A Mutant Cell with a Novel Defect in MHC Class I Quality Control1
http://www.100md.com
免疫学杂志 2005年第11期
Abstract
COS7 (African Green Monkey kidney) cells stably transfected with the mouse MHC class I allele H-2Kb were mutagenized, selected for low surface expression of endogenous MHC class I products, and subcloned. A mutant cell line, 4S8.12, expressing very low surface MHC class I (5% of parental levels) was identified. This cell line synthesized normal levels of the MHC class I H chain and 2-microglobulin, as well as normal levels of TAP, tapasin, GRP78, calnexin, calreticulin, ERp57, and protein disulfide isomerase. Full-length OVA was processed to generate presented H-2Kb-SIINFEKL complexes with equal efficiency in wild-type and mutant cells, demonstrating that proteasomes, as well as TAP and tapasin, functioned normally. Therefore, all the known components of the MHC class I Ag presentation pathway were intact. Nevertheless, primate (human and monkey) MHC class I H chain and 2-microglobulin failed to associate to form the normal peptide-receptive complex. In contrast, mouse H chains associated with 2-microglobulin normally and bound peptide at least as well as in wild-type cells. The 4S8.12 cells provide strong genetic evidence for a novel component in the MHC class I pathway. This as-yet unidentified gene is important in early assembly of primate, but not mouse, MHC class I complexes.
Introduction
Cytotoxic T lymphocyte survey tissues for the presence of foreign Ags, such as those of viral or tumor origin, presented on MHC class I molecules on the surface of cells. MHC class I molecules are trimolecular complexes, consisting of a polymorphic glycoprotein, the H chain, which is noncovalently associated with 2-microglobulin (2m) 5, and a small peptide, generally 8–10 residues long, which is usually the product of proteolysis of cellular and viral proteins in the cytosol (reviewed in Refs. 1 and 2).
H chain and 2m are cotranslationally translocated into the endoplasmic reticulum (ER), where they fold, form intrachain disulfide bonds, and associate with each other. Before binding peptide, this complex is relatively thermolabile and is retained within the ER. Peptides are transported into the lumen of the ER by the peptide transporter TAP, and if a peptide has the appropriate sequence motif, it can bind to the H chain/2m complex. Upon binding peptide, the association between H chain/2m becomes more stable, and the trimolecular complex is released from the ER and transported to the cell surface.
Each stage of MHC class I maturation is associated with ER chaperones (reviewed in Refs. 3 and 4). GRP78 (BiP) interacts with H chain shortly after its synthesis (5, 6). Protein disulfide isomerase (PDI) may help form some of the H chain intrachain disulfide bonds. Calnexin (CNX) interacts with newly synthesized H chain. CNX dissociates from human MHC when H chain interacts with 2m but remains associated with mouse H chain/2m complexes (6, 7). Calreticulin (CRT) and ERp57 both associate with H chain/2m complexes (8, 9, 10, 11, 12). Tapasin brings the peptide-receptive H chain/2m complex into physical proximity with TAP-transported peptides (13, 14, 15) and otherwise facilitates peptide binding to the H chain/2m dimer (16, 17, 18). When peptide is bound to the MHC complex, the chaperones all dissociate from MHC, and the mature H chain/2m/peptide complex is allowed to exit the ER to the cell surface.
In general, these chaperones are believed to perform quality control, monitoring protein folding and either directing folding toward the proper conformation or targeting misfolded protein to degradation. Thus, various chaperones have been shown to facilitate folding and assembly (19) and retain peptide-empty MHC class I in the ER (16, 18). However, the details of these functions are not well understood as yet.
The identity and functions of several proteins associated with Ag processing, such as TAP and tapasin, have been elucidated by studying mutant cell lines with defects in cell surface MHC class I expression. To further investigate the MHC class I pathway, we generated a series of mutant cell lines with defects in surface MHC class I expression and in this study describe one of these mutants that has a defect in ER quality control.
