Corresponding author T.-C. Hwang: Dalton Cardiovascular Research Center, 134 Research Park, University of Missouri - Columbia, Columbia, MO 65211, USA.
1Fast flickery block of the cystic fibrosis transmembrane conductance regulator (CFTR) was studied with cell-attached and whole-cell patch-clamp recordings from mouse NIH3T3 cells stably expressing a mutant CFTR channel, K1250A-CFTR. This mutant CFTR channel, once open, can stay open for minutes. Within a prolonged opening, the kinetics of fast flickery closures can be readily quantified.
2Flickering block of K1250A-CFTR channels was voltage dependent since the open probability within an opening burst decreased as the membrane was hyperpolarized.
3Mean open time (τo) and mean closed time (τc), obtained from single-channel kinetic analysis, were corrected for missed events. Our data show that corrected τc was voltage dependent while corrected τo exhibited little voltage dependence. Results from whole-cell current relaxation upon voltage jump further indicate that τc at a membrane potential of -100 mV was at least 10-fold longer than that at +100 mV.
4τc, but not τo, was sensitive to external permeant anions. After complete replacement of external Cl− with impermeant anions, τc showed little voltage dependence and approximated a value observed under strong hyperpolarization in the presence of high external permeant anions. These results suggest that the resident time of the blocker is prolonged by conditions (i.e. hyperpolarization or the absence of external permeant anions) that deplete Cl− in the CFTR pore.
5Results from macroscopic current noise analysis of both wild-type CFTR and K1250A-CFTR channels further confirm the voltage dependence and Cl− sensitivity of the fast flickery block observed with single-channel analysis.
6We conclude that the voltage dependence of the flickery block in CFTR is mainly due to the voltage-dependent occupancy of an anion-binding site in the channel pore by trans-anions. The blocker acquires a voltage-dependent off rate through an electrostatic interaction with Cl− in the pore.
CFTR channel gating has been studied extensively in its native environment as well as in heterologous expression systems (reviewed by Nagel, 1999). In cell-attached patches, CFTR channels open in bursts which are separated by long closures that last for hundreds of milliseconds to seconds depending on the degree of cAMP stimulation (Hwang et al. 1997; Al-Nakkash & Hwang, 1999). Within a burst, there are numerous brief closings that are referred to as fast flickers with a time scale of several to tens of milliseconds (Gray et al. 1989; Tabcharani et al. 1991; Haws et al. 1992; Fischer & Machen, 1994). It has been proposed that these fast flickery closings are caused by transient block of the channel from the cytoplasmic side of the membrane. This hypothesis is based on two observations. First, flickers are very prominent in cell-attached patches and are dramatically diminished upon excision of the patch into an inside-out mode (Haws et al. 1992; Fischer & Machen, 1994; Fisher & Machen, 1996). Second, flickers appear to be voltage dependent since a greater number of flickery closings are observed at more hyperpolarizing potentials (Gray et al. 1989; Haws et al. 1992). Although the molecular nature of the blockers is unknown, it has been suggested that tyrosine phosphorylation may regulate this flickery block (Fischer & Machen, 1996). Since permeabilizing the cell membrane with α-toxin, which creates pores that are permeable to molecules smaller than 4 kDa, does not alter the kinetics of flickery closings, it has been suggested that the flickers are caused by blockers that are not readily diffusible (Fischer, 1997).
Detailed kinetics of this blockade and the underlying mechanism for its voltage dependence remain unclear due to several technical hurdles. First, patches containing a single CFTR channel, essential for dwell time analysis, are rare. Second, it is difficult to separate the ATP-dependent gating events from the fast flickery closings for wild-type CFTR channels, especially in the presence of millimolar ATP. Third, extremely long single-channel recordings are required to collect a sufficient number of events for kinetic analysis because, in cell-attached mode, wild-type CFTR closes for hundreds of milliseconds between opening bursts even in the presence of a maximal cAMP stimulation (e.g. Hwang et al. 1997; Al-Nakkash & Hwang, 1999). To overcome these technical difficulties, we characterized the kinetics of the flickery blockade using the CFTR mutant K1250A in which the conserved Walker A lysine (K) at amino acid position 1250 located in NBD2 is replaced by the neutral amino acid alanine (A). Previous studies indicate that this charge neutralization dramatically prolongs an open state that can last for minutes (Gunderson & Kopito, 1995; Zeltwanger et al. 1999). Within a long opening of K1250A-CFTR, fast flickers are clearly discernible (Zeltwanger et al. 1999). Because of this prolonged opening, it is relatively easy to obtain long single-channel recordings for kinetic analysis of the flickers with minimal contamination of the ATP-dependent gating events.
