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

Abstract

The ABCC transporter subfamily includes pumps, the long and short multidrug resistance proteins (MRPs), and an ATP-gated anion channel, the cystic fibrosis transmembrane conductance regulator (CFTR). We show that despite their thermodynamic differences, these ABCC transporter subtypes use broadly similar mechanisms to couple their extracellular gates to the ATP occupancies of their cytosolic nucleotide binding domains. A conserved extracellular phenylalanine at this gate was a prime location for producing gain of function (GOF) mutants of a long MRP in yeast (Ycf1p cadmium transporter), a short yeast MRP (Yor1p oligomycin exporter), and human CFTR channels. Extracellular gate mutations rescued ATP binding mutants of the yeast MRPs and CFTR by increasing ATP sensitivity. Control ATPase-defective MRP mutants could not be rescued by this mechanism. A CFTR double mutant with an extracellular gate mutation plus a cytosolic GOF mutation was highly active (single-channel open probability >0.3) in the absence of ATP and protein kinase A, each normally required for CFTR activity. We conclude that all 3 ABCC transporter subtypes use similar mechanisms to couple their extracellular gates to ATP occupancy, and highly active CFTR channels that bypass defects in ATP binding or phosphorylation can be produced.—Wei, S., Roessler, B. C., Icyuz, M., Chauvet, S., Tao, B., Hartman IV, J. L., Kirk, K. L. 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.

The C subfamily of ATP binding cassette transporters (ABCC) includes an ATP-gated ion channel (CFTR) and active transport ATPases that confer multidrug resistance [the multidrug resistance protein (MRPs)] (1). The MRPs and CFTR are biomedically important ABC transporters whose dysfunction causes several disorders (e.g., cystic fibrosis in the case of CFTR) (2). Although MRP pumps and the CFTR channel are thermodynamically distinct transporters, each utilizes ATP binding to promote conformational changes in its translocation pathway (3, 4). For each transporter 2 ATP molecules bind at the interface of a dimer of cytosolic nucleotide binding domains (NBDs), which are linked to the transmembrane spanning domains (TMs) via their cytosolic loops (3–7). The MRPs and CFTR also possess ATPase activity that is used to energize substrate transport in the case of an MRP (3, 7) or to control ligand occupancy in the case of the ATP-gated CFTR channel (6, 8). The MRPs are further classified into short and long forms; the latter have 5 extra TMs at the N terminus in addition to the 12 TMs that are characteristic of the short MRPs and CFTR (3). CFTR has the added feature of possessing a long regulatory domain (R domain) that links NBD1 to TM7 and contains many sites for phosphorylation by cyclic nucleotide dependent kinases (e.g., PKA) (9). Phosphorylation of the cytosolic R domain normally is required for CFTR opening (9). How the R domain regulates CFTR gating is unclear and may involve multiple mechanisms (10) such as controlling NBD dimerization (10–12) and the flexibilities of the TMs/cytosolic loops (13, 14).

The evolutionary relationship between MRP pumps and the CFTR channel implies that they may use similar mechanisms for coupling ATP binding at the NBD dimer interface to those conformational changes in the TMs that underlie active substrate transport (MRPs) or channel gating (CFTR). Recently we discovered gain of function (GOF) mutations at conserved locations near the cytosolic bases of TMs 6 and 9 in CFTR and in the short MRPs (14, 15). These GOF mutations promoted ATP-free CFTR channel activity, increased the ATP sensitivity of CFTR gating, and reversed the low ATP sensitivity of a CFTR construct with an NBD2 mutation that disrupts ATP binding. Interestingly, homologous TM substitutions also rescued defective drug export by ATP binding mutants of a short MRP in yeast, the Yor1p oligomycin exporter (MRP4 ortholog).

The latter finding supports the idea that CFTR may share with certain MRP pumps a similar mechanism to link the conformation of its translocation pathway to the ATP occupancy of its NBDs. But this finding also raises a number of questions regarding how far one can push the analogy between an MRP pump and the CFTR channel and the variety of insights that can be gleaned from such a comparative analysis. Here we address the following questions: Is it possible to produce GOF mutations on the extracellular sides of the translocation pathways of the MRPs and CFTR (mutations that may reveal long-range coupling between ATP occupancy of the NBDs and structural changes at the extracellular gates of these transporters)? Do such mutations have GOF effects in long MRPs as well as in the short MRPs and the CFTR channel? How strictly correlated are the observed GOF effects of specific side chain substitutions across these various transporters? Can this information be used to produce even stronger CFTR GOF channels?

We investigated these questions by performing a detailed analysis of a conserved phenylalanine that locates to the extracellular ends of TM6 in CFTR and the short MRPs and of the analogous TM (TM11) in the long MRPs. This region of TM6 contributes to forming an extracellular gate that occludes the translocation pathway in the crystal structure of the nucleotide-free bacterial ABC exporter, MsbA (16). In the open CFTR channel, the conserved phenylalanine (F337) lies just extracellular to a narrow region of the pore where mutations affect pore blocker sensitivity and single-channel conductance (17–20). Gao and Hwang (21) recently argued that F337 sits at the distal end of an extracellular gate that occludes the CFTR pore in its closed conformation. This position also is of interest because a mutation of this phenylalanine in a long MRP in yeast (Ycf1p, an MRP1 ortholog) scored as a hit in an intragenic screen for suppressor mutations that rescue the transport activity of an ATP binding mutant of this transporter (22). The latter observation implicates this conserved position as a possible hot spot for producing GOF mutations in the MRPs and conceivably CFTR. Our results confirm that mutations at this extracellular location have GOF effects in all 3 ABCC transporter classes: Ycf1p (long MRP in yeast), Yor1p (short MRP in yeast), and the human CFTR channel. Not every side chain substitution had a GOF effect for all 3 classes, however, indicating that the local environments around this position may differ between the long MRP pumps and the short MRPs and CFTR. By combining a GOF mutation at this extracellular position with a cytosolic GOF mutation, we produced CFTR channels that were nearly maximally active in the absence of both ATP and exogenous PKA. These findings support the development of drugs that bypass defects in either ATP binding or R domain phosphorylation in cystic fibrosis mutant channels.

