CF (cystic fibrosis) is caused by mutations in CFTR (CF transmembrane conductance regulator), which cause its mistrafficking and/or dysfunction as a regulated chloride channel on the apical surface of epithelia. CFTR is a member of the ABC (ATP-binding-cassette) superfamily of membrane proteins and a disease-causing missense mutation within the ABC signature sequence; G551D-CFTR exhibits defective phosphorylation and ATP-dependent channel gating. Studies of the purified and reconstituted G551D-CFTR protein revealed that faulty gating is associated with defective ATP binding and ATPase activity, reflecting the key role of G551 in these functions. Recently, high-throughput screens of chemical libraries led to identification of modulators that enhance channel activity of G551D-CFTR. However, the molecular target(s) for these modulators and their mechanism of action remain unclear. In the present study, we evaluated the mechanism of action of one small-molecule modulator, VRT-532, identified as a specific modulator of CF-causing mutants. First, we confirmed that VRT-532 causes a significant increase in channel activity of G551D-CFTR using a novel assay of CFTR function in inside-out membrane vesicles. Biochemical studies of purified and reconstituted G551D-CFTR revealed that potentiation of the ATPase activity of VRT-532 is mediated by enhancing the affinity of the mutant for ATP. Interestingly, VRT-532 did not affect the ATPase activity of the Wt (wild-type) CFTR, supporting the idea that this compound corrects the specific molecular defect in this mutant. To summarize, these studies provide direct evidence that this compound binds to G551D-CFTR to rescue its specific defect in ATP binding and hydrolysis.
CF (cystic fibrosis), a disease that affects multiple organs including the respiratory, gastrointestinal, pancreatic and reproductive tract, is caused by mutations in the CFTR (CF transmembrane conductance regulator) gene . The normal version of this gene codes for a phosphorylation- and nucleotide-regulated chloride channel important for promoting the fluid transport necessary for proper hydration of the epithelium lining the tubular organs above. CFTR is a member of the ABC (ATP-binding-cassette) superfamily of membrane proteins and, as such, it is a multidomain membrane protein that possesses two membrane domains and two NBDs (nucleotide-binding domains; NBD1 and NBD2). CFTR also possesses a regulatory domain that confers its regulation by PKA (protein kinase A)-dependent phosphorylation [2,3].
The CF-causing missense mutation G551D-CFTR is associated with severe disease . Unlike the more common ΔF508-CFTR, G551D-CFTR is not mistrafficked and exhibits normal expression on the cell surface . However, this mutant is associated with disease because it lacks channel activation by ATP [4,5]. Molecular models and the existing crystal structure of NBD1 suggest that the glycine residue at position 551 interacts with ATP at one (the catalytically active) of the two nucleotide-binding sites located at the NBD1–NBD2 heterodimer interface and contributes to the ATPase activity of the intact protein [6–8]. Therefore it has been suggested that the disease-causing mutation G551D will interfere with ATP binding, NBD1 and NBD2 heterodimerization and the ATPase activity that results from this domain–domain interaction . In fact, we showed in studies of purified and reconstituted G551D-CFTR that it is severely defective as an ATPase [10,11]. We hypothesize that repair of the molecular defect in this mutant protein will require augmentation of ATP binding and this can be reported as an increase in ATPase activity.
Recently, small molecules have been identified on the basis of high-throughput cell-based assays [12,13]. A subgroup of these molecules is effective in modulating the functional expression of two disease-causing mutants: ΔF508-CFTR and G551D-CFTR . As we have optimized the purification and functional reconstitution of G551D-CFTR, we have an important reagent with which to evaluate whether certain of these pharmacological compounds act directly to modify the mutant CFTR protein. Further, we can assess the mechanism of action through which such ‘potentiating’ compounds may work. In the present paper, we focused on one such pharmacological modulator, VRT-532 , because unlike other modulators, VRT-532 has been reported to work specifically on mutant CFTR rather than other membrane proteins and, hence, is likely to interact directly with the mutant protein.
Vesicle-based flux assay of Wt (wild-type) CFTR and G551D-CFTR function
The measurement of iodide efflux from membrane vesicles was adapted from a previously described cellular iodide efflux method . CFTR-infected Sf9 cells were homogenized in PBS containing protease inhibitor tablets (Roche Diagnostics, Mississauga, ON, Canada) using an Emulsiflex C3 high-pressure homogenizer (Avenstin, Ottawa, ON, Canada) at 20000 p.s.i. (1 p.s.i.=6.9 kPa). Unsheared cells and large cell debris was pelleted with a low-speed centrifugation (250 g for 20 min) and discarded. The supernatant was centrifuged at 100000 g for 1.5 h, and the crude membrane pellet was resuspended in an iodide-containing buffer (150 mM KI and 20 mM Mops, pH 7.4). In order to form vesicles, the resuspended pellet was passed through a 27 gauge needle 20 times, followed by sonication. To remove iodide from the extravesicular solution, vesicle suspensions were subjected to Sephadex G-50 gel-filtration columns saturated in an iodide-free external solution (150 mM potassium glutamate and 20 mM Mops, pH 7.4). The potassium-selective ionophore valinomycin was added (20 μM) to shunt changes in potential difference generated by I−-selective conductance through CFTR.
