The FASEB Journal express article 10.1096/fj.03-0577fje. Published online October 16, 2003.
N-glycosylation-dependent block is a novel mechanism for drug-induced cardiac arrhythmia Ki-Ho Park, Suk-Mei Kwok, Chetna Sharon, Rebecca Baerga, and Federico Sesti University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Department of Physiology and Biophysics, Piscataway, New Jersey 08854 Corresponding author: F. Sesti, Ph.D., University of Medicine and Dentistry of New Jersey, Robert Wood Johnson Medical School, Department of Physiology and Biophysics, 675 Hoes Lane, Piscataway, NJ 08854. E-mail:
[email protected] ABSTRACT Voltage-gated potassium channels formed with the cardiac subunit HERG and a polymorphic variant of MinK-related peptide (MiRP1) exhibit increased susceptibility to the antibiotic sulfamethoxazole (SMX) compared with channels formed with wild-type (WT) subunits. Here the molecular bases for SMX high-affinity block are investigated. The polymorphism causes a benign T to A amino acid mutation at position 8 (T8A) that destroys an N-glycosylation site of MiRP1. In vitro disruption of glycosylation by mutagenesis or in vivo by treatment with neuraminidase is associated with increased susceptibility to SMX and to other elementary agents such as divalent cations. Defective glycosylation does not affect the ability of T8A to form stable complexes with HERG, but rather it increases drug susceptibility through structural modifications in the channel complex. We conclude that N-glycosylation may play a key role in the etiology of life-threatening arrhythmia. Key words: HERG • MiRP1 • KCNE2 • sulfamethoxazole • LQTs
F
ailures in ventricular repolarization are often characterized by a lengthening of the QT interval in the electrocardiogram that predisposes to a specific form of polymorphic, lifethreatening ventricular tachycardia known as long QT syndrome (LQTs). LQTs can be divided on clinical grounds into two main types: congenital and acquired, with the second far more common. Factors that trigger the onset of acquired LQTs include cardiac disease, metabolic disorders, and treatment with unspecific medications, such as antibiotics, antihistamines, and antiarrhythmic (1). The consensus view is that genetic predisposition may play an important role in drug-induced LQTs (2–7). For instance, the QT interval measured before drug exposure tends to be longer in patients who later develop arrhythmia than in individuals who receive the same agent safely (8, 9). The molecular bases for LQTs reside primarily in disturbances in the function of the channels that generate the electrical activity of the heart, especially of the K+ channel complex IKr (10–12). There is evidence suggesting that native IKr channels are composed of the pore-forming HERG and the MinK-related peptide 1 (MiRP1) β-subunit (2, 3, 7, 13–16). MiRPs are single-transmembrane proteins that associate with and modulate K+ channels (17). Although MiRPs were initially identified in mammals, recently they have been reported in amphibians and invertebrates, suggesting that these proteins might play a general role as K+ channel subunits (18, 19).
Several missense mutations in the gene encoding MiRP1 have been found in the genomes of individuals affected by acquired LQTs. Functional studies revealed that channels formed with these mutants exhibited decreased potassium flux in heterologous expression systems, suggesting that in native cardiomyocytes these MiRP1 mutants impair cardiac repolarization capacity and thus may predispose to LQTs (2, 7, 13, 16). Genetic screening revealed also the existence of a single-nucleotide polymorphism, T8A, which is estimated to be carried by 1-2% of the western population (7). T8A was successively identified in a patient who had developed LQTs after administration of trimethoprim/sulfamethoxazole (2). When T8A/HERG channels were studied in mammalian expression systems, they were found to be inhibited fourfold more potently by sulfamethoxazole (SMX) than WT channels. These findings provided a natural explanation for why most patients receive SMX without incident and others develop QT prolongation (2). In this study, we sought to investigate the molecular mechanisms underlying SMX high-affinity block. Here we disclose a novel mechanism that might have significant implications for the treatment of acquired LQTs: defective N-glycosylation of MiRP1 subunits. MATERIALS AND METHODS Molecular biology MiRP1 mutants were constructed by PCR and inserted in pCI-neo vector (Promega) for functional expression in CHO cells. All sequences were confirmed by automated DNA sequencing. Transcripts were quantified with spectroscopy and compared with control samples separated by agarose gel electrophoresis and stained with ethidium bromide. Biochemistry A HA epitope was fused to the N terminus of WT and T8A MiRP1 by PCR. HERG was tagged by replacing the stop codon with a cmyc sequence. Transient transfection of COS cells was with Superfect kit (Quiagen). Cells were lysed in buffer (in mM): 150 NaCl, 1% NP-40, 1% Triton, 50 Tris (pH=7.4), and a protease inhibitor cocktail (Boehringer). Immunoprecipitations were carried out with anti-cmyc monoclonal antibody 9E10 (Roche) and