N-linked glycosylation sites determine HERG channel surface membrane expression

Abstract

1. Long QT syndrome (LQT) is an electrophysiological disorder that can lead to sudden death from cardiac arrhythmias. One form of LQT has been attributed to mutations in the human ether-a-go-go-related gene (HERG) that encodes a voltage-gated cardiac K⁺ channel. While a recent report indicates that LQT in some patients is associated with a mutation of HERG at a consensus extracellular N-linked glycosylation site (N629), earlier studies failed to identify a role for N-linked glycosylation in the functional expression of voltage-gated K⁺ channels. In this study we used pharmacological agents and site-directed mutagenesis to assess the contribution of N-linked glycosylation to the surface localization of HERG channels.

2. Tunicamycin, an inhibitor of N-linked glycosylation, blocked normal surface membrane expression of a HERG-green fluorescent protein (GFP) fusion protein (HERG^GFP) transiently expressed in human embryonic kidney (HEK 293) cells imaged with confocal microscopy.

3. Immunoblot analysis revealed that N-glycosidase F shifted the molecular mass of HERG^GFP, stably expressed in HEK 293 cells, indicating the presence of N-linked carbohydrate moieties. Mutations at each of the two putative extracellular N-linked glycosylation sites (N598Q and N629Q) led to a perinuclear subcellular localization of HERG^GFP stably expressed in HEK 293 cells, with no surface membrane expression. Furthermore, patch clamp analysis revealed that there was a virtual absence of HERG current in the N-glycosylation mutants.

4. Taken together, these results strongly suggest that N-linked glycosylation is required for surface membrane expression of HERG. These findings may provide insight into a mechanism responsible for LQT2 due to N-linked glycosylation-related mutations of HERG.

Long QT syndrome (LQT) is a disorder that can cause sudden death from cardiac arrhythmias, torsade de pointes and ventricular fibrillation. Mutations in the human ether-a-go-go (eag)-related gene (HERG) underlie chromosome 7-linked LQT (Satler et al. 1998). HERG is a member of the eag family of potassium channels (Warmke & Ganetzky, 1994). Family members contain six putative membrane-spanning domains, an ion-conducting pore region and a putative cyclic nucleotide-binding domain. Expression of wild-type HERG protein in Xenopus oocytes revealed that HERG encodes the rapidly activating, inwardly rectifying K⁺ channel responsible for the cardiomyocyte current I_Kr (Sanguinetti et al. 1995), an important component of ventricular repolarization. These findings have led to the proposal that reduced I_Kr, resulting from mutations in HERG, can prolong cardiomyocyte action potential duration, thereby giving rise to LQT (reviewed in Roden et al. 1996).

It has recently been reported that HERG mutations Y611H and V822M result in an incompletely glycosylated form of the protein exhibiting a restricted perinuclear distribution with no detectable current (Zhou et al. 1998a). In addition, it has been demonstrated that N-linked oligosaccharides can act as determinants for cell surface transport of certain membrane proteins (Gut et al. 1998). The HERG amino acid sequence contains two extracellular consensus N-linked glycosylation sites, N598 and N629. In this study, we set out to determine whether N-linked glycosylation is required for surface membrane expression of HERG in a mammalian cell line, HEK 293. To do so, we generated stably transfected cell lines expressing green fluorescent protein (GFP)-tagged HERG (HERG^GFP), and singly and doubly mutated HERG^GFP that had had either one or both of the conserved N-linked glycosylation sites removed. The wild-type and N-linked glycosylation mutant HERG^GFP fusion proteins were analysed for their biochemical and electrophysiological properties as well as their subcellular localization. We found that the wild-type protein exhibited a surface membrane localization and normal electrophysiological function; however, both single (N598Q, N629Q) and double (N598Q- N629Q) mutants showed an intracellular localization, with no detectable current. These results strongly suggest that N-linked glycosylation is required for surface membrane expression of HERG channels.

