Characterization and immunolocalization of bovine N-acetylglucosamine-1-phosphodiester alpha-N-acetylglucosaminidase.

N-Acetylglucosamine-1-phosphodiester alpha-N-acetylglucosaminidase (phosphodiester alpha-GlcNAcase) has been purified 3,000-fold from bovine liver and its kinetic properties determined as described in the previous report (Mullis, K. G., Huynh, M., and Kornfeld, R. (1993) J. Biol. Chem. 269, 1718-1726). This report describes the hydrodynamic and lectin binding properties of phosphodiester alpha-GlcNAcase as well as its intracellular localization. The molecular weight of phosphodiester alpha-GlcNAcase is 204,950, as determined from density gradient centrifugation in D2O and H2O glycerol gradients and gel filtration. Enzymatically active enzyme migrates on sodium dodecyl sulfate-polyacrylamide gel electrophoresis with a molecular weight of 129,000, consistent with native phosphodiester alpha-GlcNAcase being a dimer. The lectin binding properties of phosphodiester alpha-GlcNAcase indicate that it contains sialylated species of both complex type N-linked oligosaccharides and O-linked oligosaccharides. In immunofluorescence studies phosphodiester alpha-GlcNAcase shows a perinuclear, Golgi localization in Vero cells as does the mid-Golgi marker alpha-mannosidase II. After exposure of the Vero cells to brefeldin A, phosphodiester alpha-GlcNAcase assumes an endoplasmic reticulum staining pattern. In contrast, in cells costained with the trans-Golgi marker wheat germ agglutinin, the wheat germ agglutinin marker assumed an endosomal network appearance after exposure to brefeldin A. These findings indicate that phosphodiester alpha-GlcNAcase is normally located within the Golgi stack, separate from the trans-Golgi and trans-Golgi network stained by wheat germ agglutinin.

The lectin binding properties of phosphodiester a-GlcNAcase indicate that it contains sialylated species of both complex type N-linked oligosaccharides and 0linked oligosaccharides. In immunofluorescence studies phosphodiester a-GlcNAc-shows a perinuclear, Golgi localization in Vero cells as does the mid-Golgi marker a-mannosidase 11. After exposure of the Vero cells to brefeldin A, phosphodiester a-GlcNAcase assumes an endoplasmic reticulum staining pattern. In contrast, in cells costained with the trans-Golgi marker wheat germ agglutinin, the wheat germ agglutinin marker assumed an endosomal network appearance after exposure to brefeldin A. These findings indicate that phosphodiester a-GlcNAc-is normally located within the Golgi stack, separate from the trans-Golgi and trans-Golgi network stained by wheat germ agglutinin.
In the preceding paper (1) we described the 3,000-fold purification of N-acetylglucosamine-1-phosphodiester a-Nacetylglucosaminidase (phosphodiester a-GlcNAcase)' from *This research was supported by Grants CA08759 and T32-HL07088 from the National Institutes of Health. The scanning confocal microscope was provided by a grant from the Lucille Markey Charitable Trust. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "Oduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
To whom correspondence should be addressed Depts. bovine liver. This enzyme catalyzes the second step in the formation of the mannose 6-phosphate recognition marker on lysosomal enzyme oligosaccharides by removing the N-acetylglucosamine residue from the GlcNAc-P-mannose formed by GlcNAc phosphotransferase action. The purification of bovine liver phosphodiester a-GlcNAcase has enabled us to characterize the enzyme biochemically with regard to its catalytic and kinetic properties, as described in the preceding report (1) and also its physical properties as described in this report. The molecular weight of the phosphodiester a-Glc-NAcase was determined by analysis of its hydrodynamic properties, and its oligosaccharide composition was characterized using plant lectins.
