The expression plasmid pET-trx-wt6pgdh was constructed based on a pET102-TOPO plasmid. This construct encodes a fusion protein with a N-terminal thioredoxin, which is well-known to enhance expression of soluble heterologous proteins in E. coli 21. The E. coli BL21(DE3) harboring plasmid pET-trx-wt6pgdh did not produce a detectable protein band corresponding to the size of the fusion protein under various experimental conditions (expression temperatures from 15, 20, 30 to 37°C and IPTG concentrations from 20, 100 to 500 μM). Figure 1A shows a typical result of the cell lysate of E. coli BL21(DE3) induced by 500 μM IPTG at 20°C.

Since the codon usage of E. coli BL21(DE3) is drastically different from that of T. maritima, E. coli BL21(DE3)-RIL that carries a ColE1-compatible vector encoding extra copies of tRNA genes of argU, ileY, and leuW was used to over-express soluble 6PGDH. E. coli BL21(DE3)-RIL/pET-trx-wt6pgdh produced a weak protein band corresponding to the size of the fusion protein under the conditions (500 μM IPTG and 20°C), but a majority of the fusion protein was insoluble (Figure 1B). Decreasing IPTG concentration to 20 μM did not improve the expression of soluble 6PGDH (data not shown), although lowering IPTG concentrations often improved the expression of soluble protein 27.

Because cellulose-binding module (CBM) tags have been reported to enhance heterologous protein expression and folding in E. coli 2044, we attempted to express 6PGDH by replacing the thioredoxin tag with a Clostridium thermocellum family 3 CBM tag linked with an intein. Regardless of IPTG concentrations (20 or 500 μM IPTG), the E. coli BL21(DE3)/pET-trx-wt6pgdh did not produce any obvious soluble or insoluble protein bands corresponding to the right size (Figure 1C, 20 μM IPTG). But BL21(DE3)-RIL bearing the expression plasmid pET-trx-wt6pgdh produced some soluble 6PGDH (Figure 1D, 20 μM IPTG, 20°C). These results suggested that E. coli BL21(DE3)-RIL enhanced expression of the wt6pgdh and the CBM tag helped expression of soluble 6PGDH more efficiently than did thioredoxin.

Through affinity adsorption of CBM-tagged 6PGDH on regenerated amorphous cellulose followed by intein self-cleavage 44, approximately eight mg of 6PGDH was purified per liter of the culture. But the purified 6PGDH was composed of several small-size proteins (Lane P, Figure 1D), suggesting possible proteolysis or incomplete translation. The first cause was eliminated because addition of a protease inhibitor phenylmethanesulfonyl fluoride during cell disruption and protein purification did not change the composition of the small-size proteins (data not shown).

Figure 2 shows the wt6pgdh DNA sequence, the deduced amino acids, and the codon-optimized DNA sequence (co6pgdh). The wt6pgdh gene contains 47 rare E. coli codons, accounting for ~10% of the entire sequence. They are 20 AGA, 5 AGG, 19 AUA, and 3 CUA. The AGA(Arg) and AUA(Ile) codon frequencies in the wt6pgdh gene are 4.3% and 4.0%, but are only 0.24% and 0.5% in E. coli http://www.kazusa.or.jp/codon, respectively. Moreover, two rare codons formed clusters AUA(Ile97)-AUA(Ile98) and AGA(Arg306)-AGA(Arg307). Following site-directed mutagenesis to remove these two rare-codon clusters, there were no noticeable changes in the SDS-PAGE patterns of the purified 6PGDHs before and after site-directed mutagenesis (data not shown). Therefore, the entire wt6pgdh DNA sequence was optimized to remove all 47 rare codons by using frequently-used E. coli codons, based on several rules: (i) keeping the GC ratio around 50%; (ii) avoiding cis-acting DNA sequences (internal TATA-boxes, chi-sites, and ribosomal entry sites; AT-rich or GC-rich sequence stretches; repeat sequences; and RNA secondary structures); (iii) precluding cutting sites of frequently-used restriction enzymes, and (iv) adding two stop TAATAA to ensure efficient termination of translation, and (v) using a strong terminator in the expression vector for enhancing mRNA stability. The overall GC content for the codon-optimized 6PGDH was 49% (Figure 2).

