Reduced PDZ Interactions of Rescued ΔF508CFTR Increases Its Cell Surface Mobility*

Background: Phenylalanine 508 deletion from CFTR (ΔF508CFTR) is the most common cause of CF. Results: Diffusional mobility of surface ΔF508CFTR is greater than that of wild type CFTR, irrespective of ΔF508CFTR rescue mechanism. Conclusion: Defective ΔF508CFTR-PDZ interactions result in greater ΔF508CFTR diffusion and could contribute to its enhanced internalization. Significance: Restoring ΔF508CFTR-scaffold interaction may have therapeutic benefit. Deletion of phenylalanine 508 (ΔF508) in the cystic fibrosis transmembrane conductance regulator (CFTR) plasma membrane chloride channel is the most common cause of cystic fibrosis (CF). Though several maneuvers can rescue endoplasmic reticulum-retained ΔF508CFTR and promote its trafficking to the plasma membrane, rescued ΔF508CFTR remains susceptible to quality control mechanisms that lead to accelerated endocytosis, ubiquitination, and lysosomal degradation. To investigate the role of scaffold protein interactions in rescued ΔF508CFTR surface instability, the plasma membrane mobility of ΔF508CFTR was measured in live cells by quantum dot single particle tracking. Following rescue by low temperature, chemical correctors, thapsigargin, or overexpression of GRASP55, ΔF508CFTR diffusion was more rapid than that of wild-type CFTR because of reduced interactions with PDZ domain-containing scaffold proteins. Knock-down of the plasma membrane quality control proteins CHIP and Hsc70 partially restored ΔF508CFTR-scaffold association. Quantitative comparisons of CFTR cell surface diffusion and endocytosis kinetics suggested an association between reduced scaffold binding and CFTR internalization. Our surface diffusion measurements in live cells indicate defective scaffold interactions of rescued ΔF508CFTR at the cell surface, which may contribute to its defective peripheral processing.

large regions (Ͼ50 amino acids) did increase CFTR turnover. In the Golgi, modulation of the function or abundance of PDZ-domain containing proteins has been shown to correct ⌬F508CFTR trafficking (10,22,23). Similarly, EBP50 has been implicated in r⌬F508CFTR biosynthetic processing. Knockdown of EBP50 in immortalized airway epithelial cells has been reported to reduce the surface stability of both wtCFTR and low-temperature r⌬F508CFTR (24). In contrast, overexpression of EBP50 was found to rescue ⌬F508CFTR in CFBE41ocells by a mechanism that involved cytoskeletal reorganization through EBP50-RhoA-Rho kinase-ezrin interactions (25,26).
Using single particle tracking (SPT) in live cells, we previously demonstrated that PDZ interactions mediate the formation of a CFTR-containing macromolecular complex at the cell surface (Fig. 1A, top and Refs. 27,28). SPT offers advantages over ensemble-averaged biophysical approaches used to study protein interactions in live cells. SPT provides information simultaneous on many cell surface proteins at exogenous expression levels with nanometer spatial and millisecond temporal resolution, allowing quantification of complex diffusive behaviors of protein subpopulations (29). In comparison to biochemical approaches involving cell disruption and detergent solubilization, SPT can detect weak or transient interactions in live cells.
In this study, we used SPT to test the hypothesis that ⌬F508CFTR-PDZ protein interactions are reduced compared with those of wtCFTR. We found increased cell surface mobility of r⌬F508CFTR compared with that of wtCFTR, which were found to involve reduced PDZ-interactions and elements of the peripheral quality control machinery. Disruption of CFTR-PDZ interactions enhanced endocytosis of wild type and ⌬F508CFTR, suggesting a possible role of reduced PDZ interactions in the accelerated internalization of ⌬F508CFTR.

EXPERIMENTAL PROCEDURES
Antibodies, siRNA, Additional Reagents, and cDNA Constructs-Antibodies and siGENOME SMARTpool siRNA were used as previously described (15). VX-809 was purchased from Selleckchem. CoPo-22 (13) was provided by the UCSF Cystic Fibrosis Foundation Research and Development Program. Plasmid expression vectors encoding human wt-or ⌬F508CFTR with triplet HA tags have been described and characterized previously (11,15,27). Additional CFTR constructs containing an external triplet FLAG tag were generated in pcDNA3.1 using engineered human CFTR cDNA (30); constructs were characterized functionally using a halide sensitive YFP assay as previously described (30). The C-terminal domain of CFTR with (residues 1432-1480) and without the PDZ-binding domain (residues 1432-1476) were amplified by PCR and ligated into pmCherry-C1. Plasmids encoding Myc-tagged human CHIP and CHIP-K30A were previously described (15). GFP-GRASP55 encoding plasmid was purchased from OriGene.
