Depolarization-Evoked Secretion Requires Two Vicinal Transmembrane Cysteines of Syntaxin 1A

Background The interactions of the voltage-gated Ca2+ channel (VGCC) with syntaxin 1A (Sx 1A), Synaptosome-associated protein of 25 kD (SNAP-25), and synaptotagmin, couple electrical excitation to evoked secretion. Two vicinal Cys residues, Cys 271 and Cys 272 in the Sx 1A transmembrane domain, are highly conserved and participate in modulating channel kinetics. Each of the Sx1A Cys mutants, differently modify the kinetics of Cav1.2, and neuronal Cav2.2 calcium channel. Methodology/Principle Findings We examined the effects of various Sx1A Cys mutants and the syntaxin isoforms 2, 3, and 4 each of which lack vicinal Cys residues, on evoked secretion, monitoring capacitance transients in a functional release assay. Membrane capacitance in Xenopus oocytes co-expressing Cav1.2, Sx1A, SNAP-25 and synaptotagmin, which is Bot C- and Bot A-sensitive, was elicited by a double 500 ms depolarizing pulse to 0 mV. The evoked-release was obliterated when a single Cys Sx1A mutant or either one of the Sx isoforms were substituted for Sx 1A, demonstrating the essential role of vicinal Cys residues in the depolarization mediated process. Protein expression and confocal imaging established the level of the mutated proteins in the cell and their targeting to the plasma membrane. Conclusions/Significance We propose a model whereby the two adjacent transmembranal Cys residues of Sx 1A, lash two calcium channels. Consistent with the necessity of a minimal fusion complex termed the excitosome, each Sx1A is in a complex with SNAP-25, Syt1, and the Ca2+ channel. A Hill coefficient >2 imply that at least three excitosome complexes are required for generating a secreting hetero-oligomer protein complex. This working model suggests that a fusion pore that opens during membrane depolarization could be lined by alternating transmembrane segments of Sx1A and VGCC. The functional coupling of distinct amino acids of Sx 1A with VGCC appears to be essential for depolarization-evoked secretion.

Different syntaxin isoforms sharing 23-84% identity have been described in various rat tissues, indicating distinct trafficking functions [24,25]. Unlike Sx 1A, none of the TMD of these isoforms have vicinal cysteines [24]. The involvement of Sx isoforms in secretion differs in various cells. In adipocytes and muscle cells, Sx 4 was shown to participate in GLUT-4 exocytosis [26,27]. Over expression of Sx 1A and Sx 3, but not of Sx 2 and 4, decreased insulin release in b-cells [28]. Sx 2, a Sx isoform whose TMD is less than 30% homologous to the Sx1A TMD, lowered Cav1.2 and Cav2.2 activation but had no effect on inward currents [22,23].
We have examined the role played by the two highly conserved vicinal Cys residues in Sx1A TMD on evoked-secretion, using Sx 1A mutants, Sx isoforms, a vicinal Cys block by phenyl arsene oxide (PAO), and a truncated Sx 1A. Secretion was examined by monitoring membrane capacitance (Cm) in Xenopus oocytes coexpressing Cav1.2, Sx 1A, SNAP-25, and synaptotagmin, the excitosome proteins [13]. This functional reconstitution assay detected a depolarization-triggered release under voltage-clamp conditions with high precision and time resolution [29]. It is dependent on VGCC activation and on the presence of Sx1A, SNAP-25, and synaptotagmin. Release was triggered also in the absence of synaptotbrevin 2, suggesting the involvement of an endogenous tetanus toxin-insensitive synaptobrevin. Evoked-release was also sensitive to botulinum C1 and botulinum A [13] [29].
We show that a single point mutation at the TMD of Sx 1A, C271V (CC/VC) or C272V (CC/CV), disrupted voltage-evoked secretion, and propose a model to explain the requirement of the two TMD vicinal cysteines for supporting the secretion process. Our model suggests a simultaneous interaction of the two adjacent transmembranal Cys residues of Sx 1A with two VGCC molecules, and designates the VGCC as an essential member of the exocytotic-competent macromolecular cluster.
The depolarization-induced capacitance change showed two distinct phases a fast transient component that was observed in oocytes expressing the Ca 2+ channel alone (Fig. 1A left) and a capacitance increase that was either stationary or slowly decaying, in oocytes expressing Cav1.2 in combination with syt1, SNAP-25, and Sx1A (Fig. 1A right). Because of the nature of the transient increase in Cm, and to avoid possible current changes in Cm measurements as well as any contribution of a fast transient component to our signal, which could be due to the effect of ionic current, we determined Cm just after the current had returned to its previous basal level. These considerations although underestimating the extent of the exocytosis, provide more precise and convincing measurement of the change in capacitance [29]. It should be mentioned the Sx1A cRNA concentration used in the capacitance assay was within the linear phase of its activity on Cav1.2 current amplitude [22,23].

