In comparison to chemical processes where heat exchangers as well as temperature and pressure control of the reaction are crucial parts, the key parts of bioprocesses are the biocatalysts, which can be enzymes or whole cells. These biocatalysts exhibit a certain activity and selectivity towards a specific substrate. Both activity and selectivity regarding a certain reaction can be characterized by kinetics as well as thermodynamics of the reaction [1, 2]. Apart from parameters such as temperature or pH, the solvent also plays an important role in enzyme activity and especially enzyme stability. All this needs to be quantified when setting up a bioprocess. Finally, questions regarding the choice of reactor type, catalyst recovery and downstream processing need to be answered .
Commonly accepted advantages of bioprocesses are the high selectivity of most biocatalysts as well as the usually mild reaction conditions, both resulting in low amounts of side products and thus high product quality . When setting up a bioprocess first of all the catalyst must be selected and characterized with an understanding of kinetics and appropriate reaction conditions. However, catalysts and very often also the substrates are relatively expensive. Additionally, the biocatalyst has a limited stability. Therefore, in the early stage of bioprocess development there is a strong interest in a low consumption of consumables as well as in a high reproducibility and reliability of the data obtained from the characterization [5–7]. A crucial point for bioprocesses is also the product recovery, because often relatively low product concentrations are obtained and products must be isolated from a liquid phase. Here an appropriate downstream process or in situ recovery strategy has to be chosen .
Classical approaches for the screening of enzyme activity are based on the use of multititer plates mostly made from polystyrene or glassy materials . These multi-titer plates represent an open batch system, which has two main drawbacks. Firstly, organic solvents must be circumvented because they might attack the plastic material. Secondly, volatile substrates or products cannot be used, because their evaporation results in huge errors of measurement. Therefore, the first set-up for the evaluation of a biocatalyst or reaction conditions for a bioprocess is a stirred tank reactor with a volume of usually 30–500 ml. Here, mostly standard glass hardware is used . This approach is fairly labor intensive and automation is difficult. Later, a high amount of substrate as well as enzymes is necessary to obtain the desired data in a sufficient density. However, for the process development and scale-up, (kinetic) data obtained at real process conditions are necessary.
This is where micro reaction technology can contribute its promising potentials [11, 12]. Using micro reactors a downscale of the reaction can be realized [13, 14], and also other unit operations, such as membrane separation processes as they are often used in biotechnology, can be characterized on a small scale [15, 16]. Usually the characterization of each unit operation of a process is done independently, where every single unit operation operates with its own conditions and limitations (‘process window’). However, micro system technology offers the possibility to characterize combinations of unit operations, whereby the process window of the combined systems is restricted by the overlap of the unit operations. Initially only process windows, which are acceptable for all necessary unit operations, are evaluated . Therefore, this approach can drastically reduce development times by omitting laborious investigations of conditions which are not applicable within the final process at any rate (Figure 1).
A biocatalytic process, consisting of continuous substrate feed, enzymatic reaction and continuous product removal, is presented below. The process was subdivided into its unit operations, which were considered individually.
First two different types of micro reactors for a carboligation catalyzed by benzoylformate decarboxylase [18, 19] are presented and one of these reactors is characterized in view of residence time distribution and flow conditions. In addition, the enzymatically catalyzed reaction was investigated in stirred vessel as well as in two micro reactors. Here, the comparability of stirred vessel and flow reactors should be reached .
To utilize a flow system in micro scale, a continuous product removal in a similar scale is required. Therefore, a product removal with membranes operating as contactor devices should be established and described in view of the degree of extraction and process stability.
Additionally, the usage of membranes for substrate feed was evaluated during this study. The relatively low solubility of benzaldehyde in the aqueous buffer led to a new approach for continuous substrate feed in a membrane module. A reproducible and constant benzaldehyde transfer rate across the membrane was the goal.
2 Materials and methods
2.1 Enzymatic assay
All chemicals used in this study were purchased from Sigma-Aldrich (Steinheim, Germany), unless indicated otherwise.
To obtain benzoyl formate decarboxylase (BFD) fermentation, harvesting and purification of the enzyme was done as described in [18, 19]. Enzymatic activity determination was carried out according to : one unit is defined as the amount of enzyme in milligrams transforming 1 mmol benzaldehyde and acetaldehyde to hydroxyl phenylpropanone (HPP) per min in 100 mm phosphate buffer containing 0.5 mm thiamine diphosphate (ThDP) and 2 mm MgCl2, 400 mm acetaldehyde and 40 mm benzaldehyde at pH 7.5 and 30°C. The tenfold acetaldehyde excess was chosen according to the Km values determined for BFD in previous studies .
