A Fundamental Role for KChIPs in Determining the Molecular Properties and Trafficking of Kv4.2 Potassium Channels* □ S

Kv4 potassium channels regulate action potentials in neurons and cardiac myocytes. Co-expression of EF hand-containing Ca 2 (cid:1) -binding proteins termed KChIPs with pore-forming Kv4 (cid:2) subunits causes changes in the gating and amplitude of Kv4 currents (An, W. F., Bowlby, M. R., M., Cao, J., Ling, H. P., Mendoza, G., Hinson, J. W., Mattsson, K. I., Strassle, B. W., Trimmer, J. S., and Rhodes, K. J. (2000) Nature 403, 553–556). Here we show that KChIPs profoundly affect the intracellular trafficking and molecular properties of Kv4.2 (cid:2) subunits. Co-expression of KChIPs1–3 causes a dramatic redistribution of Kv4.2, releasing intrinsic endoplasmic reticulum retention and allowing for trafficking to the cell surface. KChIP co-expression also causes fundamental changes in Kv4.2 steady-state expression levels, phosphorylation, detergent solubility, and stability that reconstitute the molecular properties of Kv4.2 surface expression of KChIP4a-containing in to the effects of on trafficking of (cid:1) Our staining of with an to the core of suggests can exist in the plasma membrane of native excitable cells.

Voltage-gated potassium or Kv channels, specifically those mediating low threshold, rapidly inactivating I to and I A currents, are known to regulate cardiac and neuronal membrane excitability, respectively (1). In cardiac cells, I to is a major determinant of the falling phase of the cardiac action potential (2). In hippocampal pyramidal neurons, dendritic I A can limit the peak of back-propagating action potentials as well as modulate incoming synaptic information, exerting profound effects on information processing (3). Kv channels are composed of homo-or heterotetramers of transmembrane pore-forming and voltage-sensing ␣ subunits (4,5). Available evidence suggests that Shal or Kv4 family ␣ subunits underlie I to in cardiac myocytes, and I A in dendrites of many central nervous system neurons. The expression of specific Kv4 family mRNAs correlates precisely with the level of I to in cardiac myocytes (6,7) and dendritic I A in neurons (8 -11). Robust staining with antibodies specific for Kv4 family members is observed in cardiac myocytes (12) and in neurons (13)(14)(15)(16). Moreover, experimental knockdown of Kv4 expression in cardiac myocytes (17)(18)(19) and in neurons (20,21) results in suppression of I to and I A , respectively.
Many Kv channels contain auxiliary subunits that can regulate the biophysical, biochemical, and cell biological characteristics of the resultant channel complexes (22). We have recently described a highly related family of four Kv4 Channel Interacting Proteins (KChIPs1-4) that binds to the cytoplasmic N-terminal domain of Kv4 ␣ subunits (16,23). When coexpressed in heterologous cells, KChIP1, KChIP2, and KChIP3 dramatically alter the inactivation kinetics and rate of recovery from inactivation of Kv4 channels and boost the amplitude of obtained current (16). The effects of the KChIP4a splice variant are distinct from other KChIPs, due to a unique N terminus (23). KChIPs may contribute to both constitutive and dynamic regulation of Kv4 channels. KChIPs are Ca 2ϩ -binding proteins. The conserved KChIP core region contains four EF hand-containing Ca 2ϩ -binding motifs (16), and studies with a minimal KChIP isoform suggest Ca 2ϩ -dependent effects on inactivation (24). Cyclic AMP-dependent protein kinase (PKA) 1 regulation of Kv4.2-encoded currents in heterologous cells also requires KChIP co-expression (25). The essential role of KChIPs was acutely demonstrated by genetic ablation of KChIP2 expression in mice, which completely eliminated the cardiac I to current resulting in a predisposition to cardiac arrhythmias (26). KChIP3 knockout mice exhibit reduced sensitivity to pain (27), although as KChIP3 has also been identified as DREAM, a calcium-dependent transcription factor (28), and calsenilin, which interacts with presenilins (29), the molecular pathophysiology of the phenotype of this knockout is not as clear.
In order to better understand the molecular mechanisms underlying KChIP regulation of cardiac and neuronal excitability, we have investigated the effects of KChIP co-expression on the intracellular trafficking and molecular characteristics of Kv4.2 channels. We found that co-expression of KChIP1, * This work was supported by National Institutes of Health Grants NS42225 (to J. S. T.) and NS37444 (to J. D. S.) and by Wyeth Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. KChIP2, and KChIP3, but not KChIP4a, leads to a conspicuous subcellular redistribution of Kv4.2 channels, from being ERretained to being expressed on the cell surface, via a mechanism that may involve masking of a cytoplasmic trafficking and/or solubility determinant. KChIPs also induce a striking transformation in the overall molecular properties (expression level, phosphorylation, detergent solubility, and stability) of Kv4.2, leading to characteristics more typical of those observed for Kv4.2 in native cells. The dramatic differences between different KChIP isoforms in inducing these effects suggest that the precise KChIP composition of Kv4 channels impacts diverse aspects of function. Thus KChIPs induce dramatic effects on not only the biophysical properties but also the cell biological and molecular characteristics of Kv4 ␣ subunits that dramatically impact the function, abundance, and distribution of Kv4 channels in excitable cells.
