Auxiliary KChIP4a Suppresses A-type K+ Current through Endoplasmic Reticulum (ER) Retention and Promoting Closed-state Inactivation of Kv4 Channels*

Background: Compared with other auxiliary KChIPs that enhance Kv4 current, KChIP4a inhibits Kv4 function. Results: We identified an ER retention motif and an adjacent VKL motif within the KChIP4a N terminus that reduces Kv4.3 surface expression and promotes closed-state inactivation (CSI), respectively. Conclusion: ER retention and CSI enhancement are two distinct mechanisms by which the N terminus of KChIP4a suppresses Kv4 function. Significance: This study provides mechanistic insight into auxiliary KChIP4a-induced inhibition of A-type Kv4 current. In the brain and heart, auxiliary Kv channel-interacting proteins (KChIPs) co-assemble with pore-forming Kv4 α-subunits to form a native K+ channel complex and regulate the expression and gating properties of Kv4 currents. Among the KChIP1–4 members, KChIP4a exhibits a unique N terminus that is known to suppress Kv4 function, but the underlying mechanism of Kv4 inhibition remains unknown. Using a combination of confocal imaging, surface biotinylation, and electrophysiological recordings, we identified a novel endoplasmic reticulum (ER) retention motif, consisting of six hydrophobic and aliphatic residues, 12–17 (LIVIVL), within the KChIP4a N-terminal KID, that functions to reduce surface expression of Kv4-KChIP complexes. This ER retention capacity is transferable and depends on its flanking location. In addition, adjacent to the ER retention motif, the residues 19–21 (VKL motif) directly promote closed-state inactivation of Kv4.3, thus leading to an inhibition of channel current. Taken together, our findings demonstrate that KChIP4a suppresses A-type Kv4 current via ER retention and enhancement of Kv4 closed-state inactivation.

Surface expression and dynamic gating are fundamental properties of rapidly inactivating (A-type) Kv4 potassium channels that play a critical role in regulating firing frequency and shaping action potential waveforms (1,2). Binding of auxiliary ␤-subunits can change the expression of Kv4 channels at the membrane surface as well as their intracellular trafficking and gating kinetics (3)(4)(5)(6)(7). Cytosolic Kv channel-interacting proteins (KChIPs), 2 a class of ␤-subunits, co-assemble with pore-forming Kv4 ␣-subunits to encode the somatodendritic A-type K ϩ current (I SA ) in neurons (3, 8 -10) and the transient outward current (I to ) in cardiac myocytes (11,12). Both I SA and I to activate at subthreshold membrane potentials, inactivate rapidly, and recover quickly from inactivation (1). Neuronal I SA plays a critical role in regulating dendritic excitability, somatodendritic signal integration, and long term potentiation (13)(14)(15)(16)(17), whereas cardiac I to is the major current for fast phase repolarization of the cardiac action potential (18,19). Genetic disruption of human Kv4 function causes temporal lobe epilepsy and spinocerebellar ataxias as well as cardiac disorders, such as atrial fibrillation and Brugada syndrome (20 -25), demonstrating the essential role of Kv4 channels in physiology and pathology.
Auxiliary KChIP1 to -4 subunits, belonging to the neuronal calcium sensor superfamily, consist of a conserved C-terminal core domain with four EF-hand-like calcium binding motifs (26) and a variable N-terminal domain that has been proposed to mediate diverse modulation on Kv4 function (27)(28)(29)(30). Alternative splicing of the four KChIP genes generates a large number of variants with distinct N-terminal domains, making the KChIP class the most diverse among the neuronal calcium sensor protein family (31). Co-expression of different KChIPs with Kv4 in heterologous expression system leads to different effects on Kv4 channel function. The majority of the KChIP isoforms dramatically increase Kv4 surface expression and peak current, slow down fast inactivation, and accelerate the recovery from inactivation (3,(32)(33)(34). Consistently, many physiological studies have shown that the current amplitudes and gating kinetics of native Kv4 channels (both I SA and I to ) are primarily determined by the expression levels of KChIPs (8,10,11,35,36). It is of interest that the KChIP4 splice variant KChIP4a, which shares a conserved C-terminal core domain with other KChIPs, functions as an inhibitory subunit and exhibits a distinct modulation on Kv4 channel surface expression and gating kinetics via its unique N-terminal domain (4,(37)(38)(39). The first N-terminal 34 residues of KChIP4a, previously known as the KIS (Kv channel inactivation suppressor) domain, slow Kv4 channel activation, disrupt fast inactivation of opened channels, and abolish its core-mediated enhancement of Kv4 current and acceleration of recovery from inactivation (4). However, the mechanisms by which the KChIP4a N terminus functions as a specific Kv4 channel inhibitory domain (KID) to suppress A-type potassium current remains unknown.
