Cystic Fibrosis-associated Mutations at Arginine 347 Alter the Pore Architecture of CFTR

Arginine 347 in the sixth transmembrane domain of cystic fibrosis transmembrane conductance regulator (CFTR) is a site of four cystic fibrosis-associated mutations. To better understand the function of Arg-347 and to learn how mutations at this site disrupt channel activity, we mutated Arg-347 to Asp, Cys, Glu, His, Leu, or Lys and examined single-channel function. Every Arg-347 mutation examined, except R347K, had a destabilizing effect on the pore, causing the channel to flutter between two conductance states. Chloride flow through the larger conductance state was similar to that of wild-type CFTR, suggesting that the residue at position 347 does not interact directly with permeating anions. We hypothesized that Arg-347 stabilizes the channel through an electrostatic interaction with an anionic residue in another transmembrane domain. To test this, we mutated anionic residues (Asp-924, Asp-993, and Glu-1104) to Arg in the context of either R347E or R347D mutations. Interestingly, the D924R mutation complemented R347D, yielding a channel that behaved like wild-type CFTR. These data suggest that Arg-347 plays an important structural role in CFTR, at least in part by forming a salt bridge with Asp-924; cystic fibrosis-associated mutations disrupt this interaction.

Studies of the effect of cystic fibrosis (CF)-associated mutations have been of value in identifying structurally and functionally important regions of CFTR. At least four CF-associated mutations have been identified at position 347 in M6: R347C, R347H, R347L, and R347P, suggesting that Arg-347 is important for CFTR structure and function (13)(14)(15). 2 Early studies by Sheppard et al. (7) showed that mutation of Arg-347 to proline significantly decreased single-channel conductance with little effect on CFTR trafficking to the plasma membrane. Other work by Tabcharani et al. (8,16) and Linsdell and Hanrahan (17) emphasized the importance of Arg-347 for anomalous mole-fraction behavior, iodide permeability, and voltagedependent block by DIDS, in addition to single-channel conductance. Interestingly, mutation of Arg-347 to a histidine (R347H) produced a channel that displayed pH-dependent conductance and anomalous mole-fraction behavior (8). These studies suggested that Arg-347 may line the pore and that a positive charge at position 347 is sufficient for wild-type conductance. Because mutation of Arg-347 eliminated anomalous mole-fraction behavior, Arg-347 itself was proposed to be an anion binding site in the CFTR pore (8), and the presumed positive charge introduced upon protonation of His-347 was thought to facilitate interactions with permeating anions. An alternative interpretation is that Arg-347 may be important for maintenance of pore architecture without contributing directly to the permeation pathway. For example, mutation of this site could lead to a change in MSD conformation and loss of an anion-binding site(s) elsewhere; protonation of His-347 might then rescue the conformation of the R347H mutant.
As discussed by Perutz (18), charged residues within proteins reside in locations where they are either solvated or can interact with and be neutralized by oppositely charged residues. These electrostatic interactions mediate an important stabilizing effect, providing increased thermostability and resistance to denaturation. Based on these considerations, it is possible that Arg-347 may line the pore where it can interact with either water or permeant anions. However, it is also possible that Arg-347 may mediate a structural role in the MSDs; there are a number of negatively charged residues with which Arg-347 might interact. Both possibilities are consistent with the present data.
To better understand the role of Arg-347 in CFTR structure and function, we examined the effect of mutating Arg-347 to cysteine, aspartic acid, glutamic acid, lysine, and leucine on CFTR conductance. We examined the cytosolic pH (pH c )-dependent behavior of CFTR-R347H and that of the other residue 347 mutants both with (R347C, R347D, R347E, and R347K) and without (R347L) a pH c -titratable residue. The conductance of CFTR-R347H is pH c -dependent. Because the site of protona-tion may exist in one of two states, either protonated or deprotonated, we tested the hypothesis that CFTR-R347H may display two pH c -dependent conductance states, which it did. If the protonatable site lines the pore, then the two ionization states of the protonatable residue might yield two distinct conductance states. Alternatively, if the protonatable site influences structure, then the two conductance states might represent two distinct conformational states of CFTR. Moreover, like CFTR-R347H, all the other residue 347 mutants, except R347K, displayed two pH c -dependent conductance states over a similar pH c range. The residue at position 347 did not influence current flow through either conductance state. These data suggested that residue 347 probably does not line the pore but likely stabilizes CFTR structure. To pursue this, we studied the effect of mutating residues elsewhere in the MSDs that might interact with Arg-347.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and Transfection-All mutants were constructed in the pTM1-CFTR4 plasmid by the method of Kunkel (19). The mutagenesis was confirmed by restriction digestion of silently introduced restriction sites and by sequencing around the introduced mutation site. In vitro transcription and translation of each mutant was performed to confirm expression of full-length protein. Wild-type and mutant channels were expressed transiently in HeLa cells using the vaccinia virus/T7 bacteriophage hybrid expression system as described previously (20). Cells were routinely studied 12 to 24 h after infection-transfection.
