Mechanism of Glibenclamide Inhibition of Cystic Fibrosis Transmembrane Conductance Regulator Cl Channels Expressed in a Murine Cell Line

Authors

  • David N. Sheppard,

    Corresponding author
    1. Human Genetics Unit, Department of Medicine and Department of Biochemistry, University of Edinburgh, Hugh Robson Building, George Square, Edinburgh EH8 9XD, UK
    • To whom correspondence should be addressed at the University of Edinburgh, Human Genetics Unit, Department of Medicine, Molecular Medicine Centre, Ground Moor, Western General Hospital, Edinburgh EH4 2XU, UK.

    Search for more papers by this author
  • Katherine A. Robinson

    1. Human Genetics Unit, Department of Medicine and Department of Biochemistry, University of Edinburgh, Hugh Robson Building, George Square, Edinburgh EH8 9XD, UK
    Search for more papers by this author

Abstract

  • 1The sulphonylurea drug glibenclamide is a widely used inhibitor of the cystic fibrosis transmembrane conductance regulator (CFTR). To investigate how glibenclamide inhibits CFTR, we studied CFTR Cl channels using excised inside-out membrane patches from cells expressing wild-type human CFTR.
  • 2Addition of glibenclamide (10–100 μM) to the intracellular solution caused a concentration-dependent decrease in the open time of CFTR Cl channels, but closed times did not change. This suggests that glibenclamide is an open-channel blocker of CFTR.
  • 3Glibenclamide is a weak organic acid. Acidification of the intracellular solution relieved glibenclamide inhibition of CFTR, suggesting that the anionic form of glibenclamide inhibits CFTR.
  • 4To begin to identify the glibenclamide binding site in CFTR, we investigated whether glibenclamide competes with either MgATP or Cl ions for a common binding site. Glibenclamide inhibition of CFTR was unaffected by nucleotide-dependent stimulation of CFTR, suggesting that glibenclamide and intracellular MgATP interact with CFTR at distinct sites.
  • 5Glibenclamide inhibition of CFTR was voltage dependent and enhanced when the external Cl concentration was decreased. The data suggest that glibenclamide and Cl ions may compete for a common binding site located within a large intracellular vestibule that is part of the CFTR pore.

Glibenclamide is a sulphonylurea drug that is widely used to treat non-insulin-dependent diabetes mellitus (Loubatières, 1977). At nanomolar concentrations it binds to the sulphonylurea receptor (SUR) of pancreatic β-cells to cause the inhibition of ATP-sensitive K+ channels (KATP channels) and promote insulin secretion (Sturgess, Kozlowski, Carrington, Hales & Ashford, 1988). Higher concentrations of glibenclamide inhibit KATP channels in other tissues (Ashcroft & Ashcroft, 1990), including skeletal and cardiac muscle (Davies, Standen & Stanfield, 1991; Ripoll, Lederer & Nichols, 1993), and Cl channels in epithelia (Wangemann et al. 1986), including the cystic fibrosis transmembrane conductance regulator (CFTR; Riordan et al. 1989; Sheppard & Welsh, 1992). Interestingly, KATP channels share other functional properties with CFTR Cl channels, the most notable of which is regulation by intracellular ATP (Ashcroft & Ashcroft, 1990; Anderson, Berger, Rich, Gregory, Smith & Welsh, 1991a). However, for both channels the effect of ATP is complex. Once phosphorylated by protein kinase A (PKA), ATP hydrolysis opens and closes CFTR Cl channels (Anderson et al. 1991a; Hwang, Nagel, Nairn & Gadsby, 1994; Carson, Travis & Welsh, 1995a; Li et al. 1996). In contrast, Mg2+-free ATP and non-hydrolysable ATP analogues inhibit KATP channels (Dunne, Illot & Petersen, 1987; Dunne, West-Jordan, Abraham, Edwards & Petersen, 1988), although after channel run-down ATP hydrolysis may reactivate KATP channels (Ohno-Shosaku, Zünkler & Trube, 1987).

A molecular explanation for the functional similarities between CFTR and KATP channels has emerged following the identification of the genes that encode KATP channels in pancreatic β-cells, and cardiac and skeletal muscle (Aguilar-Bryan et al. 1995; Inagaki et al. 1995; Sakura, Ämmälä, Smith, Gribble & Ashcroft, 1995; Inagaki et al. 1996). These KATP channels are composed of at least two subunits, an inwardly rectifying K+ channel (Kir6.2; Inagaki et al. 1995; Sakura et al. 1995) and a SUR that binds sulphonylurea drugs with either high affinity (SUB1; pancreas; Aguilar-Bryan et al. 1995) or low affinity (SUR2; cardiac and skeletal muscle; Inagaki et al. 1996). Like CFTR, SUR is a member of the ATP-binding cassette (ABC) transporter family (Hyde et al. 1990). These transporters are constructed from two motifs that each contain a membrane-spanning domain (MSD) and a nucleotide-binding domain (NBD). Typically, each MSD contains six putative transmembrane segments that span the lipid bilayer. However, SUR is predicted to have nine putative transmembrane segments in MSD1, but only four in MSD2 (Aguilar-Bryan et al. 1995). Despite differences in predicted topology, CFTR and SUR are likely to share some features in their sulphonylurea binding sites. For both KATP and CFTR channels, when sulphonylureas are added to the extracellular solution, glibenclamide inhibition develops slowly; inhibition is reversible with tolbutamide, but not with glibenclamide; and glibenclamide is a more potent inhibitor than tolbutamide (Gillis, Gee, Hammoud, McDaniel, Falke & Misler, 1989; Sheppard & Welsh, 1992). To understand better how glibenclamide interacts with CFTR, we investigated how glibenclamide inhibits CFTR Cl channels in excised inside-out membrane patches from cells expressing wild-type human CFTR.