In the present paper, fast flickery block of CFTR was studied with cell-attached and whole-cell patch-clamp techniques in NIH3T3 cells stably expressing wild-type or K1250A-CFTR. Our results demonstrate that the off rate of the block was voltage dependent while the on rate showed little voltage dependence. Furthermore, the off rate was decreased and its voltage dependence was almost completely abolished after removing external permeant anions. That the voltage-dependent off rate was mainly responsible for the voltage-dependent flickery block was supported by results from whole-cell current relaxation analysis. Using spectrum analysis of macroscopic K1250A-CFTR and wild-type CFTR currents, we obtained corner frequencies (fc) at different membrane potentials. The measured fc agreed well with the fc calculated from the single-channel kinetic parameters obtained for K1250A-CFTR channels. Our data are consistent with a model in which the voltage dependence of the blockade is mainly caused by the voltage-dependent occupancy of an anion-binding site in the channel pore by permeant anions entering the pore from the external side of the membrane. Since this blocker probably carries a negative charge, our results also support the notion that the CFTR channel pore can be occupied by multiple anions.
NIH3T3 cell lines stably expressing wild-type CFTR (Anderson et al. 1991) or K1250A-CFTR (Zeltwanger et al. 1999) were grown at 37 °C and 5 % CO2 in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10 % fetal bovine serum. Prior to patch-clamp recordings, cells were passaged and plated out on small, sterile glass coverslips in 35 mm tissue culture dishes. Recordings were made 2-5 days later.
Single-channel patch-clamp recordings
Glass coverslips with adherent cells were transferred to a continuously perfused chamber on the stage of an inverted microscope (Olympus Corp., Tokyo, Japan). Pipettes were pulled using a two-stage vertical puller (Narishige, Tokyo, Japan) and were fire polished with a homemade microforge to a pipette resistance of 3-6 MΩ in the standard bath solution. For cell-attached patch recordings, the standard bath solution contained (mm): 145 NaCl, 5 KCl, 2 MgCl2, 1 CaCl2, 5 glucose, 5 Hepes and 20 sucrose (pH 7.4 with NaOH). The standard pipette solution contained (mm): 140 NMGCl, 2 MgCl2, 5 CaCl2 and 10 Hepes (pH 7.4 with NMG). The pipette Cl− was replaced by aspartate (0 mm Cl− solution), Br− (154 mm Br− solution) or NO3− (154 mm NO3− solution) in some experiments as described in Results. The 0 mm Cl− pipette solution contained (mm): 154 NMG-aspartate, 2 MgSO4 and 10 Hepes (pH 7.4 with NMG). The 154 mm Br− pipette solution contained (mm): 154 HBr, 2 MgSO4 and 10 Hepes (pH 7.4 with NMG). The 154 mm NO3− pipette solution contained (mm): 154 HNO3, 2 MgSO4 and 10 Hepes (pH 7.4 with NMG). When non-chloride pipette solutions were used, pipettes were filled 2/3 of the entire length from the tip with the desired solution and then ≈10 μl of 6 mm Cl− pipette solution was added on top. To obtain single-channel recordings, we activated multiple K1250A-CFTR channels with 10 μm forskolin, then washed out the forskolin and observed current decay. Data were collected when there was one channel left open. Under this condition, once the K1250A-CFTR channel closes from the activated state, it remains closed unless forskolin is re-applied. Thus, flickery events can be collected at various voltages with minimal contamination of the slow gating events. For all cell-attached recordings, holding potentials indicated are negative pipette potential (-Vp).
Single-channel CFTR currents were recorded at room temperature (≈22 °C) with an EPC9 patch-clamp amplifier (Heka Electronic, Lambrecht, Germany), filtered at 100 Hz with a built-in 4-pole Bessel filter and stored on videotape. Data were played back without further filtering and captured onto the hard disk of a computer at a sampling rate of 500 Hz for kinetic analysis.
Whole-cell patch-clamp recordings
Whole-cell currents were recorded using an Axopatch-1D patch-clamp amplifier (Axon Instruments, Foster City, CA, USA). Voltage-step protocols were generated using Igor software (Wavemetrics, Lake Oswego, OR, USA) and an XOP developed by Dr Richard Bookman at the University of Miami (Miami, FL, USA). The membrane potential was held at 0 mV. To obtain instantaneous and steady-state I-V relationships, the voltage was stepped to various potentials for 100 ms from -100 to +100 mV in 20 mV increments. Net CFTR currents were determined by subtracting the current response of the cell in the absence of forskolin from that in the presence of 10 μm forskolin. To minimize the effect of the capacitance transient, the instantaneous current at each voltage was measured as the current 1 ms after the voltage jump from 0 mV. The steady-state current at each potential was measured as the current at the end of the 100 ms voltage step. Current traces in response to the voltage-step protocol were filtered at 5 kHz and digitized at 10 kHz directly onto the hard disk using an ITC-16 interface (Instrutech Corp., Greatneck, NY, USA). The pipette solution contained (mm): 101 CsCl, 2 MgCl2, 20 TEACl, 10 MgATP, 8 Tris, 10 EGTA and 5.8 glucose (pH 7.4 with CsOH). The bath solution contained (mm): 2 MgCl2, 1 CaCl2, 5 KCl, 114 isethionate, 5 Hepes, 5 glucose and 20 sucrose (pH 7.4 with NaOH).