The intracellular localization of phosphodiester a-Glc-NAcase activity has been analyzed previously by kinetic studies in intact cells and by subcellular fractionation techniques. In pulse-chase labeling studies performed in transport-impaired mouse lymphoma cells, Lazzarino and Gabel (2) showed that the initial phosphorylation of lysosomal enzyme oligosaccharides occurred in a compartment proximal to that in which they were uncovered by phosphodiester a-Glc-NAcase. Goldberg and Kornfeld (3) demonstrated that when mouse lymphoma cell membranes are fractionated on sucrose gradients, phosphodiester a-GlcNAcase activity fractionates with a lighter compartment than GlcNAc-phosphotransferase and clearly distinct from galactosyltransferase activity found in a much lighter fraction. In a similar study, Schweizer et al. (4) showed that in extracts from Vero cells, phosphodiester a-GlcNAcase activity cofractionates with GlcNAc-phosphotransferase activity at a density intermediate between the pre-Golgi marker, p53 and galactosyltransferase. These data indicate, albeit indirectly, that phosphodiester a-GlcNAcase most likely resides in the cislmedial-Golgi. We have generated and used a monoclonal antibody directed against phosphodiester a-GlcNAcase to visualize directly the intracellular location of phosphodiester a-GlcNAcase by immunofluorescence studies. We have used brefeldin A, a fungal metabolite, to define further the intracellular localization of phosphodiester a-GlcNAcase within the Golgi.
EXPERIMENTAL PROCEDURES Materials-Partially purified phosphodiester a-GlcNAcase was prepared from bovine liver as described previously (1). Methyl a -~mannopyranoside, quaternary aminoethyl Sephadex (QAE-Sephadex), and Lubrol-PX were from Sigma. Uridine diphosphate-[glucosamine,6-3H]N-acetyl-~-glucosamine (UDP-[3H]GlcNAc, 25 Ci/mM) was from American Radiolabeled Chemicals (St. Louis). Scintiverse I was from Fisher. Superose 6 and the protein standards used for gel filtration (thyroglobulin, ferritin, and catalase) were obtained from Pharmacia LKB Biotechnology Inc. Protein standards used in glycerol density gradients were obtained from Sigma. Vibrio cholera neuraminidase was obtained from Calbiochem. Prestained protein standards and nitrocellulose were obtained from Bio-Rad. The following digoxigenin-labeled lectins: Galanthus nivalis agglutinin, Sambu-cus nigra agglutinin, Maakia amurensis agglutinin, peanut agglutinin (PNA), and Datura stramonium agglutinin, as well as alkaline phosphatase-labeled anti-digoxigenin antibody, fetuin, asialofetuin, and blocking reagent were obtained in the glycan detection kit from Boehringer Mannheim. Western Blue was obtained from Promega. Retired breeder BALB/c female mice were obtained from Jackson Laboratories. Vero cells were obtained from ATCC. Brefeldin A was obtained from Sigma and was prepared as a 10 mg/ml stock in ethanol, stored at -20 "C. Labtek four-chamber permanox slides were obtained from Nunc. FITC-labeled donkey anti-mouse IgM and FITC-labeled donkey anti-rabbit IgG were obtained from Jackson Immunoresearch. Texas red-labeled wheat germ agglutinin (WGA) was obtained from EY Laboratories. All other chemicals were analytical grade.
Assay of Bovine Phosphodiester a-GlcNAcase-Phosphodiester a-GlcNAcase was incubated for 1 h at 37 "C in a reaction mixture of 0.03 ml containing 50 mM Tris maleate, pH 6.7, 0.5% (v/v) Triton X-100, and 1 mM [3H]Gl~NA~-a-P-Man-a-Me, and the reaction products were analyzed by ion exchange chromatography on QAE-Sephadex as described previously (5).
Glycerol Density Gradients-Linear 9-40% glycerol gradients of 5 ml with Hz0 or D20 as solvent were prepared and run essentially as described by Walker et al. (6) except that 0.25% (w/v) Lubrol-PX was used in the place of Triton X-100, and 150 mM NaCl was added. Calculations of sedimentation coefficients and molecular weights were carried out as described in Sadler et al. (7).
Gel Filtration Chromatography and Determination of Stokes Radius-Gel filtration chromatography of pooled phosphodiester a-GlcNAcase eluted from WGA-Sepharose was performed using Superose 6 as described previously (1). The Stokes radius of bovine phosphodiester a-GlcNAcase was determined by the method of Ackers (8).

