Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

The structures of quinidine and 4AP are shown in figure 1. They share the general feature of having a hydrophobic aromatic moiety (quinoline in quinidine and pyridine in 4AP) and a tertiary amine. The two are markedly different in size, with quinidine being much bulkier than 4AP, especially in the quinuclidine structure surrounding the tertiary amine. Their effects on native K channels have been extensively characterized (Fedida et al., 1993; Campbell et al., 1993; Furukawa et al., 1989; Roden et al., 1988; Balser et al., 1991; Imaizumi and Giles, 1987). Studies using various K channel clones further provide structural information about their actions (Tseng et al., 1996a; Yao and Tseng, 1994; Kirsch and Drewe, 1993; Yao et al., 1996; Tseng et al., 1996b; Yatani et al., 1993; Yeola et al., 1996). In a few cases, quinidine and 4AP binding preferentially occurs in the "rested" state and channel activation induces drug dissociation (Campbell et al., 1993; Tseng et al., 1996a; Roden et al., 1988; Yao et al., 1996). This type of drug-channel interaction will not be discussed here. The focus of our study is on a more general type of drug-channel interaction in which drug binding is facilitated by channel activation.

Studies on the effects of quinidine and 4AP on voltage-gated K channel clones indicate that both agents block the channels from the intracellular side of the membrane and, in most cases, binding is facilitated by channel activation. For channels that have a fast N-type inactivation process (Yao and Tseng, 1994; Yatani et al., 1993), binding of quinidine or 4AP and the closure of the inactivation gate (occlusion of the inner mouth by the N-terminal domain) interfere with each other. This points to the inner mouth region of the pore as a candidate for binding sites of both quinidine and 4AP. Indeed, site-directed mutagenesis of several K channel clones reveal important roles of residues lining the putative inner mouth region in determining the blocking potency of both quinidine (Zhang et al., 1995; Yeola et al., 1996) and 4AP (Shieh and Kirsch, 1994). However, there are important differences between quinidine and 4AP in the voltage- and time-dependencies of action (see "Results"). These differences raise two questions: 1) Do the binding sites of quinidine and 4AP overlap with each other? 2) What controls the binding site affinity and the kinetics of drug-channel interaction?

We used a K channel clone, rKv1.4, as a model to compare the actions of quinidine and 4AP. To avoid interference from the fast N-type inactivation process in the measurement of drug effects, a deletion mutant (rKv1.43-25) was used in which the N-type inactivation process was disrupted (Tseng-Crank et al., 1993). In search for a common or overlapping site involved in quinidine and 4AP binding, our focus was on the sixth transmembrane (S6) domain of rKv1.4. It has been widely recognized that the S6 domains of Na, Ca and K channels play important roles in channel gating (Hoshi et al., 1991; Zuhlke et al., 1994; Zhang et al., 1994; McPhee et al., 1995) and in drug binding (Yeola et al., 1996; Baukrowitz and Yellen, 1996; Hockerman et al., 1995; Ragsdale et al., 1996). With respect to the role of S6 in drug binding, a general theme has emerged from studies on Na, Ca and K channels: drug binding is stabilized by hydrophobic interactions between drug molecules and hydrophobic residues in the S6 domains (Yeola et al., 1996; Baukrowitz and Yellen, 1996; Hockerman et al., 1995; Ragsdale et al., 1996). For voltage-gated K channels, one position in the S6 domain seems to play an especially important role in drug binding: residue at the 13th position in the S6 sequence. In rKv1.43-25 this is threonine at position 529, T529 (fig. 1). The equivalent residue in the Shaker channel, T469, has been shown to play a key role in determining the blocking potency of a series of pore blockers (Baukrowitz and Yellen, 1996). The equivalent position in hKv1.5, T505, also affects the binding affinity of quinidine and bupivacaine (Vicente et al., 1997; Yeola et al., 1996). According to the current model of voltage-gated K channels (Durell and Guy, 1996), the S6 domain consists of two -helices due to a proline residue(s) in this region (fig. 1). The N-terminal half of S6 (5'S6) is juxtaposed with the P-region. The P-region lines the outer portion of the pore and determines ion selectivity. The C-terminal half of S6 (3'S6), together with the S4-S5 linker and the N-terminal half of the S5 domain (5'S5), lines the wide inner vestibule of the pore (Shieh and Kirsch, 1994; Slesinger et al., 1993; Lopez et al., 1994). T529 may be located in the intersection between the narrowest part of the pore and the opening into the wide inner vestibule, a "strategic" position for binding of an efficient pore blocker. Therefore, we mutated T529 of rKv1.43-25 to different residues and studied the effects of mutations on the binding of quinidine and 4AP.