Single-channel kinetic analysis
Single-channel kinetic analysis was performed using PAT version 7.0 software (Strathclyde Electrophysiology Software, University of Strathclyde, UK). Single-channel open probability (Po) was obtained from the all-point amplitude histogram. The 50 % amplitude threshold method was used for event detection in all experiments except those using Br− or NO3− pipette solution. The permeability of Br− and NO3− is higher than that of Cl− (Dawson et al. 1999); therefore, the single-channel amplitude with Br− or NO3− pipette solution is smaller than that with Cl− due to a smaller electrochemical driving force. Because of the resulting low signal-to-noise ratio, a 30 % closed level threshold was used instead of 50 % for event detection in experiments involving Br− or NO3−. The open time (τo) and the closed time (τc) constants were estimated by fitting the corresponding dwell time histogram with single exponential functions. Both τo and τc were corrected according to Colquhoun & Sigworth (1995) using the equations:
where μT and μS are measured mean open time and closed time respectively, μX and μY are corrected mean open time and closed time respectively, and ξ is the dead time of our recording, which was estimated to be 3 ms with 100 Hz filtering. Correction was performed using a C language program written by Dr Fei Wang (University of Missouri). The percentage of missed events was estimated using the equation: 1 - exp(-ξ/μS) (Colquhoun & Sigworth, 1995). Data from different patches were averaged and results are expressed as means ±s.e.m.; n indicates the number of patches. Student's t tests were performed using SigmaPlot (version 4.0) and effects were considered significant if P < 0.05.
To validate this correction method, we simulated single-channel data using the program developed by Dr Laszlo Csanady (Csanady, 2000). An opening rate of 100 s−1 and a closing rate of 40 s−1 were used to generate single-channel data at 1000 Hz corner frequency. The simulated data were then filtered at 500 or 100 Hz using the same program package (Csanady, 2000). Simulated single-channel data at these three filtering frequencies were analysed and the kinetic parameters were corrected according to the method described above. Six simulated single-channel traces were analysed. The corrected mean open times were 24.33 ± 0.37 ms for 1000 Hz filtering, 24.77 ± 0.47 ms for 500 Hz filtering and 21.79 ± 0.65 ms for 100 Hz filtering. The corrected mean closed times were 10.07 ± 0.28 ms for 1000 Hz filtering, 10.19 ± 0.25 ms for 500 Hz filtering and 9.65 ± 0.47 ms for 100 Hz filtering. These results show that the kinetic parameters, especially the mean closed time, were similar when the same data were analysed and corrected at three different filtering frequencies.
Macroscopic K1250A-CFTR or wild-type CFTR currents in cell-attached patches were elicited with 10 μm forskolin. The steady-state currents were filtered with a built-in 4-pole Butterworth filter at 1 or 5 kHz and stored on videotape. Data were played back without further filtering and captured onto a hard disk at a sampling rate of 2 or 10 kHz, respectively, for stationary noise analysis. At least 60 s of K1250A-CFTR current recordings or 180 s of wild-type CFTR current recordings were fast Fourier transformed to generate noise spectra that were further analysed in a bandwidth of 9.7-600 Hz for K1250A-CFTR or 3.9-800 Hz for wild-type CFTR with Igor software. Data from K1250A-CFTR were fitted with a single Lorentzian function to estimate the Lorentzian parameters:
where fc is the corner frequency, S0 is the zero-frequency asymptote, S1 is non-specific basal noise and S(f) is the spectral density at the frequency f. Data from wild-type CFTR were fitted with the sum of two Lorentzian components:
where S0 and S1 are the zero-frequency asymptotes corresponding to fc1 and fc2, respectively, and S2 is the basal noise.
Voltage-dependent fast flickery block of the K1250A-CFTR channels
It has been shown previously that in cell-attached patches, the inward CFTR current shows more flickering events than the outward current, and that the number of these fast flickers is dramatically reduced upon patch excision (Haws et al. 1992; Fischer & Machen, 1994, 1996). The flickering block of K1250A-CFTR channels shares these two characteristic features with wild-type channels. Figure 1A shows that the fast flickery events present in a cell-attached patch from NIH3T3 cells stably expressing K1250A-CFTR channels were reduced in frequency upon patch excision. This decrease of fast flickers was reflected in an increase of the Po after patch excision (0.93 before excision and 0.99 after excision). Since most of the flickers were lost upon patch excision, we limited our studies of the flickery block to the cell-attached configuration, where sufficient flickery events can be collected for dwell time analysis within a reasonable recording duration. Figure 1B shows representative dwell time histograms constructed from traces at -70 mV (-Vp). Both open time and closed time distributions could be fitted with single exponential functions, suggesting that this fast flickery block can be described by a simple two-state model:
Here, α is the apparent on rate of the blocker and β represents the off rate. According to this scheme, the mean open time (τo) should be equal to 1/α and the mean closed time (τc) is 1/β.