Cell Culture, Transfection, and Rescue of ⌬F508-CFTR Processing-HeLa cell lines stably expressing 3HA-tagged CFTR constructs were previously described (15). An MDCK type II cell line expressing 3HA-tagged wtCFTR was previously described (27), and ⌬F508CFTR-expressing cells were gener-ated in an analogous manner. Untransfeceted COS-7 and MDCK cells were obtained from the UCSF Cell Culture Facility. For all microscopy, cells were cultured on 18-mm coverslips in 12-well cell culture plates for 2-4 days prior to experiments. Cells were transiently transfected using Lipofectamine 2000 (Invitrogen) 2-3 days prior to experiments. Low temperature rescue of ⌬F508CFTR was performed by incubating cells at 27°C for 24 -30 h; as necessary, subsequent thermal unfolding was performed by incubation of cells for an additional 1-2 h at 37°C (15). The following pharmacological maneuvers were used to correct ⌬F508CFTR trafficking: 3 M VX-809 for 24 h (12), 10 M CoPo-22 for 24 h (13), and 1 M thapsigargin for 2 h (10).
Single Particle Tracking and Data Analysis-Qdot-labeling, SPT data collection, and analysis was performed essentially as described in prior studies (27,28,31). Data were collected with a Nikon 100ϫ/N.A 1.3 objective. Datasets were compiled from 110 -788 trajectories (average of 215 trajectories per maneuver) obtained from multiple cells on different days. Given that CFTR is endocytosed, experiments were performed within 3-5 min of labeling to reduce errors associated with inclusion of data derived from endocytosed CFTR molecules (which is assumed to be Ͻ10% in all datasets). Statistical analysis of SPT data were performed using the Kolmogorov-Smirnov test (31,32).
Determination of CFTR Surface Density, Stability, and Endocytosis Rates-Epitope-tagged CFTR surface density and endocytic rates were determined essentially as described in a prior study (15). Cell surface density of epitope-tagged CFTR variants was measured by labeling surface CFTR at 4°C with primary antibody (anti-HA or anti-FLAG) followed by: (i) a secondary antibody conjugated to fluorophore for image-based detection (after fixation), or (ii) horseradish peroxidase conjugated secondary antibody and Amplex Red (Invitrogen) for quantification in an ELISA format. Relative ⌬F508CFTR surface density presented in Fig. 4 is the normalized difference in surface density before and after incubation at 37°C for 2 h. To determine endocytic rates for CFTR variants, cells were labeled with primary antibody and incubated at 37°C for 0 or 10 min prior to labeling with secondary antibody to determine CFTR surface density. Nonspecific antibody labeling was assessed using non-transfected cells. Control experiments in HeLa cells expressing 3HA-tagged CFTR verified that imaging and Amplex Red-based analysis gave equivalent results. Datasets are presented as mean Ϯ S.E. from 3-8 independent experiments. For Amplex Red, individual measurements were performed using 6 -8 parallel samples and normalized to protein content; for image-based analysis, cell surface CFTR density measurements were performed on Ͼ30 cells per data set. Statistical analysis was by t test.

