The H,K-ATPase beta subunit as a model to study the role of N-glycosylation in membrane trafficking and apical sorting.

The role of N-glycosylation in trafficking of an apical membrane protein, the gastric H,K-ATPase beta subunit linked to yellow fluorescent protein, was analyzed in polarized LLC-PK1 cells by confocal microscopy and surface-specific biotinylation. Deletion of the N-glycosylation sites at N1, N3, N5, and N7 but not at N2, N4, and N6 significantly slowed endoplasmic reticulum-to-Golgi trafficking, impaired apical sorting, and enhanced endocytosis from the apical membrane, resulting in decreased apical expression. Golgi mannosidase inhibition to prevent carbohydrate chain branching and elongation resulted in faster internalization and degradation of the beta subunit, indicating that terminal glycosylation is important for stabilization of the protein in the apical membrane and protection of internalized protein from targeting to the degradation pathway. The decrease in the apical content of the beta subunit was less with mannosidase inhibition compared with that found in the N1, N3, N5, and N7 site mutants, suggesting that the core region sugars are more important than the terminal sugars for apical sorting.

Sorting of proteins between apical and basolateral membranes in polarized cells depends on the recognition of intrinsic sorting signals within the proteins by specific sorting machinery in the trans-Golgi network (TGN) 1 or endosomes (1)(2)(3)4). Basolateral sorting signals often contain tyrosine-based, dileucine, or other hydrophobic motifs that are recognized by clathrin coat proteins that package them into basolateral transport vesicles (5,6). Apical sorting signals are less well defined. N-Linked glycans are considered as possible candidates based on the findings that some proteins gain the ability to reach the apical membrane after recombinant addition of N-glycosylation sites (7) and that other proteins entirely or partially lose apical expression as a result of removal of Nglycans by mutation or treatment with glycosylation inhibitors (5, 8 -10). N-Glycosylation might significantly affect other trafficking steps, including protein folding and quality control in the endoplasmic reticulum (ER), endocytosis, recycling, and degradation. Because the effect of N-glycan removal on an apical sorting event has not been differentiated from a possible impairment of other trafficking steps, the role of N-glycans as apical sorting signals remains unclear.
An interesting paradigm for the study of the role of glycosylation in protein maturation, trafficking, sorting, and degradation is an apically targeted multiply glycosylated protein, the H,K-ATPase ␤ subunit. The gastric H,K-ATPase, the enzyme responsible for acid secretion in the stomach, consists of two subunits, a catalytic ␣ subunit and an accessory ␤ subunit, which has seven N-glycosylation sites. Glycosylation of the ␤ subunit has been shown to be critical for the quality control of the H,K-ATPases in the ER (11,12). Specific N-glycosylation sites have been found to be essential for plasma membrane delivery of the H,K-ATPase ␤ subunit in non-polarized HEK-293 (13) and for delivery of both ␣ and ␤ subunits in COS-7 cells (12), suggesting that they may also play a role in apical sorting. The homologous Na,K-ATPase also consists of ␣ and ␤ subunits. Of four known isoforms of the Na,K-ATPase ␤ subunits, ␤ 2 is the most homologous to the H,K-ATPase ␤ subunit, has up to nine glycosylation sites (14), and appears to result in apical sorting of the Na,K-ATPase ␣⅐␤ complex in a number of tissues (15)(16)(17). The Na,K-ATPase that contains either the ␤ 1 or ␤ 3 isoform with only two or three N-glycosylation sites localizes exclusively in the basolateral membrane (18). The high degree of glycosylation in the H,K-ATPase ␤ and Na,K-ATPase ␤ 2 subunits might imply a role of N-glycosylation in the apical sorting of the corresponding ␣⅐␤ complexes.
N-Glycosylation occurs in various stages. N-Linked oligosaccharides appear when the 14-saccharide core is transferred from the dolichol phosphate precursor to the nascent membrane protein that remains associated with the translocon in the ER. Immediately after coupling the core to the asparagine of the specific amino acid NXS or NXT motif, the N-glycosylation site, the terminal glucose residues are trimmed by ER glucosidases. Various chaperones such as calnexin bind to sugar chains as quality control elements at this and perhaps later stages. Subsequently, the mannose residues are trimmed in the ER and then in the Golgi by mannosidases I and II, respectively. This is followed by elongation of the carbohydrate chains due to addition of terminal sugars by the action of various glycosyltransferases in the trans-Golgi (19). There is considerable variation in the number and composition of terminal chains in the mature complex oligosaccharides, giving rise to heterogeneity of glycosylated proteins. The core region that contains five residues originating from the initial 14saccharide core is the same in all molecular species of mature glycoproteins. Deletion of any N-glycosylation site in a protein abrogates the whole oligosaccharide tree at that particular locus, leaving the others intact. Inhibition of glucose or mannose trimming prevents addition of the terminal sugar chains but leaves core regions intact at all N-glycosylation sites. This enables distinctions to be made concerning the roles of core region and terminal sugars at various stages of processing and trafficking of the protein.
Here, we quantitatively determined at which trafficking stages N-glycosylation deficiency affects apical polarity. We determined the effect of either deletion of individual N-glycosylation sites or treatment of the cells with N-glycan-trimming inhibitors on the steady-state distribution of the H,K-ATPase ␤ subunit between the ER, Golgi, and apical and basolateral membranes as well as on endocytosis efficiency and degradation rate in polarized LLC-PK1 cells by quantifying the immature and mature forms of the protein that differ in their molecular masses and by using surface-specific biotinylation and confocal microscopy. Glycosylation at the N1, N3, N5, and N7 sites of the gastric H,K-ATPase ␤ subunit is crucial for ER-to-Golgi trafficking, apical membrane delivery, and stability of the protein in the apical membrane. Considering the steadystate apical content of the mature H,K-ATPase ␤ subunit as a result of dynamic equilibrium between two opposite processes, the apical membrane delivery from TGN/endosomes, and endocytosis from the apical membrane, we were able to quantify the contribution of N-glycosylation to each of these steps. In addition, we show that particular carbohydrate residues of the N-glycans have different roles in apical membrane delivery and endocytosis, which is the first indication for a specific role of certain carbohydrates in apical sorting. In the mature complextype glycoprotein, the core region sugars are more important for the apical sorting event in TGN/endosomes compared with terminal sugars, whereas the terminal sugars anchor the protein on the apical membrane and also protect the protein from targeting to the degradation pathway after endocytosis.
