Defective Potassium Channel Kir2.1 Trafficking Underlies Andersen-Tawil Syndrome*

Andersen-Tawil syndrome is a skeletal and cardiac muscle disease with developmental features caused by mutations in the inward rectifier K+ channel gene KCNJ2. Patients harboring these mutations exhibit extremely variable expressivities. To explore whether these mutations can be correlated with a specific patient phenotype, we expressed both wild-type (WT) and mutant genes cloned into a bi-cistronic vector. Functional expression in human embryonic kidney 293 cells showed that none of the mutant channels express current when present alone. When co-expressed with WT channels, only construct V302M-WT yields inward current. Confocal microscopy fluorescence revealed three patterns of channel expression in the cell: 1) mutations D71V, N216H, R218Q, and pore mutations co-assemble and co-localize to the membrane with the WT and exert a dominant-negative effect on the WT channels; 2) mutation V302M leads to channels that lose their ability to co-assemble with WT and traffic to the cell surface; 3) deletions Δ95-98 and Δ314-315 lead to channels that do not traffic to the membrane but retain their ability to co-assemble with WT channels. These data show that the Andersen-Tawil syndrome phenotype may occur through a dominant-negative effect as well as through haplo-insufficiency and reveal amino acids critical in trafficking and conductance of the inward rectifier K+ channels.

From the ‡Department of Human Molecular Biology and Genetics, ¶Division of Pediatric Cardiology ʈDepartment of Neurobiology and Anatomy, **Department of Neurology and Howard Hughes Medical Institute, University of Utah, Salt Lake City, Utah 84112 Andersen-Tawil syndrome is a skeletal and cardiac muscle disease with developmental features caused by mutations in the inward rectifier K ؉ channel gene KCNJ2. Patients harboring these mutations exhibit extremely variable expressivities. To explore whether these mutations can be correlated with a specific patient phenotype, we expressed both wild-type (WT) and mutant genes cloned into a bi-cistronic vector. Functional expression in human embryonic kidney 293 cells showed that none of the mutant channels express current when present alone. When co-expressed with WT channels, only construct V302M-WT yields inward current. Confocal microscopy fluorescence revealed three patterns of channel expression in the cell: 1) mutations D71V, N216H, R218Q, and pore mutations co-assemble and co-localize to the membrane with the WT and exert a dominant-negative effect on the WT channels; 2) mutation V302M leads to channels that lose their ability to co-assemble with WT and traffic to the cell surface; 3) deletions ⌬95-98 and ⌬314 -315 lead to channels that do not traffic to the membrane but retain their ability to co-assemble with WT channels. These data show that the Andersen-Tawil syndrome phenotype may occur through a dominant-negative effect as well as through haplo-insufficiency and reveal amino acids critical in trafficking and conductance of the inward rectifier K ؉ channels.
Andersen-Tawil syndrome (ATS) 1 is a rare autosomal dominant disorder caused by mutations in the inward rectifier potassium channel gene KCNJ2, which encodes the inward rectifier potassium channel Kir2.1 protein. Two features of this disease have made this gene of high interest to scientists: (i) patients suffer from both skeletal muscle periodic paralysis and cardiac arrhythmia, a unique feature in ion channel diseases; and (ii) patients exhibit developmental problems such as cleft palate, low set ears, short stature, and development features in the limbs (clinodactyly, syndactyly, brachydactyly) (1-3). While the first feature can be easily understood because of the contribution of Kir2.1 to membrane excitability in both tissues, it is still unknown how this channel contributes to craniofacial, limb, and axial skeletal development.