In the present study, we show that auxiliary KChIP4a suppresses Kv4.3 function via an alternative N-terminal KID, which can overcome the positive modulatory effect on Kv4.3 channel function by KChIP1. Using confocal imaging, we find that the KID alone is sufficient to retain proteins in the endoplasmic reticulum (ER), and its retention ability is transferable and prominently dependent on its flanking location within the protein sequence. A novel ER retention motif (N-terminal residues 12-17, LIVIVL) within the KID overrides the KChIP4a core domain-mediated enhancement of Kv4.3 surface expression, leading to the reduction of peak current. Further experiments reveal a VKL motif (N-terminal residues 19 -21), adjacent to the LIVIVL motif, that can also suppress Kv4.3 current by facilitating Kv4.3 closed-state inactivation (CSI). Taken together, our findings demonstrate that ER retention and CSI enhancement are two distinct mechanisms by which KChIP4a suppresses Kv4 channel function.

EXPERIMENTAL PROCEDURES
Plasmid Construction-For confocal imaging experiments, cDNAs of KChIP4a and KChIP4a mutants were subcloned into a pEGFP-N1 or pEGFP-C2 vector. DsRed2-ER is a red fluorescent protein tagged with the signal sequence from calreticulin at its N terminus and a KDEL sequence at its C terminus (Clontech). The C-terminal CAAX signal of H-Ras (GCMSCKCVLS), which targets Ras proteins to the plasma membrane (PM), was added to the C terminus of EGFP to construct a PM-targeting EGFP vector (pEGFP-CAAX-N1). For biochemistry experiments, cDNAs of Kv4.3, KChIP1, KChIP4a, and KChIP4a mutants were cloned into a pcDNA3.1 vector. For electrophysiological experiments, wild-type (WT) or mutant cDNA constructs of either Kv4.3 or KChIPs were transferred into a pBluescript KSM vector.
Confocal Microscopy and Imaging Analysis-For confocal imaging experiments, HEK 293 cells were reseeded on glass coverslips coated with poly-D-lysine for detection and transfected using Lipofectamine 2000 (Invitrogen) following the manufacturer's instructions. After 24 -48 h of transfection, cells were washed twice with phosphate-buffered saline (PBS) containing 1 mM MgCl 2 and 0.1 mM CaCl 2 and fixed in 4% paraformaldehyde at 4°C for 15 min. Images were obtained using a confocal microscope (FV1000, Olympus). For staining of localization in the ER, HEK 293 cells were co-transfected with KChIP4a-EGFP or KChIP4a mutants-EGFP and DsRed2-ER. FV1000 Viewer software (Olympus) was used to calculate Pearson's correlation coefficient values for the co-localization of the fluorescence signals.
Electrophysiological Recordings-For whole-cell patch clamp recording in HEK293 cells, currents were recorded at room temperature using the EPC 10 UBS amplifier with PatchMaster software (HEKA Electronics). Patch pipettes were pulled from borosilicate glass and fire-polished to a resistance of 2-4 megaohms. The bath solution contained 135 mM NaCl, 2.5 mM KCl, 10 mM HEPES, 1 mM MgCl 2 , 1.5 mM CaCl 2 , and 10 mM glucose at pH 7.4, and the pipette solution contained 135 mM potassium gluconate, 10 mM KCl, 10 mM HEPES, 1 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM EGTA at pH 7.3.
