Most patients with cystic fibrosis (CF) have a single codon deletion (ΔF508) in the gene encoding the cystic fibrosis transmembrane conductance regulator (CFTR) that impairs assembly of the multidomain glycoprotein. The mutant protein escapes endoplasmic reticulum (ER) quality control at low temperature, but is rapidly cleared from the distal secretory pathway and degraded in lysosomes. CF cells accumulate free cholesterol similar to Niemann-Pick disease type C cells. We show that this lipid alteration is caused by the presence of misassembled mutant CFTR proteins, including ΔF508, in the distal secretory pathway rather than the absence of functional CFTR. By contrast, cholesterol distribution is not changed by either D572N CFTR, which does not mature even at low temperature, or G551D, which is processed normally but is inactive. On expression of the ΔF508 mutant, cholesterol and glycosphingolipids accumulate in punctate endosomal structures and cholesterol esters are reduced, indicating a block in the translocation of cholesterol to the ER for esterification. This is overcome by Rab9 overexpression, resulting in clearance of accumulating intracellular cholesterol. Similar but less pronounced alterations in intracellular cholesterol distribution are observed on expression of a temperature-rescued mutant variant of the related ATP-binding cassette (ABC) protein multidrug resistance-associated protein 1 (MRP1). Thus, on escape from ER quality control, misassembled mutants of CFTR and MRP1 impair lipid homeostasis in endocytic compartments.
Correct intracellular cholesterol trafficking and distribution among subcellular organelles is essential for the maintenance of cholesterol homeostasis. Cholesterol is a highly hydrophobic lipid that is transported through the aqueous cytosol either by vesicles or by a nonvesicular pathway involving soluble carrier proteins (Mukherjee and Maxfield, 2004). A major source of cellular cholesterol is endocytic uptake of lipoproteins such as low-density lipoprotein (LDL). The lipoprotein-derived cholesterol is rapidly released from endosomes and lysosomes, and is transferred to the endoplasmic reticulum (ER) for esterification, where newly synthesized cholesterol is produced. Retrograde vesicular transport through the Golgi complex has been implicated in the transport to the ER, but the detailed mechanism whereby cholesterol moves to the ER is still uncertain (Chang et al., 2006).
In the inherited lysosomal storage disorder Niemann-Pick disease type C (NPC), transport to the Golgi and ER is blocked and cholesterol and other lipids accumulate in late endosomal compartments (Kruth et al., 1986). White et al. (White et al., 2004) recently observed similar accumulation of free cholesterol in cystic fibrosis (CF). CF is primarily a disease of defective epithelial salt and fluid transport resulting from mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) chloride channel (Quinton, 1999); however, the disease phenotype is pleiotropic and includes reports of a variety of lipid alterations (Peretti et al., 2005). More than 1400 different mutations in the gene encoding CFTR have been identified in patients with CF; of these, a deletion of phenylalanine 508 (ΔF508) is the most common, being present in more than 90% of patients with CF (Sferra and Collins, 1993).
The CFTR protein is a member of the ABCC subfamily of ATP-binding cassette (ABC) membrane transport proteins, which play important functional roles in many tissues including the liver, especially the biliary tree, lung, pancreas, retina and immune system (Dean, 2005). Several ABC proteins are involved in multidrug resistance of human cancer cells, but in addition an increasing number have been implicated in genetic disease and in many cases protein misfolding has been shown to be responsible. Over time, CFTR has become a prototype for studies of ABC protein misfolding in genetic disease. Deletion of phenylalanine at position 508 in CFTR results in the temperature-sensitive ΔF508 protein that cannot mature conformationally at 37°C. Growth of cells at reduced temperature or other manipulations enable the nascent mutant protein to avoid ER quality control, but it is then very rapidly cleared from the distal secretory pathway and degraded in lysosomes (Gentzsch et al., 2004; Sharma et al., 2004).
