Purification and Characterization of Glucosidase 11, an Endoplasmic Reticulum Hydrolase Involved in Glycoprotein Biosynthesis*

Rat liver glucosidase II, an endoplasmic reticulum hydrolase involved in the biosynthesis of the N-linked class of glycoproteins, has been purified in good yield to a state approaching homogeneity. The purified en- zyme hydrolyzes p-nitrophenyl-a-D-glucopyranoside, 4-methylumbelliferyl-a-~-glucopyrmoside, maltose, and the precursor oligosaccharides glucosel-zman- nosea-acetylglucosamine, but it does not act on glu-cose3mannose&acetylglucosamine or p-nitrophenyl- B-D-glucopyranoside. The ratio of the rate at which glucose is released from p-nitrophenyl-a-D-glucopyr- anoside to that from glucose2mannosesN-acetylgluco-samine or glucoselmannosesN-acetylglucosamine re- mains constant throughout the 8-step purification procedure; thus it appears that a single enzyme is respon- sible for the activities toward both the artificial and oligosaccharide substrates. The fact that the enzyme cleaves both of the inner 1,3-linked glucosyl residues from the precursor oligosaccharides supports the view that they are linked in the a-configuration. The pH dependence of enzymatic activity is quite similar for different substrates, showing a broad optimum between pH 6 and 7.5. Activity towardp-nitropheny1-a-D- glucopyranoside is enhanced by 12

Our work with the rat liver neutral microsomal a-D-glucosidase, originally reported by Lejeune et al. (20), began shortly after dolichol-linked precursor oligosaccharides were reported to contain glucose (21)(22)(23). Since a specific function for this membrane-bound glucosidase activity was unknown, we postulated that it was the enzyme responsible for excising glucose residues from the precursor oligosaccharide after its transfer to protein. By utilizing p-nitrophenyl-a-D-glucopyranoside as a substrate, we have purified this neutral a-glucosidase to a state approaching homogeneity. The recent availability of glucosylated precursor oligosaccharides has permitted us to demonstrate that the pNP-a-glucosidase' and glucosidase I1 activities are expressed by the same enzyme. A preliminary report of this work has been previously presented (24).
Preparation of Sizing Resins and Hydroxylapatite Before use, Sephacryl S-200 and S-300 resins were exhaustively washed with a I mg/ml solution of ovalbumin in 10 INM phosphate, pH 7.0. Periodically, sizing columns were rewashed with ovalbumin. This pretreatment significantly improved the yield of glucosidase 11.

Analytical Procedures
The standard assay of p-nitrophenyl-a-D-glucosidase activity was performed by incubating enzyme in 0.5 ml of 50 mM TES-NaOH buffer, pH 6.8, containing 4 m~ of substrate at 37 "C for 30 to 60 min.
The incubation was terminated by adding 1.0 ml of 0.64% ethylenediamine, pH 10.7, and the absorbance at 400 nm was determined. A simple modification of this assay was used when the presence of membranes caused high enzyme blanks. In this modified procedure, the reaction was terminated with 0.5 ml of ice-cold 10% trichloroacetic acid, mixed by vortexing, and then centrifuged at 1500 X g for 10 min. A 0.5-ml aliquot of the supernatant solution was carefully withdrawn and mixed with 1.0 ml of 0.64% ethylenediamine and 0.150 ml of 2 M NaOH. In both assay procedures, 1 p-nitrophenyl-a-D-glucosidase unit was defined as that amount of enzyme required to release 1 pmol of p-nitrophenol per h. The standard assay for 4-methylumbelliferyla-glucosidase activity was performed in the same manner, except that 1 mM substrate was used. The cleavage of maltose by glucosidase I1 was assayed through the use of hexokinase and glucose-6-phosphate dehydrogenase. The production of NADPH from NADP was monitored either by the increase in A340 or by the increase in fluorescence at 468 nm after excitation at 300-400 nm. The incubation mixture (2.0 m l ) for this assay had the following composition: 3 p~ ATP, 1.5 p~ NADP, 3 p~ MgC12, 1.6 units of hexokinase, 1.0 unit of glucose-6-phosphate dehydrogenase, 75 mM Tris-HC1, pH 8.0, 0-15 nM maltose, and 0 to 0.30 ml of enzyme solution. The reaction was terminated after 30 min at 37 "C. Because the cleavage of maltose results in the production of two free glucose molecules, the rate at which maltose is cleaved by glucosidase I1 is presented as one-half the rate of glucose production, a value more appropriate for the comparison of the rate of cleavage of maltose with other substrates.
