Evidence That Endoplasmic Reticulum (ER)-associated Degradation of Cystic Fibrosis Transmembrane Conductance Regulator Is Linked to Retrograde Translocation from the ER Membrane*

The ubiquitin-proteasome pathway has been implicated in the degradation of newly synthesized, misfolded and unassembled proteins in the endoplasmic reticulum (ER). Using a cell-free reticulocyte lysate system we have examined the relationship between biosynthesis and ER-associated degradation of the cystic fibrosis transmembrane conductance regulator (CFTR), a polytopic protein with 12 predicted transmembrane segments. Our results provide direct evidence that full-length, glycosylated and membrane-integrated CFTR is a substrate for degradation and that degradation involves polyubiquitination and cytosolic proteolytic activity. CFTR ubiquitination was both temperature- and ATP-dependent. Degradation was significantly inhibited by EDTA, apyrase, and the proteasome inhibitors hemin and MG132. Degradation was inhibited to a lesser extent by clasto-lactacystin β-lactone, ALLN, andN α-tosyl-l-phenylalanine chloromethyl ketone and was relatively unaffected by lactacystin andN-tosyl lysyl chloromethyl ketone. In the presence of hemin, polyubiquitinated CFTR remained tightly associated with ER microsomes. However, membrane-bound ubiquitinated CFTR could be subsequently degraded into trichloroacetic acid-soluble fragments upon incubation in hemin-free, ATP-containing lysate. Thus ER-associated degradation of CFTR occurs via a membrane-bound, rather than cytosolic, intermediate and likely involves recruitment of degradation machinery to the ER membrane. Our data suggest a model in which the degradation of polytopic proteins such as CFTR is coupled to retrograde translocation and removal of the polypeptide from the lipid bilayer.

The ubiquitin-proteasome pathway has been implicated in the degradation of newly synthesized, misfolded and unassembled proteins in the endoplasmic reticulum (ER). Using a cell-free reticulocyte lysate system we have examined the relationship between biosynthesis and ER-associated degradation of the cystic fibrosis transmembrane conductance regulator (CFTR), a polytopic protein with 12 predicted transmembrane segments. Our results provide direct evidence that fulllength, glycosylated and membrane-integrated CFTR is a substrate for degradation and that degradation involves polyubiquitination and cytosolic proteolytic activity. CFTR ubiquitination was both temperature-and ATP-dependent. Degradation was significantly inhibited by EDTA, apyrase, and the proteasome inhibitors hemin and MG132. Degradation was inhibited to a lesser extent by clasto-lactacystin ␤-lactone, ALLN, and N ␣tosyl-L-phenylalanine chloromethyl ketone and was relatively unaffected by lactacystin and N-tosyl lysyl chloromethyl ketone. In the presence of hemin, polyubiquitinated CFTR remained tightly associated with ER microsomes. However, membrane-bound ubiquitinated CFTR could be subsequently degraded into trichloroacetic acid-soluble fragments upon incubation in heminfree, ATP-containing lysate. Thus ER-associated degradation of CFTR occurs via a membrane-bound, rather than cytosolic, intermediate and likely involves recruitment of degradation machinery to the ER membrane. Our data suggest a model in which the degradation of polytopic proteins such as CFTR is coupled to retrograde translocation and removal of the polypeptide from the lipid bilayer.
The rough endoplasmic reticulum (ER) 1 facilitates translocation, membrane integration, folding, and oligomeric assembly of most proteins found in the secretory pathway of eukary-otic cells (1)(2)(3)(4). The ER also functions to recognize misfolded or unassembled proteins and prevent their exit to the Golgi complex (5-7) through a process termed ER-associated degradation (ERAD) (8). Substrates for ERAD, like other secretory and transmembrane proteins, are initially targeted to and translocated across the ER membrane via the Sec61 translocon complex (1,2). However, aberrantly folded proteins are selectively translocated back into the cytosol where they undergo ubiquitination and degradation by cytosolic proteases (9 -11). Certain ERAD substrates such as major histocompatibility complex class I heavy chains (10,12,13) and unglycosylated prepro-␣ factor undergo complete retrograde translocation into a soluble, cytosolic form prior to degradation (8). In contrast, mutant forms of carboxypeptidase Y and unassembled T cell receptor ␣ subunits translocate back to the cytosol but remain associated with cytoplasmic face of the ER membrane (14,15). In both cases the generation of cytosolic intermediates is thought to occur via retrograde transport through the Sec61 translocon complex (9 -11).