Immunofluorescence Analyses of Transfected COS-1 Cells-COS-1 cells were grown in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 10% newborn calf serum (Hyclone Laboratories, Logan, UT), 50 units/ml penicillin, 50 g/ml streptomycin (both from Invitrogen) in a humidified incubator at 37°C under 5% CO 2 . Cells were maintained in plastic tissue culture dishes or on poly-L-lysinecoated glass coverslips in plastic Petri dishes. Cells were transfected with mammalian expression vectors for Kv4.2 and KChIPs by the calcium phosphate precipitation method (37) or with LipofectAMINE 2000 (Invitrogen) or Polyfect (Qiagen) transfection reagents using the manufacturers' protocols. Cells expressing Kv4.2 and/or KChIPs were stained 48 h post-transfection using a surface immunofluorescence protocol (38,39), applying the ectodomain-directed K57/1 mouse monoclonal antibody prior to detergent permeabilization to detect the cell surface Kv4.2 pool. The total cellular Kv4.2 pool was detected by EGFP fluorescence or with cytoplasmic directed rabbit polyclonal antibody Kv4.2N following detergent permeabilization. No difference in the intrinsic localization or KChIP-induced effects was observed between wild type and EGFP-tagged Kv4.2.
Bound primary antibodies were detected using Alexa 594-conjugated goat anti-mouse IgG and, if needed, Alexa 488-conjugated goat antirabbit IgG. Cells were viewed under indirect immunofluorescence on a Zeiss Axioskop 2 microscope. Cells with detectable surface staining (red) and with total green fluorescence or staining, indicating successful transfection, were scored under narrow wavelength Texas Red and fluorescein filter sets, respectively. For some experiments standard double immunofluorescence staining of permeabilized COS-1 cells was performed as described (35). Images of cells were captured into a Zeiss Axiocam (Oberkochen, Germany) cooled CCD 24-bit color digital cam-era mounted on a Zeiss Axioskop 2 microscope with a 100ϫ, 1.4 numerical aperture objective, using the software supplied with the camera.
Immunoprecipitation, SDS-PAGE, and Immunoblotting-Analyses of COS-1 cell lysates prepared from transfected cells were performed as described (37,39). In brief, to harvest COS-1 cells and prepare detergent lysates, cells were first washed twice in ice-cold PBS and then lysed for 5 min on ice in 1 ml of an ice-cold lysis buffer solution containing TBS (10 mM Tris, 150 mM NaCl, pH 8.0), 5 mM EDTA, 0.4% Triton X-100, 1 mM iodoacetamide, and a protease inhibitor mixture (2 g/ml aprotinin, 1 g/ml leupeptin, 2 g/ml antipain, 10 g/ml benzamidine, and 0.2 mM phenylmethylsulfonyl fluoride). The detergent lysate was centrifuged in a microcentrifuge for 2 min at 13,000 ϫ g to pellet nuclei and debris, and the resulting supernatant (cleared lysate) was saved for analysis. For immunoblots, the cleared lysate was added to an equal volume of 2ϫ reducing SDS sample buffer and fractionated on 9 (for Kv4.2) or 15% (for KChIPs) SDS-polyacrylamide gels. Note that "lauryl sulfate" (Sigma L-5750; 69% lauryl sulfate (SDS), 26% myristyl sulfate, 5% cetyl sulfate) was used in SDS gel recipes to accentuate electrophoretic mobility differences between different phosphorylation states of Kv4.2, as has been described previously for other proteins (37,40). Quantitation of immunoreactivity was performed by densitometry and analysis using Image J software, a public domain Java image processing program (rsb.info.nih.gov/ij/index.html).
For immunoprecipitation reactions, 200 l of detergent lysate was diluted to 1 ml in ice-cold lysis buffer. Affinity-purified polyclonal Kv4.2C antibody (5 g) was added, and the mixture was incubated on a tube rotator at 4°C for 16 h. The antibody-antigen complex was immobilized by adsorption onto 15 l of protein A-agarose (Pierce) by incubation on a tube rotator for 1 h at 4°C. Protein A beads were washed six times in lysis buffer and then incubated with calf intestinal alkaline phosphatase (AP, 0.1 unit/ml) for 16 h at 37°C. The beads were then resuspended in reducing SDS sample buffer and analyzed on 9% SDSpolyacrylamide gels.
