The 25-kDa Synaptosome-associated Protein (SNAP-25) Binds and Inhibits Delayed Rectifier Potassium Channels in Secretory Cells*

Delayed-rectifier K+ channels (KDR) are important regulators of membrane excitability in neurons and neuroendocrine cells. Opening of these voltage-dependent K+ channels results in membrane repolarization, leading to the closure of the Ca2+channels and cessation of insulin secretion in neuroendocrine islet β cells. Using patch clamp techniques, we have demonstrated that the activity of the KDR channel subtype, KV1.1, identified by its specific blocker dendrodotoxin-K, is inhibited by SNAP-25 in insulinoma HIT-T15 β cells. A co-precipitation study of rat brain confirmed that SNAP-25 interacts with the KV1.1 protein. Cleavage of SNAP-25 by expression of botulinum neurotoxin A in HIT-T15 cells relieved this SNAP-25-mediated inhibition of KDR. This inhibitory effect of SNAP-25 is mediated by the N terminus of KV1.1, likely by direct interactions with KVα1.1 and/or KVβ subunits, as revealed by co-immunoprecipitation performed in the Xenopus oocyte expression system and in vitro binding. Taken together we have concluded that SNAP-25 mediates secretion not only through its participation in the exocytotic SNARE complex but also by regulating membrane potential and calcium entry through its interaction with KDR channels.

In islet ␤ cells, glucose-mediated Ca 2ϩ -evoked insulin secretion is initiated by the closure of ATP-sensitive K ϩ channels, which in turn causes membrane depolarization and the resultant opening of L-type Ca 2ϩ channels (1). The cessation of insulin secretion is brought about by the closure of these Ca 2ϩ channels by membrane repolarization, which is primarily effected by the opening of a voltage-dependent K ϩ channels (K V ) 1 of the delayed rectifier subtype (K DR ) (2). Regulation of K DR activity therefore directly affects the duration and extent of Ca 2ϩ channel opening-ensuing Ca 2ϩ influx. In neurons, the short duration of action potentials regulates rapidly activating and inactivating N-type Ca 2ϩ channels, resulting in ultrashort (s) Ca 2ϩ fluxes acting on docked synaptic vesicles (3). In contrast to neurons that have a higher proportion of readily releasable docked vesicles (ϳ10%), less than 5% of insulincontaining secretory granules in islet ␤ cells are morphologically docked, with the vast majority of insulin granules located more distantly from the membrane within a reserve pool (4). A more sustained Ca 2ϩ influx effected by a much longer train of action potentials would be necessary to reach and mobilize this reserve pool of insulin granules to the plasma membrane and then to effect their exocytosis (3,4). This contributes to the more sustained phase of secretion, also observed in other neuroendocrine cells (5). In the islet ␤ cell, since this glucosesensitive sustained phase of insulin secretion is regulated by the K DR channel, it would be an ideal drug target for the treatment of non-insulin-dependent or type 2 diabetes. In particular, drugs that interfere selectively with it would be superior to the current treatment with sulfonylureas that act on ␤ cell K ATP channels in a glucose-independent manner, often resulting in deleterious side effects such as hypoglycemia (6). However, very little is known about the K DR channel or the identities of the molecules interacting with this channel (2).
In this study, we have begun to explore the molecular interactions between K DR channels and the critical t-SNARE, SNAP-25. Here we show that SNAP-25 interacts with insulinoma HIT-T15 cell K DR , specifically via the K v␣ 1.1 subunit. Overexpression of SNAP-25 or exogenously applied recombinant SNAP-25 protein inhibits HIT cell K DR activity. Cleavage of SNAP-25 by botulinum neurotoxin A (BoNT/A) light chain expression relieves the actions of endogenous and exogenous SNAP-25 on K DR . Furthermore, the N-terminal domain but not C-terminal domain of K V 1.1 competitively reversed the effect of SNAP-25. This work demonstrates that SNAP-25 modulates secretion not only by its involvement in membrane fusion but also by its interaction with K DR channels.

Cell Culture, Transfection, and Plasma Membrane Fractionation
HIT-T15 cells (a gift from P. Robertson, Seattle, WA) were cultured at 37°C in 5% CO 2 , 95% air in RPMI 1640 medium supplemented with 20 mM glutamine, 10% fetal calf serum (Invitrogen), penicillin (100 units/ml), and streptomycin (100 mg/ml). These cells were then transfected with plasmids (pcDNA3) containing cDNAs of full-length SNAP-25 or BoNT/A light chain (a gift from H. Niemann, Hanover, Germany) exactly as we report (22). 48 h later, we noted the transfection efficiency to be ϳ30 -40% as determined by visualization of the co-expressed green fluorescent protein (GFP) (CLONTECH, Palo Alto, CA), which was also used to identify the transfected cells for the electrophysiological studies. To determine the levels of Kv1.1 and SNARE proteins, these transfected HIT cells were further subjected to subcellular fractionation (i.e. whole-cell lysate and plasma membrane). The HIT-T15 cells were dislodged from the plates with trypsin and then washed twice with phosphate-buffered saline after alternate centrifugation (2000 rpm, 5 min). The resulting pellet was resuspended in a homogenization buffer (in mM: 150 NaCl, 20 Tris-HCl, 1 EDTA, 1 EGTA, and protein inhibitors) and homogenized using a sonicator. Large debris and nuclear fragments were removed by low centrifugation (2000 rpm for 5 min, IEC-8R centrifuge) at 4°C, and the resulting pellet was resuspended and subjected to ultracentrifugation (50,000 rpm for 1 h, TL-100 ultracentrifuge). The pellet was then suspended in SDS-loading buffer and centrifuged at 2000 rpm at 4°C for 5 min, and the supernatant containing the purified plasma membrane fraction was collected. Protein concentrations of the whole-cell lysate and plasma membrane fractions were determined by a modified Lowry method. About 15 g of protein from cell lysate and 70 g of protein from membrane preparation were separated on an 8 or 14% SDS-PAGE and transferred to a nitrocellulose membrane, and the blots were then identified with specific primary antibodies against SNAP-25 (1:1000) (22)

Immunoprecipitation
Rat Brain Homogenates-Rat brain synaptosomal heavy membrane LP1 fraction was prepared exactly as described by Huttner et al. (23). LP1 protein (350 -600 g) was incubated in lysis buffer (150 mM NaCl, 50 mM Tris, pH 7.5, 0.1% Triton X-100 containing a mixture of protease inhibitors) to a concentration of Ͼ1 mg/ml in ice for 10 min, and then 1 g of affinity-purified antibody (SNAP-25) was added. This mixture was then agitated overnight at 4°C followed by the addition of 50 -100 l of 50% slurry of protein G-Sepharose and further agitated for 4 h. The Sepharose beads were then spun-down, washed 4 times with ice-cold lysis buffer, solubilized in sample buffer, and heated at 80°C for 3 min. The proteins within the immunoprecipitated complexes were then separated on SDS-PAGE and identified by specific primary antibodies anti-SNAP-25 (1:1000) (22) and anti-K V 1.1 (1:1000, Alomone Labs).