RESULTS
Increased Cell Surface Diffusion of r⌬F508CFTR Compared with wtCFTR-Initial studies were done using HeLa cells expressing CFTR constructs containing triplet hemagglutinin tags (CFTR3HA) engineered into the fourth external loop region; these cells were used previously to characterize and identify cellular components involved in r⌬F508CFTR processing (15). For surface mobility measurements using SPT, cell surface CFTR was selectively labeled by sequential incubation with low concentrations of anti-HA antibody, secondary biotin-conjugated Fab fragment, and streptavidin-conjugated Qdots (emission 655 nm). As in prior studies (27,31), Qdot labeling specificity was confirmed by absence of Qdot labeling when primary antibody or secondary Fab fragment were omitted, or when ⌬F508CFTR was not rescued (data not shown). Diffusion of Qdot-labeled CFTR was measured at 37°C by analysis of image sequences acquired at 30 Hz over 10 s (Fig. 1B). Mobility of wtCFTR was highly confined (Fig. 1B, top), as found previously for wtCFTR expressed at endogenous levels in multiple cell types, including human airway epithelial cells (27,28). Immediately following low temperature rescue, where r⌬F508CFTR plasma membrane stability, endocytosis and ubiquitination are similar to that of wtCFTR (15), the diffusion (measured at 37°C) of r⌬F508CFTR (r⌬F508CFTR 27 ) was also similar to that of wtCFTR (Fig. 1B, middle). However, thermal unfolding of r⌬F508CFTR by 2 h of incubation at 37°C (r⌬F508CFTR 37 ), which causes accelerated endocytosis (ϳ2fold), enhanced ubiquitination and an overall reduction in cells surface r⌬F508CFTR levels (15), produced an increase in mobility (Fig. 1B, bottom). Trajectories generated from analysis of multiple image sequences ( Fig. 1, B, right and D) were summarized as cumulative probability distributions of particle range at 3 s. Fig. 1C shows significantly greater range for r⌬F508CFTR 37 diffusion (blue) in comparison to wtCFTR (black) or r⌬F508CFTR 27 (red, see Fig. 1D for representative trajectories). Diffusion of wtCFTR was not altered under the same culture conditions used to rescue and thermally unfold ⌬F508CFTR (supplemental Fig. S1A).
We previously established that confined diffusion of wtCFTR in live cells is dependent on its C terminus PDZ binding motif, PDZ domain containing scaffold proteins, and coupling to the actin cytoskeleton via ezrin (27). This finding is consistent with in vitro protein-protein interaction data suggesting the involvement of wtCFTR in a cell surface macromolecular complex. Here, to investigate the involvement of PDZ interactions in r⌬F508CFTR confinement in live cells, a chimeric protein consisting of the C terminus of CFTR (residues 1432-1480, which does not contain the endocytic motifs 1413 FLVI, 1424 YDSI, and 1430 LL) fused to mCherry (mCh-CFTR-C) was expressed to compete with CFTR for attachment to PDZ domain proteins. As expected from our previous data (27), mCh-CFTR-C expression greatly increased wtCFTR diffusion (Fig. 1, C and D, gray). Low-temperature corrected ⌬F508CFTR has been demonstrated to interact with EBP50 (24) and mCh-CFTR-C also increased the diffusion of r⌬F508CFTR 27 (Fig. 1, C and D, orange) and r⌬F508CFTR 37 (Fig. 1, C and D, green) to the same degree as wtCFTR, suggesting that PDZ interactions contribute to the confinement of cell surface ⌬F508CFTR. As previously FIGURE 1. Single-particle tracking of wild-type and low-temperature rescued (r)⌬F508CFTR in HeLa cells. A, C-terminal PDZ interactions tether wtCFTR to actin (top) but their role in r⌬F508CFTR plasma membrane stability is unknown (bottom). B, individual frames from SPT image sequences acquired at 30 fps for Qdot-labeled wtCFTR (top), r⌬F508CFTR immediately after low temperature correction (r⌬F508 27 , middle) and r⌬F508CFTR after low temperature correction and subsequent incubation at 37°C for 2 h (r⌬F508 37 , bottom). Image acquisition time is given in each frame and highlighted exemplar trajectories are shown for each condition (right, 200 nm scale bar refers to all trajectories). Images are 6.4 ϫ 6.4 m. C, cumulative distribution plots for wtCFTR (black), r⌬F508 27 (red), and r⌬F508 37 (blue) under control conditions, and for wtCFTR (gray), r⌬F508 27 (orange), and r⌬F508 37 (green) when expressed with high levels of the CFTR C-terminal region (fused to mCherry, mCh-CFTR-C), which results in mobilization of CFTR. Statistically significant differences in populations are shown (*, p Ͻ 0.001). N/⌺N is the normalized cumulative probability distribution where 1 represents the complete population of SPT data. SPT data were derived from 110 -316 trajectories. D, color-coded SPT trajectories for each condition shown in Fig. 1C (500 nm scale bar refers to all trajectories). demonstrated for wtCFTR (27), expression of a CFTR C-terminal fusion protein that lacked the PDZ-binding motif (mCh-CFTR-C-⌬PDZ) did not alter the diffusion of wtCFTR, ⌬F508CFTR 27 , and ⌬F508CFTR 37 in Hela cells (supplemental Fig. S1B). These findings show in live cells that interactions with PDZ domain-containing proteins contribute to the tethering of wt and low-temperature rescued ⌬F508CFTR, but that interactions between r⌬F508CFTR 37 and PDZ domain-containing proteins are reduced compared with wtCFTR or r⌬F508CFTR 27 .