Stable Transfection-To obtain cell lines stably expressing wild-type or mutant YFP-␤ fusion proteins, LLC-PK1 cells were grown on 10-cm plates until 20% confluent and transfected with wild-type YFP-␤ or mutant YFP-␤ using FuGENE 6 transfection reagent (Roche Applied Science). 24 h after transfection, stable cell lines were selected by addition of the eukaryotic selection marker G418 at a final concentration of 1.0 mg/ml. This concentration of G418 was maintained until single colonies appeared. 15-20 colonies were isolated, expanded, and grown in the presence of 0.25 mg of G418/ml of medium in 24-well plates. Two clones with the highest expression of YFP-␤ were selected and expanded for further studies.
Confocal Microscopy Studies-Cells stably expressing wild-type or mutant YFP-␤ were grown for at least 5 days after becoming confluent on glass bottom micro-well dishes (Mattek Corp.). Confocal microscopy images were acquired using a Zeiss LSM 510 laser scanning confocal microscope with LSM 510 Version 3.2 software.
Estimation of Surface YFP-␤ Content by Surface-specific Biotinylation-LLC-PK1 cells stably expressing wild-type or mutant YFP-␤ were maintained for at least 5 days after becoming confluent in Corning Costar polyester Transwell inserts (Corning Inc.) in 6-well plates. Biotinylation of the apical or basolateral membrane proteins was performed by previously described procedures (22,23). Briefly, cell monolayers were biotinylated with EZ-Link TM sulfosuccinimidyl-2-(biotinamido)ethyl 1,3Ј-dithiopropionate (Pierce), which was added from either the apical or basolateral side. After quenching the biotinylation reaction, cells were washed and then lysed by incubation with 200 l of 0.15 M NaCl in 15 mM Tris (pH 8.0) containing 1% Triton X-100 and 4 mM EGTA. Cell lysates were clarified by centrifugation at 15,000 ϫ g for 10 min. Samples containing 20 l of supernatant mixed with 15 l of SDS-containing sample buffer were loaded onto SDS-polyacrylamide gel to determine the total YFP-␤ content in the supernatant. To precipitate biotinylated proteins, the rest of each supernatant was incubated with 100 l of streptavidin-agarose beads (Sigma) in a total volume of 800 l of the lysis buffer for 1 h at 4°C with continuous rotation. Precipitated complexes were washed three times on the beads, and then proteins were eluted from the beads by incubation in 40 l of SDS-PAGE sample buffer (4% SDS, 0.05% bromphenol blue, 20% glycerol, and 1% ␤-mercaptoethanol in 0.1 M Tris (pH 6.8)) for 5 min at 80°C, separated on SDS-polyacrylamide gel, and analyzed by Western blotting using monoclonal antibody 2B6 against the H,K-ATPase ␤ subunit (MBL, Inc.) or the monoclonal antibody against the Na,K-ATPase ␤ 1 subunit (Novus Biologicals) as the primary antibody and anti-mouse IgG conjugated to alkaline phosphatase (Promega) as the secondary antibody according to the manufacturers' instructions.
Endocytosis Assay by Apical Surface Biotinylation-Polarized cells stably expressing wild-type or mutant YFP-␤ were biotinylated from the apical side as described above. Cells were incubated at 18°C to prevent apical membrane delivery for 20, 60, or 120 min. After that, apical biotin was stripped off by incubation with 50 mM reduced glutathione (Sigma) in 100 mM NaCl with 10% fetal bovine serum (pH 8.4) twice for 20 min. After cell lysis, the internalized biotinylated proteins were precipitated, washed, eluted from streptavidin-agarose beads, and analyzed by SDS-PAGE and Western blot analysis as described above. In the negative control, biotin was stripped off immediately after biotinylation. The initial apical content was determined by lysing cells immediately after biotinylation. In the positive control, cells were incubated at 18°C for 60 or 120 min to account for any instability of biotinylated protein and then lysed. After cell lysis, biotinylated proteins were precipitated, washed, eluted, and analyzed as described above. For the mutants with very low apical content of YFP-␤ (N1, N3, and N5), each experimental condition was repeated in three wells, and cell lysates from three identical wells were combined before precipitation on streptavidin-agarose beads. Endocytosis efficiency was calculated as a percentage of apical YFP-␤ that was internalized for 1 h.
Treatment of Cells with Glycosidase Inhibitors-Cells were incubated with castanospermine (Sigma), the Golgi mannosidase I inhibitor deoxymannojirimycin (dMAN) (Sigma), or the Golgi mannosidase II inhibitor swainsonine (Sigma) at a concentration of 1, 2, or 200 g/ml, respectively, for 48 h prior to apical biotinylation.
Glycosidase Cleavage-Where indicated, the total cell lysates or proteins precipitated by streptavidin-agarose beads were treated with peptide N-glycosidase F (PNGase F) from Flavobacterium meningosepticum (New England Biolabs Inc.) or endoglycosidase H (Endo H) from Streptomyces plicatus (Glyco-Prozyme Inc.) according to the manufacturers' instructions.