immobilized protein G (Pierce). Samples were separated by SDS-page (10-15%). Western blots were performed with anti-HA or anti-cmyc monoclonal antibody (Roche) conjugated to horseradish peroxidase (Sigma). For immunocytochemistry, cells were incubated in fresh complete media containing the monoclonal anti-HA antibody at 37°C for 1 h. Cells were washed once with phosphate-buffered saline (PBS) and fixed with paraformaldehyde (4% in PBS) for 15 min at room temperature. After being fixed, cells were washed three times for 5 min with PBS and blocked for 1 h at room temperature with 5% nonfat dry milk in PBS plus 0.1% Tween-20. Cells were incubated with the secondary antibody conjugated goat anti-mouse (Jackson ImmunoResearch) for 1 h at room temperature and subsequently washed three times for 5 min with PBS. Electrophysiology CHO cells were transiently transfected with cDNA ligated into pCI-neo using Superfect kit (Qiagen) and studied after 24-36 h. Data were recorded with an Axopatch 200B (Axon), a PC
(Dell), and Clampex software (Axon), filtered at 1.0 kHz, and sampled at 2.5 kHz. Bath solution was (in mM): 4 KCl, 100 NaCl, 10 HEPES (pH=7.5 with NaOH), 1.8 CaCl2, and 1.0 MgCl2. Pipette solution was 100 KCl, 10 HEPES (pH=7.5 with KOH), 1.0 MgCl2, 1.0 CaCl2, and 10 EGTA (pH=7.5 with KOH). SMX solutions were prepared fresh daily. RESULTS Our first step toward the investigation of the molecular mechanisms of SMX block was directed at ascertaining whether susceptibility to this drug is an intrinsic property of HERG or it is rather conferred by MiRP1. Figure 1A shows representative whole cell HERG currents recorded in the absence and in the presence of 1 mg/ml SMX (voltage protocol shown in the inset). SMX reversibly inhibits HERG current by diminishing the amplitude of the tail current and by speeding deactivation kinetics. The instantaneous decrease of the current could be due to different mechanisms, diminished unitary current for instance, and was not investigated further. Deactivation was assessed by fitting to a double exponential function tail currents elicited after preactivation by a depolarizing step at +20 mV (Fig. 1B, inset). SMX speeded both the fast and the slow components without altering the intrinsic voltage dependence of deactivation (Fig. 1B). In contrast, peak or steady-state tail currents were inhibited in a voltage-dependent fashion with hyperpolarization facilitating block (Fig. 1C). Block was also dependent on the magnitude of the activating test potential. The steady tail current at –40 mV was progressively inhibited by increasing prepulse depolarization (Fig. 1D), suggesting that SMX blocks HERG channels during or after they have transited through the open state. To include the effect of SMX on channel deactivation, in this report we use steady-state tail currents to quantify current-dose relationships (Fig. 1A, arrow). Dose-response curves for HERG channels were well fitted to the Hill equation (Fig. 1E, triangles), yielding Ki = 0.34 ± 0.02 mg/ml and n = 1.4 ± 0.2. Surprisingly, these channels are inhibited like channels formed with T8A (Ki=0.30±0.02 mg/ml, Fig. 1E, squares) and therefore are fourfold more sensitive than complexes containing WT MiRP1 subunits (Ki=1.24±0.09 mg/ml, Fig. 1E, circles). As a result of using steady-state currents, inhibition constants are decreased approximately twofold compared with those calculated using the peak current, as done previously (2). We note that the triplet of amino acids NFT at positions 6-8 constitutes one putative Nglycosylation site in MiRP1, which is predicted to be disrupted by the T to A mutation (20). Glycosylation of HA epitope-tagged MiRP1 subunits (Fig. 2, schematic), transiently expressed in COS-7, was evaluated by Western blot analysis (Fig. 2A). As described previously (7), WT subunits yielded three bands reflecting the di-, mono-, and unglycosylated state of the protein since there are two N-linked glycosylation sites in MiRP1. In contrast, T8A was detected in the mono- and unglycosylated but not in the fully glycosylated form, as expected from sequence analysis (20) (Fig. 2A). Since glycosylation is a well-recognized factor affecting intracellular retention (21–24), high affinity block of T8A mutants might be the result of lack of MiRP1 subunits at the plasma membrane with consequent formation of homomeric HERG channels. To address this possibility, we employed standard immunocytochemistry methods and used constructs epitope-tagged to the N terminus (schematic) because this region is predicted to be extracellular. Figure 2B shows examples of immunofluorescence staining of WT and T8A coexpressed with HERG in CHO cells. Unpermeabilized WT and T8A showed a clear ring of stain around the surface of the cell, suggesting normal surface expression. Similar results were obtained with cells transfected with WT or T8A alone (not shown). Quantitatively, expression of