Cycloheximide Treatment-To evaluate the stability of Kv4.2 by blocking total cellular protein synthesis, cultured COS-1 cells 24 h post-transfection were treated with cycloheximide (100 g/ml) for the indicated times and then analyzed by immunoblotting.
Primary Rat Hippocampal Neuronal Cultures-Low density primary embryonic rat hippocampal cultures were prepared according to established protocols (41,42). Briefly, an astrocyte culture was prepared from cerebral hemispheres of neonatal rats and cultured for 5 days in 6-well tissue culture plates in minimum essential medium containing 10% horse serum and 0.6% glucose prior to plating of hippocampal neurons. The day before the hippocampal dissection, the astrocyte medium was changed to neuronal maintenance medium (minimum essential medium containing N2 supplements, 0.1% ovalbumin, and 0.1 mM sodium pyruvate) for conditioning. Hippocampi dissected from embryonic day 18 rat embryos were dissociated by treatment with 0.25% trypsin at 37°C for 15 min. The hippocampal preparation were plated in minimum essential medium supplemented with 10% horse serum and 0.06% glucose onto 22-mm square coverslips (72 cells/mm 2 ) or 60-cm plastic tissue culture dishes (177 cells/mm 2 ) previously coated with 1 mg/ml poly-L-lysine. Neurons were incubated in a humidified incubator at 5% CO 2 for 4 h to attach to the coverslips, and then the coverslips transferred, inverted on wax pedestals, to the 6-well tissue culture plates that contained the previously established astrocyte cultures as a feeder layer. For neurons on plastic dishes, the media were changed to the serum-free media conditioned with a separate astrocyte culture. After 3 days, cytosine arabinoside was added to the media to 1 M final concentration. Cultures were maintained at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . One-half of the culture medium was changed weekly.
Immunofluorescence Staining of Cultured Neurons-Staining of cultured neurons was performed as described previously (43). Briefly, cultured neurons were fixed in 3% (w/v) paraformaldehyde, 3% sucrose in phosphate-buffered saline (PBS, 0.15 M NaCl, 10 mM sodium phosphate, pH 7.4) for 15 min, washed twice with PBS, and permeabilized in 0.1% (v/v) Triton X-100. Nonspecific binding sites were blocked with Blotto-T (4% (w/v) non-fat dry milk in 20 mM Tris, pH 8.0, 150 mM NaCl, and 0.1% Triton X-100). Cells were then incubated in primary antibodies for 3 h or overnight. Cells were then washed three times in Blotto-T to remove excess primary antibody, followed by incubation in the appropriate secondary antibodies (diluted to 1:2000 in Blotto-T) for 1 h. After three washes in PBS-T, coverslips were mounted on microscope slides in a 90% (v/v) glycerol solution containing 0.1 mg/ml p-phenylenediamine in PBS, pH 9.0.

KChIP Co-expression Leads to Changes in the Intracellular
Trafficking of Kv4.2-To begin to address the effects of KChIP co-expression on Kv4 channels, we analyzed the immunofluorescence staining pattern of COS-1 cells expressing either Kv4.2 alone or Kv4.2 plus KChIP1, KChIP2, KChIP3, or KChIP4a. We found that virtually all of the Kv4.2-expressing COS-1 cells had robust intracellular Kv4.2 staining in a perinuclear pattern typical for ER-retained membrane proteins (Fig. 1A). Moreover, no detectable cell surface staining of intact cells with anti-ectodomain Kv4.2 antibodies was observed (Fig.  1A). To determine whether the intracellular pool of Kv4.2 was in fact associated with the ER, we double-labeled COS-1 cells expressing Kv4.2 alone for Kv4.2 and the resident ER protein calnexin, which yielded precisely co-localizing perinuclear immunofluorescence staining patterns ( To determine whether the lack of KChIP4 effects on Kv4.2 localization was simply due to differences in steady-state expression level relative to other KChIPs, the dose dependence of the effects of KChIP2 and KChIP4a on Kv4.2 surface expression was examined (Fig. 1H). Increased KChIP2 expression yielded a dose-dependent increase in the fraction of Kv4.2expressing cells with detectable surface expression, with initial effects apparent at a KChIP2:Kv4.2 cDNA ratio of 1:128 and maximal effects at 1:8 -1:4. Interestingly, the percentage of Kv4.2-expressing cells that exhibited detectable surface staining reached a plateau at Ϸ60%. The basis for this observed lack of quantitative response is not known but is not due to simple lack of co-expression as Ͼ95% of the cells expressed both Kv4.2 and KChIP2. It should be noted that a similar ceiling was observed in our previous studies of Kv␤ subunit effects on Kv1.2 surface expression (38). KChIP4a co-expression did not yield any detectable increases in Kv4.2 surface expression at any dose tested (Fig. 1H).
