Characterization of the endomannosidase pathway for the processing of N-linked oligosaccharides in glucosidase II-deficient and parent mouse lymphoma cells.

Studies on N-linked oligosaccharide processing in the mouse lymphoma glucosidase II-deficient mutant cell line (PHAR2.7) as well as the parent BW5147 cells indicated that the former maintain their capacity to synthesize complex carbohydrate units through the use of the deglucosylation mechanism provided by endomannosidase. The in vivo activity of this enzyme was evident in the mutant cells from their production of substantial amounts of glucosylated mannose saccharides, predominantly Glc2Man; moreover, in the presence of 1-deoxymannojirimycin or kifunensine to prevent processing by mannosidase I, N-linked Man8GlcNAc2 was observed entirely in the form of the characteristic isomer in which the terminal mannose of the alpha 1,3-linked branch is missing (isomer A). In contrast, parent lymphoma cells, as well as HepG2 cells in the presence of 1-deoxymannojirimycin accumulated Man9GlcNAc2 as the primary deglucosylated N-linked oligosaccharide and contained only about 16% of their Man8GlcNAc2 as isomer A. In the presence of the glucosidase inhibitor castanospermine the mutant released Glc3Man instead of Glc2Man, and the parent cells converted their deglucosylation machinery to the endomannosidase route. Despite the mutant's capacity to accommodate a large traffic through this pathway no increase in the in vitro determined endomannosidase activity was evident. The exclusive utilization of endomannosidase by the mutant for the deglucosylation of its predominant N-linked Glc2Man9GlcNAc2 permitted an exploration of the in vivo site of this enzyme's action. Pulse-chase studies utilizing sucrose-D2O density gradient centrifugation indicated that the Glc2Man9GlcNAc2 to Man8GlcNAc2 conversion is a relatively late event that is temporally separated from the endoplasmic reticulum-situated processing of Glc3Man9GlcNAc2 to Glc2Man9GlcNAc2 and in contrast to the latter takes place in the Golgi compartment.

Studies on N-linked oligosaccharide processing in the mouse lymphoma glucosidase 11-deficient mutant cell line (PHAR2.7) as well as the parent BW6147 cells indicated that the former maintain their capacity to synthesize complex carbohydrate units through the use of the deglucosylation mechanism provided by endomannosidase. The in vivo activity of this enzyme was evident in the mutant cells from their production of substantial amounts of glucosylated mannose saccharides, predominantly GlczMan; moreover, in the presence of 1-deoxymannojirimycin or kifunensine to prevent processing by mannosidase I, N-linked MansGlcNAcz was observed entirely in the form of the characteristic isomer in which the terminal mannose of the al,3-linked branch is missing (isomer A ) . In contrast, parent lymphoma cells, as well as HepG2 cells in the presence of 1-deoxymannojirimycin accumulated ManeGlcNAcz as the primary deglucosylated Nlinked oligosaccharide and contained only about 16% of their MansGlcNAcz as isomer A. In the presence of the glucosidase inhibitor castanospermine the mutant released GlcjMan instead of GlczMan, and the parent cells converted their deglucosylation machinery to the endomannosidase route. Despite the mutant's capacity to accommodate a large traffic through this pathway no increase in the in vitro determined endomannosidase activity was evident.
The exclusive utilization of endomannosidase by the mutant for the deglucosylation of its predominant Nlinked GlczManeGlcNAcz permitted an exploration of the in vivo site of this enzyme's action. Pulse-chase studies utilizing sucrose-DzO density gradient centrifugation indicated that the GlczManeGlcNAcz to MansGlcNAcz conversion is a relatively late event that is temporally separated from the endoplasmic reticulum-situated processing of GlcaMansGlcNAcz to GlczManeGlcNAcz and in contrast to the latter takes place in the Golgi compartment.
