Novel Regulation of Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) Channel Gating by External Chloride*

The cystic fibrosis transmembrane conductance regulator (CFTR) is vital for Cl (cid:1) and HCO 3 (cid:1) transport in many epithelia. As the HCO 3 (cid:1) concentration in epithelial secretions varies and can reach as high as 140 m M , the lumen-facing domains of CFTR are exposed to large reciprocal variations in Cl (cid:1) and HCO 3 (cid:1) levels. We have investigated whether changes in the extracellular anionic environment affects the activity of CFTR using the patch clamp technique. In fast whole cell current recordings, the replacement of 100 m M external Cl (cid:1) (Cl o (cid:1) ) with HCO 3 (cid:1) , Br (cid:1) , NO 3 (cid:1) , or aspartate (cid:1) inhibited inward CFTR current (Cl (cid:1) efflux) by (cid:1) 50% in a reversible after fitting a third order polynomial using regression analysis. Mean current amplitudes calculated at E rev 60 and normalized to cell capacitance (pF) measured using the EPC-7 amplifier. relative permeability (P /P each compared with was determined from Outside out, single channel experiments were conducted on BHK cells at temperature as described previously (28) at a V m of (cid:2) mV. Single channel currents were recorded using an Axopatch 1D (Axon Instruments, City, CA), at 50 and digitized kHz using pCLAMP8 software (Axon Channel activity was quanti- fied by measuring the mean number of channels open (N P ) during recordings lasting between at each concentration. In all of the experiments, 3.2 mV ( n (cid:5) 4), which is much closer to the predicted shift. However, the mean inhibition of inward current caused by the HCO 3 (cid:2) replacement under these conditions was not significantly different to the CsCl pipette experiments (38 (cid:3) 5.3% versus 50.2 (cid:3) 5.1%, measured at E rev (cid:2) 60 mV). This observation indicates that a component of the forskolin-activated whole-cell current in CHO cells is cation-selective. However, this does not appear to prevent or affect the trans-inhibition of CFTR by changes in the extracellular anionic environment.

Cystic fibrosis transmembrane conductance regulator (CFTR) 1 is a cyclic AMP activated epithelial Cl Ϫ channel, the mutation of which causes the potentially fatal inherited disease cystic fibrosis (CF) (1). The CFTR protein is a member of the ABC transporter family and is regulated by phosphorylation (2)(3)(4) and by ATP binding and hydrolysis (5)(6)(7). Dephosphorylation of CFTR by membrane-bound protein phosphatases is also important in the physiological regulation of channel activity (2,8,9).
Although CF is generally considered to result from a defect in Cl Ϫ transport, most of the affected epithelia also transport HCO 3 Ϫ ions. Indeed, recent work suggests that there is a better correlation between defects in CFTR-dependent HCO 3 Ϫ transport than defects in Cl Ϫ transport and the severity of disease (10). HCO 3 Ϫ is an important component of epithelial secretions and, via its buffering role, controls the pH at the epithelial cell surface. It has been reported that CF-affected epithelia, particularly in the gastrointestinal tract, secrete fluid with a more acidic pH than normal epithelia (11,12). Whether this is also true for airway surface liquid is still controversial (13,14), but recent measurements of the pH and HCO 3 Ϫ concentration of fluid secreted by polarized human submucosal gland cells, Calu-3, indicate that these cells are capable of secreting substantial amounts of HCO 3 Ϫ (ϳ80 mM) under appropriate stimulation, a process that would be defective in CF (15). An acidic luminal environment affects the physical properties of mucus (16,17) and promotes bacterial binding to mucins (18,19), both of which may have important implications for CF lung disease (20). As many CF-affected epithelia normally secrete substantial amounts of HCO 3 Ϫ , the luminal HCO 3 Ϫ concentration will vary under different physiological situations. This is particularly the case for the pancreas, a tissue in which CFTR is highly expressed and is severely affected in CF. Although plasma HCO 3 Ϫ is ϳ25 mM, the concentration of HCO 3 Ϫ in pancreatic juice can reach ϳ140 mM in humans, and because pancreatic juice is isotonic with plasma, these high HCO 3 Ϫ concentrations are accompanied by a reciprocal fall in juice Cl Ϫ concentration (21,22). Thus under physiological conditions, the apical surface of pancreatic duct cells and therefore the extracellular face of CFTR will be exposed to large variations in external Cl Ϫ (Cl o Ϫ ) and HCO 3 Ϫ (HCO 3 o Ϫ ) concentrations. A similar but less extreme situation is also likely to exist in many other HCO 3 Ϫ -transporting epithelia, such as the small intestine (12), airways (15,23), liver (24), and reproductive tract (25).
