Hydrogen-driven asymmetric reduction of hydroxyacetone to (R)-1,2-propanediol by Ralstonia eutropha transformant expressing alcohol dehydrogenase from Kluyveromyces lactis

Background Conversion of industrial processes to more nature-friendly modes is a crucial subject for achieving sustainable development. Utilization of hydrogen-oxidation reactions by hydrogenase as a driving force of bioprocess reaction can be an environmentally ideal method because the reaction creates no pollutants. We expressed NAD-dependent alcohol dehydrogenase from Kluyveromyces lactis in a hydrogen-oxidizing bacterium: Ralstonia eutropha. This is the first report of hydrogen-driven in vivo coupling reaction of the alcohol dehydrogenase and indigenous soluble NAD-reducing hydrogenase. Asymmetric reduction of hydroxyacetone to (R)-1,2-propanediol, which is a commercial building block for antibacterial agents, was performed using the transformant as the microbial cell catalyst. Results The two enzymes coupled in vitro in vials without a marked decrease of reactivity during the 20 hr reaction because of the hydrogenase reaction, which generates no by-product that affects enzymes. Alcohol dehydrogenase was expressed functionally in R. eutropha in an activity level equivalent to that of indigenous NAD-reducing hydrogenase under the hydrogenase promoter. The hydrogen-driven in vivo coupling reaction proceeded only by the transformant cell without exogenous addition of a cofactor. The decrease of reaction velocity at higher concentration of hydroxyacetone was markedly reduced by application of an in vivo coupling system. Production of (R)-1,2-propanediol (99.8% e.e.) reached 67.7 g/l in 76 hr with almost a constant rate using a jar fermenter. The reaction velocity under 10% PH2 was almost equivalent to that under 100% hydrogen, indicating the availability of crude hydrogen gas from various sources. The in vivo coupling system enabled cell-recycling as catalysts. Conclusions Asymmetric reduction of hydroxyacetone by a coupling reaction of the two enzymes continued in both in vitro and in vivo systems in the presence of hydrogen. The in vivo reaction system using R. eutropha transformant expressing heterologous alcohol dehydrogenase showed advantages for practical usage relative to the in vitro coupling system. The results suggest a hopeful perspective of the hydrogen-driven bioprocess as an environmentally outstanding method to achieve industrial green innovation. Hydrogen-oxidizing bacteria can be useful hosts for the development of hydrogen-driven microbial cell factories.


Background
The importance of bioprocess reactions that work effectively under mild conditions is growing rapidly as increasing demand for green innovation of industrial process. The conversion of a ketone to a corresponding alcohol represents a common redox-reaction in organic chemistry [1,2]. Dehydrogenases and reductases are promising biocatalysts for such reactions, of which the vast majority require nicotinamide cofactors such as NADH and NADPH as reductant. Given the costs of these cofactors, their stoichiometric use is economically infeasible. Therefore, various in situ regeneration methods including chemical, photochemical, electrochemical, and enzymatic reactions have been investigated [1][2][3][4][5]. Enzymatic approaches are particularly attractive for industrial process because of their high selectivity and efficiency. For example, a bioreduction system for the production of chiral alcohols has been reported using an Escherichia coli transformant, in which NAD(P)Hdependent carbonyl reductase and glucose dehydrogenase were heterologously co-expressed [1]. Continuous production of chiral alcohol was achieved through the in vivo coupling reaction of these enzymes, but production of equimolar amounts of a waste product, gluconate, is problematic. Formate dehydrogenase is known as another useful enzyme for cofactor regeneration [3], the benefits of which include its use of formate as an inexpensive substrate for cofactor reduction and generation of gaseous CO 2 as the only by-product. It does not affect activities of the enzymes and it is easily separated. However, generation of CO 2 is problematic in global warming issues. The low catalytic activity of formate dehydrogenase is also cited as an important shortcoming.
