Molecular Proximity of Cystic Fibrosis Transmembrane Conductance Regulator and Epithelial Sodium Channel Assessed by Fluorescence Resonance Energy Transfer*

We present the evidence for a direct physical association of cystic fibrosis transmembrane conductance regulator (CFTR) and epithelial sodium channel (ENaC), two major ion channels implicated in the pathophysiology of cystic fibrosis, a devastating inherited disease. We employed fluorescence resonance energy transfer, a distance-dependent imaging technique with capability to detect molecular complexes with near angstrom resolution, to estimate the proximity of CFTR and ENaC, an essential variable for possible physical interaction to occur. Fluorescence resonance energy transfer studies were complemented with a classic biochemical approach: coimmunoprecipitation. Our results place CFTR and ENaC within reach of each other, suggestive of a direct interaction between these two proteins.

CF 2 is the most common lethal autosomal recessive disorder in Caucasians. It is caused by mutations in the CFTR protein (1-3), a Cl Ϫ channel that is widely expressed in epithelia (4). In the airways of patients with CF, mutations in the cftr gene lead to the production of poorly hydrated airway surface liquid interface, thick and sticky mucus, which in turn reduces mucociliary clearance and promotes bacterial colonization and sec-ondary infections (5). The gastrointestinal abnormalities present earlier with failure to thrive; reduced fluid secretion in the pancreatic ducts compromises digestive enzyme release into the duodenum, resulting in poor nutrition (5,6). However, the airway pathology is the major cause of CF mortality (5)(6)(7).
Although CFTR is primarily an anion channel (4, 8 -10), these major symptoms are not solely due to deficiencies in anion secretion. Absorption of the obligate cation Na ϩ across the apical membrane of the respective epithelium also contributes to the pathology (5,7,11). Some of the earliest electrophysiological studies of CF patient airways demonstrated that an amiloride-sensitive Na ϩ transport pathway contributed to the raised transepithelial potential difference across CF nasal epithelia (12). This sensitivity to the diuretic amiloride suggested an involvement of the ENaC family of proteins. Recently, further experimental evidence supporting a role for ENaC in airway hydration has been obtained in mice overexpressing ␤-ENaC (13). These animals develop a "CF-like" airway disease, which was attributed to over-absorption of Na ϩ . CFTR knockout mice recapitulated the gastrointestinal abnormalities but did not show this airway pathology (13). Thus, CFTR and ENaC both contribute to the pathology of CF, although only CFTR is affected by the genetic mutations.
These facts suggest that these two proteins are functionally interrelated. Previous studies have suggested that CFTR downregulates ENaC function (14 -18), thus restricting absorption and shifting fluid balance across the epithelium in favor of net secretion (7). In CF, however, the reverse process happens: ENaC is freed from the constraint exerted by CFTR, which is either non-functional or absent in CF, and now causes increased absorption of Na ϩ across the epithelium, with Cl Ϫ following via paracellular pathways. The result is dehydration of the airway surface fluid seen in CF patients (5).
The presence of a functional link between CFTR and ENaC has been reported by many laboratories in a number of experimental situations (14 -16, 19 -23) and is seen clinically with the nasal potential difference used to diagnose CF (12). However, the existence of this interaction in vitro has been questioned (24 -26), and the mechanism underlying this interaction is poorly defined.
In the present study, we have addressed the issue of whether or not there is direct interaction between ENaC and CFTR using an independent biophysical approach; i.e. FRET imaging. FRET exploits the exquisite sensitivity of fluorescence measurements to detect the presence of molecular complexes with near angstrom resolution. Using a standard FRET pair, with enhanced cyan fluorescent protein (ECFP) as a donor and enhanced yellow fluorescent protein (EYFP) as an acceptor, we have found that co-transfection of ECFP-CFTR with EYFPtagged ENaC subunits resulted in significant FRET efficiencies (E). In contrast, co-transfection of ENaC subunits with ECFP-CLCN1, an ion channel not found to functionally couple with ENaC, resulted in little to no FRET (E Ͻ 0.5%), an outcome comparable to negative controls (ECFP and EYFP co-transfected cells). Co-immunoprecipitation (Co-IP) experiments showed that ␣␤␥-ENaC consistently co-immunoprecipitated with CFTR. These results place CFTR and ENaC proteins in very close proximity to each other, consistent with a direct interaction of these two proteins. This renders plausible a possibility that these two proteins interact directly in vivo to contribute to CF airway pathology.