We have previously shown using fast whole cell patch clamp recordings (fWCR) that raising extracellular HCO 3 Ϫ inhibits CFTR currents in native guinea pig pancreatic duct cells (26). Our data showed that increasing HCO 3 o Ϫ caused an inhibition of both inward and outward CFTR currents. The inhibition of outward current (Cl Ϫ influx) was expected as Cl o Ϫ was replaced by the less permeant HCO 3 Ϫ ion. However, the inhibition of inward current (Cl Ϫ efflux) was surprising as the intracellular Cl Ϫ concentration should remain constant because of the large * This work was supported by a Wellcome Trust project grant (to M. A. G. and B. E. A.), a Canadian Institutes of Health research grant (to P. L.), and by a Wellcome Trust and CF Trust project grant (to A. M.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed. Cl Ϫ reservoir in the pipette and currents would not have been predicted to alter from simple theories of ion flow through channels. Thus replacing external Cl Ϫ with HCO 3 Ϫ resulted in "trans-inhibition" of Cl Ϫ efflux through CFTR. The inhibitory effect of HCO 3 Ϫ was concentration-dependent, and ϳ70% of the CFTR conductance was inhibited in the presence of 140 mM HCO 3 Ϫ . CFTR inhibition was not caused by changes in either extracellular or intracellular pH or in either the pCO 2 or CO 3 2Ϫ content of the HCO 3 Ϫ -rich solutions. We proposed that this effect of HCO 3 o Ϫ on CFTR was a novel negative feedback mechanism for the control of HCO 3 Ϫ secretion and thus the luminal surface pH in epithelia (26). That external anions can modulate CFTR function is also supported by the recent report of Shcheynikov et al. (27). These authors found that reducing Cl o Ϫ to below 20 mM caused a remarkable time-dependent increase in the OH Ϫ /HCO 3 Ϫ permeability of oocytes transfected with CFTR. This switch in OH Ϫ /HCO 3 Ϫ selectivity was suggested to arise from a conformational change in the CFTR protein, which may involve an external Cl Ϫ binding site on the ion channel (27). This effect of Cl o Ϫ is consistent with our previous proposal (26) that an extracellular "anion" binding site on CFTR is important for modulating channel function.
The main aim of this study was to clarify whether the inhibitory effect of HCO 3 Ϫ on CFTR that we previously observed was specific to the pancreatic duct. We did this by working with human CFTR heterologously expressed in Chinese hamster ovary (CHO) and baby hamster kidney (BHK) cells. We also wanted to determine whether the trans-inhibition of CFTR is caused solely by the increase in external HCO 3 Ϫ concentration, by the reciprocal fall in Cl o Ϫ , or by both mechanisms. And finally we wished to investigate whether trans-inhibition was due to an effect on channel gating or on the rate of Cl Ϫ permeation through the CFTR pore. To do this, we employed excised outside-out membrane patches and studied the effects of changing Cl o Ϫ on single channel activity.

