Glycosylation Increases Potassium Channel Stability and Surface Expression in Mammalian Cells*

N -linked glycosylation is not required for the cell surface expression of functional Shaker potassium channels in Xenopus oocytes (Santacruz-Toloza, L., Huang, Y., John, S. A., and Papazian, D. M. (1994) Biochemistry 33, 5607–5613). We have now investigated whether glycosylation increases the stability, cell surface expression, and proper folding of Shaker protein expressed in mammalian cells. The turnover rates of wild-type protein and an unglycosylated mutant (N259Q,N263Q) were compared in pulse-chase experiments. The wild-type protein was stable, showing little degradation after 48 h. In contrast, the unglycosylated mutant was rapidly degraded ( t 1 ⁄ 2 (cid:1) (cid:1) 18 h). Lactacystin slowed the degradation of the mutant protein, implicating cytoplasmic proteasomes in its turnover. Rapid lactacystin-sensitive degradation could be conferred on wild-type Shaker by a glycosylation inhibitor. Expression of the unglycosylated mutant on the cell surface, assessed using immunofluorescence microscopy and biotinylation, was dramatically reduced compared with wild type. Folding and assembly were analyzed by oxidizing intersubunit disulfide bonds, which provides a fortuitous hallmark of the native structure. Surprisingly, formation of disulfide-bonded adducts was quantitatively similar in the wild-type and unglycosylated mutant proteins. Our results indicate that glycosylation

The addition and processing of asparagine-linked oligosaccharides in the ER 1 play a pivotal role in the biogenesis and quality control of many membrane and secretory proteins (for review, see Ref. 1). N-linked glycosylation promotes proper folding and assembly and increases the stability of a variety of glycoproteins, thereby enhancing their trafficking through the cellular endomembrane system. Glycosylation begins with the cotranslational addition of a preformed 14-residue core glycan to lumenally exposed asparagine residues that are part of the consensus sequence Asn-Xaa-(Ser/Thr). Processing of the core glycan commences immediately with trimming of two of the three terminal glucose residues by ER glucosidases I and II. Generation of the monoglucosylated core glycan results in association of the newly made glycoprotein with calnexin and/or calreticulin, lectins that are part of an ER chaperone system (2,3). Other components of this system include ER glucosidase II and UDP-glucose:glycoprotein glucosyltransferase, which catalyze a cycle of deglucosylation and reglucosylation while the protein acquires its native structure (2). Properly folded molecules are released from the cycle, whereas misfolded molecules are retained in the ER and may be targeted for degradation by cytoplasmic proteasomes. Further processing of the core glycan, specifically trimming of a particular mannose residue by ER mannosidase I, has been implicated in the proteasomal targeting of several glycoproteins (4 -12).
We are studying the biogenesis and quality control of voltagedependent ion channels, using the Shaker K ϩ channel as the prototype(forreview,seeRef.13andreferencestherein).Voltagedependent ion channels are multisubunit membrane proteins that control the excitability of nerve and muscle (14). Many channel proteins are glycosylated, a modification that may influence channel function and intracellular trafficking. For instance, sialylation of voltage-dependent Na ϩ channels shifts the voltage dependence of activation through a surface charge effect (15), whereas glycosylation of the HERG (human etherà -go-go-related gene) K ϩ channel is required for cell surface expression (16,17). The significance of ion channel glycosylation is illustrated by the finding that mutations that prevent glycosylation of the HERG protein cause long QT syndrome, type 2 (18). In long QT syndrome, genetic or acquired malfunctions of ion channels delay repolarization of the cardiac action potential, predisposing affected individuals to ventricular arrhythmia and sudden death (19,20).
The Shaker K ϩ channel is particularly well suited for studies of ion channel biogenesis and quality control. Shaker channels express well in a variety of systems, including Xenopus oocytes and mammalian cell lines (21)(22)(23)(24). The wild-type protein folds and assembles into tetramers in the ER and is then efficiently transported to the Golgi apparatus where the carbohydrate chains mature prior to expression at the cell surface (22)(23)(24). In contrast, mutant proteins that fail to acquire the native structure are recognized and retained by the quality control system of the ER (24).
