Different strategies for multi-enzyme cascade reaction

0 downloads 0 Views 1MB Size Report
catalyzed by benzoylformate decarboxylase (BFD), L-selective reduction of a carbonyl group with alcohol dehydrogenase .... The kinetics were determined using the initial reaction rate method. ... enzyme activities were calculated from the increase of (S)- ..... hyde in the fed batch reactor (V0 = 10 cm3, cBFD,0 = 20 mg cm−3,.
Bioprocess and Biosystems Engineering https://doi.org/10.1007/s00449-018-1912-5

RESEARCH PAPER

Different strategies for multi-enzyme cascade reaction for chiral vic1,2-diol production Ana Vrsalović Presečki1 · Lela Pintarić2 · Anera Švarc1 · Đurđa Vasić‑Rački1 Received: 18 August 2017 / Accepted: 15 February 2018 © Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract The stereoselective three-enzyme cascade for the one-pot synthesis of (1S,2S)-1-phenylpropane-1,2-diol ((1S,2S)-1-PPD) from inexpensive starting substrates, benzaldehyde and acetaldehyde, was explored. By coupling stereoselective carboligation catalyzed by benzoylformate decarboxylase (BFD), L-selective reduction of a carbonyl group with alcohol dehydrogenase from Lactobacillus brevis (ADHLb) as well as the coenzyme regeneration by formate dehydrogenase (FDH), enantiomerically pure diastereoselective 1,2-diol was produced. Two different multi-enzyme system approaches were applied: the sequential two-step one-pot and the simultaneous one-pot cascade. All enzymes were kinetically characterized. The impact of acetaldehyde on the BFD and ADHLb stability was investigated. To overcome the kinetic limitation of acetaldehyde in the carboligation reaction and to reduce its influence on the enzyme stability, experiments were performed in two different excesses of acetaldehyde (100 and 300%). Due to the ADHLb deactivation by acetaldehyde, the simultaneous one-pot cascade proved not to be the first choice for the investigated three-enzyme system. In the sequential cascade with 300% acetaldehyde excess a 100% yield of vic 1,2-diol was reached. Keywords  Multi-enzyme cascade reaction · Asymmetric reduction · Stereoselectivity · Alcohol dehydrogenase · Benzoylformate decarboxylase

Introduction The multi-enzyme cascade reactions have become an important research area within the field of biocatalysis [1]. This approach seems to be a very promising direction for the production of optically active chemicals in pharmaceutical industry, but also show potential for the production of bulk chemicals. Multi-enzyme cascade processes result in higher yields, spend fewer chemicals, and save time, especially in the case of an unstable intermediate which does not need to be isolated. From the environmental and economical point of Electronic supplementary material  The online version of this article (https​://doi.org/10.1007/s0044​9-018-1912-5) contains supplementary material, which is available to authorized users. * Ana Vrsalović Presečki [email protected] 1



Faculty of Chemical Engineering and Technology, University of Zagreb, Savska cesta 16, 10000 Zagreb, Croatia



Faculty of Textile Technology, University of Zagreb, Prilaz baruna Filipovića 28, 10000 Zagreb, Croatia

2

view, cascade reactions are very promising since the process conducted in this manner results in significant reduction of both waste and production costs on the industrial scale [2, 3]. Additionally, in cascades, the reversible reactions could be driven to completion if the reaction product of the first reaction is used as a substrate in a following irreversible transformation. Besides advantages, some drawbacks for the implementation of such cascades could also exist. When several separate reactions are carried out separately, they are running at optimal operating conditions for each one of them. In case of multiple reactions run in one-pot, optimal conditions for each reaction often cannot be fulfilled. Therefore, in a multi-enzymatic cascade reaction a compromise pH for different enzymes should be found. More catalytic steps could complicate dependencies between the process variables [4–6], and thus mathematical modelling could be a useful tool for finding optimal conditions for a multi-enzyme reaction in a one-pot system [7, 8]. A key interest in organic chemistry is the synthesis of chiral building blocks [9]. Among these, sterically demanding chiral 1,2-diols with two chiral centers are of high interest. They have wide application as synthons for chemical

13

Vol.:(0123456789)



catalysts, agrochemicals and pharmaceuticals [10–12]. The synthesis of a specific form of a stereoisomer can be very demanding and challenging. The production of enantiomerically pure 1,2-diols has been described using chemical and enzymatic methods. The enzymatic approach seems to be a more attractive route, since the synthesis takes place under mild conditions and yields chiral diols with high chemo-, regio- and stereoselectivity. Different biocatalytic routes to chiral alcohols have been described using oxidoreductases, hydrolases and lyases, either with isolated enzymes or whole cells [13]. One powerful approach toward the synthesis of chiral 1,2-diols is the reduction of the prochiral keto group of 2-hydroxy ketones via NAD(P)H dependent oxidoreductases. The reduction of small and small-bulky 2-hydroxy ketones, such as (R)- and (S)-2-hydroxy-1-phenylpropan1-one (2-HPP) were performed using alcohol dehydrogenases from Lactobacillus brevis (ADHLb) and Thermoanaerobacter sp. (ADHT) [14–16]. In our previous work [17] we developed a method for the synthesis of two stereoisomeric forms of 1-phenylpropane1,2-diol. They were produced by (S)-2-hydroxypropiophenone reduction catalyzed by ADHLb and glycerol dehydrogenase from Cellulomonas sp. An effective regeneration system using formate dehydrogenase (FDH) was applied and complete substrate conversion was achieved. The substrate in that reaction was a chemical that is still not commercially available and was produced by the action of enzyme benzoylformate decarboxylase (BFD) from benzaldehyde and acetaldehyde. In this work we produced one stereoisomer of 1,2 diol in a three-enzyme one-pot system using cheap initial substrates by two different approaches (Fig. 1): (1) a sequential two-step one-pot cascade, and (2) a simultaneous one-pot cascade. Fig. 1  Two ways of performing a multi-enzyme cascade reaction in one-pot combining enantioselective carboligation and enantioselective reduction for the synthesis of (1S,2S)1-phenylpropane-1,2-diol

