Relationship between anion binding and anion permeability revealed by mutagenesis within the cystic fibrosis transmembrane conductance regulator chloride channel pore

Abstract

1. Anion binding within the pores of wild-type and mutant cystic fibrosis transmembrane conductance regulator (CFTR) Cl⁻ channels, expressed in two different mammalian cell lines, was assayed using patch clamp recording. Specifically, experiments measured both the conductance of different anions and the ability of other permeant anions to block Cl⁻ permeation through the pore.

2. Under symmetrical ionic conditions, wild-type CFTR channels showed the conductance sequence Cl⁻ >NO₃⁻ >Br⁻≥formate >F⁻ >SCN⁻≈ ClO₄⁻.

3. High SCN⁻ conductance was not observed, nor was there an anomalous mole fraction effect of SCN⁻ on conductance under the conditions used. Iodide currents could not be measured under symmetrical ionic conditions, but under bi-ionic conditions I⁻ conductance appeared low.

4. Chloride currents through CFTR channels were blocked by low concentrations (10 mM) of SCN⁻, I⁻ and ClO₄⁻, implying relatively tight binding of these anions within the pore.

5. Two mutations in CFTR which alter the anion permeability sequence, F337S and T338A, also altered the anion conductance sequence. Furthermore, block by SCN⁻, I⁻ and ClO₄⁻ were weakened in both mutants. Both these effects are consistent with altered anion binding within the pore.

6. The effects of mutations on anion permeability and relative anion conductance suggested that, for most anions, increased permeability was associated with increased conductance. This indicates that the CFTR channel pore does not achieve its anion selectivity by selective anion binding within the mutated region. Instead, it is suggested that entry of anions into the region around F337 and T338 facilitates their passage through the pore. In wild-type CFTR channels, anion entry into this crucial pore region is probably dominated by anion hydration energies.

Ion channels are defined primarily by their ionic selectivity, the ability to allow certain ions to pass at a high rate while effectively excluding others. The most highly selective ion channels are the voltage-gated cation channel superfamily, which may select for K⁺, Na⁺ or Ca²⁺ ions with extremely high fidelity (Hille, 1992). In these channel types, ionic selectivity is determined at a highly localised region within the pore known as the selectivity filter. In both K⁺ channels (Doyle et al. 1998) and Ca²⁺ channels (Almers & McCleskey, 1984; Hess & Tsien, 1984), selective binding of the ion of physiological interest within the selectivity filter appears to be crucial to the selective permeability of that ion through the pore. The apparent incongruity between tight binding and high conductance of the permeant ion is most often reconciled by electrostatic repulsion between ions bound simultaneously within the selectivity filter region (Almers & McCleskey, 1984; Hess & Tsien, 1984; Doyle et al. 1998), although other explanations have been put forward (e.g. Dang & McCleskey, 1998; Nonner & Eisenberg, 1998).

Much less is known about the mechanism of anion permeation and anion selectivity in Cl⁻ channels. These are generally much less selective than the cation channels described above, typically having some permeability to all small monovalent anions (see Linsdell et al. 2000). Anion permeability sequences are usually described as similar to a Hofmeister or lyotropic sequence (SCN⁻ > I⁻ > NO₃⁻ > Br⁻ > Cl⁻ > F⁻), with anions which are more easily dehydrated (lyotropes) having a higher permeability than anions which retain their waters of hydration more strongly (kosmotropes; Bormann et al. 1987; Halm & Frizzell, 1992; Linsdell & Hanrahan, 1998; Smith et al. 1999; Dawson et al. 1999; Linsdell et al. 2000). The relationship between anion permeability and anion energy of hydration supports the notion that anion dehydration is the limiting step in permeation (Dawson et al. 1999; Linsdell et al. 2000), which may reflect the fact that anions are more difficult to dehydrate than cations of a similar size (Dorman et al. 1996; Dawson et al. 1999).

Lyotropic anions, as well as showing a high permeability in most Cl⁻ channels, have also often been described as binding tightly within Cl⁻ channel pores. For example, in neuronal GABA_A and glycine receptor channels, anion conductance sequences are the inverse of (lyotropic) anion permeability sequences, suggesting that anions with the highest permeability also bind the most tightly within the pore (Bormann et al. 1987). In the cystic fibrosis transmembrane conductance regulator (CFTR) Cl⁻ channel, both anion permeability and anion binding (as determined by the ability to block Cl⁻ permeation through the pore) among halide and pseudohalide anions show a linear relationship to inverse ionic radius (Smith et al. 1999). Two possible explanations could underlie this correlation between anion permeability and anion binding. First, lyotropic anions might show the highest permeability as a consequence of the fact that they bind most tightly within the pore. Alternatively, it has been suggested that anion permeability and anion binding are independent processes which are controlled by similar forces, both being determined primarily by the small difference between anion-water interaction energies and anion-channel interaction energies (Smith et al. 1999).

The hypothesis that anion permeability and anion binding are separable facets of the permeation process in the CFTR Cl⁻ channel is supported by the fact that several mutations within the pore have been shown to alter anion binding without strongly affecting anion permeability (e.g. K335E, Anderson et al. 1991; R347D, Tabcharani et al. 1993; G314E, Mansoura et al. 1998). This has led to the proposal that while anion binding occurs at discrete intrapore sites, anion permeability is a global function of the entire pore and, as a result, relatively insensitive to mutagenesis (Dawson et al. 1999; Smith et al. 1999). However, we have recently shown that mutations at two adjacent amino acid residues within the sixth transmembrane region (TM6) of CFTR, F337 (Linsdell et al. 2000) and T338 (Linsdell et al. 1998), significantly alter CFTR channel anion permeability. Most interestingly, two mutations which reduce amino acid side chain size at position 337, F337A and F337S, virtually abolished the normal lyotropic relationship between anion permeability and energy of hydration (Linsdell et al. 2000). The present study seeks to shed new light on the relationship between anion binding and anion permeability in CFTR channels by comparing the anion binding properties of wild-type CFTR with two mutants with altered anion selectivity, F337S and T338A. Anion binding in the pore was assayed both by the relative conductance of different anions and by the ability of foreign anions to block Cl⁻ permeation.

