Role of the S4 Segment in a Voltage-dependent Calcium-sensitive Potassium (h Slo ) Channel*

We investigated the role of individual charged residues of the S4 region of a MaxiK channel (h Slo ) in channel gating. We measured macroscopic currents induced by wild type (WT) and point mutants of h Slo in inside-out membrane patches of Xenopus laevis oocytes. Of all the residues tested, only neutralizations of Arg-210 and Arg-213 were associated with a reduction in the number of gating charges as determined using the limiting slope method. Channel activation in WT and mutant channels was interpreted using an allosteric model. Mutations R207Q, R207E, and R210N facilitated channel opening in the absence of Ca 2 1 ; however, this facilitation was not observed in the channels Ca 2 1 -bound state. Mutation R213Q behaved similarly to the WT channel in the absence of Ca 2 1 , but Ca 2 1 was unable to stabilize the open state to the same extent as it does in the WT. Mutations R207Q, R207E, R210N, and R213Q reduced the coupling between Ca 2 1 binding and channel opening when compared with the WT. Mutations L204R, L204H, Q216R, E219Q, and E219K in the S4 hours later they were microinjected with 50 nl of 0.01–0.5 m g/ m l mRNA in water. Oocytes were maintained at 18 °C in ND-96 supplemented with gentamycin m g/ml). Electrophysiological Recordings— After the manual removal of the vitellin membrane, macroscopic currents were evoked in membrane patches in the inside-out configuration. Current records were obtained

Large conductance Ca 2ϩ -dependent K ϩ (MaxiK) 1 channels are sensitive to both the transmembrane potential and to the internal calcium concentration ([Ca 2ϩ ]). The probability of finding these channels in the open state increases with positive potentials or with increments of [Ca 2ϩ ] (1,2). MaxiK channels share many features with voltage-dependent Na ϩ , Ca 2ϩ , and K ϩ channels of the S4 superfamily (3). One of the most remarkable features is that MaxiK channels also contain the S4 common motif, a structure with a repeated triple sequence of one positively charged amino acid (arginine or lysine) plus two hydrophobic amino acids (4 -13). Because it is highly charged, the S4 region has been suggested to serve as the voltage sensor (14 -16). In voltage-dependent K ϩ channels the evidence supporting this view is becoming overwhelming (17)(18)(19)(20)(21)(22)(23)(24)(25).
We have recently shown that MaxiK channels possess an intrinsic voltage sensor whose rearrangement triggered by voltage induces measurable gating currents (27). Both charge movement and pore opening are facilitated by [Ca 2ϩ ] Ͼ 100 nM (26,27). We examine here the contribution of individual charged residues of the S4 region to the voltage dependence and Ca 2ϩ sensitivity of a cloned human MaxiK channel, hSlo, using site-directed mutagenesis. We conclude that: (a) the S4 region in the hSlo channel is part of the voltage sensor; (b) for all the S4 mutants tested only two charged amino acid residues appear to modify the limiting slope of the probability of opening versus voltage curves; (c) for all S4 mutants, there is a [Ca 2ϩ ] region (Յ100 nM) in which channel activity becomes independent of the [Ca 2ϩ ]. We performed an analysis of the half-activation voltages-[Ca 2ϩ ] data for the WT hSlo and the S4 mutant channels using a gating scheme drawn from the Monod-Wyman-Changeaux (MWC) model for allosteric proteins (28). The model describes the data reasonably well and indicates that shifts in the P o -V curves along the voltage axis induced by the mutations in the S4 domain are due to stabilization of the open configuration and to a decrease in the coupling between Ca 2ϩ binding and channel opening.

MATERIALS AND METHODS
Clones, Mutagenesis, and in Vitro Transcription-All experiments were performed on the high conductance, calcium-dependent K ϩ channel, hSlo, cloned from human myometrium (11). Channels were expressed from hSlo cDNAs starting from the third Kozak consensus sequence (M3). Constructs contained as 5Ј-untranslated region, 86 base pairs upstream M3 of hSlo or 223 base pairs of 5Ј-untranslated region from Shaker H4. S4 mutations were carried out on the latter version that also contained a poly(A) tail at the 3Ј-end and results in a higher level of expression (27).
Site-directed mutagenesis was performed with the polymerase chain reaction, based on the four-primer strategy (29,30). After DNA amplification, sequencing checked the presence of mutations.
