Cell Surface Rescue of Kidney Anion Exchanger 1 Mutants by Disruption of Chaperone Interactions*

Mutations in the human kidney anion exchanger 1 (kAE1) membrane glycoprotein cause impaired urine acidification resulting in distal renal tubular acidosis (dRTA). Dominant and recessive dRTA kAE1 mutants exhibit distinct trafficking defects with retention in the endoplasmic reticulum (ER), Golgi, or mislocalization to the apical membrane in polarized epithelial cells. We examined the interaction of kAE1 with the quality control system responsible for the folding of membrane glycoproteins and the retention and degradation of misfolded mutants. Using small molecule inhibitors to disrupt chaperone interactions, two functional, dominant kAE1 mutants (R589H and R901stop), retained in the ER and targeted to the proteasome for degradation by ubiquitination, were rescued to the basolateral membrane of Madin-Darby canine kidney cells. In contrast, the Golgi-localized, recessive G701D and the severely misfolded, ER-retained dominant Southeast Asian ovalocytosis (SAO) mutants were not rescued. These results show that functional dRTA mutants are retained in the ER due to their interaction with molecular chaperones, particularly calnexin, and that disruption of these interactions can promote their escape from the ER and cell surface rescue.

Distal renal tubular acidosis (dRTA) 3 is a kidney disease characterized by a defect in urine acidification and metabolic acidosis. Symptoms of dRTA include nephrocalcinosis, hypokalemia, metabolic bone disease (rickets or osteomalacia), deafness, muscle weakness, mental retardation, and progressive renal failure (1)(2)(3). dRTA arises due to mutations in the genes encoding carbonic anhydrase II, the apical H ϩ -ATPase, or the basolateral kidney anion exchanger 1 (kAE1). Human anion exchanger 1 (AE1 or band 3) is a 911-amino acid glycoprotein that exists as a dimer in the plasma membrane of red blood cells where it mediates the electroneutral exchange of chloride for bicarbonate (4). The human kidney isoform, kAE1, which lacks the N-terminal 65 amino acids of the erythroid version, is produced from a tissue-specific alternative promoter in the third intron of the AE1 (SLC4A1) gene and also undergoes alternative splicing of the mRNA (5,6). kAE1 is expressed in the basolateral membrane of the acid-secreting (␣) intercalated cells of the collecting duct of the kidney where it mediates the exchange of chloride and bicarbonate. The primary cause of dRTA is the inability of kAE1 mutants to traffic efficiently to the cell surface resulting in impaired kAE1-mediated anion transport.
Various autosomal dominant and recessive dRTA mutants of kAE1 have been identified that exhibit a wide variety of clinical severity in the heterozygous, homozygous, or compound heterozygous states (7). Dominant dRTA mutants induce either complete or incomplete dRTA and are usually asymptomatic until adulthood. When co-expressed with the WT kAE1 (kWT), the complete dominant mutations (R589H, R901stop) are able to retain the wild-type protein intracellularly in the ER by a "dominant-negative effect" (8). Interestingly, these dominant dRTA mutants retain normal transport activity when expressed in Xenopus oocytes (9,10) and the ability to bind the stilbene disulfonate inhibitor, which binds specifically to properly folded AE1 (11)(12)(13), suggesting that they are functional and not grossly misfolded.
Recessive dRTA mutants also are impaired in their trafficking to the cell surface; however, they can be rescued to the cell surface in a "dominant-positive" effect by wild-type kAE1 (8, 14 -16). The homozygous state, however, results in severe dRTA as the recessive mutants are impaired in their ability to traffic to the cell surface and in their transport activity (14,17). The presence of recessive dRTA mutations in compound heterozygotes with other recessive dRTA mutants also results in severe dRTA due to trafficking defects and lack of function.
Southeast Asian ovalocytosis (SAO) is a dominantly inherited hematological condition arising from a nine-amino acid deletion, ⌬400 -408 in AE1, resulting in a misfolded and transport-inactive protein (18). The presence of this misfolded protein at the cell surface of red blood cells result in an ovalocytic and rigid shape. However, the presence of sufficient functional wild-type AE1 in erythrocytes, or kWT in kidney cells in the heterozygous state, compensates for the lack of kSAO transport activity resulting in asymptomatic anemia or dRTA (15,19). Co-expression of kSAO and dRTA mutants in the compound heterozygote state, however, results in severe dRTA due to the enhanced intracellular retention of the dRTA mutant by heterodimer formation with kSAO and the complete lack of transport activity of kSAO (16,19).
The mechanism of intracellular retention of kAE1 mutants is yet to be examined but may involve interactions with molecular chaperones. Newly synthesized glycoproteins undergo rounds of binding and release with molecular chaperones, proteins that facilitate folding, suppress aggregation, and mediate the retention and subsequent degradation of misfolded proteins (20,21). Disruption of chaperone interactions may allow for the release of ER-retained membrane glycoproteins and permit their trafficking to the cell surface. We have demonstrated previously that dominant dRTA kAE1 mutants are retained in the ER when expressed in HEK-293 and MDCK cells (14,15). We also have shown that erythroid AE1 is able to interact with calnexin in a glycosylation-dependent manner both in vitro and in HEK and K562 cells (22,23). Chaperones therefore may play a role in the intracellular retention of dRTA mutants and may be therapeutic targets to promote ER exit and trafficking to the cell surface.