METHODS

Expression of wild-type and mutant CFTR

Experiments were carried out on two different mammalian cell lines, Chinese hamster ovary (CHO) and baby hamster kidney (BHK) cells, stably transfected with either wild-type or mutated forms of CFTR in the pNUT vector (Linsdell et al. 1998, 2000). CHO cells were used for single channel recording and BHK cells for macroscopic current recording (Linsdell & Hanrahan, 1998; Hanrahan et al. 1998). The CFTR mutants F337A, L, S, W, Yand T338A were constructed and transfected into CHO and BHK cells by Alexandra Evagelidis and Shu-Xian Zheng in the laboratory of Dr John Hanrahan (McGill University, Montréal, Québec, Canada), as described previously (Linsdell et al. 1998, 2000). In the present study, the permeation properties of two mutants, F337S and T338A, have been examined in detail. Of those mutations introduced at F337, only F337A and F337S strongly altered anion selectivity, and the effects of these two mutations were very similar (Linsdell et al. 2000). These two mutations also cause a similar reduction in Cl⁻ conductance (Fig. 5). However, of all the mutations introduced at F337, only F337S could be expressed in CHO cells and studied using single channel recording (P. Linsdell, A. Evagelidis & J. W. Hanrahan, unpublished observations). F337S was also more strongly expressed in BHK cells, allowing small macroscopic currents carried by anions with a very low conductance to be measured. In contrast to mutations at F337, all mutations previously examined at T338 (T338A, N, S, V) significantly altered anion selectivity and Cl⁻ conductance (Linsdell et al. 1998). For the present study, T338A was examined because: (1) it has a high single channel conductance, allowing single channel currents to be resolved, (2) its anion selectivity strongly follows the lyotropic sequence, in fact more strongly than that of wild-type CFTR, such that its effects on anion permeability might be considered ‘opposite’ to the effects of F337S, (3) replacing the threonine with a small, ‘neutral’ alanine is considered less likely to cause large changes in transmembrane helix structure, and (4) T338A is well expressed in both CHO and BHK cells (see Linsdell et al. 1998, for a full description of the permeation phenotype of T338A).

Figure 5. Relative conductance of channels bearing different mutations for F337, plotted as a function of amino acid side chain volume.

Current amplitude at +50 mV was estimated from macroscopic current variance analysis, as shown in Fig. 4, and is plotted as a fraction of wild-type amplitude. Different mutants are labelled according to the amino acid present at position 337; hence, F denotes wild-type. Amino acid side chain volume was estimated according to Richards (1974).

Electrophysiological recordings

Both single channel and macroscopic current recordings were made using the excised, inside-out configuration of the patch clamp technique (Linsdell & Hanrahan, 1998; Linsdell et al. 1998, 2000; Hanrahan et al. 1998). Briefly, channels were activated following patch excision by exposure of the cytoplasmic face of the patch to 20-130 nM protein kinase A catalytic subunit (prepared in the laboratory of Dr M. P. Walsh, University of Calgary, Alberta, Canada; Hanrahan et al. 1998) plus 1 mM MgATP. Both pipette (extracellular) and bath (intracellular) solutions contained (mM): 150 NaX (where X is the anion being tested), 2 MgCl_2, 10 TES, pH adjusted to 7.4 using NaOH. All chemicals were obtained from Sigma Chemical Co. (Oakville, Ontario, Canada). Current traces were filtered at 50 Hz (for single channel recording), 100-200 Hz (for macroscopic current-voltage relationships) or 500 Hz (for macroscopic current variance analysis) using an 8-pole Bessel filter, except where stated otherwise, and digitized at 250 Hz for single channel recording, 250 Hz to 1 kHz for macroscopic current- voltage relationships, and 1 kHz for current variance analysis. Macroscopic currents were analysed using pCLAMP6 computer software (Axon Instruments, Foster City, CA, USA), while single channel currents were analysed using custom-written, pCLAMP-compatible DRSCAN software (Hanrahan et al. 1998).

Measurements of macroscopic current variance were made during the slow activation of macroscopic current following addition of a low concentration of PKA (∼20 nM) (Linsdell & Hanrahan, 1998; Linsdell et al. 1998). Mean current and current variance were calculated for 5 s subrecords filtered at 500 Hz, giving a bandwidth of 0.2-500 Hz. The length of subrecords was chosen to minimise the error due to significant activation of the current during a single subrecord without omitting too much of the low frequency variance. However, this method was found to significantly underestimate unitary current amplitudes when compared with single channel recording, suggesting that part of the noise may be lost during the recording (Linsdell et al. 1998). Nevertheless, relative conductances of different anions estimated from current variance analysis were similar to those estimated by single channel recording (Fig. 7); as a result, only relative conductances from variance analysis are given. Variance versus mean current relationships were fitted by the equation:

(1)

where σ_I² is the current variance, I the mean current amplitude, i the estimated unitary current amplitude and N the estimated total number of channels in the patch.

Figure 7. Relative conductance of different anions in wild-type CFTR.

Anion conductance was estimated under symmetrical ionic conditions, using either single channel recording (×) or macroscopic current variance analysis (□). Mean of data from 5-12 patches for single channel recording, and 3-9 patches for macroscopic current variance.

Macroscopic current-voltage (I-V) relationships were constructed using depolarizing voltage ramp protocols, with a rate of change of voltage of 40-120 mV s⁻¹ (see Linsdell & Hanrahan, 1996, 1998). Each current trace is the result of a single voltage ramp. All I-V relationships have had the background (leak) current recorded before addition of PKA subtracted digitally, leaving uncontaminated CFTR currents (Linsdell & Hanrahan, 1996, 1998; Hanrahan et al. 1998; Linsdell et al. 2000). In some cases, the relative conductance of different anions was estimated by measuring the relative slope of the I-V curve for inward and outward currents under bi-ionic conditions (Figs 9 and 13). In these cases, the conductance of an anion X⁻ relative to that of Cl⁻, g_X/g_Cl, was estimated as the ratio of the slope conductance at a voltage 50 mV more negative than the reversal potential, to that at 50 mV more positive than the reversal potential. Slope conductance was estimated by fitting a straight line to all data points between 48 and 52 mV from the reversal potential (40 data points total).

