11Research Article

IntroductionIn well-polarized epithelia, the plasma membranes arepartitioned into the apical and basolateral domains byimpermeable tight junctions (the canalicular andlateral/sinusoidal domain in hepatocytes, respectively). Thisstructural design imposes a need for these cells to developsophisticated cellular machinery to correctly deliverbiomolecules (including integral membrane proteins), duringor after their biogenesis, to specific membrane compartments(Brown and Stow, 1996; Mostov et al., 2003; Rodriguez-Boulan et al., 2005). The protein sorting events result in theformation and maintenance of the individual specificcomposition of the apical and basolateral domains, which areessential for many cellular functions executed by the polarizedepithelial cells (Nelson and Yeaman, 2001). Defects in proteinsorting/trafficking are often associated with diseases(Altschuler et al., 2003).

Integral membrane proteins destined to either apical orbasolateral membrane may undergo direct sorting from thetrans-Golgi network (TGN) (Rodriguez-Boulan et al., 2005;Simons and Wandinger-Ness, 1990) or selected retentionand/or degradation that results in differential distributions atone surface domain or the other. In hepatocytes, a thirdpathway (indirect apical targeting) exists that delivers certainbasolateral proteins to the apical membrane by a transcytosisroute (Hubbard, 1991). Various signals or cell machinerieshave been shown to be involved in the protein homing process,including sorting motifs on the cargo molecules (Corbeil et al.,1992), glycosyl-phosphatidylinositol (GPI) anchorage (Brown

et al., 1989; Lisanti et al., 1988), inclusions in the lipid rafts orlipid microdomains (Simons and Ikonen, 1997) and theinteractions with proteins that are being specifically sorted.

Apart from these relatively well-recognized sortingmechanisms, the role played by protein glycosylation,especially N-glycans, in regulating protein targeting remainsomewhat controversial. In addition to the generally acceptedrole of protein glycosylation in modulating protein degradationand metabolism (Helenius and Aebi, 2004), some studies havedemonstrated a direct apical sorting effect of N-glycans insome epithelial cells (Gut et al., 1998; Scheiffele et al., 1995);however, there is also contradictory evidence to this finding(Laughery et al., 2003; Su et al., 1999).

In this study, we investigated sorting of the Na+/K+-ATPasein the polarized hepatic cells. Na+/K+-ATPase plays an importantrole in keeping ionic imbalance or homoeostasis across theplasma membrane (Skou and Esmann, 1992). Most Na+/K+-ATPase in polarized epithelia are present on the basolateralmembrane (Blitzer and Boyer, 1978); apical presences have onlybeen found in retinal pigment epithelia and choroid plexusepithelia (Gundersen et al., 1991; Rizzolo, 1999). Na+/K+-ATPase is a transmembrane enzyme that consists of a largecatalytic �-subunit, and a small type II membrane glycoprotein�-subunit (Lingrel, 1992). Which subunit dominates the sortingof the heterodimer is not well understood, and may be differentdepending on the cell types. Some reports have shown directbasolateral sorting by signals at the fourth transmembranedomain of the �-subunit (Dunbar et al., 2000), whereas otherstudies suggest that integration, metabolism, and routing of the

Polarized epithelia, such as hepatocytes, target theirintegral membrane proteins to specific apical or basolateralmembrane domains during or after biogenesis. The rolesplayed by protein glycosylation in this sorting processremain controversial. We report here that deglycosylationtreatments in well-polarized hepatic cells bydeglycosylation drugs, or by site-directed mutagenesis ofthe N-linked-glycosylation residues, all cause the Na+/K+-ATPase ��-subunit to traffic from the native basolateral tothe apical/canalicular domain. Deglycosylated ��-subunits

are still able to bind and therefore transport the catalytic��-subunits to the aberrant apical location. Such apicaltargeting is mediated via the indirect transcytosis pathway.Cells containing apical Na+/K+-ATPase appear to bedefective in maintaining the ionic gradient across theplasma membrane and in executing hepatic activitiesthat are dependent upon the ionic homeostasis such ascanalicular excretion.

Key words: Glycosylation, Protein targeting, Na+/K+-ATPase, Liver

Summary

Deglycosylation of Na+/K+-ATPase causes thebasolateral protein to undergo apical targeting inpolarized hepatic cellsWei-Nan Lian1, Tzu-Wei Wu1, Ro-Lan Dao1, Yann-Jang Chen2 and Chi-Hung Lin1,3,4,*1Institute of Microbiology and Immunology, National Yang-Ming University, 155 Sec. 2 Linong Street, Taipei 112, Taiwan2Department of Life Science, National Yang-Ming University, 155 Sec. 2 Linong Street, Taipei 112, Taiwan3Institute of Biophotonics Engineering, National Yang-Ming University, 155 Sec. 2 Linong Street, Taipei 112, Taiwan4Department of Surgery, Veteran General Hospital, 201 Sec. 2 Shih-Pai Road, Taipei 112, Taiwan*Author for correspondence (e-mail: linch@ym.edu.tw)

Accepted 21 September 2005Journal of Cell Science 119, 11-22 Published by The Company of Biologists 2006doi:10.1242/jcs.02706

Jour

nal o

f Cel

l Sci

ence

12

�-subunit are actively modulated by the �-subunit (Geering,2001; Hasler et al., 1998; Jaunin et al., 1993). In the retinalpigment epithelial cell, the apical distribution of Na+/K+-ATPaseseems to be due to the association of �-subunit with ankyrin(Gundersen et al., 1991; Nelson and Veshnock, 1987), whereasin MDCK cells, the expression and basolateral localization ofNa+/K+-ATPase are related to plasma membrane contact or theinteraction between neighboring cells (Shoshani et al., 2005).Little is known regarding the sorting of Na+/K+-ATPase inpolarized hepatic cells.

We report here that deglycosylation treatments of theheavily glycosylated Na+/K+-ATPase �-subunit by eitherpharmacological treatment or site-directed mutagenesistreatment are able to effectively cause this basolateral integralmembrane protein to be transported to the apical domain inwell-differentiated and polarized hepatic cells. The wronglytargeted �-subunit could still bind to the catalytic �-subunit,causing a proportion of the Na+/K+-ATPase proteins to beincorrectly targeted to the apical membrane. Such defects insorting of the Na+/K+-ATPase appear to confer a functionaldeficiency on these hepatic cells.

ResultsDeglycosylation by pharmacological reagents causedaberrant apical targeting of the basolateral residentNa+/K+-ATPaseWe have previously reported that Hep G2 cells cultured for 3days form bile canaliculi (or the apical domain) that arecharacterized by a high concentrations of actin filaments inthe microvilli (Lian et al., 1999). In these ‘well-polarized’hepatic cells, membrane proteins were correctly transportedto their specific membrane domains. Na+/K+-ATPase, and E-cadherin and CD147/CE 9 (Bartles et al., 1985; Spring et al.,1997) were present mainly on the basolateral membrane (Fig.1A, arrowheads). The bile canaliculi were therefore stainedred (arrows) in the color-merged panels. On the other hand,dipeptidylpeptidase IV (DPPIV) was present predominantlyon the apical domain (Abbott et al., 1994), making the bilecanaliculi in the color-merged panel appear yellow (arrows).Treating Hep G2 cells with 20 �M tunicamycin over the last24 hours of the 3 day culture period caused a redistributionof Na+/K+-ATPase from the native basolateral localizationto the apical bile canaliculi domain (Fig. 1B, doublearrowheads), but had little effect on the distribution of E-cadherin, CD147, or DPPIV (Fig. 1C). The degree of apicaltargeting was determined by examining the presence (fromthe gray-scale image) of the membrane glycoproteins ofinterest in the bile canaliculus. A total of 500 bile canaliculiwere counted; three separate experiments were included inthe statistics. Viability of the cells was not significantlyaffected by the drug exposure (data not shown).