In this study, we examined the role of chaperones in the trafficking and retention of kAE1 mutants in Madin-Darby canine kidney (MDCK) cells. Using specific small molecule inhibitors that affect chaperone binding, we have been able to rescue the plasma membrane expression of two dominant ER-retained kAE1 mutants, R589H and R901stop, but not the nonfunctional kSAO mutant or the Golgi-retained recessive G701D mutant. The mode of ER retention was glycosylation-dependent, as the absence of the single N-glycan found in kAE1 allowed these two mutants to traffic to the cell surface. Quality control in MDCK cells also involves the proteasome-mediated degradation of kAE1 mutants as the proteasomal inhibitor lactacystin slowed their rate of degradation in MDCK cells. Our results demonstrate that functional, dominant dRTA mutants (R598H and R901stop) are retained in the ER due to their interaction with chaperones, notably calnexin, and disruption of this interaction allows the mutants to escape the quality control system and traffic to the cell surface.
Drug Treatment-MDCK cells were seeded in six-well dishes for co-immunoprecipitations, on glass coverslips for nonpolarized cells, or on Transwell polycarbonate filters for polarized cells and treated with 1 mM CST, 30 M MAL3, and 10 M C3 or C4 overnight (ϳ18 h) at 37°C.
Co-immunoprecipitation and Immunoblotting-Infected MDCK cells were lysed in PBS containing 1% digitonin detergent and proteasome inhibitors (1 M leupeptin, aprotinin, pepstatin A, and 200 M PMSF) for 30 min on ice. Cell lysates were centrifuged (16,000 ϫ g for 10 min), and the supernatant was collected. Co-immunoprecipitation was then performed using either anti-kAE1 or anti-CNX antibodies, and immunoblotting using anti-HA antibodies identified co-immunoprecipitated AE1. For immunoblotting of whole cell lysate, cells were lysed directly in 2ϫ Sample Buffer and loaded directly onto 7.5% SDS-PAGE gels, followed by immunoblotting using an anti-HA antibody for protein expression.
Immunofluorescence and Microscopy-Immunofluorescence staining and confocal microscopy of MDCK cells stably expressing kAE1 was performed as described previously (15). Imaging equipment includes a laser scanning confocal Zeiss LSM 510 microscope, Zeiss AxioCam, AxioVision, and LSM image browser.
Flow Cytometry Analysis-MDCK cells stably expressing kAE1 were trypsinized and incubated with mouse anti-HA antibodies for flow cytometry analysis as described previously (14). A secondary anti-mouse Alexa 488 was used to detect cell surface kAE1 by flow cytometry using a FACSCalibur.
Cell Surface Biotinylation-Cell surface biotinylation of MDCK cells stably expressing kAE1 was performed as described previously (26). The presence of kAE1 in the total, unbound, and bound fractions was detected by immunoblotting with anti-HA antibody.
Proteasomal Degradation Assays-Infected MDCK cells were grown in six-well culture dishes and pretreated overnight with 10 M lactacystin. For the ubiquitination of kAE1, kAE1expressing MDCK cells were lysed in 1ϫ radioimmune precipitation assay buffer (1% deoxycholate, 1% Triton X-100, 0.1% SDS, 0.15 M NaCl, 10 mM Tris/HCl, pH 7.5, and 1 mM EDTA). Cell lysates were centrifuged (16,000 ϫ g for 10 min), and the supernatant was collected. Co-immunoprecipitation was performed as above, and immunoblotting confirmed immunoprecipitated kAE1 or ubiquitination using an anti-ubiquitin antibody. For cycloheximide chase experiments, cells were first pretreated with lactacystin for 1 h, treated with cycloheximide, and then lysed at various times as indicated in 2ϫ SDS sample buffer, resolved on 7.5% SDS-PAGE gels, and immunoblotted using an anti-HA antibody.

RESULTS
Expression of kAE1 in MDCK Cells-To examine the role chaperones play in the intracellular retention of kAE1 mutants, WT kAE1, kSAO, and three different kAE1 dRTA mutants (R589H, G701D, and R901stop) were expressed in MDCK cells. The wild-type kAE1 traffics efficiently to the cell surface of MDCK cells (10). Three dominant mutations in kAE1 (kSAO, R589H, and R901stop) resulted in retention in the ER in nonpolarized MDCK cells, whereas the recessive G701D is localized to the Golgi (10,14,15,27). These mutants also are retained intracellularly in polarized MDCK cells, whereas the R901stop mutant can also be seen missorted to the apical membrane (27,28). Previous immunofluorescence studies of the kAE1 mutants kSAO, R589H, and R901stop co-localize these mutants to the ER, suggesting that the ER chaperone calnexin may play a role in folding of kAE1, a membrane glycoprotein (14,15). Indeed, we have shown that erythroid AE1 is able to interact with calnexin in a glycosylation-dependent manner both in vitro and in HEK-293 cells (22,23).