Binding of Glutathione S-Transferase (GST) Fusion Proteins
The fusion proteins were synthesized as previously described (27,28). The protein concentration was estimated using the Bio-Rad protein assay kit. Purified GST fusion proteins (200 pmol) or purified GST (200 pmol) immobilized on glutathione-Sepharose beads (Amersham Biosciences; 40 l of beads were added, incubated for 30 min at 4°C, and washed 3 times in 1 ml of phosphate-buffered saline with 0.1% Triton X-100) were incubated with 5 l of the lysate containing 35 S-labeled SNAP-25 (translated on the template of in vitro synthesized RNAs using a translation rabbit reticulocyte lysate kit (Promega) according to the manufacturer's instructions) in 1000 l of phosphate-buffered saline with 0.1% Triton X-100 for 1 h at room temperature with gentle rocking. After washing, the GST fusion proteins were eluted with 15 mM reduced glutathione in 40 l of elution buffer (120 mM NaCl, 100 mM Tris-HCl, pH 8.0) and analyzed by 12% SDS-PAGE. DNAs of K v 1.1 and K v␤ 1.1 fragments to create GST fusion proteins have been previously described (27).

Electrophysiology
Previously transfected GFP-positive HIT cells (passage 65-75) were studied with standard whole-cell and cell-attached patch clamp techniques (29) as we have previously described (30). The standard pipette solution contained the following (in mM) except where noted in the text: 140 KCl, 1 MgCl 2 , 10 HEPES, 4 Na 2 ATP, 0.3 EGTA, and 8 nM CaCl 2 , pH 7.2. The external solution contained (in mM): 140 NaCl, 4 KCl, 1 MgCl 2 , 2 CaCl 2 , 10 HEPES, pH 7.4. Axopatch 1D and 200B patch clamp amplifiers were used together with pCLAMP software (v.6.0, Axon Instruments, Union City, CA) to record channel currents. Currents were typically elicited from a holding potential of Ϫ70 mV. Data were presented as mean Ϯ S.E. and compared by Student's t test for single comparisons and by analysis of variance for multiple comparisons. p Ͻ 0.05 was considered to be statistically significant. Single-channel recording and analyses were performed in the cell-attached configuration in which HIT-T15 cells were bathed in the internal solution (high K ϩ ) and pipettes contained a normal external solution. Test pulses of 30s from Ϫ70 to Ϫ10 mV were applied every 3 min. Analog signals were filtered at 2 kHz using a Bessel filter. Single channel data were analyzed using Fetchan and pSTAT software (pCLAMP 6.0, Axon Instruments). Recombinant proteins including GST-SNAP-25, GST␣ T1A, GST␣ T1B, GST-K V 1.1  , and GST-K V 1.1 411-495 were generated as previously described (27,28) and dialyzed into a cell via the patch pipette. A single voltage step to ϩ30 mV from a holding potential of Ϫ70 mV was used to assess the effects of these test proteins on K DR .

Insulinoma HIT Cell K DR Is Similar to That of Islet ␤
Cells-To determine whether the K DR in HIT-T15 (HIT) cells are similar to those reported in islet ␤ cells (2, 31, 32), we functionally and molecularly characterized this channel protein. Like islet ␤ cell K DR (2,32), HIT cell K DR possesses several similar functional properties (Fig. 1A). These properties include K ϩ ion selectivity, as demonstrated by an inhibitory effect of Cs ϩ ion, when intracellular KCl was replaced with CsCl ( Fig. 1A, CsCl), a low threshold of activation (Ϫ30 mV), and slow inactivation (Ͼ250 ms, shown in Fig. 1A, inset, and graphic summary). Characteristic of K DR , the HIT cell K DR was inhibited by high doses of the semi-selective K ϩ channel blocker tetraethylammonium (20 mM TEA in Fig. 1A, inset), which caused ϳ85% reduction (p Ͻ 0.05) in the outward current at ϩ30 mV (31). Because large conductance Ca 2ϩ -sensitve K ϩ currents (K Ca ) also contributes to the outward currents in islet ␤ cells (32), we determined the contribution of this channel to total outward currents. Like others (31), we also found that K Ca contributed little to the HIT cell outward currents, as evidenced by (a) insensitivity to low intracellular Ca 2ϩ concentrations (with 8 nM Ca 2ϩ , Ͻ8% reduction in outward currents, data not shown) or the Ca 2ϩ channel blocker nifedipine (Fig.  1A), (b) insensitivity to low doses of TEA (Ͻ10% reduction at 1 mM TEA, not shown), or (c) inhibition by 100 nM iberiotoxin, a specific large conductance K Ca blocker (Fig. 1A, iberiotoxin (IbTX)). Nonetheless, to negate even this small contribution of K Ca to the outward currents, all pipette solutions contained low Ca 2ϩ concentration (Ͻ8 nM).