On the basis of investigations carried out in a number of laboratories, a general outline for the biosynthesis of complex * This work was supported by National Institutes of Health Grant DK17477. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. N-linked carbohydrate units has emerged in recent years in which a cotranslational attachment of a glucosylated polymannose oligosaccharide (Glc3Man9GlcNAcz) is followed by a succession of processing reactions involving ER'-and Golgisituated glycosidases and glycosyltransferases (1,2). Although this established scheme dictates that removal of the 3 glucose residues, a prerequisite for complex oligosaccharide formation, is accomplished by the stepwise action of glucosidase I and I1 in the ER, it does not accommodate the frequently reported observation that the formation of these carbohydrate units persists to a substantial extent in the presence of glucosidase inhibitors (3-10). This quandary was resolved with the discovery in our laboratory of an endomannosidase which can effect deglucosylation by cleaving the linkage between the glucose-substituted mannose and the remainder of the polymannose oligosaccharide (11,12). It has been demonstrated that this enzyme, which is unaffected by agents inhibiting glucosidase action (11,12), can circumvent glucosidase blockade and thereby make possible the continued formation of complex N-linked oligosaccharides (13).
In the present investigation we have further pursued our exploration of the endomannosidase-initiated processing pathway by focusing our attention on a mutant mouse lymphoma cell line (PHAR2.7) described by Reitrnan et al. (14) which despite a deficiency in glucosidase I1 produces N-linked complex oligosaccharides. We have obtained evidence in support of our original proposal relative to these cells (11,12) that the persistent biosynthesis of complex carbohydrate units in this mutant is made possible by the alternate glucoseremoval mechanism provided by the endomannosidase pathway. By comparing and quantitating the relevant processing intermediates of the glucosidase-deficient cells with those of the parent lymphoma, it has been possible to evaluate to what extent this route functions under normal conditions. Furthermore, the mutant cells provided us with a unique opportunity to determine at what stage in the processing sequence and in which subcellular compartment the endomannosidase-mediated deglucosylation takes place in the in vivo situation. chased from ATCC, Rockville, MD. The suspended cells were grown in Dulbecco's modified Eagle's medium containing glucose (4.5 g/ liter), fetal calf serum (10% GIBCO), penicillin (100 units/ml), and streptomycin (100 pg/ml) at 37 "C in an atmosphere of 95% 02, 5% CO, as previously specified (15). The cells were harvested by centrifugation (140 X g for 5 min) and washed two times with glucose-and serum-free Dulbecco's modified Eagle's medium containing 5 mM sodium pyruvate, and radiolabeling was then accomplished by incubating 2.5 X lo' cells in 2 ml of this medium with either 180 pCi of [2-;H]mannose (30 mcilpmol) or 170 pCi of D-[U-'4C]glUCOSe (304 pCl/pmol, Du Pont-New England Nuclear) while being subjected to mild agitation on a rocker platform (Bellco, Vineland, NJ). When the glycosidase inhibitors CST (a gift from Dr. M. Kang, Merrell Dow Research Institute, Cincinnati, OH) and DMJ (purchased from Genzyme) were used, the cells were preincubated with these agents (2 mM) for 30 min before addition of the radiolabeled sugar. In the incubations used for subcellular fractionation, 1 X 1 0 ' cells were pulse labeled for 20 min with 330 pCi of [2-3H]mannose in 1 ml of the glucose-free Dulbecco's modified Eagle's medium containing 20 mM sodium pyruvate, and the chase was then initiated by the direct addition of 2 ml of the medium used for cell growth in which 37.5 mM mannose was included; throughout these pulse-chase studies as well as in a 30-min preincubation, kifunensine (kindly provided by M. Shirahashi of the Fujisawa Pharmaceutical Co. Ltd., Oska, Japan) at a concentration of 0.1 mM was present.
Subcellular Fractionation-After pulse-chase radiolabeling of PHAR2.7 cells, they were separated from the medium and were washed with ice-cold 50 mM HEPES, buffer, pH 7.4, containing 0.15 M NaC1. The cell pellets were suspended in 10 mM HEPES, pH 7.4, 0.25 M sucrose and disrupted by multiple passages of a tight fitting Dounce homogenizer until greater than 90% of the cells had been broken. Postnuclear supernatants (1 ml), prepared by centrifugation of the homogenates at 450 X g for 15 min, were layered on top of 10-50% sucrose in D20 gradients as described by Lodish et al. (16). Centrifugation was carried out at 4 "C for 3 h at 164,000 X g in Beckman SW 40 Ti rotor. The gradients were then separated into 0.9-ml fractions with an ISCO gradient collector; sucrose concentrations were determined with an ABBE-3L refractometer (Baush and Lomb, Rochester, NY).
