Destabilization of the transmembrane domain induces misfolding in a phenotypic mutant of cystic fibrosis transmembrane conductance regulator.

Two phenotypic missense mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) channel pore (L346P and R347P in transmembrane (TM) segment 6) involve gain of a proline residue, but only L346P represents a significant loss of segment hydropathy. We show here that, for synthetic peptides corresponding to sequences of CFTR TM6 segments, circular dichroism spectra of wild type and R347P TM6 in membrane mimetic environments are virtually identical, but L346P loses approximately 50% helicity, implying a membrane insertion defect in the latter mutant. A similar defect was observed in the corresponding double-spanning ("hairpin") TM5/6-L346P synthetic peptide. Examination of the biogenesis of CFTR revealed that the full-length protein harboring the L346P mutation is rapidly degraded at the endoplasmic reticulum (ER), whereas the wild type and the R347P protein process normally. Furthermore, a second site mutation (R347I) that restores in vitro membrane insertion and folding of the TM5/6-L346P peptide also rescues the folding and cell surface chloride channel function of full-length L346P CFTR. The correlated in vitro/in vivo results demonstrate that destabilizing local hydrophobic character represents a sufficient signal for marking CFTR as a non-native protein by the ER quality control, with accompanying deleterious consequences to global protein folding events.

Polytopic membrane proteins may be particularly susceptible to folding defects when non-native residues of high hydrophilicity are introduced into transmembrane positions (1). These phenomena are epitomized by the cystic fibrosis transmembrane conductance regulator (CFTR) 1 , a chloride channel protein that becomes dysfunctional in cystic fibrosis, the most common lethal autosomal recessive genetic disease in the Caucasian population (2). CFTR belongs to the ATP-binding cassette membrane transporter gene superfamily (3). Established as the CF gene product (4), CFTR contains 1480 amino acids and consists of two homologous halves (5), each with a nucleotide-binding domain (NBD1 and NBD2) and a predicted sixhelix transmembrane (TM) domain (TMD1 and TMD2). The two halves are linked by a cytoplasmic regulatory R-domain. Deletion of Phe-508 (⌬F508) in NBD1, the most common mutation among CF patients (67% among all the CF patients) (6), causes CFTR misfolding in the endoplasmic reticulum that gives rise to degradation of the protein and the severe clinical phenotype (5). However, of the more than 1000 CF-phenotypic mutations in CFTR that have now been reported, over 100 occur in membrane-spanning regions.
Critical components of the channel pore have been identified in CFTR TM helices 5 and 6 (TM5/6, residues 308 -350). TM6 has been shown to have a role in the determination of the permeation properties of CFTR (7); several residues in TM6 have been proposed to contribute to the anion binding sites, including Arg-334, Lys-335, Phe-337, Thr-338, Ser-341, Arg-347, and Arg-352, whereas the central region of TM6 has been localized as a main determinant of both anion binding and anion selectivity in CFTR (6 -9). A number of mutations occurring in TM5/6 have been found to cause mild (usually pancreatic sufficient) forms of CF, two of which involve introduction of a proline residue: L346P, a mutation that was identified in two unrelated Cypriot patients in 1994 (10); and a second sequentially adjacent CF-phenotypic mutant, R347P (11). It has been suggested that Arg-347 in TM6 forms a salt bridge with an aspartate (Asp-979) located in TM9 (12), an interaction that would be abrogated by the loss of the native TM6 Arg residue. Arg-347 has also been proposed to contribute to the pore of the CFTR Cl Ϫ channel and anion conduction (8), and the pore properties of the channel have been observed to be altered in various Arg-347 mutants (6,8). Pro residues have typically been implicated in channel gating when they occur in poreforming helices (13,14).