13

Bioprocess and Biosystems Engineering

The production of those types of 1,2-diols by the use of multi-step biocatalytic reactions has been described with only few examples involving lyases and oxidoreductases as two-pot reactions [14, 15], one-pot reaction [16, 18] and by a chemoenzymatic catalysis [19, 20].

Materials and methods Materials KH2PO4, 1,3-dihydroxyacetone dimer, benzaldehyde, acetaldehyde, ethyl acetate and thiamine diphosphate were purchased from Sigma-Aldrich (USA), K ­ 2HPO4, acetone and + ­NAD from Merck (Germany), ­MgSO4 from Fischer Scientific (UK), NADH and benzoylformate from Acros Organics and formate dehydrogenase from Candida boidinii from Fluka (Switzerland). ADHLb was provided by Forschungszentrum Jülich (Germany), while BFD with a C-terminal hexahistidine-tag was prepared as described elsewhere [21].

Enzyme activity The ADHLb activity in reactor experiments was spectrophotometrically monitored in the reaction of acetone reduction by following the decrease of NADH concentration at 340 nm. These measurements were performed in 0.1 M phosphate buffer pH 7 [cNADH = 0.14 mM, cacetone = 0.68 M, φsample = 0.01% (v/v)]. The BFD activity was followed in the reaction of decarboxylation of benzoylformate to benzaldehyde and carbon dioxide. Benzoylformate decrease was spectrophotometrically followed at 340 nm. The measurements

Bioprocess and Biosystems Engineering

were performed in 0.15  M phosphate buffer pH 6 with 2 mM ­MgSO4 and 0.5 mM thiamine diphosphate (ThDP) [cbenzoylformate = 12.5 mM, φsample = 0.05% (v/v)].

Enzyme kinetics The kinetics were determined using the initial reaction rate method. The measurements were carried out in 50 mM phosphate buffer, pH 7.5, with 2 mM ­MgSO4 and 0.5 mM ThDP at 25 °C. The ADHLb kinetics in (S)-2-HPP reduction and the FDH kinetics in formate oxidation were examined in our previous work [17]. Additionally, in this work, the impact of benzaldehyde and acetaldehyde on the ADHLb and FDH activity were tested. The BFD kinetics was measured in a 5 cm3 reactor (cBFD = 0.25 mg dm−3). The samples were taken at the regular time intervals during first 10 min of the reaction and were analyzed using HPLC system. The enzyme activities were calculated from the increase of (S)2-HPP concentration.

Reactor experiments All experiments were carried out in 50  mM phosphate buffer, pH 7.5, with 2 mM M ­ gSO4 and 0.5 mM ThDP at 3 25 °C in a 1 cm reactor. The synthesis in the sequential two-step one-pot reactor was performed in a two-step manner. First, 20 mM benzaldehyde was mixed with 40 (80) mM acetaldehyde, and the addition of BFD (γBFD = 1 mg cm−3) was considered as the start of the reaction. After all benzaldehyde was spent, or its consumptions stopped due to BFD deactivation, the second step began. In this step, the enzymes ADHLb

r1 = (

VmBFD ⋅ cbenzaldehyde ⋅ cacetaldehyde ⋅ cBFD ) ( benzaldehyde acetaldehyde Km + cbenzaldehyde ⋅ Km + cacetaldehyde +

(γADHLb = 5 mg cm−3) and FDH (γFDH = 1 mg cm−3), as well as the coenzyme NADH (cNADH = 1 mM) and co-substrate ammonium formate (c(NH4)2SO4 = 111 mM), were added into the reactor. In the simultaneous one-pot reaction all components were mixed together at the beginning of the experiment in the following concentrations: cbenzaldehyde = 13 mM, cacetaldehyde = 26 (52) mM, γ BFD  = 1  mg  cm −3 , γ ADHLb  = 5  mg  cm −3 , γFDH = 1 mg cm−3, cformate = 111 mM, and cNADH = 1 mM. In all experiments, the sampling was carried out at regular time intervals, and samples were prepared and analyzed as described below.