Sx1A/Sx 2 chimera failed to support voltageinduced secretion
To separate the role of the two vicinal Cys from the other amino acids of the Sx TMD, we studied Cav1.2 interactions using a Sx 1-2 chimera, derived from the Sx 1A cytosolic domain and Sx 2 TMD [22] The chimeric molecule unlike Sx 1A wt, had no effect on current amplitude. When the chimera was further mutated by replacing two vicinal Val with two vicinal Cys at the TMD, it restored interaction with the channel [22]. However, when used in the release-assay, despite the presence of the two Cys residue, depolarization-evoked Cm changes corresponded to basal levels (Fig. 1B). These results indicate a likely contribution of other TMD amino acids to exocytosis required perhaps, for the lining of the fusion pore [31] or for the correct folding and organization of the excitosome into a competent secretory cluster (see below: Fig. 1B,C).
Cav1.2 current amplitude was only marginally affected by substituting the various Sx1A mutants (Fig. 1C inset). In the absence of any changes in Ca 2+ influx, the large increase in Cm observed with wt Sx1A, indicate a better coupling and a more efficient assembly of the excitosome with Sx1A wt, rather than a result of a larger cation influx.

Pulse duration influences depolarization-evoked secretion
The kinetics of release triggered by varying the length of the depolarizing pulses according to von Gersdorff [32], was recorded for excitosome complexes composed of Sx1A wt or Sx 1A (CC/VV) double mutant (Fig. 1D). Oocytes expressing the corresponding excitosome complexes were depolarized by pulse durations of 250 ms, 500 ms or 26500 ms 100 ms apart. Unlike Sx 1A wt, the Sx1A (CC/VV) mutant showed capacitance transients of magnitudes similar to those induced by Cav1.2 expressed without synaptic proteins, indicating an 'inactive' secreting complex (Fig. 1C, D). We cannot rule out possible contribution to Cm that arises from an increase in intracellular divalent cations that promotes a Ca 2+dependent, SNARE independent process, similar to caffeine and ionomycin [33,34]. Therefore, as voltage pulse duration increased and more channels were opened Cm was elevated in both complexes. Even under these conditions, of larger Ba 2+ influx, the ineffectiveness of Sx1A(CC/VV) was apparent.
Expression and targeting of Cav1.2 and syntaxin 1A mutants to the cell membrane The obliteration of the capacitance response by substituting one of the Sx 1A mutants was substituted for Sx1A could result from difficulties in protein expression or targeting to the cell membrane, or both. Therefore, we prepared the fluorescent fusion proteins, RFP-Sx1A wt and RFP-Sx1A(CC/VV), that enabled us to evaluate protein expression and localization using confocal imaging.
Cav1.2 interaction with the two tagged-proteins was determined electrophysiologically by monitoring current kinetics in Xenopus oocytes injected with cRNA encoding the three channel subunits GFP-a1C/b2A/a2d with RFP-Sx1A or RFP-Sx1A(CC/VV). Inward currents were elicited from a holding potential of 280 mV to various test potentials in response to 500 ms and recorded using the two-electrode voltage-clamp assay ( [6,14,15] Fig. 2). Consistent with Sx 1A wt ( [6,15]) current amplitude elicited to +10 mV in oocytes expressing 0.8 or 1.2 ng/oocyte of RFP-Sx1A, was reduced as shown by superimposed current traces ( Fig. 2A upper). The Sx1A cRNA concentration used was within the linear phase of its activity on Cav1.2 current amplitude as previously determined [22,23].
RFP-Sx1A(CC/VV) marginally affected current amplitude ( Fig. 2B upper), similar to the effect of the un-tagged Sx1A(CC/VV) ( [22,23]). Leak-subtracted peak currents from oocytes co expressing the three channel subunits and RFP-Sx 1A, or RFP-Sx 1A (CC/VV) were plotted as current-voltage relationship, showing current modulation at various test potentials ( Fig. 2A,B middle). The rate of activation, demonstrated by the time constant of activation (tact) was accelerated to the same extent by the two RFP-Sx1A proteins ( Fig. 2A,B lower) consistent with the untagged Sx1A [22,23].
The difference in current modulation is depicted by a shift in RFP-Sx1A and RFP-Sx1A(CC/VV) concentration dependency (Fig. 2C). The absence of the Cys271 and Cys 272 interaction with the channel could be held responsible for the Sx 1A (CC/VV) inability to support release (see Fig. 1).
After establishing that the RFP incorporation into Sx1A did not interfere with the Sx1A/Cav1.2 interaction, we measured RFP-Sx1A and RFP-Sx1A(CC/VV) levels in the cells. Oocytes coexpressing the channel subunits and RFP-Sx1A or RFP-Sx1A(CC/VV) were lysed and the proteins were separated on SDS-PAGE, blotted into nitrocellulose membrane and detected using specific Sx1A antibodies (Fig. 2D). Western blot analysis showed that the protein level of the tagged-proteins was lower than non tagged Sx1A wt, most likely due to the incorporated RFP-tag, but RFP Sx1A and RFP-Sx1A(CC/VV) were equally expressed, demonstrating that the mutation did not affect protein expression (Fig. 2D). Similarly, RFP-Sx1A and RFP-Sx1A(CC/VV) levels within the excitosome (Ex) were detected (Fig. 3A).
Targeting of the excitosome complexes Cav1.2, SNAP-25, Syt I Sx1A or Sx1A(CC/VV) to the cell membrane was detected by confocal fluorescence imaging of GFP-a 1

