Alteration of the Cystic Fibrosis Transmembrane Conductance Regulator Folding Pathway OF THE MUTATION ON THE THERMODYNAMIC STABILITY AND FOLDING YIELD OF NBD1*

The cellular phenotype of the most common cystic fibrosis-causing mutation, deletion of phenylalanine 508 ( (cid:68) F508) in the amino-terminal nucleotide binding domain (NBD1) of the cystic fibrosis transmembrane con- ductance regulator (CFTR), is the inability of the mutant protein to fold and transit to the apical membrane of affected epithelial cells. Expressed NBD1s were purified and folded in vitro into soluble monomers capable of binding nucleotide. Here we report that the (cid:68) F508 mutation has little effect on the thermodynamic stabil- ity of the folded NBD1. The (cid:68) G D,GdnHCl0 is 15.5 kJ/mol for the wild type NBD1 and 14.4 kJ/mol for NBD1 (cid:68) F. In contrast, the mutation significantly reduces the folding yield at a variety of temperatures, indicating that Phe-508 makes crucial contacts during the folding process, but plays little role in stabilization of the native state. , rate off-pathway

The cellular phenotype of the most common cystic fibrosis-causing mutation, deletion of phenylalanine 508 (⌬F508) in the amino-terminal nucleotide binding domain (NBD1) of the cystic fibrosis transmembrane conductance regulator (CFTR), is the inability of the mutant protein to fold and transit to the apical membrane of affected epithelial cells. Expressed NBD1s were purified and folded in vitro into soluble monomers capable of binding nucleotide. Here we report that the ⌬F508 mutation has little effect on the thermodynamic stability of the folded NBD1. The ⌬G D,GdnHCl 0 is 15.5 kJ/mol for the wild type NBD1 and 14.4 kJ/mol for NBD1⌬F. In contrast, the mutation significantly reduces the folding yield at a variety of temperatures, indicating that Phe-508 makes crucial contacts during the folding process, but plays little role in stabilization of the native state. Under conditions that approximate the efficiency of maturation in vivo, the rate off-pathway is significantly increased by the disease causing mutation. These results establish a molecular mechanism for most cases of cystic fibrosis and provide insight into the complex processes by which primary sequence encodes the threedimensional structure.
Cystic fibrosis is a common fatal genetic disease caused by mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) 1 gene (1)(2)(3). The product of this gene is a plasma membrane cAMP-dependent Cl Ϫ channel gated in response to binding and hydrolysis of ATP (4 -7). Although more than 500 CF-causing mutations have been identified, the ⌬F508 mutation in the amino-terminal nucleotide binding domain (NBD1) is the most prevalent, accounting for approximately 70% of the disease-causing alleles (2,3). The ⌬F508 mutation leads to diminished amounts of mutant CFTR in the membranes of epithelial cells (8 -12), and, thus, the decreased chloride conductance that is a hallmark of the disease.
Two findings indicate that the basis of this phenotype is a defect in protein folding. First, the mutation destabilizes the functional conformation of a synthetic peptide containing the Phe-508 region (13). Second, when cells expressing the ⌬F508 protein are grown at reduced temperature, the maturation defect is partially corrected (14). Along with the finding that the ⌬F508 mutant protein is functional when it reaches the native state (13)(14)(15), this information suggests that correcting the maturation defect may ameliorate this form of the disease.
Recent studies suggest that the structural maturation defect in the ⌬F508 mutant occurs at an early step in vivo (16,17). In addition, these studies indicate that the ⌬508 mutation affects the rate of CFTR maturation. Moreover, CF mutations cluster in the NBDs (2,3), and many CF mutations identified as maturation-defective (18), are located in NBD1, implicating defective folding of this domain in the vast majority of CF cases. To quantitatively investigate the effect of the ⌬F508 mutation on the folding of this domain in greater detail, we have developed an in vitro folding system. In the present study, we use the system to examine the effects of this common CF-causing mutation on the thermodynamic stability and folding pathway of NBD1.