METHODS

Cell culture

For this study we used mouse mammary epithelial cells (C127 cells) that had been stably transfected with wild-type human CFTR using a bovine papilloma virus-based expression vector (Marshall et al. 1994). These cells were a generous gift of Drs A. E. Smith and S. H. Cheng (Genzyme, Framingham, MA, USA). Cells were grown in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum, and 200 mg ml−1 neomycin (all from Life Technologies Ltd, Paisley, UK), at 37 °C in a humidified atmosphere of 5% CO2. For experiments using excised inside-out membrane patches, cells were seeded onto glass coverslips and used within 48 h.

Electrophysiology

CFTR Cl channels were recorded in excised inside-out membrane patches as previously described (Hamill, Marty, Neher, Sakmann & Sigworth, 1981; Sheppard, Rich, Ostedgaard, Gregory, Smith & Welsh, 1993). An Axopatch 200A patch-clamp amplifier (Axon Instruments Inc.) was used for voltage clamping and current amplification and the pCLAMP software package (v. 6.02; Axon Instruments Inc.) was used for data acquisition and analysis. All experiments were conducted at 37 °C using a temperature-controlled microscope stage (Brook Industries, Lake Villa, IL, USA). The established sign convention was used throughout; currents produced by positive charge moving from intra- to extracellular solutions (anions moving in the opposite direction) are shown as positive currents.

Patch pipettes were fabricated from thin-walled borosilicate glass capillary tubing (Clark Electromedical Instruments, Reading, UK) using a two-stage vertical pipette puller (David Kopf Instruments, model 750; Clark Electromedical Instruments), coated with Sylgard (Merck Ltd, Lutterworth, UK), and polished using a microforge. Patch pipettes had resistances of 10–20 MΩ when filled with the low-Cl pipette solution.

The pipette (extracellular) solution contained (mM): 140 N-methyl-d-glucamine (NMDG), 140 aspartic acid, 5 CaCl2, 2 MgSO4 and 10 Tes; pH adjusted to 7.3 with Tris; [Cl], 10 mM. The bath (intracellular) solution contained (mM): 140 NMDG, 3 MgCl2, 1 CsEGTA and 10 Tes; pH adjusted to 7.3 with HCl; [Cl], 147 mM; [Ca2+]free, < 10−8 M. For pH experiments, 5 mM Bis-Tris + 5 mM Trizma base replaced 10 mM Tes in the intracellular solution. To ensure that the Cl concentrations of the pH 7.30 and pH 6.30 solutions were identical, both solutions were first titrated to pH 7.30 with HCl before the pH 6.30 solution was further titrated to pH 6.30 with H2SO4. At -50 mV, these buffers were without effect on single-channel open probability (P0; n= 4; P > 0.4).

After excision of membrane patches from C127 cells, CFTR Cl channels were activated by the addition of the catalytic subunit of PKA (75 nM) and MgATP (0.88 mM) to the intracellular solution. The MgATP concentration was subsequently reduced to 0.26 mM (∼EC50 for CFTR Cl channel activation by intracellular MgATP; Anderson et al. 1991a). To minimise channel run-down, PKA was maintained in the intracellular solution for the duration of experiments, in contrast to previous studies. Unless otherwise indicated, voltage was -50 mV.

Stock solutions of glibenclamide were prepared in dimethyl sulphoxide and stored at -20 °C until required. Stock solutions were diluted in intracellular solution to achieve final concentrations at the time of use. Dimethyl sulphoxide (0.4% v/v) did not affect the activity of CFTR Cl channels (n= 4).

To investigate the effect of glibenclamide on CFTR Cl channels, we only used membrane patches that contained ≤ 4 active channels. The number of channels in each patch was determined from the maximum number of simultaneous channel openings observed during the course of an experiment that typically lasted 30–90 min and which included interventions that significantly stimulate Po (e.g. MgATP (0.88–2.27 mM) + PKA (75 nM)). For current amplitude, Po and single-channel kinetic measurements, drug interventions of 4 min duration were compared with the average of pre- and postintervention control periods each of 4 min duration to compensate for any channel run-down during an experiment. For pH experiments, drug interventions of 4 min duration were compared with the average of pre- and postintervention control periods each of 4 min duration at pH 7.30 and pH 6.30 within a single experiment. To examine whether glibenclamide inhibition of CFTR was voltage dependent, the voltage was stepped from -100 to +100 mV in 20 mV increments of 30 s duration in the presence and absence of glibenclamide, and Po measured at each step.