MATERIALS AND METHODS

Ycf1p and Yor1p functional assays

Methods used for construction of the Yor1 mutants and analysis of oligomycin sensitivity were previously described (15). For analysis of Ycf1 mutations, the knockout strain (ycf1-∆0) from the Research Genetics (Huntsville, AL, USA) Saccharomyces cerevisiae gene deletion collection (23) was transformed by the polyethylene glycol method (24) with plasmids expressing Ycf1p alleles. The plasmids were derived by subcloning Ycf1 into pEAE93 (25) and by subsequent PCR-directed mutagenesis with verification by DNA sequencing (see primers below). Plasmid selection for transformants was on synthetic complete medium without uracil (SC-URA) (26). For gradient plate analysis, strains were arrayed in rows and grown to full density in SC-URA medium in 384-roundwell plates (Evergreen Scientific, Los Angeles, CA, USA), diluted, and spotted to fresh agar medium plates with the indicated drug (oligomycin or cadmium sulfate) concentration gradients (15). Cadmium sensitivity was assessed on SC-URA medium with a gradient produced with 10 mL 100 μM cadmium sulfate (catalog no. 383082; Sigma-Aldrich, St. Louis, MO, USA) and 30 mL plain medium. Oligomycin sensitivity was assayed in YPEG (1% yeast extract, 2% peptone, 3% ethanol, 3% glycerol), with a gradient produced with 15 ml 0.5 μg/ml oligomycin (catalog no. 75351; Sigma-Aldrich) and 25 ml plain medium. Cell arrays were created robotically with the Sciclone ALH3000 liquid handling system (Caliper Life Sciences/Perkin Elmer, Hopkinton, MA, USA), manipulating a 384-pin tool with FP6-style pins (V&P Scientific, San Diego, CA, USA) for transferring cells. Plates were incubated at 30°C and scanned manually (Yor1/oligomycin experiments) or robotically as previously described (15, 27–29). Pixel intensities were derived from time series images for each culture spot and fitted to a logistic growth equation, which was used to obtain the growth parameters (29). Functional restoration is quantified by a shift of the growth curve, reflected by change in the growth parameter, L (obtained by fitting growth curve data to a logistic growth function), corresponding to the time (hours) after which K/2 (half of carrying capacity) is reached. ∆L is the difference in L for spot cultures in the same position of a cell array on a drug gradient plate compared to the control plate without drug (29).

Analysis of CFTR channel activity in excised, inside-out membrane patches

Wild type (WT) or mutant human CFTR cDNA was transiently transfected into HEK-293T cells as described elsewhere (14, 15). All point mutations and deletion mutations were generated using mutagenic oligonucleotides (primers below), verified by DNA sequencing and subcloned into a mammalian expression vector (pCDNA, Invitrogen/Life Technologies, Carlsbad, CA, USA). Transfected cells were grown for 24–72 h at 37°C in DMEM plus 10% fetal bone serum prior to experimentation. Macroscopic or unitary currents were recorded in the excised, inside-out configuration using previously described patch-clamp techniques (14, 15). Pipettes were fabricated to tip resistances of 2–3 MΩ for macroscopic recordings and 7–10 MΩ for unitary current recordings. Pipette and bath solutions each contained (in mM): 140 N-methyl-d-glucamine, 3 MgCl₂, 1 EGTA, and 10 {2-[2-hydroxy-1,1-bis(hydroxymethyl) ethyl) amino] ethanesulfonic acid}, adjusted to pH 7.3 with HCl. All patch-clamp experiments were performed at room temperature (21–23°C). Macroscopic currents were recorded using a ramp protocol (±80 mV) with a 10 s time period. Patches were held at −60 mV for unitary current recordings. Macroscopic currents were analog filtered at 20 Hz. Unitary current signals were analog filtered at 110 Hz and then digitally filtered at 50 Hz with Clampfit 10.4 software (Axon Instruments/Molecular Devices, Sunnyvale, CA, USA).

Two methods were used to estimate single-channel open probability (P_o) values from multichannel patches. First, the product of P_o and the number of channels in the patch was estimated from gap-free records containing fewer than 8 simultaneous openings using Clampfit. The number of channels in the patch was estimated as the maximum number of simultaneous openings detected under control activation conditions (110 U/ml PKA; 1.5 mM Mg-ATP). The P_o for each subsequent condition was then calculated by dividing the NP_o product measured for that condition by this value. Second, stationary noise analysis was used to estimate P_os from multichannel records as previously described for CFTR (30, 31). This method is based on the premise that the currents that are mediated by channels with high P_os exhibit less variance (σ²) than those mediated by channels with lower P_os, which gate more dynamically. For each current record, the mean current (μ) and the current variance (σ²) were obtained from Clampfit after the baseline noise was subtracted (30, 31). The P_o for each patch was estimated from these parameters using the equation μ (1 − P_o) = σ²/i, where i is the unitary current at the experimental holding potential (−0.4 pA at −60 mV for WT-CFTR). An average P_o for the entire dataset also was derived from a plot of σ²/i vs. μ for which the slope equals 1 − P_o.

