Formation of Heteromeric Kv2 Channels in Mammalian Brain Neurons*

The formation of heteromeric tetramers is a common feature of voltage-gated potassium (Kv) channels. This results in the generation of a variety of tetrameric Kv channels that exhibit distinct biophysical and biochemical characteristics. Kv2 delayed rectifier channels are, however, unique exceptions. It has been previously shown that mammalian Kv2.1 and Kv2.2 are localized in distinct domains of neuronal membranes and are not capable of forming heteromeric channels with each other (Hwang, P. M., Glatt, C. E., Bredt, D. S., Yellen, G., and Snyder, S. H. (1992) Neuron 8, 473–481). In this study, we report a novel form of rat Kv2.2, Kv2.2long, which has not been previously recognized. Our data indicate that Kv2.2long is the predominant form of Kv2.2 expressed in cortical pyramidal neurons. In contrast to the previous findings, we also found that rat Kv2.1 and Kv2.2long are colocalized in the somata and proximal dendrites of cortical pyramidal neurons and are capable of forming functional heteromeric delayed rectifier channels. Our results suggest that the delayed rectifier currents, which regulate action potential firing, are encoded by heteromeric Kv2 channels in cortical neurons.


EXPERIMENTAL PROCEDURES
Immunohistochemistry-Immunohistochemistry was performed as described (20). Briefly, rats were anesthetized by the injection of 50 mg/kg sodium pentobarbital (intraperitoneal). Animals were then fixed by transcardiac perfusion of 4% paraformaldehyde in phosphate buffer. The fixed brain was equilibrated in 30% sucrose for cryoprotection. The brain was sectioned into 40-m thick sagittal brain sections using a cryomicrotome. Tissue sections were blocked and permeabilized for 1 h in 10% normal goat serum in phosphate buffer containing 0.3% Triton X-100. Sections were then incubated with the following primary antibodies overnight at 4°C: mouse monoclonal antibody K89 against Kv2.1 (NeuroMab, Davis, CA) (21), rabbit anti-Kv2.2 long polyclonal antibody (APC-120, Alomone Labs, Jerusalem, Israel) (22), or mouse monoclonal antibody K37 against the N terminus of Kv2.2 (NeuroMab) (23). For K37 staining, sections were preincubated in 0.1 M citric acid, pH 8.8, for 30 min at 80°C to facilitate the antibody binding. Sections were washed and probed with Alexa-conjugated secondary reagents available for each species and IgG isotype (Invitrogen). After washing, sections were mounted on glass slides for fluorescence imaging. HEK293 cells were cultured on coverslips coated with poly-L-lysine and transfected with Kv2.1 and/or Kv2.2 long cDNA using Lipofectamine 2000 (Invitrogen). Two days after transfection, HEK293 cells were fixed in 3% paraformaldehyde with 0.1% Triton X-100 in phosphate-buffered saline. Cells were blocked for 1 h in Tris-buffered saline containing 0.15 M NaCl, 4% nonfat dry milk, 0.05% Tween 20, and 20 mM Tris-HCl, pH 7.4. The cells were then incubated with primary antibodies overnight at 4°C. Cells were washed and probed with Alexa-conjugated secondary reagents. Fluorescent images were taken with a charge-coupled device camera installed on a Zeiss Axiovert 200M microscope with 10ϫ, 20ϫ, and 63ϫ lenses and a structured illumination imaging system (ApoTome) using Axiovision software (Carl Zeiss). We obtained optically sectioned images using this system. The monoclonal antibodies were obtained from the University of California Davis/National Institutes of Health NeuroMab Facility of the Department of Neurobiology, Physiology, and Behavior, College of Biological Sciences, University of California (Davis, CA). Colocalization analyses were performed using ImageJ. Pearson's coefficient takes into consideration similarity between shapes in two channels while ignoring the intensities of signals. The intensity correlation quotient was obtained as described (24). Line scan analysis was also performed using ImageJ. Fluorescence intensity was measured in 10-pixel (ϳ1 m) wide lines and is shown as a percentage of the maximum values in each plot. Rat Brain Membrane (RBM) 3 Fraction-Animals were killed by rapid decapitation (25). The brains were removed and homogenized in ice-cold buffer (320 mM sucrose, 5 mM sodium phosphate, 100 mM NaF, and 1 mM phenylmethylsulfonyl fluoride, pH 7.4) containing protein inhibitor mixture (2 g/ml aprotinin, 1 g/ml leupeptin, 2 g/ml antipain, and 10 g/ml benzamidine). The homogenates were centrifuged at 800 ϫ g for 10 min, and the resultant supernatants were centrifuged at 39,000 ϫ g for 90 min. The pellets (membrane fraction) were resuspended in the same buffer.
