Conformational change of the extracellular parts of the CFTR protein during channel gating
Cystic fibrosis can be treated by potentiators, drugs that interact directly with the cystic fibrosis transmembrane conductance regulator (CFTR) Cl− channel to increase its open probability. These substances likely target key conformational changes occurring during channel opening and closing, however, the molecular bases of these conformational changes, and their susceptibility to manipulation are poorly understood. We have used patch clamp recording to identify changes in the three-dimensional organization of the extracellularly accessible parts of the CFTR protein during channel opening and closing. State-dependent formation of both disulfide bonds and Cd2+ bridges occurred for pairs of cysteine side-chains introduced into the extreme extracellular ends of transmembrane helices (TMs) 1, 6, and 12. Between each of these three TMs, we found that both disulfide bonds and metal bridges formed preferentially or exclusively in the closed state and that these inter-TM cross-links stabilized the closed state. These results indicate that the extracellular ends of these TMs are close together when the channel is closed and that they separate from each other when the channel opens. These findings identify for the first time key conformational changes in the extracellular parts of the CFTR protein that can potentially be manipulated to control channel activity.
KeywordsCystic fibrosis transmembrane conductance regulator Chloride channel Cysteine cross-linking Conformational change Potentiator Channel structure
CF transmembrane conductance regulator
Chinese hamster ovary
Cystic fibrosis (CF) is caused by mutations in the CF transmembrane conductance regulator (CFTR), an epithelial cell Cl− channel. Because CFTR dysfunction is the root cause of CF, there is great interest in the development of so-called potentiator drugs that interact directly with CFTR to increase its function [1, 2]. As an ion channel, an increase in CFTR function would be reflected in an increase in the proportion of time the channel spends in the open, conducting state relative to the closed, non-conducting state (often referred to as channel open probability). Drugs that interact with CFTR to increase the stability of the open state (impair the conformational change to the closed state) and/or decrease the stability of the closed state (facilitate the conformational change to the open state) should, therefore, exhibit potentiator function. It is important, therefore, to understand how these conformational changes between states take place, and in particular, those localized changes in conformation that might be manipulated to increase overall channel open probability [3, 4, 5].
Of those conformational changes proposed to take place during channel opening and closing , which might be manipulated by small drugs to influence overall channel activity? Previously, we identified separation and convergence of different TMs as important structural rearrangements taking place during channel gating, and showed that interfering with the relative movement of different TMs can directly alter the stability of the open and closed states . These findings offer encouragement that the TMs themselves might be possible sites at which small molecules could directly potentiate CFTR channel activity .
The most accessible part of the TMs is their extracellular ends that are exposed to the outside of the cell (Fig. 1). On the extracellular side, the TMs are connected by extracellular loops (ECLs), most of which are very short. Using functional approaches, several TMs/ECLs have been shown to be accessible from the extracellular solution, and to contribute to the outermost part of the Cl− permeation pore, including TM1/ECL1, ECL3/TM6, and ECL6/TM12 . Furthermore, important changes in conformation have been proposed for this part of CFTR during channel gating. For example, it has been reported by several groups that access from the extracellular solution to the outer part of the pore is greater when the channel is closed and paradoxically decreases when the channel opens [8, 15, 16, 17, 18]. Based on these findings, we suggested that the outer mouth of the pore might physically constrict when the channel opens [3, 17]. To investigate conformational changes at the extracellular mouth of the pore during gating, we have now sought to engineer disulfide bonds and Cd2+ bridges between cysteine side-chains introduced at the outer ends of different TMs. We find that both disulfide bonds and Cd2+ bridges can be formed between each of TMs 1, 6, and 12. However, in all cases, these inter-TM interactions appear to occur preferentially in the closed state. We propose that the outer ends of the TMs are in close proximity in closed channels and that they separate from each other when the channel opens.
