INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion Conclusion References

In 1955 Hodgkin and Keynes proposed that K⁺ channels have long pores that can simultaneously hold multiple permeant ions as they flux in single file. This model has been expanded to the point where it is now believed that before entering the pore, each K⁺ ion in solution must first shed its waters of hydration to pass through the narrowest region. As waters of hydration are shed, K⁺ ions make intimate interactions with pore-lining residues at what are thought to be discrete ion binding sites (Hille, 1992). Attempts to locate the structural determinants of the K⁺ channel pore have provided evidence that the P-region (original mutagenesis: MacKinnon and Yellen, 1990; Hartmann et al., 1991; Yellen et al., 1991; Yool and Schwarz, 1991; Heginbotham et al., 1992; Kirsch et al., 1992; toxin binding: MacKinnon and Miller, 1989; MacKinnon et al., 1990; Goldstein et al., 1994; Aiyar et al., 1995; Hidalgo and MacKinnon, 1995; Naranjo and Miller, 1996; Ranganathan et al., 1996), the internal S4-S5 loop (Isacoff, 1991; Slesinger, 1993), and S6 (Choi et al., 1993; Isacoff et al., 1993; Aiyar et al., 1994; Lopez et al., 1994; Taglialatela et al., 1994) form the pore. A flux ratio experiment indicates that at least four K⁺ ion binding sites are located within these regions of the Shaker potassium channel (Stampe and Begenisich, 1996). The narrowest region, which determines ion selectivity, is localized to the highly conserved "signature sequence" within the P-region (Heginbotham et al., 1992, 1994; cysteine probing: Kurz et al., 1995; Lu and Miller, 1995; Pascual et al., 1995). Residues in and near this "signature sequence" have been identified that may form ion binding sites (Yellen et al., 1991; Heginbotham and MacKinnon, 1992; Kirsch et al., 1992; Choi et al., 1993; Ranganathan et al., 1996) or act as barriers to ion movement (Harris and Isacoff, 1996; Hurst et al., 1996).

The extremely fast (10⁷-10⁸/s) flux rate of K⁺ through the pore has made it difficult to study K⁺ ion binding at these sites. This problem was mainly overcome in a study of the large-conductance calcium-activated (BK) K⁺ channel by Neyton and Miller (1988a, b) through the use of Ba²⁺. Ba²⁺, a permeant ion with the same crystal diameter as K⁺, binds in the channel with a dwell time 10⁶ longer than that of K⁺, which is long enough for its occupancy to be measured as a block of K⁺ flux. By examining the influence of K⁺ ions on Ba²⁺ dissociation, Neyton and Miller defined both the electrical location and affinity of three K⁺ ion binding sites in the BK pore.

In this study, we have applied the method developed by Neyton and Miller to characterize three permeant ion binding sites in the pore of the voltage-gated Shaker K⁺ channel, which we find to closely resemble those of the BK channel. We identified pore-lining residues that appear to contribute to the formation of the two deeper sites, and ask where the sites are located with respect to the channel gates. Interestingly, we found that the deepest of the three sites binds ions most tightly, is located between the activation gate and the C-type inactivation gate, and can be occupied when either of these gates is closed. In this respect, the C-type inactivation gate is similar to the activation gate studied by Armstrong (Armstrong and Hille, 1972), in that it can close on a pore blocker that is bound in the pore. These results are consistent with a mechanism of gating that operates by pinching off access of the deep pore to the internal or external solution.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion Conclusion References

Molecular biology

All experiments were performed on the N-terminal deleted (6-46) ShB (Hoshi et al., 1990) channel and mutants that were made in this background. The gene had been cloned into pBluescript (Stratagene) + vector. Mutant plasmids were generated by the dut-ung-procedure as described by Sambrook et al. (1989). Plasmids were transformed into DH5 cells, and dsDNA was isolated by an alkaline lysis procedure. Mutant plasmids were sequenced by the Sanger dideoxy method. Dimethyl capped cRNA was prepared from mutant and wild-type plasmids by in vitro runoff transcription reactions, using either Ambion (Megascript) or Stratagene transcription kits with T7 RNA polymerase after linearization with HindIII. RNA pellets were resuspended in ultrapure water (Specialty Media, Laballette, NJ), and yields were determined by formaldehyde-agarose gel electrophoresis. Aliquots were stored at 80°C until needed for oocyte injection.

Electrophysiology

Stage V and VI Xenopus laevis oocytes were isolated, collagenased (Worthington Biochemical Corp.), and stored at 18°C in MBSH solution. MBSH contained (in mM) 88 NaCl, 1.0 KCl, 2.4 NaHCO₃, 10 HEPES (pH 7.5 with NaOH), 0.82 MgSO₄ (7 H₂O), 0.33 Ca (NO₃)₂ (4 H₂O), 0.41 CaCl₂ (2 H₂O). Pyruvate (2.5 mM final concentration), penicillin-streptomycin, gentamicin, 0.5 g bovine serum albumin (per liter), and 0.5 g Ficoll (per liter) were also added. After isolation, 50-100 nl of cRNA was injected into each oocyte, and recordings were made 2-6 days after injection.

Electrodes (0.2-1 m) were pulled on a model P-87 Flaming/Brown micropipette puller (Sutter Instrument Company) and filled with 3 M KCl. After recording, any drift of the electrode voltage offset relative to the bath solution was noted upon removal of the voltage electrode from the oocyte. Recordings with over 5-mV drift were excluded. All external solutions contained 10 mM HEPES (pH to 7.0 with either KOH or NaOH, depending on which was the major cation in solution) and 0.41 mM CaCl₂. The 2 mM KCl solution contained (in mM) 112 NaCl and 2 KCl. The 112 mM KCl solution contained 112 mM KCl. The 10 mM Ba²⁺ "low K⁺" solution contained (in mM) 10 BaCl₂, 2 KCl, and 92 NaCl. The 10 mM Ba²⁺ "high K⁺" solution contained 10 mM BaCl₂ and 92 mM KCl. The 0 mM K⁺ solution contained 112 mM NaCl. All recordings were made at room temperature (21-23°C).

