 9 Achieved adsorptions for two SPE columns loaded with HRP-Pvs. reuses applied in continuous reactions in (A) buffer and (B) with the addition of 50 g L−1AMMOENG™ 101. The dashed line represents the average loading of 95 mg in buffer and 51 mg with the addition of 50 g L−1AMMOENG™ 101 (an additional washing step with water was performed after the third reuse, see text). |
In buffer, the adsorption reached was relatively stable, whereas in buffer containing 50 g L−1AMMOENG™ 101, a loss in capacity was observed (Fig. 9). As a consequence, average adsorptions in buffer (90 mg for column 1 and 100 mg for column 2) were much higher than for buffer containing 50 g L−1AMMOENG™ 101 (50 mg for column 1 and 53 mg for column 2). Nevertheless, this trend is not due to ageing of the SPE material, but due to binding of the IL, as washing with water after the third reuse resulted in higher adsorptions. Furthermore, when changing the reaction medium, the respective average adsorption can be reached. In total, more than 30 reuses were carried out under the same conditions without an apparent loss of capacity. Moreover, recovery rates for scCO2 extraction of an average 75% in buffer and 62% with the addition AMMOENG™ 101 were achieved.
To determine the reaction progress, the following method was used: 80 °C (3 min), 10 °C min−1 to 120 °C (6 min), 40 °C min−1 to 180 °C (2 min). Retention times: 2-octanone (7.3 min), 2-octanol (9.4 min), internal standard 1-octanol (11.8 min).
To measure ee, the samples (250 μL) were mixed with N-MSTFA (50 μL), heated to 80 °C for 30 min and analysed by GC: 80 °C (3 min), 1 °C min−1 to 100 °C (5 min), 40 °C min−1 to 180 °C (2 min). Retention times: (R)-2-octanol (14.9 min), (S)-2-octanol (15.2 min).
C8: silica-based, pore size 60 Å, particle size 45 μm, specific area 500 m2 g−1, stable from pH 2 to 8, octyl phase, not end-capped, carbon content 8%.
C18ecf: silica-based, pore size 60 Å, particle size 100 μm, specific area 500 m2 g−1, stable from pH 2 to 8, octadecyl phase, end-capped, carbon content 14%.
C18 Hydra: silica-based, pore size 60 Å, particle size 45 μm, specific area 500 m2 g−1, stable from pH 2 to 8, octadecyl phase for polar analytes, carbon content 15%.
HR-X: hydrophobic polystyrenedivinylbenzene copolymer, pore size 55–60 Å, particle size 85 μm, spherical particles, specific area 1000 m2 g−1.
HR-P: highly porous polystyrenedivinylbenzene copolymer, particle size 50–100 μm, specific area 1200 m2 g−1.
Continuous reactions in pure buffer (100 mmol L−1potassium phosphate pH 7.0; 2.5 mmol L−1MgCl2) were carried out with 2-octanone (6 mmol L−1), glucose (100 mol L−1), NADP+ (0.1 mmol L−1), LbADH lyophilisate (0.6 mg) and GDH lyophilisate (10.0 mg). To adjust the residence time to 2.4 h, the volumetric flow was fixed at 6.6 mL h−1.
Continuous reactions in IL buffer mixture (100 mmol L−1potassium phosphate pH 7–2; 50 g L−1AMMOENG™ 101, 2.5 mmol L−1MgCl2) were carried out with 2-octanone (25 mmol L−1), glucose (150 mmol L−1), NADP+ (0.1 mmol L−1), LbADH lyophilisate (1.2 mg) and GDH lyophilisate (20.0 mg). To adjust the residence time to 8.0 h, the volumetric flow was fixed at 2.0 mL h−1.
A GC flow cell (mechanical workshop of the institute) was integrated into the product stream to monitor conversion. To enable integrated product separation, the outlet was further connected to a HPLC column (60 mm, ID 3 mm) filled with HR-P material (typically 0.20 g). To control the entire product separation, a second GC flow cell was integrated into the waste stream coming from the SPE column (Fig. 3). When full loading was reached, the column was exchanged and the product extracted using scCO2 (8.0 MPa, 45 °C, 0.1 mL min−1).
Footnote |
† This paper was published as part of the themed issue of contributions from the Green Solvents – Alternative Fluids in Science and Application conference held in Berchtesgaden, October 2010. |
This journal is © The Royal Society of Chemistry 2011 |