It should be noted that the tunicamycin treatment did notcause Na+/K+-ATPase to be retained in the ER (Fig. 1D). Usingcalreticulin as an ER maker, high magnification examinationof the double IF-stained cells revealed that after tunicamycintreatment the presence of Na+/K+-ATPase on the basolateralplasma membrane (arrowheads) was significantly reduced.There was an increase in Na+/K+-ATPase in the vesicularcompartments of the cytosol (arrow), but no apparentaccumulation of the deglycosylated proteins in the ER wasfound.

Different degrees of apical sorting by Na+/K+-ATPase werenoticed when the Hep G2 cells were treated with variousglycosylation inhibitors, including tunicamycin (TM), 1-deoxymannojirimycin (DMJ), 1-deoxynojirimycin (DNJ) andkifunensine (KIF). In the control conditions (CTL), the �-subunit of Na+/K+-ATPase was heavily glycosylated; majorbands were concentrated at around 50 kDa (***�, Fig. 1E).Adding 20 �M TM, 1 mM DNJ, 0.2 mM KIF or 1 mM DMJto Hep G2 cells for 24 hours resulted in a significant declinein the fully glycosylated proteins (in the cases of TM, DMJ andKIF), or an increase in the intermediately glycosylated proteins(**�, in the cases of DMJ, DNJ and KIF), or the core proteins(*�, in the cases of TM and KIF).

Localization studies revealed that 82%, 50%, 48% or 18%of the bile canaliculi (or apical domains) contained Na+/K+-ATPase �-subunit following TM, DMJ, DNJ or KIFtreatments, respectively (Fig. 1F). Note that exposure to KIFappeared to cause the most profound deglycosylation ofNa+/K+-ATPase �-subunit (with some degradation, arrow, Fig.1E) (see also Elbein et al., 1990), yet the effect on apicaltargeting was relatively minor compared with otherdeglycosylation compounds. The nature underlying this

Journal of Cell Science 119 (1)

Fig. 1. Deglycosylation by pharmacological treatments causesaberrant apical targeting of the basolateral Na+/K+-ATPase.(A) Polarized Hep G2 cells (CTL) were subjected toimmunofluorescence (IF) staining for Na+/K+-ATPase, E-cadherin,CD147 and DPPIV (green in the colour-merged channel), and co-labelled with Rhodamine-phalloidin (red) to identify F-actin-enriched bile canaliculi (arrows). Na+/K+-ATPase, E-cadherin andCD147 were present mainly on the basolateral membrane(arrowheads), whereas DPPIV was an apical marker (arrows).(B) Hep G2 cells were treated with 20 �M tunicamycin (TM) for 24hours before the IF experiments. Note that the Na+/K+-ATPase wasaberrantly targeted to the apical domain (double arrowheads, stainedyellow in the colour panel). Tunicamycin treatment had little effecton the localization of basolateral E-cadherin, CD 147 or apicalDPPIV. (C) Statistical analysis of the bile canaliculi that containedthe membrane proteins indicated, before (open bars) and after (filledbars) tunicamycin treatment. More than 500 bile canaliculi werecounted in each experiment; data are the means ± s.d. of threeexperiments. (D) Hep G2 cells were subjected to double-IF stainingto reveal Na+/K+-ATPase (green) and ER (by anti-calreticulinantibody, red) before and after tunicamycin treatment. Na+/K+-ATPase was present mainly along the basolateral membrane(arrowheads) in control cells. After tunicamycin treatment, thepresence of Na+/K+-ATPase was greatly reduced from the basolateralmembrane, and gradually increased in the cytoplasmic vesicles(arrows), with no obvious accumulation in the ER. (E) Western blotanalysis using Na+/K+-ATPase �-subunit-specific antibodies on HepG2 cells treated with mock solution (CTL) or various glycosylationinhibitors: tunicamycin (TM), 1-deoxy-mannojirimycin (DMJ), 1-deoxynojirimycin (DNJ) and Kifunensine (KIF). The fullyglycosylated (***�), intermediately glycosylated (**�) and the core(*�) proteins of Na+/K+-ATPase �-subunit are indicated. Note thatthe KIF treatment resulted in significant degradation of the �-subunitprotein (arrow). (F) After adding deglycosylating drugs for 24 hours,the percentage of bile canaliculi that contained mistargeted Na+/K+-ATPase was calculated. The bile canaliculi were recognized by F-actin staining; those positively stained for Na+/K+-ATPase �-subunitwere identified. At least 500 bile canaliculi obtained fromapproximately 25 microscopic images were included in thecalculation. Data are the means ± s.d. of three experiments. Bars, 20�m (A,B); 2 �m (D).

Jour

nal o

f Cel

l Sci

ence

13Deglycosylation mistargets Na+/K+-ATPase

Fig. 1. See previous page for legend.

Jour

nal o

f Cel

l Sci

ence

14

discrepancy was unclear. We could not rule out the possibilitythat KIF might affect the apical targeting process undergoneby the deglycosylated �-subunits.

Deglycosylation by site-directed mutagenesis alsocaused apical targeting of the Na+/K+-ATPase �-subunitIn addition to the pharmacological experiments, which areoften associated with non-specific drug effects, we performeda series of site-directed mutagenesis experiments tospecifically address the effects of protein glycosylation insorting the Na+/K+-ATPase �-subunit. Asn124 and Asn240(the two N-linked glycosylation residues of the �-subunit) orboth were changed to glutamines (N124Q, N240Q orN124/240Q, respectively); a FLAG tag was added to themutated gene for detection and biochemical purification. HepG2 cells transfected with mock solution, the wild-type, N124Q,N240Q, or N124/240Q �-subunit were subjected to westernblot analyses (Fig. 2A) and localization studies (Fig. 2B). Theexogenous wild-type Na+/K+-ATPase �-subunit containedfully glycosylated proteins (***�), intermediates (**�) and asmall amount of core protein (***�). Tunicamycin treatment(TM) resulted in a decrease in the glycosylated forms and acorresponding increase in the core protein. Hep G2 cellstransfected with N124Q Na+/K+-ATPase �-subunit containedintermediates and core proteins but very little fullyglycosylated form; after the deglycosylation by tunicamycin,only the core protein was left. In the N240Q experiments,intermediate forms were more abundant than the core proteinin the control cells; after deglycosylation the core proteinsbecame the predominant form, but there was still a significantamount of intermediates left. N124/240Q contained only coreproteins before or after tunicamycin treatment. These resultssuggested that the addition of carbohydrates to Asn124 andAsn240 accounted for the major protein glycosylation inNa+/K+-ATPase �-subunit in hepatic cells, and thatglycosylation at Asn124 was more important than that atAsn240.