The ubiquitin-proteasome pathway is also involved in the ER-associated degradation of polytopic proteins. Yeast mutants with defective ubiquitin-conjugating enzymes or proteasome subunits show decreased degradation of mutant forms of Sec61p (16,17), a resident ER protein with 10 predicted transmembrane segments (18). ERAD may also play a role in ensuring proper protein levels in the ER as suggested by mevalonatemediated regulation of 3-hydroxy-3-methylglutaryl-coenzyme A reductase (19 -21). For the cystic fibrosis transmembrane conductance regulator (CFTR), a plasma membrane protein, approximately 80% of wild type protein and 100% of common mutant forms are degraded prior to reaching the Golgi complex (22,23). Recent studies have demonstrated that ER-associated degradation of CFTR also requires an intact ubiquitin conjugation pathway and likely involves the 26 S proteasome (24,25). While the underlying reasons for CFTR degradation remain unknown, it has been proposed that the majority of wild type CFTR, like mutant CFTR, is recognized by ER quality control machinery, because it fails to fold properly (26 -29).
Degradation of polytopic proteins by cytosolic proteases poses a particular problem in that multiple transmembrane (TM) segments must be removed from the membrane. It is unclear whether polytopic proteins, like secretory and bitopic transmembrane proteins, proceed through cytosolic intermediates prior to degradation. One possibility is that the proteasome might degrade only cytosolic peptide loops, leaving lumenal and/or transmembrane segments to be degraded by other proteases (7,30). Alternatively, degradation might be coupled to the process of retrograde translocation. To distinguish among these possibilities, it will be necessary to define the temporal relationship between recognition, retrograde translocation, and degradation events involved in polytopic protein ER * This work was supported by Grants GM53457 and DK51818 from the National Institutes of Health and by the North American Cystic Fibrosis Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Present address: Division of Molecular Medicine, 3181 SW Sam Jackson Park Rd., NRC-3, Oregon Health Sciences University, Portland, OR 97201. Tel.: 503-494-7322; Fax: 503-494-7368; E-mail: skachw@ohsu.edu. 1 The abbreviations used are: ER, endoplasmic reticulum; ERAD, ER associated degradation; CFTR, cystic fibrosis transmembrane conductance regulator; DTT, dithiothreitol; HMW, high molecular weight; RRL, rabbit reticulocyte lysate; ALLN, N-acetyl-L-leucyl-L-leucinyl-norleucinal; TPCK, N ␣ -tosyl-L-phenylalanine chloromethyl ketone; TLCK, Ntosyl lysyl chloromethyl ketone; TM, transmembrane; PAGE, polyacrylamide gel electrophoresis. degradation. In this regard, a cell-free system that reconstitutes ERAD is attractive, because it is readily amenable to physical and pharmacological manipulation. Using such a system, Sato et al. (31) recently demonstrated that CFTR could be ubiquitinated cotranslationally, indicating that initial recognition events in ER degradation might occur very early in the lifetime of a polytopic protein. In the current study we use a similar cell free system to demonstrate that CFTR ubiquitination can also occur after protein synthesis, N-linked glycosylation, and membrane integration have been completed. We have also been able to separate the process of ubiquitination from that of degradation and now show that ubiquitinated CFTR remains tightly associated with the ER membrane until it is degraded into trichloroacetic acid-soluble peptide fragments. CFTR degradation is therefore physically localized to the ER. Our results are consistent with a model in which cytosolic degradation machinery is recruited to the ER membrane by polyubiquitinated substrates and suggest that degradation occurs coincident with polypeptide removal from the lipid bilayer.
Rabbit reticulocyte lysate for translation was prepared from fresh rabbit reticulocytes induced by acetylphenylhydrazine treatment as described previously (32) and stored at Ϫ80°C. Prior to freezing, hemin was added to a final concentration of 40 M. For degradation experiments, RRL was either purchased from Green Hectare Farm (Oregon, WI) or prepared as described previously (32) but without addition of hemin. Although degradation efficiency varied between different preparations of RRL, no consistent differences were observed between different RRL sources. To prepare ATP-depleted RRL, reticulocytes were washed three times with ice-cold phosphate-buffered saline, resuspended in Krebs-Ringer phosphate medium lacking glucose, and incubated in 0.2 mM 2,4-dinitrophenol and 20 mM 2-deoxyglucose at 37°C for 90 min (33). Reticulocytes were then washed twice with phosphatebuffered saline and lysed by adding 1.5 volume of 1.0 mM DTT. Samples were centrifuged at 15,000 rpm for 30 min (Beckman JA17 rotor), and the supernatant (i.e. RRL) was stored at Ϫ80°C.