In experiments with live cells, sialidase from Salmonella typhimurium recombinant in Escherichia coli (Glyco-Prozyme Inc.), ␤1,4galactosidase from Streptococcus pneumonia (Glyco-Prozyme Inc.), or PNGase F from F. meningosepticum was added to the cell medium from the apical side at a concentration of 1 unit/ml, 66 milliunits/ml, or 7500 New England Biolabs units/ml, respectively, and incubated for 16 h. 1 h prior to completion of the cleavage, cycloheximide (Sigma) at a concentration of 10 g/ml was added to the medium to inhibit de novo synthesis of YFP-␤.

Analysis of N-Glycosylation Alteration-
The H,K-ATPase ␤ subunit has seven N-glycosylation sites. The mature oligosaccharide linked to each of the seven sites consists of the threemannosyl core region and terminal chains (Fig. 1). Three approaches were used to analyze the effect of N-glycosylation deficiency in the ␤ subunit. By mutating each of seven Nglycosylation sites and expressing these mutants in polarized cells, we obtained ␤ subunits lacking all core and terminal sugars at only a single N-glycosylation site, with six sites still normally glycosylated. Alternatively, cells expressing the wildtype ␤ subunit were treated with a glucosidase inhibitor (castanospermine), a Golgi mannosidase I inhibitor (dMAN), or a Golgi mannosidase II inhibitor (swainsonine). The inhibitors prevent glucose or mannose trimming and further elongation of all the carbohydrate chains and thus result in expression of the 14-saccharide core and the high mannose-and hybrid-type oligosaccharides, respectively. These steps are illustrated schematically in Fig. 1A. Thus, the glycoprotein formed in the presence of castanospermine or dMAN lacks all the terminal sugars normally present in the complex-type ␤ subunit; and in the presence of swainsonine, the glycoprotein lacks the majority of normal terminal sugars at all seven N-glycosylation sites (Fig. 1A). The complex-, hybrid-, and high mannose-type and 14-saccharide core forms of YFP-␤ can be clearly distinguished from each other upon SDS-PAGE due to the difference in molecular masses, as shown in Fig. 1B.
Characterization of Glycosylated Forms of YFP-␤ Expressed in LLC-PK1 Cells-YFP-␤ was detected in cell lysates of LLC-PK1 cells as two bands, one at 80 -100 kDa and the other at ϳ75 kDa (Fig. 2, lane 1). After PNGase F treatment of the cell lysates, the bands at 80 -100 and 75 kDa both disappeared, and a single band was seen at ϳ55 kDa, corresponding to deglycosylated YFP-␤ (lane 3). A similar product was detected after treatment of the cell lysates with Endo H, but this treatment resulted in the disappearance of only the lower band on the Western blot, whereas the higher band was retained (lane 2). It is known that Endo H cleaves only high mannose-or hybridtype glycoproteins, whereas complex-type chains are Endo Hresistant. Therefore, the 80 -100-kDa band represents the complex-type glycosylated fraction of YFP-␤. The 75-kDa Endo H-sensitive band represents high mannose-type YFP-␤ (Fig.  1B, fourth lane), as can be concluded from comparison with the lysates prepared from cells preincubated with the glycosylation inhibitors dMAN, which led to formation of only high mannosetype YFP-␤ (second lane), and swainsonine, which resulted in formation of both hybrid-and high mannose-type glycoproteins (third lane).
In contrast to the total cell lysate, the apically biotinylated protein fraction contained only the complex-type glycosylated fraction of YFP-␤ (Fig. 2, lane 4), which was Endo H-resistant (lane 5). Thus, the mature YFP-␤ component in LLC-PK1 cells FIG. 1. A, simplified model of N-glycosylation pathways in the absence and presence of glycosidase inhibitors. The dolichol-linked precursor oligosaccharide is transferred to each of the N-glycosylation sites of the emerging polypeptide chain cotranslationally in the ER. ER glucosidases remove all three glucose residues, and the ER mannosidase removes one mannose residue and leads to formation of a high mannose-type oligosaccharide. In the Golgi, in the absence of inhibitors, specific mannose residues are sequentially trimmed from the high mannose-type oligosaccharide; terminal sugars are added; and the complex-type oligosaccharide is normally formed. Castanospermine (Cas) inhibits ER glucosidases I and II and prevents further modifications of initially attached 14-saccharide core. dMAN inhibits Golgi mannosidase I and completely prevents further trimming and terminal glycosylation. As a result, the mature glycoprotein in the presence of dMAN contains high mannose-type oligosaccharides only. Swainsonine (Sw) inhibits Golgi mannosidase II and prevents further trimming, branching, and terminal glycosylation of mannose-terminated branches, but allows further terminal glycosylation of the N-acetylglucosamine-terminated branch. Accordingly, hybrid-type oligosaccharides are formed in the presence of swainsonine. Complex-, hybrid-, and high mannosetype oligosaccharides all contain the same core region, the so-called "three-mannosyl core," but they are different from each other in the number and composition of terminal sugars. B, glycosylation pattern of YFP-␤ in the absence and presence of glycosidase inhibitors. Cells expressing YFP-␤ were incubated with glycosidase inhibitors for 48 h prior to cell lysis, SDS-PAGE, and Western blotting. The predicted oligosaccharide structure for each fraction of glycosylated YFP-␤ is shown. Cas shows the effect of castanospermine inhibition of deglucosidation; dMAN shows the result of inhibition of mannosidase I; and Sw shows the result of inhibition of mannosidase II. Each species has a distinct relative molecular mass upon SDS-PAGE.

FIG. 2. Effect of glycosidases on the apical and total fractions of YFP-␤ expressed in LLC-PK1 cells.
