Directed evolution of formate dehydrogenase from Candida boidinii for improved stability during entrapment in polyacrylamide Marion B. Ansorge-Schumacher 1 , Heike Slusarczyk 2 , Julia Schu ¨ mers 1 and Dennis Hirtz 1 1 Department of Biotechnology, RWTH Aachen University, Germany 2 Institute of Enzyme Technology, Heinrich-Heine-University Du ¨ sseldorf, Research Center Ju ¨ lich, Germany Entrapment in polymeric matrices has long since become an important technique to improve recycling, handling, and mechanical strength of delicate biocata- lysts during application in large-scale organic synthesis. It also enables reactions with otherwise instable or cofactor-dependent enzymes in organic solvents [1,2] and can thus enlarge the scope of industrial biocataly- sis. However, while entrapment matrices can in princi- ple be formed from many monomeric or polymeric, natural or synthetic compounds [3], only a few are strong enough to withstand the chemical stress and mechanical forces in a technical process. Very suitable properties can be found in polyacrylamide (PAA) gels which combine stability under almost all relevant reaction conditions [4] with high elasticity and low abrasion in stirred-tank reactors [5,6]. The network structure of PAA can easily be adapted to ensure opti- mal retention of any biocatalyst [4], while no ionic interaction with entrapped enzymes occurs [7,8]. In spite of this, PAA gels have rarely been applied as entrapment matrices for biocatalysts [9] because of the detrimental effect that the entrapment process can exert on the activity of enzymes [5,6,10,11]. An exact explanation for this effect is not known to date. How- ever, irreversible changes of amino acid residues caused by some of the compounds participating in matrix for- mation are indicated [12–16]. In this study, we explored the possibility to increase the stability of isolated enzymes during entrapment in PAA gels. This was done by means of error-prone PCR and a screening method which employed selected components of PAA gels instead of the gel itself. As an investigation system, formate dehydrogenase [(FDH) E.C.1.2.1.2] from Candida boidinii was chosen for its outstanding importance as a cofactor regener- ation system in biocatalyzed syntheses [17]. Keywords directed evolution; entrapment; formate dehydrogenase; polyacrylamide; stabilization Correspondence M. B. Ansorge-Schumacher, Department of Biotechnology, RWTH Aachen University, D-52056 Aachen, Germany Fax: +49 241 8022387 Tel: +49 241 8026620 E-mail: m.ansorge@biotec.rwth-aachen.de Website: http://www.biotec.rwth-aachen.de/ biokat (Received 4 April 2006, revised 19 June 2006, accepted 26 June 2006) doi:10.1111/j.1742-4658.2006.05395.x In two cycles of an error-prone PCR process, variants of formate dehy- drogenase from Candida boidinii were created which revealed an up to 4.4-fold (440%) higher residual activity after entrapment in polyacrylamide gels than the wild-type enzyme. These were identified in an assay using sin- gle precursor molecules of polyacrylamide instead of the complete gel for selection. The stabilization resulted from an exchange of distinct lysine, glutamic acid, and cysteine residues remote from the active site, which did not affect the kinetics of the catalyzed reaction. Thermal stability increased at the exchange of lysine and glutamic acid, but decreased due the exchange of cysteine. Overall, the variants reveal very suitable properties for application in a technical synthetic process, enabling use of entrapment in polyacrylamide as an economic and versatile immobilization method. Abbreviations APS, ammonium peroxodisulfate; FDH, formate dehydrogenase; PAA, polyacrylamide; TEMED, N,N,N¢,N¢-tetramethylethylenediamine. 3938 FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS Results and Discussion The success of error-prone PCR in the directed evolu- tion of biocatalysts strongly depends on the screen applied to identify improved properties. Best results are usually obtained when screening conditions resem- ble the conditions during application as closely as possible [18]. Consequently a screen involving entrap- ment of FDH-variants in PAA would provide most suitable conditions for the identification of variants with an improved stability towards entrapment in this matrix. Because of the high demands of this method in terms of materials and time, and the inapplicability of simple optical assays for the determination of activity in opaque gel matrices, such an approach was not feas- ible in this study. An alternative screen exerting the