View larger version (92K): [in this window] [in a new window]   Fig. 1.   Top, Cartoon of transmembrane topology of one subunit of a generic voltage-gated K channel. The structures between the end of S4 and the end of S6 are highlighted here. The linker between the S4 and S5 domains (S4-S5) is shown as a short -helix (Holmgren et al., 1996). The pore sequence is divided into 5' half (P1, shown as a short -helix) and 3' half (P2, shown as an elongated structure) (Lu and Miller, 1995). The S6 domain is divided into two -helices due to proline(s) in the sequence (Durell and Guy, 1996). The inner vestibule of the pore is lined by the S4-S5 linker (Slesinger et al., 1993), 5'S5 (Shieh and Kirsch, 1994) and 3'S6 domains (Lopez et al., 1994). The amino acid sequence of S6 in rKv1.43-25 is listed below, with threonine at position 529 highlighted. Bottom, Structures of quinidine and 4AP.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

In vitro mutagenesis and oocyte preparation. Oligonucleotide-directed mutagenesis was carried out on the rKv1.43-25 background (Tseng-Crank et al., 1993) using a commercial mutagenesis kit according to the manufacturer's instructions (Altered Sites System, Promega, Madison, WI). Mutations were confirmed by DNA sequencing using the dideoxy-mediated chain termination method.

Electrophysiological experiments. Whole cell currents were recorded using a modified two-microelectrode voltage clamp method. The voltage-sensing electrodes had a tip resistance of 1 to 2 M, and the current-passing pipettes had a tip resistance of 0.1 to 0.3 M ("cushion-pipettes," with tips filled with 1% agarose in 3 M KCl to prevent 3 M KCl leakage). An Oocyte Clamp (OC-725B) amplifier was used (Warner Instruments, Hamden, CT). During current recordings, the oocytes were superfused with a Cl-free solution to minimize interference from endogenous Cl channel currents. The composition of this solution was as follows unless otherwise stated (in mM): NaOH 96, KOH 2, Ca(OH)₂ 1.8, MgSO₄ 1, HEPES 5, Na-pyruvate 2.5, pH 7.5 with methanesulfonic acid. The bath temperature was 23 to 25°C, unless otherwise denoted. In some experiments the extracellular [K] was raised to 98 mM. This solution was made by replacing NaOH with equimolar KOH.

Data acquisition and analysis. Voltage clamp protocol generation and data acquisition were controlled by a 486 IBM computer via pClamp software and a Digidata 1200 interface (Axon Instruments). Currents were digitized at a sampling interval of 0.1 to 0.5 msec. Data analysis were performed using Clampfit (version 6.1 of pClamp), Excel (Microsoft Corporation, Redmond, WA) and PeakFit (Jandel Scientific, Corte Madera, CA) programs. Data are presented as means and S.E.s. Statistical significance was tested by unpaired t test (SigmaStat 2.0, Jandel Scientific).

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Comparison of Effects of Quinidine and 4AP on rKv1.4

The effects of quinidine on the wild-type rKv1.4 channel has been well characterized (Yatani et al., 1993). It was concluded that quinidine blocks the channel from the intracellular side of the membrane and binding preferentially occurs in the open state. Quinidine shortens the single channel open time and prolongs the closed time, indicating that the binding and unbinding kinetics are rapid. The effects of quinidine on the deletion mutant, rKv1.43-25, are shown in figure 2A. Deleting amino acids 3 to 25 from the N-terminus disrupted the fast N-type inactivation process (Tseng-Crank et al., 1993). rKv1.43-25 decayed slowly during depolarization, probably due to a slow C-type inactivation process (Hoshi et al., 1991). Quinidine slightly accelerated the decay of rKv1.43-25. However, it was difficult to dissect out a clear phase of block development. The current trace induced by the first pulse after quinidine application was similar to that recorded after repetitive pulses (upper panel of fig. 2A). This indicates that quinidine blockade of rKv1.43-25 developed rapidly and reached a steady-state during a 1-sec depolarization pulse. Unblock occurred during the interpulse interval when the membrane was repolarized to 80 mV, and blockade redeveloped during the next depolarization pulse. This is consistent with the fast binding and unbinding kinetics of quinidine described for the wild-type rKv1.4 channel (Yatani et al., 1993), and was confirmed in our single channel recordings of rKv1.43-25 (data not shown). Quinidine slowed the decay of tail current of rKv1.43-25, and induced a "cross-over" with the control tail current (data not shown). This is indicative of drug unblock upon repolarization prior to the closure of the activation gate (Yeola et al., 1996).