Single-channel currents were initially recorded on digital audiotape using a digital tape recorder (Biologic Scientific Instruments, model DTR-1204; Intracel Ltd, Royston, UK) at a bandwidth of 10 kHz. On playback, records were filtered with an eight-pole Bessel filter (Frequency Devices, model 902LPF2; SCENSYS Ltd, Aylesbury, UK) at a corner frequency of 500 Hz and acquired using a DigiData 1200 interface (Axon Instruments Inc.) and pCLAMP at a sampling rate of 5 kHz. Lists of open and closed times were created using a half-amplitude crossing criterion for event detection. Transitions < 1 ms in duration were excluded from the analysis. Single-channel open- and closed-time histograms were plotted with logarithmic x-axes with 10 bins decade−1, and the maximum likelihood method was used to fit a one-, two- or three-component exponential function to the data. The use of other combinations of ranges and bin widths gave similar results. To determine which component function fitted best, the log-likelihood ratio test was used and was considered statistically significant at a value of 2.0 or greater (Winter, Sheppard, Carson & Welsh, 1994).

Reagents

The catalytic subunit of PKA was obtained from Promega Ltd (Southampton, UK). ATP (disodium salt), Bis-Tris, glibenclamide, NMDG, pyrophosphate (tetrasodium salt), Tes and Trizma base were purchased from Sigma-Aldrich Co. Ltd. All other chemicals were of reagent grade.

Statistics

Results are expressed as means ±s.e.m. of n observations. To test for differences between groups, an analysis of variance (ANOVA) was used. To compare only two sets of data, we used Student's t test. Differences were considered statistically significant when P < 0.05. All tests were performed using SigmaStat (Jandel Scientific GmbH, Erkrath, Germany).

RESULTS

Glibenclamide alters the gating behaviour of CFTR

To investigate how glibenclamide inhibits wild-type human CFTR, we studied CFTR Cl channels in excised inside-out membrane patches from C127 cells stably expressing wild-type human CFTR. Figure 1 shows the effect of glibenclamide on the activity of a single CFTR Cl channel following phosphorylation by PKA. In the continued presence of PKA and MgATP, glibenclamide was added to the intracellular side of the membrane. Glibenclamide had no effect on single-channel current amplitude (Figs 1 and 2A).

Figure 1.

Effect of glibenclamide on the activity of a single CFTR Cl channel

Representative single-channel records from an excised inside-out membrane patch from a C127 cell stably expressing wild-type CFTR. Unless otherwise indicated, in this and subsequent figures, the following conditions were used: MgATP (0.26 mM) and the catalytic subunit of PKA (75 nM) were continuously present in the intracellular solution to which the indicated concentrations of glibenclamide were added. Voltage was -50 mV, and there was a Cl concentration gradient across the membrane (internal [Cl] = 147 mM; external [Cl] = 10 mM). Dashed lines indicate the closed channel state and downward deflections correspond to channel openings. For purposes of illustration, records were filtered at 500 Hz and digitized at 1 kHz.

Figure 2.

Effect of glibenclamide on single-channel current amplitude and Po

A, single-channel current amplitude histograms for the indicated interventions from the experiment shown in Fig. 1. In each case the closed channel amplitude is shown on the right. Linear x-axes with 20 bins decade−1 were used for all histograms and the continuous lines represent the fit of Gaussian distributions to the data. None of the current amplitudes measured in the presence of glibenclamide (Glib) differed significantly from control values (P > 0.2; n= 5–6). B, effect of glibenclamide concentration on Po. Data are means ±s.e.m.; n= 5–6 at each concentration. Voltage was -50 mV. The inset shows a Hill plot of the data. The continuous line is the fit of a first-order regression to the data.

However, glibenclamide did alter the gating behaviour of CFTR Cl channels. The pattern of gating of wild-type CFTR Cl channels is characterized by bursts of activity interrupted by brief flickery closures, separated by longer closures between bursts (Fig. 1, top trace). Visual inspection of single-channel records suggested that, as the concentration of glibenclamide increased, the gating behaviour of CFTR Cl channels changed in one of two ways: either the duration of bursts decreased, or the duration of bursts increased, but bursts were interrupted by numerous glibenclamide-induced closures. At 10 and 25 μM glibenclamide, the effect was small, but at 100 μM glibenclamide, it was dramatic (Fig. 1). Glibenclamide inhibition was readily reversible (Figs 1, 2A and 3).

Figure 3.

Open-time (A) and closed-time (B) histograms for a single CFTR Cl channel inhibited by glibenclamide

Representative histograms for the indicated interventions from a single experiment are shown. MgATP (2.27 mM) and the catalytic subunit of PKA (75 nM) were continuously present in the intracellular solution. Voltage was -50 mV and the membrane patch was bathed in symmetrical 147 mM Cl solutions. For open-time histograms, the continuous line is the fit of either a one- (Control, Wash) or a two-exponential function (10–100 μM). For closed-time histograms, the continuous line is the fit of either a two- (Control, Wash) or a three-exponential function (10–100 μM). Logarithmic x-axes with 10 bins decade−1 were used for both open- and closed-time histograms.

To quantify the effect of glibenclamide on the gating behaviour of CFTR, we measured Po. Figure 2B shows that, as the glibenclamide concentration increased, Po decreased. A Hill plot of the data is shown in Fig. 2B (inset). At -50 mV, the glibenclamide concentration causing half-maximal inhibition was 36 ± 3 μM (n= 6) and the Hill coefficient was 1.1 ± 0.1 (n= 6). These results compare well with the values we previously calculated using whole-cell data (Sheppard & Welsh, 1992), and those reported by other investigators (Tominaga, Horie, Sasayama & Okada, 1995; Schultz, DeRoos, Venglarik, Singh, Frizzell & Bridges, 1996).