During the course of experiments to examine the endocytic compartments visited by ΔF508 CFTR, we observed redistribution and accumulation of cholesterol similar to that seen in NPC and to that reported by Kelley and coworkers (White et al., 2004), who attributed this effect to the lack of CFTR function. We show in this study that the presence of several misassembled mutant CFTR proteins in the distal secretory pathway rather than the absence of functional CFTR is responsible for the disturbance of cholesterol trafficking. We present evidence that cholesterol and glycosphingolipids accumulate in endocytic compartments as a result of a block in translocation to the Golgi and ER. Interestingly, we observed similar alterations in intracellular cholesterol distribution on expression of a temperature-rescued variant of a related ABC protein, multidrug resistance-associated protein 1 (MRP1), raising the possibility that there might be a more generalized impact on lipid homeostasis caused by misfolded membrane proteins that avoid ER quality control.
ΔF508 CFTR perturbs intracellular cholesterol trafficking
Following the report of White et al. (White et al., 2004), we employed filipin labeling (Sokol et al., 1988) to detect free cholesterol in baby hamster kidney cells (BHK-21) and Chinese hamster ovary cells (CHO-K1), both of which do not express CFTR endogenously, before and after stable transfection with wild-type or ΔF508 CFTR (Fig. 1A). Both cell types expressing either wild-type or no CFTR showed staining of a condensed perinuclear region. By contrast, this structure was nearly unstained in ΔF508-expressing cells, which instead show intense labeling of punctate vesicular structures throughout the cytoplasm. Importantly, this striking change in distribution was most pronounced in cells that had been cultured at 27°C rather than 37°C, providing an initial clue to the means by which the mutant protein causes this change. To verify that ΔF508 CFTR escapes ER quality control at reduced temperature, we applied western blotting of lysates from BHK-21 cells expressing CFTR and ΔF508 CFTR at 37°C and 27°C, and detected mature ΔF508 CFTR in cells grown at low temperature (Fig. 1B). Immunofluorescence microscopy on nonpermeabilized cells detecting an external epitope confirmed cell-surface localization of ΔF508 CFTR in cells that were incubated at 27°C. At the lower temperature, the ΔF508 protein is able to transit from the ER to the Golgi, where it acquires complex oligosaccharide chains and proceeds from there to the cell surface. Thus, the change in cholesterol distribution is most dramatic under conditions where ΔF508 can enter the distal secretory pathway.
To confirm that cholesterol localizes to perinuclear Golgi-related compartments in BHK-21 cells that were not expressing wild-type or ΔF508 CFTR, we employed two commonly used Golgi markers, GM130 (Golgi matrix protein of 130 kDa) (Nakamura et al., 1995) and giantin (376 kDa Golgi complex membrane protein) (Linstedt and Hauri, 1993) and costained cholesterol with filipin (Fig. 2). Most of the intracellular cholesterol localized to a perinuclear location that was also labeled by GM130 or giantin. However, in cells expressing ΔF508 CFTR, cholesterol was distributed throughout the cells in a punctate pattern that was especially obvious at lower temperature (Fig. 1A and Fig. 2). Incubation at lower temperature did not alter the appearance of the Golgi compartment detected by GM130 and giantin antibodies in either the untransfected host BHK-21 cells or those expressing ΔF508 CFTR (Fig. 2).
Cholesterol redistribution correlates with CFTR misprocessing not dysfunction
To test the idea that the perturbation of cholesterol homeostasis was due to the presence of a misassembled mutant protein in the distal secretory pathway we examined CFTR variants (Fig. 3A) that are dysfunctional but processed normally (G55ID), misprocessed but not rescued from ER quality control at low temperature (D572N) or, like ΔF508, misprocessed but rescued at low temperatures (1410X). The diffuse punctate filipin staining pattern was clearly apparent at 27°C in 1410X as in ΔF508 expressing cells (Fig. 3C).
Mature, complex-glycosylated forms of C-terminal CFTR truncation mutants like 1410X were shown to have a five to sixfold faster degradation rate than wild-type CFTR (Haardt et al., 1999) and resemble in this aspect rescued ΔF508 (Lukacs et al., 1993; Sharma et al., 2001; Sharma et al., 2004). Neither the severe disease-causing mutation G551D, which prevents CFTR channel activation although it is processed normally (Cutting et al., 1990; Gregory et al., 1991), nor the D572N mutation, which is retained at the ER at high or low temperature, changed cholesterol distribution from normal.
