Coupling xylitol dehydrogenase with NADH oxidase improves l -xylulose production in Escherichia coli culture

Enzyme and Microbial Technology 106 (2017) 106–113

Contents lists available at ScienceDirect

Enzyme and Microbial Technology journal homepage: www.elsevier.com/locate/enzmictec

Coupling xylitol dehydrogenase with NADH oxidase improves L-xylulose production in Escherichia coli culture Qi Han, Mark A. Eiteman

MARK

⁎

School of Chemical, Materials and Biomedical Engineering University of Georgia, Athens, GA, 30602, USA

A R T I C L E I N F O

A B S T R A C T

Keywords: NAD+ regeneration L-xylulose Batch process Xylitol 4-dehydrogenase NADH oxidase

Escherichia coli expressing NAD-dependent xylitol-4-dehydrogenase (XDH) from Pantoea ananatis and growing on glucose or glycerol converts xylitol to the rare sugar L-xylulose. Although blocking potential L-xylulose consumption (L-xylulosekinase, lyxK) or co-expression of the glycerol facilitator (glpF) did not significantly affect Lxylulose formation, co-expressing XDH with water-forming NADH oxidase (NOX) from Streptococcus pneumoniae increased L-xylulose formation in shake flasks when glycerol was the carbon source. Controlled batch processes at the 1 L scale demonstrated that the final equilibrium L-xylulose/xylitol ratio was correlated to the intracellular NAD+/NADH ratio, with 69% conversion of xylitol to L-xylulose and a yield of 0.88 g L-xylulose/g xylitol consumed attained for MG1655/pZE12-xdh/pCS27-nox growing on glycerol. NADH oxidase was less effective at improving L-xylulose formation in the bioreactor than in shake flasks, likely as a result of an intrinsic maximum NAD+/NADH and L-xylulose/xylitol equilibrium ratio being attained. Intermittently feeding carbon source was ineffective at increasing the final L-xylulose concentration because introduction of carbon source was accompanied by a reduction in NAD+/NADH ratio. A batch process using 12 g/L glycerol and 22 g/L xylitol generated over 14 g/L L-xylulose after 80 h, corresponding to 65% conversion and a yield of 0.89 g L-xylulose/g xylitol consumed.

1. Introduction Rare sugars are monosaccharides and their derivatives that are found in the low abundance in nature [1]. Interest in rare sugars is attributed to their potential applications in pharmaceutical and food industries including antiviral drugs [2–4], anti-cancer or tumor treatments [5], drug building blocks [6], and as low-calorie sweeteners [7]. L-xylulose is a rare ketopentose which inhibits yeast α-glucosidase [8] and is an indicator of hepatitis or liver cirrhosis [9]. This pentose is also a precursor of other rare sugars such as L-xylose or L-lyxose [10,11]. The more readily available, less expensive, and structurally similar sugar alcohol xylitol can serve as an ideal substrate for L-xylulose. As an intermediate of xylitol metabolism, L-xylulose formation has been reported in Pantoea ananatis, Alcaligenes sp., and Bacillus pallidus [12–14]. Further investigation demonstrated that an NAD-dependent xylitol-4dehydrogenase (XDH) mediated this conversion [15]: xylitol + NAD+ → L-xylulose + NADH + H+

(1)

For the production of oxidized biochemicals using oxidoreductases such as XDH, the need for expensive cofactors (e.g., NAD+ or NADP+) as cosubstrates has discouraged strategies using purified enzymes [16].

⁎

Corresponding author. E-mail address: [email protected] (M.A. Eiteman).

http://dx.doi.org/10.1016/j.enzmictec.2017.07.010 Received 4 April 2017; Received in revised form 6 July 2017; Accepted 21 July 2017 Available online 24 July 2017 0141-0229/ © 2017 Elsevier Inc. All rights reserved.

Instead, whole-cell catalysts are used via either “resting cell” or “growing cell” approaches which essentially allows the system to regenerate cofactors. For example, a resting cell approach for L-xylulose formation involves cultivating cells in a growth medium, harvesting, centrifuging, washing and then resuspending the cells in a separate xylitol-containing medium. A growing cell approach involves cultivating cells in a growth medium which contains xylitol and therefore allows for the conversion of xylitol into L-xylulose during growth on another carbon source. The conversion of xylitol to L-xylulose has been achieved using resting cells of Escherichia coli expressing XDH from these P. ananatis, Alcaligenes sp. and B. pallidus [13,15,17,18]. The greatest conversion (defined as L-xylulose generated per mass xylitol initially supplied) of 82% was achieved using P. ananatis XDH with 5 g/ L xylitol [15]. L-xylulose formation has exclusively been studied using the resting cell method, as cells can be cultured quickly in a complex medium and then concentrated into an optimized xylitol-containing buffer for the bioconversion. Both enzymatic and microbial processes require a driving force for the conversion. Because the specific conversion of L-xylulose from xylitol involves the formation of NADH, the continued presence of NAD+ is necessary to drive the process. Presumably resting cells rely on

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

3.175 g K2HPO4·3H2O, 2.5 g K2SO4, 4.38 g NH4Cl, 25.0 mg Na2(EDTA)·2H2O, 0.15 g MgSO4·7H2O, 20 mg citric acid, 0.25 mg ZnSO4·7H2O, 0.125 mg CuCl2·2H2O, 1.25 mg MnSO4·H2O, 0.875 mg CoCl2·6H2O, 0.06 mg H3BO3, 0.25 mg Na2MoO4·2H2O, 5.5 mg FeSO4·7H2O, and 20 mg thiamine·HCl. The medium was adjusted to a pH of 7.0 with 30% (w/v) NaOH. Either medium contained 50 mg ampicillin/L (strains containing pZE12 plasmids) and/or 100 mg kanamycin/L (strains containing pCS27 plasmids).

