Diffusional Mobility of the Cystic Fibrosis Transmembrane Conductance Regulator Mutant, ΔF508-CFTR, in the Endoplasmic Reticulum Measured by Photobleaching of GFP-CFTR Chimeras*

Mutations in the cystic fibrosis transmembrane conductance regulator protein (CFTR) cause cystic fibrosis. The most common disease-causing mutation, ΔF508, is retained in the endoplasmic reticulum (ER) and is unable to function as a plasma membrane chloride channel. To investigate whether the ER retention of ΔF508-CFTR is caused by immobilization and/or aggregation, we have measured the diffusional mobility of green fluorescent protein (GFP) chimeras of wild type (wt)-CFTR and ΔF508-CFTR by fluorescence recovery after photobleaching. GFP-labeled ΔF508-CFTR was localized in the ER and wt-CFTR in the plasma membrane and intracellular membranes in transfected COS7 and Chinese hamster ovary K1 cells. Both chimeras localized to the ER after brefeldin A treatment. Spot photobleaching showed that CFTR diffusion (diffusion coefficient ∼10−9cm2/s) was not significantly slowed by the ΔF508 mutation and that nearly all wt-CFTR and ΔF508-CFTR diffused throughout the ER without restriction. Stabilization of molecular chaperone interactions by ATP depletion produced remarkable ΔF508-CFTR immobilization (∼50%) and slowed diffusion (6.5 × 10−10cm2/s) but had little effect on wt-CFTR. Fluorescence depletion experiments revealed that the immobilized ΔF508-CFTR in ATP-depleted cells remained in an ER pattern. The mobility of wt-CFTR and ΔF508-CFTR was reduced by maneuvers that alter CFTR processing or interactions with molecular chaperones, including tunicamycin, geldanamycin, and lactacystin. Photobleaching of the fluorescent ER lipid diOC4(3) showed that neither ER restructuring nor fragmentation during these maneuvers was responsible for the slowing and immobilization of CFTR. These results suggest that (a) the ER retention of ΔF508-CFTR is not due to restricted ER mobility, (b) the majority of ΔF508-CFTR is not aggregated or bound to slowly moving membrane proteins, and (c) ΔF508-CFTR may interact to a greater extent with molecular chaperones than does wt-CFTR.

The cystic fibrosis transmembrane conductance regulator protein (CFTR) 1 mediates transepithelial chloride transport across epithelial cells in the airways, intestine, pancreas, and sweat gland. Some mutations in CFTR cause the lethal genetic disease cystic fibrosis, which produces chronic lung infection, progressively impaired pulmonary function, and pancreatic insufficiency. The most common CFTR mutant, ⌬F508-CFTR, is retained at the endoplasmic reticulum (ER) and consequently is unable to function as a plasma membrane chloride channel (1). Electrophysiological studies indicate that the ER-retained ⌬F508-CFTR is at least partially functional in excised ER membranes (2). Further, growth of ⌬F508-CFTR-expressing cells under non-physiological conditions (low temperature, Ref. 3 or with chemical chaperones, Ref. 4) indicates that under certain conditions ⌬F508-CFTR can be transported to the plasma membrane and restore cell chloride permeability.
CFTR biosynthesis is an inefficient process. Newly synthesized wild type (wt)-CFTR and ⌬F508-CFTR are folded into the ER membrane where they become core-glycosylated (5,6). Only 20 -30% of newly synthesized wt-CFTR is transported to the Golgi where complex glycosylation occurs, with the remaining protein ubiquinated and degraded by the proteasome (7,8). Wild type CFTR exists initially in a protease-susceptible form (t1 ⁄2 ϳ30 min) that becomes more stable upon transport from ER to Golgi (t1 ⁄2 ϳ24 h), whereas ⌬F508-CFTR is rapidly degraded (t1 ⁄2 ϳ30 min) (5,6). Inhibition of proteasome function results in the formation of aggresomes, which are detergent-insoluble, perinuclear protein aggregates common to many degenerative diseases (9 -11).