For two-electrode voltage clamp recordings in oocytes, all cRNAs were transcribed in vitro from linearized plasmids in pBluescript KSM vectors by using the T3 mMESSAGE Machine Kit (Ambion). Xenopus laevis oocytes (stage V-VI) were selected and injected with 46 nl of cRNA solution, containing 0.5-5.0 ng of the selected cRNA, using a microinjector (Drummond Scientific). Oocytes were then kept at 17°C in ND96 solution (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 , 5 mM HEPES, pH 7.4, adjusted with NaOH). 24 h after injection, oocytes were impaled with two microelectrodes (0.5-1.0 megaohms) filled with 3 M KCl in a 40-l recording chamber. Currents were recorded in ND-96 solution at room temperature (22 Ϯ 1°C) using a GeneClamp 500B amplifier (Axon Instruments).
Data were acquired using PatchMaster software (HEKA Electronics) and digitized at 1 kHz with an LIH 8 ϩ 8 computer interface (HEKA Electronics). OriginPro version 8.6 (Origin-Lab) was used to analyze the data. To measure the peak current amplitudes and the kinetics of open-state inactivation, the currents were evoked by a 2-s depolarizing pulse to ϩ40 mV from a holding potential of Ϫ100 mV. The time constants of macroscopic inactivation were obtained by curve fitting with a single exponential function. The peak conductance-voltage (G-V) relationship was derived from peak current amplitudes evoked by depolarizing steps from Ϫ100 to ϩ60 mV at 10-mV increments, and the calculation was based on the equation, G ϭ I/(V Ϫ V rev ), where I is the peak current amplitude at the test potential V, and V rev is the reversal potential. Steady-state inactivation (SSI) was assessed by determining the peak current amplitudes at ϩ40 mV after 1-, 5-, or 10-s prepulses ranging from Ϫ120 to 0 mV. The voltage dependence of steady-state activation (G/G max ) and inactivation (I/I max ) were fitted to the following single Boltzmann relationship, y ϭ 1/(1 ϩ exp((V Ϫ V1 ⁄ 2 )/k)), where V is the test potential, V1 ⁄ 2 is the potential for half-maximal activation or inactivation, and k is the corresponding slope factor. CSI was monitored with a double-pulse protocol in which the current was evoked by a test pulse of ϩ40 mV from Ϫ100 mV (I pre ) and from subsequent Ϫ50-mV prepulses of variable durations from 5 ms to 10.4 s (I post ). Normalized currents for CSI were determined as the ratio of I post to I pre . The kinetics of CSI were obtained by fitting the normalized current amplitudes as a function of prepulse duration. All holding potentials were Ϫ100 mV in this study unless specified.
Cell Surface Biotinylation and Western Blotting Assay-Transfected HEK 293 cells were washed three times with icecold PBS and incubated for 1 h at 4°C with EZ-Link Sulfo-NHS-SS-biotin (0.5 mg/ml; Pierce) to biotinylate cell surface proteins. Excess biotin was quenched by incubating the cells for an additional 10 min with TBS containing 100 mM glycine. Cells were lysed with lysis buffer (150 mM NaCl, 20 mM Tris, 1% Triton X-100, 1% sodium deoxycholate, 0.1% SDS, 10 mM EDTA, and proteinase inhibitor mixture (Roche Applied Science), pH 8.0) at 4°C for 30 min. The cell lysates were then centrifuged at 13,000 ϫ g for 15 min to yield the protein extract in the supernatant. One fraction containing 200 g of protein was incubated with 20 l of neutravidin beads (Pierce) for 4 h at 4°C, and the other fraction was prepared as total protein. After incubation, the bead-binding surface proteins were washed four times with cell lysis buffer and eluted with loading buffer containing DTT. The total and biotinylated proteins were both subjected to Western blot analysis.

KChIP4a Suppresses Kv4.3 Function via Its N-terminal KID-
To investigate the underlying mechanism by which auxiliary KChIP4a inhibits Kv4 function, we started by examining the effect of KChIP4a coexpressed with Kv4.3 on the channel current in both mammalian cells and Xenopus laevis oocytes. By sequence alignment, KChIP4a reveals a unique N terminus, consisting of the first 34 amino acids, that has been shown to function as a KID to inhibit the channel activity (Fig. 1A). Whole-cell patch clamp recordings in HEK 293 cells and twoelectrode voltage clamp recordings in X. laevis oocytes both showed that KChIP4a co-expression resulted in a reduction of Kv4.3 current (Fig. 1, B-D). In contrast, deleting the first 34 residues of the KChIP4a N terminus (KChIP4a⌬2-34, also named KChIP4a core) led to a dramatic increase in the current amplitude ( Fig. 1, B-D). The inhibitory effect of KChIP4a KID on Kv4 current was further confirmed by fusing the KID to either the C terminus of WT Kv4.3 (Kv4.3-KID) or the N terminus of Kv4.3 N-terminal truncation mutant (KID-Kv4.3⌬2-24) (Fig. 1E). These results demonstrate that the inhibitory effect of KChIP4a on Kv4 is indeed mediated by its N-terminal KID.