Figure 9. Macroscopic current (I) -voltage (V) relationships suggest low SCN⁻ conductance.

Taken from BHK cell patches under bi-ionic conditions (intracellular SCN⁻, extracellular Cl⁻), before (Control) and after addition of 1 mM pyrophosphate (PP_i) to the intracellular solution. The outward rectification of the I-V relationship is consistent with higher Cl⁻ conductance than SCN⁻ conductance. The shape of the I-V relationship is not altered by ‘locking’ channels in the open state by addition of 1 mM PP_i, suggesting that channel gating does not contribute to this outward rectification.

Figure 13. Macroscopic I-V relationships under bi-ionic conditions (intracellular I⁻, extracellular Cl⁻), for wild-type, F337S and T338A CFTR expressed in BHK cells.

The different degrees of outward rectification suggest different relative I⁻ conductances in these three CFTR variants. The slopes of the I-V relationships were used to estimate relative I⁻ conductance as described in the text. Note the very different current reversal potentials for each channel variant under the same ionic conditions, reflecting the different I⁻ permeabilities of these different channels (Table 1).

Experiments were carried out at room temperature, 21-24°C. Values are presented as means ±s.e.m., except where stated otherwise. For graphical presentation of mean values, error bars represent ±s.e.m., where this is larger than the size of the symbol. Fitting of data and statistical analysis (Student’s two-tailed t tests) were carried out using SigmaPlot version 5.0 software (SPSS Inc., Chicago, IL, USA).

RESULTS

To compare the anion binding properties of wild-type and pore mutant forms of CFTR channels, anion binding in the pore was characterised in two ways: by measuring the relative conductance of different anions and by examining the ability of other permeant anions to block Cl⁻ permeation.

Anion conductance

The conductance of different anions in CFTR channels has been estimated using single channel recording under bi-ionic conditions (Gray et al. 1990; Tabcharani et al. 1997) and by whole-cell recording following one-sided ion substitution (Anderson et al. 1991; Gray et al. 1993; Mansoura et al. 1998). However, such results are subject to trans-ion effects (Tabcharani et al. 1997; see Discussion), and changes in channel open probability following ion replacement. To avoid these potential artifacts, in the present study anion conductance was estimated from the unitary current amplitude measured with identical solutions on both sides of the membrane (Bormann et al. 1987; Halm & Frizzell, 1992).

Unitary CFTR currents carried by Cl⁻, NO₃⁻, Br⁻ and formate were readily identified under symmetrical ionic conditions in inside-out patches excised from CHO cells (Fig. 1). However, unitary currents carried by I⁻, SCN⁻ and ClO₄⁻ were never observed, despite multiple attempts using large-bore pipettes, over a broad range of voltages (data not shown). This is in contrast to previous single channel recordings immediately following exposure to symmetrical I⁻ (Tabcharani et al. 1997) and SCN⁻ solutions (Tabcharani et al. 1993). With symmetrical F⁻ solutions, very small currents were observed at strongly hyperpolarizing potentials only (Fig. 1). Although too small to quantify accurately, the single channel conductance at −100 mV with symmetrical 150 mM F⁻ was around 1 pS.

Figure 1. Unitary CFTR channel currents carried by different anions.

Channel activity in the presence of symmetrical 150 mM NaCl, NaNO₃, NaBr and sodium formate was recorded from CHO cell patches at membrane potentials of −50 and +50 mV; the line to the left indicates all channels closed. Unitary F⁻ currents could only be resolved at strongly hyperpolarized potentials (-100 mV) and with heavy filtering (20 Hz). As noted previously (Tabcharani et al. 1997), open probability was very high in the presence of F⁻.

Mean single channel i-V curves for those anions which did carry quantifiable unitary currents are shown in Fig. 2. With symmetrical Cl⁻, NO₃⁻ and Br⁻, the i-V relationship was linear with a mean slope conductance of 7.59 ± 0.10 pS (n= 12), 4.55 ± 0.12 pS (n= 5) and 3.62 ± 0.06 pS (n= 6), respectively (Fig. 2; Table 1). In contrast, the i-V relationship with symmetrical formate was significantly inwardly rectifying, with a mean chord conductance of 3.24 ± 0.06 pS (n= 5) at −50 mV and 2.40 ± 0.06 pS (n= 6) at +50 mV (P < 0.001). Interestingly, as well as showing a higher conductance for anion efflux than anion influx (Fig. 2), organic anions such as formate show a higher permeability when present in the intracellular than the extracellular solution under bi-ionic conditions (Linsdell & Hanrahan, 1998).

Figure 2. Properties of unitary CFTR channel currents carried by different anions.

Mean single channel current (i)-voltage (V) relationships were measured in the symmetrical presence of the named anion. Mean of data from 3-9 patches.

Table 1.

Relative anion permeabilities and conductances for wild-type and mutant CFTR

	PX/PCl			gX/gCl
	WTa	F337Sa	T338Ab	WTc	WTd	F337Sd	T338Ad
Cl	1.00 ± 0.01 (10)	1.00 ± 0.08 (3)	1.00 ± 0.02 (11)	1.00 ± 0.01 (12)	1.00 ± 0.10 (9)	1.00 ± 0.16 (7)	1.00 ± 0.09 (4)
Br	1.37 ± 0.07 (8)	0.50 ± 0.04 (4)	1.74 ± 0.04 (3)	0.48 ± 0.01 (6)	0.48 ± 0.13 (4)	0.16 ± 0.03* (4)	0.44 ± 0.02 (3)
I	0.83 ± 0.03 (6)	0.23 ± 0.02 (4)	2.09 ± 0.16 (5)	—	—	—	—
F	0.10 ± 0.01 (9)	0.43 ± 0.02 (4)	0.12 ± 0.02 (4)	—	0.094 ± 0.017 (3)	0.76 ± 0.19* (4)	0.054 ± 0.011 (3)
SCN	3.55 ± 0.26 (7)	0.93 ± 0.10 (5)	5.85 ± 0.27 (4)	—	0.060 ± 0.012 (5)	0.17 ± 0.04* (3)	0.085 ± 0.007 (4)
NO2	1.58 ± 0.04 (10)	1.08 ± 0.02 (4)	2.04 ± 0.08 (3)	0.60 ± 0.02 (5)	0.73 ± 0.07 (5)	0.29 ± 0.08* (3)	0.96 ± 0.05 (3)
ClO4	0.25 ± 0.01 (8)	0.19 ± 0.00 (3)	1.35 ± 0.08 (3)	—	0.059 ± 0.014 (6)	0.041 ± 0.008 (2)	0.082 ± 0.011 (4)
Formate	0.24 ± 0.01 (9)	0.27 ± 0.02 (3)	0.45 ± 0.04 (3)	0.35 ± 0.01 (6)	0.49 ± 0.01 (5)	0.17 ± 0.02** (3)	0.46 ± 0.07 (3)