These data show that mature YFP-␤ contains complex-type oligosaccharides only. Cells were grown on Transwell filters. Upon polarization, cell layers were biotinylated from the apical side and lysed. One-tenth of the cell lysate was incubated with the glycosidase Endo H, which cleaves high mannose-and hybrid-type sugars but not the complex-type oligosaccharides from glycoproteins, or with PNGase F, which cleaves all types of oligosaccharide chains. The rest of the cell lysate was incubated with streptavidinagarose beads. Biotinylated protein-streptavidin-agarose complexes were precipitated and treated with Endo H. The proteins were then eluted from the beads and separated by SDS-PAGE. Cell lysates treated with Endo H or PNGase F as well as control samples for both cell lysates and biotinylated proteins were loaded onto the same gel. Proteins were transferred onto nitrocellulose membrane and analyzed by Western blotting using the antibody against the H,K-ATPase ␤ subunit. In the cell lysate, Endo H cleaved only the fraction of YFP-␤ corresponding to the lower band (high mannose-type) and produced deglycosylated YFP-␤ (lanes 1 and 2). The upper band in the cell lysate (lane 1) represents Endo H-insensitive mature YFP-␤, which contains complextype oligosaccharides only. PNGase F cleaved both fractions, producing deglycosylated YFP-␤ (lane 3). The fraction of YFP-␤ found on the apical membrane was insensitive to Endo H (lanes 4 and 5), indicating that it contains only complex-type oligosaccharides. C, complex-type YFP-␤; H, high mannose-type YFP-␤; D, completely deglycosylated YFP-␤. A representative blot from three independent experiments is shown.
is complex-type glycosylated protein. Only this fraction of the YFP-␤ pool was able to reach the apical membrane. The high mannose-type core glycosylated fraction of YFP-␤ present in cell lysates therefore represents the immature ER portion of the YFP-␤ pool.
Surface biotinylation was used to quantify the apical and basolateral content of YFP-␤ (Fig. 3). Under biotinylation conditions, (i) LLC-PK1 cells may have their tight junctions disrupted during the biotinylation procedure, and (ii) their apical membranes may also become leaky. Hence, in all experiments, biotinylation of basolateral membranes was used to define a specific basolateral location of the Na,K-ATPase ␤ 1 subunit (18) as a control for intact tight junctions, and the absence of high mannose-type YFP-␤ in biotinylated samples was used as an indication of apical membrane integrity. The presence of biotinylated high mannose-type YFP-␤ protein would show that the biotinylation reagent had access to the intracellular pool of high mannose-type YFP-␤ because of the apical membrane leakiness during the experiment. As shown in Fig. 3 (left panel), ϳ85% of surface YFP-␤ was detected on the apical membrane. In contrast, the endogenous Na,K-ATPase ␤ 1 subunit was predominantly labeled from the basolateral membrane, indicating integrity of tight junctions.
Effect of N-Glycosylation Site Mutations on the Relative Apical Content of YFP-␤ in LLC-PK1 Cells-Using confocal microscopy, the wild-type YFP-␤ was detected predominantly in the apical region in LLC-PK1 cells (Fig. 4, left panel), as can be seen in the confocal apical XY section of the cell monolayer and particularly in the Z section. A minor fraction of YFP-␤ was detected inside the cells (XY section through the middle of the cell layer) and in the lateral membranes (Z section). In contrast, the YFP-␤ mutant lacking the N1 glycosylation site was accumulated mostly intracellularly (right panel). The protein was predominantly localized to the perinuclear region, presumably in the ER and Golgi.
To quantify the apical content in the mutants compared with the wild-type protein, cells expressing wild-type or mutant YFP-␤ were biotinylated from the apical side in the same experiment to prevent variations in biotinylation, streptavidinagarose precipitation, and immunoblotting efficiencies. The apical content was normalized to the total YFP-␤ content in the corresponding cell lysate for each mutant and compared with that in the wild-type protein, as shown in Fig. 5. The relative apical content was dramatically decreased in N1, N3, and N5 by 14.5-, 7.1-, and 5.5-fold, respectively, and was not detectable in N7. In contrast, mutation of N2, N4, and N6 only moderately decreased the relative apical content (from 1.4-to 1.7-fold).
We found that mature YFP-␤ contains complex-type oligosaccharides only and that the high mannose-type form of the protein corresponds to the ER fraction of the cellular YFP-␤ pool (Fig. 2). Therefore, the ratio between the complex-type form and total YFP-␤ can be used as a measure of ER-to-Golgi trafficking. In all the mutants except N6, the relative content of the complex-type glycosylated fraction of YFP-␤ was decreased, indicating that these mutations slowed down ER-to-Golgi trafficking and caused more ER retention. The most significant effect on ER-to-Golgi trafficking was observed in the N1 and N7 mutants (2.3-and 6.4-fold decreases, respectively).
Effect of N-Glycosylation Site Mutations on the Internalization Efficiency of YFP-␤ in LLC-PK1 Cells-To compare the effect of mutations on the internalization efficiency of YFP-␤, the apical surface of cells expressing wild-type or mutant YFP-␤ was biotinylated, and cells were incubated at 18°C to prevent apical membrane delivery of internalized proteins. Any apical biotin was then cleaved off, and internalized biotinylated proteins were detected as described under "Experimental Procedures." Internalization efficiency (viz. the fraction of apical YFP-␤ internalized after 1 h) was increased from 2-to 3-fold in the N1, N3, and N5 mutants, but was only slightly increased in the N1, N4, and N6 mutants (Fig. 6). The internalization efficiency in N7 could not be measured due to the very low apical content of YFP-␤ in this mutant.
Effect of N-Glycosylation Site Mutations on Apical Sorting-The surface distribution of mutants lacking the N3 or N5 site was different from that of the wild-type protein. The major plasma membrane-located fraction of the mutant proteins was detected on the basolateral (but not apical) membrane (Fig. 3). This might imply that removal of the particular N-glycosylation site impaired apical sorting. However, a steady-state distribution between two distinct surface domains is not a result only of apical sorting in the TGN and/or endosomes, but also reflects a balance between apical and basolateral sorting as well as apical and basolateral endocytosis.