Construction of the Expression Vectors
Containing CFTR NBD1 and NBD1⌬F cDNAs-Expression cassette polymerase chain reaction (EC-PCR) was employed to synthesize the cDNA fragments of CFTR NBD1 and NBD1⌬F containing a 5Ј NdeI site, a 3Ј XhoI site, and a stop codon. The sequence of the primers used to define and amplify NBD1 and NBD1⌬F from Gly-404 to Ser-589 are as follows: G404 primer, 5Ј-GAAGATCGGGCATATGGGATTTGGGGAATTATTTGAG-3Ј (37 bases); S589 primer, 5Ј-GAATTGCGCCTCGAGTTAGCTTTCAAATAT-TTCTTTTTCTG-3Ј (41 bases). The plasmid pBQ2.4 and pCOF⌬F508 containing CFTR cDNA were used as templates to amplify NBD1 and NBD1⌬F, respectively. Digested PCR products were ligated into the NdeI and XhoI sites of the pET28a plasmid (Novagen). Correct recombinants were identified by restriction digestion and sequenced to confirm the fidelity of the PCR step and formation of the proper construct.
Overexpression and Purification of NBD1 and NBD1⌬F Proteins-Single colonies of BL21 (DE3) Escherichia coli transformed with pET28a⅐NBD1 or pET28a⅐NBD1⌬F were used to inoculate 100 ml of LB media containing 30 g/ml kanamycin and grown at 37°C with rigorous shaking until the absorbance at 600 nm reached 0.6 unit. Cells were then induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside, and incubation was continued for an additional 3 h at which time the cells were harvested by centrifugation at 3,200 ϫ g for 20 min at 4°C. The cell pellet was resuspended in 40 ml of 20 mM Tris-HCl, pH 7.9, containing 500 mM NaCl, 5 mM imidazole (buffer A) and sonicated on ice three times for 30 s to break the cell wall and shear the DNA. The lysate was centrifuged at 15,000 ϫ g for 30 min to collect the inclusion bodies and cellular debris. The pellet was then resuspended in 5 ml of buffer A containing 6 M GdnHCl and incubated on ice for 1 h to solubilize the proteins. The sample was centrifuged at 39,000 ϫ g for 20 min to remove any remaining insoluble material. The supernatant was then applied to a His⅐Bind resin column (2.5 ml of resin per 100 ml of culture) that was previously charged with 50 mM NiSO 4 and equilibrated with buffer A. The column was washed with 10 volumes of buffer A containing 6 M GdnHCl and 6 volumes of buffer A containing 20 mM imidazole and 6 M GdnHCl. Purified NBD1s were eluted with 6 volumes of buffer A containing 400 mM imidazole and 6 M GdnHCl. The eluate was dialyzed twice against 100 volumes of 100 mM Tris-HCl, pH 7.4, 2 mM EDTA for 12 h at 4°C. Slow removal of the denaturant led to precipitation of the recombinant proteins. After dialysis, the samples were centrifuged at 16,000 ϫ g for 20 min at 4°C. The precipitated proteins were collected, lyophilized, and stored at Ϫ70°C. Tricine SDS-PAGE (10% polyacrylamide) was carried out as described previously (19). Gels were stained with Coomassie Brilliant Blue R-250.
Folding of NBD1 and NBD1⌬F-Purified CFTR NBD1 and NBD1⌬F were dissolved in 6 M GdnHCl. The samples were diluted 30-fold with ice-cold 100 mM Tris-HCl, pH 8.0, containing 400 mM L-arginine⅐HCl, 2 mM EDTA, and 1 mM dithiothreitol (buffer B) to a final protein concentration of 18 M and incubated at 4°C overnight. Protein concentration was determined by absorbance at 280 nm using a molar extinction coefficient of 13,490 M Ϫ1 cm Ϫ1 calculated from the amino acid sequence (20).
HPLC Gel Filtration Chromatography-Folded proteins were analyzed on a Macrosphere GPC 300 7U gel filtration column that was equilibrated previously with 20 mM Tris-HCl, pH 7.4, 150 mM NaCl, 400 mM L-arginine buffer. Elution profiles were monitored at 280 nm using a flow rate of 0.5 ml/min. Molecular size markers used for gel filtration chromatography were blue dextran, bovine serum albumin, carbonic anhydrase, and aprotinin, corresponding to 2000, 69, 29, and 6.5 kDa, respectively.