Materials and methods
Experiments were carried out on Chinese hamster ovary (CHO) cells transiently transfected with human CFTR, using procedures similar to those described previously . The CFTR variant used was a so-called “cys-less” CFTR in which all cysteines have been removed by mutagenesis  and includes a mutation in the first NBD (V510A) to increase protein expression in the cell membrane . Additional mutations were introduced using the QuikChange site-directed mutagenesis system (Agilent Technologies, Santa Clara, CA, USA) and verified by DNA sequencing. Cysteine residues were substituted at sites at the extracellular ends of TM1 (R104, I106), TM6 (K329, I331, L333), and TM12 (G1127, V1129, I1131) that have previously been shown to be accessible to cysteine-reactive reagents in the extracellular solution [16, 22, 23, 24]. In some cases, cysteine substitutions were combined with the NBD2 mutation E1371Q, since mutation of this important glutamate residue results in near-permanently open channels . Previously, we have used the E1371Q mutation to study the channel state dependence of the proximity of cysteine residues introduced into different parts of CFTR [13, 18].
The proximity of pairs of cysteine side-chains introduced into different TMs was identified functionally using patch clamp recording, in two different ways. First, formation of disulfide bonds between two cysteine side-chains was inferred from changes in current amplitude resulting from treatment with the oxidizing agent copper(II)-o-phenanthroline (CuPhe) [18, 26, 27]. Since disulfide formation results in covalent attachment of two cysteines that is essentially irreversible under oxidizing conditions, this approach was used as a first screen to test which cysteine pairs showed close proximity at some stage during normal channel activity. Second, Cd2+ ions were used to form metal bridges between pairs of cysteine side-chains . Because Cd2+ bridge formation is reversible, this approach is of greater utility in studying state-dependent changes in the proximity of pairs of cysteine side-chains, from changes in apparent Cd2+-binding affinity under different gating conditions [13, 28]. For experiments using either CuPhe or Cd2+ exposure, cells were pre-treated with dithiothreitol (DTT; 5 mM) for 5 min immediately prior to the experiment, to ensure that cysteine side-chains were in a reduced state.
Experiments were carried out at room temperature, 21–24 °C. Values are presented as mean ± SEM. For graphical presentation of mean values, error bars represent SEM. Where no error bars are visible, SEM is smaller than the size of the symbol. Tests of significance were carried out using Student’s two-tailed t test, with p < 0.05 being considered statistically significant. All chemicals were from Sigma-Aldrich (Oakville, ON, Canada) except for GlyH-101 (EMD Chemicals, Gibbstown, NJ, USA).
Disulfide bond formation between the extracellular ends of different TMs
Metal bridge formation between the extracellular ends of different TMs
In fact, our results suggest that the outer ends of TMs 1, 6, and 12 are closer together when the channel is closed and that they move farther apart when the channel opens. The most direct evidence for such state-dependent changes in proximity come from the effect of the E1371Q mutation on the coordination of Cd2+ ions by two cysteine side-chains (Figs. 5, 6). In the most extreme case (L333C/G1127C), the introduction of the E1371Q mutation reduced apparent Cd2+-binding affinity by approximately 35-fold (Fig. 5d, f). If lower affinity Cd2+ binding to E1371Q-containing channels is assumed to reflect Cd2+ binding to the open state, then the affinity of Cd2+ binding must increase greatly when the channels close, suggesting much stronger Cd2+ coordination by closed channels than by open channels. The most likely explanation is that these two cysteine side-chains move close together when the channel closes (to allow tight Cd2+ coordination) and that they separate