Interestingly, KChIP4a also had a subcellular distribution distinct from the other KChIPs when expressed either alone or with Kv4.2. KChIPs1-3 expressed alone or with Kv4.2 in COS-1 cells generally exhibited a diffuse distribution throughout the cell (Fig. 2, A-C), although KChIP3-expressing cells also exhibited additional nuclear staining (Fig. 2C) consistent with its previously described DNA binding function (44). In contrast, KChIP4a staining was perinuclear (Fig. 2D), suggesting an association with the ER. KChIPs Induce Profound Changes in the Molecular Characteristics of Kv4.2-COS-1 cells expressing Kv4.2 alone exhibit a single pool of expressed protein with an apparent mass of 65 kDa on SDS gels (Fig. 3). However, two distinct populations of Kv4.2 protein, at 65 and 70 kDa, are observed when Kv4.2 is co-expressed with KChIP1, -2, or -3 (Fig. 3, A-C). The 70-kDa form is similar to the M r of native rat brain Kv4.2 (see below). The dose dependence of the induction of the surface expression and increased M r values of Kv4.2 upon co-expression of KChIPs1-3 were quite similar (Fig. 3E). Increased steady-state expression levels of Kv4.2 were also observed with increased co-expression of KChIPs, suggesting that KChIPs may also affect Kv4.2 turnover rates ( Fig. 3 and see below). Moreover, similar to its lack of effect on Kv4.2 surface expression, KChIP4a co-expression did not affect the M r or steady-state expression level of Kv4.2 at any dose tested (Fig. 3D). These observations suggest that KChIP-induced increases in Kv4.2 cell surface expression and increased M r are coupled.
As KChIP4a can bind to Kv4.2 (23), but did not induce Kv4.2 surface expression or altered M r , we next tested whether KChIP4a could suppress the inductive effects of other KChIPs. For these experiments, the Kv4.2:KChIP2 cDNA ratio was fixed at a ratio (2:1) yielding maximal KChIP2 effects (see Fig.  1), and the amount of KChIP4a cDNA was increased. We found that KChIP4a inhibited the KChIP2-induced mobility shift of Kv4.2 in a dose-dependent manner ( These profound inhibitory effects of KChIP4a raised the question as to whether KChIP4a-containing Kv4.2 complexes could be trafficked to the cell surface. We had found previously that potent ER retention signals on Kv1 subunits could be partially overcome through heteromeric assembly (39). To test whether Kv4.2 complexes with a mixed KChIP subunit composition could be found at the cell surface, we co-transfected COS-1 cells with Kv4.2, KChIP4a, and KChIP2 at a cDNA ratio of 2:1:1. The majority of cells showed ER-localized Kv4.2 and KChIP4a and diffusely distributed KChIP2 (Fig. 4C), suggesting the KChIP4a effectively competed for KChIP2 binding to Kv4.2. Interestingly, a small population of Kv4.2-expressing cells exhibited these three proteins localized at the cell periphery, suggestive of plasma membrane localization (Fig. 4D). Such plasma membrane-associated KChIP4a staining was never observed in cells expressing either KChIP4a alone or KChIP4a plus Kv4.2. These results suggest that in at least some cells KChIP4a is able to translocate from ER to plasma membrane should it co-assemble with other KChIPs (such as KChIP2) in ternary channel complexes. The basis for the differences in KChIP staining pattern in these two populations of cells is as yet unknown.
To begin to analyze the role of KChIPs in native Kv4 channels in excitable cells, we characterized Kv4.2 and KChIPs in cultured hippocampal neurons. Cultured hippocampal neurons were chosen for these experiments as Kv4.2 is highly expressed in both the hippocampus (13, 14, 16, 45) and similar hippocampal cultures (14). Moreover, given the clarity of Kv4.2 staining in these low density cultures relative to brain sections, we set out to utilize these cultures to determine whether native Kv4.2 co-localized with KChIPs. Cultured interneurons exhibited robust immunofluorescence staining for KChIP1 and KChIP4 but not KChIP2 and KChIP3 (Supplemental Material Fig. 2). Triple staining of these cells revealed that Kv4.2 and KChIPs1 and -4 precisely co-localized in large somal and dendritic clusters (Fig. 5). This precise co-localization suggests that similar to COS-1 cells expressing exogenous subunits, KChIP4 can form ternary plasma membrane complexes with Kv4.2 and other KChIPs.