CFTR diffusion was measured in additional cell types and with transiently transfected CFTR constructs to verify the robustness of observations made in HeLa cell lines. In MDCK cell lines stably expressing CFTR3HA constructs, incubation of ⌬F508CFTR-expressing cells at 27°C increased ⌬F508CFTR cell surface density to a level ϳ6-fold reduced compared with wtCFTR (supplemental Fig. S1C). As predicted from prior studies in different cell types (14,15), incubation at 37°C for 1 h to thermally denature r⌬F508CFTR produced an ϳ50% decrease in plasma membrane r⌬F508CFTR density ( Fig. 2A, left) and an ϳ2-fold increase in endocytosis compared with wtCFTR; however, ⌬F508CFTR internalization was similar to wtCFTR immediately after rescue ( Fig. 2A, right).
SPT analysis of CFTR diffusion in MDCK cell lines showed highly confined diffusion of wtCFTR (black) and r⌬F508CFTR 27 (red, Fig. 2B); however, incubation for 1 h at 37°C increased r⌬F508CFTR diffusion as seen in individual trajectories (blue) and cumulative probability distributions (Fig. 2C, black and blue). The diffusion of wtCFTR and r⌬F508CFTR 37 were significantly increased by expression of mCh-CFTR-C (Fig. 2C, gray and green). A comparable increase in channel diffusion was seen with deletion of the C-terminal PDZ-binding motif of CFTR (CFTR⌬PDZ, Fig. 2C, pink), suggesting that expression of mCh-CFTR-C mobilizes wt and r⌬F508CFTR to the same degree as deletion of the PDZ-binding motif. In control experiments, mCh-CFTR-C-⌬PDZ did not alter the diffusion of wtCFTR (Fig. 5C), r⌬F508CFTR 27 or r⌬F508CFTR 37 (supplemental Fig. S1D), verifying the involvement of PDZ-domain interactions in confinement of wtCFTR, r⌬F508CFTR 27 , and r⌬F508CFTR 37 . The smooth shape of cumulative probability distributions suggested the absence of distinct CFTR subpopulations with different mobilities. As such, data were summarized in bar graph format (Fig. 2D), showing median (blue) and inter-quartile (25 th to 75 th percentiles, red) range values. As described above, data from MDCK cell lines (Fig. 2D) shows significantly different diffusion of r⌬F508CFTR 37 versus wtCFTR or r⌬F508CFTR 27 (bars 1-3), and increased CFTR diffusion upon disruption of PDZ interactions (bars 3-6).
CFTR diffusion was further studied in COS-7 fibroblasts that were transiently transfected to express CFTR constructs containing external FLAG epitopes (CFTR3FLAG). As expected, surface density decreased and endocytosis increased after 37°C incubation of low temperature corrected ⌬F508CFTR (Fig. 2E). The diffusion of wtCFTR and r⌬F508CFTR 27 were similar in COS-7 cells, whereas incubation of r⌬F508CFTR-expressing cells at 37°C increased diffusion by ϳ30% (Fig. 2F, bars 1-3). Deletion of the PDZ binding motif from wtCFTR increased diffusion (Fig. 2F, bar 4), as did deletion from ⌬F508CFTR following low temperature rescue (⌬F⌬PDZ 27 ) and subsequently incubated at 37°C (⌬F⌬PDZ 37 ) (Fig. 2F, bars 5 and 6). The magnitude of the effect on diffusion of PDZ-domain deletion varied for wtCFTR and r⌬F508CFTR in different cell types; however, this difference likely reflects differences in plasma membrane properties as MDCK cells have been shown to reduce the diffusion of inert probe molecules as compared with other cell types (33). Further, the diffusion of wt and ⌬F508CFTR in the absence of PDZ interactions in both MDCK and COS-7 cells was comparable to that of AQP1, a membrane transport protein that has largely unhindered diffusion (33). These studies confirm that in different cell types ⌬F508CFTR rescued by low-temperature incubation interacts with PDZ surface density (left) and endocytic rates (right) in MDCK cell lines immediately after rescue (red) and after 1 h 37°C incubation (blue). B, example trajectories of wtCFTR (black), r⌬F508 27 , (red), and r⌬F508 37 , (blue) diffusion in MDCK cell lines. C, cumulative distribution curves for MDCK cell lines expressing wtCFTR (black), r⌬F508 37 (blue), or CFTR⌬PDZ (pink) and co-expressing mCh-CFTR-C with wtCFTR (gray) or r⌬F508 37 , (green). D, summary graph representing median (blue lines) and interquartile (red lines) range values for the diffusion of wtCFTR, r⌬F508CFTR and CFTR⌬PDZ-expressing MDCK cell lines. As indicated, mCh-CFTR-C was also expressed (bars 4 and 5). E, surface levels (left) and endocytic rates (right) of low temperature corrected 3FLAG-tagged ⌬F508CFTR in transiently transfected COS-7 cells immediately after rescue (red) and after 37°C incubation (blue). F, summary graphs of diffusive range for 3FLAG-tagged wtCFTR, ⌬F508CFTR, CFTR⌬PDZ, and ⌬F508CFTR⌬PDZ expressing COS-7 cells. In panels D and F, statistically significant differences in populations are shown (*, p Ͻ 0.001). SPT data for MDCK cells was derived from 113-788 trajectories and for COS-7 cells from 140 -252 trajectories. Data for CFTR surface density and endocytic rates are mean Ϯ S.E. for 3-8 independent experiments.