effects of gel formation without employing gel forma- tion itself was therefore developed. Effects of PAA building blocks on wild-type FDH Formation of PAA gels involves the cross-linking of acrylamide and bis-acrylamide monomers in a radical polymerization process employing ammonium peroxo- disulfate (APS) and N,N,N¢,N¢-tetramethylethylene- diamine (TEMED) as radical-forming agent and enhancer, respectively [19,20]. The acrylamide concen- tration for the entrapment of enzymes can range between 5% (w ⁄ v) [10] and 30% (w ⁄ v) [21], depending on the required network density [4]. When FDH was entrapped in 8% (w ⁄ v) PAA, the activity of the immobilisates was only $10% of the activity that had been expected from the amount of enzyme introduced into the matrix. This was in accord- ance with the many reports on the severe deactivation of enzymes during entrapment in PAA [5,6,10,11]. The incubation of FDH with only one or two building com- pounds of PAA, allowing no polymerization, demon- strated that this deactivation was to a large extent a result of the mere presence of monomers and auxiliaries (Table 1). Compared with the enzyme stored in plain potassium phosphate buffer (pH 7.5), the half-life of wild-type FDH (wt-FDH) in 8% (w ⁄ v) acrylamide (the term ‘acrylamide’ always referring to a mixture of acryl- amide and bis-acrylamide in a ratio of 37.5 : 1) decreased by 91.8% to 111 min. While this observation was still in accordance with the deactivation of enzymes by acrylamide monomers reported by Dobryszycki et al. [22,23], it was also observed that the deactivation of FDH was enhanced when acrylamide was combined with 0.1% (w ⁄ v) of APS or TEMED, decreasing half- life by 96 and 97.2% to 45 and 38 min, respectively. The half-life of FDH was also heavily affected in a mixture of APS and TEMED, while neither of these components alone exerted a strong effect on stability. This indicates a cooperation of the compounds in the deactivation of enzymes independent of the actual polymerization pro- cess. With regard to the identification of FDH variants with a higher stability towards the entrapment in PAA, this finding means that complete gel formation is not required for successful screening, but employment of as many building blocks and auxiliaries as possible is favourable. FDH variants with improved stability in solutions of acrylamide/TEMED Error-prone PCR applied to the wt-FDH gene at a Mn 2+ concentration of 0.05 mmolÆL )1 yielded about 3500 recombinant clones, 70% of which expressed act- ive FDH. This was estimated from a qualitative activ- ity staining on agar plates. For quick identification of active FDH variants with improved stability towards PAA entrapment, their residual activity after 30 min of incubation in a buffered solution of 8% (w ⁄ v) acryla- mide and 0.1% (w⁄ v) TEMED in relation to their activity in plain buffer was determined. The composi- tion of this activity screen was based on the findings about the effect of PAA building blocks on wt-FDH described above. The incubation time was chosen on the assumption that PAA formation is usually comple- ted within 30 min [5,6,11], i.e. the presence of devasta- ting monomers becomes negligible after this time. Among 1764 FDH variants, three were considerably less affected by acrylamide and TEMED than wt-FDH. The residual activity of one of these variants was 70%, the other two revealed a residual activity of 90%. Related to wt-FDH, which had a residual activ- ity of only 44% under the same conditions, this was an improvement of 59 and 105%, respectively. The half-life of FDH increased from 38 min to 104 min (2.7-fold), 150 min (3.9-fold), and 154 min (4.1-fold), respectively. The mutations underlying these improve- ments are given in Table 2. Table 1. Half-life (t 1 ⁄ 2 ) of wt-FDH in the presence of compounds involved in PAA formation. The term ‘acrylamide’ refers to a mix- ture of acrylamide and bis-acrylamide in a ratio of 37.5 : 1. Composition t 1 ⁄ 2 (min) Buffer 1346 0.1% APS 763 0.1% TEMED 890 0.1% APS, 0.1% TEMED 84 8% acrylamide 111 8% acrylamide, 0.1% APS 54 8% acrylamide, 0.1% TEMED 38 M. B. Ansorge-Schumacher et al. Stabilisation of FDH for entrapment FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS 3939 The considerable stabilization of FDH-K35T towards deactivation in the presence of acrylamide and TEMED by replacement of a lysine residue in position 35 confirms the findings of Cavins and Friedman [12] and Bordini et al. [16], who reported an interaction of acrylamide with the e-amino func- tion of lysine residues. Even