Table 1 Strains and plasmids used in this study. Strain or Plasmid

Relevant Characteristics

Reference

Strains MG1655 MEC342

E. coli F− λ− ilvG rfb50 rph-1 MG1655 lyxK763:Kan

wild-type This study

PLlacO1:luc ColE1 Ampr PLlacO1:MCS1 p15A Kanr Codon optimized xdh gene from Pantoea ananatis (ATCC 43072) glpF gene from E. coli MG1655

[22] [23] This study, Supplement 1 This study

glpF gene from E. coli MG1655 nox gene from Streptococcus pneumoniae

This study This study

Plasmids pZE12-luc pCS27 pZE12-xdh pZE12-xdhglpF pCS27-glpF pCS27-nox

2.4. Fermentation Growth in shake flasks was accomplished by first inoculating from plate culture a strain into 3 mL Lysogeny Broth (LB) and grown overnight at 37 °C. Then, 1 mL culture was used to inoculate 50 mL defined medium containing 5 g/L xylitol and 5 g/L carbon source (glucose or glycerol) in 250 mL baffled shake flasks maintained at 37 °C with an agitation of 250 rpm. Triplicate shake flask cultures were induced at the time of inoculation (OD ∼ 0.15) with 0.5 mM IPTG, and sampled at 48 h for measurement of xylitol, L-xylulose, and the activities of XDH and NOX. For duplicate batch or intermittent fed-batch fermentations, shake flask cultures as described above which attained an OD of 3 were used to inoculate a 2.5 L bioreactor (Bioflo 2000, New Brunswick Scientific Co., Edison, NJ) containing 1 L of defined medium with 5 g/L xylitol and 10 g/L glucose or glycerol. Cultures were induced at 2 h with 0.5 mM IPTG. The agitation was maintained at 500 rpm, and air flow rate was controlled at 1.0 L/min to ensure a dissolved oxygen concentration above 40% of saturation. The pH was maintained at 7.0 with 30% (w/v) NH4OH, and the temperature was 37 °C. For intermittent fed-batch cultures, an additional 2 g of the same carbon source in about 2 mL DI water was added to the medium every 5–6 h after cell growth had first plateaued. Mass yield of L-xylulose is the quantity of L-xylulose formed divided by the amount of xylitol consumed (g/g). Conversion is the mass of Lxylulose formed divided by the mass of xylitol initially supplied (g/g). Xylitol was never completely converted to L-xylulose, and the conversion of xylitol is also equal to the fraction of xylitol consumed multiplied by the yield. We also calculated an equilibrium ratio “R” which is the final L-xylulose formed divided by the xylitol remaining (g/g).

cellular processes to regenerate NAD+ in order to capture the electrons and use them for cell maintenance. In growing cells, metabolic processes continually turn over NADH to NAD+, and strategies which favor elevated oxidation of NADH would seem likely to enhance L-xylulose formation. Cofactor re-oxidation in growing cells can potentially be accomplished by a variety of means including delivery of an oxidizing agent (such as O2), using more oxidized substrates, promoting pathways which generate less NAD(P)H or by direct enzyme oxidization. In this study, we have expressed the xdh gene from P. ananatis and the nox gene coding water-forming NADH oxidase (NOX) from Streptococcus pneumoniae [19] to study the effect of NADH regeneration on L-xylulose production in cells growing on glucose or glycerol in shake flasks and controlled batch processes. 2. Materials and methods 2.1. Strains Escherichia coli MG1655 was used in this study (Table 1). The lyxK coding L-xylulokinase was knocked out resulting in strain MEC342 by transducing MG1655 with the corresponding Keio (FRT)Kan deletion [20]. In knockout strains, forward primers external to the target gene and reverse primers within the kanamycin resistance cassette were used to check for proper chromosomal integration.

2.5. Quantification of NAD+ and NADH NAD+ or NADH extraction was prepared by lysing cells with acid or base in 50 °C water bath. Two 1 mL aliquots from the batch culture were centrifuged (20,000 × g for 1 min), and the centrifugates separately resuspended in 250 μL 0.2 M HCl (for NAD+) or 250 μL 0.2 M NaOH (for NADH). The cofactors were extracted from cells in a 50 °C water bath for 10 min. After cooling in an ice bath, 250 μL of 0.1 M NaOH (for NAD+) or 250 μL of 0.1 M HCl (for NADH) was added to neutralize the lysates. The cell-free extracts were then centrifuged (20,000 × g for 5 min), and the supernatants were used to determine the concentration of NAD+ or NADH. The intracellular concentration of NAD+ and NADH was measured by the enzymatic cycling assay [24]. The reaction mixture contained 1.0 M Bicine buffer (pH 8.0 adjusted with 20% NaOH), 40 mM EDTA (pH 8.0 adjusted with 20% NaOH), 16.6 mM phenanzinium ethylsulfate (PES), 4.2 mM thiazolyl blue tetrazolium bromide (MTT), absolute ethanol, and 500 U/mL yeast alcohol dehydrogenase (EC 1.1.1.1, Sigma-Aldrich, St. Louis, MO, Cat. A7011). NAD+ or NADH transfers the reducing equivalents from ethanol ultimately to MTT. By measuring the reduced MTT at 570 nm, the rate of MTT reduction was proportional to the concentration of NAD+ or NADH.

2.2. Plasmid construction The xdh gene from P. ananatis (ATCC 43072) expressing NAD+dependent xylitol-4-dehydrogenase [15] was synthesized by GenScript (Piscataway, USA) as a codon optimized gene for E. coli (Supplement 1). The E. coli glycerol facilitator gene (glpF) and NADH oxidase gene (nox) from Streptococcus pneumoniae were PCR-amplified using, respectively, MG1655 genome and pTrc99A-nox [21] as the templates. Using various primers (Supplement 2), PCR products were gel-isolated, restriction enzyme digested, and ligated into pZE12-luc [22] or pCS27 [23] to yield the plasmids listed in Table 1. These plasmids, called pZE12-xdh pZE12-xdh-glpF, pCS27-glpF and pCS27-nox, were transformed into selected strains by electroporation. 2.3. Growth medium The defined medium for shake flask experiments contained (per L): 13.3 g KH2PO4, 4.0 g (NH4)2HPO4, 1.2 g MgSO4·7H2O, 13.0 mg Zn (CH3COO)2·2H2O, 1.5 mg CuCl2·2H2O, 15.0 mg MnCl2·4H2O, 2.5 mg CoCl2·6H2O, 3.0 mg H3BO3, 2.5 mg Na2MoO4·2H2O, 100 mg Fe(III) citrate, 8.4 mg Na2EDTA·2H2O, 1.7 g citric acid, and 4.5 mg thiamine·HCl. The defined medium for all modes of operation in a controlled bioreactor contained (per L): 1.8 g KH2PO4,

2.6. Enzyme activity Enzyme activities were measured from 10 mL of culture which had 107

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

been centrifuged (5000 × g for 5 min at 4 °C) and washed three times before resuspending in 1 mL of 5 mM phosphate buffer (pH 7.0). Cellfree extracts were prepared by lysing this volume with 0.5 mL of 0.1 mm glass beads in 2 mL microtubes in a Mini Bead Beater (BioSpec Products, Inc., Bartlesville, OK, USA) three times using a cycle of 40 s disruption and 1 min cooling in iced water. The mixture was then centrifuging (11,000 × g for 10 min at 4 °C), and the cell-free extract was used to determine XDH [15] and NOX [21] activities. Briefly for the XDH assay, the reaction mixture contained 50 mM Tris/glycine buffer (pH 7.0 adjusted with 20% NaOH), 500 mM xylitol, 1 mM NAD+, and 10 mM MgCl2. One unit of XDH activity was defined as the amount of enzyme to generate 1.0 μmol of NADH per min. For the NOX assay, the reaction mixture contained 125 mM phosphate buffer (pH 7.0), 2.9 mM NADH, and 5 mM dipotassium EDTA. One unit of NOX activity was defined as the amount of enzyme to consume 1.0 μmol of NADH per min. Total protein concentrations (Pierce BCA protein assay kit, Pierce Biotech, Inc.) were used to calculate specific activity using albumin as the protein standard.

