Cytosolic degradation of T-cell receptor alpha chains by the proteasome.

The T-cell antigen receptor (TCR) is an hetero-oligomeric membrane complex composed of at least seven transmembrane polypeptide chains that has served as a model for the assembly and degradation of integral membrane proteins in the endoplasmic reticulum (ER). Unassembled TCRalpha chains fail to mature to the Golgi apparatus and are rapidly degraded by a non-lysosomal "ER degradation" pathway that has been proposed to be autonomous to the ER. In these studies we show that the degradation of core-glycosylated TCRalpha is blocked by N-acetyl-L-leucyl-L-leucyl-L-norleucinal (ALLN) and lactacystin, implicating the proteasome in ER degradation. Either acute or chronic treatment of TCRalpha-transfected cells with proteasome inhibitors cause the core-glycosylated TCRalpha chains to progressively shift to an approximately 28-kDa form that lacks N-linked oligosaccharides and the N-terminal signal peptide. The susceptibility of this 28-kDa species to extravesicular protease indicates that it is not protected by the ER membrane and, hence, cytoplasmic. These data suggest a model in which TCRalpha chains that are translocated across the membrane, core-glycosylated, but fail to assemble are dislocated back to the cytoplasm for degradation by cytoplasmic proteasomes. Our data also suggest that covalent modification of TCRalpha with ubiquitin is not required for its degradation.

The T-cell antigen receptor (TCR) 1 is a hetero-oligomeric complex of at least seven polypeptide chains that has served as a model for the assembly and degradation of integral membrane proteins in the ER (1). The clonotypic ␣ subunit (TCR␣) is a type I membrane protein containing a short (ϳ5-amino acid) cytoplasmic domain and a 223-residue extracellular domain that has four potential sites for N-glycosylation. Mature TCR␣ on the surface of the antigen-specific T-cell hybridoma line 2B4 migrates as a broad 42-44-kDa band (2)(3)(4). However, when expressed in the absence of other TCR subunits, TCR␣ is synthesized as a 38-kDa core-glycosylated precursor that is sensitive to digestion with endoglycosidase H and is rapidly degraded with a half-time of ϳ50 min (5,6). This degradation process is not affected by inhibitors of autophagy, lysosomal proteolysis, or ER-Golgi traffic. Moreover, TCR␣ chains in these cells are localized to the "ER region" by immunofluorescence and electron microscopy (5). Together, these studies have led to the conclusion that TCR␣ degradation occurs at a site "within or closely associated with the ER" (5). However, efforts to identify ER-specific proteases that participate in TCR␣ degradation have been unsuccessful.
Several recent reports have suggested a role for the proteasome in the ER degradation of some membrane or lumenal proteins (reviewed in Refs. 7 and 8). For example, misfolded cystic fibrosis transmembrane conductance regulator (CFTR) molecules that fail to exit the ER are rapidly degraded by a process that requires covalent modification with ubiquitin and is blocked by lactacystin, a specific proteasome inhibitor (9). Degradation of other ER-restricted proteins including mutant human ␣ 1 -antitrypsin (10), yeast carboxypeptidase Y (11), and MHC class I heavy chains (12,13) has also recently been shown to require proteasome activity. How these proteins, which are sequestered within the ER lumen, are recognized and delivered to the cytoplasmic proteasome complex is unknown.
In this paper we have examined the role of the ubiquitinproteasome pathway in the ER degradation of newly synthesized TCR␣ chains. Our data suggest a model in which TCR␣ chains are first translocated into the ER, cleaved by signal peptidase, and N-glycosylated with core high mannose glycans. These chains are subsequently exported back to the cytoplasmic face of the ER, where they are deglycosylated and delivered to the proteasome for degradation. Moreover, our data suggest that ubiquitination of TCR␣ is not required for this process.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfection-HEK293 cells were grown and transiently transfected by calcium phosphate precipitation as described previously (14). In some experiments, N-acetyl-L-leucinal-L-leucinal-Lnorleucinal (ALLN, calpain inhibitor I, Calbiochem) or lactacystin (a kind gift from S. Omura, Kitasato Institute, Tokyo) at indicated concentrations was added to the fresh media. The cDNA corresponding to 2B4 TCR␣ (15) in pCDM8 was a kind gift from J. Bonifacino (National Institutes of Health).
