Cytosolic Pyridoxine-β-d-Glucoside Hydrolase from Porcine Jejunal Mucosa

During studies of the nutritional utilization of pyridoxine 5′-β-d-glucoside, a major form of vitamin B6 in plants, we detected two cytosolic β-glucosidases in jejunal mucosa. As expected, one was broad specificity β-glucosidase that hydrolyzed aryl β-d-glycosides but not pyridoxine β-d-glucoside. We also found a previously unknown enzyme, designated pyridoxine-β-d-glucoside hydrolase, that efficiently hydrolyzed pyridoxine β-d-glucoside. These were separated and purified as follows: broad specificity β-glucosidase 1460-fold and pyridoxine-β-d-glucoside hydrolase 36,500-fold. Purified pyridoxine-β-d-glucoside hydrolase did not hydrolyze any of the aryl glycosides tested but did hydrolyze cellobiose and lactose. Pyridoxine-β-d-glucoside hydrolase exhibited a pH optimum of 5.5 and apparent molecular mass of 130 kDa by SDS-polyacrylamide gel electrophoresis and 160 kDa by nondenaturing gel filtration, in contrast to 60 kDa for native and denatured broad specificity β-glucosidase. Glucono-δ-lactone was a strong inhibitor of both enzymes. Ionic and nonionic detergents were inhibitory for each enzyme. Conduritol B epoxide, a potent inhibitor of lysosomal acid β-glucosidase, inhibited pyridoxine-β-d-glucoside hydrolase but not broad specificity β-glucosidase, but both were inhibited by the mechanism-based inhibitor 2-deoxy-2-fluoro-β-d-glucosyl fluoride. Our findings indicate major differences between these two cytosolic β-glucosidases. Studies addressing the role of vitamin B6 nutrition in regulating the activity and its consequences regarding pyridoxine glucoside bioavailability are in progress.

A naturally occurring glycosylated derivative of vitamin B6, pyridoxine 5Ј-␤-D-glucoside (PNG), 1 was first isolated from rice bran (1). PNG is now known to exist in most fruits, vegetables, and cereal grains, in which it comprises from 5-75% of total vitamin B6 (2,3). Because of the prevalence of PNG in plantderived foods, its bioavailability as a source of available vitamin B6 is a matter of nutritional concern. The bioavailability of this glycosylated form of vitamin B6, relative to pyridoxine (PN), is ϳ25% in rats (4,5) and ϳ50% for humans (6,7), as estimated from urinary excretion of 4-pyridoxic acid after oral administration of isotopically labeled PNG. The rate-limiting phase of PNG utilization in vitamin B6 metabolism is the hydrolysis of the ␤-glycosidic bond in both rats and humans. These findings indicate that the hydrolysis of PNG is a major factor governing its bioavailability.
Glucosidases exist widely in nature. The primary ␤-glucosidases in mammalian tissues consist of a lysosomal membranebound acid ␤-glucosidase (EC 3.2.1.45; N-acylsphingosyl-␤-Dglucopyranoside glucohydrolase) which is responsible for the hydrolytic cleavage of glucosphingolipids (for review see Ref. 8), and a cytosolic ␤-glucosidase capable of hydrolyzing a variety of aryl ␤-D-glycosides (for review see Ref. 9). This cytosolic enzyme has been detected in a variety of mammalian tissues and has been designated broad specificity ␤-glucosidase. Although broad specificity ␤-glucosidase has been purified from several organs, cloned, sequenced, and intensively studied, the physiological function of this enzyme remains unclear. It has been reported that cytosolic broad specificity ␤-glucosidase from liver is capable of hydrolyzing certain cyanogenic plant glucosides such as L-picein (10,11). Enzymatic activity capable of hydrolyzing PNG has been found in nonpurified supernatant fractions of small intestinal mucosa and attributed initially to broad specificity ␤-glucosidase (12). PNG hydrolyzing activity, presumably of microbial origin, also has been detected in small intestinal contents of rodents and may contribute to the in vivo hydrolysis of dietary PNG (13,14), although the role of microbial ␤-glucosidases in PNG hydrolysis in the human small intestine probably would be less significant.
Hydrolysis, rather than intestinal absorption, is a rate-limiting factor for the utilization of PNG (4), and in vivo studies in rats and humans have indicated that much of the hydrolysis of PNG is associated with the intestine (6,15). To obtain a better understanding of the biochemistry of PNG hydrolysis, we initiated purification of broad specificity ␤-glucosidase from jejunal mucosa of the pig, a monogastric animal physiologically similar to the human. After purification of broad specificity ␤-glucosidase from porcine intestinal mucosa using procedures of Glew and associates (9) and DePetro (16) with p-nitrophenyl-␤-D-glucoside as a routine substrate, we observed that the purified enzyme does not catalyze the hydrolysis of PNG nor does PNG inhibit the hydrolysis of this aryl ␤-D-glucoside. This observation led us to undertake the isolation of the enzyme that is responsible for the hydrolysis of PNG. Using PNG as substrate, we have succeeded in purifying a previously unknown enzyme that we designate pyridoxine-␤-D-glucoside hydrolase from porcine jejunal mucosal cytosol.