SNAP-25 Inhibits Whole-cell K DR Subtype, K V 1.1 Currents-We and others previously showed that in pancreatic islets and insulinoma cell lines including HIT cells, SNAP-25 expression is required for insulin secretion (22,33,34). Furthermore, t-SNARE proteins syntaxin 1A and SNAP-25 can also directly interact with membrane ion channels involved in regulating the secretory process (15,16). To determine whether SNAP-25 can modulate K DR channels, we overexpressed SNAP-25 in HIT cells (Figs. 1B, 2B, and 3B). We already reported that overexpressed SNAP-25 proteins are appropri-ately targeted to the plasma membrane (22), which would allow for their interaction with membrane-spanning ion channel proteins such as K V 1.1. SNAP-25-or BoNT/A-transfected cells were identified by their expression of co-transfected GFP (22). All control cell studies were performed on cells transfected with GFP alone, and neither K DR activity (data not shown) nor insulin secretion was altered by GFP transfection (22). Fig. 1B shows representative whole-cell K DR current traces after transfection and a graphical summary of the peak currents, which have been normalized by cell membrane capacitance for minimizing variations in cell size. Transfection of HIT cells with SNAP-25 resulted in a 3-fold increase in SNAP-25 expression in total HIT lysates ( Fig. 2A, lane 3). Transfection efficiency as determined by the frequency of GFP-positive cells is consistently ϳ40%. With this transfection efficiency of ϳ40%, the transfected cells studied by patch clamp would be estimated to contain about a 5-6-fold increase in SNAP-25 compared with endogenous levels. In cells overexpressing SNAP-25, the peak K DR current (86.9 Ϯ 6.2 pA/picofarads, n ϭ 18, p Ͻ 0.05) was inhibited at ϩ30 mV by 22% compared with control cells (111.3 Ϯ 7.9 pA/picofarads, n ϭ 16) (Fig. 1B). We reasoned that if SNAP-25 genuinely acts to inhibit K DR , then BoNT/A cleavage of SNAP-25 should enhance K DR currents. We therefore overexpressed BoNT/A light chain in HIT cells. BoNT/A expression reduced endogenous SNAP-25 levels of HIT lysates to ϳ10% of control levels ( Fig. 2A, lane 4). Therefore, in cells expressing BoNT/A, it can be assumed that all endogenous SNAP-25 would be cleaved (22). K DR currents in these BoNT/ A-expressing cells were augmented at ϩ30 mV by 22% (135.4 Ϯ 8.9 pA/picofarads, n ϭ 14, p Ͻ 0.05) compared with control cells (Fig. 1B). Indeed, these effects of BoNT/A are opposite to the effects of SNAP-25 overexpression, which would therefore sug-FIG. 1. HIT-T15 cell K DR identification. Typical whole-cell K DR currents evoked with one-step voltage depolarization to ϩ30 mV in a control cell before (E) and after it is inhibited by TEA (20 mM) application to the extracellular solution (f) (A, inset). A, effects of various K ϩ channel blockers: iberiotoxin (IbTX; 10 Ϫ7 M, OE, n ϭ 6), TEA (20 mM, f, n ϭ 6), CsCl (replaced KCl in the pipette solution, 140 mM, q, n ϭ 5), and a Ca 2ϩ channel blocker, nifedipine (NIF; 10 Ϫ5 M, छ, n ϭ 6) on the mean steady-state current-voltage relationship recorded from control HIT cells (E, n ϭ 12). Whole-cell currents were evoked in response to a series of test pulses between Ϫ50 and ϩ50 mV in 20-mV steps from a holding potential of -70 mV. K DR currents were remarkably reduced by TEA and CsCl. B, representative families of whole-cell K DR currents recorded from a control and a SNAP-25-and a BoNT/A-transfected cell. Currents were evoked in response to a series of test pulses from -50 to ϩ50 mV in 10-mV steps from a holding potential of -70 mV. Steady-state current-voltage curves were obtained from control (E, n ϭ 16), SNAP-25-(q, n ϭ 18) or BoNT/A-transfected cells (f, n ϭ 14) (B, right panel). Data are presented as the mean Ϯ S.E. The asterisk indicates p Ͻ 0.05 against control by analysis of variance. C, effects of DTX-K, a specific K V 1.1 blocker, on K DR currents. Representative outward currents recorded from a control and a SNAP-25-and a BoNT/A-transfected cell before (E, basal) and after 300 nM DTX-K addition (q). Current traces were caused by one-step depolarization to ϩ30 mV from a holding potential of -70 mV (C, left 3 panels). The bar graph shows the effects of DTX-K on the steady-state K DR currents (mean Ϯ S.E.) in the control (n ϭ 6) and SNAP-25-(n ϭ 6) or BoNT/A-transfected cells (n ϭ 6) (C, right panel). The asterisk indicates p Ͻ 0.05 versus basal. K DR current levels after the DTX-K addition in control or SNAP-25-or BoNT/A-overexpressing cells (see graph) were very close. The dotted lines are the zero current level. Currents are normalized to cell membrane capacitance.
gest that BoNT/A cleavage of SNAP-25 removes the inhibitory actions of full-length SNAP-25. BoNT/A cleavage of SNAP-25 also revealed a smaller band ( Fig. 2A, lower unfilled arrowhead of the arrowhead doublets to the right of lanes 4 and 7), which is likely the SNAP-25 cleavage product representing amino acids 1-197 (22, 34).