Site-directed Mutagenesis-The 11 lysine residues in TCR␣ were mutated to arginine by 8 sequential rounds of polymerase chain reaction-based "megaprimer" mutagenesis (16). The mutant construct (K␣R) was verified by sequence analysis, which revealed the presence of an additional mutation of Phe 154 to Tyr. Functional comparison with a K␣R lacking this additional mutation showed that this conservative change does not influence either the kinetics of degradation or its sensitivity to proteasome inhibitors.
Immunoblotting, Metabolic Labeling, and Immunoprecipitation-HEK293 cells transiently transfected with TCR␣ were processed for immunoblot analysis as described previously (9). Samples were resolved in 11% SDS-polyacrylamide gels and electroblotted to nitrocellulose. Blots were probed with the appropriate antibody, and immunoreactive bands were detected by enhanced chemiluminescence. Metabolic labeling and immunoprecipitation was carried out as described (14) with the following modifications. Cells were pulse-labeled with 500 Ci/ml [ 35 S]Met/Cys (Ͼ1,000 Ci/mmol, NEN Life Science * This work was supported by National Institutes of Health Grant R01 DK43994. This work was done during the tenure of an established investigatorship of the American Heart Association (to R. R. K.). 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.
Cell-free Translation-Radiolabeled TCR␣ was synthesized in a coupled transcription and translation reaction using a T7-TNT kit (Promega) and 0.5 mCi/ml [ 35 S]Met (1000 Ci/mmol, Amersham). Canine pancreatic microsomes were prepared according to Walter and Blobel (19). TCR␣ translated in the presence or absence of microsomes in 50-l reactions were immunoprecipitated with mAb H28-710 and 50 l of GammaBind Sepharose beads.

Unassembled TCR␣ Chains Are Stabilized by Inhibitors of
the Proteasome-To determine if proteasomes play a role in the degradation of incompletely assembled TCR␣, we examined the effect of proteasome inhibitors on the steady state levels of TCR␣ in HEK cells transfected with cDNA encoding the 2B4␣ clonotype. Immunoblot analysis (Fig. 1A) reveals that these cells contain a predominant immunoreactive species with a mobility of 38 kDa, corresponding to the size of core-glycosylated TCR␣, as previously observed in transfected fibroblasts (5). A minor band with ϳ2-kDa faster mobility, probably corresponding to a partially glycosylated form of the protein (see below), was also detected occasionally. In the absence of proteasome inhibitors, the 38-kDa species was completely soluble in nonionic detergent. However, overnight treatment with the proteasome inhibitors ALLN or lactacystin increased the steady-state level of the 38-kDa band in the detergent-soluble fraction and also led to its appearance in the detergent-insoluble fraction. Strikingly, these proteasome inhibitors also induced the accumulation of a novel, detergent-insoluble 28-kDa band. Because the mobility of this band corresponds to the predicted mobility of core, unglycosylated TCR␣, we examined the effects of PNGase on the TCR␣ species which accumulated in proteasome-inhibited cells.
TCR␣ was immunoprecipitated from detergent-soluble and insoluble fractions of transfected HEK cells that had been metabolically labeled to steady state with [ 35 S]Met/Cys in the presence of ALLN (Fig. 1B). In this 15% acrylamide gel the 28-kDa species was resolved into a doublet of closely spaced bands, both of which were resistant to PNGase treatment, and thus, not glycosylated. The upper band of this doublet comigrates with TCR␣ translated in a cell-free protein synthesis extract in the absence of microsomes, indicating that it has an intact signal sequence. Together, these results suggest that translocation of TCR␣ is not completely efficient and that HEK and possibly other cells possess a proteasome-mediated degradation pathway that normally masks inefficiency in the process of ER translocation.