Primer sequences for cloning and site-directed mutagenesis

Ycf1p

• Forward cloning primer: CAACACAGGCATGTATATTAAGAGC

• Reverse cloning primer: TTAAACTTATGGCGTCAGAGTTGCC

• F565A: CATTGACTACTGACTTAGTTGCCCCTGCTTTGACTCTGTTC

• F565S: CATTGACTACTGACTTAGTTTCCCCTGCTTTGACTCTGTTC

• F565L: CATTGACTACTGACTTAGTTTTACCTGCTTTGACTCTGTTC

• G756D: AAGACAAACGAGCTTTTTGATCTCCAGATAAGGAGATCCC

• D777N: ACAGCTGGCAAAGGATCATTAAGTAAATAAGTGTCAGCTC

• Y1281G: GATCAAGCTCCGGCCTACCACGAGTGGAATAATTATTAAAC

Yor1p

• Forward cloning primer: CTAATTGTACATCCGGTTTTAACC

• Reverse cloning primer: TTGAGTCATTGCCCTTAAAATGG

• F468S: AGGCAACCTGGTAATATTTCTGCCTCTTTATCTTTATTTC

• F468A: AGGCAACCTGGTAATATTGCTGCCTCTTTATCTTTATTTC

• F468L: AGGCAACCTGGTAATATTCTTGCCTCTTTATCTTTATTTC

• G713D: GTGGTATTACTTTATCTGGTGATCAAAAGGCACGTATCAATTT

• Y1222G: ATAGGTAAACCAGGTCTACCGGCAAAATCAACATTTTCAA

CFTR

• Forward cloning primer: GAAGAAGCAATGGAAAAAATGATTG

• Reverse cloning primer: TCGGTGAATGTTCTGACCTTGG

• F337S: TCATCCTCCGGAAAATATCCACCACCATCTCATTCTGC

• F337A: TCATCCTCCGGAAAATAGCCACCACCATCTCATTCTGC

• F337L: TCATCCTCCGGAAAATATTAACCACCATCTCATTCTGC

• F337C: TCATCCTCCGGAAAATATGCACCACCATCTCATTCTGC

Immunoblot analysis of CFTR protein expression

Expression of the CFTR F337 mutants was verified by immunoblotting as described elsewhere (15). Forty-eight hours posttransfection, HEK-293T cells were lysed on ice in the following buffer: 130 mM NaCl, 2.7 mM KCl, 8.1 mM Na₂HPO₄, 1.5 mM KH₂PO₄, 1% w/v Triton X-100, 2 mg/ml iodoacetamide, 0.5 µM PMSF, 10 µg/ml leupeptin, 10 µg/ml pepstatin, pH 7.3. Lysates were cleared by centrifugation and protein concentrations were determined using the Micro BCA Protein Assay Kit (Thermo Scientific, Waltham, MA, USA). Lysate protein (50 µg total) was resolved by SDS-PAGE using 6% Tris-glycine gels and transferred to nitrocellulose for Western blotting. Blots were probed with anti-CFTR monoclonal antibody (MM13-4; Millipore, Billerica, MA, USA) and developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Scientific).

RESULTS

Figure 1 depicts the conserved phenylalanine at the extracellular ends of the translocation pathways of the MRPs and CFTR. Figure 1A is an alignment of TM6 for CFTR and the short MRPs and the corresponding TM11 of the long MRPs. TM6 lines the CFTR pore and the translocation pathways of the short MRPs on the basis of considerable functional and biochemical evidence (17–21, 32). Figure 1B maps onto homology models of the CFTR open and closed conformations (33) the locations of this conserved phenylalanine and of 2 previously reported GOF mutations (P355 at cytosolic base of TM6 and K978 in cytosolic loop 3 below TM9) (14, 15). These 3 sites span the CFTR pore with F337 near the presumed extracellular gate (21), where it has been argued to be extracellular to a narrow region that is a site for open channel blockers (17).

Figure 1.

The extracellular phenylalanine is conserved in short and long MRPs and resides in pore-lining TM6 near the putative CFTR extracellular gate. A) T-COFFEE multisequence alignment of human CFTR, the short and long human MRPs, and their respective yeast homologs. B) The predicted location of F337 (orange) in homology models of the CFTR channel (33). F337 is extracellular to CFTR's ligand binding site (ATP, van der Waals representation) unlike the previously identified P355 (red) and K978 (purple) allosteric sites in the full-length inward-facing (closed-state) CFTR model (left). Insets: F337 resides in pore-lining TM6 (yellow, cartoon representation) and appears to contribute to the pore architecture in outward-facing (open-state) and inward-facing (closed-state) homology models (33). The TM6 and TM12 (yellow) pore-lining helices are surrounded by an outer collar of TMs formed by TM3-4 and TM9-10 (green) that also contain allosteric sites (K978 (purple) and K190 (not shown) (14, 15).

Mutating the conserved extracellular phenylalanine has GOF effects in both long and short yeast MRPs

Figure 2 indicates that mutating the conserved extracellular phenylalanine can partially rescue the transport activities of 2 ATP binding mutants of the Ycf1p cadmium transporter (long MRP). Ycf1p confers cadmium resistance by transporting glutathione conjugates of cadmium into the yeast vacuole (22). Accordingly, Ycf1p function can be scored as cell growth on a concentration gradient of CdSO₄. To test for GOF effects, we introduced each of 3 different F565 mutations into 2 Ycf1p constructs that are expected to have reduced ATP binding affinity: a previously tested mutant that lacks an aspartate in the Walker B motif in NBD1 that coordinates the metal cofactor in Mg-ATP (D777N-Ycf1p) (16, 22, 34) and an NBD2 “A loop” mutant that lacks a conserved tyrosine that stacks against the adenine ring of ATP (Y1281G-Ycf1p) (16, 35, 36). We chose to test for GOF effects in these background constructs because F565L was identified as a hit in the aforementioned screen for mutations that suppress defective transport by the Ycf1p Walker B mutant (22). And in an earlier study (15), we observed rescue of the analogous Walker B and A loop mutants of Yor1p by cytosolic mutations that were predicted by our CFTR results (i.e., substitutions of the aforementioned proline at the base of TM6 and the cytosolic lysine below TM9). Both ATP binding mutants of Ycf1p exhibited defective cell growth on cadmium medium when compared with the WT transporter (Fig. 2). As reported previously (22), the F565L mutation partially rescued the cadmium growth phenotype of the NBD1 Walker B construct (D777N). We also observed partial rescue of the NBD2 A loop mutant (Y1281G-Ycf1p) by this side chain substitution. The 2 other F565 substitutions that were tested (A and S) had much weaker effects on the cadmium resistance phenotypes of the Ycf1p ATP binding mutants.

Figure 2.

Mutating the conserved extracellular F565 residue partially restores the function of ATP binding mutants of Ycf1p (Y1281G and D777N), but not of a signature sequence mutant (G756D). A) Representative images after 138 h of growth at 30°C. Rows consist of individual strains transformed with a plasmid containing the indicated Ycf1p alleles in a ycf1-∆0 (null allele) strain background, which were spotted across a CdSO₄ concentration gradient of 0–100 μM, right to left. The WT and ATP binding mutations, Y1281G (top panel) and D777N (middle), or the signature sequence mutation, G756D (bottom), are indicated along with the F565 substitutions, introduced alone or in combination. The vertical rectangular box indicates spot cultures analyzed in B and C. B) Restoration of Ycf1p function is observed as differential growth associated with F565 substitutions in the context of the Y1281G (top) and D777N (middle) loss of function mutations but not the G756D signature sequence mutation (bottom). Each set of growth curves is derived from the time series of data for the spot cultures at the same position in the cadmium gradient, indicated by the black rectangle in A. As illustrated in the top panel, K (average pixel density when growth ceases) and L (time at which growth reaches K/2) are growth curve parameters obtained by fitting the image data to a logistic growth function (15, 28, 29). C) Functional comparison between each single mutant, Ycf1p-Y1281G (top), Ycf1p-D777N (middle) or Ycf1p-G756D (bottom), and the respective A, S, and L substitutions for F565. Each histogram represents the mean and sem (triplicate analysis) for ∆K (red) and ∆L (blue). ∆K and ∆L were obtained by subtracting the parameter obtained for a culture spotted onto the cadmium gradient from the culture spotted onto control medium without cadmium (15, 28, 29). The statistical significance of the restoration of function was assessed by 2-tailed Student’s t test, comparing the single mutant to each double mutant. *P < 0.05.

