A Novel Metabolic Pathway for Glucose Production Mediated by α-Glucosidase-catalyzed Conversion of 1,5-Anhydrofructose*

Background: α-Glucosidase-catalyzed hydration on 1,5-anhydrofructose has not been clarified. Results: α-Glucosidases of glycoside hydrolase families 13 and 31 synthesize α-configurated product from 1,5-anhydrofructose. Combined reaction of glucan lyase and α-glucosidase enhances glucose production from starch. Conclusion: α-Glucosidases catalyze trans-addition on 1,5-anhydrofructose. Significance: Novel α-glucosidase-associated metabolic pathway of glucose formation from 1,5-anhydrofructose suggests a salvage process of unutilized 1,5-anhydrofructose to glucose in nature. α-Glucosidase is in the glycoside hydrolase family 13 (13AG) and 31 (31AG). Only 31AGs can hydrate the d-glucal double bond to form α-2-deoxyglucose. Because 1,5-anhydrofructose (AF), having a 2-OH group, mimics the oxocarbenium ion transition state, AF may be a substrate for α-glucosidases. α-Glucosidase-catalyzed hydration produced α-glucose from AF, which plateaued with time. Combined reaction with α-1,4-glucan lyase and 13AG eliminated the plateau. Aspergillus niger α-glucosidase (31AG), which is stable in organic solvent, produced ethyl α-glucoside from AF in 80% ethanol. The findings indicate that α-glucosidases catalyze trans-addition. This is the first report of α-glucosidase-associated glucose formation from AF, possibly contributing to the salvage pathway of unutilized AF.

In this study, we investigated the reaction of 13AG and 31AG on AF and report for the first time that ␣-glucosidases catalyze hydration and 1-alkoxy-2-hydro addition. A combined reaction on starch with GL and 13AG enhances glucose formation, suggesting the existence of a novel metabolic pathway for glucose production.
Hydration of AF by ␣-Glucosidases-␣-Glucosidase (DG, 0.46 g/ml; HBG2, 0.40 g/ml; ANG, 0.67 g/ml; and BWG, 0.47 g/ml) was incubated with 30 mM AF in 20 mM sodium acetate buffer (pH 5.0) for 40 min at 35°C (for DG only, ␣-glucosidase was incubated in 20 mM sodium phosphate buffer (pH 6.0) at 25°C). The reaction product was analyzed by TLC with the method of Chiba and Shimomura (23) and by high-performance anion-exchange liquid chromatography (HPAEC; Dionex, Sunnyvale, CA) with CarboPac PA1 (4 ϫ 250 mm, Dionex) and a pulsed amperometry detector (Dionex). Further analysis was done by treating the reaction product with A. niger glucose oxidase (Sigma-Aldrich). The reaction mixture containing 40 mM AF, 16 mM sodium acetate buffer (pH 4.0), and 2.84 g/ml ANG was incubated at 35°C followed by the addition of glucose oxidase (1,200 units/ml) with 0.1 M Tris-HCl buffer (pH 7.0) at 35°C for 1 h. The reaction product was analyzed with TLC as described. Assay of glucose from AF was done using the Tris-glucose oxidase-peroxidase method (Wako, Osaka, Japan) (24,25).
Combined Reaction on Starch with GL and DG-Both DG and GL (0.46 and 0.22 g/ml, respectively) were incubated with 2% soluble starch in 5 mM sodium phosphate buffer (pH 6.0) at 25°C. The reaction mixture was acquired at the indicated time, and the product was analyzed with the methods described below. Anomeric configuration of product was determined by HPLC (JASCO, Tokyo, Japan) with TSKgel Amide-80 (4.6 ϫ 250 mm; Tosoh, Tokyo, Japan) and evaporative light-scattering detector (EYELA, Tokyo, Japan).
ANG Reaction on AF in Ethanol-In 80% ethanol, ANG (0.4 g/ml) was incubated with 50 mM AF in 20 mM sodium acetate buffer (pH 4.3) at 35°C for 2 days, followed by boiling for 10 min. A TLC-purified 1-alkoxy-2-hydro addition product (ethyl ␣-D-glucopyranoside) was analyzed using field desorption mass spectrometry (JEOL JMS-SX102A) and NMR (Bruker AMX-500), chemical shifts (␦) of which were expressed in ppm relative to the external standard of [ Inhibition with GlcAL-Inhibition of GlcAL for hydrolytic reaction on pNPG was analyzed using HBG2. HBG2 (1.1 g/ml) was incubated with 100 mM GlcAL and 2.5 mM pNPG in 40 mM sodium acetate buffer (pH 5.0) at 37°C for HBG2 followed by the addition of 1 M Na 2 CO 3 . Liberated p-nitrophenol was calculated using ⑀ 1 mM, 400 nm ϭ 5.56.

