Stereoselective synthesis of (1S,2S)-1-phenylpropane-1,2-diol by cell-free extract of Lactobacillus brevis

2.1 Chemicals and organisms

Lysozyme from chicken egg white, KH₂PO₄ and thiamine diphosphate (ThDP) were purchased from Sigma-Aldrich (St. Louis, MO, USA), K₂HPO₄ and acetone from Merck (Darmstadt, Germany), MgSO₄ from Fisher Scientific (Loughborough, Leicestershire, UK), NADPH from Boehringer Mannheim GmbH (Ingelheim am Rhein, Germany), NADP⁺ from Jülich Fine Chemicals GmbH (Jülich, Germany), isopropyl alcohol from T.T.T. (Sveta nedjelja, Croatia), acetophenone from Acros Organics (Geel, Belgium) and (R)-(+)-1-phenyethanol from Merck KGaA (Darmstadt, Germany). (S)-2-HPP was synthesized as described elsewhere [32], while Lactobacillus brevis was obtained from the culture collection of the Faculty of Chemical Engineering and Technology, University of Zagreb.

2.2 Cultivation of Lactobacillus brevis

For the bacterial growth de Man, Rogosa and Sharpe (MRS) medium was prepared according to De Man et al. [33]. The prepared broth was then sterilized at 110°C for 30 min under 0.5 bar pressure. In order to make MRS agar slants, MRS agar was prepared by adding 15 g of agar in MRS broth (prepared above) and then autoclaved. Using sterile technique Lactobacillus brevis cells were inoculated from a plate culture to a fresh slant. After 72 h of incubation at 30°C slant culture was used for the preparation of inoculum for the shake flask culture. The cell growth was carried out in 500 cm³ Erlenmeyer flasks with 100 cm³ of MRS broth at 150 rpm and 30°C.

The cultivation of Lactobacillus brevis cells was carried out under anaerobic conditions at 30°C and pH 6.3 maintained by addition of concentrated ammonium hydroxide. The cultivation started by adding 100 cm³ of sterile inoculum. The initial concentration of L. brevis biomass was 0.23 g_dw dm^-3.

2.3 Disruption of Lactobacillus brevis cells

Cell disruption was performed by five different methods based on ultrasonication, lysozyme lysis and bead vortex mixing (Table 1). After each disruption process samples were centrifuged for 10 min at 14,000 rpm and 4°C. The supernatant was used for determination of LbADH activity in cell-free extract for acetone and acetophenone.

2.4 Statistical optimization of Lactobacillus brevis cell disruption using the EVOP technique

In the present study two variables, cell/beads mass ratio (-) and disruption duration (t, min), were taken into consideration for optimization through EVOP factorial design technique in order to find optimal conditions for Lactobacillus brevis cell disruption. For all used disruption methods preliminary tests were made in order to set the initial central values (or the initial optimum levels) that are needed for the analysis that follows the EVOP technique. After the central values were selected, the parameters for those experiments were arranged in both lower and higher level compared with the search-level region, which is assumed to be the initial optimum level. All of the conducted experiments were repeated two times (cycle I and cycle II). The enzyme activity was examined on two substrates (acetone and acetophenone). The magnitude of effects, error limits and the changes in the mean effect were examined as per the decision-making procedure to arrive at the optimum level [3]. When the experimental results of the first set (set I) did not reach the satisfactory level of optimum conditions, a second set (set II) of experiments was planned, selecting the best condition of the first set as the new search level for the second set. This procedure was repeated until the optimum condition was obtained [8]. The calculation steps and the decision-making procedure can be found in the literature [3, 5, 6, 8, 34, 35].

2.5 Enzyme assay

LbADH activity in cell-free extract was measured spectrophotometrically on Shimadzu UV-1800 UV-VIS Spectrophotometer at 340 nm by following the decrease of NADPH concentrations. These measurements were performed in 50 mm phosphate buffer pH 7.5. The LbADH activity in cell-free extract was monitored in the reaction of both acetone and acetophenone reduction and was determined in a total reaction volume of 1 cm³, 2.4 mm NADPH, 339.71 mm acetone or 12.58 mm acetophenone, and 0.1% (v/v) of enzyme sample. One unit of enzyme activity was defined as 1 μmol of NADPH oxidized per minute.