In this paper we report (a) the purification of these two cytosolic ␤-glucosidases from porcine intestinal mucosa, (b) initial characterization of pyridoxine-␤-D-glucoside hydrolase, and (c) comparison of the catalytic properties of these enzymes.
Concurrent with the discovery of the distinct pyridoxine-␤-Dglucoside hydrolase in porcine intestine, this laboratory has reported that vitamin B6 nutritional status regulates the activity of mucosal cytosolic ␤-glucosidases (13,14). Vitamin B6 deficiency causes elevation in mucosal activities of both broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase. A long term objective is to determine the mechanism and metabolic consequences of the nutritional regulation of these enzymes.
Purification of Pyridoxine-␤-D-glucoside Hydrolase-During this purification, all materials were kept on ice or 4°C. Pig jejunum was obtained immediately following slaughter at the University of Florida Animal Science Department. Sections were cut longitudinally and washed with 0.9% (w/v) NaCl and then stored at Ϫ80°C until use. Frozen intestine strips (approximately 200 g) were allowed to thaw on ice overnight, and mucosa was scraped off using a glass slide, yielding approximately 60 g of mucosa. The mucosa was homogenized in 180 ml of homogenization buffer comprised of 10 mM sodium phosphate, pH 7, containing 10 mM 2-mercaptoethanol and 1 mM phenylmethylsulfonyl fluoride using a Polytron PT 10 -35 device (Brinkmann Instruments, Inc., Westbury, NY). The homogenate was centrifuged at 183,000 ϫ g for 30 min at 4°C, and the pH of the supernatant was adjusted to 6 with 0.2 M acetic acid followed by centrifugation at 20,000 ϫ g for 20 min at 4°C (14,16).
The supernatant from acid precipitation was applied to a DE52 cellulose column (2.5 ϫ 25 cm) equilibrated with 10 mM sodium phosphate, pH 6. The column was washed with 1 liter of the starting buffer and the enzyme eluted with a linear gradient of 0 to 0.4 M NaCl in the same buffer (total volume of 2 liters). The fractions containing activity were pooled, concentrated by ultrafiltration in a stirred-cell apparatus (Amicon Diaflo 10 PM30 membrane, W. R. Grace and Co., Danvers, MA), purified further by chromatography on a Pharmacia Superdex 200 column (10 mm inner diameter ϫ 30 cm), equilibrated with 10 mM sodium phosphate, pH 6, containing 50 mM NaCl. Fractions that contained PNG hydrolase activity were pooled, concentrated by ultrafiltration (Ultrafree-15 centrifugal filter device, Biomax-30K NMWL membrane, 15-ml volume, Millipore Corp., Bedford, MA), and subjected to chromatography on a Rainin Hydropore AX anion exchange column (polyethyleneimine with mixed primary, secondary, and tertiary amino sites, 4.6 mm inner diameter ϫ 25 cm, Rainin Instruments, Woburn, MA) in the same buffer at a flow rate of 1 ml/min using a linear gradient of 0 -0.4 M NaCl over 30 min. Fractions that contained activity were again concentrated by ultrafiltration (Ultrafree-15 centrifugal filter device, Biomax-30K NMWL membrane, 15-ml volume, Millipore Corp., Bedford, MA) and then applied to a Bio-Rad Econo-Pac t-Butyl HIC cartridge (Macro-Prep t-Butyl HIC support, 5-ml bed volume, Bio-Rad) equilibrated in 10 mM sodium phosphate, pH 6, with 2.4 M ammonium sulfate. Elution was accomplished using a linear gradient from 2.4 to 0 M ammonium sulfate over 30 min at a flow rate of 1 ml/min. Fractions that contained pyridoxine-␤-D-glucoside hydrolase activity were pooled and stored as small portions at Ϫ80°C.
Initial attempts to purify pyridoxine-␤-D-glucoside hydrolase involved application of active fractions from DE52 cellulose to a hydroxyapatite column (2.5 ϫ 18 cm; Bio-Gel HTP, Bio-Rad) equilibrated with 10 mM sodium phosphate, pH 6, without dialysis to remove salt. Nonretained effluent fractions exhibiting activity were pooled and passed through another hydroxyapatite column under the same conditions.