We extracted cellular lipids from cells expressing the different CFTR variants grown at 27°C and separated cholesterol by thin-layer chromatography. There were increased amounts of free cholesterol in those variants where it appeared redistributed (Fig. 3B). As independent verification that altered cholesterol homeostasis was not caused by a lack of a functional CFTR, we treated cells with the CFTR channel inhibitor I-172 (Ma et al., 2002) (Fig. 3D). The compound was without effect on cholesterol localization in cells expressing wild-type CFTR. Combined with the fact that cells not transfected with any CFTR have the same normal cholesterol distribution as those transfected with the wild-type CFTR, these data argue strongly that some factor other than reduced CFTR function is responsible for accumulation of the sterol.
A related mutant ABC protein ΔF728 MRP1 affects cellular cholesterol distribution
Rescued ΔF508 CFTR that escapes ER quality control at low temperature rapidly disappears from the cell surface and is degraded in lysosomal compartments (Gentzsch et al., 2004). Our hypothesis is that the presence of the misassembled CFTR glycoprotein in the distal secretory pathway might be responsible for disturbance in cholesterol trafficking and we tested whether the counterpart of the ΔF508 mutation in CFTR in another member of the same subfamily of ABC proteins, the MRP1 multidrug transporter (Buyse et al., 2004; Deeley and Cole, 2003), would have a similar effect. That this mutation does have a comparable effect on maturation and trafficking from the ER and is temperature sensitive is shown in Fig. 4A,B. Western blotting and immunofluorescence microscopy of cells expressing MRP1 or ΔF728 MRP1 grown at 37°C and 27°C confirmed that ΔF728 MRP1 is a temperature-sensitive misfolding mutant that matures at low temperature and proceeds to the cell surface. ΔF728 MRP1, when rescued at 27°C, causes a partial redistribution of filipin labeling from the normal tightly clustered perinuclear structures to diffuse pan-cytoplasmic vesicles (Fig. 4C). The effect, however, is somewhat less pronounced than with ΔF508 CFTR rescued by low temperature. The behavior of rescued ΔF728 MRP1 in the endocytic pathway has not been characterized, but it appears that a temperature-sensitive assembly variant of a second ABC protein also affects cellular cholesterol handling. So-called folding mutations of other ABC proteins have been identified that contribute to several other human genetic diseases (Hanrahan et al., 2003) but their possible influence on the trafficking of sterols or other molecules has not been investigated.
Distribution of fluorescent sphingolipid analog is altered in cells expressing ΔF508 CFTR
Given that cholesterol was present in punctuate endosomal locations and quantification of cholesterol revealed that the total amount was increased (Fig. 1A and Fig. 3B), we tested the hypothesis that cholesterol accumulates as a result of a block in transit from the endosomes to the Golgi and ER, the site of esterification. Pulse labeling with the fluorescent analog NBD-cholesterol resulted primarily in the diffuse punctate pattern at early times, regardless of CFTR genotype (not shown), which persisted at later times in cells expressing ΔF508 but not in cells expressing wild-type or no CFTR (Fig. 5A). Although cholesterol accumulated in endosomal/lysosomal compartments in cells expressing ΔF508 CFTR, perinuclear Nile Red staining, which detects cholesteryl esters and other neutral lipids, was reduced (Fig. 5B). This might indicate a block in translocation of cholesterol to the Golgi and ER for esterification, similar to that observed in NPC.
Cholesterol and sphingolipids have high affinity for one another and are the two main components of lipid raft microdomains, and therefore accumulation of one can cause trapping and accumulation of the other (Simons and Gruenberg, 2000). Furthermore, cholesterol affects sphingolipid trafficking and modulates membrane trafficking along the endocytic pathway in sphingolipid storage diseases (Choudhury et al., 2004; Puri et al., 1999). We found in pulse-chase labeling experiments with the most commonly used fluorescent sphingolipid analog BODIPY-lactosylceramide (BODIPY-LacCer) (Marks et al., 2005) that it did not transit from the punctate cholesterol-containing vesicles to the Golgi in cells expressing rescued ΔF508 CFTR, but did transit in cells expressing wild-type or no CFTR (Fig. 5C). Thus, there is a block in glycosphingolipid as well as cholesterol trafficking.