The lower band of the 28-kDa doublet comigrates with the limit product of PNGase-deglycosylated 38-kDa TCR␣, corre-sponding to the signal-cleaved form of TCR␣. 2 This suggests that some TCR␣ chains either fail to become glycosylated following translocation and signal peptide cleavage in the ER or that a fraction of glycosylated, signal-cleaved TCR␣ chains are deglycosylated in vivo.
To distinguish between these two possibilities, we examined the effect of proteasome inhibitors on the fate of newly synthesized TCR␣ (Fig. 2). Following a 10-min pulse with [ 35 S]Met/ Cys, TCR␣ migrated primarily as a detergent-soluble 38-kDa core-glycosylated protein that was rapidly degraded with a half-time of 65 min ( Fig. 2A). In contrast, core-glycosylated TCR␣ in cells treated with the proteasome inhibitors ALLN (Fig. 2B) or lactacystin (Fig. 2C) was markedly stabilized. Both proteasome inhibitors induced the formation of bands corresponding to partially glycosylated TCR␣ intermediates and a 28-kDa species which accumulated over time in the detergentinsoluble fraction. Together, these data suggest that degradation of newly synthesized TCR␣ by the proteasome is preceded by progressive deglycosylation of the core-glycosylated protein. TCR␣ Chains Are Dislocated from the ER-We used cell fractionation and protease protection to test the possibility that TCR␣ degradation by proteasomes is associated with its dislocation from the ER to the cytoplasm. Cells were lysed by mechanical disruption,and the post-nuclear supernatant was centrifuged at 100,000 ϫ g. A small amount (Ͻ5%) of TCR␣ (both the 38-kDa and the 28-kDa forms) was recovered in the supernatant, even after a second round of 100,000 ϫ g centrifugation, suggesting that some TCR␣ had been released to the cytosolic fraction (data not shown). However, the majority of TCR␣ chains sedimented with the microsomal pellet fraction, suggesting that they are associated with ER membranes or are present as high molecular weight aggregates. To determine the orientation of these TCR␣ chains with respect to the ER membrane, the microsomal pellet fraction was subjected to digestion with proteinase K (Fig. 3). The endogenous lumenal proteins BiP and GRP94 were resistant to digestion by proteinase K in the absence, but not the presence, of detergent. By contrast, the ϳ10-kDa cytoplasmic tail of calnexin was readily cleaved by the protease indicating that this fraction contained ER vesicles that were sealed and of uniform membrane orientation. Coreglycosylated TCR␣ in the 100,000 ϫ g pellet was completely protected from protease digestion, confirming that it had been correctly translocated. Strikingly, both bands of the 28-kDa unglycosylated doublet were highly susceptible to proteinase K digestion, indicating that they must be present on the exterior, i.e. cytoplasmic side of the vesicles. These data strongly suggest that reverse translocation of TCR␣ from the ER accompanies its degradation by the proteasome.