The inward rectifier potassium channel (Kir) family is subdivided into seven members according to the extent of amino acid sequence homology (4). The predicted transmembrane topology of the Kir family is shown in Fig. 1. The protein consists of two membrane-spanning segments, namely M1 and M2, a pore-forming loop between the two transmembrane segments, and amino and carboxyl termini located intracellularly. Four subunits co-assemble in a homo-or heteromultimeric manner to form a channel (5). The KCNJ genes are widely expressed in the body. They can be found in muscle (KCNJ2, KCNJ11), heart (KCNJ2, KCNJ3, KCNJ5, KCNJ11), brain (KCNJ3, KCNJ6, KCNJ9, KCNJ11), epithelial (KCNJ1, KCNJ2), and many other tissues (6 -12). Mutations in the KCNJ gene family produce three inherited disorders in man and one in mouse. In KCNJ1 (Kir1.1), mutations have been associated with Bartter's syndrome (13), an autosomal recessive disorder characterized by hypokalemia and salt wasting. Mutations in KCNJ11 (Kir6.2) and the associated protein SUR1 have been linked to the syndrome of persistent hyperinsulinemic hypoglycemia in infancy (14,15). A mutation in Kir3.2 (GIRK2) produces the weaver phenotype in mice in an autosomal recessive manner (16). This disease results in a loss of neurons and severe ataxia. Most recently, we reported 19 mutations in the KCNJ2 gene causing ATS (17)(18)(19). The clinical manifestations in ATS patients are very variable, suggesting that different mutations may account for different phenotypes.
Early expression of mutations in Xenopus oocytes led us to suggest that all ATS mutations cause ATS through a dominant-negative mechanism. To better understand the pathogenesis of this disease, we carried out mammalian cell expression studies of previously identified mutations D71V, ⌬95-98, S136F, G144S, N216H, R218Q, G300V, V302M, E303K, and ⌬314 -315. While most of the mutations appeared to have a dominant-negative effect on the Kir2.1 channel, we observed near-normal currents when mutation V302M was co-expressed with WT. Furthermore, confocal microscopy fluorescence data implicate several amino acid residues, previously unknown, essential for channel assembly, trafficking, and membrane expression. This work provides insights into the complex functional domains of the COOH terminus of Kir2.1 as well as ATS pathogenesis.

MATERIALS AND METHODS
Mutagenesis-Human cardiac KCNJ2 was obtained in the plasmid pBluescript KS(Ϫ) as a gift from Carol Vandenberg. Mutagenic primers were used to generate all point mutations and deletions with the Gene-Editor kit (Promega) according to the manufacturer's recommendations and as described previously (17). WT cDNA was fused to Ds red fluorescent protein (DsRed) and all mutant clones (D71V, S136F, G144S, N216H, R218Q, G300V, V302M, E303K, amino acid deletion 95-98, ⌬SWLF, and deletion 314 -315, ⌬SY) were fused to enhanced green fluorescent protein (EGFP) and subcloned in the pDsRed and pEGFP vectors. These clones were subsequently PCR-amplified using primers flanked with NheI and EcoRI enzymes for the first cloning site or NotI restriction enzyme for the second cloning site and subcloned into the pIRES vector (BD Biosciences). Mutant clones were subcloned in the first multicloning site and the WT-RFP cDNA was subcloned in the second multi-cloning site. All PCR products and final constructs were sequenced to ensure fidelity.
Tissue Culture and Electrophysiology-Human embryonic kidney (HEK) 293 cells were grown and transfected using calcium-phosphate precipitation technique (20) and as described previously (21).
Current recordings were conducted in the whole-cell configuration (22) at room temperature (22°C), using an Axopatch 200B amplifier (Axon Instrument Inc., Foster City, CA). Data acquisition and analysis were performed with pClamp6 (Axon Instrument Inc.) and Igor Pro (WaveMetrics Inc., Lake Oswego, OR) programs. Pipettes (Kimax) of 0.8 -1.5 M⍀ were prepared using a Sutter P-87 puller (Novato, CA) and heat-polished prior to use. The pipette tip potential was adjusted to zero for each experiment. To minimize space clamp problems due to the electrical coupling between HEK293 cells, only isolated cells with a soma size of 10 -30 m were selected for recordings. Twenty-four to 36 h after transfection, the culture medium was removed and replaced with a bathing solution containing: 135 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 10 mM glucose, and 10 mM HEPES (pH 7.3). The pipette solution was: 140 mM KCl, 2 mM MgCl 2 , 1 mM EGTA, and 10 mM HEPES (pH 7.3). Currents were elicited by 200-ms pulses applied in 10-mV increments to potentials ranging from Ϫ120 mV to ϩ30 mV from a holding potential of Ϫ60 mV.
