Kir2.6 Regulates the Surface Expression of Kir2.x Inward Rectifier Potassium Channels*

Precise trafficking, localization, and activity of inward rectifier potassium Kir2 channels are important for shaping the electrical response of skeletal muscle. However, how coordinated trafficking occurs to target sites remains unclear. Kir2 channels are tetrameric assemblies of Kir2.x subunits. By immunocytochemistry we show that endogenous Kir2.1 and Kir2.2 are localized at the plasma membrane and T-tubules in rodent skeletal muscle. Recently, a new subunit, Kir2.6, present in human skeletal muscle, was identified as a gene in which mutations confer susceptibility to thyrotoxic hypokalemic periodic paralysis. Here we characterize the trafficking and interaction of wild type Kir2.6 with other Kir2.x in COS-1 cells and skeletal muscle in vivo. Immunocytochemical and electrophysiological data demonstrate that Kir2.6 is largely retained in the endoplasmic reticulum, despite high sequence identity with Kir2.2 and conserved endoplasmic reticulum and Golgi trafficking motifs shared with Kir2.1 and Kir2.2. We identify amino acids responsible for the trafficking differences of Kir2.6. Significantly, we show that Kir2.6 subunits can coassemble with Kir2.1 and Kir2.2 in vitro and in vivo. Notably, this interaction limits the surface expression of both Kir2.1 and Kir2.2. We provide evidence that Kir2.6 functions as a dominant negative, in which incorporation of Kir2.6 as a subunit in a Kir2 channel heterotetramer reduces the abundance of Kir2 channels on the plasma membrane.

Inward rectifier potassium (Kir2) channels are key skeletal muscle components involved in determination of muscle resting potential, regulation of electrical excitability, repolarization of the action potential, and clearance of K ϩ from the T-tubule 2 system (1)(2)(3)(4). Electrophysiological studies in skeletal muscle and immunocytochemical studies in cardiac muscle indicate that Kir2 channels are localized at the plasma membrane and T-tubules (Refs. 1 and 5-15; reviewed in Refs. 3 and 16). The abundance of Kir2 channels is critical to muscle function, and their importance is highlighted when the channels are mutated or absent. Andersen-Tawil syndrome is characterized by periodic paralysis of skeletal muscle, cardiac arrhythmia, and developmental dysgenesis, resulting from loss-of-function mutations in KCNJ2, the gene that encodes Kir2.1 potassium channels (17,18); gain-of-function mutations cause some forms of short QT syndrome (19). Targeted deletion of Kir2.1 in mice also causes defective cerebral vasodilation, and deletion of either Kir2.1 or Kir2.2 causes a loss or decrease in cardiac I K1 currents (20,21).
Recently, a search for genes involved in thyrotoxic periodic paralysis (TPP) revealed KCNJ18, which encodes a novel subtype of inward rectifier potassium channel subunit, Kir2.6 (22). Kir2.6 is expressed primarily in skeletal muscle and shares Ͼ98% identity with Kir2.2 (22). Mutations in human Kir2.6 confer susceptibility to TPP, and it has been shown that these disease-associated mutations contribute to atypical current signatures and altered cell excitability (22). Expression studies in heterologous cells demonstrate that Kir2.6 subunits are able to form inwardly rectifying channels and that disease-associated mutations include truncated subunits that do not form functional channels, as well as gain-of-function mutations that increase electrical activity because of misregulation by phosphorylation (22).
In addition to their electrophysiological properties, ion channels contribute to electrical events by their abundance on the plasma membrane. Targeting ion channels to subcellular sites and the plasma membrane is important for shaping the electrical response of muscle. Surface expression levels and channel activity are crucial for determining muscle excitability. However, trafficking of Kir2.6 to the plasma membrane has not yet been explored. Moreover, the role of wild type Kir2.6 in normal muscle physiology has not been established.
Several trafficking motifs in Kir2 subunits have been identified and are well conserved among the inward rectifier subunits that are found in human skeletal muscle (Kir2.1, Kir2.2, Kir2.3, and Kir2.6). Anterograde trafficking signals in Kir2.1 channels promote surface expression by enhancing ER and Golgi export early in the secretory pathway (23)(24)(25)(26). In particular, a di-acidic sequence motif in the C terminus of Kir2.1 facilitates traffic from the ER to the Golgi (23,24). Two additional regions within Kir2.1, a positively charged motif in the N terminus and a conserved YXX⌽ sequence on the membrane-proximal region of the C terminus, have been associated with export from the Golgi to the plasma membrane (25,26). The Kir2.1, Kir2.2, and Kir2.3 C termini also contain a PDZ-binding motif that permits binding to scaffolding and clustering MAGUK proteins that may direct channel localization and abundance (7,10,11,(27)(28)(29)(30)(31)(32).