Hydrogen, a strong, inexpensive reductant, is also innocuous to the enzymes which will be coupled to a hydrogenase reaction for production of objective products. Importantly, oxidation of hydrogen by hydrogenase causes no pollution. Therefore, a hydrogen-driven bioconversion system can present an environmentally ideal method. The NAD-reducing soluble hydrogenase (ReSH) from Ralstonia eutropha (formerly Alcaligenes eutrophus), which is classified in Group 3 of [NiFe]-hydrogenase [6], is promising for such biocatalytic application as the enzyme is notably oxygen-tolerant [7,8]. Several reports have described that an in vitro coupling reaction by ReSH and NADdependent dehydrogenases functions effectively [9][10][11]. Permeabilized cells of R. eutropha were also evaluated as whole-cell coenzyme regeneration catalysts in an organicaqueous two-phase system, in which reduction of cyclohexanone was examined using in vitro coupling reaction of a commercial horse-liver alcohol dehydrogenase and permeabilized R. eutropha cells [12]. The coupling reaction by two permeabilized cells, R. eutropha and Gluconobacter oxydans, has also been examined for reduction of 2-octanone to produce (S)-2-octanol using H 2 as reductant [13]. The NADP-reducing hydrogenase I from Pyrococcus furiosus has also been coupled with Thermoanaerobium sp. alcohol dehydrogenase in vitro for reduction of acetophenone and (2S)-hydroxy-1-phenylpropanone, which have been converted to corresponding chiral alcohols with total turnover numbers (mol product/mol consumed cofactor) of 100 and 160, respectively [14]. However, the in vivo coupling reaction performed in E. coli cells [1], which must be more advantageous as a practical system, has not been reported to date.
In this study, we expressed alcohol dehydrogenase (KlADH) from Kluyveromyces lactis [15] in R. eutropha cells and evaluated the in vivo coupling reaction of ReSH and KlADH (Figure 1). Asymmetric reduction of hydroxyacetone to (R)-1,2-propanediol, which is used as a commercial building block for synthesis of antibacterial agents, was examined using the transformant cells as H 2driven microbial catalysts in the presence of hydrogen. To confirm the coupling reaction of ReSH and KlADH in vitro, reduction of hydroxyacetone was performed in 1 ml of reaction mixture containing 0.2 U of each enzyme solution and 1 μmol of NADH (1 mM) in vials under H 2 atmosphere. When 2% (v/v, 294 μmol in the reaction mixture) of hydroxyacetone was subjected for the reaction, 235-252 μmol of the substrate was reduced to 1,2-propanediol in 20 hr. Reduction of 240 μmol of substrate is expected by the 0.2 U of each enzyme activity in 20 hr, indicating that the coupling reaction proceeded effectively without a marked decrease of either ReSH or KlADH activity during the period. This result is attributable to the characteristic of hydrogenase responsible for NADH regeneration, which creates no byproduct affecting the enzyme performance. The product was reduced to 134-158 μmol (55.8-65.8% of the expected amount; 240 μmol) when the initial concentration of hydroxyacetone was increased to 3% (v/v).

Heterologous expression of KlADH in R. eutropha H16
We consider that it is important to express KlADH in R. eutropha H16 coordinately with indigenous ReSH in a similar level for effective in vivo coupling reaction. For this purpose, expression of KlADH was examined by using ReSH promoter. When the transformant H16-KLB4 was cultivated aerobically by hydrogenase derepressing FGN medium for 24 hr, the respective ReSH and KlADH activities in the soluble fraction (SF) were 3.22 U/ml and 4.64 U/ml. The SF prepared from R. eutropha H16 cultivated in the same condition showed ReSH activity of 4.10 U/ml and no detectable ADH activity. Consequently, the cell suspension of H16-KLB4 was subjected to H 2 -driven reduction of hydroxyacetone to examine ReSH-KlADH in vivo coupling reaction. No NADH was added to the reaction system: only indigenous cofactor was used for the reaction. The production of 1,2-propanediol was increased by increasing the cell suspension used for the reaction. When 800 μl of cell suspension was used, 2% (v/v) of hydroxyacetone was reduced completely (conversion yield; calculated as 99.1%) and the reaction presumably stopped within 20 hr. The decrease of reaction velocity at higher initial concentration of hydroxyacetone was lowered markedly compared to in vitro enzymatic coupling reaction. When 2% (v/v, 294 μmol in the reaction mixture) of hydroxyacetone was subjected to the reaction using 500 μl of cell suspension, for example, the concentration of 1,2-propanediol reached 1.72% (v/v, 253 μmol in the reaction mixture) in 20 hr. Relative to this productivity, 88.4% and 74.4% of product were obtained even though the initial concentrations of hydroxyacetone were increased respectively to 3% and 5%. Goldberg et al. [16] reviewed that biocatalytic ketone reduction becomes more stable by using cell catalysts compared to reactions with isolated enzymes in most cases because the enzymes are able to react in their natural environment inside the cell. Our result also showed such advantage of utilizing cell catalyst to stabilize the conversion reaction.
