INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Voltage-gated ion channels are key elements in the generation of the nervous impulse. They consist of an ion-selective pore, several activation and inactivation gates, and a voltage-sensing machinery (Hille, 1992). The recently solved structure of the bacterial KcsA channel most likely provides a good model of the central, pore-containing part of voltage-gated potassium channels, including the gates (Doyle et al., 1998; Yellen, 1998). The positively-charged fourth transmembrane segment (S4) has been identified as voltage sensor (Mannuzzu et al., 1996; Larsson et al., 1996; Seoh et al., 1996; Aggarwal and MacKinnon, 1996; Yusaf et al., 1996), but its location in relation to the pore is not known. The purpose of the present study is to obtain information about this location. Our strategy is to use the structure of the KcsA channel as a model for the pore region of the Shaker potassium channel and to measure the effects of introduced charges in different positions in this region of the Shaker potassium channel.

Altering the charge of a surface amino acid will alter the electric field sensed by the voltage sensor and, hence, shift the voltage-dependent parameters of the ion channel along the voltage axis (McLaughlin, 1989; Hille, 1992). These shifts should be larger the closer the introduced charge is to the moving, charged parts of the protein. In the latter part of the paper, we will interpret these shifts as caused by interactions with the extracellular end of the voltage sensor S4 and translate these shifts into geometrical distances to this part of S4. We introduce a cysteine at different positions in a non-inactivating N-terminal deleted Shaker potassium channel (ShH4IR) and express these channels in Xenopus oocytes. This allows us to add a negative or a positive charge to the specific positions in situ by modifying the introduced cysteines with differently-charged reagents (MTSES or MTSET⁺, see Materials and Methods). We selected four extracellular residues that cover most of the circumference of the pore-forming part of the channel (Fig. 1): one at the extracellular end of S5 (A419) which has been suggested to be close to S4 (Elinder and Århem, 1999), two at the extracellular end of S6 (V451 and G452) which are involved in slow inactivation (Larsson and Elinder, 2000), and one at the extracellular end of the pore helix (P430). The results suggest that the extracellular end of S4 is located close to the extracellular end of S5.

View larger version (68K): [in this window] [in a new window]   FIGURE 1   Positions of the modified residues. Top view (upper part) and side view (lower part) of the bacterial KcsA channel (only backbone is shown). Mutations used in the present investigation (space filled residues; gray = C, red = O; blue = N) are labeled. A419 (at the extracellular end of S5), P430 (at the extracellular end of the pore helix), and, V451 and G452 (at the extracellular end of S6) in Shaker corresponds to R52, P63, V84, and T85 in KscA. In the center of the channel, three potassium ions (yellow) and one water molecule (red) are shown in the selectivity filter.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

Molecular biology

The experiments were performed on Shaker H4 channels (Kamb et al., 1987), made incapable of fast-inactivation by the 6-46 deletion (Hoshi et al., 1990). Cysteines were substituted using the QuikChange Kit (Stratagene, La Jolla, CA). cRNA was transcribed using the T7 mMessage mMachine kit (Ambion Inc., Austin, TX) and injected in Xenopus laevis oocytes (20-5000 pg/cell) using a Nanoject injector (Drummond Scientific Co., Broomall, PA). The oocytes were maintained at 12°C in a modified Barth's solution (MBS, in mM: 88 NaCl, 1 KCl, 2.4 NaHCO₃, 15 HEPES, 0.33 Ca(NO₃)₂, 0.41 CaCl₂, and 0.82 MgSO₄) adjusted to pH 7.5 by NaOH, and supplemented with penicillin (10 µg/ml) and streptomycin (10 µg/ml). The electrophysiological experiments were made 2-20 days after injection of mRNA.