To identify the K DR subtype upon which SNAP-25 acted, we first applied dendrotoxin (DTX), which blocks K v 1.1 and K v 1.2 channels, and subsequently applied dendrotoxin-K (DTX-K), which more specifically blocks only K v 1.1 channels. We found that both toxins exhibited identical effects, and therefore, we show only the more specific DTX-K results in Fig. 1C. DTX-K (300 nM) application had no significant additional effects in SNAP-25-expressing cells compared with controls in which there was a 33% drop in K DR current (p Ͻ 0.05). Thus, the K DR subtype upon which SNAP-25 acted was K v 1.1. We confirmed the presence of the K V 1.1 channel protein in HIT cell lysates as shown in Fig. 2A in lane 2 (HIT Cell lysate, Control). This K V 1.1 antibody (Alomone Labs) has been depleted of antibodies that reacted with closely related K V isoforms. K V 1.1 was previously identified in HIT cells (35) and also in human islets and human insulinoma ␤ cells (2). Furthermore, HIT-T15 does not express  7) were lysed by sonication or subjected to subcellular fractionation to obtain a purified plasma membrane fraction. Western analysis using specific antibodies to SNAP-25 (22) and K V 1.1 (Alomone Labs) was performed on these HIT cell lysates (15 g, lanes 2-4) and plasma membrane fractions (70 g, lanes 5-7). Rat brain synaptosomal membrane fraction (LP1, 10 g of protein) was used as the positive control. In BoNT/A-transfected HIT cells, we observed a slightly smaller (by ϳ1 kDa, indicated by lower open arrowhead on lanes 4 and 7) SNAP-25 immunoreactive band, which is likely the larger SNAP-25 cleavage product (amino acids . B (top left), representative whole-cell K DR activation current traces were evoked by depolarization to ϩ10 mV for 300 ms from a holding potential of -70 mV and normalized to the same amplitude to compare their activation time course. Time constants of activation were obtained by fitting each current trace with a monoexponential equation over test voltages from ϩ10 to ϩ40 mV (B, bottom left). Each point was expressed as the mean Ϯ S.E. The asterisk indicates p Ͻ 0.05 against control. SNAP-25 (q, n ϭ 8) decelerated K DR activation, whereas SNAP-25 cleavage by BoNT/A expression (f, n ϭ 8) accelerated K DR activation when compared with control cells (E, n ϭ 12). B (top right), representative whole-cell K DR inactivation current traces were evoked by a 10-s pulse to ϩ10 mV from a holding potential of -70 mV, and these traces were normalized to the same amplitude to compare their decay. Time constants of inactivation were fitted by a biexponential equation to calculate fast (not shown) and slow time constants, represented by the histogram (B, bottom right). Data were the mean Ϯ S.E. from control (n ϭ 8), SNAP-25 (n ϭ 6), and BoNT/A-transfected cells (n ϭ 6), respectively. The asterisk indicates p Ͻ 0.05 against control. SNAP-25 accelerated K DR inactivation, whereas SNAP-25 cleavage by BoNT/A expression decelerated K DR inactivation when compared with control cell K DR inactivation. the closely related K V 1.2 channel protein, as confirmed by us at the protein (by anti-K V 1.2 antibody, Alomone Labs) and mRNA (by reverse transcription-PCR) levels (36).
Because the K DR current in HIT cells is likely to be contributed by K V channels other than K V 1.1 (36), we next examined whether the SNAP-25-sensitive component of the K DR current is predominantly K V 1.1. In Fig. 1C, we expressed BoNT/A to cleave endogenous SNAP-25, which predictably increased the K DR currents (representative trace on the left, and summary on the right) over controls as was shown in Fig. 1B. More importantly, the application of DTX-K (300 nM) inhibited the K DR current of the BoNT/A-transfected HIT cells to precisely the same extent as the control and SNAP-25-overexpressing cells (Fig. 1C), indicating that the SNAP-25-sensitive K DR current in the HIT cells is indeed from K V 1.1 channels.
SNAP-25 Decelerates Activation and Accelerates Inactivation of Whole-cell K DR Currents-There are two possible explanations for the differences in whole-cell K DR current magnitude observed in Fig. 1B. First, SNAP-25 or BoNT/A expression could affect K DR synthesis and cell surface expression. In this way, changes in peak K DR current would reflect a change in channel density rather than in K DR activity. Fig. 2A  To assess the effects on K V 1.1 expression at the cell surface, which could be due to changes in endocytotic or exocytotic events, we determined K V 1.1 levels in the plasma membrane fraction ( Fig. 2A, upper panel, lanes  5-7). To our surprise, K V 1.1 expression in the plasma membrane preparation was reduced to 84% of control in the SNAP-25-overexpressing HIT cells (lane 6, Fig. 2A). This reduction of K V 1.1 expression at the cell surface would therefore at least contribute to the decreased current density observed in SNAP-25-transfected cells (Fig. 1B). However, if the primary action of SNAP-25 is to regulate K V 1.1 transport to the cell surface, we would expect an augmented expression of K V 1.1 in the plasma membrane after cleavage of endogenous SNAP-25 by overexpression of BoNT/A. This was not the case since BoNT/A overexpression, which reduced the membrane SNAP-25 levels to ϳ10% of controls, did not cause an increase in the plasma membrane levels of K V 1.1. In fact, the K V 1.1 levels in the plasma membrane (lane 7, Fig. 2A) were reduced to 49% of control, which we believe may in part be due to nonspecific proteolysis during the membrane preparation caused by the BoNT/A expression. Of note, the plasma membrane SNAP-25 levels of the SNAP-25-overexpressing HIT cells was greater than 5-fold normal levels, which is much greater in proportion to the reduction of membrane K V 1.1 levels, suggesting a pharmacological effect of the overexpressed SNAP-25. In the SNAP-25-and BoNT/A-transfected HIT cell lysates and plasma membrane, syntaxin 1A levels did not change (data not shown).