Effect of glibenclamide on single-channel kinetics

To determine how glibenclamide decreased Po, we examined the effect of glibenclamide on the gating kinetics of phosphorylated wild-type CFTR in membrane patches that contained only a single channel. We increased the intra-cellular MgATP concentration to decrease the duration of long closures between bursts of channel activity and asked whether we could detect a new population of channel closures that represented channels blocked by glibenclamide. Consistent with our previous results, under control conditions open- and closed-time histograms were best fitted with one- and two-component functions, respectively (Fig. 3 and Table 1; Winter et al. 1994). However, in the presence of glibenclamide we observed a new population of open times described by a fast time constant (τO1) and a new population of closed times described by an intermediate time constant (τC2; Fig. 3 and Table 1).

Table 1. Open- and close-time constants
 [Glibenclamide] (μM)
 0102550100
  1. Open- and closed-time constants were measured at the indicated glibenclamide concentrations. Under control conditions, the behaviour of CFTR Cl channels is described by one open-time constant (τO2), and two closed time constants, one fast (τC1), and one slow (τC3). In the presence of glibenclamide two additional time constants are observed, a fast open-time constant (τO1) and an intermediate closed-time constant (τC2). The total time analysed for each concentration of glibenclamide is shown, and in each patch, approximately 5000 events were analysed per intervention. Values are means ±s.e.m. of n observations. Measurements were made in the presence of PKA and MgATP (2.27 mM or 0.26 mM); voltage was -50 mV. Previous results have demonstrated that the open time of CFTR Cl channels is independent of MgATP concentration (Winter et al. 1994). Experiments were performed using external [Cl] of either 147 or 10 mM. We pooled data at different external Cl concentrations because at -50 mV, changing the external Cl concentration had only a small effect on Kd (Fig. 8B).

n 44433
τO1 (ms)9.78 ± 1.905.16 ± 0.653.47 ± 0.472.33 ± 0.27
τO2 (ms)85.8 ± 15.242.3 ± 6.124.4 ± 2.419.1 ± 1.811.8 ± 1.7
τC1 (ms)3.02 ± 0.361.83 ± 0.242.29 ± 0.482.70 ± 0.683.67 ± 0.49
τC2 (ms)10.1 ± 2.98.5 ± 1.616.4 ± 3.817.8 ± 1.5
τC3 (ms)144.7 ± 50.3151.5 ± 50.390.4 ± 17.788.8 ± 14.886.9 ± 19.2
Total time (s)1374851595430363

To determine whether open- and closed-time constants changed with glibenclamide concentration, we performed one-way repeated measures ANOVAs. This test indicated that each of the closed-time constants did not change significantly with glibenclamide concentration (Table 1; P > 0.1). In contrast, a significant difference was found between glibenclamide concentration and both of the open time constants (Table 1; P < 0.01). These results indicate that an increase in glibenclamide concentration leads to a decrease in channel open time. This suggests that glibenclamide may be an open-channel blocker of CFTR.

If glibenclamide is an open-channel blocker of CFTR then the effect of glibenclamide on CFTR Cl channels may be described by the simple kinetic model:

display math(1)

where C, O, and B represent the closed, open and blocked states of the channel, respectively; β and α are the transition rates for channel opening and closing; kon is the second-order binding constant for glibenclamide (Glib) binding to CFTR; and koff is the first-order rate constant for dissociation of glibenclamide from CFTR. The equilibrium dissociation constant for glibenclamide binding to CFTR, Kd=koff/kon; koff= 1/τC2; and kon= (1/τO1) × [glibenclamide]−1 Figure 4A shows the relationships between the reciprocals of the glibenclamide-induced open-and closed-time constants and glibenclamide concentration. From Fig. 4A, we calculate values of kon= 3.5 × 106 M−1 s−1, koff= 92 s−1 and Kd= 26 μM, at -50 mV. These values are in good agreement with those reported by Schultz et al. (1996). They are also consistent with glibenclamide (Kd= 26 μM) being a more potent inhibitor of CFTR than tolbutamide (kon= 2.8 × 106 M−1 s−1, koff= 1210 s−1, and Kd= 430 μM; Venglarik, Schultz, DeRoos, Singh & Bridges, 1996). Although both drugs bind rapidly to CFTR, glibenclamide remains bound to CFTR more than 10-fold longer than tolbutamide.

Figure 4.

Effect of glibenclamide concentration on single-channel kinetics

A, relationship between glibenclamide concentration and the reciprocals of the glibenclamide-induced open-and closed-time constants. •, 1/τO1, where τO1 is the fast open-time constant. ○, 1/τC2, where τC2 is the intermediate closed-time constant. Data points are means ±s.e.m. (n= 3–4). The continuous lines are the fit of first-order regressions to the data. kon is calculated from the slope of the 1/τO1 regression line and koff is calculated as the value of the 1/τC2 regression line at 50 μM glibenclamide. Other details as in Table 1. B, effect of glibenclamide concentration on mean burst length. Mean burst length was calculated as described in Results. Data points are means ±s.e.m.; n= 3. The continuous fine is the fit of a first-order regression to the data.

One interesting aspect of Fig. 4A is the non-zero value of the reciprocal of the glibenclamide-induced open-time constant at zero added glibenclamide. We believe that this blocking effect is unlikely to represent solution contamination. A plausible alternative explanation is that it may represent the inhibition of CFTR Cl channels by a component of the bath solution, such as the biological buffer; some buffers are known to inhibit epithelial Cl channels (Hanrahan & Tabcharani, 1990).