Results
4S8.12 cells have reduced endogenous MHC class I surface expression
We chose to mutagenize COS7 cells because these cells episomally replicate plasmids containing the SV40 origin of replication, leading to high expression of transfected genes, and because they react broadly with the many available reagents directed against components of the human Ag processing system. These characteristics make expression cloning of mutant genes practical in COS7 cells. To study presentation of defined Ags, we stably transfected COS7 with the well-characterized mouse MHC allele H-2Kb to produce COS-Kb cells. COS-Kb cells were mutagenized with ethyl methanesulfonate and repeatedly selected for low surface expression of MHC class I molecules, using a panel of anti-MHC class I Abs bound to magnetic beads. After repeated selections, the population was subcloned by limiting dilution, and subclones were analyzed by flow cytometry for low surface MHC class I expression. The 4S8.12 subclone expressed, on average, 5% of wild-type levels of MHC class I reactive with either PA2.6 or W6/32 mAbs, both of which recognize all HLA-A, -B, and -C alleles in association with 2m (Fig. 1A). As well, HLA-B-like molecules associated with 2m (mAb 4E-reactive) (Fig. 1B) and HLA-A-like molecules associated with 2m (reactive with the mAb JD12) (Fig. 1C) were both markedly lower than the parent cell line’s levels.
The decreased surface expression of endogenous MHC class I did not reflect a general defect in glycoprotein expression because transiently transfected influenza hemagglutinin was expressed at equal levels on the surface of 4S8.12 and wild-type COS-Kb cells (Fig. 1D); however, the defect was not limited to the endogenous monkey MHC class I because transiently transfected human MHC class I alleles (e.g., HLA-A*0302; Fig. 1E) were also expressed at lower levels on 4S8.12 cells than on COS-Kb.
The 4S8.12 population showed biphasic levels of endogenous surface MHC class I, with a minority of cells (5–20% under normal culture conditions) expressing nearly wild-type levels of MHC class I (Fig. 1A). This was not because of contamination with wild-type cells because repeated subcloning by limiting dilution invariably yielded biphasic populations (at least 30 subclones), rather than either uniformly high and low populations (data not shown). Furthermore, when nearly pure populations of either low- or high-MHC class I-expressing 4S8.12 cells were selected using anti-MHC class I Ab-coated magnetic beads, each population became biphasic again within 24 h of culture (Fig. 1F). This suggests that individual cells in the population rapidly switch between a normal MHC class I phenotype and a mutant (low surface MHC class I) phenotype. The trigger for this switch is unknown but may be related to environmental conditions. It is not related to the phase of cell cycle (data not shown).
Assembly of endogenous H chain/2m complexes is defective in 4S8.12 cells
Endogenous (monkey) MHC class I is expressed on the surface of 4S8.12 at markedly lower levels than in wild-type cells (Fig. 1A). This could have many causes: reduced MHC class I synthesis, failure to assemble 2m and H chain molecules in the ER, reduced peptide supply or peptide binding, increased degradation of H chain or 2m, or because mature (peptide-containing) molecules are abnormally retained in the ER. We analyzed maturation of MHC class I in wild-type and mutant cells, using the mAbs W6/32 (which recognizes MHC class I only when it is assembled with 2m, regardless of the presence or absence of peptide) and HC10, which recognizes only H chain that is not assembled with 2m. Synthesis of MHC class I H chain and 2m was similar in parent and mutant cells (Fig. 2A, compare HC10 pulse bands). However, assembly of H chain and 2m to produce W6/32-reactive complexes was reduced markedly in the mutant cells (Fig. 2A). The levels of free H chain dropped relatively rapidly in 4S8.12 cells, although free H chain was not being converted into 2m-associated, W6/32-reactive H chain. This probably represents proteolytic degradation of misfolded or unassembled H chain, as part of the normal ER quality control system. (It is formally possible that H chain aggregates, takes on a conformation that is not recognized by available mAbs, or is otherwise sequestered from detection. However, we have not been able to find any evidence for such a hidden pool of H chain.) The small amount of assembled MHC class I in 4S8.12 probably represents the small subpopulation of cells in the overall 4S8.12 population that at any time have near-normal surface MHC class I levels.