Patch-Clamp Technique-Methods used for excised, inside-out patch clamp recordings were as described previously (21)(22)(23). Voltages were referenced to the extracellular side of the membrane. All studies were done at room temperature (22-24°C) to facilitate kinetic analysis. The membrane potential was clamped at Ϫ120 mV unless otherwise indicated.
CFTR was activated by excising patches into a bath solution (pH c 7.3; Tricine or TES) containing 1 mM ATP and 75 nM catalytic subunit of cAMP-dependent protein kinase (Promega Corp., Madison, WI). During the pH c studies, cAMP-dependent protein kinase was removed, and bath (cytosolic side) ATP concentration was adjusted (usually to less than 0.05 mM ATP) to resolve single-channel bursting activity within multichannel macropatches. For experiments with excised, inside-out patches, the pipette (extracellular) solution contained (in mM): 140 N-methyl-D-glucamine (NMDG), 140 aspartic acid, 10 Bis-Tris, 5 CaCl 2 , 2 MgSO 4 , pH o 6. The bath (intracellular) solution contained (in mM): 140 NMDG, 10 Bis-Tris, 3 MgCl 2 , 4 CsOH/1 EGTA, pH c 6.5 (unless indicated) with HCl ([Ca 2ϩ ] free Ͻ 10 Ϫ8 ). We selected Bis-Tris as a buffer to avoid the blocking effect that Good's type buffers such as MOPS have on CFTR (24). pH was adjusted at room temperature (22-24°C) using a Model 10 Accumet meter fitted with a gel-filled combination electrode calibrated with pH 6.0 or 7.0 standardized buffer solutions, where appropriate (all pH analytical equipment was from Fisher Scientific, Pittsburgh, PA).
Data Analysis-Single-channel current amplitudes were determined from the peaks in all-points histograms. Single-channel conductances were derived from the slope of the single-channel linear I-V relationship. Single-channel data were filtered at 1000 Hz using an 8-pole Bessel filter (902LPF, Frequency Devices, Haverhill, MA), digitized at 5000 Hz, and digitally filtered at 500 Hz. The resolution of the A/D converter was 0.0012 pA. Events lists were generated using a halfheight transition protocol; transitions less than 1 ms were excluded. For events list generation, a prompt was positioned by eye at each current level representing the closed state, the little conductance state (O L ), and the big conductance state (O B ). The lifetimes of individual sojourns in O L or O B were collected, binned (10 bins per decade), and fitted with a one-component exponential using the maximum likelihood method. Wild-type CFTR enters a short-lived intraburst closed state; for simplicity we omitted this state in our analysis of residue 347 mutants since it is rare in wild-type CFTR in the absence of MOPS or other blocking buffers (approximately once every 213 ms within bursts of duration equalling 649 ms; n ϭ 2 at pH c 6.0, Ϫ120 mV, 22-24°C) on the time-scale of transitions between the O L and O B state (24). Singlechannel current variance analysis was done as described in the legend for Fig. 3 and in Ref. 25. Data acquisition and analysis were done using the pClamp software package (Axon Instruments Inc., Foster City, CA) and Excel 5.0 (Microsoft Corp., Redmond, WA).