2.2 Micro reactors and characterization
Two different types of micro reactors were used for this work. The μ-CaR was made of a stainless steel capillary with an inner diameter of 200 μl. Connectors and valves were provided by Swagelok. Analysis of residence time distribution was done as depicted in Figure 2. The dimensionless sum of exit age distribution F(θ) was calculated according to .
The μ-SiR was made of 500 μm borosilicate glass (Pyrex®, Plan Optik AG, Elsoff, Germany) and silicon wafers (Si-Mat, Kaufering, Germany) using standard photolithography and dry etching techniques to machine the channel pattern to the silicon. With advanced silicon etching (ASE) an anisotropic etch occurs resulting in rectangular channel geometry. The channels have a width of 300 μm, a height of 500 μm as well as a length of 77 mm. The fluidic connectors were etched to the Pyrex® using hydrofluoric acid. Finally, both wafers were mated irreversible by anodic bonding.
Enzymatic reactions were carried out as depicted in Figure 3 at 30°C. Substrate concentrations were adjusted as described for the enzymatic assay by keeping ratio of flow rates constant for all residence times. The aqueous substrate solution and the aqueous enzyme solution were pumped into the reactor and samples were taken at the outflow. The product concentration was determined using high pressure liquid chromatography (HPLC).
2.3 Membrane module and characterization
The membrane module is made of polytetrafluorethylen (Teflon®). Separated channels with an inner diameter of 2 mm and an inner length of 75 mm are available. This allows a fine adjustment of the accessible membrane area or parallel use of different membranes or different solvents. Figure 4 shows a model of one channel and one ferrule for potting and sealing the hollow fiber on both ends with an epoxide two-part glue Loctite 3340 (Henkel, Düsseldorf, Germany). The ferrules have a ¼ inch-28 UNF thread for easy capability of connecting tubes. The module can be operated co- or counter-current. Figure 5 illustrates the set-up for the analysis of the substrate feed. Figure 6 illustrates the set-up for the analysis of the product extraction.
Parameters such as temperature, enzyme concentration, flow rates and the reaction volume for the two micro reactors and the membrane processes are listed in Table 1.
Benzaldehyde (BA) and HPP concentrations in reaction buffer were determined by HPLC-UV analysis at 254 nm using a Merck KGaA (Darmstadt, Germany) Chromolith RP-8 phase at 40°C. Triethanolamine buffer and methanol (1:1) were used as an eluent with an isocratic flow rate of 1 ml/min. The reaction was quenched by adding stop solution, consisting of 90% acetonitrile and 5% phosphoric acid.
For the determination of HPP concentration in ethyl acetate a gas chromatography with flame ionization detector was used. An isocratic flow of 1.8 ml/min hydrogen was applied to a Varian CP-Chirasil-DEX CB phase from Agilent Technologies (Böblingen, Germany) at 110°C.
3 Results and discussion
Micro reactors are promising tools for screening of catalysts and reaction conditions . Here two different types of micro reactors were evaluated for the investigation of an enzymatic reaction utilizing volatile substrates.
As a model system the carboligation of BA and acetaldehyde (AA) catalyzed by the enzyme BFD from Pseudomonas putida was chosen [19, 23]. The product (S)-2-hydroxyphenyl propanone (HPP) is formed in an enantiomerically pure form (Figure 7). However, to date, the analysis of this reaction has been carried out in a glass vessel operated as a stirred tank reactor . Here, the evaporation of the highly volatile substrate acetaldehyde (boiling point 20°C) is problematic and requires additional equipment, such as reflux and/or closed pressure apparatus. Thus, using micro reactors can be highly beneficial .
The first micro reactor type used was a micro capillary reactor (μ-CaR) consisting of a steel capillary of 1.6 m and an inner diameter of 200 μm. Including connectors and valves the resulting reaction section volume was 790 μl. This reactor was characterized in view of residence time distribution (RTD) using the substrate benzaldehyde as a tracer and by measuring the absorption at 280 nm in the outflow of the reactor. The resulting F(θ) curves are shown in Figure 8.