The most plausible explanation is that endogenous SNAP-25 physiologic actions are to inhibit K DR channel activity. We tested this possibility by examining the effects of SNAP-25 on the activation and inactivation kinetics of the whole-cell K DR current. Fig. 2B demonstrates that the overexpressed SNAP-25 acts to decelerate K DR activation and accelerate K DR inactivation, with the time constants for activation and inactivation at ϩ10 mV of 21.5 Ϯ 1.0 ms (n ϭ 8) and 3.2 Ϯ 0.3 s (n ϭ 6), respectively. These values were significantly different (p Ͻ 0.05) than those for controls, which had activation and inactivation time constants of 16.3 Ϯ 1.0 ms (n ϭ 12) and 4.3 Ϯ 0.3 s (n ϭ 8), respectively. In contrast, cleavage of SNAP-25 by BoNT/A had the opposite effect, resulting in acceleration of activation and deceleration of inactivation (shown in Fig. 2B), with the time constants for activation and inactivation of 11.6 Ϯ 0.7 ms (n ϭ 8) and 5.4 Ϯ 0.5 s (n ϭ 6), respectively. These values were also significantly different (p Ͻ 0.05) than those for controls (as above). These differences in time-dependent kinetics of the K DR channel effected by SNAP-25 and SNAP-25 cleavage by BoNT/A would contribute to the reduction or augmentation, respectively, of the peak K DR current observed in cells overexpressing SNAP-25 or BoNT/A (shown in Fig. 1B).
SNAP-25 Inhibits Single K DR Channels-Single channel studies on cell-attached patches revealed single K ϩ channels in HIT cells that possess a unitary conductance of 10 picosiemens, as determined from the slope conductance (Fig. 3A), which is similar to that reported for K DR in ␤ cells (31). This conductance is like that reported for the K V 1.1 voltage-activated K ϩ channel isoform expressed in oocytes (37). We next attempted to determine whether SNAP-25 affected the probability of channel opening or the conductance of single K DR channels in cell-attached patches (Fig. 3, B-D). Sample traces are shown in Fig. 3B. We examined these single K DR channel currents evoked by a voltage pulse to Ϫ10 mV in all transfected cells. We found that neither SNAP-25 nor BoNT/A transfection had any effects on K DR channel conductance, with all single channel slope conductances calculated to be 10 picosiemens. This is reflected by the observation that the mean unitary channel current amplitude at Ϫ10 mV for each of the transfections was not significantly different from control cells, ranging from 0.47 to 0.51 pA (Fig. 3C). However, SNAP-25 overexpression greatly reduced the number of K DR channel openings (Fig. 3C), and furthermore, the mean K DR channel open time (7.6 Ϯ 3.7 ms, n ϭ 5, p Ͻ 0.05) shown in Fig. 3D was greatly shortened by 59% compared with control cells (18.2 Ϯ 2.3 ms, n ϭ 5). In contrast, cleavage of endogenous SNAP-25 by BoNT/A transfection resulted in a profound increase in the number of channel openings (Fig. 3C) and a prolongation in K DR channel open time by 150% (45.6 Ϯ 3.2 ms, n ϭ 5, p Ͻ 0.05 versus control in Fig. 3D). These altered single channel activities collectively contribute to the observed changes in the macroscopic, whole-cell currents arising from this population of K DR channels shown in Figs. 1B and 2B.
The Time-dependent Inhibitory Actions of Exogenous SNAP-25 on K DR Are Relieved by BoNT/A Proteolysis-To more clearly distinguish the effects of the overexpressed SNAP-25 on the cell surface expression of K V 1.1 (in Fig. 2A) and the direct effects of SNAP-25 on the K V 1.1 channel, we examined the effects of acutely applied exogenous recombinant SNAP-25 protein (introduced via the patch pipette after membrane rupture) in control (Fig. 4, A-C) and BoNT/A-expressing cells (Fig. 4C). This strategy also allowed us to examine the time course of action of SNAP-25 on K DR . Furthermore, SNAP-25 overexpression in other cell types leads to chronic effects such as alteration in axonal growth (38) and vesicular trafficking processes (39). It has been previously shown (40) that for a 10-megaohm access-resistance pipette, a protein with a molecular mass of 50 kDa can diffuse into the cytosol with a time constant of 5 min. Because the diffusion time constant varies with the third root of the molecular mass of the diffusing substance, we expected that the concentration of the smaller SNAP-25 (25 kDa) in cytosol would be greater than 72% of the pipette concentration within 5 min. Therefore, we expected to see exogenously applied SNAP-25-mediated effects within 5 min, which was indeed the case.