Gain of Pro in a TM helix can have several consequences. The Pro pyrolidine ring is bulky, which causes steric constraints on the conformation of the preceding residue in the helix (15). As well, Pro residues may introduce a kink in the helix structure (16,17). And, as an imino acid, Pro lacks an amide proton on the X-Pro bond for participation in helix-stabilizing intramolecular H-bonds (18). As such, the introduction of Pro into a TM helix increases the net hydrophilicity of the segment, because it results in a non-H-bonded backbone carbonyl group in the preceding turn of the helix. Furthermore, in CFTR mutant L346P, the loss of Leu significantly increases the local hydrophilicity of this TM segment, i.e. on the Liu-Deber hydropathy index where values are scaled between ϩ5 and Ϫ5, Leu ranks third (ϩ4.76), whereas Pro ranks 19th (Ϫ4.92) out of the 20 commonly occurring amino acids (19). In the case of R347P, the Arg positive charge is lost, but this mutation exchanges a polar residue (Arg ranks 14th, at Ϫ2.77) with one of comparable hydrophilicity.
In the present work, we have used solid-phase peptide synthesis to prepare sequences corresponding to TM6 segments of wild type (WT) and mutants L346P and R347P of CFTR, along with some corresponding double-spanning TM5/6 peptides for comparative structural analyses. In parallel, we examined the relative effects of L346P versus R347P on cellular processing of full-length CFTR. The results provide striking in vivo/in vitro correlates of the consequences of a missense mutation located in the predicted TM6 segment and demonstrate that destabilizing local hydrophobic character may represent a sufficient signal for recognizing CFTR as a non-native protein by the ER quality control.

EXPERIMENTAL PROCEDURES
Peptide Synthesis and Purification-Peptides were synthesized using standard Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry on a PerSeptive Biosystems Pioneer peptide synthesizer (20). Synthesis employed the use of the Pioneer's extended cycle for TM5/6 peptides or single-spanning TM6 peptides. In a typical synthesis, 0.22 mmol/g PAL-PEG-PS resin (Applied Biosystems) was used to produce amidated C terminus. The peptides were then cleaved with a mixture of 88% trifluoroacetic acid/5% phenol/5% ultrapure water/2% triisopropylsilane. The cleaved peptides were precipitated with ice-cold diethyl ether. Centrifuged pellets were redissolved in 5% acetonitrile and lyophilized. The crude peptide powder was dissolved in 30% acetonitrile, and 10 mg of the crude peptide was loaded onto a C4 preparative reversed phasehigh performance liquid chromatography column and then eluted with a water/acetonitrile gradient (30 -75% acetonitrile over 60 min). The major peak was collected and lyophilized. Mass spectrometry was used to confirm the molecular weight of the purified peptide.
Circular Dichroism Spectroscopy-CD spectra were collected using a Jasco J-720 spectropolarimeter. Lyophilized peptide was added to a buffer containing 50 mM LPC and 25 mM Tris at pH 8.0, and the mixture was vortexed to ensure complete peptide solubilization. Samples were measured at peptide concentrations between 25 and 50 M. Measurements were taken using a quartz cuvette with a path length of 0.1 mm at room temperature. Spectral scans were performed from 250 to 190 nm with a step resolution of 0.4 nm, a speed of 20 nm/min, and a bandwidth of 1.0 nm. Spectra are reported as the average of three scans.
Fluorescence Resonance Energy Transfer Analysis-FRET measurements were performed using peptides labeled with dansyl chloride as the acceptor fluorophore at the N terminus, with the Trp residue in TM6 (C terminus) serving as the donor fluorophore. Dansyl chloride labeling was carried out at room temperature by mixing 2 mg of dansyl chloride with 100 mg of dry peptide resin (before cleavage) in 7% N,N-diisopropylethylamine/93% dimethylformamide solution (v/v) overnight. Emission spectra were measured at room temperature in an LPS-220B fluorescence spectrometer. The excitation wavelength was 295 nm, with a step size of 1 s, and an integration time of 0.5 s. Spectra are reported as the average of three separate measurements.   (20); and blue represents the residues flanking the putative TM residues. The underlined Trp residues were employed in various fluorescence experiments (see text). by overlapping PCR using the appropriate mutagenic primers. The PCR products were subcloned into the BspEI/Bst1107I sites of CFTR. Baby hamster kidney (BHK) cells were stably transfected with the pNUT expression plasmids, containing the wild type (WT), L346P, or R347P CFTR, harboring an HA-epitope in the C-terminal tail of CFTR (CFTR-C int HA) (21). Following clonal selection in the presence of methotrexate (500 M), 50 -100 individual colonies were pooled and expanded for experiments. Transient expression of COS-1 cells was carried out as described previously (22).