Methylated ubiquitin was prepared by reductive methylation as described previously (34). Bovine ubiquitin was incubated at room temperature for 20 h in 12 mM formaldehyde and 20 mM cyanoborohydride followed by an additional hour in fresh reagents. The reaction was dialyzed for 48 h in 10 mM HEPES, 5 mM MgCl 2 , 3 mM DTT and concentrated to 10 mg/ml using a centricon (number 3) filter.
Carbonate Extraction-Translation products were incubated in 0.1 M NaC0 3 (pH 11.5) or in 0.25 M sucrose, 0.1 M Tris (pH 7.5) for 30 min on ice as described (37). Samples were centrifuged at 180,000 ϫ g for 30 min, and membrane pellets were dissolved directly in SDS loading buffer. Supernatants were precipitated in 20% trichloroacetic acid, pelleted at 16,000 ϫ g for 15 min, washed with acetone, and dissolved in SDS loading buffer.
Immunoprecipitation-RRL aliquots were diluted directly into 1 ml of ice-cold Buffer A (0.1 M NaCl, 1% Triton X-100, 2 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, and 0.1 M Tris (pH 8.0)) and preincubated with antisera for 10 -30 min prior to addition of 5 l of protein A Affi-Gel (Bio-Rad). CFTR antisera and nonimmune antisera were used at 1:1000 dilution. Ubiquitin antibody was used at 1:200 dilution (similar amounts of Ig were present in all reactions). Samples were mixed at 4°C for 6 -10 h, washed three times with Buffer A, twice with 0.1 M NaCl, 0.1 M Tris (pH 8.0), and taken up in SDS loading buffer.
Degradation Assay-The translation mixture was layered onto a 0.5 M sucrose buffer (0.5 M sucrose, 50 mM HEPES (pH 7.5), 0.1 M KCl, 5 mM MgCl 2 , and 1 mM DTT) and centrifuged at 180,000 ϫ g for 10 min. Membranes were resuspended in ice-cold 0.1 M sucrose buffer (0.1 M sucrose, 50 mM HEPES (pH 7.5), 0.1 M KCl, 5 mM MgCl 2 , and 1 mM DTT) at one-half the original translation volume. The degradation reaction, assembled on ice, included 20% resuspended membranes, 72% reticulocyte lysate, and a final concentration of 10 mM Tris (pH 7.5), 1 mM ATP, 5 mM MgCl 2 , 12 mM creatine phosphate, 3 mM DTT, and 80 g/ml of creatine kinase. Where indicated, RRL was replaced with water, and proteasome inhibitors were added. For ATP-depleted RRL, ATP and creatine phosphate were omitted. The degradation mixture was incubated at 37°C, and aliquots were added directly to loading buffer for SDS-PAGE, EnHance (NEN Life Sceince Products) fluorography and autoradiography. Autoradiograms were scanned on an AGFA studioscan II transmission scanner using Adobe Photoshop software. Quantitation of autoradiograms was performed using an Amersham Pharmacia Biotech DTS scanning densitometer precalibrated with a Kodak photographic step tablet (Eastman Kodak Co.). Trichloroacetic acid-soluble counts were quantitated by precipitating degradation mixture in 20% trichloroacetic acid, incubating on ice for 30 min, pelleting at 16,000 ϫ g for 15 min, and counting supernatants in 10 volumes of Ready Safe scintillation fluid (Beckman, Fullerton, CA) in a Beckman LS6500 scintillation counter.
Vesicle Pelleting and Floatation-Microsomal membranes from translation mixture were pelleted, resuspended, and incubated in ATPcontaining RRL in the presence of 40 M hemin as described above. For pelleting experiments, aliquots of this RRL/hemin mixture were taken at specified time points; KCl or Triton X-100 was added as indicated; the mixture was incubated on ice for 30 min and pelleted through 0.

RESULTS
In Vitro CFTR Biogenesis-To examine the steps of CFTR degradation in native ER membranes, we first characterized expression of full-length CFTR protein in a cell-free RRL system. Translation of plasmid pSPCFTR (35) in the absence of ER-derived microsomal membranes generated a polypeptide migrating at ϳ160 kDa (Fig. 1A). This translation product shifted 6 kDa in size when rough microsomes were added at the start of translation, consistent with the addition of two Nlinked core carbohydrates at the predicted consensus sites in the fourth extracytoplasmic loop. The presence of full-length CFTR was confirmed by immunoprecipitation with peptidespecific antisera raised against N terminus, R domain, and C terminus epitopes. Carbonate extraction of translation products demonstrated that in vitro synthesized CFTR was fully integrated into the ER membrane. Thus the RRL system is capable of efficiently reconstituting the early events of CFTR biosynthesis and processing, in agreement with results reported by Sato et al. (31). To test the stability of newly synthesized CFTR in the ER membrane, cyclohexamide was added after 2 h of translation, and samples were incubated for an additional 4 h (Fig. 1B). At 24°C, CFTR was remarkably stable, whereas at 37°C CFTR was rapidly converted into a high molecular weight (HMW) complex with an estimated size of Ͼ450 kDa (T1 ⁄2 Ͻ30 min). For convenience, this large complex will subsequently be referred to as the HMW complex, and the ϳ170 kDa band will be referred to as full-length CFTR.