For reference, we also introduced each phenylalanine mutation into the WT Ycf1p background and into an NBD1 signature sequence mutant that is expected to have negligible ATPase activity (G756D in Ycf1p) (16, 37, 38). The latter mutant should be difficult if not impossible to rescue by second site mutations in the TMs given that ATP hydrolysis is required for active substrate transport by the MRPs. The results in Fig. 2 confirm that none of the F565 substitutions rescued the Ycf1p signature sequence mutant or increased cadmium resistance in the WT background.

Figure 3 shows that mutating the conserved extracellular phenylalanine also had GOF effects on the analogous A loop mutant of the Yor1p oligomycin exporter (short MRP; Y1222G-Yor1p). We chose this ATP binding mutant of Yor1p for detailed analysis because our earlier results (15) showed that it was possible to rescue its oligomycin growth phenotype by introducing cytosolic GOF mutations that were predicted by our CFTR findings, and detailed ATP titrations can be performed for the analogous CFTR A loop mutant (Y1219G) to explore the mechanism underlying such GOF effects (15). The CFTR NBD1 Walker B mutant (D572N) is a severe processing mutant that cannot be so analyzed (39). Figure 3A–C shows that all 3 side chain substitutions that were tested (F468L, A, or S) partially rescued the oligomycin growth phenotype of Y1222G-Yor1p. None of the F468 substitutions had an apparent effect on growth in the presence or absence of oligomycin when introduced into the WT background. Also, as observed for Ycf1p, no rescue was observed for a Yor1p signature sequence mutant that should be defective for ATP hydrolysis (see G713D results in Fig. 3A) (15, 37, 38). Interestingly, the leucine substitution exerted the weakest GOF effect for Yor1p-Y1222G in contrast to it having a relatively stronger effect on Ycf1p-Y1281G (Fig. 2). The results in Figs. 2 and 3 establish that mutating the conserved extracellular phenylalanine can have GOF effects in both Ycf1p and Yor1p albeit with differences in the responses of these transporters to specific side chain substitutions.

Figure 3.

Mutating the conserved extracellular F468 residue in Yor1p partially reverses loss of function resulting from disruption of ATP binding (Y1222G), but not from a signature sequence mutation (G713D). A) Representative images after 120 h of growth at 30°C for the ATP binding (Y1222G) and signature sequence mutations (G713D) and the F468 amino acid substitutions, alone or in combination. Each row consists of a strain transformed with a plasmid containing the indicated Yor1p alleles in a yor1-∆0 (null allele) strain background and spotted across the same oligomycin concentration gradient (0–0.5 μg/ml). The vertical rectangular box indicates spot cultures analyzed in B and C. B) Representative growth curves corresponding to WT and Yor1p-Y1222G mutation, with and without the F565 amino acid substitutions. Growth curves were derived from the time series of data for the spot cultures at the same position in the oligomycin gradient (column 8 indicated in A). C) Functional comparison between the Yor1p-Y1222G single mutant and respective double mutants also harboring the F468A, F468S, or F468L substitution. The mean and sem (triplicate analysis) for ∆K (red) and ∆L (blue) are indicated. The statistical significance of the restoration of function was assessed by 2-tailed Student’s t test, comparing the single mutant to each double mutant. *P < 0.01. D) Plot of ΔL vs. gradient concentration supports quantitative differences in rescue of Yor1p-Y1222G function by different F468 substitutions. The Yor1 alleles are indicated by the legend in B. The ∆L value reports on loss of function and a reduction in ΔL indicates restoration in function at each position across the oligomycin gradient (15, 28, 29).

A subset of CFTR F337 mutants are strong GOF mutants

To test for analogous GOF effects on CFTR gating, the corresponding TM6 phenylalanine (F337) was replaced with 4 different residues. All F337 mutants expressed well in HEK-293T cells as evidenced by detection of the mature, fully glycosylated CFTR polypeptide by immunoblotting and of large CFTR-mediated currents across excised inside-out macropatches (Fig. 4). As previously reported (18) each F337 mutation substantially reduced the unitary currents (e.g., from −0.4 to −0.2 pA at −60 mV) (Fig. 5). The F337 mutations also reduced the efficacy of CFTR(inh)172, a commonly used blocker that inhibits from the cytosolic side and whose effects were previously reported to be blunted by other TM6 mutations (unpublished observations) (40). Conversely, inhibition by the extracellular pore blocker GlyH-101was only modestly affected by mutating F337 [see also Norimatsu et al. (41)]; accordingly, we chose to add this blocker at the end of most patch experiments to measure the small leak currents. GlyH-101 as well as CFTR(inh)172 inhibit CFTR currents in excised macropatches when added to either side of the membrane due to the lipophilic nature of these compounds and the unstirred aqueous layer inside the pipette near the membrane patch [unpublished observations; see also Yeh et al. (42) for discussion of CFTR potentiator activation from both sides of an excised membrane patch].

Figure 4.