Here we describe the interaction and trafficking of wild type Kir2.1, Kir2.2, and Kir2.6 in mouse skeletal muscle and in COS-1 cells. Although Kir2. 6
Transfection and Expression in COS-1-COS-1 cells were grown and maintained (10% CO 2 , 37°C) in DMEM-supplemented with 10% fetal calf serum. The cells were cotransfected with cDNAs encoding tagged Kir2 subunits and pECFP-Golgi constructs utilizing FuGENE TM 6 (Roche Applied Science), using 0.4 g of total DNA/15-mm well (0.2 g each when cotransfected with two DNAs; empty vector was not used). One day post-transfection, the cells were fixed at room temperature with 4% paraformaldehyde, 4% sucrose in PBS for 8 min and then blocked with 3% BSA, 1% goat serum, 1% donkey serum in PBS. The cells were incubated with rat anti-Myc (1:1000) or mouse anti-HA (1:500) primary antibodies for 1 h at room temperature, then washed three times with PBS, and incubated with anti-rat or anti-mouse Cy3 (1:300) secondary antibodies for detection of surface expression. To label intracellular epitopes, the cells then were washed, permeabilized with blocking buffer containing 0.15% Triton X-100, and incubated with rat anti-HA (1:500) or mouse anti-Myc (1:500) primary antibodies, washed, and incubated with anti-rat or anti-mouse Alexa Fluor 647 (1:300) secondary antibodies. In experiments that used pECFP-Golgi, the cells also were labeled with rabbit anti-GFP-conjugated Alexa Fluor 488 (1:1000).
COS-1 Cell Surface and Internal Measurements-The images were captured using widefield fluorescence microscopy with a Zeiss Axiovert 200 microscope with either Plan Apochromat 100ϫ/1.4 or Plan Apochromat 20ϫ/0.8 objectives. The images were taken with identical exposure times for all measurements of surface and internal labeling within an experiment. For each transfection, surface and internal fluorescence intensity and cell area were measured from 100 cells using AxioVision 4.6 analysis software, and the data were expressed as the average fluorescent intensity. Statistically significant differences were assessed using the unpaired Student's t test. The data are given as the averages Ϯ S.E. expressed as relative surface/internal expression level.
Electrophysiology-COS-1 cells were cotransfected with cDNAs encoding 0.5 g of Myc-tagged Kir2.x and 0.1 g of pEGFP-C1 utilizing FuGENE TM 6 (Roche Applied Science). One day post-transfection, the cell currents were recorded using standard methods for whole cell voltage clamp studies. The currents were acquired using a Dagan 3900/3911A patch clamp, low pass-filtered at 5 kHz, and sampled at 20 kHz with an ITC-16 digitizer (Instrutech). Data acquisition was con-trolled with Pulse (HEKA) software. Pipettes had resistances of 1-3 megohms. After establishing the whole cell configuration, the cell capacitance was cancelled, and series resistance was compensated (Ͼ50%). The currents were elicited by 100-ms pulses applied in 20-mV increments to potentials ranging from Ϫ100 to 60 mV from a holding potential of 0 mV. The pipette solution contained 110 mM potassium aspartate, 20 mM KCl, 10 mM EGTA, 1 mM MgCl 2 , 5 mM glucose, 10 mM Hepes, pH 7.3 (adjusted with KOH). The extracellular bath solution contained 115 mM NaCl, 30 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 5 mM glucose, 10 mM Hepes, pH 7.4 (adjusted with NaOH). Recordings were made at room temperature (22-25°C) within 23-32 h of transfection. For comparison of current magnitudes for different Kir2 channels, current amplitude at Ϫ100 mV was averaged in the interval from 5-15 ms after the onset of the voltage pulse, and leakage current was subtracted using the scaled current for pulses to 60 mV. The data were analyzed and displayed with Pulse and SigmaPlot software.

Kir2.6 Is Retained in the ER when Expressed in COS-1 Cells-
To investigate the trafficking of Kir2.1, Kir2.2, and Kir2.6 to the plasma membrane, tagged channel constructs were transfected into COS-1 cells, and their surface and internal expression were evaluated using immunocytochemistry. Surface expression was evaluated by labeling fixed cells 24 h after transfection; internal expression was assessed in the same cells by labeling subsequent to a permeabilization step. Expressing data as a surfaceto-internal ratio for each cell controlled for variations in expression levels in individual cells or between different constructs. A 24-h post-transfection time point was chosen to allow adequate time for channel biosynthesis and processing without overexpression effects that can saturate the trafficking machinery.