ATP-dependent Conversion of CFTR into a HMW Complex-To determine the requirements for conversion of CFTR into the HMW complex, CFTR was translated in RRL, and microsomal membranes were collected by pelleting through a sucrose cushion ( Fig. 2A). Microsomes were then resuspended and incubated at 37°C in sucrose buffer containing ATP (lanes 1-5), RRL supplemented with an ATP regeneration system (lanes 6 -10), ATP-depleted RRL (lanes 11-15), or ATP-depleted RRL supplemented with additional ATP (lanes [15][16][17][18][19][20]. At specified time points samples were analyzed directly by SDS-PAGE. In the presence of buffer alone, or in ATP-depleted RRL, Ͼ70% of full-length CFTR remained intact after 4 h of incubation (Fig. 2B). However, when microsomal membranes were incubated in fresh RRL containing ATP, or in ATP-depleted RRL supplemented with additional ATP, full-length CFTR was rapidly converted into the HMW complex (T1 ⁄2 Ͻ 30 min) that subsequently disappeared over the next 4 h. These results are consistent with a two-step process in which full-length CFTR is initially converted into a HMW complex via an ATP-dependent mechanism, and the HMW complex is then a substrate for degradation.
The CFTR HMW Complex Is Polyubiquitinated-We noted that degradation of the CFTR HMW complex in Fig. 2 contrasted with results in Fig. 1B where the complex accumulated over time. One difference between these two experiments is were incubated at pH 7.5 (Tris) or at pH 11.5 (Carb), and supernatant (S) and pellet (P) fractions were analyzed directly. B, CFTR translation was carried out as in A except that aurin tricarboxylic acid (0.1 mM) and cyclohexamide were added 15 min and 2 h after the start of translation, respectively. Following addition of cyclohexamide, aliquots were removed at the specified time points and analyzed directly by SDS-PAGE. Fig. 1B), after which microsomal membranes were pelleted through a 0.5 M sucrose cushion and resuspended in 0.1 M sucrose buffer as described under "Experimental Procedures." Membranes were then incubated at 37°C in buffer (lanes 1-5), RRL containing an ATP regeneration system (lanes 6 -10), ATP-depleted RRL (lanes 11-15), or ATP-depleted RRL containing an ATP regeneration system (lanes 16 -20). At specified time points, aliquots were removed from each tube and analyzed directly by SDS-PAGE. B, autoradiograms were digitized, and band intensities of full-length CFTR and the HMW complex were quantitated, normalized to the amount of full-length CFTR present at T ϭ 0, and plotted for each time point. Closed and open symbols represent CFTR and the HMW complex, respectively. that CFTR translation is performed in the presence of hemin, a required co-factor for protein expression in RRL (38), and aurin tricarboxylic acid, an inhibitor of protein synthesis initiation. Both of these compounds inhibit ATP-dependent degradation of ubiquitinated substrates in RRL. These results, together with the requirement for ATP and cytosol, suggested that the HMW complex might represent polyubiquitinated CFTR. Microsomal membranes containing newly synthesized CFTR were therefore incubated in fresh RRL supplemented with bovine ubiquitin and 40 M hemin. Aliquots were then immunoprecipitated with anti-bovine ubiquitin antisera and analyzed by SDS-PAGE at specified time points. As shown in Fig. 3A, the HMW complex, but not full-length CFTR, was reactive against antiubiquitin antisera (lanes [11][12][13][14][15]. No reactivity was observed when immunoprecipitation was performed with nonimmune sera (NIS, lanes 6 -10) or when bovine ubiquitin was omitted from the lysate (data not shown). To further confirm CFTR ubiquitination, microsomes containing newly synthesized CFTR were incubated in RRL in the presence of methylated ubiquitin, which blocks the sequential addition of ubiquitin molecules by preventing isopeptide bond formation at ubiquitin residue Lys-48. Under these conditions we observed a dose-dependent inhibition in the rate of HMW complex formation (Fig.  3B). Taken together, these data demonstrate that the sequential, covalent addition of multiple ubiquitin molecules contributes to the formation of the CFTR HMW complex.