Localization of the exogenously introduced Na+/K+-ATPase�-subunits was done by IF staining (anti-FLAG, the greenchannel of the color-merged panel, Fig. 2B); the apicaldomains were visualized by F-actin staining (arrow). Hep G2cells transfected with FLAG vector alone (CTL) exhibitedfluorescent signals inside the cell without any discernibleplasma membrane pattern. The wild-type Na+/K+-ATPase �-subunit proteins were found only along the basolateralmembrane (arrowheads) and were absent at the apical domain(arrow). All three deglycosylation mutants (N124Q, N240Qand N124/240Q) contained �-subunit signals at the apicalmembrane (double arrowheads), in addition to their basolaterallocalization.

Deglycosylated Na+/K+-ATPase �-subunit could stillinteract with the �-subunitWe wondered if the deglycosylated �-subunit could stillinteract or form a heterodimer with the �-subunit.Immunoprecipitation experiments (Fig. 3A) demonstrated thatthe immunoprecipitates pulled down by either the anti-�-subunit or anti-�-subunit antibodies contained both �-subunitand �-subunit proteins (mainly the fully glycosylated form***� and intermediately glycosylated form **�) in controlconditions. Deglycosylation of the �-subunit by tunicamycin

treatment did not interfere with its interaction with the �-subunit as evidenced by the presence of both �- and �-subunits

Journal of Cell Science 119 (1)

Fig. 2. Deglycosylation by site-directed mutagenesis also causedapical targeting of Na+/K+-ATPase �-subunit. (A) Hep G2 cellstransfected with control vector (Mock), the wild type (WT) or one ofthe three deglycosylated and FLAG-tagged �-subunit genes (N124Q,N240Q, N124/240Q) were treated with tunicamycin (TM) or not (–)before western blot analyses using anti-FLAG antibody. The fullyglycosylated (***�), intermediately glycosylated (**�), and the core(*�) proteins of the Na+/K+-ATPase �-subunit are indicated. (B) HepG2 cells transfected with control vector (CTL), the wild-type (WT),or one of the three deglycosylated and FLAG-tagged �-subunit genes(N124Q, N240Q, N124/240Q) were subjected to IF staining usinganti-FLAG antibody (green in the merged panel); F-actin staining(red) was used to localize the apical domain (arrow). Note that thewild-type �-subunit proteins were found only along the basolateralmembrane (arrowheads), whereas all three mutants exhibited apicalpresence (double arrowheads). Bar, 5 �m.

Jour

nal o

f Cel

l Sci

ence

15Deglycosylation mistargets Na+/K+-ATPase

(especially the appearance of �-subunit core proteins, *�)in the individually prepared immunoprecipitates. Similarobservations were made when the deglycosylated �-subunitmutants were applied (Fig. 3B). The immunoprecipitateisolated by the anti-FLAG antibody, which contained proteinsinteracting with the exogenous FLAG-tagged wild type or eachof the three deglycosylated �-subunit mutants, all containedthe endogenous �-subunit protein. Similarly, theimmunoprecipitate prepared by the anti-�-subunit antibodyalso contained the exogenous �-subunit protein, present atmolecular masses corresponding to their degrees ofglycosylation (Fig. 2A). Co-transfected EYFP-tagged �-tubulin was used as the transfection control.

The interactions between the �-subunit and deglycosylated�-subunits suggested that the �-subunit (although not aglycoprotein itself) could be targeted to the apical domain bythe deglycosylation treatments. This possibility was tested bythe localization studies (Fig. 3C-D). In control Hep G2 cells,both �- and �-subunits were present only along the basolateralmembrane (arrowheads, Fig. 3C CTL). After tunicamycintreatment, a significant amount of both �- and �-subunits

exhibited apical targeting (double arrowheads). In Hep G2 cellstransfected with EGFP-tagged wild-type (WT) �-subunit, theexogenous �-subunit (EGFP-�, or the green channel in thecolor-merged panel, Fig. 3D) were found mainly along thebasolateral membrane, together with the endogenous �-subunit(blue). The apical domain appeared red with F-actin staining.In N124Q-transfected cells, some mutated �-subunit proteinwas incorrectly sorted to the apical domain (doublearrowheads), accompanied by the presence �-subunits at thataberrant localization. Similar observations were made inN240Q- and N124/240Q-transfected cells (data not shown).

The rate of metabolism was similar when thedeglycosylated Na+/K+-ATPase �-subunits werecompared with wild-type proteinsTo test if the rate of protein metabolism was significantlyhigher in the deglycosylated �-subunits than in the wild type,we conducted a series of ‘pulse-chase’ experiments using[35S]methionine labeling to measure the rates of proteindegradation. As shown in the autoradiograph (Fig. 4A) and thecorresponding normalization plot (Fig. 4B), we found no

Fig. 3. Deglycosylated Na+/K+-ATPase �-subunits can still interact with the catalytic �-subunits. (A) Hep G2 cells treated withtunicamycin (TM) or untreated weresubjected to immunoprecipitation (IP)experiments using either anti-�-subunitantibody (anti-�) or anti-�-subunit antibody(anti-�). The resulting immunoprecipitateswere analyzed by western blotting. Thepositions of the �-subunit and the fullyglycosylated (***�), intermediatelyglycosylated (**�), and the core (*�)proteins of the �-subunit are indicated.(B) The interactions between the exogenousFLAG-tagged �-subunit (the wild-type andthree deglycosylated mutant constructs) andthe endogenous � protein were tested by IP-western analyses; co-transfected EYFP-�-tubulin were used as a control. (C) Thelocalization of �- and �-subunits before andafter tunicamycin treatment was visualizedby IF using antibodies specific to either �- or�-subunit (green), together with F-actinstaining (red). Note the basolateral only(arrowheads) and the apical presence (doublearrowheads) of both subunits in the control(CTL) and tunicamycin-treated (TM) cells,respectively. (D) Hep G2 cells transfectedwith EGFP-labelled �-subunit (wild-type orN124Q mutant, green), were stained forendogenous �-subunit (blue) and F-actin(red). Note both wild-type �-subunit and �-subunit were found only along the basolateralmembrane (arrowheads) in the wild-typetransfectants, whereas a portion of theN124Q �-subunit and the endogenous �-subunit was aberrantly translocated to theapical domain (double arrowheads) in HepG2 cell transfected with N124Q �-subunit.Bars, 10 �m.

Jour

nal o

f Cel

l Sci

ence

16

significant difference between the wild-type �-subunits and thethree deglycosylated mutant proteins tested. The co-transfectedEYFP-tagged �-tubulin was used as an internal control fortransfection efficiency in Hep G2 cells, and because of its longmetabolic half-life, as a reference to normalize the data takenat different time points (Fig. 4B).