Western Blotting-Samples were prepared in SDS sample buffer (2% SDS, 5% 2-mercaptoethanol, 10% glycerol, and 62.5 mM Tris-HCl, pH 6.8). Proteins were separated on 6 or 7.5% SDS-polyacrylamide gels and transferred to nitrocellulose membranes. The membranes were washed; blocked with 4% nonfat dry milk in 150 mM NaCl and 20 mM Tris-HCl, pH 8.0, for 30 min; and then incubated with primary antibodies overnight in the blocking buffer at 4°C. We used mouse monoclonal antibody K89 and the anti-Kv2.2 long antibody to detect Kv2 proteins. The membranes were washed and probed with horseradish peroxidase-conjugated secondary antibodies (KPL, Gaithersburg, MD) according to the manufacturer's instructions. Immunoreactive bands were detected with enhanced chemiluminescence (PerkinElmer Life Sciences) and visualized by exposing the membranes to x-ray films.
Immunoprecipitation-For the large-scale Kv2.1 affinity purification, RBM (50 mg of protein) was solubilized in lysis buffer containing 20 mM Tris-HCl, 0.15 M NaCl, 1 mM phenylmethylsulfonyl fluoride, 1% Triton X-100, and protease inhibitor mixture (2 g/ml aprotinin, 1 g/ml leupeptin, 2 g/ml antipain, and 10 g/ml benzamidine). A K89 affinity column was prepared using a Seize primary immunoprecipitation kit (Thermo Scientific, Rockford, IL). The lysates were incubated with the column resin for 24 h to remove nonspecific binders. The cleared lysates were then applied to either the K89 affinity column or a control column (with normal mouse IgG), followed by incubation for 4 days at 4°C. After extensive washes, bound proteins were eluted in boiling SDS sample buffer for 5 min. The eluents were subjected to SDS-PAGE, and the gels were stained with either silver or Coomassie Brilliant Blue G-255. Mass spectrometry analysis was done at the core facility of the University of Maryland (Baltimore). Briefly, excised bands from Coomassie Blue-stained gels were destained and dried in a speed vacuum concentrator. Dried gel pieces were then trypsinized at 37°C for 18 h. Peptides were extracted into 50% acetonitrile in 5% formic acid and dried in a speed vacuum concentrator. Peptides were desalted and concentrated using a reversed-phase trap column. They were separated in a reversed-phase C 18 column during a 90-min linear gradient of 5-90% acetonitrile/water containing 0.1% formic acid at a flow rate of 300 nl/ml. The eluted peptides were directly sprayed into a Finnigan LCQ ion trap mass spectrometer. Tandem mass spectra were acquired for the most intense peptide ion from the previous mass spectra. The acquired mass spectrometry scans were searched against the IPI Protein Sequence Database using the Sorcerer/Sequest search algorithm. The best hit were selected based on probability/percent of coverage and number of peptides or based on Xcore. For immunoprecipitation, HEK293 cells transiently expressing Kv2 or RBM were solubilized at 4°C for 30 min in lysis buffer. Insoluble materials were removed by centrifugation at 15,000 ϫ g for 15 min at 4°C. The supernatants were incubated with either mouse monoclonal antibody K89 or the anti-Kv2.2 long antibody at 1 g/ml for 2 h at 4°C and then incubated with protein G or A immobilized on agarose beads for 2 h at 4°C with agitation. After extensive washes, bound proteins were eluted and subjected to Western blotting.
Construction of Non-conducting Pore Mutants of Kv2.1 and Kv2.2-Rat Kv2.1 and Kv2.2 in the RBG4 vector were obtained from James Trimmer (University of California, Davis). We found that the clone encodes the genomic sequence of Kv2.2. Two pore mutations (W365C/Y380T) were introduced to Kv2.1 by PCR. Codons TGG (Trp 365 ) and TAC (Tyr 380 ) were mutated to TGC (Cys) and ACC (Thr), respectively. The mutant was cloned into the pIRES-mCherry vector (Clontech). The Kv2.1 W365C/Y380T construct was sequenced in its entirety to ensure that no additional mutations were introduced. Wild-type Kv2 sequences were also cloned into the pIRES-AcGFP vector for patch-clamp experiments.