Fig. 1. Comparison of E. coli strains for the conversion of 5 g/L xylitol to L-xylulose after 48 h in shake flask cultures using either 5 g/L glucose or 5 g/L glycerol as carbon/energy source. Conversion is the mass of L-xylulose generated divided by the initial mass of xylitol provided.

3. Results

3.3. The effect of the glycerol facilitator

3.1. The effect of glucose or glycerol on XDH activity and L-xylulose production

The glycerol facilitator (GlpF) is a membrane channel protein reported to aid xylitol transport [26]. To examine whether GlpF affects Lxylulose conversion and yield in growing cells, we transformed MG1655 with the two plasmids pZE12-xdh and pCS27-glpF and compared the bioconversion in cells growing on glucose or glycerol. MG1655/pZE12-xdh/pCS27-glpF showed no change in conversion when glucose was the carbon source and a decrease in conversion using glycerol (Fig. 1). Moreover, L-xylulose yields were markedly lower with the use of the pCS27-glpF plasmid (0.73 g/g xylitol using glucose and 0.58 g/g xylitol using glycerol). We also examined MG1655 with the two genes expressed from a single plasmid MG1655/pZE12-xdh-glpF, which led to a slight increase in conversion on either carbon source (Fig. 1) and in L-xylulose yields from xylitol of 0.81 g/g (glucose) and 0.75 g/g (glycerol). However, since GlpF showed at best only a modest benefit to conversion or yield, the glycerol facilitator protein was not further studied for the conversion of xylitol to L-xylulose using either glucose or glycerol as the carbon source.

In order to compare the effect of different carbon sources on the cell growth and L-xylulose formation from xylitol, MG1655/pZE12-xdh was grown in triplicate in 50 mL defined medium with 5 g/L glucose or 5 g/ L glycerol. The specific growth rate was greater using glucose (0.50 h−1) compared to using glycerol (0.34 h−1), and also the specific activity of XDH was greater when glucose was the carbon source (5.3 U/mg vs. 1.1 U/mg) (Table 2). Despite these differences, the final concentration of L-xylulose was significantly greater when glycerol was the carbon source (approx. 2.5 g/L versus 1.8 g/L). Although the yield based on xylitol consumed was greater for glucose (0.87 g/g) than for glycerol (0.77 g/g), the conversion of xylitol to L-xylulose (Fig. 1) was greater for glycerol (0.44 g/g) than for glucose (0.31 g/g). These results suggest that a high microbial growth rate and enzymatic activity are not necessary for optimal L-xylulose formation from xylitol. 3.2. The effect of L-xylulokinase

3.4. The effect of NADH oxidase

During growth on either glucose or glycerol, the yield of L-xylulose (i.e., 0.77–0.87 g/g) was less than the maximum theoretical yield (0.99 g/g). The lyxK gene encodes L-xylulokinase which has been reported to convert L-xylulose to L-xylulose-5P [25]. To examine whether this enzyme reduces L-xylulose yield, we constructed MEC342/pZE12xdh and again compared the bioconversion of xylitol to L-xylulose in cells growing on either glucose or glycerol. Although the L-xylulose conversion increased on glucose and decreased slightly on glycerol compared to MG1655/pZE12-xdh (Fig. 1), the yield of L-xylulose decreased on both carbon sources (0.82 g/g on glucose and 0.72 g/g on glycerol). These results indicate that L-xylulose formation is not significantly impacted from metabolism through L-xylulokinase.

The conversion of xylitol to L-xylulose by XDH requires NAD+ and generates NADH. We hypothesized that L-xylulose formation might be limited by NADH oxidation, or in other words by the availability of NAD+. In order to examine this possibility, we overexpressed nox coding water-forming NADH oxidase (NOX) and transformed MG1655 with pZE12-xdh and/or pCS27-nox. During growth on either glucose or glycerol, E. coli showed baseline NOX activity of 0.11–0.21 U/mg protein, while MG1655/pCS27-nox showed over ten times the activity (Table 2). When both pZE12-xdh and pCS27-nox were present, the NOX activity was 33–47% lower than observed with MG1655/pCS27-nox. Surprisingly, the XDH activity decreased by 85% in the two plasmid system compared to MG1655/pZE12-xdh during growth on glucose, but increased by over 70% during growth on glycerol (Table 2). It is not clear why this difference in expression occurred, but the result may be due to differences in growth on glucose and glycerol. Although MG1655/pZE12-xdh/pCS27-nox increased the NOX activity in both glucose and glycerol, compared to MG1655/pZE12-xdh the overexpression of NOX had no effect on the conversion of L-xylulose during growth on glucose but significantly increased L-xylulose conversion for cells grown on glycerol (Fig. 1). L-xylulose yield based on xylitol consumed was also greater when NOX was expressed during growth on glucose (0.89 g/g) and on glycerol (0.84 g/g). These results in shake flask cultures show that NAD+ availability improves L-xylulose

Table 2 Specific enzyme activities (Units/mg protein) for xylitol-4-dehydrogenase (XDH) and NADH oxidase (NOX) in E. coli MG1655 expressing these two enzymes on plasmids. Values represent the average (standard deviation) of activity after 48 h of growth in independent 50 mL shake flask cultures. Strains & plasmids

MG1655 pZE12-xdh MG1655 pCS27-nox MG1655 pZE12-xdh/pCS27-nox

Growth on glucose

Growth on glycerol

XDH

NOX

XDH

NOX

5.3 (0.1) N/A 0.7 (0.1)

0.11 (0.02) 6.9 (0.6) 4.6 (0.5)

1.1 (0.1) N/A 1.9 (0.1)

0.21 (0.02) 7.8 (0.4) 4.1 (0.4)

108

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

Fig. 2. L-xylulose formation from 5 g/L xylitol during batch growth of MG1655/pZE12xdh on glucose: OD (○); glucose (▲); xylitol (□); L-xylulose (■); Total NAD(H) (♦).

Fig. 4. L-xylulose formation from 5 g/L xylitol during batch growth of MG1655/pZE12xdh on glycerol: OD (○); glycerol (▼); xylitol (□); L-xylulose (■); Total NAD(H) (♦).

formation, particularly in cells grown on glycerol. The result is correlated with high expression of both XDH and NOX during growth on glycerol for the two plasmid system.