2.6 Determination of kinetic parameters

The kinetics of crude LbADH in (S)-2-HPP, acetone and acetophenone reduction and in isopropanol and (R)-1-phenylethanol oxidation were measured using the method of initial rate. The dependence of the initial rate on substrate concentrations and present compounds were determined on Shimadzu UV-1800 UV-VIS Spectrophotometer at a wavelength of 340 nm. The measurements were carried out in 1 cm³ quartz cuvettes with 50 mm phosphate buffer pH 7.5 with 2 mm MgSO₄ and 0.5 mm ThDP at 25°C. The concentration of LbADH in cell-free extract in all measurements was 0.1% (v/v). The initial rates were calculated from changes in the absorbance at 340 nm using the molar extinction coefficient for NADPH of 6.22 cm² μmol^-1.

The kinetic parameters of Michaelis-Menten kinetic models [Table 2a, Eqs. (1), (3) and (4)] were estimated from experimental data using the method of initial rate whereby the concentration of one reactant was varied while all other concentrations were kept constant. The constant k [Table 2a, Eq. (2)] was estimated from the reactor experiments. The parameters were estimated by nonlinear regression analysis using the simplex or least squares method implemented in Scientist software [36]. The set of optimum parameters was used for the simulation according to proposed models. For simulations, the built-in Episode algorithm for stiff system of differential equations was used. The standard deviations and determination coefficients as measures of curve fit were calculated by Scientist built-in statistical functions.

2.7 Enzymatic reduction of (S)-2-HPP

For the investigation of (S)-2-HPP reduction without coenzyme regeneration, a batch reactor was used. The experiment was performed in 50 mm phosphate buffer pH 7.5 with 2 mm MgSO₄ and 0.5 mm ThDP at 25°C in a reaction volume of 2 cm³ with 10.3 mm (S)-2-HPP and 10.4 mm NADPH. The reaction was initiated with addition of crude LbADH (φ_LbADH=0.2 (v/v)). The sampling was carried out at regular time intervals.

For the experiments with NADPH regeneration by isopropanol and (R)-1-phenylethanol oxidation, the concentration of crude enzyme was 0.4 (v/v). When the isopropanol-based regeneration system was employed, the substrate (S)-2-HPP, NADPH and isopropanol concentrations were 11.3 mm, 5.3 mm and 639.1 mm, respectively. For the reaction with (R)-1-phenylethanol regeneration system, the substrate concentrations were as follows: 10.9 mm (S)-2-HPP, 5.3 mm NADPH and 414.2 mm (R)-1-phenylethanol.

During the biocatalytic reaction, the product was not removed from the reactor, and the sampling was carried out at regular time intervals.

2.8 GC measurements

The (S)-2-HPP and (1S, 2S)-1-PPD concentrations were analyzed by gas chromatography with flame ionized detector (GC-FID) on Shimadzu GC-2014 Chromatograph. The separation was carried out on CP-Chirasil-Dex CB column with dimensions of 25 m length, 0.25 mm I.D. and film thickness of 0.25 μm with an injection volume of 1 μl. Helium was used as carrier gas with a linear velocity of 25 cm s^-1. The injector, detector and oven temperature were kept at 280, 240 and 140°C, respectively. Samples were prepared by mixing 10 μl of sample from reactor with 390 μl of ethyl-acetate, which was used for extracting (S)-2-HPP from the water solution, for 1 min. After 10 min of centrifugation (4°C, 14,000 rpm) the upper layer was used for the GC-FID analysis. The retention of R_(S)-2-HPP was 6.2 min, while of R_{(1S, 2S)-1-PPD} was 22.25 min.

3.1 Optimization by EVOP

In the cell disruption process, the influence of duration and cell/beads mass ratio on the enzyme activity in cell-free extract was examined. The disruption methods which include cell disruption by glass beads (Supplemental Tables S1–S4), combination of ultrasound and glass beads (Supplemental Tables S5 and S6), and combination of lysozyme and glass beads (Supplemental Tables S7–S9) were statistically optimized using the EVOP technique, while the remaining two disruption methods were not optimized using the EVOP technique due to poor activities of crude LbADH (Table 1). Following the procedure for optimization of variables by EVOP and based on our experience with Lactobacillus brevis disruption processes, experimental setups with two variables and a central point were created. After the results were collected, the error limits, the effects and the change in the mean effect were calculated and used in determining whether to continue with the optimization or to stop, for the optimal conditions were obtained (Supporting Information). The optimal conditions of examined cell disruption methods together with acquired enzyme activities are given in Table 3. According to those results, the optimal method for cell disruption was found to be the combination of lysozyme and glass beads. By using this method, time disruption is three-fold lower than by using only glass beads; the amount of glass beads is only slightly higher, but not a large amount of glass beads is being used in general. The obtained activities by using the combination of lysozyme and glass beads are insignificantly lower than the ones obtained by those methods that gave individually higher activities depending on the substrate type. In the case when acetophenone was used as substrate, the obtained activity when disruption was performed by only glass beads gave 1.1-fold higher activity than when the disruption was performed by the combination of glass beads and lysozyme. In the case with acetone, the highest enzyme activity was obtained by disruption with a combination of glass beads and ultrasonication, when the activity was 1.2-fold higher than the one obtained by the optimal method. Therefore, based on obtained results, and taking time into account, the disruption by combination of glass beads and lysozyme was found to be the optimal method for L. brevis cell disruption.