RESULTS
pH c -dependent Conductance of Residue 347 Mutants-To determine whether R347H exhibits two discrete conductance states, we studied single channels in excised, inside-out patches. Fig. 1 shows a representative single-channel trace FIG. 1. Single-channel current tracings from excised, insideout membrane patches containing the wild-type or indicated residue 347 mutant channels. All current tracings were collected with membrane voltage clamped at Ϫ120 mV at the indicated intracellular pH c ; the pipette pH o was 6.0 throughout. All-points histograms were derived from 7-40 s of data using a binwidth of 9.8 or 19.5 fA and are shown to the right of current tracings. C refers to closed state. from R347H displaying two pH c -dependent conductance states, O L and O B . Unexpectedly, all the other variants at residue 347, except R347K, showed two pH c -dependent conductance states over a similar pH range. Even R347L, which does not have a titratable side chain, displayed this behavior (Fig. 1). The conservative substitution of Arg-347 by Lys (R347K) showed only a single conductance state and no pH c dependence. Similar results were observed when NMDG-Cl was replaced with NaCl or when Bis-Tris was replaced with MES (n ϭ 2 each, not shown), suggesting that this behavior is not due to block by a pH c -dependent change in one of the pH-titratable bath reagents, NMDG or Bis-Tris. The pH of the pipette solution did influence the results (not shown). Also, there was no state-dependent bias in opening or closing. Channels opened into or closed from the O L or O B states as predicted based solely upon pH c .
The all-points histograms ( Fig. 1) illustrate qualitatively that as pH c increased, each mutant except R347K spent an increased fraction of time in O L and a correspondingly decreased fraction of time in O B . Points in the histogram from the zero current state (C), which are largely a function of ATP concentration (26 -28), are included only for purposes of clarity. Wild-type CFTR and R347K possess only one predominant conductance state over the pH c range 5.5-7.3 ( Fig. 1, data not shown, and Ref. 8). This result suggests that a large positively charged residue at position 347 (Arg or Lys) in CFTR stabilizes the O B state relative to the O L state. Visual inspection suggested that the lifetimes of O L and O B states were also influenced by the nature of the residue at position 347: R347E and R347H tended to have longer dwell times in the O L and O B states, whereas R347L, R347C, and R347D tended to display shorter dwell times. In general, the smaller the residue, the more rapid the kinetics. Perhaps steric hindrance imposed by the residue at position 347 might explain the size dependence of the kinetics. For R347D, the lifetime in the O B state was so short that a discrete O B was not apparent on the all-points histogram; instead, as pH c decreased, a shoulder developed on the O L state distribution in the all-points histogram (Fig. 1). Since R347L displayed two pH c -dependent conductance states and since leucine is aliphatic and non-ionizable, the pH c dependence of residue 347 mutants cannot be attributed simply to protonation of residue 347.
Single-channel Conductance of Residue 347 Mutants-To determine whether the residue at position 347 affects singlechannel conductance and not merely the conductance state of the channel, we examined the I-V relationship and slope conductance of the mutants with slower pH c -dependent kinetics, R347H and R347E, as well as R347K. Filtering obscured the true current amplitudes of the other pH c -dependent mutants with more rapid kinetics. Fig. 2 shows that the I-V relationships and slope conductances for the O L and O B states were not significantly affected by the nature of the residue at position 347. The single-channel conductance at pH c 6.0 of wild-type CFTR, R347K, and the O B state of R347E and R347H were all very similar (in pS): 7.7 Ϯ 0.4, 8.3 Ϯ 0.6, 7.4 Ϯ 0.4, and 6.9 Ϯ 0.2, respectively (n ϭ 3 for each). The single-channel conductance of the O L states of R347E and R347H at pH c 6.0 were also very similar (in pS): 1.5 Ϯ 0.1 and 1.6 Ϯ 0.1, respectively (n ϭ 3 and 4 for each). These data suggest that the amino acid at residue 347 does not affect single-channel current amplitude, rather the predominant effect is on the lifetime of the O L and O B conductance states. The lack of effect of the specific residue at position 347 on the rate of Cl Ϫ flow through the pore in either the O L or O B conductance state suggests that residue 347 does not interact directly with permeating anions.