Plasmid pET-ci-co6pgdh encoding the fusion protein CBM-intein-6PGDH was expressed in E. coli BL21(DE3) and BL21(DE3)-RIL, separately. Figure 3 shows that both hosts produced soluble target proteins at similar levels, suggesting that E. coli BL21(DE3)-RIL was not necessary for expression of the codon-optimized gene. After purification, no small-size protein fragments accompanied with the purified 6PGDH were observed (Figure 3), suggesting that the rare codons mainly caused the incomplete translation (Figure 1D). Approximately 15–17 mg of 6PGDH was purified per liter of culture for both hosts.

The cleaved T. maritima 6PGDH through RAC adsorption and intein cleavage was purified and characterized. The pH effects on 6PGDH activity were studied in 50 mM citric acid/sodium citrate buffers (pH 5.0 and 6.0), a Bis-Tris (pH 6.5), Tris-HCl buffers (pH 7.0, 7.5 8.0, 8.5, and 9.0), and a Hepes buffer (pH 7.5). The optimum pH was found to be around pH 7.0 (Figure 4A). About 70% of 6PGDH activities remained at pH 6.0 and 9.0. The enzyme had the similar activities in the Hepes and Tris buffers (pH 7.5).

The effects of temperature on T. maritima 6PGDH activities were measured from 20 to 95°C (Figure 4B). The optimum temperature was around 90°C. The 6PGDH had an approximately 90% of its maximum activity at 80°C, but retained only ~2 and ~20% of its maximum activity at 30 and 60°C, respectively. The activation energy was 51.4 kJ/mol at a temperature range of 20–80°C, based on the Arrhenius plot. The 6PGDH was highly thermostable (Figure 4C) in a 50 mM Hepes buffer (pH 6.8) containing 500 mM NaCl, 1 mM EDTA, and 5 mM β-mercaptoethanol. It retained more than 90% activity for 48 h at 60 and 80°C and retained ~50% enzymatic activity after 48 h at 90°C. This enzyme had half-life times of 48 and 140 h at 90 and 80°C, respectively.

The kinetics properties of T. maritima 6PGDH followed the Michaelis-Menten equation. The apparent K_m values were 11, 10, and 380 μM on 6-phosphogluconate, NADP⁺, and NAD⁺, respectively and k_cat value was 325 s^-1. Clearly, this enzyme preferred NADP⁺ to NAD⁺ as an electron acceptor.

Codon analysis indicated that the C. thermocellum CBM and the Synechocystis intein contained 21 rare codons. In addition, self-cleavage of intein during protein expression and cell disruption may result in some loss of the desired protein even at a decreased cultivation temperature 44. Expression vector pET-co6pgdh-his was constructed to express a co6pgdh gene with a C-terminal His-tag. Figure 5A shows the SDS-PAGE analysis of 6PGDH expression. More than 200 mg of 6PGDH-His was purified per liter of culture. But the specific enzymatic activity of the 6PGDH-His was approximately 80% of that of 6PGDH without the His tag.

Since T. maritima 6PGDH was extremely thermostable (Figure 4C), heat precipitation was chosen for simplifying protein purification. Plasmid pET-co6gpdh was constructed for expressing the co6gpdh gene without a His-tag. The cell lysate was treated at 90 or 100°C with time lengths ranging from 15 minute to 6 hours. The highest 6PGDH yield was obtained under heat treatment conditions (90°C for 30 min), where a 6PGDH purity was around 85%, as judged by SDS-PAGE analysis (Figure 5B). The overall recovery yield was 90% according to 6PGDH activity. Approximately 190 mg of 6PGDH was obtained from the cells harvested after 4-hour induction at 37°C.