Extraction of Radiolabeled Cells and Density Gradient Fractwns-After radiolabeling, the lymphoma cells were separated from the medium by centrifugation (140 X g for 5 min) and then extracted in a mixture (12 ml) of chloroform/methanol/uffer (3:2:1) as previously described (13) to obtain an upper phase containing the soluble oligosaccharides and an interphase fraction; the latter, after a water wash, was further extracted with chloroform/methanol/water (10103) to yield an oligosaccharide-lipid fraction and a delipidated protein pellet (17).
Fractions from the sucrose density gradients were treated in the same manner except that all volumes were halved and that carrier protein (2 mg of thyroid membranes) was added prior to the organic solvents.
Preparation of Free Oligosaccharides-After deproteinization of the media with trichloroacetic acid and removal of organic solvents from the upper phase of the cell extracts, oligosaccharides from these two sources were obtained by charcoal-Celite chromatography (30% ethanol eluate) subsequent to removal of salt by passage through Dowex 50 (H+ form) and Dowex 1 (acetate form) as previously described (13).
Preparation of Glycopeptides and Endo H Digestion-Glycopeptides were prepared in a manner previously reported (13) after Pronase digestion of the delipidated protein from the cell extracts and the sucrose density fractions as well as the ether-extracted trichloroacetic acid precipitates from the media. Since more than 90% of the radiolabeled glycopeptides were found to be associated with the cells, medium and cellular glycopeptides were pooled for most of the analyses. Polymannose oligosaccharides were released from the glycopeptides by endo H digestion (4 milliunits, Genzyme); they were desalted and separated from the endo H-resistant glycopeptides by chromatography on columns on Dowex 50 and Dowex 1 and further resolved by thin layer chromatography (13).
Endomannosidase and Glucosidase II Assays-In preparation for the enzyme assays the cells (3-6 X 10') were washed with phosphatebuffered saline, suspended in 4 volumes of 100 mM sodium phosphate buffer, pH 7.0, containing 0.5 mM dithiothreitol (14), and disrupted at 4 "C with 3 X 5-9 bursts of a Branson sonifier (setting No. 1). The resulting homogenates were separated by centrifugation (100,000 X g for 60 min) into a clear supernatant and a membrane pellet; the latter, after a washing with the homogenization buffer, was suspended therein at a protein concentration of 10 mg/ml. Both the mutant and parent cell line contained 50-60 pg of protein/106 cells, and in both cases about 55% of the total protein was present in the membrane fraction.
For the glucosidase I1 and endomannosidase assays, "C-labeled GlclMangGlcNAc (10,000 dpm), prepared in a manner previously described ( l l ) , was used as substrate in a 100-pl volume. For the glucosidase assay the reaction mixture also contained 100 mM sodium phosphate, pH 7.0,0.25% Triton X-100, and 0-50 pg of protein while the endomannosidase activity was measured in 100 mM NaMES buffer containing 1 mM CST, 10 mM EDTA, 0.25% Triton X-100, and 0-500 pg of protein; as previously indicated for the latter assay (111, an incubation at 2 "C preceded the addition of substrate. Both assays were conducted for 45 min at 37 "C at which time the incubation mixtures were deproteinized and desalted by the procedure previously described (11). The products of the enzyme reactions, namely glucose or the disaccharide Glcal+3Man, were separated by thin layer chromatography in Solvent System A and quantitated by scintillation counting after fluorographic visualization and elution from the plates (11).
Structural and Analytical Procedures-Oligosaccharides released by endo H and resolved by thin layer chromatography were coupled to 2-aminopyridine (Aldrich Chemical Co.) by the procedure of Hase et al. (18) and subsequently purified on columns of Bio-Gel P-2 (13). Reduction of oligosaccharides was carried out with NaBH, as previously described (11) while acetolysis was performed by the procedure of Varki and Kornfeld (19); radiolabeled standards for the characterization of the acetolysis fragments were prepared in a manner previously reported (11). Oligosaccharides were treated with rat liver endoa-mannosidase (100 pg of Golgi membrane protein) for 48 h at 37 "C in the presence of CST and EDTA (13), and after removal of salt and protein (12) the products were examined by thin layer chromatogra-Quantitation of radiolabeled neutral sugars was carried out after acid hydrolysis (1 N HC1,5 h at 100 "C under nitrogen). After passage of the hydrolysates through coupled columns of Dowex 50 (H+ form) and Dowex 1 (acetate form) (20), the components were separated by thin layer chromatography, and the radioactivity was determined by scintillation counting after elution. When oligosaccharides from the mutant were radiolabeled with ['4C]glucose the relative specific activities of their glucose and mannose residues was determined from a hydrolysate of purified Glc2Man9GlcNAc prepared by endo H digestion of glycopeptides from these cells. Protein was determined by the procedure of Bradford (21) using bovine serum albumin as a standard.