HPLC and GC measurements The benzaldehyde and (S)-2-HPP concentrations were measured by HPLC (Shimadzu, Tokyo, Japan) using a LiChrospher 100 RP-8 column (Hibar, 250 mm × 4 mm, Merck). The column temperature was set at 25 °C and the detection was carried out at 250 nm. Reversed-phase method was used with acetonitrile and water as mobile phases. Gradient elution was set from 15 to 65% of acetonitrile during 25 min at flow rate of 1.0 cm3 min−1. The samples taken from the reactor were quenched by the addition of acetonitrile followed by rigorous shaking on vortex and centrifuged to remove any solid. The retention times for (S)-2-HPP and benzaldehyde were 9 ± 0.1 and 11 ± 0.1 min, respectively. The (S)-2-HPP and phenylpropane 1,2-diol concentrations were analyzed by GC-FID with CP-Chirasil-Dex CB column (25 m, 0.25 mm I.D., film thickness 0.25 µm) [14, 17]. The injector, detector and oven temperature were kept at 280, 240 and 140 °C, respectively. Helium was used as a carrier gas (linear velocity of 25 cm s−1 and split ratio 10). Samples taken from the reactor were extracted by ethyl-acetate. The retention times for (S)-2-HPP, (1S, 2S)-1-PPD and (1R, 2S)-1-PPD were at 6.4, 19.4 and 22.1 min, respectively.

Mathematical model The kinetics of the production of (S)-2-HPP from benzaldehyde and acetaldehyde catalyzed by BFD from Pseudomonas putida in the presence of coenzyme ThDP was described by double-substrate Michaelis–Menten kinetics with included substrate (acetaldehyde) inhibition (Eq. 1) [22].

c2acetaldehyde acetaldehyde KiS

(1)

)

BFD deactivation was described using the biexponential model [23, 24] (Eq. 2). The model assumes the existence of two isozymes that are deactivated at different rates.

VmBFD BFD Vm0

BFD

= 𝛼 BFD ⋅ e−kd1

⋅t

BFD ( ) + 1 − 𝛼 BFD ⋅ e−kd2 ⋅t

(2)

The parameter α in Eq. (2) corresponds to the initial molar fraction of one isozyme. The kinetics of the ADHLb-catalyzed production of (1S,2S)-1-PPD from (S)-2-HPP in the presence of coenzyme NADH was described by double-substrate Michaelis–Menten kinetics with competitive product ­NAD +

13



Bioprocess and Biosystems Engineering

inhibition (Eq. 3). The reverse reaction kinetics, due to unavailability of (1S,2S)-1-PPD, was not experimentally determined, and was described by second-order kinetics: firstorder kinetics with respect to each substrate that participates in the reaction [17] (Eq. 4). r2 = (

VmADHLb ⋅ c(S)-2-HPP ⋅ cNADH ⋅ cADHLb ( ) ) ) ( c + Km(S)-2-HPP + c(S)-2-HPP ⋅ KmNADH ⋅ 1 + NAD + cNADH NAD+ Ki

(3)

(4) The kinetics of coenzyme NADH regeneration using ammonium formate as co-substrate catalyzed by FDH from Candida boidinii was described with double substrate Michaelis–Menten kinetics with competitive inhibition by product NADH and substrates from the main reaction, acetaldehyde and (S)-2-HPP (Eq. 5).

r3 = k3 ⋅ cADHLb ⋅ cNAD+ ⋅ c(1S,2S)-1-FPD

r4 = ( ( NH4 COOH Km ⋅ 1+

c(S)-2-HPP Ki(S)-2-HPP

ADHLb Vm0

ADHLb

= 𝛼 ADHLb ⋅ e−kd1

⋅t

Data processing The model parameters were estimated by non-linear regression analysis using simplex or least squares method implemented in the Scientist software, from the experimental data, the change of the initial reaction rate with substrate concentration or from the reactor experiments. The set of optimum parameters was used for the simulation according to the model equations. The Episode algorithm for the stiff system of differential equations, implemented in the Scientist software, was used for simulations. Standard deviations (σ, Eq. 7) and coefficient of determina-

VmFDH ⋅ cNAD+ ⋅ cNH4 COOH ⋅ cFDH ) ) ( ( cacetaldehyde + + acetaldehyde + cNH4 COOH ⋅ KmNAD ⋅ 1 + Ki

The deactivation of ADHLb was described using the biexponential model (Eq. 6).

VmADHLb

The mass balance equations for the batch reactor of this multi-enzyme one-pot were set according to the reaction scheme given as Fig. 2.

ADHLb ( ) + 1 − 𝛼 ADHLb ⋅ e−kd2 ⋅t

(6)

cNADH KiNADH

)

+ cNAD+

)

tion (R2, Eq. 8) of the goodness-of-fit curve were provided also by Scientist. √ √ n √1 ∑ ( )2 𝜎=√ Ycali − Yobsi (7) n i=1

Fig. 2  Reaction scheme and mass balance equations for the production of (1S,2S)-1-PPD in one-pot multi-enzymatic system

13

(5)

Bioprocess and Biosystems Engineering

R2 =

�2 ∑ n � n � �2 ∑ Yobsi − Ycali Yobsi − Y obs − i=1

i=1

n � ∑

i=1

Yobsi − Y obs

(8)

�2

In the above equations, n is the number of points, Yobs is the experimental value and Ycal is the value calculated by the model.