Figure 6 shows the fermentation profiles of E. coli BL21(DE3)/pET21-co6pgdh in 200 mL of LB medium in a 1-L Erlenmeyer flask at a constant cultivation temperature of 37°C. When the A₆₀₀ reached 0.6, 500 μM of IPTG was added. The highest A₆₀₀ was 3.9, and the total cell protein was 790 mg per liter at hour 12 (9.5 hours after IPTG induction). The 6PGDH content rose from ~10% before induction to 38% at hour 4.5 and then decreased to levels of ~30% for the remaining cultivation period. Up to 230 mg of 6PGDH was produced from one liter of the LB-grown culture.

To decrease the inducer cost, different concentrations of IPTG (20, 100 or 500 μM) as well as lactose (100 μM) were further investigated. Nearly all cells had similar growth patterns except that 500 μM IPTG slightly inhibited cell growth during the first six-hour induction. After cell lysis and heat precipitation, approximately 194, 224, 208, and 250 mg of the 6PGDH protein were obtained under the conditions of 20, 100, and 500 μM IPTG, and 100 μM lactose, respectively. The largest amount of 6PGDH was obtained when lactose was used as the inducer, suggesting that low-cost lactose was an effective inducer and worked as a supplementary carbon source for protein synthesis.

Coupling of 6PGDH and G6PDH was believed to generate one more mole of NADPH per mole of glucose-6-phosphate relative to G6PDH alone. Xylitol can be produced from xylose and NADPH mediated by xylose reductase 4546. Figure 7 shows kinetics of xylitol synthesis with glucose-6-phosphate as the NADPH regeneration substrate and G6PDH in the presence and absence of 6PGDH. In the case of three enzyme cocktails (G6PDH, 6PGDH, and xylose reductase), 35.5 mM xylitol was produced at hour six, nearly twice that of the two-enzyme system (19.5 mM, G6PDH and xylose reductase). The yields of xylitol synthesized were 175% and 97%, respectively, relative to glucose-6-phosphate consumed for the reactions mediated by the three enzymes and by the two enzymes.

All chemicals were of reagent grade, purchased from Sigma (St. Louis, MO) and Fisher Scientific (Pittsburgh, PA), unless otherwise noted. Regenerated amorphous cellulose (RAC) was prepared through cellulose dissolution by ice-cooled concentrated phosphoric acid followed by regeneration in water 55. Pfx50 DNA polymerase, Champion™ pET102 Directional TOPO^® Expression Kit with E. coli BL21 Star™ (DE3), and Ni-NTA agarose were purchased from Invitrogen (Carlsbad, CA). The T. maritima genomic DNA was purchased from the American Type Culture Collection (Manassas, VA). The strains, plasmids, and oligonucleotides used in this study are listed in Table 1.

Five expression plasmids were constructed for expressing wild-type 6pgdh (wt6pgdh) and codon-optimized 6pgdh (co6pgdh) genes under the control of T7 promoter (Table 2). Plasmid pET-trx-wt6pgdh encoding a fusion protein of thioredoxin (Trx) and 6PGDH was constructed by insertion of amplified wt6pgdh gene into pET102 Directional TOPO^® . The wt6pgdh DNA fragment was PCR amplified using primers Trx-F and Trx-R from the T. maritima genomic DNA. Plasmid pET-ci-wt6pgdh was constructed by replacing the pgm gene in the plasmid pCIP 44 by the wt6pgdh gene. The whole 6pgdh DNA sequence was optimized based on the codon usage for E. coli B http://www.kazusa.or.jp, yielding the co6pgdh DNA sequence. The co6pgdh DNA sequence with a C-terminal His-tag was synthesized by Epoch Biolabs (Sugar Land, TX) and cloned into pET21a via the restriction endonuclease sites Nde1 and BamH1 to obtain pET-co6pgdh-his. Plasmid pET-ci-co6pgdh was constructed similarly to pET-ci-wt6pgdh using the co6pgdh DNA sequence. Plasmid pET-co6pgdh encoding 6PGDH protein without a His-tag was constructed based on pET-co6pgdh-his by replacing the first codon (CAT) of the His-tag with a stop codon (TAA). The site-directed mutagenesis was conducted by using primers NH-F and NH-R following QuikChange™ Site-Directed Mutagenesis (Stratagene, La Jolla, CA).