The fast flickery block of K1250A-CFTR channels also showed clear voltage dependence when recorded in the cell-attached configuration. Figure 2A shows representative single-channel current traces at different potentials (-Vp). It appears that when the membrane patch was depolarized (i.e. -Vp= 50 mV), very few flickers were observed, whereas there were more fast flickering events at the hyperpolarizing potentials (-Vp= -50 to -130 mV), a feature noted previously for wild-type CFTR (Gray et al. 1989; Haws et al. 1992; Fisher & Machen, 1996). The open probabilities (Po) at different voltages were calculated from all-point amplitude histograms and the mean data are shown in Fig. 2B (•). As expected, Po was reduced as the membrane was hyperpolarized. The relationship between Po and -Vp could be fitted with a Boltzmann function (continuous line in Fig. 2B), which is consistent with the simple two-state model proposed in Scheme 1, with the membrane potential affecting the transition rate(s) between these two states.
The off rate, but not the on rate, of fast flickers is voltage dependent
Previous studies have shown that fast flickers in wild-type CFTR channels appear to be voltage dependent (Gray et al. 1989; Haws et al. 1992; Fisher & Machen, 1996), but it is unclear whether this voltage dependence resides in the on rate or off rate (or both) of the blocker. By merely looking at the single-channel current traces shown in those reports, it appears that more flickers are present at hyperpolarizing membrane potentials, suggesting a voltage-dependent on rate (1/τo). In the present study, similar observations were made with the K1250A-CFTR channel. As shown in Fig. 2A, it seems that as the membrane was hyperpolarized, more flickers appeared, yielding a picture of voltage-dependent τo. However, the appearance of these current traces recorded at different voltages could be misleading because of the fast nature of the blockade and the limited bandwidth of the recording system. It is possible that a significant number of the brief events have been filtered out. These missed events could affect considerably the appearance of the recorded traces as well as the accuracy of kinetic analysis. For example, if the duration of fast flickers becomes shorter as the membrane is depolarized, more flickers will be ‘missed’, resulting in an appearance of longer open times at more depolarizing potentials. In this study, we took into account the missed events resulting from filtering and corrected τo and τc using equations developed by Colquhoun & Sigworth (1995) (see Methods). Figure 3A and B (•) shows the relationship between corrected time constants and -Vp. Surprisingly, it is τc, not τo, that showed voltage dependence. Although mean data from several cells showed considerable cell-to-cell variation, for the same cell, τc at a more hyperpolarizing voltage was consistently longer than that at a more depolarizing voltage (P < 0.01), whereas τo did not change significantly with voltage (Fig. 3A and B, insets).
We used both the corrected and uncorrected time constants to calculate the Po based on Scheme 1 (Po=τo/(τo+τc)). The calculated Po values from uncorrected τo and τc (Fig. 2B, ○) were similar to those estimated from all-point amplitude histograms (Fig. 2B, •). Since the Po estimated from all-point amplitude histograms is model independent, a close match between the Po values calculated from the uncorrected time constants and the Po values estimated from all-point amplitude histograms provides additional evidence that Scheme 1 is adequate to explain our results. On the other hand, the Po values calculated from corrected time constants were consistently smaller than uncorrected Po values, as expected because of the missed events (data not shown).