Surface Mobility of Rescued ⌬F508CFTR
DECEMBER 21, 2012 • VOLUME 287 • NUMBER 52 domain proteins in a similar manner as wtCFTR, but that thermal unfolding at 37°C increases r⌬F508CFTR diffusion by partially disrupting the interactions between r⌬F508CFTR and PDZ domain-containing scaffold proteins.
⌬F508CFTR Diffusion Is Increased with Alternative Correction Methods-SPT experiments were done to investigate cell surface ⌬F508CFTR-PDZ interactions following different correction maneuvers. Treatment of MDCK cell lines expressing CFTR3HA with 3 M VX-809, a corrector drug in Phase II clinical trials (12,34), for 24 h gave ⌬F508CFTR cell surface levels similar to those produced by low-temperature rescue (supplemental Fig. S1C). SPT trajectories of ⌬F508CFTR diffusion following VX-809 correction appeared less confined than those of wtCFTR under control conditions (Fig. 3A, red and  black trajectories) and range values differed significantly (Fig.  3B, bars 1 and 3). Expression of mCh-CFTR-C increased the diffusion of VX-809-corrected ⌬F508CFTR (Fig. 3A, blue) to that found for wtCFTR with mCh-CFTR-C or for CFTR⌬PDZ (Fig. 3B, bars 4 -6), confirming the involvement of PDZ interactions in restricting r⌬F508CFTR mobility after VX-809 treatment. Control experiments verified that mCh-CFTR-C-⌬PDZ did not alter the diffusion of r⌬F508CFTR corrected with VX-809 (data not shown).
Additional experiments were performed in transiently transfected MDCK cells using VX-809 to correct 3FLAG-tagged ⌬F508CFTR to verify results generated in cell lines (Fig. 3, C  and D). In transiently transfected MDCK cells, wtCFTR mobility was reduced when compared with VX-809 rescued ⌬F508CFTR (Fig. 3C, black and red; and Fig. 3D, bars 1 and 2). The diffusion of cell surface ⌬F508CFTR was increased by expression of mCh-CFTR-C (but not mCh-CFTR-C-⌬PDZ) or deletion of the ⌬F508CFTR C-terminal PDZ-binding motif (r⌬F⌬PDZ) (Fig. 3, C and D and data not shown). VX-809 treatment did not alter the diffusion of wtCFTR or CFTR⌬PDZ, suggesting that VX-809 does not perturb plasma membrane properties (for example see Fig. 3B, bars 1 and 2; Fig. 3D, bars 5 and 6).
The diffusion of cell surface ⌬F508CFTR in MDCK cell lines was also increased compared with wtCFTR ( Fig. 3E) following correction by thaspigargin (black), the small molecule corrector CoPo-22 (red), or transient expression of GFP-GRASP55 (blue; for all datasets p Ͻ 0.001 compared with wtCFTR, Fig. 3F,  top). These treatments did not alter the diffusion of wtCFTR (Fig. 3E, left and Fig. 3F, bottom). For GRASP55-mediated ⌬F508CFTR rescue, immunocytochemistry confirmed GFP-GRASP55 co-localization with the Golgi marker giantin (Fig.  3G). Photobleaching experiments using laser scanning confocal microscopy indicated that the diffusion of the inert membrane probe glycophosphatidylinositol-anchored GFP was not altered by each of the maneuvers used to correct ⌬F508CFTR trafficking (data not shown). These studies thus demonstrate that, as found for low temperature correction, rescue by a variety of unrelated maneuvers produces cell surface ⌬F508CFTR with abnormal PDZ-domain interactions relative to wtCFTR.