To quantify the effect of mutations solely on apical sorting, we compared their effect on the relative apical content and the efficiency of endocytosis and calculated the effect of mutations on the apical sorting efficiency. The apical content of YFP-␤ normalized by comparison with the mature complextype fraction of YFP-␤ (C A /C T ) (Fig. 5) reflects a steady-state distribution between the apical and internal mature complextype pools of YFP-␤. We found that the rate of degradation of the mature complex-type fraction was not changed by Nglycosylation site mutations. Therefore, the distribution between apical and internal YFP-␤ could be shifted toward the internal pool in the mutants either because of enhanced endocytosis or because of the impaired apical membrane delivery (see Fig. 10). The relative decrease in the apical content was calculated for each mutant by dividing the apical content in the wild-type protein, (C A /C T ) wt , by the apical content in the mutant, (C A /C T ) mut (Table I, second column). Similarly, the relative increase in the efficiency of endocytosis in each mutant was calculated by dividing the endocytosis efficiency in the mutant by the endocytosis efficiency in the wild-type protein as assessed by apical biotinylation as described above (Table I, third column). If the apical content was decreased by the same factor as the endocytosis efficiency was increased, then the enhanced endocytosis would be the only reason for the lowered apical content in this mutant. However, in all mutants, the apical content was Cells were grown on Transwell filters. Upon polarization, cell layers were biotinylated from the apical (A) or basolateral (B) side and lysed. Biotinylated proteins were precipitated by streptavidin-agarose beads. Proteins were eluted from the beads, separated by SDS-PAGE, and transferred onto nitrocellulose membrane. The membrane was cut in half at the level shown by the dotted line. The upper part of the blot was developed using monoclonal antibody 2B6 against the gastric H,K-ATPase ␤ subunit, and the lower part was developed using the monoclonal antibody against the Na,K-ATPase ␤ 1 subunit (Na,K-␤ 1 ). A representative blot from three independent experiments is shown. wt, wild-type protein.
decreased to a greater extent than the endocytosis efficiency. For example, in the N5 mutant, the apical content was decreased by 4.5-fold, but the internalization efficiency was increased by only 2-fold. This indicates that the apical membrane delivery was also affected by the mutation because the relative apical content reflects a balance between apical membrane delivery and endocytosis. If the apical content in the mutant is decreased by X-fold compared with the wildtype protein and the endocytosis efficiency is increased by Y-fold, then the apical membrane delivery rate must be decreased by Z ϭ X/Y-fold. Thus, as shown in Table I, the effect of each mutation on the apical sorting efficiency was calcu-

FIG. 5. N-Glycosylation site mutations increase the ER retention of YFP-␤ and reduce the apical content of YFP-␤ in LLC-PK1 cells.
Polarized cell layers were biotinylated from the apical side and then lysed. One-tenth of the cell lysate was loaded onto SDS-polyacrylamide gel to determine total YFP-␤ expression. The rest of the cell lysate was incubated with streptavidin-agarose beads. Biotinylated proteinstreptavidin-agarose complexes were precipitated. Proteins were eluted from the beads, run on SDS-polyacrylamide gel side-by-side with total proteins from the same cell line, and immunostained using the antibody against the H,K-ATPase ␤ subunit. A, apical proteins; T, total proteins. C A , complex-type glycosylated YFP-␤ on the apical membrane; C T , complex-type glycosylated YFP-␤ in the cell lysate; H T , high mannose-type glycosylated YFP-␤ in the cell lysate. The ER-to-Golgi trafficking efficiency was estimated as a fraction of the complex-type form of total YFP-␤ in the cell lysate. Data from three independent experiments were quantified using the Kodak 1D 3.6 system and averaged and are presented as bar graphs (mean Ϯ S.D.). *, significant difference compared with the wild-type protein (wt; p Ͻ 0.05). lated as a ratio between the factor in the second column, (C A /C T ) wt /(C A /C T ) mut , and the factor in the third column, (endocytosis efficiency) mut /(endocytosis efficiency) wt .
The apical sorting efficiency was decreased in all the mutants except N2 up to 2.3-fold (Table I). The very low content of complex-type YFP-␤ in N7 did not allow determination of the relative apical content in the cell line expressing this mutant.

Effect of Inhibitors of ER Glucosidase and Golgi Mannosidases I and II on Sorting and Trafficking of YFP-␤ in LLC-PK1
Cells-Treatment of cells with castanospermine, an inhibitor of the ER glucosidase, completely blocked oligosaccharide trimming and did not allow glycoprotein processing, as shown in Fig. 1. As a result, YFP-␤ detected in the presence of castanospermine has a higher molecular mass than the high mannosetype form, as shown in Fig. 1B. Apical biotinylation of cells treated with castanospermine did not show any detectable amount of YFP-␤ on the surface (data not shown), indicating that the untrimmed protein is unable to exit the ER and to be processed to the Golgi and plasma membrane.
Preincubation of cells expressing wild-type YFP-␤ with the two Golgi mannosidase inhibitors resulted in a decrease in the molecular mass of mature YFP-␤ in cell lysates (Fig. 7). In the presence of swainsonine, two fractions of YFP-␤ were detected. The high mannose-type fraction represents the ER portion of the protein, and hybrid-type glycosylated YFP-␤ was the only form of the mature protein formed in the Golgi in the presence of this inhibitor (Fig. 1). In the presence of dMAN, the immature ER portion of the protein and the mature form of YFP-␤ were indistinguishable since dMAN led to formation of high mannose-type glycosylated YFP-␤ only (Fig. 1). The apical content was decreased as a result of preincubation with swainsonine or dMAN (Fig. 7). However, the effect of the inhibitors on the relative apical content was less than that of the N1, N3, N5, and N7 glycosylation site mutations (compare Figs. 5 and 7). In contrast, internalization in the presence of swainsonine was higher compared with that in the wild-type protein and all the mutants (Fig. 6). Both inhibitors lowered the total cellular content of YFP-␤, suggesting that they promote its degradation.