Purification of Broad Specificity ␤-Glucosidase-Chromatographic purification of broad specificity ␤-glucosidase was conducted by a minor modification of the procedures of Glew and associates (9) and DePetro (16). The pH 6 supernatant prepared and described above was applied to a DE52 cellulose column (2.5 ϫ 25 cm) previously equilibrated with 10 mM sodium phosphate, pH 6. After the column was washed with 1 liter of the starting buffer, the enzyme was eluted with a linear gradient of 0 -0.5 M NaCl in the same buffer (total volume of 1 liter). The fractions containing the activity were pooled and applied to a hydroxyapatite column (2.5 ϫ 18 cm) equilibrated with 10 mM sodium phosphate, pH 6, without dialysis to remove salt. The column was washed with five bed volumes of the same buffer, and then the enzyme was eluted with a 500-ml gradient of 10 mM to 0.1 M sodium phosphate, pH 6, followed by 200 ml of 0.1 M sodium phosphate, pH 6. The fractions that had activity were collected and applied to an octyl-Sepharose column (1.5 ϫ 25 cm) equilibrated with 10 mM sodium phosphate, pH 6, containing 0.5 M (NH 4 ) 2 SO 4 . The column was washed with 200 ml of the same buffer and then the enzyme was eluted with 10 mM sodium phosphate, pH 6, containing 60% (v/v) ethylene glycol. The fractions containing the activity were pooled and stored as small portions at Ϫ20°C until further analysis.
Enzyme Activity Assays-The standard assays for broad specificity ␤-glucosidase and pyridoxine-5Ј-␤-D-glucoside hydrolase were performed according to Nakano and Gregory (14,17). The activity of broad specificity ␤-glucosidase was determined with 2 mM p-NPGlc and pyridoxine-5Ј-␤-D-glucoside hydrolase with 0.2 mM PNG and 0.2 M sodium acetate buffer, pH 6.0. All incubations were conducted at 37°C under conditions that allowed measurement of initial rate. One unit of broad specificity ␤-glucosidase activity was defined as 1 nmol of p-nitrophenol released from the substrate per hour, using the molar absorptivity coefficient at 400 nm (18,300 M Ϫ1 cm Ϫ1 ; 18). One unit of pyridoxine-5Ј-␤-D-glucoside hydrolase activity was defined as the release of 1 nmol PN from PNG per h.
For the pH dependence studies, reactions were conducted using 0.2 M sodium acetate buffer at pH 4.5-6.0, 0.1 M sodium phosphate buffer at pH 6.5-7.5, and 0.1 M Tris-HCl buffer at pH 8.0 -9.0. For kinetic studies, standard substrates were replaced by various alternate ␤-D-glycosides at concentrations specified in the text and tables. Alternatively, a variety of concentrations of potential inhibitors or stimulating compounds as specified were added to the standard mixtures without preincubation of the enzyme.
Purified PNG hydrolase was evaluated for disaccharidase activity using 1 mM lactose, cellobiose, or sucrose individually as substrate. Incubations were conducted at 37°C and then stopped by incubating for 3 min in boiling water. Parallel blank reactions were conducted without enzyme added. Samples were analyzed for the residual disaccharides and formation of monosaccharide products by liquid chromatography (19) using a Dionex DX500 HPLC system and Dionex ED400 electrochemical detection system with pulsed amperometric detector (Dionex Corp., Sunnyvale, CA). Separations were performed using a Dionex CarboPac PA10 column eluted isocratically with 14 mM NaOH for 22 min at 1 ml/min and an additional 18 min at 100 mM NaOH. Retention times were determined and calibration curves established with authentic standards.
Protein concentration during chromatography was monitored by absorbance at 280 nm. Determination of protein concentration in the isolation of broad specificity ␤-glucosidase was determined according to Bradford (20), and the method of Markwell et al. (21) was used in the isolation of PNG hydrolase for its lower detection limits. Bovine serum albumin was the standard in each procedure.
Isoelectric focusing was performed in 5% (w/v) polyacrylamide gels as described by Garfin (23). Standard proteins (Sigma) used were soybean trypsin inhibitor (pI 4.6), bovine ␤-lactoglobulin A (pI 5.1), bovine carbonic anhydrase II (pI 5.9), and human carbonic anhydrase I (pI 6.6). pH gradients were prepared using a 60:40 (v/v) mixture of pH 3-10 and pH 2.5-5 ampholyte solutions (Pharmacia Biotech Inc.). In a parallel gel that was not fixed or stained, gel slices (3 mm) of the PNG hydrolase lane were assayed for enzymatic activity. In this procedure the gel segments were incubated in 0.2 M sodium acetate buffer, pH 5.5, for 10 h and then incubated with 125 M PNG at 37°C for 3 h.
Gel filtration chromatography was performed as described above using Pharmacia Superdex 200 column for determination of molecular mass under nondenaturing conditions. Molecular weight standards (Sigma) were cytochrome c (12.4 kDa), carbonic anhydrase (29 kDa), bovine serum albumin (66 kDa), and ␤-amylase (200 kDa). The absorbance of the column effluent was monitored at 280 nm.