Reverse-phase HPLC of desalted pyridylamino-derivatives of polymannose oligosaccharides was performed by a modification of the procedure of Hase et al. (22). Samples (-10,000 dpm "C-label) were applied to a LiChrosphr 100 RP-18 column (4.6 X 250 mm, 5 pm, Merck) and eluted with the same buffer at a flow rate of 0.4 ml/min. Monitoring of the column effluent was performed with a model 171 radioisotope detector (Beckman Instruments) as previously described (13). To each sample a pyridylamine derivative of [2-3H]mannoselabeled GlclMangGlcNAc was added to serve as an internal standard.
Thin layer chromatographic resolution of monosaccharides and small oligosaccharides was achieved on plastic sheets precoated with cellulose (0.1-mm thickness, Merck) for 6 h in pyridine/ethyl acetate/ water/acetic acid, 5:5:3:1 (Solvent System A). Separation of larger oligosaccharides was carried out on plastic sheets precoated with Silica Gel-60 (0.2-mm thickness, Merck) for 24 h in 1-propanol/acetic acid/water, 3:3:2 (Solvent System B). For the separation of hexoses from hexitols, chromatography was undertaken for 20 h on cellulosecoated plates in nitromethane/acetic acid/ethanol/water saturated with boric acid, 8l:l:l (Solvent System C). The chromatography in all systems was carried out with a wick of Whatman 3 " paper clamped to the top of the thin layer plates. The components were detected by fluorography and quantitated by scintillation counting after elution with water as previously described (13 purposes, the oligosaccharide eluates were extracted with peroxidefree ether to remove scintillants and passed through small coupled columns of Dowex 50 (H+) and Dowex 1 (acetate). The preparation of radiolabeled oligosaccharide standards has been previously reported (11)(12)(13). Radioactivity Measurements-Liquid scintillation counting was carried out in Ultrafluor with a Beckman LS 7500 instrument. Components on thin layer plates were detected by fluorography at -70 "C with X-Omatic AR film (Eastman) after spraying with a scintillation mixture containing a-methylnaphthalene (23).  Table I).

Evaluation of N-linked Complex
Complex oligosaccharide formation was also carried out by the glucosidase 11-deficient mutant as previously reported (14), but this was of a lower level than in the parent cells and was unaffected by the presence of CST (Table I). These results indicate that lymphoma cells can utilize a glucosidase-independent processing pathway for the formation of N-linked complex oligosaccharides.
Identification of Endomannosidase-generated Saccharides-After incubation with [2-'H]mannose, thin layer chromatographic examination of the combined intracellular and medium-free oligosaccharides from CST-treated parent lymphoma cells revealed components which migrated to the positions of the characteristic di-, tri-, and tetrasaccharide products of endomannosidase action (12) with the predominance of Glc'Man; these components were not present in the absence of the glucosidase inhibitor (cf. lanes 1 and 3, Fig. 1). In mutant cell incubations carried out without CST, a prominent trisaccharide was evident which comigrated with GlcPMan (lane 2, Fig. l), but in the presence of the inhibitor the amount of this component was greatly reduced with the concurrent appearance of an intense spot which moved to the position of Glc'Man (lane 4, Fig. 1). The slower moving components observed in all the lanes (Fig. 1) belong to the previously observed polymannose-GlcNAcl-2 series (25); in the glucosidase-deficient mutant and the CST-treated parent cells, these oligosaccharides remained closer to the origin due their larger size.  Radioactivity present in the total glycopeptides from the incubation; more than 90% of the radiolabeled glycopeptides synthesized by the two cell lines were associated with the cells. The identity of the trisaccharide produced by the lymphoma mutant was confirmed following isolation of this component by preparative thin layer chromatography from [I4C]glucoselabeled cells. While acid hydrolysis of the native trisaccharide yielded only glucose and mannose, the latter was released as mannitol when the NaBH,-reduced oligosaccharide was analyzed (data not shown); these observations as well as the ratio of glucose to mannose of 1.8 (after correction for differences in specific radioactivities) are consistent with a GlcpMan structure (12).