Rab9 overcomes the cholesterol block induced by ΔF508 CFTR
In the case of NPC, we had shown earlier that elevated intracellular cholesterol could be reduced by overexpression of the small GTPase Rab9 that promotes trafficking from late endosomes to the trans-Golgi network (Choudhury et al., 2002; Narita et al., 2005). More recently, we observed that overexpression of Rab9 drives rescued ΔF508 CFTR along this route (Gentzsch et al., 2004). We asked whether we could rescue cholesterol trafficking in cells expressing ΔF508 by overexpression of Rab9. BHK-21 cells expressing CFTR or ΔF508 CFTR were transfected with a plasmid encoding an enhanced green fluorescent protein (EGFP) fusion of Rab9 (Choudhury et al., 2002) and 24 hours later the cells were shifted to 27°C and incubated for another 24 hours. Three representative panels of cells expressing wild-type and ΔF508 CFTR that had been transfected with Rab9-EGFP are shown in Fig. 6 and cells that overexpress Rab9-EGFP are indicated by an arrow in the filipin panel. The punctate filipin staining is cleared from cells expressing Rab9-EGFP and ΔF508 CFTR, but remains unchanged in a perinuclear Golgi-like location in cells expressing Rab9-EGFP and CFTR. The histogram in Fig. 6 shows that Rab9 overexpression is able to clear the punctate cholesterol accumulations and this might be the consequence of Rab9 overcoming an endosome-to-Golgi cholesterol trafficking block in ΔF508-expressing cells.
Although CF is not primarily a disease of lipid metabolism, there is a very large literature describing lipid changes in patients with CF (Peretti et al., 2005). Most prominent is essential fatty acid deficiency (Strandvik et al., 2001), which is generally attributed to defective fat absorption and processing in the intestine, secondary to both the lack of pancreatic lipases and changes in the absorptive epithelium as a result of the absence of the CFTR ion channel. However, there are some suggestions that mutant CFTR might play a more direct role in epithelial essential fatty acid utilization (Bhura-Bandali et al., 2000; Freedman et al., 2004). Serum lipoprotein cholesteryl ester depletion occurs in patients with CF, and this is more pronounced in those with greater essential fatty acid deficiency (Levy et al., 1993).
Prior to the report of White et al. (White et al., 2004), little attention had been paid to the cellular handling of cholesterol and cholesterol esters in CF. We have now confirmed and considerably extended these findings. The expression of ΔF508 and other misprocessed CFTR mutants, but not dysfunctional yet correctly processed variants, clearly causes a redistribution of free cholesterol. Several observations suggest a block in the transport of free cholesterol from a vesicular compartment, probably late endosomes, to the Golgi and ER. There is an increase in cellular free cholesterol in this vesicular compartment and a decrease in cholesterol esters that are normally formed in the ER. Overexpression of Rab9, which promotes trafficking from late endosomes to the Golgi, overcomes the cholesterol accumulation and significantly reduces intracellular free cholesterol levels. When presented to cells expressing wild-type or no CFTR, fluorescent NBD-cholesterol progressed from the dispersed punctuate pattern to the perinuclear accumulations over time but remained in the endocytic vesicular compartment in cells expressing misprocessed CFTR mutants. A blockage of the movement of glycosphingolipids between the same two locations in cells expressing misprocessed CFTR was observed using BODIPY-LacCer as also occurs in NPC (Choudhury et al., 2002).
Glycolipids are of interest in CF because asialo-GM1 ganglioside is a receptor for Pseudomonas aeruginosa, which is the major colonizing microbe in the CF lung. Localization of CFTR to cholesterol- and glycosphingolipid-containing lipid rafts has been reported to be required for epithelial cell signaling in response to P. aeruginosa infection (Kowalski and Pier, 2004).
