An In Vitro Enzyme System for the Production of myo-Inositol from Starch

ABSTRACT

We developed an in vitro enzyme system to produce myo-inositol from starch. Four enzymes were used, maltodextrin phosphorylase (MalP), phosphoglucomutase (PGM), myo-inositol-3-phosphate synthase (MIPS), and inositol monophosphatase (IMPase). The enzymes were thermostable: MalP and PGM from the hyperthermophilic archaeon Thermococcus kodakarensis, MIPS from the hyperthermophilic archaeon Archaeoglobus fulgidus, and IMPase from the hyperthermophilic bacterium Thermotoga maritima. The enzymes were individually produced in Escherichia coli and partially purified by subjecting cell extracts to heat treatment and removing denatured proteins. The four enzyme samples were incubated at 90°C with amylose, phosphate, and NAD⁺, resulting in the production of myo-inositol with a yield of over 90% at 2 h. The effects of varying the concentrations of reaction components were examined. When the system volume was increased and NAD⁺ was added every 2 h, we observed the production of 2.9 g myo-inositol from 2.9 g amylose after 7 h, achieving gram-scale production with a molar conversion of approximately 96%. We further integrated the pullulanase from T. maritima into the system and observed myo-inositol production from soluble starch and raw potato with yields of 73% and 57 to 61%, respectively.

IMPORTANCE myo-Inositol is an important nutrient for human health and provides a wide variety of benefits as a dietary supplement. This study demonstrates an alternative method to produce myo-inositol from starch with an in vitro enzyme system using thermostable maltodextrin phosphorylase (MalP), phosphoglucomutase (PGM), myo-inositol-3-phosphate synthase, and myo-inositol monophosphatase. By utilizing MalP and PGM to generate glucose 6-phosphate, we can avoid the addition of phosphate donors such as ATP, the use of which would not be practical for scaled-up production of myo-inositol. myo-Inositol was produced from amylose on the gram scale with yields exceeding 90%. Conversion rates were also high, producing over 2 g of myo-inositol within 4 h in a 200-ml reaction mixture. By adding a thermostable pullulanase, we produced myo-inositol from raw potato with yields of 57 to 61% (wt/wt). The system developed here should provide an attractive alternative to conventional methods that rely on extraction or microbial production of myo-inositol.

INTRODUCTION

In recent years, the use of thermostable enzymes from (hyper)thermophilic organisms for in vitro enzyme-based production of biofuels and chemicals has been gaining attention, and a wide range of applications have been documented (1–6). These include one-pot conversions of glucose to pyruvate or lactate (7, 8), ethanol or n-butanol (9, 10), and bioplastic (11). The production of hydrogen from xylose and/or glucose (12, 13), of glucose 1-phosphate (G1P) from starch (14), and of glucose and amylose from cellulose (15) has been demonstrated. Other applications include development of enzyme-based high-energy-density sugar biobatteries (16). In these enzyme systems, competing reactions that would otherwise be present in whole-cell systems are absent, providing opportunities to elevate the yield.

The use of enzymes from (hyper)thermophiles provides a number of advantages in constructing in vitro enzyme systems (7–10, 12–16). The first advantage is that it is possible to remove or deactivate the activities of enzymes deriving from the host cell that might compete with the designed pathway simply by incubating the cell extracts at high temperature. This removes the necessity to strictly purify the enzymes, which is an attractive trait when considering use in industry, since costly downstream purification schemes would not be required. The second advantage is the high stability of the proteins, allowing their use for longer periods of time. The third advantage is that the system can be operated at higher temperature, accelerating the reaction rate. Acceleration is further achieved by the fact that substrate solubility may significantly increase at high temperatures, enhancing the initial velocity of enzyme reactions.

myo-Inositol is a cyclitol present in all three domains of life (17, 18). It is present in lipids, as phosphatidylinositol phosphate lipids in bacteria/eukaryotes and archaetidylinositol in archaea. Its phosphorylated derivatives also function as major secondary messengers in eukaryotes. The association of myo-inositol, along with d-chiro-inositol, with glucose homeostasis has been suggested, and their use as dietary supplements displayed insulin-mimetic effects (18). These results suggest that myo-inositol may provide a wide variety of benefits as a dietary supplement.

myo-Inositol can be obtained by extracting its phosphorylated derivatives such as phytic acid from corn or other grains, which can be dephosphorylated by phytases (19, 20). Studies that report the microbial production of myo-inositol are limited. Escherichia coli cells expressing the INO1 gene from Saccharomyces cerevisiae have been reported to produce myo-inositol and myo-inositol 3-phosphate (I3P) with a combined yield of 11% from glucose (21). As shown in Fig. 1A, myo-inositol production from glucose is achieved via glucose 6-phosphate (G6P) and I3P. G6P is a key node of central metabolism and is the starting compound of the pentose phosphate pathway initiating with glucose-6-phosphate dehydrogenase, of saccharide biosynthesis initiating with phosphoglucomutase, and of glycolysis with glucose-6-phosphate isomerase. We thus presumed that obtaining high yields of myo-inositol from glucose (via G6P) with in vivo metabolic engineering strategies would be very difficult.

FIG 1.