3.5. Batch processes In shake flask culture the greatest conversion of xylitol to L-xylulose occurred during growth on glycerol when both XDH and NOX were overexpressed. Shake flask cultures, however, offer limited control over environmental conditions including pH and oxygenation, which might be different in relatively fast growing cultures (for example, on glucose). Therefore, controlled batch studies at the 1 L scale were completed in duplicate with 5 g/L xylitol and 10 g/L glycerol or 10 g/L glucose using MG1655/pZE12-xdh and MG1655/pZE12-xdh/pCS27nox (Figs. 2–5). Using glucose as carbon/energy source, the specific growth rate for MG1655/pZE12-xdh was 0.50 h−1 and about 66% ( ± 3%) of the xylitol was converted to L-xylulose with a yield of 0.89 ( ± 0.04) g L-xylulose/g xylitol (Fig. 2). Ultimately, the concentrations of L-xylulose and xylitol appeared to reach an equilibrium. For MG1655/pZE12-xdh/ pCS27-nox, a specific growth rate of 0.47 h−1 was obtained, and about 53% ( ± 2%) of the xylitol was converted to L-xylulose with an average yield of 0.82 ( ± 0.04) g L-xylulose/g xylitol (Fig. 3). The equilibrium ratio R (i.e., L-xylulose formed divided by xylitol remaining) was significantly different between these two strains. For MG1655/pZE12-xdh this ratio was 2.41 ( ± 0.52) g L-xylulose/g xylitol, while for MG1655/ pZE12-xdh/pCS27-nox the value of R was 1.42 ( ± 0.09) g/g. The overexpression of NADH oxidase reduced the ratio. Using glycerol as the carbon/energy source compared to glucose, a slightly lower specific growth rate also occurred during exponential growth: 0.35 h−1 or 0.31 h−1 for MG1655/pZE12-xdh or MG1655/

Fig. 5. L-xylulose formation from 5 g/L xylitol during batch growth of MG1655/pZE12xdh/pCS27-nox on glycerol: OD (○); glycerol (▼); xylitol (□); L-xylulose (■); Total NAD (H) (♦).

pZE12-xdh/pCS27-nox, respectively. For MG1655/pZE12-xdh about 70% ( ± 2%) of the xylitol was converted to L-xylulose with a yield of 0.90 ( ± 0.05) g L-xylulose/g xylitol (Fig. 4). For MG1655/pZE12-xdh/ pCS27-nox 69% ( ± 1%) of the xylitol was converted to L-xylulose with a yield of 0.88 ( ± 0.01) g L-xylulose/g xylitol (Fig. 5). In addition to the differences between specific growth rates between growth on glycerol and on glucose, the glycerol-grown cells resulted in much greater values of R: 3.03 ( ± 0.15) g/g or 3.22 ( ± 0.06) g/g, respectively, for MG1655/pZE12-xdh and MG1655/pZE12-xdh/pCS27-nox. In other words, L-xylulose formation during growth on glycerol led to a higher final L-xylulose to xylitol ratio than growth on glucose. All bioreactor experiments suggest that the final L-xylulose and xylitol concentrations approach an equilibrium (Eq. (1)). However, the use of glycerol resulted in a greater L-xylulose/xylitol ratio. The continuous presence of NAD+ is the driving force for the conversion of xylitol to L-xylulose (Eq. (1)). We therefore additionally measured the NAD+ and NADH concentrations during each of the four sets of bioreactor experiments. For MG1655/pZE12-xdh (Figs. 2 and 4), the total NAD(H) concentration (that is, NAD+ plus NADH) at the time of carbon source depletion was about 8.5 μmol/g DCW (glucose) and 7.5 μmol/g DCW (glycerol). For MG1655/pZE12-xdh/pCS27-nox (Figs. 3 and 5), the total NAD(H) concentration was less than 6.0 μmol/ g DCW at the time of either glucose or glycerol exhaustion. In all experiments, the total NAD(H) decreased 40–50% after the carbon source was depleted. During growth on glucose, the total NAD(H) concentration was consistently lower when NADH oxidase was overexpressed (Fig. 4) compared to the endogenous NADH activity (Fig. 2). This effect was not as pronounced during growth on glycerol. We also compared the NAD+/NADH ratio for the four conditions, offset by the approximate time that the carbon source was exhausted (Fig. 6). The NAD+/NADH ratios were low (1.0–3.0) while the carbon

Fig. 3. L-xylulose formation from 5 g/L xylitol during batch growth of MG1655/pZE12xdh/pCS27-nox on glucose: OD (○); glucose (▲); xylitol (□); L-xylulose (■); Total NAD (H) (♦).

109

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

Fig. 6. The NAD+/NADH ratio in E. coli growing on glucose or glycerol: MG1655/pZE12xdh on glucose (○); MG1655/pZE12-xdh on glycerol (△); MG1655/pZE12-xdh/pCS27nox on glucose (●); MG1655/pZE12-xdh/pCS27-nox on glycerol (▲). The time is normalized for each substrate to the approximate time after the carbon/energy source was depleted.

Fig. 8. a) L-xylulose formation from 10 g/L xylitol during growth of MG1655/pZE12-xdh/ pCS27-nox on glycerol. An additional 2 g/L of glycerol was added to the fermenter three times at 7 h intervals after the initial glycerol was depleted. b) Culture dynamics focused on one time interval when an additional 2 g/L of glycerol was added during the conversion of xylitol to L-xylulose for MG1655/pZE12-xdh/pCS27-nox. OD (○); glycerol (▼); xylitol (□); L-xylulose (■); NAD+/NADH (♦). Fig. 7. The relationship between NAD+/NADH and the equilibrium ratio R of L-xylulose/ xylitol (g/g) in E. coli growing on glucose or glycerol: MG1655/pZE12-xdh on glucose (○); MG1655/pZE12-xdh on glycerol (△); MG1655/pZE12-xdh/pCS27-nox on glucose (●); MG1655/pZE12-xdh/pCS27-nox on glycerol (▲).

