Membrane-specific, Host-derived Factors Are Required for US2- and US11-mediated Degradation of Major Histocompatibility Complex Class I Molecules*

Human cytomegalovirus encodes two glycoproteins, US2 and US11, that target major histocompatibility complex (MHC) class I heavy chains for proteasomal degradation. We have developed a mRNA-dependent cell-free system that recapitulates US2- and US11-mediated degradation of MHC class I heavy chains. Microsomes support the degradation of MHC class I heavy chains in the presence of US2 or US11 in a cytosol-dependent manner. In vitro , the glycosylated heavy chain is exported from the microsomes. A deglycosylated breakdown intermediate of the heavy chain identical to that generated in intact cells accumulates in soluble form in the presence of proteasome inhibitors. Microsomes derived from the U373 astrocytoma cell line are far more effective than canine-derived membranes in supporting this US2- or US11-dependent reaction. In contrast, the HIV-encoded Vpu membrane protein can cause the destruction of CD4 from either human- or canine-derived membranes. Using the in vitro system, we show that a truncation mutant of US2 that lacks the cytosolic domain is unable to catalyze degradation, whereas a similar truncation of US11 continues to catalyze degradation of class I heavy chains. Therefore, US2 requires both transmembrane and cytosolic interactions to trigger dislocation of heavy chains, whereas US11 relies on the transmembrane domain to target heavy chains. US2 and US11 thus utilize different targeting mechanisms for class I degradation. The

I molecules on the surface of infected cells leads to recognition and destruction by cytotoxic T lymphocytes. Human cytomegalovirus (HCMV) encodes two proteins, the products of the US2 and US11 genes, which interfere with MHC class I expression by inducing rapid degradation of newly synthesized heavy chains (1)(2)(3). The destruction of heavy chains by US2 and US11 presumably enables evasion of cytotoxic T lymphocyte recognition early after infection or reactivation from latency, allowing the virus more time to replicate unnoticed. The degradation of MHC class I heavy chains by US2 and US11 is similar to the destruction of misfolded proteins that arises as a consequence of imperfections in the normal translation and folding reactions. US2 and US11 recognize the class I molecules within the lumen of the ER and dislocate them into the cytosol where they are destroyed by the proteasome (2,3).
The ER is a major site of quality control for secretory proteins in all eukaryotes. Many glycoproteins that lack the ability to fold properly or associate with partner polypeptides are retained in the ER and are eventually targeted for degradation by the proteasome (4,5). The involvement of the proteasome in the destruction of ER proteins necessitates their access to the cytosol where proteasomes are located. The mechanism by which aberrant proteins are targeted for destruction and dislocated from the ER remains unresolved and there may in fact be multiple pathways (6 -8). US2 and US11 have co-opted this degradation process to interfere with the expression of MHC class I molecules and therefore provide a model system for the study of ER-to-cytosol transport.
Genetic and cell-free systems in yeast have identified molecules that participate in the recognition, dislocation, and degradation steps of the process. Chaperones such as BiP and calnexin participate in recognition of some substrates (9,10). Several mutations in the Sec61p translocon compromise degradation and suggest that this pore may mediate dislocation into the cytosol, in addition to the import of nascent polypeptides (11). Ubiquitin-conjugating enzymes also play a role in the degradation of most ER-to-cytosol substrates in yeast, consistent with ubiquitination as a major signal for proteasomal destruction (4).
Mammalian cell-free systems have confirmed some of the observations made in yeast and intact cells. Ubiquitination and degradation of the cystic fibrosis transmembrane conductance regulator has been observed in canine microsomes (12,13). The degradation of CD4 by HIV Vpu has also been observed in canine microsomes (14). Export of a misprocessed glycosylphosphatidylinositol-anchored protein from microsomes was shown to be cytosol dependent (15). The release of ␤ 2 m from microsomes is suppressed by export of short (8 -10 residues) peptides, suggesting that such peptides and glycoproteins may both exit through the Sec61 translocon (16,17).
Our objective was to recapitulate the steps of early MHC * This work was supported in part by National Institutes of Health Grants 5R37-AI33456 and P01 AI42257. 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.
class I synthesis and recognition by US2/US11 in vitro. Dislocation and degradation in a cell-free environment would facilitate dissection of the components required for these events, but no such integrated system has been reported. Synthesis and assembly of MHC class I molecules within microsomes has been demonstrated (18). Furthermore, interactions between in vitro translated assembled class I molecules with US2 and US11 in canine pancreas-derived microsomes have likewise been shown, but in vitro dislocation by US2 or US11 has not been demonstrated (19,20).