SDS-Polyacrylamide Gel
Electrophoresis-SDS-polyacrylamide gel electrophoresis was carried out following the procedures of Laemmli (9) using either 7.5 or 10% acrylamide gels.
Protein Blotting-Protein resolved on SDS-polyacrylamide gels was transferred onto nitrocellulose following the procedures of Harlow and Lane (10).
Lectin Probe-Purified phosphodiester a-GlcNAcase was blotted onto nitrocellulose, and the filters were incubated overnight at 4 'C in Boehringer Mannheim blocking buffer. Filters were washed twice in Buffer 1 and incubated for 1 h with a digoxigenin lectin from a Boehringer Mannheim glycan detection kit: G. nivalis agglutinin, S. nigra agglutinin, D. stramonium agglutinin, M. amurensis agglutinin, and PNA. The filters were washed three times with Tris-buffered saline and incubated for 1 h with alkaline phosphatase-labeled antidigoxigenin IgG. The filters were washed three times with Trisbuffered saline, and alkaline phosphatase was developed with Western Blue.
Antibodies-Mouse monoclonal antibodies directed against bovine phosphodiester a-GlcNAcase were generated using 3,000-fold purified native phosphodiester a-GlcNAcase as an antigen. Two BALB/c mice were each injected intraperitoneally with 5 pg of phosphodiester a-GlcNAcase in complete Freund's adjuvant. Each mouse was boosted intraperitoneally at 2-week intervals for 6 weeks with 5 pg of enzyme in complete Freund's adjuvant; a final intraperitoneal boost was with 5 pg of enzyme 3 days prior to the fusion. Hybridomas were made, and primary clones were screened for both the ability to immunoprecipitate phosphodiester a-GlcNAcase activity and the ability to recognize partially purified phosphodiester a-GlcNAcase in an enzymelinked immunosorbent assay, performed by the methods of Harlow and Lane (10). Five primary clones were selected and subcloned by limiting dilution, and 5 of the 11 subclones were selected for ascites induction in mice (10). The monoclonal antibody used in this study for immunofluorescence was from subclone 5B6.3, which is an IgM.
Rabbit antibodies directed against native a-mannosidase I1 were the kind gift of Dr. Kelly Moremen.
Immunofluorescence-Vero cells were maintained in medium 199 containing 10% fetal calf serum, 100 units/ml penicillin, 100 pg/ml streptomycin, and 0.25 pg/ml fungizone. Eighteen to 24 h prior to immunofluorescence, a 100-mm confluent plate of Vero cells was split 1:60, 0.1 ml was added to an additional 0.5 ml of medium in a fourwell chamber slide, and the slide was incubated overnight in 5% C02. The cells were washed three times for 5 min with PBS, then fixed/ permeabilized for 2 min with ice-cold acetone. The cells were washed twice with PBS for 5 min and blocked for 1 h with 10% donkey serum/PBS. The slides were incubated for 1 h with either a 1:lOO dilution of mouse anti-phosphodiester a-GlcNAcase 5B6.3 ascites fluid in 10% donkey serum/PBS, a 1:1,000 dilution of rabbit anti-amannosidase I1 in 10% donkey serum/PBS serum, or a 1:50 dilution of Texas red-labeled WGA (WGA-TR) in 10% donkey serum/PBS. The cells were washed three times for 5 min with 10% donkey serum/ PBS and then incubated in secondary antibody for 1 h with shaking (secondary antibody: 1:50 FITC-labeled donkey anti-mouse IgM or 1:50 FITC-labeled donkey anti-rabbit IgG). The cells were washed three times for 5 min with 10% donkey serum/PBS. The slides were overlayed with 50% glycerol (v/v), PBS containing 0.1 M n-propyl gallate, and VWR No. 1 coverslips (24 X 60 mm) were mounted. Indirect immunofluorescence was performed using a 60 X objective lens, and photomicrographs were taken using Kodak Tmax ASA 400 film.