Wild-type CFTR is normally rapidly endocytosed in clathrin-coated vesicles, with most recycled to the plasma membrane and a small proportion routed to late endosomes and lysosomes for degradation (Bradbury et al., 1999; Lukacs et al., 1997). However, when ΔF508 CFTR is able to escape ER quality control and reach the cell surface, it proceeds through the endosomal compartment to lysosomes for degradation without significant recycling to the plasma membrane (Gentzsch et al., 2004; Sharma et al., 2004). Therefore, the presence of the mutant protein in late endosomes could contribute to retention of cholesterol and sphingolipids. How this might occur is largely a matter of speculation at this time. Nothing is known of the interactions of sterols or lipids with either wild-type or mutant CFTR, but it is a multi-spanning integral membrane protein fully integrated into the lipid bilayer. The ΔF508 mutation deletes a residue from a hydrophilic cytoplasmic domain of CFTR and this perturbs the association and conformation of the two large membrane-spanning domains of the protein that interact with lipid. Although the amount of the mutant polypeptide in endosomal membranes may be insufficient to retain lipid by direct binding it may have an impact on sterol-binding proteins involved in endosomal trafficking such as MLN64 (Alpy and Tomasetto, 2006), ORP1 (Olkkonen et al., 2006) or caveolin-1 (Ikonen et al., 2004; Pelkmans et al., 2004). We recently found a small amount of caveolin-1 co-immunoprecipitated with CFTR (A.M. and J.R.R., unpublished observation). There may be as-yet-unknown additional CFTR-interacting cholesterol-binding proteins, possibly even including NPC1 or NPC2, which when mutated cause NPC (Carstea et al., 1997; Naureckiene et al., 2000).
The molecular details of cholesterol trafficking in several lipid storage diseases such as NPC that have been investigated for many years are still not fully understood (Maxfield and Tabas, 2005). In NPC, it has become apparent that the sterol and lipid perturbations are directly caused by mutations in genes encoding NPC1 and NPC2 (Cheruku et al., 2006; Ohsaki et al., 2006; Sleat et al., 2004). By contrast, the superficially similar changes caused by mutant CFTR are obviously secondary events in the CF cells that express mutants that cause misprocessing and not just a loss of function. This raises the possibility that there could be such secondary consequences resulting from misprocessing mutations in the genes encoding other membrane proteins and might emphasize the utility of a stringent ER quality control system to avoid or minimize these effects. We found that the counterpart of the ΔF508 mutation in a second ABC protein, ΔF728 MRP1, caused a similar cholesterol redistribution and it will be of interest to examine other ABC proteins in which misprocessing mutations are responsible for several other genetic diseases (Hanrahan et al., 2003).
It is also possible that the lipid localization changes observed are reflective of a more general perturbation of vesicular trafficking, a possibility we are currently investigating. Other alterations of the endocytic pathway, caused for example by the deficiency in lysosomal-associated membrane protein 1 (LAMP-1) and LAMP-2 (Eskelinen et al., 2004) or dynamin inactivation (Robinet et al., 2006), have been reported to disturb cholesterol traffic and, similar to what we have observed, result in accumulation of unesterified cholesterol in endolysosomal compartments.
Overall, our findings have significant implications not only in CF but possibly also more generally in providing an initial indication of the impact on cells of misfolded proteins that gain access to the distal secretory pathway.
Materials and Methods
Antibodies and reagents
Anti-MRP1 and anti-CFTR mouse monoclonal antibodies (mAbs) 897.2 and 596 have been described earlier (Aleksandrov et al., 2002; Hou et al., 2000), mouse monoclonal anti-hemaglutinin (HA; 16B12) and rabbit anti-giantin antibodies were from Covance, mouse monoclonal anit-GMA130 antibody was from BD Transduction Laboratories, goat anti-mouse Alexa Fluor 568 and 488 IgG conjugate and goat anti-rabbit Alexa Fluor 568 IgG conjugate were from Molecular Probes (Invitrogen). BODIPY-LacCer was synthesized as described earlier (Martin and Pagano, 1994), filipin was from Polysciences Inc. or Sigma, and Nile Red was from Eastman Kodak. CFTR inhibitor I-172 was synthesized by A. Fauq (Chemical Synthesis Core Facility, Mayo Clinic Jacksonville, Jacksonville, FL, USA) and functionality was confirmed by inhibition of CFTR single channel currents in lipid bilayer experiments (Aleksandrov et al., 2003). Anti-mouse IgG horseradish peroxidase conjugate was from Amersham Bioscience and chemiluminescent substrate was from Pierce. NBD-cholesterol (22-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl)amino)-23,24-bisnor-5-cholen-3β-ol) was from Molecular Probes.