FIG. 2. In vitro degradation of CFTR via a HMW intermediate. A, CFTR was translated for 2 h in RRL (as in
Inhibition of CFTR HMW Complex Degradation by Proteasome Inhibitors-We next examined the ability of protease inhibitors to influence the disappearance of the CFTR HMW complex. Inhibitors of serine, cysteine, and metalloproteases had essentially no effect on the rate of disappearance of either full-length CFTR or the HMW complex (Fig. 4A). In contrast, hemin and MG132, both known inhibitors of the proteasome, resulted in significant delay in the disappearance of the HMW complex particularly at early time points (Fig. 4B). Quantitation of autoradiograms revealed that hemin effectively blocked 60% of HMW complex degradation. In contrast, MG132, ALLN, and TPCK slowed but did not prevent HMW complex disap-pearance. Surprisingly, little inhibition was observed for lactacystin or TLCK. These results suggested either that certain proteasome inhibitors might be relatively inactive in the RRL system, or alternatively, that degradation of ubiquitinated CFTR might involve additional protease activities. To distinguish between these possibilities, we therefore examined the effect of proteasome inhibitors on the degradation of methylated lysozyme, a known substrate of the 26 S proteasome. For these and subsequent experiments the rate and extent of protein degradation was quantitated by measuring the conversion of CFTR and/or lysozyme into trichloroacetic acid-soluble peptide fragments.
As shown in Fig. 5, 26% of [ 14 C]lysozyme and 65% of newly synthesized CFTR were degraded into trichloroacetic acid-soluble peptide fragments within 3 h in RRL. Lysozyme degradation was inhibited by approximately 90% in the presence of 40 M hemin or 10 mM EDTA consistent with previous studies (39). Hemin also resulted in formation of a HMW complex containing [ 14 C]lysozyme (data not shown). In contrast, the specific proteasome inhibitor lactacystin, as well as its active metabolite clasto-lactacystin ␤-lactone, had little effect on lysozyme degradation (Ͻ10% inhibition). Under these same conditions CFTR degradation was inhibited by approximately 80% in the presence of EDTA or the ATP hydrolyzing enzyme, apyrase. Hemin inhibited CFTR degradation by 56%, while lactacystin and clasto-lactacystin ␤-lactone inhibited CFTR degradation by only 20 and 35%, respectively. In these latter experiments, lactacystin-based inhibitors were preincubated with RRL for 30 -60 min prior to addition of CFTR-containing microsomal membranes. These experiments demonstrate that specific lactacystin-based proteasome inhibitors are relatively inactive at blocking degradation of these ubiquitinated substrates in RRL. It should also be noted that a significant amount of CFTR degradation in RRL (35-40%) appears to proceed through a hemin insensitive pathway distinct from that utilized by lysozyme.
Ubiquitinated CFTR Remains Tightly Associated with the ER Membrane-Sixty-five percent of in vitro synthesized CFTR protein was degraded to trichloroacetic acid-soluble fragments within 3 h of incubation in RRL (Fig. 5). Degradation could not be restricted solely to large cytosolic peptide loops (i.e. N terminus, nucleotide binding domains, and the R-domain), as these loops contain only 47% of total methionine residues (40). Rather, our data indicate that proteolytic machinery also gains access to small cytosolic loops and possibly lumenal and/or transmembrane regions of the protein. Thus prior to or coincident with degradation, some or all of CFTR is likely to be extracted from the membrane. We therefore tested whether the ubiquitinated HMW complex, which has already been targeted for degradation by ubiquitin-conjugating enzymes, might undergo retrograde translocation to a cytosolic intermediate. Ubiquitinated CFTR complexes were generated by incubation in RRL (in the presence of hemin), and vesicles were collected by pelleting through a sucrose cushion. As shown in Fig. 6A, both CFTR and the CFTR HMW complex were nearly quantitatively recovered in the microsome pellet in the absence of detergent. In addition, preincubation of CFTR-containing microsomes in high salt (up to 1.1 M KCl) failed to extract fulllength CFTR or the HMW complex into the supernatant fraction (Fig. 6B). Thus ubiquitinated CFTR does not behave as a cytosolic or peripheral membrane protein.
Results from Fig. 6 do not rule out the possibility that HMW ubiquitinated CFTR might also be pelleted as a large, detergent-sensitive protein complex. We therefore tested whether the HMW complex was physically associated with the ER by membrane floatation. Aliquots of RRL containing newly syn-  , lanes 11-15), or analyzed directly (A, lanes 1-5 and B, all lanes). thesized CFTR and/or polyubiquitinated CFTR were diluted into 2 M sucrose, layered beneath a discontinuous sucrose gradient, and centrifuged for 2 h as described under "Experimental Procedures" (Fig. 7A). For half of the samples, ribosomes were first stripped from membranes by incubation in KCl/ EDTA (Fig. 7B). The secretory protein bovine prolactin (41) was used as a standard to determine the location of microsomal membranes in the gradient. Both full-length CFTR and the polyubiquitinated complex exhibited the same floatation pattern as the secretory control. Thus in the absence of degradation, CFTR remained membrane-associated long after ubiquitination had occurred. Following ribosome stripping with EDTA/ KCl microsomes more uniformly partitioned into the lighter fractions. However ribosome removal had no detectable effect on dislocation of ubiquitinated CFTR from the ER membrane.