To compare the degradation rates of YFP-␤ preincubated with and without swainsonine, we measured the mature YFP-␤ b Decrease in apical sorting efficiency ϭ (decrease in C A /C T )/(increase in endocytosis efficiency).

FIG. 6. Effects of mutations of N-glycosylation sites on the internalization rate of YFP-␤ in LLC-PK1 cells.
A typical internalization assay (data for wild-type YFP-␤ (wt)) is shown in the upper left panel. LLC-PK1 cells expressing the wild-type protein or the N2, N4, or N6 mutant were grown on six Transwell inserts. Upon polarization, cell layers were biotinylated from the apical side. Lane 1, cells were lysed immediately after biotinylation. Lanes 2-6, cells were lysed after completion of the following procedures: cells were incubated at 18°C for 120 min (positive control; lane 2); biotin was stripped off immediately after biotinylation (negative control; lane 3); and biotin was stripped off after incubation of cells at 18°C for 20, 60, and 120 min (internalized biotinylated proteins; lanes 4 -6, respectively). An internalization assay for wild-type YFP-␤ in the cells preincubated for 48 h with swainsonine (wt ϩ sw; 2 g/ml) is shown in the upper right panel. Upon polarization, cell layers were biotinylated from the apical side. Lane 1, cells were incubated at 18°C for 60 min (positive control). Lane 2, biotin was stripped off immediately after biotinylation (negative control). Lane 3, biotin was stripped off after incubation of cells at 18°C for 60 min (internalized biotinylated proteins). A similar scheme was applied for N1, N3, and N5. After cell lysis, biotinylated proteins in all samples were precipitated by streptavidin-agarose beads, eluted, separated by SDS-PAGE, and immunoblotted. Internalization efficiencies for the wild-type protein and each mutant, calculated as a percentage of the initial apical content internalized for 1 h, are shown in the lower panel. *, significant difference compared with the wild-type protein (p Ͻ 0.05, n ϭ 3).

FIG. 7. Effects of swainsonine and dMAN on total cell expression and apical content of YFP-␤ in LLC-PK1 cells.
Cells were grown on Transwell filters, polarized, and incubated with mannosidase inhibitors for 48 h prior to biotinylation. Cell layers were biotinylated from the apical side and lysed. Biotinylated proteins were precipitated by streptavidin-agarose beads, eluted, and analyzed by SDS-PAGE and Western blotting side-by-side with total proteins of the cell lysate. C, complex-type glycoprotein; Hb, hybrid-type glycoprotein; H, high mannose-type glycoprotein. Ϫ and ϩ indicate the absence and presence of swainsonine (2 g/ml) or dMAN (200 g/ml) in the cell culture medium. Data from three independent experiments were quantified and are presented as a percentage of the control, which did not have the inhibitor (mean Ϯ S.D.). *, significant difference compared with the control (p Ͻ 0.05). content upon incubation of cells with cycloheximide to inhibit de novo protein biosynthesis. Swainsonine significantly increased the degradation rate (Fig. 8). By contrast, mutation of any single glycosylation site did not change the degradation rate of the complex-type glycosylated fraction of YFP-␤. As an example, data for the N5 mutant are shown in Fig. 8.

Effect of Treatment of the Apical Surface of LLC-PK1 Cells with Glycosidases on YFP-␤ Internalization and Degradation-
Apical treatment of cells expressing wild-type YFP-␤ with sialidase resulted in a decrease in the molecular mass of the protein (Fig. 9), indicating that terminal sialic acid residues were cleaved from the terminal chains of YFP-␤. The amount of YFP-␤ found on the apical membrane and in the total cell lysate after desialylation was the same as in the control (Fig.  9), indicating that sialic acid is not important for the stability of the protein in the apical membrane and for recycling.
Treatment of the cells with sialidase in combination with galactosidase resulted in a decrease in the molecular mass of YFP-␤ as a result of a sequential cleavage of sialic acid and ␤-galactose residues. The amount of YFP-␤ after cleavage did not change in either the apical membrane or the total cell lysate (Fig. 9), showing that galactose residues also are not important for endocytosis and recycling.
Apical treatment of cell layers with PNGase F resulted in an almost complete disappearance of YFP-␤ from the apical membrane (Fig. 9). This result indicates that the deglycosylated protein is not stable in the membrane and internalizes and is not recycled to the apical membrane. A significant decrease in the complex-type glycosylated form of YFP-␤ was detected in the cell lysate (Fig. 9), whereas the high mannose-type form reflecting the newly synthesized ER fraction of the protein remained unchanged. Surprisingly, the deglycosylated form was not detected, suggesting that it was rapidly degraded during incubation of cells with the enzyme.

N-Glycosylation and Degradation as a Quality Control System in Mammalian Cells-Glycosylation is indispensable for functional expression of the H,K-ATPase in mammalian cells.
In HEK-293 cells, prevention of ␤ subunit glycosylation due to tunicamycin treatment or mutation of all seven glycosylation sites results in almost complete loss of protein expression (12). Removal of all the sugars from apical YFP-␤ by PNGase F in LLC-PK1 cells resulted in internalization of the protein from the apical membrane and rapid degradation (Fig. 9), showing that carbohydrates are crucial for stabilization of both newly synthesized and recycling proteins. When expressed in insect cells, the H,K-ATPase ␣⅐␤ complex is not subject to degradation in the presence of tunicamycin even though the absence of glycosylation of the ␤ subunit results in a complete loss of the H,K-ATPase activity, indicating probable impairment of the proper conformation (24).