PAO interferes with syntaxin 1A /VGCC interaction and decreases depolarization-induced secretion
Phenyl arsene oxide (PAO), a thiol reagent that selectively reacts with two adjacent Cys residues to form a stable arsenic complex, interfered with the cross-talk of Sx 1A with the channel, most likely reacting with the vicinal Cys of Sx TMD and [23]. We tested the PAO effect on reconstituted secretion, applying 10 mM PAO Currents were elicited from a holding potential of -80 mV to +10 mV in response to 500 ms pulse (upper panels). Leak-subtracted peak currentvoltage relations collected data from oocytes expressing GFP-a 1 1.2/b2A/a2/d (5/5/5 ng/oocyte) without (-#-) and with RFP-Sx1A (0.8 ng/oocyte -Nand 1.6 ng/oocyte -&-) (A) and RFP-Sx1A(CC/VV). Currents were elicited in response to 500 ms pulse from a holding potential of -80 mV to various test potentials at 5-sec intervals (B) (middle). Activation rate constants (t, mean6SEM, n = 12) of currents generated in oocytes by GFP-a 1  directly into the recording bath (Fig. 4A). Membrane capacitance in oocytes expressing Cav1.2 along with the synaptic proteins, was monitored prior to PAO addition ( Fig. 2A, upper right) and then, after 10s exposure to 10 mM PAO (Fig. 4A, lower left). The DCm observed was similar to the change observed in oocytes expressing Cav1.2 without synaptic proteins (Fig. 4A, upper  left). An averaged DCm signal in 10 oocytes treated with PAO is shown in Fig. 4B, C. The oocytes were then exposed for 2.5 min to 2 mM 2,3-dimercaptopropanol (BAL), a reagent that reverses PAO reaction [35]. The external application of BAL to PAOtreated oocytes, resulted in a small but significant reversal of the PAO inhibitory effect ( Fig. 2A, right lower and Fig. 4B left).
The partial effect of BAL might be due to irreversible changes at the oocyte membrane, which could have affected directly the capacitance measurement [23]. The high selectivity of PAO for vicinal cysteines provides additional evidence for the importance of the highly conserved Cys residues of the TMD of Sx 1A to depolarization-evoked capacitance. As shown in Fig. 4B, right, oocytes treated with PAO, or PAO+BAL displayed statistically similar current amplitudes and could not account for the changes in Cm transients (see [29,36]) (n = 11-12).
Although PAO's effect on secretion could be attributed also to its other activities in the cell, it had no direct effect on Cav1.2 amplitude; it acted rapidly (,2 sec) to reverse the syntaxin effect and fully restored current amplitude [22]. The PAO results on secretion are consistent with those of Sx1A mutants, and support the possible involvement of two adjacent Cys in the release process.