Electrophysiology and Data Analysis-HEK293 cells were transiently transfected with either wild-type Kv2.1 or wild-type Kv2.2 long (in the pIRES-AcGFP vector) with or without the pore mutants. Outward Kv2 channel currents were recorded using the whole-cell voltage-clamp technique. We chose cells that express comparable levels of green fluorescent protein. Currents were recorded with an MultiClamp 700A amplifier (Molecular Devices, Sunnyvale, CA), sampled at 20 kHz, and filtered at 2 kHz with a series resistance of Ͻ5 megohms. Patch pipettes were pulled from borosilicate glass tubing to give a resistance of 1-2 megohms when filled with the pipette solu-tion (135 mM KCl, 2 mM MgCl 2 , 10 mM EGTA, 10 mM glucose, and 10 mM HEPES, pH 7.3). The extracellular buffer contained 140 mM NaCl, 4 mM KCl, 2 mM CaCl 2 , 2 mM MgCl 2 , 10 mM glucose, and 10 mM HEPES, pH 7.4. All recordings were performed at room temperature (ϳ25°C). The membrane potential was held at Ϫ80 mV and depolarized (or hyperpolarized) from the holding potential of Ϫ80 mV to voltages between Ϫ90 and ϩ70 mV in 10-mV increments for 200 ms. Linear leak components were identified from the hyperpolarizing pulse and subtracted. Peak outward sustained K ϩ currents were measured 100 ms into the pulse at each depolarizing potential.

RESULTS
Two Different Sequences of rat Kv2.2-Kv2.2 mRNA was first cloned from a rat brain cDNA library (9) and identified to encode an 802-amino acid long protein. However, we recently found that there are two mRNA sequences of rat Kv2.2 registered in the NCBI Database. One originated from the cloned cDNA sequence (accession number M77482), and the other from the rat genomic sequence (accession number NM_054000). They differ after nucleotide 2288 (number based on the coding region) because of a single-nucleotide deletion (Fig. 1A). The resultant two protein sequences have either 802 (from the original cDNA) or 907 (from the genomic sequence) amino acid residues. In the previous analyses of Kv2.2 localization in the brain, antibodies raised against the cytoplasmic C terminus of the shorter form (Kv2.2 short ) were used because this region differs substantially from that of Kv2.1 (data not shown). Studies reported that Kv2.2 short exhibits a subcellular localization distinct from that of Kv2.1 (14,26). However, the epitope sequence is present only in Kv2.2 short but not in the novel longer form (Kv2.2 long ) (Fig. 1B). Therefore, we evaluated the subcellular localization of Kv2.2 long in the cerebral cortex of rat using a rabbit antibody whose epitope is specific in Kv2.2 long (Fig. 1B) (22).

Colocalization of Kv2.1 and Kv2.2 long in Rat Brain Neurons-
First, the specificity of the anti-Kv2.2 long polyclonal antibody was verified as described previously (27,28). RBM fractions were prepared and subjected to Western blotting using the anti-Kv2.2 long antibody. As shown in Fig. 2A, a major band was observed at ϳ140 kDa, which is larger than the size expected from the primary amino acid sequence (102 kDa), indicating post-translational modifications of Kv2.2. Indeed, alkaline phosphatase treatment of RBM resulted in a downward shift of the immunoreactive band from 140 to 110 kDa (data not shown), indicating constitutive phosphorylation of Kv2.2 in the brain, like Kv2.1 (29,30). Preincubation with the antigen peptide completely blocked the binding of the antibody, verifying  the specificity of the antibody ( Fig. 2A). To examine whether the anti-Kv2.2 long antibody cross-reacts with Kv2.1, lysates of HEK293 cells solely expressing Kv2.1 or Kv2.2 long were probed with either the mouse anti-Kv2.1 (K89) or anti-Kv2.2 long antibody by Western blotting (Fig. 2B). Rat Kv2.1 and Kv2.2 long exhibit 48% identity in the C termini, but the epitope sequence of the anti-Kv2.2 long antibody is missing in Kv2.1 (data not shown). Likewise, the K89 and anti-Kv2.2 long antibodies detected Kv2.1 and Kv2.2, respectively, expressed in HEK293 cells, but there was no detectable crossreactivity of the antibodies. This was also verified in the immunostaining of HEK293 cells coexpressing Kv2.1 and Kv2.2 long (Fig. 2C).
