If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Ion channels use charged amino-acid residues to attract oppositely charged permeant ions into the channel pore. In the cystic fibrosis transmembrane conductance regulator (CFTR) Cl− channel, a number of arginine and lysine residues have been shown to be important for Cl− permeation. Among these, two in close proximity in the pore—Lys95 and Arg134—are indispensable for anion binding and high Cl− conductance, suggesting that high positive charge density is required for pore function. Here we used mutagenesis and functional characterization to show that a nearby pore-lining negatively charged residue (Glu92) plays a functionally additive role with these two positive charges. While neutralization of this negative charge had little effect on anion binding or Cl− conductance, such neutralization was able to reverse the detrimental effects of removing the positive charge at either Lys95 or Arg134, as well as the similar effects of introducing a negative charge at a neighboring residue (Ser1141). Furthermore, neutralization of Glu92 greatly increased the susceptibility of the channel to blockage by divalent S2O32− anions, mimicking the effect of introducing additional positive charge in this region; this effect was reversed by concurrent neutralization of either Lys95 or Arg134. Across a panel of mutant channels that introduced or removed fixed charges at these four positions, we found that many pore properties are dependent on the overall charge or charge density. We propose that the CFTR pore uses a combination of positively and negatively charged residues to optimize the anion binding and Cl− conductance properties of the channel.
) have highlighted the important role of positively charged amino acid side chains in Cl− permeation through the channel pore (Fig. 1). Several positively charged side chains are clustered around the internal and external mouths of the pore (
). Additionally, neutralization of either practically abolishes the normal permeant anion conductance selectivity sequence, in which lyotropic permeant anions (such as SCN−) normally show a lower conductance than Cl− (
In spite of their important functional roles, the absolute positioning of the positive charges at these two sites, in the first (Lys95) and second (Arg134) of CFTR’s 12 transmembrane α-helices (TMs), does not appear essential. Thus, mutagenesis can be used to “transplant” either of these positive charges to other nearby sites in the inner vestibule, such as Ile344 and Val345 (TM6) or Ser1141 (TM12) (
) suggests that these other sites are permissive hosts for these functionally important charges. However, introduction of additional, ectopic positive charge (by mutagenesis of these permissive sites in the presence of endogenous charges at both Lys95 and Arg134) does not appear drastically to alter anion conductance or anion-binding properties (
)). This finding has been used to suggest that the number (or density) of fixed positive charges in the inner vestibule of the wild-type CFTR pore is well optimized to select for monovalent over divalent anions (
One puzzle is: why should two fixed positive charges in this region result in a monovalent-selective anion channel, whereas adding a third favors binding of divalent anions? The present work investigates the role of a nearby pore-lining negative charge—that of Glu92 in TM1 (Fig. 1)—in acting in an additive way with the positive charges of Lys95 and Arg134 to determine CFTR anion binding and conductance.
Expression of Glu92-mutant CFTR
Previous work has indicated difficulty in obtaining functional expression of Glu92-mutant forms of CFTR in mammalian cells (
). Western blot analysis showed that several different Glu92 mutants were able to generate mature, complex glycosylated (Band C) protein when expressed in baby hamster kidney (BHK) cells (Fig. 2), implying that at least some protein should be trafficked to the cell membrane. However, the relative abundance of this mature protein form appeared sensitive to the charge of the side chain at position 92: negative charge (wild-type, E92D) produced mostly mature protein, neutral charge (E92N, E92Q) significantly less, and positive charge (E92K, E92R) almost no detectable Band C protein (Fig. 2). One deviation from this apparent pattern was E92T, which also produced almost no Band C protein (Fig. 2). To test for possible interactions with nearby native positive charges, we also introduced E92K into backgrounds where these charges were neutralized (K95Q, R134Q) or reversed (K95E, R134E). Of these, only K95E was able to reverse the apparent effects of E92K on protein processing, leading to the appearance of significant amounts of Band C protein (Fig. 2). Mutation of Lys95 alone had little apparent effect on CFTR protein processing (see Supporting information). Recently, we showed that R134E and R134Q also had only moderate effects on CFTR protein processing under identical conditions (
). Similar effects were observed following removal of the negative charge at position 92, as illustrated by the effect of different mutations on block by S2O32− (Fig. 3). Thus, in E92N and E92Q (E1371Q background—see Experimental procedures)—but not in charge-conservative E92D/E1371Q—this divalent anion caused a strong, highly voltage-dependent block that is practically absent in control E1371Q-CFTR channels (Fig. 3). At the most negative voltages studied, S2O32− blocked E92N/E1371Q and E92Q/E1371Q with KDs of ∼1 mM (Fig. 3, C–E), more than an order of magnitude lower than for E1371Q (Fig. 3B). A similar strong blocking effect was also observed in E92T/E1371Q (Fig. 4B), although this mutant gave only very small macroscopic currents that precluded detailed analysis. E92K/E1371Q and E92R/E1371Q expression produced no noticeable current under these conditions, consistent with the Western blot results (Fig. 2).