(A) Pathways related to glucose 6-phosphate metabolism. Enzyme reactions responsible for the conversion of glucose to myo-inositol are indicated with bold arrows. For simplicity, cofactors are not shown for some of the reactions. (B) The three enzyme reactions responsible for the conversion of glucose to myo-inositol. In the MIPS reaction, NAD⁺ is once reduced to NADH in the first half of the MIPS reaction but is regenerated in the second half of the reaction. (C) The designed pathway for the in vitro conversion of starch or amylose to myo-inositol. Abbreviations: GK, ATP- or ADP-dependent glucokinase; MIPS, myo-inositol-3-phosphate synthase; MalP: maltodextrin phosphorylase; PGM, phosphoglucomutase; IMPase, inositol monophosphatase.

Thermococcus kodakarensis (22, 23), Archaeoglobus fulgidus (24), and Thermotoga maritima (25) are all hyperthermophiles that display optimal growth temperatures of 80 to 85°C. T. kodakarensis and A. fulgidus are archaea, whereas T. maritima is a bacterium. The complete genome sequences of these organisms have been determined (26–28), and gene manipulation systems have been established for T. kodakarensis (29–33) and T. maritima (34). We have been examining the metabolism of T. kodakarensis in detail and have identified a number of enzymes and pathways that differ from the conventional pathways identified in bacteria and eukaryotes (35–37). Glycolysis in this archaeon is brought about by a modified Embden-Meyerhof pathway, which is found in members of the Thermococcales (38). Sugar phosphorylation is catalyzed by ADP-dependent glucokinase (GK) and ADP-dependent phosphofructokinase (PFK) (39, 40). The conversion of glyceraldehyde 3-phosphate (GAP) to 3-phosphoglycerate is catalyzed by a single-step oxidation coupled to the reduction of ferredoxin (GAP:ferredoxin oxidoreductase) (41) or by a nonphosphorylating GAP dehydrogenase (42). T. kodakarensis does not utilize glucose as an external carbon source but can utilize maltooligosaccharides composed of three or more glucose units (31). After the oligosaccharides are transported via an ABC transporter, it is presumed that the sugars are hydrolyzed to glucose within the cell, which is then phosphorylated by ADP-dependent GK. On the other hand, the genome of T. kodakarensis also harbors a gene (TK1406) predicted to encode a maltodextrin phosphorylase (MalP). The function of MalP would convert the maltooligosaccharides to G1P using free phosphates. T. kodakarensis also harbors a phosphoglucomutase (PGM), which we have previously characterized, encoded by TK1108 (43). It is therefore possible that this organism converts maltooligosaccharides to G6P via MalP and PGM.

In this study, we used the thermostable MalP and PGM from T. kodakarensis to produce G6P from maltodextrin. By utilizing phosphorolysis (MalP) and phosphomutase (PGM) reactions instead of hydrolysis (amylase) and phosphorylation (glucokinase) reactions, this enables us to generate G6P without the consumption of phosphate donors such as ATP. By combining the two enzymes with a thermostable myo-inositol-3-phosphate synthase (MIPS) from A. fulgidus (29) and an inositol monophosphatase (IMPase) from T. maritima (30), we designed and developed an in vitro enzyme system that displays rapid and high-yield production of myo-inositol from starch compounds.

RESULTS

Design of an enzyme system to produce myo-inositol from amylose.

Based on low observed yields of myo-inositol from engineered E. coli, we expected difficulties in obtaining high yields of myo-inositol from glucose via microbial production due to the number of competing enzymes and pathways (Fig. 1A). One solution would be to utilize the enzymes involved in myo-inositol biosynthesis in an in vitro system. The enzymes necessary would be GK, MIPS, and IMPase (Fig. 1B). However, in this system, the synthesis of one molecule of myo-inositol would require the hydrolysis and consumption of one molecule of ATP (or ADP) and would not be practical for scaled-up myo-inositol production. By using the products of a putative MalP gene (TK1406) and a PGM encoded by TK1108, it would be possible to convert maltodextrin to myo-inositol without the stoichiometric consumption of a phosphate donor such as ATP or ADP (Fig. 1C). The four-enzyme system utilizes (i) MalP, which converts the glucose unit on the nonreducing end of a maltodextrin chain to G1P with free phosphate, (ii) PGM, which catalyzes the phosphotransfer of G1P to G6P, (iii) MIPS, which catalyzes the isomerization of G6P to I3P, and (iv) IMPase, which releases the phosphate from I3P to generate myo-inositol. The phosphate is thus regenerated and can be used for the next conversion from maltodextrin to myo-inositol. The MIPS reaction is dependent on NAD⁺, but the reaction consists of two steps in which electrons are first transferred from G6P to NAD⁺ (oxidation at C-5) and later returned to the intermediate myo-inosose phosphate to generate I3P (44). There is therefore no net consumption of NAD⁺ during the MIPS reaction, and in principle, the four-enzyme reactions provide a sustainable system for the conversion of maltodextrin to myo-inositol without the consumption of phosphate donors such as ATP.

Cloning and expression of the four enzyme genes.