We have previously shown that the Shaker K ϩ channel protein is glycosylated on two asparagine residues, Asn 259 and Asn 263 , in the first extracellular loop (21). Mutating these as-paragine residues to glutamine abolishes glycosylation but does not prevent the cell surface expression of functional channels in Xenopus oocytes. Furthermore, the voltage dependence of unglycosylated channels is similar to that of wild-type Shaker channels (21). These results indicate that glycosylation is not required for the assembly of functional channels or their transport to the cell surface in Xenopus oocytes (21).
Residues Asn 259 and Asn 263 are also glycosylated in mammalian cells (22), where we have demonstrated that the Shaker protein interacts with calnexin shortly after synthesis (25). This interaction is transient and requires carbohydrate modification of the Shaker protein. Interestingly, ER-retained mutant proteins with defects in subunit folding and assembly also interact transiently with calnexin, with a time course similar to that of wild-type Shaker (25). These results indicate that the mutant proteins are not retained in the ER by stable association with calnexin.
Our previous studies did not directly address whether glycosylation increases the stability, cell surface expression, or folding efficiency of the Shaker protein. We have now investigated these issues in mammalian cells in culture. We find that glycosylation dramatically increases the stability and cell surface expression of the Shaker protein but is not required for acquisition of the native structure.
Cell Culture and Metabolic Labeling-Human embryonic kidney cells (HEK293T) were grown as described (24) and transfected with Cytofectene reagent (Bio-Rad). For transfection experiments, the Shaker B cDNA (27) and a mutant construct, N259Q,N263Q, which eliminates N-linked glycosylation of the Shaker protein (21), were transferred into the vector pcDNA1/AMP (Invitrogen, Carlsbad, CA). A S346C mutation was engineered into wild-type and unglycosylated mutant Shaker cDNAs using polymerase chain reaction methods as described previously (24). A total of 1 g of DNA was used per 35-mm well of cells. Forty-eight hours after transfection, HEK293T cells were incubated for 30 min in methionine-and cysteine-free Dulbecco's modified Eagle's medium (Mediatech, Herndon, VA), pulsed for 1 h with 200 Ci/ml [ 35 S]methionine and [ 35 S]cysteine (Tran 35 S-Label, ICN, Irvine, CA), and chased for various times in complete, nonradioactive medium. The cells were washed once with ice-cold phosphate-buffered saline (PBS) and lysed in ice-cold lysis buffer (1% Triton X-100, 150 mM NaCl, 50 mM HEPES-NaOH, pH 7.4) supplemented with protease inhibitors as described previously (21). The cell lysates were rotated at 4°C for 10 min and then centrifuged at 100,000 ϫ g for 30 min at 4°C to remove insoluble material.
Immunoprecipitation, Electrophoresis, and Fluorography-Immunoprecipitation of Shaker protein was performed as described previously (21) using an antibody directed against a Shaker-␤-galactosidase fusion protein (26). Protein samples were resuspended in Laemmli sample buffer containing either 10% 2-mercaptoethanol (reducing conditions) or 16 mM iodoacetamide (nonreducing conditions) and boiled for 5 min prior to electrophoresis on denaturing 5 or 7.5% polyacrylamide gels with 3 or 4% stacking gels (24). To normalize the amount of protein in electrophoresis samples, 5% of the lysate supernatant was subjected to trichloroacetic acid precipitation and counted in a scintillation counter. Lanes were loaded with samples containing equal trichloroacetic acidprecipitable counts per minute. The gels were stained with Coomassie Blue, soaked in fluorographic enhancer, dried, and exposed to film.
Oxidation of Intersubunit Disulfide Bonds-Specific disulfide bonds between Shaker subunits were oxidized in intact cells by exposing them to 20, 50, or 100 mM H 2 O 2 for 15 min as described previously (28). The reaction was quenched by adding 5 mM N-ethylmaleimide for 5 min (28). Shaker protein was then solubilized, immunoprecipitated, and boiled in sample buffer containing either 10% 2-mercaptoethanol (reducing conditions) or 16 mM iodoacetamide (nonreducing conditions). Electrophysiology-For electrophysiological analysis, Xenopus oocytes were injected with 4 -10 ng of cRNA and incubated for 24 h at 17°C. Whole cell currents were recorded at room temperature (22-23°C) in modified Barth's saline using a two-electrode voltage clamp (Warner Instruments, Hamden, CT) (29). The data were acquired using pCLAMP version 5.5.1 (Axon Instruments, Foster City, CA) and analyzed using Origin version 5.0 (Microcal, Northampton, MA). Statistical analysis was performed using InStat version 2.0.1 (Graphpad Software, San Diego, CA).