HA-tagged wild-type and kAE1 mutants were expressed in MDCK cells using a Moloney murine leukemia viral infection system, and their intracellular localization was confirmed by immunofluorescence ( Fig. 1A). We have shown previously that the HA tag does not affect the trafficking or folding of kAE1 (12,14), whereas others have shown that it has no affect on transport function (29,30). Wild-type kAE1 was localized to the plasma membrane, whereas the four mutants all exhibited an intracellular localization as expected. Wild-type kAE1 expressed in MDCK cells migrates as multiple bands on SDS gels. The major upper bands (Ͼ80%) contains complex oligosaccharide, and the minor lower band corresponds to kAE1 with high mannose oligosaccharide (Fig. 1B, total and ␣AE1 immunoprecipitate), as characterized previously (15). During kAE1 biosynthesis, the core Glc 3 Man 9 GlcNAc 2 oligosaccharide is transferred en bloc to Asn-642 in the lumen of the ER and is then trimmed and processed in the ER and as kAE1 traffics through the Golgi adopting a complex oligosaccharide (14,27,31). The upper bands (Fig. 1, closed circle) represent kAE1 that has exited the ER, undergone oligosaccharide processing in the Golgi, and trafficked to the cell surface, whereas the lower band ( Fig. 1, open circle) represents kAE1 that is still in the ER. In contrast to the wild-type kAE1, the immunoblots of the kSAO, R589H, and R901stop mutants expressed in MDCK cells show very little upper complex form of kAE1, with the lower high mannose form dominating. This is consistent with their predominant ER localization. The G701D mutant exhibits an intermediate pattern, with a reduced amount of upper complex bands (50%) relative to wild-type kAE1, consistent with its FIGURE 1. Expression of HA-tagged kAE1 mutants in MDCK cells. A, localization of wild-type and kAE1 mutants in permeabilized nonpolarized MDCK cells by confocal microscopy using anti-HA antibodies. Scale bar, 10 m. B, co-immunoprecipitation of wild-type and kAE1 mutants with calnexin. Confluent MDCK cells were lysed in 1% digitonin for co-immunoprecipitation with rabbit antibodies against the N terminus of human kAE1 (␣kAE1) or the luminal domain of human calnexin (␣CNX). Immunoprecipitated proteins were run on 7.5% SDS-polyacrylamide gels and immunoblotted using a mouse monoclonal anti-HA antibody to detect HA-tagged AE1. Closed circles correspond to kAE1 carrying complex oligosaccharide, whereas open circles correspond to high mannose oligosaccharide. C, comparison on the percent of high mannose kAE1 interacting with calnexin by co-immunoprecipitation relative to the amount of high mannose kAE1 present in kAE1 immunoprecipitates. IP, immunoprecipitation. OCTOBER 22, 2010 • VOLUME 285 • NUMBER 43 impaired ability to exit the ER and predominant Golgi localization.

Rescuing kAE1 dRTA Mutants
kAE1 Interacts with Calnexin-Co-immunoprecipitation studies using an anti-calnexin antibody (␣CNX) demonstrates that calnexin interacts preferentially with the high mannose versions of wild-type kAE1 and kAE1 mutants (Fig. 1B, open circle), consistent with calnexin role as an ER chaperone. Indeed, a heightened interaction of the high mannose-containing mutant glycoproteins kSAO, R589H, G701D, and R901stop with calnexin was seen by co-immunoprecipitation indicative of their ER retention (Fig. 1C, percent kAE1 immunoprecipitated by the anti-calnexin antibody relative to the amount immunoprecipitated by the anti-kAE1 antibody). Calnexin may therefore play a role in the folding of kAE1 and in the retention and degradation of kAE1 mutants.
Castanospermine Blocks Calnexin Recognition of kAE1-To test whether calnexin is responsible for the retention of kAE1 mutants in the ER of MDCK cells, we used castanospermine (CST), an inhibitor of glucosidase I, to block glucose trimming and the subsequent lectin interaction of substrates with calnexin (32). Calnexin recognizes the transient Glc 1 Man 9 GlcNAc 2 intermediate present on membrane glycoproteins during biosynthesis (32). Calnexin has been shown to bind to monoglucosylated N-linked oligosaccharides present on many glycoproteins in the ER until the mature protein conformation is achieved (33)(34)(35). Immunoblotting of whole cell lysate biochemically confirmed the effect of CST as processing of the oligosaccharide is inhibited retaining the initial Glc 3 Man 9 GlcNAc 2 oligosaccharide ( Fig. 2A). In the presence of CST, wild-type kAE1 is mainly in the high mannose form, compared with wild-type kAE1 in the absence of CST, which is mainly in the complex form (Fig. 1B). A shift in oligosaccharide from high mannose to complex form in the G701D mutant also is inhibited effectively by CST with no detectable effect on the migration of the SAO, R589H, or R901stop mutants, which are already predominantly in the high mannose form (Fig. 2A). The presence of some complex oligosaccharide on wild-type kAE1 even after overnight CST treatment may represent kAE1 already at the cell surface. Previous pulse-chase studies done in HEK-293 cells have shown that wild-type kAE1 has a half-life of 15 h, indicating that it is quite stable at the cell surface and not targeted for rapid degradation (11).