EXPERIMENTAL PROCEDURES
Cells and Electrophysiology-CHO and BHK cells, stably transfected with human wild type CFTR, were grown as previously described (2,28). Cells were cultured on glass coverslips for use in patch clamp experiments between 2 and 5 days after plating. Whole cell patch clamp recordings were carried out at room temperature as described previously (26). An EPC-7 amplifier (List Electronic, Darmstadt, Germany) was used to record currents from CHO or BHK cells. Two configurations of the whole cell technique were used: 1) fWCR in which physical and electrical access to the cell is obtained by rupturing the patch of membrane underneath the patch pipette and 2) the slow whole cell (sWCR), a perforated patch recording technique in which a pore-forming antibiotic is included in the pipette solution. In this case, electrical access is obtained after sufficient integration of the antibiotic into the plasma membrane. In both configurations, whole cell currents were monitored using two different voltage protocols. For monitoring time-dependent changes in currents, the membrane potential (V m ) was held at 0 mV and alternately clamped to Ϯ60 mV for 1 s with a 1-s hold at 0 mV between each pulse. Steady-state current-voltage (I/V) relationships were obtained by holding V m at 0 mV and clamping to Ϯ100 mV in 20-mV increments for 500 ms with an 800-ms interval between each pulse. Data were filtered at 1 kHz and sampled at 2 kHz with a CED 1401 interface (Cambridge Electronic Design, Cambridge, United Kingdom). Reversal potentials (E rev ) and current densities were determined from I/V plots after fitting a third order polynomial using regression analysis. Mean current amplitudes were calculated at E rev Ϯ 60 mV and normalized to cell capacitance (pF) measured using the EPC-7 amplifier. The relative permeability (P x Ϫ/P Cl Ϫ) for each anion (X Ϫ ) compared with Cl Ϫ was determined from changes in E rev (26). Outside out, single channel experiments were conducted on BHK cells at room temperature as described previously (28) at a V m of Ϫ50 mV. Single channel currents were recorded using an Axopatch 1D amplifier (Axon Instruments, Union City, CA), filtered at 50 Hz, and digitized at 1 kHz using pCLAMP8 software (Axon Instruments). Channel activity was quantified by measuring the mean number of channels open (NP o ) during recordings lasting between 1-3 min at each Cl Ϫ concentration. In all of the experiments, V m was corrected for liquid junction potentials.
Solutions-For fWCR, the standard pipette solution contained as follows: 110 mM CsCl; 5 mM EGTA; 10 mM HEPES; 2 mM MgCl 2 ; and 1 mM Na 2 ATP (pH 7.2 with CsOH) (240 mosm). For experiments in which intracellular [Cl Ϫ ] was varied, CsCl was replaced with equimolar Nmethyl-D-glucamine-Cl) or iso-osmolar mannitol. To achieve a free Ca 2ϩ concentration in the pipette solution of 10 Ϫ7 M, 2.12 mM CaCl 2 was added to the standard solution. For perforated patch sWCR, 240 g/ml amphotericin B was added to the standard pipette solution with the Na 2 ATP omitted. The pipette solution used in outside out patch experiments contained the following: 110 mM CsCl; 10 mM HEPES; 5 mM EGTA; 1 mM MgCl 2 ; 2.12 mM CaCl 2 ; 1 mM MgATP; and 50 nM protein kinase A (Promega, Madison, WI), pH 7.2 with CsOH. The standard bath solution in all of the experiments contained 145 mM NaCl, 4.5 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , 10 mM HEPES, and 5 mM glucose, pH 7.4 with NaOH. For anion replacement studies, NaCl in the standard bath solution was replaced with equimolar concentrations of sodium HCO 3 Ϫ , Br Ϫ , NO 3 Ϫ , or aspartate Ϫ . When using NaHCO 3 solutions, CaCl 2 was omitted to prevent precipitation. HCO 3 Ϫ containing solutions were not routinely gassed with CO 2 , so their pH was higher than that of the standard bath solution (pH 7.4). For the reduced [Cl Ϫ ] bath solutions, NaCl was removed from the standard bath solution and replaced with sufficient mannitol to maintain osmolarity as measured using an osmometer. 2 mM CaCl 2 was omitted to ensure that the solution was comparable with the HCO 3 Ϫ substitution experiments. During WCR, CFTR was activated using 5 M forskolin (Tocris, Avonmouth, United Kingdom).