View larger version (21K): [in this window] [in a new window]   Fig. 2.   Quinidine and 4AP suppressed rKv1.43-25 with different time and voltage dependencies. A, Effects of quinidine (50 µM) on rKv1.43-25. Top, Quinidine slightly accelerated the decay of rKv1.43-25 but there was no clear phase of block development. Currents were recorded during 1-sec depolarization pulses from Vh 80 mV to Vt +40 mV before and after drug application. Bottom, Effect of quinidine on rKv1.43-25 was enhanced by membrane depolarization. Drug effect was quantified by the ratio of current amplitude recorded in the presence of quinidine to control (Id/Ic, right ordinate). Currents were measured at the end of 1-sec depolarization pulses from Vh 80 mV to different test voltages applied once every 15 sec. Data were average from three experiments. Data of Id/Ic were compared with the activation curve of rKv1.43-25, that was constructed in the following manner. In 98 mM [K]o, the membrane was depolarized from Vh 80 mV to different Vt (from 55 to +60 mV in 5 mV increments) for 100 msec, followed by repolarization to 65 mV. Peak inward tail currents were measured 2 msec after return to 65 mV, and then normalized to the maximal tail current following Vt to +60 mV. The relationship between Vt and the normalized tail current (as an estimate of the fraction of channels activated at Vt, right ordinate) was fit with a double Boltzmann function (1) where A1 and (1  A1) are fractions of the two Boltzmann components, V1 and V2 are the corresponding half-maximum activation voltages and k1 and k2 are the slope factors. The curve was calculated using equation (1) and mean parameter values averaged from 11 experiments: A1 = 0.64 ± 0.02, V1 = 39.7 ± 2.0 mV, k1 = 4.3 ± 0.5 mV, V2 = 6.7 ± 4.5 mV, k2 = 13.5 ± 1.1 mV. B, Effects of 4AP (0.5 mM) on rKv1.43-25. Top, 4AP suppressed rKv1.43-25 during depolarization with a clear phase of block development, and there was no unblock between depolarization pulses. Currents were recorded during 1-sec depolarization pulses from Vh 80 mV to Vt 20 mV under the control conditions and during the first to third pulses (labeled 1, 2 and 3) after 4AP application. Between the control and the first pulse after 4AP application, the cell was superfused with the 4AP-containing solution for 5 min during which the membrane was held at 80 mV. The interpulse interval in the presence of 4AP was 1 min. Bottom, Effect of 4AP on rKv1.43-25 was reduced by membrane depolarization. The drug effect was quantified by the current ratio (Id/Ic, left ordinate). Currents were measured at the end of 5-sec depolarization pulses from Vh 80 mV to different Vt applied once every 15 sec (control) or every 60 sec (in the presence of 4AP). Data were average from three experiments. Data of Id/Ic are superimposed on the activation curve of rKv1.43-25 (right ordinate).

In the lower panel of figure 2A, the ratios of current amplitudes measured at the end of 1-sec pulses in the presence of quinidine to control (I_d/I_c) are plotted against the test pulse voltage. The suppressing effect of quinidine on rKv1.43-25 was enhanced by membrane depolarization. The voltage-dependence of drug effect was analyzed using the Woodhull formalism and data points in the voltage range of +30 to +70 mV, when channel activation approached a plateau (Woodhull, 1973). The analysis suggests that the quinidine binding site in rKv1.43-25 lies within the membrane electrical field and has an equivalent electrical distance of 0.30 ± 0.01 from the intracellular surface of the membrane (n = 10).

The effects of 4AP on the wild-type and the deletion mutant of rKv1.4 have been described previously (Yao and Tseng, 1994). It was concluded that 4AP blocks rKv1.4 from the intracellular side of the membrane and binding is facilitated by channel activation. Figure 2B illustrates the key features of 4AP actions on rKv1.43-25 to contrast them with those of quinidine's actions. When the membrane was held at 80 mV during wash-in of 4AP (0.5 mM) for 5 min, modest block developed as suggested by the change in the peak current amplitude during the first pulse after 4AP application. In contrast to the modest degree of block that had developed at 80 mV for 5 min, marked block developed rapidly during depolarization to 20 mV, as reflected by the accelerated decay of rKv1.43-25 during the first pulse after 4AP application. The apparent separation between channel activation and block development indicates that the rate of 4AP blockade of rKv1.43-25 is slow relative to the rate of channel activation. This is consistent with the effects of 4AP on single channel currents of rKv1.43-25: 4AP does not alter the single channel open time but shortens the burst duration (Yao and Tseng, 1994). 4AP did not unblock from rKv1.43-25 during the interpulse interval when the membrane was repolarized to 80 mV, as reflected by the marked reduction of peak current amplitudes during the second and the third pulses after 4AP application. Therefore, data in figure 2 indicate two differences between quinidine and 4AP in their actions on rKv1.43-25: 1) the kinetics of drug-channel interaction is much slower for 4AP than for quinidine and 2) upon repolarization the bound 4AP molecule is trapped inside the channel by the closure of the activation gate, although quinidine leaves the channel followed by deactivation.