Co-immunoprecipitation studies of cells treated with CST to inhibit glucosidase I showed impaired calnexin interactions with the wildtype and mutant forms of kAE1 as expected (Fig. 2B). Even though overnight treatment of MDCK cells with 1 mM CST results in an increase in the high mannose form of wild-type and mutant kAE1 (Fig.  2B, total and ␣AE1), minimal amounts of wild-type and the various mutants of kAE1 were seen in the co-immunoprecipitation fractions with calnexin ( Fig. 2B, ␣CNX) compared with the amount readily detected in the absence of drug (Fig. 1B). This indicates that blocking oligosaccharide processing of kAE1 effectively inhibits the interaction with the chaperone calnexin.
Castanospermine Rescues kAE1 dRTA Mutants to the Cell Surface of MDCK Cells-Confocal microscopy was used to examine the effect of calnexin binding inhibition by CST on the localization of kAE1 and dRTA mutants (Fig. 3). Preventing the interaction of kAE1 with calnexin may allow ER-localized mutants to escape this quality control step and traffic to the cell surface. In the first column, the cell surface expression of wildtype kAE1 and the mutants expressed in MDCK cells after overnight treatment with CST as detected using the extracellular HA epitope is shown. Wild-type kAE1 was readily detected at the cell surface of nonpolarized MDCK cells, indicating that blocking its interaction with calnexin does not prevent its trafficking to the cell surface. Thus, an interaction of kAE1 with calnexin is not required for its trafficking to the plasma membrane.
Remarkably, the two dominant dRTA mutants R589H and R901stop were detected at the cell surface after CST treatment of cells (Fig. 3, green staining, first column). kSAO and G701D, however, were not detected at the cell surface following CST treatment. Following fixation and permeabilization, the total intracellular levels of kAE1 were seen, indicating that CST had no effect on the intracellular retention of the kSAO and G701D mutants (Fig. 3, second column). Immunofluorescence of calnexin (Fig. 3, blue) also was used to stain for the ER. These results are not unique to MDCK cells, as similar results were seen when HEK-293 cells expressing the various dRTA mutants were treated with CST (data not shown), where the R589H and R901stop mutants were rescued to the cell surface, whereas the kSAO and G701D mutants were not.
CST also was able to rescue both the R589H and R901stop mutants to the basolateral membrane in polarized MDCK cells (Fig. 4). Note the diffuse intracellular staining of these two Detergent extracts were immunoprecipitated using either rabbit anti-kAE1 or anti-CNX antibodies. Immunoprecipitated kAE1 was detected by immunoblotting using a mouse anti-HA antibody. CST treatment blocks the interaction of kAE1 with calnexin resulting in a decrease in kAE1 co-immunoprecipitated with anti-calnexin antibodies. Closed circles correspond to kAE1 carrying complex oligosaccharide, whereas open circles correspond to high mannose oligosaccharide. FIGURE 3. Cell surface rescue of two dRTA mutants in nonpolarized MDCK cells. Nonpolarized MDCK cells expressing wild-type and dRTA mutants were incubated with 1 mM CST overnight. A mouse anti-HA antibody was used to detect cell surface kAE1 using an engineered HA tag in the third extracellular loop, seen as a green staining in kWT, R589H, and R901stop in intact cells. Cells were then fixed and permeabilized, and the total kAE1 pool and calnexin were detected by immunofluorescence, seen in red and blue, respectively. OCTOBER 22, 2010 • VOLUME 285 • NUMBER 43 mutants in the absence of CST, but the strong basolateral staining after treatment of the MDCK cells with CST. Again, CST was unable to rescue the trafficking of the SAO and G701D mutants to the basolateral membrane in polarized MDCK cells. Therefore, not only does CST allow the mutants to exit the ER, it also allows them to traffic to their proper site in polarized cells. These results indicate that calnexin plays an essential role in their intracellular retention of these dominant dRTA mutants, resulting in a lack of functional protein at the cell surface.

Rescuing kAE1 dRTA Mutants
To confirm that the high mannose form of kAE1 is able to traffic to the cell surface in MDCK cells treated with CST, cell surface biotinylation experiments were performed using a membrane-impermeant biotinylation reagent, NHS-SS-biotin. Biotinylated proteins were isolated using streptavidin beads and the unbound and bound fractions were analyzed by immunoblotting for kAE1 using an anti-HA antibody (Fig. 5, A and  B). In the absence of CST, the wild-type kAE1 adopts a complex oligosaccharide (Fig. 5A, ϪCST, closed circle), and this form can be biotinylated, indicative of cell surface expression. Higher molecular weight species migrating near the top of the gel also were found in the biotinylated fraction, likely corresponding to SDS-resistant kAE1 dimers. No biotinylated high mannose form of wild-type kAE1 was detected, confirming the intracellular localization of this form of kAE1. In the presence of CST, biotinylation confirmed that the high mannose form of kAE1 (Fig. 4A, ϩCST, open circle in bound fraction) is able to traffic to the cell surface, showing that cell surface expression is not dependant on oligosaccharide processing. Biotinylation experiments also were performed on the dominant dRTA mutants expressed in MDCK cells in the presence of CST; however, no R589H or R901stop kAE1 was detectable in the bound biotinylated fraction (Fig. 5B, R901stop, data not shown). The inabil-ity to detect rescued dRTA mutants by this method is likely due to the low level of rescue or the inefficiency of the biotinylation reaction.