The genes we selected for MalP and PGM production were TK1406 and TK1108 from T. kodakarensis. For production of MIPS, we selected AF1794 from the hyperthermophilic archaeon Archaeoglobus fulgidus. The protein has previously been demonstrated to catalyze the conversion of G6P to I3P dependent on NAD⁺ (44). For IMPase, we selected TM1415 from the hyperthermophilic bacterium Thermotoga maritima. The protein was shown to be a phosphatase with activity specific toward I3P (45). We have previously characterized the IMPase from T. kodakarensis, but the enzyme was not specific toward I3P and also displayed high phosphatase activity toward fructose 1,6-bisphosphate (46). We thus chose not to use the enzyme from T. kodakarensis, as there was the possibility that it would act toward G1P and/or G6P. All of the enzymes selected here are derived from hyperthermophiles and thus display the advantages described in the introduction.

The four genes were individually expressed in E. coli. The cells were collected, disrupted by sonication, and incubated at 90°C for 10 min in order to denature/inactivate the enzymes from the host E. coli cells. After removing precipitated proteins, the supernatants were directly used as the enzyme samples. SDS-PAGE analyses of the heat-treated protein samples are shown in Fig. 2. Although individual protein levels varied, we observed significant and consistent amounts of each individual protein in the heat-treated protein samples. The content of each protein in the protein samples was as follows: MalP, 67.1% ± 6.9%; PGM, 79.6% ± 5.6%; MIPS, 93.8% ± 2.2%; and IMPase, 95.7% ± 2.5%. Taking into account the industrial potential of this system, we directly used these heat-treated protein samples as the source of enzyme in order to avoid as many steps as possible in enzyme preparation.

FIG 2.

SDS-PAGE analysis of the protein samples used in the study. Cell extracts of E. coli were subjected to heat treatment at 90°C, followed by centrifugation. The supernatants (1 μg protein per lane) were applied to SDS-PAGE and stained with Coomassie brilliant blue. Lane M, molecular mass marker; lane 1, MalP sample; lane 2, PGM sample; lane 3, MIPS sample; lane 4, IMPase sample. Bands corresponding to the individual recombinant proteins are indicated with arrows.

Production of G6P from amylose with MalP and PGM.

We first examined the production of G6P from amylose using MalP and PGM. Amylose is a glucose polymer linked only with α-1,4-linkages, which is the bond recognized by MalP. The amylose chains used in this study have an average length of approximately 18 glucose units. MalP and PGM samples were incubated with 5 mM amylose and 10 mM sodium phosphate (a detailed description of the reaction components is in Materials and Methods) at 70°C for 10 min, and G6P production was quantified using G6P dehydrogenase. Only in the presence of all components could we observe the generation of G6P (Fig. 3A). No production was observed when MalP, PGM, or amylose was omitted from the reaction mixture.

FIG 3.

(A) G6P production from amylose using MalP and PGM. The concentration of G6P was measured after 10 min of reaction at 70°C. (B) myo-Inositol production from G6P. The reaction products were examined by HPLC after reactions with both substrate and enzyme (red), with only substrate (green), or with only enzyme (blue). The dashed line is the result of applying a myo-inositol standard. (C) myo-Inositol production from amylose. The reaction products were examined with HPLC after 0 min (black), 30 min (green), 60 min (blue), and 120 min (red) of reaction with components shown in Table 1 for system A.

Production of myo-inositol from G6P with MIPS and IMPase.

We next examined whether G6P could be converted to myo-inositol using the A. fulgidus MIPS and T. maritima IMPase. Both enzyme samples were incubated with 100 mM G6P and 10 mM NAD⁺ at 90°C for 30 min, and myo-inositol was detected with high-performance liquid chromatography (HPLC). We could clearly observe the increase of a peak that displayed the same retention time as the myo-inositol standard (Fig. 3B).

Production of myo-inositol from amylose with MalP, PGM, MIPS, and IMPase.

As the combinations of two enzymes displayed the expected conversions, we used all four enzymes and incubated them with 5 mM amylose along with 50 mM sodium phosphate and 1 mM NAD⁺ at 90°C. In initial trials, we added relatively high concentrations of enzyme, 1.5- to 2-fold higher than those indicated in Table 1 for system A. We observed the generation of a peak that displayed the same retention time as the myo-inositol standard. By individually decreasing the amount of each protein added to the reaction mixture, we were able to reduce the amount of proteins added while maintaining the conversion rate. MIPS was added in excess (2-fold) in order to prevent the MIPS reaction from being rate limiting. When MIPS was rate limiting, we observed an increase in the generation of glucose, due to the thermal breakdown of G6P, the substrate of MIPS. Supporting this, when we incubated 100 mM G1P or G6P at 90°C for 3 h in our reaction mixture, we observed the generation of approximately 1 mM glucose from G1P, while over 4 mM glucose was generated from G6P, indicating that G6P is the more labile of the two.

TABLE 1.