Another important difference between 4AP and quinidine is the voltage-dependence of drug action. In contrast to quinidine, the suppressing effect of 4AP on rKv1.43-25 was reduced by membrane depolarization in the voltage range from 40 to +60 mV (fig. 2B, lower panel). One possible explanation for such a voltage-dependence is that, although 4AP binding to rKv1.43-25 was facilitated by channel activation, membrane depolarization actually induced 4AP unblock by reducing the binding site affinity. This was tested by the experiments shown in figure 3. In these experiments, the membrane was held at 45 mV during wash-in of 4AP and during interpulse intervals. This holding voltage was 5 to 10 mV above the threshold of channel activation and therefore would allow 4AP binding to occur (evidenced by a 4AP-induced reduction in the outward holding current level, data not shown). Under these conditions, currents induced by depolarization pulses in the presence of 4AP had completely different time courses than those shown in figure 2B: the initial current amplitude was small but the current level gradually increased during depolarization. Therefore, consistent with the prediction based on figure 2B, a depolarization pulse could induce 4AP unblock if 4AP blockade had already occurred. The rate and extent of 4AP unblock were enhanced by stronger depolarization (fig. 3A). As shown in figure 3B, the voltage-dependence of the rate and extent of 4AP unblock from rKv1.43-25 tracked the voltage-dependence of channel activation, suggesting that channel activation induced 4AP unblock.

View larger version (25K): [in this window] [in a new window]   Fig. 3.   4AP dissociated from rKv1.43-25 in the open state. A, Superimposed rKv1.43-25 current traces recorded during 2-sec depolarization pulses from Vh 45 to Vt 30 mV (left) or to +50 mV (right) before (control) and after 4AP (0.5 mM) application. The interpulse interval was 15 sec under the control conditions and 60 sec in the presence of 4AP. B, Comparison of voltage-dependence of rate and extent of 4AP unblock from rKv1.43-25 and voltage-dependence of channel activation. The degree of 4AP unblock was evaluated by the ratio (Id/Ic, solid circles and left ordinate) of currents measured at the end of 2-sec depolarization pulses from Vh 45 mV to different Vt. The rate of 4AP unblock was the reciprocal of time constant of single exponential fit to the rising phase of current traces in the presence of 4AP (open circles, right ordinate, n = 3). The activation curve of rKv1.43-25 (fraction activated, left ordinate) was calculated as described for figure 2.

Data presented in figure 3 support the notion that activation-induced 4AP unblock is an important factor in determining the voltage-dependence of 4AP action on rKv1.43-25 (fig. 2B). However, these data do not allow us to determine whether 4AP bound within the channel could experience the membrane electrical field or not. We addressed this issue by examining the effects of elevating [K]_o on drug actions. Elevating [K]_o will impede drug binding by an electrostatic repulsion and reduce the apparent blocking potency if: 1) the binding site is within the pore and 2) drug binds in the positively charged form (as is the case for quinidine binding to rKv1.43-25). This was verified by the experiments shown in figure 4. On average, elevating [K]_o from 2 to 98 mM increased quinidine's K_d value (at +40 mV) from 117.2 ± 27.3 µM (n = 15) to 663.1 ± 107.1 µM (n = 6), i.e., a 5.7-fold increase in the K_d value (P < .001). In contrast, elevating [K]_o from 2 to 98 mM had much lesser effects on the apparent blocking potency of 4AP on rKv1.43-25. This can be seen from the similar degree of current reduction at the end of the pulses shown in figure 5A. However, elevating [K]_o slowed the development of 4AP blockade. In the experiment shown in figure 5A,  of 4AP block was 203 msec in 2 mM [K]_o and 500 msec in 98 mM [K]_o. The apparent rate of 4AP block development (reciprocal of  of block) is plotted against 4AP concentration in figure 5B. Linear regression was used to estimate the apparent binding (K_on) and unbinding (K_off) rate constants based on equation (3) (fig. 5, legend). In 2 mM [K]_o, K_on was 3.35 mM¹sec¹ and K_off was 3.65 sec¹. In 98 mM [K]_o both rate constants were reduced (K_on = 1.96 mM¹sec¹, K_off = 2.92 sec¹). The K_d value estimated from K_on/K_off was 1.09 vs. 1.49 mM in 2 vs. 98 mM [K]_o. Therefore, elevating [K]_o from 2 to 98 mM caused only a 50% increase in 4AP's K_d value. This is much less than the 5.7-fold increase in quinidine's K_d value caused by the same degree of [K]_o elevation. There are two possible explanations for such a low [K]_o sensitivity in 4AP's blocking potency: 1) the 4AP binding site is outside the pore or 2) 4AP's charge is not exposed once bound inside the channel. Results from our mutagenesis experiments to be presented below can help resolve this issue.

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Elevating [K]o from 2 to 98 mM reduced the apparent potency of quinidine suppression of rKv1.43-25. A, Superimposed current traces before and after quinidine (200 µM) application in 2 (left) and 98 (right) mM [K]o. Currents were recorded during 1-sec depolarization pulses from Vh 80 mV to Vt +80 mV. Data were obtained from the same cell. B, Summary of quinidine's Kd values measured in 2 and 98 mM [K]o, with numbers of measurements in parentheses. The Kd values were estimated using current ratio (Id/Ic) of currents measured at the end of 1-sec depolarization pulses to +40 mV, based on (2)