The apical targeting of deglycosylated Na+/K+-ATPase �-subunits was mediated via the indirect transcytosis routeThe route of apical trafficking by the deglycosylated Na+/K+-ATPase �-subunit was investigated. As shown in Fig. 5A-B,the �-subunits residing on the basolateral membrane were

labeled with antibody probes at 4°C (to prevent the probefrom being internalized) for 15 minutes and subjected toimmunofluorescence staining. After increasing the incubationtemperature to 37°C, the dynamic redistributions of the �-subunits, which were fluorescently labeled, were monitored atthe basolateral membrane over a 60 minute observation period.The percentage of bile canaliculi that contained the labeled �-subunits was calculated and plotted as a function of time (Fig.5C). In the control Hep G2 cells (Fig. 5A), we observeda gradual decrease of fluorescence from the basolateralmembrane (arrowheads). There was no discernibleaccumulation of the fluorescently labeled �-subunit in theapical domain (open circles, Fig. 5C). On the other hand, incells treated with tunicamycin (Fig. 5B), we observed not onlya reduction of basolateral fluorescence, but also a progressiveappearance (double arrowheads, Fig. 5B) and an increase in thenumber (filled squares, Fig. 5C) of bile canaliculi thatcontained the �-subunits transported from the basolateraldomain. These results indicated that the transcytosis pathwaywas involved in targeting deglycosylated Na+/K+-ATPase �-subunits to the apical membrane.

Aberrant apical targeting of Na+/K+-ATPase perturbedthe maintenance of ionic homeostasis in polarizedhepatic cellsTo address the functional consequence of the aberrant apicaltargeting of Na+/K+-ATPase, we performed a series of livecell experiments to test the maintenance of ionic homeostasis.Intracellular Na+ concentrations ([Na+]i) of the well-polarizedHep G2 cells were continuously monitored before, during andafter washout of the solution containing high sodium (300mM NaCl for 10 seconds) (Fig. 6). In the experiments usingnon-ratio Sodium Green to measure [Na+]i (Fig. 5A), wefound that the control cells (CTL) were able to keep [Na+]iwithin a very limited range during the high-Na+ challengeprocess, except for a brief fluorescence intensity downturnimmediately after the addition of high-Na+ solution (thereason for the deflection is unknown but might be due tomorphogenesis of the cell in the transient hyperosmoticenvironment). On the other hand, cells treated withtunicamycin (TM) for 12 hours, or with ouabain (a Na+/K+-ATPase inhibitor) for 10 minutes exhibited an apparent andprogressive increase in [Na+]i in response to the elevatedextracellular Na+ concentration. Treating Hep G2 cells withboth tunicamycin and ouabain did not cause any furtherincrease in [Na+]i during the high-Na+ challenge whencompared with ouabain treatment alone. This finding arguesthat the tunicamycin effects are ‘confined’ by the ouabaineffects, suggesting strongly that the defective ionichomeostasis observed in tunicamycin-treated cells isprobably due to defects in sodium excretion, rather thanmechanisms such as an unregulated increase of sodiuminflux.

To quantify [Na+]i in molar concentrations, we applied ratiosodium indicator SBFI to the measurements (Fig. 6B,C). Notethat the control Hep G2 cells or cells expressing the wild-typeNa+/K+-ATPase �-subunit (WT) both maintained their [Na+]iwithout any notable increase during the high-Na+ exposure. Incontrast, the Hep G2 cells treated with tunicamycin (TM),ouabain, or expressing deglycosylated forms of Na+/K+-ATPase �-subunit (N124Q, N240Q and N124/240Q), all

Journal of Cell Science 119 (1)

Fig. 4. The rate of protein degradation was similar between the wild-type and the deglycosylated mutants of �-subunits. (A) Hep G2 cellstransfected with control vector (Mock), the wild type (WT) or one ofthe three deglycosylated and FLAG-tagged �-subunit genes (N124Q,N240Q, N124/240Q) were labelled with [35S]methionine for 15minutes, followed by incubation in unlabelled culture medium for thetime periods indicated. Whole-cell lysates were extracted andexamined by anti-FLAG IP and autoradiography. (B) Quantitativeanalysis of the autoradiograph. The co-transfected EYFP tagged �-tubulin was used as a control for transfection efficiency. The 35Ssignals of the �-subunit were first normalized by the signals of theco-transfected EYFP-�-tubulin. The data recorded at 1, 3 or 6 hourswere normalized against that recorded immediately after the washoutof [35S]methionine label (0 hour).

Jour

nal o

f Cel

l Sci

ence

17Deglycosylation mistargets Na+/K+-ATPase

responded to the high-Na+ challenges with [Na+]i upsurges.The measured [Na+]i before the high-Na+ challenge in cellstreated with tunicamycin, ouabain (Fig. 6C, asterisks) ortransfected with deglycosylated mutants (crosses) were allsignificantly higher than the control cells or cells expressingthe wild-type �-subunit, respectively. The relatively high [Na+]iin the mutant cells did not seem to affect their viability (datanot shown).

Bile canalicular secretion by the transcytotic transportwas inhibited by the deglycosylation treatmentsAlthough the Hep G2 cells containing the deglycosylatedapical Na+/K+-ATPase �-subunit were alive, some of theirfunctions were defective. We and other labs have previouslydemonstrated that polarized Hep G2 cells were capable ofexcreting FDA from the culture medium to the bile canaliculithrough a transcytotic transport pathway, whereas 3 kDa

Fig. 5. Apical targeting by the deglycosylated Na+/K+-ATPase �-subunitwas mediated via the indirect transcytosis pathway. (A) Polarized Hep G2cells were treated with monoclonal anti-Na+/K+-ATPase �-subunitantibody at 4°C for 15 minutes. After washout of the primary antibody,the cells were warmed to 37°C for the time periods indicated, then fixedand stained with fluorescently labelled secondary antibody. Within the 60minute observation period, no fluorescent signal appeared at the bilecanaliculi (arrows). (B) The same experiments were applied to the HepG2 cells treated with tunicamycin. The fluorescent signals were originallyfound only along the basolateral membrane (arrowheads); over time, therewas a progressive increase in fluorescence at the apical canaliculardomain (double arrowheads). (C) The percentage of bile canaliculi (BC)that contained the fluorescence generated from the mistargeted Na+/K+-ATPase �-subunit was calculated in the control (�) and tunicamycin-treated cells (�). The bile canaliculi were recognized by F-actin staining;those also positively stained for Na+/K+-ATPase �-subunit were identified. At least 500 bile canaliculi obtained from approximately 20microscopic images were included in the calculation. Results shown are the means ± s.d. of three independent experiments. Bars, 10 �m.