Immunofluorescence Microscopy-Indirect immunofluorescence was performed as described previously (30). Briefly, HEK293T cells growing on glass coverslips were transiently transfected with various cDNAs and, 48 h after transfection, washed twice with PBS and then fixed in PBS containing 4% paraformaldehdye for 20 min. The fixative was removed by washing (three times for 5 min each time) in PBS, and then the cells were permeabilized with 0.25% Triton X-100 (Sigma) for 5 min, followed by washing in PBS (twice for 5 min each time). Nonspecific antibody binding was blocked by incubating the coverslips for 1 h at 37°C in PBS containing 10% bovine serum albumin (Sigma). After washing in PBS (three times for 10 min each time), the cells were incubated (for 2 h at room temperature) with a polyclonal rabbit antiserum directed against a Shaker ␤-galactosidase fusion protein (1:200 dilution in 3% bovine serum albumin in PBS) and a mouse monoclonal antibody against calnexin, an ER-resident protein (1:250 dilution). After washing the antibody-labeled HEK293T cells in PBS (three times for 5 min each time), the cells were incubated with fluorescent-conjugated secondary antibodies (Alexa-488-conjugated goat anti-rabbit (1: 1000) and Alexa-568-conjugated goat anti-mouse (1:1500)) for 1 h at room temperature. Finally, the cells were washed in PBS (three times for 5 min each time) and then rinsed briefly in distilled water and mounted in SlowFade (Molecular Probes). Images were obtained with a 63ϫ quartz objective on an inverted laser scanning confocal microscope (Leica Dm IRB/E, Meyer Instruments Inc., Houston, TX).
Cysteine-specific Cell Surface Biotinylation Assay-Cell surface expression of wild-type and mutant Shaker proteins was analyzed by biotinylation of an engineered extracellular cysteine using an impermeant reagent (31). Briefly, at 24 -36 h post-transfection, HEK293T cells were washed twice with PBS (warmed to 37°C) and then incubated at 37°C and 5% CO 2 for 30 min with HPDP-biotin (0.03 mg/ml; Pierce) in PBS solution. The cells were then washed two to three times with warm PBS to remove the biotinylating reagent. The cells were collected in 400 l of ice-cold lysis buffer containing protease inhibitors, rotated for 15 min at 4°C, and then centrifuged at 100,000 ϫ g for 30 min at 4°C to remove insoluble material. Aliquots corresponding to 5% of the total volume of the solubilized material were reserved. The rest of the solubilized material was incubated with ϳ25 l of streptavidinlinked agarose beads (Pierce) at 4°C for 2 h. The beads were collected by brief centrifugation and washed three to five times with 1% Triton X-100-containing buffer (21). Bead-precipitated protein was eluted by boiling in 50 l of gel loading buffer followed by brief centrifugation. The streptavidin-precipitated protein, together with the aliquots of the total solubilized material, were subjected to electrophoresis on 7.5% acrylamide gels. The proteins were transferred to a nitrocellulose membrane, immunoblotted using a rabbit polyclonal anti-Shaker antibody, and visualized using enhanced chemiluminescence. Protein bands were quantified by densitometry (model GS-700 Imaging Densitometer; Bio-Rad) using Molecular Analyst software (version 1.4; Bio-Rad).

Unglycosylated Shaker Protein Is Unstable and Degraded by
Cytoplasmic Proteasomes-To determine whether N-linked glycosylation stabilizes K ϩ channel protein in mammalian cells, the wild-type Shaker protein and an unglycosylated mutant, N259Q,N263Q, were expressed in HEK293T cells for pulse-chase analysis. The cells were metabolically labeled during a 1-h pulse and were either harvested immediately or incubated in nonradioactive medium for chase periods up to 48 h. After detergent extraction, Shaker protein was immunoprecipitated and subjected to electrophoresis and fluorography.