View larger version (23K): [in this window] [in a new window]   Fig. 5.   Elevating [K]o from 2 to 98 mM slowed 4AP suppression of rKv1.43-25 with little effects on the apparent blocking potency. A, Superimposed current traces recorded under the control conditions (thin traces) and during the first pulses after 4AP (0.5 mM) application (open circles). Currents were recorded during 2-sec depolarization pulses from Vh 80 mV to Vt 20. Between the "control" and "4AP" traces the cell was superfused with the 4AP solution for 5 min while the membrane was held at 80 mV. Currents recorded in 98 mM [K]o were inward. The current traces are inverted to facilitate comparison with current traces in 2 mM [K]o. The data points of open circles are superimposed on curves calculated from a single exponential function with best-fit time constants (203 and 500 msec in 2 and 98 mM [K]o, respectively). B, Summary of the apparent rates of 4AP block development at different 4AP concentrations in 2 and 98 mM [K]o. The reciprocal of  of 4AP block development (determined as described for A) is plotted against [4AP]. Each data point was average from four to eight measurements. Linear regression was applied to estimate the apparent 4AP binding (Kon) and unbinding (Koff) rate constants according to (3) The superimposed lines were calculated from equation (3) with best-fit parameter values: 2 mM [K]o, Kon = 3.35 mM1sec1, Koff = 3.65 sec1; 98 mM [K]o, Kon = 1.96 mM1sec1, Koff = 2.92 sec1.

The slow kinetics of 4AP binding and unbinding suggest that drug-channel interaction may not be a simple, diffusion-limited process. Instead, 4AP binding and unbinding may involve or require conformational changes in the channel protein. If this is the case, the kinetics of 4AP actions should display a temperature-sensitivity higher than that of a diffusion-limited event (Hille, 1992). This was tested in the experiments shown in figure 6. The time constants of 4AP blockade were estimated by fitting a single exponential function to the decay phase of current traces induced by the first pulses after 4AP application at V_h 80 mV (fig. 6A, left panels). Elevating the bath temperature from 23 to 35°C markedly shortened  of block development (241.3 ± 20.5 vs. 80.2 ± 7.3 msec at 23 and 35°C, fig. 6A, right panel). This difference in the rate of 4AP block had a Q₁₀ of 2.5. The  values of 4AP unblock were estimated by fitting a single exponential function to the rising phase of current traces induced by the first pulse after 4AP application at V_h 45 mV (fig. 6B, left panels). The average  of 4AP unblock was 489.3 ± 37.0 vs. 187.8 ± 17.0 msec at 23 vs. 35°C (Q₁₀ = 2.2, fig. 6B, right panel). Therefore, the Q₁₀ values of 4AP block and unblock are higher than the expected value of diffusion-limited events (Q₁₀  1.5), but are similar to the reported values of Q₁₀ for channel gating processes (Hille, 1992). This supports the notion that 4AP interaction with rKv1.43-25 involves or requires conformational changes in the channel protein.

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Temperature-sensitivity of kinetics of interactions between 4AP and rKv1.43-25. A, Rates of 4AP (0.5 mM) blockade at 23 and 35°C. Left, Superimposed current traces recorded during 1-sec depolarization pulses from Vh 80 mV to Vt 20 mV before (thin traces) and during the first pulses after 4AP application (open circles). 4AP was applied for 5 min during which the membrane was held at 80 mV before the first pulses were applied. The data points of open circles are superimposed on curves calculated from a single exponential function with best-fit time constants (254 msec at 23°C and 89 msec at 35°C). Right, Summary of  of 4AP block development at 23 and 35°C. The voltage clamp protocol and data analysis were as described for the left panel. Numbers in parentheses are numbers of measurements. The estimated Q10 value for the rate of 4AP blockade is 2.5. B, Rate of 4AP unblock from rKv1.43-25 at 23°C and 35°C. Left, Superimposed current traces recorded during 2-sec depolarization pulses from Vh 45 to Vt 20 mV before (thin traces) and during the first pulses after 4AP (0.5 mM) application (open circles). 4AP was applied for 5 min during which the membrane was held at 45 mV before the first pulses were applied. The data points of open circles are superimposed on curves calculated from a single exponential function with best-fit time constants (512 msec at 23°C and 177 msec at 35°C). Right, Summary of  of 4AP unblock at 23 and 35°C. The voltage clamp protocol and data analysis were as described for the left panel. Q10 of 4AP unblock is 2.2.

The above data indicate that although both quinidine and 4AP bind to rKv1.43-25 from inside the cell membrane and binding is facilitated by channel activation, the kinetics and voltage-dependence of drug-channel interaction are distinctly different. The differential effects of elevating [K]_o on the blocking potency of quinidine and 4AP suggest that these two blockers experience different degrees of electrostatic effects of K ions from the opposite end of the pore. We then examined whether the binding sites of quinidine and 4AP in rKv1.43-25 overlap and what controls the affinity of their binding sites. Our focus was on the S6 domain and in particular, threonine at position 529 (T529).

Effects of T529 Mutations on the Blocking Potency of Quinidine and 4AP

T529 was mutated to 13 other residues. Five of them (T529D, T529K, T529Q, T529N and T529Y) did not produce currents. Currents through a sixth mutant, T529W, were very small. The remaining seven mutants (T529G, T529S, T529A, T529V, T529I, T529L and T529F) produced robust currents and were used to test the effects of T529 mutations on drug binding.