A subset of CFTR F337 mutations are GOF mutations. A) Macroscopic current record for excised patch containing hundreds of F337S-CFTR channels showing detectable current after ATP removal and strong AMP-PNP activation. Voltage ramp protocol (±80 mV; red dotted line indicates 0 current level). Control condition was 110 U/ml PKA and 1.5 mM Mg-ATP. PKI (inhibitory PKA peptide; 1.4 μg/ml) was added to block continuous phosphorylation resulting in partial dephosphorylation due to membrane associated phosphatases (14, 15). Hexokinase (24 U/ml) and glucose (10 mM) were added to the bath to scavenge ATP followed by bath perfusion with an ATP-free solution. AMP-PNP (2 mM) and a CFTR pore blocker (100 μM GlyH-101) were added where indicated. B) Macroscopic current record for excised patch containing approximately the same number of F337L-CFTR channels showing that this mutation does not increase ATP-free channel activity or subsequent activation by AMP-PNP. Conditions were identical to A. C) Scatter plot showing the fractional ATP-free current measured at −80 mV normalized to the control current (1.5 mM ATP) for the indicated constructs. Each symbol represents an individual experiment. Mean ± SEM percent ATP-free currents were as follows: WT (0.5 ± 0.2%; n = 5); F337L (0.6 ± 0.3%; n = 5); F337C (2.5 ± 1.4%; n = 5), F337A (9.6 ± 1.4%; n = 5), and F337S (15.8 ± 4.5%; n = 10). The results for F337A and F337S were significantly different from WT by unpaired Student’s t test (P < 0.05). D) Mean relative AMP-PNP activation normalized to the control current (1.5 mM ATP) for the indicated constructs. Numbers (n) are indicated in parentheses. *P < 0.05 compared with WT by unpaired Student’s t test. E) Immunoblot of indicated constructs transiently expressed in HEK-293T cells. See Materials and Methods for details. Upper bands represent mature, fully glycosylated proteins. Lower bands represent the immature, endoplasmic reticulum forms. The relative abundance of the mature form was not obviously reduced by the F337 substitutions. Identical results were obtained for F337A in separate immunoblots (not shown).

Figure 5.

F337S-CFTR channels have increased P_os under all conditions examined. A) Unitary current recordings at a single holding potential (−60 mV; record inverted for presentation) for an excised inside-out patch containing 6 F337S-CFTR channels. PKA, ATP, and AMP-PNP concentrations were the same as for Fig. 4. PKA and ATP were removed by extensive bath perfusion (-ATP condition) prior to AMP-PNP addition. CFTR(inh)172 was added at 10 μM. Indicated above each trace is the P_o estimated for that record as described in Materials and Methods. B) Corresponding unitary current recordings for a patch containing 5 WT-CFTR channels. Conditions and methods were identical to A. Note the different current scales indicating the lower unitary currents exhibited by the F337S mutant. C–E) Mean P_os for WT-CFTR and F337S-CFTR channels estimated for the indicated conditions. Errors are sems. Numbers (n) are 4 and 5 for WT and F337S, respectively. *P < 0.05 compared with WT by unpaired Student’s t test.

Two of the four F337 mutants were strong CFTR GOF constructs (Fig. 4). F337S-CFTR and F337A-CFTR exhibited robust GOF properties that included 1) substantial currents that remained following the removal of bath ATP with a scavenger (hexokinase/glucose) and subsequent perfusion with an ATP-free solution and 2) strong activation by the poorly hydrolyzable β,γ-imidoadenosine 5′-triphosphate (AMP-PNP), which is a weak agonist for WT CFTR (example record in Fig. 4A, data summaries in Fig. 4C, D) (4, 43). Because the F337S mutation exhibited strong GOF effects at the macroscopic current level, its P_os under various activation conditions (ATP, no nucleotide, AMP-PNP) were estimated for inside-out patches that were obtained using smaller tip pipettes to reduce the number of channels per patch (<8; see Materials and Methods). The results in Fig. 5 confirm that the extracellular F337S mutation increased P_o especially following ATP removal and the subsequent addition of AMP-PNP. Interestingly, the leucine substitution (F337L), which was a strong GOF mutation in Ycf1p but a weaker GOF mutation in Yor1p, had no detectable GOF effect on CFTR channel activity (Fig. 4B–D). In sum, the data of Figs. 4 and 5 show that mutating the conserved extracellular phenylalanine also can have strong GOF effects on CFTR activity. But, as noted above for Ycf1p and Yor1p, the specific side chain substitutions that exert these effects vary among transporters.

The extracellular F337S substitution enhances the ATP sensitivities of CFTR channels including the Y1219G ATP binding mutant

GOF mutations that increase the ligand-free activities of allosteric proteins such as hormone receptors and neurotransmitter-gated channels also reciprocally enhance ligand sensitivity by biasing the equilibrium toward those conformations with the higher ligand affinities (i.e., the activated receptor or open channel) (14, 15, 44–46). The ATP titration data in Fig. 6 indicate that such reciprocity is also apparent for the F337S mutation. This substitution substantially increased the ATP sensitivity of CFTR activation when introduced either into the WT background or into the NBD2 A loop mutant that lacks the conserved tyrosine that stacks against the adenine ring of ATP at site 2 (Y1219G-CFTR). The latter ATP binding mutant of CFTR, which is analogous to the Y1281G-Ycf1p and Y1222G-Yor1p mutants, has a low apparent affinity for ATP compared with WT-CFTR [EC₅₀ of 1.95 vs. 0.12 mM; see also Wei et al. (15) and Zhou et al. (36)]. These titration results provide good evidence for long-range coupling between the ATP occupancies of the NBDs and structural perturbations at the extracellular gate of CFTR. The titration data also support the interpretation that ATP binding mutants of Ycf1p and Yor1p can be rescued by analogous GOF mutations in the TMs because these mutations reciprocally/allosterically increase ATP occupancy.

Figure 6.

The extracellular F337S mutation increases the ATP sensitivity of CFTR activation either in the WT background or in the Y1219G ATP binding mutant. ATP titrations for the indicated constructs in excised macropatches were performed as described (14, 15). Currents are normalized to the maximal currents measured at −80 mV for the highest ATP concentration tested. Note the break in the X-axis. Symbols are means ± sems. n = 4 patches for each construct except the Y1219G single mutant (n = 5). Curves are best fits to the Hill equation. *P < 0.05 compared with WT by unpaired Student’s t test. ^#P < 0.05 compared with Y1219G by unpaired Student’s t test.

The F337S substitution markedly increases the activities of CFTR constructs that cannot be activated by ATP

We previously observed that cytosolic GOF mutations increased the channel activity of the most common CF regulation mutant G551D-CFTR (14, 15). This signature sequence mutant, which is analogous to the NBD1 signature sequence mutants of Yor1p and Ycf1p discussed above, has exceptionally low channel activity that is unresponsive to ATP (47). G551D-CFTR channel activity in excised membrane patches is strongly increased by bath addition of the natural compound curcumin (14, 15, 48), which reveals the presence of gating-defective channels in the patch (Fig. 7A). Introducing the F337S substitution markedly increased the control currents mediated by G551D-CFTR channels and correspondingly reduced the relative stimulation by curcumin (Fig. 7B–D). The currents mediated by this double mutant remained ATP-insensitive as we observed earlier for P355 and K978 substitutions (unpublished observations) (14, 15).

Figure 7.

