Modulation of a Brain Voltage-gated K+ Channel by Syntaxin 1A Requires the Physical Interaction of Gβγ with the Channel*

Recently we suggested that direct interactions between voltage-gated K+ channels and proteins of the exocytotic machinery, such as those observed between the Kv1.1/Kvβ channel, syntaxin 1A, and SNAP-25 may be involved in neurotransmitter release. Furthermore, we demonstrated that the direct interaction with syntaxin 1A enhances the fast inactivation of Kv1.1/Kvβ1.1 in oocytes. Here we show that G-protein βγ subunits play a crucial role in the enhancement of inactivation by syntaxin 1A. The effect caused by overexpression of syntaxin 1A is eliminated in the presence of chelators of endogenous βγ subunits in the whole cell and at the plasma membrane. Conversely, enhancement of inactivation caused by overexpression of β1γ2 subunits is eliminated upon knock-down of endogenous syntaxin or its scavenging at the plasma membrane. We further show that the N terminus of Kv1.1 binds brain synaptosomal and recombinant syntaxin 1A and concomitantly binds β1γ2; the binding of β1γ2 enhances that of syntaxin 1A. Taken together, we suggest a mechanism whereby syntaxin and G protein βγ subunits interact concomitantly with a Kv channel to regulate its inactivation.

Voltage-gated K ϩ (Kv) 1 channels participate in a host of cellular processes, from setting the resting membrane potential and shaping action potential wave-form and frequency to controlling synaptic strength (1). Recently, we challenged the commonly accepted concept that presynaptic Kv channels participate in neurotransmitter release simply by virtue of their ability to shape action potentials that invade nerve terminals (2,3), and suggested that the fine tuning of transmitter release might be attributable to direct interaction between Kv channels and proteins of the exocytotic machinery (4). We demonstrated that the Kv channel composed of the pore forming Kv1.1 and auxiliary Kv␤ subunits interact in fresh brain synaptosomes with syntaxin 1A, SNAP-25, and synaptotagmin, and this interaction is relieved following triggering of transmitter release. Furthermore, in insulinoma HIT-T15 ␤ cells the activity of Kv1.1 channel was inhibited by SNAP-25 (5). Also, we showed, in Xenopus oocytes, that the direct interaction of the Kv1.1/Kv␤1.1 (␣␤) channel with syntaxin 1A enhances the fast inactivation of the channel (4) that is conferred by the N-terminal part of ␤, in a mechanism termed "ball and chain" inactivation (6). The reciprocal effects of ␣, syntaxin 1A, and SNAP-25 are reminiscent of the finding that presynaptic Nand L-type voltage-gated Ca 2ϩ channels interact directly with proteins of the exocytotic apparatus in neurons, and that their interaction with syntaxin 1A and SNAP-25 causes feedback effects on the channel function in oocytes (reviewed in Ref. 7) and in synaptosomes (8). Recent studies have shown that disruption of the interaction with syntaxin 1A in neurons has functional implications for transmitter release, reducing the efficacy of both Ca 2ϩ -dependent (7,9) and Ca 2ϩ -independent (10) release.
Previous studies by our group have shown that the extent of inactivation depends on cellular mechanisms leading to phosphorylation (11) and dephosphorylation (12) of Kv1.1 and on the interaction of the channel with microfilaments (11). Furthermore, we identified G protein ␤␥ subunits (G␤␥) as the main regulators of the interaction with microfilaments and consequently as regulators of the extent of inactivation (13).
Here we report that the effect of syntaxin on the inactivation of the ␣␤ channel requires the presence of G␤␥. Intriguingly, G-protein modulation of Ca 2ϩ currents in nerve terminals was eliminated by cleavage of syntaxin (14). Also, a major part of syntaxin-induced inhibition of N-type Ca 2ϩ channel expressed in HEK cells crucially depends on the presence of G␤␥, which tonically inhibits the Ca 2ϩ channel (15,16). Thus, the finding of inter-relationships between syntaxin 1A and G␤␥ interactions with a Kv channel points to further analogy with the interaction of the N-type Ca 2ϩ channel with these proteins, suggesting a general complex pattern of regulation of the activity of presynaptic voltage-gated channels, which may be related to the fine-tuning of presynaptic activity.
Preparation of Phosducin-C-terminal hexahistidine-tagged wildtype phosducin (phosducin-His 6 ), was expressed in Escherichia coli strain BL21(DE3) pLysS as previously described (17). The induced cells were lysed in 50 mM sodium phosphate buffer, pH 7.4, by sonication. The lysate was centrifuged at 19,000 ϫ g for 30 min and the His 6 -tagged proteins were purified from the supernatant to Ͼ95% homogeneity by chromatography on nickel-nitrilotriacetic acid columns (Qiagen) followed by gel filtration on Superdex 200 column (Amersham Biosciences) in 50 mM Na phosphate buffer, pH 7.4. The purified protein was aliquoted, frozen in liquid N 2 , and stored at Ϫ80°C.