A defining characteristic of the closed inline image open inline image blocked kinetic scheme shown in eqn (1) is that in the presence of the blocker the mean open time per burst is unaltered; the burst is merely prolonged by numerous intervening visits to the blocked state (Neher & Steinbach, 1978). To determine whether the closed inline image open inline image blocked kinetic scheme well describes glibenclamide inhibition of CFTR, we examined the effect of glibenclamide concentration on mean burst inline image length. Mean burst length can be calculated using the relationship (Neher & Steinbach, 1978):

display math(2)

In the absence of glibenclamide, α= 1/τO2 and Kd=koff/kon, with kon and koff calculated as described above. Figure 4B shows the effect of glibenclamide concentration on mean burst length. Consistent with the predictions of eqn (1) (Neher & Steinbach, 1978), mean burst length increased linearly with the concentration of glibenclamide (Fig. 4B). This result demonstrates that glibenclamide inhibition of CFTR is well described by the closed inline image open inline image blocked kinetic scheme.

The anionic form of glibenclamide inhibits CFTR Cl channels

Glibenclamide (pKa= 6.3) is a weak organic acid that exists in solution as either anionic or undissociated forms, the proportions of each depending on the pH. Previous studies using a benzenesulphonic acid derivative of glibenclamide that is completely dissociated at pH 7.40 suggest that the anionic form of glibenclamide interacts with SUR in pancreatic β-cells (Schwanstecher, Schwanstecher, Dickel, Chudziak, Moshiri & Panten, 1994). The structure of the anionic form of glibenclamide is shown above.

To investigate whether the anionic or undissociated form of glibenclamide inhibits CFTR, we compared the effect of glibenclamide on CFTR Cl channels at pH 7.30 and pH 6.30. Acidifying the intracellular solution from pH 7.30 to pH 6.30 decreases the proportion of the anionic form of glibenclamide from about 90% to about 50% of the total available drug. Figure 5A and B shows that acidification of the intracellular solution was without effect on the activity of wild-type CFTR Cl channels. However, at pH 6.30 there was a small, but significant (P < 0.05), decrease in the potency of glibenclamide inhibition (Fig. 5A and B). This result suggests that the anionic form of glibenclamide inhibits CFTR Cl channels.

Figure 5.

Acidification of the intracellular solution relieves glibenclamide inhibition of CFTR Cl channels

A, representative recordings show the effect of glibenclamide (50 μM)on the activity of two CFTR Cl channels at pH 7.30 (top) and pH 6.30 (bottom). Voltage was -50 mV. B, relationship between P0 and intracellular pH in the absence (filled symbols) and presence (open symbols) of glibenclamide (50 μM). Different symbols connected by fines represent individual experiments. The filled and open circles show mean ±s.e.m. values; n= 5. Percentage inhibition of P0 by glibenclamide (50 μM)at pH 7.30 was 65.4 ± 2.8% (n= 5) and at pH 6.30 was 53.4 ± 3.1% (n= 5; P < 0.05).

Glibenclamide inhibits pyrophosphate-stimulated CFTR Cl channels

To begin to identify the glibenclamide binding site in CFTR, we investigated whether the glibenclamide binding site is located within the NBDs. To test the possibility that glibenclamide may bind to NBD2 and promote channel closure, we examined whether glibenclamide could inhibit CFTR Cl channels stimulated by pyrophosphate (PPi, a compound that binds with high affinity to CFTR and greatly slows the rate of NBD2-mediated channel closure (Gunderson & Kopito, 1994; Carson, Winter, Travis & Welsh, 1995b). As previously observed (Gunderson & Kopito, 1994; Carson et al. 1995b), PPi greatly prolonged the duration of bursts of channel activity (Fig. 6A). Nevertheless, glibenclamide potently inhibited PPi-stimulated CFTR Cl channels (Fig. 6A and B). This result suggests that when channel open time is prolonged CFTR is more available to be blocked.

Figure 6.

Glibenclamide inhibits pyrophosphate-stimulated CFTR Cl channels

A, effect of glibenclamide (50 μM) on the activity of a single CFTR Cl channel stimulated by pyrophosphate (PPi). MgATP (0.26 mM), the catalytic subunit of PKA (75 nM), and PPi (5 mM) were continuously present in the intracellular solution. Voltage was -50 mV. Each trace is 20 s long. B, effect of glibenclamide (50 μM) on the P0 of CFTR Cl channels stimulated by the indicated concentrations of MgATP and PPi. Data are means ±s.e.m. of either n= 6 (0.26 mM MgATP) or n= 3 (2.27 mM MgATP and 0.26 mM MgATP ± 5 mM PPi). Percentage inhibition of P0 by glibenclamide (50 μM) was 58.8 ± 6.6% (n= 6) for 0.26 mM MgATP; 52.1 ± 7.9% (n= 3) for 2.27 mM MgATP; and 76.4 ± 4.3% (n= 3) for 0.26 mM MgATP ± 5 mM PPi.