Robust plasma membrane fluorescence was consistently observed for Kir2.1 and Kir2.2, together with a low level of internal fluorescence that colocalized with the Golgi markers pECFP-Golgi or GM130, suggesting that Kir2.1 and Kir2.2 are trafficked through the Golgi, then highly expressed on the plasma membrane of COS-1 cells (Fig. 1A, top and middle rows). Conversely, Kir2.6 channels displayed very low surface fluorescence and were predominantly intracellular in the perinuclear region colocalized with the ER marker PDI (Fig. 1A, bottom row). Note that the ER localization of Kir2.6 was not affected by the epitope tag on the subunit, and identical results were obtained for channels tagged with Myc on the extracellular M1-to-P linker ( Fig. 1), with GFP on the intracellular N terminus (see Fig. 9), or with V5 on the N terminus (not shown). Quantitative analysis of channel expression for a large number of cells revealed that the relative ratio of surface/internal fluorescence was highest for Kir2.1, was decreased to ϳ40% for Kir2.2 compared with Kir2.1, and was only ϳ20% for Kir2.6 compared with Kir2.2. These data suggest that in COS-1 cells, Kir2.6 channels are mainly retained in the ER (Fig. 1B).
To ascertain whether the ER retention of Kir2.6 was dependent on the cellular context, we compared its trafficking in the HL-1 mouse cardiac muscle cell line. Kir2.6 was similarly retained in the ER and colocalized with PDI, whereas Kir2.1 and Kir2.2 were trafficked to the plasma membrane as in COS-1 cells, a monkey kidney cell line (data not shown).
Kir2.6 Forms Functional Homotetrameric Channels in COS-1 Cells-Although Kir2.6 traffics poorly to the cell surface, immunofluorescence data suggested that a small fraction of the protein was present on the plasma membrane (Fig. 1B). We used whole cell voltage clamp recording in COS-1 cells to assess the ability of Kir2.6 subunits to form functional homotetrameric channels. The hallmark of Kir2 channels is their strong inward rectification, which in muscle allows them to carry currents near the resting potential but not at positive potentials. Both Kir2.2 and Kir2.6 exhibit strong inward rectification, displaying larger inward currents at potentials negative to the potassium equilibrium potential and small outward currents at more positive voltages, in agreement with previous findings ( Fig. 2A) (22,46). Because the Kir2.6 surface/internal ratio by immunofluorescence was significantly lower than Kir2.1 and Kir2.2 ratios, we expected to find differences in current magnitude between Kir2.2 and Kir2.6, and indeed the whole cell current magnitude for Kir2.6 was ϳ15% compared with Kir2.2 (Fig. 2B). Because single channel conductance and open probability are similar for Kir2.6 and Kir2.2 (22), this reduced Kir2.6 current most likely corresponds to a decrease in the number of channels on the plasma membrane.
Kir2.2 Proline 156 Is a Critical Determinant of Channel Trafficking to the Plasma Membrane-Because of the large differences in surface expression observed between Kir2.6 and Kir2.2, despite the conservation of known forward trafficking signals, we compared channel sequences to identify potentially new amino acid regions that might be important for trafficking. Site-directed mutagenesis was used to alter each of the nine amino acids that varies between Kir2.2 and Kir2.6 by replacing sites in Kir2.6 with the corresponding Kir2.2 counterparts. These mutations targeted sites in the N-terminal region (three sites: Kir2.6-L15S, Kir2.6-Q39R, and Kir2.6-H40R), the M1 helix (Kir2.6-V100I), between the M1 and the P loop (Kir2.6-H118R), between the P-loop and the M2 helix (Kir2.6-L156P), and three sites in the C terminus (Kir2.6-H192Q located next to the PIP 2 binding domain, Kir2.6-V249I, and Kir2.6-G430E positioned adjacent to the PDZ binding domain). The constructs were transfected into COS-1 cells and labeled for internal and surface expression as described above.
Immunocytochemical analysis revealed that Pro-156 is essential for efficient channel trafficking to the plasma membrane (Fig. 3A, top row). Kir2.6-L156P displays a surface-tointernal ratio significantly higher than Kir2.6 and more than half that of Kir2.2. The reverse Kir2.2 mutation, Kir2.2-P156L, is sufficient to retain the channel in the ER (Fig. 3). Further, we found that a combination of Pro-156 and Gln-192 is able to confer full plasma membrane expression, and the high expression ratio of Kir2.6-L156P/H192Q was not significantly different from Kir2.2 (Fig. 3). In contrast, mutations at other sites were unable to overcome the ER retention of Kir2.6 ( Fig. 3 and data not shown).