The F337S mutation increases the activity of the common CF regulation mutant, G551D-CFTR. A, B) Macroscopic currents mediated by G551D-CFTR without (A) or with (B) the F337S substitution. Activation conditions were the same as for Fig. 4. Curcumin (30 μM) was added where indicated to provide a rough measure of the number of channels in the patch (14, 15, 48). Note the much larger control current and lower relative activation by curcumin for F337S/G551D-CFTR. C) Scatter plot showing the generally larger macroscopic control currents at +80 mV for G551D-CFTR channels containing the F337S substitution. Each symbol represents an individual macropatch experiment. The scatter is due in part to variable numbers of channels in the macropatches. Note break in y axis. Mean ± SEM control currents were as follows: G551D (0.5 ± 0.1 pA; n = 5) and F337S/G551D (115.5 ± 59.2; n = 5). D) Mean relative activation by curcumin normalized to control currents for the indicated constructs. The much lower relative activation of F337S/G551D-CFTR is due to its higher control or baseline currents. Numbers (n) are indicated in parentheses. Errors are sems. *P < 0.05 by unpaired Student’s t test.

Figure 8 shows that the F337S substitution also substantially increased the activity of a CFTR truncation mutant that lacks NBD2 (Δ1198-CFTR). This construct behaves similarly to the G551D CF mutant in that it exhibits very low, ATP-unresponsive currents in excised patches under control conditions but can be stimulated by curcumin (Fig. 8A) (14, 15, 48). As for G551D-CFTR, the F337S substitution substantially increased the control currents and correspondingly reduced the relative stimulation by curcumin when introduced into this NBD2 deletion construct (Fig. 8B–D). The currents mediated by F337S/Δ1198-CFTR were insensitive to addition of the ATP scavenger as expected for a construct lacking 1 of the 2 NBDs that dimerize to form the composite ATP binding sites (Fig. 8E). But the currents mediated by this NBD2 deletion mutant remained strongly dependent on PKA phosphorylation (Fig. 8F). The latter finding supports earlier evidence that R domain phosphorylation regulates channel activity by a mechanism that is independent of controlling NBD dimerization or ATP binding (14). The results of Figs. 7 and 8 confirm that the F337S substitution is a bona fide GOF mutation that promotes the unliganded activities of CFTR constructs that cannot be stimulated by ATP.

Figure 8.

The F337S mutation enhances the ATP-independent activities of channels lacking NBD2. A, B) Macroscopic currents mediated by Δ1198-CFTR channels without (A) or with (B) the F337S substitution. Activation conditions and curcumin concentrations were identical to Fig. 7. B) The F337S/Δ1198-CFTR current was inhibited by adding a high concentration of the voltage-dependent CFTR pore blocker, glibenclamide (300 μM). C) Scatter plot showing the larger macroscopic control currents at +80 mV for Δ1198-CFTR channels with the F337S substitution. Mean ± SEM control currents were as follows: Δ1198 (2.7 ± 2.3 pA; n = 6) and F337S/ Δ1198 (177.0 ± 75.6; n = 7). D) Mean relative activation by curcumin normalized to control currents for the indicated constructs. Numbers (n) are indicated in parentheses. *P < 0.05 by unpaired Student’s t test. E) Macroscopic current record showing that the control currents mediated by F337S/Δ1198-CFTR are not reduced by scavenging ATP with hexokinase/glucose (see Fig. 4 legend for hexokinase/glucose concentrations). This experiment was repeated 3 times with identical results. F) Macroscopic current record showing that F337S/Δ1198-CFTR channel activity is strongly dependent on PKA phosphorylation. Increasing concentrations of PKA were added to the bath in the presence of 1.5 mM Mg-ATP. This experiment was repeated 4 times with identical results.

PKA sensitivity is also enhanced by the extracellular F337S mutation

A GOF mutation that increases unliganded channel opening might also affect the PKA sensitivity of channel activation given that R domain phosphorylation is required to open the channel even in the absence of ATP binding (Fig. 8) (14). This prediction is analogous to the effect of a GOF mutation to increase ligand sensitivity, as was verified for the F337S mutant in the ATP titration experiments shown in Fig. 6. The PKA titration results in Fig. 9 indicate that this GOF mutation also substantially increased the sensitivity of channel activation by PKA.

Figure 9.

The extracellular F337S mutation also increases CFTR sensitivity to cytosolic PKA. A, B) Representative macroscopic current records for WT-CFTR (A) and F337-CFTR (B) at increasing PKA concentrations in the continued presence of 1.5 mM Mg-ATP. C) Mean PKA titration data for F337S-CFTR and WT-CFTR. Each symbol represents the mean ± sem of 5 experiments. Currents were measured at −80 mV and normalized to the currents observed at 110 U/ml PKA. Note break in x axis. *P < 0.05 by unpaired Student’s t test.

Combining GOF mutations at opposite ends of the CFTR pore produces channels that are highly active in the absence of both ATP and PKA

GOF mutations at different locations in the polypeptide may have additive effects on channel activity if they perturb channel structure in distinct ways. To test this idea, the extracellular F337S mutation was combined with the previously characterized K978C mutation, which locates to the cytosolic side below TM9 (14, 15). The K978C substitution increases ATP-free channel activity and enhances the ATP and PKA sensitivities of CFTR activation similar to that shown here for the F337S mutation (14). Figure 10 shows that a double mutant with both GOF mutations had high activity in the absence of either ATP or PKA. Removing bath ATP in the standard excised macropatch protocol decreased the currents mediated by F337S/K978C-CFTR by less than 20% (Fig. 10A, B). Subsequent addition of AMP-PNP completely reversed the small current decrease following ATP removal indicating that this poorly hydrolysable analog has the same efficacy as ATP for activating the double mutant. The data in Fig. 10B show that the double GOF mutant exhibited substantially larger fractional ATP-independent currents than either single GOF mutant. More strikingly, F337S/K978C-CFTR appeared to be nearly maximally active in the absence of both PKA and ATP (Fig. 10C–F). Substantial F337S/K978C-CFTR-mediated currents could be detected for macropatches that were excised in the absence of both PKA and ATP in the bath. These currents were only slightly increased by bath addition of ATP, PKA, curcumin or VX-770, a Food and Drug Administration-approved CFTR potentiator (49), implying that this construct is nearly maximally active in the absence of any exogenous stimulus (Fig. 10C). In support of this interpretation, the P_o of F337S/K978C-CFTR estimated from multichannel records in the absence of both PKA and ATP (example record in Fig. 10D) ranged from 0.34 to 0.95 with a mean ± SEM P_o of 0.53 ± 0.08 (n = 7 patches; estimated using the conventional Clampfit protocol; see Materials and Methods). The P_o of this double mutant in the absence of PKA and ATP also was quite high (>0.9) when estimated by stationary noise analysis using a previously described protocol for CFTR multichannel analysis (Fig. 10F and Materials and Methods) (30, 31). The exceptionally high activity of this CFTR channel construct in the absence of both ATP and exogenous PKA is unprecedented to our knowledge.