more effective was the replacement of glutamic acid in position 53, which was detected in both the variants with the best resid- ual activity. This was an interesting result, as gluta- mic acid had not been recognized as a special target of acrylamide or TEMED before. It was also surpri- sing that such good results were obtained with valine as replacement for glutamic acid, given that Perez et al. [15] had found a strong interaction between valine residues and acrylamide. Possibly, an even higher stability towards acrylamide would be created if this randomly inserted valine was exchanged for a more inert amino acid by site-directed mutagenesis. In contrast, no obvious effect on stability was achieved when additional to the replacement of glutamic acid in position 53, a lysine residue in posi- tion 56 was replaced by arginine. As the 3D struc- ture of FDH from C. boidinii has not been solved to date, the distinct locations of the mutations were not directly accessible. However, homology modelling according to Slusarczyk et al. [26], implies that all mutations are peripheral to the protein and remote from the active site. A salt bridge that is probably formed by the residues E53 and K56 in the wild- type FDH would be destroyed in the variants FDH- E53V and FDH-E53V ⁄ K56R. As a consequence, the flexibility of the enzyme could be increased. The gene coding for variant FDH-E53V ⁄ K56R was used as template for a second error-prone PCR, con- sidering that the replacement of the lysine residue had no negative effect on the residual activity of FDH in the presence of acrylamide and TEMED, and might exert a positive effect under reinforced screening condi- tions. This second error-prone PCR induced a higher mutation rate by using a Mn 2+ concentration of 0.15 mmolÆL )1 [24,25] and yielded 50% of recombinant clones expressing active FDH. For identification of FDH variants with further improved stability towards PAA entrapment, the concentration of acrylamide in the activity screen was increased to 15% (w ⁄ v). At this concentration, the template FDH-E53V ⁄ K56R had a residual activity of 22%. Among 1092 FDH-variants of the second genera- tion, again three with improved stability were found. All of them exerted an activity of 120% when incuba- ted in 15% (w ⁄ v) acrylamide ⁄ 0.1% (w ⁄ v) TEMED (Table 2), which was an improvement of 5.5-fold compared with the template FDH-E53V ⁄ K56R. The half-life of these variants in a solution of 8% (w ⁄ v) acrylamide and 0.1% (w ⁄ v) TEMED was about 1600 min, which was 42.1-fold more than that of wt- FDH (38 min). In all three variants, this considerable improvement was caused by an identical mutation, the exchange of cysteine in position 23 for serine. Consid- ering that Chiari et al. [13] observed an at least two- fold stronger effect of acrylamide on cysteine than on other amino acids, this is quite intelligible. It is an interesting result, however, that the randomly inserted mutation C23S should be identical to the one that Slusarczyk et al. [26] had identified as being most effective for stabilizing FDH during technical applica- tion. This finding indicates that the cysteine residue in this position might play a general role in FDH-stabil- ity. It is again a residue located rather at the periphery of the enzyme molecule than anywhere near the active site [26]. Biochemical properties of mutant FDHs Before investigating the performance of the new FDH-variants in PAA, their overall suitability for application was checked by comparing their kinetics and thermal stability to the properties of wt-FDH. The results of this study are summarized in Table 3. It should be noted that discrepancies of the values pre- sented herein for wt-FDH from formerly published data is due to the use of a different analysis software: In contrast to scientist for windows, which was used by Slusarczyk et al. [26], excel for windows, employed herein, takes into account the inhibition Table 2. Characteristics of FDH variants with increased activity towards acrylamide (AA) ⁄ TEMED. Percentage values of half-life (t 1 ⁄ 2 ) are related to the half-life of wt-FDH under the same condi- tions (38 min, see Table 1). Codon exchanges 8% AA ⁄ 0.1% TEMED Fold increase Residual activity t 1 ⁄ 2 (min) Variants of first generation K35T aaa fi aca 70% 104 2.7 E53V gaa fi gta 90% 150 4 E53V ⁄ K56R gaa fi gta 90% 154 4.1 aaa fi aga Variants of second generation E53V ⁄ K56R ⁄ C23S gaa fi gta 120% a 1600 42 aaa fi aga tgt fi agt a Residual activity was measured in 15% AA ⁄ 0.1% TEMED. Stabilisation of FDH for