TNP-ATP Binding-Nucleotide binding was determined using fluorescence enhancement of the ATP analog TNP-ATP upon binding to NBD1 and NBD1⌬F (13,21,22). Samples containing 0.9 M NBD1 or NBD1⌬F in buffer B at the indicated concentration of TNP-ATP were excited at 410 nm with a 2 nm band pass. The emission was then measured at 555 nm (with a 4 nm band pass).
Unfolding of NBD1 and NBD1⌬F by GdnHCl and Heat-Samples containing 1.8 M folded NBD1 or NBD1⌬F in 20 mM Tris-HCl, pH 8.0, were incubated at room temperature for 2 h in the indicated concentration of GdnHCl. Intrinsic tryptophan fluorescence emission spectra were collected using 282 nm exciting light. Thermal unfolding of NBD1 and NBD1⌬F was monitored by light scattering at 400 nm. Scattered light due to aggregation of unfolded protein in the absence of a chemical denaturant was measured at a 90°angle. The temperature of samples containing 0.9 M NBD1 or NBD1⌬F in buffer B was ramped from 6 to 75°C at a rate of 0.5°C/min.
Temperature Dependence of NBD1 and NBD1⌬F Folding-For studies of the temperature dependence of folding, NBD1 and NBD1⌬F were dissolved in 6 M GdnHCl at the desired protein concentration. Unfolded proteins were diluted 30-fold with buffer B to final concentrations of 18 M or 2 M and incubated at the indicated temperature overnight. The protein samples were centrifuged at 16,000 ϫ g for 10 min to remove insoluble, misfolded protein. Folded, soluble protein was determined from the intrinsic tryptophan fluorescence at room temperature. All protein was soluble and folded after dilution of GdnHCl at 4°C.
Aggregation of NBD1 and NBD1⌬F-For the off-folding pathway studies, NBD1 and NBD1⌬F were solubilized in 6 M GdnHCl, then were diluted 30-fold with buffer B to a final protein concentration of 18 M and incubated at 23°C. Scattered light at 400 nm was measured at a 90°angle to monitor the aggregation progress.

NBD1 and NBD1⌬F Expression, Purification, and Folding in
Vitro-A pET based expression vector capable of directing the expression of exons 9 through 12 of CFTR (Gly-404 to Ser-589) as a fusion with a polyhistidine sequence was constructed. A 1-liter culture of BL21 (DE3) E. coli transformed with the plasmid expressed 120 mg of NBD1 in 3 h in response to isopropyl-1-thio-␤-D-galactopyranoside. This high rate of expression leads to the formation of insoluble NBD1 inclusions which can be isolated and dissolved in 6 M GdnHCl prior to further purification on a nickel chelate affinity column. The resulting denatured NBD1 is greater than 95% pure as assessed by laser densitometry of Coomassie-stained SDS-PAGE (Fig. 1A).
Denatured NBD1 and NBD1⌬F were folded into nucleotide binding monomers in vitro. The key to the success of the folding was the inclusion of L-arginine in the folding buffer (23,24). Possibly, the guanidinium group of arginine increases the solubility of exposed polar side chains in the denatured NBD1 and the amphipathic character of the amino acid protects exposed hydrophobic interaction surfaces in the folded domains. In this regard, it is important to note that NBD1 has been removed from the context of the large multidomain membrane protein in which it normally resides. Thus, surfaces that are normally involved in domain-domain interactions in the intact protein, and which may be responsible for the tendency of NBD1 to form polymers (21) or interact with membranes in vitro (25), may be shielded by arginine. However, in the presence of 400 mM arginine, NBD1 can be folded and maintained as a soluble, functional monomer as assessed by intrinsic tryptophan fluorescence, size exclusion chromatography, and TNP-ATP binding (Fig. 1, B-D).