METHODS

DNA constructs and transfection of HEK 293 cells

HERG cDNA (kindly provided by Dr G. A. Robertson, University of Wisconsin, USA) was subcloned into BamHI/EcoRI sites of the pBK-CMV expression vector (Stratagene). GFP cDNA (kindly provided by Dr J. Orlowski, Department of Physiology, McGill University) was subcloned into the Not I/XbaI sites of the pBK-CMV expression vector. The HERG-GFP fusion construct (designated HERG ^GFP) was generated by ligating the 3′ end of the HERG cDNA to the 5′ end of the GFP cDNA. In order to do so, a Not I site at the 3′ end of the HERG cDNA was introduced by polymerase chain reaction (PCR) and the GFP cDNA was sublconed into the Not I/XbaI sites of the pBK-CMV-HERG expression vector. The final fusion construct was purified using a Wizard Plus Midiprep Purification System (Promega) and sequenced (Sheldon Biotechnology, Montréal, Québec, Canada) in order to verify its integrity.

The following primers were used to mutate either singly or in combination the two N-linked glycosylation sites located in the extracellular loop between transmembrane segments S5 and S6 using the unique site elimination technique (Deng & Nickoloff, 1992): N598Q, 5′ GGCAAACCCTACCAGAGCAGCGGCCTG; N629Q, 5′ GTGGGCTTCGGCCAGGTCTCTCCCAAC (mutated codons are underlined). The N-linked glycosylation mutant HERG^GFP fusion constructs were then purified using a Wizard Plus Midiprep DNA Purification System and sequenced (Sheldon Biotechnology).

HEK 293 cells were transfected with wild-type and mutant constructs using Lipofectamine (Gibco). After selection in 800 μg ml⁻¹ geneticin (G418, 50 % active; Gibco) for 10-15 days, single colonies were selected, grown and assayed for HERG^GFP expression. The stably transfected cells were maintained in α-minimum essential medium, 10 % fetal bovine serum, 1 % penicillin-streptomycin (Gibco) and 800 μg ml⁻¹ geneticin.

Immunocytochemistry

For transient transfectants, native GFP fluorescence was used to localize HERG^GFP fusion protein. Cells plated on coverslips were rinsed in phosphate-buffered saline (PBS), fixed in 4 % paraformaldehyde for 20 min at room temperature, rinsed in PBS and mounted using Immuno Fluore (Fisher Scientific). For tunicamycin (Sigma) treatment, 1 μg ml⁻¹ tunicamycin was added to cells 5 h post-transfection for an additional 28 h. Cells were then fixed as above.

For the stable transfectants, however, an anti-HERG antibody was used to localize HERG^GFP as the native GFP fluorescence was too low to allow for subcellular localization. The low level of native GFP fluorescence is due not to a low level of fusion protein expression but to a reduction in the absorption, and thus emission, of the chromophore due to dimerization of the GFP molecule. This dimerization results from high levels of fusion protein expression (information from Clontech). Briefly, stably transfected cells plated on coverslips were fixed in 2 % paraformaldehyde for 30 min followed by five rinses in PBS. Cells were then incubated in 0.5 % Triton X-100-0.5 % BSA in PBS for 30 min followed by incubation with an anti-HERG antibody generated against the amino terminus of HERG, anti-N (kindly provided by Drs A. Pond & J. M. Nerbonne, Department of Pharmacology and Molecular Biology, Washington University, St Louis, MO, USA; Pond & Nerbonne, 1996), for 1 h and subsequently rinsed in PBS. The cells were then incubated with fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc., Bar Harbor, ME, USA) for 1 h followed by three rinses in PBS. All incubations were performed at room temperature. Coverslips were mounted using Immuno Floure.

All imaging was performed using a Zeiss LSM 410 inverted confocal microscope. GFP and FITC were imaged by exciting the sample with a 488 nm line from an argon-krypton laser and the resulting fluorescence was collected on a photomultiplier after passage through FT510 and BP515-540 filter sets. Optical sections were taken using a × 25, 0.8 NA (optical thickness, 3.1 μm) objective and a × 63, 1.4 NA (optical thickness, 1.0 μm) objective. All images were printed on a Kodak XLS8300 high resolution (300 DPI) printer.