^aFrom Linsdell et al. (2000)

^bfrom Linsdell et al. (1998)

^cby single channel recording

^dby current variance analysis. Relative permeabilities (PX/PCl) for different anions present in the intracellular solution under biionic conditions were taken from Linsdell et al. (1998, 2000). Relative conductances (g_X/g_Cl), were estimated either by single channel recording or by current variance analysis, as described in the text. Numbers in parentheses indicate the number of patches examined in each case. *Significant difference in conductance from the corresponding value in wild-type

^*P < 0.05

^**P < 0.001.

Unitary Cl⁻ currents carried by wild-type, F337S and T338A CFTR are compared in Fig. 3A. Mean slope conductance (Fig. 3B) was reduced in F337S (from 7.59 ± 0.10 pS (n= 12) to 1.76 ± 0.03 pS (n= 7)), and significantly increased in T338A (to 9.94 ± 0.14 pS, n= 6). Because of the low conductance of F337S, anion conductance was also estimated from the increase in current variance associated with activation of macroscopic CFTR currents at +50 mV in inside-out patches excised from BHK cells (Fig. 4), as described in Methods.

Figure 3. Unitary properties of F337S and T338A CFTR.

A, single channel currents carried by wild-type, F337S and T338A, expressed in CHO cells, with symmetrical 150 mM NaCl, at a membrane potential of −50 mV. The line to the far left indicates all channels closed. B, mean i-V relationships under these ionic conditions, wild-type •, F337S ○ and T338A □; mean of data from 3-9 patches.

Figure 4. Estimation of unitary current amplitude from macroscopic current variance.

A, activation of macroscopic wild-type, F337S and T338A CFTR currents in BHK cell patches at +50 mV, in the symmetrical presence of 150 mM NaCl, by addition of PKA in the presence of ATP (see Methods). B, relationship between CFTR current amplitude (I) and current variance (σ_I²) for the patches shown above. All three have been fitted by eqn (1) (see Methods), giving i= 0.118 pA and N= 374 for wild-type, i= 0.0387 pA and N= 1359 for F337S and i= 0.279 pA and N= 175 for T338A.

The conductance of other mutants with substitutions for F337 could not previously be examined because these mutants could not be expressed in CHO cells (Linsdell et al. 2000). The relative Cl⁻ conductance of each of these mutants could, however, be estimated from macroscopic current variance analysis following expression in BHK cells, as described in Fig. 4. Relative Cl⁻ current amplitudes at +50 mV for each of these mutants are shown in Fig. 5. Relative Cl⁻ conductance appears to be weakly correlated with the size of the amino acid side chain present at position F337, with a large side chain favouring high conductance; this correlation results mainly from the low (and similar) conductances of the two mutants which strongly disrupted normal lyotropic anion selectivity, F337A and F337S. However, conductance was reduced to a similar extent in F337Y, a mutation which results in only a small change in amino acid side chain size. Interestingly, Cl⁻ conductance was strongly correlated with amino acid side chain size at position 338, with small side chains (such as alanine) favouring high conductance (Linsdell et al. 1998).

Macroscopic current variance analysis was also used to compare the relative conductances of different anions in wild-type, F337S and T338A (Fig. 6; Table 1). Although unitary conductances estimated in this way were considerably lower than those measured using single channel recording (see Methods), relative conductances in wild-type CFTR measured by the two methods were similar (Fig. 7; Table 1) and allowed the conductance sequence to be extended to Cl⁻ > NO₃⁻ > Br⁻≡ formate > F⁻ > SCN⁻≡ ClO₄⁻. Macroscopic wild-type CFTR currents were not observed with symmetrical I⁻ solutions. Relative conductances in wild-type estimated by these two methods (Fig. 7), as well as those in F337S and T338A estimated from macroscopic current variance analysis, are summarised in Table 1. For F337S, a number of significant differences from the wild-type pattern were observed; the relative conductance of NO₃⁻, Br⁻ and formate were all significantly decreased, while the relative conductance of F⁻ and SCN⁻ increased. Although smaller differences are apparent for T338A, in no case were these significantly different from wild-type (Table 1). The conductance sequences for wild-type, F337S and T338A are summarised in Table 2.

Figure 6. Macroscopic current variance analysis of relative anion conductance in wild-type, F337S and T338A.

Macroscopic currents were activated in BHK cell patches at +50 mV, in the symmetrical presence of the anion named on the far left, by addition of PKA in the presence of ATP (see Fig. 4). In each case, the scale bar represents 5 pA on the ordinate and 1 min on the abscissa. Below each trace is the corresponding relationship between current amplitude (I) and current variance (σ_I²), fitted by eqn (1) (see Methods). Results of such analyses are summarised in Table 1.

Table 2.

Permeability and conductance sequences for wildtype and mutant CFTR

	WildType	F337S	T338A
Permeability sequence	SCN > NO 3 > Br > Cl > I > ClO4 ≈ formate > F	NO 3 > Cl ≥ SCN > Br > F > formate > I >ClO4	SCN > I ≥ NO3 > Br > ClO4 > Cl > formate > F
Conductance sequence	Cl > NO 3 ≥ Br ± formate > F > SCN ≈ ClO4	Cl > F > NO3 > SCN ≈ formate ≈ Br > ClO4	Cl ≥ NO3 > formate ≈ Br > SCN ≈ ClO4 > F

Permeability and conductance sequences are derived from data given in Table 1.