CFTR folding in the ER appears to be facilitated by interactions with the cellular quality control machinery. Wild type CFTR and ⌬F508-CFTR have been shown to interact with the molecular chaperones Hsc70, Hdj2, Hsp70, Hsp90, and calnexin but not with BiP or Grp94 (12)(13)(14)(15). Recent data suggest that the Hsc70 co-chaperone CHIP (C-terminus of Hsc70-interacting protein) targets immature CFTR for proteasome-mediated degradation (16). Biochemical data indicate that calnexin and Hsp70 bind ⌬F508-CFTR more avidly than wt-CFTR (12,13). The differential processing of ⌬F508-CFTR and wt-CFTR, however, is probably not accounted for by gross structural differences. Proteolytic cleavage experiments have suggested that the conformation of ⌬F508-CFTR is similar to that of the early, relatively unstable form of wt-CFTR (17). In vitro folding studies of the isolated nucleotide binding domain 1 (containing the phenylalanine 508 residue) support the view that the ⌬F508 mutation does not affect structure (18,19) but may affect folding kinetics.
Here we have used fluorescence recovery after photobleaching to measure the diffusional mobility of wt-CFTR and ⌬F508-CFTR at the ER to investigate whether ⌬F508-CFTR immobilization or interaction with chaperones is responsible for its ER retention. Several mechanisms have been proposed to explain the failure of ⌬F508-CFTR to be exported from the ER: immobilization of ⌬F508-CFTR, potentially by association with chaperones or by self-aggregation; efficient ⌬F508-CFTR recycling from the Golgi; and/or failure of the ER export machinery to recognize ⌬F508-CFTR. Photobleaching measurements were done on cells expressing green fluorescent protein (GFP) chimeras of wt-CFTR and ⌬F508-CFTR; a series of maneuvers was used to probe for interactions with CFTR processing machinery. It was shown previously that GFP fusion to the N terminus of CFTR did not effect CFTR localization, processing, or function (9, 20 -22). We used photobleaching previously to quantify aqueous phase rheology in cellular and organellar compartments (23)(24)(25) and to investigate protein-protein interactions of aquaporin (AQP) water channels in the ER and plasma membranes (26,27). We find here that the ER retention of ⌬F508-CFTR is not due to immobilization, aggregation, or binding to slow moving membrane components; however, a greater fraction of ⌬F508-CFTR than wild type CFTR interacted with molecular chaperones after ATP depletion, and inhibition of CFTR processing by a proteasome inhibitor resulted in CFTR aggregation and immobilization.

EXPERIMENTAL PROCEDURES
Cell Culture, Plasmids, Transfection, and Treatments-COS7 (ATCC CRL-1651) and CHO-K1 (ATCC CCL-61) cells (obtained from the University of California, San Francisco Cell Culture Facility) were cultured in DMEH-21 and Ham's F12 media, respectively, supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100 g/ml streptomycin at 37°C in a 5% CO 2 /95% air atmosphere. Cells were grown on 18-mm diameter glass coverslips in 12-well plates. Cells were transiently transfected with plasmid DNA using 6 l of lipofectamine (Invitrogen) and 1 g of plasmid DNA according to the manufacturer's instructions.
cDNA encoding wt-CFTR was fused downstream of the enhanced GFP (CLONTECH) as described previously, and the ⌬F508 mutation was introduced by site-directed mutagenesis (20,22). A glycosylationnull mutant of GFP-wt-CFTR, in which asparagine residues were replaced by glutamines at positions 894 and 900, was also generated by site-directed mutagenesis. For GFP expression in the ER lumen, the GFP sequence was flanked by an N-terminal preprolactin leader sequence and a C-terminal KDEL motif (25). For GFP expression in the cytoplasm, the coding region of pEGFP (CLONTECH) was ligated into pcDNA3.1 as a HindIII/EcoRI fragment.