Strong block by S2O32− suggests that the functional effect of removing a fixed negative charge in the pore may be similar to that of introducing a fixed positive charge (
). To test this, mutagenesis was used to alter the fixed charge at two sites simultaneously. The strong blocking effect of S2O32− on E92Q/E1371Q was effectively reversed by mutations that either remove a positive charge (K95Q, R134Q) or introduce an ectopic negative charge at a nearby site (S1141E) (Fig. 4). These results suggest that the balance between positive and negative charges in this part of the pore determines the binding affinity for intracellular divalent anions. In support of this idea, the double pore-mutant E92K/K95E (E1371Q background)—which effectively reverses the positions of negatively and positively charged side chains in TM1—showed similar S2O32− block as E1371Q (Fig. 4B). E92K/K95Q/E1371Q, E92K/R134E/E1371Q, and E92K/R134Q/E1371Q failed to generate any current under these conditions, again consistent with Western blot results (Fig. 2).
Positive and negative charges control unitary conductance
The positive charges at Lys95 and Arg134 play key roles in maximizing unitary conductance in CFTR (see the start of the text); mutations that remove either of these positive charges reduce conductance by ≥90% (
) (Fig. 5, A, B and E). In contrast, mutagenesis of Glu92 (E92D, E92N, E92Q, E92T) had no significant effect on conductance (Fig. 5, A and E). However, neutralization of this negative charge (in E92Q) was able to partially reverse the effects of K95Q and R134Q on conductance (Fig. 5, A, B, D and E). Whereas the conductance of K95Q was only 9.8 ± 0.5% (n = 5) of wild-type, this was increased to 87 ± 2% (n = 5) of wild-type in E92Q/K95Q (p < 0.001) (Fig. 5, A, B, D and E). The conductance of R134Q is too low to be quantified using single-channel recording (
); however, E92Q/R134Q channels had a conductance 52 ± 2% (n = 4) of wild-type (Fig. 5, D and E). Introduction of an extra negative charge (in S1141E) was highly disruptive to channel function, leading to a reduction in conductance to below the limits of resolution, similar to the effects of removing the positive charges of Lys95 or Arg134 (Fig. 5E). This dramatic reduction in conductance was partly reversed by the E92Q mutation (i.e., in E92Q/S1141E) that effectively “moves” the negative charge from TM1 (Glu92) to TM12 (Ser1141), although the conductance of E92Q/S1141E (26 ± 2% (n = 6) of wild-type) was noticeably lower than either E92Q/K95Q or E92Q/R134Q (see above) (Fig. 5, D and E). These results are consistent with the overall charge at positions 92, 95, 134, and 1141 controlling conductance (Fig. 5D). Consistent with this idea, the charge-switching E92K/K95E mutant showed retained channel function, with a conductance 60 ± 1% (n = 7) of wild-type (Fig. 5, A, C and E).