For assays involving precursor oligosaccharide, the enzyme aliquot, adjusted to 0.150 m l , was pipetted into a small (8 X 10 m m ) test tube and kept on ice. To this solution was added 0.050 ml of 200 m~ TES-NaOH, pH 6.8, and lo00 to 1500 cpm of the appropriate [3H]glucoselabeled precursor oligosaccharide (see below) in 0.050 ml of H20 to give a final volume of 0.250 ml . The tube was incubated at 37 "C for the designated time (5 to 20 min) and then heated to 100 "C to terminate the reaction. The release of glucose was determined in either of two different ways. In Method 1, the released glucose was separated from large oligosaccharide by ascending paper chromatography according to the procedure of Grinna and Robbins (7). Radioactive samples were mixed with Aquasol and counted in a Beckman LS-7OOO programmable liquid scintillation counter (with automatic quench control) using the appropriate blanks and standards. Method 2 was a slight modifcation of the procedure of Michael and Kornfeld (10). After termination of the incubation, 0.5 ml of a solution containing 1 pmol of ATP and 1.8 units of hexokinase were added to each tube, and the mixture was reincubated at 37 "C for 20 min. Each sample was then diluted to 2.0 ml and passed through a column (0.5 x 3.0 cm) of Amberlite IRA-400 (Cl-). The effluent and a 2.0-ml water wash were fmt collected, and then the bound glucose 6-phosphate was eluted with 2.0 ml of 2 M NaC1. Method 2 was a reproducible and rapid assay of glucosidase I1 activity, but Method 1 was used whenever it was possible that added compounds, e.g. maltose, 2deoxyglucose, mannose,p-chloromercuribenzenesulfonate, etc., might inhibit hexokinase appreciably. It was possible to express glucosidase I1 activity only in terms of the percentage of [3H]glucose label released in a given amount of time, since neither the chemical quantitation of the amount of oligosaccharide substrate present nor of the amount of glucose released has yet proven possible. In a manner analogous to past reports (7,8,10,(12)(13)(14)(15), 1 unit of glucosidase I1 activity was defined as that amount of enzyme that released 1% of the [3H]glucose from the designated precursor oligosaccharide in 1 h. Bradford's "microprotein" dye-binding assay (25) was used to determine protein in solutions containing Ampholines or Tris buffer. However, throughout the purification steps in which phosphate rather than Tris buffers were used (Steps I-VI11 of the procedure detailed below), protein was determined with fluorescamine by the procedure of Anderson and Desnick (26).

Analytical Gel Electrophoresis
Polyacrylamide gel electrophoresis was performed using the discontinuous system of Davis (27). Glucosidase activity was visualized in the following manner. Gels were incubated in 100 m phosphate buffer, pH 6.5, for 15 min at 37 "C. This solution was decanted and replaced by a solution of 100 m~ phosphate, pH 6.8, containing 0.5 mM 4-methylumbelliferyl-a-~-glucopyranoside and 5 m~ 2-mercaptoethanol. Within 15 to 30 min enzyme activity could be observed as a fluorescent band under ultraviolet illumination. The released 4methylumbelliferone was washed from the gels by incubating overnight in 20% methanol at 37 "C with shaking. The gels were then stained for protein.
SDS-polyacrylamide gel electrophoresis was performed under reductive conditions according to the method of Laemmli (28). Prior to application to the gels, samples were dialyzed against two changes of 100 volumes of 0.2% SDS, lyophilized to dryness, and then resuspended in 0.30 ml of 10 m~ Tris/HCl, pH 6.6, containing 750 mM 2mercaptoethanol, 0.2% SDS, and 10% glycerol. After incubation at 100 "C for 5 min, the sampie was layered on the stacking gel. Gels were stained for protein with 0.5% Coomassie blue in 20% methanol/ 10% acetic acid and destained in 10% methanol/7% acetic acid. Staining for glycoprotein was accomplished using the periodic acid-Schiffs base technique according to Kapitany and Zebrowski (29).

Isoelectrofocusing
Analytical isoelectrofocusing was carried out in 3.5% polyacrylamide gels containing 0.1% Triton X-100 and a 1% (w/v) final concentration of Biolyte pH 3-10 Ampholines. The upper gel reservoir was 200 mM NaOH, and the lower reservoir was 150 m~ phosphoric acid. Preparative isoelectrofocusing was performed in a 1 1 0 4 capacity LKB column, with a 10-65% (w/v) sucrose density gradient employing 2.5% (w/v) pH 3-10 Ampholines. In either gels or density gradients, isoelectrofocusing was performed following the instructions given in LKB technical manual number 8100, Appendix I. A maximum of 2 watts of power was applied to 12 polyacrylamide gels, and a maximum of 5 watts was applied to the column.

Molecular Weight Determinations
The oligomeric molecular weight of glucosidase I1 was determined by polyacrylamide gel electrophoresis according to the method of Hedrick and Smith (30) using the Davis system (27) and also by molecular sizing on a calibrated column of Sephacryl S-300. The subunit molecular weight of the enzyme was determined by SDSpolyacrylamide electrophoresis according to the method of Laemmli (28). All standard plots of M, versus relative mobility (or V,) were fitted to the experimental data by the method of least squares.