The use of the system presented here reveals a heterogeneity in the membrane-bound components required for dislocation of heavy chains by US2 and US11. The intermediates that occur in the course of degradation by US2 and US11 in vitro are biochemically indistinguishable from the deglycosylated, cytosolic intermediates observed in intact cells. However, US2 and US11 do not rely equally on cytosolic domains to trigger degradation. Further, the mechanism for US2/US11 and class I appears to be distinct from that used by Vpu for the degradation of CD4. Here we also show that microsomes of human origin support US2/US11 degradation, whereas canine pancreas microsomes do not. Our data are consistent with wide divergence in the manner by which proteins are exported from the ER for cytosolic degradation.
Antibodies and Reagents-Polyclonal rabbit anti-HC, which recognizes unfolded heavy chains, has been described (24). W6/32, a mouse monoclonal antibody that recognizes properly folded MHC class I molecules in association with ␤ 2 m, has been described (25). MG132 was synthesized in the laboratory as described (26).
The anti-US2 polyclonal antibody was generated by immunizing rabbits with peptides corresponding to amino acids 20 -38, 92-110, and 146 -164 of US2. The Sec61␤ antibody was prepared as described (27). Cow liver cytosol was kindly provided by C. Shamu. In Vitro Transcription and Translation-In vitro transcription was performed as described (22). In vitro translations were performed for 45 min (except where otherwise noted) at 30°C in a reaction mixture containing 17.5 l of Flexi rabbit reticulocyte lysate (RRL) (Promega), 0.8 l of KCl (2.5 M), 0.5 l of amino acid mixture minus methionine (1 mM, Promega), 2.5 l of [ 35 S]methionine (10 mCi/ml, translation grade; PerkinElmer Life Sciences), and 0.5 l human astrocytoma or 1 l canine pancreas microsomal membranes, where indicated.
Preparation of Microsomes-Canine pancreas microsomes were prepared as described (28).
In Vitro Degradation Assay-In vitro translations of HLA-A2 and/or other polypeptides were performed in the presence of microsomes (0.5 l of astrocytoma microsomes/21.5 l of translation reaction, 1 l of canine pancreas microsomes/21.5 l of translation reaction). Transla-tion reactions were centrifuged at 20,000 ϫ g for 15 min. The supernatant was removed, and microsomes were resuspended in 5 l of homogenization buffer and 15 l of Flexi RRL containing 1 mg/ml RNase (Roche Molecular Biochemicals) and 50 M MG132 where indicated. Cytosol extracts containing an ATP-regenerating system (29) were added where indicated in place of RRL. Reactions were incubated at 30°C, and aliquots were withdrawn at indicated time points. Microsomal pellet was separated from RRL supernatant by centrifugation at 20,000 ϫ g for 15 min. Reducing sample buffer was added to supernatant and pellet fractions followed by separation by SDS-PAGE, or samples were mixed with lysis buffer (see below) for immunoprecipitation where indicated.
Pulse-Chase Analysis, Immunoprecipitation, and Endoglycosidase Digestion-Pulse-chase analysis, immunoprecipitation, and endoglycosidase H digestion was performed as described (24). Peptide N-glycanase (PNG) (New England Biolabs) digestion was performed as described by the manufacturer.

Microsomes Derived from Human Astrocytoma Cells Support
Proper Insertion, Signal Sequence Cleavage, Glycosylation, and Folding of in Vitro Translated Membrane Glycoproteins-The HCMV US2 and US11 gene products induce proteasomal degradation of MHC class I heavy chains. To generate a cell-free system with which to study US2-and US11-mediated degradation, microsomes were isolated from U373-MG cells, a human astrocytoma cell line that supports rapid degradation when transfected with US2 or US11. We first characterized the ability of these microsomes to insert and process glycoproteins in the course of in vitro translation reactions (Fig. 1). The addition of membranes to an in vitro translation reaction of HLA-A2 yielded a polypeptide with a slower mobility than the product translated in the absence of membranes (Fig. 1A, compare lanes 1 and 3). This product was sensitive to digestion with endoglycosidase H (EH), indicating the presence of an N-linked glycan (Fig. 1A, lane 2). The faster mobility of the EH-treated microsomal species relative to the HLA-A2 polypeptide translated in the absence of microsomes indicates that the signal sequence had been cleaved from the microsomal HLA-A2. This HLA-A2 polypeptide was identical in molecular size to HLA-A2 translated into canine pancreas microsomes (Fig. 7). To ascertain proper assembly of HLA-A2 molecules in astrocytoma microsomes, HLA-A2 was co-translated with mRNA that encodes the light chain, ␤ 2 m (Fig. 1B). Immunoprecipitations with a rabbit polyclonal antiserum that recognizes unfolded MHC class I heavy chains, or with W6/32, a monoclonal antibody that specifically recognizes properly folded MHC class I molecules in association with ␤ 2 m (Fig. 1B,  lanes 2 and 3) showed that proper assembly of translated HLA-A2 and ␤ 2 m readily occurred. Therefore folded and unfolded species can be immunoprecipitated from the translation reactions, indicating that these microsomes fully support protein folding.