Syntaxin isoforms affect the kinetic properties of Cav1.2 To expand and further complement our insight of Sx1A role in depolarization evoked-release, we compared the ability of three Sx 1A isoforms, Sx 2, Sx 3, and Sx 4 to modulate Cav1.2 kinetics. Cav1.2 was expressed with each one of the Sx isoforms and inward  currents were elicited in response to a 160 ms pulse, from a holding potential of 280 mV to various test potentials, according to the voltage protocol [6]; Fig. 5A). Superposition of current traces and current-voltage relationship of several oocytes are shown (Fig. 5A,B). Cav1.2 current amplitude was significantly decreased by Sx 1A, and to a smaller extent, by Sx 3 and Sx 4, but not by Sx 2 (Fig. 5A). All four isoforms showed a similar shift towards positive potentials as depicted in the I/Imax ratios (Fig. 5C). All Sx isoforms caused an increase in the time constant of activation (tact) at voltages above 0 mV, while Sx 1A increased tact also at negative potentials (Fig. 5D).
For assessment of syntaxin isoform modulation of Cav1.2 kinetics, representative current traces of Cav1.2 co-expressed with each Sx isoform were superimposed. Currents were evoked by a voltage step from a holding potential of -80 mV to +5 mV in response to a 160 ms pulse (Fig. 6). A graded effect on the PLoS ONE | www.plosone.org decrease in current amplitude was observed with Sx 3,Sx 4,Sx 1A. Sx 2 displayed no inhibitory effect ( Fig. 6A; [22,23]). All Sx isoforms shifted considerably the Cav1.2 normalized current, confirming functional interaction with Cav1.2 (Fig. 6B). The ratio of G/Gmax was shifted towards positive potentials by all Sx isoforms (Fig. 6C) and the largest shift of I/Imax was observed for Sx 2 (Fig. 6D). Consistent with previous reports [24][25][26][27][28], the functional modulation of the channel kinetics by the Sx isoforms similar to Sx 1A, shows a correct functional engagement with the channel.

Sx 1A supports depolarization-induced capacitance increase
Given this framework, we next tested each of the syntaxin isoforms for supporting depolarization-evoked release. Oocytes expressing Cav1.2 without, and with Sx 1A, SNAP-25, and SytI (excitosome) were depolarized according to the protocol in Fig. 1, and Cm monitored (Fig. 7A left). Using the same depolarizing protocol, Cm was monitored in oocytes injected with a cRNA mixture of Cav1.2, SytI, SNAP-25 and one of the Sx isoforms, Sx 2, Sx 3 or Sx 4 (Fig. 7A). The amino acid sequences of the TM domains of Sx isoforms are indicated in Fig. 7A, right. The averages of capacitance transients (DCm) were: for Sx 1A, 2.6560.17 nF (n = 18), Sx 2, 1.0460.13 nF (n = 16), Sx 3, 0.8660.12 nF (n = 12) and Sx 4, 0.4860.11 nF, while DCm of Cav1.2 alone was 0.760.1 nF (Fig. 7B left panel). Thus, except for Sx 1A, none of the Sx isoforms supported depolarization-evoked release. Cav1.2 current amplitude recorded in the presence of the synaptic proteins was only slightly smaller, and not correlated to the calculated Cm (Fig. 7B, inset).