Confocal Microscopy-Fluorescence microscopy was performed with a Nikon Eclipse TE300 inverted microscope, and images were captured and manipulated with a CoolSNAP camera and software. The confocal is a Bio-Rad MRC 1024, and the software is LaserSharp version 3.2.

Andersen-Tawil Syndrome Mutants Do Not Produce Current in HEK293
Cells-To correlate functional expression data with confocal microscopy fluorescence imaging, KCNJ2 wild type (WT) or mutant cDNAs (D71V, G144S, N216H, R218Q, G300V, V302M, E303K, amino acid deletion ⌬95-98, and ⌬314 -315) were fused to DsRed or EGFP at the COOH-terminal end and subcloned into the pIRES vector ( Fig. 1). All constructs were subsequently introduced and expressed in HEK293 cells using whole cell patch clamp technique. Only cells transfected with WT-DsRed cDNA exhibited K ϩ conductance when cells were subjected to potentials ranging from Ϫ120 to ϩ30 mV (Fig. 2). The current-voltage relationship of the WT-EGFP or WT-DsRed clones is identical to published data on KCNJ2 gene expressed in HEK293 cells (23). This suggests that the KCNJ2 protein product retained its normal functional properties when fused to EGFP or DsRed such as subunit assembly, trafficking, and rectification. However, there was no current expressed in HEK293 cells 24 h or even 36 h post-transfection when mutant cDNAs were used. All currents recorded have the same amplitude and characteristics of those recorded from mock-transfected cells (Ϫ117 Ϯ 22 pA at Ϫ120 mV, n ϭ 8).
ATS Mutants Exert a Dominant-negative Effect on WT Chan-nels-Because ATS is an autosomal dominant trait and KCNJ2 channels can assemble as homo-tetramers, we introduced WT and mutant channels in the same cell using a bi-cistronic expression vector. A pIRES plasmid was constructed that directs the co-expression of two proteins with the first protein fused to EGFP (mutant cDNA) and the second protein fused to DsRed (WT cDNA). An encephalomyocarditis virus internal ribosomal entry site (IRES) mediates translational initiation from this 3Ј cassette, and a cytomegalovirus promoter drives bi-cistronic transcripts expression. Current recorded from cells transfected with the construct containing WT-GFP/WT-RFP had an amplitude higher (Ϫ2,788 Ϯ 355 pA at Ϫ120 mV; n ϭ 8) than that recorded from cells transfected with WT-GFP (Ϫ2,347 Ϯ 316 pA at Ϫ120 mV; n ϭ 8) or WT-DsRed only, suggesting that transcripts are produced from both sites. When mutant channels (mutant-EGFP) were used at the first site and WT-DsRed at the second site in the same vector no current was detected from all constructs except V302M-GFP/WT-DsRed (Ϫ700 Ϯ 135 pA at Ϫ120 mV; n ϭ 8) (Fig. 2). These results suggest that all ATS mutants FIG. 1. Topology of the Kir family and mutations associated with the KCNJ2 gene. A, two helical structures form the membranespanning segments (M1 and M2), the pore region P, and two intracellular fragments (NH 2 and COOH termini) form a subunit that oligomerize to form a channel. All ATS missense mutations (filled circles, R67W, D71V/N, T75R, S136F, G144S, G146D, P186L, R189I, N216H, R218Q/W, G300V/D, V302M, E303K, R312C) and deletions (triangles, ⌬95-98 and ⌬314 -315) are represented on the channel. B, pIRES constructs used for HEK293 cells transfection. WT cDNA was fused to DsRed fluorescent protein (DsRed) and mutant clones (D71V, T75R, S136F, G144S, N216H, G300V, V302M, E303K, amino acid deletion 95-98, ⌬SWLF, deletion 314 -315, ⌬SY) were fused to EGFP and subcloned in the pDsRed and pEGFP vectors. These clones were subsequently PCR-amplified and subcloned into the pIRES vector. Mutant clones were subcloned in the first site and the WT-RFP cDNA was subcloned in the second site. exert a dominant-negative effect on WT channels except the V302M mutation, which may cause the disease through a haploinsufficiency mechanism.