Characterization of N-Linked Polymannose Oligosaccharides-Chromatographic examination of the endo H-released oligosaccharides from parent lymphoma cells revealed MangGlcNAc and ManRGlcNAc as the predominant components (lane I , Fig. 2). In the mutant cells three oligosaccharide species (designated a-c) were evident (lane 2, Fig. 2), which on the basis of their chromatographic migration and the products released by in vitro endomannosidase treatment (Fig.  3) were identified as GlczMan9GlcNAc, GlcnMansGlcNAc, and Glc2Man7GlcNAc, respectively, intermixed with small amounts of monoglucosylated components (GlclMangGlcNAc and GlclManRGlcNAc).
To inhibit further processing of deglucosylated polymannose units which would be formed in the mutant cells by endomannosidase action, DMJ, a mannosidase I inhibitor (24), was added to the incubations. In the presence of this agent, oligosaccharides appeared among the endo H-released components of the mutant cells which migrated to the position of MansGlcNAc and ManTGlcNAc (f and g in lane 4, Fig. 2.) and which were distinguished from glucosylated species by their resistance to in vitro digestion with endomannosidase (Fig. 3). However, addition of DMJ to parent cell incubations resulted in a pronounced accumulation of MangGlcNAc ( l a n e 3, Fig. 2), a component, which consistent with a glucosidase I1 deficiency, was not present in the mutant cell line. Indeed a quantitation of the various endo H-released oligosaccharides formed by the mutant in the presence of DMJ indicated that a substantial portion (-35%) was present as nonglucosylated species among which, in marked contrast to the parent cells, MangGlcNAc was almost completely absent (Table 11); the latter component was also missing when CST was added to DMJ-treated parent cells (data not shown).
Characterization of Man8GlcNAc Isomers-Since the action of endomannosidase on glucosylated MangGlcNAc leads to the formation of a characteristic ManRGlcNAc variant (isomer A ) , in which the terminal mannose of the al,3-linked branch is missing (ll), an examination of this oligosaccharide should provide information in regard to the processing routes employed by the parent and mutant lymphoma cells. While HPLC analysis of the Man8GlcNAc from the parent cells indicated that isomer B, in which the terminal mannose from the middle branch of the polymannose unit is missing (11,13) was almost exclusively present, essentially only isomer A was detected in the mutant (Fig. 4, Table 111). Addition of DMJ to the parent cell incubations led to a substantial increase in the proportion of isomer A (Fig. 4, Table 111) indicating that this component is more rapidly processed in vivo by mannos- Endo H-released components a-g (5000 dpm) obtained from CONTROL and DMJ-treated mutant lymphoma cells (see Fig. 2) were digested with Golgi rat liver endomannosidase under conditions described under "Experimental Procedures" and then submitted to chromatography on a cellulose-coated plate for 6 h in Solvent System A. The abbreviations for the Glcl-zMan standards shown to the left of the chromatogram are the same as in Fig. 1. The oligosaccharide composition of each component as determined from the endomannosidase products is indicated above each lane, and the system of abbreviations is the same as in Fig. 2. The radioactive material evident at the origins primarily represents the polymannose-GlcNAc product of endomannosidase action.  Fig. 3. The endo H-released oligosaccharides from pooled cellular and medium glycoprotein were separated by preparative thin layer chromatography in Solvent System B.
bThe molar distribution of oligosaccharides was determined by scintillation counting of the eluted components. The extent of glucosylation of the slower components was determined by endomannosidase digestion (Fig. 4); under the conditions of digestion employed monoglucosylated species were completely hydrolyzed to yield GclMan while diglucosylated oligosaccharides were cleaved to the extent of 47% to yield GlczMan. The values for glucosylated oligosaccharides were furthermore corrected for the experimentally determined difference in specific activity of the glucose and mannose residues (Glc/Man = 1.3) as described under "Experimental Procedures.'' Sum of diglucosylated components; primarily Glc2MangGlcNAc in mutant while below detection (-) in parent cells.