Construction and Expression of CFTR Variants in Mammalian
Immunoblotting and Metabolic Pulse-chase Studies-CFTR immunoblotting was performed with the mouse monoclonal anti-HA Ab (Covance) using enhanced chemiluminescence (ECL) detection, and immunoblots were quantified with densitometry, as described previously (23). The Na ϩ /K ϩ -ATPase was visualized by the a6F Ab (Developmental Studies Hybridoma Bank, University of Iowa). The stability of CFTR variants was monitored by the pulse-chase technique. First, the cellular methionine and cysteine content was depleted in Met-and Cys-free medium (37°C, 30 min) and then pulse-labeled for 15 min in the presence of 0.1 mCi of [ 35 S]methionine and [ 35 S]cysteine (Amersham Biosciences) at 37°C. Following the indicated chase period in complete medium, membrane proteins were solubilized in 1 ml radioimmune precipitation assay buffer (150 mM NaCl, 20 mM Tris-HCl, 1% Triton X-100, 0.1% SDS, and 0.5% sodium deoxycholate, pH 8.0) supplemented with protease inhibitors (10 g/ml leupeptin and pepstatin, 0.5 mM phenylmethylsulfonyl fluoride, and 10 mM iodoacetamide). Immunoprecipitates, obtained with anti-HA Ab, were analyzed by SDS-PAGE and fluorography. The radioactivity incorporated into CFTR was quantified using a PhosphorImager (Amersham Biosciences) with the Image-QuaNT software (Molecular Dynamics) as described (24).
Iodide Efflux Assay-The plasma membrane cAMP-dependent halide conductance of stably transfected BHK cells was determined by the iodide efflux assay (25). Iodide efflux was initiated by replacing the loading buffer with efflux medium (composed of 136 mM nitrate in place of iodide). The extracellular medium was replaced every minute with efflux medium (1 ml). After a steady state was reached, the intracellular cAMP level was raised by agonists (10 M forskolin, 0.2 mM CPT-cAMP, and 0.2 mM isobutylmethyl xanthane) to achieve maximal phosphorylation of CFTR. The collection of the efflux medium was resumed for an additional 6 -9 min. The amount of iodide in each sample was determined with an iodide selective electrode (Orion).

RESULTS
Hydrophobicity Threshold of CF-phenotypic Mutant TM Segments-Although both L346P and R347P represent a gain of a Pro residue in CFTR TM6, the resulting local 346/347 diads (PR and LP, respectively) differ significantly in hydrophobic character. This situation is highlighted using TM Finder, a web-based program that reliably predicts whether a stretch of amino acids has sufficient hydrophobicity to exist as a stably inserted transmembrane helix (26). The outputs for the wild type (WT) and L346P CFTR sequences support an initial hypothesis that the L346P mutation decreases the average net hydrophobicity of the original TM6 segment sufficiently to prevent the proper membrane insertion of the full TM6 segment (Fig. 1, A and B). Specifically, the program identifies a 20amino acid stretch (residues 330 -349) in the WT CFTR sequence that likely corresponds to the membrane-embedded residues of TM6. In contrast, the output for the L346P mutant sequence (residues 330 -341) fails to predict a sufficiently long stretch of amino acids that are above the threshold hydrophobicity required for a TM helix (27,28). However, unlike L346P, the R347P mutation doesn't involve a significant change in hydrophobicity, and TM Finder predicts that the R347P mutant has the same membrane-inserted amino acid stretch (residues 330 -349) as the WT TM6 sequence (Fig. 1C). It may be noted that programs such as TM Finder may underpredict the actual total of membrane-inserted residues in a given TM segment, because interfacial residues at both termini of the segment tend to be rich in hydrophilic residues, such as Lys and Arg, which strongly drive the local average hydropathy down (26).
Synthesis and Circular Dichroism Spectra of CFTR TM6 Peptides-To experimentally test the TM-Finder prediction, we synthesized Lys-tagged versions of the TM6-WT and the two mutant (L346P and R347P) sequences (Table I). The Lystagged methodology involves adding several Lys residues (in the present case, three Lys) to both the N and C termini, a strategy that greatly facilitates peptide purification and characterization while maintaining the native-like structural properties of the TM segment they flank (20). The circular dichroism (CD) spectra of these two peptides are shown in Fig. 2A. Although the TM6-WT peptide adopted an ␣-helical structure in the presence of LPC micelles, the TM6-L346P peptide displays only ϳ50% of the helicity observed for the TM6-WT sequence. In contrast, the CD spectra of TM6-R347P and the TM6-WT peptides are virtually superimposable.