There are at least two possible explanations for the pH c -dependent behavior of residue 347 mutants. One explanation is that mutations at residue 347 reveal the effect of a protonatable residue in the permeation path that directly interacts with anions. In this case, the lifetime of the O B and O L state represent the lifetime of the protonated and deprotonated ionization states of this unknown site, respectively. A second explanation is that when Arg-347 is mutated, the membrane-spanning domains that form the pore fluctuate between two conformational states, O L and O B ; protonation or deprotonation of some unknown site favors the O B or O L conformational states, respectively. The first explanation seems unlikely because the reciprocal lifetime of the O L state which represents the "on" rate for the proton is very slow (e.g. it is 6 ϫ 10 7 M Ϫ1 s Ϫ1 for R347E). This rate is ϳ200-fold slower than proton transfer onto an imidazole in free solution (29,30) and ϳ10-fold slower than transfer onto an imidazole in the pore of ROMK1 (25). The slow rate, however, might be explained in part by residence of the protonatable site in a sterically and electrostatically shielded position. In addition, the site of protonation revealed by a residue 347 mutation would have to perfectly replace the lost anion-binding site incurred through the mutagenesis; this seems exceedingly fortuitous. This consideration convinced us to favor a model involving a conformational change in CFTR.
Dwell-time Analysis of O L and O B States-We performed a dwell-time analysis of the lifetimes of the O L and O B states of R347E and R347H to enable more quantitative comparisons between them and to better understand their pH c dependence. Both O L and O B dwell-time histograms were best fit by a single exponential function at all pH c values examined. This suggests that the distribution between O L and O B state does not change markedly throughout the gating cycle; this can also be observed visually in Fig. 1. Fig. 3A shows that the reciprocal lifetime of the O L state increased relatively linearly with increasing proton concentration. This indicates that the rate of entry into the O B state increased with increasing proton concentration. This linear, first order dependence on proton concentration suggests that protonation of only a single site limits the rate of movement from the O L to the O B state. The rate of exit from the O B state decreased non-linearly with increasing proton concentrations. The kinetics and the equilibria between O L and O B were very similar for both mutants (Fig. 3A). The observable pK (0 mV) for the equilibrium between O L and O B of R347E and R347H were 6.4 and 6.3, respectively. The faster kinetics of R347D, R347C, and R347L made dwell-time analysis for these mutants less reliable. Therefore, to obtain an estimate of the approximate pK of these mutants, we examined changes in the variance of the open-state current with changes in pH c (25). The current variance should go through a maximum when the pH c equals the observable pK. Each mutant displayed an increase in current variance with increasing proton concentration (i.e. decreasing pH). Fig. 3B shows that R347C, R347D, and R347L did not reach a peak variance over the range of pH c studied, suggesting that their apparent pK is less than 5.0 -5.5. Visual inspection of the tracings suggested that the mutants were largely in their O L state over the range of pH c employed such that we did not miss a peak in the variance. In this analysis, we assumed that the mutants would display minimal pH-independent movement between O L and O B states at very acidic pH c so that the variance should pass through a maximum as pH c decreases. As a control for the variance analysis, we examined the R347E mutant on which we had also done dwell-time analysis (Fig. 3A); as expected, Fig. 3B shows that the variance of R347E goes through a maximum between pH c 6.5 and 5.5.

Voltage Dependence of O L and O B -
To learn more about the nature of the two conductance states of residue 347 mutants, we examined the voltage dependence of the O L and O B lifetimes. Fig. 4A shows qualitatively that both conductance states of R347E were voltage-dependent. Fig. 4B shows quantitatively that at pH c 6.0 the O L and O B states for R347E and R347H were both influenced by the transmembrane voltage. The voltage dependence may derive from alterations in the distribution of protons near the protonatable site or from charged regions of CFTR moving through the voltage field. We assumed the Boltzman distribution to quantify the effect of voltage on the equilibrium between O L and O B states such that pK͑mV͒ ϭ pK͑0 mV͒ ϩ zFE/2.303RT (Eq. 1) in which ϭ electrical distance from the cytosolic surface, z ϭ valence, F ϭ Faraday's constant, E ϭ transmembrane potential, r ϭ gas constant, T ϭ temperature, pK ϭ the negative log of the equilibrium constant between O L and O B conformations. The degree of voltage dependence was similar for both mutants despite the charge differences at residue 347 and yielded a z of 0.25 and 0.21 for R347E and R347H, respectively. The voltage dependence was asymmetrically disposed between the rate of entry into the O B state and the rate of exit from O B (Fig. 4B).