Components of the Peripheral Protein Quality Control System Modulate ⌬F508CFTR Mobility-Experiments were performed to identify cellular components that are responsible for reduced ⌬F508CFTR-PDZ interactions. Using a small interfer-ing (si)-RNA screen, Okiyoneda et al. (15) previously identified constituents of the peripheral protein quality control (PPQC) system that recognize and eliminate low-temperature corrected and thermally denatured ⌬F508CFTR from the cell surface. This approach identified the E3 Ub ligase CHIP (C terminus of Hsc70-interacting protein) and the chaperone Hsc70 (heat shock cognate of 70 kDa) as key determinants of r⌬F508CFTR 37 turnover such that knockdown of either protein increased cell surface stability. Here, we tested whether altering the abundance or activity of CHIP and Hsc70 would also affect ⌬F508CFTR mobility. r⌬F508CFTR 37 diffusion was  (bar 1, top). G, GFP-GRASP55 is Golgi-localized by immunocytochemical analysis (scale bar, 10 m). Scale bar in A (500 nm) refers to all trajectories. In panels B and D, statistically significant differences in populations are shown (*, p Ͻ 0.001). SPT data for MDCK cell lines was derived from 140 -788 trajectories and for transiently transfected MDCK cells from 131-298 trajectories. studied in HeLa cells following siRNA reduction of CHIP or Hsc70 expression by Ͼ75% (Fig. 4A, left top). In agreement with prior studies (15), knockdown of CHIP or Hsc70 increased cell surface stability of r⌬F508CFTR by ϳ50% during subsequent 37°C incubation (Fig. 4A, left bottom). Increasing r⌬F508CFTR 37 cell surface density by CHIP or Hsc70 knockdown was also accompanied by a reduction in r⌬F508CFTR 37 diffusion toward that of wtCFTR (Fig. 4A, right).
To further investigate the role of CHIP in ⌬F508CFTR surface diffusion, wild type, and mutant CHIP constructs were transiently expressed in COS-7 cells. As expected based on earlier observations (15), overexpression of CHIP in COS-7 cells reduced r⌬F508CFTR 37 surface stability (Fig. 4B, bottom) and also increased r⌬F508CFTR 37 surface diffusion (Fig. 4B, top). CHIP binds to Hsc70 and Hsp90 (heat shock protein of 90 kDa) via an N-terminal tetratricopeptide repeat domain, and these associations are blocked in the CHIP-K30A mutant (35). Expression of CHIP-K30A in COS-7 cells did not alter r⌬F508CFTR 37 diffusion or stability (Fig. 4B), indicating that CHIP recognition of thermally destabilized r⌬F508CFTR is dependent upon Hsc70 and Hsp90. In VX-809-treated HeLa cells r⌬F508CFTR diffusion was similar to that of r⌬F508CFTR 37 , as would be expected from prior studies in MDCK cells (Fig. 4C, top left). Further, siRNA knockdown of CHIP reduced r⌬F508CFTR mobility and increased surface stability (Fig. 4C, left). CHIP expression also reduced confinement and cell surface stability of r⌬F508CFTR in transiently transfected MDCK cells treated with VX-809; however, expression of CHIP-K30A did not alter diffusion or stability of VX-809-corrected r⌬F508CFTR, confirming that CHIP recognition of r⌬F508CFTR required Hsc70 and Hsp90 (Fig. 4C,  right). In control experiments, wtCFTR mobility was not altered by CHIP knockdown in HeLa cells or CHIP overexpression in COS-7 cells; non-targeted siRNA did not alter wtCFTR or r⌬F508CFTR diffusion in HeLa cells (supplemental Fig.  S1E). These studies indicate that components of the PPQC machinery including CHIP, Hsc70 and Hsp90, contribute to the increased diffusion of cell surface ⌬F508CFTR rescued by lowtemperature or VX-809.