The fed-batch conversion of hydroxyacetone was investigated further using a jar fermenter under continuous flow of hydrogen ( Figure 3). Production of 1,2-propanediol continued for long period with an almost constant rate, which reached 6.54% (v/v, 67.7 g/l) after 76 hr reaction with average reaction velocity of 0.89 g/l/hr. According to this average reaction velocity, 2% (v/v, 294 mmol/l) of substrate periodically added to the reactor was converted completely in every 25.2 hr. The conversion yield of substrate supplied in total (8%, v/v) was calculated as 75.6% in 76 hr reaction. The obtained average reaction velocity corresponds to 28.3% of the expected value (3.15 g/l/hr) calculated from the lower enzyme unit (KlADH) in the SF prepared from a cell suspension used for the reaction (ReSH, 2.84 U/ml; KlADH, 1.38 U/ml). The 76 hr reaction sample showed 99.8% enantiomeric excess of (R)-1,2-propanediol. The reaction velocities varied depending roughly on the ReSH and KlADH activities present in the cells freshly cultivated for each reaction. For example, average reaction velocities of 1.08 g/l/h (2.77 U/ml ReSH and 2.45 U/ml KlADH in the cell suspension) and 0.78 g/l/h (1.15 U/ml ReSH and 1.36 U/ml KlADH in the cell suspension) were obtained in the same working conditions (data not shown). However, reactions continuously proceeded almost linearly in each batch reaction, as also observed in the experiments shown by Figures 4 and 5.
Two reports have described a H 2 -driven coupling reaction using R. eutropha cells for cofactor regeneration, but they are in vitro coupling systems and performed in an organic-aqueous two-phase system with vials. Andersson et al. [12] combined commercial horse-liver alcohol dehydrogenase and permeabilized R. eutropha cells for reduction of cyclohexanone to cyclohexanol. The conversion yield of 200 μmol cyclohexanone in 2 ml heptane (0.2 ml aqueous phase) reached 98% (9.8 g/l product) in the presence of 1 μmol NAD, but the reaction velocity decreased greatly after 24 hr. Rundbäck et al. [13] studied reduction of 2-octane to (S)-2-octanol by the combination of permeabilized cells of Gluconobacter oxydans and R. eutropha.  Figure 4 Influence of P H2 on the H 2 -driven reaction by R. eutropha transformant with a jar fermenter. The reaction mixture (120 ml) contained 110 ml cell suspension of H16-KLB4 (1 g wet cell/5 ml reaction buffer) and 2% (v/v) hydroxyacetone. (A) As the figure shows, the P H2 supplied to the reaction mixture decreased from 100% to 10% in a stepwise manner. The SF prepared from the cell suspension used for the reaction contained 1.90 U/ml and 2.85 U/ml of ReSH and KlADH activities, respectively. (B) As the figure shows, the P H2 supplied to the reaction mixture was increased from 10% to 100%. The SF prepared from the cell suspension used for the reaction contained 1.86 U/ml and 3.61 U/ml of ReSH and KlADH activities, respectively.
The conversion yield of 200 mM substrate in 0.25 ml ndodecane (1 ml aqueous phase) reached about 75% (ca. 20 g/l) in the presence of 2 mM NAD. In reaction of hydrophobic substances, the use of a second phase of an organic solvent acts as a reservoir delivering substrate to aqueous phase. The organic phase also helps to reduce possible toxicity of substrate and/or product to biocatalysts present in aqueous phase [17]. It is intriguing to consider how the in vivo coupling system by transformant cells stabilizes such reactions of hydrophobic substrates using an organic-aqueous two-phase system.
