Modulation of slow inactivation in human cardiac Kv1.5 channels by extra- and intracellular permeant cations

Abstract

1. The properties and regulation of slow inactivation by intracellular and extracellular cations in the human heart K⁺ channel hKv1.5 have been investigated. Extensive NH₂- and COOH-terminal deletions outside the central core of transmembrane domains did not affect the degree of inactivation.

2. The voltage dependence of steady-state inactivation curves of hKv1.5 channels was unchanged in Rb⁺ and Cs⁺, compared with K⁺, but biexponential inactivation over 10 s was reduced from ∼100% of peak current in Na⁺ to ∼65% in K⁺, ∼50% in Rb⁺ and ∼30% in Cs⁺. This occurred as a result of a decrease in both fast and slow components of inactivation, with little change in inactivation time constants.

3. Changes in extracellular cation species and concentration (5-300 mM) had only small effects on the rates of inactivation and recovery from inactivation (τ_recovery∼1 s). Mutation of residues at a putative regulatory site at R487 in the outer pore mouth did not affect slow inactivation or recovery from inactivation of hKv1.5, although sensitivity to extracellular TEA was conferred.

4. Symmetrical reduction of both intra- and extracellular cation concentrations accelerated and augmented both components of inactivation of K⁺ (K_d = 34.7 mM) and Cs⁺ (K_d = 20.5 mM) currents. These effects could be quantitatively accounted for by unilateral reduction of intracellular K⁺ (${\text{K}}_{\text{i}}^{\text{+}}$) (K_d = 43.4 mM) or ${\text{Cs}}_{\text{i}}^{\text{+}}$ with constant 135 mM external ion concentrations.

5. We conclude that inactivation and recovery from inactivation in hKv1.5 were not typically C-type in nature. However, the ion species dependence of inactivation was still closely coupled to ion permeation through the pore. Intracellular ion modulatory actions were more potent than extracellular actions, although still of relatively low affinity. These results suggest the presence of ion binding sites capable of regulating inactivation located on both intracellular and extracellular sides of the pore selectivity filter.

Slow inactivation is responsible for eventual current decay during prolonged depolarizations in many rapidly activating but slowly inactivating delayed rectifier K⁺ channels. Understanding the regulation of this process is important in understanding the role that these channels have in vivo. In the human heart, the K⁺ channel hKv1.5 is active during atrial action potential repolarization (Wang et al. 1993), and is present in significant amounts in both human (Mays et al. 1995) and rat ventricular muscle (Dixon & McKinnon, 1994; Barry et al. 1995). Modulation of outward K⁺ current inactivation at this time will have important effects on action potential repolarization, duration, and therefore contractility. The most prevalent form of slow inactivation is C-type (Hoshi et al. 1991; Kukuljan et al. 1995), which is thought to result from a slow constriction of the outer mouth of the pore (Liu et al. 1996), as evidenced by changes in Cd⁺ affinity (Yellen et al. 1994) and studies on the subunit dependence of inactivation (Ogielska et al. 1995; Panyi et al. 1995), and may be associated with a change in the permeability of the channel towards Na⁺ ions (Starkus et al. 1997). A number of amino acid residues in the S6 or outer pore mouth have been shown to be important in the control of C-type inactivation: the first is A463 in the Shaker B (ShB) channel (Hoshi et al. 1991), or A413 in Kv1.3 (Panyi et al. 1995); the second is T449 in ShB, nearer the pore region (Lopez-Barneo et al. 1993; Ogielska et al. 1995; Schlief et al. 1996); H401 is significant for inactivation in Kv1.3 (Busch et al. 1991; Pardo et al. 1992). Other forms of slow inactivation do exist, most notably P-type, present in Kv2.1 (DeBiasi et al. 1993) where inactivation is ascribed to a single I369L change, and can be modulated in the deep pore region (Kiss & Korn, 1998).

Evidence indicates that in K⁺ channels, permeating ions can modulate channel gating kinetics (Pardo et al. 1992; Lopez-Barneo et al. 1993), which is in contrast with the original assumptions that ion channel gating proceeds independently of the presence or species of permeant ion. C-type inactivation is modified by extracellular ion concentrations in both native (DeCoursey, 1990) and cloned Shaker (Lopez-Barneo et al. 1993; Baukrowitz & Yellen, 1995), Kv1.3 (Marom & Levitan, 1994), and Kv1.4 channels (Pardo et al. 1992; Rasmusson et al. 1995), and by extracellular tetraethylammonium (TEA) ions (Grissmer & Cahalan, 1989; MacKinnon & Yellen, 1990; Kavanaugh et al. 1991). This is not a universal phenomenon; in Kv2.1 inactivation is not slowed by extracellular TEA or extracellular K⁺ (${\text{K}}_{\text{o}}^{\text{+}}$) (Klemic et al. 1998). In Kv1 channels, recovery from inactivation is modulated by extracellular K⁺ in Kv1.3 (Levy & Deutsch, 1996) and Kv1.4 (Rasmusson et al. 1995). These experiments all indicate that during the onset of C-type inactivation, occupancy of a site in the mouth of the K⁺ channel pore by ions can slow channel closing in a ‘foot-in-the-door’ manner (Marchais & Marty, 1979; Swenson & Armstrong, 1981; Lopez-Barneo et al. 1993). Ions like Cs⁺, Na⁺, or NMG⁺, substituted for K⁺ in the extracellular solution lead to an acceleration of C-type inactivation due to their lower affinity for this regulatory site in the outer pore mouth (Lopez-Barneo et al. 1993).

Much of the data on slow inactivation has been obtained from channels such as ShB or Kv1.4 mutants which have had their pre-existing rapid inactivation removed. Here we have examined slow inactivation in human Kv1.5 channels expressed in human embryonic kidney cells. hKv1.5 channels have no N-terminal ball sequence to cause fast inactivation (Philipson et al. 1991; Tamkun et al. 1991; Fedida et al. 1993) and the slow inactivation that is observed has been ascribed to C-type inactivation (Snyders et al. 1993). Our data indicate that the slow inactivation shares some features, but lacks other properties of ‘classical’ C-type inactivation. Changes in the extracellular cation species or concentration had relatively minor effects on the inactivation rate, and on the recovery from inactivation. Conferring TEA sensitivity did not affect the time course of slow inactivation, and a comparison of the different inactivation rates induced by permeating Cs⁺ and K⁺ suggested that intracellular or permeating ions were more important in the control of inactivation in this channel, with significant regulation occurring at a site more accessible from the inner mouth of the pore.