Measurements of iodide efflux were made using an iodide-selective electrode (Lazar Research Laboratories, Los Angeles, CA, U.S.A.). Tracings were recorded and analysed using the Digidata 1320A Data Acquisition System with Clampex 8 software (Axon Instruments, Sunnyvale, CA, U.S.A.).
Purification and reconstitution of Wt CFTR and G551D-CFTR
Detailed protocols regarding the generation of Wt and mutant CFTR–His proteins are described elsewhere . A frozen Sf9 cell pellet from a 1 litre expression culture was thawed and solubilized in PBS. Cells were then disrupted using a French press set at 1000 p.s.i. and centrifuged at 500 g for 20 min at 4 °C. Supernatant containing crude membranes was centrifuged for 2 h at 100000 g at 4°C to yield a plasma membrane and microsome-enriched fraction. Membranes were solubilized for 2–4 h in 8% PFO (pentadecafluoro-octanoic acid) and 25 mM phosphate (pH 8.0). Procedures for purification, reconstitution and phosphorylation of purified CFTR–His were as described elsewhere [16,17]. Phosphorylated samples were pelleted using an airfuge, washed twice with 50 mM Tris/HCl, 50 mM NaCl, 2.5 mM MgCl2 and 1 mM DTT (dithiothreitol) at pH 7.5 and then dialysed overnight against 4 litres of washing buffer.
ATPase assay of purified Wt CFTR and G551D-CFTR protein
ATPase activity was measured as the production of [γ-32P]Pi from [γ-32P]ATP as described by Gross et al. [17a]. The assay was carried out in a reaction mixture (50 μl) containing 1 μg of NBD protein, 10% (v/v) glycerol, 50 mM Tris/HCl, 50 mM NaCl, 2 mM MgCl2, 0.1 μM dodecyl maltoside and 8 μCi of [γ-32P]ATP (3000 Ci/mmol) and unlabelled ATP at pH 7.5. The reaction mixture (25 μl) contained 1 μg of reconstituted protein, 50 mM Tris/HCl, 50 mM NaCl, 2.5 mM MgCl2 and 1 mM DTT at pH 7.5. Pi and ADP were separated from ATP by TLC using the same elution conditions as described by Ramjeesingh et al. .
ATP dose response for the ATPase activity of the full-length protein was fitted with the Michaelis–Menten function using the curve-fitting programs in Prism (GraphPad). Comparison between three or more groups was performed using the one-way ANOVA program with secondary analysis of data pairs using the Tukey test. P values less than 0.05 were considered statistically significant.
A novel iodide efflux assay reports potentiating activity of VRT-532 on G551D-CFTR
VRT-532 has been shown to potentiate cAMP-activated chloride ion flux through G551D-CFTR in Fischer rat thyroid epithelial cells grown in monolayers and studied in voltage-clamp experiments conducted in the Ussing chamber apparatus . The increase in chloride flux described above could report changes in the function of other membrane proteins that have an impact on the driving force for chloride ion flux through G551D-CFTR rather than changes in the intrinsic activity of the mutant protein. Furthermore, this increase could reflect changing in trafficking of the mutant protein. Therefore, as the first step towards understanding the mechanism of action of this small molecule in G551D-CFTR function, we were prompted to re-evaluate the effect of this modulator on the channel activity of G551D-CFTR in a more simplified system.
Typically, activation of the channel function of CFTR and CF-causing mutants (including G551D-CFTR) by membrane-permeable agonists of cAMP-dependent PKA has been studied in intact cells using a convenient method known as the iodide efflux assay [18–20]. In this assay, cells were preloaded with iodide and efflux through open CFTR channels was measured after activation with agonists of PKA phosphorylation at 1 min intervals. Although this is a very useful assay, it is limited with respect to sensitivity and kinetic analysis. At the other end of the spectrum, patch–clamp studies of single channels in isolated membrane patches provide exquisite resolution of the kinetics of channel gating; however, as the channel activity of certain mutants, including G551D-CFTR, is severely impaired, acquisition of sufficient data to draw conclusions regarding the pharmacological effects of interesting compounds is time-consuming and often limited to a small number of channel proteins. For these reasons, we developed a modified efflux assay using iodide-loaded, inside-out membrane vesicles prepared from Sf9 cells expressing CFTR (Figure 1A). Vesicles were prepared from Sf9 cells infected with baculovirus containing cDNA coding for Wt CFTR or G551D-CFTR using methods that have been described previously . The vesicles were loaded with a buffer containing KI and passed through a gel-filtration column saturated with potassium glutamate in order to generate an outward iodide gradient. One would predict that CFTR in the membrane will mediate iodide efflux once activated by addition of PKA (200 nM) and ATP (1 mM) and this efflux can be detected using an iodide-sensing electrode as shown in Figure 1. The potassium ionophore, valinomycin (20 μM), will be added to ensure that any intravesicular potential difference that would reduce iodide efflux was dissipated. Therefore the rate of PKA+ATP-activated iodide efflux from these vesicles should be limited by the open time of the CFTR channels in these membranes.