Immunoblot analysis

Parallel 100 mm plates of similarly confluent cultures were used to isolate crude membrane fractions. All steps were performed at 4°C. Briefly, the cells were rinsed 5 times with PBS, lysed with 2 ml of a solution containing (mM): 200 NaCl, 33 NaF, 10 EDTA and 50 Hepes (pH 7.4) and Complete protease inhibitor cocktail (Boehringer Mannheim, diluted according to manufacturer's instructions), and harvested with the aid of a rubber scraper. Cells were then homogenized and centrifuged at 3600 g for 10 min. The supernatants were collected and the membrane fractions were pelleted by centrifugation at 110 000 g for 40 min. The membrane-enriched pellets were solubilized in 50 mM Tris-HCl, 15 mM β-mercaptoethanol and 1 % SDS. The membrane proteins were then heated to 70°C in sample buffer (0.12 M Tris-HCl (pH 6.8), 2 % SDS, 2 %β-mercaptoethanol, 20 % glycerol and 0.001 % Bromophenol Blue) for 2 min and resolved on a 6.25 % SDS polyacrylamide gel. The proteins were then transferred onto a PVDF membrane (Amersham) overnight. Membranes were quenched in PBS with 5 % dried milk-0.1 % Tween 20 for 1 h at room temperature and subsequently incubated with the anti-N antibody at a 1:500 dilution for 2 h at room temperature, followed by three 10 min rinses in PBS with 0.1 % Tween 20. Membranes were then incubated in a 1:3000 dilution of horseradish peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch Laboratories, Inc.) for 1 h at room temperature. Horseradish peroxidase-conjugated secondary antibodies were detected using the chemiluminescence ECL⁺ detection kit (Amersham).

For N-glycosidase F treatment, 1 % NP-40, Complete protease inhibitor cocktail, and 2 U N-glycosidase F (Boehringer) were added to ∼30 μg of the resuspended membrane fractions. The mixture was then incubated at 37°C for 21 h. The reaction was stopped by adding sample buffer and boiling for 2 min.

Electrophysiological analysis

The whole-cell patch clamp technique was used to record membrane currents. Cells were plated on 35 mm Petri dishes and placed on the stage of an inverted microscope (Zeiss IM35). All experiments were carried out at 35 ± 1°C. Patch clamp electrodes were filled with medium containing (mM): 130 KCl, 1 MgCl₂, 5 EGTA, 5 MgATP and 10 Hepes (pH 7.2 with KOH). The pipette tip resistance was 2-4 MΩ. The external medium contained (mM): 137 NaCl, 4 KCl, 1.8 CaCl₂, 1 MgCl₂, 10 glucose and 10 Hepes (pH 7.4 with NaOH). Membrane currents were recorded using an Axopatch amplifier (Axopatch-1D, Axon Instruments Corp.). Voltage clamp pulses were delivered from a custom-designed software package (Alembic Software Co., Montréal, Canada) implemented on a personal computer equipped with an analog-to-digital card (Omega Corp., Stanford, CT, USA). The data were stored on a computer hard disk and analysed using the same software. A two-step voltage clamp protocol was imposed from a holding potential of -80 mV to assess the presence of HERG current. The first step, which served to activate HERG current, consisted of a 4 s depolarizing pulse to potentials between -60 and +50 mV in increments of 10 mV. At the end of the first step the membrane was repolarized back to -60 mV for 2 s, in order to assess HERG tail currents, before returning to the -80 mV holding potential. Cell capacitance was estimated by analog measurements using the patch clamp amplifier. Graphics and statistical analysis were carried out using Origin software (Microcal, Northampton, MA, USA). E-4031 was a generous gift from Eisai Co. (Tsukuba Research Laboratories, Japan).

RESULTS

Role of N-linked glycosylation in surface membrane expression of HERG^GFP

Fluorescence confocal microscopy was used to determine the subcellular localization of HERG^GFP transiently expressed in HEK 293 cells. Native GFP fluorescence revealed that HEK 293 cells transiently expressing HERG^GFP showed abundant surface membrane expression of HERG^GFP (Fig. 1A, arrows indicate localization between cells; Fig. 1B, arrows indicate surface membrane expression). When the cells were treated for 28 h with tunicamycin, an inhibitor of N-linked glycosylation, the transfected protein was no longer seen at the surface membrane but showed an intracellular accumulation with a predominant perinuclear subcellular localization (Fig. 1C, the arrow indicates an endoplasmic reticulum (ER)-Golgi-like distribution pattern; Fig. 1D). These results indicate that treatment of cells with tunicamycin results in an intracellular retention of HERG^GFP with no surface membrane expression.

Figure 1. Effect of tunicamycin on the subcellular localization of transiently expressed HERG^GFP in HEK 293 cells.