To test further whether glibenclamide interacts with the NBDs, we examined the effect of glibenclamide at different MgATP concentrations. Figure 6B shows that at 2.27 mM MgATP, Po values in both the presence and absence of glibenclamide were greater than those at 0.26 mM MgATP. However, the percentage inhibition by glibenclamide did not change (P > 0.2). An Eadie–Hofstee plot of the data suggested that glibenclamide is a non-competitive inhibitor of MgATP-dependent gating of CFTR Cl channels (not shown). This suggests that glibenclamide and intracellular MgATP do not compete for a common binding site.

The glibenclamide binding site is located, within the channel pore

To investigate whether the glibenclamide binding site is located within the electric field of the membrane, we examined whether glibenclamide inhibition of CFTR is voltage dependent. Membrane patches were bathed in symmetrical 147 mM Cl solutions and P0 was measured in the presence and absence of glibenclamide over the voltage range ± 100mV. As previously observed (Sheppard et al. 1993; Hanrahan, Tabcharani, Chang & Riordan, 1994), the activity of wild-type CFTR was voltage independent (Fig. 7A and B). In the presence of glibenclamide, channel activity decreased significantly at negative voltages, but at positive voltages inhibition was relieved (Fig. 7A and B).

Figure 7.

Glibenclamide inhibition of CFTR Cl channels is voltage dependent

A, representative recordings showing the effect of ghbenclamide (50 μM) on the activity of four wild-type CFTR Cl channels at the voltages -100 mV (top) and +100 mV (bottom). Each trace is 10 s long. The membrane patch was bathed in symmetrical 147 mM Cl solutions. Under these conditions, wild-type CFTR had a linear current-voltage (I–V) relationship with a reversal potential, Vrev, of -1.5 ± 2.3 mV (n= 5); chloride equilibrium potential (ECl) = 0 mV. B, relationship between Po and voltage in the absence (•) and presence (○) of glibenclamide (50 μM) when the membrane patch was bathed in symmetrical 147 mM Cl solutions. Data points are means ±s.e.m. (n= 3–5) at each voltage. C, relationship between the voltage-dependent dissociation constant (Kd) and voltage for data shown in B.Kd was calculated as described in Results. The continuous line is the fit of a first-order regression to the data.

The voltage-dependent dissociation constant (Kd)for glibenclamide inhibition can be calculated with data from multichannel patches using the relationship (Benham, Bolton, Lang & Takewaki, 1985):

display math(3)

where Kd is the voltage-dependent dissociation constant at voltage V, and Pdrug and P are the open probabilities in the presence and absence of drug, respectively. Figure 7C shows that Kd was strongly voltage dependent. At -50 mV Kd was 15 ± 2 μM (n= 5), in good agreement with the value of 26 μM calculated using single-channel kinetic data and that reported by Schultz et al. (1996). This result also suggests that eqn (1) is a good description of CFTR inhibition by glibenclamide.

The voltage dependence of glibenclamide inhibition suggests that glibenclamide binds within the electric field of the membrane. The electrical distance sensed by glibenclamide (δ) can be calculated using the relationship (Woodhull, 1973):

display math(4)

where Kd(0 mV) is the voltage-dependent dissociation constant at 0 mV, z is the valency of glibenclamide, and F, R and T have their usual meanings. Using the data in Fig. 7C and assuming a valency of -1 and a single binding site for glibenclamide, δ= 0.48 ± 0.08 (n= 5) measured from the inside of the membrane over the voltage range -100 to-40 mV.

Because glibenclamide inhibition of CFTR is voltage dependent, we speculated that glibenclamide binds within the CFTR pore. If the binding site is located within the CFTR pore, the passage of Cl ions through the pore would be predicted to interfere with glibenclamide inhibition. To test this hypothesis, we investigated the effect of reducing the external Cl concentration on the voltage dependence of glibenclamide inhibition. Figure 8A and B shows that when the external [Cl] was 10 mM, the potency of glibenclamide inhibition was increased (external [Cl] = 147 mM, Kd(0 mV) = 37 ± 6 μM (n= 5); external [Cl] = 10 mM, Kd(0 mV) = 16 ± 2 μM (n= 4); P < 0.01). Reducing the external Cl concentration also decreased the electrical distance sensed by glibenclamide (δ= 0.25 ± 0.05; n= 4). We interpret these results to suggest that glibenclamide and Cl ions may compete for a common binding site. This suggests that the glibenclamide binding site is located within the CFTR pore.

Figure 8.

Low external Cl concentrations enhance glibenclamide inhibition of CFTR Cl channels

A, relationship between P0 and voltage in the absence (•) and presence (○) of glibenclamide (50 μM) when external [Cl] was 10 mM. We have previously shown that, under these conditions, wild-type CFTR has a rectifying I–V relationship with Vrev at about +60 mV; ECl=+71 mV (Sheppard et al. 1993). Data points are means ±s.e.m. (n= 4) at each voltage except +20 mV (n= 3) and -100 mV (n= 2). B, relationship between Kd and voltage for data shown in A. The continuous line is the fit of a first-order regression to the data. For comparison, the dotted line shows the line fitted to the Kd–voltage data when external [Cl] was 147 mM.

DISCUSSION

Our data show that the sulphonylurea drug glibenclamide is an open-channel blocker of CFTR Cl channels. As the glibenclamide concentration increased, open times decreased, but closed times did not change. The anionic form of glibenclamide inhibited CFTR Cl channels, and inhibition was voltage dependent and enhanced when the external Cl concentration was reduced. These results suggest that glibenclamide inhibits CFTR by occluding the pore and preventing Cl permeation.