entrapment M. B. Ansorge-Schumacher et al. 3940 FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS constants for formate and NAD + and thus leads to slightly different, but probably more exact results. It was found that none of the mutations in FDH exerted a considerable effect on its kinetics. FDH- E53V ⁄ K56R and FDH-E53V ⁄ K56R ⁄ C23S had a slightly decreased K M of formate, while K M of NAD + increased slightly in FDH-E53V ⁄ K56R ⁄ C23S. Inactivation by NADH remained almost unchanged in all variants, and temperature optima were in the same range as of wt-FDH. The same holds true for the inactivation temperature. Thus, the catalytic properties of all FDH-variants are comparable to those of wt-FDH. Thermal half-life of all three first-generation FDH- variants at 50 °C increased by 12–27% compared with wt-FDH. The best improvement was found in FDH-E53V ⁄ K56R, which had also revealed the best stability towards the presence of acrylamide ⁄ TEMED. Thus, the possibly higher molecular flexi- bility of this variant due to the disruption of a per- ipheral salt bridge (see above) had no negative influence on the thermostability of the enzyme. In contrast to the stabilization towards acrylamide ⁄ TEMED, however, thermal stabilization was slightly affected by the exchange of lysine in position 56 for arginine. This can be concluded from the additional improvement in thermostability of FDH-E53V ⁄ K56R compared to FDH-E53V (Table 3). Mutation C23S, which dominated the second generation of FDH var- iants and had a highly beneficial effect regarding the stability in acrylamide ⁄ TEMED, had a detrimental effect on thermal half-life of the enzyme. Compared with wt-FDH, half-life decreased by 2.5-fold; further- more, in the template enzyme, FDH-E53V ⁄ K56R the decrease was 3.1-fold. This result is in accordance to the findings of Slusarczyk et al. [26], who observed a decrease in half-life of their mutant FDH-C23S of 80% at 50 °C, and confirms the special relevance of C23 for FDH stability. The better performance of our FDH variant compared with FDH-C23S is obvi- ously a result of the stabilizing effects of the addi- tional mutations E53V and K56R which were introduced during the first step of evolution. Stability of mutant FDHs in PAA Finally, the performance of the FDH variants after entrapment in PAA was investigated by measuring the activity in comparison to equally entrapped wt-FDH. For entrapment, acrylamide concentrations of 8% (w ⁄ v) as well as 15% (w ⁄ v) were used to ensure a close rela- tionship to the formerly employed screening conditions. However, the concentrations of TEMED and APS in the polymerization mix had to be increased to 1% (w ⁄ v) and 5% (w ⁄ v), respectively, in order to obtain stable and reproducible gel beads within a polymerization time of 30 min. Activity was measured after a reaction time of 60 min. The FDH variants of both generations had a clearly increased residual activity after entrapment in PAA compared to wt-FDH (Fig. 1). The improvement was more pronounced in the variants of the second evolu- tionary step, and the difference in stability between variants from the first and second generation increased with increasing concentration of PAA. These findings were in very good accordance with the behaviour of the variants under screening conditions. Of course, based on the observed half-lives in a solution of 8% Table 3. Kinetic constants, temperature optima and half-life of FDH-variants. FDH variant K M,formate (mmolÆL )1 ) K M,NAD (lmolÆL )1 ) K I,NADH (lmolÆL )1 )T opt (°C) T m (°C) t 1 ⁄ 2 at 50 °C (min) wt-FDH 5.9 50 4 60 62 365 FDH-K35T 6.2 49 5 55 58 408 FDH-E53V 5.2 49 3 55–60 55 433 FDH-E53V ⁄ K56R 5.0 52 8 60 60 462 FDH-E53V ⁄ K56R ⁄ C23S 4.5 58 5 55–60 59 148 Fig. 1. Activity of wt-FDH and FDH-variants in 8% (w ⁄ v) PAA (grey columns) and 15% (w ⁄ v) PAA (black columns). The activity was determined after 1 h of reaction at 30 °C, activity of wt-FDH was set to 100% after determination in the respective gel. M. B. Ansorge-Schumacher et al. Stabilisation of FDH for entrapment FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS 3941 acrylamide and 0.1% TEMED, a lower activity of FDH-K35T and a more similar performance of vari- ants FDH-E53V and FDH-E53V ⁄ K56R in PAA had been expected. The differences could be explained by inaccuracies in determination of CO 2 . Also, however, it is possible that the presence of APS, the increased concentrations of APS and TEMED, and the polymer- ization process itself affect the FDH variants in differ- ent ways, depending on their