tagged WT and T8A in CHO cells was also similar, resulting in ~6% of the cells in both cases (Fig. 2C). Coexpression of T8A MiRP1-HA with HERG-cmyc allowed recovery of both by immunoprecipitation (IP) with an anti-cmyc monoclonal antibody (Fig. 2D). Recovery was specific because anti-cmyc IP gave no signal when T8A was expressed alone (Fig. 2D). We conclude that disruption of N-glycosylation does not affect the ability of T8A to co-assemble with HERG, which is also consistent with previous studies that provided functional evidence for the formation of heteromeric T8A/HERG complexes (2, 7). Since T8A appears to be processed and to associate normally with HERG, we next investigated whether high-affinity block might arise from altered electrostatic interactions and/or allosteric rearrangements. To explore these possibilities, we employed a mutagenesis approach with the idea that in the presence of these mechanisms, mutations of T and/or its neighbor residues would alter SMX blockage to different extent. Conversely, all mutations we tested, S5A, N6A, F7A, T8A, Q9E with the exception of N6A (and obviously T8A), did not alter blockage (Fig. 3A). The fact that channels formed with N6A and T8A but not with F7A are more sensitive to the drug suggests that defective glycosylation might be the primary cause for high-affinity block. In fact, glycosylation sequences are triplets in the form: N-X-S/T, with any residue in the middle (20). Consistent with this notion, T8S subunits exhibited WT susceptibility to the drug (Ki=1.28±0.10 mg/ml, Fig. 3A), whereas channels formed with N6C were blocked like N6A and T8A mutants (Ki=0.30±0.11 mg/ml, Fig. 3A). To further test our hypothesis we treated the channels with the enzyme neuraminidase, which hydrolytically cleaves sialic acids from complex glycoproteins (25). Neuraminidase was chosen because it accesses only the proteins already on the surface and thus would not affect HERG trafficking (26). Treatment with the enzyme did not affect SMX block of HERG channels alone (Fig. 3B). In contrast, affinity was increased fourfold in heteromeric WT MiRP1/HERG complexes (Fig. 3C). Thus, while the Ki of HERG channels remained fairly unaffected by neuraminidase (Ki=0.34±0.02 mg/ml before and Ki=0.32±0.04 mg/ml after incubation), the Ki of WT MiRP1/HERG channels measured before incubation was Ki = 1.24 ± 0.1 mg/ml and decreased to Ki = 0.37 ± 0.02 mg/ml after incubation, a value characteristic of T8A and/or N6A mutants. Moreover, the enzyme had no effect on T8A mutants (Ki=0.26±0.08 mg/ml, Fig. 3D). Taken together these data demonstrate that defective N-glycosylation of MiRP1 is the primary cause for increased susceptibility to SMX. The effect appears to depend mostly on the N6 site because complete cleavage by neuraminidase does not further increase or alter blockage. HERG channels exhibit a peculiar sensitivity to diverse drugs, divalent cations, and specific toxins that act to block the channel from the external side of the membrane (27–33). To ascertain whether this susceptibility might arise from common structural determinants, we elected cadmium as the test ion and characterized its inhibitory effect. As reported by others (29), we found HERG channels being blocked by external cadmium ions in a dose-dependent manner with a Ki = 0.14 ± 0.01 mM (Fig. 4A and E). Cadmium shifted the threshold for activation toward more positive voltages (∆V=13.6±3.6 mV, n=5 with 0.3 mM Cd2+, not shown) and modified the rate of deactivation of HERG currents without altering the intrinsic voltage dependence of the process (Fig. 4B). Blockade was facilitated by hyperpolarization (Fig. 4C and D) and by increasing the magnitude of the conditioning test pulse (not shown). Thus, the modifications that cadmium exerts on the channel are remarkably similar to those of SMX molecules, suggesting that both agents might act to inhibit the protein through common mechanisms and shared
molecular determinants. Consistent with this notion, the presence of SMX in the test solution (0.5 mg/ml) caused an apparent decrease of HERG susceptibility to cadmium approximately fourfold (as a result of current normalization, Ki=0.61±0.08 mM, Fig. 4F), a change not expected with independent binding processes. Moreover, we found that, like with SMX, channel sensitivity to cadmium was diminished by co-expression with WT MiRP1 (Ki=0.28±0.07 mM, n=4, Fig. 4G) and retrieved with T8A (Ki=0.15±0.01 mM, n=3, Fig. 4G). HERG channels expressed in CHO cells appear approximately threefold more sensitive to cadmium than the same channels expressed in Xenopus oocytes (29). We attribute this discrepancy to endogenous xMiRPs that have been shown to associate with and modulate the properties of heterologously expressed HERG channels (18). We conclude that block induced by SMX and divalent ions exhibit several similarities, suggesting that both interactions might take place within a common domain of the channel protein. DISCUSSION MiRP1 protects HERG from the arrhythmic effect of external agents Factors that delay the termination of the ventricular action potential, such as decreased potassium flux for instance, can be a primary cause for cardiac disease (12). In this context, block of cardiac IKr channels as a side effect of generic drugs represents a frequent source of acquired arrhythmia (12). Since pharmacological concentrations of SMX (0.1-0.3 mg/ml; refs 34, 35) significantly inhibit HERG channels, SMX therapy should place patients at serious risk of cardiac disease. Clearly, this