Effect of glibenclamide on KATP and CFTR channels

The characteristics of glibenclamide inhibition of KATP channels in pancreatic β-cells (Kir6.2–SUR1 complex; Inagaki et al. 1995) differ from that of glibenclamide inhibition of CFTR Cl channels in cells expressing wild-type human CFTR in several important ways. First, the affinity of glibenclamide for KATP channels in pancreatic β-cells far exceeds the affinity of glibenclamide for CFTR Cl channels. Glibenclamide inhibited (1) KATP currents in CRI-G1 cells, a rat pancreatic islet cell line, and (2) 86Rb+-efflux activated by metabolic inhibition in COS cells coexpressing recombinant wild-type murine Kir6.2 and hamster SUR1, with half-maximal inhibition of 27 and 2 nM, respectively (Sturgess et al. 1988; Inagaki et al. 1995). In contrast, glibenclamide inhibited (1) cAMP-stimulated Cl currents in guinea-pig ventricular myocytes expressing endogenous CFTR, and (2) CFTR Cl channels in cells expressing recombinant wild-type human CFTR, with half-maximal inhibition of 25–38 μM (Tominaga et al. 1995; Schultz et al. 1996; present study). These results suggest that some feature(s) of the glibenclamide binding site in SUR1, not shared with that of CFTR, strongly stabilize the interaction of glibenclamide with KATP channels in pancreatic β-cells.

Second, glibenclamide inhibition of CFTR Cl channels was strongly voltage dependent. In the presence of glibenclamide, channel activity decreased significantly at negative voltages, but at positive voltages inhibition was relieved. In contrast, sulphonylurea inhibition of KATP channels in pancreatic β-cells is voltage independent. Tolbutamide (10–500 μM) inhibited KATP channels in rat pancreatic β-cells and KATP currents in HEK293 cells coexpressing recombinant wild-type murine Kir6.2 and hamster SUR1 with similar potency at all voltages tested over the range -120 to +30 mV (Gillis et al. 1989; Sakura et al. 1995). Because the anionic form of sulphonylureas interacts with SUR (Schwanstecher et al. 1994), these results suggest that the sulphonylurea binding site of KATP channels in pancreatic β-cells is located outside the electric field of the membrane. In contrast, the present results indicate that the glibenclamide binding site of CFTR is located about halfway across the membrane electric field.

Third, glibenclamide alters the gating behaviour of KATP and CFTR channels in different ways, although it is without effect on the single-channel current amplitude of both channels (Gillis et al. 1989; Schultz et al. 1996; present study). In cell-attached patches from rat pancreatic β-cells, tolbutamide (10 μM) dramatically increased the duration of long closures separating bursts of KATP channel activity, but had little or no effect on the open time of KATP channels (Gillis et al. 1989); similar results were observed with glibenclamide (Gillis et al. 1989). In contrast, as the glibenclamide concentration increased the open time of CFTR Cl channels decreased, but closed times did not change (Schultz et al. 1996; present study). These results suggest that the mechanism of glibenclamide inhibition of KATP channels in pancreatic β-cells differs significantly from that of CFTR Cl channels in cells expressing wild-type human CFTR.

Accessibility of the glibenclamide binding site

Sulphonylureas inhibit KATP channels in CRI-G1 cells when applied to either the intra- or extracellular side of the cell membrane (Sturgess et al. 1988). In our previous studies using the whole-cell configuration, glibenclamide inhibition of CFTR Cl currents developed slowly and was irreversible when glibenclamide was added to the extracellular solution (Sheppard & Welsh, 1992). In contrast, when glibenclamide is added to the solution bathing the intracellular side of excised inside-out membrane patches glibenclamide inhibition is readily reversible (Schultz et al. 1996; present study). These observations suggest that glibenclamide gains access to its binding site in CFTR from the intracellular side of cell membranes. Consistent with this idea, studies using glibenclamide analogues with reduced membrane permeability suggest that glibenclamide approaches its binding site in SUR from the intracellular side of β-cell membranes (Schwanstecher et al. 1994).

The finding that glibenclamide gains access to its binding site in CFTR and SUR from the intracellular side of the cell membrane suggests that the glibenclamide binding site may be located on an intracellular domain. In CFTR and SUR, the NBDs are predicted to lie on the intracellular side of the cell membrane (Riordan et al. 1989; Aguilar-Bryan et al. 1995). The NBDs of CFTR hydrolyse ATP to regulate the gating behaviour of CFTR Cl channels (Anderson et al. 1991a; Hwang et al. 1994; Carson et al. 1995a; Li et al. 1996). Therefore, to investigate the possibility that the glibenclamide binding site is located within the NBDs, we examined whether glibenclamide competes with MgATP for a common binding site. However, glibenclamide inhibition of CFTR was unaffected by nucleotide-dependent stimulation of CFTR. This suggests that glibenclamide and intracellular MgATP interact with CFTR at distinct sites.

Topology of the CFTR pore

Our findings that the anionic form of glibenclamide inhibits CFTR Cl channels, that glibenclamide inhibition is voltage dependent, and that glibenclamide inhibition is enhanced when the external Cl concentration is reduced suggest that the glibenclamide binding site is located within the CFTR pore. These characteristics of glibenclamide inhibition of CFTR are reminiscent of those of diphenylamine-2-carboxylate (DPC), which blocks CFTR when added to either the intra- or extracellular solution (McDonough, Davidson, Lester & McCarty, 1994), and those of glutamate and gluconate, two large anions that only block CFTR when added to the intracellular solution (Linsdell & Hanrahan, 1996a). Based on these results, we suggest that external Cl ions may compete with the anionic form of glibenclamide for a common binding site or interact with a site close to the glibenclamide binding site. Decreasing the external Cl concentration would therefore promote and stabilize glibenclamide binding. This suggests that the glibenclamide binding site is located within the CFTR pore.