somewhat when the channel opens (resulting in a weakening of Cd2+ coordination). Qualitatively similar, although somewhat less striking, results were found with R104C/L333C (Fig. 6c, e) and I106C/G1127C (Fig. 6d, f), again consistent with these side-chains moving apart when the channel opens. In contrast, we found no evidence for Cd2+ bridge formation between I331C and G1127C (Fig. 5c, e), in spite of the fact that disulfide bonds apparently form readily between two cysteines at these positions (Fig. 2). As discussed below, we believe that any potential Cd2+ bridge formation between these two side-chains may be masked by the very high apparent Cd2+-binding affinity of the I331C side-chain in isolation (Fig. 5a, c, e). In all cases where there is evidence for Cd2+ bridge formation—between TM1 and TM6 (R104C/L333C; Fig. 6c, e), between TM1 and TM12 (I106C/G1127C; Fig. 6d, f), and between TM6 and TM12 (L333C/G1127C; Fig. 5d, f)—Cd2+ bridge formation inhibits channel function, consistent with these Cd2+ bridges acting to stabilize the closed state. Since Cd2+ interacts more strongly with the closed state, its inhibitory effects on channel function presumably reflect changes in channel gating rather than inhibition of Cl− permeation through the open channel. The consistent stabilization of the closed state observed in the present study contrasts with our earlier finding that a Cd2+ bridge between residues located in the inner vestibule of the pore, between TM1 (K95C) and TM12 (S1141C), acts to stabilize the channel open state . In the present study, at those sites that were tested, we found no evidence for Cd2+ bridges that could stabilize the open state.
We found that all constructs tested with a cysteine substituted for I331 were potently inhibited by low concentrations of Cd2+ (Fig. 5a, c, e). Indeed, I331C itself was strongly inhibited by Cd2+ (Ki ~ 5 µM), in contrast with other single-cysteine mutants with Kis > 100 µM (Figs. 5, 6) which is more typical for Cd2+ interaction with a single-cysteine side-chain . One possible explanation for this result is that Cd2+ might be strongly coordinated by I331C and another, nearby side-chain. While the use of cys-less CFTR ensures that there are no other cysteine side-chains present, it has been shown that Cd2+ ions can be coordinated by cysteine and other side-chains such as histidine, aspartate, or glutamate [29, 30, 31]. However, the possibility that such side-chains might exist in close proximity to I331C was not investigated directly.
Consistent with the suggestion that Cd2+ bridges could be formed more readily in the closed state, we also found that the functional effects of disulfide bond formation between cysteine side-chains in different TMs were significantly reduced by introduction of the E1371Q mutation (Fig. 4). We believe the most likely explanation is that disulfide bonds form less readily in the open state, which again is consistent with TMs approaching closer together in the closed state and moving apart in the open state. In fact, in two double-cysteine mutants studied—I331C/G1127C and R104C/L333C—CuPhe had only very small effects in the presence of the E1371Q mutant, even though it strongly inhibited current in 1371E-background channels that presumably were opening and closing during the experiment (Figs. 2, 3, 4). In these two cases, therefore, it is plausible that disulfide bonds cannot form in the open state but only in the closed state, consistent with a relative movement of the two cysteine side-chains in question during channel gating. Significant effects of the E1371Q mutation on sensitivity to CuPhe were also observed for R104C/L333C-bearing cysteines in TMs 1 and 6 (R104C/L333C), again consistent with the outer ends of these TMs all moving apart from each other when the channel opens. In contrast, disulfide bonds apparently formed readily between I106C (TM1) and G1127C (TM12) in the open state, suggesting that the outer ends of these TMs remain close together in open channels.