Microsomes Derived from Human Astrocytoma Cells Support US2-and US11-mediated Degradation of Class I Molecules-We next examined the ability of the membranes to induce US2/US11-mediated degradation of MHC class I heavy chains. HLA-A2 was translated into microsomes derived from control, US2-expressing, or US11-expressing astrocytoma cells ( Fig. 2A). RNase was added after 45 min to terminate ongoing translation, and reactions were sampled at 0, 30, 60, and 90 min after addition of RNase. Microsomal fractions from each time point were isolated by centrifugation and analyzed directly by SDS-PAGE. The amount of the HLA-A2 polypeptide recovered decreased over time in the US2 and US11 microsomes ( Fig. 2A, lanes 5-12), whereas it was stable in control microsomes over 90 min ( Fig. 2A, lanes 1-4). Amounts of HLA-A2 were quantitated and plotted as the percentage remaining relative to time 0. Approximately 60% of HLA-A2 molecules were degraded in US2 and US11 microsomes after 90 min, with a more rapid decline in HLA-A2 levels in US2 microsomes at earlier time points. Additionally, less HLA-A2 was visible in US2 microsomes at time 0 relative to control and US11 microsomes ( Fig. 2A, compare lanes 1, 5, and 9). These results have been observed consistently in three experiments and indicate that US2-mediated degradation of HLA-A2 was active during the translation period. US2 targets HLA-A2 for degradation at an earlier step during HLA-A2 maturation than US11.
Next, we modified the assay by performing translations in the presence of microsomes, followed by isolation of the microsomes by centrifugation, removal of the translation supernatant, and resuspension of microsomes in a fresh batch of cytosol containing RNase to terminate translation and start the chase. HLA-A2 was translated into control or US2 microsomes, and CD4 was co-translated to serve as a control (Fig. 2B). CD4 is a protein of a size comparable with that of MHC class I products, and likewise a member of the Ig superfamily, yet not a substrate for US2 or US11. After translation for 45 min at 30°C, the microsomes were pelleted by centrifugation and resuspended in a small volume of homogenization buffer. RRL containing an ATP-regenerating system and RNase was added to resuspended microsomes as chase cytosol. The mixture was then incubated at 30°C and sampled at 0, 30, and 60 min after the addition of cytosol. The membrane fractions were analyzed directly by SDS-PAGE (Fig. 2B). Whereas the amount of radiolabeled HLA-A2 polypeptide present in the US2 microsomes decreased over the chase period as compared with the control microsomes, CD4 was equally stable over 60 min in US2 and control microsomes (Fig. 2B, lanes 1-3 and 7-9). No significant loss of HLA-A2 or CD4 occurred in control microsomes (Fig. 2B, lanes 1-6). Therefore, US2 specifically destabilized HLA-A2. US2/11-mediated degradation is sensitive to agents that alter redox potential (21). Similar to what is observed in intact cells, no dislocation of HLA-A2 was observed from US2 microsomes when microsomes were resuspended in RRL containing 10 mM N-ethylmaleimide (NEM), to alkylate free sulfhydryls (Fig. 2B, lanes 4 -6 and 10 -12). The addition of 1 mM NEM or another oxidant, diamide, blocked dislocation by more than 50% in RRL as opposed to 100% in intact cells. 2 The lower sensitivity to NEM and diamide in the in vitro reactions may be attributable to the high (100 -200 mg/ml) protein concentration of RRL.
