Inactivation of a Voltagedependent K+ Channel by β Subunit

Kv1.1/Kvβ1.1 (αβ) K+ channel expressed in Xenopus oocytes was shown to have a fast inactivating current component. The fraction of this component (extent of inactivation) is increased by microfilament disruption induced by cytochalasins or by phosphorylation of the α subunit at Ser-446, which impairs the interaction of the channel with microfilaments. The relevant sites of interaction on the channel molecules have not been identified. Using a phosphorylation-deficient mutant of α, S446A, to ensure maximal basal interaction of the channel with the cytoskeleton, we show that one relevant site is the end of the C terminus of α. Truncation of the last six amino acids resulted in αβ channels with an extent of inactivation up to 2.5-fold larger and its further enhancement by cytochalasins being reduced 2-fold. The wild-type channels exhibited strong inactivation, which could not be markedly increased either by cytochalasins or by the C-terminal mutations, indicating that the interaction of the wild-type channels with microfilaments was minimal to begin with, presumably because of extensive basal phosphorylation. Since the C-terminal end of Kv1.1 was shown to participate in channel clustering via an interaction with members of the PSD-95 family of proteins, we propose that a similar interaction with an endogenous protein takes place, contributing to channel connection to the oocyte cytoskeleton. This is the first report to assign a modulatory role to such an interaction: together with the state of phosphorylation of the channel, it regulates the extent of inactivation conferred by the β subunit.

In contrast with the well understood structural determinants within the core of voltage-gated K ϩ (Kv) channel proteins that underlie many functional properties (e.g. voltage-sensing regions, pore region; Refs. [1][2][3][4][5], much less is known about the role of extrinsic factors in determining Kv channel function. One such factor is the interaction of ion channels with the cytoskeleton, which could affect structural and functional organization of neuronal elements. Recently, a family of membrane-associated putative guanylate kinases were shown to bind directly to K ϩ channels of the Shaker (Kv1) family and to induce their clustering in COS7 cells (6,7). This family includes PSD-95, an abundant synaptic protein found both preand postsynaptically, and SAP97, a protein found in axons of neurons as well as in non-neuronal cells (for reviews, see Refs. 8 and 9). These data together with the demonstration of colocalization of the channels with PSD-95 (6,10) in the brain, particularly in nerve terminal plexuses of basket cells in the cerebellum, provide correlative evidence of a direct association in vivo. Apart from a probable role in clustering, it is not known whether the association with PSD-95 has a functional role of modulating the biophysical characteristics of the channel.
For several years now our interest has been the modulation by direct phosphorylation of the voltage-gated K ϩ channel Kv1.1 (originally cloned from rodent brain cDNA libraries; Refs. [11][12][13][14], which is expressed in Xenopus oocytes as a delayed rectifier type (15,16). Recently it became clear that part of the diversity of Kv channels may arise from association of pore-forming K v 1 subunits with auxiliary K v ␤ subunits (for review, see Ref. 17) and that the functional consequence of the association between K v 1.1 (␣) and rat brain K v ␤1.1 (␤) is the appearance of rapid inactivation of the current (18 -20). Therefore, we studied also the Kv1.1/Kv␤1.1 (␣␤) channel and its modulation by direct phosphorylation (21). We demonstrated that the inactivation of the ␣␤ current is not complete, even under conditions where the ␣ polypeptide is saturated with the ␤ polypeptide, and has an inherent sustained component, indistinguishable from a pure ␣ current. The extent of the inactivation (the fraction of the inactivating component) is increased either by phosphorylation of the ␣ subunits at Ser-446 or by depolymerization of the microfilaments by cytochalasins; the latter effect occludes the former. To account for our findings we proposed a simple model that assumes the existence of two modes of the ␣␤ channels, in one mode inactivation is conferred on the channels by the ␤, in the other mode no inactivation is observed despite the fact that ␣ is physically associated with ␤. Interaction of the channels with microfilaments shifts the equilibrium between the two modes toward the noninactivating mode, and phosphorylation (which impairs the interaction of the channels with microfilaments by an unknown mechanism) shifts the equilibrium toward the inactivating mode. The ␣ protein has at its very C terminus a TDV sequence (amino acids 493-495), which presumably mediates the coclustering of Kv1.1 channels with PSD-95 and SAP97 in COS7 cells (7). Thus, we assumed that this site might mediate the interaction of the ␣␤ channels with the oocyte microfilaments; phosphorylation of Ser-446, which is also in the cytoplasmic C terminus of the ␣ protein, seemed an attractive mechanism to modulate this interaction. Supportive of such a notion is the recent demonstration that the binding of the inwardly rectifying K ϩ channel Kir 2.3 to PSD-95 is regulated by phosphorylation of the channel; however, in this case the phosphorylation occurs at the binding site itself (22).