Because the L346 locus appeared to be most affected by the Pro mutation, we further examined the effect of the L346P mutation in an expanded context by synthesizing the helixloop-helix TM5/6-WT and TM5/6-L346P peptides. As "helical hairpins" when folded, these constructs constitute the minimal tertiary contact systems representative of the CFTR TM domain (29). CD spectra indicate that the TM5/6-L346P construct exhibits a 25% decrease in helicity when compared with the TM5/6-WT construct in SDS micelles, consistent with a 50% decrease in one TM helix (Fig. 2B).
Fluorescence Studies of CFTR Single TM6 and Double TM5/6 Peptides-The proposition that the L346P TM segment is only partially inserted into micellar membranes was further examined by fluorescence experiments. To this end, we synthesized two additional peptides containing a TM-embedded fluorescent probe introduced through the conservative F342W mutation (TM6-WT(F342W) and TM6-L346P(F342W), respectively) (Table I). Characteristic Trp fluorescence spectra of these two peptides in detergent micelles are presented in Fig. 2C. Noting that a membrane-embedded Trp residue will typically display increased fluorescence intensity with a blue-shifted position versus an aqueous-located counterpart, the data indicate that the Trp residue in the WT species resides in an apolar environment (maximum near 320 nm) whereas the Trp in the L346P peptide is largely aqueous exposed (shoulder near 340 nm), supporting the notion that the apolar-to-polar mutation prevents proper TM6 insertion.
Fluorescence resonance energy transfer (FRET), the transfer of the excited-state energy from the initially excited donor to an acceptor, can provide further information as to the proximity of donor and acceptor chromophores (30). The donor-acceptor distance can be determined from the efficiency of energy transfer, which can be derived from steady-state measurements of the extent of donor quenching due to the acceptor. We exploited this phenomenon to detect differences in the distance between the N and C termini in TM5/6 peptides. To perform this analysis, we modified the sequences of the peptides through inser- Core-and complex-glycosylated forms are indicated by empty and filled arrowheads, respectively. D, relative translational rate of the full-length L346P CFTR. The biosynthetic rate of the full-length, core-glycosylated CFTR was measured by pulse labeling. To avoid clonal variations, COS-1 cells were transiently transfected with the WT and L346P CFTR. The rate of radioactive amino acid incorporation was measured after 10-min labeling. CFTR was immunoprecipitated with anti-HA Ab and visualized by fluorography. The higher mobility band is conceivably due to alternative initiation. The radioactivity associated with the WT and L346P CFTR was normalized for cellular protein and was not more than 4% variant in two experiments. tion of a Trp residue (donor moiety) near the peptide C terminus (Table I) and subsequently labeled the WT and mutant peptides with a dansyl (acceptor moiety, N-terminal) group. The FRET spectra obtained indicate a significant decrease in the donor quench in case of mutant TM5/6-L346P(W) peptide as compared with its WT counterpart (Fig. 2D). Note that the diagram depicts the quench in donor (Trp) intensity near 338 nm; a corresponding increase is observed in dansyl fluorescence intensity near 550 nm (not shown). The FRET results suggest that (i) the WT sequence is likely folded into a helical hairpin and (ii) a substantial increase in fluorophore separation occurs between the N and C termini of the L346P mutant hairpin compared with the WT sequence.