could be prolonged by reintroducing the carbon source after its initial depletion. We therefore completed experiments in which an additional 2 g carbon source was intermittently added into the 1 Liter bioreactor three times at 5–6 h intervals after its initial depletion. These experiments were completed for both glucose and glycerol using MG1655/ pZE12-xdh with or without pCS27-nox, and the results with glycerol using MG1655/pZE12-xdh/pCS27-nox are shown in Fig. 8a. Each addition of glycerol was accompanied by a decrease in L-xylulose and an increase in xylitol. In other words, the addition of the carbon source resulted in a reversal of the desired reaction: L-xylulose was converted into xylitol while the additional carbon source remained present. Despite three additions of glycerol, the ultimate (68 h) L-xylulose conversion was 58% and the yield was 0.85 g/g, less than the values observed for the batch processes. Analogous results were also observed with glucose and with the absence of NOX: L-xylulose was converted into xylitol while the freshly added carbon source was present (data not shown). We also carefully measured the NAD+ and NADH concentrations during one experiment and during one interval that glycerol was added back into the vessel (glycerol added at 42.2 h; sampling interval was 40–48 h). Fig. 8b shows the NAD+/NADH ratio over that brief time interval, in addition to the glycerol, xylitol and L-xylulose concentrations. These results clearly show that the introduction of glycerol led to an immediate decrease in NAD+/NADH, which corresponded to the conversion of a portion of the L-xylulose back into xylitol. The formation of L-xylulose commenced again only when glycerol became depleted, a time which also corresponded to an increasing NAD+/NADH

source was present, but increased soon after the carbon source was depleted. Among the four experiments, the highest NAD+/NADH ratio (5.0–6.2) was obtained by MG1655/pZE12-xdh/pCS27-nox growing on glycerol, which suggested a relationship between high NAD+/NADH and the equilibrium ratio R. We therefore compared the final NAD+/ NADH ratio with this final equilibrium ratio (L-xylulose/xylitol), and Fig. 7 shows these results. These two ratios are correlated: greater values of NAD+/NADH led to greater final ratios of L-xylulose/xylitol. Surprisingly, overexpression of NADH oxidase reduced the NAD+/ NADH ratio and the equilibrium ratio of L-xylulose/xylitol for cells growing on glucose, but increased both ratios for cells growing on glycerol. The relationship between NAD+/NADH ratio and equilibrium ratio showed the appearance of saturation kinetics, with the equilibrium ratio approaching a finite maximum. Additional increase in NAD+/NADH ratio may have limited effect on the formation of L-xylulose. 3.6. Intermittent fed-batch processes During the batch processes, conversion of xylitol to L-xylulose began during the mid-exponential growth phase, and continued after the carbon and energy sources were depleted. However, the total amount of NAD(H) declined (Figs. 2–5). We speculated that L-xylulose formation 110

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

concentrations in well-aerated bioreactors. In the current study, growing cells in baffled shake flasks to an OD of 5 benefited by NOX overexpression whereas the same cells in well-aerated bioreactors did not exhibit as significant an improvement, suggesting oxygen limitation occurs in shake flasks even after nutrient exhaustion. Resting cells invariably undergo a processing step (e.g., 30 min of handling before xylitol conversion begins [18]), a time during which the NAD+/NADH ratio likely changes, particularly in the absence of available xylitol and oxygen. We speculate that under such conditions cells could drain intracellular NADH for cell maintenance and thus might be comparatively primed for the oxidative conversion. The resting cell approach generated 22 g/L L-xylulose from 250 g/L xylitol, a conversion of 8.8% [18], whereas the growing cells method resulted in 14 g/L L-xylulose from 22 g/L xylitol, a conversion of 65%. Finally, the productivity attained in the current study of 0.19 g/L h includes cell growth, whereas calculations of productivity for resting cell methods seldom include the time used for cell growth and the intermediate preparation and processing steps. These studies demonstrate that the equilibria between a pair of oxidized and reduced biochemicals can be shifted by altering the NAD+/NADH ratio. Indeed, the intermittent fed-batch results (Fig. 8) demonstrate how quickly the equilibrium between biochemicals can be shifted simply by providing a carbon source which alters the balance between reduced and oxidized cofactors. Metabolic flux analysis of E. coli showed glyceraldehyde-3P-dehydrogenase (GAPDH) has the highest contribution to NADH generation under aerobic steady-state growth on glucose [30,31]. Continuous cultivation of E. coli also demonstrated that the expression and activity of GAPDH were positively correlated with introduced glucose [32]. In other words, simply providing a carbon source triggers an increase of activity of native NADdependent enzymes, leading to the prompt generation of more NADH. Maintaining a high NAD+/NADH ratio is essential for the biocatalytic formation of oxidized chemicals, but this balance between NAD+ and NADH is altered by the selection of carbon/energy source. For example, under anaerobic conditions, the NAD+/NADH ratio and metabolite distribution of ethanol and acetate were different for growth on glucose compared to growth on the more reduced sorbitol [33]. In the current study, growth of E. coli on glycerol led to a greater NAD+/ NADH ratio compared to growth on glucose, and this ratio was correlated to an elevated equilibrium ratio R of L-xylulose to xylitol. This result may be due to an intrinsic difference in the cells’ regulation of NAD+/NADH during growth on glycerol compared to glucose, due to the lower growth rate of E. coli on glycerol, or simply because glycerol is more reduced than glucose. In anaerobic culture a more reduced carbon/energy source like glycerol is associated with a lower NAD+/ NADH ratio obtained during growth on a more reduced substrate [34], not a greater NAD+/NADH ratio as observed here. It is not clear that such anaerobic results can be extended to fully aerobic results. Previous research has also demonstrated that NAD+/NADH is correlated to the specific growth rate: as E. coli specific growth rate increased from 0.3 h−1 to 0.5 h−1 the NAD+/NADH ratio decreased by a factor of four [21]. Thus, glucose-grown cells may simply have lower NAD+/NADH ratios because of the faster growth rate of E. coli on this substrate. The interpretation is further complicated by overflow metabolism in E. coli, which is a consequence of an imbalance between NADH generation and the tricarboxylic acid cycle flux during growth on glucose, resulting in acetate formation of acetate and substantially lower NAD+/NADH at high growth rate [19,35]. In this study, over 0.5 g/L acetate was typically formed during shake flask glucose culture, while acetate formation was 90% less during growth on glycerol. The potential for oxygen limitation during fast growth in shake flask further distinguishes growth on glucose compared to growth on glycerol. The expression of a water-forming NOX would be expected to assist oxidization of intracellular NADH in cases where normally oxidation processes were saturated. NOX overexpression exhibited different performance on the redox balance and L-xylulose generation for the two

Fig. 9. L-xylulose formation from 22 g/L xylitol during batch growth of MG1655/pZE12xdh/pCS27-nox on glycerol. OD (○); glycerol (▼); xylitol (□); L-xylulose (■).