The effects of quinidine on WT (=rKv1.43-25) and T529 mutant channels were quantified by K_d values at +40 mV. At this voltage, the degrees of channel activation reached or approached a plateau in all cases, avoiding possible alterations in the apparent blocking potency among channels introduced by different degrees of channel activation. Figure 7 illustrates current traces of WT and T529 mutants recorded during 1-sec depolarization pulses to +40 mV before and after quinidine (100 µM) application. Quinidine reduced current amplitudes in all cases. There was no clear phase of block development in any of the channels. Quinidine's effects approached a steady-state at the end of the 1-sec pulses. The current ratios (I_d/I_c) measured at the end of the pulses were used to construct concentration-response relationships.

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Effects of 100 µM quinidine on rKv1.43-25 (WT) and mutants in which threonine at position 529 was replaced by different residues. The mutants were designated as T529X, with X = substituting residue using the one-letter code. Currents were recorded during 1-sec depolarization pulses from Vh 80 mV to Vt +40 mV applied once every 15 sec.

Figure 8 shows the mean values of I_d/I_c plotted against the quinidine concentration for WT and T529 mutant channels. The data points are superimposed on curves calculated from equation (2) with mean K_d values averaged from 4 to 11 measurements each (listed in fig. 11 legend). Replacing T529 with phenylalanine (T529F) reduced quinidine's blocking potency, although all the other mutant channels displayed a higher sensitivity to quinidine compared to that of the WT channel.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   Concentration-response relationships of quinidine suppression of rKv1.43-25 (WT) and T529 mutants. Each cell was exposed to increasing concentrations of quinidine (from 5 to 1000 or 2000 µM) and the current amplitude was monitored by 1-sec depolarization pulses from Vh 80 mV to Vt +40 mV applied once every 15 sec. The relationship between current ratio (Id/Ic) and [quinidine] was fit with equation (2) to estimate the Kd value. Shown are data points (mean values from 4 to 11 measurements, S.E. bars omitted for the sake of clarity) superimposed on curves calculated from equation (2) with mean Kd values listed in figure 11 legend.

The K_d values of 4AP were determined using the voltage clamp protocol illustrated in figure 9 (except for T529L and T529I): the current amplitude was monitored by 40-msec depolarization pulses from V_h 80 mV to V_t 20 mV applied once every 15 or 30 sec. The cell was exposed to increasing concentrations of 4AP and the current ratio (I_d/I_c) measured at the steady-state of 4AP's effect at each concentration was used to construct the concentration-response relationship. The rationale for this voltage clamp protocol is the following. 4AP's blocking potency is reduced by membrane depolarization due to blocker dissociation from an activated channel (figs. 2B and 3). Because T529 mutations alter the voltage-dependence of channel activation (authors' unpublished data), we needed to avoid the interference from differences in the degree of channel activation when measuring 4AP's blocking potency. Therefore, short (40 msec) depolarization pulses to a low voltage (20 mV) were used to monitor the progression of drug effects without inducing significant 4AP dissociation. Because once bound 4AP molecules were trapped inside the channel by the closure of the activation gate at V_h (80 mV), the degree of 4AP blockade would accumulate over a train of such short pulses. Figure 9 shows that, under the conditions described above, 4AP blocked the WT channel with a K_d of 126 µM.

View larger version (18K): [in this window] [in a new window]   Fig. 9.   Concentration-response relationship for 4AP suppression of rKv1.43-25 (WT). The main plot illustrates the time course of changes in current ratio (Id/Ic) when the cell was exposed to increasing concentrations of 4AP (in µM, marked on top). The current amplitudes were monitored by 40-msec depolarization pulses from Vh 80 mV to Vt 20 mV applied once every 30 sec. Inset on right, Superimposed original current traces recorded before 4AP application and at the steady-state of 4AP effect (concentrations marked to the right). Inset on left, Concentration-response relationship (Id/Ic vs. [4AP]) from the same experiment. The superimposed curve was calculated from equation (2) with Kd = 126 µM.

Data from representative experiments on T529F and T529L are shown in figure 10. 4AP blocked T529F with a much higher potency (K_d = 14.3 µM) than its effect on the WT channel. As for the WT channel, the 4AP's effect on T529F was partially reversible after washing out of 4AP. In contrast, T529L was insensitive to 1 or 2 mM 4AP. In the experiment shown in figure 10B, the current amplitude was even slightly increased in 10 mM 4AP. Elevating the 4AP concentration to 20, 50 and 100 mM caused a rapid reduction in the current amplitude and, at 100 mM 4AP, this was followed by a secondary decrease in current amplitude along with an increase in the membrane "leak" conductance. Washing out 4AP caused a rapid but incomplete reversal of blocker effects. The low 4AP sensitivity of T529L made it difficult to construct a complete concentration-response relationship. The secondary reduction in current amplitude seen in high 4AP concentrations might be due to effects unrelated to pore blockade. Therefore, the K_d value for T529L was estimated from the initial current ratio in the presence of 100 mM 4AP (arrow in fig. 10B) by equation (2). For T529I that also had a low sensitivity to 4AP, the K_d value was similarly estimated by current ratio in 10 mM 4AP.