Vesicular assay of CFTR function in Sf9 membranes
As shown in Figure 1(B), we confirmed that vesicles formed from native Sf9 membranes, which are not infected with baculovirus containing CFTR, failed to show any iodide efflux after addition of PKA and ATP, as expected for membranes lacking regulated CFTR channels. Furthermore, this absent response did not reflect inefficient vesicular loading with iodide or a paucity of loaded vesicles, as addition of the vesicle lytic reagent, Triton X-100, caused a large and immediate bolus of iodide release. On the other hand, as shown in Figure 1(C), inside-out vesicles prepared from Sf9 cells expressing Wt CFTR steadily effluxed iodide after PKA plus ATP addition. As shown in Figure 1(D), the PKA- and ATP-activated iodide flux was only observed in vesicles containing CFTR, in multiple comparative studies, supporting the claim that this response reports CFTR channel activation and the rate of efflux provides a measure of the open time of a large population of CFTR channels.
Figure 2 shows an example of the regulated efflux from membrane vesicles expressing G551D-CFTR. As expected for this mutant , there is negligible flux mediated by addition of PKA (200 nM) and MgATP (1 mM). Importantly, the addition of VRT-532 (15 μM) enhanced iodide efflux significantly, supporting the idea that this compound acts on the mutant protein directly to cause opening of the channel gate.
VRT-532 enhances channel activity in G551D-CFTR
VRT-532 directly binds reconstituted G551D-CFTR and modifies its ATPase activity
It has been shown that ATP-dependent heterodimerization of the two NBDs of CFTR is required for their activity as an ATPase and is permissive for subsequent conformational changes leading to opening of the chloride channel gate [22–24]. Further, it has been suggested that the failure of G551D-CFTR to bind ATP accounts for its defective ATPase activity and channel gating . These findings prompted us to test the hypothesis that VRT-532 potentiates channel opening of G551D-CFTR by directly binding to enhance ATP affinity and hydrolysis. As previously determined, purified and reconstituted G551D-CFTR exhibits a very low level of ATPase activity relative to the Wt protein (0.1±0.1 versus 1.4±0.1 nmol·μg·h−1 respectively) in the presence of 1 mM MgATP . As predicted, addition of VRT-532 caused an increase in the intrinsic ATPase activity of the purified, reconstituted mutant, measured as the increased generation of radiolabelled Pi from radiolabelled ATP (Figure 3A). Different concentrations of VRT-532 were tested, and a Boltzmann fit of the corresponding ATPase activities yielded an EC50 of approx. 3 μM (Figure 3B).
VRT-532 enhances intrinsic ATPase activity of G551D-CFTR
Although the ATPase activity of G551D-CFTR was significantly enhanced by the addition of VRT-532, the activity of the treated mutant protein failed to achieve the level of activity exhibited by the Wt CFTR protein (measured at 1 mM MgATP) as shown in Figure 4(A). Interestingly, the ATPase activity of the Wt protein was not further enhanced by the addition of VRT-532 (10 μM), suggesting that the consequences of VRT-532 binding to the two proteins are different (Figure 4B).
VRT-532 significantly enhances the ATPase activity of G551D-CFTR but not Wt CFTR
In Figure 5 and Table 1, the apparent ATP dependence of the ATPase activity exhibited by G551D-CFTR after treatment with VRT-532 (5 μM) is shown. In contrast with the dose response obtained for untreated G551D-CFTR, which failed to exhibit saturable ATP dependence, the data corresponding to the VRT-532-treated protein could be fitted using a Michaelis–Menten function (r2=0.9). These findings suggest that VRT-532 acts to enhance the ATPase activity of G551D-CFTR by increasing the apparent affinity for ATP.
VRT-532 enhances the ATPase activity of G551D-CFTR by increasing the apparent affinity for ATP
Promising compounds have been identified in high-throughput cell-based screens of chemical libraries [12,13]. However, the molecular mechanism of action of these compounds is unknown and understanding such mechanisms remains a major challenge for development of such lead compounds into specific drugs. In the present paper, we have shown the potential value of membrane vesicle transport assays and biochemical studies of purified and reconstituted CF-mutant proteins for evaluating direct interactions with small-molecule modulators and their mode of action. In these ‘proof of principle’ studies, we revealed the direct action of one such compound, VRT-532, to enhance the ATP affinity of the CF-causing mutant, G551D-CFTR.