Jour

nal o

f Cel

l Sci

ence

18

dextran enters the bile canaliculi via permeable tight junctionsin a process termed ‘paracellular transport’ (Lian et al., 1999).When applying these functional assays to the deglycosylationexperiments (Fig. 7A), we found that about half of the bilecanaliculi in the control Hep G2 cells were loaded with thefluorescent dye (arrows, or the yellow bile canaliculi in thecolor-merged panel) after adding FDA to the culture media for

30 minutes. The canalicular excretion of FDA was profoundlyreduced by tunicamycin treatment; in the drug-treated Hep G2cells, only about 20% of the bile canaliculi were capable ofexcreting FDA, compared with 48% in control cells (P<0.05by Student’s t-test; filled bars, Fig. 7C). On the other hand, theentrance of 3 kDa dextran to the bile canaliculi via theparacellular transport pathway was not significantly affected by

Journal of Cell Science 119 (1)

Fig. 6. Cells containingapical Na+/K+-ATPase weredefective in maintainingsodium homeostasis.(A) Polarized Hep G2 cellswere loaded with non-ratiosodium indicator, SodiumGreen and then treated withmock solution (CTL),tunicamycin (TM) for 12hours, ouabain for 10minutes, or both. Thechanges in fluorescenceintensity before, during andafter washout of the highNa+ solution (Bar, 10seconds) were monitoredunder a fluorescencemicroscope, and plotted as afunction of time. The mean(line) ± s.d. (dashed line)curves are shown. (B) Thesame experiment as in A, butthe cells were loaded withthe ratio sodium indicator,SBFI. Ratio imaging by theintensity of emissionfluorescence excited by 340nm and 380 nm was plottedas a function of time. Thepharmacological andmutagenesis treatments areas indicated. (C) Thequantification of intracellularsodium concentrations forthe data set shown in B, invivo calibration for SBFIloading was performed.Cells treated withtunicamycin and ouabain (*),or transfected with threedeglycosylated �-subunitmutant constructs (+), allexhibited a higher [Na+]ithan the control cells beforethe high-Na+ challenge(P<0.05 by Student’s t-test).Means ± s.d. are shown.

Jour

nal o

f Cel

l Sci

ence

19Deglycosylation mistargets Na+/K+-ATPase

the drug treatment (Fig. 7B). An hour afteradding 3 kDa dextran to the culture media, 52%and 45% of the bile canaliculi were found tocontain the dextran markers in the controland tunicamycin-treated cells, respectively.Similar conclusions were drawn from thefunctional studies using cells transfected withdeglycosylated mutants of �-subunit (data notshown).

DiscussionIn many epithelia other than hepatocytes (suchas MDCK or Caco-2 cells), basolateraltargeting of the integral membrane protein isgoverned mainly by specific sorting motifs inthe cytoplasmic domain (Casanova et al., 1991;Hunziker et al., 1991; Matter et al., 1992;Mostov et al., 1986; Rodriguez-Boulan et al.,2005), whereas the physical properties of theprotein itself (e.g. GPI anchorage orassociations with lipids) appear to mediateselective apical sorting (Brown and Rose,1992; Lisanti et al., 1989; Shvartsman et al.,2003). It is generally accepted that thebasolateral sorting signals are predominantover the apical ones; the latter are revealed onlywhen the former are ablated. Some studies alsosuggest that N-glycan is an active apical sortingsignal, as the carbohydrate moieties may berecognized by (lectin-like) receptors presentalong the apical sorting pathway. Based on thisgeneral view, it is surprising to find in thisstudy that the removal of N-glycan from abasolateral membrane protein results in itsapical targeting.

The basolateral sorting signals for theNa+/K+-ATPase �-subunit, if present as theconventional views predict, should remain intacteven after the deglycosylation treatments usedhere. Yet, rather than stay in the basolateralmembrane, the N-glycan-depleted Na+/K+-ATPase �-subunit was found to travel from thebasolateral membrane to the apical domain viathe conventional ‘transcytosis’ or indirecttransport (Fig. 5) that is typically used forsorting apical resident proteins in hepatic cells.There is no evidence indicating the involvementof direct apical pathway (such as the sortingroute for canalicular ABC transporter) (Kippand Arias, 2000) for transporting deglycosylated�-subunits. The unusual journey taken by thedeglycosylated membrane protein involvesselective internalization from the basolateralmembrane, sorting among the endosomalcompartments, transportation within thevesicular compartments and reincorporation tothe apical membrane. How the removal of N-linked glycosylation in the �-subunit triggers orregulates these processes is unknown; however, it is intriguingto consider the possibility that removing the carbohydrates mayreveal new recognition epitopes or change the physical

properties of the glycoprotein, and thereby facilitate itsinteractions with the apical sorting machinery (Huet et al.,2003).

Fig. 7. Hep G2 cells containing apical Na+/K+-ATPase are defective in canalicularexcretion. (A) Polarized Hep G2 cells were treated with 20 �g/ml tunicamycin for24 hours (TM) or untreated (CTL). Fluorescein diacetate (FDA, green) was added tothe culture media for 30 minutes; the locations of bile canaliculi were visualized byF-actin staining (red). The bile canaliculi capable of excreting FDA via transcytotictransport were stained yellow (arrows), or stained red otherwise (arrowheads).(B) Fluorescently labelled 3 kDa dextrans (green) were added to the culture mediafor 1 hour; the locations of bile canaliculi were visualized by F-actin staining (red).The fluorescent dextran could enter the canalicular lumen via permeable tightjunctions between the neighbouring cells (the paracellular pathway). Bile canaliculiactive in paracellular transport were stained yellow (arrows), and stained red ifinactive (arrowheads). (C) Quantification of the bile canaliculi (BC) apical domainsactive in transcytotic (grey bars) or paracellular transport (open bars) in the presence(TM) or absence (–) of tunicamycin treatments. More than 500 apical domains werecounted for each experiment and results show the means ± s.d. *P<0.05 byStudent’s t-test. Bars, 20 �m.

Jour

nal o

f Cel

l Sci

ence

20

Our results argue against the view that the apical presenceof the Na+/K+-ATPase �-subunit may be due to selectiveretention of the glycoprotein which is randomly targeted toboth apical and basolateral domains. First, the aberrant apicalsorting was only found in the deglycosylated Na+/K+-ATPase�-subunit, but not in two other basolateral membrane proteins(E-cadherin and CD147) or the apical marker DPPIV; thisspecificity is unlikely to be accounted for by relatively non-specific control mechanisms such as protein degradation.Secondly, we found that the rates of protein degradation arealmost indistinguishable between the wild type and thedeglycosylated �-subunit mutants (Fig. 4) (see also Beggah etal., 1997). This finding is not unprecedented, there are severalexamples demonstrating that deglycosylation has no obviouseffect on metabolism, targeting, molecular complex formationor the function of the glycoprotein (Ackermann and Geering,1990; Laughery et al., 2003; Zamofing et al., 1988). Thirdly,although KIF treatment induced the most significant �-subunitdegradation, this drug is least effective in causing apical sortingof the �-subunit compared with the other glycosylationinhibitors tested (Fig. 1C,D). Although the mechanismsunderlying the lack of apical presence of the �-subunit uponKIF treatment are unknown, we argued that the KIF-mediateddeglycosylation did cause apical targeting of the �-subunit;however, the majority of the apical �-subunit was furtherdegraded perhaps because of other KIF effects. Interestingly,KIF has been shown to cause little glycoprotein degradationin many cell types (Belbeoc’h et al., 2003; Wang andAndrolewicz, 2000). The profound �-subunit degradation byKIF observed here might be related to its apical targeting effectin our hepatic cell model. Taken together, our results favor thenotion that the N-linked carbohydrates of the Na+/K+-ATPase�-subunit may play an active role in modulating theglycoprotein sorting process, and this modulation is notmediated though the regulation of protein degradation.