To compare the μ-CaR to the stirred tank reactor plug performance of the flow reactor was evaluated. According to the first order boundary condition for open-open systems, an analytical solution from the dispersion model was used to determine Bodenstein numbers. Therefore the E(θ, Dax) curves were fitted to experimental data. Figure 9 illustrates the comparison between experimental data and the analytical solution, resulting in a Bodenstein number of 95. Various flow rates were adjusted to validate the flow conditions. The resulting Bodenstein numbers are summarized in Table 2.
The resulting Bodenstein numbers vary between 78 and 110. Thus, laminar plug-flow can be assumed for all evaluated Reynolds numbers . A RTD analysis of the stirred vessel is not required because of premixed substrates and the addition of enzyme dissolved in reaction buffer.
Because the investigated biocatalytic reaction (see Figure 7) has a typical initial reaction rate of 4 mm/min at a substrate concentration of 20 mm BA and 200 mm AA residence times of up to 15 min should be realized, thus flow rates were varied in the range of 0.05 to 1.00 ml/min.
The second reactor (μ-SiR) was made from borosilicate glass and silicon. Rectangular channels were etched into the silicon, capped with glass on both sides. This allows a visual access to the reactor. The reaction section volume was 115 μl and a Y-shaped connector was implemented directly on the microchip.
Figure 10 depicts concentration over time curves for the two investigated micro reactor types in comparison to the formerly used stirred vessel (‘batch’) (30 ml). To compare the stirred vessel with flow systems the reaction conditions must be comparable. Owing to the reactor set-up and enzyme solubility an equal enzyme concentration in all reactors was not achieved. For that reason, a normalized time tn, multiplying the applied enzyme activity in units per milliliter (U×ml) reaction volume with the batch and residence time, is used for comparison.
Evidently, the resulting curves are very similar and it can be concluded that the downscale of reaction volume has no impact on the reaction progress. When comparing the results of the continuous flow experiments with the batch experiments one has to keep in mind that each data point necessitates at least one residence time. Therefore, for five data points a reduction of enzyme and substrate consumption in comparison to the batch experiment at 30 ml scale by a factor of 7.6 and 52.2 for the μ-CaR and the μ-SiR, respectively, was achieved. The reactor volume was reduced at these experiments by a factor of 38 and 261, respectively.
A crucial point for many bioprocesses is product recovery, because often relatively low product concentrations are obtained and products must be isolated from a liquid phase. Here, the choice of an appropriate downstream process or in situ recovery strategy has to be done. To analyze the substrate supply as well as the product extraction, the respective unit operations were implemented on small scale applying membrane contactors [27, 28].
The biocatalytic reaction of BFD requires the substrate benzaldehyde. However, this substrate has a limited solubility of only 40 mm in aqueous solution. Dissolving the substrate in batch mode is a time-consuming step that requires intensive mixing for around 30 min. Therefore, a feed via a membrane is a good way to circumvent this procedure and enables the direct integration of the substrate supply into the micro system. Benzaldehyde is a hydrophobic compound that easily wets hydrophobic membrane materials such as polyvinylidene fluoride (PVDF), whereas the aqueous/polar reaction phase cannot go into the pores. The membrane acts as an interface between the two liquid phases and benzaldehyde diffuses into the reaction system. These membranes are available as flat membrane filters as well as micro hollow fibers. The huge advantage of micro hollow fibers is their low inner diameter of around 700 μm, which excellently fits to the micro reactors illustrated above . Therefore, this membrane type was chosen for the investigated system. The volume of the used hollow fiber was 30 μl at a length of 7.5 cm. For the biocatalytic reaction the pore size is a also crucial parameter, because a leaching of the enzyme through the membrane reduces the catalytic activity in the reactor. However, two immiscible liquid phases, the hydrophilic enzyme containing phase (aqueous) and the hydrophobic substrate phase, were applied, no leaching of the enzyme in the direction of the organic phase was observed and the micro hollow fiber was therefore operated as a liquid-liquid contactor .
As illustrated in Figure 11 saturation of the membrane with the substrate benzaldehyde and thus a constant benzaldehyde transfer rate (156 μmol/min in this case) is achieved within 90 min at a constant volume flow of 0.3 ml/min representing a residence time of 0.1 min . Both the residence time of 0.1 min as well as the transfer rate of 156 μmol/min are completely sufficient to provide enough substrate for the biocatalytic reaction. Carrying out the reaction with an activity of around 4 mm/min in a volume of 115 μl and at a residence of 0.1 min, a maximal consumption of 0.046 μmol is expected whereas a feed of 15.6 μmol is ensured.