METHODS

Cell preparation and transient transfection

In the majority of experiments, the human embryonic kidney cell line 293 (HEK 293) was used as a transient expression system due to the ease of transfection of cells, the ability to express hKv1.5 at high levels (Wible & Fedida, 1994), and the lack of endogenous β-subunits in these cells (Uebele et al. 1996). The cells were grown in Minimal Essential Medium (MEM) at 37°C in an air-5% CO₂ incubator. One day before transfection, cells were plated on glass coverslips in 25 mm Petri dishes with 30-50% confluence. On the day of transfection, cells were washed twice with MEM. In order to identify the transfected cells efficiently, hKv1.5 DNA in pcDNA3 was co-transfected with the vector pHook-1 (Invitrogen, San Diego, CA, USA). This plasmid encoded the production of an antibody to the hapten phOX, which when expressed is displayed on the cell surface. The hKv1.5 plasmid was mixed with pHook-1 (in a 1:1 ratio, 1.25 μg each) plus 15 μl lipofectin, and then added to the dishes containing HEK cells in 1 ml MEM. After 4-6 h of transfection, the medium was changed to MEM supplemented with 10% fetal bovine serum and an antifungal agent. Cells were cultured for a further 24-36 h to allow channel expression, before recording. One hour prior to experiments cells were treated with beads coated with phOX. After 15 min, excess beads were washed off with cell culture medium and cells which had beads stuck to them were used for electrophysiological experiments. All cell culture supplies were obtained from Canadian Life Technologies (Bramalea, Ontario, Canada). In some experiments a stable line expressing hKv1.5 at a high level was used. We have described the fabrication and use of such lines previously (Fedida et al. 1993; Wible & Fedida, 1994).

Solutions

The standard pipette filling solution contained (mM): KCl, 135; EGTA, 5; MgCl₂, 1; Hepes, 10; and was adjusted to pH 7.2 with KOH. When KCl was replaced with either RbCl, NaCl, or CsCl, pH was adjusted with RbOH, NaOH, or CsOH, respectively. When ${\text{K}}_{\text{i}}^{\text{+}}$ or ${\text{Cs}}_{\text{i}}^{\text{+}}$ were reduced as in Fig. 7, N-methyl-D-glucamine (NMG) was used as a replacement. The standard bath solution contained (mM): Hepes, 10; MgCl₂, 1; CaCl₂, 1; the balance of ions was made up with 140 mM NMG and was adjusted to pH 7.4 with HCl. For recordings in the presence of different external Na⁺, K⁺, Rb⁺, or Cs⁺ concentrations, the NMG base external solution was used and the concentration of NMG was reduced as the cation concentration was elevated to maintain constant osmolarity (except of course in 300 mM ${\text{K}}_{\text{o}}^{\text{+}}$), and was adjusted to pH 7.4 with the appropriate hydroxide solution. All chemicals were from Sigma.

Figure 7. Effect of different symmetrical K⁺ and Cs⁺ permeant ion concentrations on the components of slow inactivation.

A, rapid and complete inactivation of currents with symmetrical 5 mM K⁺ in both pipette and bath. Currents obtained during 2 s depolarizations from -80 mV to between -20 and +70 mV. B, fast time constants of current inactivation (τ₂) at +70 mV at different symmetrical K⁺ or Cs⁺ concentrations in pipette and bath. For K⁺ data, n = 4-9 at each concentration; for Cs⁺ data, n = 4-8 at each concentration. For practical recording reasons, data at 300 mM ${\text{K}}_{\text{o}}^{\text{+}}$ could only be obtained with 130 mM ${\text{K}}_{\text{i}}^{\text{+}}$. Data were fitted with a Hill equation (continuous line) which gave a K_d of 34.7 mM for K⁺ (n_H= 2.5), and a K_d of 20.5 mM for Cs⁺ (n_H= 0.9). C, τ₁ time constants of current inactivation at +70 mV at different symmetrical K⁺ or Cs⁺ concentrations, as for B. D, residual non-inactivating K⁺ and Cs⁺ currents obtained from bi-exponential fits of eqn (2). Values are expressed as a fraction of peak current.

Electrophysiological procedures

Coverslips containing cells were removed from the incubator before experiments and placed in a superfusion chamber (volume 250 μl) containing the control bath solution at 22-23°C. Whole-cell recordings were made using Axopatch 200A or 1D amplifiers, with 1/100 or 0.1/100 headstages (Axon Instruments). Patch electrodes were pulled from thin-walled borosilicate glass (World Precision Instruments) on a horizontal micropipette puller, fire polished, and filled with appropriate solutions. Electrodes had resistances of 1.5-3.0 MΩ when filled with control filling solution. Analog capacity compensation and 80-90% series resistance compensation were used in all whole-cell measurements. Membrane potentials have not been corrected for small junctional potentials that arose between pipette and bath solutions. For the bath solutions containing NMG and 5 mM concentrations of other monovalent cation (K⁺, Rb⁺, Na⁺, or Cs⁺) this amounted to -6.2 ± 0.5 mV (Cs⁺) and -6.1 ± 0.4 mV (Rb⁺) (means ±s.e.m., n = 6 trials) against internal solutions. For all other solutions measured junction potentials were less than 1 mV. Data were filtered at 1-10 kHz before digitization and stored on a microcomputer for later analysis using pCLAMP 6 software (Axon Instruments).

Data analysis

The voltage dependence of steady-state hKv1.5 channel inactivation with different permeating ions was fitted with a Boltzmann function:

(1)

where V_½ represents the voltage at which 50% inactivation occurred, V is the membrane potential, and k is the slope factor that reflects the steepness of the voltage dependence. Because not all channels are inactivated at the end of a 10 s pulse, C represents the fraction of non-inactivated channels.

The inactivation kinetics of current carried by K⁺, Rb⁺ or Cs⁺ were determined by double exponential fits obtained at different potentials during 10 s pulses or 10-20 s pulses for Cs⁺, according to the equation:

(2)

in which τ₂ and τ₁ are the fast and slow time constants of inactivation, A₂ and A₁ are the corresponding amplitudes and C indicates the non-inactivating current amplitude at the end of the pulse. For data in Figs 7 and 8 when the K⁺ concentrations were reduced, inactivation was greatly accelerated and pulse durations were shortened accordingly. Experimental values are given as means ±s.e.m. One-way ANOVA was used to compare K⁺, Rb⁺ and Cs⁺ current kinetics. A value of P < 0.05 was considered statistically significant.

RESULTS

Slow inactivation in hKv1.5 is determined within the central core of membrane-spanning domains

In Kv2.1, large amino terminal deletions of 100-139 residues slow activation but also completely prevent the development of slow inactivation over 10 s (Van Dongen et al. 1990; Pascual et al. 1997). Kv1.4, which shows 70% amino acid homology to hKv1.5, has a ball and chain motif responsible for fast inactivation (Tseng & Tseng-Crank, 1992). When this is deleted, other domains in the N-terminus can also mediate a slower form of inactivation (Kondoh et al. 1997), although the predominant residual form of inactivation in this channel is C-type (Rasmusson et al. 1995). hKv1.5 lacks these motifs but does exhibit a relatively fast form of inactivation, and has a long intracellular N-terminal domain, so it was important to ensure that the slow inactivation process in hKv1.5 was limited to the core region of the channel.