is not the case and the drug is administered safely, although induced LQT is a wellrecognized side effect of the antibiotic (1, 36–38). We note that HERG channels are not likely to exist in homomeric form in vivo but rather to form heteromeric complexes with the β-subunits MiRP1, MinK, or variably both (7, 39). The presence of MiRP1 in native IKr channels may naturally explain why SMX is generally well tolerated since MiRP1/HERG complexes are significantly less susceptible to SMX (~4-fold) than HERG channels alone. Thus, factors that impair the protective action of MiRP1 may augment channel susceptibility to SMX and increase the risk for arrhythmia. Imperfect N-linked glycosylation is the primary cause for SMX high-affinity block Our data reveal a novel mechanism for drug-induced LQTs: defective N-glycosylation of MiRP1 subunits. The T to A mutation occurs in the first glycosylation site (NFT) located near the N terminus and disrupts the ability of the site to become glycosylated. Moreover, mutations of other amino acids increase susceptibility to SMX only when the mutation is predicted to constitutively suppress glycosylation of the site. Consistent with this notion, modification of fully glycosylated channels by neuraminidase shifts SMX affinity of WT/HERG channels from low to high but has no effect on HERG alone or with T8A. The presence of multiple glycosylation consensus sequences, including one near the N terminus, is a characteristic feature of MiRPs; for example MPS-1, a C. elegans MiRP, has a NIS sequence at position 4 (although carbohydrate modifications can affect various organisms to a different extent; ref 40) and so all other MiRPs do display similar motifs (7, 18, 19). Although the role of glycosylation for MiRP function remains elusive, we note that the presence of two consensus sequences in MiRP1 might account for the normal properties of T8A mutants.
Molecular and structural bases for glycosylation-dependent block Intracellular retention by defective glycosylation can be a cause of genetic disease; for example, mutations of N-glycosylation sites or inhibition of glycosylation with tunicamycin prevent trafficking to the surface for HERG in heterologous expression systems (22, 41). Similar mechanisms, however, cannot be evoked to explain the arrhythmogeneticity of T8A, which appears to be processed and function normally. Structurally, T8A differs from WT by the lack of carbohydrate groups that are attached to the N6 site. Thus, it seems plausible to attribute to some structural effect the primary cause for glycosylation-dependent block (Fig. 5). We note that the unusual reactivity of HERG channels to multiple agents of diverse physical and chemical properties, such as SMX molecules and cadmium ions, might disclose unique structural features. Both agents inhibit HERG channels alone, suggesting that this subunit provides the receptors. SMX and cadmium exert similar effects on the channel by decreasing the repolarizing current and accelerating the rate of deactivation. In both cases, WT lessens channel sensitivity whereas T8A does not. Further, the observation that SMX influences the interaction of cadmium with the channel supports the notion that their binding sites must lie in physical proximity. The picture that emerges from these observations is one in which multiple binding sites might contribute to form a variable receptor site able to interact with diverse blocking agents (the existence of diverse sites is very likely; for instance, block of HERG channels by divalent cations is a complex process probably involving two or more sites; ref 27). The biophysical characteristics of block, its voltage dependence and state dependence, especially the fact that it requires the channel to transit through the open state, support a transmembrane location of the variable receptor, probably along the permeation pathway at the mouth of the pore. We propose that under normal conditions the oligosaccharides attached to the N6 site lie in proximity of the variable receptor so that their absence, like with T8A, might facilitate drug accessibility (Fig. 5). Alternative mechanisms are, however, possible; for instance, the carbohydrates might induce structural alterations of the variable receptor or of its surroundings and will require further investigation. The consequences of the peculiar structural features of MiRP1/HERG channels could be significant. Our findings insinuate the possibility that individuals carrying T8A might be at risk of arrhythmia when treated not only with SMX but also with drugs of diverse therapeutic and structural classes. Should this scenario be confirmed, routine genetic screening should be considered in the clinical management of cardiac disease. ACKNOWLEDGMENTS We thank Dr. Laura Bianchi for critical discussion and helpful comments. We also thank Cinzia Sesti for help with the graphics. HERG-cmyc was a gift from Dr. Thomas MacDonald. This work was supported by an AHA grant (#0235470T) and a UMDNJ foundation award to F. Sesti. Present address for R. Baerga: Pontifical Catholic University of Puerto Rico, 2250 Ave. Las Americas, MARC Honor Program, Suite 643, Ponce, Puerto Rico 00717-0777. REFERENCES 1.