Proteasome Inhibition Reveals Accumulation of Soluble Heavy Chain Intermediates Released from US2 Microsomes-When US2-expressing cells are treated with proteasome inhibitors, deglycosylated MHC class I heavy chains accumulate in the cytosol (3). To determine whether the same intermediates can be observed in vitro, the dislocation assay was performed with proteasome inhibitors added to the RRL chase mix. Membrane-associated and soluble class I products were analyzed separately (Fig. 3). When the supernatant fraction was analyzed, two distinct polypeptide species were observed in inhibitor-treated reactions, but at later chase points only (Fig. 3A, lane 8). These species were detected in incubations only in the presence of the proteasome inhibitor, MG132, and when US2 was present in the microsomes. Identical species were observed in supernatants recovered from US11-containing microsomes (Fig. 6A, lanes 9 and 12) but not from control microsomes. 2 Other specificity controls confirm that these intermediates were a product of functional US2/US11-mediated degradation only (see below). The loss of HLA-A2 from microsomes was unaffected by the inclusion of MG132 in the chase cytosol (Fig.  3A, lanes 1-4). The amount of soluble intermediates that can be recovered was always less than the amount lost from the microsomes, indicating incomplete inhibition of proteasome activity or the involvement of other proteases present in rabbit reticulocyte lysate. Release of glycosylated CD4 into the supernatants was not observed in the presence of US2 (Fig. 3A, lanes [5][6][7][8]. Immunoblotting of membrane and supernatant fractions for Sec61␤, an integral ER protein, shows that supernatants did not contain ER-membrane proteins, and therefore the release of HC into the supernatant cannot be attributed to a failure to retrieve small microsomal vesicles (see below; immunoprecipitations were performed and analyzed by both SDS-PAGE and IEF (Fig. 3B). The polypeptides found in the supernatants of the in vitro reactions in the presence of MG132 (Fig. 3B, lane 6) were immunoprecipitable with anti-heavy chain antiserum and co-migrate with HLA-A2 recovered from US2 cells treated with MG132 (Fig. 3B, lane 2). Furthermore, the mobility of the faster-migrating species was identical to the mobility of HLA-A2 treated with PNG both on SDS-PAGE and on IEF gels (Fig. 3B, lanes 4 and 8).
The heavy chain intermediate that accumulates in US2-and US11-expressing cells in the presence of proteasome inhibitors is generated by the removal of the N-linked glycan by an N-glycanase and therefore has a faster mobility on SDS-PAGE (2, 3). Deglycosylated heavy chains acquire an additional negative charge, because of the conversion of asparagine to aspartic acid upon hydrolysis of the glycoamide linkage by N-glycanase, and therefore have a distinct mobility on IEF gels (3). This identification was confirmed by digestions of immunoprecipitates with EH and recombinant PNG (Fig. 3B, IEF gel). The latter is particularly relevant, as the time-dependent conversion of glycosylated HLA-A2 in the in vitro reactions to the more acidic species identifiable by IEF paralleled that seen in US2 cells (except that less of the material was recovered from the in vitro chase supernatants (lane 6) than from intact cells (lane 2)) (Fig. 3B). Thus, biochemical analysis confirmed that the intermediates generated in this system were identical to the HLA-A2 intermediates produced in intact cells. In the absence of a source of the reportedly cytosolic mammalian N-glycanase that could be added to the in vitro system described here, the role of N-glycanase in the reaction remains to be explored.
In Vitro Degradation of MHC Class I Molecules by US2 Is Allele-specific-For intact cells, the recognition of heavy chains by US2 and US11 is to a large extent specific for particular MHC locus products. To establish the specificity of the in vitro reactions, degradation assays were performed in US2 microsomes with another MHC class I allele, HLA-C. Unlike HLA-A2, HLA-C was stable in US2 microsomes and no intermedi-  1-4), US2-transfected (lanes 5-8) or US11-transfected (lanes 9 -12) U373 cells for 30 min. RNase was added to each reaction, and aliquots were withdrawn at 0, 30, 60, and 90 min after the addition of RNase. Microsomes were analyzed directly by SDS-PAGE (12.5%). Autoradiograms were quantitated with ALPHAIMAGER software (Alpha Innotech, San Leandro, CA). Band intensities were plotted as percentage of HLA-A2 remaining relative to intensity at 0 min chase point. B, CD4 and HLA-A2 mRNAs were co-translated into control microsomes (lanes 1-6) or US2 microsomes (lanes 7-12). The degradation assay was performed in the presence (lanes 4 -6 and 10 -12) or absences (lanes 1-3 and 7-9) of 10 mM NEM added to chase cytosol. Reactions were incubated at 30°C and sampled at 0, 30, and 60 min. Microsomes were recovered by centrifugation and analyzed directly by SDS-PAGE (12.5%). Autoradiograms were quantitated as in A, and HLA-A2 and CD4 band intensities were plotted as percentage remaining relative to time 0.
ates of HLA-C were released from the microsomes (Fig. 4, lanes  7 and 8), consistent with the resistance of the C-locus product to degradation in intact cells (20). Furthermore, HLA-E and HLA-G, when analyzed in the in vitro system, were stable, again consistent with observations in intact cells (20,32). 2 We conclude that the in vitro system faithfully recapitulates the specificity of dislocation described in living cells.