Channel Assembly and Trafficking Are Targeted in the Andersen-Tawil Syndrome-To understand how these mutants exert their dominant-negative effect and why the V302M substitution leads to ion conductance when co-expressed with WT channels, all constructs fused to either EGFP or DsRed were expressed in HEK293 cells on a coverslip and confocal microscopy study was conducted. We used single constructs to monitor channel distribution in the cell and double constructs (mutant-EGFP/WT-DsRed) to monitor both channel distribution and assembly. The same constructs used in electrophysiology were used for confocal microscopy imaging. All single constructs showed that mutant channels had a cellular distribution similar to that of WT channels except ⌬95-98-EGFP, V302M-EGFP, and ⌬314 -315-EGFP. These three mutants exhibited a scattered pattern of expression (Fig. 3). Double constructs with mutant-EGFP in the first site and WT-DsRed in the second site allowed us to investigate how our mutants interact with WT KCNJ2. Fig. 4 shows the distribution of the WT and mutant channels in HEK293 cells. While D71V, G144S, N216H, R218Q, G300V, and E303K (-EGFP) co-localize to the membrane with the WT-DsRed producing a yellow color, ⌬95-98-EGFP and ⌬314 -315-EGFP co-localize in the cytoplasm with WT-DsRed and also produce a yellow color (Fig. 4). Interestingly, V302M channels do not co-localize with WT (green fluorescence in the cytoplasm), and only WT-DsRed is expressed in the membrane.
According to the crystal structure of the a bacterial potassium channel (KirBac1.1), which has structure representative of the Kir family, amino acids 95-98 are found in the outer helix (24). A deletion in this segment may cause misfolding of the protein and, consequently, retention in the endoplasmic reticulum or degradation. Amino acids 302, 314, and 315 are located in the complex functional COOH terminus domain of Kir2.1. For ⌬314 -315 channels, the yellow color seems to be mainly localized to the cytoplasm. This indicates that these mistrafficked mutant channels retain their ability to co-assemble with the WT-DsRed protein, thus trapping functional WT-DsRed protein in the cytoplasm. Finally, the V302M substitution seems to affect both channel trafficking and assembly, since the V302M-EGFP channels are found in the cytoplasm and the WT-DsRed channels are located in the membrane (Fig.  4). This expression pattern further supports our electrophysiological data showing that only the V302M-EGFP/WT-DsRed construct was able to give rise to potassium conductance, due to the free WT-DsRed subunits that were unable to co-assemble with the mutant subunits.
To identify the relative contribution of the deleted amino acids at positions 314 and 315, we mutated each of them to alanine and introduced them into the pIRES vector with and without WT-DsRed in the second site. Confocal microscopy images of transfected HEK293 cells showed that Y315A substitution caused a more dramatic effect than substitution S314A (Fig. 5). The WT and mutant channels seem to be sequestered together in the cytoplasm.