Electrophysiological Recordings-After the manual removal of the vitellin membrane, macroscopic currents were evoked in membrane patches in the inside-out configuration. Current records were obtained 2-6 days after injection of the oocytes. The pipettes had 1-2 megohm resistances, and no series resistance compensation was used. The error due to uncompensated series resistance was never greater than ϳ5 mV. Pipettes were filled with (mM): 110 potassium methanesulfonate, 10 HEPES, 2 MgCl 2 , pH 7.0. The free [Ca 2ϩ ] in the bath solution, which faces the intracellular side of the membrane in the inside-out patch configuration, was buffered with Ca 2ϩ chelators and calculated with the software CHELATOR (31). From  Acquisition and Analysis-The current signal was digitized to a frequency equal to 5 times the filter cutoff frequency. The acquisition and basic analysis of the data were performed with pClamp 6.0 software (Axon Instrument, Inc.). To obtain the maximum conductance, G max , membrane patches were stimulated with rectangular voltage pulses and steady-state I-V curves were constructed. Conductance (G) versus voltage relationships were obtained from the relation G ϭ I/(V Ϫ E K ) where I is the measured K ϩ current and E K is the K ϩ equilibrium potential. The data were well described by a Boltzmann function of the form where P o is the probability of channel opening, V1 ⁄2 is the voltage at which P o ϭ 0.5; z eq is the equivalent gating charge associated with the closed to open transitions, and R, T, and F have their usual meanings. G max , V1 ⁄2 , and z eq were obtained using a non-linear curve fitting procedure. Maximum conductance was an adjustable parameter in those patches in which artifacts introduced by fast Ca 2ϩ block and nonlinearity of the instantaneous current-voltage relation (32,33) were minimal. This G max was used to fit V1 ⁄2 and z eq at those [Ca 2ϩ ] where the determination of G max was unreliable. The validity of this method of determining G max was checked using tail current measurements (33).

RESULTS
Point Mutations in the S4 Segment of hSlo-We have tested a series of S4 mutants in a quest to find the role of this transmembrane segment in hSlo channels. Fig. 1 shows two possible alignments for hSlo S4 segment. Alignment 2 predicts an hSlo S4 segment with only 3 positive residues (arginines) at 207, 210, and 213, compared with 7 in Shaker K ϩ channels. We have substituted these positively charged residues of S4 by either neutral (glutamine or asparagine) or negative (glutamate) residues. Other positions, that are basic amino acids in Shaker, are hydrophobic (Leu-204), polar (Gln-216, Gln-222), or acidic (Glu-219) in hSlo. In these positions, the following point mutations were performed: L204R, L204H, Q216R, E219Q and E219K. In alignment 1 (34), residue Arg-201 could also form part of hSlo S4 segment. We have neutralized this residue to glutamine (R201Q). In all the cases tested, mutations in the S4 segment of hSlo produced functional channels.
FIG. 2. Macroscopic currents for wild type hSlo and S4 mutant channels. Currents were recorded from inside-out macropatches. Holding voltage was 0 mV. A, internal [Ca 2ϩ ] was 5.7 M. Currents were elicited by voltage steps to Ϫ100, Ϫ50 and from 0 to 100 mV with increments of 20 mV for WT hSlo channels; and by voltage steps to Ϫ200, Ϫ150, Ϫ50 and from 0 to 60 mV in increments of 20 mV for R207Q and R207E mutants. For R207E an additional current trace elicited by a voltage step to Ϫ250 mV is shown. B, internal [Ca 2ϩ ] was 103 M. WT currents were induced by voltage steps to Ϫ140, Ϫ80, Ϫ50 and from 0 to 60 mV in increments of 20 mV, and R210Q and R213Q currents were induced by voltage steps from Ϫ100 to Ϫ25, in increments of 25 mV, and from 0 to 100 mV in increments of 20 mV. C, internal [Ca 2ϩ ] was 1.2 M. Currents were induced by voltage steps to Ϫ100 mV and from 0 to 100 mV in increments of 20 mV for WT, and to Ϫ180, Ϫ50 mV and from 0 to 100 mV in increments of 20 mV for R210N.