Cells and cell culture
BHK-21 and CHO-K1 cells were obtained from the American Type Culture Collection and grown in DMEM/F12 supplemented with 5% fetal bovine serum (FBS) or MEM Alpha Medium supplemented with 8% FBS, respectively, in humidified incubators with 5% CO2 at 37°C, or 24-48 hours at 27°C if indicated. BHK-21 and CHO-K1 cells stably expressing CFTR or ΔF508 CFTR have been described in our earlier publications (Chang et al., 1993; Seibert et al., 1995). BHK-21 clones stably expressing CFTR and ΔF508 CFTR with HA epitope have been described earlier (Gentzsch et al., 2004). Stable BHK-21 cell lines expressing G551D and D572N variants of CFTR or the C-terminal truncation 1410X CFTR were established as described previously (Chang et al., 1993; Gentzsch and Riordan, 2001; Loo et al., 1998). BHK-21 clones stably expressing MRP1 or ΔF728 MRP1 were constructed by transfection of cells with the corresponding cDNAs in the pNUT expression vector. To create ΔF728 MRP1 cDNA, MRP1 cDNA cloned into pNUT (pNUT/MRP1/His) (Chang et al., 1997) was used as a template for in vitro mutagenesis. The phenylalanine residue at position of 728 was deleted by using the following forward and reverse primers and the QuikChange site-directed mutagenesis kit (Hou et al., 2000). The forward and reverse primers for ΔF728 are: ΔF728 forward, 5′-CTC CGA GAA AAC ATC CTT ΔΔΔ GGA TGT CAG CTG GAG GAA-3′; and ΔF728 reverse, 5′-TTC CTC CAG CTG ACA TCC ΔΔΔ AAG GAT GTT TTC TCG GAG-3′. The ΔΔΔ signs indicate that the three nucleotides TTT, the codon for F728, were deleted in the forward and reverse primers. pNUT/ΔF728/MRP1/His was confirmed by sequencing.
For fluorescence microscopy, cells were fixed with 4% paraformaldehyde for 10 minutes. Cholesterol was stained with 50 μg/ml filipin as described previously (Choudhury et al., 2004). To monitor internalization of the fluorescent glycosphingolipid analog, cells were incubated 24 to 48 hours at 27°C and labeled with BODIPY-LacCer for 30 minutes at 4°C and subsequently chased for 1 hour at 37°C as described earlier (Choudhury et al., 2002). For Nile Red staining, cells were fixed for 30 minutes with 3% paraformaldehyde and were then treated for 5 seconds with ice-cold phosphate-buffered saline (PBS) containing 0.05% Triton X-100, washed and further incubated with 100 nM Nile Red in PBS for 10 minutes. Nile Red fluorescence was detected at red wavelengths and quantified by image analysis. GM130 and giantin were detected after permeabilization with 0.1% saponin in PBS using mouse anti-GM130 mAb (BD Transduction Laboratories) followed by goat anti-mouse Alexa Fluor 568 IgG conjugate or rabbit anti-giantin (Covance Research Products) followed by goat anti-rabbit Alexa Fluor 568 IgG conjugate. Fluorescence microscopy was carried out using an Olympus IX70 or a Leica DMRB fluorescence microscope and a Zeiss LSM510 confocal laser scanning microscope. Image analysis was performed using MetaMorph V4.6.9 (Universal Imaging Corp.) and ImageJ 1.34s (ImageJ; http://rsb.info.nih.gov/ij/) image-processing programs. In any given experiment, all photomicrographs were exposed and processed identically for a given fluorophore.
Cells were lyzed and subjected to SDS gel electrophoresis on 6% polyacrylamide gels. Proteins were blotted to nitrocellulose membranes and probed with anti-CFTR mAb 596 or monoclonal anti-MRP1 antibody 897.2 and visualized by enhanced chemiluminescence detection.
Quantitative cholesterol analysis
Cellular lipids were extracted with chloroform/methanol (2/1) from 5×106 cells of each clone and cholesterol was separated by thin-layer chromatography, stained and quantified as described previously (Puri et al., 2003).
We thank Christine L. Wheatley, Victor V. Ozols and Yue-xian Hou for assistance with tissue culture. We are grateful to Mark Ruona and Alicia Orth for advice with figure preparation. A.C. was supported by a fellowship from the American Heart Association. This work was supported by the NIH and CFF.