In the colocalization studies, cells were prepared essentially as described above. Cells were incubated for 1 h with a 1:50 dilution of WGA-TR in 10% donkey serum/PBS. The cells were washed three times for 5 min with 10% donkey serum/PBS, incubated for 1 h in a 1:lOO dilution of mouse anti-phosphodiester a-GlcNAcase 5B6.3 ascites fluid, and washed three times for 5 min in 10% donkey/PBS before incubation for 1 h in a 1:50 dilution of FITC-labeled donkey anti-mouse IgM secondary antibody. The cells were washed three times for 5 min in 10% donkey serum/PBS. The slides were overlayed with 50% glycerol (v/v), PBS containing 0.1 M n-propyl gallate, and VWR No. 1 coverslips (24 X 60 mm) were mounted. Confocal microscopy was performed using a MRC-500 scanning laser confocal microscope (Bio-Rad). Photographs of images from video monitors were recorded on Tmax 100 ASA film.

RESULTS
In the accompanying report (1) the general properties, substrate specificity, and kinetic parameters of purified bovine phosphodiester a-GlcNAcase were described, and in this report a number of its physical and hydrodynamic properties as well as its subcellular localization are presented.
Hydrodynamic Analysis of Bovine Phosphodiester a-Glc-NAcase-Phosphodiester a-GlcNAcase has a molecular weight of 129,000 as determined by its migration in nonreducing SDS-polyacrylamide gel electrophoresis (Table I). However, when subjected to gel filtration on Superose 6, the enzyme eluted near the ferritin standard with a M, equivalent of 400,000, corresponding to a Stokes radius of 70.75 A (Fig.  1). The diffusion coefficient (Dzo,w) calculated from these data is 3.07 X cmz/s. Since phosphodiester a-GlcNAcase is a membrane-bound glycoprotein requiring detergent for solubilization, we performed hydrodynamic analysis to determine the amount of detergent that is bound to the enzyme. Bound detergent causes a marked shift in apparent sedimentation in solvents of different densities (11). Fig. 2 shows the results obtained when partially purified phosphodiester a-GlcNAcase was subjected to density gradient centrifugation in 9-40% glycerol in either HzO (panel A ) or DzO (panel B ) . In both gradients, phosphodiester a-GlcNAcase sedimented between the malate dehydrogenase and lactate dehydrogenase standards. The relative sedimentation coefficient of phosphodiester a-Glc-NAcase was 6.4 S in HzO and 5.4 S in DzO. When the values were corrected to H 2 0 at 20 "C, the sedimentation coefficient of phosphodiester a-GlcNAcase was 7.4 S in both gradients. These data indicated that phosphodiester a-GlcNAcase binds  very little detergent, and they were used to calculate a partial specific volume of 0.714 cm3/g for the detergent-protein complex.
A molecular weight for phosphodiester a-GlcNAcase was calculated using the Svedberg equation which takes into account the diffusion coefficient derived from the Stokes radius and the sedimentation coefficient and partial specific volume calculated from the density gradient centrifugation. Using these values the molecular weight obtained for phosphodiester a-GlcNAcase was 204,550 (Table I).