PDZ Interactions Stabilize wtCFTR at the Cell Surface-The present study indicates that reduced r⌬F508CFTR surface stability is associated with increased diffusive mobility. Given that endocytosis is the initial step in reducing r⌬F508CFTR surface density, an experimental strategy was developed to quantitatively investigate whether the observed increase in diffusion associated with reduced PDZ interactions of ⌬F508CFTR could alter CFTR internalization rates. Our approach relied upon exogenous expression of the CFTR C terminus (mCh-CFTR-C) to increase wtCFTR mobility and thus mimic r⌬F508CFTR surface dynamics (Fig. 5A). CFTR3HA-expressing MDCK cells were enriched for mCh-CFTR-C content by fluorescence-activated cell sorting (FACS) to generate a cell population with ϳ80% mCh-CFTR-C-positive cells that were readily identified in image-based assays of CFTR trafficking and surface stability (Fig. 5B, left). Expression of mCh-CFTR-C in FACS-enriched cells was relatively homogeneous (Fig. 5B, right), albeit ϳ40fold less than in prior experiments that used transient transfection of the same construct. Surface expression of wtCFTR was similar in control and mCh-CFTR-C-enriched cells, suggesting that expression of exogenous CFTR C-terminal domains does not alter CFTR stability (supplemental Fig. S1C). This result is consistent with data from CFTR with short C-terminal deletions that show similar stability to wtCFTR (19). SPT confirmed that wtCFTR diffusion in mCh-CFTR-C-enriched cells was greater (by ϳ25%) than that in control cells (Fig. 5C, bars 1 and  3), and similar to that of ⌬F508CFTR rescued by low-temperature or VX-809 (Fig. 5C, bars 4 and 5).
Endocytosis of wtCFTR in mCh-CFTR-C-enriched cells was increased by ϳ30% as compared with wtCFTR (Fig. 5D, bar 1  and 3, p Ͻ 0.005, ϳ16 versus 21%), suggesting that PDZ interactions stabilize CFTR at the cell surface over time scales of minutes. In support of this conclusion, deletion of the CFTR C-terminal PDZ binding domain (CFTR⌬PDZ), which is expected to fully account for any PDZ-dependent component of CFTR internalization (and increased diffusive by ϳ50%, Fig.  2), increased endocytosis by ϳ60% as compared with wtCFTR ( Fig. 5D, bar 4, p Ͻ 0.001, ϳ16 versus 26%). Therefore, compared with wtCFTR, the percentage increase in surface diffusion and in endocytosis rate after deletion of the CFTR PDZbinding domain or upon partial disruption of PDZ interactions to model r⌬F508CFTR diffusion are in reasonable agreement. Control experiments confirmed that CFTR diffusion (Fig. 5C, bar 2) and internalization (Fig. 5D, bar 2) were not altered by enrichment of cells for expression of the PDZ binding-domain deleted CFTR C terminus (mCh-CFTR-C-⌬PDZ). Finally, in MDCK cells transiently expressing VX-809-rescued ⌬F508CFTR3FLAG, co-expression of mCh-CFTR-C at high levels enhanced ⌬F508CFTR internalization, confirming that PDZ interactions can also influence r⌬F508CFTR endocytosis (Fig. 5E). These studies indicate that (i) PDZ interactions stabilize CFTR at the cell surface over time scales of minutes, and (ii) increasing the diffusion of wtCFTR to that of r⌬F508CFTR is sufficient to alter internalization kinetics.

DISCUSSION
SPT provides definitive information about the motions of individual channels, receptors or lipids in the plasma membrane of live cells, which can reveal complex diffusive behaviors over extended time and distance scales (29,36). Our prior SPT studies on wtCFTR showed that it integrates into an actin-associated, macromolecular complex that includes PDZ domaincontaining proteins and ezrin (27,28). Here, SPT analysis of the dynamics and interactions of rescued ⌬F508CFTR revealed novel properties relative to wtCFTR that were related to the peripheral quality control mechanism and, potentially, ⌬F508CFTR internalization dynamics.
In live cells at physiological temperature, we found consistently greater cell surface mobility of r⌬F508CFTR (by ϳ30%) compared with wtCFTR, regardless of correction mechanism, which was due to distinct association with PDZ scaffold proteins. Incubation of low temperature-corrected ⌬F508CFTR at 37°C causes increased ubiquitination and endocytosis, and consequent reduced plasma membrane stability (15,37). Though r⌬F508CFTR-PDZ interactions were similar to those of wtCFTR just after low-temperature correction, incubation at 37°C increased r⌬F508CFTR diffusion and reduced PDZ binding. Unfolded ⌬F508CFTR is eliminated from the cell surface by the PPQC system. We found that modulation of the PPQC system also altered the diffusion of r⌬F508CFTR, suggesting an interdependence of r⌬F508CFTR surface stability and mobility.