Composition of the in vitro enzyme reaction systems with amylose as the substrate

Component (gene) or parameter	Concn or amt in system:
	A	B	C	D	E
Tris-HCl, pH 7.5	100 mM	100 mM	100 mM	100 mM	100 mM
MgCl2	10 mM	10 mM	10 mM	10 mM	10 mM
Disodium hydrogen phosphate	50 mM	50 mM	50 mM	50 mM	50 mM
Glucose 1,6-bisphosphate	50 μM	0 or 50 μM
NAD+	1 mM	1 mM	Xa mM	1 mMb	1 mMc
Amylose	5 mM	5 mM	5 mM	5 or 10 mMb	5 mM
MalP (TK1406)	100 μg (1.0 nmol)	100 μg	100 μg	100 μg	10 mg (0.1 μmol)
PGM (TK1108)	400 μg (8.0 nmol)	400 μg	400 μg	400 μg	40 mg (0.8 μmol)
MIPS (AF1794)	500 μg (11.5 nmol)	500 μg	500 μg	500 μg	50 mg (1.15 μmol)
IMPase (TM1415)	800 μg (28 nmol)	800 μg	800 μg	800 μg	80 mg (2.8 μmol)
Total vol	2 ml	2 ml	2 ml	2 ml	200 ml
Figure(s)	3C	4, 5A	5B	6, 7A	7B

^aX = 0, 0.5, 1, 2, 5, or 10; see Fig. 5B legend for details.

^bInitial concentration; see Fig. 6 and 7A legends for details.

^cInitial concentration; see Fig. 7B legend for details.

Using a reaction mixture with the composition shown in Table 1 for system A, we observed a steady increase in the peak presumed to correspond to myo-inositol (Fig. 3C). We purified the compound and examined its structure with proton nuclear magnetic resonance (¹H NMR). The obtained ¹H NMR spectrum was identical to that obtained with a myo-inositol standard (Fig. 4), confirming that the conversion from amylose to myo-inositol was achieved with the four-enzyme in vitro system.

FIG 4.

¹H NMR spectra of a myo-inositol standard (top) and the product (bottom) of the in vitro enzyme system. myo-Inositol standard: ¹H NMR (600 MHz, D₂O), δ 3.08 (1H, t, J = 9.6 Hz), δ 3.34 (2H, dd, J = 10.2 Hz, J = 3.0 Hz), δ 3.43 (2H, t, J = 9.9 Hz), δ 3.87 (1H, t, J = 2.7 Hz). Product: ¹H NMR (600 MHz, D₂O), δ 3.08 (1H, t, J = 9.6 Hz), δ 3.34 (2H, dd, J = 10.2 Hz, J = 3.0 Hz), δ 3.43 (2H, t, J = 9.6 Hz), δ 3.87 (1H, t, J = 3.0 Hz). Products were obtained from a reaction (2 h) with the composition shown in Table 1 for system B without GBP.

We next varied the concentrations of various components of the system in order to avoid adding them in excess. Glucose 1,6-bisphosphate (GBP) is often added to activate PGM. The compound provides a phosphate group to the active center and acts as the starter phosphate group in the phosphate relay reaction. Omitting GBP from our enzyme system (Table 1, system B) did not affect myo-inositol production rates (Fig. 5A). The T. kodakarensis PGM may not require a starter phosphate group, or the enzymes recovered from E. coli may already be phosphorylated. When we varied the concentration of NAD⁺ that is utilized in the MIPS reaction (Table 1, system C), we found that the rates of myo-inositol production were comparable within a concentration range of 0.5 to 10 mM (Fig. 5B). In both experiments, we also measured the concentrations of glucose formed during the reaction. The concentrations were much lower than those of myo-inositol, with a maximum of approximately 2.8 mM. Intriguingly, although it was at a lower rate, myo-inositol production was observed in the absence of NAD⁺ addition (Fig. 5B), and we presumed that this was due to the NAD⁺ present in the enzyme samples. We measured the NAD⁺ and NADH in a reaction mixture without exogenous NAD⁺ and detected 0.013 mM NAD⁺ and 0.02 mM NADH.

FIG 5.

(A) Effects of the enzyme sample and the presence or absence of GBP in the reaction system. The composition of the reaction system was that shown in Table 1 for system B. Blue circles and blue triangles display myo-inositol and glucose, respectively, in the presence of GBP. Red circles and red triangles display myo-inositol and glucose, respectively, without GBP. Open circles and open triangles display myo-inositol and glucose, respectively, using purified enzymes. (B) Effects of NAD⁺ concentration in the reaction system. The composition of the reaction system was that shown in Table 1 for system C. myo-Inositol (circles) and glucose (triangles) were measured. NAD⁺ was added to a final concentration of 0 mM (white), 0.5 mM (black), 1 mM (red), 2 mM (gray), 5 mM (green), or 10 mM (blue). The symbols represent the averages from three independent reactions. The averages ± standard deviations of myo-inositol concentration at 2 h are 37.0 ± 1.1 mM (0 mM NAD⁺), 82.4 ± 7.3 mM (0.5 mM NAD⁺), 78.5 ± 0.9 mM (1 mM NAD⁺), 78.2 ± 2.7 mM (2 mM NAD⁺), 82.0 ± 3.3 mM (5 mM NAD⁺), and 75.6 ± 4.0 mM (10 mM NAD⁺). Standard deviations of glucose concentrations at 2 h were at maximum ±1.20 mM.