Oocytes were typically held at 80 mV, except where stated otherwise in the text, and stepped to test potentials. The duration of test voltages was 800 ms for the experiments measuring open channel Ba²⁺ dissociation and 30 ms for experiments containing both closed and open channels (closed/open). Leak currents were subtracted on line with a P/4 subtraction protocol for the closed/open channel experiments. No leak subtraction was done for the open channel Ba²⁺ dissociation experiments. Data were acquired with an Axoclamp-2A amplifier (Axon Instruments, Foster City, CA), a TL-1 AD/DA computer interface (Axon Instruments, Foster City, CA), and an IBM 486 AST PC. Data were sampled at 1 or 5 kHz and filtered first through a 1-pole low-pass filter (Axoclamp-2A) at 30 kHz, and then filtered again at 200 Hz or 1 kHz through an 8-pole Bessel filter (Frequency Devices). Data were analyzed with pClamp software (Axon Instruments). To determine the time course of Ba²⁺ wash-out from open channels in the presence of significant C-type inactivation (such as with 0 K_out⁺), a control trace acquired in the absence of Ba²⁺ was subtracted from the experimental trace exhibiting Ba²⁺ unbinding.

Model of Ba²⁺ open channel dissociation

The voltage dependence of open-channel Ba²⁺ dissociation was modeled with a single-file three permeant ion binding site model (Fig. 3 C). Site 1 is the "lock-in" site and the shallow fast equilibrating Ba²⁺ block site. Site 2 is the low-affinity "enhancement" site. Site 3 is the deep Ba²⁺ binding site. The total off rate for Ba²⁺ from site 3 of open channels is equal to the off rate in the outward direction plus the off rate in the inward direction:

(1)

where _v,c is the off rate for Ba²⁺ in the outward direction, and _v,c is the off rate for Ba²⁺ in the inward direction. Both of these rate constants are dependent on membrane voltage (v) and external K⁺ concentration (c).

The Ba²⁺ off rate in the outward direction can be expressed as

(2)

where f₁(v, c) is the fraction of channels with site 1 occupied by an external K⁺ ion, f₂(v, c) is the fraction of channels with only site 2 occupied by a K⁺ ion, and f₁₊₂(v, c) is the fraction of channels with both sites 1 and 2 occupied by K⁺ ions. ₀ is the outward off rate at 0 mV, and _off is the electrical distance from the Ba²⁺ binding site to the rate-limiting barrier to Ba²⁺ exit in the outward direction. F, V, R, and T have their usual meanings. The same equation written with the fractional states occupancies, f_x, expressed explicitly in terms of binding constants, becomes

(3)

where and where k_{1(0 mV)} is the dissociation constant at 0 mV membrane potential for K⁺ binding to site 1 (when Ba²⁺ occupies site 3), ₁ is the electrical distance for K⁺ to bind to site 1 from the external solution (fixed at 0.18 based upon the  of fast Ba²⁺ block; see below), k_{2(0 mV)} is ratio of the number of channels with site 1 occupied by K⁺ to the number of channels with K⁺ occupying site 2 at 0 mV, ₂ is the electrical distance from site 1 to site 2, k_{3(0 mV)} is the dissociation constant at 0 mV membrane potential for K⁺ to bind at site 1 when there is a K⁺ ion at site 2, and ₃ is the electrical distance for K⁺ to bind at site 1 when site 2 is occupied by a K⁺ ion. For simplicity, ₁ = ₃.

The rate of Ba²⁺ dissociation in the inward direction can be expressed as

(4)

where ₀ is the dissociation rate at 0 mV in the inward direction for Ba²⁺ when site 1 or site 2 is occupied by K⁺ or when both sites are unoccupied by K⁺. We set this value to zero because there is little inward dissociation with 0 mM K_out⁺ and 2 mM K_out⁺ (Fig. 2 C). '₀ is the Ba²⁺ off rate in the inward direction at 0 mV when sites 1 and 2 are occupied by K⁺. _'off is the electrical distance from the Ba²⁺ ion binding site to the rate-limiting barrier to its exit in the inward direction. The same equation written with the fractional states occupancies, f_x, expressed explicitly in terms of binding constants, becomes

(5)

Therefore the total Ba²⁺ off rate (Eq. 1) expressed in terms of external K⁺ concentration and voltage is simply the sum of Eqs. 3 and 5. The Ba²⁺ ion dissociation rates for 0 mM K⁺, 2 mM K⁺, and 112 mM K⁺ were fit simultaneously (Madonna 5.1.2) with the sum of Eqs. 3 and 5, yielding the solid line fits in Fig. 3 B. The four fixed parameters were ₀ (6.77 s¹), _off (0.355), ₁ (0.18), and ₃ (0.18). ₀ and _off were determined from experiments, respectively, as the 0 mV off rate and _off for Ba²⁺ with 0 mM K_out⁺. ₁ was assumed to be 0.18, the voltage dependence of external Ba²⁺ binding to its fast site (see below), which is site 1 in the model. For simplicity, ₁ = ₃. The values for the free parameters derived from a fit of the model to the data (Fig. 3 B) were

Model for Ba²⁺ block

The calculation of the voltage dependence of Ba²⁺ unblock in Fig. 5 C required a model for Ba²⁺ blockade. In our model channels are in one of three possible states: where O is the unblocked conducting state of the channel, B₁ is Ba²⁺ bound to the fast site (site 1), and B₃ is Ba²⁺ bound to the slow site (site 3). (Site 2 is the low-affinity site, which we assume is occupied minimally under 10 mM Ba²⁺.) , , , and  are rate constants. Differential equations for the disappearance of these states can be written and solved for at equilibrium. This yields the following equations:

(6)

(7)

where _fast is the electrical distance to site 1 from the external solution, ₁₃ is the electrical distance between site 1 and site 3, K_dfast is the K_{d(0 mV)} for Ba²⁺ at site 1, and K is the 0 mV equilibrium constant for Ba²⁺ movement between sites 1 and 3. Using Eq. 6, one can calculate the fraction unblocked for site 1: A fit of the voltage dependence of the fraction unblocked for site 1 to the observed fraction unblocked of the fast component (Fig. 5, A and C) gives the values of K_dfast (10.0 mM) and _fast (0.18). The fast component of block was measured as in Fig. 5 A (fraction unbound of the fast component = 1  fast). Obtaining ₁₃ and the K_{d(0 mV)} for site 3 is more complex and requires calculation of the total fraction of unblocked channels. Using Eqs. 6 and 7, the total fraction of unblocked channels is

(8)

where K_dslow is KK_dfast, which is the K_{d(0 mV)} for the slow site (site 3), and _slow is _fast + ₁₃, which is the electrical distance between the external solution and site 3. The total fraction of unblocked channels was measured as in Fig. 5 A (total unblocked = 1  fast  slow). This data were plotted against step potential and fit with Eq. 9 (Fig. 5 C; total unblocked). The fit gives the values for K_dslow (1.11 mM) and _slow (0.38), using the values of K_dfast and _fast obtained above from the fit of Eq. 8 to the observed fraction unblocked of the fast component. For simplicity, we have ignored K⁺ binding to sites 1 and 3. This experiment could not be carried out in K⁺-free solutions because we find that Ba²⁺ increases current in K⁺-free solutions before blocking conductance (data not shown).