Preparation of [3H]Glucose-labeled Precursor Oligosaccharide Microsomal membranes were prepared from commercially obtained frozen chicken liver for use in the in vitro oligosaccharide labeling system of Grinna and Robbins (7). Following incubation of the membranes (20 mg of protein/ml) with 400 pM MnC12, 16 pM UDP-GlcNAc, 2 GDP-Man, and 1 p~ UDP-[6-'H]glucose for exactly 9 min, the dolichol pyrophosphoryl oligosaccharide was extracted, hydrolyzed with mild acid to cleave the pyrophosphate bond, and incubated with endonuclease H (7). The released oligosaccharide was lyophilized and applied to a column (1 X 310 cm) of Bio-Gel P-4 (-400 mesh) equilibrated with 10 m~ phosphate, pH 7.0, containing 0.5 mM sodium azide. The precursor oligosaccharides G k -MangGlcNAc, GlcnMangGlcNAc, and GlclMan9GlcNAc eluted as distinct peaks of radioactivity which were identified according to their agreement with previously published results (7). GlczMangGlcNAc and GlclMangGlcNAc were the dominant species obtained, with GlcnMangGlcNAc comprising most of the remaining radioactivity recovered. There was some contamination of the GlcsMansGlcNAc component with GlceMangGlcNAc (about 15% contamination).

Preparation of Subcellular Fractions for the Purification of
Glucosidase ZZ 20 male Wistar rats (150 g) were fasted for 36 h prior to sacrifice by decapitation. The livers were minced and washed extensively in 0.25 M sucrose containing 10 m~ Tris-HCI, pH 8.0, and 5 m~ 2-mercaptoethanol before they were blotted and weighed. The livers were suspended in 4 volumes of the same sucrose buffer and carefully homogenized in a Waring blender (ten 4-s pulses, separated by 20-s pauses to minimize foaming). The nuclear fraction was prepared by centrifugation at 1500 X g for 10 mi n, and the fraction was homogenized and centrifuged again. The supernatant solution was centrifuged at 8000 X g for 15 min, and the pellet was homogenized and centrifuged again to yield the mitochondrial-lysosomal fraction. This procedure minimizes the loss of glucosidase I1 to this fraction. The postmitochondrial-lysosomal supernatant solution was diluted with the sucrose buffer to a final ratio of 20 ml per g of liver and made 10 mM in freshly prepared CaC12 in order to precipitate microsomes (31). The suspension was stirred for 30 min at 4 "C and then centrifuged at 12,000 X g for 15 min. The microsomal fraction was resuspended in 0.25 M sucrose containing 10 mM Tris-HC1, pH 7.2, and 5 rn 2mercaptoethanol. This suspension was then made 0.01% (v/v) in Triton X-100 by dropwise addition of 1% (v/v) Triton X-100 with stirring in order to remove all remaining lysosomal contamination while leaving the microsomes intact. (Additional CaClz is not necessary to keep the microsomes aggregated.) Following centrifugation, the washed microsomes were carefully resuspended in 50 nm phosphate, pH 7.0, and 5 m~ 2-mercaptoethanol for further use. These washed microsomes are approximately 5-fold enriched in glucose-6phosphatase activity over the homogenate, with a recovery of about 70% of the glucose-6-phosphatase activity of the homogenate.
Purification of the Rat Liver Glucosidase ZZ Since the enzyme is unstable during most chromatographic steps, all buffers contained 5 m~ 2-mercaptoethanol and were of pH 6 to 9, and all steps were performed at 4-6 "C unless otherwise noted. A representative purification experiment is summarized in Table I. Step I: Triton X-I00 Extraction of Washed Microsomes-The washed microsomes were resuspended in 8 volumes of 50 m~ phosphate buffer, pH 7.0. After ten 4-s bursts in a Waring blender, the suspension was made 1% (v/v) in Triton X-100 by dropwise addition of detergent with stirring. Stirring was continued for 2 h, and the mixture was then centrifuged at 78,000 X g for 4 h. The supernatant solution routinely contained 95-100% of the original microsomalpNPa-glucosidase activity enriched approximately 8-fold over the homogenate.
Step ZZ: Batch Adsorption with Hydroxylapatite-A suspension of hydroxylapatite (25) was slowly added with stirring to the microsomal Triton extract to give a final 3:l (w/w) ratio of hydroxylapatite to protein. This mixture was brought to room temperature, stirred for 15 min, and then centrifuged at 1500 X g for 5 min. The pellet was resuspended in 500 ml of 50 m~ phosphate, pH 7.0, stirred for 15 min at room temperature, and then centrifuged as before. Further washing of the hydroxylapatite successively with 500-ml volumes of 60 mM, 70 mM, and 80 mM phosphate, pH 7.0, was performed before elution of the enzyme was effected with 200 ml of 250 mM phosphate, pH 7.0.