Lectin Blot Analysis of Phosphodiester a-GlcNAcase-During the purification of phosphodiester a-GlcNAcase, it became apparent that the bovine liver enzyme is a glycoprotein (1). The fact that the enzyme bound to concanavalin A-Sepharose indicated that it contained asparagine-linked oligosaccharides, and its binding to WGA-Sepharose indicated that it contained either sialic acids or GlcNAc residues. To define the types of oligosaccharides on phosphodiester a-GlcNAcase, the enzyme was subjected to SDS-polyacrylamide gel electrophoresis, blotted onto nitrocellulose, and the blots were probed with five plant lectins of different carbohydrate binding specificities. As shown in Fig. 3 and Table 11, phosphodiester a-GlcNAcase was reactive with the Galp1,4GlcNAc binding lectin D. stramonium agglutinin and the sialic acid binding lectins M. amurensis agglutinin and S. nigra agglutinin. When the enzyme was digested prior to SDS-polyacrylamide gel electrophoresis with V. cholera neuraminidase, which removes sialic acids, it lost reactivity with M. amurensis agglutinin and S. nigra agglutinin, retained reactivity with D. stramonium agglutinin, and gained reactivity with PNA (Fig.   3). Table I1 shows the carbohydrate binding specificity of the lectins used and presents a summary of the lectin reactivities of phosphodiester a-GlcNAcase. The acquisition of reactivity to PNA, which recognizes Galp1,SGalNAc linkages typical of 0-linked oligosaccharides, indicates that the enzyme contains a sialylated form of the 0-linked disaccharide. The fact that S. nigra agglutinin binds to phosphodiester a-GlcNAcase shows that the enzyme contains sialic acid a2,6GalNAc linkages on its 0-linked chains. G. nivalis agglutinin does not react with phosphodiester a-GlcNAcase, indicating that terminal mannose residues found on high mannose oligosaccharides are not present. D. stramonium agglutinin, which recognizes Galpl,4GlcNAc, an underlying structure in complex type N-linked oligosaccharides, reacts with phosphodiester a-GlcNAcase. However, when phosphodiester a-GlcNAcase was first treated with N-glycanase, to remove the N-linked oligosaccharides, binding of D. stramonium agglutinin was abolished (data not shown). Phosphodiester a-GlcNAcase is bound by M. amurensis agglutinin, which recognizes sialic acid a2,3Gal residues on complex type N-linked and perhaps 0-linked oligosaccharides. Taken together these data suggest that phosphodiester a-GlcNAcase contains complex type asparagine-linked oligosaccharides and 0-linked oligosaccharides as shown in Table 11. These data are in agreement with the lectin affinities observed during purification of bovine phosphodiester a-GlcNAcase (1).
Bovine Phosphodiester a-GlcNAcase Is Localized to the Golgi Apparatus-To define more precisely the intracellular localization of phosphodiester a-GlcNAcase, we used a mouse monoclonal antibody directed against the native enzyme in immunofluorescence studies. Table I11 shows that the mouse monoclonal antibody 5B6.3, which is an IgM, immunoprecipitated phosphodiester a-GlcNAcase activity, whereas an irrelevant mouse monoclonal antibody did not. In both human fibroblasts (data not shown) and Vero cells, phosphodiester a-GlcNAcase is localized in a juxtanuclear region of the cells as shown in the lower right panel of Fig. 4 Fig. 4 shows that a-mannnosidase 11, an enzyme of the medial-Golgi, has a similar juxtanuclear localization in Vero cells.
WGA binds to highly sialylated glycoproteins present in the trans-Golgi and trans-Golgi network (TGN). WGA-TR staining in Vero cells is juxtanuclear but has a much tighter localization than phosphodiester a-GlcNAcase as shown in the leftpanels of Fig. 5 . In cells labeled for both phosphodiester a-GlcNAcase and WGA-TR, there are areas of overlap, but the staining of WGA-TR and the localization of phosphodies- Fraction ter a-GlcNAcase are not identical as shown in the bottom left panel of Fig. 5 .
The right panels of Fig. 5 show the results obtained when cells were exposed to 10 pg/ml brefeldin A (BFA) for 15 min before fixation and permeabilization. Phosphodiester a-GlcNAcase disperses into an ER pattern after BFA treatment consistent with its localization in the Golgi stacks (12). WGA staining in BFA-treated Vero cells is less dispersed than the staining seen for phosphodiester a-GlcNAcase. In cells both labeled for phosphodiester a-GlcNAcase and stained for  TABLE 111 Immunoprecipitation of phosphodiester a-GlcNAcase activity Reaction mixtures containing 4 pl of phosphodiester a-GlcNAcase activity (0.19 nmol/h) and 100 pl of tissue culture supernatant were incubated on ice for 30 min. Rabbit anti-mouse secondary antibody (2 pl) was then added to the immune complex and incubated for 30 min. The entire complex was precipitated using staphylococcal A cells, and 3 X washed pellets were directly assayed for phosphodiester a-GlcNAcase activity.