Results and discussion Enzyme kinetics The BFD kinetics in the carboligation reaction of benzaldehyde and acetaldehyde was measured using the initial reaction rate method and was described by double-substrate Michaelis–Menten kinetics. The Michaelis–Menten constants for acetaldehyde and benzaldehyde were taken from the work of Peper et al. (2011) [22], where they indicated the existence of acetaldehyde inhibition at values higher than 400 mM. The maximal activity and the substrate inhibition constant were estimated using the experimental data (Table 1). Substrate inhibition was described using the common non-competitive inhibition (Eq. 1), which means that the substrate binds with the enzyme–substrate (ES) complex making the newly formed complex (ESS) less active or even inactive. From the estimated constants (Km and KiS for acetaldehyde) it can be concluded that acetaldehyde is an effective substrate inhibitor. The high value of the Michaelis constant for acetaldehyde and the value of the substrate inhibition constant imply that the acetaldehyde concentration in the reaction of carboligation of benzaldehyde and acetaldehyde catalyzed by BFD is a very important initial parameter. Also, it is known that acetaldehyde affects the BFD stability [21]. Thus, the impact of different acetaldehyde concentration on BFD deactivation was investigated (Supplement Fig. 1S). The obtained data indicated that deactivation occurs with the existence of small amounts Table 1  Kinetic parameters of the BFD-catalyzed carboligation reaction of benzaldehyde and acetaldehyde Parameter

Value

VmBFD (U mg−1) benzaldehyde Km (mmol dm−3)* acetaldehyde Km (mmol dm−3)* acetaldehyde (mmol dm−3) KiS

19.40 ± 2.37

*Data taken from Peper et al. (2011) [22]

19.40 775.00 514.17 ± 52.88

of acetaldehyde in the solution. What is interesting, the increase of acetaldehyde concentration did not result with a higher deactivation rate. The enzyme deactivation was described by the biexponential model (Eq. 2). The reason for choosing such type of model was because the first and second order deactivation kinetics did not give satisfactory results. Also, the use of this type of equation for describing enzyme deactivation is very common because the simple three-parameters biexponential equation describes enzyme activity decay, whatever the mechanism is. It applies to complex mechanisms but also covers the simple “one step—two states” situation as it was in our case. The proposed expression is considered to be a model-free empiric equation. The series deactivation model or the Lumry–Eyring model could also describe this two-state deactivation mechanism but it contains four parameters and doesn’t contain any intrinsic control of its correctness. Furthermore, series model that assume no activity for the intermediate form of enzyme has the same mathematical expression as the biexponential model [23, 24]. The kinetics of ADHLb in the reaction of (S)-2HPP reduction was measured in our previous work [17] (Table 2). The kinetics was described by double-substrate Michaelis–Menten kinetics with competitive product ­NAD+ inhibition (Eq. 3). Additionally, the impact of benzaldehyde and acetaldehyde on the initial reaction rate was examined, and no inhibition was found. The impact of acetaldehyde on ADHLb stability was measured as well Table 2  Kinetic parameters of the reaction of (S)-2-HPP reduction catalyzed by ADHLb  Parameter

Value

VmADHLb (U mg−1)

0.53 ± 0.00

Km(S)-2-HPP (mmol dm−3) KmNADH (mmol dm−3) + KiNAD (mmol dm−3)

36.42 ± 3.31 0.08 ± 0.01 1.37 ± 0.28

*Data taken from the Švarc et al. (2015) [17]

Table 3  Kinetic parameters for FDH in the reaction of formate oxidation Parameter

Value

VmFDH (U mg−1]* NH COOH Km 4

0.52 ± 0.02 −3

(mmol dm )*

+ KmNAD (mmol dm−3)* KiNADH (mmol dm−3)* Ki(S)-2-HPP (mmol dm−3)* acetaldehyde Ki (mmol dm−3)

4.34 ± 1.00 0.01 ± 0.00 0.003 ± 0.001 4.10 ± 0.48 23.11 ± 2.09

*Data taken from the Švarc et al. (2015) [17]

13



Bioprocess and Biosystems Engineering

(Supplement Fig. 2S). It was found that the ADHLb deactivation increases with the increase of acetaldehyde concentration, and was also described by the biexponential model (Eq. 6) for the same reason mentioned above regarding the BFD deactivation. The FDH kinetics was also presented in our previous work [17]. As for ADHLb, the impact of acetaldehyde and benzaldehyde on the initial reaction rate of FDH was tested. It was found that acetaldehyde slightly inhibits the FDH-catalyzed reaction of formate oxidation, whereas benzaldehyde does not have any influence on the reaction (Table 3). To describe the FDH kinetics, the double-substrate Michaelis–Menten kinetics with competitive inhibition by co-product NADH and main substrate acetaldehyde and intermediate (S)-2-HPP (Eq. 5) was used.