DISCUSSION
We have used a functional reconstituted assay of secretion in Xenopus oocytes [29] as an approach to studying the crosstalk Figure 6. Comparison of syntaxin isoforms effects on the kinetics of Cav1.2. Cav1.2 subunits were co-injected with syntaxin isoforms (data from Figure 3). Ba 2+ currents were elicited from a holding potential of -80 mV by a voltage step applied to +4 mV in response to 160 ms pulse duration.  between Sx 1A and VGCC in exocytosis, and a means of examining the role of VGCC activation during the exocytotic events.
Previously, it was shown that when a single Val residue was used to replace either one of the two highly conserved Cys 271 or Cys 272 residues of the Sx 1A TMD, modifications of Cav1.2 kinetics was similar to wt Sx 1A [22,23]. A loss of current modulation was observed only when both vicinal Cys were mutated.
Despite the similar modulation of Cav1.2 kinetics by either one of the the single Cys mutants or wt Sx1A, we showed that neither CC/CV nor CC/VC supported depolarization-evoked release. These results reveal that both of the adjacent Cys residues are essential for promoting release.

Sx 1A/Sx 2 chimera
Sx 2 has two vicinal Val residues, instead of two Cys at the corresponding position in the Sx 2 TMD. As previously shown, Sx 2 did not lower VGCC current amplitude, consistent with the need of two vicinal Cys residues for modulating inward current [24]. Hence, if the two Cys were solely responsible for decreasing VGCC current amplitudes, mutating Val residues to Cys at the Sx 1-2 chimera (constructed by cytosolic Sx1A and transmembrane domain of Sx 2) would restore inhibitory activity. Indeed, the mutated chimera, Sx1-2 (VV/CC), restored the predicted inhibitory activity [2,22]. Despite this gain of activity, Sx1-2 (VV/CC) did not support release, indicating that in addition to Cys 271 and Cys 272, other amino acids at Sx 1A TMD play a role in the process of evoked release. These results reflect more stringent requirements for the release process, and are consistent with amperometry studies in PC 12 cells, where over-expression of only selective TMD Sx 1A mutants were reported to alter Ca 2+evoked release [31].

Sx isoforms
The essential role of two vicinal Cys residues in depolarization evoked-release was further corroborated by the use of Sx isoforms, lacking the TMD vicinal Cys residues. Here we have demonstrated that none of the isoforms, supported evoked-release in the functional secretion assay. Unique sequences of both the cytosolic and transmembrane domains of Sx 1A appear to act in concert to support depolarization-evoked release. In further support of this, when Sx 1A is cleaved by botulinum C1, both evoked-release and modulation of VGCC currents [22] are lost [29,37].
Previously, it was reported that in the retina, Sx 3 mediates presynaptic transmitter release from ribbon synapses, as opposed to the presynaptic Sx 2 and Sx 4 which are likely to mediate postsynaptic trafficking [38]. In an insulinoma cell line [28] and in pancreatic b cells, Sx 4 facilitates secretion of insulin over expression of Sx 1A and Sx 3 but not Sx 2 and Sx 4, strongly inhibited actions on insulin release [39]. In view of the data presented, it will be of interest to test whether Sx 2, Sx 3 and Sx 4 can support secretion perhaps, by interacting with different SNAP-25 and/or VAMP isoforms [40] .