Photobleaching experiments were performed 2-3 days after transfection for CFTR-transfected cells and 1-2 days after transfection for cytoplasmic or ER-targeted unconjugated GFP. For most experiments GFP-CFTR-transfected cells were treated with brefeldin A (BFA) by 16 -24 h of incubation at 5 g/ml at 37°C in the CO 2 incubator. For ATP depletion, cells were incubated in phosphate-buffered saline containing 6 mM 2-deoxyglucose and 0.02% sodium azide at 37°C for 30 min or 1 h. Geldanamycin (Calbiochem) was used at 0.1 g/ml for 90 min (14) and tunicamycin (Calbiochem) at 5 g/ml for 20 h. Clasto-lactacystin ␤-lactone (the active form of lactacystin, Sigma) was used at 6 M for 1.5, 6, or 14 h. In heat shock experiments, cells were incubated at 42°C for 1 h followed by 16 -20 h recovery at 37°C. Paraformaldehyde (for GFP-CFTR immobilization) was used at 4% for 30 min in phosphate-buffered saline at 23°C. ER membranes in untransfected cells (subjected to various treatments above) were fluorescently labeled with 3,3Ј-dibutyloxacarbocyanine iodide (diOC 4 (3), Molecular Probes) by incubation for 10 min with 10 M dye in phosphate-buffered saline at 37°C. In some experiments transfected cells were identified by their green fluorescence prior to diOC 4 (3) labeling, and those cells were photobleached. The relatively small amount of GFP had little effect on the recovery of the much brighter and more rapidly diffusing diOC 4 (3).
Fluorescence Recovery after Photobleaching-For spot photobleach-ing, an argon ion laser beam (488 nm, Innova 70 -4, Coherent) was modulated by an acousto-optic modulator and the first-order beam was directed onto cells through an oil immersion (Nikon fluor) objective lens (ϫ40, numerical aperture (N.A.) 1.3; ϫ60, N.A. 1.4; or ϫ100, N.A. 1.4) using an inverted epifluorescence microscope. To visualize cells and to select cellular regions to be bleached, full-field epi-illumination was accomplished using the expanded zero-order laser beam directed onto the objective lens by a fiber optic. The beam was reflected onto the objective by a dichroic mirror (510 nm), and emitted fluorescence was filtered by serial 530 Ϯ 15 nm bandpass and 515 nm longpass filters. Emitted fluorescence was detected by a photomultiplier and digitized by a 14-bit analog-to-digital converter. The photomultiplier was transiently gated off during the bleach pulse. Fluorescence was sampled over 200 ms before the bleach pulse, then at rates of up to 1 MHz. For measurement of slow recoveries, an electronic shutter was used to collect fluorescence at 1 Hz, averaging 10,000 acquisitions during the shutter open time of 20 ms.
For photobleaching experiments with image detection, a Leitz upright microscope with a cooled CCD camera detector (Photometrics) was used to record full-field epifluorescence images. An electronically shuttered bleach beam from the argon laser (488 nm) was directed onto the sample (from below) using a Nikon ϫ25 long working-distance air objective. Cells were viewed (from above) using an oil immersion ϫ60 objective (N.A. 1.4) and GFP filter set (HQ filters, Chroma). Software was written (in LabVIEW, National Instruments) to coordinate the bleach pulse, illumination and camera shutters, and image acquisition. Fluorescence loss in photobleaching (FLIP) experiments were performed using the same apparatus in which repeated laser photobleaching was performed with specified delays between bleach pulses.
All photobleaching experiments were done at 37°C. Cultured cells were mounted in custom built perfusion chambers designed to fit a PDMI-2 microincubator (Harvard Apparatus) controlled by a TC-201A temperature regulator (Harvard Apparatus). Objective lens temperature was also maintained at 37°C using a lens thermoregulator (Bioptechs Inc., Butler, PA). For maneuvers on GFP-CFTR-expressing cells, single recovery curves from 5-12 different cells were averaged to generate a single averaged recovery curve, and this was repeated 4 -10 times using different coverslips. For cytoplasmic GFP and ER-targeted GFP-expressing cells, each averaged recovery curve was generated from 10 individual recoveries, each from a different cell. For diOC 4 (3) measurements, averaged recovery curves were generated from the average of 5-8 cells. Results are reported as means Ϯ S.E. for the number of different coverslips, and representative averaged recovery curves from single coverslips are shown in figures for the reader to evaluate curve shape and signal-to-noise ratio. Statistical analysis was performed by analysis of variance against control conditions, and values of p Ͻ 0.05 were considered significant.