Flow cytometry was used to determine the changes in cell surface expression of kAE1 mutants following CST treatment. Using the external HA epitope on the third extracellular loop, cell surface expression of kAE1 was monitored on intact transfected cells. In the presence of CST, an increase in cell surface expression corresponding to an upward shift in fluorescence was seen for the R589H and R901stop mutants (Fig. 5C, blue histogram). No such shift was observed for the wild-type protein or the SAO mutant. A slight upward shift for the G701D mutant in cells treated with CST was observed in this experiment; however, this effect was not reproducible. The effect of CST on the amount of cell surface expression of kAE1 dRTA mutants is, however, quite subtle, as the level of cell surface protein does not achieve the same fluorescence intensity as seen in the kWT histograms (Fig. 5C). Thus, the amount of mutant protein rescued to the cell surface was much lower than the level of wild-type protein, as was also indicated by the biotinylation experiments.
The changes in cell surface expression following CST treatment in five separate experiments were quantified by two separate methods. Overton subtraction (36) comparing histograms in the absence and presence of CST shows a 21 Ϯ 8% increase in R589H cell surface expression as well as a 19 Ϯ 4% increase in R901stop cell surface expression. There also was a slight increase in kWT cell surface expression 8 Ϯ 2% but a negligible change in kSAO and G701D cell surface expression. The percent of cells above background levels of MFI (Fig. 5C, mutants versus MDCK cells alone), also were examined with a 3 and 5% increase in cells positive for R589H and R901stop at the cell surface in the presence of CST. This modest increase in cell surface expression of dRTA mutants in the presence of CST explains the inability to detect rescued dRTA mutants in the cell surface biotinylation experiments.
Absence of the N-Linked Glycan Prevents Calnexin Interactions and Leads to Cell Surface Rescue of kAE1-The effect of CST indicates the involvement of calnexin in the ER retention of these dRTA mutants. The interaction of calnexin with AE1 is N-glycosylation-dependent (23), suggesting that the oligosaccharide found in the fourth extracellular loop of AE1 plays a role in the ER retention of the dominant R589H and R901stop dRTA mutants. To confirm the lectin role of calnexin in the ER-mediated retention of the mutants, the asparagine at position 642 was mutated to an aspartate residue to prevent oligosaccharide attachment. The absence of oligosaccharide has been shown to have no effect on the trafficking of AE1 to the cell surface in transfected HEK-293 cells (37) or when expressed in Xenopus oocytes (38). Mutation of the native N-linked glycosylation site, Asn-642, resulted in expression of unglycosylated kAE1 (N642D) as confirmed by immunoblotting of whole cell lysate from MDCK cells expressing the N642D mutants (Fig.  6A). No complex oligosaccharide could be detected on the N642D mutants, and bands with a slightly faster mobility relative to the high mannose form were seen (Fig. 6A, compare kSAO and kSAO N642D lanes, open circle bands and X). These bands were insensitive to treatment with endoglycosidase H (data not shown), indicating that the N642D kAE1 mutants were indeed nonglycosylated.
Immunofluorescence staining and confocal microscopy confirmed that the absence of oligosaccharide on the wild-type protein did not affect the localization of kAE1 to the cell surface in nonpolarized MDCK cells (Fig. 6B, first panel). Thus, N-linked oligosaccharide is not required for the trafficking of kAE1 from the ER to the plasma membrane in MDCK cells. Interestingly, both the R589H/N642D and R901stop/N642D mutants were seen at the cell surface, whereas the nonfunctional kSAO/N642D and G701D/N642D were retained intracellularly (Fig. 6B), as was seen with the CST rescue experiments. This demonstrates that the mode of ER retention of the two dominant mutants is glycosylation-dependent, as the absence of the single N-glycan allows trafficking of these two mutants to the cell surface.
To confirm that kAE1 is able to traffic to the cell surface in MDCK cells in the absence of the single N-glycan, cell surface biotinylation experiments were performed. When the oligosaccharide attachment is prevented, the wild-type kAE1 protein is present as a single band and can be biotinylated, indicative of cell surface expression (Fig.  6C, kWT N642D bound fraction). In this experiment, the R589H/ N642D and R901stop/N642D mutants were biotinylated and now detectable in the bound fraction, indicative of cell surface expression. The kSAO/N642D and G701D/ N642D were not, however, found in the bound fraction, confirming the confocal microscopy data that the lack of oligosaccharide has no effect on the intracellular retention of these two mutants. The ability to detect biotinylation of the dominant mutants lacking their N-glycan chain suggests that the level of these mutants at the cell surface is higher than in the CST rescue experiments (Fig. 5).