There were two possible explanations for the reduction in H chain/2m complexes in immunoprecipitates from 4S8.12. H chain and 2m could have failed to assemble altogether. Alternatively, H chain and 2m might have assembled but failed to associate with the peptide. Because peptide-empty complexes are less stable than peptide-loaded complexes, the empty complexes could have dissociated during the immunoprecipitation procedure. To differentiate between these possibilities, we needed to confirm that, under our immunoprecipitation protocol, we were able to detect peptide-empty (unstable) complexes. Therefore, we performed immunoprecipitations in COS-Kb cells expressing the herpes simplex virus protein ICP47 (Fig. 2A), which binds to the TAP peptide transporter and blocks peptide transport into the ER (35, 36), using the adenovirus construct AdICP47. Surface expression of MHC class I is dependent generally on peptide supply to the ER, and ICP47 drastically reduced MHC class I surface expression in the parent COS-Kb cells (Fig. 2B). The reduction in surface expression of MHC class I in COS-Kb cells expressing ICP47 was comparable to that in COS-Kb cells treated with the ER-to-Gogli transport blocker brefeldin A (data not shown), demonstrating that peptide supply to nascent MHC class I complexes was almost completely blocked. Even under these conditions, when the H chain/2m complexes must have been peptide-empty and relatively unstable, we were able to immunoprecipitate W6/32-reactive complexes (Fig. 2A, COS-Kb + ICP47), even though some of the complexes dissociated during the 37°C incubation of the 45-min chase.
Additionally, we confirmed that W6/32 could recognize peptide-empty MHC class I by examining the thermostability of the complexes this Ab recognized (Fig. 2, C and D). Cell lysates were either kept on ice and immunoprecipitated immediately or incubated at 40°C for 1 h before immunoprecipitation. Under these conditions MHC, class I complexes could be immunoprecipitated from COS-Kb cells by W6/32 even after a 15-min pulse. However, these complexes were predominately thermolabile because incubation at 40°C for 1 h abrogated recognition by W6/32. The complexes became progressively more thermostable during the chase period, although at 15 min of chase and even after 45 min of chase a significant proportion of the complexes failed to immunoprecipitate after 40°C incubation. The increasing stability of the complexes presumably corresponds to acquisition of peptide as the complexes mature, and therefore, the dissociating (thermolabile) fraction represents peptide-empty complexes. Lysates from 4S8.12 cells, immunoprecipitated in parallel with those from the parent COS-Kb cells (but without the 40°C incubation), again showed few or no W6/32-reactive complexes. Thus, although W6/32 immunoprecipitated peptide-empty complexes immediately after the pulse, this Ab failed to recognize complexes from 4S8.12 cells.
These experiments confirmed that, with our immunoprecipitation conditions, we could detect assembly of MHC class I complexes even when the complexes were not stabilized by association with high-affinity peptide. Therefore, the inability to detect any W6/32-reactive complexes in 4S8.12 cells, in the same experiments, means that in these mutant cells H chain and 2m never assemble into their initial, peptide-empty configuration. Mutations in Ag presentation genes that prevent peptide transfer into the ER (TAP) or peptide loading (tapasin) are already known, while to our knowledge, no mutant cell lines in which normal H chain and 2m fail to assemble into a peptide-receptive complex have been described previously.
The reduced assembly of endogenous MHC class I and 2m could be due to a mutation in one of these molecules; for example, a dominant-negative mutant of 2m has been described which disrupts H chain/2m association (37). Levels of 2m were similar in 4S8.12 and COS-Kb cells (see below, Fig. 5B). We cloned and sequenced 2m from 4S8.12 and parent COS-Kb cells and found that the sequences were identical (GenBank accession no. AY570381). Both HLA-A-like and HLA-B-like molecules showed reduced expression in 4S8.12 cells (Fig. 1, B and C), showing that the effect was not a mutation in a single H chain allele. As well, transiently transfected human MHC class I molecules (e.g., HLA-A*0302; Fig. 1E) showed reduced assembly in 4S8.12 compared with parent COS-Kb cells. To further confirm that the reduction in assembly was not due to a primary mutation in endogenous 2m or MHC class I genes, we transfected the parent and mutant cells with a plasmid expressing human 2m covalently attached through a flexible linker to the N terminus of HLA-A*0302 (single-chain HLA-A3, "SC-A3"). In these transfectants, although both members of the complex were of wild-type sequence, assembly and surface expression of the HLA-A3/human 2m molecule (assessed with the conformation-sensitive Ab GAP-A3) was markedly lower in the mutant 4S8.12 than the parent COS-Kb cell line (Fig. 3A). Therefore, the mutation in these cells must affect an accessory factor that facilitates assembly of 2m with human and monkey MHC class I H chain.
Discussion
In this article, we describe a mutant cell line, 4S8.12, with a defect in quality control that manifests differently for primate and mouse MHC class I molecules. In 4S8.12 cells, monkey and human MHC class I H chains fail to assemble with 2m so that few peptide-receptive complexes form in the ER. In contrast, mouse MHC class I H chains apparently assemble normally with 2m and can bind to peptide as well as in normal cells. 4S8.12 cells show no growth abnormalities, and other glycoproteins (e.g., influenza hemagglutinin; Fig. 1D) are apparently unaffected by these defects, reaching the cell surface at wild-type levels and in a mature conformation.