Viskin, S. (1999) Long QT syndromes and torsade de pointes. Lancet 354, 1625–1633
2.
Sesti, F., Abbott, G. W., Wei, J., Murray, K. T., Saksena, S., Schwartz, P. J., Priori, S. G., Roden, D. M., George, A. L., Jr., and Goldstein, S. A. (2000) A common polymorphism associated with antibiotic-induced cardiac arrhythmia. Proc. Natl. Acad. Sci. USA 97, 10613–10618
3.
Cui, J., Kagan, A., Qin, D., Mathew, J., Melman, Y. F., and McDonald, T. V. (2001) Analysis of the cyclic nucleotide binding domain of the HERG potassium channel and interactions with KCNE2. J. Biol. Chem. 276, 17244–17251
4.
Schulze-Bahr, E., Schwarz, M., Hoffmann, S., Wedekind, H., Funke, H., Haverkamp, W., Breithardt, G., Pongs, O., and Isbrandt, D. (2001) A novel long-QT 5 gene mutation in the C-terminus (V109I) is associated with a mild phenotype. J. Mol. Med. 79, 504–509
5.
Napolitano, C., Schwartz, P. J., Brown, A. M., Ronchetti, E., Bianchi, L., Pinnavaia, A., Acquaro, G., and Priori, S. G. (2000) Evidence for a cardiac ion channel mutation underlying drug-induced QT prolongation and life-threatening arrhythmias. J. Cardiovasc. Electrophysiol. 11, 691–696
6.
Kuo, H. C., Cheng, C. F., Clark, R. B., Lin, J. J., Lin, J. L., Hoshijima, M., Nguyen-Tran, V. T., Gu, Y., Ikeda, Y., Chu, P. H., et al. (2001) A defect in the Kv channel-interacting protein 2 (KChIP2) gene leads to a complete loss of I(to) and confers susceptibility to ventricular tachycardia. Cell 107, 801–813
7.
Abbott, G., Sesti, F., Splawski, I., Buck, M., Lehman, M., Timothy, K., Keating, M., and Goldstein, S. (1999) MiRP1 forms IKr potassium channels with HERG and is associated with cardiac arrhythmia. Cell 97, 175–187
8.
Roden, D. (1998) Taking the "idio" out of "idiosyncratic": predicting torsades de pointes. Pacing Clin. Electrophysiol. 21, 1029–1034
9.