In this study we set out to test the site of interaction with the cytoskeleton by studying the effect of truncation of the last C-terminal amino acids of the ␣ protein on the extent of inactivation of the phosphorylated and nonphosphorylated ␣␤ channels and their susceptibility to microfilament depolymerization. The results are in agreement with a scenario where the very C terminus of the ␣ subunits in ␣␤ channels interacts with the microfilaments via a PSD-like protein; the interaction is phosphorylation-dependent and affects the extent of channel inactivation.

EXPERIMENTAL PROCEDURES
Materials-Chemicals were from Sigma (Rishon Le-Zion, Israel) unless stated otherwise. Vanadate (sodium orthovanadate) and okadaic acid were from Alomone Laboratories (Jerusalem). [ 35 S]Methionine/ cysteine mix and [␥-32 P]ATP were from Amersham Corp. K v 1.1 antiserum was generated against a 23-amino acid peptide that corresponds to the N terminus of RCK1 (SGENADEASAAPGHPQDGSYPRQ), as described (15).
DNA Constructs and mRNAs-K v 1.1 cDNA was subcloned into a vector to yield a SupEx-RCK1 construct that confers high levels of expression in oocytes (15). All substitution mutants of K v 1.1␣ were subcloned into the same vector. Oligonucleotide-mediated mutagenesis was performed using a mutagenesis kit (CLONTECH) according to the manufacturer's instructions, using double-stranded DNA as templates and oligonucleotide primers (synthesized by General Biotechnology Inc., Rehovot, Israel) encoding the desired mutations. S446A was generated as described (16); K490s.c and T493A were generated on either the WT 1 channel or S446A mutated molecules using the primers (underlined nucleotides encode the mutated residues): 5Ј-CGTTAATAA-GAGCTAGCTCCTGACCG-3Ј; 5Ј-GCAAGCTCCTGGCCGATGTTTAA-AAAAAGC-3Ј, respectively. mRNAs were transcribes in vitro using T7 RNA polymerase.
Oocytes, Drug Treatments, and Electrophysiological Recording-Frogs (Xenopus laevis) were maintained and dissected, and their oocytes were prepared as described (23). Oocytes were injected with 2-5 ng/l K v 1.1 and 200 -500 ng/l K v ␤1.1 mRNA for two electrode voltage clamp studies and 100 -200 ng/l K v 1.1 and 3-7 g/l K v ␤1.1 mRNAs for macropatch and biochemical studies. The concentrations of the injected mutant RCK1 mRNAs deviated slightly from the above concentrations as they were adjusted to give similar current amplitudes with K v ␤1.1. Injected oocytes were incubated at 22°C for 1-3 days in ND96 solution (96 mM NaCl, 2 mM KCl, 1 mM MgCl 2 , 1 mM CaCl 2 , 5 mM Hepes, pH 7.5) supplemented with 2.5 mM sodium pyruvate, 100 g/ml streptomycin, and 100 units/ml penicillin (NDE solution) and then assayed either electrophysiologically or biochemically. Cytochalasins were added 4 -6 h before the electrophysiological assay.