N- and C-terminal mutant hKv1.5 channels were constructed by deleting the initial 209 amino acid residues (ΔN209-hKv1.5), removing as much of the NH₂-terminus prior to the core membrane-spanning domains as possible, or the terminal 60 residues (ΔC60-hKv1.5), as indicated in Fig. 1A. After transient expression of these mutant constructs in HEK cells, the kinetics were tested using a steady-state inactivation protocol. Cells were given an initial pulse (P1) to a range of potentials up to +80 mV for 10 s and then a second pulse (P2) to +60 mV to measure residual current. Raw data from ΔN209-hKv1.5 are shown in Fig. 1B and steady-state voltage dependence of inactivation is shown for both the ΔN209 and ΔC60 mutants along with wild-type (WT) channels in Fig. 1C. The currents from both N- and C-terminal mutant constructs inactivated in a very similar manner to the WT channels (e.g. Fig. 2A). Inactivation in ΔC60 was unchanged from the WT channel. The V_½ inactivation was -14.5 ± 1.4 mV (n = 6) with a slope, k, of 6.3 ± 0.8 mV. In the ΔN209, the amount of inactivation was unchanged from the WT channel, but the voltage dependence was shifted to more negative potentials with V_½ inactivation of -32.6 ± 1.4 mV, and slope factor of 4.0 ± 0.37 mV (n = 7). From these results we conclude that while the amino terminus may have a modulatory role in determining the voltage dependence of slow inactivation, neither intracellular domain can regulate the amplitude or presence of inactivation itself.

Figure 1. Inactivation of N-terminal and C-terminal deletion mutants of hKv1.5.

A, diagrammatic representation of hKv1.5 wild-type channels (WT), the N-terminal deletion of 209 residues (ΔN209), and the C-terminal deletion of 60 residues (ΔC60). B, protocol and original data from ΔN209 to measure steady-state inactivation of hKv1.5. Cells were held at -80 mV and pulsed for 10 s to between -60 and +80 mV in 10 mV steps (P1). Residual currents were measured during a subsequent test pulse (P2) to +60 mV for 400 ms. A 30 s interpulse interval at -80 mV allowed full recovery from inactivation. C, steady-state inactivation relations obtained using the protocol described in B. Test pulse currents were normalized to the largest available current after the prepulse. Data are from WT channels (•), from ΔC60 (▵) and ΔN209 (^). The lines are fits to eqn (1) using the averaged V_½ and k from individual data sets. For WT (n = 5), ΔC60 (n = 6) and ΔN209 (n = 7), the V_½ values were -13.8 ± 0.2, -14.5 ± 1.4 and -32.6 ± 1.4 mV, respectively, and k values were 6.0 ± 0.7, 6.3 ± 0.8 and 4.0 ± 0.4 mV, respectively.

Figure 2. Inactivation of K⁺, Rb⁺, Cs⁺ and Na⁺ currents through hKv1.5 channels.

In each case only a single monovalent cation species was present in both bath (5 mM) and pipette (135 mM) solutions. A and B, K⁺ and Cs⁺ currents obtained using the two-pulse protocol described for Fig. 1. P1 pulses were for 10 s to between -60 and +80 mV in 10 mV steps, followed by a P2 pulse to +60 mV. C, K⁺, Rb⁺, Cs⁺ and Na⁺ currents during P1 pulses to +60 mV, normalized to peak values to compare the degree of inactivation. For the Na⁺ current tracing, the time base has been expanded and runs from 0 to 1 s. D, steady-state inactivation relations obtained using the protocol described in A and B. Test pulse currents were normalized to the largest available current after the prepulse. The lines are fits to eqn (1) using the averaged V_½ and k from individual data sets. The V_½ values for K⁺, Rb⁺ and Cs⁺ were -13.8 ± 0.2, -10.4 ± 1.2 and -10.3 ± 0.6 mV, respectively, and k values were 6.0 ± 0.7, 5.3 ± 0.6 and 9.6 ± 3.3 mV, respectively. Data points in each relation were means ±s.e.m. of 4-5 experiments.

Permeant ion species determines the kinetics of biexponential inactivation

In the experiments described below we have used the WT channel and manipulated the cation species and concentration on both sides of the cell membrane in order to define how intracellular and extracellular cations can modulate slow inactivation. With standard 135 mM ${\text{X}}_{\text{i}}^{\text{+}}$-5 mM ${\text{X}}_{\text{o}}^{\text{+}}$ cation gradients, K⁺, Rb⁺ and Cs⁺ currents through hKv1.5 channels inactivated differently as shown in Fig. 2A-B. Currents were recorded for up to 10 s during P1 pulses at each depolarizing potential, and then the membrane potential was changed to +60 mV for a 400 ms P2 pulse to assess the residual current. The inactivation was more obvious at the larger depolarizing voltages, and was incomplete, reaching a relative steady state at the end of 10 s pulses. It was apparent that inactivation was less complete when K⁺ was replaced by Rb⁺ and Cs⁺. To compare the inactivation of K⁺, Rb⁺, Cs⁺ and Na⁺ currents, peak current traces at +60 mV have been normalized to the same peak level (Fig. 2C). It appeared that it was an early substantial phase of current inactivation that was prominent with K⁺, and was much reduced in Rb⁺ and Cs⁺ solutions, and that a second slower phase of inactivation was less affected. Additional experiments using Na⁺ as the charge carrying cation revealed extremely rapid and complete inactivation (Fig. 2C). The voltage dependence of steady-state inactivation of hKv1.5 channels under different permeating ion conditions was determined by plotting the normalized peak current during +60 mV P2 pulses against P1 pulse potentials. From Boltzmann fits (eqn (1)), the averaged half-inactivating voltages (V_½) were -13.8 ± 0.2, -10.4 ± 1.2 and -10.3 ± 0.6 mV and slope factors (k) were -6.0 ± 0.7, -5.4 ± 0.6 and -9.7 ± 3.4 for K⁺, Rb⁺ or Cs⁺ current, respectively. Neither parameter varied significantly. These values were obtained from averages of fits to individual data sets. The curves in Fig. 2D were then generated from these averaged parameters and provide a good fit to averaged steady-state inactivation data. It was apparent that at the end of 10 s depolarizations, much less inactivation resulted in the presence of Cs⁺ than K⁺, and that this effect was not particularly potential dependent. It appeared that the larger crystal radius cation was more effective at impeding inactivation than K⁺. Rb⁺ data were much like K⁺ data, but it was the marked difference between Cs⁺ and K⁺ inactivation that we have used in later experiments aimed at separating intracellular and extracellular regulation of inactivation (Figs 4–8).

Figure 4. Effects of elevation of ${\text{K}}_{\text{o}}^{\text{+}}$ on the amplitude of slow inactivation.

A-C, three examples from cells expressing hKv1.5 at high density (A), at lower density (B) and data from an outside-out patch recording (C). The left-hand panels show original data obtained at 5 and 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$ and the right-hand panels show data after normalization to peak current values. Voltage steps were from -80 to +80 mV.

Figure 8. Effect of changes in the permeant ion concentration on the inactivation rate.

A, pipette contained 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$, and bath contained 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$. Predicted reversal potential was ≈+58 mV. Cell was held at -80 mV and pulsed to +120 mV, to +40 mV and then to +60 mV, in each case for 10 s with 30 s between pulses. B, pipette contained 5 mM ${\text{K}}_{\text{i}}^{\text{+}}$, and bath contained 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$. Predicted reversal potential was ≈+85 mV. Cell was held at -80 mV and pulsed to +100 and then +40 mV for 3 s with 120 s between pulses. C, fast time constants (τ₂) of current inactivation at +70 mV at different K⁺ concentrations in pipette and 135 mM K⁺ in the bath; n = 6-9 at each concentration, except for 5 mM ${\text{K}}_{\text{i}}^{\text{+}}$, when n = 2. Continuous line through data points represents fit to a Hill equation, K_d = 37.6 mM. Dashed line is symmetrical ${\text{K}}_{\text{o}}^{\text{+}}$-${\text{K}}_{\text{i}}^{\text{+}}$ fit from Fig. 7B. D, as for C but with different Cs⁺ concentrations in pipette and 135 mM Cs⁺ in the bath; n = 5-9 at each concentration. Data have not been fitted with a Hill equation, but dashed line is symmetrical ${\text{Cs}}_{\text{o}}^{\text{+}}$-${\text{Cs}}_{\text{i}}^{\text{+}}$ fit from Fig. 7B.