Oocytes and Electrophysiological Recording-Oocytes of Xenopus laevis were prepared as described (18). Oocytes were injected (50 nl/oocyte) with 150Ϫ300 ng/l K v 1.1 and 1Ϫ3 g/l K v ␤1.1 mRNAs for biochemical experiments, and with 5Ϫ10 ng/l K v 1.1 and 15Ϫ1000 ng/l K v ␤1.1 mRNAs for electrophysiological experiments. G␤ 1 , G␥ 2 , and c-␤ARK mRNAs (100 ng/l each) were injected for both biochemical and electrophysiological experiments, 25 ng/l syntaxin mRNA was injected for electrophysiological experiments and 25-100 ng/l was injected for biochemical experiments. Dihydrocytochalasin B (DHCB; Sigma) treatment was done as described (11): oocytes were incubated in 40 -60 M of the drug for several hours prior to electrophysiological assay. Stock solution of DHCB was made in ethanol (kept at Ϫ20°C); control solutions always included 40 -60 M ethanol. Two-electrode voltage clamp recordings were performed as described (19). To avoid possible errors introduced by series resistance, only current amplitudes up to 4 A were recorded. Currents were elicited by stepping up the membrane potential from a holding potential of Ϫ80 to ϩ50 mV for 350 ms. Net current was obtained by subtracting the scaled leak current elicited by a voltage step from Ϫ80 to Ϫ90 mV. Oocytes with a leak current of more than 3 nA/1 mV were discarded.
Biochemistry in Oocytes-Oocytes were subjected to immunoprecipitation as described (19). For co-precipitation of G␤␥ only, oocyte lysate was incubated for 30 min at 37°C with the anti-␣ antibody followed by 30 min incubation with 30 -40 l of protein A-Sepharose. Immunoprecipitates from 1% Triton X-100 homogenates of either 5 oocytes or 20 plasma membranes (separated mechanically, as described in Ivanina et al. (29) were separated by SDSϪPAGE and either subjected to Western blot analysis using the ECL detection system (Amersham Biosciences) or digitized scans of [ 35 S]methionine/cysteine (Met/Cys)-labeled proteins were derived by PhosphorImager (Amersham Biosciences) and relative intensities were quantitated by ImageQuant (as described in Refs. 4 and 13).
In Vitro Binding of GST Fusion Proteins with Syntaxin 1A-The fusion proteins were synthesized and reacted with syntaxin as described (4,13). Briefly, purified GST fusion proteins (150 pmol) immobilized on glutathioneϪSepharose beads were incubated with either 1-10 l of lysate containing 35 S-labeled G␤␥ or syntaxin 1A (translated on the template of in vitro synthesized RNAs using a translation rabbit reticulocyte lysate kit (Promega) according to the manufacturer's instructions) or 200 pmol of recombinant syntaxin peptide in 1 ml of phosphate-buffered saline with 0.1% Triton X-100 and 0.5 mg/ml bovine serum albumin, for 1 h at room temperature (except for Fig. 5D, see legend), eluted with 15 mM reduced glutathione in 30 l of elution buffer (120 mM NaCl, 100 mM TrisϪHCl, pH 8), and then subjected to SDSϪPAGE.
Statistical Analysis-Data are presented as mean Ϯ S.E. Student's t test was used to calculate the statistical significance of differences between two populations.

Functional
Interactions of G␤␥ and Syntaxin 1A with the Kv1.1/Kv␤1.1 Channel Are Similar-Co-expression of ␣ with ␤ in oocytes injected with the corresponding mRNAs results in the formation of a heteromultimeric ␣␤ channel that conveys a rapidly inactivating current with a fast inactivating component (I i ) and a sustained, noninactivating component (I s ) (see Fig.  1A). The extent of inactivation (the inactivating fraction) is defined as I i /I p (I p ϭ peak current). Inactivation of the ␣␤ current, but not the inactivation rate constant, depends on the level of ␤, which provides the "ball" for inactivation. The extent of inactivation increases up to saturation at I i /I p ϭ 0.5Ϫ0.8 as the ratio of ␤-mRNA to ␣-mRNA injected into oocytes is increased to about 50:1 (depending on the batches of mRNAs and oocytes) (11). As previously shown by our group (4, 13) ( Further similarity between the functional effects of G␤␥ and syntaxin was observed by examining the action of DHCB, a microfilament-disrupting agent previously shown to increase the extent of inactivation (11). As shown in Fig. 1C, the effect of treatment of oocytes with DHCB and the effect of syntaxin were additive when channels were expressed with nonsaturating ␤-mRNA to ␣-mRNA ratios (low ␤ to ␣), whereas upon saturation with ␤ (high ␤ to ␣), which increased the inactivation, both effects were occluded. These results were similar to those obtained previously for the G␤␥ effect (Ref. 13; see also Functional Interaction of Syntaxin with the ␣␤ Channel Requires G␤␥, and Functional Interaction of G␤␥ with the ␣␤ Channel Requires Syntaxin-The similarity between the functional interactions of G␤␥ and syntaxin with the channel led us to examine the possibility of coupling between the effects of G␤␥ and syntaxin. Fig. 1D shows electrophysiological analysis of channels co-expressed with G␤␥, with syntaxin, or with both. The effects of G␤␥ and syntaxin on the extent of inactivation are nonadditive (black bars), implying a possible convergence of signaling pathways. Moreover, injection of 30 pg of the antisyntaxin AS linker into oocytes co-expressing the channel with G␤␥ abolished the effect of G␤␥ on inactivation. These results indicate that endogenous syntaxin is required for generation of the G␤␥ effect. We then looked for symmetry by examining whether G␤␥ is required for generation of the syntaxin effect. We analyzed the effect of co-expressed syntaxin in the absence or presence of co-expressed c␤ARK. c␤ARK abolished the effect of syntaxin on inactivation, indicating that endogenous G␤␥ is needed for generation of the syntaxin effect (Fig. 1D).
