l-Ribose Production from l-Arabinose by Using Purified l-Arabinose Isomerase and Mannose-6-Phosphate Isomerase from Geobacillus thermodenitrificans

Two enzymes, l-arabinose isomerase and mannose-6-phosphate isomerase, from Geobacillus thermodenitrificans produced 118 g/liter l-ribose from 500 g/liter l-arabinose at pH 7.0, 70°C, and 1 mM Co²⁺ for 3 h, with a conversion yield of 23.6% and a volumetric productivity of 39.3 g liter⁻¹ h⁻¹.l-Ribose, a potential starting material for the synthesis of many l-nucleoside-based pharmaceutical compounds, is not abundant in nature (4, 15, 20). l-Ribose has been synthesized primarily from l-arabinose, l-xylose, d-glucose, d-galactose, d-ribose, and d-mannono-1,4-lactone (1, 13, 20). Recombinant cells containing a NAD-dependent mannitol-1-dehydrogenase produced 52 g/liter l-ribose from 100 g/liter ribitol after fermentation for 72 h (14). However, the volumetric productivity of l-ribose was 26-fold lower than that of the chemical synthetic method starting from l-arabinose (6). l-Ribose isomerase from an Acinetobacter sp., which is most active with l-ribose, showed poor efficiency in the conversion of l-ribulose to l-ribose (9). Recently, l-ribulose was produced with a conversion yield of 19% from the inexpensive sugar l-arabinose using l-arabinose isomerase (AI) from Geobacillus thermodenitrificans (18). l-Ribose has been produced from l-ribulose using mannose-6-phosphate isomerase (MPI) from Bacillus subtilis with a conversion yield of 70% (17). In this study, the production of l-ribose from l-arabinose was demonstrated via a two-enzyme system from G. thermodenitrificans, in which l-ribulose was first produced from l-arabinose by AI and subsequently converted to l-ribose by MPI.The analysis of monosaccharides and the purification and thermostability of AI and MPI from G. thermodenitrificans (2) isolated from compost were performed as described previously (7, 18, 19). The cross-linked enzymes were obtained from the treatment of 0.5% glutaraldehyde (10, 16). The reaction was performed by replacing the reaction solution with 100 g/liter l-arabinose and 1 mM Co²⁺ every 6 h at 70°C and pH 7.0. The reaction volume of 10 ml contained 5 g of the cross-linked enzymes with 8 U/ml AI and 20 U/ml MPI. One unit of AI or MPI activity, which corresponded to 0.0625 or 2.5 mg protein, respectively, was defined as the amount of enzyme required to produce 1 μmol of l-ribulose or l-ribose, respectively, per min at 70°C, pH 7.0, and 1 mM Co²⁺. Unless otherwise stated, the reaction was carried out in 50 mM piperazine-N,N′-bis(2-ethanesulfonic acid) (PIPES) buffer (pH 7.0) in the presence of 1 mM Co²⁺ at 70°C for 4 h. All experiments were performed in triplicate.The recombinant Escherichia coli ER2566 (New England Biolabs, Ipswich, MA) containing pTrc99A plasmid (Pharmacia Biotech, Piscataway, NJ) and the AI or MPI gene was cultivated in a 7-liter fermentor containing 3 liters of chemically defined medium (11). When the cell mass reached 2 g/liter, 10 g/liter lactose was added for enzyme induction. After 14 h, 40 g/liter cells with 13,400 U/liter of AI or 34 g/liter cells with 630 U/liter of MPI was obtained. The enzyme was purified by heat treatment and Hi-Trap anion-exchange chromatography. The purification yields of AI and MPI were 21 and 78%, respectively, and the levels of purity for the concentrated AI and MPI by gene scanning were 48 and 92%, respectively. Maximum l-ribose production from l-arabinose by AI and by MPI in 10 ml of total volume was observed at pH 7.0, 70°C, and 1 mM Co²⁺ (data not shown). Half-lives for the two-enzyme system containing 10 mM l-arabinose, 0.2 U/ml AI, and 0.5 U/ml MPI at 60, 65, 70, 75, and 80°C were 1,216, 235, 48, 26, and 12 h, respectively. The use of Co²⁺ may be disadvantageous, as it is fairly toxic. This problem can be solved by using Mn²⁺ instead of Co²⁺. When Mn²⁺ was used in the reaction with the same amounts of enzymes, the conversion yield was the same as that obtained with Co²⁺, even though the volumetric productivity was lower than that with Co²⁺ (data not shown).The effect of the ratio of AI to MPI in the two-step enzymatic production of l-ribose from l-arabinose was investigated by mixing the enzyme solutions (8 U/ml AI and 20 U/ml MPI) to obtain AI/MPI ratios ranging from 10:90 to 90:10 (vol/vol) (Fig. ​(Fig.1).1). The reactions were run with 300 g/liter l-arabinose. Maximum l-ribose production was observed at a volume ratio of 50:50 of the enzyme solutions. The effects of enzyme concentration on l-ribose production were investigated at the