The off rate is affected by trans-anion concentration
A voltage-dependent off rate and a voltage-independent on rate (Fig. 3A and B, •) suggest that the classic Woodhull model (Woodhull, 1973), which places the binding site of the blocker inside the electric field, may not be adequate to explain the flickery block of CFTR (see Discussion for details). Furthermore, it seems inappropriate to apply the Woodhull model to a channel with a multi-ion pore such as the CFTR channel (Tabcharani et al. 1993). For multi-ion pores, a voltage-dependent off rate of the blocker can be caused by a voltage-dependent binding of permeant ions from the opposite side of the membrane (e.g. MacKinnon & Miller, 1988). The electrostatic repulsion between the permeant ion and the blocker in the pore enhances the dissociation of the blocking ion. This trans-enhancement effect predicts that the off rate is sensitive to changes in external permeant ion concentrations. To test this hypothesis, we replaced Cl− ions in the pipette (external) solution with aspartate (0 mm Cl− solution), an impermeant anion, or Br− (154 mm Br− solution) or NO3− (154 mm NO3− solution), both permeant anions for CFTR. Figure 4 shows the current traces recorded with 0 mm Cl− pipette solution at different voltages. Compared to current traces in Fig. 2A (154 mm Cl− pipette solution), more flickery block events were observed at a given voltage in the absence of external Cl− and the voltage dependence of the fast flickery block was diminished significantly. With 0 mm external Cl−, τo (Fig. 3A, ○) showed no obvious voltage dependence similar to that with high external Cl−. However, τc (-50 to -90 mV) was prolonged after removal of external Cl−. For instance, at -50 mV τc was 8.1 ± 0.4 ms n = 10 with 0 mm external Cl−, which is ≈2-fold longer than that with a high external Cl− (τc= 4.4 ± 0.6 ms, n= 5) (Fig. 3C). Moreover, τc exhibited little voltage dependence in the absence of external Cl− (Fig. 3B, ○). On the other hand, when permeant anions, such as Br− or NO3−, were substituted for external Cl− ions, τc at -50 mV was 3.8 ± 0.3 ms n = 4 and 4.0 ± 0.3 ms n = 4, respectively (Fig. 3C), neither of which was significantly different from that with a high external Cl−. These results strongly suggest that the trans-enhanced dissociation may account for the voltage-dependent off rate of this flickery block.
Voltage-dependent block of the whole-cell K1250A-CFTR current
Our single-channel recordings were filtered at 100 Hz (dead time = 3 ms) in order to obtain a reasonable signal-to-noise ratio for dwell time analysis. Since both τo and τc of the flickery block were several to tens of milliseconds at more hyperpolarizing voltages, we predicted that this block should produce resolvable whole-cell current relaxation upon voltage jump. Because of the macroscopic nature of the signal in whole-cell recordings, a much higher bandwidth (5 kHz, dead time = 60 μs) can be used. Scheme 1 predicts a steady-state fraction of unblock =β/(α+β) (or fraction of block =α/(α+β)), and a relaxation time constant (τr) upon voltage jump = 1/(α+β). Under symmetrical Cl− conditions with 125 mm Cl− in both the external and the internal solutions, the whole-cell instantaneous I-V relationship of K1250A-CFTR was linear whereas the steady-state I-V relationship was somewhat outwardly rectifying (data not shown). The fraction of block under this symmetrical Cl− condition was small (18.0 ± 6.1 %, n= 5, at -100 mV) while the current relaxation was fast; both features preclude accurate estimation of τr. However, the hypothesis of trans-enhanced dissociation predicts a slower current relaxation and a larger fraction of block when external [Cl−] is lowered. We therefore studied whole-cell current relaxation under an asymmetrical Cl− condition with 11 mm external Cl− and 125 mm internal Cl− (the fraction of block was 29.2 ± 2.9 %, n= 4, at -100 mV).
Figure 5A shows net whole-cell CFTR current traces at different membrane potentials (Vm). At more positive Vm, the current was essentially time independent. However, current relaxation was clearly observed when the voltage was stepped from 0 mV to more negative Vm (e.g. -100 mV). The appearance of this current relaxation is indicative of the existence of a voltage-dependent process. Figure 5B shows both the instantaneous (○) and steady-state (•) I-V relationships constructed from the current traces shown in Fig. 5A. The instantaneous I-V relationship showed inward rectification as expected for an asymmetrical [Cl−], while the steady-state I-V relationship exhibited outward rectification (especially at more negative Vm). The difference between the instantaneous current (Iinst) and the steady-state current (Iss) is due to a voltage-dependent block (Scheme 1).
To better resolve the current relaxation, we used a revised voltage-step protocol. Figure 5C shows a current trace in response to a voltage step to 100 mV for 100 ms from a holding potential of 0 mV (to relieve the block) and subsequently to -100 mV (to develop the block). The τr at -100 mV, estimated by fitting the time course with a single exponential function, was 11.3 ± 1.5 ms n = 4. A different voltage-step protocol was used to examine the time course of relief of the block. The Vm was first stepped from 0 to -100 mV for 100 ms to reach a steady-state block, and then stepped to +100 mV to relieve the block. However, the current relaxation was apparently too fast to be resolved completely and was mostly masked by an incomplete subtraction of the capacitance current (Fig. 5D). Only a minor fraction of the current relaxation can be seen as a small bend at the initial part of the current trace upon voltage jump to +100 mV. Although we cannot exclude the existence of an ultra-fast kinetic component, buried in the capacitance transient, in our macroscopic current relaxation analysis (see below), we conclude that both the fraction of block and τr are voltage dependent, and that the current relaxation is much slower at -100 mV than that at +100 mV.