Transient Transfection-Transient transfections of the cells with construct(s) of interest were performed using Lipofectamine 2000 reagent (Invitrogen) following the manufacturer's recommended protocol (27). Briefly, 1 l/200 ng of DNA of Lipofectamine 2000 reagent was diluted with 100 l of Opti-MEM I (Invitrogen) and incubated at room temperature for 5 min. Separately, 200 ng of each DNA construct was diluted with 100 l of Opti-MEM I (Invitrogen). After 5-min incubation, the diluted DNA constructs were combined with Lipofectamine 2000 and allowed to complex for 20 min. After incubation the transfection solution was added to the cells in Opti-MEM I. The cells were incubated for 5 h at 37°C in a CO 2 incubator followed by a change of the transfection solution to regular growth medium without antibiotics. Forty-eight hours after transfection, the cells were gently rinsed with 1ϫ phosphate-buffered saline (PBS, Invitrogen) and fixed for 15 min at room temperature with 4% paraformaldehyde (prepared from 20% EM Grade solution, Electron Microscopy Services, Hatfield, PA). Cells then were rinsed three times with 1ϫ PBS and mounted on glass microscope slides (Fisher Scientific) using 0.2% n-propyl gallate (Sigma) in 9:1 glycerol/PBS (v/v). The same protocol was used to transfect the cells for co-immuno-precipitation experiments; the DNA/Lipofectamine 2000 ratio was 1 g/2.5 l.
Generation of the Fluorophore-tagged CFTR and ENaC cDNAs-We have restricted our study to the N-terminal tail of CFTR and C termini of ␣␤␥-ENaC. Previous studies have shown that an interference with the C-terminal tail of CFTR might lead to its improper localization due to disruption of a PDZ binding motif (28). Similarly, manipulation of N termini of ␣␤␥-ENaC affects its gating properties (18,29). The CFTR N-terminally tagged with enhanced green fluorescent protein (EGFP-CFTR) was a kind gift of Dr. B. Stanton, Dartmouth Medical School (30). The EGFP was exchanged with enhanced cyan fluorescent protein CFTR (ECFP-CFTR) following NheI/ XhoI (New England Biolabs, Ipswich, MA or Promega, Madison, WI) digestion (see Fig. 1C). Human Chloride Channel 1 (CLCN1) tagged N-terminally with ECFP (ECFP-CLCN1) was a kind gift of Dr. Christoph Fahlke (Dept. of Neurophysiology, Medizinische Hochschule Hannover, Hannover, Germany) (31). cDNA constructs of human ␣-ENaC, ␤-ENaC, and ␥-ENaC subunits with C-terminal EYFP were generated by PCR amplification of the genes of interest and cloning into the vector pEYFP-N1 (BD Biosciences Clontech, Palo Alto, CA) using XhoI and BamHI restriction sites (27). Positive control (ECFP-EYFP) was generated by fusion of ECFP to EYFP-N1 vector using XhoI and BamHI restriction sites through a 7amino acid linker (SGLRSRA). All constructs were verified by direct sequencing prior to use.
ENaC Antibodies-Rabbit polyclonal antibodies (Abs) were generated against synthetic peptides in collaboration with Dr. Mark Knepper (National Institutes of Health). Antigenic peptides with the sequences, ␣-hENaC, LMKGNKREEQGLGPEP-AAPQQPTC; ␤-hENaC, CNYDSLRLQPLDVIESDSEGDAI; and ␥-hENaC, CNTLRLDSAFSSQLTDTQLTNEF, were synthesized by Lofstrand Laboratories (Gaithersburg, MD). The peptides correspond to human sequences, chosen to maximize immunogenicity, to minimize likely cross reactions with other eukaryotic proteins, and to avoid common post-translational modifications as described (32). All peptides were high-performance liquid chromatography-purified. The peptides were conjugated to a carrier protein keyhole limpet hemocyanin and injected into rabbits using a standard protocol (Antibodies Inc., Davis, CA). Antisera from test bleeds, production bleeds, and final exsanguination were affinity-purified and tested on immunoblots from human kidney cortex and medulla (not shown). These immunoblots showed single bands in the range of 80 -90 kDa consistent with previous anti-ENaC Abs directed to rodent proteins (33).