View larger version (22K): [in this window] [in a new window]   Fig. 10.   Comparison of effects of 4AP on T529F (A) and T529L (B). Each panel shows the time course of changes in current ratio (Id/Ic) when the cell was exposed to increasing concentrations of 4AP (marked on top) and after washout. The current amplitudes were monitored using the same voltage clamp protocol as described for figure 9. Insets, Superimposed current traces recorded before and after 4AP application (concentrations marked to the right). The concentration-response relationship of 4AP's effect on T529F was fit with equation (2) to estimate Kd (14.3 µM). The initial Id/Ic value of T529L in the presence of 100 mM 4AP (arrow) was used to calculate Kd from equation (2) (Kd = 67 mM).

Figure 11 shows a summary of K_d values of quinidine and 4AP in blocking the WT and T529 mutant channels. The K_d values are listed in figure 11 legend. For T529G, T529S, T529V, T529I and T529L, the quinidine's K_d values were reduced (blocking potency enhanced) while the 4AP's K_d values were increased relative to those of the WT channel. T529F had the opposite effects: quinidine's K_d value was increased while 4AP's K_d value was reduced. Finally, for T529A the K_d values of both quinidine and 4AP were reduced.

View larger version (21K): [in this window] [in a new window]   Fig. 11.   Comparison of effects of T529 mutations on the blocking potency of quinidine and 4AP. The voltage clamp protocols and methods of Kd estimation were as described in figures 7 to 10. The Kd values of quinidine (left ordinate) and those of 4AP (right ordinate) are plotted on a logarithmic scale against channel types along the abscissa (WT = rKv1.43-25). The following are the Kd values (n = number of measurements): Quinidine (µM) n 4AP (µM) n WT 63.2  ± 5.4 10 124.2  ± 6.7 3 T529G 6.9  ± 0.9a 6 192.1  ± 27.9 5 T529S 31.6  ± 6.4a 10 311.9  ± 86.9 4 T529A 20.4  ± 3.4a 8 33.2  ± 7.5a 3 T529V 13.0  ± 3.8a 4 195.6  ± 41.5 6 T529I 12.3  ± 3.4a 6 10000 1 T529L 18.6  ± 3.1a 6 190000  ± 64799a 5 T529F 479.6  ± 109.8a 5 18.4  ± 1.9a 6 a P < .05 vs. WT.

Data in figure 11 indicate that position 529 is an important determinant of blocking potency of both quinidine and 4AP. This, in turn, suggests that the binding sites of quinidine and 4AP in rKv1.4 at least partially overlap with each other. Replacing T529 with G, S, V, I, L and F had opposite effects on quinidine and 4AP binding, indicating that different factors are involved in determining the binding sites' affinity for the two blockers.

Correlating Effects of T529 Mutations on Quinidine and 4AP Blocking Potency with Side Chain Properties or with Changes in Channel Gating

Factors affecting quinidine's binding affinity. To elucidate the factors involved in determining the effects of T529 mutations on quinidine binding, we tried to correlate changes in quinidine's blocking potency with side chain properties at position 529. Two factors were considered: side chain polarity and residue volume (fig. 12). It has been suggested, based on a limited mutagenesis study of quinidine's effect on hKv1.5, that the primary factor in determining drug binding affinity is the hydrophobicity of side chain at position 505 (equivalent to 529 of rKv1.4) (Yeola et al., 1996). However, when examined with more extensive mutations and with the side chain hydrophobicity evaluated quantitatively, we could not identify a clear correlation between quinidine's K_d value and side chain polarity at 529. This is shown in figure 12A: the K_d values of quinidine are plotted on a logarithmic scale against 529 side chain polarity. The side chain polarity was estimated by the free energy (kcal/mol) needed to transfer the side chain from an aqueous phase to octanol (Guy, 1985).

View larger version (25K): [in this window] [in a new window]   Fig. 12.   Correlating side chain properties of residues at position 529 with quinidine's blocking potency. Quinidine's Kd values from figure 11 are plotted on a logarithmic scale against 529 side chain polarity [(Guy, 1985) with the direction of increasing hydrophobicity or increasing hydrophilicity marked] (A) or 529 residue volume (Zamyatnin, 1972) (B). The line in B was a linear regression of data points from WT, T529G, T529S, T529A and T529F.