Interaction of Both Syntaxin and G␤␥ with the Functional ␣␤ Channel at the Plasma Membrane Is Required for the Increase in the Extent of Inactivation-Furthermore, we addressed the question whether both G␤␥ and syntaxin have to be present at the plasma membrane to jointly regulate the inactivation. We first considered the interaction between the channel and syntaxin. A previous study showed that the effect of co-expressed syntaxin requires the physical interaction of syntaxin with the functional channel (4). This was achieved by taking advantage of the fact that the N 718 -963 ("synprint") peptide, a Ca 2ϩ -channel domain (found to block co-immunoprecipitation of native N-type Ca 2ϩ channels with syntaxin (21)) competed efficiently, in an in vitro binding assay, with the binding of syntaxin to the ␤ subunit of the channel. Thus, by microinjection of this peptide into oocytes co-expressing syntaxin and the ␣␤ channel, we were able to acutely reverse the effect of syntaxin on the channel. Because the synprint peptide was injected 20Ϫ60 min prior to the electrophysiological assay it could not affect significantly ␣␤ synthesis or assembly in this time frame so that it must affect properties of plasma membrane channels. As a control we used the shorter N 718 -859 peptide, which did not compete for syntaxin binding and did not rescue the channel from the effect of syntaxin (Ref. 4; see also Fig. 2, inset). These results indicated that interaction of the channel with exogenous syntaxin, resulting in increased inactivation, occurs at the plasma membrane. In this study we further examined whether interaction of the channel with endogenous syntaxin, which is needed for increasing the extent of inactivation in oocytes co-expressing G␤␥, also occurs at the plasma membrane. To this end we tried to rescue the functional channel from the effect of G␤␥ by injecting the synprint peptide into oocytes co-expressing G␤␥ with the channel. We found that the increased inactivation by G␤␥ was indeed acutely reversed by this peptide and not by the control peptide (Fig. 2, left panel). The effect of the synprint peptide in oocytes expressing ␣␤ alone was statistically insignificant, however, when compared with the effect of the control peptide that enhanced inactivation, there was a clear reduction in inactivation, as was shown before (Ref. A, inactivation of the ␣␤ current is increased by overexpression of syntaxin and decreased by antisense ODN knock-down of syntaxin. Current traces evoked by depolarization to ϩ50 mV from single oocytes of the same batch injected with ␣ and ␤ mRNAs (in a ␤ to ␣ mRNA ratio of 5:1), either alone (␣␤) or together with syntaxin 1A mRNA (ϩSyx), or injected with antisense syntaxin (30 pg; ϩAS linker) 2 days before the assay. I i , I s , and I p illustrate the inactivating, noninactivating, and peak current components of ␣␤, respectively, as defined in the text. B, inactivation of ␣␤ current is increased by overexpression of G␤ 1 ␥ 2 (ϩG␤␥) and decreased by overexpression of c␤ARK (ϩc␤ARK); experiment as described in A. C, normalized and averaged effects of DHCB treatment in oocytes injected with ␣ and ␤ mRNAs in low (5-10:1, left) and high (50 -100:1, right) ␤ to ␣ ratios, without or with syntaxin 1A mRNA. The inset shows the normalized and averaged effects of DHCB treatment in oocytes injected with ␣ and ␤ mRNAs in high ␤ to ␣ ratios with and without G␤ 1 ␥ 2 . Numbers above the bars refer to the number of oocyte batches; numbers in parentheses refer to oocytes. *, p ϭ 0.001. D, averaged effects of syntaxin 1A (ϩSyx) or G␤ 1 ␥ 2 (ϩG␤␥) or both (ϩG␤␥ϩSyx) in oocytes co-injected with ␣␤ mRNAs (low ␤ to ␣ ratio), compared with oocytes injected with ␣␤ alone (␣␤). The effect of co-injected c␤ARK on the effect of G␤␥ or syntaxin is shown in those oocytes. panel). These results indicated that endogenous syntaxin (like exogenously expressed syntaxin) interacts with the functional channel at the plasma membrane to support the G␤␥ effect.