In neurons, KChIP4a co-exists with other KChIPs (10, 32), suggesting that KChIP4a may compete with other KChIPs to modulate Kv4 expression and gating (40). To test this hypothesis, we co-injected varying amounts of KChIP4a cRNAs with a fixed amount of Kv4.3 and KChIP1 cRNAs into Xenopus laevis oocytes and recorded the currents ( Fig. 2A). As shown in Fig.  2B, with a fixed amount of KChIP1, increasing KChIP4a expression yielded a dose-dependent inhibition of Kv4.3 current amplitude. Consistent with previous reports (4), increasing amounts of KChIP4a gradually slowed the macroscopic openstate inactivation of Kv4.3 current (Fig. 2C). This dose-dependent, KChIP4a-mediated inhibition of Kv4.3-KChIP1 current was further confirmed by parallel biochemical experiments. We observed that total Kv4.3 protein expression was significantly reduced when co-expressed with KChIP4a (Fig. 2, D and E). However, KChIP4a co-expression resulted in a greater decrease in cell surface Kv4.3 protein, as compared with total expression (Fig. 2, F and G). Together, these results show that the N-terminal KID of KChIP4a acts as a dominant negative regulator to suppress the function of Kv4.3-KChIP1 channel complexes by inhibiting Kv4.3 expression as well as forwarding trafficking to the cell surface.
The ER Retention Capacity of the KID Is Transferable and Dependent on Its Location-The above results show that the N-terminal KID of KChIP4a results in the decreased anterograde trafficking of Kv4 channels, and previous studies have suggested that the KChIP4a N terminus is required for ER localization of KChIP4a in heterologous cells (32,37). To find potential ER retention motifs within the KID, we began testing whether the retention ability of the KID is transferable and dependent upon its location within the protein sequence. We created a series of constructs with insertions of the KID at the proximal N terminus, the intermediate region, and the C-terminal end of the entire fusion protein. As a positive control, N-terminal attachment of full-length KChIP4a to EGFP (KChIP4a-EGFP) completely altered the subcellular distribution of EGFP and led to co-localization with the ER marker DsRed2-ER (Fig. 3A). By contrast, the KID deletion mutant of KChIP4a (KChIP4a⌬2-34-EGFP) showed a diffuse distribution throughout the cytoplasm and nucleoplasm of HEK 293 cells (Fig. 3B). Transplanting the KID to the C terminus of EGFP (EGFP-KID) was sufficient to retain EGFP in the ER (Fig.  3C), whereas inserting the KID into the middle part of intact protein (EGFP-KChIP4a) disrupted its ER retention ability (Fig.  3D). To establish whether the KID could redirect other membrane proteins to the ER, we fused the KID with EGFP to the C terminus of CD4 (CD4-EGFP-KID). Our results showed that CD4 fused with the KID caused localization of the chimeric proteins to the ER (Fig. 3E). However, fusing CD4 to the N terminus of KChIP4a (CD4-KChIP4a-EGFP, with equivalent insertion of the KID into the intermediate region of the intact protein) had no effect on trafficking and maintained the ability of CD4 proteins targetting to the PM instead of being retained in the ER (Fig. 3F). Quantitative analysis of fluorescence intensity for co-variance of WT KChIP4a or KChIP4a mutants against the ER marker DsRed2-ER further confirmed the effect of ER retention mediated by its KID (Fig. 3G). These results indicate that the KID functions as an ER retention signal when located at either the N or C terminus of a protein but not when buried in the intermediate region.