respective mutations. Indeed, an additional effect of these factors is indica- ted by the overall lower stabilization of all variants in PAA than in polyacrylamide ⁄ TEMED. Conclusions In principle, screening with a mixture of acrylamide and TEMED turned out to be well suited to resembling the deactivating forces of PAA formation on FDH, and thus enable the identification of variants with consider- ably improved stability. Thus, the adaptation of syn- thetically valuable enzymes for a technically and economically reasonable immobilization in PAA is now possible. Further improvements could probably be achieved when additional combinations of PAA forming compounds, such as acrylamide ⁄ APS or TEMED ⁄ APS, and higher concentrations of both APS and TEMED were employed. Also, combinations of a variety of favourable mutations by in vitro recombination methods such as DNA-shuffling [27] or staggered extension [28] could lead to considerably improved stability. Experimental procedures Materials Buffer salts and chemicals were of analytical grade and pur- chased from Fluka (Neu-Ulm, Germany), Roth (Karlsruhe, Germany), or Merck (Darmstadt, Germany). Restriction enzymes, DNA-modifying enzymes, and dNTPs were obtained from Roche Diagnostics (Mannheim, Germany). Markers for DNA and protein analysis, and PCR buffer were purchased from Invitrogen (Karlsruhe, Germany). Error-prone PCR and cloning Ten nanograms of the expression plasmid pBTac-FDH [26] and 20 pmol of each of the primers pBTacF1 (5¢-TG CCTGGCAGTTCCCTACTC-3¢) and pBTacR2 (5¢-CGA CATCATAACGGTTCT GG-3¢) were added to a mixture of 10 lL mutagenesis buffer [0.1 molÆL )1 Tris ⁄ HCl, pH 8.3; 0.5 molÆL )1 KCl, 70 mmolÆL )1 MgCl 2 , 0.1% (w ⁄ v) gelatine], 0.05 or 0.15 mmolÆL )1 of MnCl 2 , and 10 lL of dNTP-mix (10 mmolÆL )1 dCTP, 10 mmolÆL )1 dTTP, 2 mmolÆL )1 dATP, 2 mmolÆL )1 dGTP), 99 lL double distilled water and 1 lL (1 U) Taq-polymerase. Amplification was conducted in a Biometra 500 PCR-cycler (Biometra, Go ¨ ttingen, Germany) applying one cycle at 95 °C for 5 min, 25 cycles of 5 min at 94 °C, 1 min at 50 °C, and 1 min at 72 °C each, and at last one cycle at 72 °C for 5 min. The PCR fragments were puri- fied, ligated into expression vector pBTac2 and transformed into Escherichia coli JM101. The analysis of the nucleotide sequence was done by Sequiserve (Vaterstetten, Germany). Pre-selection of active clones Colonies of E. coli were fixed on agar plates by overlaying them with a solution of agar (1.6% w ⁄ v in 0.1 molÆL )1 potas- sium phosphate buffer, pH 7.5, containing 0.2% v ⁄ v Triton- X-100 and 10 mmolÆL )1 EDTA) at a maximum temperature of 65 °C. When the layer of agar had cooled down to room temperature and became solid, the plates were treated three times with a solution of 0.2% (v ⁄ v) Triton-X-100 and 10 mmolÆL )1 of EDTA in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5) and another three times with potas- sium phosphate buffer (0.1 molÆL )1 , pH 7.5). Each treatment was performed for 10–15 min. The plates were then overlaid with a first dyeing solution (1.25 molÆL )1 sodium formate, 0.2 gÆL )1 phenazinethosulfate, and 2 gÆL )1 nitrotetrazolium- blue chloride in 0.1 molÆL )1 of potassium phosphate buffer, pH 7.5) and incubated in the dark for 10 min. A second dye- ing solution (50 mmolÆL )1 NAD + in 0.1 molÆL )1 potassium phosphate buffer, pH 7.5) was added in a ratio of 1 : 100 (v ⁄ v) and the plates were gently moved in the dark until vio- let spots became visible. These spots marked colonies expres- sing active FDH. The dyeing mixture was removed, the plates were washed with water and left to dry. Determination of FDH activity Activity of native FDH was determined in a spectropho- tometer (Beckmann DU Ò series; Beckmann, Fullerton, CA, USA) or in a microtitre plate reader (ThermoMax; Molecu- lar Devices, Ismaning, Germany) at 340 nm and 30 °C. For measurements in the spectrophotometer, the assay mixture contained 0.25 molÆL )1 sodium formate and 1.7 mmolÆL )1 NAD + in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5). For measurements in the microtitre plate reader, the assay mix was composed of 0.6 molÆL )1 sodium formate and 3.6 mmolÆL )1 NAD + in 0.1 molÆL )1 potassium phos- phate buffer (pH 7.5). The reaction was started by adding FDH and was monitored over 2 min. From the increase in extinction, activity was calculated using the equation A½U=ml¼ DE à V total V enzyme à e à d with DE being the extinction increase within 1 min, V total the total volume of the assay mixture, V enzyme the volume Stabilisation of FDH for