Involvement of COOH-terminal Residues of Kir2.1 in Trafficking-Our study of ATS mutations revealed three amino acid residues in the COOH-terminal domain of the Kir channels that may be part of a binding motif essential for channel trafficking. This region is indeed well conserved among Kir family members (Fig. 6B). To test our hypothesis, we extended our mutagenesis studies to residues surrounding the amino acids involved in ATS. We performed alanine scanning mutagenesis of most of the residues from 290 to 325. All Kir clones were constructed in the first cloning site of the pIRES vector in-frame with the EGFP. After cell transfection, confocal microscopy images were taken to assess the localization of the channels. Our data show that mutations of residues Leu 316 , Glu 319 , Ile 320 , Leu 321 , and Trp 322 show an intracellular distribution pattern (Fig. 6A) very similar to that observed for either the deletion ⌬Ser 314 -Tyr 315 (Fig. 4) or the substitutions S314A-GFP and Y315A-GFP (Fig. 5). Although these channels retain their ability to co-assemble with WT, they seem to be sequestered in the cytoplasm. DISCUSSION ATS is characterized by periodic paralysis, ventricular dysrhythmias, with long-QT and skeletal developmental abnor-malities. We have linked the KCNJ2 gene to ATS and identified mutations in families with ATS phenotypes (17)(18)(19). Three of these mutations are located in the NH 2 terminus (R67W, D71V/N, and T75R), one deletion occurs in the first transmembrane segment M1 (⌬95-98), three mutations occur in the pore region (S136F, G144S and G146D), and the remaining mutations reside in the COOH terminus of the Kir 2.1 channel (P186L, R189I, N216H, R218Q/W, G300V/D, V302M, E303K, R312C, and ⌬314 -315). Combining functional expression and confocal microscopy imaging, we were able to identify amino acids that abolish channel function, affect homomeric channel assembly, and play a role in channel trafficking.
D71V, N216H, R218Q, R218W, G300V, and E303K mutant channels seem to co-assemble with the Kir2.1-WT channels and to co-localize to the plasma membrane in HEK293 cells. In so doing, the non-functional mutant protein likely "poisons" the Kir2.1 channel tetramer. We anticipate that several mechanisms underlie channel silencing, since these amino acid residues localize to different functional domains of the Kir2.1 protein as shown on the crystal structure of the Kirbac1.1 (Fig. 7). Changes occurring in the pore region likely alter the conformation of the pore or its potassium ion selectivity to produce a non-functional channel as has already been shown for a Kir3.2 mutation causing the weaver phenotype in mice (16) and those introduced to the pore of the channel (25). The D71V mutation does not alter channel assembly or trafficking or seem to be degraded, suggesting that the slide helix plays a role in the Kir2.1 channel gating. Alternatively, mutations occurring in the NH 2 terminus cytoplasmic region of the Kir2.1 may interfere with the correct insertion of segment M1 into the membrane (26).
Mutations in the COOH terminus of the protein produce diverse effects on the whole cell current confirming multiple functions of this domain. This is expected since the distal part of the channel starting at the M2 segment has been implicated in rectification, sensitivity to phosphorylation, phosphatidylinositol 4,5-bisphosphate (PIP 2 ) binding, PSD-95/SAP-90 binding, and has an endoplasmic reticulum export motif (23,(27)(28)(29)(30)(31)(32). While N216H, R218Q, R218W, G300V, E303K, and ⌬314 -315 produce no current when expressed alone or with the normal WT channel, V302M does not produce any measurable current when expressed alone but yields an inwardly rectifying current in the presence of WT channels in the same cell. By altering this site, the mutant channels could not reach the plasma membrane. This may be due to a trafficking defect or a consequence of channel misfolding. Only channels composed of free WT reached the membrane and could produce detectable current (Fig. 2).
Using anti-GFP antibodies, we were able to detect two major bands at 75 and 90 kDa for all fusion constructs except that of the V302M mutation that showed only 75-kDa band (Fig. 8). This finding may indicate instability of the V302M mRNA or protein. Alternatively, this mutation may cause misfolding of Kir2.1 leading to incomplete post-translational modifications. Further characterization of this mutation is necessary to elaborate the associated mechanism. In any case, our data suggest that residue Val 302 is part of a critical motif responsible for proper channel folding and may produce ATS through a haploinsufficiency mechanism.