To understand the mechanism underlying the KID-mediated ER retention, we utilized a CAAX signal from the H-Ras proteins that functions as a membrane anchor to evaluate the retention ability of KChIP4a and its mutants. The CAAX signal was appended to the C terminus of KChIP4a-EGFP fusion proteins. We then tested the effect of the KID on CAAX signalmediated forward trafficking to the cell surface. The result showed that the intact KID overcame the CAAX signal-mediated PM translocation (Fig. 5A). Consistent with our previous results shown in Figs. 3B and 4, the mutants KChIP4a⌬2-34-CAAX, KChIP4a⌬2-15-CAAX, KChIP4a 11-14A-CAAX, and KChIP4a 15-18A-CAAX lost their ability to stay in the ER as illustrated by their distribution at the PM (Fig. 5, B, D-F, and I), whereas the mutants KChIP4a⌬2-10-CAAX, KChIP4a 19-22A-CAAX, and KChIP4a⌬23-34-CAAX retained their ER retention ability (Fig. 5, C and G-I). These results demonstrate that residues 11-18 (VLIVIVLF) of the KID function as an ER To determine whether the VLIVIVLF (residues 11-18) is sufficient for ER retention, we replaced the whole KID (34 amino acids) with VLIVIVLF and appended it to the KChIP4a core. Our results showed that VLIVIVLF alone is not only necessary but also sufficient to retain proteins in the ER, despite the presence of a CAAX signal (Fig. 6, A, F, and K). To further circumscribe the retention motif, we replaced the KID with VLIVI (residues 11-15), LIVIV (residues 12-16), IVIVL (residues 13-17), or VIVLF (residues 14 -18). The VLIVI (residues 11-15) did not alter the cytoplasm and nucleoplasm-diffused distribution of the KChIP4a core domain, as compared with the core expressed alone (Fig. 6, B and K). The PM distribution of VLIVI-core-CAAX fusion proteins was also similar to the core-CAAX (KChIP4a⌬2-34-CAAX) (Fig. 6, G and K). Either LIVIV (residues 12-16) or IVIVL (residues 13-17) alone was sufficient to trap the chimeric proteins in the ER, with or without the CAAX signal (Fig. 6, C, D, H, I, and K). VIVLF (residues 14 -18) allowed effective targeting and retention within the ER (Fig. 6, E and K) but was not sufficient to prevent the CAAX signal-mediated PM localization of the chimeric proteins (Fig. 6, J and K). To further narrow down residues critical for ER retention, we replaced the KID with different combinations of four amino acid sequences, such as VLIV (residues 11-14), LIVI (residues 12-15), IVIV (residues 13-16), VIVL (residues 14 -17), and IVLF (residues 15-18), respectively. All chimeras carrying the four amino acid sequences, except VLIV (residues [11][12][13][14], stayed in the ER (Fig. 7, A-E and K), whereas those chimeras fused with CAAX signal traveled to the cell surface (Fig. 7, F-J and K), suggesting that the four-amino acid ER retention motifs are not strong enough to overcome the PM translocation mediated by the CAAX signal. These results reveal a hydrophobic and aliphatic motif, consisting of six residues 12-17 (LIVIVL), that functions as an ER retention signal responsible for KChIP4a localization in the ER.
Enhanced CSI of Kv4.3 by the N-terminal KID of KChIP4a-To examine whether the ER retention signal within the N-terminal KID of KChIP4a affected total and surface expression of Kv4.3, surface biotinylation and Western blot were carried out on HEK 293 cells with Kv4.3 co-expressed with WT or mutated KChIP4a. The results showed that mutating residues 11-18 (11-14A and 15-18A) relieved ER retention and reversed the inhibitory effect of KChIP4a on total and surface expression of Kv4.3 (Fig. 8, A and C). The reduction of total Kv4.3 expression resulted from the decreased expression of KChIP4a, as compared with the KChIP4a⌬2-34 (Fig. 8, E  and F), suggesting that the ER retention signal within the KID accelerates the rate of protein degradation. Quantitative analysis showed greater changes of Kv4.3 proteins in surface expression, as compared with total expression (Fig. 8, B and  D), indicating that the KID inhibits the surface expression of Kv4.3 via reducing protein stability as well as anterograde trafficking. Electrophysiological recordings showed a dramatic increase in Kv4.3 current amplitudes when residues 11-18 were disrupted (Fig. 8G), consistent with their role in mediating Kv4.3 retention in the ER.