TCR␣ Degradation Does Not Require Ubiquitination of Lysines-Substrates destined for degradation by the 26 S proteasome are commonly "tagged" by the covalent attachment of multiubiquitin chains (20). Inhibition of proteasome function in vitro or in vivo usually induces the accumulation of a signifi-cant fraction of highly ubiquitinated proteins, including ER degradation substrates like CFTR (9). As TCR␣ is a small protein, attachment of even a single ubiquitin moiety (ϳ7 kDa) would result in a readily detectable decrease in gel mobility. In the present study, no such mobility shift was observed in proteasome-inhibited cells (Figs. 1 and 2). However, ubiquitinated TCR␣ could have been missed if the ubiquitin linkages were labile to cleavage by cellular isopeptidases. To directly test whether TCR␣ ubiquitination is required for its degradation by the proteasome, we constructed a TCR␣ mutant (K␣R) in which all 11 lysine residues were substituted by arginine. Attachment of ubiquitin to substrates occurs via an isopeptide linkage between a lysine ⑀-amino group on the substrate and the Cterminal glycine of ubiquitin (21). Cells transfected with K␣R were pulse-labeled with [ 35 S]Met/Cys for 10 min, and the K␣R protein was immunoprecipitated from both the detergent-soluble and insoluble fractions with anti-TCR␣ antibody (Fig. 4A). Like wild-type TCR␣, K␣R was core-glycosylated and rapidly degraded. Remarkably, this degradation was efficiently inhibited by the proteasome inhibitor ALLN, giving rise to the appearance of partially and completely deglycosylated forms in both detergent-soluble and insoluble fractions. K␣R degradation was similarly inhibited by 5 M clasto-lactacystin ␤-lactone, the active form of lactacystin (22). These data suggest that either ubiquitination of TCR␣ is not required for its degradation by the proteasome or ubiquitin moieties can be attached to TCR␣ at alternate non-lysine residue(s). Future studies will be required to distinguish between these two possibilities.

DISCUSSION
Selective proteolysis is the final step in the elaborate network of proofreading and editing processes that have evolved to protect eukaryotic cells against the potentially deleterious consequences of errors that can accrue between genes and proteins. These include alterations in primary sequence due to mutation or to transcriptional and translational errors, as well as the effects of inappropriate spatial and temporal expression. Selective degradation is also required to eliminate unassembled or misassembled subunits of hetero-oligomeric plasma membrane complexes such as the heptameric TCR (4). Eukaryotic cells contain two major proteolytic systems: proteasomes, which are present in the cytoplasm and the nucleus, and lysosomes. In contrast to the lysosome-mediated disposal of mature or partially assembled TCR oligomers, the rapid, nonlysosomal degradation of unassembled TCR␣ subunits had suggested the existence of a unique degradation system associated with the endoplasmic reticulum (23). However, several recent studies have demonstrated that some misfolded proteins in the ER can

FIG. 2. TCR␣ degradation is blocked by proteasome inhibitors.
HEK293 cells transfected with TCR␣ were pulse-labeled for 10 min and chased for the times indicated without protease inhibitor (A) or in the presence of 20 g/ml ALLN (B) or 50 M lactacystin (C). Cells were separated into detergent-soluble and insoluble (pellet) fractions as indicated and immunoprecipitated with mAb H28-710. Band intensity was quantified by densitometry and plotted as a percentage of the signal at 0 min. In C, cells transfected with vector (lanes 1 and 4) or TCR␣ cDNA (lanes 2, 3, 5, and 6) were pulsed for 10 min (P) and chased for 180 min (C). The sizes of partially glycosylated TCR␣ forms are indicated by the asterisks. be degraded by cytoplasmic proteasomes following their reverse translocation from the ER (7,8). The data in this paper demonstrate that unassembled TCR␣ subunits that have been biosynthetically translocated into the ER and core-glycosylated are exported or "dislocated" into the cytoplasm, where they are deglycosylated and degraded by the proteasome.
TCR␣ was synthesized in HEK cells as a 38-kDa core-glycosylated precursor that was rapidly degraded. Our data show that lactacystin and ALLN stabilize this core-glycosylated form of TCR␣, implicating the proteasome in its degradation. Although the effect of ALLN in stabilizing TCR␣ has been previously reported (24), neither its activity against the proteasome nor its ability to induce the accumulation of dislocated and deglycosylated forms were recognized at that time. Our data show that either acute or chronic treatment of TCR␣-transfected cells with proteasome inhibitors cause the core-glycosylated 38-kDa TCR␣ chains to progressively shift to an ϳ28-kDa form that also lacks both N-linked oligosaccharides and an N-terminal signal peptide. As signal peptidase has its active site at the lumenal face of the ER, these data establish that some TCR␣ chains must have been at least partially translocated across the ER membrane. The susceptibility of the 28-kDa species to extravesicular protease indicates that it is not protected by the ER membrane and, hence, is cytoplasmic.