To compare the effect of the mutations on the fast and slow components of Ba²⁺ block (Fig. 7), the fractions unblocked for the fast and slow components were calculated for both wild-type and mutant channels as follows. The fraction unblocked for the fast component was calculated as above (fraction unblocked fast = 1  fast; Fig. 5 A). The fraction unblocked of the slow component was taken as the fractional amount of current at steady state of the slow component divided by the amount of current after fast block (fraction unblocked slow = unblocked/slow + unblocked; Fig. 5 A).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion Conclusion References

Open-channel Ba²⁺ dissociation

At negative voltages, external Ba²⁺ entered the pore of N-terminal deleted Shaker channels (ShB; Hoshi et al., 1990) and remained bound after wash-out of Ba²⁺ from the external solution, as measured by a depolarizing step after the wash-out (Fig. 1 A). This depolarization evoked current from a fraction of the channels at the normal opening rate, whereas the bulk of the current rose much more slowly, consistent with the idea that the majority of channels were open, but blocked by Ba²⁺, and that they then slowly became unblocked (Fig. 1 B). The open channel Ba²⁺ off rate was taken as the reciprocal of the time constant for the slow rising phase, which was fit well by a single exponential. In all of our analysis, we assume that the channel gates normally with a Ba²⁺ bound in the pore (but see Miller et al., 1987; Neyton and Pelleschi, 1991).

View larger version (16K): [in this window] [in a new window]   FIGURE 1   Open channel Ba2+ dissociation from wild-type ShB channels. (A) The Oocyte was held at 80 mV as the perfusing bath solution was switched from 0 mM Ba2+/112 mM K+ to 10 mM Ba2+/2 mM K+/92 mM Na+ and back to 0 mM Ba2+/112 mM K+ while holding at 80 mV. Two minutes after Ba2+ wash-out, a 800-ms depolarizing step opened the channels. (B) Current traces are in response to a 800-ms step to +95 mV before and after perfusion of 10 mM Ba2+. The post-Ba2+ trace contains a fast component reflecting opening of channels that were unbound to Ba2+, and a slow component reflecting Ba2+ unbinding from open channels. The slow component was fit with a single exponential, off = 280 ms.

These observations in the ShB channel are consistent with those found in the delayed rectifier K⁺ channel of squid giant axon (Armstrong et al., 1982), in which external Ba²⁺ enters the closed channel at negative voltages, binds at a deep high-affinity site, and is prevented from dissociating inwardly by the closed activation gate and outwardly by the electric field. These results are also similar to observations in the BK channel, where Ba²⁺ can be trapped in the channel when closure of the activation gate prevents its inward dissociation (Miller, 1987).

Dependence of open-channel Ba²⁺ dissociation on voltage and external K⁺

In 0 mM [K_out⁺], the rate of open channel Ba²⁺ dissociation increased with increasingly depolarized step potentials (Fig. 2, A and C), suggesting that at positive potentials Ba²⁺ dissociated primarily in the outward direction. The electrical distance from the Ba²⁺ ion binding site to its rate-limiting barrier for exit, _off (Woodhull, 1973), was 0.35, indicating that the binding site lies deep in the membrane electric field. Increasing [K_out⁺] from 0 mM to 2 mM slowed Ba²⁺ dissociation at all voltages studied, with little effect on _off (_off = 0.373; Fig. 2, B and C). The slowing of Ba²⁺ dissociation by external K⁺ (by threefold at 0 mV) suggests that a more external K⁺ ion binding site must be empty for outward Ba²⁺ dissociation. This behavior is very similar to that described for K⁺ occupancy of the high-affinity "lock-in" site seen in BK channels (Neyton and Miller, 1988a).

View larger version (13K): [in this window] [in a new window]   FIGURE 2   Ba2+ dissociation from wild-type ShB open channels is dependent on voltage and external K+ concentration. (A) Post-Ba2+ currents at two step potentials in 2 mM Kout+. The slow components of the traces were fit with single exponentials. The +60-mV trace (off = 72.0 ± 4.43 ms, n = 12) was significantly faster (p = 3.01 × 1010) than the +20-mV trace (off = 210 ± 11.6 ms, n = 16), suggesting outward Ba2+ dissociation in 2 mM Kout+. (B) Post-Ba2+ current in 0 mM Kout+ (off = 44.4 ± 8.83 ms, n = 4) was significantly faster (p < 1 × 104) than 2 mM Kout+ (off = 130 ± 8.83 ms, n = 12), Vstep = +40 mV. (C) Voltage dependence of open channel Ba2+ off rates for 0 mM Kout+ () and 2 mM Kout+ (). Lines are fits to the equation Koff = Koff(0 mV)*exp(zoffFV/RT), where Koff(0 mV) is the off rate with no applied voltage, off is the electrical distance from the Ba2+ binding site to the rate-limiting barrier to Ba2+ exit, z is the valence of Ba2+, and F, R, and T have their usual meanings. The 0 mV Ba2+ off rates are 6.65 s1 for 0 mM Kout+ and 2.40 s1 for 2 mM Kout+. off is 0.348 for 0 mM Kout+ and 0.373 for 2 mM Kout+.

Ba²⁺ dissociation was very different when the external concentration of [K⁺] was raised to 112 mM. At this high external K⁺ concentration, the rate of Ba²⁺ dissociation slowed with increased depolarization between +20 mV and +100 mV, and then increased at more positive potentials (Fig. 3, A and B). This suggests that in 112 mM [K_out⁺], Ba²⁺ has a propensity to dissociate in the inward direction, but at extreme depolarizations the electric field becomes strong enough to reverse this trend and drive Ba²⁺ outward. This behavior is very similar to that described for the "enhancement" site in BK channels that binds K⁺ with low affinity (Neyton and Miller, 1988b).