Proposed model
The most commonly accepted model for exocytosis is the SNAREs' model where the SNARE proteins (Sx.1A, SNAP-25 and synaptobrevin 2) are suggested to fuse with each other forcing the membranes into a close proximity to form a bilayer [41]. Initial reconstitution experiments using liposome containing recombinant SNAREs supported this model and led to the hypothesis that SNAREs by themselves are sufficient to account for the membrane fusion process [42]. However, un-physiologically high concentrations of SNAREs were required for the fusion and membranes breakage during fusion, have suggested that SNAREs alone are not sufficient to account for biological membrane fusion [43,44]. These results raised the question whether the SNAREs just bring the membranes together or actually initiate fusion [45].
A more integral role of Sx 1A in the SNARE complex was implicated by its interaction with Munc-18 [46][47][48]. A tentative working model suggested that Munc18-1 promotes vesicle delivery most likely through the actin network [49] and then leads to vesicle docking [48]. According to this model, Sx1A/Munc18-1 dimer is formed and by unknown mechanism, transits to Munc18-1-SNARE complex. This complex was suggested to promote vesicle fusion by accelerating the fusion reaction [50].
Our previous work has indicated that a minimal set of proteins assemble into a complex composed of the Ca 2+ channel, tSNAREs and Syt1 [1,5,13,14,16,23,29,53]. A functional assay was used that mimics several of the characteristics of exocytosis found in excitable cells; it is driven by membrane depolarization, sensitive to botulinum toxins (C and A), non-linear with divalent cation concentrations, and responds differently to various types of VGCC. In the functional secretion assay however, unlike in neuronal cells, synaptobrevin-2 contributed to an increase but was not absolutely necessary for secretion. This suggested the involvement of endogenous tetanus toxin-insensitive synaptobrevin (or cellubrevin) in the fusion process [29].
We propose here a working model that is based on a non-linear allosteric functional interactions of Sx1A with VGCC [29]. We have scrutinized the model in light of our results with the functional secretion assay that measures the fusion event. We assume that in our assay the vesicles in the oocytes are properly primed and associated with the VGCC [22,23].
Consistent with our findings of the necessity of a minimal fusion complex, each Sx1A is in a complex with SNAP-25, Syt1, and the Ca 2+ channel. The failure to support release when either of the two adjacent Cys residues in Sx 1A TMD are substituted by Val most likely results from structure alteration of Sx1A, that could affect the interaction and the allosteric regulation of the channel [23].
Based on a Hill coefficient of n H = 2.0260.2, indicating the non-linear dependency of Sx 1A concentration on current modulation [23], we propose that three (or more) Sx1A molecules assemble with equal number of Ca 2+ channels to generate an exocytotic complex. The adjacent two Cys residues of Sx1A are proposed to play a role in the supra molecular assembly of channel molecules into a secreting competent complex [54][55][56].
Located far apart on a helical wheel model, we propose that one Cys residue could interact with one channel and the other, with a second channel (Fig. 8A left). The two vicinal Cys at the Sx 1A TMD would enable one Sx 1A molecule to connect with two adjacent VGCC molecules. The association of three Sx1A with three channels would generate a cluster that mediates secretion ( Fig. 8A right), and a top view (Fig. 8B). This would explain why a single Cys mutation could disrupt the assembly of the multiprotein complex thereby, obliterating secretion. The observation of an apparently lower number of clusters detected with confocal imaging of RFP-Sx1A(CC/VV) substituted for RFP-Sx 1A, would support this model. We believe that the hetero-oligomer complex generates a fusion pore ( Fig. 6B; shaded area) that is lined by alternating transmembrane segments of Sx 1A and the calcium channel. Previously, a fusion pore formed by the circular arrangement of 5 to 8 Sx1A TMD segments was suggested [31,57] Our working model, which is based mainly on secretion of vesicles assembled with VGCC, does not preclude the participation of Munc18-1, Rab, or other proteins in the multiprotein complex. Further studies should clarify how and when does the Munc18-1/ closed Sx1A dimer becomes a Munc18-1 trans SNARE complex, and whether Munc18-1/SNARE complex accelerates vesicle fusion via the multiprotein complex presented above.
In summary, unlike existing models of secretion, our model incorporates the VGCC as a pivotal member of the secretory complex. The close proximity of the channel to the exocytotic proteins allows for fast transmission of conformational changes from the channel to the release machinery. Secretion triggered by conformational change induced by cation bound at the channel selectivity filter prior to cation entry, implies a role of the VGCC as the Ca 2+ sensor protein of secretion [7,8]. Lastly, the channel as part of the exocytotic complex could be a constituent of the fusion pore, directly controlling the fusion process.