Direct measurement of surface and internal expression revealed that Kir2.6 subunits associate with Kir2.1 and Kir2.2 subunits and result in their partial retention in the ER (Fig. 5A).
To illustrate the effect of Kir2.6 on trafficking of other Kir2.x, each of these images shows a pair of cells, one of which expresses Kir2.6 (arrow), and both of which express either Kir2.1 or Kir2.2. The cells expressing Kir2.6 (arrow) show an increase in the intracellular abundance of Kir2.1 or Kir2.2 that colocalizes with Kir2.6 and a corresponding decrease in surface expression of Kir2.1 or Kir2.2 compared with the neighboring cell without Kir2.6. Quantitative analysis showed that coexpression of Kir2.6 with Kir2.1 reduced Kir2.1 surface/internal expression ratio by ϳ30% (Fig. 5B). Similarly, coexpression of Kir2.6 with Kir2.2 reduced Kir2.2 surface/internal ratio by ϳ50% (Fig. 5C). Total protein levels of the subunits, evaluated by immunoblot, were comparable (supplemental Fig. S1). Hence, Kir2.6 impedes surface expression of Kir2.1 and Kir2.2 when coexpressed in COS-1 cells, suggesting that Kir2.6 subunits are dominant in their ability to retain heterotetrameric channels in the ER.
Endogenous Kir2.1 and Kir2.2 Localize at T-tubules and the Plasma Membrane in Skeletal Muscle-Targeting of ion channels to subcellular sites is important for shaping the response of muscle. We evaluated the localization of endogenous Kir2.1 and Kir2.2 channels in rat tibialis cranialis skeletal muscle using channel-specific antibodies. Kir2.6 is present in humans and other primates, but the gene encoding Kir2.6 has not been identified in the data bank for rodents or other nonprimate species (22), so endogenous Kir2.6 was not assessed. Kir2.1 labeling was abundant on the cell surface and as transverse striations regularly spaced at intervals of ϳ2.5 m in moderately stretched fibers (Fig. 6A). Optical confocal sectioning showed that Kir2.1 labeling was high on striations near the surface membrane and was lower toward the center of muscle cells. At high magnification in moderately stretched fibers, the transverse labeling could be resolved as double rows that colocalized with rows of dihydropyridine receptor ␣2 subunit (DHPR), a marker of T-tubules (Fig. 6A, insets, arrowheads). The double rows of Kir2.1 labeling bracketing the Z line (identified with anti-␣actinin; Fig. 6A, insets) were continuous with DHPR and with the same spacing and were slightly narrower than double rows   MARCH 18, 2011 • VOLUME 286 • NUMBER 11 of ryanodine receptor that mark the sarcoplasmic reticulum at the triad junction (data not shown). In mammalian skeletal muscle, there are two closely spaced transverse tubules per sarcomere, and the striated Kir2.1 labeling pattern is consistent with localization at T-tubules (47). Altogether the data suggest that endogenous Kir2.1 is abundant on the plasma membrane and on T-tubules near the periphery of muscle fibers and with lower abundance on T-tubules in the center of fibers.

Trafficking of Kir2.6 Potassium Channel
Endogenous Kir2.2 showed a distribution that is distinct but overlapping with Kir2.1. Kir2.2 was present in transverse striations with a distribution that extended into the center of muscle fibers (Fig. 6B). Colabeling with antibodies to markers indicated   that these striations corresponded to T-tubules (47); Kir2.2 labeling resolved as double transverse rows that colocalize with rows of DHPR (Fig. 6B, insets, arrowheads), that bracket the Z-line identified with anti-␣-actinin (Fig. 6B), and that are slightly narrower than double rows of ryanodine receptor (not shown). Interestingly, the distribution of both Kir2.1 and Kir2.2 was punctate along the T-tubules and appeared adjacent to and with some overlap with DHPR, suggesting that the Kir2 channels and DHPR are spatially segregated in discrete clusters on the T-tubules. In addition, Kir2.2 was abundant at the neuromuscular junction and was present in a punctate Golgi distribution both adjacent to the neuromuscular junction and throughout the muscle fibers (data not shown; see also Ref. 7).