View larger version (18K): [in this window] [in a new window]   FIGURE 3   Ba2+ dissociates in the inward direction in 112 mM Kout+. (A) Post-Ba2+ traces are shown for steps to +20 mV and +80 mV step potentials in 112 mM Kout+. The time constants for Ba2+ dissociation were 20.3 ± 2.87 ms (n = 5) at +20 mV and 228 ± 5.49 ms (n = 4) at +80 mV (p = 3.52 × 109). (B) Voltage dependence of Ba2+ off rates are plotted against the step potential for 0 mM Kout+ (), 2 mM Kout+ (), and 112 mM Kout+ (). The lines are simultaneous fits to a model with three permeant ion binding sites (see Materials and Methods and panel), where Ba2+ outward dissociation from the deep site, site 3, is prevented when a K+ ion is located at the external "lock-in" site, site 1, and Ba2+ dissociation is speeded up in the internal direction when K+ ions occupy the "lock-in" site and the "enhancement" site, site 2, simultaneously. (C) A diagrammatic model of Ba2+ dissociation from its deep site is shown. In 0 mM Kout+, Ba2+ dissociates more easily in the outward than in the inward direction. In 2 mM Kout+, K+ sometimes is bound to the "lock-in" site, preventing outward Ba2+ dissociation. In 112 mM Kout+, K+ ions occupy the "lock-in" site and the "enhancement" site (which has lower affinity for K+ because of repulsion by Ba2+ at the deep site 3), and Ba2+ is repelled in the inward direction.

These observations indicated that, like BK, the deep pore of Shaker has three binding sites that are in ready exchange with the external solution: a deep Ba²⁺ binding site and two shallower sites, one with low affinity ("enhancement" site) and the other with higher affinity ("lock-in" site) for K⁺ in the Ba²⁺ bound channel.

Ion selectivity of the "lock-in" site

To examine the selectivity of the "lock-in" site, the 0 mV Ba²⁺ off rate was determined with different external ions at the low (2 mM) concentration (Fig. 4 A). Compared to 112 mM Na_out⁺, the 0 mV Ba²⁺ off rate was not slowed by 2 mM NH₄⁺, but was slowed threefold by 2 mM K⁺, and sevenfold by 2 mM Rb⁺. Replacement of 112 mM Na⁺ with 112 mM NMDG⁺, a large cation that presumably does not bind to the pore, increased the Ba²⁺ off rate by 1.5-fold. This indicates that Na⁺ itself bound to the "lock-in" site, but so weakly that it had an effect smaller than that of K⁺ at 56 times the concentration. The affinity series for the "lock-in" site was, therefore, Rb⁺ > K⁺  Na⁺ > NMDG⁺. This series is identical to the BK channel studied by Neyton and Miller (1988a). The high selectivity of the "lock-in" site for permeant ions suggests that this site lies in the pore. Changing the external ion concentration had little effect on _off, except in the case of Rb⁺ (Fig. 4 B).

View larger version (14K): [in this window] [in a new window]   FIGURE 4   Ion selectivity of the "lock-in" site. (A) Ba2+ off rates at 0 mV with different external ions. Values of the mean off rates are for 112 mM NMDG+: 10.4 ± 1.20 s1 (n = 6); 112 mM Na+: 7.06 ± 0.450 s1 (n = 23); 2 mM NH4+/112 mM Na+: 5.51 ± 0.182 s1 (n = 3); 2 mM K+/112 mM Na+: 3.27 ± 0.383 s1 (n = 12); and 2 mM Rb+/112 mM Na+: 1.00 ± 0.125 s1 (n = 3). Stars represent values that are significantly different (p < 0.05) from 112 mM Na+ by t-test. (B) Ba2+ off rates are plotted against step potential for different external ion concentrations. off values are 0.293 for 112 mM NMDG+ (), 0.335 for 112 mM Na+ (), 0.337 for 2 mM NH4+/112 mM Na+ (), 0.373 for 2 mM K+/112 mM Na+ (), and 0.239 for 2 mM Rb+/112 mM Na+ (). (C) Post 10 mM Ba2+ traces are shown for 2 mM Kout+ and 2 mM Baout2+. Values of  for 2 mM K+ and 2 mM Ba2+ are 59.3 ± 6.01 ms (n = 9) and 145 ± 11.1 ms (n = 7), respectively (p < 1 × 105). Vstep = +80 mV.

Ba_out²⁺ (2 mM) was also found to bind to the "lock-in" site and slow the outward dissociation of a second Ba²⁺ bound at the deep binding site (Fig. 4 C). In 2 mM external Ba²⁺, Ba²⁺ can associate with its deep site in channels that are not blocked, which should accelerate the slow rise phase of current, because  = 1/[Ba²⁺] +  (where  = on rate and  = off rate) compared to 0 Ba²⁺ during the washout, where _off = 1/. However, because the rise of current was slower in 2 mM Ba_out²⁺ than that in 2 mM K_out⁺, it seems instead that a Ba²⁺ ion located at the "lock-in" site was able to slow the dissociation of a second Ba²⁺ at the deep site. This effect was similar to the effect of monovalent ions at low concentration, in that greater depolarization accelerated Ba²⁺ dissociation ( = 406 ± 28.0 ms V_step = +20 mV n = 8;  = 145 ± 11.1 s V_step = +80 mV n = 7). This suggests that external Ba²⁺, like external K⁺, can bind at the "lock-in" site and block outward dissociation of a second Ba²⁺ bound at the deep high-affinity blocking site.

Electrical location of Ba²⁺ binding sites

To determine the relative locations within the electric field of the two Ba²⁺ binding sites, we examined the rates with which external Ba²⁺ reached and dissociated from them. Short repeated depolarizations applied during the wash-in of 10 mM external Ba²⁺ revealed two kinetic components of block onset, each representing about half of a total of 80% block (Fig. 5 A). The fast component was faster than the rate of exchange of Ba²⁺ in the bath, and was rapidly reversed after short exposures to Ba²⁺ (Fig. 5 B), whereas longer exposures that brought the slow component to steady-state reversed more slowly. Wash-out after a long exposure contained one component. These observations are consistent with those of Hurst et al. (1995), who suggested that Ba²⁺ binds to two sites, a shallow, rapidly equilibrating site and a deep, slowly equilibrating site, which are sequential in the pore (i.e., a shallow fast site and a deep slow site). This idea was supported by an examination of the dependence of the two components on the step potential. The electrical distance from the external solution to the Ba²⁺ ion binding sites () was smaller for the fast site (0.18) than for the slow site (0.38) (Fig. 5 C). Because the fast Ba²⁺ block site is situated shallowly in the membrane electric field, we concluded that this is likely to be the "lock-in" site, and because the slow block site lies deep in the membrane electric field, we concluded that this is likely to be the deep Ba²⁺ binding site. The  for the deep site (0.38) is very close to the _off (0.37) observed in the open channel off rate. This indicates that the majority of the voltage dependence for Ba²⁺ binding is in the off rate at this site.