Ablation of the Cytosolic Domain Ablates US2 Function but Not US11 Function-Microsomes derived from US2 and US11 transfectants contain high levels of the viral protein. The ability of the pre-existing viral protein in these microsomes to target MHC class I heavy chains for dislocation shows that US2 and US11 need not be translated simultaneously with heavy chains to trigger dislocation. Do in vitro translated US2 and US11 catalyze the degradation of heavy chains? If so, this would facilitate the analysis of US2 and US11 mutants by translation. Detection of radiolabeled viral protein would assist in future identification of interacting proteins in the system. We translated HLA-A2 into control microsomes, divided the translation mixture in half, and added either wild type US2 mRNA or US2-186 mRNA, which lacks the putative cytosolic domain. Translation was continued for 30 min, after which the assay was continued as above with membrane and supernatant fraction sampled at the indicated times. HLA-A2 was unstable in the presence of co-translated US2 but not US2-186 (Fig. 5A,  lanes 1-6), indicating that the cytosolic portion of US2 is critical for inducing heavy chain degradation. Additionally, deglycosylated HC intermediates appeared in the supernatants of reactions that contained wild type US2 but not US2-186 (Fig.  5A, lanes 7-12). A portion of each reaction was immunoblotted for Sec61␤ (Fig. 5A). The Sec61␤ polypeptide was observed only in pellet fractions, indicating that the appearance of HLA-A2 in the supernatant fractions was not because of an inability to retrieve membrane proteins. The luminal domain of US2 is sufficient for interaction with HLA-A2 (33), and therefore the lack of dislocation by US2-186 is not attributable to an inability to bind to HLA-A2 molecules. Pulse-chase analysis of proteasome-inhibitor treated US2-186 astrocytoma transfectants confirmed the inactivity of this mutant also in intact cells, as no conversion of HC from glycosylated to deglycosylated species was observed (Fig. 5B, lanes 1 and 2). A corresponding set of experiments was done with US11. When wild type US11 or US11-201, which lacks the putative cytosolic domain, were translated into microsomes which contain radiolabeled HLA-A2 and chased (as described for Fig. 5) (Fig. 6A), deglycosylated intermediates accumulated in the supernatant from both US11-and US11-201-containing reactions (Fig.  6A, lanes 7-12). These HLA-A2 intermediates migrated at positions identical to intermediates released from US2containing microsomes. 2 Similarly, MHC class I intermediates found in US11-and US2-expressing cells are biochemically indistinguishable. Significant loss of HLA-A2 was not apparent from microsome pellets (Fig. 6A, lanes 1-6), as US11 degradation is less efficient in vitro and quantities of in vitro translated US11 were less than the amounts present in microsomes derived from US11 transfectants. Again, the ability of both molecules to degrade HLA-A2 was identical to the result observed in intact US11/US11-201 transfectants (Fig. 6B). HC immunoprecipitations indicated that rapid conversion to deglycosylated HC was observed in both wild type US11 and US11-201 transfectants (Fig. 6B, lanes 1-4). The apparent increase of wild type US11/US11-201 over the chase was caused by delayed cleavage of the US11 signal sequence (Fig. 6B, lanes 1-4) (24). Both wild type US11 and US11-201 are stable in intact cells and in microsomes. Therefore the stability of these molecules does not correlate with their ability to degrade MHC class I. We conclude that US2 requires its cytosolic domain for function whereas US11 does not, suggesting that the two polypeptides utilize distinct mechanisms to facilitate degradation of the MHC class I molecules.