Next we considered the interaction between the channel and G␤␥. In a previous study (13) we could not determine whether the effect of G␤␥ requires interaction of G␤␥ with functional channels at the plasma membrane, or if the interaction with the channel is limited to the biosynthesis and assembly phase at the endoplasmic reticulum. In an attempt to answer this question in this study we used phosducin, a cytoplasmic protein characterized by its ability to specifically bind G␤␥ subunits with high affinity and to efficiently scavenge G␤␥ (22). We were able to show that microinjection of 1 M (final concentration) recombinant phosducin (17) into oocytes co-expressing G␤␥ could acutely reverse the effect of G␤␥ on the inactivation, whereas heat-denatured (2 h incubation at 65°C) phosducin could not (Fig. 3, left panel). This indicated that exogenous G␤␥ exerts its effect by interacting with the channel at the plasma membrane. We further showed that phosducin, but not the denatured phosducin, also reverses the increase in extent of inactivation in oocytes co-expressing syntaxin (Fig. 3, right panel), indicating that endogenous G␤␥ also exerts its effect by interacting with the channel at the plasma membrane. Injection of either intact or denatured phosducin into oocytes expressing the channel alone increased the extent of inactivation. This effect, probably irrelevant to the ability of the channels to bind G␤␥, was manifested only with oocytes devoid of overexpressed G␤␥ and could mask the effect relevant to the ability of phosducin to scavenge the endogenous G␤␥. Overall, these data are consistent with a mechanism by which the concomitant interaction of G␤␥ and syntaxin with the functional channel enhances the extent of inactivation. Overexpression of at least one of these agents, syntaxin or G␤␥, is necessary to produce a maximal effect.
Physical Interaction of Syntaxin with the ␣␤ Channel Is Similar to That of G␤␥-The similarity and coupling between the functional effects of syntaxin and G␤␥ with the ␣␤ channel led us to compare their physical interactions with the channel. A previous study by our group (4) demonstrated that syntaxin 1A associates with the ␣␤ protein complex in brain synaptosomes and in a ϳ1:1 stoichiometry with the ␣␤ channel in plasma membranes of oocytes. We specifically characterized the interaction of ␤ with syntaxin and showed that immobilized ␤ϪGST fusion protein could pull down syntaxin from synaptosomal lysates and bind recombinant syntaxin. G␤␥ binds in vitro not only to ␤ but also to the N terminus of ␣ (13), therefore, we examined the interaction of syntaxin with ␣. Toward this end three approaches were adopted. First, we performed a pull-down assay, using synaptosomal lysates and immobilized GST fusion proteins corresponding to the major intracellular parts of ␣. The following fusion proteins were constructed (Fig.  4A): ␣C, the full-length C terminus of ␣; ␣N, the full-length N terminus of ␣; ␣T1"A" and ␣T1"B", regions of the N terminus of ␣ that participate in tetramerization of Kv␣ subunits (23)(24)(25); the latter also participates in Kv␤ subunit binding. Two types of detergents were used, 2% CHAPS or 4% Triton X-100, to control for artifacts because of possible aggregations of proteins. The results of the two experimental conditions were similar. The pulled down syntaxin was visualized using Western blot analysis. Fig. 4B shows one experiment (two left panels) and a summary of values obtained in five experiments (right panel) demonstrating that ␣N and ␣T1B but not ␣C or GST itself pulled down syntaxin. The second approach was an in vitro binding assay using the same immobilized GST fusion proteins and either the recombinant cytoplasmic part of syntaxin (see "Experimental Procedures") or 35 S-labeled fulllength syntaxin synthesized in reticulocyte lysate. The results of both settings were similar. Syntaxin interacted with the full-length N terminus of ␣ and no interaction was detected with the full-length C terminus (Fig. 4C, middle panel). More specific analysis of this interaction showed that syntaxin bound to both the T1A and T1B domains, more strongly (by about 1.5-fold) to T1A (Fig. 4C, left and right panels). It thus appears that syntaxin, similarly to G␤␥, interacts in vitro not only with ␤ (4) but also with the N terminus of ␣. To further compare the binding of G␤␥ with that of syntaxin we examined the interaction of G␤␥ with the N-terminal domains. The binding pattern was somewhat different from that of syntaxin, as it was hard to detect any preferential binding of G␤␥ to one of the domains (Fig. 4C, left and right panels).