Electrophysiology, solutions, and reagents

The currents were measured with the two-electrode voltage-clamp technique (CA-1 amplifier, Dagan Corporation, Minneapolis, MN). Microelectrodes were made from borosilicate glass and filled with a 3M KCl solution. The resulting resistance varied between 0.5 and 2.0 M. The amplifier's capacitance and leak compensation were used, and the currents were low-pass filtered at 1 kHz. All experiments were carried out at room temperature (20-23°C). For the electrophysiological experiments, we used the MBS solution described above. To screen surface charges, we added 20 mM MgCl₂ to the MBS solution. Mg²⁺ ions have been shown to shift voltage-dependent parameters equally and, therefore, have been suggested to exert pure screening of surface charges without directly binding to potassium channels (Elinder and Århem, 1998; Elinder et al., 1998). To alter the charge at the substituted cysteines, the membrane-impermeant thiol reagents, positively-charged MTSET ([2-(trimethylammonium)ethyl]methanethiosulfonate, bromide) and negatively-charged MTSES (sodium (2-sulfonatoethyl)methanethiosulfonate) (Toronto Research Chemicals Inc., North York, Ontario, Canada), were applied continuously in the bath solution by a gravity-driven perfusion system, and the modification was assayed functionally in two-electrode voltage-clamped oocytes, as described earlier (Larsson et al., 1996; Baker et al., 1998). The cysteine reagents were applied to saturation (typically 100 µM MTSET or 1 mM MTSES for 200 s). The molecular structures of the attached reagents are shown in Fig. 2. The sizes are close to that of the positively charged amino acid residue arginine.

View larger version (14K): [in this window] [in a new window]   FIGURE 2   Schematic structures of cysteine residues without and with attached MTS reagents. For comparison, a positively charged arginine residue is shown. The residue in KcsA corresponding to 419A in Shaker is an arginine (see Fig. 1).

Shift measurements

The steady-state potassium conductance G_K(V) was calculated as

(1)

where I_K(V) is the steady-state current, V is the membrane potential measured in the bulk solutions, and E_K is the equilibrium potential (80 mV). For 419C, 451C, and 452C, the shifts of the G(V) curves caused by either MTS reagents or Mg²⁺ were measured by sliding the control G(V) to overlap (most importantly the lower part of the curves, because the curves were not normalized) the Mg²⁺ or MTS G(V). For 430C, we had to normalize the maximum conductance (at ~+60 mV) before sliding the G(V) curves.

Calculation of the surface potential

The change in the surface potential caused by the cysteine reagents was estimated directly from the voltage shift of the G(V) curve (V_MTS). To minimize the errors in our calculations caused by possible nonelectrostatic effects of the MTS reagents, we calculated the change in surface potential, , as (V_MTSET  V_MTSES)/2. To confirm that this shift was caused by a change in the surface charge density and consequently in the surface potential, we measured the voltage shift of the G(V) curve induced by an extracellular application of 20 mM Mg²⁺ (V_Mg) and used the Grahame equation (Grahame, 1947) for a quantitative evaluation:

(2)

where  is the surface charge density, _r is the dielectric constant of the medium (80 in water), ₀ is the permittivity of free space (8.85 · 10¹² F·m¹), c_i is the bulk concentration, and z_i is the valence of the ith ionic species in the extracellular solution.  is the surface potential. R, T, and F have their usual thermodynamic significance. To relate the Mg-induced shift, V_Mg, to the surface potential, , Eq. 2 has to be solved numerically. In brief, one has to find the (constant)  value that, for the two different Mg²⁺ concentrations (c_Mg2+), will give two values of  that differ by V_Mg (see pp 457-470 in Hille (1992) for more detailed explanations of surface charge calculations). The final concentrations of the different valencies that were used in the calculations were (in mM, note that approximately 10 mM NaOH was added for the pH adjustment) 100 cation⁺, 1.6 cation²⁺, 93 anion, and 0.8 anion². Although Eq. 2 is strictly valid only for a uniformly smeared charge, it has been shown that the equation can be used as an approximation for charge densities more negative than 0.16 elementary charges per nm² (Peitzsch et al., 1995). The surface charge density of the Shaker potassium channel has previously been estimated to be 0.27e nm² (Elinder et al., 1998). Using Eq. 2, we calculated the predictions of the changes in the Mg²⁺-induced shifts (V_Mg in Table 1) from the measured shifts induced by the MTS-reagents ( in Table 1) and the Mg²⁺-induced shifts of the unmodified channels (V_Mg in Table 1).