entrapment M. B. Ansorge-Schumacher et al. 3942 FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS of added enzyme solution, d the layer thickness of the cuvette and e the extinction coefficient of NADH + H + (6.22 mLÆlmol )1 Æcm )1 ). One unit is defined as the amount of enzyme that catalyses the reduction of 1 lmol of NAD + per min at pH 7.5 and 30 °C. Screening for improved FDH stability E. coli colonies expressing active FDH were transferred from agar plates into 1.2 mL LB medium (containing 100 lgÆmL )1 ampicillin) in 96-well plates and incubated at 37 °Cona rotary shaker. When OD 550 reached 0.4, expression of FDH was induced by addition of 50 lL isopropyl thio-b-d-galacto- side (20 mmolÆL )1 ) and the cultivation proceeded for 4 h. Afterwards, the cells were harvested and cell lysis was per- formed in 0.1 molÆL )1 potassium phosphate buffer (pH 7.5) with 0.1% (v ⁄ v) Triton-X-100 and 0.2 molÆL )1 EDTA. The cell debris was removed by centrifugation and 35 lLof the resulting crude extract, which was of comparable FDH activity for all variants, were transferred into a 96-well micro- titre plate, mixed with 35 lL acrylamide (16% w ⁄ v and 30% w ⁄ v, respectively) and TEMED (0.2% w ⁄ v) in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5), and incubated at 30 °C for 30 min 70 lL FDH-assay mix for microtitre plate readers were then added and the extinction at 340 nm was monitored at 30 °C. The initial performance was determined by replacing the acrylamide ⁄ TEMED solutions by 35 lL potassium phosphate buffer (0.1 molÆL )1 , pH 7.5). The resid- ual activity was calculated by dividing activity after incuba- tion in acrylamide ⁄ TEMED by the activity after incubation in buffer. For determination of half-life, the activity of crude extracts was determined after incubation in the desired mix- tures of acrylamide, TEMED, and APS at 30 °C for 5–180 min, according to the protocol described above. The measured inactivation was adapted to the time law A ¼A 0 *e ()kt) , with A 0 being the activity after incubation time t and k the inactivation constant. The half-life was then calculated using the equation s ¼ ln 2/k. Expression and purification of FDH Recombinant E. coli were cultivated at 37 °C in LB med- ium containing 100 lgÆmL )1 of ampicillin. Expression was induced by addition of 0.5 mmolÆL )1 isopropyl thio-b-d- galactoside when the OD 620 nm was about 0.5. between 6 and 8 h after induction, the cells were harvested by centrif- ugation, resuspended in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5) to give a concentration of 50% (w ⁄ v), and PEG 400 was added to a final concentration of 30% (w ⁄ v). This solution was then mixed and incubated at 37 °C for 2 h. The resulting crude extract was cooled down to 20 °C and H 2 O and K 2 HPO 4 were added to final con- centrations of 21% (w ⁄ w) and 5% (w ⁄ w), respectively. After complete dissolution of K 2 HPO 4 , PEG 1550 and NaCl were added to final concentrations of 7% (w ⁄ w) and 6% (w ⁄ w), respectively, and the solution was stirred for 30 min. The solution separated into two phases within a settling time of 2 h. The upper phase of the system was removed and mixed with PEG 6000 and H 2 O at final con- centrations of 20% (w ⁄ w) and 10% (w ⁄ w), respectively. FDH precipitated from this solution after 2–3 h, was collec- ted by centrifugation, and redissolved in a 1 : 1 (v ⁄ v) solu- tion of potassium phosphate buffer (0.1 molÆL )1 , pH 7.5) and glycerine. It was stored at )20 °C until use. Determination of kinetic constants Kinetic constants were derived from duplicate measure- ments of initial velocities under conditions where only one substrate (formate or NAD + ) was limiting. The data obtained were fitted to the mathematical model of a dou- ble-substrate kinetic given below, using exel for windows (Micromath. Inc., Salt Lake City, UT, USA) for analysis. In this model, v 0 describes the initial velocity of the reac- tion, v max the maximum reaction rate, [x] the concentration of the two substrates, K Mx the Michaelis–Menten constant of substrate x, and K INADH2 the inhibition constant of NADH 2 . All experiments were performed with purified enzyme. Measurement of temperature influences For determination of T m (midpoint of thermal inactiva- tion), inactivation experiments were carried out by incuba- ting purified FDH in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5) at different fixed temperatures between 30 °C and 71 °C for 20 min and then assayed for residual activity and protein concentration. The tempera- ture at which the residual