Reactor experiments and model validation The production of optically active (1S,2S)-PPD from benzaldehyde and acetaldehyde was carried out in the multi-enzyme system in a batch reactor in two ways: as a sequential two-step one-pot multi-enzyme system and as a simultaneous one-pot multi-enzyme system (Fig. 1). The enzymes BFD and ADHLb are active in the same pH range and it was demonstrated that both reactions can be performed at the same pH [17, 22]. Therefore, for this onepot multi-enzyme system it was not necessary to choose a compromise pH. All experiments were conducted in phosphate buffer at pH 7.5 with the addition of cofactor ThDP and magnesium that are necessary for the BFD activity [25]. The reaction of vic-1,2-diol production carried out in a sequential [2, 3] two-step one-pot multi-enzyme system consisted of two steps. In the first step, the BFD-catalyzed carboligation reaction of benzaldehyde and acetaldehyde was performed, and the formed product (S)-2-HPP was used as a substrate for the second reaction step. In the second step, the enzyme ADHLb and coenzyme NADH were added to the reaction mixture from the first step to produce the vic-1,2diol by ADHLb-catalyzed (S)-2-HPP reduction. To carry out the coenzyme regeneration, FDH and ammonium formate were added. The point of this system was to test the possibility of this type of multi-enzyme reaction and to get an insight of the detailed behavior regarding the consumption and production of the components. Two experiments were performed in such way with different excess of acetaldehyde. It has been shown that the presence of acetaldehyde, even in a small concentration, has a negative impact on the BFD stability (Supplement, Fig. S1), but it is necessary to emphasize that increase of acetaldehyde concentration does not have a further negative influence on its stability. So, regarding the BFD enzyme, because of the high value of the Michaelis constant for acetaldehyde (Table 1), it is better to use higher

13

Fig. 3  Sequential two step one-pot multi-enzyme system for 1,2-diol production from acetaldehyde and benzaldehyde with the initial 100% acetaldehyde excess. a benzaldehyde (circles—experiment, solid— model), (S)-2-HPP (squares—experiment, dotted—model), (1S,2S)1-PPD (triangles—experiment, dashed—model) time change, b the BFD activity during the experiment, line—model

acetaldehyde excess to achieve higher reaction rates of the reaction of acetaldehyde and benzaldehyde carboligation. A limitation regarding the BFD enzyme is acetaldehyde inhibition and according to the kinetic measurement of the impact of acetaldehyde concentration on the initial reaction rate of the BFD enzyme (Supplement Fig. S3) concentration of acetaldehyde higher than 600 mM reduces the enzyme activity. Also, it is necessary to take into consideration the inhibition of FDH by acetaldehyde (Table 3) and its negative impact on the ADHLb (Supplement Fig. S2). Therefore, it was decided to examine two different experimental setups: one with a 100% acetaldehyde excess (acetaldehyde:benzal dehyde = 2:1), and the other with a 300% excess (acetaldehy de:benzaldehyde = 4:1), whereby all other initial concentrations were the same. Results of these experiments are shown in Figs. 3 and 4. During both experiments, the BFD deactivation was noticed and was described by biexponential model (Eq. 2). The deactivation rate in both experiments was approximately the same (Table 4).

Bioprocess and Biosystems Engineering

approximately 20 mM when the enzyme ADHLb, coenzyme NADH, and components needed for coenzyme regeneration (formate dehydrogenase and ammonium formate) were added to the reaction mixture. After 6.5 h, complete (S)2-HPP conversion was achieved, which is approximately 4 h longer than in the experiment with lower acetaldehyde excess, due to acetaldehyde presence in the reaction mixture: acetaldehyde caused ADHLb deactivation (Fig. 4b), as well as inhibition on FDH (Table 3), the enzyme involved in NADH regeneration. The ADHLb deactivation was described by the biexponential model (Eq. 6). Although in this second experiment ADHLb was deactivated, 20 mM of vic-1,2-diol was produced twice as fast as it was in the experiment with lower aldehyde excess. For this multienzyme system, it is evident that acetaldehyde concentration is very important, and that it is necessary to find a balance between its effects on this three-enzyme system. For the carboligation reaction catalyzed by BFD, higher acetaldehyde concentration led to faster reaction. On the other hand, the ADHLb was deactivated in the presence of acetaldehyde,

Fig. 4  Sequential two-step one-pot multi-enzyme system for (1S,2S)1-PPD production from acetaldehyde and benzaldehyde with the initial 300% acetaldehyde excess. a benzaldehyde (circles—experiment, solid—model), (S)-2-HPP (squares—experiment, dotted—model), (1S,2S)-1-PPD (triangles—experiment, dashed—model) time change, b BFD (black circles) and ADHLb (grey circles) activity during the experiment, line—model

Table 4  Parameters of the biexponential model for BFD and ADHLb deactivation in the sequential two-step one-pot multi-enzyme system for 1,2diol production

Parameter

Enzyme BFD

ADHLb

α (−) 0.36703 0.1012 kd1 ­(min−1) 0.00142 0.29332 kd2 ­(min−1) 0.24595 0.00044

In the experiment with lower acetaldehyde excess (Fig. 3), the substrate benzaldehyde was not completely converted, even after 33 h, due to BFD deactivation (Fig. 3b). In that moment, the concentration of (S)-2-HPP was approximately 19 mM, and the second enzyme ADHLb, together with NADH, FDH and ammonium formate were added to the reaction mixture. The (S)-2-HPP was completely reduced within two and a half hours, and ADHLb deactivation did not occur. In the reaction carried out with 300% acetaldehyde excess (Fig. 4), a 100% conversion of benzaldehyde was reached significantly faster (after 6.5 h), than in the experiment with 100% acetaldehyde excess (Fig. 3). In the second experiment, the concentration of (S)-2-HPP was

Fig. 5  Simultaneous one-pot multi-enzyme system for (1S,2S)-1-PPD production from acetaldehyde and benzaldehyde with the initial 100% acetaldehyde excess. a benzaldehyde (circles—experiment, solid— model), (S)-2-HPP (squares—experiment, dotted—model), (1S,2S)-1PPD (triangles—experiment, dashed—model) time change, b activity of BFD (circles) and ADHLb (squares) during the experiment, line— model