Amino Acid Analysis-The purified enzyme preparations were subjected to SDS-polyacrylamide gel electrophoresis in a Tris-Tricine discontinuous buffer system (24) and then electrophoretically blotted to polyvinylidene difluoride membranes and stained with Coomassie Blue R-250 (25). The band corresponding to each enzyme was excised and subjected to gas-phase acid hydrolysis (26) followed by amino acid analysis as the phenylisothiocarbamate derivative using an Applied Biosystems 420A instrument (Foster City, CA).
Kinetic Analysis-Kinetic constants (K m , V max , and K i ) were calculated by nonlinear regression using EZ-FIT software (27).

Purification of Pyridoxine-␤-D-Glucoside Hydrolase-
The separation of pyridoxine-␤-D-glucoside hydrolase and broad specificity ␤-glucosidase activities by ion exchange on DE52 cellulose provided clear evidence of the existence of a distinct enzyme able to hydrolyze PNG (Fig. 1). The subsequently developed purification scheme of four consecutive chromatographic steps is shown in Fig. 2, and a summary of the purification of pyridoxine-␤-D-glucoside hydrolase is shown in Table  I. HPLC gel filtration and ion exchange steps each yielded further purification, and the final hydrophobic interaction chromatography step yielded an essentially pure enzyme with 36,500-fold purification relative to the supernatant from acid precipitation. This procedure has been repeated several times with consistent results.
Although PNG hydrolase activity is quite stable in frozen intestine, the fully purified enzyme exhibits much less stability. Only about 25% retention of activity is observed following a single cycle of frozen storage at Ϫ80°C (2 days) followed by thawing and assay. In contrast, the purified enzyme loses about 15% activity per day at 4°C. Table II. The enzyme was purified ϳ1,460-fold relative to the supernatant from acid precipitation.

Purification of Broad Specificity ␤-Glucosidase-A summary of the purification is shown in
The enzyme activity was stable in 10 mM sodium phosphate, pH 6, containing 60% (v/v) ethylene glycol at Ϫ20 or Ϫ80°C for over 2 months. The activity toward p-NPGlc did not change in the presence of 0.1% (w/v) taurocholic acid with or without 0.05% (v/v) Triton X-100 in the assay medium.
Molecular Mass and Isoelectric Point-SDS-polyacrylamide gel electrophoresis analysis indicated that broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase had been purified to homogeneity (Fig. 3). The apparent molecular mass under denaturing conditions as determined by SDS-polyacrylamide gel electrophoresis was 60 kDa for broad specificity ␤-glucosidase and 130 kDa for pyridoxine-␤-D-glucoside hydrolase. Determination of native molecular mass by gel filtration chromatography purification indicated 57 kDa for broad specificity ␤-glucosidase and 160 kDa for pyridoxine-␤-D-glucoside hydrolase (Fig. 4). The isoelectric point of broad specificity ␤-glucosidase was 4.8, and pyridoxine-␤-D-glucoside hydrolase exhibited a cluster of at least 4 bands all with pI Յ 4.8 (Fig. 5). Direct assay of gel slices for PNG hydrolase activity revealed activity across this cluster, which confirmed the heterogeneity of charged species in this preparation.
Amino Acid Composition-The results presented in Table III indicate similarity but also substantial compositional differences between these two enzymes. Most notably, broad specificity ␤-glucosidase exhibited ϳ2-fold greater proline and ϳ30% greater serine, whereas pyridoxine-␤-D-glucoside hydrolase exhibited ϳ80% greater lysine and over 40% greater alanine content. Both enzymes were high in glycine (ϳ12 mol%) and Asx ϩ Glx (18 -21 mol%).
Effects of Various Compounds on Activity-Effects of various FIG. 1. Elution profile of broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase activity from DE52 cellulose column chromatography. The supernatant from acid precipitation was loaded onto DE52 column and pyridoxine-␤-D-glucoside hydrolase (E) and broad specificity (Broadspec.) ␤-glucosidase (q) was eluted with a gradient of 0 -0.25 M NaCl in 10 mM sodium phosphate buffer, pH 6. Protein (‚) was monitored as the absorbance at 280 nm.