Co-immunoprecipitation studies of the N642D mutants also was performed to determine whether or not calnexin is able to interact with these unglycosylated kAE1 membrane proteins via the polypeptide binding site of calnexin. It has been proposed that calnexin has two mechanisms of action, a lectin-only and a dual binding, lectin-and protein-mediated chaperone function (32). We found that a small amount of kWT N642D can be co-immunoprecipitated with calnexin (Fig.  6D, kWT N642D ␣CNX), confirming the ability of calnexin to mediate protein-protein interactions. We have shown previously that the amount of AE1 N642D that was co-immunoprecipitated with the anti-calnexin antibody was, however, lower than what was found for wild-type AE1 (23). Similarly, little or no N642D forms of the kAE1 mutants were detectable in the anti-calnexin immunoprecipitates relative to the anti-kAE1 immunoprecipitates. We conclude that the interaction of the polytopic kAE1 membrane glycoprotein with calnexin in MDCK cells is highly N-glycosylation-dependent.
kAE1 Can Be Rescued with Other Small Molecule Inhibitors-A variety of small molecules and pharmacological drugs have been used to rescue the trafficking of intracellularly retained mutant proteins such as the cystic fibrosis transmembrane conductance regulator (CFTR) (39). Confocal microscopy was used to examine the effect of small molecules on the cell surface open circles correspond to high mannose oligosaccharide; and X is nonglycosylated kAE1. B, cell surface expression of N642D wild-type and kAE1 mutants in nonpolarized MDCK cells by confocal microscopy using an anti-HA antibody. Scale bar, 10 m. C, cell surface biotinylation of N642D mutants. In the absence of N-glycan, the wild-type, R589H, and R901stop can be detected at the cell surface by biotinylation. D, calnexin interacts weakly with nonglycosylated kAE1. Confluent MDCK cells were lysed in 1% digitonin for co-immunoprecipitation with antibodies against kAE1 (␣kAE1) and calnexin (␣CNX). Immunoprecipitated proteins were run on 7.5% SDS-polyacrylamide gels and immunoblotted using an anti-HA antibody to detect AE1. B, bound biotinylated kAE1; U, unbound biotinylated kAE1; T, total biotinylated kAE1. expression of wild-type and kAE1 dRTA mutants in MDCK cells. MAL3, a small molecule modulator of Hsc70 binding, and two small molecule correctors of CFTR, C3 and C4, also were used to examine their effect on the trafficking of dRTA mutants. MAL3 is a structural analog of NSC 630668-R/1 and 15-deoxyspergualine that inhibits the Hsp40-stimulating ATPase activity of Hsc70, resulting in decreased peptide affinity (40). Although the mechanism of action of the small molecule correctors, C3 and C4, is unknown, they also have been shown to promote trafficking of CFTR to the cell surface (41,42). By confocal microscopy, all three compounds were able to rescue the R589H and R901stop mutants to the cell surface (Fig.  7, columns 2-4, green staining), similar to the effect observed by disruption of the calnexin interaction. The severely misfolded, nonfunctional kSAO and the Golgi-localized G701D mutants, however, were not rescued to the cell surface following treatment with the various drugs, as can be seen by the complete absence of these mutants at the cell surface (Fig. 7, green stain-ing). To confirm that these MDCK cells are indeed expressing the mutant dRTA kAE1 proteins despite their lack of cell surface expression, these cells were permeabilized and stained for intracellular kAE1 (Fig. 7, red staining). Calnexin also was visualized for co-localization of intracellular kAE1 in the ER but was omitted for clarity. The effect of these drugs on the trafficking of these dRTA mutants also was examined in polarized MDCK cells by confocal microscopy. The cell surface rescue of R589H and R901stop by MAL3, C3, and C4 to the basolateral membrane in polarized MDCK cells was confirmed (data not shown).
kAE1 Is Ubiquitinated and Degraded by the Proteasome-To determine the fate of the kAE1 mutants in MDCK cells, we used the highly specific and irreversible proteasome inhibitor, lactacystin (43), to examine whether or not the proteasome plays a role in the degradation of wild-type or mutant forms of kAE1. Proteins that are targeted for degradation by the proteasome typically are modified covalently by the attachment of one or more ubiquitin molecules. We set out to see whether kAE1 is ubiquitinated by immunoprecipitating kAE1 from MDCK cells treated with lactacystin to inhibit the degradation of ubiquitinated proteins and then by blotting the immunoprecipitated kAE1 for covalently attached ubiquitin. Wild-type and dRTA mutants were immunoprecipitated from MDCK cells and then resolved on SDS-PAGE gels. Western blotting was then used to identify either kAE1 (Fig. 8A) or ubiquitinated proteins (Fig. 8A) in the immunoprecipitated fractions. Both the wild-type and mutant kAE1 proteins were found to be ubiquitinated in cells treated with lactacystin, which would indicate that they could be targeted for degradation. Deglycosylation of ubiquitinated kAE1 using endoglycosidase H (data not shown), showed that the form of kAE1 that was ubiquitinated does not contain high mannose oligosaccharide, suggesting that the oligosaccharide had been removed from kAE1. A cytosolic N-glycanase may be responsible for the removal of kAE1 oligosaccharides as ER-associated degradation substrate intermediates have been found to be deglycosylated when they reach the cytosol prior to their degradation by the proteasome (44).