The components of inactivation over a 10 s depolarization varied significantly under the different ionic conditions. K⁺ and Rb⁺ current inactivation during 10 s pulses and Cs⁺ inactivation over 20 s were well fitted using a double exponential function (eqn (2)) and the resulting amplitudes corresponding to each component have been plotted as a fraction of total current amplitude at the start of the fit (Fig. 3). As stated above, the quasi-steady-state current component was strongly ion species dependent, with a larger residual current component present in Cs⁺, compared with Rb⁺ and then K⁺. As a proportion of total current at 10 s residual currents were 0.38 ± 0.02, 0.53 ± 0.03 and 0.70 ± 0.02 with K⁺, Rb⁺ and Cs⁺, respectively, at +80 mV. The main cause of this increased steady-state component was an ion-dependent decrease in the amplitude of the fast inactivating component (A₂). The fast inactivation component comprised between 25 and 35% of the total K⁺ current amplitude at 10 s, but only ∼5% of the total Cs⁺ current. The slower inactivation process (A₁) did not show as marked ion dependence, but there was still less Cs⁺-dependent than K⁺-dependent inactivation. A₁ values were significantly different between K⁺ and Cs⁺ at potentials >+50 mV. Fast (τ₂) time constants of inactivation were similar in Cs⁺, Rb⁺ and K⁺ at 446 ± 40, 587 ± 37 and 568 ± 37 ms, respectively (n = 5, 9 and 4, respectively), at +80 mV and showed little voltage dependence (data not shown).

Figure 3. Three components of inactivating K⁺, Rb⁺ and Cs⁺ currents.

Currents during 10 s depolarizations at different potentials (e.g. Fig. 2AB and ) were fitted with a double exponential function (eqn (2)). Corresponding fast (A), slow (B) and steady-state (C) inactivating current amplitudes were expressed as a fraction of the total current. Using ANOVA, for A₂, all three data sets were significantly different from each other (P < 0.05) at all potentials. The A₁ and non-inactivating component amplitudes were significantly different from each other at voltages positive to +50 and +30 mV, respectively (P < 0.05). Data for each bar were from 4-6 cells.

Slow inactivation is relatively insensitive to extracellular ion concentration

Extracellular cations are usually strong modulators of the rate of C-type inactivation in ShakerK⁺ channels (Lopez-Barneo et al. 1993; Baukrowitz & Yellen, 1995). In hKv1.5 there were relatively small changes in inactivation in response to increases in the extracellular cation concentration. In these experiments ${\text{K}}_{\text{i}}^{\text{+}}$ (or ${\text{Cs}}_{\text{i}}^{\text{+}}$) was kept constant at 135 mM and the external cation concentration was varied. We attempted to prevent extracellular ion accumulation as a result of outward ion flux through channels (Baukrowitz & Yellen, 1995), by using rapid solution changes and examining currents from cells expressing hKv1.5 at widely different levels. Three examples are shown in Fig. 4 when ${\text{K}}_{\text{o}}^{\text{+}}$ was changed from 5 to 135 mM (in Fig. 4B change was from 135 to 5 mM). In each case elevation of ${\text{K}}_{\text{o}}^{\text{+}}$ reduced current amplitude due to a reduction in driving force and reduced the amount of inactivation, so that when currents were normalized to peak values the reduction of inactivation in 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$ was small but consistent. The examples in Fig. 4 represent the largest changes in normalized current generally observed. Data from whole-cell (A and B) and outside-out patch (C) recordings were very similar, suggesting that bulk ${\text{K}}_{\text{o}}^{\text{+}}$ accumulation was not a significant problem. For 5 s duration pulses the mean amplitude of the A₂ component of inactivation was 0.13 ± 0.02 in 5 mM ${\text{K}}_{\text{o}}^{\text{+}}$ and 0.09 ± 0.015 for 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$.

As shown in Fig. 4, elevation of ${\text{K}}_{\text{o}}^{\text{+}}$ from 5 to 135 mM tended to reduce the magnitude of inactivation, but further elevation of ${\text{K}}_{\text{o}}^{\text{+}}$ to 300 mM actually accelerated inactivation to a degree (Fig. 5A). The significance of the 300 mM ${\text{K}}_{\text{o}}^{\text{+}}$ data is less certain, since cell shrinkage tended to occur on exposure to the high osmolarity solution. Still, it was clear that only limited slowing of inactivation occurred when ${\text{K}}_{\text{o}}^{\text{+}}$ was elevated to high levels. Currents inactivated less with ${\text{Cs}}_{\text{i}}^{\text{+}}$ and ${\text{Cs}}_{\text{o}}^{\text{+}}$ as shown in Fig. 2, but there was still only a small effect of elevating ${\text{Cs}}_{\text{o}}^{\text{+}}$ over the range 5-135 mM (35-135 mM in Fig. 5B). No significant differences in inactivation time constants were observed over a wide range of extracellular K⁺ concentrations (${\text{K}}_{\text{o}}^{\text{+}}$, Fig. 5C), and although a decrease in the amplitude of the A₂ component of inactivation was observed at 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$, this was reversed at the highest concentration of ${\text{K}}_{\text{o}}^{\text{+}}$ used (300 mM, Fig. 5D). Similar results were observed for mean Cs⁺ data (n = 10, not shown).

Figure 5. Modulation of inactivation by changes in the concentrations of external K⁺ and Cs⁺.

Traces were elicited by a 10 s step to +80 mV from a holding potential of -80 mV. A, currents were recorded from the same cell in 5, 135 and 300 mM ${\text{K}}_{\text{o}}^{\text{+}}$ at constant 135 mM ${\text{K}}_{\text{i}}^{\text{+}}$. B, current from another cell recorded in 35, 70 and 135 mM ${\text{Cs}}_{\text{o}}^{\text{+}}$ at constant 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$. In both A and B, traces have been normalized to the peak current value. C and D, summary data for the modulation of inactivation kinetics in hKv1.5 by ${\text{K}}_{\text{o}}^{\text{+}}$. In all cases the intracellular K⁺ concentration was constant at 135 mM. The current decay was fitted using a double exponential eqn (2). Data are means ±s.e.m. (n = 7-14). There were no significant differences between K⁺ concentrations for τ data, or the amplitude data.