Factors affecting 4AP's binding affinity. Plotting 4AP's K_d value against the side chain polarity or residue volume at position 529 did not reveal any clear pattern (data not shown). Therefore we needed to consider other factor(s) that might affect 4AP binding. It has been shown for 4AP's effects on chimeric channels of Kv2.1 and Kv3.1 that there is an empirical linear relationship between 4AP's K_d value and the rate of channel deactivation: a slower deactivation is coupled with a lower 4AP sensitivity (Shieh and Kirsch, 1994). A slower rate of deactivation reflects a longer dwell time in the open state. The above relationship therefore is consistent with the notion that 4AP tends to dissociate from channels in the open state that have a low binding affinity. We tested whether a relationship between 4AP's K_d value and the rate of channel deactivation could be identified among WT and T529 mutant channels.

View larger version (26K): [in this window] [in a new window]   Fig. 13.   A, Effects of T529 mutations on the rate of deactivation at 80 mV. Recordings were made in 98 mM [K]o using the voltage clamp protocol shown on top left: the membrane was depolarized to +60 mV for 100 msec followed by repolarization to 80 mV. Inward tail currents at 80 mV are shown. In each panel, the tail current from one of the T529 mutants (depicted by the thick trace) is superimposed on the WT tail current (thin trace). To facilitate comparison, the gain of display is adjusted so that the tail current amplitudes match each other. Note the differences in the time calibration. B, Correlating effects of T529 mutations on 4AP's blocking potency with effects on the rate of deactivation. 4AP's Kd values from figure 11 are plotted against  of deactivation at 80 mV on a double logarithmic scale. The dotted lines are drawn by eye. The following are values of  of deactivation (in msec, numbers of measurements in parentheses): WT 5.7  ± 1.1 (4) T529G 7.1  ± 0.9 (3) T529S 3.8  ± 0.6 (6) T529A 21.3  ± 1.7a (7) T529V 70.0  ± 7.9a (7) T529I 300.0  ± 52.8a (5) T529L 403.2  ± 135.3a (3) T529F 12.1  ± 3.5 (3) a P < .05.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Our data suggest that both quinidine and 4AP can bind deep in the pore, and position 529 in the middle of the S6 domain is an important determinant for their binding affinity. This may be somewhat surprising for 4AP, for which a more superficial binding site has been assigned based on mutagenesis work on chimeric channels of Kv2.1 and Kv3.1 (Durell and Guy, 1996; Shieh and Kirsch, 1994). In our study, replacing T529 with leucine practically abolished the 4AP sensitivity. Such a dramatic effect argues that T529 indeed directly interacts with 4AP.

Does residue at position 529 face the lumen of the pore? It has been shown that the S6 segments of Na, Ca and K channels can play important roles in channel gating (Hoshi et al., 1991; Zuhlke et al., 1994; Zhang et al., 1994; McPhee et al., 1995). This suggests that the S6 segments do not simply serve a "packing" function in stabilizing the protein structure, but can actively engage in conformational changes in the channel gating processes. Our data from T529 mutations are consistent with such a role of the S6 segment in rKv1.4 (authors' unpublished data).

Steric hindrance influences quinidine, but not 4AP, binding. For a bulky molecule like quinidine, steric hindrance is an important factor in determining the apparent binding affinity. It is possible that quinidine binds to the narrow part of the inner vestibule between T529 and the ion selectivity filter (fig. 1) and causes a deformation of this part of the pore. Mutations that reduce side chain volume here can relieve the steric constraint and stabilize quinidine binding. This is supported by data presented here (fig. 12). Furthermore, we have shown that mutating the threonine at position 501 [in the middle of the P-region and adjacent to the ion selectivity filter (Durell and Guy, 1996)] to serine and thus reducing side chain volume enhances quinidine binding (K_d at +40 mV: 63.2 ± 5.4 to 29.2 ± 3.4 µM, P < .05) (Zhang et al., 1995). The mechanism is probably also a relief of steric constraint around the quinidine binding site.

Channel activation increases the accessibility but reduces the affinity of 4AP binding site. Our data suggest that activation of the rKv1.43-25 channel has two effects on 4AP binding: increasing the accessibility of the binding site (fig. 2B) and reducing the binding site affinity (fig. 2B and 3). The correlation between 4AP's K_d value and of deactivation shown in figure 13B lends further support to the latter contention. Therefore, 4AP binding is hindered when the rKv1.43-25 channel is fully-closed or when the channel is fully-open, and optimal 4AP binding probably occurs in a "transitional" state between these two extremes.

S6 as an important determinant of drug binding to voltage-gated ion channels. Our data are consistent with a common theme of drug-channel interactions that has emerged from studies of drug binding to Na, Ca and K channels: the S6 segment plays an important role in drug binding, and hydrophobic interactions between drug molecules and S6 residues stabilize drug binding (Baukrowitz and Yellen, 1996; Hockerman et al., 1995; Ragsdale et al., 1996). Furthermore, our data show that T529 in rKv1.4 plays a key role in determining the binding affinity of pore blockers. This is similar to the situation in the Shaker (Baukrowitz and Yellen, 1996) and hKv1.5 (Yeola et al., 1996) channels.

View larger version (10K): [in this window] [in a new window]   Scheme 1.

View larger version (10K): [in this window] [in a new window]   Scheme 2.