The interactions between the �- and �-subunit of Na+/K+-ATPase is an intriguing cell biology topic that involves a rangeof different and somewhat controversial views. With respect tothe basolateral residence of this ion pump, there are reportssuggesting the fourth transmembrane domain of �-subunit isthe active basolateral sorting signal (Muth et al., 1998),whereas other studies have demonstrated a role played by the�-subunit in guiding the transport of � component (Hasler etal., 1998; Laughery et al., 2003). The results shown in Fig. 3support the latter notion: namely, when the deglycosylatedNa+/K+-ATPase �-subunit is ectopically delivered to the apicaldomain, there are interactions between the deglycosylated�-subunit and �-subunit (which is not a glycoprotein andtherefore should not be directly affected by the deglycosylationtreatments applied) that appear to override the native sortingsignals and transport the heterodimer to the apical membrane.The simplest explanation of our results is that the Na+/K+-ATPase �-subunit can bind to the �-subunit and activelyregulate its sorting after biogenesis. Such an interactionbetween the �- and �-subunits may occur transiently or onlyalong certain steps of the sorting process; the functional �-subunits may exist on the plasma membrane dissociated fromthe �-subunit (Laughery et al., 2003).

Given a fixed amount of Na+/K+-ATPase �-subunit in a cell,and no evidence of increased protein synthesis followingthe deglycosylation treatments, the apical sorting of the

deglycosylated �-subunits is likely to reduce the presence ofcatalytic �-subunits at the native basolateral membranebecause a proportion of �-subunits have been translocated tothe apical domain, where the very different physicochemicalenvironment is able to render the apical catalytic enzymesinactive (Devonald et al., 2003; Koivisto et al., 2001). It istherefore reasoned that polarized hepatic cells containingapical Na+/K+-ATPase may experience defects in maintainingenough Na+/K+-ATPase to maintain ionic homeostasis.Indeed, this is what we have observed here. After thedeglycosylation treatments by either drugs or mutagenesisprocedures, the cells appear to maintain a relatively high[Na+]i (21-35 mM, Fig. 6C) compared with the control cells(8-19 mM), and their adjustments to the transient high-Na+

influxes are less effective (Fig. 6A,B). Furthermore, such ionichomeostatic defects may lead to the reduction in biliarysecretion (as indicated by the decreased transcytosis transportof FDA, Fig. 7A) because the sodium gradient across theplasma membrane plays a role in driving canalicular secretion(Burwen et al., 1992). On the other hand, the paracellulartransport, which is governed by tight junction permeability, isnot affected by the deglycosylation treatments (Fig. 7) (seealso Ihrke et al., 1993).

Defects in the trafficking of membrane proteins in polarizedepithelial cells are often associated with diseases, includingcystic fibrosis, Liddle’s syndrome, nephrogenic diabetesinsipidus and Dubin-Johnson syndrome (Altschuler et al.,2003). Apical mislocation of Na+/K+-ATPase is associated withhuman polycystic kidney disease (Ogborn et al., 1993; Wilson,1997; Wilson et al., 1991). The results shown in this reportsuggest that there is a more complex targeting role for theN-glycans of Na+/K+-ATPase (and other membraneglycoproteins) than has been previously thought.

Materials and MethodsCell culture and immunofluorescence stainingThe culture of the hepatoblastoma cell line, Hep G2 and immunofluorescence (IF)staining was according to methods described previously (Lian et al., 1999). Theprimary antibodies included those made against FLAG M2 and �-tubulin (fromSigma), Na+/K+-ATPase �-subunit (from Affinity Bioreagents), Na+/K+-ATPase �-subunit and CD147 (from BD Bioscience), calreticulin (ER marker, from Abcam,Cambridge, UK). The fluorophore-conjugated secondary antibodies were all fromJackson Immuno Research (West Grove, PA). Rhodamine-phalloidin (MolecularProbes) was used for F-actin staining at a concentration of 1 U/ml. The stainedsamples were observed under a confocal microscope (TCS-SP2, Leica Microsystem,Germany).

Immunoprecipitation and western blot analysisFor immunoprecipitation (IP) experiments, Hep G2 cells were transfected usingFuGENE 6 transfection Reagent (Roche), and lysed after 48 hour incubations withNET buffer (150 mM NaCl, 0.5% NP-40, 50 mM Tris, 1 mM EDTA and 1% TritonX-100, supplemented with protease inhibitors). The soluble proteins werecollected after centrifugation at 12,000 g at 4°C for 30 minutes, and incubatedwith 40 �l protein A-agarose beads coated with antibodies of interest. Theresulting immunoprecipitates were separated by 10.5% SDS-PAGE andtransfected to a nitrocellulose membrane (Bio-Rad). After confirmation of thepresence of proteins by Ponceau-S staining, standard western blotting procedureswere performed (Lian et al., 1999). The blotting signal was detected bySuperSignal Chemiluminescent substrate (Pierce) and recorded by Hyperfilm(Amersham-Pharmacia).

Mutagenesis of the Na+/K+-ATPase �-subunitPoint mutations of Na+/K+-ATPase �-subunit was created by the PCR overlappingmethod using the human sequence as the template. Asn124 and Asn240 werereplaced by Gln (resulting in two single mutants N124Q and N240Q, and one doublemutant N124/240Q). The constructed cDNA was transfer to the donor vector(pDNR-dual) of the Creator system (Clontech Laboratories) by fusion PCR, withor without the incorporation of FLAG or EGFP tags at the N-terminus of the

Journal of Cell Science 119 (1)

Jour

nal o

f Cel

l Sci

ence

21Deglycosylation mistargets Na+/K+-ATPase

polypeptide chain. Plasmids were transformed to E. coli (DH5�) for amplificationand purified by NucleoBond plasmid purification kits (Clontech).

Protein degradation assayMetabolism of the wild-type or mutant Na+/K+-ATPase �-subunits were determinedby a pulse-chase assay after radio-isotope labeling (Rajasekaran et al., 2001). Aftertransfection, Hep G2 cells were grown for 48 hours and starved in Met- and Cys-free DMEM containing 1% dialyzed FBS for 2 hours at 37°C before metaboliclabeling with 2 mCi/ml trans-[35S]methionine (Amersham-Pharmacia) for 15minutes at 37°C. The cells were washed twice with PBS and chased with isotope-free culture medium for the periods indicated, before the IP experiments. Theresulting autoradiograph was scanned and analyzed with Phoretix 1D standardsoftware (NonLinear Dynamics), and quantified by densitometry analysis.

Kinetic tracing of Na+/K+-ATPase �-subunits in living cellsKinetic tracing of Na+/K+-ATPase �-subunits in living cells was performed aspreviously described (Ihrke et al., 1993) with some modifications. Briefly, 3 daycultured Hep G2 cells were incubated with 0.1 �g/ml anti-Na+/K+-ATPaseantibodies at 4°C for 15 minutes. Cells were washed twice with cold serum-freemedium then incubated with complete medium at 37°C for the time periodindicated. Cells were then fixed and subjected to IF staining.

Intracellular Na+ measurementsThe intracellular sodium concentration [Na+]i of the living cells was monitored bySodium Green and SBFI-AM (Molecular Probes). The loading of dye was facilitatedby adding 0.02% Pluronic F-127 to the culture medium at 37°C for 60 minutes.Emission fluorescence of SBFI at 510 nm by dual excitation at 340/380 nm (usingHamamatsu monochromater) was recorded and analyzed with Acqua-Cosmossoftware (Hamamatsu) equipped with documented ratio algorithms (Gilon andHenquin, 1993). Calibration of SBFI-AM fluorescence was performed by addingcells with known concentrations of Na+ in the presence of 10 �g/ml of Na+

ionophore gramicidin D (Harootunian et al., 1989). The high-Na+ challenge wasperformed using PBS supplemented with 300 mM NaCl, pH 7.2. The challengeperiod was 10 seconds.