All mutations examined by Linsdell et al. (1998) involving substitution of T338 altered both anion selectivity and conductance. However, of those mutations involving substitution of F337 (Fig. 5), only F337A and F337S strongly disrupted selectivity (Linsdell et al. 2000). These two mutations also greatly reduced Cl⁻ conductance (Fig. 5). In order to determine whether the changes in relative anion conductance observed in F337S (Table 1) were associated with the change in selectivity or the change in conductance in this mutant, relative anion conductance was also examined in F337Y (Fig. 8). This mutant shows a reduction in Cl⁻ conductance similar to that seen in F337S (Fig. 5), but its anion selectivity is identical to that of wild-type (Linsdell et al. 2000). Two of the most striking effects of F337S on relative anion conductance, the 67 % reduction (compared with wild-type) in g_Br/g_Cl and the 710 % increase in g_F/g_Cl, were not reproduced in F337Y (Fig. 8). This strongly suggests that the alteration in anion relative conductance observed in F337S is related to the alteration in anion selectivity previously reported in this mutant (Linsdell et al. 2000).

Figure 8. Anion conductance changes observed in F337S are not seen in F337Y.

Relative Cl⁻, Br⁻ and F⁻ conductances were estimated under symmetrical ionic conditions by macroscopic current variance analysis. Mean of data from 3-9 patches.

Thiocyanate conductance

Macroscopic current variance analysis suggested a much lower conductance for SCN⁻ than for Cl⁻ (Fig. 7, Table 1). Although this result is in contrast to previous single channel recordings from CFTR-transfected CHO cells immediately after SCN⁻ exposure (Tabcharani et al. 1993), it is in agreement with macroscopic I-V relationships recorded under bi-ionic conditions (SCN⁻ intracellular, Cl⁻ extracellular) (Fig. 9). The conductance ratio, g_SCN/g_Cl, estimated from the slope of the I-V curve as described in Methods, was 0.27 ± 0.05 (n= 4). This is significantly higher than the value of 0.060 ± 0.012 (n= 5) (P < 0.01) estimated from current variance analysis (Table 1), probably reflecting block of Cl⁻ currents by trans-SCN⁻ ions (Tabcharani et al. 1993). Estimation of relative conductance from bi-ionic I-V relationships (Fig. 9) is also sensitive to anion-dependent changes in channel open probability. However, this presumably did not affect the relative SCN⁻ conductance estimated in this way, because ‘locking’ CFTR channels open by adding 1 mM pyrophosphate (PP_i) to the intracellular solution, which should circumvent changes in channel gating, did not alter the shape of the I-V curve (Fig. 9). Following addition of PP_i, g_SCN/g_Cl was 0.32 ± 0.06 (n= 4), not significantly different from the pre-PP_i value given above (P > 0.5).

Anion block of Cl⁻ currents

Permeant anions which bind relatively tightly within the channel pore may also reduce Cl⁻ currents when added at low concentrations (Tabcharani et al. 1993; Mansoura et al. 1998). Figure 10 shows CFTR single channel Cl⁻ currents recorded at −50 mV with 150 mM NaCl on both sides of the membrane, and following the addition of other permeant anions to the intracellular solution. The lyotropic anions SCN⁻, I⁻ and ClO₄⁻ significantly reduced Cl⁻ current amplitude when added to the intracellular solution at a concentration of 10 mM (Fig. 11A), suggesting that all bind relatively tightly within the pore. In contrast, other permeant anions (Br⁻, formate, NO₃⁻, F⁻) were without significant effect, even when added at 25 mM (Fig. 11A). Block by these anions is likely to be voltage dependent; however, at −50 mV entry of these blockers into the pore from the intracellular solution will be favoured. These results support the hypothesis that anion binding within the CFTR pore, like anion permeability, follows a lyotropic sequence.

Figure 10. Block of Cl⁻ permeation in wild-type CFTR by other permeant anions.

Single channel currents recorded from CHO cell patches with symmetrical 150 mM NaCl, and following the addition of 10 mM NaSCN, 10 mM NaI, 10 mM NaClO₄ or 25 mM NaF to the intracellular solution. Channel activity was recorded at a membrane potential of −50 mV; the line to the left indicates all channels closed.

Figure 11. Block of wild-type, F337S and T338A CFTR by intracellular permeant anions.

Relative current amplitudes (at −50 mV) in the presence of different intracellular permeant anions were estimated from single channel recording (for wild-type, see Fig. 10, and also for T338A), or from macroscopic current variance analysis for F337S. *Significant difference from control in each case (*P < 0.05, **P < 0.001). Mean of data from 3-7 patches in each case.

The ability of permeant anions to block Cl⁻ currents was also examined in selectivity altering mutants, using macroscopic current variance experiments (for the low-conductance F337S) or single channel recording (for the high-conductance T338A), at −50 mV (Fig. 11). In F337S, block by SCN⁻ and I⁻ was somewhat weakened, and block by 10 mM ClO₄⁻ was abolished (although 25 mM ClO₄⁻ did significantly reduce Cl⁻ current amplitude; data not shown). However, in contrast to wild-type, F337S Cl⁻ currents were blocked by 25 mM Br⁻ (Fig. 11B), although they were not significantly affected by 10 mM Br⁻ (data not shown). T338A Cl⁻ currents were only weakly blocked by SCN⁻ and ClO₄⁻, and not significantly blocked by other permeant anions including I⁻ (Fig. 11C). Since block was examined only at a single voltage (-50 mV), differences between CFTR variants may reflect changes in blocker affinity and/or voltage dependence. T338A Cl⁻ currents were significantly increased in the presence of 25 mM intracellular NO₃⁻, suggesting that the high permeability and conductance of NO₃⁻ in this mutant (Table 1) allowed some degree of summation of Cl⁻ and NO₃⁻ currents through the channel.