Western Blot Analysis of CFTR and ENaC (34,35). CFTR-HEK293T cells were transfected with 2 g of ECFP-CFTR construct. 48 h post-transfection, cells were lysed with radioimmune precipitation assay buffer (Pierce) complemented with Complete protease inhibitor mixture (Roche Biochemical, Mannheim, Germany) at 4°C. After centrifugation (15,800 ϫ g for 10 min at 4°C) non-soluble material was discarded. 100 g of total cell lysate was analyzed on a 6% protein gel (Invitrogen). The protein gel was blotted onto a polyvinylidene difluoride membrane (Bio-Rad) and probed with 1:10,000 diluted polyclonal Ab against the second nucleotide binding domain of CFTR (University of Alabama at Birmingham Cystic Fibrosis Research Center) and probed with anti-rabbit horseradish peroxidase (Dako, Denmark A/S) using chemiluminescence.
CFTR and ENaC Co-immunoprecipitation-CFTR and ENaC co-immunoprecipitation were performed as described previously (35). Briefly, HEK293T cells were transfected with CFTR and appropriate combinations of ENaC subunits. Fortyeight hours post-transfection, cells were lysed with 0.2% Triton X-100 in PBS complemented with Complete protease inhibitor mixture (Roche Applied Science) at 4°C. After centrifugation (15,800 ϫ g for 10 min at 4°C) non-soluble material was discarded, and the supernatant was incubated with 1 g of C-terminal CFTR monoclonal Ab clone 24-1 (R&D Systems, Minneapolis, MN) cross-linked to A/G-agarose beads (Santa Cruz Biotechnology, Santa Cruz, CA) for 2 h. After incubating the lysates with cross-linked Ab, the beads were pelleted and washed three to five times in PBS containing 1% Triton X-100. Precipitated CFTR and bound proteins were analyzed on SDS-PAGE, transferred to a polyvinylidene difluoride membrane and the samples were processed for Western blotting using Abs raised against ENaC and CFTR.
FRET Imaging-The acceptor photobleaching approach was performed as previously described (27,36). Briefly, images were acquired using the Leica Confocal Scanning System (Exton, PA) TCS SP2 outfitted with a Leica DMRXE upright microscope, an argon laser, and 100 ϫ 1.4 numerical aperture Plan Apochromatic oil immersion objective. ECFP and EYFP fluorescence were excited with 458 and 514 nm laser light, respectively. Dual excitation was achieved by a dual dichroic (DD458/514) mirror in the excitation path. The emissions for ECFP and EYFP were collected using 465-505 and 525-600 nm bandpass windows, respectively. Fixed samples were imaged within 3 days. We noticed an increase in background fluorescence if fixed samples were processed after a long period (Ͼ7-10 days) of storage. The regions of interest in cells expressing both fluorophores were photobleached using 514 nm laser light to 30% of original intensity. ECFP and EYFP images were taken both before and after acceptor photobleaching. FRET efficiency (E) was calculated using Leica software: E ϭ D post Ϫ D pre /D post , where D pre and D post are ECFP emission before and after regional photobleaching.
FLIM-The images were taken with the LIFA system (Lambert Instruments, The Netherlands), a frequency domain FLIM system, attached to a Nikon TE2000 wide field fluorescence microscope with 100 ϫ 1.49 numerical aperture objective. A 443 nm modulated LED was used as a light source, and 12 phase-step images were taken at the frequency of 40 MHz with the modulated ICCD (intensified charge-coupled device) camera. By using the filter cyan GFP only the donor emission was measured. In the resulting images, the intensity is shown in grayscale analog-to-digital units, and the lifetime from phase is in pseudocolors (nanoseconds).