KChIP Co-expression Causes Increased Phosphorylation of Kv4.2-We next set out to determine the molecular basis for the decreased electrophoretic mobility of Kv4.2 observed in cells co-expressing KChIPs1-3. Kv4.2 does not have N-linked glycosylation sites and is not glycosylated (46). As such, the observed biosynthetic shift in M r cannot be due to conversion of Endo H-sensitive high mannose to Endo H-resistant complex oligosaccharide chains, as is observed for other potassium channel polypeptides (47,48). However, treatment with the broad spectrum AP from calf intestine yielded a clear shift in the electrophoretic mobility of Kv4.2 in extracts prepared from cells co-expressing KChIP2 (Fig. 6A). AP treatment of native Kv4.2 in adult rat brain membranes yielded a similar shift in M r (Figs. 6A and 7E). The M r of Kv4.2 in these AP-treated samples resembled that of Kv4.2 in cells expressing Kv4.2 alone (Fig. 7E). These results suggest that in brain and in COS-1 cells co-expressing KChIP2, but not in cells expressing Kv4.2 alone, the Kv4.2 polypeptide is covalently modified by AP-sensitive phosphorylation.
The Detergent Solubility and Stability of Kv4.2 Is Affected by KChIP Co-expression-We next began to analyze the consequences of KChIP co-expression, and the KChIP-dependent increase in Kv4.2 phosphorylation, on the overall biochemical properties of the Kv4.2 protein. We first analyzed the solubility of Kv4.2 in the nonionic detergent Triton X-100, which can be used an indicator of the folding state of membrane proteins (49). We found that in cells expressing Kv4.2 alone, the majority of the total Kv4.2 protein was insoluble at all Triton X-100 concentrations tested (0.1-2.0%; Fig. 6B). However, in cells expressing Kv4.2 plus KChIP2, virtually all of the Kv4.2 was soluble in Triton X-100, even at the lowest Triton X-100 concentration tested (0.1%). Interestingly, both the 65-and 70-kDa forms of Kv4.2 in co-expressing cells exhibited a enhanced solubility in Triton X-100. This result suggests that KChIP association, and not the resulting increase in phosphorylation, underlies any changes in folding state that are reflected in increased detergent solubility.
Should the changes in the inherent detergent insolubility of Kv4.2 upon KChIP co-expression reflect differences in folding of the Kv4.2 polypeptide, then the association with KChIPs may also affect the stability of Kv4.2, as misfolded proteins are generally targeted for degradation. To examine the stability of the Kv4.2 protein, COS-1 cells expressing Kv4.2 alone, or Kv4.2 and KChIP2, were treated with the protein synthesis inhibitor cycloheximide. In the absence of further Kv4.2 synthesis, the stability of the existing cellular Kv4.2 pool could then be assayed by immunoblot. Fig. 6C shows the results of one such assay. The upper panel, from cells expressing Kv4.2 alone, shows that in the absence of KChIP co-expression Kv4.2 has a t1 ⁄2 of Ϸ2 h. In COS-1 cells expressing Kv4.2 and KChIP2, the 65-kDa form has a similar t1 ⁄2 (Ϸ2 h), whereas the 70-kDa form has an increased t1 ⁄2 of Ϸ4 h. This increase in stability presumably underlies the dose-dependent increases in steady-state expression levels of Kv4.2 upon KChIP co-expression (Fig. 3).
Serine 552 PKA Phosphorylation of Kv4.2 Was Induced by KChIP Co-expression-Kv4.2 is phosphorylated in vitro by purified cAMP-dependent protein kinase (PKA) at Ser-552; rat brain Kv4.2 is also phosphorylated at this site (34). We therefore examined whether the 70-kDa form of Kv4.2 in cells coexpressing KChIP2 is phosphorylated at this site using phospho-specific antibodies (34). On immunoblots of COS-1 cells expressing Kv4.2 alone, or Kv4.2 plus KChIP2, the CT-PKA antibody (specific for phospho-Ser-552) recognized Kv4.2 only from the co-expressing cells, and specifically the 70-kDa form of Kv4.2 found in these cells (Fig. 7A). A parallel immunofluorescence experiment revealed that in cells permeabilized to allow for access of this antibody to its cytoplasmic epitope, no staining was observed to cells expressing Kv4.2 alone, even though robust staining was observed to the same cells with a phosphorylation-independent anti-Kv4.2 antibody (Fig. 7B). However, robust CT-PKA staining was observed in COS-1 cells expressing Kv4.2 and KChIP2 (Fig. 7C). Interestingly, the CT-PKA staining was predominantly associated with the plasma membrane, such that the Golgi-associated intracellular pools of Kv4.2 were not stained by CT-PKA (compare the left and right panels of Fig. 7C). These data suggest that the 70-kDa form of Kv4.2 is phosphorylated at Ser-552 and that this modification may be spatially restricted to the cell surface pool of Kv4.2.
Phosphorylation of Kv4.2 at Ser-552 has been observed previously (45) in adult rat hippocampus in situ. To determine whether PKA phosphorylation at Ser-552 is associated with the cell surface Kv4.2 in neurons, as it is in COS-1 cells, we doublestained cultured hippocampal neurons with the phospho-specific CT-PKA antibody and a phosphorylation-independent anti-Kv4.2 antibody. As shown in Fig. 7D, interneurons were robustly stained with both antibodies, and the staining patterns co-localized precisely throughout the soma and dendrites. Consistent with this robust CT-PKA immunofluorescence staining, lysates prepared from cultured neurons contained a predominant 70-kDa population of Kv4.2 (Fig. 7E). These results together show that Kv4.2 in neurons, and in COS-1 cells co-expressing either KChIPs1-3, exhibits similar characteristics (cell surface localization and phosphorylation at Ser-552, 70 kDa).