Taking into account the results described above, we further examined our enzyme system using the reaction composition shown in Table 1 for system D. We first examined the conversion rate and yield. With 5 mM amylose, we observed a linear generation of myo-inositol until approximately 2 h (Fig. 6). The concentration of the product at 2 h was approximately 74 mM, which corresponds to 93% conversion of the glucose units in the substrate amylose. When we increased the initial concentration of amylose to 10 mM, we observed an increased rate of myo-inositol production, but production rates decreased after 2 h of conversion. We presumed that this was due either to the thermal degradation of NAD⁺ or to the thermal inactivation of one of the enzymes. We thus added NAD⁺ corresponding to a final concentration of 1 mM at 1-h intervals. As a result, we found that supplementing NAD⁺ to the reaction mixture resulted in the production rates being maintained for longer periods of time. The concentration of the product at 4 h was approximately 145 mM, which corresponds to 91% conversion of the glucose units in the substrate amylose. The formation of glucose and maltose was observed during the reaction, with concentrations of approximately 5 mM each.

FIG 6.

Effects of supplementing NAD⁺ during the reaction. myo-Inositol (circles), glucose (triangles), and maltose (squares) were measured. Open symbols represent product concentrations in reaction systems without supplemented NAD⁺. Closed symbols represent product concentrations in reaction systems supplemented with NAD⁺ every hour, corresponding to a 1 mM increase in final concentration. The composition of the reaction system was as shown in Table 1 for system D. Amylose was added to a final concentration of 5 mM (black) or 10 mM (red). The symbols represent the averages from three independent reactions. The averages ± standard deviations of myo-inositol concentration at 6 h are 152 ± 11 mM (10 mM amylose, with NAD⁺), 122 ± 9 mM (10 mM amylose, without NAD⁺), 79.5 ± 3.2 mM (5 mM amylose, with NAD⁺), and 77.5 ± 3.3 mM (5 mM amylose, without NAD⁺). The standard deviations of glucose and maltose concentrations at 6 h were at maximum ±0.94 mM and ± 1.74 mM, respectively.

Increasing the level of myo-inositol production with the four-enzyme system.

We next examined the possibilities for increasing the amount of myo-inositol produced by the four-enzyme system. We added amylose so that the final concentrations would increase by 5 mM or 10 mM every hour with or without NAD⁺ (final concentration increase, 1 mM). The results are shown in Fig. 7A. When NAD⁺ was not added, we observed a linear increase in myo-inositol until 2 h, and the production rate decreased thereafter. When NAD⁺ was added together with amylose, production rates were maintained for longer periods of time, and the myo-inositol concentration reached 200 mM in about 4 h. As the productivities with 5 mM and 10 mM amylose were similar, the amylose concentration seems to be saturated at 5 mM. As the addition of NAD⁺ maintained the production rates, it was the thermal degradation of NAD⁺, and not the thermal inactivation of enzyme, that caused the decrease in production rates after 2 h. As the production rates did not display a large decrease for 6 h, we can presume that the enzymes maintained a substantial portion of their activity during this period.

FIG 7.

(A) Effects of supplementing NAD⁺ and amylose during the reaction. myo-Inositol (circles), glucose (triangles), and maltose (squares) were measured. Formation of maltose could not be detected and thus is not shown. Red closed symbols, 10 mM amylose and 1 mM NAD⁺ added every hour; black closed symbols, 5 mM amylose and 1 mM NAD⁺ added every hour; red open symbols, 10 mM amylose and no NAD⁺ added every hour; black open symbols, 5 mM amylose and no NAD⁺ added every hour. The composition of the reaction system was as shown in Table 1 for system D. The symbols represent the averages from three independent reactions. The averages ± standard deviations of myo-inositol concentration at 6 h are 268 ± 4 mM (10 mM amylose, with NAD⁺), 218 ± 12 mM (5 mM amylose, with NAD⁺), 142 ± 8 mM (10 mM amylose, without NAD⁺), and 111 ± 19 mM (5 mM amylose, without NAD⁺). The standard deviations of glucose at 6 h were at maximum ±2.25 mM. (B) myo-Inositol production in a 200-ml reaction system. The composition of the reaction system was as shown in Table 1 for system E. myo-Inositol (circles), glucose (triangles), and maltose (squares) were measured. The reaction was started with 5 mM amylose, and NAD⁺ (1 mM) was supplemented every 2 h. The standard deviations of glucose and maltose at 7 h were at maximum ±0.31 and ±0.28 mM, respectively.

We next increased the scale of the system to 200 ml from 2 ml. The concentration of each component was maintained (Table 1, system E), and NAD⁺ (final concentration 1 mM) was added every 2 h. After 6 h of the reaction, the myo-inositol concentration reached 81 mM (Fig. 7B). This corresponded to the production of 2.9 g of myo-inositol from 2.9 g of amylose, achieving gram-scale production with a yield of approximately 96%.

Production of myo-inositol from soluble starch or raw potato.