As previously mentioned, VRT-532 is a particularly interesting small-molecule modulator as it enhances the channel activity of at least two different CFTR genotypes, including ΔF508 and G551D-CFTR, suggesting that it may have a general role in repairing the molecular defects in multiple disease variants . In addition, it appears to be multifunctional, correcting the misfolding defect in ΔF508-CFTR as well as its altered channel gating . In addition, these consequences of VRT-532 treatment appear to be relatively specific for CFTR, as compared with misfolding mutants of the closely related ABC protein, P-glycoprotein . This is not the case for other related small molecules such as VRT-325 [13,25]. Therefore the findings of the present study suggest that VRT-532 directly binds to purified reconstituted G551D-CFTR to modulate its activity, providing the impetus for future studies to determine whether it binds directly to ΔF508-CFTR and to define the site for binding on these two mutants.
Both the chloride channel gating and ATPase functions of CFTR likely report multiple dynamic conformational changes occurring throughout the whole molecule and, hence, the specific site at which VRT-532 binds to the protein is difficult to predict . For example, ATPase activity requires the interaction between NBD1 and NBD2 and it is modulated by connections with the membrane-spanning domains. Both the primary site at which ATP catalysis is mediated and the modulatory site at which ATP binding and positive regulation of the catalytic site occurs are situated at the NBD1 and NBD2 interface [15,27]. Interestingly, we know that ATP catalysis is also modified by changes in the structure of the membrane-spanning domains. For example, channel pore blockers, which are known to bind to the membrane-spanning domains, cause inhibition of ATPase activity . Furthermore, the intramolecular interactions that mediate channel gating have yet to be fully understood. It has been suggested that gating probably involves a series of transient changes in the interactions between the coupling helices extending from the membrane domains and the cytosolic domains, including the NBDs and the ‘R’ domain as well as probable changes in the interaction among intramembrane helical segments . Therefore the effect of VRT-532 on ATPase activity and channel opening could be mediated by binding to any of the domains or domain–domain interfaces discussed above.
Since the VRT-532 compound exerts a profound effect on ATPase activity by the G551D-CFTR protein rather than on the Wt protein, we speculate that it may be binding to or indirectly stabilizing an intramolecular interaction which is important for ATPase activity and is altered in the mutant protein. The domain–domain interface that is most likely to be altered in this mutant is that interface in between NBD1 and NBD2 according to the putative location of Gly-551 based on structural models of CFTR generated using the prokaryotic ABC protein, Sav1866, as a template [7,29–31]. Therefore we suggest that VRT-532 may be binding directly to stabilize the NBD1 and NBD2 interface and/or to bind to either the coupling helices or regions of the ‘R’ domain which participate in stabilizing this NBD1–NBD2 interface. Interestingly, as VRT-532 also partially repairs the folding defect of the major mutant ΔF508 and this defect is thought to reflect an altered interaction between NBD1 and the fourth intracellular coupling helix extending from MSD2 (membrane-spanning domain 2) , it is conceivable that VRT-532 binds in a region that modifies this interface and that modulation of this interface also promotes NBD1 and NBD2 interaction. We are currently optimizing the purification and reconstitution of the major CF mutant for the purpose of testing this hypothesis directly.
While G551D-CFTR is not a common mutation in the CF patient population, it does account for disease in approx. 5% of cases and the clinical phenotype is considered very severe. Hence, the insights regarding the molecular mechanisms for functional repair of the G551D-CFTR mutant by the small molecule, VRT-532, generated in the present study will provide the template for future drug development and potentially improvement in the clinical care of these severely affected patients. Currently, a clinical trial is in progress that is investigating the efficacy of a different molecule (VX-770) in patients bearing the G551D-CFTR mutation (http://www.cff.org/research/ClinicalResearch/FAQs/VX-770). These trials are garnering a great deal of attention and excitement. This molecule is not yet available to academia and its structure and mechanism of action remain unknown, but once it is available, we will apply our unique research tools to understanding how it works.
We acknowledge the gift of purified catalytic subunit of PKA from Dr S. S. Taylor (Howard Hughes Medical Institute and Department of Chemistry and Biochemistry, University of California, San Diego, CA, U.S.A.) as well as Dr Robert Bridges (Department of Physiology and Biophysics, Rosalind Franklin University, North Chicago, IL, U.S.A.) for VRT-532, a component of a library of CFTR modulators.
This work was supported by the Canadian Cystic Fibrosis Foundation; the Canadian Institutes of Health Research (through the BREATHE programme); and the Cystic Fibrosis Foundation U.S.A. [grant number BEAR06DDSO].