RESULTS
Expression, Fluorescence Imaging, and Functional Analysis of the Fluorophore-tagged Constructs-Western blot detection of tagged CFTR and ENaC constructs was performed to test whether the fusion of the 27-kDa ECFP or EYFP tags affected protein expression. The blot in Fig. 1C demonstrates the bands  DECEMBER 14, 2007 • VOLUME 282 • NUMBER 50 corresponding to the expected molecular mass of untagged CFTR (lane 1) and the N-terminally tagged construct with a 27-kDa fluorescence tag, ECFP (lane 2, ECFP-CFTR). Similarly, the bands corresponding to the molecular mass of ENaCs with 27-kDa fluorescence tags were observed in cells expressing tagged ENaC subunits (Fig. 1D). We next tested whether fusion of the ECFP or EYFP tags altered the proper plasma membrane targeting of CFTR and ENaC constructs. We found that the expression of CFTR construct predominantly delineated the plasma membrane. This staining suggests that the addition of fluorescence tags does not interfere with the trafficking of CFTR (data not shown). The functional expression of fluorophore-tagged CFTR constructs revealed no alteration in its function as a Cl Ϫ channel (data not shown) in agreement with previously published findings (28). A predominant endoplasmic reticulum staining was observed for ENaC by immunofluorescence (data not shown) and is characteristic of the poor trafficking of these proteins. However, amiloride-sensitive currents characteristic of functional ENaC channels were recorded from cells transiently transfected with tagged ENaC subunits (data not shown and Refs. 27,37,38). These data suggest that a population of ENaC proteins reaches the plasma membrane of transfected cells. Cotransfection of CFTR and ENaC constructs showed colocalization of the two proteins suggesting that they can traffic to the plasma membrane (supplemental Fig. S1).

CFTR-ENaC FRET
ENaC Subunits Interact with CFTR-In a recent report (18) we have presented a systematic study of the kinetic properties of ENaCs formed by different combinations of subunits. We also correlated our kinetic data with the effects of CFTR on ENaC. We found that CFTR inhibits ENaCs at a single channel level, suggestive of a direct physical interaction. To further validate these findings, we have employed a biochemical approach, Co-IP, to assess a physical interaction between CFTR and ENaC.
When CFTR and ␣-, ␤-, and ␥-ENaC were co-expressed, each ENaC subunit was co-immunoprecipitated with CFTR using a C-terminal CFTR Ab, suggesting that CFTR and ENaC channels are in a complex (Fig. 2). No co-IP signal was detected when ENaC subunits or CFTR were expressed alone indicating that the co-IP signals are specific and demonstrate that ENaC subunits do not bind nonspecifically to the beads. The expression of the ENaC subunits was confirmed by Western blot and could not be detected in untransfected cells. The bottom panels of Fig. 2 show the same blot reprobed with anti-CFTR Ab to confirm the presence of CFTR in the IP. Taken together, these results indicate that CFTR and ENaC do interact in HEK293T cells.
As an additional control, we assayed for the possibility that the Co-IP signals observed resulted from random aggregation of CFTR and ENaC subunits due to overexpression. Thus, one culture of cells was transfected with CFTR while a second culture was transfected with ␤-ENaC. After mixing these two lysates, co-IP experiments were performed as described above. No ␤-ENaC signal was detected suggesting that the interaction between CFTR and ␤-ENaC do not occur post-lysis (data not shown). This finding also reduces the possibility that this interaction is due to overexpression and random interactions of the proteins.
Next, we substituted the CFTR with ECFP-CLCN1, a Cl Ϫ channel that is not thought to interact with ENaC. As can be seen from Fig. 2D, we were unable to Co-IP ␤-ENaC with ECFP-CLCN1. In this approach we used anti-GFP monoclonal Ab JL-8 (BD Living Colors) that recognizes CFP protein to IP CLCN1, and then probed with the ␤-ENaC Ab. The expression of CLCN1 in the cells was confirmed using GFP Ab (Fig. 2D). The cells expressing ␣␤␥-ENaC and/or ECFP-CLCN1 were  1-3) shows that ␥-ENaC is detected only in cells transfected with ␥-ENaC construct. D, ␤-ENaC does not Co-IP with the Cl Ϫ channel ECFP-CLCN1 when co-expressed in cells. Anti-GFP monoclonal Ab JL-8 (BD Living Colors) was used to IP CLCN1, and the blot was probed with the ␤-ENaC Ab. The expression of CLCN1 in the cells was confirmed using GFP Ab (lanes 2 and 4). An Ab against ␤-ENaC failed to detect a co-IP signal in IP of cells co-expressing ␤-ENaC and CLCN1 (lane 4, bottom panel). The expression of ␤-ENaC in the lysates was confirmed (lanes 1 and 2). These experiments were repeated at least three times, and similar results were obtained.