Analysis of Photobleaching Recovery Data-Spot photobleaching recovery curves were analyzed as described in Levin et al. (27). Briefly, recovery half-times (t1 ⁄2 ), the time for fluorescence to recover by 50% due to diffusion, were determined from fluorescence recovery curves, F(t), by non-linear regression. Fluorescence recovery curves were taken to be a combination of reversible (as determined from fixed cell analysis) and irreversible (due to diffusion) processes, F(t) ϭ a 1 (F(t) reversible ϩ a 2 (F(t) irreversible ) where a 1 ϩ a 2 ϭ 1. The diffusion-mediated recovery was fitted using the semiempirical equation: where F 0 is prebleach fluorescence, F inf is fluorescence at infinite time, and R is the fractional fluorescence recovery. Absolute diffusion coefficients were calculated from t1 ⁄2 using a mathematical model of ER diffusion (25,28).

RESULTS
The diffusion of wt-CFTR and ⌬F508-CFTR in the ER of living cells was determined using GFP as a fluorescent reporter (Fig. 1A). Transfection of COS7 and CHO cells with the GFPwt-CFTR chimera showed GFP localization to the plasma membrane and intracellular membranes (including Golgi and ER) but exclusively to the ER after BFA treatment (Fig. 1B). The GFP-⌬F508-CFTR chimera was seen in an ER-like pattern without or after BFA treatment (Fig. 1C). Similar patterns of ER staining were observed in COS7 and CHO cells transfected with ER-localized (aqueous phase) GFP or stained with the fluorescent ER lipid probe diOC 4 Photobleaching experiments with image detection were done to evaluate the gross mobility of GFP-CFTR chimeras in BFA-treated COS7 ( Fig. 2A) and CHO (Fig. 2B) cells. A large region of the cell, ϳ5 m in diameter (marked by cross-hairs), was bleached by the laser pulse, and fluorescence recovery was recorded by serial imaging. In each case a darkened region is seen immediately postbleach (t ϭ 0 s). Recovery into this bleached region is diffusive in nature from the edges of the bleach spot inward and appears to be nearly complete. The kinetics of fluorescence recovery were qualitatively similar for cells expressing wt-CFTR and ⌬F508-CFTR and in both cell types. In each case the fluorescence recovery was Ͼ 80% complete as determined from the ratio of fluorescence intensity in the bleach spot to the whole cell immediately after versus 4 min after the bleach. Quantitative spot photobleaching was done in the brighter COS7 cells. Fig. 3A shows spot photobleaching of GFP-labeled wt-CFTR (top) and ⌬F508-CFTR (bottom) using a ϫ60 oil immersion objective to produce a small spot (ϳ1.2-m diameter) in which GFP fluorescence was bleached to ϳ50% of its initial value. To distinguish between processes that produce irreversible photobleaching (authentic GFP diffusion) from those that produce reversible photobleaching (from triplet-state relaxation or other photophysical phenomena), fluorescence recovery was measured in fixed cells (Fig. 3B) and using different spot sizes (Fig. 3C). As found in prior photobleaching studies (26,27), the fluorescence recovery remaining after fixation represents reversible (diffusion-independent) photobleaching because fixation abolishes diffusion. Fig. 3B (top curve) shows a small component (5-10%) of reversible photobleaching with ϳ1-s exponential time constant. Similar recoveries were found under the same photobleaching conditions for unconjugated GFP in cytoplasm (middle curve) and aqueous phase (unconjugated) GFP in the ER lumen (bottom curve). As expected for reversible fluorescence recovery, the recovery rates did not depend on spot size (not shown). Fig. 3C shows that (in unfixed cells) the fluorescence recovery was spot size-dependent as expected for diffusive processes, because the recovery time for a larger spot increases approximately with the square of the spot diameter. Using a regression procedure to determine (t1 ⁄2 ) for irreversible fluorescence recovery (see "Experimental Procedures"), t1 ⁄2 values for GFP-wt-CFTR were 1040 Ϯ 60 ms (100ϫ), 3.0 Ϯ 0.4 s (60ϫ), and 4.9 Ϯ 0.2 s (40ϫ) (n ϭ 4 -10 sets of measurements). The t1 ⁄2 values for GFP-⌬F508-CFTR (in BFA-treated cells) were 1170 Ϯ 120 ms (100ϫ), 3.3 Ϯ 0.5 s (60ϫ), and 6.4 Ϯ 0.3 s (40ϫ) (n ϭ 4 -10). The relative t1 ⁄2 are as expected for a diffusive process.