A, subcellular localization of HERG^GFP. Native GFP fluorescence overlayed on a transmitted-light image. Arrows indicate HERG^GFP localization at surface membranes. B, fluorescence image revealing the subcellular localization of HERG^GFP. Arrows indicate HERG^GFP localization at surface membranes. C, subcellular localization of HERG^GFP in cells treated with 1 μg ml⁻¹ tunicamycin for 28 h. Native GFP fluorescence overlayed on a transmitted-light image. Arrows indicate the perinuclear localization of HERG^GFP. D, fluorescence image revealing the subcellular localization of HERG^GFP in cells treated with 1 μg ml⁻¹ tunicamycin for 28 h. Representative images of three separate experiments are shown.

Biochemical analysis of HERG^GFP stably transfected into HEK 293 cells

Immunoblot analysis of HERG^GFP stably transfected into HEK 293 cells is shown in Fig. 2. An antibody directed against the amino terminus of HERG, anti-N, was used to probe for HERG^GFP protein. The anti-N antibody recognized multiple species of HERG^GFP in stably transfected cells; an upper band with an apparent molecular mass of ∼185 kDa and a lower band with an apparent molecular mass of ∼165 kDa. A degradation product at ∼70 kDa was also evident (Fig. 2, lane 2). Neither band was present in untransfected HEK 293 cells (Fig. 2, lane 1). Similarly, Zhou et al. (1998b) reported that two bands of HERG are evident upon Western blot analysis; an upper band with an apparent molecular mass of 155 kDa and a lower band with an apparent molecular mass of 135 kDa. This is consistent with our results as we have an ∼30 kDa GFP molecule tagged to the carboxyl terminus of HERG.

Figure 2. Immunoblot analysis of HERG^GFP protein stably expressed in HEK 293 cells.

Enriched membrane fractions were prepared as described in Methods. Each lane was loaded with ≈30 μg of protein. An anti-HERG antibody was used to probe for HERG^GFP. The control lane shows results with untransfected cells. The HERG^GFP lanes show results with cells stably expressing HERG^GFP before (-) and after (+) treatment with N-glycosidase F. The data are representative of four separate experiments.

To determine the extent of N-linked carbohydrate modification, protein isolated from HEK 293 cells stably expressing HERG^GFP was treated with N-glycosidase F, which removes N-linked carbohydrate from proteins. This treatment converted the 185 kDa band to an ∼170 kDa band and shifted the lower 165 kDa band to an ∼162 kDa band (Fig. 2, lane 3). The ∼70 kDa degradation product was also shifted, suggesting that it is derived from a glycosylated portion of the protein. These results suggest that both molecular mass species of HERG^GFP undergo N-linked glycosylation, with the higher molecular mass species most probably representing a fully glycosylated form of the protein.

Biochemical analysis of HERG^GFP and N-linked glycosylation mutants of HERG^GFP stably transfected into HEK 293 cells

HERG contains two potential extracellular consensus N-linked glycosylation sites: one site (N598) near the putative pore region, and the second (N629) within the pore region. To determine the role of N-linked glycosylation at each of these sites three constructs were generated by site-directed mutagenesis: N598Q, N629Q and N598Q-N629Q. Each of these three N-linked glycosylation mutant HERG^GFP fusion proteins was stably expressed in HEK 293 cells.

Immunoblot analysis of total membrane protein isolated from wild-type HERG^GFP, N598Q-HERG^GFP, N629Q-HERG^GFP and N598Q-N629Q-HERG^GFP stably expressed in HEK 293 cells is shown in Fig. 3. For each of the three N-linked glycosylation mutant HERG^GFP proteins the anti-N antibody recognized a protein species with an apparent molecular mass of ∼162 kDa (Fig. 3, lanes 3-5). The apparent molecular mass of each of the mutant HERG^GFP proteins was similar to the lower band of the enzymatically deglycosylated wild-type HERG^GFP protein (Fig. 2, lane 3), suggesting that none of the mutant proteins was glycosylated, and that both N-linked glycosylation sites must be available for either site to be glycosylated as mutations at either N-linked glycosylation site led to a non-glycosylated form of the protein. An ∼70 kDa degradation product was also evident for the wild-type protein and a shifted correlate product was evident for all three mutant proteins, indicating that the degradation product is derived from a glycosylated portion of the protein.

Figure 3. Immunoblot analysis of wild-type and N-linked glycosylation mutants of HERG^GFP stably expressed in HEK 293 cells.