Figure 10.

The F337S/K978C double mutant is nearly fully active in the absence of both exogenous PKA and ATP. A) Representative macroscopic current record showing that ATP removal by scavenger addition followed by bath perfusion with an ATP-free solution only modestly decreases the activities of F337S/K978C-CFTR channels. Conditions were identical to Fig. 4. B) Mean fractional ATP-free currents for the indicated single and double mutants. Numbers (n) are indicated in parentheses. Errors are sems. The F337S data set is from Fig. 4C. *P < 0.05 compared with either single mutant by unpaired Student’s t test. C) Representative macroscopic current record showing that F337S/K978C-CFTR channels are nearly maximally active in excised patches in the absence of both ATP and exogenous PKA. PKA (110 U/ml), Mg-ATP (1.5 mM), curcumin (30 μM), VX-770 (10 μM), and GlyH-101 (100 μM) were added where indicated. D) Unitary current recording at a single holding potential for an excised patch containing 3 F337S/K978C-CFTR channels. Neither PKA nor ATP was present in the bath. A P_o of 0.68 was estimated for this patch using the standard method of analysis (see Materials and Methods for details). The mean ± SEM P_o averaged over 7 such patches (minus both PKA and ATP) was 0.53 ± 0.21. E) Gap-free record at a single holding potential for an excised patch containing 80–100 F337S/K978C-CFTR channels showing substantial activity in the absence of bath PKA and ATP. F) Stationary noise plot for F337S/K978C-CFTR channels in the absence of PKA and ATP. Each symbol represents an individual excised patch. Some patches contained relatively few channels (<5; e.g., the record in D) and others had considerably more channels (>20; e.g., E). The data were fit to a straight line by linear regression the slope of which (1 − P_o) gives an independent overall estimate of the P_o in the absence of PKA and ATP of 0.94 (see Materials and Methods for details). When estimated from the ratio of σ²/μ for each individual experiment, the P_os ranged from 0.74 to 0.99.

DISCUSSION

Our findings support the view that all 3 subtypes of ABCC transporters use similar global mechanisms to couple the conformations of their presumed extracellular gates to the ATP occupancies of their NBDs. Mutating a conserved phenylalanine at the extracellular ends of the translocation pathways of a long MRP, a short MRP and CFTR had GOF effects for each despite the fact that the latter is a channel and the former are active transport ATPases. These transporters did exhibit varied responses to different side chain substitutions at this location. Thus, it would appear that this sort of comparative analysis can reveal similarities in transport mechanisms at a global level, although the local structures surrounding the residues in question may vary in detail among these transporters. Another major outcome was our success in exploiting this type of analysis to produce CFTR channels that are highly active in the absence of any exogenous activator (i.e., ATP and PKA). This finding supports the feasibility of discovering allosteric drugs that circumvent defects in ATP binding or R domain phosphorylation caused by CF mutations (VX-770 or ivacaftor being one such example) (49).

Allosteric repair of ATP binding mutants of a long and short MRP

The conserved extracellular phenylalanine was chosen for detailed study for 2 reasons: its location at the extracellular ends of the translocations pathways of CFTR and the long and short MRPs permitted us to test for long-range allosteric effects of extracellular perturbations on the apparent ATP occupancies of the cytosolic NBDs of these different transporters (and on the apparent phosphorylation state of the CFTR R domain), and the F565L substitution in Ycf1p (long MRP in yeast) scored as a hit in a previous intragenic screen for mutations that rescued poor growth on cadmium-containing media for yeast expressing the Walker B mutant, D777N-Ycf1p (22). We reasoned that the latter result might be related to an allosteric effect of the F565L substitution on ATP occupancy given that the D777N mutation is expected to reduce Mg-ATP binding affinity (15, 22, 36). We confirmed this finding and further showed that the same F565L substitution partially rescued the cadmium growth defect exhibited by a second ATP binding mutant, Y1281G-Ycf1p. The latter mutant is expected to have low ATP affinity at binding site 2 because the conserved tyrosine stacks against the adenine ring of ATP in other ABC transporters (34, 35). This interpretation is consistent with the strong inhibitory effect of the analogous mutation (Y1219G) on the ATP sensitivity of CFTR channel activation (15, 36), which could be reversed by introducing a second site GOF mutation of the corresponding phenylalanine.

GOF effects were also observed for the short MRP, the Yor1p oligomycin exporter, when the analogous extracellular phenylalanine was mutated. All 3 tested phenylalanine mutations reproducibly rescued growth on oligomycin medium for the A loop mutant of Yor1p. In contrast, only the originally identified F565L substitution reproducibly rescued the ATP binding mutants of Ycf1p. The other tested Ycf1p mutations (F565A, S) exhibited weak to undetectable rescue. Importantly, none of the phenylalanine substitutions rescued NBD1 signature sequence mutants of Yor1p or Ycf1p, which are expected to have negligible ATPase activity (15, 37, 38). This is consistent with the essential role of ATP hydrolysis to energize active substrate export by these pumps. GOF mutations such as the extracellular phenylalanine substitutions analyzed in this study are unlikely to reverse strong defects in ATP hydrolysis or in the coupling between the NBDs and the translocation pathway. However, such GOF mutations can at least partially rescue defective substrate export by transporter constructs for which ATP binding is rate limiting. The underlying mechanism likely involves allosteric/reciprocal effects of these GOF mutations on ATP occupancy, as supported by our ATP titration data for CFTR.