Statistics-Data are presented as mean Ϯ S.E. The significance of difference between groups was determined using Student's paired or unpaired t tests where appropriate. Alternatively for multiple comparisons, the data were analyzed using one-way analysis of variance (ANOVA). A probability of p Յ0.05 was considered statistically significant.
Chemicals-Stock solutions of forskolin (50 mM) were made up in 100% ethanol. Unless otherwise stated, all of the chemicals were from Sigma.

External HCO 3
Ϫ Inhibits Human CFTR Currents-Results presented in Fig. 1 demonstrate that extracellular HCO 3 Ϫ inhibits human CFTR expressed in CHO or BHK cells in a manner similar to that observed for CFTR in native guinea pig pancreatic duct cells (26). The data shows dose-response curves for the inhibitory effect of external HCO 3 Ϫ on human CFTR currents expressed in CHO cells (open circles) and for guinea pig CFTR (closed circles). The data are expressed as the percent inhibition of the inward current measured at E rev Ϫ60 mV using fWCR, and the curves have been fit to a simple one-site Michaelis-Menten equation. Overall, there is no significant difference in the maximal inhibition achieved between species (human, 70 Ϯ 10.3%; guinea pig, 65 Ϯ 4.0%); however, human CFTR had a significantly higher K m value (human, 37.4 Ϯ 14; guinea pig, 5.2 Ϯ 1.4 mM), suggesting that human CFTR is less sensitive to block by external HCO 3 Ϫ . CFTR expressed in BHK cells (Fig. 1, closed triangle) behaved in a similar fashion to CFTR expressed in CHO cells. In all of the cases, the block by external HCO 3 Ϫ was voltage-independent over the potential range Ϯ100 mV and fully reversible (data not shown), similar to our previous results from guinea pig duct cells (26).
Effect of Other External Anions on CFTR Currents in CHO Cells-Trans-inhibition of CFTR currents was also observed with Br Ϫ , aspartate Ϫ , and NO 3 Ϫ (Table I). The efficacy sequence was HCO 3 Ϫ ϭ Br Ϫ ϭ aspartate Ϫ Ͼ NO 3 Ϫ (Table I). The extent of trans-inhibition caused by these anions was not obviously related to their relative permeabilities (Table I) as calculated from reversal potential shifts. For instance, Br Ϫ is more permeant through CFTR than HCO 3 Ϫ (P x Ϫ/P Cl Ϫ ϭ 1.53) yet has an equivalent inhibitory effect on inward currents.
External Cl Ϫ Concentration Underlies the Inhibition of CFTR in CHO Cells-All of the anions tested caused some inhibition of the inward CFTR Cl Ϫ current ( Table I), suggesting that trans-inhibition of CFTR was not anion-specific. To investigate this further, Cl o Ϫ was reduced without replacing the ionic component and osmolarity maintained at 300 mosm by the addition of mannitol. As shown in Fig. 2, A and B, using fWCR both the outward and inward CFTR currents were inhibited by the replacement of 100 mM Cl o Ϫ with mannitol. In this representative experiment, the inward CFTR Cl Ϫ current was reduced by 57.9% at E rev Ϫ60 mV in the presence of 51.5 mM Cl o Ϫ compared with the current magnitude in the presence of 155.5 mM Cl o Ϫ . Note that the degree of CFTR inhibition is very similar regardless of whether extracellular Cl Ϫ is replaced with mannitol or with HCO 3 Ϫ (compare Figs. 1 and 2A). Fig. 2B shows I/V plots for the experiment illustrated in Fig. 2A. In these experiments, the average change in E rev following replacement of 100 mM extracellular Cl Ϫ with mannitol was 12.8 Ϯ 0.8 mV (n ϭ 6), somewhat less than the predicted value (27.8 mV), which is probably explained by the presence of a cation conductance in the CHO cells (see legend to Table I). Fig. 2C shows the Cl Ϫ dose-inhibition curve for inward CFTR currents measured using fast WCR when extracellular NaCl is replaced with mannitol. Note that the plot is curvilinear and appears biphasic. The degree of current inhibition increased markedly to ϳ50% as Cl o Ϫ was reduced from 155 to 71.5 mM, but further reduction in Cl o Ϫ to 10.5 mM caused rather less inhibition (Fig. 2C). Note that a substantial fraction (ϳ30%) of the CFTR current could not be inhibited, even when Cl o Ϫ was reduced to 10.5 mM. Internal Cl Ϫ Concentration Does Not Modulate CFTR in CHO