Minardo, J., Heger, J., Miles, W., Zipes, D., and Prystowsky, E. (1998) Clinical characteristics of patients with ventricular fibrillation during antiarrhythmic drug therapy. N. Engl. J. Med. 319, 257–262
10. Bennett, P. B., Yazawa, K., Makita, N., and George, A. L., Jr. (1995) Molecular mechanism for an inherited cardiac arrhythmia. Nature 376, 683–685 11. Curran, M., Spalwski, I., Timothy, K., Vincent, G., Green, E., and Keating, M. (1995) A molecular bases for cardiac arrhythmia: HERG mutations cause long QT syndrome. Cell 80, 795–803 12. Keating, M., and Sanguinetti, M. (2001) Molecular and cellular mechanisms of cardiac arrhythmia. Cell 104, 569–580 13. Isbrandt, D., Friederich, P., Solth, A., Haverkamp, W., Ebneth, A., Borggrefe, M., Funke, H., Sauter, K., Breithardt, G., Pongs, O., et al. (2002) Identification and functional characterization of a novel KCNE2 (MiRP1) mutation that alters HERG channel kinetics. J. Mol. Med. 80, 524–532
14. Larsen, L. A., Andersen, P. S., Kanters, J., Svendsen, I. H., Jacobsen, J. R., Vuust, J., Wettrell, G., Tranebjaerg, L., Bathen, J., and Christiansen, M. (2001) Screening for mutations and polymorphisms in the genes KCNH2 and KCNE2 encoding the cardiac HERG/MiRP1 ion channel: implications for acquired and congenital long Q-T syndrome. Clin. Chem. 47, 1390–1395 15. Mazhari, R., Greenstein, J. L., Winslow, R. L., Marban, E., and Nuss, H. B. (2001) Molecular interactions between two long-QT syndrome gene products, HERG and KCNE2, rationalized by in vitro and in silico analysis. Circ. Res. 89, 33–38 16. Burton, D., Song, C., Fishbein, I., Hazelwood, S., Li, Q., DeFelice, S., Connolly, J., Perlstein, I., Coulter, D., and Levy, R. (2003) The incorporation of an ion channel gene mutation associated with the long QT syndrome (Q9E-hMiRP1) in a plasmid vector for sitespecific arrhythmia gene therapy: in vitro and in vivo feasibility studies. Hum. Gene Ther. 14, 907–922 17. Abbott, G., and Goldstein, S. (1998) A superfamily of small potassium channel subunits: form and function of the MinK-related peptides (MiRPs). Q. Rev. Biophys. 31, 357–398 18. Anantharam, A., Lewis, A., Panaghie, G., Gordon, E., McCrossan, Z. A., Lerner, D. J., and Abbott, G. W. (2003) RNA interference reveals that endogenous Xenopus minK-related peptides govern mammalian K+ channel function in oocyte expression studies. J. Biol. Chem. 278, 11739–11745 19. Bianchi, L., Kwok, S. M., Driscoll, M., and Sesti, F. (2003) A potassium channel-MiRP complex controls neurosensory function in Caenorhabditis elegans. J. Biol. Chem. 278, 12415–12424 20. Marshall, R. D. (1972) Glycoproteins. Annu. Rev. Biochem. 41, 673–702 21. Sumikawa, K., Parker, I., and Miledi, R. (1988) Effect of tunicamycin on the expression of functional brain neurotransmitter receptors and voltage-operated channels in Xenopus oocytes. Brain Res. 464, 191–199 22. Petrecca, K., Atanasiu, R., Akhavan, A., and Shrier, A. (1999) N-linked glycosylation sites determine HERG channel surface membrane expression. J. Physiol. 515, 41–48 23. Freeman, L. C., Lippold, J. J., and Mitchell, K. E. (2000) Glycosylation influences gating and pH sensitivity of I(sK). J. Membr. Biol. 177, 65–79 24. Gehle, V. M., Walcott, E. C., Nishizaki, T., and Sumikawa, K. (1997) N-glycosylation at the conserved sites ensures the expression of properly folded functional ACh receptors. Brain Res. Mol. Brain Res. 45, 219–229 25. Cassidy, J. T., Jourdian, G. W., and Roseman, S. (1965) The sialic acids. VI. Purification and properties of sialidase from Clostridium perfringens. J. Biol. Chem. 240, 3501–3506
26. McDonagh, J. C., and Nathan, R. D. (1990) Sialic acid and the surface charge of delayed rectifier potassium channels. J. Mol. Cell. Cardiol. 22, 1305–1316 27. Ho, W. K., Kim, I., Lee, C. O., Youm, J. B., Lee, S. H., and Earm, Y. E. (1999) Blockade of HERG channels expressed in Xenopus laevis oocytes by external divalent cations. Biophys. J. 76, 1959–1971 28. Ho, W. K., Kim, I., Lee, C. O., and Earm, Y. E. (1998) Voltage-dependent blockade of HERG channels expressed in Xenopus oocytes by external Ca2+ and Mg2+. J. Physiol. 507, 631–638 29. Johnson, J. P., Jr., Balser, J. R., and Bennett, P. B. (1999) Enhancement of HERG K(+) currents by Cd(2+) destabilization of the inactivated state. Biophys. J. 77, 2534–2541 30. Shimizu, H., Toyoshima, C., and Oiki, S. (2003) Interaction between tetraethylammonium and permeant cations at the inactivation gate of the HERG potassium channel. Jpn. J. Physiol. 