Assembly and quality control in the ER is the function of chaperones, and we suggest that the phenotype of 4S8.12 cells is the result of a defect in a chaperone that facilitates assembly of MHC class I complexes. This putative chaperone has not yet been identified but does not appear to be one of the chaperones known to associate with MHC class I during its maturation: those chaperones are all present at normal levels and do not rescue the defect when transfected.
Mutations in several genes known to be involved in MHC class I Ag presentation (e.g., TAP and tapasin) prevent peptide from associating with H chain/2m complexes in the ER. These peptide-empty complexes, which are retained in the ER, are unstable and consequently can be difficult to detect under some conditions. However, we confirmed that in our experiments we were able to detect peptide-empty H chain/2m complexes in the ER. Therefore, the failure to detect such complexes in 4S8.12 cells was because H chain does not assemble with 2m in these cells.
Free HC in 4S8.12 cells is degraded more rapidly than in COS-Kb cells (Fig. 2, B and C; compare HC10 lanes). ER proteins that are misfolded or that fail to assemble with their appropriate binding partner are translocated normally from the ER and destroyed by ER-associated degradation pathways. The rapid degradation seen in 4S8.12 is probably because H chain fails to fold properly in these cells and is treated like any other misfolded protein. In fact, at very early time points (e.g., <10 min; data not shown), H chain in normal COS-Kb cells is degraded rapidly, at a rate very similar to that in 4S8.12, but in the normal cells, this rapid degradation quickly slows down, probably as H chain folds into a more native conformation. It is also conceivable that rapid degradation is the primary defect in 4S8.12 cells, analogous to the situation in human CMV-infected cells expressing the viral immune evasion proteins US2 and US11, which induce rapid ubiquitin-proteasome-mediated degradation of MHC class I H chain (47, 48). However, because a significant amount of free H chain is still present in 4S8.12 cells even after 45 min of chase (e.g., Fig. 2A), yet very few H chain/2m complexes are formed at that time, it is unlikely that rapid degradation is the sole abnormality in these cells.
The defect in 4S8.12 cells is partial, because a low level (5% of wild type) of apparently properly assembled monkey MHC class I complexes are present at the cell surface. As well, a minority of cells in the population express normal levels of endogenous MHC class I on their surface and some environmental conditions, such as growth in glucose-depleted medium or treatment with deoxyglucose or the calcium ionophore A23187 (data not shown), all treatments that strongly induce many ER chaperones, at least partially reverse the defect. We do not know whether these normal-appearing cells are genuinely completely normal or whether they may still have subtle defects in MHC conformation or function. As well, the reversal of the phenotype under these conditions may be due to correction of the underlying defect or may involve overexpression of other chaperones that compensate for the defect: redundancy in chaperone function is common.
MHC class I molecules undergo a precisely orchestrated program of maturation in the ER that results in expression of a mature complex at the cell surface. If any of the three components of the complex (H chain, 2m, or peptide) is missing, the incomplete complex is normally retained within the ER and ultimately degraded. Assembly of H chain and 2m in vivo is presumably facilitated by chaperones, and indeed, the MHC class I H chain associates with a series of ER chaperones during its maturation: GRP78, CNX, CRT, and ERp57. These chaperones associate with many other nascent ER proteins and glycoproteins (49, 50) and generally facilitate proper protein folding (51, 52). However, while the general function of these chaperones is clear, their precise roles in MHC class I maturation are not well understood. CNX has been shown to enhance folding and assembly of MHC class I (19) and may also help retain misfolded complexes in the ER. However, even in a cell line that lacks CNX, MHC maturation proceeds apparently normally (53). After H chain binds to 2m, CNX dissociates from human H chain, and the H chain/2m dimer associates with several other proteins to form a peptide-binding complex, containing at least CRT, ERp57, tapasin, and TAP. CRT plays a role in loading MHC complexes with optimal peptide (11). ERp57, which has PDI motifs, is required for the production of a disulfide bond between tapasin and the MHC class I peptide-binding groove (54); however, the physiological importance of this disulfide link is not clear. ERp57 has also been implicated in the production of intrachain disulfide bonds in the H chain (55), but again, the importance of this is not clear.