As previously reported, the wild-type Shaker protein was detected after the pulse as a core-glycosylated precursor, which matured to a complex glycosylated form after processing in the Golgi apparatus ( Fig. 1A) (21)(22)(23)(24). The unglycosylated N259Q,N263Q protein migrated slightly faster than the coreglycosylated immature form of the wild-type protein (21,23) and underwent a small change in mobility during the chase because of an unidentified post-translational modification (Fig.  1B). The wild-type and unglycosylated mutant proteins were differentially stable during the chase period. The wild-type protein was stable up to 48 h, whereas the unglycosylated mutant protein was degraded at a significantly faster rate ( Fig.  1). At 24 h, ϳ98% of the wild-type protein but only 24% of the unglycosylated mutant protein remained (Fig. 1C).
To determine whether cytoplasmic proteasomes play a role in the degradation of the unglycosylated Shaker mutant, transfected HEK293T cells were incubated in lactacystin, a proteasome-specific inhibitor (32). Treatment with lactacystin significantly slowed degradation of unglycosylated Shaker protein; at 24 h, 65% of the unglycosylated mutant protein remained (Fig. 1). In contrast, the wild-type protein was unaffected by lactacystin. These results indicate that in the absence of glycosylation, the Shaker protein is targeted for degradation by proteasomes. However, alternative degradation pathways may also be involved because lactacystin did not completely prevent degradation of the unglycosylated mutant protein.
To investigate whether lysosomes are involved in Shaker protein turnover, the pulse-chase experiment was repeated in the presence of the lysosomotropic agent NH 4 Cl. At 36 h of chase, NH 4 Cl had no effect on the turnover of either wild-type or unglycosylated mutant proteins (Fig. 2). Similarly, the serine/cysteine protease inhibitors leupeptin and pepstatin A, which are known to inhibit lysosomal proteases (33), had no effect on the degradation rate of the wild-type or mutant proteins (Fig. 2). These results indicate that lysosomes do not contribute significantly to the degradation of Shaker proteins during a 36-h chase period.
Proteins in the secretory pathway that are targeted to the proteasome for degradation are thought to undergo retrograde translocation directly from the ER to the cytoplasm (34,35). In a few cases, however, the substrate protein must be transferred to a post-ER compartment prior to degradation (36,37). Therefore, we characterized the turnover of the unglycosylated mutant protein after trapping it in the ER with BFA in the presence of NOC. Under these conditions, the protein was still degraded rapidly by the lactacystin-sensitive proteasomal pathway (Fig. 3). These results indicate that proteasomal targeting of the unglycosylated mutant protein occurs directly from the ER.
Decreased Stability of Unglycosylated Mutant Protein Can Be Mimicked by Inhibiting Glycosylation of the Wild-type Protein-The data presented so far do not rule out the possibility that the N259Q,N263Q protein is rapidly degraded because of the asparagine-to-glutamine mutations per se rather than the lack of glycosylation. To investigate this issue, we blocked N-linked glycosylation of the wild-type protein with a tripeptide inhibitor, Asn-Tyr-Thr-NH 2 (38). Wild-type Shaker protein that has not been glycosylated was degraded by the lactacystinsensitive proteasome pathway with kinetics similar to that of the N259Q,N263Q double mutant (Fig. 4). These results confirm that lack of glycosylation targets the Shaker protein for degradation by cytoplasmic proteasomes.
Glycosylation Promotes Cell Surface Expression of Shaker Protein-Glycosylation is not absolutely required for cell surface expression of functional Shaker channels in Xenopus oocytes (21). However, our previous study did not address whether glycosylation increases the rate and/or amount of channel protein inserted into the plasma membrane. To test the hypothesis that glycosylation promotes cell surface expression of Shaker channels, we performed confocal microscopy and surface biotinylation experiments. HEK293T cells expressing wild-type or the unglycosylated mutant proteins were permeabilized with 0.1% Triton X-100 and then double-labeled with antibodies directed against Shaker and calnexin, an ER-resident protein. Wild-type Shaker protein showed robust cell surface expression with little or no overlap with calnexin (Fig. 5,  top row). Similarly, there was no overlap between labeling for the wild-type protein and other ER markers such as BiP, GRP94 (glucose-regulated protein 94), and protein disulfide isomerase or the Golgi marker, ␤-COP1 (data not shown). These results indicate that the wild-type protein exits the ER and trafficks to the cell surface efficiently (Fig. 5, middle row). Cell surface expression was also observed for the unglycosylated mutant protein. In contrast to the wild-type protein, however, a significant fraction of the unglycosylated protein was retained intracellularly, overlapping in distribution with the ER marker calnexin (Fig. 5, arrow). In addition, some of the unglycosylated Shaker protein was found in punctate clusters that resembled aggresomes (39).