Degradation of Membrane-bound, Ubiquitinated CFTR-To test whether membrane-bound CFTR remained a substrate for degradation and not simply a protease resistant aggregate, CFTR-containing microsomes were incubated in RRL in the presence of hemin to allow accumulation of the ubiquitinated intermediate. Vesicles were isolated at sequential time points, reincubated in the presence or absence of fresh RRL, and deg-radation of CFTR was monitored by release of trichloroacetic acid-soluble counts. As shown in Fig. 8A, 70% of total CFTR was degraded into trichloroacetic acid soluble counts following 3 h incubation in ATP-containing RRL. When vesicles were collected after 20 min of preincubation in hemin-containing RRL and resuspended in fresh RRL lacking hemin, 56% of the total CFTR was subsequently degraded. This degradation remained sensitive to both hemin and EDTA. Incubation of CFTR in hemin-containing RRL for longer periods of time (i.e. 1-2 h) resulted in the near complete conversion (ϳ90%) of full-length protein into the HMW complex as in prior experiments (data not shown). Even after 2 h of preincubation, 57% of total CFTR protein was still able to be degraded upon removal of hemin. In each of these experiments, the degradation of polyubiquitinated CFTR was thus 80% as efficient as the degradation of newly synthesized protein. Furthermore, with longer preincubation times the reversible component of CFTR degradation became progressively more sensitive to inhibition such that after 2 h of preincubation, hemin and EDTA each blocked degradation of polyubiquitinated CFTR by approximately 90%. When microsomes containing pre-ubiquitinated CFTR were similarly resuspended and incubated in the presence of buffer alone (Fig. 8B), no degradation of the HMW polyubiquitinated CFTR was observed.
Together these results demonstrate that: (i) the inhibitory effects of hemin on CFTR degradation are reversible, (ii) most, but not all, of the membrane-bound ubiquitinated CFTR remains a substrate for degradation, (iii) CFTR degradation, like ubiquitination, requires the presence of both ATP and cytosolic factors, and (iv) degradation is physically localized to the ER membrane and likely occurs coincident with removal of at least a portion of CFTR from the lipid bilayer. DISCUSSION The ER is a primary site of cellular quality control involved in the recognition and disposal of misfolded and unassembled proteins (5). While it was initially proposed that ER proteases were responsible for degrading ER substrates, recent evidence indicates that ERAD is mediated largely by cytosolic proteases including the 26 S proteasome (6,7). This topologic disparity between substrate and degradation machinery has resulted in a reevaluation of protein movement into and out of the ER compartment (6). It now appears that translocation across the ER membrane is a bidirectional process that is regulated, in part, by the folded state of a given protein substrate. Whereas, nascent secretory and transmembrane proteins are targeted to, translocated across, and/or integrated into the ER membrane by ER translocation machinery (2,42), proteins that fail to fold properly may also be recognized by ER quality control machinery and transported out of the ER and back to the cytosol (6,9,10,15). These observations have raised important questions regarding the process of retrograde transport across the ER membrane. At what stage of biogenesis are protein substrates identified? How do recognition events target proteins back to the translocon and/or facilitate movement into the cytosol? And how is this retrograde transport coupled to the process of degradation? In the current study, we address these issues by developing an in vitro system that reconstitutes de novo synthesis and ER-associated degradation of the polytopic protein CFTR. Full-length, glycosylated and membrane-integrated CFTR was a substrate for ATP-dependent ubiquitination. In addition, polyubiquitinated CFTR remained tightly associated with the ER membrane until it was degraded into trichloroacetic acid-soluble peptide fragments. Thus ubiquitination precedes retrograde translocation, and degradation of polytopic proteins such as CFTR is physically localized to the ER membrane rather than the cytosol as proposed for some secretory and transmembrane substrates (8,10,13).