We tested 4S8.12 cells for the presence and function of multiple chaperones. All those tested showed equal levels by Western blots (Fig. 5A), and transfection of 4S8.12 cells with wild-type chaperones did not restore expression of MHC class I (Fig. 5, C–F). In addition, we confirmed that the primary sequence of several genes critical to MHC class I folding and assembly (ERp57, CRT, and 2m) were all wild type.
Tapasin is involved in loading of H chain/2m complexes with peptide and in retention of peptide-empty complexes in the ER (13, 14, 18, 45, 46). Because tapasin deficiency leads to a marked reduction in peptide binding to newly synthesized MHC class I, it was a strong candidate as the mutant gene in 4S8.12 cells. However, several lines of evidence rule this possibility out. First, presentation of SIINFEKL on H-2Kb in 4S8.12 cells was similar to, or higher than, that in parent cells (Fig. 3, D and E), whether the SIINFEKL was delivered directly into the ER (bypassing TAP) or generated by proteasome-mediated degradation of full-length OVA. Efficient generation of H-2Kb-SIINFEKL complexes has been shown previously to be dependent on tapasin (45, 46). Second, MHC class I complexes are associated physically with TAP in parent and mutant cells (Fig. 5B), which requires tapasin (13, 14). Third, sequencing of cDNA from parent and mutant cells showed no change in the primary sequencing of tapasin. Finally, transfection of 4S8.12 cells with wild-type tapasin did not restore surface MHC class I to wild-type levels (Fig. 5G).
Primate and mouse MHC class I H chains behaved very differently in the 4S8.12 cell line. Whereas human and monkey MHC class I complexes failed to associate with 2m and were retained in the ER until they were degraded, mouse MHC class I H chains reached the cell surface at least as well as in the wild-type COS-Kb cells; in fact, the mouse alleles tested reached higher levels at the surface of 4S8.12 than on COS-Kb cells (Fig. 4, A–C). As well, although human MHC class I failed to interact with binding peptide (Fig. 3), mouse H-2Kb in 4S8.12 cells presented peptide at least as well as did wild-type cells (Fig. 4, D and E). Therefore, the putative chaperone that is defective in 4S8.12 cells is essential for folding and assembly of human MHC class I but is not required (or is redundant) for the assembly of mouse MHC class I. Mouse and human MHC class I alleles are known to interact differently with ER chaperones: in particular, mouse alleles remain associated with CNX even after 2m association, whereas CNX dissociates from human alleles at this stage (6). However, CNX is present in 4S8.12 at normal levels (Fig. 5A), and transfection of wild-type CNX into 4S8.12 does not alter expression of human MHC class I (Fig. 5C) or of H-2Kb (data not shown); as well, the phenotype of a mutant human cell line lacking CNX is quite different from that of 4S8.12 (53), and disruption of CNX association with human MHC class I does not inhibit association with 2m (56) as occurs in 4S8.12.
Although most if not all glycoproteins interact with chaperones during their maturation, MHC class I complexes may be unusual in their chaperone requirements because for their normal function the complexes must achieve a nearly fully folded conformation to bind peptides in the ER. The difference in conformation between peptide-empty and peptide-loaded H chain/2m complexes is very subtle; for example, almost all mAbs that recognize conformational determinants on MHC class I recognize both the peptide-empty and loaded complexes, although not always with equal efficiency. Yet the peptide-empty complex is normally efficiently retained within the ER, whereas the peptide-loaded form is released. Presumably, there is some subtle conformational change that permits MHC class I to exit the ER.
The gene that is affected in 4S8.12 cells is not yet identified. Most known chaperones in the MHC class I pathway were identified based on finding a physical association with MHC class I, but in immunoprecipitations, we have not detected differences in molecules that coprecipitate with MHC class I in 4S8.12 and wild-type COS-Kb cells (data not shown). However, such biochemical analyses are limited by the ability to preserve and detect stable intermolecular associations. In contrast, this is not a limitation of mutational analysis, which allows genetic screens to identify essential new steps in complex pathways. The 4S8.12 cells provide strong genetic evidence for a new component in the MHC class I pathway. This novel gene is important in the folding and assembly of primate but not mouse MHC class I.
Footnotes
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.