RESULTS AND DISCUSSION
Hirano et al. (26) reported that ␣-glucosidase could not hydrate AF, allowing us to analyze the hydrations of AF using 13AG (HBG1) and 31AG (ANG and BWG). As shown in Fig.  2A, only glucose-like monosaccharide was detected as a product in the reactions on AF with all ␣-glucosidases examined. The product of 13AG or 31AG was also analyzed by HPAEC, which verified glucose formation (Fig. 2B). The product was sensitive to glucose oxidase treatment, as evident by the absence of a glucose spot upon TLC analysis. No product was formed by CBE-inactivated ANG, which was prepared by incubation of 2.84 g/ml ANG with 0.1 M CBE in 20 mM sodium acetate buffer (pH 4.5) at 35°C for 2 h (27). These results indicate that 13AG and 31AG catalyze hydration of AF to produce glucose. Because Hirano et al. (26) used N-glycan-trimmingtype ␣-1,3-glucosidase (a 31AG belonging to EC 3.2.1.130; glycoprotein glucosylmannohydrolase) for AF hydration, it is possible that this enzyme has a different substrate specificity from 13AG and 31AG.
The time course exhibited a plateau (Fig. 2C) at DG-catalyzed glucose formation. AF formed keto-enol tautomers and hydrated compound under the aqueous condition (Fig. 1C), indicating that 1,2-enol AF was a possible substrate for ␣-glucosidases. To prove this possibility, we performed a combined reaction on soluble starch with GL and DG (Fig. 1D). DG can act on AF but not a soluble starch, whereas the other ␣-glucosidases used will be able to release glucose from soluble starch (3,4). Therefore, the formation of glucose will provide the evidence that the GL released AF from soluble starch. Generated AF can be acted on by the DG present. As shown in Fig. 2D, our combined reaction eliminates the plateau during glucose production with a Ͼ40-fold enhancement of hydration as com-  -catalyzed reactions (A, B, E, and F), conversion of AF under aqueous condition (C), combined reaction of GL and DG (D), and structure of ethyl ␣-glucoside (G). A, 31AG-catalyzed hydrolytic reaction on p-nitrophenyl 2-deoxyglucoside to form ␣-2dGlc and p-nitrophenol (pNP) (16). B, 31AG-catalyzed trans-addition to GlcAL (17,18). In water (RϭH), 31AG produces ␣-2dGlc with newly formed axial proton (shown by italic boldfaced H). Alkyl alcohol (Rϭalkyl) is also used to form alkyl ␣-2-deoxyglucoside (28). C, tautomerization of AF (19 -21). D, GL produces 1,2-enol AF from starch followed by conversion to ␣-glucose with DG. E, ␣-glucosidase-catalyzed trans-addition on 1,2-enol AF. F, ANG-catalyzed 1-alcohoxy-2-hydro addition to 1,2-enol AF; G, the TLC-isolated product from an ANG-mediated reaction on AF and 80% ethanol was determined to be an ethyl ␣-glucoside by MS and NMR analyses. pared with the glucose formation at 5 min of Fig. 2C. HPLC analysis of resulting glucose revealed its ␣-anomeric configuration (Fig. 2E), indicative of trans-addition by DG belonging to 13AG (Fig. 1E). It is considered that the plateau (Fig. 2C) was formed by restricted supply of 1,2-enol AF from hydrated AF (Fig. 1C). These findings indicate that ␣-glucosidase hydrates 1,2-enol AF that has been newly produced by GL from soluble starch to form the ␣-glucose, also suggesting the presence of a novel metabolic pathway (Fig. 1D) in which the combined reaction with GL and DG generates glucose from starch in all organisms having GL.
We tried to apply the same combined reaction to 31AG, but all of our 31AGs exhibited hydrolytic activity on soluble starch. Therefore, we performed a 1-alkoxy-2-hydro addition to AF using highly concentrated ethanol-stable ANG (28) (Fig. 1F). The TLC-isolated product from an ANG-mediated reaction on AF and 80% ethanol was determined to be an ethyl ␣-glucoside by MS and NMR analyses (Fig. 1G). Formation of ethyl ␣-glu-coside indicated that 31AG also catalyzed the trans-addition on AF (Fig. 1E).
It is interesting that GlcAL, having no 2-OH group of AF (Fig.  1B), could not inhibit the HBG2-catalyzed hydrolytic reaction on pNPG, although we used 100 mM GlcAL for this experiment. HBG2 is a 13AG, and 13AGs do not have any hydration ability at GlcAL. Therefore, GlcAL does not enable incorporation to the catalytic site of any 13AG. Frandsen et al. (29) also suggested that 2-OH is not involved in stabilizing the transition state by GH 13 enzymes, one of which is 13AG.
There are two AF formation systems: (i) GL-associated starch degradation and (ii) pyranose 2-oxidase-catalyzing oxidation at the 2-OH group of glucose, forming a 2-keto AF (Fig.  1C) (30), followed by tautomerization. Pyranose 2-oxidases, which are found in many fungi, are extracellular (or periplasmic) flavoproteins (30 -32), indicating a presence of AF out of cells. In nature, therefore, extracellular ␣-glucosidases (e.g. 13AGs secreted from bacteria) may convert AF to glucose, which is immediately incorporated into other cells (e.g. 13AGsecreting bacteria cells). AF, synthesized by extracellular GL, is also altered to glucose by the same mechanism. Recently, Yu et al. (33) found that fungus A. niger was able to use AF as the sole carbon source for growth, whereas Bakers' yeast could not grow in the presence of 0.47 M AF in medium. This observation is of interest because A. niger secretes large amounts of 31AG into medium (9), but ␣-glucosidase (13AG) of yeast stays inside of cells (34,35). In nature, microorganisms receive glucose from AF by secreting ␣-glucosidase, implying the presence of a salvage pathway of unutilized AF.

CONCLUSION
The data indicate that (i) ␣-glucosidases catalyze a transaddition on AF; (ii) GlcAL without the 2-OH group of AF cannot bind to the catalytic site of 13AG, although 31AG can; (iii) alkyl ␣-glucosides are synthesized by ␣-glucosidase-associated 1-alkoxy-2-hydro addition to AF; and (iv) the combined reaction of GL and ␣-glucosidase produces glucose from soluble starch. Our study also suggests a novel metabolic pathway of ␣-glucosidase-associated conversion of AF to glucose in nature, salvaging unutilized AF.