13



Bioprocess and Biosystems Engineering

and the regeneration reaction of formate oxidation was inhibited by it. This simultaneous one-pot multi-enzyme system is much simpler than the sequential multi-enzyme system. It is not necessary to follow the reaction trend, since all substrates and enzymes are added at the beginning of the reaction. Two experiments were performed in the same manner as in the sequential multi-enzyme system. The concentration of all initial components was the same except the acetaldehyde concentration. The same excess of acetaldehyde was applied (100 and 300%). The results of the experiments are shown in Figs. 5 and 6. The BFD deactivation also occurred (Figs.  5b, 6b; Table 5), but it was somewhat lower than in the sequential experiment (Table 4). A possible reason is the presence of ADHLb and FDH. Namely, it is known that protein concentration can enhance enzyme stability [26]. Even benzaldehyde was spent slower in the reaction with the 100% acetaldehyde excess (Fig. 5a), than in the reaction with higher excess (Fig. 6a), the course of both experiments was almost the same. The reason for that is very fast ADHLb deactivation (Figs. 5b, 6b; Table 5) that occurred at the same rate in both experiments. Because of it, 1,2-diol production was controlled by the ADHLb, i.e., (S)-2-HPP reduction. In both experiments, complete (S)-2-HPP conversion was not achieved even after 8 days. The reason for such fast loss of the ADHLb activity could be due to benzaldehyde and its negative impact on ADHLb. In the experiments that were conducted in a sequential manner negligible enzyme deactivation was noted (Table 4). In these experiments, ADHLb was added to the reaction mixture when benzaldehyde was nearly spent and could not affect the ADHLb stability. The stereoselectivity in all experiments were in accordance with the previous work (ee > 97%) [17]. The goodness-of-fit curve for each individual variable in all experiments, as well for the overall experiments are given in Table 6. According to the coefficient of determination, the fitting is better in the sequential experiments. Somewhat higher standard deviation in sequential experiments can be attributed to having more experimental points than in the simultaneous one. The weakest goodness-of-fit was noted

Fig. 6  Simultaneous one-pot multi-enzyme system for (1S,2S)-1-PPD production from acetaldehyde and benzaldehyde with the initial 300% acetaldehyde excess. a benzaldehyde (circles—experiment, solid— model), (S)-2-HPP (squares—experiment, dotted—model), (1S,2S)-1PPD (triangles—experiment, dashed—model) time change, b activity of BFD (circles) and ADHLb (squares) during the experiment, line— model Table 5  Parameters of biexponential model for BFD and ADHLb deactivation in simultaneous one-pot multi-enzyme system for diol production

Parameter

Table 6  Goodness-of-fit curve for the experiments shown in Figs. 3, 4, 5, 6

Approach

Sequential

Acetaldehyde excess

100%

Statistic parameter

σ (−)

R2 (−)

σ (−)

R2 (−)

σ (−)

R2 (−)

σ (−)

R2 (−)

Benzaldehyde (S)-2-HPP (1S,2S)-1-PPD BFD ADHLb Total

0.79 0.61 0.02 1.99 – 1.02

0.99 0.99 0.98 0.99 – 0.99

2.58 2.01 0.02 2.79 0.01 1.20

0.85 0.98 0.99 0.99 0.93 0.98

0.93 0.61 0.29 3.66 0.02 0.65

0.93 0.96 0.99 0.98 1.00 0.97

1.12 0.96 0.28 1.22 0.19 0.87

0.89 0.95 1.00 1.00 0.91 0.97

13

α (−) kd1 ­(min−1) kd2 ­(min−1)

Enzyme BFD

ADHLb

0.36703 0.00013 0.00770

0.10120 0.00003 0.08650

Simultaneous 300%

100%

300%

Bioprocess and Biosystems Engineering

when describing the changes in benzaldehyde concentration, while diol fitting was the best. Considering the coefficient of determination, BFD deactivation fitting was better than the deactivation of ADHLb, especially in the experiments with higher excess of acetaldehyde. Namely, the BFD decay does not depend on the concentration of acetaldehyde, as is the case with ADHLb (Supplement, Fig. S1 and S2). Regardless, standard deviations for the fitting of ADHLb indicate that data points are close to the ones calculated by the model. Overall statistics of all experimental points for each experiment are satisfactory. This indicated that the proposed mathematical model with the estimated parameters described well all the experimental data. Motivated by the satisfactory statistic output and wanting to put the model into the perspective, we did some model simulations. Since the results of the experiments that were carried out using simultaneous approach indicated strong ADHLb deactivation in the presence of benzaldehyde, we decided to simulate this kind of experiment in a fed-batch reactor with constant substrate addition. The aim was to keep