1 These studies were supported by grants from the National Institutes of Health (to K.L.R.).
2 Address correspondence and reprint requests to Dr. Ian A. York, Department of Pathology, University of Massachusetts Medical Center, 5 Lake Avenue, Worcester MA 01655. E-mail address: Ian.York{at}umassmed.edu
3 Current address: Millennium Pharmaceuticals, Cambridge, MA 02139.
4 Current address: Ipsen, London, U.K.
5 Abbreviations used in this paper: 2m, 2-microglobulin; ER, endoplasmic reticulum; PDI, protein disulfide isomerase; CNX, calnexin; CRT, calreticulin.
Received for publication April 9, 2004. Accepted for publication March 29, 2005.
References
]
York, I. A., A. L. Goldberg, X. Y. Mo, K. L. Rock. 1999. Proteolysis and class I major histocompatibility complex antigen presentation. Immunol. Rev. 172: 49-66.
Shastri, N., S. Schwab, T. Serwold. 2002. Producing nature’s gene-chips: the generation of peptides for display by MHC class I molecules. Annu. Rev. Immunol. 20: 463-493.
Solheim, J. C.. 1999. Class I MHC molecules: assembly and antigen presentation. Immunol. Rev. 172: 11-19.
Paulsson, K., P. Wang. 2003. Chaperones and folding of MHC class I molecules in the endoplasmic reticulum. Biochim. Biophys. Acta 1641: 1-12.
Kahn-Perles, B., J. Salamero, O. Jouans. 1994. Biogenesis of MHC class I antigens: involvement of multiple chaperone molecules. Eur. J. Cell Biol. 64: 176-185.
Nossner, E., P. Parham. 1995. Species-specific differences in chaperone interaction of human and mouse major histocompatibility complex class I molecules. J. Exp. Med. 181: 327-337.(Ian A. York2,*, Ethan P. )
COS7 (African Green Monkey kidney) cells stably transfected with the mouse MHC class I allele H-2Kb were mutagenized, selected for low surface expression of endogenous MHC class I products, and subcloned. A mutant cell line, 4S8.12, expressing very low surface MHC class I (5% of parental levels) was identified. This cell line synthesized normal levels of the MHC class I H chain and 2-microglobulin, as well as normal levels of TAP, tapasin, GRP78, calnexin, calreticulin, ERp57, and protein disulfide isomerase. Full-length OVA was processed to generate presented H-2Kb-SIINFEKL complexes with equal efficiency in wild-type and mutant cells, demonstrating that proteasomes, as well as TAP and tapasin, functioned normally. Therefore, all the known components of the MHC class I Ag presentation pathway were intact. Nevertheless, primate (human and monkey) MHC class I H chain and 2-microglobulin failed to associate to form the normal peptide-receptive complex. In contrast, mouse H chains associated with 2-microglobulin normally and bound peptide at least as well as in wild-type cells. The 4S8.12 cells provide strong genetic evidence for a novel component in the MHC class I pathway. This as-yet unidentified gene is important in early assembly of primate, but not mouse, MHC class I complexes.
Introduction
Cytotoxic T lymphocyte survey tissues for the presence of foreign Ags, such as those of viral or tumor origin, presented on MHC class I molecules on the surface of cells. MHC class I molecules are trimolecular complexes, consisting of a polymorphic glycoprotein, the H chain, which is noncovalently associated with 2-microglobulin (2m) 5, and a small peptide, generally 8–10 residues long, which is usually the product of proteolysis of cellular and viral proteins in the cytosol (reviewed in Refs. 1 and 2).
H chain and 2m are cotranslationally translocated into the endoplasmic reticulum (ER), where they fold, form intrachain disulfide bonds, and associate with each other. Before binding peptide, this complex is relatively thermolabile and is retained within the ER. Peptides are transported into the lumen of the ER by the peptide transporter TAP, and if a peptide has the appropriate sequence motif, it can bind to the H chain/2m complex. Upon binding peptide, the association between H chain/2m becomes more stable, and the trimolecular complex is released from the ER and transported to the cell surface.
Each stage of MHC class I maturation is associated with ER chaperones (reviewed in Refs. 3 and 4). GRP78 (BiP) interacts with H chain shortly after its synthesis (5, 6). Protein disulfide isomerase (PDI) may help form some of the H chain intrachain disulfide bonds. Calnexin (CNX) interacts with newly synthesized H chain. CNX dissociates from human MHC when H chain interacts with 2m but remains associated with mouse H chain/2m complexes (6, 7). Calreticulin (CRT) and ERp57 both associate with H chain/2m complexes (8, 9, 10, 11, 12). Tapasin brings the peptide-receptive H chain/2m complex into physical proximity with TAP-transported peptides (13, 14, 15) and otherwise facilitates peptide binding to the H chain/2m dimer (16, 17, 18). When peptide is bound to the MHC complex, the chaperones all dissociate from MHC, and the mature H chain/2m/peptide complex is allowed to exit the ER to the cell surface.