Given this tight coupling between Ser-552 phosphorylation and Kv4.2 cell surface expression, we next examined whether eliminating the PKA phosphorylation site at Ser-552 would affect Kv4.2 trafficking. Surprisingly, eliminating the CT-PKA phosphorylation site with an S552A point mutation did not eliminate the 70-kDa band of Kv4.2 observed upon KChIP2 co-expression (Fig. 8A) nor the ability of KChIP2 to induce trafficking of Kv4.2 to the plasma membrane (Fig. 8B). A similar mutation at the other consensus Kv4.2 PKA site (T38A, not shown) or the double T38A/S552A mutation also did not eliminate the KChIP2-dependent appearance of the higher M r Kv4.2 band on immunoblots (Fig. 8A) or increased Kv4.2 surface expression (Fig. 8C). A number of other phosphorylation site mutants (extracellular signal-regulated kinase: T602A, T607A, and S616A; and casein kinase II: S502A) were also not affected in their trafficking response to KChIP2 co-expression (data not shown). These results suggest that although phosphorylation of Kv4.2 at Ser-552 correlates with KChIP-induced surface expression, phosphorylation at this site does not in itself regulate intracellular trafficking of Kv4.2.
The Cytoplasmic N Terminus of Kv4.2 Contains an Atypical Retention Motif-Certain potassium channel subunits contain dibasic ER retention signals, in the form of Arg-Xaa-Arg, that are masked upon assembly with auxiliary subunits (50). Kv4.2 contains an RKR sequence (amino acids [35][36][37] in the cytoplasmic N terminus. Deletions in the N-terminal domain of Kv4.2 eliminating the RKR motif yield KChIP-independent increases in Kv4.2 currents (51). As KChIPs bind to the N terminus of Kv4.2, they could exert their trafficking through masking of this dibasic motif. To test this model, we generated an RKR-AAA triple point mutant to eliminate this putative retention signal. However, this mutant was indistinguishable from wildtype Kv4.2 in both its ER localization in the absence of KChIP co-expression and the KChIP-induced trafficking to the cell surface (Supplemental Material Fig. 1). These data suggest that the RKR sequence in the cytoplasmic N terminus of Kv4.2 does not act as an ER retention signal.
We next examined a deletion mutant lacking the 29 amino acids immediately following the initiation methionine (Kv4.2⌬2-31). This mutant was efficiently expressed on the surface of COS-1 cells in the absence of KChIP co-expression (Fig. 9A). Moreover, this mutant, unlike wild-type Kv4.2, was recognized by the CT-PKA antibody in cells lacking KChIPs (Fig. 9B). These data suggest that phosphorylation at Ser-552 is not a consequence of KChIP association per se but occurs in response to plasma membrane localization. Kv4.2⌬2-31 also exhibited heterogeneous electrophoretic mobility on SDS gels, although the ratio of the upper band to the lower band was less than that observed in cells expressing wild-type Kv4.2 and KChIP2 (Fig. 9C).

KChIPs Dramatically Impact the Molecular Properties of Kv4
␣ Subunit Polypeptides-We show here that in addition to their previously reported effects on the gating and amplitude of Kv4 currents, KChIP co-expression dramatically changes the molecular properties of Kv4.2 ␣ subunit polypeptides. Taken together, our data indicate that in the absence of KChIPs, Kv4.2 tends to misfold and/or aggregate in the ER, as evidenced by ER localization, lack of cell surface expression, and insolubility in nonionic detergents. Moreover, Kv4.2 exhibits a markedly shorter half-life in the absence of KChIP co-expression, consistent with its misfolded and ER-retained phenotype. In the absence of KChIP co-expression, Kv4.2 also exists in a hypophosphorylated state relative to that observed for native brain Kv4.2, as evidenced by both a reduced M r on SDS-PAGE, and a lack of immunoreactivity with a phospho-specific Ser-552 CT-PKA site antibody. However, upon co-expression of KChIPs1-3 in COS-1 cells, each of these aberrant molecular attributes of the Kv4.2 ␣ subunit is rescued. Thus, KChIPs can profoundly alter the biophysical, molecular, and cell biological characteristics of co-expressed Kv4.2 channels to better approximate their native properties in excitable cells.