compounds on the activities of these enzymes are shown in Table IV. The effects of detergents were examined because ionic detergents (bile salts) are known markedly to activate lysosomal acid ␤-glucosidase (28 -30) and are needed for extracting this enzyme (31); however, these are inhibitors for cytosolic ␤-glucosidases. Nonionic detergents were found to be activators for broad specificity ␤-glucosidase, whereas these detergents slightly inhibited the activity of pyridoxine-␤-D-glucoside hydrolase. The ionic detergents were found to be inhibitory for both of these mucosal cytosolic enzymes. For broad specificity ␤-glucosidase, concentrations of approximately 0.1% (w/v) detergents caused ϳ50% inhibition, although approxi- Mucosal cytosolic fraction following acid precipitation was applied to a DE52 diethylaminoethyl cellulose column. Protein concentration was monitored as absorbance at 280 nm. Elution was accomplished using a 0 -0.4 M NaCl gradient in 10 mM sodium phosphate buffer, pH 6, following a 1-liter wash. Fractions (15 ml) were collected throughout the NaCl gradient only. B, gel filtration chromatography on Superdex 200. Fractions with activity from the previous column were concentrated and then separated (750-l injections) at 0.5 ml/min flow rate and 1 ml/fraction. Absorbance (solid line) was measured continuously at 280 nm. C, anion exchange HPLC on Hydropore AX column. Fractions with activity from the previous column were concentrated and then separated (750-l injections) at 1 ml/min with a gradient of 0 -0.4 M NaCl in 10 mM sodium phosphate buffer, pH 6, in 30 min. Absorbance (solid line) was measured continuously at 280 nm. Fractions of 1 min/tube were collected. D, hydrophobic interaction chromatography on Econo-Pac t-Butyl HIC column. Fractions with activity from the previous column were concentrated and then separated (100-l injections) at 1 ml/min using a linear gradient from 2.4 to 0 M ammonium sulfate over 60 min in 10 mM sodium phosphate, pH 6. The column was initially equilibrated in this buffer containing 2.4 M ammonium sulfate. Absorbance (solid line) was measured continuously at 280 nm. Fractions of 1 min/tube were collected. mately 1% (w/v) was necessary to achieve the same degree of inhibition to pyridoxine-␤-D-glucoside hydrolase. Deoxycholate was a more potent inhibitor of both enzymes than taurocholate, similar to the report by Raghavan et al. (32) regarding the lysosomal acid ␤-glucosidase. Thiol compounds either stimulated or lacked inhibition of these enzymes (Table IV). Sulfhydryl reagents (e.g. N-ethylmaleimide) were inhibitory to the activities of both enzymes (Table IV).  Alcohols (ethyl and n-butyl alcohol; Table IV) were activators for the enzymes at low concentration (less than 1% (v/v)); however, they were inhibitory at higher concentrations (2-10%) as reported by Gopalan et al. (33). Ethylene glycol was used to elute broad specificity ␤-glucosidase from the hydrophobic gel (octyl-Sepharose), and it stabilized the activity of the enzyme in this study and others (16,34). However, ethylene glycol was an inhibitor for pyridoxine-␤-D-glucoside hydrolase, and the inclusion of 10% (v/v) of this solvent in the assay medium yielded 50% inhibition.
Substrate Specificity and Kinetics-Pyridoxine-␤-D-glucoside hydrolase did not hydrolyze any of the aryl ␤-D-glycosides examined above; thus, comparative K m values were not determined. Pyridoxine-␤-D-glucoside hydrolase exhibited a K m for PNG of 0.88 Ϯ 0.12 mM (ϮS.E.), and the V max of the purified enzyme was 13.2 Ϯ 0.8 mol/h⅐mg protein (Fig. 7).
Although broad specificity ␤-glucosidase does not hydrolyze cellobiose or other disaccharides (9), we examined pyridoxine-␤-D-glucoside hydrolase for potential disaccharidase activity. HPLC analysis indicated that pyridoxine-␤-D-glucoside hydro-lase cleaved 1 mM lactose and 1 mM cellobiose to their respective monosaccharides at 30 and 16 mol/h/mg protein, respectively (Table V). Equivalent rates of substrate disappearance and product formation were observed within each reaction. The enzyme did not exhibit sucrase activity.

TABLE V Hydrolsis of disaccharides by pyridoxine-␤-D-glucoside hydrolase
Values presented are concentrations of glucose, galactose, and fructose following incubation of 1 mM lactose, 1 mM cellobiose, and 1 mM sucrose with purified pyridoxine-␤-D-glucoside hydrolase for 1 h. Reaction rates, expressed as a specific activity, are also presented. Reaction mixtures (500 l, containing 0.015 g of purified enzyme or corresponding blanks with no enzyme) were incubated for 60 min, and reactions were terminated by incubation for 3 min in boiling water and then analyzed by HPLC. ND, not detected. Values without enzyme represent contaminants in substrates and/or products of nonenzymatic hydrolysis. Investigation of Inhibitors-Various potential inhibitors were examined to gain further insight into the specificity of these enzymes and to further differentiate the nature of their catalytic properties. Aldono-(1, 5)-lactones were examined because they are potent, highly specific transition state analogue inhibitors of glucosidases (35)(36)(37). The carbohydrate moiety of a substrate and the lactone ring of corresponding configuration of these inhibitors bind competitively to the same site of the active catalytic center (38). It has been previously reported that the cytosolic broad specificity ␤-glucosidase is inhibited by micromolar concentrations of glucono-␦-lactone (38,39). In the present study, both the broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase were similarly inhibited by glucono-␦-lactone, with K i values of 10.3 and 7.4 M, respectively, as shown by Dixon plots (Fig. 8).