Next, we applied a third, in vivo, approach to establish the interaction of the ␣ subunit with syntaxin. In a previous study (4), co-immunoprecipitation experiments with Xenopus oocyte lysates (1% Triton X-100), using anti-␣ antibody, showed that syntaxin co-precipitates with the ␣␤ channel from lysates of oocytes expressing syntaxin with ␣ and ␤ subunits. In this study, binding of syntaxin to ␣ itself was substantiated by successful co-precipitation of syntaxin with ␣ from lysates of oocytes co-expressing syntaxin with ␣ alone (Fig. 4D). To compare this interaction of syntaxin with that of G␤␥ we first had to determine the conditions for obtaining specific co-immunoprecipitation of G␤␥ with the channel in oocytes, which we were previously unable to do (13). By incubating oocyte lysate for a short period at 37°C with the anti-␣ antibody (see "Experimental Procedures"), we were able to demonstrate G␤-immunoreactive bands that were precipitated together with ␣ (or with ␣␤; not shown) from oocytes co-expressing G␤␥ with the channel, but not from oocytes expressing G␤␥ alone (Fig. 4E). These results show that G␤ can co-precipitate with the ␣ subunit. Taken together, the in vitro and in vivo findings point to a similarity between the physical interactions of G␤␥ and of syntaxin with ␣.  , lower left panel). 45 g of synaptosomes were loaded on the Lysate lane. Normalized relative ECL signal intensities of bound syntaxin (derived from IB Syx) for each of the GST fusion proteins normalized to its relative amount (derived from IB GST) were averaged from five experiments (in one of which we used 4% Triton X-100 instead of CHAPS lysate) (right panel). Numbers on the right refer to the mobility of prestained molecular weight standards. C, in vitro interaction of syntaxin 1A or G␤ 1 ␥ 2 with GST fusion proteins of ␣ fragments. In vitro synthesized 35 S-labeled syntaxin 1A or G␤ 1 ␥ 2 were incubated with 200 pmol of the indicated GST fusion proteins immobilized on GSHϪagarose beads in a 1-ml 0.1% Triton X-100 reaction volume for 1 h. The gluthatione-eluted proteins were analyzed by SDSϪPAGE. Upper left and upper middle panels are digitized PhosphorImager scans and the lower left and lower middle panels are scans of Coomassie Blue staining. Left panel shows results from one of three similar experiments. Numbers on the left refer to the mobility of prestained molecular weight standards. Right panel shows syntaxin (ϩSyx) and G␤␥ (ϩG␤␥) ratio of binding to T1A over binding to T1B. ECL signal intensities of syntaxin or G␤␥ bound to GST-T1A were normalized to those for GST-T1B. The signal for each GST-protein was normalized to its relative amount (derived from Coomassie Blue staining). The results shown are averaged values from three experiments (one of which is shown in right panel; * p Ͻ 0.05). Middle panel is one of three similar experiments. One more experiment was done with recombinant syntaxin. D, co-immunoprecipitation (IP) of syntaxin 1A with ␣ from Xenopus oocytes. SDSϪPAGE analysis of [ 35 S]Met/Cys-labeled ␣ and syntaxin 1A proteins co-precipitated by anti-␣ antibody (IP␣) from homogenates of plasma membranes of oocytes that were injected with ␣ or syntaxin 1A mRNAs only (␣ or Syx), or co-injected with syntaxin 1A and ␣ (␣ϩSyx). Arrows indicate the relevant proteins. The results shown are from one of three independent experiments. E, co-immunoprecipitation of G␤ 1 ␥ 2 with ␣ from Xenopus oocytes. Proteins were precipitated by anti-␣ antibody from homogenates of whole oocytes that were injected with G␤ 1 ␥ 2 mRNA alone (G␤␥) or together with ␣ (␣ ϩ G␤␥) mRNAs. Protein samples were analyzed on a 5Ϫ15% gradient gel. The results shown are from one of three independent experiments. Numbers on the right in D and E refer to the mobility of prestained molecular weight standards.

Syntaxin and G␤␥ Are Able to Concomitantly Bind to the
Channel-The next question to be addressed was whether the ␣ channel binds G␤␥ and syntaxin concomitantly. To answer this question, two approaches were employed. First, we carried out a co-immunoprecipitation assay, using antibody against ␣, to examine the interactions of the ␣␤ channel with syntaxin and with G␤␥ in oocytes co-expressing ␣, ␤, syntaxin, and G␤␥. These experiments showed that both syntaxin-and G␤-immunoreactive bands precipitate with the channel proteins (Fig. 5). This result supports the notion that the channel interacts with both syntaxin and G␤␥ to form a ternary complex, however, it does not rule out the possibility that two distinct K ϩ channel populations were immunoprecipitated, each forming a binary complex, one bound to syntaxin alone and the other bound to G␤␥ alone.