Noise analysis of macroscopic K1250A-CFTR and wild-type CFTR currents
Ideally, one should analyse single-channel kinetics on data that are lightly filtered in order to obtain more reliable kinetic parameters. However, to obtain a reasonable signal-to-noise ratio for channels with a fairly small single-channel conductance, such as CFTR channels (e.g. < 0.25 pA at -50 mV in Fig. 2A), we have to filter the data heavily (e.g. 100 Hz in the present study). We performed noise analysis on macroscopic currents as an independent way to confirm our single-channel results. Macroscopic K1250A-CFTR currents were evoked by 10 μm forskolin in cell-attached patches from NIH3T3 cells stably expressing K1250A-CFTR. Steady-state current recordings at various voltages were collected and fast Fourier transformed to generate noise spectra. Figure 6 shows representative noise spectra of K1250A-CFTR macroscopic currents recorded at -50 mV in a cell-attached patch. Data were filtered at either 1 kHz (Fig. 6A) or 5 kHz (Fig. 6B). The corresponding noise spectra were fitted with single Lorentzian functions and the resulting corner frequencies (fc) were similar under these two conditions (30.01 Hz for 1 kHz and 35.20 Hz for 5 kHz). These values are similar to the predicted fc based on single-channel time constants for fast flickers (see below). Moreover, we did not find evidence for higher frequency Lorentzian components in either of the spectra. These results indicate that our noise analysis, even at 1 kHz filtering, indeed detected the kinetic events related to flickery block. Therefore, we used 1 kHz filtering for all the following noise analysis experiments.
Since fc= (1/τo+ 1/τc)/2π, we can predict fc from the corrected τo and τc derived from single-channel kinetic analysis. Our single-channel kinetic data suggest that τc decreases, while τo remains unchanged, with membrane depolarization in the presence of high external Cl−; therefore, we expect fc should increase when the membrane is depolarized. Furthermore, this voltage dependence of fc will be abolished when all the external Cl− ions are replaced by the impermeant anion aspartate since τc becomes voltage independent under this condition. In addition, at the same voltage, fc should be smaller when external Cl− ions are replaced by impermeant anions, since τc is longer under this condition. Data from our noise analysis essentially confirmed these predictions. With 154 mm external Cl−, fc increased with membrane depolarization (Fig. 7A and C). With 0 mm external Cl−, fc exhibited little, if any, voltage dependence (Fig. 7B and C). At the same voltage, fc in the presence of 154 mm external Cl− was higher than that in the absence of external Cl− (Fig. 7C). Figure 8 shows that values of fc obtained directly from noise analysis (•) are comparable to those calculated from single-channel kinetic parameters (○).
Macroscopic noise analysis also allowed us to examine the voltage dependence of flickery block in wild-type CFTR. Figure 9A shows a representative current noise spectrum at -50 mV. The noise spectrum was fitted with the sum of two Lorentzian components: a slow component, fc1, that was less than 4 Hz and a fast component, fc2, of 38.96 Hz. Although fc1 cannot be estimated accurately because of a limited frequency range for curve fitting, the low frequency nature of this component suggests that it may reflect the slow ATP-dependent gating events reported previously (Fischer & Machen, 1994; Hwang et al. 1997). On the other hand, fc2 was in the frequency range of the fast flickery events seen in K1250A-CFTR channels. Furthermore, fc2 showed a similar voltage dependence to that observed in K1250A-CFTR channels (Fig. 9B). These results suggest that the mechanism of the fast flickery block of K1250A-CFTR channels is probably applicable to wild-type CFTR channels.
In the current study, we investigated the underlying mechanism of intrinsic voltage-dependent flickery block of CFTR channel current. Single-channel kinetic analysis indicated that the off rate, but not the on rate, of the blocker was voltage dependent and this off rate can be affected by the concentration of permeant trans-anions. Results from macroscopic current noise analysis further confirmed the conclusion drawn from the single-channel kinetic analysis. Thus a significant portion of the voltage dependence may result from electrostatic interactions between permeant ions and the blocker in the conducting pore.
Previous studies indicate that the fast flickery block of CFTR is voltage dependent (Gray et al. 1989; Haws et al. 1992; Fischer & Machen, 1996). The appearance of single-channel current traces, reported by these authors and also shown in this study (Fig. 2A), gives one the impression that there are more flickery events at more hyperpolarizing potentials, suggesting an apparent voltage dependence of the on rate. However, these single-channel recordings, obtained under heavy filtering, may have missed a significant number of brief events. For example, at a filtering frequency of 100 Hz, we estimate that ≈20 % of the flickers are missed at -130 mV (-Vp). The percentage of missed events increases to ≈30 % as the membrane potential is depolarized to -50 mV (-Vp). Therefore, the appearance of single-channel current traces could be misleading and proper corrections of the missed events are necessary.