Estimation of the S4 location

The potential _r, at a distance r, from an elementary charge e located at the border between a low-dielectric (membrane) and a high-dielectric (water) medium is (McLaughlin, 1989; Elinder and Århem, 1999)

(3)

where  is the inverse of the Debye length in the aqueous phase (9.6 Å in the MBS solution, see Elinder and Århem (1999) for calculation). Because Shaker is a homotetramer and we changed the charge of a residue in all four subunits, Eq. 3 is modified to calculate the predicted change in potential at a specific position (x, y) in the surface plane of the channel

(4)

where (x_i, y_i) is the position of a surface charge residue in the ith subunit. Assuming a helical screw motion of S4, we can estimate the location of the extracellular end of S4 with no deeper knowledge of the interior of the protein (see the Results section). We assume that the four S4s are symmetrically located in the channel. To determine the position of the extracellular end of S4, we compared the experimentally obtained change in surface potential () with the predicted value from Eq. 4 by

(5)

where n indicates the identity of the mutation (a total of four in the present study). The r.m.s. values were determined for an area of 100 × 100 Å² at every 2 × 2 Å² (see Fig. 5). The extracellular end of S4 is most likely to be found at the position where the r.m.s. has its smallest value.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

MTSET⁺ and MTSES shift the G(V) of 419C in opposite directions

Figure 3 A shows the effect of differently-charged cysteine reagents on 419C channels. MTSES shifted the conductance versus voltage curve G(V) in the negative direction along the voltage axis on average 10.2 mV (Table 1). MTSET⁺ shifted the G(V) in the positive direction on average +6.1 mV (Table 1). The amplitude of the maximum conductance was only marginally affected (<5%). The shifts caused by MTSES and MTSET⁺ are in opposite directions but not perfectly symmetrical, suggesting that the structure of the MTS-reagent itself slightly affects the stability of the channel in some of its conformational states. To separate the G(V) shifts caused by electrostatic effects () from nonelectrostatic effects of the MTS reagents, we assumed that the nonelectrostatic effects were the same for the two MTS reagents and that the difference between the MTSET⁺-induced and the MTSES-induced shifts is equal to 2. This gave  = (+6.1  (10.2))/2 = 8.2 mV. To verify that this value was due to electrostatic interactions and reflected a change in surface potential, we analyzed the effects of Mg²⁺ on the G(V) curves.

View larger version (30K): [in this window] [in a new window]   FIGURE 3   Effects of charged cysteine reagents and Mg2+ on the G(V) curve of A419C. (A) G(V) curves of 419C channels before ( and ) and after MTSET+ () or MTSES () treatment. The negatively-charged MTSES shifts the G(V) in the negative direction, and the positively-charged MTSET+ shifts the G(V) in the positive direction. The dotted lines are the control data shifted as indicated. Note that the data are from two different cells. (B) G(V) curves of 419C channels () and MTSES-modified 419C channels () in MBS solution (open symbols) and in MBS + 20 mM Mg2+ (filled symbols). 20 mM Mg2+ shifts the control G(V) less than the G(V) of the MTSES+-treated channels. (C) G(V) curves of 419C channels () and MTSET+-modified 419C channels () in MBS solution (open symbols) and in MBS + 20 mM Mg2+ (filled symbols). 20 mM Mg2+ shifts the control G(V) more than the G(V) of the MTSET+-treated channels. (D) Schematic figure showing the effects of a negative surface potential and changes in the surface potential on the electric field. The extracellular (e.c.) potential far from the membrane is 0 mV and the intracellular (i.c.) potential is set to 70 mV. The potential approaches the negative surface potential () as one is approaching the channel/membrane that has a net negative surface charge. Modification of an extracellularly-exposed cysteine with MTSES makes the surface charge density more negative, and, consequently, the surface potential becomes more negative (by ). 20 mM Mg2+ (incompletely) screens the surface charges and changes the surface potential by VMg (control channels) or V*Mg (MTSES-modified channels). Note that V*Mg > VMg.