Fluorescence emission spectra of the wild type and ⌬F508 NBD1s reveal a pronounced blue shift in the peak position and an increase in fluorescence intensity upon removal of the chemical denaturant, consistent with burial of the single tryptophan at position 496 in a hydrophobic environment as the domain folds (Fig. 1B). Four additional lines of evidence argue that both NBD1s are folded. First, they are soluble in the absence of denaturant. Second, both elute from a molecular sizing column intermediate to carbonic anhydrase (29 kDa) and aprotinin (6.5 kDa), a position consistent with a globular monomer with a predicted molecular weight of 22,000 (Fig. 1C). Third, as has been described previously for expressed soluble NBD1 (15,21), in vitro folded NBD1 and NBD1⌬F bind the nucleotide TNP-ATP (Fig. 1D). The apparent K d of both NBD1s for TNP-ATP is 3 M under these conditions in agreement with previous results (21). Finally, denaturation of the folded NBD1s is highly cooperative, indicating disruption of an ordered structure ( Fig. 2A,  inset).
Thermodynamic Stability of NBD1 and NBD1⌬F-Notably, the concentration of GdnHCl required to induce cooperative unfolding is similar for the two domains. The C m values (halfmaximum denaturant concentration for unfolding) are 1.5 M for NBD1 and 1.3 M for NBD1⌬F ( Fig. 2A). Comparison of the

Mutational Effects on CFTR Folding 7262
thermodynamic stabilities of NBD1 and NBD1⌬F calculated from this reversible transition indicates that ⌬⌬G D is only 1.1 kJ/mol ( Fig. 2A). In comparison, the ⌬G D of the wild type NBD1 is 15.5 kJ/mol. Thus, the effect of the ⌬F508 mutation on the stability of NBD1 under these conditions is minimal. This is in marked contrast to the effects of ⌬F mutation on the urea-dependent unfolding of peptide fragments of NBD1 (13). The cooperativity of the transition is reflected in the m GdnHCl values (the slope of natural log of the fraction folded versus the denaturant concentration) ( Fig. 2A, inset). The m GdnHCl values of 10.3 kJ mol Ϫ1 M Ϫ1 for the wild type and 11.2 kJ mol Ϫ1 M Ϫ1 for the ⌬F508 mutant NBD1 indicate that the change in solventaccessible surface area upon folding is similar for the two NBD1s and appropriate for a protein of this size. Finally, thermal denaturation of the expressed domains reveals only a minor difference in the stability of the two domains (Fig. 2B). Wild type NBD1 has a melting temperature, T m , of 49°C, and NBD1⌬F508 a T m of 46°C. Clearly, the ⌬F508 mutation affects the temperature dependence of the domain structure (Fig. 2) to a lesser degree than it affects the temperature dependence of CFTR maturation (14). Thus, these results indicate that the inability of the ⌬F508 CFTR to transit to the apical membrane cannot be explained by a reduction in the native state stability of the mutant protein.
Temperature Sensitivity of the Folding Yield-In contrast, the temperature sensitivity of the folding process was dramatically altered by the CF causing mutation (Fig. 3A). When the concentration of the chemical denaturant GdnHCl is reduced by rapid dilution, two competing processes occur. The first, progression from the denatured state to the native state along the folding pathway, is a first order process and protein concentration independent. The second, formation of inappropriate associations between molecules that lead off the normal folding pathway, is a second order process, and, thus, depends upon protein concentration. At low temperature and protein concentrations, the intramolecular interactions responsible for folding of NBD1 into soluble, functional, monomers predominate. At elevated temperatures and protein concentrations, intermolecular interactions occur that result in the formation of an insoluble conformer that scatters 400 nm of light. Thus, at 2 M final NBD1 concentration and 37°C, 63% of the wild type polypeptide folds into the soluble conformation, while only 38% of the ⌬F508 assumes the folded conformation. At 18 M final polypeptide concentration and 25°C, 29% of the wild type domain reaches the native state in contrast to 19% of the ⌬F508 mutant.