Canine Pancreas Microsomes Do Not Support Degradation of HLA-A2 in the Presence of Cotranslated US2-We were able to observe US2/US11-specific destabilization of HLA-A2 in the astrocytoma-derived microsomes. We examined canine pancreas as a more readily available source of microsomes to support US2-mediated degradation. Degradation assays were performed with HLA-A2 and US2 co-translated in control astrocytoma or canine pancreas membranes (Fig. 7). Whereas the HLA-A2 polypeptide was unstable in the presence of co- translated US2 in the astrocytoma microsomes (Fig. 7, lanes  1-3), we were unable to observe destabilization of HLA-A2 in the canine microsomes (Fig. 7, lanes 4 -6). No deglycosylated species were released from the canine microsomes in the presence of US2. 2 One possible explanation would be the inability of preexisting canine ␤ 2 m to support assembly of HLA-A2. However, addition of human ␤ 2 m mRNA to translation reactions did not enable HLA-A2 degradation from canine membranes even though complete assembly of the HLA-A2 molecules was readily demonstrable. 2 To examine whether canine-derived membranes were competent to dislocate another ER-resident protein susceptible to degradation, we performed in vitro assays with the CD4 and Vpu polypeptides. Vpu is an accessory protein encoded by HIV that causes ER-resident CD4 to be destroyed by the proteasome in a manner similar to MHC class I by US2, although many of the details of Vpu-and US2/US11-mediated degradation remain to be worked out (34,35). CD4 was translated into control or canine pancreas membranes and mRNA encoding Vpu or a mutant of Vpu lacking the ability to catalyze CD4 degradation (23) was added (Fig. 8). CD4 was destabilized by wild type Vpu in both types of microsomes (Fig. 8, lanes 1-3 and 7-9), although the destabilization was more pronounced in the human microsomes. However, CD4 was stable in the presence of mutant Vpu in both sets of microsomes, indicating that the instability of CD4 was not the result of a nonspecific property of the membranes or the presence of an additional in vitro translated polypeptide (Fig. 8, lanes 4 -6 and 10 -12). Degradation of CD4

FIG. 4. In vitro degradation of MHC class I molecules is allele-specific.
CD4 mRNA was cotranslated with HLA-A2 mRNA or HLA-C mRNA in the presence of US2 microsomes. The degradation assay was performed in the presence of 50 M MG132 (as in Fig. 3A). The microsomal pellet (lanes 1-4) and supernatant fractions (lanes 5-8) were analyzed by SDS-PAGE (10%). Lanes 1-4 were exposed for 4 days, and lanes 5-8 were exposed for 14 days. did not result in the accumulation of soluble intermediates in this system, consistent with what has been observed in vitro and in intact cells (14,36). We conclude that the inability of canine microsomes to support US2-mediated degradation is specific to US2. This observation points to the existence of multiple mechanisms for recognition and/or dislocation of type I membrane proteins, with a species-specific component to the host membrane protein involved.
Degradation of Heavy Chains in Vitro Is Dependent on Cytosol-To examine the requirements for in vitro degradation, FIG. 7. Canine microsomes do not support US2-mediated degradation in vitro. HLA-A2 mRNA was translated into control astrocytoma microsomes or canine pancreas microsomes for 30 min. US2 mRNA was added and translations continued for 30 min. Microsomes were pelleted and resuspended in RRL ϩ RNase ϩ MG132 and sampled at 0, 30, and 60 min. Astrocytoma pellet fractions (lanes 1-3) and canine pellet fractions (lanes 4 -6) were analyzed directly by SDS-PAGE (12.5%). Autoradiograms were quantitated and plotted as in Fig. 2.   FIG. 8. Canine pancreas and human astrocytoma microsomes support the degradation of CD4 by HIV Vpu in vitro. CD4 mRNA was translated into canine or astrocytoma microsomes for 30 min, followed by the addition of Vpu or Vpu2/6 mRNA to each, and translations were continued for 30 min. The degradation assay was performed in the presence of MG132 and sampled at 0, 30, and 60 min. Pellet fractions from each time point were analyzed by SDS-PAGE (12.5%).
FIG. 6. US11 lacking the cytosolic domain is able to degrade MHC class I molecules. A, HLA-A2 mRNA was translated into control microsomes for 30 min followed by the addition of wild type US11 or US11-201 mRNA and continued translation for 30 min. The degradation assay was performed in the presence of MG132 and sampled at 0, 30, and 60 min. Pellet (lanes 1-6) and supernatant (lanes 7-12) fractions were analyzed by SDS-PAGE (12.5%). Lanes 1-6 were exposed for 15 h, and lanes 7-12 were exposed for 72 h. B, astrocytoma transfectants of US11 or US11-201 were starved for 60 min in 50 M MG132, metabolically pulsed with [ 35 S]methionine for 10 min, and chased with unlabeled medium for 30 min. MHC class I and US11 molecules were recovered from cell lysates with anti-HC and anti-US11 serum and analyzed by SDS-PAGE (12.5%).