The component of the G(V) shifts are due to electrostatic interactions

According to the surface charge theory, raising the external concentration of divalent cations will screen the exposed surface charges and reduce the absolute value of the (negative) surface potential (McLaughlin, 1989; Hille, 1992) (see Fig. 3 D). This reduction in surface potential will lead to a shift of the voltage-dependent parameters along the voltage axis, which is roughly proportional to the absolute value of the surface potential. Therefore, to test if we have altered the surface charge density and consequently the surface potential by the introduced charged MTS reagents, we perfused the channels with an extracellular solution containing an increased concentration of MgCl₂ (the Cl ions having negligible effects according to Eq. 2).

The application of 20 mM Mg²⁺ shifted the G(V) curve (V_Mg) for 419C channels in the positive direction along the voltage axis (Fig. 3, B and C) on average +13.8 mV (Table 1). V_Mg increased to 16.1 mV (Table 1) when 419C channels were modified with MTSES (Fig. 3 B) and decreased to 11.7 mV (Table 1) when 419C channels were modified with MTSET⁺ (Fig. 3 C). These results are qualitatively what is expected from a change in surface potential by the MTS-modifications of 419C. To make a quantitative evaluation, we used Eq. 2 to calculate the predicted V_Mg for the MTS-modified channels based on the V_Mg for the unmodified channels (+13.8 mV) and on the of the MTS-modified channels (8.2 mV). The predicted shifts were +16.1 mV and +11.5 mV, for MTSES- and MTSET⁺-modified channels respectively, which are in agreement with the experimentally found shifts showing that is of electrostatic origin.

Electrostatic effects of other residues

We have shown that a charge at residue 419 in a Shaker potassium channel affects the voltage sensing of the channel with ±8.2 mV, the sign depending on the sign of the charge. With the same method we could also show that a charge at residue 451 affects the voltage sensing of the channel with ±3.6 mV and a charge at residue 452 affects the voltage sensing of the channel with ±5.6 mV (Table 1). The effect of the MTS-modification of residue 430C deviated from those of the other residues. First, the MTS reagents reduced the maximum conductance considerably: MTSES by 67 ± 11% (mean ± SD, n = 7), and MTSET⁺ by 81 ± 12% (mean ± SD, n = 5). Second, the MTS-induced shifts were both in the positive direction: +19.1 mV for MTSES and +26.7 mV for MTSET⁺ (Table 1). This suggests that the MTS molecules have a major nonelectrostatic effect on the G(V) curve in addition to a minor electrostatic effect: a +22.9 mV shift caused by nonelectrostatic effects and a ±3.8 mV shift by electrostatic effects. The V_Mg of 430C was not significantly altered by MTSES (Table 1). This suggests that residue 430 contributes very little, or not at all, to the surface potential felt by the voltage sensor. In the following calculation, we used the values as the estimate of the electrostatic effect of the introduced charges (for position 430, see also Appendix).

The location of S4

One simple interpretation of our results is that, of the residues studied, residue 419 should be located closest to the voltage sensor S4, whereas residue 452 is located further away from S4, and residues 430 and 451 are located even further from S4. This suggests that S4 is located close to S5 (see Fig. 1). The exact effect of the introduced surface charges on the voltage dependence of the channel depends on the exact movement of all the charged residues of the channel in relation to the introduced charge. In the following quantitative analysis, we will assume that it is mainly the charges in S4 that are moving and that all other charges are relatively immobile. In addition, to make a quantitative estimation of the position of S4, we have to make some assumptions about the structure and the movement of S4.