Step IIZ: Chromatography on Concanavalin A-Sepharose 4B-The 250 n m phosphate eluate was applied to a concanavalin A-Sepharose 4B column (1 X 6 cm) at 4 "C. The column was washed at room temperature with 200 ml of 200 m~ phosphate, pH 7.0, and the glucosidase activity was then eluted with 150 ml of 200 mM phosphate, pH 7.0, containing 1 M a-methylmannoside. The eluate was immediately concentrated to 3 ml by ultrafitration employing an Amicon apparatus fitted with a PM-30 membrane.
Step IV: Gel Filtration on Sephacryl S-2OU"The concentrate was made 20% in glycerol and chromatographed on Sephacryl S-200 (Fig.  ZA). The enzymatic activity usually eluted as two components. The major component eluted just after the center of the void volume peak (V,/Vo = 1.05). The minor component eluted at VJVO = 1.36 and was very unstable. (Although this minor component sometimes accounted for 20% of the activity recovered from the column, it often was present in only trace amounts.) The major component was pooled by combining column fractions with highest enzymatic activity so that the pool possessed a specific activity of at least 11 units/mg.
Step V: Chromatography on DEAE-Cellulose-The pool of the fractions from the S-200 column was diluted 1:1 with 10 nm phosphate, pH 7.0, containing 0.1% Triton X-100 and applied to a column (1 X 6 cm) of Whatman DE-52 that had been equilibrated with 10 mM phosphate, pH 7.0. Because the enzyme exhibited extreme instability on this column, it was necessary to carry out this step as quickly as possible. Elution of the enzyme occurred at 150 m~ NaCl when a 200-ml linear NaCl gradient (10 to 300 m~) in 10 IIIM phosphate buffer, pH 7.0, was applied. The fractions with highest enzymatic activity were combined so that the pool possessed a specific activity of at least 18 units/mg.
Step VI: Chromatography on Hydroxylapatite-The DE-52 eluate pool was diluted to 130% of its volume with 10 mM phosphate, pH 7.0, and applied to a column of Bio-Rad HTP hydroxylapatite (1 X 4 cm). The enzyme eluted as a single symmetrical peak at approximately 200 m~ phosphate in a 200-ml linear phosphate gradient (10 to 400 mM), pH 7.0. The column fractions were combined to give a specific activity of at least 30 units/mg for the pool.
Step VZZ: Gel Filtration with Sephacryt S-300-The pooled eluate from the hydroxylapatite column was concentrated as before to 3 ml , made 20% in glycerol, and applied to a column ( Step VZZZ: Second Chromatography on Hydroxylapatite-The pool from the Sephacryl S-300 column was applied to a second hydroxylapatite column identical with the one used in Step VI. Elution was accomplished with 200 ml of a linear phosphate gradient (10 to 300 m~) , pH 7.0. The enzyme eluted at 200 mM phosphate as umn results in the apparent high yield and high purification from Step 111. For this reason, these values are given in parentheses. During Step IV, a minor form is sometimes observed. This activity may represent artifactual modification of glucosidase 11, and so its values are reported in parentheses. a single symmetrical peak. The activity and protein elution profiles followed each other closely (Fig. 1B). The specific activity across the peak was 76 to 80 units/mg and after pooling and concentration was 80 units/mg. This constituted a 344-fold purification over the microsomal Triton extract and corresponded to a 2650-fold purification over the homogenate. The overall yield from the microsomal fraction was nearly 13% (Table I).

Subcellular Distribution of Neutral Glucosidase Lejeune et al. (20)
, using maltose as substrate, reported that rat liver contains both membrane-bound and soluble neutral a-glucosidase activities. Using pNP-a-glucoside as substrate, we found that the 105,000 X g microsomal fraction contained 45% of the activity in the homogenate; the remainder was present in the cytosol. That the microsomal enzyme was present in the endoplasmic reticulum rather than in Golgi or plasma membranes was strongly suggested by the work of Dewald and Touster (32). Their experimental results were confirmed in the present study (data not shown). Moreover, the amount of neutral pNP-a-glucosidase activity obtained from a highly purified Golgi fraction (33) was less than 1% of the amount found in washed microsomes and had a specific activity of 0.040 unit/mg, approximately one-fourth the specific activity obtained with the endoplasmic reticulum fraction. Therefore, a washed microsomal fraction was used as the source of the enzyme for purification procedures.