FIG. 1. Primary structure alignment of hSlo and Shaker S4 segments. Alignment 1 was obtained from a multiple sequence alignment with other voltage gated K ϩ channels (34), but the S4 segment includes residues R362 in Shaker and R201 of hSlo. Alignment 2 was obtained from a multiple sequence alignment with voltage gated K, Ca, and Na channels, using 400 N-terminal amino acids from hSlo (34). The Pileup program of the Genetics Computer Group software and default settings were used. Gray boxes, conserved positively charged residues. Also highlighted are the polar (ellipse), the negative (gray ellipse), and the hydrophobic (black box) residues of hSlo that in Shaker are positively charged. Arrows mark the mutations studied.
Effects of S4 Point Mutations on the Voltage Activation Curves: Shifts along the Voltage Axis-The effect of five point mutations in the S4 segment on hSlo channel activation is depicted in Figs. 2 and 3. In Fig. 2A, the [Ca 2ϩ ] was 5.7 M and the holding voltage was 0 mV. Both R207Q and R207E mutations induce a clear change in activation kinetics of the hSlo channel. In this case currents can be elicited at very negative voltages and all channels are open at Ϫ20 mV, a voltage at which less than 10% of the channels are open in the wild type hSlo channel (Fig. 3A). Fig. 2B shows that at 103 M Ca 2ϩ , glutamine substitution for arginines 210 and 213 (R210Q, R213Q) results in channels that could be opened only by applying very positive potentials. At this [Ca 2ϩ ], mutants R210Q and R213Q have half-activation potentials, V1 ⁄2 , of 96 and 98 mV, respectively, while the V1 ⁄2 for WT channel was -59 mV ( Fig. 3B). It is of interest to compare the effects of mutations R210Q and R210N on the activation curves. For [Ca 2ϩ ] ϭ 1.2 M, V1 ⁄2 of the R210N construct was about the same as the one obtained for the wild type channel (Fig. 3C). However, the position of the P o -V curve for the R210N mutant relative to the WT channel is a function of the [Ca 2ϩ ], and this observation will be discussed in detail below. Summarizing, mutations R207Q and R207E produced leftward shifts along the voltage axis of the P o -V curve compared with the wild type channel; R210Q and R213Q induced rightward shifts of the P o -V curve; and mutations L204R, L204H, Q216R, E219Q, and E219K were of minor consequences (Table I).
Number of Gating Charges in Wild Type and Mutant hSlo Channels-Several lines of experimental evidence support the notion that the S4 transmembrane segment in Na ϩ , Ca 2ϩ and K ϩ voltage-dependent channels is the voltage sensor (18,21,22,(37)(38)(39)(40). A prediction of this model is that a reduction in the number of charges in the S4 segment should lead to a reduction in the number of gating charges (z eq ; Equation 1) that determine the slope of the Boltzmann distribution around the half activation voltage. Mutations R207Q, R207E, R210N, R210Q, and R213Q decrease the slope of the P o -V curves (Fig. 3, A-C) by 49, 50, 65, 65, 51, and 59%, respectively (Table I). These results support the notion that the S4 basic amino acids are part of the voltage sensor in hSlo. By contrast, five mutations of the S4 segment did not significantly alter the slope of the P o -V curve: L204H, L204R, Q216R, E219Q, and E219K (Table  I). Additionally, the neutralization of the arginine 201, which in alignment 2 (34) is located in the S3-S4 linker, did not change the slope of the P o -V curves ( Table I).
The use of a Boltzmann function to fit the P o -V data assumes a simple two-state model: closed-open. There is compelling evidence that the hSlo channel contain several closed and open states (27,28,(41)(42)(43)(44)(45)(46). In this case, a fit to the P o -V data using Equation 1 may result in an underestimation of the number of gating charges per channel. The total number of gating charges  for channels having an arbitrary number of closed states can be obtained through measurements of gating currents (47). However, hSlo gating current measurements require large channel expression and an adequate methodology to measure them, and this has been successfully done only recently (27). Alternatively, a widely used method to obtain the equivalent number of gating charges for a linear model containing any number of closed states is the limiting slope method (LSM) (48). We used LSM to further investigate the mutants for which z eq is lower than that of WT. In order to use the LSM, the P o (V) data were plotted as log(P o ) versus V (Fig. 4), and the P o values in the range between 0.001 to 0.01 were fitted to the equation where z eq lim is the equivalent number of gating charges at very low P o . The LSM was applied to the different S4 mutants used in this study, and the results are shown in Fig. 5. The z eq lim was found to be the same as z eq obtained using Equation 1 for the wild type hSlo channel. Using the LSM, we could only confirm that neutralization of arginines in positions 210 and 213 were able to significantly reduce the number of gating charges directly coupled to the hSlo pore opening (Table I, Fig.  5). The notion that arginine 207 is located outside of the electric field is strengthened by the result obtained with the double mutant R207Q/R213Q. For this mutant, z eq lim was the same as the one found for the single mutant R213Q, suggesting that the neutralization of arginine 207 is unable to reduce the gating charge (Table I).