View larger version (20K): [in this window] [in a new window]   FIGURE 5   Open and closed channel Ba2+ block of ShB. (A) Membrane voltage was held at 80 mV and a 30-ms step to +40 mV was made at 0.2 Hz. Steady-state current at the end of the step was normalized and plotted against time. Switching the external solution from 2 mM K+/112 mM Na+ to 10 mM Ba2+/2 mM K+/92 mM Na+ led to a biphasic decay. The fast component of decay was too fast to determine its time course. The slow component of decay was well fit with a single exponential ( = 90.1 s; amp. = 0.352). Washout of Ba2+ was also fit well with a single exponential ( = 30.8 s; amp = 0.671). The amounts of block due to the fast and slow components are shown as Fast and Slow. The fraction of channels unbound to the fast site (site 1) is 1  Fast. The fraction of channels unbound to the slow site (site 3) is Unblocked/(Unblocked + Slow). The fraction of channels unbound to site 1 and site 3 is Unblocked = 1  Fast  Slow. (The washout should have two exponentials; however, we see only one, for two possible reasons. First, the washout is faster than the slow component of wash-in, making it hard to distinguish two components. Second, of the channels that are bound at site 1, half are also bound at site 3. This lowers the number of channels that can contribute to a fast component of washout.) (B) Membrane voltage was held at 80 mV and was stepped for 30 ms to 0 mV at 0.2 Hz. Switching external solution from 2 mM K+/112 mM Na+ to 10 mM Ba2+/2 mM K+/92 mM Na+ for a brief period of time (solid bar) caused a quickly relievable block of current. (C) Voltage dependence of block for the fast and slow components of Ba2+ wash-in. The fraction unblocked of the fast component was calculated as in A (Fraction unblocked fast = 1  Fast) and plotted against step potential. This is the fraction of channels that are unbound to site 1. The values for Kdfast and fast for the fast component are 10.0 mM and 0.18, respectively, which were obtained from the fit to the equation: Fraction unblocked fast = (1 + [Ba2+]efast FV2/RT/Kdfast)1. (see Materials and Methods). The fraction of channels that were not blocked at site 1 or site 3 was calculated as in A (Total Unblocked = 1  Fast  Slow) and plotted against step potential. The values of Kdslow and slow for the slow component are 1.11 mM and 0.38, respectively, which were obtained from the fit to the equation Total Unblocked fraction = (1 + [Ba2+]efastFV2/RT/Kdfast + [Ba2+]eslow FV2/RT/Kdslow)1, where Kdfast and fast were used from the fit to the fraction unblocked of the fast component. Kdfast and Kdslow are the Ba2+ dissociation constants at 0 mV, and  is the fraction of the membrane electric field traversed by Ba2+ from the external solution to its binding site. Solid lines are fits to the above equations. Membrane voltage and external solutions were controlled as in A. (D) Determination of closed channel Ba2+ off rate. Ba2+ off rates were determined from single exponential fits to the Ba2+ wash-out. This protocol gives Ba2+ dissociation from both open and closed channels; however, the contribution to the total off rate (open plus closed channel) from the open channels decreases with shortening of the step duration. The linear regression extrapolated value of the Ba2+ off rate at 0 step duration is the closed-channel off rate (0.01322 s1).

Closed channel unblock of wild-type ShB

The fact that the fraction unblocked for both fast and slow block increased with greater step depolarization indicates that significant Ba²⁺ dissociation occurs during the brief step (i.e., from the open state). Therefore the total Ba²⁺ dissociation rate reflects unbinding from both open channels during the step and closed channels during the interval between steps. To estimate the closed channel Ba²⁺ off rate from its deep binding site, the duration of the step potential was decreased over a range from 90 to 20 ms (Fig. 5 D). As the step duration is decreased, the contribution of open channels to the total Ba²⁺ off rate (open plus closed channels) is decreased. Extrapolation of the linear regression fit of the Ba²⁺ off rate versus step duration to 0 ms gives the upper limit of the Ba²⁺ off rate for closed channels at the 80 mV holding potential. This value is the upper limit, because some Ba²⁺ dissociation will occur during the tail. This value was 60-fold slower than the extrapolated value of the open channel off rate from the deep site at 80 mV. This suggests either that opening of the activation gate admits internal K⁺, which knocks Ba²⁺ in the outward direction, accelerating the exit of Ba²⁺ from the channel, or that a conformational change in the pore caused by opening lowers an energetic barrier that is rate-limiting for outward Ba²⁺ dissociation.

C-type inactivation traps Ba²⁺ in the pore of wild-type channels

Recent studies have suggested that C-type inactivation occurs by the squeezing shut of the outer mouth of the pore (Yellen et al., 1994; Liu et al., 1996) after the evacuation of a K⁺ binding site with an affinity of 2 mM (Baukrowitz and Yellen, 1995, 1996). We asked whether the C-type inactivation gate is analogous to the activation gate (Armstrong and Hille, 1972) in pinching off the outer end of the conduction pathway, rather than producing a general collapse of the pore. We tested this by loading Ba²⁺ into its deep binding site, inducing C-type inactivation, washing external Ba²⁺ away, and then examining whether conduction returned with the kinetics of recovery from C-type inactivation, or with the relatively slower kinetics of Ba²⁺ dissociation.