The second approach we employed was an in vitro binding assay. We reasoned that if we could show a synergistic interaction between the binding of G␤␥ and syntaxin to one of the channel domains this will not only demonstrate a possible concomitant binding of syntaxin and G␤␥ to the channel, but will also provide insight into the molecular mechanism underlying the mutual requirement of syntaxin and G␤␥ for function. To this end we used the GST fusion protein corresponding to the full-length N terminus of ␣ (␣N) that binds both G␤␥ and syntaxin, however, with different binding patterns (Fig. 4C). We performed a series of experiments in which we assayed the simultaneous binding of 35 S-labeled G␤␥ and syntaxin at different protein ratios to ␣N and compared with the binding of comparable amounts of each one of them alone. The labeling intensities of both proteins (each having about the same number of labeled amino acids) were normalized to the corresponding amounts (intensities of Coomassie Blue staining) of ␣N. The binding assays were performed in two different detergents (0.1% Triton X-100 and 0.5% CHAPS) to exclude artifacts. When each were assayed separately, the binding of syntaxin was generally weaker than that of G␤␥. When assayed simultaneously, the binding of syntaxin did not enhance the binding of G␤␥ (in 13 of 15 experiments performed at different protein ratios of G␤␥ and syntaxin), rather it reduced it by 34 Ϯ 3% (n ϭ 15; see an extreme example for reduced G␤␥ binding in Fig. 6C, however, one exception in Fig. 6A). In contrast, the binding of G␤␥ did enhance the binding of syntaxin by 2.89 Ϯ 0.49-fold (p Ͻ 0.004; n ϭ 9; range 1.56 -5.7). The enhancing effect of G␤␥ was observed when the ratio of the reacting proteins was adjusted so that the initial G␤␥ binding was severalfold larger than that of syntaxin (which was set at the limit of the experimental resolution) (see Fig. 6A, left panel for one representative experiment and right panel for a summary of nine experiments). Importantly, under the same conditions we could not detect any enhancement by G␤␥ of syntaxin binding to the partial fragment of the N terminus, ␣T1A (which binds syntaxin; rather, G␤␥ reduced the syntaxin binding (Fig.  6A, left panel). In fact, such an apparent reduction in the binding of a protein to a GST-fused peptide was observed also in the presence of another protein that does not bind the peptide (13) and thus is probably an experimental artifact. The enhanced binding of syntaxin to ␣N was G␤␥ dose-dependent and was saturable (Fig. 6B). The enhancement was also dependent on the amount of syntaxin, as only above a certain concentration the enhancement could be detected (Fig. 6C). Analysis of the time dependence of G␤␥ and syntaxin bindings revealed that the enhancement of syntaxin binding in the presence of G␤␥ followed the binding of G␤␥ (Fig. 6D), again indicating that syntaxin binding is dependent on G␤␥ binding.
As it was demonstrated that G␤␥ binds physically to syntaxin (15), the obvious interpretation of the synergistic effect of G␤␥ on syntaxin binding would be that syntaxin is simply recruited to the channel by binding physically to G␤␥. However, the fact that in several experiments, while enhancing syntaxin binding, G␤␥ binding decreased substantially (e.g. Fig. 6C), together with the finding that the binding patterns of G␤␥ and syntaxin to the N terminus were not identical (Fig.  4C), leads us to favor a mechanism in which binding of G␤␥ allosterically increases the binding affinity of the N terminus for syntaxin. It is possible that both mechanisms, recruitment of syntaxin via the binary complex syntaxin-G␤␥ and allosteric interaction, contribute in concert to the enhancement of syntaxin binding. These experiments support the idea that G␤␥ and syntaxin can interact concomitantly with the channel to produce the observed effect on inactivation. DISCUSSION We reported the existence of a physical interaction in brain synaptosomes between an ␣␤ channel complex and syntaxin 1A (syntaxin), occurring at least partially within the context of a larger macromolecular complex that also contains synaptotagmin and SNAP-25. The interaction was sensitive to the physiological state of the synaptosomes, being relieved upon stimulation of neurotransmitter release (4). We further demonstrated, in Xenopus oocytes, that the direct interaction of the ␣␤ channel with syntaxin affects the fast inactivation of the channel. In this study we focus on another protein, G␤␥, which may exist in a complex with both the channel and syntaxin in synaptosomes, and plays a role in mediating the functional interaction of syntaxin with the channel, leading to enhanced inactivation. A somewhat similar interplay between G␤␥ and syntaxin 1A was described for the voltage-dependent inhibition of presynaptic N-type Ca 2ϩ channels (14,15). Thus, the emerging picture is one in which the activities of presynaptic voltage-dependent ion channels are regulated by proteins that participate in presynaptic neurotransmitter release, possibly with the function of finetuning presynaptic activity.