The role of CFTR-PDZ interactions is incompletely understood. PDZ interactions were initially reported to mediate CFTR polarization, but subsequent contradictory evidence was reported (18,19). Inconsistent data has also been reported on the role of PDZ interactions in CFTR turnover (19 -21). Several studies have concluded that multivalent PDZ scaffolds facilitate compartmentalization of proteins to regulate CFTR activity (17,38). Our experimental approach provided evidence in live cells for involvement of PDZ interactions in CFTR complexation and stabilization at the cell surface over time scales of minutes. This conclusion is based on experimental approaches that either deleted the PDZ-binding domain from CFTR or used expression of an exogenous peptide (mCh-CFTR-C) to com- pete with CFTR for PDZ domain-containing scaffold proteins. Previously, knockdown of EBP50 was shown to reduce surface stability of wtCFTR and low-temperature r⌬F508CFTR over hours (24). However, interpretation of this study is difficult because EBP50 is a multi-domain protein that interacts with distinct PDZ-binding domain proteins and the actin cytoskeleton. Further, altering EBP50 expression profoundly alters the actin cytoskeleton (26).
In experiments designed to quantitatively relate CFTR diffusion to internalization dynamics in MDCK cells, deletion of the CFTR PDZ-binding domain (ϳ50% increase in diffusion compared with wtCFTR) enhanced endocytosis by ϳ60% whereas partial disruption of PDZ interactions to mimic r⌬F508CFTR diffusion (ϳ30% increase in diffusion) increased internalization by ϳ30%. Endocytosis of ⌬F508CFTR in MDCK cells after lowtemperature correction and thermal unfolding, or VX-809 treatment, were ϳ1.8-fold greater than that of wtCFTR. Therefore, the proportion of r⌬F508CFTR internalization that relates to increased r⌬F508CFTR diffusion (Fig. 5D, blue dashed line) could be ϳ50%. Although reduced scaffold binding and consequent accelerated endocytosis would predispose r⌬F508CFTR to lysosomal trafficking and degradation, the overall elimination of r⌬F508CFTR is ultimately dependent on channel ubiquitination (14,15). Indeed, there are several disease-causing CFTRs with C terminus deletions yet relatively mild disease, which has suggested that severe disease, as found for the ⌬F508 mutation, requires CFTR structural destabilization and consequent ubiquitination in addition to enhanced endocytosis (14,15,19,21).
A limited number of prior studies have demonstrated that PDZ interactions stabilize proteins at the cell surface by reducing endocytic uptake. For instance, stimulated G-protein coupled receptors that contain PDZ-binding domains segregate into specific clathrin-coated pits to greatly extend their surface lifetime through actin association (39). Similar to the present study, a correlative analysis of protein diffusion and internalization has been performed for the neuronal ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPAR) (40). The AMPAR forms a complex with Stargazin and the PDZ scaffold PSD-95, which result in receptor immobilization at the neuronal postsynaptic density. Disruption of Stargazin-PSD-95 interactions with a cell permeable peptide mimetic of the Stargazin PDZ binding domain increased AMPAR diffusive range by ϳ1.3-fold and endocytosis by ϳ4-fold (40). For CFTR, Thelin et al. (41) reported that the S13F mutation increased CFTR diffusion coefficient by only ϳ20%, but increased internalization by ϳ2-fold. However, in that study CFTR was reported as mobile under control conditions, presumably due CFTR expression at super-physiological levels (27), making interpretation of these findings difficult.
Components of the PPQC network, including CHIP, Hsc70 and Hsp90, were found here to alter the plasma membrane dynamics of r⌬F508CFTR after low temperature or chemical correction. Although the mechanism responsible for PPQC modulation of r⌬F508CFTR diffusion was not established, we speculate that r⌬F508CFTR-PDZ interactions might be reduced by ubiquitination, or perhaps sterically hindered by r⌬F508CFTR-CHIP/Hsp90/Hsc70 complex-ation. Further work is required to address these hypotheses. CHIP and associated quality control components have been shown to mediate ubiquination of ⌬F508CFTR at the cell surface and the ER, and inhibition of CHIP causes ⌬F508CFTR accumulation in the ER (15,42,43). Functional inhibition of CHIP or a CHIP-binding partner thus represents a potential therapeutic approach to stabilize r⌬F508CFTR at the cell surface.