Our results indicate that after correction for the missed events, τo of this fast blockade showed little voltage dependence while τc was voltage dependent (Fig. 3A and B). When these results are applied to Scheme 1, we can conclude that dissociation, but not association, of the blocker constitutes a voltage-dependent step. This conclusion derived from single-channel kinetic analysis is further supported by whole-cell recordings performed with less filtering (i.e. 5 kHz). Using voltage-step protocols, we were able to observe macroscopic current relaxation caused by voltage-dependent block (Fig. 5). τr at -100 mV (τr(-100)) was slow and thus can be estimated whereas τr at +100 mV (τr(+100)) was too fast to be quantified. However, the initial part of the current relaxation can be reconstructed by estimating the theoretical value of Iinst at +100 mV (see Fig. 5 legend). Using this estimated value for Iinst, we predict that τr(+100) is less than 1 ms. Therefore, τr(-100) is at least 10-fold larger than τr(+100). On the other hand, the steady-state Po at +100 mV (Vm) was ≈25 % higher than that at -100 mV (Fig. 2B). Since τr= 1/(α+β) and Po=β/(α+β), we conclude that the off rate at -100 mV is at least 10-fold slower than that at +100 mV. τc at -100 mV (τc(-100)) was ≈8 ms (Fig. 3); therefore, the closed time constant at +100 mV (τc(+100)) is probably < 1 ms. Thus, after a steady-state block is achieved at -100 mV, stepping the voltage to +100 mV destabilizes the blocker in the pore. The resulting fast off rate of the blocker allows fast recovery of the blocked current. Since τc(+100) is estimated to be < 1 ms, more than 95 % of the flickers at +100 mV will be missed should a single-channel recording be made at a corner frequency of 100 Hz!
Results from noise analysis of macroscopic K1250A-CFTR current further verify our conclusions based on single-channel kinetic studies. The fc obtained from noise analysis matches the fc calculated from corrected τo and τc over the range of voltages tested (Fig. 8). Results from noise analysis also showed similar voltage dependence and Cl− sensitivity of the fast flickery block as predicted from single-channel kinetic parameters (Fig. 7). In short, the agreement between the results from noise analysis and those from single-channel kinetic analysis provide strong evidence that the kinetic parameters obtained from single-channel analysis are valid.
One concern of using the CFTR mutant K1250A-CFTR to study the mechanism of fast flickery block is whether K1250A-CFTR and wild-type CFTR share the same mechanism of voltage-dependent flickery block. To address this issue, we performed noise analysis on wild-type CFTR currents as well. Since there are significant long closed events in addition to the fast flickery events in wild-type CFTR channels, the power spectrum was fitted by the sum of two Lorentzian components (Fig. 9A). Since fc2 of wild-type CFTR channels shows a similar voltage dependence to that of K1250A-CFTR (Fig. 9B), we believe that the voltage-dependent mechanism proposed for the fast flickery block of K1250A-CFTR channels is probably also applicable to wild-type CFTR channels. It is worth noting, however, that fc2 of wild-type CFTR channels is slightly higher than the fc of the fast flickers seen in K1250A-CFTR channels. The mechanism responsible for this difference is unclear, but it would be interesting to investigate how a mutation in NBD2 changes the kinetics of the fast flickers in the future.
The voltage-dependent off rate, but not the on rate, of CFTR fast flickery block reported here is reminiscent of voltage-dependent block of several potassium channels by tetraethylammonium derivatives (Armstrong, 1971), Na+ (Yellen, 1984) and charybdotoxin (MacKinnon & Miller, 1988). In those reports, the voltage-dependent off rate was interpreted as entirely or partly resulting from ion-ion interactions in the permeation pathway. Potassium channels are known to have multi-ion pores (Hodgkin & Keynes, 1955; Doyle et al. 1998). Therefore, all or part of the blocker and potassium ions can enter the aqueous pathway from the opposite sides of the channel and occupy the pore simultaneously. Since the steady-state occupancy of the potassium ion binding site(s) is affected by the transmembrane potential, the strength of the electrostatic repulsion between potassium and the positively charged blocker is voltage dependent. This then results in a voltage-dependent off rate of the blocker. Furthermore, since the steady-state occupancy of the ion-binding site in the pore can be affected by the concentration of permeant ions in the bulk solution, this model predicts that the off rate of the blocker should also depend on the concentration of permeant trans-ions. Indeed we observed that the off rate of the flickery block in CFTR was slower and showed little voltage dependence when permeant trans-anions were replaced by impermeant anions (Fig. 3B and C and 7B and C). Therefore, our data on the flickery block of CFTR are consistent with a mechanism that can place the unknown blocker and permeant Cl− ions in the aqueous pore simultaneously, further supporting the notion that CFTR has a multi-ion pore (Tabcharani et al. 1993; cf. Smith et al. 1999).