Purification of Glucosidase 1 1 Table I shows the results of a typical purification experiment (see under "Experimental Procedures"). Variability in the final two steps (VI1 and VIII) sometimes resulted in a lower extent of purifkation and a lower yield. The best results were obtained when the purification was accomplished rapidly. Unless otherwise indicated, the studies reported below utilized enzyme purified through Step VIII.

Characteristics of Glucosidase 11
Evidence that Glucosidase 11 and the Neutral pNP-a-Glucosidase Are Identical-The purified enzyme was analyzed by electrophoresis on 6% polyacrylamide tube gels (see under "Experimental Procedures"). When enzymatic activity was visualized using 4-methylumbeUiferyl-a-~-glucoside, only a discrete doublet was observed ( Fig. 2A). Subsequent staining of the same or duplicate gels for protein with Coomassie blue demonstrated one closely spaced doublet identical in position and appearance with the activity band. In 5 to 10% gels, the protein and activity bands always comigrated exactly; furthermore, the spacing of the doublet did not change. Overloaded gels gave the same results. When samples from different parts of the pNP-a-glucosidase activity peak obtained with the hydroxylapatite column (Step VIII; Fig. lB, fractions  19, 21, and 25) were analyzed by electrophoresis, the same discrete doublet of activity and protein banding was observed. The two portions of the doublet were still approximately equivalent in intensity. The same results were obtained with samples from the Step IV Sephacryl S-200 column (Fig. lA,  fractions 65, 69, 71, and 77).
The purity of the enzyme was further investigated by reductive SDS-polyacrylamide electrophoresis in the Laemmli system (see under "Experimental Procedures"). Under these conditions, the enzyme migrated as one major band on 8 and 10% gels (Fig. 2B). There were minor bands present on overloaded gels that migrated faster than the major bands; these contaminants constituted about 10% of the total protein seen on the gels (Fig. 2 B , fmt gel). These bands may be derived from the glucosidase I1 subunit, or they may truly represent contaminants that were not observed on the overloaded native gels. The major band was always rather broad, and it is In each case, the major protein band constitutes 90% or more of the total protein detected. It is more common to see the low molecular weight contaminants of the type visible in the first (8%) gel, but occasionally higher molecular weight contaminants, such as those in the second (10%) gel, were observed.
possible that it was an unresolved doublet (Fig. 2B, second The highly purified enzyme efficiently catalyzed the release of glucose from GlczMangGlcNAc and GlclMangGlcNAc, but not from Glc3MangGlcNAc. Table I1 demonstrates the ability of the enzyme to cleave these oligosaccharides at different stages of the purification scheme. In general, 70 to 80% of the glucose was released from Glcl-zMan9GlcNAc. The ability to release glucose from GlcsMansGlcNAc (glucosidase I activity) was lost during the Sephacryl S-200 sizing step (Step IV).
Due to incomplete separation of GlcsMan9GlcNAc and GlczMansGlcNAc during their preparation (see under "Experimental Procedures"), the Glc3MangGlcNAc used in these experiments was known to contain a 10-15% contamination of Glc2Man9GlcNAc; this probably accounted for the low amounts (less than 5-6%) of glucose released from GlcaMansGlcNAc during the latter stages of enzyme purification. The release of about 80% of the glucose label from the Glc2Man9GlcNAc and GlclMangGlcNAc agreed very well with previous reports (7,8, 10, 13, 14). Table I1 also presents data pertaining to the ratio between pNP-&-glucosidase activity and the cleavage of glucose from Glc,_,MangGlcNAc throughout the purification procedure. It is apparent that 1 unit of pNP-a-glucosidase activity possessed about 3000 "Glc, units" of activity throughout the purification and about 6000 "Glcz units" of activity. Purified glucosidase TI releases glucose from GlczMangGlcNAc at a higher rate than from GlclMan9GlcNAc; this finding is in agreement with observations in previous reports (7, 10, 13, 14). More importantly, since the ratio of activity toward pNP-a-glucoside and the oligosaccharides remains essentially constant throughout the purification, it appears that a single enzyme is responsible for both types of activity. Enzyme purified 2000-fold by an alternate puification scheme involving sucrose density gradient isoelectrofocusing gave identical results (data not shown).
The Oligomeric Molecular Weight of Purified Glucosidase 11-To determine the molecular weight of the native enzyme, 5 through 9% Davis system polyacrylamide gels were run in order to generate Ferguson plots (33) according to the method of Hedrick and Smith (30). The average molecular weight obtained was 262,000 k 4,000 (Fig. 3.4). Gel filtration with a Sephacryl 5-300 column (Fig. 3B) gave a value of 288,000 -t 10,000 daltons. gel).