Two electrode voltage clamp recordings were performed as described (16). To avoid possible errors introduced by series resistance, only current amplitudes up to 4 A were recorded, and in a given experiment the amplitudes of WT and mutant currents were similar. Oocytes were placed in a 1-ml bath continuously perfused with ND96 solution. Currents were elicited by stepping the membrane potential from a holding potential of Ϫ80 mV to ϩ50 mV for 70 ms. Net current was estimated by subtraction of the scaled leak current elicited by a voltage step to Ϫ90 mV.
Metabolic Labeling with [ 35 S]Methionine/Cysteine ([ 35 S]Met/Cys), Homogenization, and Immunoprecipitation-This was done essentially as described (16). Following injection, six to eight oocytes were incubated at 22°C for 4 h in NDE solution, then for 3 days in NDE containing 0.2 mCi/ml [ 35 S]Met/Cys, and then were homogenized in 150 -300 l of medium consisting of 20 mM Tris, pH ϭ 7.4, 5 mM EDTA, 5 mM EGTA, 100 mM NaCl, 50 g/ml phenylmethylsulfonyl fluoride, 1 mM iodoacetamide, 1 M pepstatin, 1 mM 1,10-phenanthroline supplemented with protein phosphatase inhibitors: 50 nM okadaic acid, 0.5 mM vanadate, and 50 mM KF. Debris was removed by centrifugation at 1000 ϫ g for 10 min at 4°C. After addition of Triton X-100 to a final concentration of 4%, followed by centrifugation for 15 min at 4°C, antiserum was added to the supernatant for 16 h at 4°C. The antibodyantigen complex was incubated for 1 h at 4°C with protein A-Sepharose and then pelleted by centrifugation for 1 min at 8000 ϫ g. Immunoprecipitates were washed four times with immunowash buffer (150 mM NaCl, 6 mM EDTA, 50 mM Tris, pH ϭ 7.5, 0.1% Triton X-100); the final wash contained no Triton X-100.
Quantification of Labeling Intensities and Generation of Digitized PhosphorImager Scans-Gels were dried and placed in a PhosphorImager (Molecular Dynamics) cassette for about 1 day. Using the software ImageQuant, a digitized scan was derived, and relative intensities of protein bands were estimated quantitatively by the software Image-Quant as described (16).
Statistical Analysis-Data are presented as means Ϯ S.E.; N denotes number of frogs assayed, n denotes the number of oocytes assayed. The Student's t test was used to calculate the statistical significance of differences between two populations.

RESULTS
The mRNAs of Kv1.1 (␣) and Kv␤1.1 (␤) were coinjected at a ␤-to-␣ mRNA ratio of 100, which ensures saturation of ␣ polypeptides with ␤. As described before (21), the resulting outward current assayed by the two-electrode voltage clamp technique had a fast inactivating component (I i ) and a noninactivating sustained component (I s ) remaining at the end of a depolarizing pulse to ϩ50 mV. The apparent extent of inactivation (inactivating fraction) was defined as I i /I p (I p ϭ peak current). Previously, we showed that the extent of inactivation of the ␣ S446A ␤ channel (in which Ser-446 of the ␣ subunit was substituted with alanine; S446A) is significantly smaller (by about 70%) than that of the WT (␣ WT ␤) channel and related it to the fact that the mutant, unlike the wild-type, channel is not phosphorylated in its basal state in the oocyte (21). In this study the average extent of inactivation of ␣ WT ␤ currents was 0.57 Ϯ 0.02 (mean Ϯ S.E.; N ϭ 3 oocyte batches, n ϭ 41 oocytes) and that of ␣ S446A ␤ currents was 0.12 Ϯ 0.025 (N ϭ 6 oocyte batches, n ϭ 78 oocytes).