With the specificity of the antibodies verified, subcellular localizations of Kv2.1 and Kv2.2 long in cortical neurons were investigated. Rat brain sagittal sections were doubly immunostained with the anti-Kv2.1 (K89) and anti-Kv2.2 long antibodies. Kv2.1 was relatively homogeneously expressed throughout the cerebral cortex as reported previously (17), whereas Kv2.2 expression was detected in a subset of cortical pyramidal neurons (ϳ50% of Kv2.1positive neurons). The expression of Kv2.2 was especially apparent in the layer V pyramidal neurons. This is consistent with previous findings by others, using single-cell reverse transcription-PCR, that ϳ60% of cortical neurons express Kv2.2 (16). In the pyramidal neurons, Kv2.1 formed large clusters in the somata and proximal dendrites (Fig. 3A) as reported previously (18). Our immunostaining results showed that Kv2.2 is also localized in large clusters in the somata and proximal dendrites, in contrast to the previously reported localization of Kv2.2 short . In neurons positive for both Kv2.1 and Kv2.2 long , we found a substantial overlap of Kv2.1 and Kv2.2 long in discrete clusters (Fig.  3B). Quantitative image analyses revealed that Kv2.1 and Kv2.2 long exhibit a high degree of colocalization with the following colocalization coefficients: a Pearson's coefficient (the correlation of the intensity distributions between two channels) of 0.66 Ϯ 0.03 and an intensity correlation quotient (the degree of synchronous changes in signal intensity between two channels) (24) of 0.29 Ϯ 0.01 in a colocalization analysis (p Ͻ 0.01, when the intensity correlation quotient was compared with that obtained from random noise images; n ϭ 5). Line scanning analysis also clearly indicated their colocalization in clusters (Fig. 3C). These results indicate that Kv2.1 and Kv2.2 long coexist in these microstructures in the same membrane domains of cortical neurons.  It has been previously reported that Kv2.1 forms large neuron-like clusters in HEK293 cells (31,32). Using this model system, we further confirmed the clustered localization of Kv2.2 long and its colocalization with Kv2.1. As shown in Fig. 4A, Kv2.2 long formed large clusters reminiscent of those formed by Kv2.1. Furthermore, Kv2.1 and Kv2.2 long were nearly completely colocalized in clusters in cells expressing both of the subunits (Fig. 4, B and C). Together, these results strongly indicate that Kv2.2 long exhibits a subcellular localization similar to that of Kv2.1 and that they may form heteromeric delayed rectifier channels, in contrast to the current view of Kv2 channels.
Finally, we examined the colocalization of Kv2.1 and Kv2.2 long in neurons using mouse monoclonal antibody K37 raised against the N-terminal 61 residues of Kv2.2, which are identical between Kv2.2 short and Kv2.2 long . Therefore, this antibody was expected to label both Kv2.2 short and Kv2.2 long . As shown in Fig. 5A, antibody K37 labeling showed clustered localization of Kv2.2 in the somata and proximal dendrites of cortical pyramidal neurons, similar to that observed with the rabbit anti-Kv2.2 long antibody. We did not detect Kv2.2 labeling in distal dendrites, which was observed in previous studies using different antibodies. Double labeling with the K37 and anti-Kv2.2 long antibodies showed virtually identical staining patterns in cortical neurons (Fig. 5B). These results indicate that Kv2.2 long is the predominant form of Kv2.2 expressed in rat cortical pyramidal neurons.