3.2 Kinetic characterization of LbADH-catalyzed reactions

In order to obtain LbADH in cell-free extract, which has been used for the production of (1S, 2S)-1-PPD, based on obtained results for cell disruption, Lactobacillus brevis cells were pretreated by lysozyme (1 mg ml^-1, 0.1 m phosphate buffer pH 7) for 1 h at 37°C followed by addition of glass beads in cell/beads mass ratio of 0.05. The whole solution was vortexed at high speed for 10 min with constant ice cooling, and the obtained crude LbADH was then used for the production of (1S, 2S)-1-PPD.

The kinetics of crude LbADH in (S)-2-HPP reduction was described by two-substrate Michaelis-Menten kinetics with competitive co-substrate inhibition. The impact of components involved in the regeneration reactions on the main reaction kinetics was studied. It was found that isopropanol and (R)-1-phenylethanol inhibit the reaction rate of (S)-2-HPP reduction in a competitive manner; the proposed kinetic equation of (S)-2-HPP reduction was complemented with competitive alcohol inhibition [Table 2a, Eq. (1)]. The kinetics of the reverse reaction was described with second-order kinetics, because (1S, 2S)-1-PPD is not commercially available [Table 2a, Eq. (2)]. The estimated values of inhibition constants from Eq. (1) (Table 2a) are shown in Table 4.

The kinetics of crude LbADH in regeneration reaction was determined, because only efficient cofactor regeneration enables a successful use of cofactor [37]. It was found that the initial rate of alcohol oxidation was inhibited by NADPH, ketone and (S)-2-HPP. The reaction rate of (R)-1-phenylethanol oxidation was additionally inhibited by NADP⁺ [Table 2a, Eq. (3)]. Acetone, NADP⁺ and isopropanol inhibited the initial rate of acetone reduction, while NADP⁺ inhibited the initial rate of acetophenone reduction [Table 2a, Eq. (4)].

The parameters of Eqs. (1), (3) and (4) (Table 2a) were estimated by nonlinear regression from the experimental data (Table 5).

To complete mathematical models of (S)-2-HPP reduction with and without coenzyme regeneration, mass balance equations for a batch reactor were set up based on shown reactions (Figures 1 and 2). The mass balance equations for (S)-2-HPP reduction with coenzyme regeneration are presented in Eqs. (5)–(10) (Table 2b); for the reaction without coenzyme regeneration, r₃=r₄=0 mmol dm^-3 min^-1.

Figure 1:

Reaction scheme of the (S)-2-HPP reduction with coenzyme regeneration by isopropanol oxidation.

Figure 2:

Reaction scheme of the (S)-2-HPP reduction with coenzyme regeneration by (R)-1-phenylethanol oxidation.

The maximum activities of LbADH in cell-free extract in reactions of isopropanol and (R)-1-phenylethanol oxidation were lower than in reverse reactions. The V_m values of alcohol oxidation were two-fold lower than in reactions of ketone reduction. This was expected since it is known that the activities of dehydrogenases in the oxidation reactions are higher under alkaline conditions and that pH around 7 is preferred for reduction reactions [38]. The estimated K_m values suggested that the enzyme was more specific for (R)-1-phenylethanol, acetophenone, NADP⁺ and NADPH than for (S)-2-HPP, isopropanol and acetone. In the reactions of alcohol oxidation, the enzyme was significantly inhibited by NADPH than in reactions of ketone reduction by NADP⁺. In the reaction of (R)-1-phenylethanol, oxidation inhibition by main substrate (S)-2-HPP was noted as well as by co-substrate (R)-1-phenylethanol in the reaction of acetophenone reduction. The ratio of product inhibition constant of acetone and the Michaelis constant of isopropanol was found to be greater than one, which indicates that isopropanol oxidation could successfully serve for NADPH regeneration, while (R)-1-phenylethanol oxidation could not [39].