Deletion ⌬314 -315 also lost its ability to traffic to the membrane but did not alter subunit assembly. WT subunits were trapped in the cytoplasm by the mutant subunits to form nonfunctional channels. Alanine scanning mutagenesis of these two amino acids showed that substitutions at residue 315 may have a more pronounced effect on channel trafficking than those at residue 314 (Fig. 5). These data also support our hypothesis that this region is part of an essential binding motif for channel trafficking.
Mutations N216H, R218Q, G300V, and E303K do not seem to affect trafficking or assembly of the Kir2.1 but still produce non-functional channels. These amino acids reside close to a region thought to be involved in rectification and in PIP 2 binding (33)(34)(35)(36)(37). Residue 216 and Arg 218 have been shown to be involved in Kir2.1-PIP 2 interactions. PIP 2 seems to be ligand for a Kir2.1 channel activation (34). Channels that interact strongly with the PIP 2 are constitutively active at negative potentials and contribute to setting rest membrane potential (38,39). Additionally, we have recently shown that PIP 2 -binding residues of the Kir2.1 are common targets of mutations causing ATS (18). Mutations in PIP 2 -related residues accounted for over 60% of reported KCNJ2 probands.
It is difficult from these experiments to determine the exact contribution of each residue in the COOH-terminal domain of the protein to channel trafficking or assembly, despite the availability of the crystal structure. The COOH terminus of the Kir2.1 contains the endoplasmic reticulum export sequence FCYENE and trafficking-related acidic clusters EEDDSE at positions 374 -379 and 386 -391, respectively (28,40). Our data show that these two sequences are not sufficient for efficient surface expression and that additional trafficking signals are required. Our alanine scanning mutagenesis identified two amino acid clusters (residues 314 -316 and 319 -322) and further supports the idea of complex protein folding of the residues surrounding Ser 314 -Tyr 315 to create a three-dimensional interaction site essential for Kir trafficking. This is supported by the crystal structure of Kibac1.1, which shows ␤ sheets formed with these residues. Either other Kir domains are required or other Kir2.1 partner proteins are necessary for surface expression of these channels. Alternatively, residues located between amino acids 310 and 330 may be involved in channel heteromerization as suggested recently for these channels (41).
ATS patients exhibit the triad of muscle weakness, cardiac ectopy, and developmental abnormalities. The excitability features of this disease can be understood as Kir2.1 is expressed in the membrane of excitable cells and plays an essential role in maintaining the resting membrane potential and shaping the cardiac action potential (39). Mutations preventing proper expression of Kir2.1 in the membrane can understandably interfere with normal cell excitability and cause the repolarization defects of this disease. Our data indicate ATS can be caused by such a mechanism and implicate novel regions of Kir2.1 essential for membrane expression. In addition, our findings may have implications for other loci underlying ATS as previous studies have indicated mutations in other genes likely cause nearly 40% of diagnosed cases of ATS (19). Genes encoding proteins which facilitate membrane expression of Kir2.1 by interacting with the trafficking regions of the COOH terminus or localize Kir2.1 to specific regions of the membrane may be involved. Recently, mutations in the first non-ion channel, nonaccessory subunit (Ankyrin-B), were found to cause a form of long QT syndrome, LQT4 (42). This adaptor protein anchors ion channels to the T-tubules and a loss of its function seems to cause long-QT by resultant disruption in localized ion channel function. Another study has shown that filamin-A (actin-binding protein) may interact with Kir2.1 protein at the COOH terminus domain (43). It is reasonable to speculate that mutations in interacting, trafficking, or anchoring proteins might cause unexplained cases of ATS.
Experiments are under way to determine the role of the amino acid residues in the COOH terminus and for a potential binding motif that may shed light on inward rectifier potassium channel modulation, assembly, and trafficking. Our data clearly identify residues crucial for KCNJ2 channel trafficking that can be useful to understanding the involvement of KCNJ2 in the developmental pattern seen in ATS patients.