A Single Conductance Pore for Chloride Ions Formed by Two Cystic Fibrosis Transmembrane Conductance Regulator Molecules*

The cystic fibrosis transmembrane conductance regulator (CFTR) is a cAMP-dependent protein kinase (PKA)- and ATP-regulated chloride channel, whose gating process involves intra- or intermolecular interactions among the cytosolic domains of the CFTR protein. Tandem linkage of two CFTR molecules produces a functional chloride channel with properties that are similar to those of the native CFTR channel, including trafficking to the plasma membrane, ATP- and PKA-dependent gating, and a unitary conductance of 8 picosiemens (pS). A heterodimer, consisting of a wild type and a mutant CFTR, also forms an 8-pS chloride channel with mixed gating properties of the wild type and mutant CFTR channels. The data suggest that two CFTR molecules interact together to form a single conductance pore for chloride ions.


The cystic fibrosis transmembrane conductance regulator (CFTR) is a cAMP-dependent protein kinase (PKA)-and ATP-regulated chloride channel, whose gating process involves intra-or intermolecular interactions among the cytosolic domains of the CFTR protein.
Tandem linkage of two CFTR molecules produces a functional chloride channel with properties that are similar to those of the native CFTR channel, including trafficking to the plasma membrane, ATP-and PKA-dependent gating, and a unitary conductance of 8 picosiemens (pS). A heterodimer, consisting of a wild type and a mutant CFTR, also forms an 8-pS chloride channel with mixed gating properties of the wild type and mutant CFTR channels. The data suggest that two CFTR molecules interact together to form a single conductance pore for chloride ions. CFTR 1 is a multi-functional protein, which provides the pore of a linear conductance chloride channel (1)(2)(3)(4)(5) and also functions to regulate other membrane proteins (6,7). Mutations in CFTR leading to defective regulation or transport of chloride ions across the apical surface of epithelial cells are the primary cause of the genetic disease of cystic fibrosis (8 -10). Comprehensive genotype-phenotype studies have indicated possible contribution of protein-protein interactions to the severity of the disease (11), but little is known on the stoichiometry of CFTR as a chloride channel.
The native CFTR chloride channel is activated by PKAphosphorylation of serine residues in its regulatory or R domain and then gated by binding and hydrolysis of ATP by the nucleotide binding folds (12). The actual pore of the chloride channel is presumably formed by portions of the two membrane spanning domains of CFTR, each consisting of six transmem-brane segments (13), with an ohmic conductance of ϳ8 pS in 200 mM KCl solution (14,15). An early study by Rich et al. (16) showed that deletion of amino acids 708 -835 from the R domain (⌬R) renders the CFTR channel PKA independent. The open probability of ⌬R-CFTR is approximately one-third that of the wt channel and does not increase upon PKA phosphorylation (17,18). Based on these different gating properties of the wt and ⌬R channels, we set out to test the intermolecular interactions of CFTR by constructing tandem cDNAs between the wt-CFTR and the ⌬R-CFTR. Our rationale is as follows. If a monomer of CFTR is sufficient to form an 8-pS chloride channel, the dimeric CFTR molecules would form either a 16-pS chloride channel or two 8-pS chloride channels that may gate independently or together. On the other hand, if two CFTR molecules are required to function as a chloride channel, we expected the tandem construct to form a single 8-pS chloride channel, provided that the linker sequence does not affect the CFTR channel function. Furthermore, we predicted that the wt-⌬R (or ⌬R-wt) channel should exhibit mixed properties of the wt and ⌬R channels.

EXPERIMENTAL PROCEDURES
Subcloning of CFTR cDNAs-The wild type and ⌬R (708 -835) CFTR cDNAs were cloned into the NheI/XhoI sites of the pCEP4 expression vector (14,17). The tandem constructs, wt-wt, wt-⌬R, ⌬R-wt, and ⌬R-⌬R, were generated in three steps. First, site-directed mutageneses were used to remove the stop codon and to introduce a BssHII restriction site at the 3Ј end of the CFTR cDNA, to create C-BssH. Second, similar approach was taken to remove the Kozak sequence and to introduce a BssHII site at the 5Ј end of the CFTR cDNA, to yield N-BssH. Third, the entire CFTR cDNA from N-BssH was released from the pCEP4 vector through digestion with BssHII and XhoI and ligated into the BssHII/XhoI sites of the C-BssH, to create the wt-wt dimer. This represents a direct head-to-tail linkage of two CFTR cDNAs. A double stranded oligonucleotide containing the recognition sequence for thrombin (underlined), Arg-Ala-Ala-Ser-Leu-Val-Pro-Arg-Gly-Ser-Gly-Gly-Gly-Gly, was ligated to the BssHII site of wt-wt, to yield the wt-e-wt construct.