Modulations by Syntaxin and G␤␥ Are Linked-Different lines of evidence point to a possible coupling between G␤␥induced and syntaxin-induced modulations of the fast inactivation of the ␣␤ channel in Xenopus oocytes. First, the functional effects of both agents seem similar and are nonadditive FIG. 5. G␤ 1 ␥ 2 and syntaxin 1A interact with the ␣␤ channel in oocytes. Immunoblot analysis of ␣, ␤, syntaxin 1A (Syx), and G␤ 1 (G␤) immunoprecipitated (IP) from homogenates of oocytes uninjected (control), injected with ␣ and ␤ mRNAs alone (␣␤), co-injected with either syntaxin 1A (␣␤ϩSyx) or G␤ 1 ␥ 2 and syntaxin 1A (␣␤ϩG␤␥ ϩSyx) mRNAs, or injected with G␤ 1 ␥ 2 and syntaxin 1A without the channel (␤␥ϩSyx) mRNAs, with antibody against ␣ (IP␣, left panel). Homogenates of oocytes from the same groups were solubilized directly in SDS sample buffer (Total, right panel). Protein samples were electrophoresed on 12% gel and transferred to nitrocellulose membranes, and the resulting blots were probed with the appropriate antibodies, as indicated (IB), to monitor the co-immunoprecipitated (left panel) or total (right panel) amount of the corresponding proteins. Signals were visualized using ECL. Note the weak signals of endogenous G␤␥ in lysates of oocytes that do not express exogenous G␤␥.
( Fig. 1, A, B, and D). Thus, overexpression of each of them resulted in enhanced inactivation and, conversely, knock down of endogenous syntaxin or sequestration of endogenous free G␤␥ resulted in decreased inactivation. Also, the effects exerted both by G␤␥ and syntaxin were occluded in the same way by disruption of microfilaments and by saturation with ␤ subunits of the channel (Fig. 1C), treatments both known to enhance inactivation. Second, the physical interactions of G␤␥ or syntaxin with the channel are similar. Each agent was coimmunoprecipitated from oocyte lysates, both with ␣␤ or with ␣ alone and interacted in vitro with both ␤ and the N terminus of ␣ (Fig. 4, C-E) (4,13).
To prove a link between the two modulations by G␤␥ and syntaxin we first showed that the effect of overexpressed syntaxin is absolutely dependent on the level of free endogenous G␤␥ (which can be scavenged by c␤ARK). The converse was also true, i.e. the effect of overexpressed G␤␥ was absolutely dependent on the level of endogenous syntaxin (which can be knocked down by syntaxin antisense ODN). These findings indicated that interaction of the channel with both G␤␥ and syntaxin is a requirement for the enhanced inactivation. However, this approach did not distinguish between a functional channel that is already in the plasma membrane and a channel that is still in the process of biosynthesis. Therefore, we verified that the enhanced inactivation in oocytes co-expressing G␤␥ is abolished by acute administration of the synprint peptide, previously shown (4) to impair the effect of syntaxin (Fig. 2). Symmetrically (Fig. 3), the enhanced inactivation in oocytes co-expressing syntaxin was abolished by the acute administration of phosducin, which also abolished the enhanced inactivation in oocytes co-expressing G␤␥, by binding to and thereby sequestering G␤␥ (22,17). These results strongly suggest that both synprint and phosducin reverse the changes in inactivation by acutely disrupting physical interactions within the ternary complex comprised of the G␤␥ϪsyntaxinϪfunctional channel, implying that the enhanced inactivation is the result of proteinϪprotein interactions at the cell surface. To substantiate this conclusion, we demonstrated that the N terminus of ␣ can concomitantly interact with syntaxin and G␤␥ in vitro, to form a ternary complex (Fig. 6). Also, we provided evidence supporting the notion that the channel can simultaneously interact with syntaxin and G␤␥ in oocytes (Figs. 5).
Mechanism of Syntaxin/G␤␥ Modulation-In the case of syntaxin-mediated G␤␥ inhibition of N-type Ca 2ϩ channels, syntaxin optimized the G-protein modulation of the Ca 2ϩ channel, rather than acting as an essential participant. It was therefore suggested that syntaxin mediates a co-localization of G␤␥ and the Ca 2ϩ channel by virtue of its ability to bind G␤␥ (15). Our electrophysiological results concerning the modulation of the inactivation of a Kv channel show an absolute requirement for both syntaxin and G␤␥. This, together with the in vitro binding results that demonstrate enhanced syntaxin binding to the channel by G␤␥, are consistent with a mechanism in which the concomitant physical interaction of the channel with both syntaxin and G␤␥ is mandatory to enhance inactivation.