Cells-We next used fWCR to investigate whether a reduction in internal [Cl Ϫ ] also regulated CFTR. The intracellular [Cl Ϫ ] in epithelial cells is usually in the 30 -60 mM range (29,30). Fig. 3A shows that we could not detect any significant change in forskolin-stimulated CFTR current density when pipette [Cl Ϫ ] was either raised above or decreased below the standard value of 114 mM. Thus changes in internal [Cl Ϫ ] down to 54 mM has no effect on Cl Ϫ influx (outward current), in marked contrast to the effect of lowering external Importantly, the degree of trans-inhibition of CFTR currents following reduction in Cl o Ϫ was not affected by these changes in internal [Cl Ϫ ] (Fig. 3B). The mean percent inhibition of the inward current in the presence of 54, 79, 114, and 154 mM internal Cl Ϫ following the replacement of 100 mM external [Cl Ϫ ] with mannitol was 49.8 Ϯ 9.8 (n ϭ 3), 71.5 (n ϭ 2), 63.2 Ϯ 7.0 (n ϭ 7), and 54.9 Ϯ 9.2% (n ϭ 8), respectively (p ϭ 0.96). Taken together, these observations suggest that trans-inhibition of CFTR is mediated solely by the concentration of external Cl Ϫ and that neither the internal [Cl Ϫ ] nor the [Cl Ϫ ] gradient across the cell membrane are important factors.
External Cl Ϫ Inhibits CFTR Activity in Cell-free Membrane Patches from BHK Cells-To gain insight into the mechanism of the Cl o Ϫ effect, we performed similar experiments using cellfree outside out membrane patches. Because of the density of channels in the transfected cells, only multi-CFTR-containing patches were obtained; therefore, kinetic analysis was limited to a comparison of the channel activity (NP o ) before and after changes in external Cl Ϫ concentration. Nonetheless, Fig. 4A shows that a reduction in Cl o Ϫ from 155.5 to 35.5 mM led to a marked decrease in the probability of channel opening without any obvious change in the size of the single channel currents. In three experiments, NP o decreased by 82.9 Ϯ 12% and Student's paired t tests showed a highly significant change in this parameter (Fig. 4B, p ϭ 0.012). The reverse experiment was also conducted in which the bath solution was changed from the lower to higher Cl Ϫ concentration, and results are summarized in Fig. 4C. From four such experiments, there was a significant change in NP o (p ϭ 0.013) and an overall 62.3 Ϯ 11.1% increase in channel activity, a value not significantly TABLE I Effect of external anion substitution on inward CFTR current and relative anion selectivity ⌬E rev values represent the mean change in reversal potential following substitution of 100 mM NaCl with other anions and were obtained from I/V plots. Relative permeability (P x Ϫ/P Cl Ϫ) and percent inhibition of inward current were determined as stated under "Experimental Procedures." Asp Ϫ represents aspartate. Note that the mean shift in E rev following substitution of 100 mM Cl Ϫ with HCO 3 Ϫ is much lower than the 27.8 mV predicted by the Nernst equation for a Cl Ϫ -selective current. Some of this difference can be explained by the fact that HCO 3 Ϫ can permeate CFTR, albeit to a lesser extent than Cl Ϫ (26,27,32). However, in additional experiments in which the CsCl was replaced by NMDG-Cl in the pipette solution (i.e. no internal permeant cations), the replacement of 100 mM Cl o Ϫ with HCO 3 Ϫ caused E rev to shift by 19.7 Ϯ 3.2 mV (n ϭ 4), which is much closer to the predicted shift. However, the mean inhibition of inward current caused by the HCO 3 Ϫ replacement under these conditions was not significantly different to the CsCl pipette experiments (38 Ϯ 5.3% versus 50.2 Ϯ 5.1%, measured at E rev Ϫ60 mV). This observation indicates that a component of the forskolin-activated whole-cell current in CHO cells is cation-selective. However, this does not appear to prevent or affect the trans-inhibition of CFTR by changes in the extracellular anionic environment. different to that obtained for the high to low Cl o Ϫ experiments. Therefore, these results suggest that the marked inhibition in whole cell inward current upon bath anion-replacement depicted in Figs. 1 and 2 is primarily due to a reduction in open state probability of CFTR channels.