The relationship between anion permeability and anion conductance

The results summarised in Tables 1 and 2 indicate that the CFTR mutation F337S, which virtually abolishes the lyotropic pattern of anion permeability (Linsdell et al. 2000), also alters the relative conductance of different anions in the pore, whereas T338A, which strengthens the lyotropic nature of anion permeability (Linsdell et al. 1998), has no significant effect on relative conductance. To understand the interactions between permeant anions and this region of the pore, and how they affect anion permeation, it is useful to analyse how permeability and conductance are related in these channels.

Anion permeability in wild-type and T338A CFTR follows an approximate lyotropic sequence, with relative anion permeability being correlated with energy of hydration (Linsdell & Hanrahan, 1998; Linsdell et al. 1998; Fig. 12A). In contrast, there is no obvious relationship between relative anion conductance and energy of hydration in either wild-type or T338A (Fig. 12B). In both cases, conductance is optimal for Cl⁻, high for those anions with similar energies of hydration (NO₃⁻, Br⁻, formate) and lower both for more kosmotropic (F⁻) and more lyotropic (SCN⁻, ClO₄⁻) anions. However, the small number of anions studied precludes any strong conclusions in this area. The relationship between anion permeability and energy of hydration observed in wild-type and T338A is lost in F337S (Fig. 12A), which we previously suggested reflected a reduction in the relative importance of anion dehydration in determining anion permeability in this mutant (Linsdell et al. 2000). However, anion conductance does appear weakly correlated with energy of hydration in F337S, although this is mainly due to the dramatic increase in relative conductance of the highly kosmotropic F⁻ ion (Fig. 12B). Interestingly, Cl⁻ remains the anion with the highest conductance in this mutant, and the difference in conductance between Cl⁻ and anions with similar energies of hydration (Br⁻ and NO₃⁻) is greater than for wild-type or T338A (Fig. 12B).

Figure 12. Dependence of relative anion permeability (A) and relative anion conductance (B) on anion-free energy of hydration in wild-type, F337S and T338A CFTR.

Values of P_X/P_Cl and g_X/g_Cl in each case are as given in Table 1. Free energies of hydration (G_h) were taken from Marcus (1997).

For the halides Br⁻ and F⁻, as well as NO₃⁻ and ClO₄⁻, there is a clear correlation between relative permeability and relative conductance among the three CFTR variants studied here; variants showing a high permeability to these anions also show a high relative conductance (Table 1). This applies both to lyotropic anions with increased permeability in T338A (Br⁻, NO₃⁻, ClO₄⁻) and the kosmotropic anion F⁻, which shows increased permeability in F337S. A similar relationship is seen with I⁻, when its relative conductance was estimated from the slope of the macroscopic I-V curve (Fig. 13). Since I⁻ currents were not observed in either CHO or BHK cell membranes under symmetrical ionic conditions, this was the only way in which some indication of changes in I⁻ conductance could be obtained. The conductance ratio, g_I/g_Cl, was 0.20 ± 0.03 (n= 4) for wild-type, 0.14 ± 0.05 (n= 4) for F337S, and 0.59 ± 0.09 (n= 4) for T338A. Although it is possible that these differences are affected by changes in trans-ion effects between different CFTR variants, the overall pattern is consistent with that seen with Br⁻, F⁻, NO₃⁻ and ClO₄⁻; high permeability is correlated with high conductance.

This relationship between permeability and conductance was not observed for formate or SCN⁻. Formate permeability was not strongly affected by mutation of F337, which affected mainly the selectivity between small anions (Linsdell et al. 2000). SCN⁻ conductance was low in all mutants (as was ClO₄⁻ conductance), and SCN⁻ was an effective blocker of all three CFTR variants studied; perhaps SCN⁻ binding to other sites within the pore limits its conductance.

DISCUSSION

The precise molecular architecture of the Cl⁻ channel pore of CFTR remains unclear, although TM6 certainly plays a key role in forming the pore and determining its permeation properties (Dawson et al. 1999; McCarty, 2000). While other TMs presumably also contribute to the pore, their identity and roles are uncertain (Dawson et al. 2000; McCarty, 2000).

Permeant ions have been described as uniquely useful probes of the CFTR channel pore, as they presumably experience all of the forces which act on Cl⁻ ions during the normal permeation process (Mansoura et al. 1998; Dawson et al. 1999). Several mutations within the pore of CFTR alter permeant anion binding without strongly affecting anion selectivity (in terms of permeability ratios) (e.g. K335E, Anderson et al. 1991; G314E, Mansoura et al. 1998), suggesting that these mutations affect anion binding sites not intimately involved in the anion selectivity process (Smith et al. 1999; Linsdell et al. 2000). The present study sought to investigate the role of anion binding within the pore in determining selectivity, by comparing the anion binding properties of CFTR variants with differing anion selectivity sequences.

Comparison with previous work

Previous studies have estimated the relative conductance of different anions in the CFTR pore (Gray et al. 1990, 1993; Anderson et al. 1991; Overholt et al. 1993; Tabcharani et al. 1997; Mansoura et al. 1998; Illek et al. 1999). However, these studies have generally used only one-sided substitution of Cl⁻, and as such the results are subject to trans-ion effects, i.e. block of the current carried by one ion by the ion present on the opposite side of the membrane. This problem is exemplified in Fig. 9, where one-sided substitution of SCN⁻ yields an apparent relative conductance for this anion, g_SCN/g_Cl, of 0.27, compared with 0.06 estimated by current variance analysis with SCN⁻ present on both sides of the membrane. One-sided substitution by SCN⁻ is presumed to reduce Cl⁻ conductance, due to SCN⁻ block of Cl⁻ currents, leading to an overestimation of relative SCN⁻ conductance. In the present study, trans-ion effects were avoided by measuring the conductance to different anions under symmetrical ionic conditions.