Our data indicate that the majority of dislocated TCR␣ sediments at relatively low speed and is insoluble in nonionic detergent. This change in detergent solubility is probably the result of the formation of high molecular weight aggregates. TCR␣ contains an unconventional transmembrane domain that is interrupted by four polar or potentially charged amino acids. In the absence of oligomeric partners that could shield these side chains from the hydrophobic core of the lipid bilayer, these polar residues have a dominant destabilizing influence over the rest of the molecule (6,25,26). It is unlikely, therefore, that nascent TCR␣ chains are able to effectively partition from the hydrophilic environment of the translocon into the lipid bilayer. At the same time the remaining 16 hydrophobic residues that constitute the TCR␣ transmembrane domain are unlikely to be able to effectively partition into the cytosol and may facilitate aggregation of the undegraded dislocated chains. It is possible that the inability of this heterodox transmembrane to effectively partition into the lipid bilayer may facilitate its dislocation without ever fully dissociating from the translocon.
The data presented in this paper suggest that ubiquitination of TCR␣ chains is not required for their degradation by cytoplasmic proteasomes. Although the attachment of high molecular weight ubiquitin polymers has been demonstrated to increase the susceptibility of substrate for degradation by 26 S proteasome, modification by ubiquitin is neither a necessary (27,28) nor a sufficient signal (29) for degradation. The requirement for ubiquitination of membrane and secretory proteins degraded by the proteasome is also variable. For example, inhibition of proteasome-mediated degradation of ␣ 1 -antitrypsin (10) or MHC class I heavy chain (13,30) does not appear to lead to the accumulation of ubiquitinated forms, although the lack of an evident ubiquitin "ladder" is not sufficient evidence upon which to exclude a role for ubiquitin. By contrast, there is evidence supporting a requirement for substrate ubiquitination in the degradation of other membrane and secretory proteins including connexin 43 (31) and CFTR in mammalian cells (9) and Sec61p (32) and carboxypeptidase Y (11) in yeast.
In the absence of ubiquitination, what signals are used to target ER degradation substrates to the proteasome? In cytomegalovirus-infected cells, two gene products appear to possess the capacity to induce the dislocation of MHC class I heavy chains from the ER and accompany them to the proteasome. We speculate that in non-virus-infected cells such targeting could be accomplished by direct coupling of proteasomes to the dislocation apparatus. Possibly, the presence of a misfolded protein in association with the dislocation apparatus could provide a signal that would recruit the docking of proteasome. Such a signal could be transmitted via a transmembrane chaperone like calnexin, as has been suggested recently (10). For this model to be true, dislocation of substrate would be predicted to depend on proteasome activity. In our studies Ͻ30% TCR␣ was dislocated (as measured by the appearance of deglycosylated chains) after 3 h in the presence of proteasome inhibitor, even though Ͼ75% TCR␣ would have been degraded during the same interval in the absence of proteasome inhibitors. Although preliminary, these data suggest that dislocation of TCR␣ from the ER may be coupled to the activity of the proteasome.
Taken together, the data presented above support the conclusion that TCR␣ chains are dislocated from the ER for degradation by cytoplasmic proteasomes. Thus, TCR␣ joins a growing number of membrane and secretory proteins which appear to be disposed of by a process involving dislocation from the ER and subsequent degradation by cytoplasmic proteasomes. Since TCR␣ has served as a prototype that has largely defined the process of ER degradation, we propose that the cytosolic degradation pathway may be the major pathway for degradation of misfolded or unassembled proteins in the ER.