Gopalan et al. (40) reported that amphiphilic alkyl ␤-D-glucosides consistently displayed 100 -250-fold lower inhibition constants (K i ) with the broad specificity ␤-glucosidase of human liver compared with a placental lysosomal acidic ␤-glucosidase. In their study, the inhibitory effects of alkyl ␤-D-glucosides increased in proportion to the alkyl chain length. In the present study, we observed that n-octyl-␤-D-glucoside inhibited the activities of both broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase more effectively than did n-amyl(pentyl)-␤-D-glucoside (Table VI). These glucosides were potent competitive inhibitors for broad specificity ␤-glucosidase; K i values of 1.1 mM for n-amyl-␤-D-glucoside and 0.12 mM for n-octyl-␤-D-glucoside, respectively, were observed (Table  VI). Differences in the K i values were noted between the two enzymes. Pyridoxine-␤-D-glucoside hydrolase was found to be less sensitive to these alkyl glucosides, like lysosomal ␤-glucosidase (40), with inhibition constants of 33 mM for n-amyl-␤-Dglucoside and 7.5 mM for n-octyl-␤-D-glucoside (Table VI). These K i values are 30 -60-fold higher than those observed for broad specificity ␤-glucosidase. Modes of inhibition were found to be competitive. Dixon plots for broad specificity ␤-glucosidase with n-amyl-␤-D-glucoside and for pyridoxine-␤-D-glucoside hydrolase with n-octyl-␤-D-glucoside are shown in Fig. 8, A and B, respectively.
Previous studies have shown that 2-deoxy-2-fluoro-␤-D-glucosyl fluoride is a mechanism-based inhibitor of ␤-glucosidases, including hepatic cytosolic broad specificity ␤-glucosidase, by forming a stable covalent linkage of the 2-deoxy-2-fluoro-␤-Dglucosyl moiety to the active site (47). We examined pyridoxine-␤-D-glucoside hydrolase to evaluate its susceptibility to this mechanism-based inhibitor. Pyridoxine-␤-D-glucoside hydrolase exhibited a mixed mode of inhibition when evaluated in the presence of 0.4 mM 2-deoxy-2-fluoro-␤-D-glucosyl fluoride (Fig.  7). In contrast to the uninhibited reaction which yielded a K m for PNG of 0.88 Ϯ 0.12 mM and V max of 13.2 Ϯ 0.8 mol/h⅐mg protein, the presence of 0.4 mM 2-deoxy-2-fluoro-␤-D-glucosyl fluoride yielded a K m for PNG of 3.7 Ϯ 2.5 mM and V max of 7.5 Ϯ 3.5 mol/h⅐mg protein. DISCUSSION Because we did not anticipate the existence of a novel enzyme that is responsible for the hydrolysis of PNG, we initially attempted a purification of broad specificity ␤-glucosidase from the cytosolic fraction of jejunal mucosa and monitored the purification using a conventional assay with a nonphysiological substrate often used for this purpose. However, the activity catalyzing PNG hydrolysis was not co-eluted with broad specificity ␤-glucosidase, and the final preparation was completely  devoid of activity toward PNG. Using PNG as substrate in assays to monitor the chromatographic fractionations, we were able to partially purify pyridoxine-␤-D-glucoside hydrolase and separate it from broad specificity ␤-glucosidase on DE52 cellulose (Fig. 1). We initially observed that this enzyme does not bind to hydroxyapatite gel under the conditions we used to retain broad specificity ␤-glucosidase. Subsequent attempts to use this as a step in further purification of pyridoxine-␤-Dglucoside hydrolase failed because current commercially available forms of hydroxyapatite retain pyridoxine-␤-D-glucoside hydrolase apparently irreversibly. The sequence of four chromatographic steps (DE52 cellulose, Superdex 200, Hydropore AX, and Econo-Pac t-Butyl HIC columns) successfully purified pyridoxine-␤-D-glucoside hydrolase to homogeneity. The effectiveness of hydrophobic interaction chromatography in this purification is illustrated in the SDS-polyacrylamide gel electrophoresis comparison of Hydropore AX fractions that contained high enzyme activity but many contaminating bands with t-butyl HIC fractions showing the purified enzyme (Fig.  3).