GST-SNAP-25 inhibited K DR currents in a dose-and timedependent manner. Fig. 4A shows a representative series of K DR currents demonstrating a time-dependent reduction in K DR currents in a control HIT cell dialyzed with 10 Ϫ9 M GST-SNAP-25. Fig. 4B is a summary (n ϭ 5-6 cells each) of the dose-dependent inhibition of K DR by SNAP-25. 10 Ϫ10 M GST-SNAP-25 had minimal effects on K DR even 5 min after membrane rupture when compared with GST (10 Ϫ8 M) alone, which had no effect (even after up to 30 min, as shown in Fig. 4C). In contrast, 10 Ϫ9 M GST-SNAP-25 caused a significant 25% reduction of K DR currents 4 min after membrane rupture (Fig. 4B). At a higher concentration of GST-SNAP-25 (10 Ϫ8 M), the inhibition reached a significant level just 2 min after membrane rupture, with a maximal level of 39% occurring at 5 min (Fig.  4B). In Fig. 4C, we recorded for up to 30 min and found that this maximal inhibition of K DR by 10 Ϫ8 M GST-SNAP-25 was maintained. These acute changes on K DR currents by exoge-nous application of SNAP-25 precisely mirrored those caused by SNAP-25 overexpression (Fig. 1B), which supports our thinking that a dominant action of the excess SNAP-25 is to inhibit K DR currents.
We next wanted to examine the effects of the active cleavage of SNAP-25 by BoNT/A on K DR activity. To demonstrate this, we dialyzed GST-SNAP-25 into BoNT/A-expressing HIT cells (Fig. 4C). In vivo expression of BoNT/A light chain ensured continued cleavage of SNAP-25 from both endogenous synthesis and exogenous application. The initial K DR current recorded immediately after membrane rupture in BoNT/A-expressing cells was 23 Ϯ 1% (n ϭ 5, p Ͻ 0.05), higher than that recorded in control cells (Fig. 4C). The initial higher K DR values in the BoNT/A cells is due to cleavage of endogenous SNAP-25, which is expected to be the case before dialysis of exogenous GST- SNAP-25 into the cytosol (Fig. 1B). As we dialyzed GST-SNAP-25 (10 Ϫ8 M) into these BoNT/A-transfected cells, we observed a decline in K DR currents that paralleled the decline seen with GST-SNAP-25 dialyzed into control cells over a similar time period of 12 min after membrane rupture (Fig. 4C). This reached a maximum level of 84 Ϯ 1% (n ϭ 5, p Ͻ 0.05) of the initial control cell value, which when compared with the initial value of BoNT/A-transfected cells (123%), was a reduction of 39% (Fig. 4C). This percentage reduction is identical to that seen with GST-SNAP-25 dialysis into control cells. Of note, after 12 min, the K DR currents in the BoNT/A-transfected cells began to recover as a result of BoNT/A action (Fig. 4C). The time course of this recovery of K DR currents resulting from BoNT/A cleavage of the dialyzed SNAP-25 is consistent with the predicted enzymatic activity of BoNT/A, as reported by in vitro proteolysis of recombinant SNAP-25 by BoNT/A (41). By 30 min, K DR activity completely recovered, reaching levels identical to those recorded immediately after initial membrane rupture (Fig. 4C). This indicates complete cleavage of SNAP-25, rendering it incompetent to modulate K DR .
SNAP-25 Interacts with K V␣ 1.1 and K V␤ 1.1 Subunits-These functional data therefore suggested the possibility that SNAP-25 could interact with K v 1.1 channel proteins. To test this possibility, we examined whether SNAP-25 is in a complex with the ␣-subunit of the K V 1.1 channel (Fig. 5A). To demonstrate the general applicability of these findings to neurons and since these proteins are less abundant in HIT cells, we performed immunoprecipitation studies using rat brain LP1 synaptosomal fractions (s-LP1) shown in lane 1 to be enriched in SNAP-25 and K V 1.1 proteins (Fig. 5A). Of note, lane 2 shows that a rabbit anti-SNAP-25 antibody generated against the full-length SNAP-25 (22) immunoprecipitated SNAP-25 from the solubilized LP1 fraction and co-precipitated K V 1.1 proteins. In control studies, an ϳ50-kDa protein was detected that is likely an IgG heavy chain (lane 3) and which was also observed in lane 2. Neither SNAP-25 nor K V 1.1 proteins were precipitated by either the preimmune IgG (lane 3) or protein-G-Sepharose in the lysis buffer (lane 4).
We showed earlier that K v 1.1 is in complex with K v␤ auxiliary subunits in brain synaptosomes and that this channel exists in multimolecular complex consisting also of the t-SNARE, syntaxin 1A (21). Also, we showed that the K v 1.1/ K v␤ 1.1 channel interacted physically with syntaxin 1A expressed in Xenopus oocytes. Thus, syntaxin 1A could mediate  5-6). The single asterisk indicates p Ͻ 0.05, and the double asterisk indicates p Ͻ 0.01 compared with 10 Ϫ8 M GST alone. C, inhibitory effect of exogenous SNAP-25 on K DR is relieved by BoNT/A proteolysis. K DR currents were recorded over a 30-min time period with an initial interpulse interval of 12 s followed by a slower interpulse interval of 3 min after the first 3 min. To directly compare these effects, K DR currents obtained from GST-SNAP-25 dialysis in both control and BoNT/A-expressing cells were normalized as a percentage of the initial effect (immediately after cell membrane rupture) of inactive GST-alone dialyzed into control cells. Data are the mean Ϯ S.E. Dialysis of 10 Ϫ8 M GST into control cells (E, n ϭ 5) had no effect on K DR . GST-SNAP-25 dialyzed into control cells (q n ϭ 4) resulted in prolonged inhibition of K DR currents. GST-SNAP-25 dialyzed into BoNT/A-expressing cells (f, n ϭ 5) resulted in an initial inhibition of K DR currents, which paralleled that seen in control cells. This inhibition reversed after the onset of BoNT/A cleavage of GST-SNAP-25, resulting in the complete removal of the inhibitory effect of GST-SNAP-25 on K DR currents after 30 min. the interaction between K V 1.1 and SNAP-25 in synaptosomes. To examine whether K v 1.1 could interact directly with SNAP-25 without the mediation of other SNARE proteins or neuronal presynaptic proteins, we employed the heterologous expression system of Xenopus oocytes. We showed by SDS-PAGE analysis of metabolically labeled proteins that SNAP-25 expressed in oocytes co-immunoprecipitated using K v 1.1 antibody with K v 1.1 (␣) co-expressed alone (␣; Fig. 5B) or together with K v␤ 1.1 (␤; not shown). The specificity of the channel interaction with SNAP-25 was verified by reciprocal co-immunoprecipitation of K v 1.1 and K v␤ 1.1 with SNAP-25 using antibody against SNAP-25 (Fig. 5C). The faint immunoreactive bands corresponding to K v 1.1 and K v␤ 1.1 in oocytes injected with ␣␤ without SNAP-25 are probably the result of coprecipitation with the endogenous SNAP-25 homolog(s) 2 (previously we also detected endogenous syntaxin homolog(s) (21)) of the channel proteins (which were better expressed in these oocytes than in oocytes coexpressing also SNAP-25; compare the expression of K v 1.1 in Fig. 5C, bottom panel, lanes ␣␤ϩSNAP-25 and ␣␤). As previously shown (42), the K v 1.1 protein is expressed in oocytes mainly in the form of a doublet of ϳ57 and ϳ54 kDa polypep-tides that are N-glycosylated and represent functional channels.