Positive and negative charges control anion binding
The positive charges of Lys95 and Arg134 are also important for anion binding in the inner vestibule, and neutralization of these charges affects the binding affinity of permeant and blocking anions (see the start of the text). Block of Cl− permeation by intracellular permeant Au(CN)2− ions was slightly weakened in E92Q-mutant channels (mean KD (at −40 mV) increased from 40.0 ± 8.5 μM (n = 8) in E1371Q to 123 ± 32 μM (n = 6) in E92Q/E1371Q) (Fig. 6, A–C), consistent with the moderate effects of mutations that increase the number of positive charges (Q98K, I344K, V345K, S1141K (
)). Addition of negative charge (in S1141E/E1371Q) drastically weakened Au(CN)2− block (mean KD (−40 mV) 5990 ± 870 μM (n = 4)) (Fig. 6, A–C), to an even greater extent than removal of endogenous positive charge (K95Q/E1371Q, R134Q/E1371Q; Fig. 6, D and E; (
)). This apparent loss of Au(CN)2− binding in S1141E/E1371Q was mostly (>90%) reversed in E92Q/S1141E/E1371Q (mean KD (−40 mV) 359 ± 105 μM (n = 6)) (Fig. 6, A–D). The E92Q mutation was also able to reverse the effects of K95Q and R134Q by >90% (Fig. 6D), indicating that a balance between fixed positive and negative charges determines Au(CN)2− binding affinity. Consistent with this, the charge-switching E92K/K95E mutant (E1371Q background) showed retained Au(CN)2− binding (mean KD (−40 mV) 235 ± 70 μM (n = 4)) (Fig. 6E).
Block of Cl− permeation by intracellular 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB) showed a similar pattern. Thus, E92Q itself (E1371Q background) did not significantly alter the blocking effects of 50 μM NPPB (Fig. 7). S1141E/E1371Q showed very weak block (Fig. 7), similar to K95Q/E1371Q and R134Q/E1371Q (Fig. 7B) (
). The E92Q mutation effectively reversed the effects of K95Q, R134Q, and S1141E, resulting in similar overall block as in E1371Q (Fig. 7B). Block of E92K/K95E/E1371Q was not significantly different from E1371Q (Fig. 7B).
Another apparent manifestation of anion binding in the inner vestibule is the relative conductance of different permeant anions (
). Thus, mutations that remove positive charge (K95Q, R134Q) have been reported to disrupt the normal anion conductance selectivity of the channel and in particular to increase the normally very low relative conductance of lyotropic SCN− ions (
) (Fig. 8). E92Q/E1371Q showed a slightly elevated relative SCN− conductance (GSCN/GCl) compared with E1371Q (Fig. 8). In S1141E/E1371Q, SCN− showed a considerably higher conductance than Cl− (Fig. 8), similar to previous findings with K95Q/E1371Q and R134Q/E1371Q (
) (Fig. 8). The elevated GSCN/GCl seen in each of S1141E, K95Q and R134Q (E1371Q background) was effectively reversed by the comutation E92Q, restoring the wild-type pattern of low relative SCN− conductance (Fig. 8). E92K/K95E/E1371Q also showed low GSCN/GCl (Fig. 8B).
Considering only the effects of single mutations, Glu92 does not appear to play a major role in Cl− permeation in CFTR. Three different mutations that neutralize this negative charge—along with the charge-conserving E92D—failed to significantly alter single-channel conductance (Fig. 5), consistent with an earlier report that E92C has a similar conductance to wild-type (
). Furthermore, the charge-neutralizing E92Q mutant had only a moderate impact on permeant anion binding (assessed by Au(CN)2− block of Cl− permeation; Fig. 6), no effect on NPPB block (Fig. 7), and led to only a moderate weakening of Cl−/SCN− conductance selectivity (Fig. 8). The only hint that this negative charge might play a role in controlling pore properties is the very strong block of charge-neutralizing E92N, E92Q, and E92T (but not charge-conserving E92D) by intracellular S2O32− ions (Fig. 3). This dramatic strengthening of S2O32− block is highly reminiscent of the effect of mutations that introduce an ectopic positive charge within the inner vestibule of the pore, I344K and S1141K (
). However, ectopic positive charge has been shown to be able to restore the very low conductance, weak anion binding, and disrupted anion conductance selectivity associated with mutations that remove important positive charges in the inner vestibule—Lys95 and Arg134 (
). However, this ability to rescue the disrupted pore function of K95Q and R134Q is also shared by E92Q, leading to relatively normal wild-type (or E1371Q)-like single-channel conductance (Fig. 5), Au(CN)2− block (Fig. 6), NPPB block (Fig. 7), and GSCN/GCl (Fig. 8) in both E92Q/K95Q and E92Q/R134Q mutated channels. Conversely, K95Q and R134Q are able to reverse the effects of E92Q on S2O32− block (Fig. 4). Thus, rather than the absolute number of positive charges lining the inner vestibule, it appears that the overall balance between fixed positive and negative charges controls anion binding in this region and thereby affects Cl− conductance. Consistent with this, the double mutant E92K/K95E—which reverses the location of positive and negative charges—has functional properties quite similar to the wild-type/E1371Q pore (Figure 4, Figure 5, Figure 6, Figure 7, Figure 8).