Nonionic detergents (30,44,48) and bile salts (31) are required to extract the lysosomal acid ␤-glucosidase in a soluble form, and it requires incorporation of neutral detergents or ionic lipids such as sodium taurocholate or acidic phospholipids into the assay medium to achieve maximum glucocerebrosidase activity (29,49). Although nonionic detergents are only weak activators of lysosomal ␤-glucosidase, ionic detergents are potent activators (32). Nonionic detergents were modest activators of broad specificity ␤-glucosidase in this study (Table IV). Similar activation by nonionic detergents of broad specificity ␤-glucosidase from rat kidney cortex was also seen by Glew et al. (50). In contrast, nonionic detergents provided no activation of pyridoxine-␤-D-glucoside hydrolase activity; they were actually slightly inhibitory. As stated above, ionic detergents were inhibitors to both of the enzymes in this study, although less so for pyridoxine-␤-D-glucoside hydrolase. Overall, these results indicate that the mucosal cytosolic enzymes examined here differ markedly from the lysosomal acid ␤-glucosidase (i.e. glucocerebrosidase). The cytosolic broad specificity ␤-glucosidase is known as a neutral ␤-glucosidase, whereas the lysosomal enzyme is an acidic ␤-glucosidase on the basis of pH optima. Cytosolic pyridoxine-␤-D-glucoside hydrolase exhibited a similar pH optimum (pH 5.5) to the cytosolic broad specificity ␤-glucosidase in human liver (43), and the mucosal broad specificity ␤-glucosidase exhibited a slightly higher pH optimum (pH 6.5).
The overall results indicate that pyridoxine-␤-D-glucoside hydrolase has an acidic pH preference and is much less sensitive to alkyl glucosides than broad specificity ␤-glucosidase, although it is affected greatly by the potent inhibitor of the lysosomal enzyme, conduritol B epoxide. However, pyridoxine-␤-D-glucoside hydrolase is as sensitive to glucono-␦-lactone as cytosolic broad specificity ␤-glucosidase, and bile salts and nonionic detergents are inhibitory, not stimulatory as seen with the lysosomal ␤-glucosidase. Also this enzyme does not require the presence of detergents to obtain a soluble form. To purify ␤-glucosidases, DEAE-cellulose is often used for the cytosolic enzyme and CM-Sephadex for the lysosomal enzyme. The cytosolic enzyme generally does not show affinity for CM-Sephadex (29). Pyridoxine-␤-D-glucoside hydrolase exhibits some similarity to the lysosomal enzyme but binds readily to DEAE-cellulose (Fig. 1). Pyridoxine-␤-D-glucoside hydrolase appeared to exist as a monomer of approximately 130 kDa. Gel filtration behavior in nondenaturing conditions suggested possible rapid associative-dissociative behavior (apparent native mass 160 kDa). As reviewed by Glew et al. (46), lysosomal ␤-glucosidases aggregate to form a dimer or tetramer as an active form. Pyridoxine-␤-D-glucoside hydrolase differs in many ways from lysosomal acid ␤-glucosidase, and in many respects it has similarities to cytosolic ␤-glucosidases. Perhaps the most novel aspect of the identification and initial characterization of pyridoxine-␤-D-glucoside hydrolase is the high degree of substrate specificity of this enzyme and the fact that this specificity is for a naturally occurring substrate that comprises a significant fraction of dietary vitamin B6. The observations that pyridoxine-␤-D-glucoside hydrolase also hydrolyzes cellobiose, a glucosyl ␤-D-glucoside, and lactose, a glucosyl ␤-Dgalactoside, with rates similar to that of PNG hydrolysis and that nonglycosylated PN is not an inhibitor at the low concentration used (i.e. 50 M) suggest that pyridoxine-␤-D-glucoside hydrolase may have little recognition at its active site for the pyridoxyl aglycone of PNG.
The initial amino acid analysis of these enzymes revealed substantial differences in the relative quantity of several amino acids. In addition, substantial differences were observed for both enzymes from that reported by LaMarco and Glew (34) for broad specificity ␤-glucosidase from human liver, although the porcine broad specificity ␤-glucosidase exhibited the greater similarity to that from human liver. The molecular mass of ϳ57-60 kDa for porcine jejunal mucosal broad specificity ␤-glucosidase differs from the 68-kDa value reported by LaMarco and Glew (34) for the human hepatic enzyme and from 53 kDa deduced from the cDNA sequence of broad specificity ␤-glucosidase from guinea pig liver (50). The observed isoelectric point of porcine intestinal broad specificity ␤-glucosidase of 4.8 is in close agreement with those of 4.5-5.2 reported for mammalian cytosolic ␤-glucosidases (9, 51). The multiple bands observed in isoelectric focusing of pyridoxine-␤-D-glucoside hydrolase suggest charge heterogeneity, potentially related to glycosylation, with low pI (Յ4.8) consistent with the acidic character of many glycohydrolases. Until the full amino acid sequence is determined, the overall differences cannot be fully assessed. Analysis of sequence will permit an evaluation of the relationship, if any, between mucosal cytosolic broad specificity ␤-glucosidase, pyridoxine-␤-D-glucoside hydrolase, lysosomal acid ␤-glucosidase, broad specificity ␤-glucosidase from other tissues, and the many ␤-glucosidases from plant and microbial sources. Since pyridoxine-␤-D-glucoside hydrolase seems to prefer quite hydrophilic substrates (PNG and disaccharides), its amino acid composition, especially at the active site, may differ considerably from known cytosolic or lysosomal ␤-glucosidases. Of additional interest is the fact that the molecular mass of pyridoxine-␤-D-glucoside hydrolase (130 kDa) is similar to the family of higher molecular mass glycohydrolases such as mammalian intestinal brush border lactasephlorizin hydrolase (52). Our data suggest that both broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase have at least one essential cysteine residue at or near the active site, since thiol compounds stimulate but sulfhydryl reagents are potent inhibitors, as noted by Glew et al. (46). It should also be noted that 2-deoxy-2-fluoro-␤-D-glucosyl fluoride inhibited pyridoxine-␤-D-glucoside hydrolase in this study, affecting both K m and V max consistent with the mechanism-based covalent modification of the enzyme (47). This compound also inhibits cytosolic broad specificity ␤-glucosidase (53), although kinetic data have not been reported to our knowledge.