To investigate interactions of wt-CFTR and ⌬F508-CFTR with elements of the ER quality control machinery, we used a series of maneuvers to modulate molecular chaperone interactions. Fig. 4A shows wt-CFTR mobility after ATP depletion (a maneuver that increases interactions between CFTR and chaperones in the ER), which was produced by 30 min of incubation with 6 mM 2-deoxyglucose and 0.02% sodium azide (top curve). The diffusion coefficient of GFP-wt-CFTR was similar to control conditions, (10.6 Ϯ 0.9) ϫ 10 Ϫ10 cm 2 /s (n ϭ 7), and the percentage recovery was reduced significantly to 80 Ϯ 2%. The diffusion coefficient of GFP-⌬F508-CFTR was reduced significantly after a 30-min ATP depletion to (6.6 Ϯ 0.5) ϫ 10 Ϫ10 cm 2 /s (n ϭ 5) (Fig. 4A, bottom curve), and substantially more GFP-⌬F508-CFTR became immobilized (mobile percentage, 55 Ϯ 5%). ATP depletion for 60 min had no further effect. Table I summarizes averaged diffusion coefficients and percentages of mobile GFP. To investigate whether the reduced ⌬F508-CFTR mobility in ATP-depleted cells was related to restructuring or fragmentation of the ER, the mobility of the fluorescent ER membrane marker diOC 4 (3) was measured. Fig. 4B shows that diffusion of diOC 4 (3) was essentially complete with a diffusion coefficient of (2.4 Ϯ 0.1) ϫ 10 Ϫ8 cm 2 /s (n ϭ 10). diOC 4 (3) diffusion was unaffected by BFA treatment ((2.4 Ϯ 0.1) ϫ 10 Ϫ8 cm 2 /s, n ϭ 5, middle curve) or ATP depletion ((2.7 Ϯ 0.3) ϫ 10 Ϫ8 cm 2 /s, n ϭ 4, bottom curve). Further, photobleaching of transfected cells expressing GFP-labeled wild type CFTR or ⌬F508-CFTR and subsequently labeled with diOC 4 (3) in situ showed that the expression of GFP-CFTR chimeras did not affect ER structure or fluidity (not shown). Image photobleaching experiments similar to those shown in Fig. 2 were also performed on diOC 4 (3)-labeled cells (not shown), and it was demonstrated that ATP depletion did not alter ER continuity or gross structure.
To identify the cellular location of the immobilized GFP-⌬F508-CFTR in ATP-depleted cells, fluorescence loss in photobleaching (FLIP) experiments were performed (Fig. 4C). A fixed spot in the cell was repetitively bleached with delays between each bleach pulse to permit diffusion of unbleached GFP-labeled CFTR into the location of the bleach spot. Under these conditions GFP-wt-CFTR freely diffused throughout the ER with ϳ10% of cell fluorescence remaining after 35 bleach pulses (top). In contrast, a substantial amount of fluorescence remained after bleaching of GFP-⌬F508-CFTR in ATP-depleted cells (bottom) with the remaining fluorescence having an ER pattern. These findings suggest that ATP depletion does not alter ER continuity or gross structure and that the immobilization of ⌬F508-CFTR in ATP-depleted cells results from interactions with ER proteins.
Additional maneuvers were carried out to modulate putative CFTR interactions with molecular chaperones and elements of protein processing and degradation ( Fig. 5 and Table I). Heat shock, which nonspecifically up-regulates the expression of molecular chaperones, produced small reductions in the diffusion coefficients and percentage mobilities of wt-CFTR and ⌬F508-CFTR (Fig. 5, top curves). Proteasome inhibition by lactacystin resulted in mild slowing and immobilization of wt-CFTR and ⌬F508-CFTR after 6 h of incubation (second curves), with complete immobilization after 14 h (third curves). GFP was seen in an ER pattern at 6 h but was concentrated in aggresomes at 14 h (not shown). The ansamy-sin compound geldanamycin, which disrupts associations with Hsp90, caused comparable slowing of wt-CFTR and ⌬F508-CFTR with somewhat greater immobilization of ⌬F508-CFTR (fourth curves). Last, the role of glycosylation was investigated by incubation with tunicamycin (to inhibit addition of oligosaccharide chains), using a glycosylation-null mutant of wt-CFTR (asparagines 894 and 900 replaced by glutamines). Somewhat more immobilization of ⌬F508-CFTR than wt-CFTR was found with tunicamycin (fifth curves). Although the glycosylation-null mutant of wt-CFTR expressed poorly and a substantial proportion of fluorescence was associated with aggresomes, the mobility of ERassociated CFTR could be measured in some cells and was found to be similar to that produced by tunicamycin (sixth curve).