Anion binding in the inner vestibule is not the sole determinant of Cl− conductance (see the start of the text). The uncharged, narrow region is thought to be a major barrier to anion conductance (
), removal of a positive charge from the inner vestibule (as in K95Q and R134Q) is highly disruptive to Cl− conductance (Fig. 5) and anion-binding properties (Figure 6, Figure 7, Figure 8). These effects were mimicked—and in some cases exceeded—by introducing an additional negative charge in the inner vestibule, in S1141E mutant channels. Thus, S1141E reduced Cl− conductance below the limit of resolution (Fig. 5), drastically weakened block of Cl− permeation by both Au(CN)2− (Fig. 6) and NPPB (Fig. 7), and reversed Cl−:SCN− conductance selectivity (Fig. 8). It is perhaps not surprising that introducing a pore-lining negative charge should be disruptive to the function of an anion channel (although this is what nature has done in placing Glu92 close to an important anion-binding site in the CFTR pore). We hypothesize that the strong disruption of pore function observed in S1141E-mutant channels may reflect the important position of this amino acid close to an important anion-binding site involving Lys95 and Arg134 (Fig. 1), although we have not investigated the effects of introducing a negative charge at other sites in the inner vestibule. Importantly, all of the functional effects of S1141E can be “rescued” to some extent by the comutation E92Q (Figure 5, Figure 6, Figure 7, Figure 8), and conversely, the strong block of E92Q/E1371Q by S2O32− can be reversed by S1141E (Fig. 4). In effect, in E92Q/S1141E-mutant channels, a negative charge has been “transplanted” from TM1 to TM12 (Fig. 1), generally preserving pore conductance and anion-binding properties.
The nearby residues mutated in the present study—Glu92 (TM1), Lys95 (TM1), Arg134 (TM2), and Ser1141 (TM12) (Fig. 1)—contribute two fixed positive charges and one fixed negative charge to the inner vestibule of the pore. Mutations that decrease the “positivity” in this region—either by removing positive charges (K95Q, R134Q) or adding negative charge (S1141E)—drastically decrease conductance (by >90%) (Fig. 9A). Mutations that increase positivity—by adding positive charge (S1141K) or removing negative charge (E92N, E92Q, E92T)—have almost no effect on conductance (Fig. 9A). Mutations that retain the balance between positive and negative charge—either by conservative substitution (E92D), or by “moving” charges around different sites (E92Q/S1141E, E92K/K95E, K95S/S1141K, R134Q/S1141K), or by removing both negative and positive charges (E92Q/K95Q, E92Q/R134Q)—retain conductance (between 52 and 101% of wild-type conductance) (Fig. 9A), with the exception of E92Q/S1141E, which has a conductance 26 ± 2% of wild-type (n = 4); as discussed above, this may reflect the highly detrimental effects of having a negatively charged side chain at position 1141. We propose that it is the balance between positive and negative charges—rather than the number of charges—that controls conductance (Fig. 9A); an excess of positive charges is necessary for normal conductance, but increasing the excess from +1 to +2 causes little or no increase in conductance. Nevertheless, even for channels with one excess positive charge, there is a large variation of conductance (Fig. 9A). This may reflect that the exact position of the positive and negative charges within the inner vestibule has some influence on conductance; for instance, maximal conductance is obtained with positive charge at positions 95 and 134 and a negative charge at position 92 (wild-type, E92D), and other positions may be suboptimal “hosts” for these charges. Alternatively, unstudied factors other than charge may also be at play within the inner vestibule to modulate conductance.