We then proceeded to determine the domain within the K v 1.1/K v␤ 1.1 protein with which SNAP-25 would directly interact. This study is particularly important since we had detected a small amount of endogenous syntaxin homolog(s) in oocytes (21), which raised the possibility that SNAP-25 binding to the K v 1.1 could be mediated by the binding to the endogenous syntaxin. We have measured the in vitro binding of 35 S-labeled SNAP-25 synthesized in reticulocyte lysate to the following bacterially expressed recombinant proteins: GST ␣C, corresponding to amino acids 411-495, i.e. the whole-length C terminus of K v 1.1; GST␣T1A and GST␣T1B, corresponding to amino acids 1-71 (containing 30 amino acids that comprise T1A domain) and 72-143 of the N terminus of K v 1.1, respectively (both are involved in tetramerization of K v␣ subunits (43Ϫ45), and the latter is involved also in K v␤ subunit binding (46,47); GST ␤c, corresponding to amino acids 75-397, a region in K v␤ that is conserved among the different K v␤ isoforms and is involved in binding to K v 1.1; and ␤full, corresponding to the whole length of K v␤ 1.1 (amino acids 1-397). Fig. 6A shows that SNAP-25 bound to full-length K v␤ 1.1 and to its C terminus, to the N terminus, and preferentially to the T1A domain but not 2 S. Tsuk, D. Chikvashvili, and I. Lotan, unpublished results. The results shown are from one of three independent experiments. C, top, reciprocal co-immunoprecipitation using SNAP-25 antibody, carried out in oocytes that were either uninjected (control), injected with K V 1.1 (10 ng/oocyte) together with K V␤ 1.1 (30 ng/oocyte) without or with SNAP-25 (10 ng/oocyte) mRNAs (␣␤ and ␣␤ϩSNAP-25, correspondingly), or injected with SNAP-25 alone (SNAP-25). Bottom, homogenates of each group of oocytes were subjected to immunoprecipitation by K V 1.1 antibody. Oocytes used in this experiment had a low level of endogenous proteins migrating as K V 1.1 and precipitated by the K V 1.1 antibody, as we report previously (42). The protein samples were analyzed on an 8% PAGE. Arrows indicate the relevant proteins. The electrophoretic mobilities of molecular mass standards (size in kDa) are shown along the right (B) or left (C) of each autoradiogram.
to the C terminus of K v 1.1. Next we set out to test if there was a causative relationship between the direct interaction of the channel with SNAP-25 and the functional interaction that leads to inhibition of the current. To this end we tried to acutely rescue the channel in HIT cells overexpressing SNAP-25 from the functional effects of SNAP-25 by dialysis of a peptide corresponding to a domain of the channel that is involved in SNAP-25 binding. The N-terminal domain of K v ␣1.1 was chosen for the following considerations. The oocyte results demonstrated that K v 1.1 alone can bind SNAP-25, and the in vitro binding results pointed to its N terminus. Also, the in vitro binding results showed that the C-terminal domain of K v␤ , a region that is conserved among the different K v␤ isoforms, does bind SNAP-25. Because this domain binds the N-terminal domain of K v 1.1 (46,47), we assumed that the impact of its possible physical interaction with SNAP-25 could be transferred to the N terminus of K v 1.1, conferring its functional effect. In any case, whether it interacts directly with SNAP-25 or indirectly via interaction with K v␤ , excess soluble N-terminal K v 1.1 is likely to interfere with the interaction of the endogenous membrane-bound K v 1.1 protein with SNAP-25. Indeed, dialysis of the recombinant ␣T1A or N termini (amino acids 1-168) but not the ␣T1B (data not shown), C-termini (amino acids 411-495), or GST resulted in the augmentation of the outward current (28 -35% over control levels), presumably preventing the functional effect of SNAP-25 by disrupting its physical interaction with the channel proteins (Fig. 6B). We further showed that neither the N-nor C-terminal domains had any effect on K DR currents of BoNT/A-transfected HIT cells (data not shown). Taken together, these binding and functional data point to a link between the functional effects of SNAP-25 on the K v 1.1 channel and its physical interaction with this channel protein. DISCUSSION Here we present the first evidence that SNAP-25 interacts with and inhibits the K DR (subtype K v 1.1) channel. Although the overexpressed SNAP-25 had a small effect in reducing the cell surface expression of K v 1.1 ( Fig. 2A), dialysis of excess recombinant SNAP-25 protein acutely inhibited K DR currents in both control and BoNT/A-transfected HIT cells (Fig. 4), which supports a primary inhibitory role of SNAP-25 on the K v 1.1 channel. More importantly, the experiments on BoNT/A cleavage of endogenous SNAP-25 (Figs. 1-3) demonstrated the specific actions of the endogenous SNAP-25 in inhibiting K DR currents.