Fig. 7  Simulation of simultaneous one-pot multi-enzyme system for (1S,2S)-1-PPD production from acetaldehyde and benzaldehyde in the fed batch reactor (V0 = 10  cm3, cBFD,0 = 20  mg  cm−3, cADHLb,0 = 25  mg  cm−3, cFDH,0 = 5  mg  cm−3, cNADH,0 = 5  mM, cformate,0 = 500 mM, q = 0.05 cm3 min−1, cbenzaldehyde,feed = 25 mM) (a) yield on diol (circles) and benzaldehyde concentration (triangles) at the different acetaldehyde excess (t = 800  min) (b) benzaldehyde (solid), (S)-2-HPP (dotted), (1S,2S)-1-PPD (dashed) time change (260% acetaldehyde excess)

the benzaldehyde concentration as low as possible (below 1  mM). For this purpose, a fed-batch reactor model was set (Supplement Fig. S4). The impact of benzaldehyde on ADHLb deactivation was neglected due to its low concentration. The parameters of the biexponential deactivation model for ADHLb (α and kd1) were taken from the sequential one-pot batch experiment (Table 4). Taking into account the results gathered from the independent measurements of the impact of acetaldehyde on the ADHLb deactivation (Supplement Fig. S2), kd2 was calculated from the concentration of acetaldehyde, while deactivation constants for the BFD enzyme were taken from the simultaneous one-pot experiments (Table 5). Simulation results for varying acetaldehyde excess are given in Fig. 7a. Benzaldehyde concentration was below 1  mM when acetaldehyde excess was 200% or higher. Regarding the diol yield (Ydiol, calculated diol/theoretical diol), at the tested conditions, optimal acetaldehyde excess was in the range between 200 and 400%, where the yield was at its highest (approximately 0.9). Higher concentration of acetaldehyde caused higher deactivation of ADHLb that led to the slight decrease of the final diol concentration. Complete simulations of substrate and product for the 260% acetaldehyde excess are presented in Fig. 7b. Almost the same results were obtained in the sequential experiment in a batch reactor with 300% acetaldehyde excess (Fig. 4a). From this, it can be concluded that simultaneous diol production can be successfully carried out in a fed-batch reactor with acetaldehyde and benzaldehyde feed. The production of optically active α-aryl vicinal 1,2-diols in a similar system that combines carboligation and reduction was already done by the benzaldehyde lyase and oxidoreductase. The process utilizing lyophilized whole cells contained mentioned enzymes, done by Jakoblinnert and Rother (2014) [18], was also tested in a simultaneous and sequential mode. In this process, the product was a (R,R) enantiomeric form of the diol. In that case, the simultaneous mode seemed to be the better one; even though the sequential mode yielded higher final concentration than the simultaneous mode because of the side reaction of benzaldehyde reduction. The process ran in the simultaneous mode was faster and resulted with higher STY (amount of product synthesized per volume; per day). The enzyme deactivation in that case was probably omitted because of using enzymes in whole cells [27–29]. Shamangutan et al. (2012) [16] also used benzaldehyde lyase in combination with ADHLb and GDH to produce α-aryl vicinal diols, but with only one optically active center. They have carried out experiments in one-pot in a sequential manner to avoid enzyme deactivation.

13



Bioprocess and Biosystems Engineering

Conclusion The investigated stereoselective three-enzyme cascade for the one-pot synthesis of (1S,2S)-1-phenylpropane-1,2-diol should be run in a sequential manner rather than simultaneous in a batch reactor. The acetaldehyde excess of 300% in the sequential one-pot multi-enzyme system gave the best results regarding the conversion and productivity. A possible solution for the simultaneous one-pot multi-enzyme system, in this case, could be to perform the process in a fed-batch mode with controlled addition of substrates into the reaction mixture. By doing this, if benzaldehyde is spent simultaneously by the action of BFD, its impact on the ADHLb deactivation could be minimized. Acknowledgements  This work was supported by University of Zagreb short-term financial scientific research support under the title “Mathematical modeling of the biocatalytic synthesis of industrially interesting products”. The authors would like to thank Prof. Martina Pohl from the Institute of Bio- and Geosciences, IBG-1: Biotechnology, Research center Jülich, Germany for the gift of alcohol dehydrogenase from Lactobacillus brevis and Davor Valinger from Faculty of Food Technology and Biotechnology, University of Zagreb for the isolation of enzyme benzoylformate decarboxylase.

Compliance with ethical standards  Conflict of interest  The authors declare that they have no conflict of interest.

References 1. Ricca E, Brucher B, Schrittwieser JH (2011) Multi-enzymatic cascade reactions: overview and perspectives. Adv Synth Catal 353:2239–2262 2. Enoki J, Meisborn J, Müller AC, Kourist R (2016) A multienzymatic cascade reaction for the stereoselective production of γ-oxyfunctionalyzed amino acids. Front Microbiol 7:425 3. Monti D, Ferrandi EE, Zanellato I, Hua L, Polentini F, Carrea G, Riva S (2009) One-pot multienzymatic synthesis of 12-ketoursodeoxycholic acid: subtle cofactor specificities rule the reaction equilibria of five biocatalysts working in a row. Adv Synth Catal 351:1303–1311 4. Burton SG, le Roes-Hill M (2008) In: Garcia-Junceda E (ed) Multi-step enzyme catalysis—biotransformations and chemoenzymatic synthesis. Wiley, Weinheim 5. Fischer T, Pietruszka J (2010) Key building blocks via enzymemediated synthesis. Top Curr Chem 297:1–43 6. Hailes HC, Dalby PA, Woodley JM (2007) Integration of biocatalytic conversions into chemical syntheses. J Chem Technol Biotechnol 82:1063–1066 7. Xue R, Woodley JM (2012) Process technology for multi-enzymatic reaction systems. Bioresour Technol 115:183–195 8. Rios-Solisa L, Morrisa P, Granta C, Odeleyea AO, Hailesb HC, Warda JM, Dalbya PA, Baganza F, Lyea GJ (2015) Modelling and optimisation of the one-pot, multi-enzymatic synthesis of chiral amino-alcohols based on microscale kinetic parameter determination. Chem Eng Sci 122:360–372 9. Hall M, Bommarius AS (2011) Enantioenriched compounds via enzyme-catalyzed redox reactions. Chem Rev 111:4088–4110 10. Lavandera I, Kern A, Ferreira-Silva B, Glieder A, de Wildeman S, Kroutil W (2008) Stereoselective bioreduction of bulky-bulky

13

11. 12.