Influence of P H2 on H 2 -driven reaction by R. eutropha transformant H16-KLB4 Influence of P H2 of the headspace gas to the H 2 -driven reaction was investigated in vials. The reaction mixture (1 ml) contained 500 μl of cell suspension and 3% (v/v) hydroxyacetone. The amounts of 1,2-propanediol produced in 20 hr reaction were almost identical between 100% and 10% P H2 (gas mixture of H 2 and N 2 , 1 atm) (data not shown). The result was confirmed further by reaction with a jar fermenter (Figure 4). The reaction velocity did not change greatly by the P H2 value of the introduced gas higher than 10%. In Figure 4A, the reaction was initiated by supplying 100% hydrogen, and the P H2 was reduced to 30%; then to 10% in a stepwise manner after several hours in each condition. The average reaction velocity during the respective condition was 1.84 g/l/hr (100% H 2 ), 1.35 g/l/hr (30% P H2 ) and 1.24 g/ l/hr (10% P H2 ). When gas P H2 was increased from 10% to 100% in reverse, 1.21 g/l/hr (10% P H2 ), 1.13 g/l/hr (30% P H2 ) and 1.17 g/l/hr (100% H 2 ) average reaction velocities were obtained ( Figure 4B). Gas mixtures of P H2 of less than 10% could not be prepared because of the range limit of the thermal mass flow meter, but it was confirmed that the presence of 10% H 2 in gas mixture was sufficient to maintain the reaction velocity equivalent to that by 100% hydrogen. The saturated H 2 concentration in water phase is 68 μM at 30°C under gas phase (1 atm) containing 10% P H2 . The result is consistent with the K m value of ReSH, which is obtained experimentally as 6.1-11.9 μM [18,19].

Repetitive batch reaction by R. eutropha transformant H16-KLB4 with a jar fermenter
Repetitive H 2 -driven reaction by the transformant was investigated using a jar fermenter under continuous flow of hydrogen. The reaction was started with 2% (v/v) of hydroxyacetone, which was added successively to the reaction mixture as the reaction proceeded. Cells were harvested by centrifugation after a batch reaction of 22-23 hr, and were subjected to the next batch reaction ( Figure 5). When the procedure of cell harvest and resuspension in a fresh reaction buffer was operated anaerobically to protect ReSH from oxidative inactivation, average reaction velocities were maintained above 1 g/l/hr during five batch reactions. The product reached 189 g/l in all by six batch runs (total net reaction time; 132 hr), in which the average reaction velocity obtained using the sixth batch reaction corresponded to 39.6% of the first batch ( Figure 5A). Most hydrogenases are sensitive to oxygen, which causes loss of catalytic activity by oxidation of active site and/or FeS clusters under air atmosphere [20]. However, the cell activity was not greatly affected by handling in air. When cell harvest and resuspension procedures were operated under air, average reaction velocities higher than 1 g/l/hr persisted until the fourth batch reaction. The total 1,2-propanediol produced by six batch reactions reached 166 g/l (total net reaction time; 135 hr), in which 32.3% of average reaction velocity relative to that of the first batch was retained in the sixth batch reaction ( Figure 5B). The results are attributable to the high oxygen tolerance of ReSH [7,8]. In addition, there will be an advantage of in vivo coupling system in maintaining lower oxygen level inside the cell.

Conclusions
The coupling reaction of ReSH and KlADH continued in the presence of hydrogen in both in vitro and in vivo systems. This report of a hydrogen-driven bioconversion by in vivo coupling system is the first describing the use of R. eutropha transformant expressing KlADH (Figure 1). The in vivo coupling reaction proceeded continuously only by cell suspension, which did not require exogenous addition of a cofactor. The concentration of (R)-1,2-propanediol reached 67.7 g/l by periodical addition of hydroxyacetone without a marked decrease of reaction velocity. Reuse of transformant cells became available by in vivo coupling reaction. Moreover, the presence of 10% P H2 retained a reaction velocity equivalent to that by 100% hydrogen, indicating the availability of crude hydrogen gas from various sources. These results suggest a hopeful perspective of hydrogen-driven bioprocesses using in vivo coupling system to achieve industrial green innovation.