As high production rates and yields were achieved with amylose, we next explored the possibilities of producing myo-inositol from soluble starch or raw potatoes. These substrates contain α-1,6-linkages in addition to the α-1,4-linkages found in amylose chains. As MalP does not recognize α-1,6-linkages, it would thus provide an advantage to add an enzyme that could cleave the α-1,6-linkages and release the individual chains linked with α-1,4-glycosidic bonds. We selected the thermostable pullulanase from T. maritima encoded by TM1845. The enzyme displays hydrolytic activity only toward α-1,6-linkages and not toward α-1,4-linkages (47). The closely related enzyme (92% identical) from Thermotoga neopolitana also exhibits similar properties (48). The TM1845 gene was cloned and expressed in E. coli, the cell extracts with the recombinant protein were heat treated at 90°C for 10 min, and the supernatant after centrifugation was used as the enzyme sample.

With soluble starch, we used a reaction mixture with the composition indicated in Table 2 for system F. The amount of soluble starch in the reaction is equivalent to that of 5 mM amylose in terms of glucose units. The reaction was carried out for 6 h at 90°C. After 4 h, we observed the production of approximately 58 mM myo-inositol, which corresponds to a yield of 73% from soluble starch (Fig. 8A). The addition of pullulanase had a large effect, as the same reaction without the enzyme led to the production of only 40 mM myo-inositol, corresponding to a yield of 50% from soluble starch.

TABLE 2.

Composition of the in vitro enzyme reaction systems with soluble starch or raw potato as the substrate

Component (gene) or parameter	Concn (mM) or amt in system:
	F	G
Tris-HCl (pH 7.5)	100	100
MgCl2	10	10
Disodium hydrogen phosphate	50	50
NAD+	1a	1b
Soluble starch	29 mg
Raw potato		300 mg (wet wt)
MalP (TK1406)	100 μg (1.0 nmol)	100 μg
PGM (TK1108)	400 μg (8.0 nmol)	400 μg
MIPS (AF1794)	500 μg (11.5 nmol)	500 μg
IMPase (TM1415)	800 μg (28 nmol)	800 μg
Pullulanase (TM1845)	200 μg (2.1 nmol)	200 μg
Total vol	2 ml	2 ml
Figure	8A	8B

^aInitial concentration; see Fig. 8A legend for details.

^bInitial concentration; see Fig. 8B legend for details.

FIG 8.

(A) myo-Inositol production from soluble starch. The composition of the reaction system was as shown in Table 2 for system F. myo-Inositol (circles), glucose (triangles), and maltose (squares) were measured. NAD⁺ (1 mM) was supplemented every 2 h. (B) myo-Inositol production from raw potato. Two types of potato were used, Danshaku potato (D) and May Queen potato (M). Yields were calculated on the presumption that 87% of the dry potato mass was composed of starch. The composition of the reaction system was as shown in Table 2 for system G. myo-Inositol (black), glucose (white), and maltose (gray) were measured after a 6-h reaction. NAD⁺ (1 mM) was supplemented every 2 h.

We next attempted to produce myo-inositol from raw Danshaku or May Queen potatoes. The amount of potato added to the reaction mixture corresponded to a dry weight of 100 mg. We used a reaction mixture with the composition indicated in Table 2 for system G. After a 6-h reaction, myo-inositol accumulated to a concentration of 139 mM (Danshaku) or 146 mM (May Queen), which corresponds to 50 mg or 53 mg of product, respectively. Taking into account that the starch content in potatoes is approximately 87%, the conversion rate from raw potato to myo-inositol was approximately 57% for Danshaku potatoes (32% without pullulanase) and 61% for May Queen potatoes (54% without pullulanase) (Fig. 8B).

DISCUSSION

We have established an in vitro enzyme system that efficiently converts starch compounds such as amylose, soluble starch, and raw potato to myo-inositol. The system can produce myo-inositol on the gram scale within 3 h. By using the MalP and PGM reactions, we were able to introduce the phosphorylated glucose while avoiding the consumption of ATP. We had expected that the overall reaction was energetically favorable, as the conversion of the glucopyranose structure to the C₆ cyclitol structure and the inositol-phosphate hydrolysis were expected to be exergonic. The phosphorylase reaction and the phosphomutase reactions are well documented (49, 50), and their Gibbs free energy differences are +3.19 kJ/mol and −7.34 to −7.09 kJ/mol, respectively. Thermodynamic modeling and calculation tools have also been published (51, 52) and estimate that the MIPS reaction and the IMPase reaction display free energy differences of −50.3 kJ/mol and −23.4 to −20.7 kJ/mol. The sum of these values indicates that the conversion from starch to myo-inositol is highly favorable.

There are still a number of points in our system that can or need to be improved to obtain even higher levels of efficiency. Concerning enzyme preparation, we used enzyme samples that were partially purified by heat treatment. This was to avoid, as much as possible, laborious and costly steps in preparation that would eventually hamper large-scale or industrial use of our system. Indeed, the conversion rates and yield using partially purified enzyme samples were indistinguishable from those obtained with equivalent amounts of purified proteins (Fig. 5A). One way to simplify enzyme preparation even further would be to express all genes in a single host cell or to at least mix E. coli cells that express individual genes prior to sonication. Another point is to enable the reuse of the enzymes, and this should be possible with enzyme immobilization techniques. Immobilization may also lead to further enhancing the stability of the enzymes. A further point to address is the thermal degradation of NAD⁺. Although intrinsic stabilization of the cofactor may be difficult, synthetic pathways to regenerate NAD⁺ have already been developed (53), and integrating these systems should alleviate the amount of cofactor needed to sustain myo-inositol synthesis.