The L346P Mutation Impairs the Folding of CFTR in Vivo-ER-retained, core-glycosylated (or incompletely folded) CFTR can be readily distinguished from the mature, complex-glycosylated (or folded) CFTR by immunoblot analysis, based on the faster electrophoretic mobility of the core-glycosylated form compared with the complex-glycosylated CFTR. Because impaired post-translational folding of the CFTR usually causes its biosynthetic processing arrest, we examined the processing of full-length L346P-and R347P-CFTR by immunoblotting and pulse-chase analysis of BHK cells, which stably express CFTR. To facilitate the detection of CFTR, an HA-epitope tag was inserted at the C terminus of the channel (CFTR-C int HA) (21). As shown in Fig. 3A, immunoblot analysis of equal amounts of cell extracts demonstrated that the L346P, but not the R347P, mutation, prevented the expression of the complex-glycosylated CFTR. The R347H missense mutation, associated with a mild functional defect of CFTR channel activity, was also expressed at the same level as the WT CFTR, confirming previous reports (31). Similar results were obtained in transiently transfected COS-1 cells, indicating that the cellular phenotype of the L346P CFTR is independent of the expression system used (Fig. 3B).
Impaired steady-state expression of L346P-CFTR could be a consequence of its rapid degradation at the ER and/or of accelerated removal of the channel from post-Golgi compartments (32). To distinguish between these scenarios, the biogenesis of L346P CFTR was monitored in BHK cells by the metabolic pulse-chase technique. Although the accumulation of the complex-glycosylated WT-CFTR was obvious after 2 h of chase, the L346P mutation prevented the appearance of the complexglycosylated form as shown by autoradiography (Fig. 3C). These results suggest that the L346P mutation imposes a folding defect on CFTR, leading to the retention and degradation of the mutant at the ER. On the other hand, the L346P mutation does not appear to cause premature translational termination or failure of the TM5-6 or TM7-8 segments to insert into the ER, because similar amounts of radioactively labeled coreglycosylated WT and L346P CFTR were accumulated during a 10-min radioactive labeling (Fig. 3D).
A Second Site Mutation in TM6 Restores Biosynthetic Processing of CFTR-To test the assumption that destabilization of local TM5/6 hairpin formation may inhibit post-translational folding in the context of full-length CFTR, we searched for a second site mutation that could re-establish the stability of the L346P TM5/6. Consideration of amino acid replacements in the vicinity of the L346P mutation identified a second site mutation (R347I) that restored the hydrophobicity of the TM6 L346P-containing segment to the threshold level that ensured membrane insertion according to TM Finder (Fig. 1D) (26). The L346P/R347I single spanning TM6 peptide was first synthesized, and analysis of its CD spectrum confirmed that this mutation restored the ␣-helical content of this TM6 double mutant to its WT counterpart (Fig. 4A).
We then assessed whether the R347I mutation could restore hairpin formation of the TM5/6-L346P polypeptide by the FRET assay. Using a TM5/6-L346P/R347I peptide in which a Trp residue was inserted near the C terminus (Table I), and in which the N terminus was labeled with a dansyl group, we found that the donor fluorescence quench in the double mutant was now similar to that of the WT TM5/6 ( Fig. 4B).
If destabilization of the TM5/6 hairpin accounts for the processing defect of the L346P CFTR, introducing the second site mutation should correspondingly restore the folding and biosynthetic processing of full-length CFTR. This was indeed the case. Immunoblot analysis demonstrated the appearance of the complex-glycosylated L346P/R347I CFTR in both transiently transfected COS-1 and stably transfected BHK cells, whereas no detectable amount of L346P CFTR was present (Figs. 3B and 5A, respectively). Functional assessment of the plasma membrane protein kinase A-activated halide conductance confirmed the partial reversion of the processing defect by demonstrating that the cAMP-stimulated iodide release of the L346P CFTR (6.3 Ϯ 0.2 nmol/min) was increased by 3-fold in the presence of the second site mutation (18.4 Ϯ 03 nmol/min) (Fig.  5B). The detection of L346P CFTR by functional assay, but not by immunoblotting, is conceivable due to the higher sensitivity of the iodide efflux assay.