The S4 segment is most likely an -helix (Yang et al., 1997; Cha et al., 1999; Glauner et al., 1999; Li-Smerin et al., 2000a), which would make the charges on S4 wrap around this helix like a helical screw (Catterall, 1986). The movement of S4 during the opening of the channel exposes three charged residues into the extracellular solution (Baker et al., 1998; Tiwari-Woodruff et al., 2000), probably reflecting three consecutive activation steps (Keynes and Elinder, 1999). If we assume that the S4 motion is that of a helical screw moving in three steps (Catterall, 1986; Keynes and Elinder, 1999), then this motion is electrostatically equivalent to moving the three bottom charges from the cytosol to the extracellular solution while keeping the charges inside the protein fixed (Fig. 4). Thus, even though the electrical potential caused by an extracellular surface charge is not negligible in the membrane, it does not affect the gating because the intramembranous charge pattern is unaltered during gating in this model. Furthermore, the electrical potential caused by an extracellular surface charge is negligible at the intracellular side of the membrane (see Fig. 5 in Mathias et al., 1992). For the following quantitative analysis, we used this helical-screw model of S4 motion. In this model, it is only the electrical potential at the extracellular surface where the three top S4 charges (R362, R365, and R368) emerge, that determines the size of the G(V) shift caused by changes of the extracellular surface charges. The three top charges on S4 would all lie on the same side of an -helix (Fig. 4) and we have, in the following approximate calculation, treated them as a point charge (for quantitative justifications see Appendix).

View larger version (31K): [in this window] [in a new window]   FIGURE 4   Schematic model of S4 movement. Positions of S4 in resting and activated states. Dotted side chains indicate that the -carbons of the positively charged residues on S4 are located on the back of the helical core. Note that the charge pattern in the membrane does not change upon activation. The three charged extracellular residues in the activated position are located on one side of the helical core. The labels indicate the residue numbers of the S4 charges in Shaker.

We used Eq. 4 to predict the surface potential at different positions _pred(x, y) caused by the introduction of a charged residue in all four subunits. We calculated the differences between our experimental data, , and the predicted values _pred(x, y) for all possible positions (x, y) of the top charges of S4 on the channel surface (see Materials and Methods). The smaller the difference is for a certain position of S4, the higher the probability that S4 is located at that position. Fig. 5 A shows the r.m.s. values of these differences calculated for the entire channel surface. The resulting landscape shows four distinct valleys (red) at a distance of ~11 Å outside the center of the extracellular end of S5. These valleys suggest, under the assumption of the helical-screw model above, the location of the top charges of S4. The low error value at the minimum (0.77 mV) indicates that it is a good fit to our data.

View larger version (57K): [in this window] [in a new window]   FIGURE 5   Structural models of a potassium channel. (A) Estimated location of the extracellular end of S4. Top view of the bacterial KcsA channel with r.m.s deviations between the experimental and theoretical surface potential values for different positions of S4 calculated with Eq. 5. The total area is 100 × 100 Å2. The symbols ( for 419,  for 451,  for 452, and  for 430) mark the assumed positions of the charges (C or the tip of the residue) used in the electrostatic calculations. The yellow circle in the center is a potassium ion in the selectivity filter. The color bar covers r.m.s. values from 0.77 to 5.60 mV. The black area in the center of the figure has an r.m.s. deviation 5.6 mV. The extracellular ends of the four S4 segments are most likely located in the dark red areas. In the upper left subunit, we have suggested positions of S1-S4 that are consistent with ours and other groups' experimental results (Tiwari-Woodruff et al., 1997; Baker et al., 1998; Li-Smerin et al., 2000a,b; Hong and Miller, 2000). The large + indicate the extracellular positions of the three top charges in the activated state. The small + indicate the positions of the intramembraneous charges. (B) Side view of the model in A. S5 and S6 from different subunits are shown in different colors (red, yellow, blue, and green). The proposed localization of S1-S4 in the green subunit is indicated. S4 is shown in both the resting (dashed) and the activated position.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION APPENDIX REFERENCES