Kir2.6 Is Predominantly Retained in the ER in Mouse Skeletal
Muscle-To study Kir trafficking in muscle in vivo, we electroporated tagged channel cDNAs into mouse tibialis cranialis skeletal muscle, and 7 days later evaluated channel expression and localization by immunocytochemistry. We observed that channel expression was spatially restricted in muscle fibers to regions of the fiber near electroporated nuclei. Kir2.1 was particularly abundant on the plasma membrane and peripheral T-tubules (supplemental Fig. S2) and showed a clear punctate intracellular Golgi distribution that colabeled with the Golgi marker GM130 (45) (Fig. 7). Kir2.1 was in low abundance in the ER, which is positioned as a perinuclear region closely associated with each of the nuclei in skeletal muscle (43)(44)(45). Thus, the trafficking of tagged Kir2.1 (Fig. 7 and supplemental Fig. S2) closely matched the localization of endogenous Kir2.1 (Fig. 6), and its relative abundance suggests that Kir2.1 is rapidly trafficked out of the ER to the Golgi, peripheral T-tubules, and plasma membrane.
The expression pattern of electroporated Kir2.2 (Fig. 7) gave high expression in Golgi and low expression in ER, which corresponds with labeling of endogenous Kir2.2 channels (Fig. 6B), indicating that Kir2.2 traffics efficiently from the ER through the Golgi (Fig. 7, middle row). Kir2.2 also was present in a striated pattern consistent with localization in T-tubules (Fig. 7).
In contrast, we observed a strikingly different localization for Kir2.6, which was restricted to the ER, with no detectable expression in Golgi or plasma membrane. Colocalization with the ER marker PDI confirmed retention of Kir2.6 to the ER (43, 44) (Fig. 7, bottom row).
Kir2.6-L156P Is Trafficked to the Golgi in Mouse Skeletal Muscle-In vivo electroporation was utilized to investigate trafficking and localization of mutant Kir2 channels in skeletal muscle. We found that forward trafficking of the channels could be most clearly assessed by comparison of the abundance of Kir2.x subunits in Golgi relative to the ER because of the concentrated and discrete nature of those cellular compartments and because of their close proximity in individual confocal sections. Of all of the Kir2.6 single amino acid mutants eval-uated, only Kir2.6-L156P resulted in partial release from the ER and trafficking to the Golgi, indicating that among the amino acids that differ between Kir2.2 and Kir2.6, Pro-156 is the most critical determinant for forward trafficking (Fig. 8, top row). As in COS-1 cells, forward trafficking of Kir2.6-L156P was less complete than Kir2.2, and the double mutation Kir2.6-L156P/ H192Q was required to promote maximal transfer from the ER to the Golgi (Fig. 8, bottom row).
Kir2.1 and Kir2.2 Are Partially Retained in the ER when Coexpressed with Kir2.6 in Mouse Skeletal Muscle-The affect of Kir2.6 on the subcellular localization of Kir2.1 and Kir2.2 trafficking was further studied in vivo by coelectroporation of pairs of cDNAs into mouse skeletal muscle. By itself Kir2.1 was rarely observed in the ER. However, when Kir2.1 and Kir2.6 were coexpressed, Kir2.1 was present in both the ER and Golgi in most coelectroporated nuclei, and the fraction of Kir2.1 that was retained in the ER colocalized with Kir2.6 ( Fig. 9). Affects of Kir2.6 on Kir2.2 trafficking were even more severe. Coexpression of Kir2.2 and Kir2.6 resulted in complete retention of Kir2.2 in the perinuclear ER region in the vast majority of the coelectroporated nuclei examined (Fig. 9). In contrast to the strong effect of Kir2.6 on the localization of Kir2.1 and Kir2.2, these subunits had only a weak influence on Kir2.6 localization. The ER localization of Kir2.6 was not affected by coexpression with Kir2.2; Kir2.6 displayed only minor trafficking to the Golgi in some nuclei when coexpressed with Kir2.1 (Fig. 9).
These results were extended by measuring the effect of coelectroporated Kir2.6 on the surface/internal ratio of expressed Kir2.1 or Kir2.2 in muscle (Fig. 10, A-D). Nuclei expressing Kir2.6 (Fig. 10, A and C, arrows) showed increased internal expression and/or decreased surface expression of coexpressed Kir2.1 or Kir2.2 compared with nearby nuclei in which Kir2.6 expression was low or absent (Fig. 10, A and C,  arrowheads). Quantitative analysis demonstrated that surface/ internal ratios were reduced from muscle coelectroporated with Kir2.6 and Kir2.1 or Kir2.2 compared with muscle electroporated with Kir2.1 or Kir2.2 alone (Fig. 10, B and D).