Quantitative Description of Results-Cox et al. (28) used a voltage-dependent version of the Monod-Wyman-Changeaux model of allosteric proteins considering the MaxiK channel as a homotetramer with a single Ca 2ϩ binding site in each subunit. Each subunit can be in a reluctant (R) or in a permissive (P) state, and only those channels containing the four subunits in the P configuration are conductive. For each subunit, we have the scheme below.
and K 2 (V) are the voltage-dependent equilibrium constants connecting R and P states with and without Ca 2ϩ bound, respectively, and K c and K o are the dissociation constants for Ca 2ϩ from the R and P configurations, respectively.
In the present study, we used the model of Cox et al. (28) in which four Ca 2ϩ ions bind to the channel and the microscopic dissociation constants depend only on whether the subunit is in the R or P configurations. This model is sufficient to give a good fit to the data obtained with the wild type hSlo and to interpret the S4 mutant results mechanistically. The model predicts that the probability of opening is given by the relation (28)  The change in conformational energy difference between open and closed configuration of the channel induced by Ca 2ϩ binding, A, is given by the following equation.
The parameter A can be viewed as a measure of the coupling between Ca 2ϩ binding and pore opening. A is calculated from the ratio K c /K o , which can be obtained fitting the V1 ⁄2 -[Ca 2ϩ ] data using Equation 4. Fig. 6A (Fig. 7A). These observations further support the notion that the hSlo channel is intrinsically voltage-dependent. As stated before, mutations R207Q, R207E, R210Q, and R213Q in the S4 domain of hSlo produced pronounced shifts of the voltage activation curve along the voltage axis ( Fig. 3 and Table I). Specifically, mutations R207Q and R207E shift the P o -V curves by about 150 mV to the left along the voltage axis. In the V1 ⁄2 -[Ca 2ϩ ] plane, the curves for these mutants are displaced downward and are essentially parallel to that obtained for the wild type channel (Fig. 6B). A fit to the data using Equation 4 indicates that mutants R207Q and R207E mainly decrease ⌬G o compared with the WT channel (Table II; Fig.  7A). These mutations also induced a reduction in parameter A (Table II; Fig. 7A). For the R210N mutant, the V1 ⁄2 is 50 mV higher than the WT channel at [Ca 2ϩ ] below 10 Ϫ7 M (Fig. 7B). Due to the smaller z eq of the R210N mutant, a crossover of the mutant and the WT channel V1 ⁄2 - [Ca 2ϩ ] curves occurs at about 1 M Ca 2ϩ (Fig. 6C). The parameter A also decreases in this mutant (Table II; Fig. 7A). Mutations R210Q and R213Q (Fig.  3B, Table I) decrease z eq and shifted the activation curves to very positive potentials at all [Ca 2ϩ ] tested (Fig. 6D)  ]. Energy at the peak is arbitrary. Ba, schematic representing a WT hSlo channel subunit in its four possible configurations. The subunit contains three gating particles that can be in the reluctant (circles) or in the permissive (squares) configurations. Voltage induces the configurational change from reluctant to permissive to in a single step. Calcium binding decreases the energy to perform the change from the reluctant to the permissive configuration (solid symbols) mediated by voltage. Bb, in the R213Q mutant, one of the gating particles was uncoupled to the calcium binding. In this case, if a is the free energy necessary to induce the configurational change of one gating particle, the channel subunit in the absence of Ca 2ϩ behaves very similarly to the WT hSlo subunit. However, it takes a kJ/mol more to change the subunit from the reluctant to the permissive configuration in the presence of Ca 2 . Bc, in the R210N mutant one of the gating particles was uncoupled to the calcium binding but has the configuration of a Ca-bound gate (solid symbols). Therefore, in this case it takes a kJ/mol less energy to displace the equilibrium from the reluctant to the permissive forms in the absence of Ca 2ϩ . The Ca 2ϩ -bound subunit behaves like the WT hSlo subunit. Bd, the R207Q channel subunit has two gates uncoupled to Ca binding as in R210N. In this case, it takes 2a less energy to induce the change from the reluctant to the permissive form in the absence of Ca 2ϩ . The Ca 2ϩ -bound subunit behaves like the WT hSlo subunit.