Sum of monoglucosylated components; primarily GlclMang-GlcNAc in mutant as well as parent cells.  Fig. 2, endo H-released oligosaccharides which migrated to the position of MansGlcNAc on thin layer chromatography were derivatized with 2-aminopyridine and analyzed (10,000 dpm) on a LiChrosphr 100 RP-18 column as described under "Experimental Procedures"; radioactivity was determined by scintillation counting with a flow-through detector. The position of elution of the three isomers of ManaGlcNAc have been designated by the letters A, B, and C; these have been shown to have a mannose residue missing on the al,3-linked, middle and cul,6-linked chain of the polymannose unit, respectively (13). The pyridylamino derivative of [3H]mannoselabeled GlclMan9GlcNAc which was used as an internal standard emerged from the column at 75 min.

. Separation by HPLC of the pyridylamino derivatives of MansGlcNAc isomers from parent and mutant lymphoma cells. After incubation of the cells with ['4C]glucose under conditions specified in
idase I than the B isomer which would also account for the absence of Man,GlcNAc in the mutant when this agent was not present (Fig. 2). Indeed when the mannosidase inhibitor was added to HepG2 cells, this increase in the proportion of isomer A was also evident especially in the secreted glycoproteins (Table 111).
Acetolysis studies on the MansGlcNAc from mutant and parent cells were in agreement with the HPLC analyses. While the reduced oligosaccharide from mutant cells, yielded only products characteristic of isomer A, namely mannobiose, mannotriose, and Man3GlcNAcHn, the fragments obtained from the Man,GlcNAcH, of parent cells reflected the presence inhibitor, kifunensine, (26)" to prevent'further processing of  20 80 Cells were radiolabeled with ['4C]glucose in the absence (-) or presence of DMJ (2 mM) under conditions described in Fig. 3, and the endo H-released oligosaccharides which migrated to the position of MansGlcNAc on thin layer chromatography were analyzed by HPLC after derivatization (12,000 dpm) with 2-aminopyridine (see Fig. 5). The oligosaccharides from the lymphoma cells were derived from pooled cellular and medium glycoproteins while in the HepG2 cells they were separately examined as indicated by the term in the parentheses.
* Isomers A and B have a mannose residue missing on the d , 3linked and middle chain of the polymannose residue, respectively; the third isomer, C, was not detected (Fig. 5).
Calculated from the radioactivity in the HPLC-separated components.
of DMJ.
Insufficient ManaGlcNAc for analysis were found in the absence of both isomers B and A with the latter becoming more prominent when DMJ was added to the incubations (Fig. 5 ) .
Quantitation of the Endomannosidase Processing Pathway in Mutant and CST-treated Parent Lymphoma Cells-Upon incubation with [2-3H]mannose, a close correlation was observed in both types of cells between the sum of the distinctive endomannosidase-derived di-, tri-, and tetrasaccharides and the total molar equivalents of deglucosylated N-linked oligosaccharides among which complex carbohydrate units predominated (Table IV). From these data it becomes evident that endomannosidase can provide a route in lymphoma cells by which a glucosidase deficiency or blockade can be circumvented.
I n Vitro Assessment of Endomannosidase Actiuity in Mutant and Parent Cells-In order to determine if the mutant cell line compensated for its glucosidase I1 deficiency by increasing its expression of endomannosidase, enzyme analyses were carried out on its soluble and membrane fractions and compared to those from parent cells. These studies clearly indicated that while glucosidase I1 activity was, as anticipated (14), essentially absent in the mutant, the level of endomannosidase was similar to that observed in the parent cells (Fig.  6). Furthermore, it became evident that endomannosidase was predominantly associated with the membrane fraction in contrast to glucosidase I1 which was recovered to a substantial extent (-50%) in the soluble portion of the parent cells (Fig.  6); the total in vitro glucosidase I1 activity toward the GlclMan9GlcNAc substrate was calculated to be about 50 times as great as that of endomannosidase.