Biliary secretion assayBiliary secretions through transcytosis or paracellular transport route were measuredby loading the culture medium with Fluorescein diacetate (FDA, 5 �g/ml at 37°Cfor 30 minutes), or 3 kDa FITC-dextran (5 �g/ml at 37°C for 1 hour) as previouslydescribed (Lian et al., 1999). The bile canaliculi were visualized by Rhodamine-phalloidin staining. The percentage of bile canaliculi that contained excreted FDA(indicating competent transcytosis transport activity) or 3 kDa FITC-dextran(indicating competent paracellular transport activity) were calculated. Themicroscopy was done on a Leica DMIRBE inverted fluorescence microscopeequipped with a cool-CCD (ORCA-1394, Hamamatsu, Japan) and MetaMorphimaging system (Universal Imaging, West Chester, PA).

We thank Ming-Ta Hsu, Ting-Ting Liu and Cheng-Po Hu forfruitful discussions and Weber Chen for technical support. This workis supported by grants from National Science Council, UST-CNST,National Research Program for Genomic Medicine and National NanoScience and Technology Program, Taiwan, awarded to C.H.L.

ReferencesAbbott, C. A., Baker, E., Sutherland, G. R. and McCaughan, G. W. (1994). Genomic

organization, exact localization, and tissue expression of the human CD26 (dipeptidylpeptidase IV) gene. Immunogenetics 40, 331-338.

Ackermann, U. and Geering, K. (1990). Mutual dependence of Na,K-ATPase alpha- andbeta-subunits for correct posttranslational processing and intracellular transport. FEBSLett. 269, 105-108.

Altschuler, Y., Hodson, C. and Milgram, S. L. (2003). The apical compartment:trafficking pathways, regulators and scaffolding proteins. Curr. Opin. Cell Biol. 15,423-429.

Bartles, J. R., Braiterman, L. T. and Hubbard, A. L. (1985). Endogenous andexogenous domain markers of the rat hepatocyte plasma membrane. J. Cell Biol. 100,1126-1138.

Beggah, A. T., Jaunin, P. and Geering, K. (1997). Role of glycosylation and disulfidebond formation in the beta subunit in the folding and functional expression of Na,K-ATPase. J. Biol. Chem. 272, 10318-10326.

Belbeoc’h, S., Massoulie, J. and Bon, S. (2003). The C-terminal T peptide ofacetylcholinesterase enhances degradation of unassembled active subunits through theERAD pathway. EMBO J. 22, 3536-3545.

Blitzer, B. L. and Boyer, J. L. (1978). Cytochemical localization of Na+, K+-ATPase inthe rat hepatocyte. J. Clin. Invest. 62, 1104-1108.

Brown, D. and Stow, J. L. (1996). Protein trafficking and polarity in kidney epithelium:from cell biology to physiology. Physiol. Rev. 76, 245-297.

Brown, D. A. and Rose, J. K. (1992). Sorting of GPI-anchored proteins to glycolipid-

enriched membrane subdomains during transport to the apical cell surface. Cell 68,533-544.

Brown, D. A., Crise, B. and Rose, J. K. (1989). Mechanism of membrane anchoringaffects polarized expression of two proteins in MDCK cells. Science 245, 1499-1501.

Burwen, S. J., Schmucker, D. L. and Jones, A. L. (1992). Subcellular and molecularmechanisms of bile secretion. Int. Rev. Cytol. 135, 269-313.

Casanova, J. E., Apodaca, G. and Mostov, K. E. (1991). An autonomous signal forbasolateral sorting in the cytoplasmic domain of the polymeric immunoglobulinreceptor. Cell 66, 65-75.

Corbeil, D., Boileau, G., Lemay, G. and Crine, P. (1992). Expression and polarizedapical secretion in Madin-Darby canine kidney cells of a recombinant soluble form ofneutral endopeptidase lacking the cytosolic and transmembrane domains. J. Biol.Chem. 267, 2798-2801.

Devonald, M. A., Smith, A. N., Poon, J. P., Ihrke, G. and Karet, F. E. (2003). Non-polarized targeting of AE1 causes autosomal dominant distal renal tubular acidosis.Nat. Genet. 33, 125-127.

Dunbar, L. A., Aronson, P. and Caplan, M. J. (2000). A transmembrane segmentdetermines the steady-state localization of an ion-transporting adenosinetriphosphatase. J. Cell Biol. 148, 769-778.

Elbein, A. D., Tropea, J. E., Mitchell, M. and Kaushal, G. P. (1990). Kifunensine, apotent inhibitor of the glycoprotein processing mannosidase I. J. Biol. Chem. 265,15599-15605.

Geering, K. (2001). The functional role of beta subunits in oligomeric P-type ATPases.J. Bioenerg. Biomembr. 33, 425-438.

Gilon, P. and Henquin, J. C. (1993). Activation of muscarinic receptors increases theconcentration of free Na+ in mouse pancreatic B-cells. FEBS Lett. 315, 353-356.

Gundersen, D., Orlowski, J. and Rodriguez-Boulan, E. (1991). Apical polarity ofNa,K-ATPase in retinal pigment epithelium is linked to a reversal of the ankyrin-fodrinsubmembrane cytoskeleton. J. Cell Biol. 112, 863-872.

Gut, A., Kappeler, F., Hyka, N., Balda, M. S., Hauri, H. P. and Matter, K. (1998).Carbohydrate-mediated Golgi to cell surface transport and apical targeting ofmembrane proteins. EMBO J. 17, 1919-1929.

Harootunian, A. T., Kao, J. P., Eckert, B. K. and Tsien, R. Y. (1989). Fluorescenceratio imaging of cytosolic free Na+ in individual fibroblasts and lymphocytes. J. Biol.Chem. 264, 19458-19467.

Hasler, U., Wang, X., Crambert, G., Beguin, P., Jaisser, F., Horisberger, J. D. andGeering, K. (1998). Role of beta-subunit domains in the assembly, stable expression,intracellular routing, and functional properties of Na,K-ATPase. J. Biol. Chem. 273,30826-30835.

Helenius, A. and Aebi, M. (2004). Roles of N-linked glycans in the endoplasmicreticulum. Annu. Rev. Biochem. 73, 1019-1049.

Hubbard, A. L. (1991). Targeting of membrane and secretory proteins to the apicaldomain in epithelial cells. Semin. Cell Biol. 2, 365-374.

Huet, G., Gouyer, V., Delacour, D., Richet, C., Zanetta, J. P., Delannoy, P. andDegand, P. (2003). Involvement of glycosylation in the intracellular trafficking ofglycoproteins in polarized epithelial cells. Biochimie 85, 323-330.

Hunziker, W., Harter, C., Matter, K. and Mellman, I. (1991). Basolateral sortingin MDCK cells requires a distinct cytoplasmic domain determinant. Cell 66, 907-920.