Pyridoxine-␤-D-glucoside hydrolase has similarities and differences from mammalian cytosolic and lysosomal ␤-glucosidases and, in terms of size and lactose hydrolysis, with mammalian lactase-phlorizin hydrolase. Aside from catalyzing the intracellular mucosal hydrolysis of absorbed PNG, we do not know yet if this enzyme can hydrolyze other natural glucosides.
LaMarco and Glew (10) reported that some natural glucosides such as L-picein and dhurrin can inhibit the broad specificity ␤-glucosidase activity in guinea pig liver, and L-picein is itself a substrate for this enzyme. Cyanogenic plant glucosides could also be hydrolyzed by the broad specificity enzyme, although their rates of hydrolysis are relatively low. For example, Damygdalin was hydrolyzed at 35% of the rate at which p-NPGlc was hydrolyzed by the cytosolic ␤-glucosidase (11).
Iwami and Yasumoto (54) reported that pyridoxine 4Ј-␤-Dglucoside, a minor form of vitamin B6 in certain plants, was absorbed by simple diffusion and that rat intestinal ␤-glucosidase activity did not hydrolyze this glucoside. We have shown in the present study that pyridoxine 5Ј-␤-D-glucoside is hydrolyzed by a specific enzyme present in pig intestinal mucosal cytosol but not by the cytosolic broad specificity ␤-glucosidase. We have also shown that pyridoxine-␤-D-glucoside hydrolase activity exists in rat, guinea pig, and human intestinal mucosal cytosolic fractions (12,54). As shown in the present study, our previous assumption that the intestinal cytosolic broad specificity ␤-glucosidase was solely responsible for hydrolysis of PNG in rat, guinea pig, and human intestine (12) is incorrect. Whether the disaccharidase activity of pyridoxine-␤-D-glucoside hydrolase has physiological function in mucosal cells is unclear. This enzyme differs markedly from several membrane-bound glycosidases involved in trimming newly synthesized glycoproteins in endoplasmic reticulum. No role of cytosolic pyridoxine-␤-D-glucoside hydrolase in disaccharide digestion is expected because of its intracellular location.
The specific activity of pyridoxine-␤-D-glucoside hydrolase in human jejunum is approximately 5.8-fold greater than that of rat intestine (12). This difference is consistent with the greater in vivo bioavailability of orally administered PNG in humans than rats. However, guinea pigs and rats exhibit similar specific activity of intestinal cytosolic pyridoxine-␤-D-glucoside hydrolase but guinea pigs exhibit much greater PNG bioavailability than either rats or humans (55). The high intralumenal ␤-glucosidase activity in guinea pig small intestine may be responsible for this high bioavailability. The fact that some tissues other than small intestine showed this activity, for example, kidney in rats (14), suggests that these organs may contribute to the in vivo post-absorptive hydrolysis of PNG and, thus, contribute to the bioavailability of PNG. Studies of intravenously injected PNG in humans have shown bioavailability approximately half that of the orally administered compound (6). This suggests that organs in addition to the intestine in humans contribute to the partial hydrolysis and utilization of dietary PNG.
Currently efforts are directed toward the preparation of antibodies against pyridoxine-␤-D-glucoside hydrolase and broad specificity pyridoxine-␤-D-glucoside hydrolase. Glew et al. (9) have reported the existence of immunoreactive broad specificity ␤-glucosidase in a variety of tissues, including intestine. It will be of interest to determine the distribution of pyridoxine-␤-D-glucoside hydrolase among tissues as well as to determine the cross-reactivity among the various mammalian cytosolic and lysosomal ␤-glucosidases. In addition, immunochemical analysis will be essential in clarifying the mechanism by which vitamin B6 deficiency causes increased activity of both intestinal broad specificity ␤-glucosidase and pyridoxine-␤-D-glucoside hydrolase (7).