optimal unit ratio (AI/MPI ratio, 1:2.5) with 500 g/liter l-arabinose and AI and MPI concentrations from 0.4 and 1.0 U/ml, respectively, to 9.2 and 23.0 U/ml, respectively (Fig. ​(Fig.2A).2A). l-Ribose production increased with increasing amounts of enzymes until reaching a plateau at 8 U/ml AI and 20 U/ml MPI. The effect of substrate concentration on l-ribose production was evaluated at l-arabinose concentrations ranging from 15 to 500 g/liter with 8 U/ml AI and 20 U/ml MPI (Fig. ​(Fig.2B).2B). The production of both l-ribose and l-ribulose, an intermediate, increased with increasing substrate level. The results suggest that concentrations of substrate above 500 g/liter l-arabinose might cause the increased production. The conversion yields of l-ribose and l-ribulose from l-arabinose were constant at 32% and 14%, respectively, within an initial concentration of 100 g/liter l-arabinose, indicating that the reactions reached equilibrium at an l-arabinose/l-ribulose/l-ribose ratio of 54:14:32, which was in agreement with the calculated equilibrium (17). However, at l-arabinose concentrations above 100 g/liter, the conversion yields of l-ribose and l-ribulose from l-arabinose decreased with increasing l-arabinose concentration. The l-arabinose/l-ribulose/l-ribose ratio, with an initial l-arabinose concentration of 300 g/liter, was 71:6:23 after 4 h of reaction. To obtain near-equilibrium (54:14:32) at this high concentration of l-arabinose, more effective enzymes are required.Open in a separate windowFIG. 1.Effect of the ratio of AI to MPI on l-ribose production from l-arabinose by the purified AI and MPI from G. thermodenitrificans. Data are the means for three separate experiments, and error bars represent standard deviations. Symbols: •, l-ribose; ▪, l-ribulose.Open in a separate windowFIG. 2.(A) Effect of enzyme concentration on l-ribose production from l-arabinose at the optimal unit ratio (AI/MPI ratio, 1:2.5). Symbols: •, l-ribose; ▪, l-ribulose; ○, l-arabinose. (B) Effect of l-arabinose concentration on l-ribose production. Symbols: •, l-ribose; ▪, l-ribulose. Data are the means for three separate experiments, and error bars represent standard deviations.A time course reaction of l-ribose production from l-arabinose was monitored for 3 h with 8 U/ml AI and 20 U/ml MPI (Fig. ​(Fig.3).3). As a result, 118 g/liter l-ribose was obtained from an initial l-arabinose concentration of 500 g/liter after 3 h, with a conversion yield of 23.6% and a productivity of 39.3 g liter⁻¹ h⁻¹. Recombinant E. coli containing MDH yielded 52 g/liter l-ribose from an initial ribitol concentration of 100 g/liter after 72 h, with a productivity of 0.72 g liter⁻¹ h⁻¹ (14). The production and productivity obtained in the current study using AI and MPI from G. thermodenitrificans were 2.3- and 55-fold higher, respectively, than those obtained from ribitol and 17- and 21-fold higher than those obtained with the production of l-ribose from l-arabinose using resting cells of recombinant Lactobacillus plantarum (5). The chemical synthetic method is capable of producing 56.5 g/liter l-ribose from 250 g/liter l-arabinose after 3 h, corresponding to a productivity of 18.8 g liter⁻¹ h⁻¹ (6). Still, both the production and productivity of l-ribose using the method described herein were 2.1-fold higher. Thus, the method of production of l-ribose in the present study exhibited the highest productivity and production, compared to other fermentation methods and chemical syntheses.Open in a separate windowFIG. 3.Time course of l-ribose production from l-arabinose by purified AI and MPI from G. thermodenitrificans. Data are the means for three separate experiments, and error bars represent standard deviations. Symbols: •, l-ribose; ▪, l-ribulose; ○, l-arabinose.Several rounds of conversion reusing the cross-linked enzymes were performed (Fig. ​(Fig.4).4). The immobilized enzymes showed more than 20% conversion of l-ribose from l-arabinose for the 9th batch, and the concentration of l-ribose was reduced to 43% after the 20th batch. These results suggest that the immobilization of enzyme facilitates separation of product and enzyme, and it enables the enzyme to function continuously, as reported previously (3, 8, 12). Thus, the reuse of enzyme by immobilization improves the economic viability of this enzymatic process.Open in a separate windowFIG. 4.Reuse of immobilized AI and MPI from G. thermodenitrificans for l-ribose production from 100 g/liter l-arabinose. Data are the means for three separate experiments, and error bars represent standard deviations.