immunoprecipitated with an Ab against GFP. Although this Ab detected ECFP-CLCN1 in both crude lysates and IP of ECFP-CLCN1-expressing cells, an Ab against ␤-ENaC failed to detect a band in IP of co-expressing cells (top panel and right hand lanes of the bottom panel). However, this Ab did detect ␤-ENaC in the crude lysates from cells expressing either ␣␤␥-ENaC alone or those co-expressing ECFP-CLCN1. These data suggest that ␤-ENaC does not interact with the chloride channel CLCN1. All together, these data suggest that CFTR and ENaC can interact in HEK293T cells and that this interaction is specific.
FRET Measurements by Acceptor Photobleaching-Because ENaC subunits Co-IP with CFTR (Fig. 2), their possible interaction was probed by FRET imaging. FRET occurs when donor and acceptor fluorescent proteins are sufficiently close (Յ10 nm) that radiationless (i.e. not mediated by photons) transfer of energy from donor to acceptor (39 -41) can be observed after excitation of the donor (Fig. 1A). Both the rate and the efficiency of energy transfer are directly related to the distance between the donor and acceptor. If the fluorophores are more than 10 nm apart, no FRET will be detected (Fig. 1B).
In acceptor photobleaching FRET, the acceptor is selectively photobleached (36). As a result, if the acceptor is close to the donor and FRET had been occurring, an increase in the fluorescence of the donor is observed because the photobleached acceptor is unable to accept energy. Apparent FRET efficiency (E) is determined from the ratios of ECFP fluorescence emission before and after EYFP photobleaching.
Control FRET Measurements-As a positive control we used ECFP-EYFP fusion protein separated by a 7-amino acid linker (Fig. 1A). The regions of interest in cells expressing both fluorophores were photobleached using the 514 nm laser to 30% of original intensity. ECFP and EYFP images were taken both before and after acceptor photobleaching. The cells transfected with this construct showed an obvious increase in CFP fluorescence after photobleaching (Fig. 3A), and E averaged ϳ30% (supplemental Fig. S2). In contrast, our negative control, cells transfected with two unlinked ECFP and EYFP constructs, did not show any change in CFP fluorescence after regional bleaching (Fig. 3B), with an E of Ͻϳ1% (supplemental Fig. S2). Also, with acceptor bleaching, another internal negative control is built into each sample, the unbleached areas of the cell. After each bleaching, we compared the bleached area with unbleached regions of the cell to avoid introduction of noise as a result of imaging per se and/or drift of the stage. These two controls confirmed our ability to measure FRET by acceptor photobleaching and always preceded our FRET measurements.
Detection of FRET between CFTR and ENaC-Cells were cotransfected with vectors encoding ECFP-CFTR and ␣␤␥-ENaC. One of the three ENaC subunits was tagged with EYFP. Also, because ␣-ENaC is known to form a functional homomeric channel, FRET between CFTR and ␣-ENaC alone was measured. Fig. 3C shows a set of images acquired before and after regional photobleaching when ECFP-CFTR was cotransfected with ␣-EYFP. In Fig. 3C the bleached area is shown as a green box. To calculate E, regions of interest were drawn inside the bleached area and were restricted to the delineated plasma membrane. These regions of interest allow us to exclude FRET occurring in the endoplasmic reticulum or other internal compartments. We found an appreciable increase in CFP fluorescence after selective photobleaching with all combinations of ENaC and CFTR, indicative of FRET occurrence. The E averaged ϳ7%, but values as high as ϳ17% were observed. Thus these values are significant when compared with the negative controls (Fig. 3B) and are comparable to E reported in other FRET studies (42,43). No increase in CFP fluorescence was observed in non-bleached areas. Supplemental Fig. S2 also demonstrates E values for cells co-transfected with constructs encoding ECFP-CFTR and other ENaC combinations. Again, E was significant relative to the negative control. These measurements indicate that FRET occurs between CFTR and ENaC.