Rate of Aggregation-When the rate of formation of the off- FIG. 2. Denaturation of NBD1 and NBD1⌬F. The native structures of folded NBD1 and NBD1⌬F were denatured by addition of the chemical denaturant GdnHCl or by increasing the temperature. In the absence of GdnHCl, the denatured domains form insoluble associations that scatter 400 nm light. A, denaturation of NBD1 resulted in decreased tryptophan fluorescence and a red shift in the emission maximum as the single tryptophan, Trp-496, is exposed to solvent (see Fig.  1B). 1.8 M folded wild type NBD1 (EOOE) and NBD1⌬F (Ⅺ---Ⅺ) in buffer B were incubated with GdnHCl at the indicated concentration for 2 h. The sample was excited at 282 nm and fluorescence emission spectra were collected. The equilibrium dependence of the fluorescence emission peak position on the denaturant concentration (inset) reveals cooperative unfolding of the domain. As this is a reversible process which approximates a two-state conversion, the free energy of denaturation (⌬G D ) can be calculated from the fraction folded over the transition region. Extrapolation to the absence of denaturant indicates that ⌬⌬G D between NBD1 and NBD1⌬F is 1.1 kJ/mol. B, thermal denaturation of folded NBD1 and NBD1⌬F. NBD1 (0.9 M, solid line) and NBD1⌬F (0.9 M, dashed line) in buffer B were heated from 6 to 75°C at a rate of 0.5°C/min. Scattered light was measured at 400 nm at an angle of 90°.   (17), a lag phase followed by an increase in light scattering is observed. Significantly, the ⌬F508 mutation decreases the length of the lag phase and increases the rate of change in light scattering indicating that the rate of formation of the off-pathway conformer is enhanced by the disease-causing mutation (Fig. 3B). However, the altered kinetics may not be due to a direct effect of the mutation on the rate constant for an off-pathway step. Rather, they may reflect an increased concentration of a folding intermediate which is prone to self-association at high concentration. In this regard, in the cell the proteasome is responsible for degradation of both wild type and ⌬F508 CFTR (26,27). However, when proteolysis is inhibited in vivo, the efficiency of maturation is not enhanced as misfolded full-length CFTR simply aggregates into a detergent-insoluble form (26 Previous results demonstrated that a peptide fragment of NBD1, P67, containing the A consensus region and a region of homology around Phe-508, capable of binding ATP, is destabilized by the ⌬F508 mutation (13). These results suggested that conditions counteracting the destabilizing effects may allow maturation of the mutant CFTR and rescue of the disease phenotype (13,28). It was subsequently observed that reduction of the temperature at which expressing cells were grown increased the efficiency of CFTR maturation (14). Moreover, functional CFTR chloride channels could be observed in the plasma membrane at the sensitive temperature of 35°C after growth at the permissive temperature of 26°C. These findings are consistent with the current results which indicate that a step on the folding pathway, rather than the stability of the native state is affected by the mutation. Thus, destabilization of the peptide P67 by the ⌬F508 mutation suggests it may provide a model of a kinetically trapped folding intermediate (28).
The fact that Phe-508 is critical for interactions that direct folding, but makes little contribution to the stability of the native state is not unique. For example, a large number of mutations of the P22 tail spike protein affect its folding but not its native state stability, indicating that non-native state interactions may be important for directing the folding pathway (29,30). These mutations are, thus, temperature-sensitive for folding (tsf), as is ⌬F508. Interestingly, global second site suppressor mutants of these tsf mutations have been isolated (30). It remains an intriguing possibility that intragenic suppressors of the ⌬F508 phenotype (31, 32) may act by correcting the folding defect.
Understanding the interactions that take place on the CFTR folding pathway may have profound importance for describing the mechanisms of both the disease process and how primary sequence determines the final native structure of proteins. Therapies directed at correcting the folding defect deserve further consideration as treatments for cystic fibrosis. Potentially, alteration of molecular chaperone expression or of cellular conditions to increase the on-pathway rate or decrease the offpathway rate may prove useful. However, it is important to remember that simply inhibiting the final proteolytic off-pathway step is apparently not adequate to correct the disease phenotype (26,27) as might be expected if the steps prior to proteolysis were in equilibrium with the native state. More likely, positive impact on the disease state will require intervention at the initial off-pathway steps.