Our biophysical analysis showed that SNAP-25 inhibited K DR activity in part by altering channel kinetics so that the channel assumes a reluctant state by slowing channel activation and enhancing channel slow (C-type) inactivation (Figs.  1-3). Biochemical analysis in Xenopus oocytes and in vitro binding studies revealed that the K v 1.1 channel interacts physically with SNAP-25 both via the membrane pore-forming K v␣ and the peripheral auxiliary K v␤ subunits (Figs. 5 and 6A). Both the functional and the physical interactions indicate that the cytosolic N-terminal fragment of K v␣ , particularly the ␣T1A domain, acutely reverses the inhibitory effects of SNAP-25 on K DR currents (Fig. 6B). The crystal structures of the K v␤ subunit and the T1 assembly domain of the N terminus of K v␣ subunit (48,49) reveals that submembrane modulating com- 1 (see precise sequences under "Results") immobilized on GSH-agarose beads were incubated with 5 l of in vitro synthesized 35 S-labeled SNAP-25 in a 1-ml reaction volume. After elution with glutathione, proteins were separated on 12% gel and analyzed by PhosphorImager to detect the co-precipitation of SNAP-25. A, bottom, Coomassie Blue staining of the same gel shows each of the eluted GST fusion proteins. Numbers refer to mobility of prestained molecular weight standards. Each panel is a representative experiment of three similar experiments. B, Nterminal domains of K V ␣1.1 reverses the inhibitory effects of SNAP-25 on K DR currents in HIT cells. Dialysis of recombinant ␣T1A (10 Ϫ8 M, n ϭ 5) or K V 1.1 1-168 (10 Ϫ8 M, n ϭ 6) via the patch pipette enhanced the K DR current in SNAP-25transfected cells. In contrast, dialysis of GST (10 Ϫ8 M, n ϭ 6) or the cytoplasmic C-terminus of K V 1.1 (amino acids 411-495, K V 1.1 411-495 , 10 Ϫ8 M, n ϭ 6) had no effect on the K DR current in SNAP-25transfected cells. Whole-cell currents were evoked in response to ϩ30 mV from a holding potential of -70 mV just after cell membrane rupture (t ϭ 0) and 6 min later (t ϭ 6 min). Data are the mean Ϯ S.E. summarized in the bar graph (B, lower panel). The asterisk indicates p Ͻ 0.05 versus t ϭ 0. pF, picofarads.
plexes interact with the core domain of a channel responsible for the activation and C-type inactivation gating (50). Very recently it has been demonstrated that a cytosolic C-terminal complex conveys inhibition of HCN pacemaker channel gating (51). Cleavage of SNAP-25 by BoNT/A expression relieved inhibition of K DR by SNAP-25 by reversing the SNAP-25 effects on K DR kinetics. This indicates endogenous SNAP-25 has a clamping effect on K DR .
Even though SNAP-25 can interact with the K v 1.1 channel directly, it is likely that the molecular complex also involves other proteins. In fact, we recently report direct interactions of syntaxin 1A with K v 1.1 channel enhancing its rapid inactivation (21). In that report, we showed that in brain synaptosomes, the K v␣ 1.1 and K v␤ subunits exist in a macromolecular complex also comprising syntaxin 1A, SNAP-25, and synaptotagmin. Such an interaction between K v 1.1 and the exocytotic SNARE complex proteins is analogous to those described for the Ca 2ϩ channels. For example, SNAP-25 inhibited P/Q-type Ca 2ϩ channels, and this inhibition could be overcome by the formation of a SNARE complex of SNAP-25 with syntaxin 1A and synaptotagmin I (52). Similar modulation of N-and L-type Ca 2ϩ channels by the molecular interactions of these SNARE proteins was also reported (13,15). Further studies are required to examine whether these SNARE proteins and/or their associated proteins could modulate K DR channels independently or in complex with SNAP-25. Notably, we also report that the G protein ␤␥ subunits directly interact with the N terminus of K v␣ 1.1 and with K v␤ 1.1 to affect fast (N-type) inactivation gating (27). It will be of interest to examine possible complex functional and physical interactions between SNAP-25, syntaxin 1A, and G protein ␤␥ subunits.
Delayed-rectifier K ϩ channels (K DR ) are important regulators of membrane excitability in neurons and neuroendocrine cells. Because the other major t-SNARE, syntaxin, has been shown to interact directly with and modulate other ion channels such as Na ϩ (17,18), Ca 2ϩ (12,13,15,16), and Cl Ϫ (19,20) channels, our findings now extend the repertoire of known SNARE-channel interactions to a member of the important voltage-activated K ϩ channel (K V ) family. Our study shows that BoNT/A-mediated inhibition of secretion is likely due to both the inhibitory effects of SNAP-25 cleavage and also from the release of SNAP-25 inhibition of the K DR channel protein, which ultimately leads to an inhibition of the sustained phase of Ca 2ϩ influx. Wei et al. (53) show that the BoNT/A cleavage product SNAP-25⌬C9 exhibited a novel distinct action on reducing the replenishment of the readily releasable pool in the neuroendocrine chromaffin cell, which also reduced the sustained phase of exocytosis as measured by membrane capacitance. Taken together, these studies suggest that SNAP-25 acts on multiple targets to orchestrate the sequence of events to effect optimal exocytosis, including modulation of the membrane potential as a mechanism to provide the appropriate feedback between these events.