In addition to the identified residues 11-18 that are critical for ER retention, we were surprised to find that mutating residues 19 -22 (19 -22A) increased the current density of Kv4.3 without affecting its surface expression (Fig. 8, D and G). Recent pharmacological studies showed that some compounds can inhibit Kv4.3 function by accelerating channel CSI, suggesting that Kv4 inhibition might result from an increased number of unavailable channels that are in closed inactivated states (41)(42)(43)(44)(45). To test the gating effect of KChIP4a, we co-expressed Kv4.3 with the KChIP4a core (KChIP4a⌬2-34) and found that pref-erential CSI was reduced (Fig. 9, A and B). Depolarizing to ϩ40 mV after holding for 10.4 s at Ϫ50 mV resulted in only about 38% of Kv4.3-core complex current being inactivated (Fig. 9B). In contrast, co-expression of Kv4.3 with KChIP4a increased CSI to about 90%, as compared with Kv4.3 alone (ϳ72%) and Kv4.3-core co-expression (ϳ38%) (Fig. 9B and Table 1). The SSI curve for Kv4.3 under a 5-s prepulse had a V1 ⁄ 2 of Ϫ52.0 Ϯ 0.3 mV and a k of 5.5 Ϯ 0.3 mV. The KChIP4a core significantly shifted the inactivation curve (V1 ⁄ 2 ) to the right in a depolarization direction (V1 ⁄ 2 of Ϫ46.4 Ϯ 0.2 mV and k of 4.6 Ϯ 0.2). Consistent with its effect on CSI, the KID overrode the rightshift effect of the core and caused the curve to shift in the hyperpolarization direction (V1 ⁄ 2 of Ϫ58.5 Ϯ 0.5 mV and a k of 5.2 Ϯ 0.4 mV) (Fig. 9C and Table 1). We also examined the conductance-voltage (G-V) curves and found no significant differences (Fig. 9D). These data suggest that the modulation of Kv4.3 inactivation by the KID is dominant over that by the KChIP4a core To determine which part of the KID is involved in modulating Kv4.3 CSI, we generated mutants and found that residues 19 -22 of the KID played a critical role in facilitating the development of CSI and causing a hyperpolarizing shift of SSI (Fig.  10, A and B, and Table 1). The 19 -22A mutant completely eliminated the enhancement effect of the KID on CSI, consis- tent with the effect of the KID deletion mutant (⌬2-34). Further dissection revealed that residues 19 -21, VKL, are responsible for promoting CSI, and residue 21 (leucine) is the most critical (Fig. 10, C and D, and Table 1). A previous study suggested that either the N terminus of KChIP2x (2xNt) or KChIP3x (3xNt) functions like the KID (37). To test the effect of 2xNt or 3xNt on CSI, we replaced the KID with 2xNt or 3xNt, where their amino acid residues 19 -21 are VKV or IAV, respectively. The VKV from 2xNt exerted an effect similar to that of the KChIP4a 21A mutant on CSI and SSI. The IAV from 3xNt, analogous to the KChIP4a 19 -22A or ⌬2-34 mutation, had no effect on promoting CSI (Fig. 10, E and F,   confirming the specific effect of the VKL motif from the KID on CSI. Together, these results demonstrate that the VKL motif of the KID inhibits Kv4.3 function by facilitating the channel CSI.

DISCUSSION
The goal of this study was to understand how auxiliary KChIP4a, which only differs from other KChIPs in the N terminus, inhibits Kv4 function. We found that KChIP4a acts as an inhibitory subunit to suppress Kv4 A-type current (I SA ). The inhibitory property of KChIP4a on I SA is primarily mediated by its unique N-terminal KID. Our findings reveal two distinct mechanisms by which KChIP4a exerts its inhibition on Kv4 function. First, an LIVIVL motif (N-terminal residues 12-17) mediates ER retention of KChIP4a and Kv4 channels; second, the VKL motif (N-terminal residues 19 -21) of the KID directly inhibits Kv4.3 function by increasing the number of channels in the closed inactivated state.
Previous studies on Kv4 modulation by KChIPs have mostly focused on the effects of a single KChIP on Kv4 function.