53, 25–34 31. Moreno, I., Caballero, R., Gonzalez, T., Arias, C., Valenzuela, C., Iriepa, I., Galvez, E., Tamargo, J., and Delpon, E. (2003) Effects of irbesartan on cloned potassium channels involved in human cardiac repolarization. J. Pharmacol. Exp. Ther. 304, 862–873 32. Zitron, E., Karle, C. A., Wendt-Nordahl, G., Kathofer, S., Zhang, W., Thomas, D., Weretka, S., and Kiehn, J. (2002) Bertosamil blocks HERG potassium channels in their open and inactivated states. Br. J. Pharmacol. 137, 221–228 33. Gurrola, G. B., Rosati, B., Rocchetti, M., Pimienta, G., Zaza, A., Arcangeli, A., Olivotto, M., Possani, L. D., and Wanke, E. (1999) A toxin to nervous, cardiac, and endocrine ERG K+ channels isolated from Centruroides noxius scorpion venom. FASEB J. 13, 953–962 34. Joos, B., Blaser, J., Opravil, M., Chave, J., and Luthy, R. (1995) Monitoring of cotrimoxazole concentrations in serum during treatment of pneumocystis carinii pneumonia. Antimicrob. Agents Chemother. 39, 2661–2666 35. Blaser, J., Joos, B., Opravil, M., and Luthy, R. (1993) Variability of serum concentrations of trimethoprim and sulfamethoxazole during high dose therapy. Infection 21, 206–209 36. Lopez, J., Harold, J., Rosenthal, M., Oseran, D., Schapira, J., and Peter, T. (1987) QT prolongation and torsades de pointes after administration of trimethoprim-sulfamethoxazole. Am. J. Cardiol. 59, 376–377 37. Wiener, I., Rubin, D., Martinez, E., Postman, J., and Herman, M. (1981) QT prolongation and paroxysmal ventricular tachycardia occurring during fever following trimethoprimsulfamethoxazole administration. Mt. Sinai J. Med. 48, 53–55 38. Ackerman, M. (1998) The long QT syndrome: ion channel diseases of the heart. Mayo Clin. Proc. 73, 250–269
39. McDonald, T., Yu, Z., Ming, Z., Palma, E., Meyers, M., Wang, K., Goldstein, S., and Fishman, G. (1997) A MinK-HERG complex regulates the cardiac potassium current IKr. Nature 388, 289–292 40. Herman, T., and Horvitz, H. (1999) Three proteins involved in Caenorhabditis elegans vulval invagination are similar to components of a glycosylation pathway. Proc. Natl. Acad. Sci. USA 96, 974–979 41. Gong, Q., Anderson, C. L., January, C. T., and Zhou, Z. (2002) Role of glycosylation in cell surface expression and stability of HERG potassium channels. Am. J. Physiol. Heart Circ. Physiol. 283, H77–H84 Received June 17, 2003; accepted August 11, 2003
Fig. 1
Figure 1. HERG channels alone exhibit high susceptibility to SMX. A) Families of whole-cell HERG currents in the absence (left), presence of 1 mg/ml SMX (center), and after washout (right). Currents were evoked by the voltage protocol illustrated in the inset (0.5 s conditioning pulses from –80 mV to +40 mV in 20 mV intervals followed by a 1 s repolarizing pulse at –40 mV). Holding voltage was –80 mV for all experiments and interpulse interval 1 s if not otherwise stated. B) Voltage dependence of deactivation kinetics in the absence or presence of 1 mg/ml SMX was −t / τ −t / τ calculated by fitting deactivating currents with a 2 exponential function, I o + I f e f + I s e s . Currents were elicited by the voltage protocol shown in the inset (a 0.2 s conditioning pulse to +20 mV followed by 0.5 s test pulses from –120 to +40 mV in 20 mV steps). Data are from groups of 5-6 cells. C) Voltage dependence of block. Currents in the presence or in the absence of 1 mg/ml SMX were elicited by the voltage protocol shown in B. Fractional peak currents are shown in the upper panel, steady-state currents in the lower panel. Data are from groups of 5 cells. D) Conditioning pulse dependence of block. Steady-state fractional current at –40 mV elicited by the indicated preconditioning voltage steps (voltage protocol shown in A). Data are from groups of 7-8 cells. E) Current-dose relationships for HERG alone (triangles), WT (circles), and T8A (squares) in the presence of the indicated amounts of SMX in the bath. Currents were elicited by the voltage protocol shown in A and current were measured at the end of the repolarizing pulse (arrow). Data fit to the Hill function Kin/(Kin + [SMX]n) with Ki = 0.34 ± 0.03 mg/ml and n = 1.43 ± 0.14 for HERG, Ki = 1.2 ± 0.06 mg/ml and n = 1.4 ± 0.09 for WT, and Ki = 0.31 ± 0.05 mg/ml and n = 1.7 ± 0.05 for T8A. Data are from groups of 5-6 cells.