Recording Configuration Affects the Regulation of CFTR by External Cl Ϫ in CHO Cells-All of the experiments described so far were performed under situations in which the cytosolic solution was controlled by the composition of the pipette solution. The question we next asked was whether CFTR inhibition by changes in Cl o Ϫ also occurred under more physiological conditions where the cytoplasmic compartment remains intact. To address this possibility, we turned to the sWCR-perforated patch recording technique (see "Experimental Procedures"). Using this approach, CFTR current density following stimulation with 5 M forskolin was approximately twice as large as that found for fWCR (sWCR, 440.6 Ϯ 57.7 and Ϫ387.9 Ϯ 46.4 pA/pF (n ϭ 14); fWCR, 238.2 Ϯ 39.3 and Ϫ186.3 Ϯ 25.1 pA/pF (n ϭ 19) at E rev ϩ60 mV and E rev Ϫ60 mV, respectively (p Ͻ 0.01 at both voltages)), indicating that cytoplasmic integrity is important for full CFTR current activation. Fig. 5 shows the dose-inhibition curve for Cl o Ϫ obtained using sWCR (closed triangles) with Cl o Ϫ being replaced iso-osmotically with mannitol. For comparison, fWCR data (closed circles), previously shown in Fig. 2C, are also included. In contrast to the data obtained in fWCR, the dose-inhibition curve for sWCR clearly shows a different profile. In sWCR, the percent reduction of inward CFTR current following a decrease in Cl o Ϫ was significantly less over the range 116.5-71.5 mM Cl Ϫ . Moreover, the concentration of Cl o Ϫ causing a 50% inhibition of the forskolin-stimulated CFTR current at E rev Ϫ60 mV in sWCR was 31.8 mM compared with a value of 77.5 mM in fWCR, giving a leftward shift of the dose-inhibition curve of ϳ47 mM. Nevertheless, the maximum degree of CFTR current inhibition was similar in fast and slow WCR (Fig. 5). The same difference in the Cl Ϫ dose-inhibition curves for fast and slow WCR was observed when extracellular Cl Ϫ was replaced with HCO 3 Ϫ . In sWCR, the response curve was shifted to the left by ϳ39 mM (data not shown).
Internal Ca 2ϩ Concentration Modulates Trans-inhibition in CHO Cells-The pipette solutions used for the fWCR experiments contained 5 mM EGTA, which will buffer the intracellular free [Ca 2ϩ ] at ϳ10 Ϫ9 M. However, EGTA is unable to cross the amphotericin-permeabilized membrane during sWCR; thus the internal free [Ca 2ϩ ] in these experiments is unknown but is expected to be within the physiological range. To determine whether internal free [Ca 2ϩ ] was modulating the inhibitory response, cytosolic Ca 2ϩ was fixed at a more physiological resting value of 10 Ϫ7 M during fWCR experiments (see "Experimental Procedures").