Truncation of the very end of the C terminus of S446A ␣ polypeptide by deletion of the last six amino acids (substituting Lys-490 with stop codon: K490s.c), or substitution of Thr-493 with Ala (T493A), increased by 155 and 65% the extent of inactivation of the respective ␣ S446A ␤ channels (see Fig. 1, A  and B, open bars). The same mutations in the WT ␣ polypeptide did not increase the extent of inactivation of the corresponding ␣␤ channels (Fig. 2, A and B, open bars). Previously, we showed that treatment of oocytes with 40 M of the microfilamentdisrupting agent dihydrocytochalasin B (DHCB) increases the extent of inactivation, the effect being severalfold larger in the ␣ S446A ␤ than in the ␣ WT ␤ channels. If, as we believed, the above increases in the extent of inactivation of the ␣ S446A ␤ channels in which the C terminus of the ␣ subunit was truncated are due to some impairment of the interaction of the channels with the cytoskeleton, one would expect that these effects will be occluded by the DHCB treatment. Indeed, in oocytes of the same frogs the inactivations of ␣ S446A ␤ and each one of its C-terminal mutants were practically the same after DHCB treatment (Fig. 1B, hatched bars). Thus, whereas the effect of DHCB was about 400% on ␣ S446A ␤, it was more than 2-fold smaller (about 185%) on either one of the mutants. The effects of DHCB on the ␣ WT ␤ channels (about 10%) and on its C-terminal mutations were much smaller (Fig. 2B, hatched  bars).
One possibility to account for the larger extent of inactivation of the ␣ S446A ␤ mutants (as compared with ␣ S446A ␤ itself) could be that the ␤-binding capacities of the mutant polypeptides were larger. This possibility was excluded in the concomitant biochemical analysis. SDS-PAGE analysis of [ 35 S]Met/ Cys-labeled ␣ and ␤ polypeptides coprecipitated by anti-␣ antibody showed that the ␤-binding capacity of all the ␣ S446A polypeptide species is similar, namely, the same amount of ␤ was coprecipitated per equal amounts of any of the ␣ polypeptides (Fig. 3). This result is in accord with our previous data showing that under the conditions of such an experiment the ␤-to-␣ ratio of the coprecipitated polypeptides was the maximal possible, when Ͼ90% of ␣ was saturated with ␤ (21). The same SDS-PAGE analysis (Fig. 3) shows that, as demonstrated before (16), the S446A ␣ polypeptides migrate as a 54-kDa polypeptide, which is the nonphosphorylated form, whereas the WT ␣ polypeptide migrates as a doublet of 54-and 57-kDa polypeptides, the latter being the phosphorylated form. It should be noted that the WT ␣ polypeptide was basally phosphorylated to a large extent, as is evident from the relative intensities of the 57-and 54-kDa bands. DISCUSSION Our preceding study (21) demonstrated that the inactivation of the Kv1.1/Kv␤1.1 (␣␤) channel is hampered by interaction of the channel with microfilaments and that this interaction can be disrupted to a significant extent by phosphorylation of a single C-terminal residue (Ser-446) of the ␣ subunit. Thus, in the present study, the ␣ S446A ␤ channel having the S446A ␣ subunit, which, unlike the WT, is not basally phosphorylated in the oocyte, had a substantially weaker inactivation (ϳ10% of total current) than the ␣ WT ␤ channel (ϳ60%). Consequently, ␣ S446A ␤ displayed a much stronger increase (by ϳ300%) in inactivation than ␣ WT ␤ (by ϳ10%) upon treatment with DHCB, which disrupts the microfilaments (21). Therefore, to look for the interaction between the ␣␤ channel and the cytoskeleton, rather than using the wild-type channel, we utilized the ␣ S446A ␤ mutant in which the channel-cytoskeleton interaction is functionally intact and which responds to any reduction in the strength of this interaction by a marked change in inactivation kinetics.