Similarly to conductance, an excess of positive charge—achieved by any combination of positively and negatively charged side chains that has been investigated—is clearly required for high-affinity Au(CN)2− binding (Fig. 9B), for NPPB block (Fig. 9C), and for Cl−:SCN− conductance selectivity (Fig. 9D). Since each of these parameters likely reflects aspects of anion binding in the inner vestibule (see the start of the text), it appears that an excess of positive charge is required to form a functional anion-binding site in the inner vestibule. Increasing the excess of positive charge (as in E92Q and S1141K) has practically no impact on these anion-binding parameters (or conductance, see above) (Fig. 9), but does lead to a drastic increase in the binding of the test divalent anion S2O32− (Fig. 3; (
). The apparent lack of strong position dependence in the functional role of these charges is consistent with an electrostatic effect rather than, for example, structural changes due to charge-dependent interactions with other TMs.
Mutation of Glu92 also appears to have charge-dependent effects on the production of mature, complex-glycosylated CFTR protein (Fig. 2), suggesting that the presence of a negative charge at this position in important for CFTR protein folding and/or trafficking to the membrane. E92K is a CF-causing mutation (CFTR2 database, https://cftr2.org) that has previously been shown to have very low function and plasma membrane expression using other techniques (
). Mistrafficking of E92K was reversed by the comutation K95E, which restores the overall balance of negative and positive side chains to TM1 (Fig. 2). While we have not studied the effect of mutations on CFTR protein processing in detail, one possibility is that this charge balance is important for the correct α-helical configuration of TM1 and/or its proper interaction with other TMs.
Our investigation is not the first example of an ion channel pore incorporating the “wrong” fixed charge in an ion-binding site. For example, voltage-gated Na+ channels have both negative and positive charges surrounding their selectivity filters (the well-known “DEKA” locus); the positively charged lysine is thought to be important for high Na+:K+ selectivity (
). Among anion channels, ClCs have a conserved negatively charged glutamate side chain that can move to occupy a Cl− binding site in the pore, which is thought to be important in permeation and gating in these channels (
). However, in these examples, charged side chains make specific interactions with bound permeating ions, unlike the situation we describe in the CFTR pore, where it appears that the pore uses a combination of positive and negative charges to titrate the valence selectivity of a permeant ion-binding site. Is there any advantage to having an anion-binding site formed by a combination of two positive and one negative charges (as is apparently the case for wild-type CFTR), when a channel with one positive and no negative charges (such as E92Q/K95Q or E92Q/R134Q) is also able to sustain relatively high Cl− conductance (Fig. 5)? We can only speculate that, since wild-type has a (slightly) higher conductance than other channel variants studied here (Fig. 5), this arrangement of positive and negative charges may be well suited for optimizing conductance without allowing detrimental strong binding of divalent anions (as is observed in E92N and E92Q). The wild-type pore also appears well optimized for tight binding of permeant anions (as quantified using Au(CN)− block) (Fig. 6), which has recently been proposed as being important in maximizing pore occupancy and thereby increasing conductance (
). Thus, the somewhat counterintuitive presence of a negatively charged side chain in an anion-binding site appears, in this case at least, to confer optimal anion binding and conductance properties to the channel.