activity was 50% was defined as T m and was calculated from the point of inflection in a plot of residual activity at different temperatures versus the respective temperatures. Additionally, thermal inactivation of FDH at 50 °C was monitored until inactivation reached 80%. From the time course, half-life at 50 °C was calcula- ted analogous to the calculation of half-life in the presence of acrylamide. Temperature optima were determined by plotting the initial velocity of FDH activity at temperatures between 15 and 80 °C versus the temperature. Formation of PAA immobilisates A solution of 8% (w ⁄ v) or 15% (w ⁄ v) acrylamide (concen- tration of bis-acrylamide: 2.67% w ⁄ w) in potassium phos- phate buffer (0.1 molÆ L )1 , pH 7.5) was degassed for 15 min 1% TEMED and 5% APS were added, and the mixture was dropped into silicone oil (M50; Roth, Karlsruhe, Ger- many). Polymerization took place at 15 °C within 30 min. The average bead volume was 7.4 ± 2.5 mm 3 . For calibra- tion of the CO 2 measurement, solutions of NaHCO 3 were M. B. Ansorge-Schumacher et al. Stabilisation of FDH for entrapment FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS 3943 added before polymerization. For entrapment of FDH, 0.33 UÆmL )1 of the purified enzyme were included before polymerization and the beads were equilibrated in a solu- tion of 50 mmolÆL )1 NAD + in potassium phosphate buffer (0.1 molÆL )1 , pH 7.5) for 16 h after the polymerization pro- cess was complete. Determination of activity of entrapped FDH The activity of entrapped FDH was determined by monit- oring the formation of CO 2 . For this, 1 g PAA beads and 3 mL hexane were placed into an 8 mL GC-vial, which thereafter was sealed with a rubber cap and left to equili- brate to a temperature of 30 °C for 30 min. The reaction was started by injecting 100 lmol of formate into the vial and stopped after 60 min by injecting 1 mL of HCl (1 molÆL )1 ). After another 30 min of incubation, a gas sam- ple of 100 lL was withdrawn from the headspace with a gas-tight syringe, injected into a GC ⁄ HCD (Perkin Elmar, Connecticut, USA) and analysed isothermally at 40 °C (detector temperature at 140 °C; injector temperature at 65 °C; carrier gas helium 5.0 at 62.5 mLÆmin )1 ) on a CTR- 1-column (Alltech, Munich, Germany). The retention time of CO 2 under this conditions was 1.25 min. The peak areas obtained were converted into CO 2 content by use of the calibration curve illustrated in Fig. 2. This calibration was obtained by entrapping defined concentrations (1–60 mmolÆ L )1 ) of NaHCO 3 in PAA beads instead of FDH and per- forming the same procedure as described for the FDH immobilisates. The measured peak area was related to the known concentrations of CO 2 resulting from the quantita- tive transformation of NaHCO 3 into CO 2 at acidic pH. Acknowledgements We thank PD Dr. A. Eisentra ¨ ger and Dipl Ing. C. Grundke, both at the Institute of Hygiene and Envi- ronmental Medicine, Klinikum Aachen, for provision of the gas chromatograph and expert help with the determination of CO 2 concentration, and Professor M R. Kula (Institute of Enzyme Technology Hein- rich-Heine-University Du ¨ sseldorf) and Professor W. Hartmeier (Department. of Biotechnology, RWTH. Aachen University) for provision of laboratory space and equipment. References 1 Metrangolo-Ruiz de Temino D, Hartmeier W & Ansorge-Schumacher MB (2005) Entrapment of the alcohol dehydrogenase from Lactobacillus kefir in poly- vinyl alcohol for the synthesis of chiral hydrophobic alcohols in organic solvents. Enzyme Microb Technol 36, 3–9. 2 Hischer T, Gocke D, Ferna ´ ndez M, Hoyos P, Alca ´ ntara AR, Sinisterra JV, Hartmeier W & Ansorge-Schumacher MB (2005) Enzyme catalysed synthesis of chiral 2-hydroxy ketones in a two-phase system stabilised with polyvinyl alcohol. Tetrahedron 61, 7378–7383. 3 Willaert RG & Baron GV (1996) Gel entrapment and microencapsulation: methods, application and engineer- ing principles. Revs Chem Eng 12, 1–205. 4 Pizarro C, Fernandez-Torroba MA, Benito C & Gonza- lez-Saiz JM (1997) Optimization by experimental design of polyacrylamide gel composition as support for enzyme immobilization by entrapment. Biotechnol Bioeng 53, 497–506. 5 Nilsson H, Mosbach R & Mosbach K (1972) The use of bead polymerization of acrylic monomers for immobilization of enzymes. Biochim Biophys Acta 268, 253–256. 6 Johansson AC & Mosbach K (1974) Acrylic copolymers as matrices for the immobilization of enzymes 1. Cova- lent binding or entrapping of various enzymes to bead- formed acrylic copolymers. Biochim Biophys Acta 370, 339–347. 