Subcellular Localization of Endomannosidase Action in the in Viuo State-The glucosidase 11-deficient mutant cells, which apparently achieve deglucosylation of their N-linked oligosaccharides exclusively by the action of endomannosidase, provided an attractive system for exploring the subcellular locale in which this enzyme functions since all the MansGlcNAc produced is in the form of the A isomer. When pulse-chase incubations were carried out with the mutant in the Dresence of the recentlv described Dotent mannosidase I deglucosylated N-linked carbohydrate units, thin layer chromatographic examination of the endo H-released oligosaccharides clearly revealed that the conversion of Glc3Man9GlcNAc to GlcPMansGlcNAc is essentially complete at a time when the formation of ManRGlcNAc is just beginning (Fig. 7). Resolution of the postnuclear supernatants of the radiolabeled cells on the sucrose-D20 gradient described by Lodish et al. (16) indicated that ManRGlcNAc-containing glycoproteins, which become evident only after a 90-min chase, are primarily located in two peaks which correspond in density to Golgi vesicles (16), in contrast to the oligosaccharide-lipids which are situated in the heavier ER membranes (Fig. 8). Due to the presence of the mannosidase inhibitor, the N-linked carbohydrate units remained endo H susceptible throughout the pulse-chase study, and thin layer chromatographic examination of the oligosaccharides released by this enzyme indicated that the ratio of ManRGlcNAc to its GlcZMansGlcNAc pre- ' The values represent the radioactivity present in the component from the combined cells and medium; figures in parentheses represent the molar equivalent radioactivity which was calculated by dividing the dpm in the oligosaccharide by the number of mannose residues which it contains. 'The tri-, di-, and monoglucosylated mannose saccharides were assayed by scintillation counting after elution from thin layer chromatographs (see Fig. 1).
-, indicated that component could not be detected. e Deglucosylated polymannose oligosaccharides (Man5.7GlcNAc) were resolved by thin layer chromatography after their release from glycopeptides by endo H digestion and quantitated by scintil1atic.n counting; no detectable amounts of Man9GlcNAc or MansGlcNAc were present.
' Complex oligosaccharides (multiantennary plus biantennary) were resolved by concanavalin A-Sepharose chromatography as described under "Experimental Procedures"; correction has been made for the radiolabeled fucose present in these oligosaccharides as determined by thin layer chromatography after acid hydrolysis.
The molar equivalent radioactivity of complex oligosaccharides was calculated on the basis of 3 mannose residues/carbohydrate unit. cursor was highest in the lighter membrane fractions of the gradient; a plot of this ratio clearly defined the two Golgi peaks and indicated that no substantial endomannosidasemediated deglucosylation took place in the ER compartment (Fig. 9).

DISCUSSION
From the results presented in this paper, it is apparent that the glucosidase 11-deficient mouse lymphoma cell line mutant (PHA"2.7) described by Reitman et al. (14) employs the endomannosidase-initiated processing pathway to maintain its capacity to synthesize complex N-linked oligosaccharides. Indeed the observations made in this study and in a previous investigation with glucosidase inhibitors in HepG2 cells (13) support the proposal made upon discovery of endomannosidase, that the unique deglucosylation mechanism provided by this enzyme can make possible circumvention of a glucosidase blockade whether imposed by genetic factors or exogenous agents (11,12).
The in vivo operation of endomannosidase in the mutant was most clearly demonstrated by the observation that the characteristic products of this enzyme, namely GlczMan and N-linked ManRGlcNAc isomer A, were produced by these cells. The finding that the addition of CST to mutant incubations resulted in the formation of GlcnMan instead of GlczMan indicated that these cells have the capacity to attach as well as remove the outer al-&!-linked glucose residue in a normal manner; indeed in the parent lymphoma cell line, as in HepG2 cells (13), triglucosylmannose is the primary product released in the presence of CST. On the other hand, the failure of the mutant to form MangGlcNAc, even in the presence of DMJ which causes it to accumulate in the parent cell line, is consistent with the in vitro demonstrable absence of glucosidase I1 activity determined with GlclMangGlcNAc as substrate. The occurrence of small amounts of Glc,Man and monoglucosylated polymannose units in the mutant, however, suggests the continued presence of some glucosidase function; such residual activity may preferentially degrade the diglucosylated oligosaccharides as it has been reported that at least in vitro glucosidase I1 has higher activity toward GlZMangGlcNAc than Glc,MangGlcNAc (27). Substantial quantities of deglucosylated N-linked polymannose oligosaccharides were produced by the mutant, but because of their rapid further processing by mannosidase I they only become evident when an inhibitor of this enzyme was added to the incubations. The major deglucosylated oligosaccharide species was Man8GlcNAc (isomer A ) although also present were Man7GlcNAc and Man6GlcNAc which are presumed to have originated by endomannosidase action on GlczMansGlcNAc and GlczMan7GlcNAc, respectively. The apparently exclusive utilization of the endomannosidase pathway by the mutant cells for the deglucosylation of the Nlinked GlcpMan9GlcNAcz facilitated our exploration of the in vivo locale of this enzyme's action. The pulse-chase studies carried out with the mutant cells clearly indicated that the GlcpMangGlcNAcz to ManHGlcNAcz conversion is a rather late event which is temporally separated from the ER-situated processing of Glc3MangGlcNAc2 to GlczMangGlcNAcp and, in contrast to the latter, takes place in the Golgi compartment.