How do the binding of G␤␥ and syntaxin enhance inactivation? Various modulators of the ␣␤ channel (protein kinases, phosphatases, cytoskeleton disrupters, PSD-95-like proteins, syntaxin, G␤␥) alter the extent of inactivation, but not its rate (4,(11)(12)(13). Therefore, we suggested that the interaction with each of these agents shifts the equilibrium between two gating modes of the ␣␤ channel, the one inactivating and the other noninactivating, toward the inactivating mode (4,11,13,26). We demonstrated such a shift in the case of phosphorylationinduced enhancement of ␣␤ inactivation on the single-channel level (26). The determined stoichiometry of Kv␣1 and Kv␤1 is consistent with the ␣ 4 ␤ n model, where n ϭ 0 -4, depending on the relative concentrations of ␣ and ␤ (as opposed to that with Kv␤2 which is ␣ 4 ␤ 4 only (27)). The inactivating ball particle in these channels can be provided by the N terminus of the ␤ subunit. The G␤␥ modulation (13) that is occluded by microfilament disruption and saturation of ␣ with ␤, suggested to us that in order for the ␣␤ channel to exist in the inactivating mode, one ␤ subunit per tetramer of ␣ subunits is sufficient, as long as the ball is detached from the microfilaments. However, with only one ball per tetramer, chances are that the ball is attached to the microfilaments and that the channel therefore resides mostly in the noninactivating mode. Increasing the ␤/␣ ratio increases the probability that at least one ball per channel will be detached from the microfilaments to implement inactivation and shift the channel to the inactivating mode. In support of our theory, compelling evidence that the N terminus of ␤ strongly associates with the microfilaments was presented (28). Furthermore, on the basis of our biochemical analysis of the G␤␥ modulation (13) showing enhanced assembly between ␣ and ␤ in the presence of co-expressed G␤␥, we assigned a role for G␤␥, early in biosynthesis, in enhancing the assembly of ␣ tetramers with an increased number of ␤ subunits, thereby shifting the channel to the inactivating mode.
In view of the results of the present study that show the inter-relationship between G␤␥ and syntaxin interactions with the channel, this model should now be modified to account for the following facts: (i) concomitant interaction of G␤␥ and syn- FIG. 7. Model for the G␤␥/syntaxin-mediated increase in extent of inactivation. A schematic diagram illustrating the postulated ␣␤ channel assembly and the resultant effects on channel interaction with microfilaments, leading to shifts in the equilibrium between two existing modes of behavior: fast inactivating (I) and noninactivating (NI), without (flow chart to the left) or with (flow chart to the right) G␤␥ and syntaxin. In the inactivating mode at least one ball domain is detached from the microfilaments and confers inactivation. Shown are the ␣ subunits with their transmembrane core region and their N terminus (having the T1 domain that includes T1A and T1B) and ␤ subunits with their core region (C) and variable (V) region that includes the inactivating ball domain. The model may not represent the actual stoichiometry of ␣␤ complexes. taxin with the channel at the plasma membrane is necessary for the functional effect (this study); (ii) concomitant binding of G␤␥ and syntaxin to the channel is suggested from co-precipitation and in vitro binding results (this study); (iii) physical association of syntaxin with the channel occurs mainly in plasma membranes of oocytes (as evident from co-immunoprecipitation experiments (4)); (iv) no reliable indication of an enhanced assembly of ␤ with ␣ in oocytes overexpressing syntaxin, in contrast to oocytes overexpressing G␤␥, could be obtained (4). Thus, in the modified model the binding of G␤␥ at the endoplasmic reticulum to enhance assembly may be necessary but is insufficient to shift the channel from the noninactivating (NI) to the inactivating mode (I); the subsequent binding of syntaxin at the plasma membrane to the G␤␥-bound channel is essential for the enhanced inactivation to occur. In oocytes overexpressing syntaxin alone, the endogenous G␤␥ is probably sufficient to ensure the enhanced inactivation driven by syntaxin. This is indicated by the ability of overexpressed c␤ARK and injected phosducin to remove the effect of syntaxin. Furthermore, in co-immunoprecipitation experiments, we observed a tendency for overexpressed c␤ARK, which reduces endogenous G␤␥ levels (13), to reduce the basal ratio of ␤ assembled with ␣ by 0.77 Ϯ 0.155 (n ϭ 5) in oocytes injected with a nonsaturating mRNA ratio of ␤/␣. This effect was, however, statistically insignificant (p Ͼ 0.14), possibly because the changes in basal assembly of the reduction in endogenous G␤␥ levels were close to the resolution limit (data not shown). In any scenario, whether or not it includes G␤␥-induced assembly of ␤ with ␣ (because of the interaction of G␤␥ with both the ␣ and ␤ subunits (13)), the data presented here strongly suggest that, to enhance inactivation, both G␤␥ and syntaxin are required to be bound to the functional channel. We further suggest, in view of the finding that G␤␥ enhances syntaxin binding to the N terminus of ␣ in vitro, that G␤␥, which binds to the channel already in the early stage of biosynthesis (13), facilitates allosterically the binding of syntaxin to the channel at the plasma membrane (Fig. 7). However, we cannot exclude the possibility that syntaxin is recruited to the channel via formation of the binary complex syntaxin-G␤␥ (15).