Subunit Molecular Weight-As indicated above, electrophoresis of the enzyme on reductive SDS-10% polyacrylamide gels demonstrated one band. The R,s of a series of standards were plotted against their reported molecular weights as shown in Fig. 4. From this plot, the subunit M, of the enzyme was estimated to be between 64,000 and 66,000. When less stringent conditions were used in the treatment of sample prior to its loading onto the gels (150 m~ 2-mercaptoethanol rather than 750 m 2-mercaptoethanol and heating a t 40 "C rather than boiling), a second band sometimes appeared. This band migrated in a 10% gel as if it were approximately twice the sue of the subunit (i. e. 130,000 to 150,000 daltons); therefore, this band may represent a relatively stable dimer of the enzyme subunits.
Glycoprotein Nature of the Enzyme-Periodic acid-Schiff s base staining (29) was positive with 60 pg of purified glucosidase I1 after reductive SDS-electrophoresis in 10% gels and with 50 pg of purified enzyme after electrophoresis on 8% Davis gels. This was consistent with the behavior of the enzyme on a column of concanavalin A-Sepharose 4B (Step 111 of the purification procedure) and with the glycosidase studies described below.
Isoelectric Point of Glucosidase II-A PI of 3.2 was obtained with a fresh 1% Triton X-100 extract of washed microsomes (Fig. 5A). Isoelectrofocusing of freshly purified glucosidase I1 using a pH gradient of 3 to 10 in either sucrose density gradients (Fig. 5B) or 3.5% polyacrylamide gels showed one sharply focused component with a pl of 3.5-3.8. When partially purified enzyme preparations (purified through Step 111) were stored at 4 "C, they generally exhibited a higher p1 value; the PI of such stored preparations was sometimes as high as 4.8 (data not shown).
Since it seemed likely that sialic acid residues might be responsible for the low PI of glucosidase XI, the Triton extract was treated with several enzymes in the presence or absence of their inhibitors. The glucosidase I1 (1.  droxylapatite column (Step VIII) Expressed as the percentage of total radioactivity released by 0.02 min by 0.02 pNP-glucosidase unit using the standard assay conditions, pNP-glucosidase unit of glucosidase I1 using the standard assay '' A Glc, unit is that amount of glucosidase 11 that will cleave of conditions described under "Experimental Procedures," except that the radioactivity from Glc,Man&lcNAc in 1 h. A Glcz unit is defined incubation time was extended to 6 h.   67,000). The arrow indicates the elution volume of purified glucosidase 11; the estimated molecular weight from this experiment was 282,000. The elution buffer was the same as that used for the Sephacryl S-200 column (Fig. 2 4 ) .
absence of their inhibitors caused basic PI shifts of approximately 1.3 pH units. These shifts were greatly reduced when the appropriate inhibitors of the enzymes were present. On the other hand, endo-/3-N-acetylglucosaminidase H (0.01 unit, 50 mM citrate, pH 5.5) had no effect on the PI of glucosidase 11. The results suggested that modification of sidic acid-containing oligosaccharides may be largely responsible for the PI shifts observed. Stability of Glucosidase 11-The enzyme was stable (greater than 85% retention of activity for at least 1 week) between pH 6.0 and 8.0 only in the presence of 5 mM 2mercaptoethanol. Phosphate buffer stabilized the enzyme better than borate, barbital, Tris, or TES buffers.
The 1% Triton X-100 extract of washed microsomes was subjected to heat treatment in 50 m~ phosphate, pH 7.0, and 5 m~ 2-mercaptoethanol in order to study the thermal inactivation of the enzymatic activity and also to discern whether there might be more than one pNP-a-glucosidase present in the extract. The enzyme was completely stable for a t least 2 h a t 36 "C but was slowly inactivated at 40 "C. The apparent f i t order inactivation kinetics a t 44 "C indicated that there was present only one major enzymatic form capable of cleav-  ing pNP-a-glucoside. The heat inactivation of a 2-week-old 1% Triton extract was considerably different. Although this extract had retained 90% of its original activity, the enzymatic activity was rapidly inactivated at 40 "C. The enzyme was best stored for several weeks by quickfreezing in liquid nitrogen in a solution of 100 ~l l~ phosphate buffer, pH 7.0,lOO m maltose, and 5 m 2-mercaptoethanol.
Enzymatic Properties of Isolated Glucosidase II pH Dependence-As shown in Fig. 6 A , the observed pH dependence of glucosidase 11 activity toward pNP-a-glucoside was nearly identical with that reported initially by Lejeune et al. (20) for the hydrolysis of maltose. The pH dependence for the hydrolysis of 4-methylumbelliferyl-a-~-glucopyranoside was essentially the same (data not shown). With the precursor oligosaccharides as substrates, the pH activity curves were more sharply defined than those with the above three substrates (Fig. 6A). With GlclMangGlcNAc, the optimal pH was 6.6, and with GlcnMangGlcNAc it was 7.0. It should be noted that purified enzyme showed no activity in the pH range in which the lysosomal acid a-glucosidase is active (pH 3.0 to 5.0, with the optimum at pH 4.5 (37)).