ratio. 3.7. Batch process using greater concentrations Our results demonstrate that NADH oxidase benefits L-xylulose formation when the production strain is grown on glycerol (Fig. 5), but that intermittent feeding of carbon source is not effective in increasing final L-xylulose concentration. Because an equilibrium exists between Lxylulose and xylitol, the best approach for increasing the final L-xylulose titer in a growing cell method seems simply to use a higher concentration of xylitol. We therefore performed batch processes using about 12 g/L glycerol and 22 g/L xylitol with MG1655/pZE12-xdh/ pCS27-nox (Fig. 9). The process converted 65% of the xylitol to L-xylulose in about 72 h, with a productivity of 0.19 g/L h and a yield of 0.89 g L-xylulose/g xylitol. The final equilibrium ratio was approximately 2.40 ( ± 0.05). 4. Discussion Whole-cell synthesis of biochemicals mediated by NAD(H)-dependent enzymes is influenced by the NAD+/NADH ratio. In this study, an NAD-dependent xylitol dehydrogenase was examined to convert xylitol to L-xylulose, and the final ratio R of L-xylulose to xylitol was positively correlated with the NAD+/NADH ratio. Maintaining a relatively high NAD+/NADH ratio has previously been shown to favor oxidized biochemical formation. For example, altering the NAD+/NADH ratio of Bacillus subtilis from 0.83 to 2.5 led to a 2-fold increase in the conversion of 2,3-butanediol to the more oxidized acetoin [27]. Similarly, in a recombinant E. coli, the conversion of glycerol to dihydroxyacetone increased from 0.1 to 0.8 g/g DCW by increasing the NAD+/NADH ratio 2-fold [28]. Conversely, anaerobic cells generate more reduced products with a diminished NAD+/NADH ratio. For example, decreasing the E. coli NAD+/NADH ratio from 0.8 to 0.4 as a result of the overexpression of formate dehydrogenase led to a 40% increase in ethanol generation when grown on sorbitol, gluconate or glucose [29]. The NAD+/NADH ratio is also important for oxidations using resting cells. In a previous study of the conversion of xylitol to L-xylulose with the same enzyme using resting cells, the NAD+/NADH ratio was 8–10 one hour after exposing the concentrated cells to xylitol [18]. In the current study, the NAD+/NADH ratio was about 3 times lower (3.2) in MG1655/pZE12-xdh cells one hour after glucose exhaustion (Fig. 6). Also, the L-xylulose/xylitol ratio was much greater in the present study (1.5–3.5) than reported for the extracellular ratio during resting cell conversion (0.07–0.08) [18]. Another similar study using resting cells reported NAD+/NADH ratios of 3–7 [28]. Numerous differences between resting and growing cell approaches exist which complicate the interpretation of these differences. The Usvalampi et al. study measured cofactor concentrations using (unbaffled) shake flasks in a culture having an OD of > 10 [18], whereas the current study reports cofactor 111

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

superior to shake flasks under conditions when other phenomena such as XDH activity are not limiting. Considering the impact of the NAD+/ NADH ratio on the performance of whole-cell biocatalysts, our findings provide guidance for improving NAD+-dependent bioprocesses.

carbon sources examined. In particular, the NAD+/NADH ratio was slightly raised during growth on glycerol but reduced during growth on glucose during NOX overexpression. The cellular response to NOX overexpression likely involves a complex interplay between growth rate, aeration, metabolic fluxes and differential expression level during growth on different carbon sources. During the conversion of xylitol to L-xylulose, growth on glycerol might not only have contributed to a greater NAD+/NADH ratio, but this substrate might also have improved xylitol transport. The uptake of xylitol in E. coli depends on a glycerol facilitator (GlpF, encoded by glpF), which prefers nonchiral linear alditols sharing similar structure as glycerol [26,36]. Glycerol promotes the transcription of glpF and ultimate level of GlpF [37]. Furthermore, a high xylitol uptake rate was observed when both glycerol and xylitol were present [26], and the membrane expression level of GlpF significantly affects glycerol transportation rate [38]. A greater transport rate could cause glycerol-grown cells to maintain a greater intracellular xylitol concentration, which could result in a higher conversion to L-xylulose. Both MG1655/pZE12xdh and MG1655 pZE12-xdh/pCS27-nox attained a greater equilibrium ratio R when grown on glycerol compared to growth on glucose. However, overexpression of glpF either on the same plasmid as xdh or on a second plasmid did not improve xylitol conversion in shake flasks, suggesting that this membrane protein was saturated or that its overexpression did not increase the activity of this membrane-bound protein. Continued availability of NAD+ provides a driving force for efficient NAD+-dependent biotransformations. When the intracellular NAD(H) level is high, NAD+ biosynthesis is repressed and NAD+ degradation is activated [39]. Present results show a significant decrease in total NAD (H) after carbon source depleted, as well as a decline of NAD+/NADH ratio. Synthesis of NAD(H) depends on ATP [40]. However, in the absence of remaining carbon source, NAD+ degradation would likely dominate, limiting ultimate L-xylulose formation. Previous efforts to manipulate the pool of NAD(H) and the NAD+/NADH ratio were able to increase the specific cell production rate [28,41,42]. Since NAD+ degradation will affect the equilibrium between xylitol and L-xylulose, maintaining high total NAD(H) would likely improve the whole-cell biotransformation of reduced substrates to oxidized products. In this study, xylitol was the substrate for L-xylulose formation, while glycerol or glucose was the sole carbon/energy source for biomass formation. Thus, the selection of xylitol and glucose/glycerol maintained a physiological separation of substrates between the formation of biomass and the desired product. Numerous previous studies have focused on the fermentative production of xylitol from lignocellulose-derived D-xylose [43]. These studies typically involved either D-xylose as the sole carbon source [44], or used engineered strains of S. cerevisiae [45], C. tropicalis [46,47] or E. coli [48] for the simultaneous utilization of glucose/glycerol (for biomass formation) and D-xylose (for xylitol formation). Although D-xylose could theoretically be used as both a carbon/energy source for biomass formation while being partially converted to L-xylulose via xylitol, such an approach would require the careful partitioning of this single compound between biomass and the product. Furthermore, the conversion of D-xylose to xylitol is a reduction while xylitol to L-xylulose is an oxidation. Therefore, a redox-based driving force for the conversion of D-xylose to L-xylulose does not exist under either aerobic or anaerobic conditions, making simultaneous utilization of D-xylose for biomass and L-xylulose formation problematic. To our knowledge no study has reported xylitol formation from glucose or glycerol. In summary, the NAD+/NADH ratio is important for driving the formation of L-xylulose from xylitol. Expression of a water-forming NADH oxidase can enhance this NAD-dependent reaction, but also the selection of carbon/energy source impacts the formation of L-xylulose: aerobic growth on glycerol facilitates a high NAD+/NADH ratio which improves the biosynthesis of L-xylulose. Adequate aeration can also be important for the conversion, suggesting that controlled bioreactors are