Mammalian cells apparently have more stringent quality control that responds to more subtle changes in protein conformation, such as changes due to lack of N-glycosylation, and that allows for selective elimination of nonconforming proteins. It appears also that degradation pathways were evolved in parallel with complex-type glycosylation as additional quality control steps for additional regulation of trafficking and sorting. For example, the hybrid-and high mannose-type glycosylated ␤ subunits formed in the presence of swainsonine and dMAN in LLC-PK1 cells were degraded much faster compared with the normal complex-type glycosylated protein (Figs. 7 and 8). In insect cells, however, the expressed ␤ subunit is not degraded but instead forms a functionally active complex with the ␣ subunit (24) even though it contains only high mannose-type carbohydrate chains due to the lack of terminal glycosylation.

N-Glycosylation Site Mutations Slow Down ER-to-Golgi Trafficking and Apical Membrane Delivery and Enhance
Internalization of the ␤ Subunit-Mutation of any one of the seven N-glycosylation sites reduced the apical content of YFP-␤ (Fig.  5). The degradation rate was approximately the same in the wild-type protein and mutants (Fig. 8, compare wt and N5). This leaves three possible steps that could be affected by mutations: ER-to-Golgi trafficking, apical membrane delivery from the TGN (or endosomes), and internalization from the Cells expressing wild-type or N5 mutant YFP-␤ were incubated with or without cycloheximide (CHX; 20 g/ml), a protein synthesis inhibitor. In half of the wells containing wild-type YFP-␤-expressing cells, swainsonine (2 g/ml) was applied 48 h prior to addition of cycloheximide. After 3 or 6 h of incubation with cycloheximide, the apical and total cellular contents of complex-type YFP-␤ (wild-type protein (wt) and N5) or hybrid-type YFP-␤ (wild-type protein ϩ swainsonine) were determined (see "Experimental Procedures") as a fraction of the YFP-␤ content in a corresponding control well with no cycloheximide. A decreased amount of YFP-␤ upon incubation of cells with cycloheximide reflected the degradation rate of the protein. The bar graphs demonstrate that, in the presence of the glycosylation inhibitor swainsonine, the intracellular degradation of wildtype YFP-␤ was significantly enhanced, whereas the degradation rate of the mutant lacking the N5 glycosylation site remained unchanged compared with that of wild-type YFP-␤ in the absence of swainsonine. *, significant difference compared with the YFP-␤ content in a corresponding control well with no cycloheximide (p Ͻ 0.05, n ϭ 3). apical membrane (Fig. 10). Since the wild-type and mutant YFP-␤ fusion proteins were expressed in the same cell line, the components of cell machinery such as protein complexes recognizing sorting signals, vesicle movement, and cytoskeleton that are involved in trafficking and sorting of mutants are the same as for the wild-type protein. Therefore, the only difference between the mutants and the wild-type protein is the sorting or trafficking information encoded within the protein or carbohydrate structure. The changes in distribution of YFP-␤ between cellular compartments due to mutations must result from the changes in signal sorter molecular recognition, which in turn affect the rates of corresponding trafficking steps. Therefore, the N-glycosylation site mutations might affect proper protein folding and quality control in the ER, sorting signal recognition in the TGN and endosomes, and/or endocytosis from the apical membrane (Fig. 10).
ER-to-Golgi trafficking was impaired in all the mutants except N6. The most significant effect, a 6.4-fold decrease, was detected in N7; and in the other mutants, the ER-to-Golgi trafficking was inhibited from 1.1-to 2.3-fold (Fig. 5). Mutations probably result in a longer interaction of the protein with chaperones in the calnexin-calreticulin cycle and hence in a longer retention time in the ER (19). However, the ER retention of the mutants is not the only reason for a dramatic decrease in apical content. In all the mutants, the apical content was decreased to a greater degree than the ER-to-Golgi trafficking (Fig. 5). Therefore, another reason for the decrease in the apical ␤ subunit is an increase in endocytosis efficiency in the mutants from 1.1-to 3.1-fold ( Fig. 6 and Table I). In addition, apical sorting was impaired by up to 2.3-fold as shown by a decrease in the apical/basolateral ratio in the mutants in Fig. 3 as quantified in Table I. Therefore, the 1.5-15-fold decrease in apical content in the mutants is a result of decreased ER-to-Golgi trafficking, impaired apical sorting, and enhanced endocytosis.
Distinct Roles of Core Region and Terminal Sugars in Apical Sorting, Internalization, and Degradation of the ␤ Subunit-Mature complex-type glycosylated YFP-␤, which undergoes sorting and trafficking from the TGN to the plasma membrane, internalization, recycling, and degradation, contains the threemannosyl core and terminal chains (Fig. 1). The effect of the Golgi mannosidase inhibitors compared with that of the Nglycosylation site mutations enables us to distinguish between the roles of core region and terminal carbohydrates in these steps. Inhibitors prevent trimming and further terminal glycosylation and result in either the absence of all (dMAN) or most (swainsonine) terminal sugars normally present at all seven N-glycosylation sites in the glycoprotein. Mutations result in the absence of core region and terminal sugars at one of the seven N-glycosylation sites.