in V1 ⁄2 at [Ca 2ϩ ] Ͻ 10 Ϫ7 M is fully accounted for by a change in z eq since ⌬G o does not change with respect to the WT hSlo (Table II, Fig. 7A). We expect a crossover of the mutant and wild type V1 ⁄2 -[Ca 2ϩ ] curves if the effect of the mutation is a change in z eq only (cf. Fig. 6C). This crossover does not occur in this mutant due to a decrease in the coupling parameter A (Table II; Fig. 7A). We do not have enough data to interpret the effects of the R210Q mutant in terms of the model. DISCUSSION Voltage-sensing Residues in the S4 Segment of hSlo and Shaker K ϩ Channels-The results shown in this work demonstrate that the S4 segment of the MaxiK hSlo is an important part of the voltage sensor of the channel. In contrast to voltagegated K ϩ channels (K V ) hSlo seems to have only three positive residues in its S4 segment: the arginines in positions 207, 210, and 213 (Fig. 1, alignment 2). The neutralization of these residues decreases the slope of the G-V curves and shifts the curves in the voltage axis, suggesting that these residues are part of the voltage sensor. However, if the number of equivalent gating charges is determined by the limiting slope method, we found that only arginines in positions 210 and 213 contribute significantly to the number of charges per channel. In the voltage-dependent K ϩ channel, Shaker, the equivalent residues in the S4 segment are also arginines: Arg-365, Arg-368, and Arg-371 (Fig. 1). These three basic residues in the S4 segment contribute significantly to the gating charge (24). On the other hand, Aggarwal and MacKinnon (17) have counted wild type and S4 mutant Shaker channels using radiolabeled agitoxin, a high affinity blocker, and measured the total gating charge in each case. They found that only residues Arg-362, Arg-365, Arg-368, and Arg-371 sense the transmembrane voltage. It is surprising, therefore, that although hSlo possesses three equivalent positively charged residues out of the four that are responsible for the voltage dependence in Shaker, the voltage dependence of hSlo is much smaller (see below and Table I). One possible explanation for the smaller number of gating charges in hSlo is that residue Arg-207 in hSlo appears not to contribute significantly to the total charge per channel. The residue equivalent to Arg-207 in Shaker K ϩ channels, Arg-365 (Fig. 1, alignment 2), contributes with five electronic charges. 2 Moreover, in Shaker K ϩ channels, residue Glu-293 in S2 also constitutes a significant part of the voltage sensor (24,35,50). The mutation E293Q in Shaker reduced the charge per channel from 13 e o to 7 e o . According to Wallner et al. (34) alignment, for hSlo and dSlo S2 region, the residue Glu-293 in Shaker is not conserved in these channels, but a tyrosine (Tyr-163) is present. The lack of this important negative charge in MaxiK channels may also contribute to their weaker voltage dependence. It is also important to note here that Papazian et al. (35) proposed, on the basis of rescue experiments with double mutations, that in Shaker K ϩ channels Lys-374 might form a charge network with Glu-293 in S2 and Asp-316 in S3. In Shaker, the mutant channel K374Q is not functional and is rescued by either neutralizing Glu-293 to a Gln (equivalent to Tyr-163 in hSlo) or Asp-316 to Asn (35). In hSlo, in the equiv-alent position to Lys-374, there is a glutamine at 216 (Fig. 1,  alignment 2). It is tempting to suggest following the analogy between these two K ϩ channels, that hSlo, despite the fact it contains a glutamine in position 216, matures properly because there is a neutral residue, a tyrosine, in position 163.