By what mechanism do KChIPs1-3 exert these dramatic effects on Kv4.2? KChIPs were originally identified as Kv4interacting proteins in a yeast two-hybrid screen using the cytoplasmic N terminus of Kv4.3 as bait (16). Our analyses of the effects of N-terminal deletion mutants of Kv4.2 mutants on trafficking reveal that in addition to providing the KChIP binding domain, the N terminus contains the signal for ER retention/retrieval. One attractive possibility for such a signal is a canonical N-terminal RKR motif (Kv4.2 amino acids 35-37). Such motifs have been shown to regulate the trafficking of K ATP (50) and cystic fibrosis transmembrane regulator (52) channels, and ␥-aminobutyric acid, type B, receptors (53). Previous studies had shown that a large N-terminal deletion (⌬2-40) eliminating the RKR sequence caused an increase in Kv4.2 current amplitude and changed subcellular localization (51). However, we found that site-directed mutation of the RKR motif of Kv4.2 in itself did not lead to release from the ER. However, deletions in the relatively hydrophobic N terminus of Kv4.2 upstream of the RKR motif released the ER retention and allowed for KChIP-independent cell surface expression and CT-PKA phosphorylation of Kv4.2. This suggests that in the absence of bound KChIPs, the hydrophobic domains of multiple Kv4.2 subunits could oligomerize, leading to aggregation, misfolding, and subsequent recognition by the quality control machinery of the ER resulting in retention and degradation. Such a model would be consistent with the observed insolubility of Kv4.2 in nonionic detergents. KChIP binding could lead to the masking of these hydrophobic sequences and allow for successful rescue from the trapped, detergent-insoluble Kv4.2 phenotype. Consistent with this model is the observation that the amplitude of Kv4.2 currents progressively increased with the increased size of deletions in the Kv4.2 N terminus (51). Such an observation is inconsistent with a mechanism based on a discrete retention signal (such as an RKR motif) which would be predicted to lead to all-or-none effects, and more consistent with an broad motif, such as represented by the extended hydrophobic sequences in the highly conserved N termini of Kv4 ␣ subunits, as the primary determinant of Kv4 channel trafficking.
Heterogeneity in KChIP Effects on Kv4.2 Trafficking-It is somewhat surprising that KChIPs1-3 induced such dramatic effects on Kv4.2, whereas co-expression of the highly related KChIP4a did not. KChIP4a also exerts distinct effects on the gating of co-expressed Kv4 channels, due to the presence of a unique N-terminal domain not found in KChIPs1-3 or in other KChIP4 splice variants (23). Our immunofluorescence staining showed that unlike KChIPs1-3, which exhibited a diffuse cytoplasmic localization, KChIP4a was localized near or at the ER in both the absence and presence of Kv4.2. Preliminary experiments indicate that KChIP4 splice variants (e.g. KChIP4b) that lack the N-terminal KIS domain (23) exhibit properties more similar to KChIPs1-3 than to KChIP4a, 3 suggesting that this domain may confer KChIP4a-specific effects on both channel gating and trafficking. However, both the lack of observed effects of KChIP4a co-expression and the KChIP4a inhibition of the effects of other KChIPs, to rescue the ERretained and hypophosphorylated phenotype of Kv4.2, could be indirect and simply due to competition of inactive KChIP4a subunits with active KChIPs1-3 for the common KChIP-binding site on the N terminus of Kv4.2.
Our data suggest that Kv4 channels containing only KChIP4a subunits would be inefficiently trafficked to the cell surface of mammalian cells. However, heteromeric assembly of channel complexes containing KChIP4a with other KChIPs yields channels with intermediate gating properties (23) as seen previously for Kv␤ subunits (54). Moreover, our data show plasma membrane-associated KChIP4a in certain cells co-expressing KChIP2 and Kv4.2, suggesting that co-assembly of KChIP4a with Kv4.2 and other KChIPs may allow for surface expression of KChIP4a-containing channels in mammalian cells. This suggests that heteromeric assembly of KChIPs results in channels with intermediate trafficking characteristics, similar to the effects of subunit composition on trafficking of Kv1 ␣ subunits (39). Our staining of cultured hippocampal neurons with an antibody to the core domain of KChIP4 suggests that Kv4 channel complexes with mixed KChIP composition can exist in the plasma membrane of native excitable cells.