FIG. 2. Trans-inhibition of CFTR is observed when external Cl o
Ϫ alone is reduced. Representative fWCR current recordings from CHO cells transfected with CFTR measured between Ϯ100 mV in 20-mV steps. A(i), unstimulated current records; (ii), currents following stimulation with 5 M forskolin; (iii), stimulated currents after Cl o Ϫ was reduced to 51.5 mM (the removal of 100 mM NaCl and replaced with mannitol). B, corresponding I/V relationship for this experiment.   5 (open circles) shows that increasing the intracellular [Ca 2ϩ ] from 10 Ϫ9 to 10 Ϫ7 M in fWCR changes the shape of the dose-inhibition response to Cl o Ϫ depletion. When intracellular Ca 2ϩ is raised, the inhibition curve moves to the left, closely resembling that obtained in sWCR (Fig. 5, triangles). In fWCR, with 10 Ϫ7 M Ca 2ϩ in the pipette, the mean percent inhibition of the inward CFTR current at E rev Ϫ60 mV was 9.3 Ϯ 7.7%, 23.5 Ϯ 3.5%, and 30.7 Ϯ 4.9% at 116.5, 91.5, and 71.5 mM Cl o Ϫ , respectively (n ϭ 3-7). Two of these values (116.5 and 71.5 mM Cl Ϫ ) are significantly different from the values obtained in fWCR with 10 Ϫ9 M Ca 2ϩ in the pipette solution (p ϭ 0.005 and 0.01, respectively, at E rev Ϫ60 mV), and the other value (91.5 mM Cl Ϫ ) just fails to reach significance (p ϭ 0.06). These results suggest that differences in the cytoplasmic Ca 2ϩ concentration are a major factor contributing to the apparent difference between trans-inhibition observed using fast and slow WCR.

DISCUSSION
The external face of CFTR is exposed to wide variations in luminal pH, [HCO 3 Ϫ ], and [Cl Ϫ ] that occur normally at the epithelial cell surface. We have asked whether such changes in the composition of surface/secreted fluids can influence the activity of CFTR. Here we show that human CFTR expressed in CHO or BHK cells is inhibited by a reduction in external Cl Ϫ through a mechanism that can be modulated by internal [Ca 2ϩ ]. How such a signal is transmitted across the luminal membrane remains to be explained. Our data show that the replacement of extracellular Cl Ϫ with another anion (as will occur physiologically when HCO 3 Ϫ secretion occurs) causes a trans-inhibition of Cl Ϫ efflux through CFTR. Trans-inhibition of CFTR currents was also observed in fWCR following isoosmotic replacement of extracellular NaCl with mannitol. This finding suggests that a reduction in external Cl Ϫ per se is the factor that causes inhibition. Similar changes in internal Cl Ϫ concentration had no effect on CFTR activity, illustrating the strict sidedness of the effect.
In fWCR, a substantial proportion of the CFTR current (ϳ45%) was inhibited when extracellular Cl Ϫ was reduced from a supra-physiological value of 155 mM to the normal plasma value of ϳ110 mM. Further reduction in extracellular Cl Ϫ below 110 mM (as will occur in the luminal secretions produced by HCO 3 Ϫ -transporting epithelia) caused little additional inhibition of the CFTR channels. A substantial proportion of the CFTR current (ϳ30%) remained, even when external Cl Ϫ was reduced to 10.5 mM, which is approximately the minimum value observed in pancreatic juice (21). Under the more physiological conditions of the sWCR configuration where the cytoplasmic compartment remains essentially intact, we found that the Cl Ϫ dose-inhibition curve was shifted to the left by ϳ45 mM. Thus Cl o Ϫ needed to be reduced to ϳ32 mM to cause a 50% inhibition of CFTR current in sWCR compared with ϳ78 mM in fWCR. This means that, in sWCR, the inhibitory effect of Cl Ϫ removal is more apparent over the normal range of Cl Ϫ concentrations found in the fluids secreted by epithelia. Nevertheless, the maximal levels of CFTR current inhibition were similar in fast and slow WCR. Increasing the intracellular [Ca 2ϩ ] to physiologically relevant resting levels (100 nM) during fWCR shifted the Cl Ϫ dose-inhibition curve to the left mimicking the curve obtained using sWCR. The mechanism by which an in-

FIG. 5. Inhibition of inward CFTR current by Cl o
؊ is dependent on recording configuration and intracellular Ca 2؉ concentration from CHO cells stably transfected with CFTR. Data shows the Cl Ϫ concentration-response curve for the fWCR data shown in Fig.  2C, filled circles. Results from similar experiments using the sWCR configuration are also shown (filled triangles) (n ϭ 3-7). Slow WCRs were made with amphotericin B in the pipette solution (see "Experimental Procedures"). The mean percent inhibition was determined from I/V data at E rev Ϫ60 mV. Asterisk indicates data for which there is a significant difference between the fast and slow WCR (p Յ 0.05 crease in cytosolic Ca 2ϩ blunts the response of CFTR to reduced Cl o Ϫ is unclear. CFTR activity has not been shown to be directly dependent on Ca 2ϩ (32) and thus the effect is likely to be indirect. Calcium could modulate the interaction of CFTR with regulatory protein kinases, such as protein kinase C, which is known to enhance CFTR activity through the facilitation of CFTR phosphorylation by PKA (4,32). Alternatively, Ca 2ϩ may alter the interaction of CFTR with regulatory protein phosphatases (2,8,9,32).