Membrane Insertion Defects in CF-phenotypic Mutants-
Misfolding of the membrane domains of polytopic proteins arising from genetic mutations can account for the molecular basis of human disease. In the present work, we have performed parallel in vitro and in vivo studies on selected CF-phenotypic TM-based mutations of CFTR of which both involve a gain of proline, with the outcome that the introduction of a Pro residue per se is not the determinant of folding. Rather, the context of FIG. 4. Circular dichroism and fluorescence spectra for CFTR TM6 peptides. A, CD spectra of TM6-WT and TM6-L346P/R347I in LPC micelles. B, Trp fluorescence spectra for the unlabeled and labeled TM5/ 6-WT and TM5/6-L346P/R347I peptides in LPC micelles. FRET measurements were performed using peptides labeled with dansyl chloride as the acceptor fluorophore with the Trp residue in TM6 serving as a donor fluorophore. The samples were excited at 295 nm. Each result was the average of three separate measurements. Spectra of the two peptides were essentially indistinguishable. the mutation within the TM segment as a whole, as well as the character of the residue lost, are also critical for the global folding of the protein. Thus, despite the similarity of the two mutations and their adjacent positions in the sequence, the TM6-L346P peptide displayed only ϳ50% of the helicity observed for the TM6-WT sequence, whereas the TM6-R347P retained WT character, indicating that the ability of the L346P peptide to properly insert into the apolar milieu has been significantly compromised (Fig. 2A).
Given that, in intact CFTR, structural effects within TM6 likely are influenced by the sequentially vicinal TM5 segment, it is conceivable that interactions with such neighboring helices could stabilize the inserted state of a mutant TM6 sequence. Previous authors have speculated that in certain cases, the insertion of TM helices can be aided by interactions with neighboring TM segments (33,34). To address this situation, we synthesized and compared helix-loop-helix peptides corresponding to the CFTR TM5/6-WT and the TM5/6-L346P sequence. However, we observed a corresponding decrease in the helical content for the mutant sequence (in this experiment, ϳ25% of total helical content), supporting the findings with the single-spanning species (Fig. 2B) that approximately half of TM6 becomes aqueous-exposed. Fluorescence resonance energy transfer experiments on labeled TM5/6 constructs further suggested that L346P prevents formation of proper TM6 topology, because FRET effects were significantly reduced in the TM5/ 6-L346P peptide versus the corresponding WT construct.
The physiological role of the TM hairpins, which represent the basic tertiary unit in the topogenesis and folding of multispanning membrane proteins, is not fully understood. Although the formation of helix-helix interactions in TM hairpins is likely to be required for co-or post-translational integration (35), the cytosolic surface of TM domains may serve as a platform for the binding of the large cytosolic domains (including the NBD domains), as indicated by the crystal structure of MsbA and BtuD (35,36). Therefore, disrupting or destabilizing the topology of the TM5-TM6 may impair the global conformation of CFTR by perturbing interdomain interactions in vivo. The biogenesis of wild type CFTR itself is inefficient. Depending on the expression system used, only 20 -60% of newly synthesized WT is converted into the fully processed complexglycosylated form (24,37). This inefficient maturation of wild type CFTR has been attributed to the fact that the polarresidue-rich TM6 fails to behave as a proper membrane anchor (11). Thus, a mutation in TM6 such as L346P may further reduce the efficiency of proper membrane anchoring of the protein.
Consistent with these considerations, we found that fulllength CFTR protein harboring the L346P mutation is subjected to core glycosylation but was unable to fold and was rapidly degraded in vivo. There are two suggested mechanisms to counteract this anchoring deficiency: the ribosome and the ER translocon co-operate to prevent TM6 from passing through the membrane co-translationally and/or cytosolic domains of the ion channel post-translationally maintain TM6 in a membrane-spanning topology (11). Because the full-length L346P protein is indeed synthesized, an additional possibility is that the protein is able to compensate, at least in part, for the topological defect at the TM5/6 locus via TM-packing interactions with the second (TM7-12) CFTR TM domain. Based on our study, it appears that L346P may affect local CFTR TM5/6 structure to such an extent that the ER-associated qualitycontrol mechanism recognizes the mutant as non-native and marks it for degradation. As a result, escape from the ER and cell surface delivery of L346P CFTR is severely compromised FIG. 5. The effect of a second site mutation on the expression and function of L346P CFTR. A, CFTR expression of BHK cells, stably expressing the indicated construct, was visualized by immunoblotting with the mouse monoclonal M3A7 and L12B4 anti-CFTR Ab (top panel). Equal amounts of protein were loaded on each lane. Filled arrowhead, complex-glycosylated form; empty arrowhead, core-glycosylated form. B, cAMPstimulated iodide efflux of BHK cells expressing wild-type (wt), L346P, or L346P/ R347I CFTR. Basal (empty symbols) and protein kinase A-activated iodide release was determined by iodide-sensitive electrode as described under "Experimental Procedures." Protein kinase A agonist mixture was added at 0 min.  (Table I) (excluding Lys tags) arranged into two putative TM helices with an intervening loop at 328 -330. Yellow circles correspond to residues predicted by TM Finder (Fig. 1) to be the membrane-embedded TM5 and TM6 components. Light blue circles correspond to adjacent putatively aqueous-based residues. Gray circles correspond to residues that may be positioned at the water/membrane interface. B, introduction of Pro-346 into TM6 is depicted as unfolding the TM6 helix between residues 341 and 349, likely placing them in the microenvironment of the water/membrane interface. The loop residues may similarly be pulled toward the water/ membrane interface. See text for a further discussion. (Fig. 3A). Although insertion of unpaired charged residues in the TM domain of membrane proteins is recognized as a signal for ER degradation (38), our results suggest that destabilization of local segmental hydrophobic character is also sufficient to induce misfolding of CFTR. The partial reversion of the L346P CFTR processing defect by a second site mutation (R347I) (Fig. 3B), which restores full-length TM6 insertion potential (Fig. 1D), suggests that segment hydrophobicity is prominent among the factors that play determining roles in the post-translational folding of CFTR.