Among those anions which gave measurable single channel currents under symmetrical ionic conditions, the conductance sequence was Cl⁻ (1.0) > NO₃⁻ (0.60) > Br⁻ (0.48) > formate (0.32-0.43) > F⁻ (∼0.12) (Figs 1 and 2, Table 1). Current variance analysis yielded similar results for these anions while allowing the range of anions tested to be extended, giving an overall conductance sequence (at +50 mV) of Cl⁻ (1.0) > NO₃⁻ (0.73) > Br⁻ (0.48) = formate (0.49) > F⁻ (0.09) > SCN⁻ (0.06) = ClO₄⁻ (0.06). Since some anions (e.g. formate) may show a rectifying i-V relationship, the conductance sequence may be different at other voltages. The shape of the current-voltage relationship with symmetrical F⁻, SCN⁻ or ClO₄⁻ was not investigated. These conductance sequences are quite similar to those obtained from whole cell CFTR currents when extracellular Cl⁻ was replaced by other anions, both in rat pancreatic duct cells (Cl⁻ (1.0) > Br⁻ (0.83) > ClO₄⁻ (0.41); Gray et al. 1993) and in Xenopus oocytes (Cl⁻ (1.0) > NO₃⁻ (0.75) > Br⁻ (0.64) > SCN⁻ (0.14); Mansoura et al. 1998).

A major discrepancy between the present and previous studies was the inability to record single channel currents carried by SCN⁻ or I⁻. Thiocyanate currents with a conductance of 10.3 pS, and I⁻ currents with a conductance of 5.4 pS have previously been reported for CFTR expressed in CHO cells immediately following symmetrical exposure to SCN⁻- or I⁻-containing solutions (Tabcharani et al. 1993, 1997). In the present study, only a low SCN⁻ conductance state was identified using current variance analysis, with a g_SCN/g_Cl of 0.06 (Fig. 6, Table 1), similar to that reported for whole cell CFTR currents in Xenopus oocytes (Mansoura et al. 1998). Iodide currents could not be recorded under symmetrical conditions, but macroscopic I-V relationships showed the conductance sequence Cl⁻ (1.0) > SCN⁻ (0.27) > I⁻ (0.20) (Figs 9 and 13). While subject to the kind of trans-ion effects described above, this apparent low I⁻ conductance is consistent with previous whole cell (Anderson et al. 1991; Gray et al. 1993; Overholt et al. 1993; Mansoura et al. 1998) and transepithelial current results (Illek et al. 1999). One possible explanation for the discrepancy with the single channel results of Tabcharani et al. (1993, 1997) is that CFTR may exist in different states with different conductances to highly lyotropic anions such as SCN⁻ and I⁻ (Tabcharani et al. 1997). There is additional evidence that CFTR may exist in multiple open states with different permeation properties (Gunderson & Kopito, 1995; Ishihara & Welsh, 1997; Linsdell & Hanrahan, 1998). CFTR may exist only transiently in a state with a high conductance to anions such as SCN⁻ and I⁻, a state which may have been missed in the present and other studies. Thus, single channel currents carried by I⁻, showing both high relative permeabilities P_I/P_Cl (1.8-2.0) and high g_I/g_Cl (0.8), disappear within 20-120 s following exposure to I⁻-containing solutions, depending on the experimental conditions (Tabcharani et al. 1997). Under steady-state conditions, both P_I/P_Cl and g_I/g_Cl are considerably lower (Tabcharani et al. 1997; Table 1). The high SCN⁻ conductance state previously described by Tabcharani et al. (1993) is only observed for a similarly short period following exposure to SCN⁻-containing solutions (J. A. Tabcharani & J. W. Hanrahan, personal communication). The results shown in Fig. 6 indicate that, under steady-state conditions, CFTR exists in a stable, low SCN⁻ conductance state.

Previous studies have also shown that the lyotropic anions SCN⁻ (Tabcharani et al. 1993; Mansoura et al. 1998) and I⁻ (Tabcharani et al. 1992) block Cl⁻ permeation through the CFTR channel. Similar effects were observed in the present study with these two anions and also ClO₄⁻ (Figs 10 and 11A). No other permeant anion tested (Br⁻, formate, NO₃⁻, F⁻) significantly affected Cl⁻ permeation through the wild-type CFTR pore at concentrations of 25 mM (Fig. 11A).

Anions bind within the pore

Anion binding has previously been described as a signal feature of the CFTR Cl⁻ channel pore (Dawson et al. 1999), consistent with the notion that permeating ions bind briefly to discrete sites as they traverse ion channel pores (Hille, 1992). The present results reiterate two key pieces of evidence for permeant anion binding within the CFTR pore. First, several anions with a higher permeability than Cl⁻ (Br⁻, NO₃⁻, SCN⁻) have a lower conductance than Cl⁻ (Table 1), suggesting that their passage through the pore is retarded by relatively tight binding to intrapore sites. As noted previously (Mansoura et al. 1998), the discrepancy between relative permeability and relative conductance is particularly striking for SCN⁻, which enters the pore readily (P_SCN/P_Cl= 3.55) but passes through the pore very slowly (g_SCN/g_Cl= 0.06). Secondly, low concentrations of some permeant anions (SCN⁻, I⁻, ClO₄⁻) are able to reduce Cl⁻ flux through the pore (Figs 10 and 11), suggesting that they bind relatively tightly within the pore and slow the overall rate of ion permeation.

CFTR shows a lyotropic anion selectivity sequence, with anion permeability being correlated with the ease of anion dehydration (Fig. 12A; Linsdell & Hanrahan, 1998; Smith et al. 1999; Dawson et al. 1999; Linsdell et al. 2000). Chloride permeation through the wild-type CFTR pore was significantly reduced only by the most lyotropic anions studied (SCN⁻, I⁻, ClO₄⁻), consistent with the hypothesis that the tightness of anion binding within the pore shows a similar dependence on anion dehydration to that shown by anion permeability. This might suggest that high permeability in CFTR results from tight binding, as is thought to occur in Ca²⁺ channels (Almers & McCleskey, 1984; Hess & Tsien, 1984; Tsien et al. 1987; Ellinor et al. 1995). Alternatively, it has been suggested that this apparent correlation exists because anion permeability and anion binding, although independent processes, are both controlled primarily by small differences between anion-water interaction energies and anion-channel interaction energies (Smith et al. 1999).