The interaction of the ␣␤ channel with microfilaments could occur at sites on both the ␣ and the ␤ polypeptides. A highly probable candidate was the consensus TDV sequence at the end of the C terminus, recognized by the PDZ domain of the Dlg/ PSD-95 family of proteins (for reviews, see Refs. 8 and 24), thus providing cytoskeletal interaction (9) and inducing clustering of Kv1.1 channels in COS7 cells (7). To test this hypothesis, we studied the effects of truncation of the very C terminus of ␣ subunits, and of a point mutation in the TDV sequence, on the extent of inactivation, both basal and following DHCB treatment, of the ␣ S446A ␤ channels. One would expect that if the interaction between the channel and the cytoskeleton was disrupted by the mutations, the extent of inactivation would increase and the enhancement of inactivation by DHCB would be weakened. Indeed, the two different C-terminal mutations in the S446A ␣ subunit (a deletion of the last six amino acids and a replacement of the crucial Thr-493 with alanine) significantly increased the basal extent of inactivation of ␣ S446A ␤ and reduced the magnitude of the inactivation enhancement by DHCB. As expected, under the conditions of these experiments, in which the ␣ WT ␤ channels were essentially detached from the cytoskeleton in their basal state in the oocyte (as indicated by the large basal inactivation and small DHCB effect), the same C-terminal mutations of the WT ␣ subunit did not increase the inactivation of the ␣ WT ␤ channels. This is also in accord with the large extent of basal phosphorylation in the oocytes used in these experiments (Fig. 3). In conclusion, the data presented in this report corroborate the hypothesis that the very C terminus of ␣ interacts with the microfilaments.
The interaction of the very C terminus of the ␣ subunit with the oocyte's cytoskeleton is via an endogenous protein in the oocyte, possibly resembling protein members of the dlg/PSD-95 family. However, truncation of all last six amino acids in the C-terminal tail was more effective than substitution of Thr-493 in the TDV sequence with alanine, a substitution that was shown to abolish the interaction of a related ␣ protein, Kv1.4, with PSD-95 (6). Thus, it seems that the TDV motif in Kv1.1 is not the only determinant of ␣-subunit binding to the postulated endogenous protein. In view of the demonstration that the interaction of Kv1.1 with PSD-95 was much weaker than that of Kv1.4 in COS7 cells, and that Kv1.1 is missing a glutamate that precedes the TDV sequence in Kv1. 4 and was shown to be critical for the Kv1.4 interaction with PSD-95 (7), it seems plausible that Kv1.1 interacts in vivo with a protein having a somewhat different recognition sequence, yet to be identified. An alternative explanation to the stronger interaction of Kv1.4 with PSD-95 could be that the basal phosphorylation of this channel is weaker than that of Kv1.1 in the cells tested.
It should be pointed out that the DHCB effect occluded the effects caused by truncations of the C terminus of ␣ in ␣ S446A ␤ channels, much as it occluded (21) the effect of Ser-446 phosphorylation of ␣ in ␣ WT ␤ channels. This suggests that disruption of microfilaments by DHCB causes maximal enhancement of inactivation extent of the ␣␤ channel experimentally achievable, probably by a massive disruption of interactions between the channels and the microfilaments. However, the fact that the effect of DHCB was still prominent, though significantly reduced, in the C terminus-truncated mutants suggests that there are additional sites of interaction with the microfilaments on the ␣␤ channel molecules that modulate the inactivation.
The present study correlates the extent of inactivation of the Kv1.1/Kv␤1.1 channel with its interactions with PSD-95-like proteins. This is the first report that assigns a modulatory role to such interactions. Such a functional interaction might be physiologically relevant at synaptic sites in different brain areas where PSD-95 family members were shown to be concentrated and colocalized with Shaker-type K ϩ channels including Kv1.1 (6).
FIG. 3. The ␤-binding capacities of the S446A and its C terminus mutants are similar. SDS-PAGE analysis of ␣␤ channels composed of either WT, S446A, or one of the two C terminus mutants of S446A ␣ polypeptides. [ 35 S]Met/Cys-labeled ␣ (different variants, indicated above the lanes) and ␤ polypeptides were coimmunopurified by the ␣ antibody from oocytes coinjected with ␣ and ␤ mRNAs at a ␤-to-␣ ratio of 35 (lanes 2-5). Lane 1 is noninjected oocytes. Proteins were immunopurified upon solubilization in 4% Triton X-100 in the presence of protease and phosphatase inhibitors. Arrows indicate migration of molecular weight markers.