Time Course-The release of glucose from Glcz-MangGlcNAc by glucosidase I1 reaches a maximum of approximately 80% (Fig. 6B), in agreement with the data in Table 11, and is linear for 35 to 40 min. Glucose release from GlclMangGlcNAc attained the same maximum, but the rate of release was somewhat slower.
Kinetics: Activators and Inhibitors-The cleavage of pNP-Glc by purified glucosidase I1 follows Michaelis-Menten kinetics. Double reciprocal plots of the data yielded a K, of 0.85 lll~ and VmaX of 0.090 pmol/h/pg of enzyme (Fig. 7). The cleavage of maltose also followed Michaelis-Menten kinetics, with a K , of 4.8 mM and a VmaX of 0.125 pmol/h/pg of enzyme (Fig. 7). We have not yet been able to determine the K, and V,,, for the enzyme with Glcl-ZMangGlcNAc oligosaccharides because of the limited amounts of substrate available.
As shown in Fig. 7 Fig. 8A shows the activation of pNP-Glc activity as a function of the concentration of 2-deoxy-~-glucose and mannose. The former compound was effective at a concentration as low as 12.5 m. Mannose also enhanced enzymatic activity but to a more modest level and at higher concentrations than 2-deoxy-~-glucose. It seemed possible that the apparent activation might reflect transglucosylation similar to that exhibited by lysosomal acid a-glucosidase (37) or the transgalactosylation shown by /%galactosidase (38). However, neither the expected product of this process, glucosyl-2-deoxyglucose, nor any higher molecular weight product could be detected by thin layer chromatography.
The present report describes a procedure for isolating rat liver glucosidase I1 in highly purified form, permitting further study of its substrate specificity and response to effectors, and the determination of its molecular weight, subunit composition, and other properties. The availability of the purified enzyme should facilitate future studies of its structure and function. Glucosidase I1 is a neutral a-glucosidase that excises both of the 1,3-linked glucose residues of precursor oligosaccharides after the terminal 1,2-liked glucose has been removed. Since the ratio of the rate of glucose released from precursor oligosaccharides to the rate of glucose released from pNP-a-glucoside remained constant throughout the purification procedure, and since the recovery a t each step is reasonably good, it is very likely that we have isolated the major glucosidase I1 from rat liver microsomes. That this enzyme may be involved in regulating the levels of lipid-linked glucosecontaining oligosaccharides was first suggested by the work of , who found in testing rat spleen lymphocytes that deglucosylated lipid-linked precursor oligosaccharides were preferentially catabolized through the action of a phosphodiesterase to form a free phosphooligosaccharide. In regard to the roles of the glucose residues in precursor oligosaccharides, a recent report (43) on the biosynthesis of human chorionic gonadotrophic hormone in cultured human choriocarcinoma (JAR) cells may also be mentioned. Only the monoglucosyl derivative could be detected in precursors of the a-subunit of the hormone. It would be of interest to determine whether this observation is a result of the occurrence of a biosynthetic pathway that does not involve di-and triglucosylated intermediates, or, alternatively, an unusually rapid intracellular conversion of these intermediates to the monoglucosyl derivatives.
The fact that a purified a-glucosidase efficiently releases both of the inner 1,3-linked glucoses verifies previous findings with cruder enzyme preparations (7, 8, 10, 13) and provides further evidence that these residues are linked in the a-configuration. Spiro et al. (17) reported that the chromium trioxide analysis of the calf thyroid precursor oligosaccharide indicated that the glucose residues were a-linked. Glucosidase I1 hydrolyzes GlcpMangGlcNAc at a higher rate than it hydrolyzes GlclMangGlcNAc. This finding is consistent with data obtained using partially purified enzyme (7,10,13,14); the GlczMangGlcNAc activity (expressed as a percentage of labeled glucose released per unit of time) generally appeared to be 1.5-to 2-fold higher than the Glc,MangGlcNAc activity. It is difficult to compare the concentrations of the GlczMan&lcNAc and GlclMan9GlcNAc substrates used and the actual rate of glucose release. However, assuming a uniform labeling of the glucose residues and noting that the concentration of GlczMangGlcNAc in terms of terminal glucose is only half that of GlclMangGlcNAc, then it may be concluded from our studies ( Table 11) that GlczMangGlcNAc is hydrolyzed several times faster than GlclMangGlcNAc. This conclusion is in harmony with the results of in vivo pulse and

Purification and
Characterization of Glucosidase 11 9999 pulse-chase radiolabeling studies with chicken embryo fibroblasts (5, 6), which have shown that the first and second glucose residues are removed rapidly, while the last glucose is removed substantially more slowly. (Although compartmentalization may contribute to this observation, it very likely also reflects the substrate preference of glucosidase 11.) The tentative conclusion about the relative substrate activity of GlczMangGlcNAc and GlclMan9GlcNAc would not be valid if the two glucose residues were cleaved from the former compound in a concerted sequential manner without the monoglucosyl derivative leaving the active site of the enzyme. Whether or not this occurs cannot as yet be answered.