DISCUSSION
The hydrodynamic studies of phosphodiester a-GlcNAcase have revealed that this membrane-bound glycoprotein has a calculated molecular weight of 204,550 as derived from the Svedberg equation. Many detergent-solubilized membrane proteins are isolated as protein-detergent complexes containing substantial amounts of detergent which causes the complex to have a high partial specific volume compared with standard globular proteins. Therefore, we subjected the bovine liver phosphodiester a-GlcNAcase to density gradient centrifugation in both D20 and H20 to derive values for the partial specific volume, 17, and the sedimentation coefficient s~, , ,~ of the enzyme-detergent complex. If a protein-detergent  (+) or not (-) with BFA (10 pg/ml) for 15 min a t 37 "C, the cells were fixed and permeabilized with ice-cold acetone, and phosphodiester a-GlcNAcase was detected using antiphosphodiester a-GlcNAcase monoclonal antibody 5B6.3 and FITCconjugated donkey anti-mouse IgM secondary antibody, as described under "Experimental Procedures." Control panels were treated only with secondary antibody. Other cells, prepared as described under "Experimental Procedures," were stained using WGA-TR. Cells in the two bottompanels were colabeled for phosphodiester a-GlcNAcase and WGA-TR. FITC staining is shown on the left; Texas red staining is shown on the right. The fluorescence was viewed by confocal microscopy. complex contains substantial detergent it will sediment to a "lighter" position of a D20 gradient compared with its position on an H20 gradient relative to standard proteins (11). The phosphodiester a-GlcNAcase sedimented in the same position relative to the standard proteins in both the D20 and H20 gradients indicating that the enzyme binds little detergent, and the value of its sedimentation coefficient was the same in both gradients when corrected to water a t 20 "C. The value for the partial specific volume of phosphodiester a-GlcNAcase calculated from density gradient centrifugation is 0.7142 cm3/ g. This value is similar to the partial specific volumes of the protein standards used in the density gradient centrifugation which range from 0.72-0.75 cm3/g and much lower than the partial specific volume of 0.958 cm3/g for the detergent Lubrol-PX.
The Stokes radius for phosphodiester a-GlcNAcase was determined using gel filtration of the enzyme in comparison with globular protein standards of known molecular weigbt and Stokes radius. The enzyme has a Stokes radius of 70.7 A, a value that is inconsistent for a globular protein with a molecular weight of 204,550, which should have a Stokes radius of approximately 45 A. The frictional ratio ( f / f~ min), calculated for phosphodiester a-GlcNAcase from the values for its molecular weight, partial specific volume, and Stokes radius, is 1.8, which indicates that the molecule is significantly more asymmetric than typical globular proteins which have frictional ratios of 1-1.3. Therefore, we propose that phosphodiester a-GlcNAcase is not globular, but instead has an extended structure, the exact nature of which cannot be elucidated from the data given here.
In contrast to the molecular weight of 204,550 derived from hydrodynamic measurements, enzymatically active phosphodiester a-GlcNAcase migrates on SDS-polyacrylamide gel electrophoresis like a protein of molecular weight 129,000. This observation suggests that native enzyme may be a dimer that dissociates into monomers in SDS. Since the enzyme eluted from the SDS-gels is active, either the monomer is enzymatically active or may be able to reform dimers under the elution conditions used to extract enzymatic activity. We are unable to distinguish between these two possibilities. It is interesting that Ben-Yoseph et al. (13), using the method of radiation inactivation, report a molecular size of 129 kDa for human placental phosphodiester a-GlcNAcase.
In the purification of bovine liver phosphodiester a-Glc-NAcase, enzyme molecules with carbohydrates containing asparagine-linked oligosaccharides and sialic acid were selected because of the binding affinities of concanavalin A and WGA. Those enzyme molecules that do not bind to these lectins because they lack the necessary carbohydrate epitopes constitute 26-40% of the total enzyme activity in different preparations (1). The lectin binding experiments allowed us to examine the carbohydrate composition of the glycosylated phosphodiester a-GlcNAcase further. The enzyme bound not only to concanavalin A, but also to M. amurensis agglutinin and D. stramonium agglutinin, although not G. nivalis agglutinin, a finding consistent with the presence of complex type N-linked oligosaccharides. The enzyme also bound WGA and S. nigra agglutinin and, after treatment with neuraminidase, bound to PNA. Taken together these data suggest the presence of 0-linked oligosaccharides on phosphodiester a-GlcNAcase. Although, as pointed out in Table 11, the lectin blot experiments cannot provide complete oligosaccharide structure determination, the specificity of the lectins unequivocally demonstrated the presence of sialic acid and galactose on the enzyme. Since phosphodiester a-GlcNAcase has sialic acid a2,3Gal on N-and possibly 0-linked oligosaccharides and sialic acid a2,6GalNAc residues on 0-linked oligosaccharides, it presumably has been acted upon by galactosyl and sialyltransferases which are localized to the trans-Golgi and TGN.