Kv4.2 Phosphorylation and Trafficking-The surface expression of Kv4.2 and phosphorylation at Ser-552 were strongly correlated by a number of criteria. Both events were dependent on co-expression with KChIP1-3 and exhibited identical KChIP dose dependence. KChIP4a co-expression did not induce either surface expression or S552A phosphorylation and inhibited the ability of KChIPs1-3 to induce these events with a similar dose dependence. Moreover, staining of cells with the Ser-552 phospho-specific CT-PKA antibody was specific for the plasma membrane-associated pool of Kv4.2. However, Kv4.2 N-terminal deletion mutants that exhibited KChIP-independent trafficking to the cell surface were also phosphorylated at Ser-552, showing that Ser-552 phosphorylation correlated better with cell surface expression than for KChIP association per se. These observations together led us to originally hypothesize that phosphorylation at Ser-552 acts as a binary molecular switch regulating Kv4.2 trafficking. However, analysis of the S552A point mutant showed that this was clearly not the case. Given the pronounced correlation between Kv4.2 surface expression and phosphorylation at Ser-552, phosphorylation at this site must be spatially defined. As delivery of Kv4.2 to the plasma membrane appears to be required for phosphorylation at this site, association of cell surface Kv4.2 with a plasma membrane-localized pool of PKA may be required for Ser-552 phosphorylation. Interestingly, in the mammalian hippocampus, spatially restricted subpopulations of dendritic Kv4.2-displayed CT-PKA immunoreactivity exist, suggesting a highly localized, presumably synaptic input-specific regulation of Ser-552 phosphorylation (45). Recent studies have shown that phosphorylation at Ser-552 is also necessary for the modulation of Kv4.2 gating kinetics by PKA and that KChIPs (in this case KChIP3) are required for the PKA effects (25). These studies suggest a complex, spatially regulated relationship between the functional characteristics of Kv4.2, KChIP interaction, Ser-552 phosphorylation, and synaptic activity.
Our experiments have also definitively shown that phosphorylation at Ser-552 is not responsible for the KChIP-induced and phosphorylation-dependent increase in the M r of Kv4.2 on SDS gels. In the course of these experiments, we have also eliminated consensus Kv4.2 phosphorylation sites at Thr-38 (PKA), Thr-602, Thr-607, Ser-616 (extracellular signal-regulated kinase), and Ser-502 (casein kinase II) as contributing to the changes that yield KChIP-induced increases in the M r of Kv4.2. Phosphorylation of Kv4.2 at other sites(s) presumably underlies the alkaline phosphatase-sensitive, KChIP-dependent shift in M r that we observed to correlate strongly with surface expression, and increased stability, of Kv4.2. Identification of this site will lead to insights into the regulatory switch whereby Kv4.2 trafficking is regulated by phosphorylation.
Biological Role of KChIPs in Regulating Kv4.2-Our results suggest that KChIPs dramatically impact not only the gating and trafficking of Kv4.2 channels but also diverse aspects of the molecular characteristics of the Kv4.2 polypeptide itself. Moreover, we found extensive co-localization of endogenous KChIPs and Kv4.2 in cultured neurons, where Kv4.2 exhibits molecular properties consistent with KChIP association. Are KChIPs the primary determinant of Kv4.2 expression in mammalian neurons, cardiac myocytes, and other excitable cells? Studies in heart provide the clearest evidence for a primary role of KChIPs in determining the level of expression of Kv4 channels. A steep gradient of the Kv4-encoded I to exists across human and canine ventricle (55). Interestingly, this functional gradient is not established by the level of Kv4 mRNA but by a parallel gradient in expression of KChIP2, the major cardiac KChIP isoform (56). A similar dependence on KChIP, not Kv4, expression levels was observed for functional expression of Kv4 channels in gastrointestinal smooth muscle (57). Moreover, mice lacking KChIP2 exhibit a complete loss of the cardiac Kv4-encoded I to , leading to prolonged action potentials in ventricular myocytes and susceptibility to cardiac arrhythmias (26). The fact that it is the expression level of KChIPs, and not the pore-forming Kv4 subunits, that determines the level of the Kv4-encoded current in diverse excitable cells is consistent with our studies demonstrating that Kv4 channels require KChIPs for diverse aspects of their native form, not the least of which is efficient trafficking the cell surface.
The role of KChIPs in determining the functional expression of Kv4-encoded currents in mammalian neurons is less clear. Kv4.2 mRNA abundance and A-type potassium current amplitudes are linearly related in basal ganglia and basal forebrain neurons (11), whereas A current levels, and pacemaker frequency, of individual dopaminergic neurons in substantia nigra appears to mirror both Kv4.3 and KChIP3 expression level (58). The situation in mammalian neurons may be further complicated by the presence of numerous other proteins that can modulate Kv4 channels in heterologous systems, such as the CD26-related dipeptidyl aminopeptidase-like protein DPPX (59), frequenin/NCS-1 (60, 61), Kv␤2 (62), and PSD-95 (30). In our experiments we did not observe any effects of co-expression of either Kv␤2 or PSD-95 on any aspect of Kv4.2 trafficking.
Moreover, we found that in cultured hippocampal neurons the subcellular localizations of Kv4.2 and PSD-95, in stark contrast to Kv4.2 and KChIPs, do not overlap (Supplemental Material Fig. 3). Although the role of these other putative Kv4 interacting proteins is not yet clear, our studies have shown that KChIP co-expression leads to diverse and fundamental changes in the molecular properties of Kv4.2 that confer the characteristics typical of native Kv4.2 channels in neurons and other excitable cells.