3.3 Reduction of (S)-2-HPP

The experiment of (S)-2-HPP reduction without coenzyme regeneration was conducted to find the equilibrium conversion and kinetic constant of the reverse reaction. The synthesis of (1S, 2S)-1-PPD was conducted with equimolar (10 mm) (S)-2-HPP and NADPH substrate concentrations in batch reactor. In the reaction without coenzyme regeneration, equilibrium conversion of 27.5% was achieved within 1020 min (Figure 3A). The calculated equilibrium constant was 0.379. The estimated rate constant for the reverse reaction of (1S, 2S)-1-PPD oxidation was 0.043 dm³ mmol^-1 min^-1 [Table 2a, Eq. (2)]. Based on the conversion value, it could be concluded that the regeneration of NADPH is highly needed for shifting the equilibrium in the desired direction. In an attempt to shift the reaction in the direction of (S)-2-HPP reduction, the (1S, 2S)-1-PPD synthesis was carried out with two independent in situ coenzyme regeneration systems by isopropanol and (R)-1-phenylethanol oxidation. In the reaction with coenzyme regeneration by isopropanol oxidation (Figure 3B) (11.3 mm (S)-2-HPP, 5.3 mm NADPH and 1533.8 mm isopropanol), the equilibrium conversion was obtained within 1580 min and equaled 99.0%, while by (R)-1-phenylethanol oxidation (Figure 3C) (10.9 mm (S)-2-HPP, 5.3 mm NADPH and 414.2 mm (R)-1-phenylethanol) was 70.1% within 3470 min.

Figure 3:

(S)-2-HPP reduction by LbADH in cell-free extract obtained by using the optimal cell disruption method and conditions in batch reactor: (A) reaction without coenzyme regeneration (φ_LbADH=0.2, c_(S)-2-HPP=10.3 mm, c_NADPH=10.4 mm, 50 mm phosphate buffer pH 7.5, 2 mm MgSO₄, 0.5 mm ThDp, T=25°C, V_reactor=2.5 cm³), (B) reaction with coenzyme regeneration by isopropanol oxidation (φ_LbADH=0.4, c_(S)-2-HPP=11.3 mm, c_NADPH=5.3 mm, c_isopropanol=1533.8 mm, 50 mm phosphate buffer pH 7.5, 2 mm MgSO₄, 0.5 mm ThDp, T=25°C, V_reactor=2.5 cm³), (C) φ_LbADH=0.4, c_(S)-2-HPP=10.9 mm, c_NADPH=5.3 mm, c_{(R)-1-phenylethanol}=414.2 mm, 50 mm phosphate buffer pH 7.5, 2 mm MgSO₄, 0.5 mm ThDp, T=25°C, V_reactor=2.0 cm³).

Black circles, experiment; line, model simulation; black, (S)-2-HPP; grey, (1S, 2S)-1-PPD.

Despite that both systems shifted the reaction in the direction of (S)-2-HPP reduction, isopropanol proved to be the perfect co-substrate for in situ substrate-coupled NADPH regeneration catalyzed by crude LbADH. The use of isopropanol as co-substrate for NADPH regeneration resulted in higher equilibrium conversion in a shorter period of time and presents a commercially affordable chemical which is easily removed from the reaction mixture (e.g. pervaporation) [40]. There are numerous studies concerning ADH- and LbADH-catalyzed NAD(P)H regeneration by isopropanol, since isopropanol is one of the most used co-substrates for substrate-coupled NADPH regeneration [12, 14, 19, 24, 27, 28, 41–44], but there are no information regarding production of (1S, 2S)-1-PPD by LbADH in cell-free extract as well as regarding usage of phenylethanol as co-substrate for NAD(P)H regeneration.

All three batch experiments (Figure 3) were performed to validate mathematical models of used systems (Table 2 and Figures 1 and 2). The proposed mathematical models for (S)-2-HPP reduction without NADPH regeneration, with NADPH regeneration by isopropanol oxidation (Figure 1) and by (R)-1-phenylethanol oxidation (Figure 2), were in good agreement with experimental data. According to the goodness-of-fit statistics for the reactions without coenzyme regeneration, with coenzyme regeneration by isopropanol and (R)-1-phenylethanol oxidation, the coefficients of determination were 0.895, 0.949 and 0.859, while standard deviations were 0.897, 0.872 and 1.045, respectively.