Endogenous Kir2.2 Is Partially Retained in the ER by Kir2.6 Expression-Strong support for a regulatory role of Kir2.6 in Kir2.x channel trafficking was further obtained by examining the effect of Kir2.6 expression on the localization of endogenous channels in mouse muscle. In control skeletal muscle fibers, endogenous Kir2.2 was not observed in the ER, but after in vivo electroporation of Kir2.6, nearly all of the nuclei expressing Kir2.6 showed retention of endogenous Kir2.2 in the ER surrounding those nuclei, which colocalized with Kir2.6 ( Fig.  10, E and F). Nearby nuclei not expressing Kir2.6 were unaffected (Fig. 10E, arrowheads), and muscle fibers electroporated with unrelated constructs (myristoylated-GFP or GFP-Ca 2ϩ -ATPase) were unaffected (data not shown). In contrast to the effects on endogenous Kir2.2, we did not observe ER retention of endogenous Kir2.1 (data not shown), perhaps because of a smaller effect on Kir2. 1 (Figs. 5,9,and 10) or the lower sensitivitiy of the Kir2.1 antibody to endogenous mouse Kir2.1. Taken together, these data indicate that Kir2.6 has a dominant negative affect on the trafficking of other Kir2.x channel subunits in skeletal muscle, with the strongest effect on Kir2.2 and a more moderate effect on Kir2.1.

DISCUSSION
In this study we have investigated the trafficking, cellular localization, and subunit interactions of a recently identified human inward rectifier potassium channel Kir2.6. This subunit was discovered as a gene that when mutated confers susceptibility to TPP, a condition characterized by muscle weakness or paralysis accompanied by hypokalemia and thyrotoxicosis (22). Here we characterize wild type Kir2.6 to understand how the typical native Kir2.6 contributes to muscle cell biology. Surprisingly, we find that despite its high sequence similarity to other Kir2.x subunits (Ͼ98% identity with Kir2.2), Kir2.6 is very poorly trafficked to the surface membrane and instead primarily resides in the endoplasmic reticulum. Whole cell currents of Kir2.6 by itself are very small relative to other inward rectifier subunits because of its low surface expression. However, inward rectifier channels are composed of a heterotetrameric assembly of subunits, and we show that Kir2.6 readily coassembles with other Kir2.x subunits. Because of its dominant trafficking phenotype, Kir2.6 exerts an important regulatory control on the trafficking of inward rectifier channels through dominant negative retention in the ER. To our knowledge, this is the first demonstration of regulation of the trafficking of strong inward rectifier Kir2 potassium channels by dominant negative interaction with wild type channel subunits.
Localization of potassium channels in skeletal muscle is important to their function; Kir2 channels play a major role in setting the cell resting potential, in controlling muscle cell excitability by determining the extent of sodium channel inactivation, and in clearing activity-dependent accumulation of K ϩ from T-tubules. We have shown by immunocytochemistry that the major skeletal muscle Kir channels Kir2.1 and Kir2.2 are well positioned for these functions, with Kir2.1 channels located on the plasma membrane and the peripheral regions of T-tubules, and Kir2.2 located in central regions of T-tubules and neuromuscular junction (Fig. 6). These results agree well with the proposed localization of inward rectifier channels based on electrophysiological studies in frog skeletal muscle (8, The error bars indicate Ϯ S.E. E, mouse tibialis cranialis skeletal muscle fibers were electroporated in vivo with GFP-Kir2.6. Seven days later, tissue was fixed, permeabilized, and labeled with rabbit anti-Kir2.2 and anti-rabbit Cy3 (red) to label endogenous Kir2.2, mouse anti-GFP, and anti-mouse Cy2 (green) to label GFP-Kir2.6 and DAPI (blue). Endogenous Kir2.2 is retained in the ER only surrounding nuclei that express GFP-Kir2.6 (arrow) and not for untransfected nuclei (arrowheads). Bar, 20 M. F, localization of endogenous Kir2.2 from control skeletal muscle fibers or from muscle electroporated with GFP-Kir2.6. Endogenous Kir2.2 subunit distribution in skeletal muscle was visually scored for presence in the ER and T-tubules or only in the T-tubules from widefield fluorescence images from control fibers or for nuclei expressing GFP-Kir2.6 from electroporated fibers (n ϭ 100 nuclei for each condition). [12][13][14] and on biochemical membrane fractionation studies in rat skeletal muscle (1). Indeed a compelling case for Kir channel localization on T-tubules and plasma membrane was made by Wallinga et al. (15), who showed by modeling that maintenance of skeletal muscle excitability and excitation-contraction coupling during repetitive action potential firing requires the presence of Kir channels in these sites for clearance of K ϩ and prevention of positive shifts in the resting membrane potential. Our results also are in agreement with studies in cardiac muscle in which T-tubule and plasma membrane localization, and their roles in K ϩ accumulation have been demonstrated (5,6,10,11).