It should also be mentioned that the yields of glucose released from Glcl_nMangGlcNAc never exceeded 8076, although recoveries of added labeled free glucose in these assays were nearly quantitative. Previous investigators (7,8,10,(12)(13)(14) also generally obtained less than quantitative release of potentially susceptible glucose residues. Whether these results reflect the presence of minor amounts of a different isomer of the oligosaccharide substrate with internal glucosyl residues or the conversion of radiolabeled glucose into mannose during the in vitro microsomal labeling assay (7) is unknown at the present time.
The action of glucosidase I1 on pNP-a-glucoside or GlclMangGlcNAc was not affected by 30 n" turanose, an effective inhibitor of the lysosomal a-glucosidase at 1 to 5 mM (20), nor was it affected by 25 mM kojibiose, an inhibitor of glucosidase I (14).
The effects observed with maltose, glucose, p-chloromercuribenzene sulfonate, and Tris were very similar to those reported for the rat liver glucosidase I1 activity investigated by Ugalde et al. (13,14). Similar inhibition of rat liver glucosidase I1 activity by glucose was also reported by Grinna and Robbins (7). The observations that the highly purified rat liver glucosidase I1 was not activated by 500 mM phosphate, pH 7.2, and was only slightly activated by 60% (w/v) ammonium sulfate are in contradistinction to the results obtained with the thyroid glucosidase (17). Glucosidase I1 was not influenced by Na' or K' , in marked contrast to lysosomal acid a-glucosidase, which is greatly stimulated by mono-and divalent metal ions (37). Mannose and 2-deoxyglucose activate the pNP-glucosidase activity of glucosidase 11. On the other hand, these two sugars inhibit, rather than activate, glucosidase I1 when the precursor oligosaccharides are used as substrates. Mannose had previously been reported to inhibit the cleavage of Glc2-MangGlcNAc and GlclMangGlcNAc (7). A likely explanation for the contrary effects of the two sugars on the hydrolysis of pNP-glucoside, on one hand, and oligosaccharides on the other, is that the two sugars are recognized by a site on the enzyme that normally binds the a-1,2-linked mannosyl residues of a branch of the oligosaccharide adjacent to the one containing glucose. Grinna and Robbins (7,8) have reported that the glucosidase I1 activity was three to four times higher with Glcl-PMan9GlcNAc as substrate than with Glcl-zMan7GlcNAc, which lacks the two terminal mannoses of the middle branch. Moreover, they report that the cleavage of Glcl_nMangGlcNAc was inhibited 80% by the addition of 1 mM MangGlcNAc to the incubation mixture, a result also suggesting recognition of the a-1,2-linked mannosyl branch (8). In addition, Michael and Kornfeld (10) found that the cleavage of Glcl.zMan4GlcNAc was many-fold slower than the cleavage of Glcl-ZMansGlcNAc. Spiro et al. (17) have reported similar results.
While the differing effects of mannose and 2-deoxyglucose on the cleavage of the artificial and oligosaccharide substrates and the slightly different pH optima observed with different substrates (Fig. 6A) might be considered suggestive of the presence of more than one glucosidase in our preparation, the evidence from both our studies and those of others rather strongly suggests that only one enzyme is involved.
Isoelectrofocusing of purified glucosidase I1 in either sucrose density gradients or polyacrylamide gels showed an apparent PI of 3.5-3.8 for freshly prepared glucosidase 11; only one major form was apparent. In our experience, a PI of 3.5-3.8 is low for microsomal proteins, most of which exhibit PI values between 6 and 8 by isoelectrofocusing. This PI is higher than that obtained with freshly prepared Triton extract of washed microsomes (PI = 3.2). These results differ slightly from those recently published by Ugalde et al. (14) in which glucosidase I1 was isoelectrofocused after elution from a column of concanavalin A-Sepharose. These investigators reported one major form with a PI of 4.2. The fact that the PI increases when glucosidase I1 is stored prior to further purification suggests the presence of a labile acidic group or the modification of the glycoprotein by some other component of the preparation. The fact that a substantial increase in PI was produced by incubation of the Triton extract with neuraminidases or with endonuclease D, but not with endonuclease H, suggests that oligosaccharide of the complex type may be responsible to some extent for the low PI of glucosidase 11. It is possible that neuraminidase present in crude glucosidase I1 preparations may be the cause of the increase in PI when the preparations are stored.