The principle of a liquid-liquid contactor in the form of a micro hollow fiber was also applied for product extraction. Here, an organic solvent (ethyl acetate) was used to extract the product from the aqueous phase. The degree of extraction at different residence times up to 1 min was investigated using a 20 mm HPP solution (Figure 12). At a residence time of 1 min a degree of extraction of 50% was obtained. Comparing these results again with results obtained for the downscaled biocatalytic reaction, a product concentration of 2 mm HPP is to be expected in the case of a residence time of 1 min, illustrating the effectiveness of this established procedure.
In this study, the successful downscale of different unit operations for the biocatalytic carboligation of benzaldehyde and acetaldehyde was demonstrated. The reactor volume was reduced to 115 μl, thus enabling a substrate and enzyme saving by a factor of 52 in comparison to standard laboratory techniques. Additionally, the successful downscale of membrane based liquid-liquid contactors was shown, which allow, for example, the screening of solvents for extraction. Comparable volumes as well as residence times were realized, enabling the integration of all three unit operations.
Future work will deal with the combination of different unit operations (Figure 13) and the analysis of these combinations compared to the analysis of single unit operations. A more detailed analysis of the flow stream of the individual components within the channel is aimed for both micro reactors .
We thank Michael Hensges (Hamburg University of Technology) for technical support and Dr. Anne van den Wittenboer for helpful comments on this manuscript. We thank Prof. Dr. M. Pohl for providing the BFD plasmid.
List of abbreviations
Advanced silicon etching
Benzoyl formate decarboxylase
High pressure liquid chromatography
Micro capillary reactor
Micro silicon reactor
Residence time distribution
List of symbols
Axial dispersion coefficient
Density curve of dimensionless residence time
Dimensionless residence time
Sum curve of dimensionless residence time
Strompen S, Weiß M, Ingram T, Smirnova I, Gröger H, Hilterhaus L, Liese A. Biotechnol. Bioeng. 2012, 109, 1479–1489 Google Scholar
Vasic-Racki D, Kragl U, Liese A. Chem. Biochem. Eng. Q. 2003, 17, 7–18 Google Scholar
Liese A, Seelbach K, Wandrey C, Eds., Industrial Biotransformations, Wiley-VCH: Weinheim, 2006 Google Scholar
Hilterhaus L, Liese A. In White Biotechnology. Advances in Biochemical Engineering/Biotechnology, Ulber R, Sell D, Eds., Springer-Verlag: Berlin/Heidelberg, 2007, Vol. 105, pp. 133–173 Google Scholar
Fernandes P, Carvalho F, Marques MPC. Rec. Patents Biotechnol. 2011, 5, 160–173Google Scholar
Schmid A, Dordick JS, Hauer B, Kiener A, Wubbolts M, Witholt B. Nature 2001, 409, 258–268 Google Scholar
Lutz S, Bornscheuer UT, Eds., Protein Engineering Handbook, 1st ed., Wiley-VCH: Weinheim, 2008 Google Scholar
Schügerl K, Zeng AP, Eds., Tools and Applications of Biochemical Engineering Science. Advances in Biochemical Engineering Biotechnology, Springer-Verlag: Berlin/Heidelberg, 2002 Google Scholar
Mills PL, Quiram DJ, Ryleyc JF. Chem. Eng. Sci. 2007, 62, 6992–7010 Google Scholar
Chen BH, Micheletti M, Baganz F, Woodley JM, Lye GJ. Chem. Eng. Sci. 2009, 64, 403–409 Google Scholar
Klemm E, Rudek M, Markowz G, Schütte R. In Winnacker-Küchler Chemische Technik – Prozesse und Produkte, 5th ed., Dittmeyer R, Keim W, Kreysa G, Oberhol A, Eds., Wiley-VCH: Weinheim, 2004, Vol. 2, pp. 759–819 Google Scholar
de Bellefon C, Pestre N, Lamouille T, Grenouillet P, Hessel V. Adv. Synth. Catal. 2003, 345, 190–193 Google Scholar
Kolfschoten RC, Janssen AEM, Boom RM. J. Sep. Sci. 2001, 34, 1338–1346 Google Scholar
Wojik A, Marr R. Chem. Ingen. Techn. 2005, 77, 653–668 Google Scholar
Hessel V, Cortesea B, de Croona MHJM. Chem. Eng. Sci. 2011, 66, 1426–1448 Google Scholar
Kara S, Long WS, Berheide M, Peper S, Niemeyer B, Liese A. J. Biotechnol. 2011, 152, 87–92 Google Scholar