Our overall findings are consistent with a model in which the portion of native TM6 (Fig. 6A) unfolds upon introduction of Pro-346 and likely enters an aqueous-based local microenvironment (Fig. 6B). As discussed in the introduction, this consequence is promoted when introduction of Pro significantly increases the hydrophilicity of the local segment. Because the resulting TM6 is now too short to span the cellular bilayers, a compensatory "pull" on the residues within the TM5/6-L346P loop region may also occur (Fig. 6B). This latter effect, in turn, could misalign and/or tilt TM5 and TM6 and thus abrogate (some) native TM5-TM6 side-chain-side-chain packing interactions; for example, the loss of Leu-346 may contribute to the weakening of native TM5-TM6 van der Waals packing interactions.
Implications for the Molecular Defects in CF-phenotypic Mutants of CFTR-Thermodynamic models proposed for membrane protein folding involve steps related to interfacial partitioning, interfacial folding, and insertion (39,40). With respect to insertion, studies on model TM peptide systems have identified a "threshold hydrophobicity," which, along with sufficient segment length and helical propensity in non-polar environment, serve as parameters that characterize a given protein strand as an incipient TM segment (26 -28). Therefore, mutant segments within the TM domain of CFTR that fall below this threshold are likely to mitigate against proper membrane insertion and disrupt protein folding, with consequences in vivo such as blocking processing, blocking regulation, altering conductance, and/or reducing synthesis. Based on our results, impaired post-translational folding of CFTR is the primary defect in the case of the L346P mutation.
In contrast, the effects of the R347P mutation present an alternate scenario. Polar residues arising in WT membrane domains, such as Arg-347, are often linked to protein function, including participation in translocation of polar substrates (e.g. chloride transport in CFTR) and/or in stabilizing the tertiary structure of the TM domain through side-chain-side-chain electrostatic interactions (41,42). Thus, the loss of a WT interhelical salt bridge (12) in R347P may contribute to the underlying defect. Note that this salt bridge would similarly be abolished in the L346P/R347I mutant, perhaps explaining, in part, why this "rescue mutant" is not fully functional.
Our overall findings indicate that introduction of Pro-346 or Pro-347 per se did not significantly affect overall CFTR structure/function, as any (new) properties the Pro residue imparts must be weighed against the properties/function(s) attributable to the residue lost. Thus, the loss of the hydrophobic Leu-346 residue concomitant with the introduction of the hydrophilic Pro residue drops TM6 below its threshold hydrophobicity with direct impact on the biogenesis of CFTR. However, loss of an Arg residue at the adjacent position appears to induce a more downstream event, viz., R347P may promote changes in the selectivity or effectiveness of the channel pore of CFTR stemming from loss of side-chain positive character (43)(44)(45)(46), rather than preventing post-translational folding. Our work provides insight into the diversity of local phenomena that can produce dysfunctional forms of CFTR and of membrane proteins generally and suggests that similar circumstances will contribute to molecular defects that underlie human diseases.