The effect of swainsonine and dMAN on the apical content of YFP-␤ was smaller compared with that found with the mutants, suggesting that the core region sugars contribute more to apical sorting compared with the terminal sugars. Swainsonine significantly increased the rate of internalization of YFP-␤. The internalization rate of wild-type YFP-␤ in the presence of swainsonine was higher compared with that of any of the mutants. This suggests that terminal sugars are important for stabilization of the ␤ subunit in the apical membrane. The enhanced rate of endocytosis in the N1, N3, and N5 mutants is probably a result of a lack of terminal sugars in these particular sites. To find out which of the terminal sugars are essential, we treated cells with specific terminal glycosidases. It appears that terminal sialic acid and ␤-galactose residues are not essential for stabilization of the ␤ subunit in the apical membrane since sialidase and galactosidase treatment of the cells did not change the amount of apical YFP-␤ (Fig. 9). Thus, some other terminal carbohydrate residues such as ␣-galactose, fucose, and terminal N-acetylglucosamine might be important for ␤ subunit stabilization.
Both swainsonine and dMAN significantly decreased the total amount of YFP-␤ (Fig. 7), suggesting that they promote degradation of the protein. Indeed, the rate of YFP-␤ degradation that was seen as a decrease in protein content upon incubation with cycloheximide was much higher after swainsonine treatment (Fig. 8). Therefore, terminal glycosylation protects the protein from being targeted to degradation pathways. The degradation rate was not increased in the mutants lacking both core region and terminal sugars at one of seven N-glycosylation sites (Fig. 8). This might imply that specific mannose residues exposed at the chain termini in the high mannose-and hybridtype glycoproteins (after dMAN and swainsonine treatment) act as degradation targeting signals. Without inhibitors, in complex-type wild-type or mutant YFP-␤, these mannose residues are not present due to the trimming by Golgi mannosidases, and the mannoses of the three-mannosyl core are concealed by the terminal sugars (see Fig. 1). The mannose residues are also concealed in the products of deglycosylation by sialidase and galactosidase, perhaps explaining why sialidase and galactosidase treatment of the cells did not decrease the total content of YFP-␤ (Fig. 9).
Role of N-Glycans as Apical Sorting Signals-Certain Nglycosylation sites in the ␤ subunit appear to be more important for trafficking and apical sorting than the others. For example, the N7 site seems to be the most critical in the ␤ subunit since its removal resulted in complete ER retention of the protein. The other mutations that significantly increased the ER retention of the protein (N1, N3, and N5) also affected the other two trafficking steps, apical sorting and internalization. On the other hand, the N2, N4, and N6 mutations only slightly affected all three steps. This might suggest that the carbohydrate chains located at N1, N3, N5, and N7 are critical for the conformation of the protein necessary for the molecular recognition of trafficking signals for ER exit and then of apical sorting signals in the TGN and endosomes. A specific conformation might also be required for stability of the protein in the apical membrane, which would explain why the N1, N3, and N5 mutants were internalized faster. This is compatible with the suggestion that N-glycans provide a structural support for apical sorting (25).
However, the data presented here show that particular carbohydrate residues may also play a specific role in trafficking and apical sorting. We found that terminal sugars are essential for stabilization of the protein in the apical membrane, whereas the core region sugars are more important for apical sorting in the TGN and endosomes. This might imply that specific carbohydrate residues act as sorting signals. It is possible that terminal sugars anchor the protein in the apical membrane by interacting with unknown apical proteins or lipids in lipid rafts, whereas the particular core region carbohydrate residues (core N-acetylglucosamine or mannose) act as primary apical sorting signals that are recognized by lectin cargo receptors in the TGN and endosomes, in agreement with the proposed earlier model (8).

Role of N-Glycosylation of the ␤ Subunit in Stability and Apical Sorting of the Gastric H,K-ATPase ␣⅐␤ Holoenzyme-
The assembly of the ␣ and ␤ subunits of the gastric H,K-ATPase occurs in the ER (26). Expression studies in various cell types have shown that the presence of the ␤ subunit is required for the correct folding and membrane insertion of the ␣ subunit in the ER and its stabilization and subsequent trafficking (12,26,27). If expressed alone, the ␣ subunit is retained in the ER and is degraded. In contrast, even in the absence of the ␣ subunit, the ␤ subunit is able to fold properly, to undergo full maturation, and to traffic to the plasma membrane in both non-polarized and polarized cells (13,28,29). Carbohydrate residues linked to the ␤ subunit play a significant role in quality control of the ␤ subunit alone in non-polarized cells (13) and in polarized LLC-PK1 cells or in assembled ␣⅐␤ complexes (12,24) in non-polarized cells. Mutation of particular N-glycosylation sites impairs membrane targeting of the ␤ subunit in non-polarized HEK-293 cells (13) and apical sorting in LLC-PK1 cells. Recent findings indicate that mechanisms of membrane targeting in non-polarized cells are similar to sorting in polarized cells (5, 7, 30 -32) and that apical and basolateral proteins even in non-polarized cells are sorted into different transport containers. Data on H,K-ATPase expression in nonpolarized cells favor the primary role of the ␤ subunit and not the ␣ subunit in the apical sorting of the H,K-ATPase since the presence of the ␣ subunit does not restore the plasma membrane targeting of the H,K-ATPase ␣⅐␤ complexes containing N-glycosylation site-mutated ␤ subunits (12). This might suggest that N-glycosylation in the ␤ subunit is essential for apical sorting of the H,K-ATPase ␣⅐␤ holoenzyme in parietal cells.
In conclusion, N-glycosylation, in particular at the N1, N3, N5, and N7 glycosylation sites, is crucial for apical localization of the H,K-ATPase ␤ subunit in polarized LLC-PK1 cells. Core glycosylation and trimming in the ER are critical for proper ␤ subunit folding and quality control. Core region sugars are more important than terminal sugars for apical sorting of the protein in the TGN and/or endosomes. Terminal glycosylation is important for stabilization of the protein on the apical membrane and protection of the internalized protein from degradation. Presumably, these conclusions extend to other apically targeted glycosylated proteins.

Olga Vagin, Shahlo Turdikulova and George Sachs
Membrane Trafficking and Apical Sorting -Glycosylation in N Subunit as a Model to Study the Role of β The H,K-ATPase