To compare quantitatively the levels of glycosylated and unglycosylated Shaker protein at the cell surface, we performed cell surface biotinylation at an engineered cysteine at position 346 in the extracellular S3-S4 loop. First, we confirmed that the S346C mutation does not affect the stability, functional expression, or localization of the Shaker protein (Fig. 6). Pulse-chase experiments revealed that the S346C and N259Q,N263Q,S346C proteins turn over with kinetics similar to those of the wild-type and N259Q,N263Q proteins, respectively (Fig. 6A). Confocal analysis indicated that the S346C mutation does not alter the distribution of the wild-type and unglycosylated Shaker proteins (Fig. 6B; compare with Fig. 5). Furthermore, upon expression in Xenopus oocytes, S346C and N259Q,N263Q,S346C proteins generated functional channels (Fig. 6C).
HEK293T cells expressing the S346C or N259Q, N263Q,S346C proteins were incubated with the cysteine- specific, impermeant reagent HPDP-biotin (31). After detergent solubilization, biotinylated protein was precipitated with streptavidin-agarose beads and subjected to electrophoresis and immunoblot analysis with the Shaker-specific antibody. An unfractionated aliquot (5%) of the solubilized material was analyzed in parallel to determine the fraction of Shaker protein biotinylated by the procedure (Fig. 7). The results indicate that expression at the cell surface was significantly higher (ϳ7-fold, p Ͻ 0.001, paired Student's t test) for the glycosylated S346C protein than for the unglycosylated N259Q,N263Q,S346C protein (Fig. 7). Biotinylation occurred specifically at S346C, because no Shaker immunoreactive material was precipitated Forty-eight hours after transfection, the cells were permeabilized and then labeled with a rabbit polyclonal antiserum directed against a Shaker-␤-galactosidase fusion protein and a mouse monoclonal antibody directed against calnexin (an ER-resident protein) and visualized by incubation with fluorescent-conjugated Alexa-488 goat anti-rabbit (Shaker; green) and Alexa-568 goat anti-mouse (calnexin; red) secondary antibodies, respectively. In each panel, the same confocal plane was used for acquisition. Yellow regions in the merged images represent colocalization of Shaker protein and calnexin. The calibration bar applies to all panels. Little overlap of the wild-type protein and calnexin is evident in the merged images, consistent with efficient trafficking of wild type to the cell surface. In contrast, the unglycosylated mutant accumulates intracellularly with significant overlap with calnexin (arrow). C, wild-type (WT), S346C, N259Q,N263Q, and N259Q,N263Q,S346C Shaker cRNAs were expressed individually in Xenopus oocytes for voltage clamp analysis. Currents, recorded 24 h after injection, were evoked by stepping for 100 ms from a holding potential of Ϫ100 mV to potentials ranging between Ϫ100 and ϩ80 mV in 20-mV increments. Representative traces are shown and have been scaled for comparison. Scale bars represent 5 A and 10 ms. Inactivation time constants, determined at ϩ60 mV by fitting data with a single exponential function, were: wild type, 3.5 Ϯ 0.5 ms (n ϭ 5); S346C, 3.8 Ϯ 0.4 ms (n ϭ 6); N259Q,N263Q, 5.4 Ϯ 1.3 ms (n ϭ 6); and N259Q,N263Q,S346C, 6.0 Ϯ 0.8 ms (n ϭ 8).
with streptavidin beads after biotinylation of intact cells expressing the wild-type or N259Q,N263Q mutant proteins (data not shown). The wild-type Shaker protein was biotinylated on endogenous intracellular cysteines if cells were permeabilized with 0.1% Triton X-100 prior to application of HPDP-biotin, however (data not shown). We conclude that glycosylation significantly increases surface expression of the Shaker protein in HEK293T cells, consistent with the results of confocal microscopy.