Heterologous protein expression in Xenopus oocytes
cRNA Injection into Oocytes-cRNAs were prepared using the T7/T3 Stratagene transcription kit, and the product was monitored by gel electrophoresis and optical density measurements.
Stage V and VI oocytes were surgically removed from female Xenopus Laevis and were injected with cRNA mixtures encoding Ca 2+ channel subunits that were adjusted empirically to make the inward current lower than 24 mA. cRNA mixtures were injected into oocytes using a microdispenser (Drummond 510, Broomall, PA), in a final volume of 40 nl. Xenopus oocytes were injected with a mixture of cRNAs encoding the Cav1.2 Ca 2+ channel subunits a 1 1.2 (Lc-type (D1733); rabbit; 5 ng/oocyte) a2/d (rabbit; 5 ng/oocyte) and b2A (rat; 10 ng/oocyte) and one day later either with water (for controls) or with a mixture of cRNAs encoding SNAP-25 (rat; 0.5 ng/oocyte, SytI (rat; 1.0 ng/oocyte) and Sx1A (0.5 ng/ oocyte), or the equivalent amount of the corresponding isoform (as previously described [15]. Oocytes were kept in 18 uC until depolarization-induced exocytosis or inward currents were recorded after further incubation for 4 or 6 days.

Capacitance monitoring in Xenopus oocytes
Membrane capacitance (Cm) was monitored in the two-electrode voltage-clamp configuration as published elsewhere [36]. Briefly, Cm was determined from the current response to a triangular, symmetrical voltage command of ''Paired ramps'' [45]. In a typical voltage stimulus, command voltage V com increases by 40 mV within 20 ms, equivalent to a ramp slope of 2 V/s and with this slope, capacitance is obtained from the differencê  [36]. Starting from a holding potential of 280 mV, depolarizing stimuli were applied by clamping the cells to 0 mV for 26500 ms, separated by 100 ms at 280 mV (unless stated otherwise). Capacitance was monitored before and after the stimulus, together with membrane potential (Vm) and current (Im).

Confocal imaging
Single optical sections through the oocytes were acquired with an Olympus FV1000 (Olympus, Japan) equipped with a 406 oil objective (N.A. 1.3). Two excitation lasers were used sequentially: 488 nm for the GFP and 543 nm for RFP. Narrow-band emission filters 505-525 nm were used in the GFP channel and 560-620 nm in the RFP channel. Sequential scanning was performed with a resolution set to 5126512 pixels (0.621 mm/pixel), and single optical sections ,0.5 mm thick were captured. Exposure time was 8 msec/pixel. Protein expression and co-localization were analyzed using Image J software

Electrophysiological assays
Whole-cell voltage clamp recordings were acquired from oocytes at day 5 or six after cRNA injection, as described previously [13].
To minimize Ca 2+ -activated Cl 2 currents, oocytes were injected with BAPTA (final 5 mM) prior to recordings. Membrane currents were recorded by a two-electrode voltage clamp method using a DAGAN 8500 amplifier (Dagan). Bath solution contained (in mM): 5 Ba(OH) 2 , 50 N-methyl-D-glucamine, 1 KOH, 40 tetraethylammonium, 5 HEPES, titrated to pH 7.5 with methanesulfonic acid CH 3 SO 3 H. Pulse duration for activation was 160 ms in 10 s intervals. Current traces were leak-subtracted on-line by the Clampex 8 software, and channel activation rates were analyzed by applying a mono-exponential fit (Axon instruments, Foster City, CA) to the current traces at the relevant ranges.
Fit~A exp½t : =t actzB where A = current amplitude, t = time constant, t = time to peak. Activation was determined from the beginning of the trace just after the capacitative transient to the peak-current region.
Data Presentation and Statistical Analysis-Peak current and time constant values analyzed by Clampfit 8 and transferred as ASCI file to an Excel worksheet (Microsoft Inc.). Data were averaged for each group of oocytes, and S.E. was determined. Data are presented as means6S.E. Statistical significance relative to the control group in each experiment was done by Student's t test by the Excel software. Statistical significance between multiple groups in each experiment was determined using one-way analysis of variance test using the Origin 6 software (Microcal). Final data was transferred to Origin 6 worksheet, plotted, and printed as final figures.