polypeptides were translated into US2 microsomes and these microsomes were then further incubated in either RRL, cow liver cytosol with an ATP-regenerating system, or homogenization buffer with an ATP-regenerating system. Cytosol is required to induce the loss of heavy chains from the microsomes as well as the appearance of the deglycosylated intermediate in the cytosolic supernatant (Fig. 9). Both rabbit and cow liver cytosol supported heavy chain degradation from US2 microsomes (Fig. 9, lanes 3-6). Cytosol prepared from astrocytoma cells also supported degradation and formation of the deglycosylated intermediate (data not shown). No disappearance of HLA-A2 was seen in from US2 microsomes chased in buffer (ϩATP regenerating system) (Fig. 9, lanes 1 and 2). Therefore, although species-specific membrane factors may be required to interact with US2 and US11 in the membrane, the cytosolic factors required for dislocation are less species-specific. RRL was more effective than cow liver cytosol in extraction of HLA-A2 from membranes but equal amounts of intermediate were observed in the supernatant (Fig. 9, lanes 10 and 12). Either proteasome inhibition was more effective in the less concentrated cow liver cytosol, or RRL contained proteases not inhibited by proteasome inhibitors, that also destroy HLA-A2. DISCUSSION Here we describe a cell-free system that duplicates the ERto-cytosol dislocation and degradation of MHC class I molecules triggered by the HCMV US2 and US11 proteins in a specific manner. We use this in vitro system to dissect the dislocation process. First, we demonstrate that membranes prepared from a human cell line perform comparably to canine microsomal membranes in the insertion and processing of glycoproteins. Next, we show that class I heavy chains are unstable in astrocytoma-derived membranes that contain US2 or US11. The half-life of heavy chains in US2 and US11 microsomes is ϳ40 -60 min compared with less than 5 min in intact cells. This is comparable with the delayed kinetics for the degradation of CD4 by Vpu in microsomes (14). For the Vpu-CD4 co-translation, the degradation in vitro occurred three times more slowly than in intact cells. The US2-mediated degradation in vitro is ϳ8 -10 times slower than in intact cells. The relative amounts of US2 present in microsomes prepared from US2 cells is high, as the cell line overexpresses US2, and therefore a shortage of US2 is unlikely to be the reason for the delay. Additionally, supplementation of US2 microsomes with additional US2 by in vitro translation does not enhance degradation. 2 Many factors may explain the reduced rate of degradation in the in vitro system. A specialized subcompartment of the ER in which US2 rapidly triggers dislocation may be disrupted in the micro-somes used. The microsomal vesicles most proficient in import of in vitro translated polypeptides may not support efficient dislocation. Local concentrations or assembly of membrane or cytosolic factors that drive extraction may be limiting. Although energy in the form of ATP is required, it remains to be established that ATP is in fact the high energy intermediate directly powering the reaction and therefore metabolic steps that may occur in vitro may be rate-limiting.
Proteasome inhibition permits the accumulation of a breakdown intermediate that is identical to the breakdown intermediate generated in intact US2/US11 cells. It is sufficient to treat the cytosol with proteasome inhibitors to observe this effect in vitro. The appearance of glycosylated and non-glycosylated heavy chains released from the microsomes suggests that fully glycosylated material is exported through the ER membrane and that this material is released from the membrane prior to exposure to N-glycanase. This observation required the use of the in vitro system described here. In US2/ US11 cells, only deglycosylated material is detected in cytosolic fractions. N-Glycanase may be significantly more active in intact cells than in RRL. Alternatively, glycosylated material that has exited may not be immediately accessible to N-glycanase and the steps that render it accessible may be less efficient in RRL. The molecular identity of the N-glycanase involved in this reaction in mammalian cells remains to be established.
The ER-luminal portion of US2 contains an immunoglobulinlike fold that mediates the interaction with folded MHC class I molecules but is not itself sufficient to induce degradation. Therefore elements in the transmembrane and cytoplasmic tail regions of US2 must be important for its function (37). Alignment of the US11 sequence with US2 suggests that US11 contains a similar Ig-like fold, which is also likely to dictate association with MHC class I (37). Surprisingly, deletion of the cytosolic tail ablated US2 function but did not ablate US11 function, a result verified both in vitro and in intact cells. Therefore, the two molecules may use different machinery to target heavy chains, or may depend to a different degree on regions of the heavy chain cytosolic tail to stimulate extraction. Despite the proposed Ig fold for both molecules, there is only 25% amino acid similarity between US2 and US11, very little of which maps to the transmembrane and tail regions. A difference in mechanism may also explain different kinetics of degradation between US2 and US11 in vitro.