Changes in the extracellular cation species with constant ${\text{K}}_{\text{i}}^{\text{+}}$ or ${\text{Cs}}_{\text{i}}^{\text{+}}$

Although the extracellular cation concentration did not appear to modulate slow inactivation in hKv1.5 very strongly, it was obvious from data in Figs 2 and 5 that inactivation with Cs⁺ as the cation was much reduced compared with K⁺. We therefore tested the effects of different external cation species while keeping ${\text{K}}_{\text{i}}^{\text{+}}$ or ${\text{Cs}}_{\text{i}}^{\text{+}}$ constant at 135 mM (Fig. 6). In this situation we found that currents inactivated similarly despite the presence of different external cations, so that it appeared that inactivation was more closely defined by the species of intracellular ion. In the presence of ${\text{K}}_{\text{i}}^{\text{+}}$, currents adopted a K⁺-like inactivation amplitude and time course (see Fig. 2A). In Fig. 6A and B we illustrate data from two cells where either ${\text{Cs}}_{\text{i}}^{\text{+}}$ or ${\text{K}}_{\text{i}}^{\text{+}}$ was constant at 135 mM and the external cation species was varied in sequence, from NMG⁺ to Cs⁺ to K⁺ to Na⁺. Apart from some slowing of inactivation when ${\text{Na}}_{\text{o}}^{\text{+}}$ was used, only relatively small effects of ion species were observed, and the major determinant of inactivation was the intracellular ion species, so that with 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$ (Fig. 6A) outward currents showed the decreased inactivation typically seen with Cs⁺ (Fig. 2B). Conversely, with ${\text{K}}_{\text{i}}^{\text{+}}$, outward currents showed marked inactivation which was independent of the extracellular cation species (Fig. 6B). Cells were fully washed in each external solution before measurements were made to prevent contamination by other ions, and exposed in different sequences to ensure reproducibility of data. Mean data for the non-inactivating outward current fraction at the end of 10 s depolarizations confirmed that there was little difference in the amount of inactivation in the different external solutions (Fig. 6C). More importantly, it was clear that all outward currents recorded with ${\text{Cs}}_{\text{i}}^{\text{+}}$ inactivated less than with ${\text{K}}_{\text{i}}^{\text{+}}$, independently of the external cation type.

Figure 6. Effects of changes in the external ion species in the presence of constant ${\text{K}}_{\text{i}}^{\text{+}}$ or ${\text{Cs}}_{\text{i}}^{\text{+}}$.

A, pipette contained 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$, and external solution was changed in sequence to NMG⁺, then Cs⁺, then K⁺, each at 135 mM. Cell was held at -80 mV and pulsed to +60 mV for 10 s. Note transient inward tail current in ${\text{K}}_{\text{o}}^{\text{+}}$. B, as for A except that pipette contained 135 mM ${\text{K}}_{\text{i}}^{\text{+}}$ and external cation sequence was NMG⁺, then K⁺, then Cs⁺, then Na⁺. C, cumulative data on non-inactivating current at 10 s with 135 mM different external cation species, with constant 135 mM ${\text{K}}_{\text{i}}^{\text{+}}$ (n = 6, 3 for NMG), or ${\text{Cs}}_{\text{i}}^{\text{+}}$ (n = 4) as indicated. All data with ${\text{Cs}}_{\text{i}}^{\text{+}}$ were significantly different from data with ${\text{K}}_{\text{i}}^{\text{+}}$ as indicated by the asterisks, but within the ${\text{K}}_{\text{i}}^{\text{+}}$ group, only ${\text{Na}}_{\text{o}}^{\text{+}}$ data were significantly different from other external cations (all pairwise multiple comparisons, P < 0.05).

The permeating ion determines the inactivation time course

Bulk extracellular cation concentration and species are not very important factors in the regulation of inactivation, but dominant modulation by the intracellular cation species may reflect internal cation ‘leak’ or flux from the intracellular side of the channel during outward ionic currents. Such a flux could saturate all regulatory sites throughout the pore, both intracellular and extracellular, and prevent any modulatory action of extracellular cations (Baukrowitz & Yellen, 1995). The results do suggest that regulation occurs within, or close to the pore, otherwise the bulk extracellular cation concentration would have been expected to have effects.

Symmetrical K⁺ or Cs⁺ ion concentrations were varied on both sides of the pore (Fig. 7) to characterize the concentration dependence of ion modulation of pore-dependent inactivation. When the ${\text{K}}_{\text{i}}^{\text{+}}$ and ${\text{K}}_{\text{o}}^{\text{+}}$ concentrations were reduced to 5 mM, inactivation was complete and much faster, as shown by the current-voltage data (Fig. 7A). Bi-exponential fits gave a time constant for the fast component of inactivation of 34.3 ms at +70 mV with a non-inactivating fraction of 0.056. The increased rate of inactivation was caused by acceleration of both the fast (τ₂, Fig. 7B) and slower (τ₁, Fig. 7C) components of inactivation. At K⁺ concentrations of < 30 mM inactivation became much more rapid and complete, and τ₂ data were fitted with a Hill equation (continuous line) which gave a K_d of 34.7 mM (Fig. 7B). This acceleration of τ₂ was accompanied by increases in τ₁ (Fig. 7C) and a decreased proportion of non-inactivating current (Fig. 7D). With ${\text{Cs}}_{\text{i}}^{\text{+}}$ and ${\text{Cs}}_{\text{o}}^{\text{+}}$, an increase in the amplitude and decrease in the time constant of fast inactivation was also observed as Cs⁺ was lowered (Fig. 7B), and data were fitted with a K_d of 20.5 mM. The slope of the relationship was shallower for Cs⁺ than for K⁺, with a Hill coefficient of 0.9. The slow time constant (τ₁) was also accelerated, but channels in which Cs⁺ was permeating were more resistant to inactivation, even down to quite low concentrations of ${\text{Cs}}_{\text{i}}^{\text{+}}$ and ${\text{Cs}}_{\text{o}}^{\text{+}}$. This is shown in the graph illustrating the relative amplitude of non-inactivating current (Fig. 7D). These data were obtained during 7 s depolarizations, and even at 1 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$ and ${\text{Cs}}_{\text{o}}^{\text{+}}$ (not plotted in Fig. 7D) this component of current remained at 0.38 ± 0.06 (n = 5) of peak current.

Separation of intracellular cation effects on inactivation

Changes in symmetrical ion concentrations can have large effects on the components of inactivation, whereas variations of extracellular ion concentration and species only had small effects on inactivation (Figs 4–6). We have used the difference in inactivation rates and permeabilities of ${\text{K}}_{\text{i}}^{\text{+}}$${\text{Cs}}_{\text{i}}^{\text{+}}$ and ${\text{K}}_{\text{i}}^{\text{+}}$ shown in Figs 2 and 6 to test the hypothesis that intracellular cations provide important modulation of inactivation in this channel.

In the experiment shown in Fig. 8A, with 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$ and 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$, the predicted reversal potential was ∼+58 mV given a Cs⁺:K⁺ permeability ratio of ∼0.1 (Chen et al. 1997). For the pulses to +120 or +60 mV, slowly inactivating outward Cs⁺ currents were observed. In the same cell, a pulse to +40 mV gave a more rapidly inactivating inward current, presumably carried by K⁺. The inward and outward currents are of almost identical amplitude so the data clearly demonstrate that the nature of the permeating ion, rather than that present in the extracellular or intracellular medium, or the rate of the ion flux, was the best determinant of the rate and amplitude of inactivation in hKv1.5. Although the ion species within the pore can now be seen to be important in the time course of inactivation, the surprising observation here is that the inward K⁺ current inactivates so rapidly and completely, given 135 mM K⁺ at the external mouth of the channel.