Apparent separation of the outer ends of TMs 1, 6, and 12 suggests that the outermost part of the channel pore exhibits a relatively “closed” conformation when the channel is closed and that it undergoes a relative opening transition when the channel as a whole opens (Fig. 7). Those residues at the far extracellular limits of these three TMs shown in Fig. 7a may be close enough together for disulfide bond formation when the channel is closed (although not necessarily when the channel is open), which as described above places important structural constraints on the physical dimensions of the extracellular part of the pore when the channel is closed. This close approach of the outer ends of pore-forming TMs in the closed state might be considered consistent with the existence of a functional “gate” that controls opening and closing, located relatively close to the extracellular ends of the TMs [3, 8, 9, 32]. However, it would seem to refute our previous suggestion [3, 17] that the outer mouth of the pore physically constricts during channel opening. This apparent discrepancy might suggest that reduced accessibility to residues in the outer part of the pore during channel opening could reflect not overall pore constriction, but other, more specifically localized changes in pore architecture. It has been suggested that individual TMs might undergo translational  and/or rotational [33, 34] movements during opening and closing, and such movements could increase or decrease the accessibility of individual side-chains independent of overall pore dimensions. Furthermore, residues close to the extracellular “gate” might show strongly state-dependent accessibility resulting from gate movement . In addition, movement of other parts of the outer pore (such as TM8 ) could influence access to the outer pore during channel gating.
Our results suggest that dynamic conformational changes of the outermost part of the Cl− permeation pathway take place as the channel opens and closes (Fig. 7a, d). While the extent of these conformational changes cannot be discerned from our present study, large-scale rearrangements of the extracellular face of the protein are not predicted from structures of the channel in inactive) (Fig. 7b) and near-open states (Fig. 7c). Of course, since the structure shown in Fig. 7c remains closed at its extracellular end, it is possible that a further, local conformational rearrangement is associated with the final channel opening step and that this could be associated with a relative separation of the outer ends of TMs 1, 6, and 12 as predicted by our results. Furthermore, our results suggest that, even if they are relatively minor, conformational changes at the outer mouth of the pore are important for channel opening, since manipulations that prevent these changes (such as Cd2+ bridge formation) appear sufficient to stabilize a non-conducting state of the channel. A simple cartoon model of pore opening and closing based on these results is shown in Fig. 7d. In this model, separation of the outer ends of the TMs leads to opening of an extracellular gate [3, 8, 11, 14] and widening of the outer mouth of the pore. This model is consistent with the suggestion, based on cryo-EM structures, that the outer TMs undergo a rigid-body movement during channel gating . Disulfide bond and/or metal bridge formation has previously been reported between more central and cytoplasmic parts of TMs 1 and 6 [13, 27, 35, 36], 1 and 12 [13, 36], and 6 and 12 [13, 18, 19, 36], suggesting that these core components of the pore remain in relatively close proximity throughout the pore, at least at some point in the gating cycle. Interestingly, metal bridge formation has suggested that the central and inner parts of TMs 6 and 12 may also separate when the channel opens, whereas the corresponding parts of TMs 1 and 12 have been suggested to move closer together on channel opening [13, 36], potentially pointing to region-specific differences in inter-TM movements during channel gating.
Since we were unable to identify any disulfide bonds or metal bridges between different TMs that appeared capable of stabilizing the channel in a conducting, open state, our results do not directly suggest a mechanism by which manipulation of outer pore mouth structure by potentiator drugs might increase channel activity. In theory, if channel closure is associated with these TMs moving together, then a physical manipulation that could hold them apart might be predicted to stabilize the channel open state. However, how this could be achieved by a small molecule without physical occlusion of the outer mouth of the pore is not clear. Alternatively, as the outer ends of these TMs separate during channel opening, this separation movement might bring these TMs into closer proximity with other parts of the protein, for example, other TMs or ECLs that are more peripheral from the pore mouth. If so, it might be predicted that stabilization of these other putative associations could, in theory, stabilize the channel open state and, therefore, provide the possibility for channel potentiation.
We would like to thank Christina Irving for technical assistance. This work was supported by Cystic Fibrosis Canada.
- 9.Wei S, Roessler BC, Icyuz M, Chauvet S, Tao B, Hartman JL, Kirk KL (2016) Long-range coupling between the extracellular gates and the intracellular ATP binding domains of multidrug resistance protein pumps and cystic fibrosis transmembrane conductance regulator channels. FASEB J 30:1247–1262CrossRefPubMedGoogle Scholar