The ability of newly translated US2 to target class I molecules for degradation further strengthens the hypothesis that US2 recruits, rather than induces, factors that trigger dislocation. The mammalian ER thus contains all of the factors required for US2 to function. Several studies in yeast show that the degradation process is assisted by the induction of an unfolded protein response (38,39). Up-regulation of chaperones during an unfolded protein response is thought to assist protein folding and/or facilitate recognition of misfolded proteins for dislocation. The mechanism used by US2/US11 may bypass this requirement by recognizing the target soon after synthesis. An important distinction between US2/US11-mediated degradation and the degradation of misfolded proteins is the difference in kinetics; class I molecules are rapidly degraded in US2/US11-expressing cells, whereas misfolded proteins may be retained in the ER for hours before being recognized and disposed of. Additionally, the ability of newly translated US2 to target pre-existing heavy chains suggests that US2 recruits factors to stimulate dislocation rather than triggering dislocation of heavy chains while they are still in the Sec61 pore.
indicates that membrane-associated factors play a role in steps subsequent to recognition of class I heavy chains. Whether this is the result of a species-specific or tissue-specific factor is unclear at present. Degradation of HLA molecules by vaccinia virus-expressed US2/US11 in a porcine endothelial cell line has been observed (20). Vpu activity may not be as dependent on membrane components because it directly interacts with a component of an ubiquitin-protein isopeptide ligase complex, ␤-TRCP, which is proposed to recruit Skp1 and direct ubiquitination of CD4 (40). This model does not address the mechanism by which CD4 is physically extracted from the ER, but it is notable that no cytosolic intermediates of CD4 are detected in vitro or in cells in the presence of proteasome inhibitors (14,36). The interactions between translocating heavy chains or CD4 and the Sec61 complex need to be examined. The soluble heavy chain intermediates generated in vitro are found in a compartment in which no Sec61␤ is detected (Fig. 5A). It appears that fully glycosylated material is released from the microsomes. Therefore, a fraction of the glycosylated membrane-associated heavy chains in US2 microsomes may be associated with translocation channels. The delayed kinetics of dislocation in this system and the ability to manipulate heavy chains after translation and before dislocation may allow us to trap molecules in association with these channels in future experiments.
Another observation from this work addresses the potential heterogeneity of the ER. Canine pancreas is often used as a convenient and reliable source of microsomes that will support all manner of ER-associated reactions. Generally, the results obtained by in vitro translation using canine pancreas have compared well with results obtained in living cells, where relevant. However, the exocrine portion of the pancreas is a tissue that may well display highly tissuespecific specializations and limitations. Not necessarily all ER-dependent reactions may be catalyzed with equal efficiency. Assembly and processing reactions for multimeric glycoproteins are supported well, but as we show here, not all aspects of the dislocation reaction can be recapitulated as effectively in canine microsomes.
The dislocation reaction observed in vitro is specific. First, HLA-A2, but not HLA-C locus products (as well as HLA-E and HLA-G) 2 are dislocated. Second, a mutant of US2 that lacks only 13 residues of the cytoplasmic tail is inactive. Third, we observe no dislocation of the CD4 glycoprotein, also a member of the Ig superfamily. Conversely, the HIV-encoded Vpu protein is capable of extracting CD4 from the ER in vitro. Thus, CD4 is not uniquely resistant to dislocation in vitro. The cytosol dependence of ER-to-cytosol translocation and degradation is well established (10,15) and is observed in this system also. Interestingly, the species dependence of the US2/US11-mediated process is specific to the membranes but not to the cytosol, as cow liver and rabbit cytosols both support the reaction.
Ubiquitinated class I heavy chains have been detected in US11 and US2 cells (41), and ubiquitination is required for dislocation of heavy chains in US11 cells (42,43). A similar requirement for ubiquitination during US2-mediated dislocation is likely, as no differences in cytosolic requirements between US2 and US11 have been noted thus far, other than the presence of cytosolic domains of the US2 molecules (see above) and class I molecules 2 themselves. The functional relevance of ubiquitinated heavy chains as an obligate intermediate in the extraction process remains unclear. Ubiquitination of the cytoplasmic tail of heavy chains is not required for their disposal, as removal of all lysines from the tail prevents neither ubiquitination nor degradation of the class I molecules (41). ER-resi-dent class I heavy chains remain resistant to protease in the absence of a functional ubiquitin system, suggesting that they are not partially dislocated (42,43). Therefore, it remains unclear how luminal regions are ubiquitinated and direct extraction. We have so far not observed ubiquitinated heavy chains in this in vitro system. Because we always observed less soluble intermediates than the amount degraded from the microsomes, a fraction of the intermediates may be in a form not readily observable by SDS-PAGE. Heterogeneous polyubiquitinated species would fall into this category. It is estimated that less than 10 -20% of heavy chains are ubiquitinated in semipermeable, proteasome-inhibitor-treated US11 cells (41). Given the complexity of the ubiquitin conjugation machinery and the diversity of ubiquitin ligases, identification of the enzymes that are active during the dislocation of heavy chains represents a major undertaking. An in vitro system such as this one that recapitulates the dislocation reaction may be helpful in such endeavors.