Enriched membrane fractions were prepared as described in Methods. Each lane was loaded with ≈30 μg of protein. An anti-HERG antibody was used to probe for HERG^GFP. The control lane shows results with untransfected cells. Results from protein isolated from cells stably expressing HERG^GFP, N598Q-HERG^GFP (N598Q), N629Q-HERG^GFP (N629Q) or N598Q-N629Q-HERG^GFP (N598Q-N629Q) are shown in their respective lanes. The data are representative of four separate experiments.

Functional analysis of N-linked glycosylation mutants

To evaluate the functional role of N-linked carbohydrate modification of HERG^GFP, the whole-cell patch clamp technique was used to study the membrane currents in HEK 293 cells stably expressing wild-type HERG^GFP and the N-linked glycosylation mutants. Figure 4Aa shows current traces recorded during a series of depolarizing voltage clamp steps from a -80 mV holding potential to potentials ranging from -60 to +50 mV. A tail current can be observed at the termination of the initial depolarizing step when the membrane is suddenly stepped to -60 mV. Figure 4Ab shows that the time-dependent current induced during the step and the current tail were almost completely blocked by E-4031, a selective inhibitor of I_Kr and I_HERG (Sanguinetti et al. 1995). In the presence of E-4031 a small amplitude rapidly activating transient outward current persisted along with a residual steady-state background current. Control experiments carried out using untransfected HEK 293 cells (n = 32 cells) showed that the small amplitude transient current was present in most cells but the large time-dependent outward current observed in cells transfected with HERG^GFP was never observed. These results indicate that the HERG^GFP fusion protein is functional and that the time-dependent outward current in cells transfected with HERG^GFP (I_HERG^GFP) has properties similar to I_HERG. This is further illustrated in Fig. 4Ac, which shows the current-voltage relationship measured at the end of the initial 4 s depolarization step in HEK 293 cells stably transfected with HERG^GFP and in control untransfected cells. The current-voltage relationship indicates that I_HERG^GFP is activated at membrane potentials positive to -40 mV. At potentials positive to 0 mV the current-voltage relationship showed a marked inward rectification with a negative slope conductance. In untransfected cells a small residual linear background current persisted with reversal close to 0 mV suggesting a degree of non-selectivity. The above observations are very similar to those recently reported for HEK 293 cells transfected with HERG (Zhou et al. 1998b) and demonstrate that HERG^GFP encodes an ion channel with pharmacological and electrophysiological characteristics similar to the channel encoded by native HERG.

Figure 4. Whole-cell outward currents in HEK 293 cells stably expressing wild-type and N-linked glycosylation mutants of HERG^GFP.

A two-step voltage clamp protocol was imposed from a holding potential of -80 mV to assess the presence of HERG current in HEK 293 cells stably transfected with wild-type HERG^GFP or HERG^GFP mutants. The outward currents were evoked during an initial 4 s depolarizing pulse to potentials between -60 and +50 mV in increments of 10 mV. At the end of the first step the membrane was repolarized back to -60 mV for 2 s before returning to the -80 mV holding potential. Aa, a series of current traces recorded in HEK 293 cells stably transfected with HERG^GFP showing an inwardly rectifying delayed rectifier current. Ab, in the presence of E-4031 (1 μM) the HERG^GFP-induced delayed rectifier current was completely blocked. The residual transient outward current and background current were similar to those observed in untransfected cells. Ac, current-voltage relationship of untransfected HEK 293 cells (n = 6 cells, ▴) and HEK 293 cells stably transfected with HERG^GFP (n = 6, ▪) normalized to the cell capacitance (16.3 ± 1.5 pF; n = 6). B, each panel shows a series of current traces recorded from HEK 293 cells stably transfected with N598Q-HERG^GFP, N629Q-HERG^GFP or N598Q-N629Q-HERG^GFP. HERG^GFP current was absent in all three mutants.

The current recordings from HEK 293 cells stably expressing the N-linked glycosylation mutants N598Q-HERG^GFP, N629Q-HERG^GFP and N598Q-N629Q-HERG^GFP are shown in Fig. 4B. The currents were induced using the same voltage clamp protocol as in Fig. 4A. It was found that the large I_HERG^GFP evident in wild-type HERG ^GFP stably transfected cells was absent in all three N-linked glycosylation mutants (n = 40 for each mutant). The recordings indicate that HEK 293 cells transfected with the mutants retain the small transient outward current. Current-voltage plots of the three mutants (data not shown) were similar to those observed in untransfected HEK 293 cells (Fig. 4Ac), consistent with the presence of a small amplitude endogenous background current.