7 Trevan MD (1980) Immobilized enzymes. An Introduct- ion and Application in Biotechnology. Wiley-VCH, New York. 8 Hemachander C, Niranjan B & Rengarajula P (2001) Whole cell immobilization of Ralstonia pickettii for lipase production. Process Biochem 36, 629–633. 9 Kiatkamjornwong S, Siwarungson N & Nganbunsri A (1999) In situ immobilization of alkaline protease during inverse suspension polymerization of polyacrylamide and poly(acrylamide-co-methacrylic acid) hydrogel beads. J Appl Pol Sci 73 , 2273–2291. Fig. 2. Calibration for determination of CO 2 release from PAA immobilisates. m, total CO 2 release from gels with an acrylamide content of 8% (w ⁄ v), and r, total CO 2 release from gels with an acrylamide content of 15% (w ⁄ v). Stabilisation of FDH for entrapment M. B. Ansorge-Schumacher et al. 3944 FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS 10 Mosbach K & Mosbach R (1966) Entrapment of enzymes and microorganisms in synthetic cross-linked polymers and their application in column techniques. Acta Chem Scand 20, 2807–2810. 11 Degani Y & Miron T (1970) Immobilization of choli- nesterase in cross-linked polyacrylamide. Biochim Biophys Acta 212 , 362–364. 12 Cavins JF & Friedman M (1967) New amino acids derived from reactions of epsilon-amino groups in pro- teins with alpha, beta-unsaturated compounds. Biochem 6, 3766–3770. 13 Chiari M, Righetti PG, Negri A, Ceciliani F & Ronchi S (1992) Preincubation with cysteine prevents modifica- tion of sulhydryl groups in proteins by unreacted acryl- amide in a gel. Electrophoresis 13, 882–884. 14 Bordini E, Hamdan M & Righetti PG (1999) Matrix- assisted laser desorption ⁄ ionisation time-of-flight mass spectrometry for monitoring alkylation of beta-lacto- globulin B exposed to a series of N-substituted acryl- amide monomers. Rapid Commun Mass Spectrometry 13, 2209–2215. 15 Perez HL, Cheong HK, Yang JS & Osterman-Golkar S (1999) Simultaneous analysis of hemoglobin adducts of acrylamide and glycidamide by gas chromatography- mass spectrometry. Anal Biochem 274, 59–68. 16 Bordini E, Hamdan M & Righetti PG (2000) Probing acrylamide alkylation sites in cysteine-free proteins by matrix-assisted laser desorption ⁄ ionisation time-of-flight. Rapid Commun Mass Spectrometry 14, 840–848. 17 Chenault HK, Simon ES & Whitesides GM (1988) Cofactor regeneration for enzyme-catalysed synthesis. Biotechnol General Eng Rev 6, 221–270. 18 Arnold FH & Moore JC (1997) Optimizing industrial enzymes by directed evolution. Adv Biochem Eng ⁄ Biotechnol 58, 1–14. 19 Hermanson GT, Mallia AK & Smith PK (1992) Immo- bilized Affinity Ligand Techniques. Academic Press, Inc, San Diego, CA, London. 20 Menter P (1999) Acrylamide Polymerization – a Practical Approach. Bio-Rad Laboratories, Inc, Hercules, CA. 21 Van Duijn P, Pascoe E & Van der Ploeg M (1967) The- oretical and experimental aspects of enzyme determina- tion in a cytochemical model system of polyacrylamide films containing alkaline phosphatase. J Histochem Cytochem 15, 631–645. 22 Dobryszycki P, Rymarczuk M, Bulai G & Kochman M (1999a) Effect of acrylamide on aldolase structure 1. Induction of intermediate states. Biochim Biophys Acta 1431, 338–350. 23 Dobryszycki P, Rymarczuk M, Gapinski J & Kochman M (1999b) Effect of acrylamide on aldolase structure 2: characterization of aldolase unfolding intermediates. Biochim Biophys Acta 1431, 351–362. 24 Cadwell RC & Joyce GF (1992) Randomization of genes by PCR mutagenesis. PCR Methods Appl 2, 28–33. 25 Cadwell RC & Joyce GF (1994) Mutagenic PCR. PCR Methods Appl 3, 136–140. 26 Slusarczyk H, Felber S, Kula M-R & Pohl M (2000) Stabilization of NAD-dependent formate dehydrogen- ase from Candida boidinii by site-directed mutagenesis of cysteine residues. Eur J Biochem 267, 1280– 1289. 27 Stemmer WPC (1994) DNA shuffling by random frag- mentation and reassembly: in vitro recombination for molecular evolution. Proc Natl Acad Sci USA 91, 10747–10751. 28 Zhao H, Giver L, Shao Z, Affholter JA & Arnold FH (1998) Molecular evolution by staggered extension pro- cess (StEP) in vitro recombination. Nature Biotechn 16, 258–261. M. B. Ansorge-Schumacher et al. Stabilisation of FDH for entrapment FEBS Journal 273 (2006) 3938–3945 ª 2006 The Authors Journal compilation ª 2006 FEBS 3945 . Directed evolution of formate dehydrogenase from Candida boidinii for improved stability during entrapment in polyacrylamide Marion. were introduced during the first step of evolution. Stability of mutant FDHs in PAA Finally, the performance of the FDH variants after entrapment in PAA