These observations are consistent with previously reported in vitro studies which demonstrated that Golgi membranes have the radioactivity in the ManaGlcNAc and Glc2MangGlcNAc components was determined by scintillation counting, and the ratio of these two values was plotted for each fraction. The radioactivity recovered from the gradient in MansGlcNAc and Glc2MangGlcNAc was 2.73 X lo5 dpm and 19.9 X lo5 dpm, respectively. a considerably higher endomannosidase-specific activity than ER fractions (11).
Although it is apparent from the present investigation and our previous study (13) that endomannosidase can provide glucosidase-deficient or inhibited cells with an alternate processing route, it is not yet clear to what extent this enzyme functions under physiological conditions; in normal cells the appearance of the characteristic glucosylated mannose saccharides (Glcl_3Man) cannot be used as an index of endomannosidase processing as they are rapidly degraded by glucosidase action. On the other hand, if the assumption is made that Man,GlcNAc isomer A is exclusively formed by endomannosidase, an evaluation of the in vivo operation of this enzyme under normal conditions can be made from a quantitation of this component. Our analyses of the MansGlcNAc from parent lymphoma cells revealed that about 16% is present in the form of isomer A and a similar observation was made in HepG2 cells. However, the production of this MansGlcNAc variant was not evident unless DMJ was added to the incubations indicating that the A isomer is preferentially processed by mannosidase I. Even if all of the Man,GlcNAc and Man,GlcNAc which are formed by the parent lymphoma cells in the presence of DMJ are presumed to have arisen by endomannosidase action on glucosylated Man,GlcNAc and MasGlcNAc, respectively, only about 14% of the total deglucosylated species (see Table 11) could have arisen by this processing route. It is unlikely that this relatively modest involvement of endomannosidase in the deglucosylation process of normal cells is due to the low total cellular activity of this enzyme when compared to glucosidase 11, as we have shown that mutant lymphoma cells as well as CST-treated HepG2 cells (13) can accommodate a large traffic through this alternate route. Indeed in the glucosidase IIdeficient lymphoma cells the increase in the operation of the endomannosidase pathway did not even require a higher level of this enzyme as determined by in vitro assay. The limiting factor to endomannosidase function may be the number of proteins with glucosylated polymannose oligosaccharides which reach the Golgi complex where this enzyme is located. Although in glucosidase-inhibited or deficient cells, all of the glycoproteins which enter this organelle still retain their glucose residues, under normal circumstances extensive removal of these sugars would already have taken place by the ER-situated glucosidase I and 11, and only a small portion of the total molecules would still be in the glucosylated form. It is possible that under physiological conditions endomannosidase acts on carbohydrate units with truncated mannose branches for which it is known to have a particularly high affinity (12). Since such partially processed oligosaccharides (e.g. GlclMan8GlcNAc and GlclMan7GlcNAc) are poor substrates for cleavage by glucosidase I1 (28, 29) they would be more likely to reach the Golgi complex.
While this paper was in preparation, a study by Fujimoto and Kornfeld (30) was published which demonstrated that the glucosidase 11-deficient mouse lymphoma cell line produces Glc,Man and GlclMan in a sufficient amount to account for its complex N-linked oligosaccharides formation. While our study is in agreement with these findings, it addresses itself to a substantial number of additional aspects of the glucosidase-independent pathway in lymphoma cells. Our characterization of the deglucosylated N-oligosaccharides has shown that the mutant cells, in contrast to the parent, produces exclusively the distinctive endomannosidase-generated isomer A of Man,GlcNAc. This finding has permitted us to determine the subcellular locale in which this enzyme acts and, furthermore, from a measurement of the MansGlcNAc isomers in normal cells, we have obtained some insight into the extent to which endomannosidase functions under physiological conditions.