Mutations which affect anion selectivity also affect anion binding

Previously we showed that mutation of adjacent TM6 residues F337 and T338 strongly affected the anion selectivity of CFTR, as judged by changes in the relative permeabilities of different anions and in the anion permeability sequences (Linsdell et al. 1998, 2000). These findings were in contrast to other studies showing that mutations affecting other putative pore regions had very little effect on selectivity (see Dawson et al. 1999), leading us to suggest that CFTR anion selectivity is determined predominantly over a physically discrete pore region close to F337/T338 (Linsdell et al. 2000). In the present study, the anion binding properties of two mutants, F337S and T338A, were examined in detail. F337S strongly altered the relative conductance of different anions, leading to significant decreases in the conductance of NO₃⁻, Br⁻ and formate, and significant increases in the relative conductance of F⁻ and SCN⁻ (Table 1). These changes altered the conductance sequence for permeant anions (Table 2). In contrast, T338A did not significantly affect the relative conductance of different anions (Table 1), although some small changes in the conductance sequence were apparent (Table 2). The effects of the mutation F337S on relative anion conductance were not mimicked by another mutation, F337Y, which shows a similarly reduced Cl⁻ conductance but unaltered anion selectivity (Fig. 8). This indicates that the mutation F337S causes a change in the architecture of the pore which alters anion permeability and anion conductance simultaneously, perhaps by modifying an anion binding site.

Both mutations also affected the block of Cl⁻ permeation by other permeant anions. Block by SCN⁻, I⁻ and ClO₄⁻ was significantly weakened in both F337S and T338A, and F337S showed significant block by high concentrations of Br⁻ not evident in wild-type (Fig. 11). These results suggest that lyotropic anion binding is weakened in both F337S and T338A, in spite of the ‘opposite’ effect of these two mutations on anion selectivity. Thus, the ability of permeant anions to block Cl⁻ permeation does not appear to be strongly correlated with their permeability.

Although these results implicate the region around F337 and T338 in anion binding within the pore, other intrapore anion binding sites undoubtedly exist. Mutation of other sites has previously been shown to affect anion binding (e.g. K95, K335, Anderson et al. 1991; K335, R347, Tabcharani et al. 1993; S341, S1141, McDonough et al. 1994; R347, Linsdell & Hanrahan, 1996; G314, K335, Mansoura et al. 1998; S1118, Zhang et al. 2000), in some cases indirectly via a gross change in pore architecture (Cotten & Welsh, 1999). Interestingly, S1118 in TM11 has been suggested to occupy a position similar to that of T338 in TM6, and the mutations S1118A and S1118F cause small alterations in anion permeability (Zhang et al. 2000). Anion binding to other sites may modify anion conductance without strongly affecting anion permeability; in fact, tight binding to other sites may be mainly responsible for the low conductance of lyotropic anions such as SCN⁻ (Smith et al. 1999).

Implications for the mechanism of anion selectivity

The mutations F337S and T338A caused co-ordinated changes in the relative permeability and relative conductance of Br⁻, F⁻, NO₃⁻, ClO₄⁻ and I⁻ ions, with high permeability being associated with high conductance (Table 1; Fig. 13). This indicates that the CFTR channel pore does not achieve lyotropic anion selectivity via selective lyotropic anion binding in the region around these crucial amino acid residues, as weakening of anion binding (increased conductance) is associated with increased permeability, and tighter binding with lower permeability, for these anions. Thus, although wild-type CFTR is characterised by a lyotropic anion selectivity sequence and also shows tight binding of lyotropic anions, there does not appear to be a causal relationship between these two aspects of pore function (see also Smith et al. 1999). This is in contrast with the accepted model of cation selectivity in Ca²⁺ channels, which suggests that Ca²⁺ selectivity is achieved by selective, high affinity binding of Ca²⁺ ions within the pore (Tsien et al. 1987; Ellinor et al. 1995). Instead, lyotropic anion selectivity in CFTR channels may be achieved by a mechanism more reminiscent of that of K⁺ selectivity in K⁺ channel pores, where the physiological ion of interest enters the selectivity filter more readily than other ions (Doyle et al. 1998). Thus, not only in wild-type CFTR but also both F337S and T338A, Cl⁻ is the anion with the highest conductance (Table 1), even though other anions may have higher permeabilities. An analogous situation occurs in K⁺ channels; K⁺ shows the highest conductance through the channel, even when other cations (such as Tl⁺) are more permeant (Coronado et al. 1980; Eisenman et al. 1986; Heginbotham & MacKinnon, 1993; Choe et al. 2000). In contrast, although Ca²⁺ channels are highly selective for Ca²⁺ ions in terms of permeability, in pure solutions the conductance to other cations may greatly exceed Ca²⁺ conductance (Hess et al. 1986; Tsien et al. 1987). Optimization of Cl⁻ conductance, rather than Cl⁻ permeability, is likely to be of greater importance to the physiological function of CFTR and other Cl⁻ channels, although in CFTR the mutations T338A and T338S increase Cl⁻ conductance, apparently without adversely affecting other channel properties (Linsdell et al. 1998). In neuronal GABA_A and glycine receptor channels, which also show a lyotropic anion permeability sequence, Cl⁻ is also the anion with the highest conductance (Bormann et al. 1987; Fatima-Shad & Barry, 1993).

The present results support a model whereby entry of anions into the region around F337 and T338 facilitates their passage through the pore; anions which enter into this ‘selectivity’ region more easily will show both high relative permeability and high relative conductance. The conductance of highly lyotropic anions, in particular SCN⁻, may be reduced by binding relatively tightly to other sites within the pore which have little or no impact on permeability. In wild-type CFTR, the ability of anions to enter the ‘selectivity region’ appears to be dominated by anion free energy of hydration, perhaps because some degree of anion dehydration occurs as the anions enter this region. In F337S, a reduction in the relative importance of anion dehydration allows kosmotropic anions such as F⁻ to enter more easily, leading to a dramatic increase in both P_F/P_Cl and g_F/g_Cl in this mutant. In contrast, entry of Br⁻, NO₃⁻ and I⁻ is impeded in F337S, decreasing both the permeability and conductance of these relatively lyotropic anions.