Our results are consistent with a model in which CFTR degradation is tightly coupled to removal of the polypeptide from the lipid bilayer. At least 65% of [ 35 S]methionine-labeled CFTR was degraded into trichloroacetic acid-soluble fragments, consistent with proteasomal degradation in which the resulting polypeptides are 6 -9 residues in length (43). Because a small fraction of full-length CFTR (approximately 10 -15%) remained intact in nearly all experiments, the extent of degradation for individual CFTR polypeptides actually approached 80%. Thus CFTR degradation could not be limited to large cytosolic regions containing the N terminus, nucleotide binding domains, and R domain, which together comprise 61% of the total mass but only 47% of methionine residues (40). If all cytosolic loops within 8 -15 residues from the membrane were cleaved, then degradation of all full-length CFTR polypeptide chains would liberate only 60% total CFTR methionine into the trichloroacetic acid-soluble fraction (40). It therefore seems likely that proteolytic machinery such as the proteasome might participate in removal of CFTR from the lipid bilayer to gain access to transmembrane segments and/or lumenal peptide regions. This possibility is supported by the observation that: (i) proteasome-like particles are physically localized to the cytosolic surface of the ER membrane (44) and (ii) our results that CFTR degradation occurs via a membrane-bound, rather than cytosolic, intermediate. If retrograde translocation of CFTR occurred through the Sec61 translocon, then extraction from the lipid bilayer could potentially involve the reverse process of membrane integration, namely lateral movement of TM segments from the membrane back into the translocation channel and into the cytosol (45,46).
While this study provides additional support that the ubiquitin-proteasome pathway is involved in ER-associated degradation of CFTR, it is unclear why different proteasome inhibitors exhibited different effects on CFTR degradation. Lactacystin and clasto-lactacystin ␤-lactone, which covalently inactivate the active site threonine residue on the ␤ 5 subunit of the 20 S proteasome (subunit X of the mammalian proteasome) FIG. 5. Degradation of lysozyme and CFTR. Degradation mixtures were assembled as described in the legend to Fig. 2 and added to resuspended microsomal membranes containing CFTR or [ 14 C]lysozyme (0.5 Ci/ml). Hemin (40 M), EDTA (10 mM), or apyrase (0.1 unit/l) was added at T ϭ 0 as indicated. Lactacystin (100 M) or clasto-lactacystin ␤-lactone (100 M) were preincubated with RRL for 30 -60 min prior to assembling the reaction. Aliquots were precipitated in 20% trichloroacetic acid, and soluble counts were calculated by the following formula: % counts released ϭ (C n Ϫ C 0 )/(T Ϫ C 0 ), where C n ϭ soluble counts at each time point, C 0 ϭ soluble counts at T ϭ 0, and T ϭ total counts per aliquot. Results represent the average (ϮS.E.) of two to six separate experiments. (47,48), had relatively minor effects on CFTR degradation as determined either by disappearance of the CFTR HMW complex (Fig. 4) or generation of trichloroacetic acid-soluble fragments (Fig. 5). It is interesting that clasto-lactacystin ␤-lactone, the active form of lactacystin, is inactivated by glutathione in cells (49,50). Our observation that neither lactacystin nor its ␤-lactone form inhibited the degradation of lysozyme, a known proteasome substrate, suggests that RRL is capable of rendering these agents inactive, even when preincubated in the absence of exogenous reducing agents. We do not know why the results reported here differ from those of Qu et al. (51) that demonstrated 20 M lactacystin inhibited the degradation of both ␣ 1 -antitrypsin and lysozyme in a similar RRL system. However, our results were consistent for two different preparations of lactacystin and at multiple inhibitor concentrations up to 100 M. It is possible that these differ-FIG. 6. HMW complex pelleting. CFTR-containing microsomes were incubated in RRL in the presence of hemin. A, at times specified, aliquots were analyzed directly (lanes 1-4), pelleted through 0.5 M sucrose buffer (lanes 5-9), or incubated on ice in 1% Triton X-100 for 30 min prior to pelleting (lanes 10 -14). Pellets were dissolved directly into SDS loading buffer. B, aliquots of RRL degradation mixture were taken at specified times and analyzed directly in SDS loading buffer (lanes 1-3) or incubated for 30 min on ice following addition of KCl to the final concentrations indicated (lanes 4 -21). Samples were then pelleted through 0.5 M sucrose buffer, and equal fractions of pellet and supernatant were analyzed by SDS-PAGE.
FIG. 7. HMW complex floatation. Microsomal membranes containing newly synthesized CFTR or bovine prolactin were incubated at 37°C in RRL degradation mixture for 0, 20, or 120 min in the presence of hemin. A, samples were mixed to a final concentration of 2 M sucrose, layered beneath a discontinuous sucrose gradient, and centrifuged at 350,000 ϫ g for 2 h as described under "Experimental Procedures." B, EDTA (10 mM) and KCl (0.6 M) were added prior to centrifugation. Aliquots were taken from top to bottom of the gradient (lanes 1-10, respectively). The HMW complex, CFTR, and prolactin control are indicated. ences reflect varying amounts of proteasome in different RRL preparations.