The acquisition of terminal glycosylation by residents of the cis-and medial-Golgi has been reported. GIMP,, a cis-Golgi marker (14), MG-160 (15), and a-mannosidase I1 (16), both medial-Golgi markers, all have sialic acid residues on their N-linked carbohydrates. However, all resident Golgi glycoproteins do not undergo terminal glycosylation. Mannosidase IA, concentrated in the medial-Golgi, and p58, a cis-Golgi marker, lack terminal processing by sialyltransferase or galactosyltransferase (17,18). Furthermore, the glycosylation of Golgi membrane glycoproteins may vary from cell type to cell type. a-Mannosidase 11, a medial-Golgi processing enzyme, contains complex N-linked oligosaccharides and 0linked oligosaccharides in 3T3 cells but only N-linked oligosaccharides in HeLa cells (16). Assuming that phosphodiester a-GlcNAcase is localized to the &/medial-Golgi, it is formally possible that the terminal processing of its oligosaccharides occurs as the newly synthesized processing enzymes are traversing the Golgi. Alternatively, a percentage of phosphodiester a-GlcNAcase may cycle through the Golgi stacks and be retrieved in the trans-Golgi or TGN, while maintaining residence in the ck/mediaGGolgi.
Immunofluorescence studies using a monoclonal antibody directed against phosphodiester a-GlcNAcase have allowed us to analyze the intracellular location of this enzyme. In Vero cells, phosphodiester a-GlcNAcase is concentrated in a perinuclear "Golgi-like'' structure. A very similar staining pattern was observed with a-mannosidase 11, a medial-Golgi marker. In the presence of BFA, phosphodiester a-GlcNAcase disperses into an ER pattern, consistent with the finding that BFA causes the Golgi cisternae to disperse into the ER pattern (12,19). In contrast WGA staining is less dispersed in the presence of BFA, and the staining is clearly distinct from that obtained for phosphodiester a-GlcNAcase. Recently, others have reported the formation of a hybrid network of TGN and early endosomes in cells treated with BFA (20-22), a network distinguishable from that formed by Golgi stack proteins. WGA stains heavily sialylated glycoproteins of the trans-Golgi and TGN and most likely accumulates in this hybrid network of TGN and early endosomes after treatment with BFA. In contrast, we believe phosphodiester a-GlcNAcase is most likely a resident of the Golgi stacks because, after treatment with BFA, it distributes to an "ER-like" compartment, distinct from the hybrid network of TGN and endosomes. Interestingly, Sampath et al. (23) found that cells treated with 0.1 pg/ml BFA showed an 80% decrease in their phosphorylated oligosaccharides (derived from lysosomal enzymes), and the majority of the phosphorylated species contained phosphodiesters, i.e. were not uncovered. In an earlier study, Radons et al. (24) found that the phosphorylation of oligosaccharides on the lysosomal enzyme cathepsin D was diminished in BFA-treated cells and that the uncovering of the phosphate groups was abolished. Both studies indicate that phosphodiester a-GlcNAcase is relatively inactive on oligosaccharides of lysosomal enzymes in BFA-treated cells, and both groups suggest that phosphodiester a-GlcNAcase is, therefore, in a compartment distal to the BFA block. However, since our data show that the enzyme is dispersed into the ER in the presence of BFA, its low activity may be caused by dilution or unfavorable conditions in that compartment. To identify the intracellular localization of phosphodiester a-GlcNAcase precisely it will be necessary to perform immunoelectron microscopy using antibodies directed against phosphodiester a-GlcNAcase.