A subset of F337 substitutions are CFTR GOF mutations that reveal long-range coupling between the extracellular gate and ATP occupancy

There is good evidence that the conserved phenylalanine in CFTR (F337) lies just extracellular to a narrow region of the pore that is a binding site for various open channel blockers. Cui et al. (18) previously showed that mutating this residue reduced single-channel conductance and the affinities of several blockers that access the pore from the cytosolic side. More recently, Gao and Hwang (21) argued that F337 contributes to the formation of an extracellular gate that occludes the CFTR pore. The present results confirmed these earlier findings and also revealed that a subset of F337 substitutions are strong GOF mutants, notably, F337S and F337A. GOF effects on CFTR channel gating were operationally defined as large fractional currents that persist following ATP removal, robust activation by the normally weak agonist, AMP-PNP, and substantial increases in channel activities when introduced into CFTR constructs that cannot be activated by ATP, namely, the most common CF regulation mutant (G551D-CFTR) or a truncation mutant lacking NBD2 (Δ1198-CFTR). The F337S substitution also increased the ATP sensitivity of CFTR channel activation when introduced either into the WT background or into the Y1219G-CFTR ATP binding mutant. Classic allosteric activation models predict an increase in ligand sensitivity for a GOF mutation that increases unliganded activity because such a mutation biases the equilibrium toward the conformation with the higher ligand affinity (i.e., an activated receptor or an open channel) (44). Similar increases in ligand sensitivity have been reported for a number of GOF mutations in neurotransmitter-gated ion channels and hormone receptors (44–46). This likely explains the ATP titration data for CFTR as well as the functional rescue of analogous mutations that inhibit ATP binding in Ycf1p and Yor1p (see also above).

One notable difference was our finding that the F to L substitution was the strongest GOF mutation in Ycf1p but the weakest in Yor1p and CFTR. The lack of a GOF effect of this mutation on CFTR function seems unlikely to be due to a general effect on protein folding or channel gating. The F337L mutation did not obviously disrupt CFTR protein maturation or inhibit the control ATP-dependent currents in macropatches (Fig. 4). Instead, it seems plausible that this difference in side chain preference reflects local structural differences in the vicinity of the conserved phenylalanine between the long MRP (Ycf1p) and the 2 shorter ABCC transporters, Yor1p (short MRP) and CFTR. Currently we do not know the structural basis of how certain mutations of this conserved extracellular phenylalanine have GOF effects on MRP or CFTR function whereas others do not. The underlying structural details are also unknown for the great majority of GOF mutations that have been reported for neurotransmitter-gated channels such as the nicotinic acetylcholine receptor (46). The value of this type of comparative analysis is that it can reveal global similarities in transport mechanism even when the local structural details are unknown and in fact may differ to some extent among the transporters that are examined.

The F337S GOF mutation also provides insights into how PKA phosphorylation regulates CFTR

PKA-mediated phosphorylation is essential for WT CFTR channel activity (9). The underlying mechanisms are unclear and possibly complex (10). Two findings from our analysis of the F337S GOF mutant impact our understanding of how PKA regulates CFTR channel activity. First, the strong PKA dependence of the channel activity of the F337S/Δ1198-CFTR truncation construct (Fig. 8F) argues for a regulatory mechanism that is independent of any effect that phosphorylation might have on NBD dimerization (see also 14). A widely held view is that PKA phosphorylation increases CFTR activity by promoting NBD dimerization (11, 12). There is evidence that the unphosphorylated R domain, which contains nearly all of the consensus PKA sites, can inhibit NBD dimerization and that this inhibition is relieved by PKA phosphorylation of this large cytosolic domain (11, 12). The F337S/Δ1198-CFTR construct lacks NBD2, cannot form an NBD1-NBD2 dimer and is unresponsive to ATP. Yet its activity was almost completely dependent on PKA phosphorylation. This indicates there must be another mechanism at work in addition to or instead of an effect of PKA phosphorylation on NBD dimerization. Conceivably, elements of the R domain physically interact in a phosphorylation-dependent manner with certain cytosolic loops/TMs to regulate their stiffness or flexibilities downstream of ATP-induced NBD dimerization [see also Hegedüs et al. (13)].

The second finding that impacts our understanding of the PKA regulatory mechanism was the long-range allosteric effect of the extracellular F337S mutation on PKA sensitivity (Fig. 9). The higher PKA sensitivity of this GOF mutant supports the view that the R domain is an allosteric modulator of CFTR channel activity whose phosphorylation state is conformationally coupled to the channel open state. A GOF mutation that biases the equilibrium to the open state should reciprocally increase PKA sensitivity (in addition to ligand sensitivity) because the phosphorylation states of its various PKA sites are allosterically coupled to the open channel conformation (14, 50). Similar arguments have been proposed to explain the regulation of neurotransmitter-gated channels and other allosteric proteins by phosphorylation (51, 52). The interesting aspect here is that an effect on PKA sensitivity over such a long-range could be detected (i.e., between a structural perturbation on the extracellular side of the pore and the apparent phosphorylation of a cytosolic domain).

Engineering a superactive CFTR

The F337S/K978C double mutant has the highest single-channel activity in the absence of exogenous PKA and ATP of any CFTR construct that we have characterized to date. PKA and ATP are essential for WT channel activity in excised membrane patches. GOF mutant channels that exhibit fairly high activities in the absence of ATP have been reported previously by us (14, 15) and others (30) but to our knowledge those CFTR mutants must first be phosphorylated by exogenous PKA for this activity. We anticipated that the F337S and K978C mutations would have additive GOF effects on ATP-free CFTR activity because they locate to opposite sides of the pore where they presumably impact CFTR structure in different ways. Earlier we observed additive effects of other mutant combinations of CFTR (15) as has the Auerbach group (46) in their mutational analysis of the nicotinic acetylcholine receptor. The bigger surprise was the virtually complete independence of the activity of the F337S/K978C double mutant on exogenous PKA. Either this construct is highly active when unphosphorylated or it is basally phosphorylated to a greater degree by endogenous kinases in unstimulated cells prior to patch excision. Either explanation would be consistent with the concept of the R domain as an allosteric modulator whose phosphorylation state is sensitive to mutations that affect the equilibrium between closed and open states. Irrespective of the specific mechanism, the ability to produce CFTR channels that are highly active in the absence of both exogenous PKA and ATP supports the feasibility of discovering CF drugs that function like allosteric modulators to bypass defective regulation of CF mutant channels by either of these factors.

Glossary

ABC

ATP binding cassette

AMP-PNP

β,γ-imidoadenosine 5′-triphosphate

CFTR

cystic fibrosis transmembrane conductance regulator

GOF

gain of function

MRP

multidrug resistance protein

P_o

single-channel open probability

NBD

nucleotide binding domain

SC-URA

synthetic complete medium without uracil

TM

transmembrane spanning domain

WT

wild-type