If the mutations cause kAE1 to misfold, the mutants may be more sensitive to degradation and turnover. To compare the FIGURE 7. Cell surface rescue of dRTA mutants using small molecules in nonpolarized MDCK cells. Nonpolarized MDCK cells expressing wild-type and dRTA mutants were incubated with MAL3, C3, or C4 overnight. Cell surface expression of kAE1 using an external HA epitope was detected in nonpermeabilized cells by immunofluorescence (green). Cells were then fixed and permeabilized and stained for total kAE1 (red) and the ER chaperone calnexin (omitted for clarity). Note the presence of cell surface kAE1 (green) in MDCK cells expressing R589H and R901stop treated with MAL3, C3, and C4.
rates of degradation of the dRTA mutants relative to the wildtype protein, cycloheximide was used to block protein expression, and then immunoblotting was used to assess the levels of protein remaining over time. The wild-type protein was found to be quite stable, with protein being detected even 17 h later (Fig. 8B,  kWT, Ϫlactacystin). This is in agreement to our previously published results indicating that kAE1 has a half-life of 15 h as determined by pulse-chase studies in HEK-293 cells (11). The levels of the mutants, kSAO and R589H, also were assessed following 17 h of cycloheximide treatment and were found to be completely absent at that time indicating their more rapid degradation (Fig. 8, kSAO and R589H, Ϫlactacystin). Lactacystin also was used to reduce the proteasome-mediated degradation of kAE1. It was found to have a protective effect on the levels of wild-type and mutant kAE1 indicating that the proteasome is responsible for the degradation of kAE1 and the ER-localized mutants (SAO, R589H, and R901stop) but not the Golgi-localized mutant (G701D) in MDCK cells.

DISCUSSION
We have shown previously (8, 11, 12, 14 -16) that in polarized and nonpolarized MDCK cells, kAE1 mutants (SAO and dRTA) are retained intracellularly and are unable to traffic efficiently to the plasma membrane to carry out their transport function. To assess the role molecular chaperones play in the intracellular retention of dRTA mutants, we used a variety of drugs and nonglycosylated mutants that inhibit chaperone binding to substrate glycoprotein intermediates. Of the mutants studied, the dominant R589H and R901stop showed plasma membrane expression upon treatment with various drugs whereas kSAO and G701D did not (Figs. 3, 4, and 7). This indicates that ER retention is chaperone-mediated, as disruption of chaperone binding allowed these two dominant mutants to escape to the plasma membrane. Disruption of the interaction of the dominant kAE1 mutants with calnexin, either by blocking N-glycan processing with CST or by removal of the N-glycosylation site, N642D, promoted rescue of the traf- A, wild-type kAE1 and kAE1 mutants are ubiquitinated after lactacystin-mediated inhibition of the proteasome. MDCK cells expressing wild-type and kAE1 mutants were solubilized in radioimmune precipitation assay buffer, and kAE1 was immunoprecipitated using an antibody against the N terminus of kAE1. Immunoprecipitated proteins were run on a 7.5% SDS-PAGE gel and immunoblotted for kAE1 using an anti-HA antibody (left) or ubiquitin (right). Closed circles correspond to kAE1 carrying complex oligosaccharide, open circles correspond to high mannose oligosaccharide, and X is nonglycosylated kAE1. B, cycloheximide chase experiment to monitor protein degradation. Control cells, or cells that were pretreated for 1 h with 10 M lactacystin, were then incubated with 10 M cycloheximide to inhibit protein synthesis. Cells were lysed at various time points in 2ϫ sample buffer and run on SDS-PAGE gels. Immunoblotting for AE1 was then used to monitor protein degradation over time. C, quantification of degradation rates in the absence (red) and presence (blue) of lactacystin (y axis in logarithmic scale). Following cycloheximide treatment, the only protein seen after 17 h was the wild-type kAE1 protein, and this time point has been omitted in the quantification of degradation for the various mutants.
ficking of these mutants to the cell surface. By preventing the interactions of the mutant protein with the quality control chaperones, the newly synthesized, slightly misfolded, but functional kAE1 protein can escape recognition by the mechanisms responsible for its retention and ultimate degradation.
The use of small molecules as correctors of ER retention of dominant dRTA mutants is potentially feasible, as the R589H and R901stop mutants retain anion-transport activity in patient red blood cells and when expressed in Xenopus oocytes (9, 10). These dRTA mutants can be classified as class 2 mutants, mutations that arise due to defective protein processing yet are still functionally active (45). dRTA caused by these mutations arise due to the retention of functional kAE1 protein in the ER. The Hsc70 inhibitor and small molecule correctors (MAL3, C3, and C4) also were able to rescue cell surface expression of the two dominant mutants confirming the role of the quality control system of chaperones in the retention of these dRTA mutants. MAL3, a disruptor of Hsc70 action, suggests that this cytosolic chaperone may play a role in the folding of kAE1, likely by interaction with its large N-terminal cytosolic domain or cytosolic loops located in the membrane domain. We also have been able to demonstrate an interaction between Hsc70 and kAE1 by co-immunoprecipitation (data not shown). The Vertex compounds, C3 and C4, also were shown to rescue the cell surface expression of R589H and R901stop, and although their exact mechanism of action is unknown, these compounds have been shown to interact directly with CFTR in the ER increasing its folding efficiency (41,46). Our results indicate that these compounds also can rescue dRTA mutants of kAE1 and therefore are not specific for CFTR indicating a more general rescue mechanism. Immunoblots of MDCK cells expressing dRTA mutants showed no change in glycosylation pattern in the presence of MAL3 or the small molecule correctors (data not shown). As in previous studies with erythroid AE1 (47), and as we have now shown with the CST results, formation of complex oligosaccharide is not required for the trafficking of kAE1 to the cell surface. It is therefore the interaction with the quality control mechanisms in the cell that determine the intracellular localization of this membrane glycoprotein. It is not surprising that disruption of either calnexin or Hsc70 interaction with kAE1 can rescue it to the cell surface, as kAE1 is a polytopic membrane protein with regions exposed on both the ER lumen and cytosol. As previously suggested (48), protein folding pathways are both intertwined and redundant, and these results suggest that multiple chaperones are involved in the folding and intracellular retention of kAE1.