Conflict of interest The authors declare no financial or commercial conflict of interest. Acknowledgment The authors thank Sarah Lee for technical assistance. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.enzmictec.2017.07.010. References [1] T.B. Granstrom, G. Takata, M. Tokuda, K. Izumori, Izumoring: a novel and complete strategy for bioproduction of rare sugars, J. Biosci. Bioeng. 97 (2) (2004) 89–94. [2] L.S. Jeong, R.F. Schinazi, J.W. Beach, H.O. Kim, S. Nampalli, K. Shanmuganathan, A.J. Alves, A. McMillan, C.K. Chu, R. Mathis, Asymmetric synthesis and biological evaluation of. beta.-L-(2R, 5S)-and. alpha.-L-(2R, 5R)-1, 3-oxathiolane-pyrimidine and-purine nucleosides as potential anti-HIV agents, J. Med. Chem. 36 (2) (1993) 181–195. [3] T. Ma, J.S. Lin, M.G. Newton, Y.C. Cheng, C.K. Chu, Synthesis and anti-hepatitis B virus activity of 9-(2-deoxy-2-fluoro-beta-L-arabinofuranosyl) purine nucleosides, J. Med. Chem. 40 (17) (1997) 2750–2754. [4] C. Mathé, G. Gosselin, L-Nucleoside enantiomers as antivirals drugs: a mini-review, Antivir. Res. 71 (2) (2006) 276–281. [5] E.J. Moran, J.E. Tellew, Z. Zhao, R.W. Armstrong, Dehydroamino acid derivatives from D-arabinose and L-serine: synthesis of models for the azinomycin antitumor antibiotics, J. Org. Chem. 58 (27) (1993) 7848–7859. [6] T.E. Goodwin, K.R. Cousins, H.M. Crane, P.O. Eason, T.E. Freyaldenhoven, C.C. Harmon, B.K. King, C.D. LaRocca, R.L. Lile, S.G. Orlicek, Synthesis of two new maytansinoid model compounds from carbohydrate precursors, J. Carbohydr. Chem. 17 (3) (1998) 323–339. [7] M. Kroger, K. Meister, R. Kava, Low-calorie sweeteners and other sugar substitutes: a review of the safety issues, Compr. Rev. Food Sci. Food Saf. 5 (2) (2006) 35–47. [8] S. Muniruzzaman, Y.T. Pan, Y. Zeng, B. Atkins, K. Izumori, A.D. Elbein, Inhibition of glycoprotein processing by L-fructose and L-xylulose, Glycobiology 6 (8) (1996) 795–803. [9] O. Hiroshi, S. Seui, S. Hiroshi, O. Toshitsugu, Increased urinary excretion of L-xylulose in patients with liver cirrhosis, Clin. Chim. Acta 67 (2) (1976) 131–136. [10] S.H. Bhuiyan, Z. Ahmed, M. Utamura, K. Izumori, A new method for the production of L-lyxose from ribitol using microbial and enzymatic reactions, J. Ferment. Bioeng. 86 (5) (1998) 513–516. [11] J. Jokela, O. Pastinen, M. Leisola, Isomerization of pentose and hexose sugars by an enzyme reactor packed with cross-linked xylose isomerase crystals, Enzyme Microb. Technol. 31 (1–2) (2002) 67–76. [12] R.C. Doten, R.P. Mortlock, Production of D- and L-xylulose by mutants of Klebsiella pneumoniae and Erwinia uredovora, Appl. Environ. Microbiol. 49 (1) (1985) 158–162. [13] A.R. Khan, H. Tokunaga, K. Yoshida, K. Izumori, Conversion of xylitol to L-xylulose by Alcaligenesp. 701B-cells, J. Ferment. Bioeng. 72 (6) (1991) 488–490. [14] W. Poonperm, G. Takata, K. Morimoto, T.B. Granström, K. Izumori, Production of Lxylulose from xylitol by a newly isolated strain of Bacillus pallidus Y25 and characterization of its relevant enzyme xylitol dehydrogenase, Enzyme Microb. Technol. 40 (5) (2007) 1206–1212. [15] J.S. Aarnikunnas, A. Pihlajaniemi, A. Palva, M. Leisola, A. Nyyssola, Cloning and expression of a xylitol-4-dehydrogenase gene from Pantoea ananatis, Appl. Environ. Microbiol. 72 (1) (2006) 368–377. [16] A. Berenguer-Murcia, R. Fernandez-Lafuente, New trends in the recycling of NAD (P)H for the design of sustainable asymmetric reductions catalyzed by dehydrogenases, Curr. Org. Chem. 14 (10) (2010) 1000–1021. [17] G. Takata, W. Poonperm, K. Morimoto, K. Izumori, Cloning and overexpression of the xylitol dehydrogenase gene from Bacillus pallidus and its application to L-xylulose production, Biosci. Biotechnol. Biochem. 74 (9) (2010) 1807–1813. [18] A. Usvalampi, K. Kiviharju, M. Leisola, A. Nyyssölä, Factors affecting the production of L-xylulose by resting cells of recombinant Escherichia coli, J. Ind. Microbiol. Biotechnol. 36 (2009) 1323–1330. [19] G.N. Vemuri, E. Altman, D.P. Sangurdekar, A.B. Khodursky, M.A. Eiteman, Overflow metabolism in Escherichia coli during steady-state growth: transcriptional regulation and effect of the redox ratio, Appl. Environ. Microbiol. 72 (5) (2006) 3653–3661. [20] T. Baba, T. Ara, M. Hasegawa, Y. Takai, Y. Okumura, M. Baba, K.A. Datsenko, M. Tomita, B.L. Wanner, H. Mori, Construction of Escherichia coli K-12 in-frame, single-gene knockout mutants: the Keio collection, Mol. Syst. Biol. 2 (1) (2006)