Although levels were low, we observed the production of glucose, or in some cases maltose, in our enzyme system. A major portion of these sugars most likely derives from the substrate amylose. The MalP reaction is a phosphorolysis reaction of the α-1,4-linkage in amylose chains, and cleavage proceeds from the nonreducing ends of the chains, releasing G1P. The final phosphorolysis reaction acts on maltose and results in the formation of G1P and glucose. This glucose, which corresponds to the terminal glucose at the reducing ends of the amylose chains, in principle should remain in the reaction system. As the amylose chains used in this study were composed of approximately 18 glucose units, the majority, if not all, of the glucose and maltose produced in our reactions most likely derives from the reducing ends of the substrates.

In this study, we selected myo-inositol as the end product of our system. However, there are a number of other candidate compounds that can be produced simply by changing the metabolism downstream of G6P. For example, an isomerization reaction to fructose 6-phosphate (F6P) would enable us to target mannose production. This would be a cofactor-free pathway, and activity may be maintained for even longer periods of time.

This study has demonstrated that the MalP/PGM pathway can function in vitro for the production of G6P. In our systems, phosphate concentrations were set at 50 mM, and the intracellular concentration of phosphate in T. kodakarensis remains to be determined. It would be of interest to determine whether this pathway actually functions in vivo in T. kodakarensis and, if so, its contribution relative to the conventional hydrolysis and phosphorylation pathway catalyzed by amylases and ADP-dependent GK.

MATERIALS AND METHODS

Strains and growth conditions.

E. coli strains DH5α and BL21-CodonPlus(DE3)-RIL were cultivated at 37°C in lysogeny broth (LB) medium containing 100 mg liter⁻¹ ampicillin. Unless mentioned otherwise, all chemicals were purchased from Wako Pure Chemicals (Osaka, Japan) or Nacalai Tesque (Kyoto, Japan).

Gene expression in E. coli and preparation of enzyme samples.

The gene encoding MalP was amplified by PCR from the chromosomal DNA of T. kodakarensis with the primer set TK1406F/TK1406R (5′-AAAGGATCCCATATGGTGAACGTTTCCAATGCCGTTGAGGAT-3′/5′-AAAGAATTCTCAGTCAAGTCCCTTCCACTTGACCAGA-3′). The gene encoding MIPS was amplified by PCR from the chromosomal DNA of Archaeoglobus fulgidus with the primer set AF1794F/AF1794R (5′-GGAGGTGATGCCATATGAAGGTCTG-3′/5′-AAAGAATTCTTATTTCAGATTTGAATACC-3′). The gene encoding IMPase was amplified by PCR from the chromosomal DNA of Thermotoga maritima with the primer set TM1415F/TM1415R (5′-AAACATATGGACAGACTGGACTTTTCTAT-3′/5′-AAAGAATTCTCACTTTCCTCCTATTTCTTCTACC-3′). The gene encoding pullulanase was amplified by PCR from the chromosomal DNA of T. maritima with the primer set TM1845F/TM1845R (5′-AAACATATGAAGACTAAACTCTGGTT-3′/5′-AAAGAATTCTCACTCTCTGTACAGAACGT-3′). Using NdeI and EcoRI, whose recognition sites are underlined, the amplified fragments and pET21a(+) expression vector (Merck KGaA, Darmstadt, Germany) were digested and ligated. The region encoding the secretion signal of pullulanase was removed by inverse PCR with the primer set TM1845invF/TM1845invR (5′-ATGTCAGAAACCACCATCGTAGTCCACTAT-3′/5′-ATGTATATCTCCTTCTTAAAGTTAA-3′). After confirming the absence of unintended mutations, the constructed plasmids were individually introduced into E. coli BL21-CodonPlus(DE3)-RIL. The plasmid for expression of PGM was previously constructed elsewhere (43). Procedures for gene expression were the same for all proteins. Transformants were inoculated into LB medium containing ampicillin and cultivated at 37°C until the optical densities at 660 nm reached above 0.4. Isopropyl β-d-1-thiogalactopyranoside was added to a final concentration of 0.1 mM to induce expression, and cells were cultivated for a further 4 h. Cells were harvested by centrifugation (4°C, 5,000 × g, 15 min) and suspended in 50 mM Tris-HCl (pH 7.5) with 150 mM NaCl. After centrifugation (4°C, 5,000 × g, 15 min), cells were suspended in 50 mM Tris-HCl buffer (pH 7.5), disrupted by sonication, and centrifuged again (4°C, 5,000 × g, 15 min). The soluble cell extract was heated at 90°C for 10 min and then centrifuged (4°C, 5,000 × g, 15 min). The supernatant was used as the enzyme sample for myo-inositol production. Samples were also applied to SDS-PAGE and stained with Coomassie brilliant blue. Gels were scanned with Gelscan-2 (iMeasure, Matsumoto, Japan) to obtain high-resolution images of the gels. The percentage of recombinant protein in the supernatant was calculated by quantifying the relative band intensities using ImageQuant TL (GE Healthcare Biosciences, Chicago, IL). The amounts of NAD⁺/NADH in the enzyme samples were measured using the NAD/NADH quantitation kit (Sigma-Aldrich, St. Louis, MO).