The E. coli strain harboring the expression plasmid was grown in a 200 mL LB medium in a 1-L flask with appropriate antibiotics at 37°C until the A₆₀₀ reached ~0.6. After addition of the inducer (isopropyl β-D-1-thiogalactopyranoside – IPTG or lactose), the culture were grown at 37°C or a lower temperature (e.g. 20°C). For the CBM-tag proteins, they were purified through affinity binding on RAC followed by intein self-cleavage 44. For the His-tag proteins, the cells were disrupted by sonicator in a 50 mM Tris-HCl (pH 7.5) containing 500 mM NaCl, 5 mM imidazol, and 20% (w/v) glycerol. After centrifugation at 8,000 g for 5 min, the supernatant was collected and purified by using affinity binding on Ni-NTA resin. For purification through heat precipitation, the soluble fraction of E. coli cell lysate was incubated at 90°C for 30 min in a 50 mM Tris-HCl (pH 7.5) buffer containing 100 mM NaCl and 20% (w/v) glycerol. After centrifugation, the supernatant contained relatively pure 6PGDH.

The T. maritima glucose-6-phosphate dehydrogenase (G6PDH) was produced by E. coli BL21(DE3)-RIL/pET-ci-g6pdh. The G6PDH was purified through affinity binding on RAC followed by intein self-cleavage 44 and characterized as described 56. The Neurospora crassa xylose reductase was expressed and purified as described elsewhere 57.

T. maritima 6PGDH activity was measured in a 50 mM Hepes buffer (pH 7.5) containing 2 mM 6-phosphogluconate, 1 mM NADP⁺, 5 mM Mg²⁺, 0.5 mM Mn²⁺, and 0.5 mg bovine serum albumin per mL at 80°C for 5 min. The reaction product NADPH was measured at 340 nm by DU^® 800 UV/visible spectrophotometer (Beckman Coulter, Fullerton, CA). The enzyme unit was defined as one μmole of NADPH produced per min. For determining enzyme kinetic parameters, the K_m of 6-phosphogluconate was measured in a 50 mM Hepes (pH 7.5) buffer containing 1 mM NADP⁺, 5 mM Mg²⁺, 0.5 mM Mn²⁺, along with 2.5 to 50 μM 6-phosphogluconate; the K_m of NADP⁺ was measured in the same buffer 2 mM 6-phosphogluconate with various concentrations of NADP⁺ from 2.5 to 50 μM. The K_m of NAD⁺ was measured using 50 to 1000 μM NAD⁺.

Synthesis of xylitol from xylose mediated by xylose reductase was conducted in a 200 μL reaction volume in the presence of G6PDH or G6PDH/6PGDH at 25°C. The reaction mixture contained a 50 mM Hepes buffer (pH 7.5) with 50 mM xylose, 1 mg xylose reductase/mL, 2 mM NADP⁺, 20 mM glucose-6-phospahte, 0.16 mg T. maritima 6PGDH/mL, 0.3 mg T. maritima 6PGDH/mL, 1 mg BSA/mL, 0.5 mM MnCl₂ and 5 mM MgCl₂. A 10 μL of the sample was withdrawn and diluted 20-fold in 5 mM H₂SO₄. Xylitol was measured by a HPLC equipped with the Bio-Rad Aminex HPX-87H column 59.