A variation of the experiment in Fig. 8A is shown in Fig. 8B. Here, in a different cell, the ${\text{Cs}}_{\text{i}}^{\text{+}}$ was replaced by 5 mM ${\text{K}}_{\text{i}}^{\text{+}}$ which resulted in a reversal of the physiological ${\text{K}}_{\text{i}}^{\text{+}}$-${\text{K}}_{\text{o}}^{\text{+}}$ gradient so that the reversal potential was ∼+85 mV. At both +100 and +40 mV, outward and inward currents inactivated more rapidly and completely than inward K⁺ current in Fig. 8A. The mean fast inactivation time constant (τ₂) at +100 mV was 85.2 ± 13 ms (n = 4), and at +40 mV the time constant was 109 ± 6 ms (n = 10). The current inactivated to within 15% of baseline over the short pulse duration. The main point to be made from these data is that reduction of the intracellular K⁺ concentration by substitution with ${\text{Cs}}_{\text{i}}^{\text{+}}$ (Fig. 8A), or directly to 5 mM ${\text{K}}_{\text{i}}^{\text{+}}$ (Fig. 8B), produced far more potent changes in inactivation than changes in the extracellular cation concentration or species (Figs 4–6). The concentration dependence of the ${\text{K}}_{\text{i}}^{\text{+}}$ and ${\text{Cs}}_{\text{i}}^{\text{+}}$ effects on the rate of fast inactivation are shown in Fig. 8C and D. For variations in ${\text{K}}_{\text{i}}^{\text{+}}$, with a constant 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$, the data, plotted as columns, form a sigmoid relation that has been fitted with a Hill equation (continuous line) to give a K_d of 43.4 mM (Fig. 8C). For comparison, the fit to the symmetrical ${\text{K}}_{\text{i}}^{\text{+}}$-${\text{K}}_{\text{o}}^{\text{+}}$ data from Fig. 7B has been plotted as the dashed line. When ${\text{Cs}}_{\text{i}}^{\text{+}}$ was changed in isolation at constant 135 mM ${\text{Cs}}_{\text{o}}^{\text{+}}$ (open columns in Fig. 8D), the effects on the inactivation rate were very similar to the effects of symmetrical changes in ${\text{Cs}}_{\text{i}}^{\text{+}}$-${\text{Cs}}_{\text{o}}^{\text{+}}$ (dashed line replotted from Fig. 7B). As ${\text{Cs}}_{\text{i}}^{\text{+}}$ was lowered below 30 mM, the positive potential shift in E_Cs (Cs⁺ equilibrium potential) combined with its low permeability made measurements of inactivating Cs⁺ currents unreliable. For this reason, no data are shown for ${\text{Cs}}_{\text{i}}^{\text{+}}$ less than 17 mM.

The data in Fig. 8 not only strongly point to the importance of intracellular cations as the dominant regulator of inactivation in this channel, but also demonstrate that acceleration of inactivation could be observed when intracellular cation concentrations alone were reduced. When the effects of the different ions were compared, again it appeared that although ${\text{Cs}}_{\text{i}}^{\text{+}}$ underwent qualitatively similar regulation to ${\text{K}}_{\text{i}}^{\text{+}}$, permeating Cs⁺ ions (Fig. 8A and D) tended to cause a relative destabilization of the inactivated state such that channels always inactivated more slowly and less completely than when K⁺ ions permeated the channels (Fig. 8B and C). This is an effect that closely parallels that seen with symmetrical ion concentrations (Fig. 7), and was noted earlier (Fig. 2). At very low concentrations of intracellular K⁺ or Cs⁺ it appeared that the relationships between the two ions were converging (Figs 7 and 8). Unfortunately, data at Cs⁺ concentrations in the range 1-5 mM were difficult to obtain and fit with confidence.

Properties of the point mutants R487Y and R487V

In many Shaker channels, mutations of T449 or the equivalent site to Y or V increases the time constant for C-type inactivation to > 10 s (Lopez-Barneo et al. 1993), whereas the positively charged residue K gives a rapid C-type inactivation (Schlief et al. 1996). Kv1.5 shows an intermediate rate of pore-dependent inactivation with R487 at the equivalent site, and is only mildly sensitive to extracellular TEA at concentrations up to 50 mM (Fedida et al. 1993; Snyders et al. 1993; Grissmer et al. 1994). We used the expected TEA sensitivity of the Y residue at the site to confirm the success of the prediction. In Fig. 9A, control currents are shown from R487Y-hKv1.5 and the first four current tracings after switching to a bath solution containing 5 mM TEA. Despite the appearance of marked TEA sensitivity, currents inactivated in a very similar manner to the WT channel. Rapid current block was observed (traces 1-2), and normalization of the control and trace 2 resulted in an identical time course of inactivation (data not shown). In Fig. 9B, TEA data at 1 and 2.5 mM have been compared with the control current tracing, by normalization to their respective peak current values. Despite an increased noise level there was little change in the time course of inactivation in TEA-blocked tracings. An example of data obtained using the steady-state inactivation protocol (Fig. 1B) is shown in Fig. 9C and the mean steady-state inactivation relations for R487Y are shown in Fig. 9D. The data illustrate that the time course, amplitude and voltage dependence of inactivation were very similar to that seen in WT channels (compare with Fig. 2A). When the valine mutation was introduced into hKv1.5 (R487V), the process of slow inactivation was again unaffected (data not shown).

Figure 9. Inactivation in the mutant channel R487Y.

A, effect of 5 mM extracellular TEACl. Control pulses to +60 mV from -80 mV were given at 0.1 Hz and TEA was added to the bath just prior to the recording of trace 1. Traces labelled 2 and steady-state were the next two current tracings recorded. B, normalized inactivating R487Y current in control and during exposure to 1 mM or 2.5 mM TEACl in the external bath. Peak currents before normalization were 2.3 nA in control, and 0.82 nA and 0.38 nA in 1 mM and 2.5 mM TEA, respectively. C, currents from cell transfected with R487Y recorded during two-pulse steady-state inactivation protocol. Holding potential was -80 mV and currents are illustrated for prepulses between -60 and +60 mV in 10 mV steps. Residual current was measured during the test pulse to +60 mV. D, steady-state inactivation relations obtained from same data as in C. Test pulse currents were normalized to the largest available current after the prepulse. The line was fitted to eqn (1) using the averaged V_½ (-23 mV) and k (5.4 mV) from individual data sets. Data points are means ±s.e.m. of 4 experiments.