112

Enzyme and Microbial Technology 106 (2017) 106–113

Q. Han, M.A. Eiteman

[35] M.A. Eiteman, E. Altman, Overcoming acetate in Escherichia coli recombinant protein fermentations, Trends Biotechnol. 24 (11) (2006) 530–536. [36] D. Fu, A. Libson, L.J. Miercke, C. Weitzman, P. Nollert, J. Krucinski, R.M. Stroud, Structure of a glycerol-conducting channel and the basis for its selectivity, Science 290 (2000) 481–486. [37] K. Martínez-Gómez, N. Flores, H.M. Castañeda, G. Martínez-Batallar, G. HernándezChávez, O.T. Ramírez, G. Gosset, S. Encarnación, F. Bolivar, New insights into Escherichia coli metabolism: carbon scavenging, acetate metabolism and carbon recycling responses during growth on glycerol, Microb. Cell Fact. 11 (1) (2012) 46. [38] G. Sweet, C. Gandor, R. Voegele, N. Wittekindt, J. Beuerle, V. Truniger, E.C. Lin, W. Boos, Glycerol facilitator of Escherichia coli: cloning of glpF and identification of the glpF product, J. Bacteriol. 172 (1) (1990) 424–430. [39] J.W. Foster, A.G. Moat, Nicotinamide adenine dinucleotide biosynthesis and pyridine nucleotide cycle metabolism in microbial systems, Microbiol. Rev. 44 (1) (1980) 83. [40] J.W. Foster, A.M. Baskowsky-Foster, Pyridine nucleotide cycle of Salmonella typhimurium: in vivo recycling of nicotinamide adenine dinucleotide, J. Bacteriol. 142 (3) (1980) 1032–1035. [41] S.J. Berrı́os-Rivera, K.Y. San, G.N. Bennett, The effect of NAPRTase overexpression on the total levels of NAD the NADH/NAD+ ratio, and the distribution of metabolites in Escherichia coli, Metab. Eng. 4 (3) (2002) 238–247. [42] L. Liang, R. Liu, X. Chen, X. Ren, J. Ma, K. Chen, M. Jiang, P. Wei, P. Ouyang, Effects of overexpression of NAPRTase NAMNAT, and NAD synthetase in the NAD(H) biosynthetic pathways on the NAD (H) pool, NADH/NAD+ ratio, and succinic acid production with different carbon sources by metabolically engineered Escherichia coli, Biochem. Eng. J. 81 (2013) 90–96. [43] Y.-C. Park, E.J. Oh, J.-H. Jo, Y.-S. Jin, J.-H. Seo, Recent advances in biological production of sugar alcohols, Curr. Opin. Biotechnol. 37 (2016) 105–113. [44] H. Horitsu, Y. Yahashi, K. Takamizawa, K. Kawai, T. Suzuki, N. Watanabe, Prodcution of xylitol from xylose by Candida tropicalis: optimization of production rate, Biotechnol. Bioeng. 40 (1992) 1085–1091. [45] J.H. Jo, S.Y. Oh, H.S. Lee, Y.C. Park, J.H. Seo, Dual utilization of NADPH and NADH cofactors enhances xylitol production in engineered Saccharomyces cerevisiae, Biotechnol. J. 10 (2015) 1935–1943. [46] B.S. Ko, J. Kim, J.H. Kim, Production of xylitol from D-xylose by a xylitol dehydrogenase gene-disrupted mutant of Candida tropicalis, Appl. Environ. Microbiol. 72 (2006) 4207–4213. [47] W. Jeon, B. Yoon, B. Ko, W. Shim, J. Kim, Xylitol production is increased by expression of codon-optimized Neurospora crassa xylose reductase gene in Candida tropicalis, Bioprocess Biosys. Eng. 35 (2012) 191–198. [48] B. Su, M. Wu, Z. Zhang, J. Lin, L. Yang, Efficient production of xylitol from hemicellulosic hydrolysate using engineered Escherichia coli, Metab. Eng. 31 (2015) 112–122.

0008 (2006). [21] G.N. Vemuri, M.A. Eiteman, E. Altman, Increased recombinant protein production in Escherichia coli strains with overexpressed water-forming NADH oxidase and a deleted ArcA regulatory protein, Biotechnol. Bioeng. 94 (3) (2006) 538–542. [22] R. Lutz, H. Bujard, Independent and tight regulation of transcriptional units in Escherichia coli via the LacR/O, the TetR/O and AraC/I1-I2 regulatory elements, Nucleic Acids Res. 25 (6) (1997) 1203–1210. [23] C.R. Shen, J.C. Liao, Metabolic engineering of Escherichia coli for 1-butanol and 1propanol production via the keto-acid pathways, Metab. Eng. 10 (6) (2008) 312–320. [24] B. Bernofsky, M. Swan, An improved cycling assay for nicotinamide adenine dinucleotide, Anal. Biochem. 53 (2) (1973) 452–458. [25] E. Di Luccio, B. Petschacher, J. Voegtli, H.T. Chou, H. Stahlberg, B. Nidetzky, D.K. Wilson, Structural and kinetic studies of induced fit in xylulose kinase from Escherichia coli, J. Mol. Biol. 365 (3) (2007) 783–798. [26] K.B. Heller, E.C. Lin, T.H. Wilson, Substrate specificity and transport properties of the glycerol facilitator of Escherichia coli, J. Bacteriol. 144 (1) (1980) 274–278. [27] T. Bao, X. Zhang, Z. Rao, X. Zhao, R. Zhang, T. Yang, Z. Xu, S. Yang, Efficient wholecell biocatalyst for acetoin production with NAD+ regeneration system through homologous co-expression of 2, 3-butanediol dehydrogenase and NADH oxidase in engineered Bacillus subtilis, PLoS One 9 (7) (2014) e102951. [28] Y.J. Zhou, W. Yang, L. Wang, Z. Zhu, S. Zhang, Z.K. Zhao, Engineering NAD+ availability for Escherichia coli whole-cell biocatalysis: a case study for dihydroxyacetone production, Microb. Cell Fact. 12 (1) (2013) 103. [29] A.M. Sánchez, G.N. Bennett, K.Y. San, Effect of different levels of NADH availability on metabolic fluxes of Escherichia coli chemostat cultures in defined medium, J. Biotechnol. 117 (4) (2005) 395–405. [30] M. Emmerling, J.E. Bailey, U. Sauer, Glucose catabolism of Escherichia coli strains with increased activity and altered regulation of key glycolytic enzymes, Metab. Eng. 1 (2) (1999) 117–127. [31] L.M. Blank, B.E. Ebert, K. Buehler, B. Bühler, Redox biocatalysis and metabolism: molecular mechanisms and metabolic network analysis, Antioxid. Redox Signal. 13 (3) (2010) 349–394. [32] A. Gschaedler, N. Robas, J. Boudrant, C. Branlant, Effects of pulse addition of carbon sources on continuous cultivation of Escherichia coli containing a recombinant E. coli gapA gene, Biotechnol. Bioeng. 63 (6) (1999) 712–720. [33] K.Y. San, G.N. Bennett, S.J. Berrı́os-Rivera, R.V. Vadali, Y.T. Yang, E. Horton, F.B. Rudolph, B. Sariyar, K. Blackwood, Metabolic engineering through cofactor manipulation and its effects on metabolic flux redistribution in Escherichia coli, Metab. Eng. 4 (2) (2002) 182–192. [34] A.M. Sánchez, G.N. Bennett, K.Y. San, Effect of different levels of NADH availability on metabolic fluxes of Escherichia coli chemostat cultures in defined medium, J. Biotechnol. 117 (4) (2005) 395–405.

113