Absence of FRET between CLCN1 and ENaC-Next, we addressed the issue of specificity, i.e. whether the observed FRET was specific between CFTR and ENaC or results of random association between two membrane proteins. To address this issue we replaced CFTR with another Cl Ϫ channel, CLCN1 tagged with ECFP (31). To our knowledge, there is no report of association between CLCN1 and ENaC using either electrophysiology or biochemistry. Fig. 3D demonstrates representa-

. FRET acceptor photobleaching of the cells transfected with ECFP-EYFP fusion construct (A), ECFP and EYFP constructs co-transfected separately (B), ECFP-CFTR and ␣-EYFP-ENaC (C), and ECFP-CLCN1 and ␣␤-EYFP-␥-ENaC (D).
The area of photobleaching is highlighted by green boxes. ECFP and EYFP images were taken both before and after acceptor photobleaching. FRET efficiency is displayed as a pseudocolor representation. Scale, 5 m. tive images taken from cells expressing ECFP-CLCN1 and ␣␤-EYFP-␥-ENaC. No FRET was observed between CLCN1 and ENaC. A summary bar graph of E values for other combinations of tagged ENaC subunits and CLCN1 is given in (supplemental Fig. S3). These results, namely the presence of FRET between CFTR and ENaC and its absence between CLCN1 and ENaC, underscore the specificity of the energy transfer between CFTR and ENaC and support the hypothesis of specific CFTR-ENaC interaction.
FRET Measurements by FLIM-As an alternative and independent measure of CFTR-ENaC FRET we used FLIM-FRET. FLIM is fluorescence lifetime imaging microscopy and detects the average time that a group of fluorophores spend in the excited state (44). FRET is marked by a decrease in fluorescence lifetime of the donor fluorophore as its energy is transferred to the acceptor fluorophore. FRET efficiencies (E) are directly calculated from the lifetime donor ratios with and without acceptor. Fig. 4A demonstrates the intensity and corresponding lifetime from phase images of the cells expressing CFTR tagged with ECFP. The FLIM image shows that, in the absence of the EYFP-ENaC, the ECFP tagged to CFTR had an average lifetime (ECFP) of ϳ1.9 ns, within a range of previously reported values (45). When ECFP-CFTR and any ENaC subunit (Fig. 4, B-D) tagged with EYFP were co-expressed, a decrease in fluorescence lifetime to FRET of ϳ1.7 ns was observed, which reflects the predicted energy transfer from ECFP to EYFP. The values are averaged over 6 -10 cells, obtained from 2 or 3 samples. The FLIM-FRET efficiency is calculated as 1 Ϫ ( FRET / ECFP-N-CFTR ) (supplemental Fig. S4).

DISCUSSION
Even though the role of amiloride-sensitive Na ϩ pathways in airway CF were delineated at the beginning of 1980s (12,46), coupling between CFTR and ENaC only became evident after molecular cloning of ENaC (47)(48)(49) following the identification of CFTR (1)(2)(3). Soon after the molecular cloning of ␣␤␥-ENaC subunits, the remarkable ability of CFTR to influence other transporters became evident. This regulatory ability of CFTR is best exemplified by influence of CFTR on ENaC activity (7). Although the evidence overwhelmingly points toward a possible interaction, the nature of this interaction, as well as its existence, is still debated, partly due to contradictory electrophysiological findings and limited biochemical data.
In this study, multiple independent approaches were used to examine the potential interaction of CFTR with ENaC. Using CFTR and ENaC molecules tagged with fluorescent variants of GFP, we were able to observe a clear delineation of the plasma membrane and strong overlap of the two tagged proteins when they were coexpressed in HEK293T cells (supplemental Fig. S1). We next used Co-IP experiments to determine if the two proteins could interact physically (Fig. 2). Lastly, we used two independent applications of the FRET technique (Figs. 3 and 4) to determine if these two important transport molecules were in close proximity in transfected cells. Using these independent techniques, our findings are consistent with the hypothesis that CFTR and ENaC are in close proximity and likely interact directly.