Fig. 2
Figure 2. Defective N-glycosylation of T8A mutants. A) SDS-PAGE (15%) and Western blot visualization of total cell lysates performed with anti-HA antibody of untransfected (1) or transiently transfected COS-7 cells with pCI-neo (2), WT-HA MiRP1, and T8A-HA MiRP1 (schematic). Three WT bands correspond to the di-, mono-, and unglycosylated form. In contrast, T8A is detected only in partially but not fully glycosylated states. B) Immunolocalization of WT and T8A at the cell surface. CHO cells were transiently transfected with cDNA encoding WT and T8A modified with a HA epitope in the N terminus (schematic). Staining with primary and secondary antibodies was performed without membrane permeabilization. No staining was observed with untransfected cells. C) Protein expression of WT and T8A subunits in CHO cells. Immunostained cells were counted over populations of 200 cells in average. Each data are an average of 10 or more groups. D) Co-immunoprecipitations of T8A/HERG channels. A cmyc epitope was fused to the C terminus of HERG for immunoprecipitation (IP). Transiently transfected COS-7 cells with T8A-HA alone or with HERG-cmyc were immunoprecipitated with anti-cmyc antibody. Left and center: visualization of IP with anti HA antibody and Western blot (15% SDS) of T8A alone and T8A+HERG. Right: visualization with anti-cmyc (10% SDS). Scale bars = 15, 21, and 50 kDa (15% gels) and 11, 40, and 115 kDa (10% gels).
Fig. 3
Figure 3. Defective N-glycosylation of MiRP1 is associated with increased affinity to SMX. A) Inhibition constants (Ki) calculated by fitting dose-response curves of the indicated MiRP1 mutants to the Hill function (Fig. 1E). The Hill coefficient ranged from 1.3 ± 0.17 to 1.7 ± 0.25. Mutations of N6 or T8 (boxes) disrupt glycosylation of the site. Data are calculated from groups of 4-7 cells. Dotted lines indicate inhibition constants of HERG and WT. B-D) Dose-response curves calculated as essentially described in Fig. 1E from cells expressing HERG channels alone, with WT or T8A treated with the N-glycosylation inhibitor neuraminidase. Cells were preincubated with the enzyme (0.3 U/ml) at 37°C for 30 min before the experiment and were used for not longer than 2 h. Data fitted to the Hill function with: Ki = 0.3 ± 0.06 mg/ml and n = 1.6 ± 0.13 for HERG alone, Ki = 0.39 ± 0.08 mg/ml and n = 1.6 ± 0.14 for MiRP1/HERG and Ki = 0.29 ± 0.11 mg/ml and n = 1.6 ± 0.14 for T8A. Dotted lines correspond to the fits of dose-response relationship in the absence of neuraminidase. Data are from groups of 3-4 cells.
Fig. 4
Figure 4. SMX and cadmium block are similar. A) Whole HERG currents in the absence and in the presence of 0.3 mM cadmium in the bath. Currents were evoked by the voltage protocol shown in the inset. A) Voltage dependence of deactivation kinetics (fast and slow components) in the absence and in the presence of cadmium. Deactivation currents were fitted as essentially described in Fig. 1. Data from groups of 3-5 cells. C-D) Voltage-dependence of cadmium block. Fractional currents were calculated by using peak (C) or steady-state (D) values. Data are from groups of 3-5 cells. E-F) HERG dose-response curves for cadmium (E) or with the addition of 0.5 mg/ml SMX in the test solutions (F). Fits to the Hill function gave Ki = 0.11 ± 0.01 and Ki = 0.61 ± 0.11 without or with SMX, respectively. The Hill coefficient ranged from 1 to 1.2. Data are from groups of 3-5 cells. The theoretical dotted line in F was calculated assuming that SMX and cadmium bind to the same site so that 0.5 mg/ml SMX would be the equivalent of adding 0.25 mM cadmium to the test solutions. Current normalization would give an apparent Ki = 0.41. The theoretical line was constructed assuming a Hill coefficient, n = 1. G) Dose-response curves for cadmium, for WT/HERG and T8A/HERG channels. Fit of the data yielded Ki = 0.3 ± 0.05 mM and Ki = 0.14 ± 0.01 mM, respectively; n = 1.0 in both cases. Data are from groups of 3 cells.
Fig. 5
Figure 5. A proposed structural model for glycosylation-dependent block. Four HERG subunits (peach), form a single, central ion-conduction pathway. For the sake of the representation, there are four MiRP1 subunits (green) facing the conduction pathway. The variable receptor is at the mouth of the pore and contains distinct binding sites (not indicated in the drawing). Under normal conditions, the carbohydrates attached to MiRP1 (green branches) shield the variable receptor and thus impair SMX (blue balls) binding. Conversely, in channels formed with T8A mutants, SMX accessibility to the receptor is facilitated by the absence of the oligosaccharide groups.