To quantify these differences we used the following: in which ␦ ϭ symmetry factor which partitions voltage dependence (31,32). The rate of exit from the O B state ( B Ϫ1 ) was more voltage-dependent than the rate of exit from the O L state ( L Ϫ1 ) for both mutants (␦ ϭ 0.8 versus 1 Ϫ ␦ ϭ 0.2 for R347E and ␦ ϭ 0.7 versus 1 Ϫ ␦ ϭ 0.3 for R347H). This observation may be explained by the protonatable site moving through the voltage field; protonation and deprotonation of this site before and after the conformational change will affect the net charge migrating through the applied voltage. Since voltage dependence  arises from charge movement through a voltage field and since R347E and R347H displayed similar voltage dependences and carry different charges at position 347, residue 347 is not likely moving through a transmembrane potential during interchange between O L and O B states.
The Phenotype of R347D Is Suppressed by the D924R Mutation-The data suggest that Arg-347 and Lys-347 may stabilize the structure of the pore; in their absence, the channel "flickers" between two conductance states. Arginine and lysine residues through electrostatic interactions with anionic residues are important for the structure of membrane-spanning domains in other proteins such as the Lac permease and the inward-rectified K ϩ channel, IRK1 (33)(34)(35). As discussed by Perutz (18), salt bridges within proteins are an important structural feature that confers thermostability and resistance to denaturation. We hypothesized that Arg-347 may mediate a stabilizing influence by contributing to a salt bridge within the MSDs. There are multiple glutamates and aspartates within transmembrane (M) regions with which Arg-347 or Lys-347 might interact: Glu-92 (M1), Glu-873 (M7), Asp-924 (M8), Asp-993 (M9), and Glu-1104 (M11). To identify the Arg-347 interaction partner, we replaced Arg-347 with an anionic residue (R347E or R347D) and introduced an arginine residue in the place of candidate partners in a salt bridge. We studied the conductance properties of the following double mutants: R347D/D924R, R347D/D993R, and R347E/E1104R. The R347D/D993R and R347E/E1104R mutants each had two conductance states with pH c -dependent behavior (Fig. 5). For R347D/D993R the increased entry into the O B state was apparent as a shoulder on the amplitude histogram at pH c 5.5. Accordingly, for R347D/D993R and R347E/E1104R the current variance in the open state increased with decreasing pH c (Fig.  5B). Qualitatively, the lifetimes of the O L and O B conductance states in the R347D/D993R and R347E/E1104R were similar to that of the R347D and R347E mutants, respectively. The amplitude of the O L state was larger for both of these double mutants as compared with the single mutants (Figs. 1B and 5B). We also observed an infrequent, additional small conductance state in the R347E/E1104 mutant (see amplitude histogram in Fig. 5A); this is likely due to the E1104R mutation itself.
In contrast to the other double mutants, the R347D/D924R mutant did not display the pH c -dependent flicker found in the R347D single mutant (Fig. 5, A and B), and there was no effect of pH on open-channel variance (Fig. 5B). The single-channel conductance was similar to that of wild-type CFTR (6.1 Ϯ 0.1 pS; n ϭ 3; Fig. 2). These data suggest that the D924R mutation compensates for or rescues the phenotype of the R347D mutation. This result predicts that the D924R mutation alone (with Arg at position 347) would generate an unstable channel with at least two open conductance states. Fig. 6 shows that the D924R single mutant displayed multiple (ϳ3) conductance states that appeared to be pH c -independent.

Function of Arg-347 in CFTR-Previous work from our and
Hanrahan's laboratories (7,8,17) has led to the speculation that Arg-347 may be an anion-binding site in the CFTR pore. However in contrast to earlier interpretations, our current data show that residue 347 does not influence permeation properties via a direct interaction with permeating anions. Instead they suggest that Arg-347 may be more important for maintenance of pore architecture.
We found that mutation of residue 347 to glutamate, aspartate, cysteine, histidine, or leucine all produced channels with two distinct conductance states, O L and O B . pH c , and voltage influenced the movement between these two states over a sim-ilar range for all mutants. Additionally, the single-channel slope conductances of R347H and R347E were the same in both O B and O L states. The O B state for each had the same conductance as wild-type CFTR. The average calculated pK a for a glutamate and a histidine within a protein are 4.0 and 6.9 (36), respectively. If His-347 or Glu-347 line the permeation pathway, at pH c 6.0 (bath solution) or pH o 6.5 (pipette solution) His-347 is predicted to be protonated and to have at least a partial positive charge, and Glu-347 should be fully deprotonated and have a negative charge. Since the charge or the structure of the residue at position 347 failed to affect singlechannel conductance, it seems unlikely to be either an anionbinding site or to be positioned such that it interacts sterically or electrostatically with permeating anions. Thus, mutation of Arg-347 may decrease single-channel conductance and anomalous mole-fraction behavior by disrupting pore architecture and the function of some other anion-binding site(s).