The most potent inhibitors of CFTR degradation were EDTA, apyrase, and hemin. EDTA and hemin have been shown previously to inhibit the degradation of other proteasome substrates in RRL by stabilizing HMW ubiquitin conjugates (39,52,53). Hemin, a heme precursor, exerts pleiotropic effects in RRL. Its stimulatory effect on protein translation (38) is proposed to be mediated by inactivation of the translation factor eif-2, possibly by inhibiting an endogenous RRL kinase (54,55). In addition, hemin directly inhibits ATPase activities of the purified 19 S proteasome regulatory complex (PA700) as well as the 26 S proteasome itself (56). Because the catalytic core of the 20 S proteasome cannot accommodate folded or ubiquitinated polypeptides, it has been proposed that ATPases contained within the 19 S regulatory subunit are required for substrate unfolding as well as deubiquitination (43, 48, 56 -59). It is therefore likely that hemin inhibits proteasome-mediated CFTR degradation by blocking ATPase activity and thereby preventing CFTR unfolding and/or deubiquitination. If CFTR unfolding were required for removal from the bilayer, such a mechanism would explain the accumulation of membranebound, polyubiquitinated CFTR observed in the current study.
Our results do not rule out the possibility that additional peptidase activities either in the ER lumen or within the membrane also participate in CFTR degradation as has been suggested previously (25,60). One such candidate is signal peptidase complex, an ER-localized protease that is capable of cleaving transmembrane proteins (61) and free signal peptides (30) at sites located within the plane of the lipid bilayer. Interestingly, CFTR contains cryptic signal peptidase cleavage sites within or adjacent to TM segments of the N terminus hydrophobic domain (62,63). If similar proteases facilitated CFTR degradation, then TM segments and small peptide loops could potentially be released from the membrane. Such a mechanism might explain why a significant amount of CFTR degradation was ATP-independent and hemin insensitive. It is also possible that the membrane-bound HMW complex might actually consist of polyubiquitinated CFTR cleavage products that are then degraded by the ubiquitin-proteasome pathway.
Cell-free systems have significantly enhanced our understanding of protein folding (64) and translocation (2). However we do not yet know whether the folded state of newly synthesized CFTR in the RRL system is the same as that found in the ER of cells or whether the mechanism of CFTR recognition by ER quality control machinery in vitro is the same as in vivo. Processing in the RRL system is limited to early events that occur at the ER membrane, and it is possible that CFTR might require specialized cellular components that are limiting in RRL and/or canine pancreas microsomes. Recently, RRL and yeast-derived cell-free systems have been used to examine the degradation of other secretory and transmembrane proteins (8,51), including CFTR (31). Our observation that 85-90% of newly synthesized wild type CFTR is degraded in the RRL system is consistent with studies that demonstrate most wild type CFTR is also degraded in the ER membrane of intact cells (22,23). In addition, preliminary experiments in our laboratory have demonstrated that other polytopic proteins expressed in the RRL system, such as human P-glycoprotein and aquaporin water channels, are much more stable than CFTR. 2 These studies indicate that rapid CFTR degradation in vitro is not simply a cell free artifact, but rather reflects specific structural features recognized by ER quality control machinery. In vitro systems such as RRL will thus likely provide a versatile tool for investigating the events and cellular machinery involved in 2 E. Chong and W. R. Skach, unpublished observations. FIG. 8. Degradation of preubiquitinated CFTR. CFTR-containing microsomal membranes were incubated in RRL in the presence of hemin as indicated. A, at each time point, samples were pelleted through 0.5 M sucrose buffer and resuspended in fresh RRL in the presence or absence of hemin or EDTA. Aliquots were precipitated in 20% trichloroacetic acid at the indicated times, and percent soluble counts released was determined by scintillation counting. B, samples were pelleted through 0.5 M sucrose buffer in the presence or absence of 1 mM ATP and resuspended in 10 mM Tris, 5 mM MgCl 2 , 3 mM DTT, and 80 g/ml creatine kinase, with or without 1 mM ATP/12 mM creatine phosphate. Time scale for each graph is based on T ϭ 0 for the starting RRL degradation mixture. For T ϭ 20, 60, and 120 min graphs, the base line soluble counts reflect the percent of counts released from RRL in the presence of hemin. Total trichloroacetic acid-soluble counts plotted therefore represent the counts released during initial incubation plus the counts released following resuspension and reincubation of microsomes.
ER-associated degradation. In the case of CFTR, they may also provide screening assays to identify compounds that improve folding and trafficking efficiency.