Energy Diagrams for the Wild Type and S4 Mutant Channels: A Pictorial Representation- Fig. 7A shows the free energy diagrams of the channel in the absence of and after Ca 2ϩ binding at zero applied voltage taking the closed state as a reference. In the absence of Ca 2ϩ , we need to supply the hSlo WT type channel with 29 kJ/mol in order to reach a P o ϭ 0.5. This energy is about the same for the R213Q mutant but it is less for the R210N, becoming negligible for the R207Q mutant. In order to have a pictorial view of the changes in channel conformation introduced by the mutations, we fancy each channel subunit as shown in Fig. 7B. In the allosteric model of Cox et al. (28), the voltage sensor from each subunit moves in a highly concerted way such that their movement is represented by a single voltage dependent step. Consider for the sake of illustration a single channel subunit that is in the reluctant state with a voltage sensor consisting of three gating particles represented in Fig. 7B (a) as open circles. The simultaneous displacement of the particles induced by the applied voltage modifies the subunit conformation from the reluctant to the permissive state. For a channel to open, it is necessary that the four subunits reach the permissive state. The voltage-dependent conformational change is represented in Fig. 7B (a) with a change in the gating particle shape from open circles to squares (permissive). The free energy associated with the displacement of one particle, ␦g, is modulated by Ca 2ϩ in such a way that if the subunit binds Ca 2ϩ , ␦g is decreased by an amount a. In other words, binding of Ca 2ϩ decreases the energy difference between the reluctant (closed circles) and the permissive (closed squares) states of the subunit by 3a. In the mutant R213Q, the equilibrium between reluctant and permissive forms of the subunit in the absence of Ca 2ϩ is unchanged with respect to the WT hSlo channel (Fig. 7B, b). However, one of the particles of the voltage sensor of this particular mutant fails to be modulated by Ca 2ϩ (right side of Fig. 7B, b). In this case, 2 Notice, that in the case of neutralization of residue 365 in Shaker leads to an unexpected result, since we expect that R365 should contribute no more than four charges, i.e. if one charge per subunit is eliminated, four charges per channel are eliminated only if they move through the entire electric field. As pointed out by Seoh et al. (24), this result can be explained if the neutralization induces a voltage drop that occurs over a large extent of the protein. This implies that the results of charge neutralizations should be taken cautiously. However, the results of Larsson et al. (18) and Seoh et al. (24) are strong evidence that, upon channel activation, there is a displacement of residues Arg-365, Arg-368, and Arg-371 in Shaker K ϩ channels and that these residues contribute to the total charge per channel.  Fig. 6, using the z eq values from Table I. A was calculated from K c /K o ratio using equations 9 and 10. Since K c and K d are correlated, the error of A was obtained rerunning the curve fitting using ⌬G o , K o , and K c /K o ratio as adjustable parameters. Errors given were read from the diagonal elements of the final error matrix of the least square fitting procedure. Ca 2ϩ binding to the channel subunit is less effective in modulating the configuration of the voltage sensor (i.e. energy decreases by only 2a). On the other hand, in the mutant R210N, one of the gating particles has the Ca 2ϩ -modulated conformation even if no Ca 2ϩ is bound to the subunit (Fig. 7B, c). In this case, the energy supplied by the electric field needed to induce the conformational change that leads to the permissive subunit state is a kJ/mol smaller than in the case of the WT channel. Binding of Ca 2ϩ to the subunit modulates the remaining two gating particles, and the resulting structure behaves similarly to the WT channel. Fig. 7B (d) shows that in the absence of Ca 2ϩ two of the gating particles of the R207Q mutant are in the Ca 2ϩ -modulated conformation. The conformation of the voltage sensor of the R207Q mutant is such that the channel subunit is in the permissive conformation even in the absence of an applied voltage. The R207Q mutant behaves like the wild type subunit when it binds Ca 2ϩ (right side of Fig. 7B, d).
Model Adequacy-As discussed recently by Rothberg and Magleby (44), the MWC model we have used in the present study could approximate the basic features of the Ca 2ϩ dependence of the kinetic structure. Cox et al. (28) found that the MWC model gives an adequate description for the effects of voltage and Ca 2ϩ on P o for mSlo channels. We think this model is appropriate to account for the data obtained with the WT and the S4 mutants described in the present study. Furthermore, it provided a heuristic device to interpret the data in terms of thermodynamic parameters. However, it is clear that more complex models need to be used when trying to give a detailed description of the channel voltage-dependent gating kinetics (27,44,46). In this regard, the MWC model fails to predict the observed Cole-Moore shift, the charge movement between open states (27), and to account for the brief closed intervals (44).
Our results suggest that the S4 segment in the hSlo channel plays the role of the voltage sensor and that Ca 2ϩ acts by decreasing the intrinsic configurational energy difference that separates the closed from the open state at zero applied voltage. This conclusion is further supported by the observation that for the wild type hSlo channel the same amount of gating charge is moved in the Ca 2ϩ -dependent and in the Ca 2ϩ -independent activation regions (27).