Although Kir2.2 and Kir2.6 share more than 98% identity and are thought to have arisen from gene duplication (22) with several diversifications acquired throughout evolution, their trafficking and localization are widely different. Exogenous expression of Kir2.6 in both COS-1 cells and mouse skeletal muscle showed colocalization with the ER marker PDI, indicating that Kir2.6 is retained in the ER and trafficked poorly to the plasma membrane (Figs. 1, 7, and 9). In contrast Kir2.1 and Kir2.2 trafficked out of the ER through the Golgi, shown by colocalization with Golgi markers, and finally expressed on T-tubules and the plasma membrane (Figs. 1, 7, and 10 and supplemental Fig. S2).
Even though Kir2.6 primarily is retained in the ER, a small proportion of the channel is trafficked to the cell surface in COS-1 cells. We speculated that this surface Kir2.6 formed functional homotetrameric channels. Indeed, electrophysiological investigation confirmed that Kir2.6 can form functional channels on the plasma membrane, in agreement with findings from Ryan et al. (22). Kir2.6 currents display characteristic inward rectification, with more pronounced inward currents at potentials negative to E K than at potentials positive to E K . However, when compared with Kir2.2, whole cell currents reveal a marked reduction in magnitude, implying that fewer channels are present on the plasma membrane.
It is possible that high expression levels and longer expression times in some cultured cells may allow Kir2.6 channels to escape cellular trafficking and quality control mechanisms and may account for larger Kir2.6 current magnitudes observed by Ryan et al. (22) in 293T cells. In mouse skeletal muscle, however, even after 7-14 days of expression in vivo, we found no evidence for Kir2.6 trafficking beyond the ER when Kir2.6 was expressed alone, suggesting that it is effectively retained in the ER (Fig. 7). When expressed with Kir2.1 in mouse skeletal muscle, a very low level of Kir2.6 trafficked to the Golgi (Fig. 9). In primate skeletal muscle, it is possible that a tissue-specific and species-specific accessory subunit or chaperone may aid the forward trafficking of Kir2.6; thus, a greater fraction of Kir2.6 might traffic to surface membranes in human skeletal muscle.
Although the Kir2.6 sequence includes all known anterograde trafficking signals that are present in other Kir2.x channels, it is still retained in the ER (Figs. 1, 5, 7, 9, and 10). Our studies show that a previously unidentified site, proline 156, is the most essential amino acid for surface trafficking of Kir2.2, and notably, the leucine at position 156 of Kir2.6 caused its retention in the ER (Figs. 3 and 8). Alignment of all known human Kir family sequences to identify conserved amino acids shows that the amino acid pair cysteine 155 and proline 156 is conserved among nearly all eukaryotic Kir subunits, but a leucine is present in place of the proline in Kir2.6 (Fig. 11A). Prolines have an established role in ␣-helical turn structure and are commonly found as the first residue of an ␣-helix (48). Indeed, in the three-dimensional structure of Kir2.2 (Protein Data Bank file 3JYC), proline 156 is located at a turn that forms the beginning of the M2 inner helix (Fig. 11B) (49). Significantly, the adjacent cysteine 155 is part of an intrasubunit disulfide with conserved cysteine 123 that is essential for channel function and that bridges the two extracellular loops in eukaryotic Kir2 channels (49 -52). It has been shown that mutation of either of these conserved cysteines of Kir2.1 in one subunit of a channel tetramer is sufficient to eliminate wild type Kir2.1 currents in a dominant negative manner (50). We assume that proline 156 in Kir2.2 induces bending that initiates the M2 helix. Leucine at this position in Kir2.6 likely impacts channel conformation and may alter folding efficiency or disulfide bond formation that could result in retention of Kir2.6 in the ER.
Interestingly, the absence of a conserved proline has been implicated in a loss of function of two voltage-gated potassium channel subunits. Kv6.3 and Kv8.1 are electrically silent homotetrameric channels because of retention in the ER, a phenomenon likely resulting from divergence of a proline-containing motif. Study of chimeric subunits between Kv6.3 and Kv2.1 and point mutation analysis in Kv8.1 revealed that a PXP motif, located in the S6 transmembrane domain in the gating hinge, is altered to PXT or PXA in the silent channels, respectively (53, 54) (Protein Data Bank file 2A79).