Channels were blocked by exposure to external Ba²⁺, and the membrane was held at 0 mV to induce C-type inactivation. Ba²⁺ was then removed from the external solution, and the membrane was held at 0 mV for an additional 10 min. This 10-min wash-out period represents 4000 times the time constant of open channel Ba²⁺ dissociation at 0 mV (K_off 0 mV = 6.65 s¹). This should permit all of the channels to recover from block, unless closure of the C-type inactivation gate prevents dissociation of the bound Ba²⁺. After the wash-out period, the membrane was returned to 80 mV and stepped for short pulses to 0 mV, as before, to test the conducting state of the channels. If C-type inactivation does not retard outward Ba²⁺ dissociation at all, the channels should all lose their Ba²⁺ during the 10-min wash-out, and the current should recover with the kinetics of recovery from C-type inactivation ( = 12.9 ± 1.05 s, n = 11; Fig. 6). Instead, after the 10-min wash-out of Ba²⁺, channels recovered monoexponentially with a  = 63.8 ± 4.5 s (n = 11), fivefold (p < 1 × 10⁹) more slowly than recovery from C-type inactivation alone. We observed a similar rate of 63.2 ± 4.6 s (n = 7) when we monitored recovery with the same protocol, but omitting the long step to 0 mV that induced C-type inactivation. This demonstrates that the ShB channel can close its C-type inactivation gate with Ba²⁺ occupying the deep site and trap Ba²⁺ in the pore.

View larger version (20K): [in this window] [in a new window]   FIGURE 6   C-type inactivation slows Ba2+ dissociation from its deep site in wild-type ShB channels. Perfusion of 10 mM Baout2+ (solid bar) and switch from Vhold = 80 mV with steps to 0 mV to Vhold = 0 mV (open bar) to induce C-type inactivation. After 10 min of wash-out of Ba2+ at 0 mV, holding the channels in the C-type inactivated state, 66.5 ± 0.02% (n = 4) of the channels recovered and conducted in 50 s, compared to 100% (n = 4) of the channels conducting 50 s after a 0-mV depolarization for 3 min in 0 Ba2+.

Mutations in the P-region affect the binding of Ba²⁺

In an effort to identify the molecular constituents of the Ba²⁺ and K⁺ binding sites, mutations were made in the P-region of ShB. We focused our mutagenesis on the polar residues in the K⁺ channel "signature sequence" (Heginbotham et al., 1994). Not all mutants produced functional channels. Of the mutations made, T439Y, Y445C, D447N, and D447T did not give functional homomultimers, whereas T441C, V443G, and Y445F did. Mutations T441C and Y445F preserved selectivity for K⁺ over Na⁺ (not shown), whereas V443G destroyed channel selectivity, as has been shown earlier (Heginbotham et al., 1994). Cysteines substituted at each of these positions (T441, V443, and Y445) have been shown earlier to be accessible to either internal or external thiol reagents, suggesting that the native residues face the permeation pathway (Lu and Miller, 1995; Pascual et al., 1995).

T441C greatly reduced the slow component of Ba²⁺ block (from 63% to 19%, p < 10⁵; Fig. 7 A), whereas the amount of fast block remained unaffected. This suggests that the external end of the channel, where fast block occurs, was not altered by the mutation, but that a region deep in the pore, where slow block occurs, was affected. The fact that the residual slow component of wash-out of Ba²⁺ was twofold faster (p = 0.046) and the fraction slow block was much less for the mutation than for wild-type ShB, indicates that the Ba²⁺ affinity for the deep binding site was decreased because of destabilization of Ba²⁺ binding.

View larger version (15K): [in this window] [in a new window]   FIGURE 7   Mutations in the P-region affect barium block. Solutions are 0 Ba2+ (2 mM K+/112 mM Na+/0 Ba2+) and 10 mM Ba2+ (2 mM K+/92 mM Na+/10 mM Ba2+). The fraction unblocked of the fast component was measured as the fraction of the total current in the absence of Ba2+ after equilibration of the fast component as in Fig. 5 C (Fraction unbound Fast = 1  Fast). Fraction unblocked of the slow component was measured as the current value at steady state, expressed as a fraction of the current remaining after fast block (Unblocked/Unblocked + Slow). (A) Mutation T441C had a fast Ba2+ block similar to that of wild-type ShB (T441C Fraction unblocked = 0.566 ± 0.024, n = 6; wild-type ShB Fraction unblocked = 0.650 ± 0.013, n = 6); however, the fraction unblocked of the slow component was significantly increased (p = 1 × 105) for T441C (0.813 ± 0.042, n = 3) compared to wild-type ShB (0.375 ± 0.0313, n = 6). The time constant for the slow component of wash-in is 32.7 ± 11.71 s (n = 2). The time constant for wash-out for T441C is 17.00 ± 4.98 s (n = 3), which is significantly shorter (p = 4.56 × 102) than that of wild-type ShB (27.86 ± 1.75 s, n = 6). The amplitudes for the wash-in and wash-out of the slow component were 0.10 ± 0.02 and 0.21 ± 0.076 (n = 2), respectively (Vhold = 80 mV; Vstep = +40 mV). (B) Mutation V443G increased Ba2+ blocking kinetics so much that the time constants of block could not be determined. Fraction unblocked = 0.126 ± 0.056 (n = 2) (Vhold = 100 mV; Vstep = +40 mV). (C) The mutation Y445F had both fast and slow ( = 69.1 ± 8.63 s; amp = 0.625 ± 0.069; n = 5) components of block for Ba2+ wash-in. The fraction unblocked for the fast component for Y445F 0.610 ± 0.0239 (n = 7) was comparable (p = 0.173) to wild-type ShB 0.650 ± 0.0149 (n = 7); however, the slow component was slower and had a smaller fraction unblocked. The fraction unblocked of the slow component for Y445F was 0.101 ± 0.0227 (n = 7), which is significantly smaller (p < 105) than wild-type ShB 0.395 ± 0.0282 (n = 7). The Ba2+ wash-out was significantly slower (p < 0.01) for Y445F (1670 ± 390 s, n = 8) compared to 27.9 ± 1.75 s (n = 5) for wild-type ShB (Vhold = 80 mV; Vstep = +40 mV).

V443G accelerated the kinetics of block without altering the total fraction bound (Fig. 7 B). One possibility is that, unlike T441C, which altered binding of Ba²⁺ at its deep site, V443G left binding normal, but eased the entry and exit of Ba²⁺ to and from the deep site, by reducing an energy barrier to Ba²⁺ movement. However, we cannot distinguish this model from a more complex model in which both the deep and shallow sites are altered, with an increased affinity of the shallow site and a decreased affinity of the deep site.