Not only was the onset of slow inactivation unaffected in the R487Y mutant compared with the WT channel, but recovery was unaffected as well (Fig. 10). Recovery from inactivation was tested by giving a 5 s prepulse to +40 mV and then assessing current recovery at different intervals. Prepulses were given every 30 s to allow full recovery from inactivation between trials. Two examples of data obtained using this protocol are shown in Fig. 10A and B, for WT and the R487Y mutant, respectively. Currents during the prepulses superimposed in both cases to show uniformity of control data and recovery can be seen to occur in a similar manner in the two channel types at intervals up to 3 s. Mean data on recovery from inactivation in WT and R487Y channels are shown in Fig. 10C. No difference was observed between WT and R487Y channels. Data in both cases have been fitted to a single exponential recovery function with similar time constants of 1107 and 976 ms in WT and R487Y, respectively. These results suggested that slow inactivation in hKv1.5 was not strongly regulated by residues in the outer pore equivalent to those in Shaker channels. In addition, hKv1.5 recovery from inactivation was not affected by changing the cation concentration at the outer pore mouth. Extracellular K⁺ can modulate the recovery from C-type inactivation by interaction at this site in the external vestibule of the Kv1.3 channel (Levy & Deutsch, 1996) and ferret Kv1.4 (Rasmusson et al. 1995). We have tested recovery from inactivation in elevated ${\text{K}}_{\text{o}}^{\text{+}}$ (Fig. 10D-F). Control recovery in 5 mM ${\text{K}}_{\text{o}}^{\text{+}}$ is shown in Fig. 10D with plotted data in Fig. 10F. When ${\text{K}}_{\text{o}}^{\text{+}}$ was elevated to 135 mM in the same cell no obvious difference in the recovery from inactivation was noted (Fig. 10E). Recovery time constants of 1217 ± 104 ms in 5 mM ${\text{K}}_{\text{o}}^{\text{+}}$ and 1163 ± 54 ms in 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$ were observed in paired exposures on five cells. These values were not significantly different (P < 0.05).

Figure 10. Recovery from inactivation in WT and R487Y.

A 5 s prepulse to +40 mV from a holding potential of -80 mV was given to inactivate the channels, and current recovery was then assessed by giving short 60 ms test pulses at different intervals after repolarization from the prepulse. The intervals were from 180 ms to 4 s at increments of 400 ms. Pulses were given every 30 s to allow full recovery from inactivation between tests. A and B, original data from WT and R487Y channels. C, fractional recovery from inactivation. Mean test pulse data were normalized to preceding control peak current amplitudes and plotted as a function of the interpulse interval. Single exponential recoveries had time constants of 1107 ± 80 and 976 ± 41 ms (mean ±s.e.m., n = 8) in WT and R487Y, respectively. D and E, data from a single cell obtained using the same recovery protocol as in A, except that ${\text{K}}_{\text{o}}^{\text{+}}$ was changed from 5 mM (D) to 135 mM (E). F, mean data for recovery time course in 5 and 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$. Data were analysed as in C. Single exponential recoveries had time constants (τ_recovery) of 1217 ± 104 and 1163 ± 54 ms (mean ±s.e.m., n = 5) in 5 and 135 mM ${\text{K}}_{\text{o}}^{\text{+}}$, respectively.

DISCUSSION

Slow inactivation in Kv1.5

Data obtained from hKv1.5 suggest that the channel has a slow inactivation process limited to the membrane-spanning domains and the pore region of the channel, based on evidence from experiments with different sized permeant ions. Unlike for Kv2.1, extensive N- and C-terminal deletions did not modify the degree of inactivation (Fig. 1). The steady-state inactivation relationships of hKv1.5 showed strong voltage dependence, with V_½ inactivation of ∼-10 mV (Fig. 2). This is in contrast to the slight voltage dependence of inactivation time constants and related amplitudes (Fig. 3; Snyders et al. 1993). As for Shakermutant channels (Hoshi et al. 1991), this suggested that the slow inactivation process derived its voltage dependence not from its rate constants, but from the voltage dependence of channel opening. Other evidence also supported inactivation occurring predominantly from the open state, or kinetically adjacent closed states: (1) the effect of permeating ions on the voltage dependence of steady-state inactivation curves was negligible (Fig. 2D), despite effects on the amplitude of inactivation, and (2) inactivation voltages overlapped with activation potentials, which suggests that inactivation was coupled to states close to channel opening (DeCoursey, 1990). Inactivation in hKv1.5 was bi-exponential with the fast component of inactivation (A₂) comprising up to 40-50% of the total current at more positive potentials. The amplitudes of both fast and slow components of inactivation were strongly affected by the permeating cation species and by changes in symmetrical and intracellular ion concentrations.

Modulation of inactivation and recovery by extracellular cation species and concentration

In other channels like ShB, Kv1.3 and Kv1.4, most data have suggested that the species and concentration of extracellular cations can strongly modify the rate of C-type inactivation (Pardo et al. 1992; Lopez-Barneo et al. 1993; Marom & Levitan, 1994; Kukuljan et al. 1995) and the recovery from inactivation (Levy & Deutsch, 1996). In ShBΔ-T449K, 200 mM external Rb⁺ and K⁺ showed similar rates and degrees of inactivation, whereas 200 mM ${\text{Cs}}_{\text{o}}^{\text{+}}$ and NMG⁺ showed accelerated, more complete inactivation (Lopez-Barneo et al. 1993). It was suggested that equivalent concentrations of extracellular ions like Cs⁺, NH₄⁺ or Na⁺ were less able to delay inactivation at the outer pore mouth than readily permeant K⁺ or Rb⁺. In these experiments only the extracellular cation concentration and species were elevated. Not all studies have observed strong modulation of slow inactivation in K⁺ channels when ${\text{K}}_{\text{o}}^{\text{+}}$ is changed. Baukrowitz & Yellen (1995) observed that in the absence of N-type inactivation, C-type inactivation in ShΔ was slow and rather insensitive to ${\text{K}}_{\text{o}}^{\text{+}}$, as also seen recently in ShBΔ (Yang et al. 1997). In Kv2.1, neither raised ${\text{K}}_{\text{o}}^{\text{+}}$ nor extracellular TEA slowed inactivation (Klemic et al. 1998). Most studies agree, though, that mutation of the specific T449 in the outer pore mouth of Shaker can modulate these kinetics significantly, which is interpreted in terms of a regulatory site in the outer mouth of the pore at which K⁺ has a longer dwell time than other ions. Modulation at such a site can also explain the actions of elevated ${\text{K}}_{\text{o}}^{\text{+}}$ rather than ${\text{Cs}}_{\text{o}}^{\text{+}}$ on recovery from inactivation (Levy & Deutsch, 1996). A second important effect of raising ${\text{K}}_{\text{o}}^{\text{+}}$ in Shaker channels is that current amplitude increases despite a reduction in driving force. This has been interpreted as reduced closed-state inactivation (Lopez-Barneo et al. 1993) or some other mechanism by which the number of activatable channels is increased. This action will accentuate any slowing of inactivation when currents obtained in different ${\text{K}}_{\text{o}}^{\text{+}}$ concentrations are normalized.

In hKv1.5, the inactivation time course was slowed by elevation of ${\text{K}}_{\text{o}}^{\text{+}}$, but the effect was small when currents were normalized to peak outward current level. An important question that must be addressed is the bulk extracellular ion accumulation from the large currents generated in transfected cells. When channels are expressed in oocytes or mammalian cells, the high expression levels can lead to accumulation of ions flowing through channels at the extracellular mouth of the channels or even to bulk increases in extracellular cation concentration (Baukrowitz & Yellen, 1995; Yang et al. 1997). We do not think that this can explain our results for two reasons. (1) We illustrated data from cells with high and low expression levels and current densities, and also outside-out patch data with small currents (Fig. 4). In all these situations, there was only minor modulation of inactivation by changes in ${\text{K}}_{\text{o}}^{\text{+}}$. (2) When ${\text{Cs}}_{\text{i}}^{\text{+}}$ was the permeant cation, channel permeability was < 10% of that of K⁺, and thus significant ${\text{Cs}}_{\text{o}}^{\text{+}}$ accumulation was unlikely. Still, even with ${\text{Cs}}_{\text{i}}^{\text{+}}$, elevation of ${\text{Cs}}_{\text{o}}^{\text{+}}$ had only small effects on inactivation (Fig. 5B).