Purification of recombinant MalP, PGM, MIPS, and IMPase.

The supernatant prepared as described in the previous section containing each protein was applied to an anion-exchange column (Resource Q; GE Healthcare Biosciences) equilibrated with 50 mM Tris-HCl (pH 7.5). Proteins were eluted with a linear gradient of 0 to 1.0 M NaCl, and relevant fractions were collected and concentrated with an Amicon Ultra centrifugal filter unit (molecular weight cutoff [MWCO], 3,000; EMD Millipore, Billerica, MA). The resulting protein solution was applied to a gel filtration column (Superdex 200; GE Healthcare Biosciences) equilibrated with 50 mM Tris-HCl (pH 7.5) containing 0.15 M NaCl.

Conversion of amylose to G6P with MalP and PGM.

The conversion of amylose to G6P with MalP and PGM was measured by a discontinuous assay coupled with the G6P dehydrogenase reaction. The initial reaction mixture (100 μl) consisted of 100 mM Tris-HCl (pH 7.5), 10 mM disodium hydrogen phosphate, 10 mM MgCl₂, 50 μM glucose 1,6-bisphosphate (GBP) (Sigma-Aldrich), 5 mM amylose, and 1.0 μg each of heat-treated MalP and PGM. The reaction was carried out at 70°C for 10 min. Proteins were removed by ultrafiltration using an Amicon Ultra centrifugal filter unit (MWCO, 10,000) (EMD Millipore), and a 10-μl aliquot was added to the G6P dehydrogenase reaction mixture. The G6P dehydrogenase reaction mixture (200 μl) consisted of 100 mM Tris-HCl (pH 7.5), 0.5 mM NADP⁺ (Oriental Yeast, Tokyo, Japan), 2 U of G6P dehydrogenase (Sigma-Aldrich), and the 10-μl aliquot of the MalP/PGM reaction. The reaction was carried out at room temperature, and the amount of NADPH generated was measured at 340 nm.

Conversion of G6P to myo-inositol with MIPS and IMPase.

Conversion of G6P to myo-inositol with MIPS and IMPase was measured by monitoring the generation of myo-inositol using high-performance liquid chromatography (HPLC) (Shimadzu, Kyoto, Japan). The reaction mixture (100 μl) consisted of 100 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 10 mM NAD⁺ (Oriental Yeast), 100 mM G6P (Sigma-Aldrich), 10.8 μg of heat-treated MIPS, and 99.6 μg of heat-treated IMPase. The reaction was carried out at 90°C for various periods of time. Proteins were removed by ultrafiltration as described above, and a 10-μl aliquot was applied to an Asahipak NH2P-50 4E column (Showa Denko, Kanagawa, Japan). Compounds were separated with 70% (vol/vol) acetonitrile at a flow rate of 1.0 ml min⁻¹ at 40°C. A refractive index (RI) detector was used for measurement.

Production of myo-inositol from amylose, soluble starch, or raw potato.

The in vitro production of myo-inositol from amylose with MalP, PGM, MIPS, and IMPase was measured by the same method used for MIPS and IMPase activity measurement. The reaction mixture (2 ml) consisted of 100 mM Tris-HCl (pH 7.5), 10 mM MgCl₂, 50 mM disodium hydrogen phosphate, 50 μM GBP, 1 mM NAD⁺, 100 μg of heat-treated MalP, 400 μg of heat-treated PGM, 500 μg of heat-treated MIPS, and 800 μg of heat-treated IMPase samples (Table 1, system A). In another experiment, we replaced the heat-treated enzyme samples with the same amounts of purified enzyme. The reactions were carried out at 90°C. Samples (100 μl) were taken at various periods of time and put on ice. Proteins were removed by ultrafiltration, and a 10-μl aliquot was used for HPLC analysis. Reaction conditions were varied depending on the experiment, and the reaction compositions are indicated in Table 1 (systems A to E). Experiments using soluble starch instead of amylose were also performed (Table 2, system F). When using raw potato (Danshaku potato or May Queen potato) as the substrate, potatoes were crushed with a mill and water was removed by centrifugation. A portion of the pellet (300 mg wet weight, 100 mg dry weight) was used for the production of myo-inositol with the reaction composition indicated in Table 2 for system G.

NMR analyses of the product.

The product of the reaction was purified and collected by HPLC (FRC-10A, Shimadzu). The NMR sample was prepared by mixing the dried product (approximately 1.4 mg) with D₂O (Cambridge Isotope Laboratories, Andover, MA). The NMR measurement was carried out with ECA600P (JEOL, Tokyo, Japan) at 600 MHz and room temperature. The chemical shifts of the ¹H NMR spectra are given in ppm relative to the signals of solvent using external standards of D₂O at 4.6 ppm.

ADDENDUM

During the review process, You et al. published a paper on this subject (54). Although some source organisms for the enzymes differed, they accomplished large-scale myo-inositol production in 20,000-liter reactors. A stark difference from the present study is that the reaction system used by You et al. does not require the addition of exogenous NAD⁺. This may be due to the lower reaction temperature (70°C) used in that study.