Subcellular localization of N-linked glycosylation mutant proteins

Fluorescence confocal microscopy was used to determine the subcellular localization of the wild-type and N-linked glycosylation mutants of HERG^GFP stably expressed in HEK 293 cells. Wild-type HERG^GFP exhibited a surface membrane distribution pattern (Fig. 5A), consistent with the functional electrophysiological recordings. In contrast, N598Q-HERG^GFP, N629Q-HERG^GFP and N598Q-N629Q-HERG^GFP revealed a perinuclear subcellular localization consistent with the absence of a HERG current upon electrophysiological analysis (Fig. 5B-D) indicating that both N-linked glycosylation sites must be available for surface membrane expression of HERG^GFP.

Figure 5. Subcellular localization of wild-type and N-linked glycosylation mutants of HERG^GFP stably expressed in HEK 293 cells.

Cells were immunolabelled with an anti-HERG antibody followed by a FITC-conjugated goat anti-rabbit IgG antibody. Representative immunofluorescence images are overlayed on transmitted-light images. A, surface membrane localization of wild-type HERG^GFP. B, perinuclear localization of N598Q-HERG^GFP. C, perinuclear localization of N629Q-HERG^GFP. D, perinuclear localization of N598Q-N629Q-HERG^GFP. Representative images of four separate experiments are shown.

DISCUSSION

In the present study we set out to determine the role of N-linked glycosylation in surface membrane expression of HERG. In part, this was assessed using an inhibitor of N-linked glycosylation, tunicamycin. As a means to visualize the subcellular localization of HERG, we constructed a HERG-GFP fusion protein, HERG^GFP. We found that treatment of HERG^GFP transiently expressed in HEK 293 cells with tunicamycin led to an absence of surface membrane expression of the protein and an intracellular accumulation exhibiting a perinuclear subcellular distribution. In order to determine whether the HERG^GFP fusion construct functioned normally, whole-cell patch clamp recordings of HERG^GFP stably expressed in HEK 293 cells were performed. The HERG^GFP current that was generated revealed similar properties to HERG stably expressed in the same cell line (Zhou et al. 1998b). These findings indicate that GFP tagged to the carboxyl terminus affects neither HERG^GFP targeting to the cell surface nor its function.

Immunoblot analysis of HERG^GFP revealed that it exists as two molecular mass species of ∼185 and ∼165 kDa. Upon enzymatic deglycosylation with N-glycosidase F, the larger molecular mass species shifted from ∼185 to ∼170 kDa, and the smaller molecular mass species shifted from ∼165 to ∼162 kDa. These findings are consistent with those of Zhou et al. (1998b) and indicate that the upper band most probably represents the fully glycosylated form of HERG. Since treatment with the enzyme did not convert the two protein bands to a single band, post-translational modifications other than N-linked glycosylation may be involved. Possibilities include palmitylation, sulfation, phosphorylation or acylation. O-linked glycosylation does not seem to be a possibility as HERG does not contain such a consensus site.

In order to elucidate the role of each of the two consensus N-linked glycosylation sites, N598 and N629, we constructed three N-linked glycosylation mutants, N598Q-HERG^GFP, N629Q-HERG^GFP and N598Q-N629Q-HERG^GFP, and stably expressed them in HEK 293 cells. Replacement of the asparagine at amino acid 598 with a glutamine resulted in the loss of glycosylation not only at site 598, but also at 629, yielding a molecular mass species of ∼162 kDa. Similarly, replacement of the asparagine at amino acid 629 with glutamine resulted in the loss of glycosylation at sites 629 and 598, yielding a molecular mass species of ∼162 kDa. The double mutant also resulted in a molecular mass species of ∼162 kDa. These findings indicate that both glycosylation sites, N598 and N629, must be available for glycosylation to occur. Immunofluorescence analysis revealed that each of the N-linked glycosylation mutants showed an absence of surface membrane expression, with a perinuclear subcellular distribution. Moreover, no HERG current was detectable in any of the three N-linked glycosylation mutants. These results indicate that N-linked glycosylation is required for proper trafficking of HERG^GFP as the unglycosylated 162 kDa form of the protein does not appear to exit the ER-Golgi apparatus. An alternative explanation is that the replacement of the asparagine with a glutamine yields a three-dimensional configuration of the protein rendering it susceptible to degradation prior to exiting the ER-Golgi apparatus. However, treatment of cells with tunicamycin resulted in an ER-Golgi-like retention of HERG^GFP that was similar to that observed with the N-linked glycosylation mutants. This further supports the notion that N-linked carbohydrate is required for trafficking of HERG^GFP to its final destination at the cell surface. Moreover, in one of the cells stably expressing N598Q-HERG^GFP we were able to record a small HERG-like current indicating that at least some of the protein retains its proper three-dimensional configuration in the non-glycosylated state, and that a small percentage is able to reach the surface membrane. Thus it appears that for HERG^GFP, the consensus N-linked glycosylation sites must both be available for glycosylation to occur, and that the non-glycosylated form of the protein is retained intracellularly.