CFTR blockade by exogenously applied blockers has been studied extensively in excised inside-out patches expressing wild-type CFTR. Most exogenously applied blockers, such as sulfonylureas (Sheppard & Welsh, 1992; Venglarik et al. 1994; Schultz et al. 1996; Sheppard & Robinson, 1997), disulfonic stilbenes (Linsdell & Hanrahan, 1996a) and large anions such as gluconate (Linsdell & Hanrahan, 1996b), can only block CFTR channels from the intracellular side. Other blockers such as arylaminobenzoates can block CFTR channels from either side (McCarty et al. 1993). Most of these studies interpret the steady-state voltage dependence of the block using the Woodhull model (Woodhull, 1973), which assumes that the blocker's binding site is located in the electrical field; therefore, the blocker can enter and exit the channel pore in a voltage-dependent manner. From the relationship between voltage and the degree of block, estimated electrical distances range from 0.16 to 0.60 measured from the inside of the membrane (McCarty et al. 1993; Overholt et al. 1995; Linsdell & Hanrahan, 1996a,b; Sheppard & Robinson, 1997), suggesting that the blocker binding site is located fairly deep in the pore. Our results of fast flickery block are consistent with the idea that the internal blocker physically occludes the pore, but due to a lack of apparent voltage-dependent on rate, this blocker does not necessarily reside deep in the pore. This idea is further supported by the results from experiments with 0 mm external Cl−. Presumably in the absence of external Cl−, any observed voltage dependence should be purely due to the effect of membrane potential on the blocker if the blocker's binding site is inside the electrical field. However, we observed little voltage dependence in the absence of external Cl− (Fig. 3B, ○). It is worth noting that τc with high external Cl− saturated around -110 mV to -130 mV and that this saturating value of τc approximated the τc with 0 mm Cl−. This result can be explained by the hypothesis that strong hyperpolarizing voltages deplete the CFTR pore of resident Cl− even in the presence of Cl− in the bulk solution, resulting in a dissociation of the blocker that is not affected by voltage or the presence of trans-anions. Therefore, the blocker acquires most of the voltage dependence of its off rate through electrostatic interaction with a neighbouring Cl− in the pore. Several other groups (Hanrahan & Tabcharani, 1990; Linsdell & Hanrahan, 1996b;Sheppard & Robinson, 1997) also report that the blockade by exogenously applied CFTR channel blockers, such as Hepes, gluconate and glibenclamide, is sensitive to extracellular Cl− concentration. However, the Woodhull model, which is based on a single binding site, is used to explain the block in these studies. It would be interesting to see whether the novel mechanism of voltage-dependent block of CFTR channels we proposed in the present study may also apply to those exogenous blockers.
It has been suggested that the CFTR pore is structurally asymmetrical with a wide internal vestibule but a shallow external mouth (Hwang & Sheppard, 1999; cf. Cheung & Akabas, 1997). We propose that the internal blocker characterized in the current work occludes the surface of the internal vestibule without moving deep into the pore (Fig. 10). Although the molecular nature of the internal blocker is unclear, this blocker is likely to be a large molecule because it is not readily diffusible (Fisher, 1997), and the block is observed in whole-cell recordings. Alternatively, this internal blocker may be part of the channel itself since a significant number of fast flickers are still observed after the membrane patch is excised. It is likely that the binding site of the internal blocker is also a Cl− binding site, since the number of flickers decreased dramatically when millimolar SCN− was introduced to the cytoplasmic side of the channel in excised inside-out patches (Z. Zhou & T.-C. Hwang, unpublished observations), suggesting that SCN−, a Cl− surrogate, and the internal blocker may compete for a common binding site.
In this study, we took the advantage of the CFTR mutant K1250A-CFTR, which can be ‘locked’ open for minutes to allow more accurate kinetic analysis of the flickery block of CFTR channels. The strategy of using mutant channels to tackle specific biophysical questions of ion permeation and blockade has been used widely in the ion channel field. For instance, a Shaker potassium channel mutant with part of the N-terminus (the ‘ball’) removed has been used successfully to provide valuable structure-function information about the potassium channel pore (e.g. MacKinnon & Yellen, 1990; Yellen et al. 1994). We believe future studies of channel blockade in K1250A-CFTR channels by exogenously applied blockers could provide useful information on the mechanism of CFTR blockade as well as on the structure of the CFTR pore.
We are grateful to Dr Kevin Gillis and Dr Allan Powe for helpful comments. This research was supported by the National Institutes of Health and the Cystic Fibrosis Foundation. Dr Zhou is supported by a postdoctoral fellowship from the Molecular Biology Program, University of Missouri - Columbia and the American Heart Association, Missouri Affiliate.