There are many artifacts of technique that have clouded the literature regarding potential interactions between CFTR and ENaC (24 -26). One caveat in the present study concerns the use of an overexpression system. This is an unavoidable complication of the fluorescently tagged FRET-based approach to studying protein-protein interaction. In such systems, there is a possibility of observing nonspecific interactions due to a simple mass balance effect. If occurring, this would be seen as a positive FRET signal due to increased close packing of tagged molecules, where due to high levels expression, there is a FRET interaction in constrained volumes such as in the endoplasmic reticulum or at the plasma membrane. The use of a fluorescently tagged membrane protein such as the sarcolemmal Cl Ϫ channel, CLCN1, can assess the specificity of a positive FRET signal. In our studies there was no evidence to support an interaction between CLCN1 and ENaC subunits (Fig. 3D). The absence of FRET between CLCN1 and our tagged proteins further suggests that the interaction we observed between CFTR and the ENaC subunits is specific.
An overexpression artifact can also be a confounding factor in Co-IP experiments: random interactions may occur due solely to high concentrations of two proteins. This is unavoidable in heterologous expression systems, which are required because endogenous CFTR or ENaC proteins are difficult to detect. To address overexpression concerns in this system, lysates from cells expressing either CFTR or ␤-ENaC were mixed in vitro. Under these conditions, the two proteins could not be co-immunoprecipitated, suggesting that the Co-IP of CFTR and ENaC was not a random artifact of the overexpression system. It may also suggest that an interaction requires the two proteins to be simultaneously processed and trafficked to the membrane. Although there are known limitations regarding the sensitivity and specificity of the anti-human CFTR (clone 24-1) Ab, the inclusion of appropriate positive and negative controls and confirmation of our findings by Western blot analysis support our conclusions.
It can also be appreciated that the calculated FRET efficiency between CFTR and ENaC proteins is significantly lower than the positive control in which the FRET pair is physically linked together. Although the values are lower, they are significantly higher than those of negative controls. These efficiencies are also comparable to that found by other investigators (42,43) and may speak to the nature of the interaction. As our experiments focused upon interactions between the N terminus of CFTR and the C-terminal tails of the ENaC subunits, their separation may be greater than that of the 37-amino acid linker found in the positive control but less than that of the 10 nm threshold for FRET. As ENaC trafficking and function is disrupted by manipulations of the N terminus, it is difficult to use FRET to probe the interaction of the N termini of ENaC proteins with CFTR; however, the interaction may be mediated by the N termini of ENaC protein rather than the C termini. It is also very possible that the proteins do not interact directly, but rather through an intermediary, forcing the distance between them to be larger and decreasing the efficiency of FRET. This is rather unlikely, but possible.
Interestingly, the identity of the tagged ENaC subunit had a minimal effect upon the measured FRET efficiency. Because ENaC proteins must form homo-or heteromers to form functional channels, this suggests that the CFTR interacts with a sole ENaC, rather than multiple channels. For example, ␣-ENaC is known to form a functional homomeric channel; however, there is not an increase in the FRET efficiency between ␣-ENaC and CFTR as would be expected if multiple ␣-ENaC subunits interacted with a single CFTR molecule. Although these FRET data cannot easily address the stoichiometry of the interaction, these data suggest a 1-CFTR:1-ENaC arrangement rather than 1-CFTR:Many-ENaC.
The transport of multiple ions and solutes has been proposed to be under the overall control of CFTR, but the potential reciprocal relationship between CFTR and ENaC has been the most widely investigated. However, the present study is the first to present evidence supporting a direct physical interaction between these two important proteins. These studies, which were carried out under conditions of heterologous expression, create the foundation for experiments in cells endogenously expressing both ENaC and CFTR proteins as well as experiments assaying for FRET between CFTR mutants and ENaC proteins. Furthermore, this system allows for mapping of the CFTR and ENaC subunits to define interaction domains or residues. The functional significance of these findings and their implications for CF treatment require further study. However, a direct functional interaction may suggest the most efficacious therapies would target ENaC as well or place functioning CFTR at the membrane, rather than solely affecting Cl Ϫ transport.