Molecular Interaction of Kv2.1 and Kv2.2 long -Considering the high degree of colocalization, we hypothesized that Kv2.1 and Kv2.2 long form functional heteromeric channels. To test this, we examined whether they can physically interact in HEK293 cells using co-immunoprecipitation analysis. We first confirmed the specificity of the antibodies in the immunoprecipitation application. Lysates from HEK293 cells expressing either Kv2.1 or Kv2.2 long were subjected to immunoprecipitation with either the anti-Kv2.1 (K89) or anti-Kv2.2 long antibody. There was no detectable cross-reactivity in this type of application (Fig. 6, A and B). We then performed co-immunoprecipitation assays using cells coexpressing Kv2.1 and Kv2.2 long . As shown in Fig. 6C, the anti-Kv2.1 antibody (K89) co-immunoprecipitated Kv2.2 long along with Kv2.1. Furthermore, the anti-Kv2.2 long antibody also co-immunoprecipitated Kv2.1 together with Kv2.2 long . These results were highly reproducible in three independent experiments (data not shown). To exclude the possibility that the interaction occurs following cell lysis but not in situ, we separately prepared lysates from cells expressing either Kv2.1 or Kv2.2 long , mixed these lysates together, and performed the co-immunoprecipitation assay. There was no detectable co-immunoprecipitation (supplemental Fig. 1), indicating that the binding of Kv2.1 and Kv2.2 long occurs in situ, but not after cell lysis. We then examined whether they are associated in brain neurons. Kv2.1 was purified from rat brain lysates using an affinity column conjugated with antibody K89. Bound proteins were eluted and separated by SDS-PAGE, and proteins were then stained with Coomassie Brilliant Blue. Multiple bands appeared at ϳ100 kDa but not in the eluent from the control column (Fig. 6D). These bands were excised, trypsinized, and subjected to mass spectrometry analysis for protein identification ( Table 1). The analysis revealed that the 120-kDa major band contained Kv2.1, showing the high efficiency of Kv2.1 affinity purification. Intriguingly, peptides of the Kv2.2 subunit were  1% Triton X-100 lysis buffer. Either the K89 or anti-Kv2.2 long antibody was added to the lysates and incubate for 2 h at 4°C. Protein A or G beads were added, and the lysates were further incubated for 2 h at 4°C. Proteins precipitated with beads were eluted and detected by Western blotting using the K89 and anti-Kv2.2 long antibodies. D, co-purification of Kv2.2 with Kv2.1 from rat brain tissues. A silver-stained gel image of affinity-purified samples for mass spectrometry analysis is presented. For the mass spectrometry analysis, a Coomassie Blue-stained gel was used. See "Experimental Procedures" for details. E, reciprocal co-immunoprecipitation of Kv2.1 and Kv2.2 long from rat brain lysates. Control experiments in C and D were performed without the addition of the antibodies during immunoprecipitation. recovered from the band above the Kv2.1 band (ϳ140 kDa). This was consistent in two independent affinity-purified samples. We also confirmed their interaction using co-immunoprecipitation assays. The K89 and anti-Kv2.2 long antibodies reciprocally immunoprecipitated Kv2.1 and Kv2.2 long from rat brain lysates as shown in Fig. 6E. These results demonstrate that Kv2.1 and Kv2.2 long interact in brain neurons.
Kv2.1 and Kv2.2 long Form Heteromeric Delayed Rectifier K ϩ Channels-The immunoprecipitation experiments suggested the interaction of the two Kv2 subunits. However, it was not conclusive whether or not the Kv2.1 and Kv2.2 subunits form heteromeric channels. To directly address this, we took advantage of the non-conducting pore mutant of Kv2.1 (W365C/ Y380T) (19,33). If they truly form heteromeric channels, the non-conducting mutant of Kv2.1 should be able to suppress the conductance of the wild-type Kv2.2 long channels. We therefore expressed either wild-type Kv2.1 or Kv2.2 long with Kv2.1 W365C/Y380T in HEK293 cells and recorded ionic currents in voltage-clamp recordings. Kv2.2 long expressed in HEK293 cells exhibited delayed rectifier currents (Fig. 7A), which were similar to those of homomeric Kv2.1 channels in the current-voltage relationship (Fig. 7B). Kv2.1 W365C/Y380T itself did not conduct detectable currents (data not shown), as reported previously. Coexpression of Kv2.1 W365C/Y380T significantly inhibited the current amplitude of wild-type Kv2.1 channels (p Ͻ 0.01; n ϭ 6) (Fig. 7B), demonstrating the inhibitory efficacy of the mutant. This non-conducting mutant also significantly and dramatically suppressed the conductance of the wild-type Kv2.2 long channels (Fig. 7, B and C) by nearly 70% (p Ͻ 0.001; n ϭ 6). Western blot analysis confirmed that this inhibition was not due to a reduction in the expression level of wild-type Kv2.2 long . The level of wild-type Kv2.2 long was not affected by the coexpression of the Kv2.1 pore mutant (Fig. 7D). These results strongly suggest that Kv2.1 and Kv2.2 long , which are colocalized in large surface clusters in neurons, form heteromeric delayed rectifier channels.