To gain some insight into the mechanism of inhibition of CFTR by Cl o Ϫ , outside out single channel patch experiments were employed. The results presented in Fig. 4  Ϫ brings about this specific effect is unknown, but modulation of ATP binding and/or hydrolysis at the nucleotide binding domains of CFTR is one possibility. We found that including 5 mM pyrophosphate in the pipette solution during fWCR experiments to lock open CFTR (33) did not affect the ability of low Cl o Ϫ to inhibit CFTR (data not shown). This finding suggests that the gating effect we observe is not because of a change in open-time duration and thus probably occurs by stabilizing the closed state of the channel (32). Gating of the CFTR channel is also known to be affected by phosphorylation/dephosphorylation of the R domain (2-4, 8, 9), and a Cl Ϫ -dependent conformational change in CFTR could well affect the phosphorylation status of CFTR, which is known to alter channel activity (32). That changes in Cl o Ϫ can induce a structural change in CFTR is supported by the recent data of Shcheynikov et al. (27). Indeed, the phosphorylation of the R domain itself has been shown to cause a pronounced conformational change in CFTR monitored by fluorescence spectroscopy (34). Clearly, establishing whether changes in Cl o Ϫ alter the structure of CFTR is an important area for future work.
The mechanism by which the CFTR protein senses Cl o Ϫ remains to be determined. Our observation that the inhibitory effect of anion substitution and low Cl o Ϫ on CFTR currents was voltage-independent argues against a Cl Ϫ sensor being located within the selectivity filter of the channel pore. This is in marked contrast to the Cl Ϫ -dependent gating of the ClC channel pore (35,36). There are a number of positively charged amino acid residues on the extracellular loops of CFTR, e.g. Arg-104, Lys-114, and Arg-117 in extracellular loop (EL) 1; Lys-329 in EL3; Lys-892, His-897, and Arg-899 in EL4; and Arg-1128 in EL6. It is possible that one or more of these residues form a Cl Ϫ sensor. The disease-causing mutation R117H is known to reduce the open state probability of CFTR by ϳ30% (37). However, Gong and Linsdell (38) have identified Arg-334, which is probably located at the outer mouth of the CFTR pore as an important arginine residue involved in high affinity Cl Ϫ binding. Because this residue is very important in coordinating ion-ion interactions, it may well have a role to play in the Cl Ϫ -sensing mechanism.
Although the exact mechanism of epithelial HCO 3 Ϫ secretion is still not fully understood, it is generally accepted that CFTR has a fundamental role in this process (10, 21-27, 29, 39). Physiologically, we propose that, during agonist-stimulated anion secretion, CFTR activity is reduced by the reciprocal changes in luminal Cl Ϫ and HCO 3 Ϫ concentrations. The reduction in CFTR activity as Cl o Ϫ falls would help to maintain the electrical driving force for anion exit by preventing excessive depolarization of the apical membrane. This effect, together with the recently described increase in HCO 3 Ϫ permeability of CFTR at low Cl o Ϫ (27) would help to drive net HCO 3 Ϫ secretion. Thus changes in luminal [Cl Ϫ ] has two major effects on CFTR. It modulates both gating and anion permeability and therefore represents an important novel signal for regulating anion and fluid secretion in epithelial cells.