In hKv1.5 there was also little change in the time constants of recovery from inactivation when ${\text{K}}_{\text{o}}^{\text{+}}$ was elevated from 5 to 135 mM. In addition, alterations of the external cation species only produced minor modification of the inactivation rate, which was principally determined by the intracellular cation species, ${\text{Cs}}_{\text{i}}^{\text{+}}$ or ${\text{K}}_{\text{i}}^{\text{+}}$. This seems an entirely reasonable result because when outward currents flow they are likely to occupy cation binding sites throughout the pore and shield any such sites from the action of extracellular cations, as suggested by Baukrowitz & Yellen, (1995).

Permeating ion regulation of inactivation in hKv1.5

The experiments discussed above strongly support the idea that the permeating ion regulated the inactivation amplitude by saturation of cation binding sites within the permeation pathway. Convincing support for this idea was provided by the experiment shown in Fig. 8A. Here, with 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$/135 mM ${\text{K}}_{\text{o}}^{\text{+}}$, a reversal potential close to +60 mV provided the opportunity to observe inactivating currents carried by both ions in different directions through the pore. The result was clear, Cs⁺ currents inactivated much less than K⁺ currents, despite currents of similar magnitude (∼2.5 nA).

Both Cs⁺ and Rb⁺ reduced inactivation amplitude over 10 s compared with K⁺ and Na⁺. The averaged V_½ (∼-10 to -15 mV) and slope factors (∼-7 mV) for K⁺, Rb⁺ and Cs⁺ varied only slightly, but at the quasi-steady state, the non-inactivating proportion of current was increased from < 40% of the total (in K⁺) to > 70% (in Cs⁺), principally due to a reduction in the amplitude of the fast component of inactivation. At +80 mV only 5% of the inactivation in Cs⁺ was attributable to fast inactivation, whereas this amounted to more than 30% in K⁺ (Fig. 3). This had the result of leaving a much larger non-inactivating current in Cs⁺ conditions at the end of 10 s pulses. When results for the four ions were compared (Fig. 2C), it was clear that the larger cations inactivated less, and that once channels were open the effects were not voltage dependent (Fig. 2D). This suggests destabilization of the inactivated state with larger cations, and an inverse relationship between ion crystal radius and the amplitude of inactivation. The experiments suggest an important regulation of inactivation within the permeation pathway and suggest that inactivation in hKv1.5 is of a ‘pore-type’.

Intracellular regulation of slow inactivation

The possibility of intracellular regulation of slow inactivation has not been widely considered in different channels, and indeed it can be difficult to obtain unequivocal evidence for intracellular regulation, as there must always be ions permeating the channel to observe currents. A number of results that we have obtained suggest that intracellular regulation is an important mechanism in this channel. It was noted in Fig. 8A that when 135 mM ${\text{Cs}}_{\text{i}}^{\text{+}}$-135 mM ${\text{K}}_{\text{o}}^{\text{+}}$ solutions were used, slowed inactivation might be expected with an elevated ${\text{K}}_{\text{o}}^{\text{+}}$ when K⁺ permeated the channel. However, the inward K⁺ current at +40 mV inactivated rapidly and more completely than expected, to only ∼15% of the peak current level (cf. Fig. 2). One explanation for this could be that the local effective intracellular K⁺ concentration was low. Similarly we observed that with the normal K⁺ gradient reversed (5 mM ${\text{K}}_{\text{i}}^{\text{+}}$-135 mM ${\text{K}}_{\text{o}}^{\text{+}}$), inactivation was further accelerated and more complete. This would not be expected if an extracellular cation binding site was the primary determinant of the inactivation rate. The inactivation appeared to conform to that seen when low symmetrical concentrations of ${\text{K}}_{\text{i}}^{\text{+}}$ and ${\text{K}}_{\text{o}}^{\text{+}}$ were tested and therefore suggested that in hKv1.5 different intracellular cation concentrations and species could strongly modulate the rate and amplitude of inactivation. For both Cs⁺ and K⁺ the acceleration of inactivation observed with different symmetrical ion concentrations (Fig. 7B) could be reproduced when only the intracellular concentration was lowered and the external ion concentration was maintained at 135 mM (Fig. 8C and D). We believe that this is strong evidence, in hKv1.5 at least, that intracellular cation access to a specific binding site is the dominant regulator of the fast component of inactivation.

Inactivation and recovery from inactivation in the mutants R487Y and R487V

In many K⁺ channels a regulatory site for external cations has been shown to lie in the outer pore mouth. The specific residues within the K⁺ channel pore that mediate C-type and P-type inactivation in other channels are R487 and A501 (T449 and A463 in ShBΔ), and V476 (I369 in Kv2.1). In hKv1.5, both A501 and V476 are already the residues that would be expected to yield the least inactivation, and so we concentrated on the widely studied T449 site, which also comprises part of the external TEA binding site (MacKinnon & Yellen, 1990; Heginbotham & MacKinnon, 1992). Mutations to Y and V slow inactivation greatly in ShBΔ (Lopez-Barneo et al. 1993; but see Schlief et al. 1996), and, as an added property, the Y residue confers increased sensitivity to extracellular TEA (Heginbotham & MacKinnon, 1992). We observed that the mutation R487Y did confer high TEA sensitivity to the normally TEA-insensitive hKv1.5 (Fedida et al. 1993; Snyders et al. 1993) with > 70% block at 1 mM extracellular TEA (Fig. 9A). This provided important confirmation of the adequacy of our mutation, but two important features of slow inactivation were not altered. The rate of inactivation in TEA-blocked channels was not appreciably slowed, and the inactivation kinetics of unblocked R487Y channels were not significantly different from WT channels. This result suggested that, in hKv1.5, TEA block and slow inactivation could be completely dissociated, which has also been suggested recently in ShB (Molina et al. 1997). Not only was the onset of inactivation unaffected in the mutant, but recovery from inactivation could not be distinguished from that in the WT channel (Fig. 10). These data are strong evidence that the residues in the outer pore mouth of hKv1.5 may not determine inactivation in the same way as in Shaker mutant channels.

Conclusions

Here we have analysed the properties and ion dependence of slow inactivation in the human channel hKv1.5. We suggest the following conclusions. (1) Ions with a larger crystal radius impede inactivation when they are permeating the channel. (2) Extracellular cation concentration and species only partially regulate the amplitude and the rate of inactivation in this particular channel; the rates and amplitudes of outward current inactivation are more determined by the species of intracellular cation. (3) More specific localization of inactivation by point mutations of a residue in the outer pore mouth reveal that hKv1.5 is not regulated, although sensitivity to extracellular TEA could be conferred. (4) Since the permeation properties of different ions are controlled at the selectivity filter of the channel, the possibility exists that the sites at which ion concentration and species modulate inactivation lie at or close to this filter, and are more accessible from the intracellular mouth of the pore. Further studies will be required to pinpoint the specific intracellular or extracellular domains that mediate the particular form of ‘pore-type’ inactivation in this channel.