Construction of KlADH expression vector and transformation of R. eutropha H16
Standard DNA techniques were used for DNA manipulation [24]. The plasmid pSE-KLB2 containing the complete 1,155-bp KlADH gene [15] was used as a PCR template to introduce Nde I site at the initiation codon ATG and Xba I site at the end of the amplified PCR fragment. The following oligonucleotides were used as primers. Therein, designed restriction sites are underlined and mismatched nucleotides to the original sequence are shown in lower case letters: 5 0 -GAATT CTcatATGCGTGCATTAGCTTATTTCGG-3 0 and 5 0 -AACAGTTcTAGAACCCTCTTTGACAAGC-3 0 . The amplified 401-bp fragment was phosphorylated using a Mighty cloning reagent set (Takara Bio Inc.) and cloned to Eco RV site of pBluescript KS (+) (Stratagene Cloning Systems Inc.) to confirm the sequence. A thermosequenase primer cycle sequencing kit (GE Healthcare UK Ltd., Buckinghamshire, England) and DNA sequence LIC-4200S (Li-Cor Inc., Lincoln, NE, USA) were used for sequencing. The PCR fragment of the obtained recombinant plasmid pBlueKLB3F, whose authenticity of the sequence was confirmed, was digested by Nde I and Xba I and was ligated to the corresponding site of the pCH 591 [25]. The resulting plasmid pCH-KLB2 contained the initial 385-bp fragment of KlADH gene at the downstream of ReSH promoter. The plasmid was digested by Pfl MI and Xba I, into which the 1,005-bp Pfl MI-Xba I fragment from pSE-KLB2 was introduced. The resulting plasmid pCH-KLB3 contained a complete set of the ReSH promoter and KlADH gene. The Eco RV-Sna BI fragment from pCH-KLB3 containing the set was introduced to Swa I site of pEDY 309 [25]. The resulting plasmid pEDY-KLB4 was transformed into E. coli S17-1 and transferred to R. eutropha H16 by spot mating, by which R. eutropha transformant H16-KLB4 was obtained. Purification of the PCR amplified fragment and extraction of DNA fragment from the gel strip were done using gel extraction kit (QIA quick; Qiagen GmbH, Hilden, Germany). The plasmids pCH591 and pEDY309 were kindly donated by Dr. B. Friedrich and Dr. O. Lenz (Humboldt Universität zu Berlin, Germany).

Preparation of SF
The cells were washed using 50 mM potassium phosphate (KP) buffer (pH 7.0) and centrifuged at 7,000 × g for 20 min at 4°C. The washed cells were suspended homogeneously in the same buffer to be 1 g wet cell/5 ml and were broken twice (50 W, 5 min, Sonifier 250; Branson Ultrasonics Corp., CT, USA). Cell debris and unbroken cells were removed by centrifugation at 7,000 × g for 30 min at 4°C. Membranes in the cell-free extract were removed by ultracentrifugation at 100,000 × g for 1 hr at 4°C. The supernatant was used as SF.

Enzyme assays
NAD-reducing hydrogenase activity was determined spectrophotometrically at 30°C by following the H 2dependent reduction of NAD as an electron acceptor using glass cuvettes sealed with a rubber stopper and an aluminum cap [26]. The assay mixture contained 2 mM NAD in 50 mM KP buffer (pH 7.0). NAD-reducing activity of KlADH was similarly measured spectrophotometrically at 30°C in the presence of 20 mM (R)-1,2propanediol and 2.5 mM NAD in 50 mM KP buffer (pH 7.0). One unit of activity was defined as the amount of enzyme that reduced 1 μmol of NAD in a minute. The amount of reduced NAD was determined by the increase of absorbance at 340 nm (ε 340 nm = 6.22 mM -1 cm -1 ). Protein concentrations were estimated routinely [27] using established procedures of Bio-Rad Protein Assay (Bio-Rad Laboratories Inc., CA, USA).