DISCUSSION
Previous studies have shown that Kv2.2 is uniformly localized in the dendrites of mammalian neurons (14,16,26), whereas Kv2.1 is highly concentrated in the somata and proximal dendrites in discrete surface clusters (17). Together with the fact that the non-conducting mutant of rat Kv2.1 does not inhibit rat Kv2.2 channels expressed in HEK293 cells (19), it has been thought that Kv2.1 and Kv2.2 do not form heteromeric channels. It was puzzling, however, because Kv2 subunits of Xenopus do form functional heteromeric channels (33). We believe that our results have solved this conundrum. In this work, we report that there are two different mRNA sequences of rat Kv2.2 in the data base: the original cDNA sequence (9) and one derived from the genomic sequence. We found that the original sequence has a single-nucleotide deletion compared with the genomic sequence. The cDNA sequence with the deletion results in a frameshift in the protein sequence after amino acid residue 764, providing a 105-amino acid shorter Kv2.2 polypeptide (Kv2.2 short ) than that deduced from the genomic sequence (Kv2.2 long ). We currently do not know whether the deletion is due to a cloning artifact or is the result of endogenous RNA editing. However, immunolabeling using antibody K37 indicated   Using specific immunoreagents, we found that Kv2.2 long is highly colocalized in discrete surface clusters with Kv2.1 in the somata and proximal dendrites of rat cortical neurons. Clustering might be an intrinsic characteristic of Kv2.2 long , like Kv2.1, because it can form clusters in HEK293 cells in the absence of Kv2.1. It has been previously shown that the clustering of Kv2.1 requires the internal sequence in the cytoplasmic C-terminal tail, the proximal restriction and clustering signal (15). The critical residues in the proximal restriction and clustering signal are well conserved in the C-terminal tail of Kv2.2 (data not shown) in both the shorter and longer forms. However, it has been shown that Kv2.2 short does not form clusters in HEK293 cells (32). We speculate that the altered protein sequence in the C terminus of Kv2.2 short results in a large structural rearrangement, which prevents it from forming clusters.
The substantial colocalization of Kv2.1 and Kv2.2 long brought about the possibility that they are indeed expressed as heteromeric channels. Our co-immunoprecipitation assays and the experiments using the non-conducting mutant clearly demonstrated that Kv2.1 and Kv2.2 long are physically associated in brain neurons in situ and form heteromeric channels when coexpressed. This is highly consistent with the findings of Blaine and Ribera (33) that the channel current of Xenopus Kv2.2 is significantly inhibited by a Xenopus Kv2.1 non-conducting mutant. The tetramerization (T1) domain of Kv1 channel subunits has been located in the cytoplasmic N-terminal tail (34). It has been recently reported that the N-terminal domain of Kv2.1 is also critical in forming a heterotetramer with Kv6.3, a modulatory subunit (35). In addition to the difference in the C-terminal tail, two amino acid residues at positions 161 and 162 in the N-terminal domain also differ between Kv2.2 short (Asp and Gly) and Kv2.2 long (Gln and Arg). This may account for the lack of heteromerization of Kv2.1 and Kv2.2 short , although these residues are not close to the conserved histidine residue (at amino acid 109 in Kv2.2) that is critical for the binding of Kv6.3 to Kv2.1. However, it should be noted that Schnitzler and co-workers (35) also reported that the N-terminal tail of Kv2.2 short can in fact interact with that of Kv2.1 in the yeast two-hybrid assay. There is also accumulating evidence that the N-and C-terminal tails of Kv2.1 subunits interact to change the biophysical and biochemical properties of the channel (36 -38). Therefore, it is possible that the C termini of Kv2 subunits facilitate the formation of channel tetramers and that this may be disrupted in Kv2.2 short .
One interesting aspect of the Kv2.1-Kv2.2 interaction is that the non-conducting mutant of Kv2.1 showed a greater inhibition on wild-type Kv2.2 long than on Kv2.1 (p Ͻ 0.05; n ϭ 6) (Fig.  5C). Because it has been elegantly shown that Kv2.1 and Kv9.3, another modulatory subunit, have a fixed stoichiometry of 3:1 (39), we speculate that Kv2.1 and Kv2.2 long also have a similar stoichiometry, in which Kv2.1 is dominant in a tetramer. With wild-type Kv2.1 and its non-conducting mutant, their stoichiometry follows a binomial distribution, and thus, the degree of inhibition would be variable. However, if Kv2.1 and Kv2.2 long preferably have a fixed stoichiometry of 3:1, the Kv2.1 non-conducting mutant becomes dominant in Kv2.2 long /Kv2.1 mutant tetramers. Therefore, the mutant would exhibit a much greater inhibitory efficacy on the Kv2.2 long channel current than on wild-type Kv2.1 (Fig. 7B). This is still a speculation and requires further research.