Because all members of the KChIP family share a high homology in their C-terminal core regions and each can form two interfaces associated with two neighboring Kv4 N-terminal domains (46 -48), this raises the possibility that different KChIPs may compete for common KChIP-binding sites, resulting in a diversity of modulation of Kv4 function by KChIPs. This hypothesis is supported by our observation that KChIP4a inhibits the surface expression and peak current of Kv4 coexpressed with KChIP1 in a dose-dependent and competitive manner by antagonizing the binding of KChIP1 to Kv4 (Fig. 2). Whether these observations imply that Kv4-KChIP complexes in native cells contain heteromeric KChIPs or only one KChIP isoform is presently unknown and deserves further investigation.
Cell surface expression of proteins depends on the delicate balance of forward trafficking and intracellular retention (49). ER retention/retrieval motifs are dominant trafficking signals that are often encoded in individual subunits of multiprotein complexes, thereby providing a negative regulatory mechanism for cell surface protein expression or a quality control mechanism to ensure proper oligomerization or correct folding (50). In most cases, ER retention/retrieval motifs are located in the ␣-subunits of ion channels and can be masked by subunit assembly (51)(52)(53) or antagonized by auxiliary subunits (54,55) or regulatory proteins (56 -58). Some accessory ␤-subunits or regulators contain ER retention motifs and down-regulate surface expression of ␣-subunits (59, 60). In the current study, we

TABLE 1 Inactivation kinetics of Kv4.3 alone and its co-expression with KChIP4a or its mutants
Values are means Ϯ S.E.; n ϭ number of oocytes. Mean data points obtained from the steady-state inactivation protocol with 5-s prepulse were best fitted by (OriginPro version 8.6 software) a standard single Boltzmann curve, in which V1 ⁄ 2 is the half-inactivation potential, and k is the slope factor. Inactivation time constants () of closed-state inactivation were obtained by fitting single exponential function to the decaying phases of Kv4.3 current evoked by ϩ40-mV depolarization after a subthreshold Ϫ50-mV prepulse with variable durations. Remaining current is the fraction of non-inactivated channel current.

Steady-state inactivation
Closed-state inactivation  (34). In addition, we utilized the CAAX signal from H-Ras that functions as a PM targeting signal to evaluate the retention ability of this newfound motif. The H-Ras-CAAX signal uses the exocytosis pathway in a similar fashion to the trafficking pathway of membrane proteins (61). Therefore, our results support that the ER retention motif identified in this study plays a dominant negative role in classical vesicular trafficking pathway, thus leading to the reduction of Kv4 surface expression. Due to its hydrophobic nature, the LIVIVL motif identified and reported here differs from canonical motifs consisting of basic amino acids. A series of hydrophobic ER retention signals have been identified in different parts of proteins, including the N terminus, C terminus, transmembrane segments, and cytoplasmic linker of transmembrane segments (53,(62)(63)(64)(65)(66)(67)(68). Unlike well characterized basic motifs that consist of RXR, RR, and KK, the molecular identity of the hydrophobic motifs still remains to be further explored. The hydrophobic ER retention motifs vary in their location and composition, and the only common feature is membrane adjacency or association. Another interesting observation is that the retention ability of the LIVIVL motif is transferable and dependent on its flanking location within the protein sequence. This feature is similar to the H/KDEL motif that is often located in the C-terminal end of soluble ER-resident proteins, such as Bip (Grp78), calreticulin, endoplasmin (Grp94), protein-disulfide isomerase, etc. (50).
The addition of the KDEL to the C terminus of various proteins can also lead to their ER retention (69 -72). The H/KDEL motif is recognized by the KDEL receptor ERD2, which is mainly localized to the cis-Golgi and targets the KDEL-containing proteins into the retrograde COPI-mediated transport pathway (73,74). However, how the LIVIVL motif causes ER retention of LIVIVL-containing proteins requires further investigation.