C-type inactivation with Ba²⁺ in the pore of Y445F channels

Y445F had a fast component of block similar to wild-type ShB and an increased sensitivity to slow block (Fig. 7 C). This increase in affinity was due to a dramatic (60-fold, p < 0.01) slowing of Ba²⁺ dissociation, evident in the wash-out. This mutant also exhibited an unusually fast C-type inactivation (Fig. 8 A). Recovery from this inactivation was slow enough so that, unlike wild-type ShB, 5-s intervals at 80 mV were insufficient to prevent accumulation of inactivation during the repeated 30-ms steps (Fig. 8 B). This raised the possibility that the apparent increase in the Ba²⁺ affinity of site 3 is actually due to trapping of Ba²⁺ at that site by closure of the C-type inactivation gate. We examined this possibility further.

View larger version (14K): [in this window] [in a new window]   FIGURE 8   Y445F channels can C-type inactivate with a Ba2+ ion in the pore. (A) Mutation Y445F increases the rate of C-type inactivation compared to wild-type ShB. The time constant for inactivation in 2 mM Kout+ for wild-type ShB (4190 ± 93.9 ms, n = 4) was 150-fold (p < 1 × 1011) slower than that of Y445F (25.2 ± 3.81 ms, n = 6). For Y445F the time constant for inactivation in 2 mM Kout+ was significantly faster (p < 1 × 107) than 112 Kout+ (144 ± 9.58 ms, n = 6) (Vhold = 80 mV; Vstep = +40 mV). (B) The membrane voltage was held at 80 mV and stepped to +40 mV for 30 ms at 0.2 Hz. Repeated depolarizations cause a decrease in Y445F current, unlike in wild-type ShB. This is probably due to accumulated C-type inactivation in Y445F. (C) Ba2+ wash-out of Y445F is faster (p < 0.05) with 112 mM Kout+ (off = 116 ± 19.3 s, n = 4) than 2 mM Kout+ (off = 1670 ± 390 s, n = 8), suggesting that 112 mM Kout+ slows C-type inactivation, allowing Ba2+ to dissociate more quickly (Vhold = 80 mV; Vstep = +40 mV).

Increasing external K⁺ from 2 mM to 112 mM slowed C-type inactivation in Y445F by sixfold (Fig. 8 A), comparable to the effect observed in wild-type ShB and ShB channels (Lopez-Barneo et al., 1993; Baukrowitz and Yellen, 1996). This indicates that the K⁺ site that influences C-type inactivation in Y445F is intact. To determine if the fast closure of the C-type inactivation gate in Y445F was responsible for the slow time course of Ba²⁺ wash-out rate, we compared Ba²⁺ wash-out in low and high external [K⁺]. If the slow wash-out of Ba²⁺ from Y445F channels was due to the channel C-type inactivating on Ba²⁺, then Ba²⁺ wash-out should be faster in 112 mM K_out⁺, where C-type inactivation is slowed. Indeed, the rate of Ba²⁺ wash-out was 14-fold (p = 2 × 10²) faster when external K⁺ was raised to 112 mM compared to 2 mM (Fig. 8 C).

The increase in the rate of Ba²⁺ dissociation in high K_out⁺ could not have been due to enhanced Ba²⁺ inward dissociation, because dissociation was faster at +80 mV than at +20 mV (Fig. 9 A). This voltage dependence indicates that the bulk of the Ba²⁺ dissociations for Y445F in 112 mM K_out⁺ were in the outward direction. This is in contrast to wild-type ShB channels, where 112 mM K_out⁺ accelerates inward dissociation (Fig. 3 B). This result suggests that the "enhancement" site is disrupted in Y445F channels, whereas the preservation of fast Ba²⁺ block, as shown above, suggests that the "lock-in" site remains intact.

View larger version (15K): [in this window] [in a new window]   FIGURE 9   Mutation Y445F disrupts a K+ ion binding site. (A) Ba2+ off rates for Y445F were significantly faster (p < 0.01) for steps to +80 mV (n = 9) than +20 mV (Vhold = 80 mV; Vstep = +40 mV; step duration = 30 ms; 0.2 Hz). (B) Absence of slow Ba2+ block in wild-type ShB channels in 112 mM Kout+ (Vhold = 80 mV; Vstep = +40 mV). (C) Wash-in of Ba2+ has two components for Y445F in 112 mM Kout+ (1 = 19.6 ± 0.847 s; amp1 = 0.618 ± 0.028; 2 = 232.9 ± 19.4 s; amp2 = 0.284 ± 0.026; n = 7).

Mutation Y445F lowers the affinity of a K⁺ ion binding site

To test the above interpretation that Y445F disrupts the "enhancement" site, we compared wild-type ShB and Y445F with respect to the onset of Ba²⁺ block in high external K⁺. This test was motivated by the results of Hurst et al. (1995), which showed that high external K⁺ eliminates the slow component of Ba²⁺ block and attenuates the fast component in N-terminal deleted ShH4 channels, an observation that we repeated (Fig. 9 B). This result suggests that high K⁺ occupancy of the "enhancement" site or the deep Ba²⁺ binding site prevents external Ba²⁺ from reaching the deep site. In contrast to wild-type ShB, Y445F preserved both the fast and slow components of block in 92 mM K_out⁺, although both were reduced in magnitude and in rate compared to 2 mM K_out⁺ (compare Figs. 7 C and 9 C). This indicates that in Y445F channels Ba²⁺ can reach its deep site, even in 92 mM K_out⁺, and is consistent with a decreased occupancy of a low-affinity K⁺ binding site in Y445F channels. Because the fast and slow Ba²⁺ binding sites were intact in Y445F channels (although the affinity of the deep site could not be ascertained because of the trapping by C-type inactivation), the low-affinity K⁺ binding site affected by this mutation seemed most likely to be the "enhancement" site.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion Conclusion References

Ion binding sites

Our results on the Shaker channel are consistent with the existence of at least three single-file permeant ion binding sites in the pore that can be simultaneously occupied. All of these sites are in exchange with the external solution when the activation gate is closed. As shown earlier for BK channels (Neyton and Miller, 1988a, b), we find for Shaker that low concentrations of external permeant ion slow outward Ba²⁺ dissociation, and high concentrations drive Ba²⁺ inward, except at very strong depolarizations. Neyton and Miller referred to these two permeant ion effects as "lock-in," which they attributed to block of the pathway for Ba²⁺ outward dissociation from its deep site due to occupancy of a distal site, and "enhancement," which they attributed to repulsion of Ba²⁺ in the inward direction out of its binding site due to occupancy of a nearby site. Such repulsion of a blocker by a permeant ion has also been observed in BK channels by other workers (Yellen, 1984) and in the delayed rectifier of the squid giant axon (Bezanilla and Armstrong, 1972). Because the repulsion is mutual, binding of the permeant ion to the "enhancement" site is unfavorable when the deep site is occupied by Ba²⁺, so that this only occurs at high external concentration of permeant ion. A simple model that incorporates these principles accounted well for the voltage dependence of open channel Ba²⁺ dissociation in three different external K⁺ concentrations (see below).