We have tested the effect of an introduced charge at different positions in the extracellular end of the pore domain of Shaker potassium channels. The key finding was that a charge in S5 (residue 419) electrostatically affects the voltage sensor more than a charge at other positions. Assuming that the electrostatic effects are with the charges in S4, this suggests that S4 is located close to S5. We also made a quantitative estimation of the location of S4, under the assumption of helical-screw motion, that indicates that the top S4 charges in the activated position are located ~11 Å from the center of the extracellular end of S5 (Fig. 5 A). It should be pointed out that, in our study, we do not obtain any information about the location of the intracellular end of S4. S4 could very well be tilted, and we have in Fig. 5 B suggested a possible three-dimensional location of S4.

The presented location of S4 deviates from recent suggestions. Studies based on tryptophan or alanine scanning, or energy-minimizing computer calculations tentatively locate S4 close to the interface between the subunits in voltage-gated potassium channels (Durell et al., 1998; Li-Smerin et al., 2000a,b; Hong and Miller, 2000). However, these investigations were not directly aimed at localizing S4 in relation to the pore.

Our calculation above was done with some simplistic assumptions: the exact positions of the side chains containing the charged groups, the local dielectric environment around the surface charges, and the position of S4 in the z-direction. However, calculations with reasonable variations of these assumptions show that the top charges of S4 are always located closer to the extracellular end of S5 than to S6 or the P helix (see Appendix). The calculations above were also done under the assumption of a helical-screw motion of S4 (Catterall, 1986; Baker et al., 1998; Keynes and Elinder 1999). Other models of the S4 motion have also been suggested (Aggarwal and MacKinnon, 1996; Papazian and Bezanilla, 1997; but see Horn, 2000). However, the fact that charges introduced in S5 have larger electrostatic effects than charges in other parts of the pore region suggests, for most models of S4 motion, that S4 is closest to S5.

In Fig. 5 we have, in addition to the position of S4, indicated tentative positions of the S1-S3 segments, based on experimental results from other studies (Tiwari-Woodruff et al., 1997; Baker et al., 1998; Li-Smerin et al., 2000a,b; Hong and Miller, 2000). The top S4 charges (R362, R365, and R368) are located in the extracellular solution (Baker et al., 1998), close to the red spot, while the S4 charges in the membrane (R371, K374, and R377) can make the interactions, with negative charges in S2 and S3, that were identified in intragenic suppression investigations (Tiwari-Woodruff et al., 1997). Large parts of S1 and S2 are facing the lipids, whereas S3 (placed in the groove between subunits) is mainly facing protein surfaces (Li-Smerin et al., 2000a; Hong and Miller, 2000). Finally, a portion of S5 is facing the lipids in contrast to the rest of the S5-S6 pore domain (Li-Smerin et al., 2000b). The suggested location of S4 would allow for large scale movements of S4 (e.g., helical screw motion) because one side (the hydrophobic side) of the transmembrane portion of the S4 helix would be facing the fluid lipid bilayer (Fig. 5). The main difference with the earlier work (e.g., Li-Smerin et al., 2000a,b) is that we place S3 at the center of the subunit instead of S4. Li-Smerin and coworkers mapped the residues of the pore region that is interacting with the voltage sensor module (i.e., S1-S4). In our model, this would suggest that it was S3's interaction surface with the pore region that was mapped in their work.

The present estimation of the location of S4 is important for future studies in determining the molecular coupling between the movement of the voltage sensor and the opening of the pore. It may also serve as guideline for future mapping of the outer segments S1, S2, and S3, as well as auxiliary -subunits (Gulbis et al., 1999) in relation to the pore. Furthermore, the present method to determine structural features of the Shaker potassium channel can be used for other proteins where electrostatic effects are important.