The data indicate that glibenclamide traverses about half of the electrical distance across the membrane to reach its binding site. Although it is not known how electrical potential is distributed along the length of the CFTR pore, two studies have demonstrated a good correlation between electrical distance and physical distance along the length of the sixth transmembrane segment (M6) that lines the CFTR pore. Tabcharani et al. (1993) found that the binding site for SCN was located about 20% of the electrical distance across the membrane from the intracellular end, in good agreement with the predicted location of R347 (assuming an α-helix) with which it interacted. Similarly, McDonough et al. (1994) found that the binding site for DPC was located about 40% of the electrical distance across the membrane from the intracellular end, in good agreement with the predicted location of S341 with which it interacted. However, Cheung & Akabas (1997) suggest that electrical potential is not distributed linearly along the length of M6. Based on studies using sulfhydryl-specific reagents and cysteine scanning mutagenesis, Cheung & Akabas (1997) suggest that the intracellular end of M6 between S341 and T351 forms a major barrier to current flow through the CFTR pore.

The finding that the glibenclamide binding site is located about halfway across the membrane electric field is surprising because the energy-minimized structure of glibenclamide occupies an elliptical cylinder of length 1.95 nm, and cross-sectional diameters of 1.37nm and 0.73nm, respectively (S. J. Yewdall, personal communication). In contrast, the diameter of an unhydrated Cl ion is only 0.36 nm (Hille, 1992) and the estimated diameter of the CFTR Cl channel is 0.55–0.6 nm (Hanrahan et al. 1994; Cheung & Akabas, 1997). These dimensions indicate that glibenclamide is too large to permeate the CFTR pore.

To resolve the apparent contradiction between the location of the glibenclamide binding site within the membrane electric field and the dimensions of glibenclamide, we suggest that the glibenclamide binding site is located within a large intracellular vestibule that is part of the CFTR pore. Between the vestibule and the outside of the channel, the pore narrows. We speculate that this is the location of the selectivity filter in the CFTR pore. At the intracellular end of the CFTR pore, glibenclamide may pass directly through the open channel to reach its binding site. Consistent with this idea is the rapid onset and reversal of inhibition when glibenclamide is added to the intracellular solution (Schultz et al. 1996; present study). In contrast, at the extracellular end of the CFTR pore, glibenclamide is too large to pass through the selectivity filter. Instead, extracellular glibenclamide reaches its binding site by permeating through the membrane. Consistent with this idea is the slow onset and irreversibility of inhibition when glibenclamide is added to the extracellular solution (Sheppard & Welsh, 1992).

This model of the topology of the CFTR pore is supported by a recent study of CFTR inhibition by disulphonic stilbenes (Linsdell & Hanrahan, 1996b). A distinguishing feature of CFTR Cl channels is that disulphonic stilbenes are without effect on CFTR when added to the extracellular solution. However, when added to the intracellular solution disulphonic stilbenes caused a voltage-dependent block of CFTR Cl currents and interacted with R347, a residue that contributes to the CFTR pore. Based on these results, Linsdell & Hanrahan (1996b) suggest that the CFTR pore may contain a large intracellular vestibule where large anions bind and prevent Cl permeation.

Implications for CFTR

Previous work has shown that mutation of a number of residues within the MSDs alters the pore properties of CFTR (Anderson et al. 1991b; Sheppard et al. 1993; Tabcharani et al. 1993; McDonough et al. 1994; Cheung & Akabas, 1997). Therefore, we speculate that sequences within the MSDs may contribute to the glibenclamide binding site. Those sequences most likely lie in MSD1, based on the finding that SCN, DPC and disulphonic stilbenes all interact with residues located in M6 (Tabcharani et al. 1993; McDonough et al. 1994; Linsdell & Hanrahan, 1996b). Finally, if sequences in MSD1 form the glibenclamide binding site in CFTR, MSD1 may also contribute to the glibenclamide binding site in SUR. Consistent with this idea, the binding site for an iodinated derivative of glibenclamide is located near the amino-terminus of SUR1 (Aguilar-Bryan et al. 1995).

In conclusion, our data indicate that glibenclamide is an open-channel blocker of CFTR. Our data suggest that the CFTR pore contains a large intracellular vestibule, where the glibenclamide binding site is located. Based on the present results, future studies may identify the glibenclamide binding site in CFTR.

Acknowledgements

We thank Professor M. J. Welsh and colleagues at the University of Iowa, and Drs M. A. Valverde, C. A. R. Boyd, and R. H. Ashley for advice and critical comments. We thank Drs A. E. Smith and S. H. Cheng for the generous gift of C127 cells and Dr G. McLachlan for assistance with cell culture. We thank Dr S. J. Yewdall for molecular modelling and the University of Edinburgh Biomedical Sciences Planning Unit Mechanical Workshop and M. D. McGregor for their services. This work was supported by the BBSRC and the Cystic Fibrosis Trust.

Ancillary