Expression of CFTR in HEK 293 Cells-The human embryonic kidney (HEK 293) cells were used for expression of CFTR proteins. The different CFTR cDNAs were introduced into the HEK 293 cells using the LipofectAMINE reagent. Two days after transfection, the cells were used for Western blot assay, SPQ measurement, or isolation of membrane vesicles followed by reconstitution studies in the lipid bilayer membranes, as described previously (14,15).
SPQ Assay of Chloride Transport-The CFTR-mediated chloride transport was measured by SPQ assay with HEK 293 cells expressing the wt, ⌬R, wt-wt, ⌬R-⌬R, and wt-⌬R CFTR proteins, following the procedure described previously (14). Basically, the cells were loaded with SPQ dye (Molecular Probes) using hypotonic shock, and chloride flux across the plasma membrane was measured upon stimulation with 10 M forskolin.
Lipid Bilayer Reconstitution of CFTR Channel-The procedure for single channel measurements of CFTR using the lipid bilayer reconstitution technique has been described elsewhere (17). Briefly, microsomal membrane vesicles were isolated from HEK 293 cells transiently expressing either the wt-, ⌬R-, wt-wt, wt-⌬R, ⌬R-wt, or ⌬R-⌬R proteins and added to the cis To be able to manipulate the oligomerization state of the CFTR proteins, we engineered an enzymatic digestion site for thrombin in the linker sequence of the wt-wt dimer. This construct is named wt-e-wt (Fig. 1A, lane 10). Digestion of wt-e-wt with thrombin resulted in reduction of the apparent size from dimer to monomers (lane 9), whereas thrombin had no effect on the wt monomer (lane 8) or the wt-wt dimer (lane 11).
SPQ assays indicate that the wt-wt dimer, similar to the wt monomer, supports chloride transport in HEK 293 cells upon stimulation with forskolin (Fig. 1B). Those cells expressing the ⌬R and ⌬R-⌬R proteins exhibit basal chloride transport activities without stimulation with forskolin, which is consistent with the studies of Rich et al. (16). Interestingly, the cells expressing wt-⌬R have basal chloride transport activity in the absence of forskolin, which became significantly higher upon stimulation by forskolin (relative changes in fluorescence per minute, 0.095 Ϯ 0.010, Ϫforskolin; 0.203 Ϯ 0.007, ϩforskolin, n ϭ 65). These results indicate that CFTR dimers can traffic properly to the plasma membrane of HEK 293 cells.
To study the single channel functions of the CFTR dimers, microsomal membrane vesicles containing the wt-wt or ⌬R-⌬R proteins are incorporated into the lipid bilayer membrane. Fig.  2A shows representative current traces from the wt, wt-wt, ⌬R, and ⌬R-⌬R channels, and their corresponding current-voltage relationships are plotted in Fig. 2B. As can be seen, all four constructs give rise to chloride channels with unitary conductances of ϳ8 pS. In 9 out of 13 experiments with wt-wt, and 8 out of 12 experiments with ⌬R-⌬R, we only observed openings of a single channel (not two channels) in the bilayer membrane. Similar to the wt channel, opening of the wt-wt channel absolutely requires the presence of both ATP and PKA in the cytosolic solution; and similar to ⌬R, opening of the ⌬R-⌬R channel is independent of PKA phosphorylation. Interestingly, the activity of the wt-wt channel appears to be significantly lower than that of the wt channel (Fig. 2C). Studies from other laboratories have shown that the amino-and carboxyl-terminal tails of CFTR contribute to the overall function of the CFTR channel (19,20). We speculate that the head-to-tail connection in the dimeric construct probably constrains the movement of the amino-and carboxyl-terminal portions of CFTR, reducing activity of the wt-wt channel (see also Fig. 4).
Thus, it appears that the dimeric constructs of CFTR form functional chloride channels with conduction properties that are indistinguishable from the monomers of CFTR, i.e. all of them have single channel conductance of ϳ8 pS. The dimeric constructs could in principle have two separate pores with conductance of 8 pS for chloride ions, and because of some physical constraint due to the linker sequence, opening of one pore could prevent opening of the other pore, which would result in the overall appearance of a single CFTR channel. The other possibility is that the 8-pS channel normally recorded in single channel measurements (2,4,5,14,17) actually represents dimeric complexes of CFTR that naturally assemble in the cell surface membrane. Data from the following sets of experiments support the latter hypothesis.