Previously, we proposed that the core of KChIPs increases total expression level of Kv4.3 by stabilizing Kv4.3 tetramerization via a clamping action (48). In the current study we show that the N-terminal KID of KChIP4a, a non-core region, exerts an inhibitory effect on total Kv4.3 and KChIP4a by reducing their protein stability (Fig. 8). Considering the ER retention effect of the KID, it is likely that the ER retention motif within the KID causes ER-associated protein degradation of KChIP4a itself and its associated proteins, such as Kv4.3. Kv4 expression and functional properties are regulated by KChIPs as well as dipeptidyl peptidase-like proteins containing DPP6 and DPP10 (5,6). Considerable evidence suggests that native neuronal I SA channels function in macromolecular protein complexes composed of Kv4 pore-forming subunits together with accessory KChIPs and DPP6/10 subunits (75,76). Different from cytosolic KChIPs, the accessory DPP6/10 subunits are transmembrane proteins and selectively increase surface expression of Kv4.2 but not total expression (77). In this study, we demonstrate that the ER retention of the KID is dominant over surface trafficking of multitransmembrane protein Kv4.3, single transmembrane protein CD4, and the secretory CAAX signal from H-Ras protein, so the DPP-mediated augmentation effect on E, CSI of Kv4.3 co-expressed with KChIP4a, core (KChIP4a⌬2-34), 2xNt-core, or 3xNt-core (n ϭ 12-23 oocytes). F, voltage dependence of SSI of Kv4.3 co-expressed with KChIP4a, core (KChIP4a⌬2-34), 2xNt-core, or 3xNt-core (n ϭ 11-13 oocytes).
Kv4 surface expression is likely to be inhibited when co-assembled with KChIP4a.
The molecular mechanism of Kv4 fast inactivation differs significantly from other rapidly inactivating Kv channels, such as Shaker and Kv1.4 (78 -82). Kv4 channels exhibit a preferential CSI, implying that upon voltage-dependent activation, inactivation occurs preferentially from partially activated closed states that precede the open state (82,83). In contrast to N-type and C-type inactivation, the molecular basis of CSI remains to be determined. The KID has been previously characterized as an open-state inactivation suppressor (so-called KIS domain) of Kv4 in heterologous cells and neurons (4,84). Surprisingly, in this study, we found that the KID promotes CSI and shifts the voltage dependence of SSI in a hyperpolarization direction, whereas the core of KChIP4a-mediated Kv4 modulation on CSI and SSI kinetics is opposite to that of the KID (Fig. 9). Moreover, we demonstrated that the existence of the VKL motif (N-terminal residues 19 -21) is responsible for the KID modulation on Kv4 CSI. Therefore, identification of a putative receptor site for the VKL motif binding to Kv4 channels may provide structural insights into the mechanism of Kv4 CSI.
Similar to Kv4 channels, Kv2 channels (Kv2.1 and Kv2.2) also inactivate preferentially from partially activated closed states, and its steady-state inactivation curve exhibits a U-type shape (85). The functional properties of Kv2 are regulated by forming heteromers with the otherwise "silent" ␣-subunits of the Kv5, Kv6, Kv8, or Kv9 subfamily. The modulatory effects of these "silent" ␣-subunits on Kv2 are analogous to those of KChIP4a on Kv4, such as a reduction in peak current amplitude, inhibition of open-state inactivation, acceleration of CSI, and promotion of hyperpolarized shifts in the voltage dependence of SSI (86 -89). Altogether, these findings suggest that bidirectional modulation of Kv4 open-state and closed-state inactivation mediated by the KID may be a highly conserved function among various ion channels and ancillary subunits.
The channel subunit composition is a determinant in shaping the repertoire of A-type K ϩ channels present at the surface of neurons. Differential regulation of auxiliary KChIPs can modify the subunit composition of native Kv4 channel complexes, altering their biophysical and physiological properties as well as neuronal excitability. Recently, an interesting study has demonstrated that activation of noncoding RNA 38A induced by inflammatory stimuli causes a gene-specific alternative splicing shift from KChIP4bl to KChIP4a, leading to an inhibition of Kv4 function as well as altered amyloid production associated with neurodegenerative etiology (90). Therefore, it is possible that the expression level of KChIP4a may vary under pathological conditions that are affected by inflammation, pain, meningitis, multiple sclerosis, Alzheimer disease, Parkinson disease, or stroke-mediated neuronal dysfunction (91), resulting in perturbation of I SA and neuronal excitability. Furthermore, because KChIP4a can directly inhibit Kv4 function by promoting channel CSI, a specific disruption of the interaction between the KChIP4a KID and Kv4 channels aimed at enhancing I SA may provide a new strategy for treat-ing hyperexcitable neurological disorders, such as pain and epilepsy.