H 2 -driven reduction of hydroxyacetone by in vitro enzymatic coupling of ReSH and KlADH
SFs containing ReSH and KlADH were prepared respectively from R. eutropha H16 and E. coli HB101 possessing pSE-KLB2 cells as described above. They were added to give 0.2 U/ml of each enzyme to the reaction mixture containing 1 mM NAD and 2% or 3% (v/v) of hydroxyacetone in 50 mM KP buffer (pH 7.0). The reaction was performed in a 10 ml vial containing 1 ml of reaction mixture, which was sealed with a rubber stopper and an aluminum cap, at 30°C for 20 hr under shaking at 130 rpm. The reaction was initiated by adding hydroxyacetone with a syringe after hydrogen was filled into the vial head space. The cell suspension of R. eutropha H16-KLB4, which was cultivated under hydrogenase derepressing condition, was This study prepared to be 1 g of wet cell per 5 ml reaction buffer (50 mM KP buffer, pH 7.0) and subjected as microbial catalyst. ReSH and KlADH activities present in the cell suspension were measured by preparing SF as described above. The reaction with 10-ml vials was done as in vitro enzymatic coupling reaction under hydrogen atmosphere at 30°C for 20 hr. Gas mixtures of H 2 and N 2 (1 atm) were prepared using thermal mass flow meters (Emerson Japan, Ltd., Tokyo, Japan) when the influence of headspace P H2 on the reaction was investigated. The reaction mixture (1 ml) contained 50-800 μl cell suspension and 2-5% (v/v) of hydroxyacetone. The reaction with a 250 ml jar fermenter (BMJ-25; Able & Biott Co., Ltd., Tokyo, Japan) was performed at 30°C under agitation at 1,000 rpm. Hydrogen gas was supplied to the reaction mixture with a thermal mass flow meter (Emerson Japan, Ltd.) at a flow rate of 250 ml/min. Gas mixtures of H 2 and N 2 (1 atm) were supplied at a total flow rate of 250 ml/min when influence of P H2 was investigated. The reaction mixture (120 ml) contained 60 ml or 110 ml cell suspension and 2% (v/v) of hydroxyacetone at the beginning of reaction. When the reaction was continued to reach higher product concentration, hydroxyacetone (2%, v/v) was added periodically to the reaction mixture as the reaction proceeded. The reaction pH was controlled using a pH controller (FC-2000; Tokyo Rikakikai Co., Ltd., Tokyo, Japan).

Measurement of 1,2-propanediol and chirality analysis
The reaction mixtures of in vitro enzymatic reaction were taken to sampling tubes, which were boiled for 5 min to stop the reaction and centrifuged. The supernatants were subjected to quantification of 1,2-propanediol using a gas chromatograph (GC-14B; Shimadzu Corp., Kyoto, Japan) equipped with a column (2.1 m × 3.2 mm internal diameter) packed with 5% Thermon-3000 (Chromosorb W 80/100 AW-DMCS; Shimadzu GLC Ltd., Tokyo, Japan) and a flame ionization detector. The column temperature was 150°C. The samples of in vivo coupling reaction were centrifuged to remove R. eutropha H16-KLB4 cells, and the supernatants were boiled for 5 min and centrifuged again. The supernatants were subjected to gas chromatography as described above. Optical purity of 1,2-propanediol produced was analyzed using a high performance liquid chromatograph (L-6320; Hitachi, Ltd., Tokyo, Japan) equipped with an OD-H column (25 cm × 4.6 mm internal diameter, Chiralcel; Daicel Corp., Tokyo, Japan) at 40°C. Solvent consisting of n-hexane:2-propanol (4:1) was used as mobile phase at a flow rate of 1.0 ml/min. Detection was done at 254 nm. The sample prepared for GC analysis containing ca. 25 μl of 1,2-propanediol was taken to a test tube and dissolved in ethyl acetate after evaporation of the water phase. The solvent was evaporated after filtration with a 0.45 μm filter (Millex LH; Merck KG, Darmstadt, Germany). Then the remaining sample was dissolved in 25 μl of phenylisocyanate and was left to stand for 30 min at 50°C. Ethanol (25 μl) was added to the sample and incubated for 10 min at 50°C. Solvent of the sample was exchanged to the HPLC solvent by evaporation, and was subsequently subjected to HPLC after filtration (Millex LH; Merck KG).