To date, 19 different HERG mutations in LQT patients have been reported (Curran et al. 1995; Akimoto et al. 1996; Benson et al. 1996; Dausse et al. 1996; Satler et al. 1996, 1998; Tanaka et al. 1997). Importantly, it was recently reported that HERG protein resulting from mutations Y611H and V822M was incompletely glycosylated and no current was detectable (Zhou et al. 1998a). In addition, a recent study has demonstrated that N-linked oligosaccharides can act as determinants for cell surface transport of certain membrane proteins (Gut et al. 1998). Recently two mutations resulting in LQT were found at a consensus extracellular N-linked glycosylation site, N629D and N629S (Satler et al. 1998). Taken in the context of the results presented in this paper, these observations suggest that mutations at N-linked glycosylation sites or mutations that inhibit N-linked glycosylation may affect HERG protein trafficking and surface expression.

The role of carbohydrates in the trafficking and function of certain ion channels has been reported by several groups. Waechter et al. (1983) demonstrated that tunicamycin reduced the number as well as the function of sodium channels on the cell surface. The effect of tunicamycin was not mediated solely by increasing the rate of proteolytic degradation of internalized sodium channels but must also reduce the rate of sodium channel appearance on the cell surface and/or increase the rate of internalization of channels from the cell surface. Merlie et al. (1982) reported that when muscle cells were treated with tunicamycin, the α1 subunit of the muscle-type nicotinic acetylcholine receptor failed to assemble into toxin-binding receptors on the cell surface. In addition, Mishina et al. (1985) reported that when the asparagine at amino acid 141 of the Torpedo α1 subunit was replaced with aspartic acid, there was no acetylcholine response and no detectable α-bungarotoxin-binding sites on the cell surface. N-linked glycosylation has also been shown to be necessary for proper trafficking of the GLYT1 glycine transporter to the plasma membrane (Olivares et al. 1995).

The role of N-linked glycosylation has not been clearly established for K⁺ channels. Santacruz-Toloza et al. (1994) reported that for the Shaker K⁺ channel, N-linked glycosylation is not required for the assembly of functional channels or for their transport to the cell surface. Similarly, Deal et al. (1994) reported that glycosylation at the single extracellular N-linked glycosylation site is not required for subunit assembly, transport to the surface, protein stability or function of the Kv1.1 channel.

Early studies with tunicamycin demonstrated that when N-linked carbohydrate addition to protein was blocked, most non-glycosylated forms of the proteins accumulated in the ER, aggregated, and did not exit. This led to the concept that carbohydrate addition aids protein folding and stabilizes the protein conformation rendering it less susceptible to proteolytic degradation (Fiedler & Simons, 1995). Further studies have demonstrated that glycoproteins associate with ER resident chaperones, including calnexin, calreticulin, PDI, Bip and Grp94, and that this association promotes correct folding and oligomeric assembly, prevents degradation and supports quality control (Hammond & Helenius, 1995; Hebert et al. 1997). ER retention of misfolded or unassembled proteins has been demonstrated for the ΔF508 cystic fibrosis transmembrane conductance regulator, the truncated α chain of β-hexosaminidase and unassembled subunits of the T cell receptor (Bonifacino & Lippincot-Schwartz, 1991; Ward & Kopito, 1995).

In summary, we report that inhibition of N-linked glycosylation using tunicamycin results in a perinuclear accumulation of HERG^GFP in HEK 293 cells. Replacement of the asparagines at the each of the consensus N-linked glycosylation sites with glutamine leads to a similar perinuclear accumulation of HERG^GFP with no surface membrane expression and no detectable current. These findings may help to provide insight into a mechanism responsible for the lack of HERG current resulting from certain mutations observed in individuals with LQT2.