We have shown by flow cytometry that only a small fraction of dominant dRTA mutants can be rescued to the cell surface. In the case of CFTR, the minimal requirements for gene therapy have demonstrated that as few as 6 -10% of corrected cells can generated sufficient chloride transport in vivo, suggesting that low levels of CFTR rescue are sufficient to correct the CF epithelial chloride transport defect (49,50). It remains controversial however, that disruption of the calnexin interaction with ER retained forms of mutant CFTR leads to the maturation of CFTR (51,52). Nevertheless, small molecule drugs may be used to therapeutically target chaperone interactions in diseases associated with defective protein folding and trafficking. In the case of dRTA, symptoms arise due to a reduced amount of functional kAE1 at the basolateral membrane of the acid-secreting (␣) intercalated cells of the collecting duct of the kidney, not only due to the lack of mutant kAE1 at the cell surface, but also due to the dominant R589H protein ability to retain the wild-type protein in the ER (11). The use of these drugs would therefore be able to promote the cell surface expression of these dRTA mutants and the retained wild-type protein that arise due to simple trafficking defects, increasing the overall level of functional protein at the cell surface.
The severely misfolded kSAO is retained however in the ER and is not rescued by CST or the other compounds. Thus, disruption of calnexin interactions is not sufficient to rescue this severely misfolded mutant in MDCK cells. We have now demonstrated that kAE1 is targeted for degradation by the proteasome via ubiquitination, consistent with the theory that excessive interactions of unfolded proteins with molecular chaperones target them for degradation when the native conformation is not achieved. Misfolded kAE1 may therefore be a substrate for ER-association protein degradation. This is not surprising, as the ER chaperone calnexin and the cytosolic chaperone Hsc70, which we have shown to play a role in the retention of dRTA mutants, have been implicated in the recognition of misfolded proteins for the ER-associated degradation pathway (53). We have now confirmed the protective effect of the proteasome inhibitor lactacystin on the degradation of kAE1 in MDCK cells; lactacystin has been shown previously to protect the erythroid SAO protein against degradation in HEK-293 cells (15). The lack of cell surface rescue of kSAO through disruption of chaperone interactions and the fact that lactacystin provides protection against proteasomal degradation agrees with previous studies (15,18,38) that have shown that kSAO is severely misfolded and nonfunctional and reinforces the idea that there are multiple mechanisms of quality control active within the cell to retain misfolded proteins. Quality control in MDCK cells therefore is highly effective in recognizing severely misfolded membrane proteins and targeting them for retention.
Although disruption of chaperone interactions may allow rescue of dominant dRTA mutants to the cell surface but not the severely misfolded kSAO, chaperones likely do not play a major role in the Golgi retention of recessive mutants such as G701D. Even though the G701D dRTA mutant was shown to have a heightened interaction with calnexin, castanospermine treatment did not rescue its cell surface expression. This is not surprising, because G701D can exit the ER and traffic to the Golgi, yet is severely impaired in moving from the Golgi to the cell surface. When co-expressed with glycophorin A in Xenopus oocytes or in erythrocyte precursors that endogenously express glycophorin A, the G701D mutant is able to traffic to the cell surface showing that glycophorin A facilitates the cell surface expression of G701D kAE1 (17). Kidney cells, however, do not contain glycophorin A, which explains the lack of cell surface expression of this mutant and the development of dRTA. Thus, the cellular context and nature of interacting proteins is very important in producing a phenotype of mutations with the same genotype. Indeed, human red cell precursors lose specific chaperones like calnexin during differentiation (22). This loss of protein quality control may allow misfolded forms of AE1, such as the SAO mutant, to traffic to the cell surface during erythropoiesis and be expressed in the mature red cell. Plasma membrane rescue therefore is not a general phenomenon for a protein but is specific for a particular set of mutants in a particular cellular context. This study adds significant new insights into the role chaperones play in the molecular basis and manifestation of certain forms of dRTA. Small molecules that inhibit chaperone-mediated retention and promote cell surface expression of certain dominant dRTA mutants (R5899H, R901stop) in kidney cells are therefore potential therapeutic agents, which may be used to treat this group of dRTA patients.