13. 14.

15. 16.

17.

18. 19. 20. 21.

22.

23.

24. 25.

26. 27. 28.

29.

ketones by a novel ADH from Ralstonia sp. J Org Chem 73:6003–6005 Patel RN (2008) Synthesis of chiral pharmaceutical intermediates by biocatalysis. Coord Chem Rev 252:659–701 Choudary BM, Chowdari NS, Madhi S, Kantam ML (2003) A trifunctional catalyst for one-pot synthesis of chiral diols via heck coupling—n-oxidation—asymmetric dihydroxylation: application for the synthesis of diltiazem and taxol side chain. J Org Chem 68:1736–1746 Goldberg K, Schroer K, Lütz S (2007) Biocatalytic ketone reduction–a powerful tool for the production of chiral alcohols-part II: whole-cell reductions. Appl Microbiol Biotechnol 76:237–255 Kulig J, Simon R, Rose C, Husain S, Häckh M, Lüdke S, Zeitler K, Kroutil W, Pohl M, Rother D (2012) Stereoselective synthesis of bulky 1,2-diols with alcohol dehydrogenases. Catal Sci Technol 2:1580–1589 Kihumbu D, Stillger T, Hummel W, Liese A (2002) Enzymatic synthesis of all stereoisomers of 1-phenylpropane-1,2-diol. Tetrahedron: Asymmetry 13:1069–1072 Shanmuganathan S, Natalia D, Greiner L, Dominguez de Maria P (2012) Oxidation-hydroxymethylation-reduction: a one-pot threestep biocatalytic synthesis of optically active a-aryl vicinal diols. Green Chem 14:94–97 Švarc A, Valinger D, Vasić-Rački Đ, Presečki AV (2015) Stereoselective synthesis of phenylpropane 1,2-diols from (S)2-hydroxypropiophenone by NADH-dependent oxidoreductases. Biochem Eng J 103:250–255 Jakoblinnert A, Rother D (2014) A two-step biocatalytic cascade in micro-aqueous medium: using whole cells to obtain high concentrations of a vicinal diol. Green Chem 16:3472–3482 Bencze LC, Paizs C, Tosa ML, Dan Irimie F, Retey J (2011) Chemoenzymatic one-pot synthesis of both (R)-and (S)-aryl1,2-ethanediols. ChemCatChem 3:343–346 Kamal A, Sandbhor M, Ahmed K, Adil SF, Ali Shaik A (2013) Chemoenzymatic synthesis of enantiomerically pure terminal 1,2diols. Tetrahedron: Asymmetry 14:3861–3866 Iding H, Dünnwald T, Greiner L, Liese A, Müller M (2000) Benzoylformate decarboxylase from Pseudomonas putida as stable catalyst for the synthesis of chiral 2-hydroxy ketones. Chem Eur J 6:1483–1495 Peper S, Kara S, Long WS, Liese A, Niemeyer B (2011) Immobilization and characterization of benzoylformate decarboxylase from Pseudomonas putida on spherical silica carrier. Bioprocess Biosyst Eng 34:671–680 Aymard C, Belarbi A (2000) Kinetics of thermal deactivation of enzymes: a simple three parameters phenomenological model can describe the decay of enzyme activity, irrespectively of the mechanism. Enzym Microb Technol 27:612–618 Polakovič M, Bryjak J (2002) Modelling of the kinetics of thermal inactivation of glucoamylase from Aspergillus niger. J Mol Catal B: Enzym 19–20:443–450 Kara S, Long WS, Berheide M, Peper S, Niemeyer B, Liese A (2011) Influence of reaction conditions on the enantioselectivity of biocatalyzed C–C bond formations under high pressure conditions. J Biotechnol 152:87–92 Kuznetsova IM, Turoverov KK, Uversky VN (2014) What macromolecular crowding can do to a protein. Int J Mol Sci 15:23090–23140 Hollmann F, Arends IW, Holtmann D (2011) Enzymatic reductions for the chemist. Green Chem 13:2285–2314 Dominguez de Maria P, Stillger T, Pohl M, Kiesel M, Liese A, Groger H, Trauthwein H (2008) Enantioselective C–C bond ligation using recombinant Escherichia coli-whole-cell biocatalysts. Adv Synth Catal 350:165–173 Jakoblinnert A, Mladenov R, Paul A, Sibilla F, Schwaneberg U, Ansorge-Schumacher MB, Dominguez de Maria P (2011) Asymmetric reduction of ketones with recombinant E. coli whole cells in neat substrates. Chem Commun 47:12230–12232