Experiments were carried out on BHK cells transiently transfected with CFTR, as described previously (
). Additional mutations were introduced using the QuikChange site-directed mutagenesis system (Agilent Technologies) and verified by DNA sequencing. Mutations were introduced at four pore-lining sites that are predicted to be in close proximity in the inner vestibule of the pore—Glu92 (TM1), Lys95 (TM1), Arg134 (TM2), and Ser1141 (TM12) (Fig. 1). Previous work has suggested that the positive charges of both Lys95 and Arg134 can be “transplanted” to Ser1141 (by comutagenesis) with little disruption of pore function, leading to the suggestion that these neighboring residues are in functionally analogous positions in the pore (
). All macroscopic current recordings (Figs. 3, 4 and Figure 6, Figure 7, Figure 8) were carried out on very high open probability channels bearing the E1371Q mutation that effectively isolates effects on Cl− permeation (
). Briefly, immunoblotting was performed on total cellular protein (5–30 μg, depending on the transfection) with monoclonal mouse anti-CFTR antibody (M3A7, Sigma-Aldrich) (1:2000 dilution) followed by secondary horseradish peroxidase-conjugated goat anti-mouse (Abcam Inc) (1:20,000 dilution). Relative expression of mature, complex glycosylated CFTR protein (“Band C”, ∼175 kDa) and immature, core glycosylated protein (“Band B”, ∼150 kDa) was assessed by densitometry analysis of scanned Western blot images using ImageJ (Version 1.48, National Institutes of Health).
Functional characterization of CFTR channels was carried out using single channel and macroscopic patch clamp recordings from inside-out membrane patches, as described in detail previously (
). For most experiments, the extracellular (pipette) solution contained (mM): 150 Na gluconate, 2 MgCl2, 10 N-tris[hydroxymethyl]methyl-2-aminoethanesulfonate (TES), pH 7.4, and the intracellular (bath) solution contained (mM): 150 NaCl, 2 MgCl2, 10 TES, pH 7.4, to generate an outwardly-directed [Cl−] gradient. To quantify the relative conductance of Cl− and SCN− ions (Fig. 8), the intracellular face of inside-out patches was perfused with the intracellular solution described above, or (mM): 154 NaSCN, 2 Mg(OH)2, 10 TES, pH 7.4. In all experiments, activity of constitutively active E1371Q-containing channels was sustained using a low concentration of intracellular ATP (50 μM) to obviate ATP block (
). Single channel activity was stimulated by exposure to protein kinase A catalytic subunit (PKA; 1–5 nM) plus 50 μM ATP in the intracellular solution. Current traces were filtered at 150 Hz (for macroscopic currents) or 50 Hz (for single channel currents) using an eight-pole Bessel filter, digitized at 1 kHz, and analyzed using pCLAMP10 software (Molecular Devices). Measurement of single channel and macroscopic current amplitudes and construction of leak-subtracted current–voltage (I-V) relationships were carried out as described previously (
). Membrane voltages were corrected for liquid junction potentials calculated using pCLAMP software.
Block of Cl− permeation by intracellular S2O32−, Au(CN)2−, and NPPB was investigated by direct application of these substances to the intracellular face of inside-out patches from high-concentration stocks made up in normal intracellular solution, as described recently (
For S2O32−, the relationship between KD and membrane potential (V) was fitted by the equation:
where zδ is the measured effective valence of the blocking ion (actual valence (z) multiplied by the fraction of the transmembrane electric field apparently experienced during the blocking reaction (δ)), and F, R, and T have their usual thermodynamic meanings.
Relative conductance of SCN− and Cl− ions (GSCN/GCl) was estimated by perfusion of the intracellular face of inside-out patches containing constitutively active E1371Q-CFTR channels with solutions containing high concentrations of these two anions (see above), as described previously (
). Relative conductance was quantified by measuring the slope conductance of the relatively linear part of the current–voltage relationship between −100 and 0 mV.
Experiments were carried out at room temperature, 21 to 24 °C. Where shown, mean values are presented ±standard deviation (SD). For graphical presentation of mean values, error bars represent SD, and where no error bars are visible, SD is smaller than the size of the symbol. In some cases (Figs. 2C, 4B, 5E, 6E, 7B and 8B), individual data points are shown together with bars showing mean ± SD. Tests of significance were carried out using one-way ANOVA with Bonferroni correction, with p < 0.05 being considered statistically significant. Unless otherwise stated, chemicals were from Sigma-Aldrich except for KAu(CN)2 (Strem Chemicals) and PKA (Promega).
Original data generated or analyzed during this study are available from the corresponding author, Dr Paul Linsdell ( [email protected] ) upon reasonable request.