We have shown that Kir2.6 associates with Kir2.1 and Kir2.2 by immunoprecipitation (Fig. 4), with important consequences on channel trafficking, plasma membrane abundance, and localization (Figs. 3, 5, 9, and 10). Kir2 channels are known to form by homotetrameric and heterotetrameric assembly of Kir2.x subunits, with the resulting channels reflecting the physiological properties of their subunits (3). However, the subunit composition, the extent of heteromultimerization, and the roles of individual subunits in different cell types are not known. Gene knock-out studies in mice showed that Kir2.1 and Kir2.2 contribute to K ϩ currents in heart in a nonadditive manner, suggesting that these subunits can coassemble in vivo (21). In cardiac myocytes, the diversity of channel conductances, Ba 2ϩ block properties, and pH sensitivity also support a model in which channels are composed of homotetrameric and heterotetrameric subunits (33)(34)(35)(36). Our data showing the overlapping distribution of endogenous Kir2.1 and Kir2.2 channels in skeletal muscle T-tubules are consistent with the idea that a fraction of these subunits coassemble in vivo in skeletal muscle (Fig. 6).
Our trafficking studies, in which Kir2.6 causes ER retention of other Kir2.x subunits and colocalization of subunits, provide strong support for heteromultimeric formation of Kir2 channels in skeletal muscle, cardiac muscle cells, and COS-1 cells (Figs. 3, 5, 9, and 10). Kir2.6 is the dominant subunit when it forms a heteromultimer with Kir2.1 or Kir2.2 in vivo and has a stronger dominant negative trafficking effect on Kir2.2 compared with Kir2.1, suggesting that those subunits that are most similar in sequence may have a greater propensity to coassemble. A dominant negative trafficking phenomenon also is a well known explanation for disease-associated mutations in Andersen-Tawil syndrome where Kir2.1 ⌬95-98 and Kir2.1 ⌬314 -315 do not traffic to the plasma membrane. Instead, they are distributed inside the cell, although they retain their ability to coassemble with wild type channels (18). Bendahhou et al. (18) speculate that deletion of amino acids 95-98, located in the outer M1 helix, results in protein misfolding and consequent ER retention. Coexpression of wild type Kir2.1 with Kir2.1 ⌬95-98 or Kir2.1 ⌬314 -315 trapped the wild type subunit at intracellular sites. Reduction in Kir2.1 expression on the plasma membrane can interfere with regulation of electrical excitability and muscle resting potential (18).
It is interesting to note that another gene encoding Kir2.5/ Kir2.2v, also closely related to Kir2.2, may act as a negative regulator of Kir2.2 channel activity as well, but in that case the Kir2.5 subunit is electrically silent possibly because of alterations of the conserved GYG selectivity filter or C terminus (55).
Although our studies do not address the function of Kir2.6 mutations that are associated with TPP, we speculate that some of those mutations may alter channel trafficking, leading to changes in the magnitude of Kir2 currents on surface membrane and T-tubules. Two of the TPP-associated mutations, Kir2.6 R399X and Kir2.6 Q407X, cause truncation of the Kir2.6 C terminus and result in the absence of a putative PDZ-binding motif (22), a region known to be important for subcellular localization of Kir2 channels (7,10,11,28,30,31).
The Kir2.6 gene contains a thyroid-responsive element that may regulate gene transcription (22) and consequently lead to changes in its protein abundance, ER retention of Kir2.x subunits, and reduction in inward rectifier currents. Thus, thyrotoxicity may alter a delicate balance of electrical activity in muscle. Modeling studies and human disease-associated mutations in the genes encoding Kir2.1 and Kir2.6 suggest that too little or too much inward rectifier current can alter the electrical excitability of muscle and lead to paralysis (15,17,22,56). Because TPP patients normalize when treated with antithyroid agents (reviewed in Ref. 57), the consequences of high T3 on muscle conductance is of interest.
In summary, we suggest that Kir2.6 is a regulatory subunit that is mainly retained in the ER as a consequence of leucine 156. Association between Kir2.6 and Kir2.1 or Kir2.2 promotes ER retention of the heterotetramer. The amount of inward rectifier channels on the plasma membrane may depend on the ratio between the Kir2.1, Kir2.2, and Kir2.6 subunits. We speculate that homeostasis is maintained in healthy individuals, enabling a stable amount of cell surface channel expression. Together, our data support the concept that ER retention is an important mechanism in controlling skeletal muscle excitability.