Our results indicate that the "lock-in" site is highly selective for permeant ions, suggesting that it lies within the permeation pathway. Because it can also be occupied by Ba²⁺, it is most likely the shallow site ( = 0.18) that rapidly equilibrates with extracellular Ba²⁺ before Ba²⁺ goes deeper into the pore and binds tightly to its deeper site ( = 0.38). This is in agreement with the earlier results of Hurst et al. (1995). Both the electrical location of the "lock-in" site, and the site's affinity for K⁺ predicted from the fit of our three binding site model (K_d = 0.75 mM; see below), are quantitatively similar to measurements on BK channels ( = 0.15, K_d = 0.3 mM; Neyton and Miller, 1988a, b). (It should be noted that in our experiments and in the studies of Neyton and Miller, the affinity of the "lock-in" site is measured with a Ba²⁺ already bound to the deep site. This might cause the affinity of the K⁺ site to be less than if Ba²⁺ was not present because of any repulsion between the ions; however, we do not think this is so. See below.) The selectivity series of the "lock-in" site is also similar between these channels (Rb⁺ > K⁺ > NH₄⁺), except that Shaker appears to be more selective against Na⁺. Replacing 112 mM Na_out⁺ with 112 mM NMDG_out⁺ had little effect on the Ba²⁺ off rate for ShB, whereas Neyton and Miller (1988a) observe a fourfold increase in the Ba²⁺ off rate when Na⁺ is replaced with NMDG⁺.

The K⁺ affinity of the "enhancement" site is also similar between these two channels, requiring over 100 mM external K⁺ to raise occupancy to the point that inward Ba²⁺ dissociation is favored at modest depolarizations. The voltage dependence of the off rate for outward and inward Ba²⁺ dissociation was also similar for the two channels. For outward dissociation in 0 mM K_out⁺, the _off for BK was 0.42 and the _off for ShB was 0.35. For inward dissociation the _off was 0.32 for BK (150 mM K_out⁺) and 0.34 for ShB (112 mM K_out⁺). These results demonstrate that there is a high degree of conservation of pore structure between these subfamilies of K⁺ channels.

Three ion binding site model

A simple three binding site model (see Materials and Methods) of the Shaker pore, of the kind proposed qualitatively by Neyton and Miller for the BK channel (1988a, b), accounted well for the observed voltage dependence of open channel Ba²⁺ dissociation in the three different external K⁺ concentrations. The "lock-in" site has the experimentally determined electrical position of the fast Ba²⁺ site of 0.18. The "enhancement" site lies between the "lock-in" site and the deep, slow Ba²⁺ binding site, which has the experimentally determined electrical position of 0.38.

Wash-in and prolonged wash-out of external Ba²⁺ from closed channels at negative voltage leaves one Ba²⁺ loaded in the deep Ba²⁺ site, whereas Ba²⁺ is washed out of the "lock-in" and "enhancement" sites, where it binds weakly. (Ba²⁺ is held weakly at the "lock-in" site because of its rapid equilibration (Fig. 5 A), and repulsion by Ba²⁺ at the deep site prevents a second Ba²⁺ from binding tightly at the nearby "enhancement" site.) Therefore, when the external solution is devoid of external K⁺, the "lock-in" and "enhancement" sites are vacant. At 2 mM external K⁺ concentration, K⁺ occupies the "lock-in" site a large fraction of the time, blocking the pathway for outward Ba²⁺ dissociation from the deep site, so that the Ba²⁺ must wait until the "lock-in" site is vacated before it can exit. Repulsion by Ba²⁺ at its deep site prevents K⁺ from occupying the "enhancement" site at low external [K⁺]_out. However, when external K⁺ concentration is raised to 112 mM (56-fold higher), the "enhancement" site becomes significantly occupied. The mutual repulsion between the K⁺ at the "enhancement" site and Ba²⁺ at its deep site drives Ba²⁺ in the inward direction.

Simultaneous fits of the model to the curves relating open channel Ba²⁺ off rate to voltage in 0, 2, and 112 mM [K⁺]_out fit the data well (Fig. 3 B, solid lines). The affinity of the "lock-in" site for K⁺ predicted from fits of our three binding site model (K_d = 0.75-0.80 mM) agrees well with that measured BK channels (K_d = 0.3 mM; Neyton and Miller, 1988a).

The C-type inactivation gate is external to the deep Ba²⁺ binding site

Armstrong et al. (1982) showed that Ba²⁺ can enter the closed delayed rectifier K⁺ channel from the outside and bind with high affinity at its blocking site. Armstrong et al. also showed that internal Ba²⁺ was prevented from access to its binding site when the activation gate was closed. This indicated that activation gating operates by opening and closing access of the pore to ions in the internal solution. We find here that another gate, the C-type inactivation gate, can function in a similar manner from the opposite end of the pore. Access to the external solution of a Ba²⁺ ion bound at its deep site in the pore was cut off by closure of the C-type inactivation gate. This indicates that the C-type inactivation gate pinches off the permeation pathway between the deep Ba²⁺ site and the external mouth of the pore, and suggests that C-type inactivation does not produce a general collapse of the pore. The rearrangement that brings residues in the P-S6 of one subunit closer to those of another when the C-type inactivation closes (Liu et al., 1996) seems to therefore represent a local structural event. Recent work by Molina et al. (1997) suggests that the pinching off is internal to residue T449, which places the C-type inactivation gate between T449 and T441, the deep Ba²⁺ binding site.

The rate of C-type inactivation depends on the external concentration of K⁺ (Lopez-Barneo, 1993; Baukrowitz and Yellen, 1995, 1996). Using organic blockers with differing bound times at the internal mouth of the channel, Baukrowitz and Yellen (1996) showed that block of outward current flow increased the sensitivity of C-type inactivation to external K⁺