Ihrke, G., Neufeld, E. B., Meads, T., Shanks, M. R., Cassio, D., Laurent, M., Schroer,T. A., Pagano, R. E. and Hubbard, A. L. (1993). WIF-B cells: an in vitro model forstudies of hepatocyte polarity. J. Cell Biol. 123, 1761-1775.

Jaunin, P., Jaisser, F., Beggah, A. T., Takeyasu, K., Mangeat, P., Rossier, B. C.,Horisberger, J. D. and Geering, K. (1993). Role of the transmembrane andextracytoplasmic domain of beta subunits in subunit assembly, intracellular transport,and functional expression of Na,K-pumps. J. Cell Biol. 123, 1751-1759.

Kipp, H. and Arias, I. M. (2000). Newly synthesized canalicular ABC transporters aredirectly targeted from the Golgi to the hepatocyte apical domain in rat liver. J. Biol.Chem. 275, 15917-15925.

Koivisto, U. M., Hubbard, A. L. and Mellman, I. (2001). A novel cellular phenotypefor familial hypercholesterolemia due to a defect in polarized targeting of LDLreceptor. Cell 105, 575-585.

Laughery, M. D., Todd, M. L. and Kaplan, J. H. (2003). Mutational analysis of alpha-beta subunit interactions in the delivery of Na,K-ATPase heterodimers to the plasmamembrane. J. Biol. Chem. 278, 34794-34803.

Lian, W. N., Tsai, J. W., Yu, P. M., Wu, T. W., Yang, S. C., Chau, Y. P. and Lin, C.H. (1999). Targeting of aminopeptidase N to bile canaliculi correlates with secretoryactivities of the developing canalicular domain. Hepatology 30, 748-760.

Lingrel, J. B. (1992). Na,K-ATPase: isoform structure, function, and expression. J.Bioenerg. Biomembr. 24, 263-270.

Lisanti, M. P., Sargiacomo, M., Graeve, L., Saltiel, A. R. and Rodriguez-Boulan, E.(1988). Polarized apical distribution of glycosyl-phosphatidylinositol-anchoredproteins in a renal epithelial cell line. Proc. Natl. Acad. Sci. USA 85, 9557-9561.

Lisanti, M. P., Caras, I. W., Davitz, M. A. and Rodriguez-Boulan, E. (1989). Aglycophospholipid membrane anchor acts as an apical targeting signal in polarizedepithelial cells. J. Cell Biol. 109, 2145-2156.

Matter, K., Hunziker, W. and Mellman, I. (1992). Basolateral sorting of LDL receptorin MDCK cells: the cytoplasmic domain contains two tyrosine-dependent targetingdeterminants. Cell 71, 741-753.

Mostov, K. E., de Bruyn Kops, A. and Deitcher, D. L. (1986). Deletion of thecytoplasmic domain of the polymeric immunoglobulin receptor prevents basolaterallocalization and endocytosis. Cell 47, 359-364.

Mostov, K., Su, T. and ter Beest, M. (2003). Polarized epithelial membrane traffic:conservation and plasticity. Nat. Cell. Biol. 5, 287-293.

Jour

nal o

f Cel

l Sci

ence

22

Muth, T. R., Gottardi, C. J., Roush, D. L. and Caplan, M. J. (1998). A basolateralsorting signal is encoded in the alpha-subunit of Na-K-ATPase. Am. J. Physiol. 274,C688-C696.

Nelson, W. J. and Veshnock, P. J. (1987). Ankyrin binding to (Na+ + K+)ATPase andimplications for the organization of membrane domains in polarized cells. Nature 328,533-536.

Nelson, W. J. and Yeaman, C. (2001). Protein trafficking in the exocytic pathway ofpolarized epithelial cells. Trends Cell Biol. 11, 483-486.

Ogborn, M. R., Sareen, S. and Grimm, P. C. (1993). Renal tubule Na,K-ATPase polarityin glucocorticoid-induced polycystic kidney disease. J. Histochem. Cytochem. 41, 555-558.

Rajasekaran, S. A., Palmer, L. G., Quan, K., Harper, J. F., Ball, W. J., Jr, Bander,N. H., Peralta Soler, A. and Rajasekaran, A. K. (2001). Na,K-ATPase beta-subunitis required for epithelial polarization, suppression of invasion, and cell motility. Mol.Biol. Cell 12, 279-295.

Rizzolo, L. J. (1999). Polarization of the Na+, K(+)-ATPase in epithelia derived from theneuroepithelium. Int. Rev. Cytol. 185, 195-235.

Rodriguez-Boulan, E., Kreitze, G. and Musch, A. (2005). Organization of vesiculartrafficking in epithelia. Nat. Rev. Mol. Cell. Biol. 6, 233-247.

Scheiffele, P., Peranen, J. and Simons, K. (1995). N-glycans as apical sorting signalsin epithelial cells. Nature 378, 96-98.

Shoshani, L., Contreras, R. G., Roldan, M. L., Moreno, J., Lazaro, A., Balda, M. S.,Matter, K. and Cereijido, M. (2005). The polarized expression of Na+,K+-ATPasein epithelia depends on the association between beta-subunits located in neighboringcells. Mol. Biol. Cell 16, 1071-1081.

Shvartsman, D. E., Kotler, M., Tall, R. D., Roth, M. G. and Henis, Y. I. (2003).

Differently anchored influenza hemagglutinin mutants display distinct interactiondynamics with mutual rafts. J. Cell Biol. 163, 879-888.

Simons, K. and Ikonen, E. (1997). Functional rafts in cell membranes. Nature 387, 569-572.

Simons, K. and Wandinger-Ness, A. (1990). Polarized sorting in epithelia. Cell 62, 207-210.

Skou, J. C. and Esmann, M. (1992). The Na,K-ATPase. J. Bioenerg. Biomembr. 24, 249-261.

Spring, F. A., Holmes, C. H., Simpson, K. L., Mawby, W. J., Mattes, M. J., Okubo,Y. and Parsons, S. F. (1997). The Oka blood group antigen is a marker for the M6leukocyte activation antigen, the human homolog of OX-47 antigen, basigin andneurothelin, an immunoglobulin superfamily molecule that is widely expressed inhuman cells and tissues. Eur. J. Immunol. 27, 891-897.

Su, T., Cariappa, R. and Stanley, K. (1999). N-glycans are not a universal signal forapical sorting of secretory proteins. FEBS Lett. 453, 391-394.

Wang, Y. and Androlewicz, M. J. (2000). Oligosaccharide trimming plays a role in theendoplasmic reticulum-associated degradation of tyrosinase. Biochem. Biophys. Res.Commun. 271, 22-27.

Wilson, P. D. (1997). Epithelial cell polarity and disease. Am. J. Physiol. 272, F434-F442.

Wilson, P. D., Sherwood, A. C., Palla, K., Du, J., Watson, R. and Norman, J. T. (1991).Reversed polarity of Na(+)-K(+)-ATPase: mislocation to apical plasma membranes inpolycystic kidney disease epithelia. Am. J. Physiol. 260, F420-F430.

Zamofing, D., Rossier, B. C. and Geering, K. (1988). Role of the Na,K-ATPase beta-subunit in the cellular accumulation and maturation of the enzyme as assessed byglycosylation inhibitors. J. Membr. Biol. 104, 69-79.

Journal of Cell Science 119 (1)

Jour

nal o

f Cel

l Sci

ence