Marangoni AG, Ed., Enzyme Kinetics: A Modern Approach, John Wiley & Sons: New York, 2002 Google Scholar
Levenspiel O, Ed., Chemical Reaction Engineering, 3rd ed., John Wiley & Sons: New York, 1999 Google Scholar
Kara S. Online Monitoring of Biocatalytic 2-Hydroxy Ketone Synthesis under Ambient and High Pressure, Mensch und Buch: Berlin, 2012 Google Scholar
Song H, Ismagilov RF. J. Am. Chem. Soc. 2003, 125, 14613–14619 Google Scholar
Schlüter M, Hoffmann M, Räbiger N. Chem. Ingen. Techn. 2004, 76, 1682–1688 Google Scholar
Muller DH, Liauw MA, Greiner L. Chem. Eng. Technol. 2005, 28, 1569–1571 Google Scholar
Kopf F, Schlueter M, Kaufhold D, Hilterhaus L, Liese A, Wolff C, Beutel S, Scheper T. Chem. Ingen. Techn. 2011, 83, 1066–1073 Google Scholar
Gostoli C. In Mass Transfer – Advanced Aspects, Nakajima H, Ed., InTech: Rijeka, 2011 Google Scholar
Cussler EL. Diffusion – Mass Transfer in Fluid Systems, Cambridge University Press: Cambridge, 2009 Google Scholar
Tišma M, Zelić B, Vasić-Rački D, Žnidaršič-Plazlc P, Plazl I, Chem. Eng. J. 2009, 149, 383–388 Google Scholar
About the article
Janosch Fagaschewski (born 1983) studied chemical engineering at the Hamburg University of Technology. He obtained his diploma in 2010 working on a modular enzymatic process in micro scale. Since 2010, he has been working as a PhD student at the Institute of Technical Biocatalysis at the Hamburg University of Technology. His research interests include the scale down of enzymatically catalyzed reactions, the combination of several unit operations in micro scale and the development of screening tools for enzymes and process conditions in micro fluidic devices.
Sven Bohne (born 1979) studied electrical engineering specializing in microsystem technology at the University of Rostock and Hamburg University of Technology. He obtained his diploma in electrical engineering in 2009 and worked as a research assistant at the Institute of Microsystems Technology in Hamburg. There he started working on his PhD thesis ‘The Development of a Microfluidic Enzymatic Multistep Bioreactor’.
Dennis Kaufhold (born 1984) studied chemical engineering at the Hamburg University of Technology. He obtained his diploma in chemical engineering in 2009 working on the continuous enzymatic hydrolysis of natural oils. Since 2009, he has been working as a PhD student at the Institute of Technical Biocatalysis at the Hamburg University of Technology. He engages in the field of aeration of biocatalytic reaction systems, with major research on bubble free aeration with hollow fiber membrane contactors.
Jörg Müller (born 1944) studied electrical engineering at the Technical University Braunschweig, Germany. He obtained his diploma in 1969 and started his PhD studies at the Institute of HF-Technology at TU Braunschweig (H.-G. Unger), receiving his PhD in 1972. In 1979, he finished his Habilitation there. He worked as head of R&D ‘high-frequency diodes’ at Siemens AG, Germany, until 1983. From then to 2012 he was head of the Institute of Microsystems Technology at the Hamburg University of Technology, Germany. His research interest includes micro systems for biotechnology, gas analytics and minimal invasive medicine, as well as integrated optics for telecommunications and sensors. In 2009, his research on a planar integrated micro mass spectrometer (PIMMS) was honored with the AMA innovation award.
Lutz Hilterhaus (born 1978) studied chemistry specializing in biochemistry at the University of Münster, Germany. He obtained his diploma in chemistry in 2003 and worked as a research assistant at the Institute of Biochemistry in Münster. From 2004 to 2007 he did his PhD studies in the working group of Prof. Andreas Liese at the Institute of Technical Biocatalysis at Hamburg University of Technology (TUHH). After a Postdoc in the working group of Prof. Uwe Bornscheuer (University of Greifswald) in 2008, he returned to Hamburg where he is head of a junior research group ‘MicroBioTechnology’ since 2009.
Published Online: 2012-08-25
Published in Print: 2012-08-01