Lack of Glycosylation Does Not Dramatically Impair Proper Folding of Shaker Protein-In the Shaker channel, intracellular cysteine residues in the amino (Cys 96 ) and carboxyl (Cys 505 ) termini of adjacent subunits form disulfide bonds in intact cells exposed to mild oxidizing conditions (24,28). The disulfide bonds are generated between adjacent Shaker monomers, resulting in the sequential formation of dimers, trimers, linear tetramers, and circular tetramers (28). Previous results indicate that disulfide-bonded adducts are generated in Shaker proteins that are able to fold and assemble properly but not in mutant proteins that are recognized as abnormal and retained by the ER quality control system (24). Therefore, the capacity to form intersubunit disulfide bonds provides a useful hallmark of the native structure of the Shaker channel and can be used as an indirect measure of folding and subunit assembly (13,24,28). Intact HEK293T cells expressing the wild-type or unglycosylated mutant proteins were metabolically labeled and exposed to the oxidizing agent hydrogen peroxide (20 -100 mM) to favor disulfide bond formation. The reaction was quenched with 5 mM N-ethylmaleimide to protect remaining free sulfhydryl groups prior to solubilization and immunoprecipitation. Under nonreducing conditions, most of the monomer band was converted to higher molecular mass adducts corresponding to a dimer, trimer, linear tetramer, and a circular tetramer (Ref. 28 and data not shown). As expected, adducts formed by the unglycosylated Shaker mutant were smaller than those of the glycosylated wild-type protein (Ref. 24 and data not shown). Fig. 8 summarizes the percentage of the total Shaker protein converted to higher molecular mass adducts in cells expressing the wild-type (open bars) or unglycosylated (filled bars) proteins exposed to 20, 50, or 100 mM H 2 O 2 for 10 min. Overall, more than 60% of the Shaker protein was converted to higher molecular mass adducts under these conditions, with larger adducts predominating at higher concentrations of H 2 O 2 (Ref. 28 and data not shown). Importantly, similar levels of disulfide-bonded adducts were observed for the wild-type and ung-lycosylated mutant proteins ( Fig. 8; p Ͼ 0.05, Student's t test). By this measure of channel folding and assembly, glycosylation of the Shaker protein does not significantly promote acquisition of the native state. DISCUSSION We have investigated the role of N-linked glycosylation in the stability, folding, cell surface expression, and proteasomal targeting of the Shaker K ϩ channel protein. The data indicate that glycosylation dramatically increases stability and cell surface expression, without significantly affecting folding and assembly of the protein. In the absence of N-linked glycosylation, a significant fraction of the protein is retained in the ER and degraded. Cytoplasmic proteasomes contribute to the degradation of the unglycosylated protein, although other nonlysosomal proteolytic pathways may also be involved. Our results imply that proteasomal targeting of the unglycosylated protein is not the result of gross misfolding or improper assembly, although we cannot rule out subtle changes in the structure of the protein resulting from the lack of carbohydrate modification.
How does N-linked glycosylation stabilize the Shaker protein? One possibility is that interaction with calnexin and/or calreticulin protects newly made wild-type Shaker protein from retrotranslocation and proteasomal degradation. We have previously shown that calnexin interacts transiently with the wild-type Shaker protein shortly after its synthesis in the ER (25). In contrast, the unglycosylated N259Q,N263Q mutant protein fails to associate with calnexin, consistent with the expectation that calnexin binds specifically to proteins containing monoglucosylated glycans (1,3).
An alternative hypothesis is that the wild-type and unglycosylated Shaker proteins are differentially sensitive to ER-associated degradation because they are transported out of the ER at different rates. Even though glycosylation has little apparent effect on the eventual acquisition of the native structure, it may increase the rate of folding and assembly of the Shaker protein, thereby allowing a more rapid exit from the ER. Although the oxidation assay provides a way to assess the net outcome of biogenesis, it does not monitor the kinetics of the HEK293T cells expressing S346C in the wild-type or N259Q,N263Q backgrounds were labeled with HPDP-biotin. 24 -36 h after transfection, biotinylated proteins were precipitated with streptavidin-agarose beads. After electrophoresis, Shaker protein was identified by immunoblot analysis using a polyclonal Shaker-␤-galactosidase fusion antibody, and the percentage of Shaker protein biotinylated was determined as described under "Experimental Procedures" (n ϭ 4, performed in quadruplicate). ***, p Ͻ 0.001 compared with S346C; paired Student's t test.