The O B and O L Conductance States-All the mutants except R347K showed two pH-dependent conductance states. The equilibria between O B and O L states were similar despite the FIG. 5. A, single-channel current tracings from excised, inside-out membrane patches containing R347E/E1104R, R347D/D924R, and R347D/D993R. Membrane voltage was Ϫ120 mV, and pH c is indicated; pipette pH o was 6.5 throughout. All-points histograms were derived from 6 -39 s of data using a binwidth of 9.8 or 19.5 fA and are shown on the right. B, current variance of R347E/E1104R, R347D/D924R, and R347D/D993R at the indicated pH c was collected as in Fig. 3. Each data point was derived from 2-3 patches. Error bars are smaller than the symbols. nature of the mutation at residue 347 with values of pK ranging from ϳ5-7. 3 Because the values of pK were roughly similar, the data suggest that the structure, titratability, and pK of residue 347 do not markedly influence the distribution between O L and O B states. The linear dependence on proton concentration for O B entry suggests that a single protonatable site largely determines entry into this state.
The O B and O L states may represent two protonation states of the CFTR molecule or two distinct conformational states that are influenced by protonation. For the reasons discussed above, we favor the latter possibility, that alternating residence in two discrete conformations is responsible for the two conductance states. How might this occur? We hypothesize that Arg-347 forms a salt bridge with another negatively charged residue in CFTR. Mutation of Arg-347 would leave a negatively charged residue unpaired within the membrane. Charged molecules within a low dielectric constant are unstable (37). Therefore, the protein might reorient to solvate the charge, i.e. enter the O L state. Subsequent protonation of that site would neutralize its charge and allow the protein to reorient back to its native conformation, i.e. return to the O B state.
This model suggests that the residue(s) with which Arg-347 interacts may be the protonatable site. Several observations are consistent with this hypothesis. First, we found that a positively charged lysine, which can support a salt bridge, was able to supplant arginine. However, the other residues tested were not able to substitute for arginine. Perhaps histidine was not able to replace arginine because its side chain was not long enough or it may have an anomalously acidic pK a within the low dielectric constant of the membrane.
Second, and more importantly, we found that a second-site complementary mutation at position 924 (D924R) largely eliminated the pH c -dependent flickering phenotype of the R347D mutation and restored current amplitude to near wild-type values. We also recognize the possibility that Arg-347 may interact with additional yet untested residues, e.g. Glu-873 in M7.
Third, a salt bridge between Arg-347 and Asp-924 should be disrupted by mutation of Asp-924. As predicted, we found that the D924R mutant displayed erratic flickery, pH c -independent behavior. Presumably, this mutation also generates an unstable channel. The pH c independence of D924R is also consistent with the hypothesis that Asp-924 is the site of protonation in the residue 347 mutants. Interestingly, mutation of a glutamate in the putative salt bridge in the P-loop of the IRK1 channel leads to a single-channel flickering phenotype reminiscent of residue 347 mutations (35).
Tertiary Structure of the Membrane-spanning Domains-At the simplest level, the data suggest that both MSDs functionally interact in a manner that influences permeation. The studies of R347D/D924R are consistent with a salt bridge between Arg-347 and Asp-924 and thus an interaction between M6 and M8. This further suggests that mutations in MSD2 may alter the phenotype of mutations in MSD1. Consistent with this, mutation of D993R and E1104R in MSD2 increased the relative amplitude of the O L conductance state in the context of Arg-347 mutations.
The Arg-347 residue is targeted by several CF-associated mutations, R347C, R347H, R347L, and R347P (13-15). 2 Our data suggest that CF-associated as well as other mutations at residue 347 affect CFTR similarly. They disrupt pore architecture by disrupting an interaction between Arg-347 and another residue(s), one possibly being Asp-924 in M8. These data highlight the importance of residue 347 for CFTR function and may explain in part why this residue is targeted by multiple CFassociated mutations.
FIG. 6. Single-channel currents from the D924R variant. Currents were obtained at indicated pH c with membrane voltage maintained at Ϫ120 mV. Similar results were obtained in three patches.