As was done for the ATP depletion experiments, the diffusion of the ER marker diOC 4 (3) was measured in transfected cells to establish that ER restructuring or fragmentation was not responsible for reduced CFTR mobility. The diffusion of diOC 4 (3) in cells that were heat-shocked or treated with lactacystin, geldanamycin, or tunicamycin was essentially complete (Ͼ96% mobile) and similar to that measured in control cells, with diffusion coefficients of 2.2-2.7 ϫ 10 Ϫ8 cm 2 /s (n ϭ 4 for each treatment, data not shown).

FIG. 2. Photobleaching of GFP-CFTR chimeras in transiently transfected COS7 (A) and CHO (B) cells.
Photobleaching with image detection in cells expressing GFP-labeled wt-CFTR and ⌬F508-CFTR (after BFA treatment). Images are shown before bleaching, with the bleach spot indicated by white crosshairs, and at the indicated times after the bleach pulse. All experiments were done at 37°C. Each image series is one set of experiments representative of 4 -8. Scale bar, 10 m.

DISCUSSION
This study has shown that although ⌬F508-CFTR is retained in the ER, it diffuses at a comparable rate to wt-CFTR in BFA-treated cells. Similar observations were made for the T126M mutant of AQP2 (a water channel that is partially misfolded and similarly retained in the ER, Ref. 27) and for a temperature-sensitive folding mutant of the vesicular stomatitis virus G protein (29). Therefore, slowed or restricted ⌬F508-CFTR mobility in the ER cannot account for its failure to be exported. Efficient recycling from the Golgi or failure to be recognized by the ER export machinery remain possible explanations for the ER localization of ⌬F508-CFTR. The observation of small amounts of ⌬F508-CFTR in post-ER membranes suggests that efficient recycling may be responsible for the retention of ⌬F508-CFTR in the ER (30).
The small immobile pool of ⌬F508-CFTR and the substantial immobilization after ATP depletion suggest that a greater frac-tion of ⌬F508-CFTR interacts with molecular chaperones than does wt-CFTR. Approximately 50% of ⌬F508-CFTR was immobile after ATP depletion, with only minor effects on the mobility of wt-CFTR. The simplest interpretation of this finding is that ATP depletion, which stabilizes protein interactions with molecular chaperones, is able to reveal ⌬F508-CFTR interactions that are not observable under normal conditions. As discussed in the Introduction, ⌬F508-CFTR structure is thought to be similar to that of an early intermediate of wt-CFTR, and biochemical evidence supports ⌬F508-CFTR associations with molecular chaperones.