The heterodimers of CFTR, wt-⌬R and ⌬R-wt, also form functional chloride channels with unitary conductance of 8 pS, which display mixed gating properties of the wt and ⌬R channels (Fig. 3). First, opening of the wt-⌬R and ⌬R-wt channels exhibit bursting kinetics, but unlike either the wt or ⌬R channels, these bursting patterns are interrupted by fast closing transitions (compare traces of Fig. 3, A and B, with Fig. 2A).  (n ϭ 7). Second, open probability of the wt-⌬R and ⌬R-wt channels display a clear PKA dependence (Fig. 3C). The channels exhibit constitutive activity in the absence of PKA, which becomes significantly higher upon PKA phosphorylation (Fig. 2C). In contrast, open probabilities of the ⌬R and ⌬R-⌬R channels are completely independent of PKA phosphorylation (Fig. 3D). Fig. 4 shows the effect of thrombin on the wt-e-wt channel. Compared with the wt-wt and wt-⌬R constructs, the wt-e-wt dimer contains 14 extra amino acids in the linker sequence corresponding to the thrombin cleavage site (underlined) (R-A-A-S-L-V-P-R-G-S-G-G-G-G). As shown in Fig. 4A, the wt-e-wt channel opens predominantly to a single 8-pS conductance state, with an average P o of 0.186 Ϯ 0.045 (n ϭ 11) at Ϫ100 mV. 3-5 min following the addition of 10 units/ml of thrombin to the cytosolic solution, the activity of the wt-e-wt channel increases approximately 4-fold, but the apparent number of channels in the bilayer membrane remains unchanged. During the course of the experiment, simultaneous opening of two channels were never observed, even though open probability of the thrombintreated wt-e-wt channel was as high as p ϭ 0.723 (Fig. 4B). Thrombin would separate the halves of the CFTR dimer and presumably remove constraints on the movement of the aminoand carboxyl-terminal tails of CFTR. If such constraints limit channel openings, this may explain why the wt-wt channel has a lower open probability than the wt channel (Fig. 2C).
Taken together, we have shown that the dimeric constructs of CFTR can be expressed in the plasma membrane of HEK 293 cells, and these CFTR dimers form functional chloride channels with unitary conductance of 8 pS, similar to the native CFTR channel (2,4). The gating properties of the wt-wt and ⌬R-⌬R channels are similar to the wt and ⌬R channels, whereas the wt-⌬R and ⌬R-wt channels exhibit mixed properties of the wt and ⌬R channels. The fact that both wt-⌬R and ⌬R-wt channels exhibit similar PKA dependence and similar gating kinetics (Fig. 3, A and B) suggests that both halves of the CFTR dimer are properly expressed and inserted in the membrane of HEK 293 cells. The tandem linkage of CFTR apparently reduces the overall activity of the chloride channel due to physical constraints introduced at the junction of the two CFTR molecules, but does not affect the conduction property of the chloride channel, since cleavage of wt-e-wt with thrombin did not change the conductance state of the channel. Our data suggest that two CFTR molecules are required for the chloride channel to open to the 8-pS conductance state.
Marshall et al. (21) used co-immunoprecipitation to search for protein-protein interactions between different mutant forms of CFTR and concluded that CFTR exists predominantly in a monomeric state. It may be that the intermolecular interactions between the CFTR molecules are weak and do not survive strong detergent solubilization (i.e. SDS) or that the CFTR dimers only represent a small percentage of the total CFTR proteins that could not be detected by the co-immunoprecipitation procedure. A recent study by Eskandari et al. (22) established structural evidence for a dimeric complex with the CFTR proteins. These investigators used freeze fracture electron microscopy to investigate the oligomeric assembly of membrane proteins expressed in Xenopus oocytes, and they concluded that the intramembrane structure of CFTR was consistent with a dimeric assembly of 12-transmembrane helix of the CFTR monomers.
Besides being of fundamental importance in understanding the mechanism by which the CFTR channel works, the concept that CFTR functions as a dimer may have implication for persons who are compound heterozygotes for CFTR mutations. Some mutant pairs may be able to complement each other thereby increasing the overall CFTR function, but other mutant pairs may not. This may explain some of the phenotypic variations in patients with CFTR mutations, especially those with some residual function. It will be important to know at what stage in the biosynthetic pathway of CFTR the dimers are formed, i.e. before leaving the endoplasmic reticulum or after trafficking to the apical membrane. More specifically, which portions of the CFTR molecule are involved in the contact interaction or constitute the binding site(s) for accessory proteins that could interact with CFTR?