FIG. 8. Folding and assembly of Shaker channels is not impaired in the absence of glycosylation. HEK293T cells expressing the wild-type or N259Q,N263Q mutant proteins were subjected to metabolic labeling for 30 min, followed by a 3-h chase in nonradioactive medium. Intact cells were then incubated in 20, 50, or 100 mM H 2 O 2 for 10 min. The oxidation reaction was quenched with 5 mM N-ethylmaleimide. Shaker protein was immunoprecipitated, and electrophoresis was performed under nonreducing conditions. Fluorographs were quantified by densitometry to determine the extent of formation of higher molecular mass adducts at differing H 2 O 2 concentrations. The amount of wild-type (open bars) or N259Q,N263Q mutant (filled bars) protein found in adducts (dimer or larger) under nonreducing conditions was divided by the total protein (monomer plus adducts) in the same lane and plotted as percentage of adducts at the indicated H 2 O 2 concentrations (n ϭ 5-7).
process. Timely exit from the ER may be a key factor in protecting the wild-type but not the unglycosylated protein from degradation. Although we did not ascertain the kinetics of ER exit of the wild-type and unglycosylated Shaker proteins, our finding that a significant fraction of the unglycosylated Shaker protein is located in the ER (Fig. 5) supports the hypothesis that residence time in the ER is a determinant of degradation. Interestingly, mannose trimming by ER mannosidase I, which is implicated in the proteasomal targeting of some glycoproteins, has been proposed as a timing mechanism that regulates degradation (12). We propose that the kinetics of ER exit could serve as an alternative timing mechanism to regulate turnover. In this scenario, calnexin and/or calreticulin could, by accelerating folding and assembly, still play a role in protecting the wild-type protein from proteolysis.
We have identified a variety of mutations that disrupt folding or assembly of the Shaker protein, preventing exit from the ER (24). These mutant proteins are core glycosylated and, similarly to wild type, interact transiently with calnexin (25). Studying the turnover of these mutant proteins should shed light on the relative roles of calnexin and ER retention in regulating degradation of the Shaker protein.
In HEK293T cells, glycosylation stabilizes the Shaker protein and dramatically enhances its trafficking to the cell surface. A significant fraction of the unglycosylated protein is retained intracellularly. Yet functional channels are readily detected when the N259Q,N263Q mutant is expressed in Xenopus oocytes (Ref. 21 and this study). Although we have not quantitatively compared protein amounts or current amplitudes produced by the wild-type and unglycosylated mutant cRNAs in Xenopus oocytes, the levels of expression appear to be similar. This is consistent with the results of oxidation experiments, which indicate that the unglycosylated protein folds and assembles as efficiently as wild type in HEK293T cells. Taken together, these data suggest that the unglycosylated Shaker protein trafficks to the plasma membrane much more efficiently in Xenopus oocytes than in HEK293T cells. One factor that may be relevant is that oocytes and HEK293T cells are maintained at different temperatures, 17 and 37°C, respectively. This will affect the kinetics of folding and assembly, intracellular trafficking, and degradation, with the observed outcome of similar expression levels for the wild-type and N259Q,N263Q proteins. Given these results, it would be interesting to know whether the unglycosylated HERG K ϩ channel mutant, which is trapped intracellularly in mammalian cells, is able to reach the plasma membrane in Xenopus oocytes. Differential trafficking of other proteins, including cystic fibrosis transmembrane conductance regulator, in oocytes and in mammalian cells has been previously observed (40,41).
In the past few years, high resolution structures have been reported for the KcsA bacterial potassium channel as well as several soluble domains from K ϩ channels (42)(43)(44)(45). Structural analysis of a voltage-dependent K ϩ channel is now eagerly awaited but will require biochemical overexpression and purification of homogeneous protein preparations. Our results have implications for biochemical overexpression of eukaryotic K ϩ channel proteins for structural analysis, where unglycosylated protein might be useful for reducing heterogeneity of the protein preparation. Although our results suggest that genetic removal of glycosylation may not impair acquisition of the native state, lack of glycosylation may reduce levels of protein because of degradation in some cell types.