Additional maneuvers were performed to investigate specific interactions between wt-CFTR and ⌬F508-CFTR and the cellular protein processing machinery. Inhibition of N-linked glycosylation by tunicamycin resulted in significant slowing and immobilization of both wt-CFTR and ⌬F508-CFTR, similar to findings for the VSVG glycoprotein (29). Similar results were FIG. 4. Slowed GFP-⌬F508-CFTR diffusion after ATP depletion. A, photobleaching after cellular ATP depletion produced by a 30-min incubation with 6 mM 2-deoxyglucose and 0.02% sodium azide (ϫ60 objective). B, photobleaching of the ER lipid probe diOC 4 (3) in control, BFA-treated, and ATP-depleted cells (ϫ100 objective). C, fluorescence depletion experiments of GFP-wt-CFTR under control conditions (top) and of GFP-⌬F508-CFTR after ATP depletion (bottom). Images shown before bleaching, with the bleach spot indicated by white cross-hairs, and after the indicated number of bleach pulses. Scale bar, 10 m. found with a glycosylation-null wt-CFTR mutant, which was poorly expressed in the plasma membrane as determined by biotinylation studies (not shown). Protein glycosylation can facilitate folding, sorting, and quality control (31). Our data are thus consistent with a role for glycosylation in CFTR folding. Proteasome inhibition causes the accumulation of deglycosylated wt-CFTR and ⌬F508-CFTR in detergent-insoluble, high molecular weight aggregates (9,32). Consistent with the observations here, Bebök et al. (32) demonstrated an association between the Sec61 translocation system and CFTR that was enhanced by proteasome inhibition. After prolonged (14 h) incubation with lactacystin, wt-CFTR and ⌬F508-CFTR were found in aggresomes (9,10). Aggresomes are vimentin-enclosed, pericentriolar, electron-dense structures that contain aggregates of misfolded, often ubiquinated proteins associated with molecular chaperones and proteasomes. As expected, the GFP-CFTR chimeras were completely immobile in aggresomes. Inhibition of interaction between Hsp90 and CFTR by geldanamycin also slowed the diffusion of wt-CFTR and ⌬F508-CFTR, with greater immobilization of ⌬F508-CFTR. Geldanamycin has been shown to disrupt interactions of Hsp90 and CFTR, increasing the degradation rates of wt-CFTR and ⌬F508-CFTR (14). Hsp90 has thus been proposed to have a potential role in protecting CFTR from rapid degradation. The relative immobilization of wt-CFTR and ⌬F508-CFTR after geldanamycin treatment supports a role of Hsp90 for CFTR processing. As suggested for VSVG by Nehls et al. (29), we believe that the physical state of wt-CFTR and ⌬F508-CFTR after perturbation of CFTR processing by tunicamycin or geldanamycin may be different from that produced by ATP depletion. The slowing and immobilization of CFTR by ATP depletion may represent CFTR interactions with molecular chaperones existing in large complexes (33), whereas the immobilization after tunicamycin and geldanamycin, or molecular deglycosylation, may represent CFTR self-aggregation. Because of the logarithmic relationship between molecular radius and diffusion rate for a protein in a bilayer, substantial aggregates (10 -100 units) must be formed to significantly slow diffusion. The similar responses of wt-CFTR and ⌬F508-CFTR to some of the pharmacological maneuvers are consistent with the intrinsically inefficient folding of wt-CFTR, which when compared with P-glycoprotein (another member of the ABC cassette family) is poorly folded.
The diffusion coefficient for CFTR diffusional mobility in the ER, ϳ10 ϫ 10 Ϫ10 cm 2 /s, is within the range of other ER membrane proteins that have been studied. The membrane water channel AQP2 (retained in the ER with BFA) and the ERretained mutant AQP2-T126M have diffusion coefficients of 2.6 -3.0 ϫ 10 Ϫ10 cm 2 /s (27). Cytochrome P450 2C2, an intrinsic ER protein, has a diffusion coefficient of 5.8 ϫ 10 Ϫ10 cm 2 /s (34). VSVG in its native and misfolded states and two ER resident transmembrane proteins (lamin B receptor and the ␤-subunit of the signal recognition particle receptor) have diffusion coefficients of 26 -50 ϫ 10 Ϫ10 cm 2 /s, depending upon temperature (29). Two Golgi resident transmembrane proteins (galactosyltransferase and the KDEL receptor) retained in the ER with BFA have diffusion coefficients of 21-48 ϫ 10 Ϫ10 cm 2 /s, depending upon cell type and temperature (29,35). The diffusion coefficient of the ER-retained MHC class 1 molecule H2L d is 20 -46 ϫ 10 Ϫ10 cm 2 /s and that of TAP1 (transporter associated with antigen presentation 1) is 12 ϫ 10 Ϫ10 cm 2 /s (36). The variation in these diffusion coefficients probably reflects different protein sizes, interactions with ER and cytoplasmic proteins, and ER composition in different cells.
In summary, the photobleaching data provide the following evidence: (a) the ER retention of ⌬F508-CFTR is not due to restricted ER mobility, (b) the majority of ⌬F508-CFTR is not aggregated or bound to slowly moving membrane proteins, and (c) ⌬F508-CFTR may interact to a greater extent with molecular chaperones than does wt-CFTR. Measurements of ER membrane protein diffusion provide a unique in vivo approach to study protein-protein associations that complements classical biochemical and 2-hybrid methods.