In general, these chaperones are believed to perform quality control, monitoring protein folding and either directing folding toward the proper conformation or targeting misfolded protein to degradation. Thus, various chaperones have been shown to facilitate folding and assembly (19) and retain peptide-empty MHC class I in the ER (16, 18). However, the details of these functions are not well understood as yet.
The identity and functions of several proteins associated with Ag processing, such as TAP and tapasin, have been elucidated by studying mutant cell lines with defects in cell surface MHC class I expression. To further investigate the MHC class I pathway, we generated a series of mutant cell lines with defects in surface MHC class I expression and in this study describe one of these mutants that has a defect in ER quality control.
Results
4S8.12 cells have reduced endogenous MHC class I surface expression
We chose to mutagenize COS7 cells because these cells episomally replicate plasmids containing the SV40 origin of replication, leading to high expression of transfected genes, and because they react broadly with the many available reagents directed against components of the human Ag processing system. These characteristics make expression cloning of mutant genes practical in COS7 cells. To study presentation of defined Ags, we stably transfected COS7 with the well-characterized mouse MHC allele H-2Kb to produce COS-Kb cells. COS-Kb cells were mutagenized with ethyl methanesulfonate and repeatedly selected for low surface expression of MHC class I molecules, using a panel of anti-MHC class I Abs bound to magnetic beads. After repeated selections, the population was subcloned by limiting dilution, and subclones were analyzed by flow cytometry for low surface MHC class I expression. The 4S8.12 subclone expressed, on average, 5% of wild-type levels of MHC class I reactive with either PA2.6 or W6/32 mAbs, both of which recognize all HLA-A, -B, and -C alleles in association with 2m (Fig. 1A). As well, HLA-B-like molecules associated with 2m (mAb 4E-reactive) (Fig. 1B) and HLA-A-like molecules associated with 2m (reactive with the mAb JD12) (Fig. 1C) were both markedly lower than the parent cell line’s levels.
The decreased surface expression of endogenous MHC class I did not reflect a general defect in glycoprotein expression because transiently transfected influenza hemagglutinin was expressed at equal levels on the surface of 4S8.12 and wild-type COS-Kb cells (Fig. 1D); however, the defect was not limited to the endogenous monkey MHC class I because transiently transfected human MHC class I alleles (e.g., HLA-A*0302; Fig. 1E) were also expressed at lower levels on 4S8.12 cells than on COS-Kb.
The 4S8.12 population showed biphasic levels of endogenous surface MHC class I, with a minority of cells (5–20% under normal culture conditions) expressing nearly wild-type levels of MHC class I (Fig. 1A). This was not because of contamination with wild-type cells because repeated subcloning by limiting dilution invariably yielded biphasic populations (at least 30 subclones), rather than either uniformly high and low populations (data not shown). Furthermore, when nearly pure populations of either low- or high-MHC class I-expressing 4S8.12 cells were selected using anti-MHC class I Ab-coated magnetic beads, each population became biphasic again within 24 h of culture (Fig. 1F). This suggests that individual cells in the population rapidly switch between a normal MHC class I phenotype and a mutant (low surface MHC class I) phenotype. The trigger for this switch is unknown but may be related to environmental conditions. It is not related to the phase of cell cycle (data not shown).
Assembly of endogenous H chain/2m complexes is defective in 4S8.12 cells
Endogenous (monkey) MHC class I is expressed on the surface of 4S8.12 at markedly lower levels than in wild-type cells (Fig. 1A). This could have many causes: reduced MHC class I synthesis, failure to assemble 2m and H chain molecules in the ER, reduced peptide supply or peptide binding, increased degradation of H chain or 2m, or because mature (peptide-containing) molecules are abnormally retained in the ER. We analyzed maturation of MHC class I in wild-type and mutant cells, using the mAbs W6/32 (which recognizes MHC class I only when it is assembled with 2m, regardless of the presence or absence of peptide) and HC10, which recognizes only H chain that is not assembled with 2m. Synthesis of MHC class I H chain and 2m was similar in parent and mutant cells (Fig. 2A, compare HC10 pulse bands). However, assembly of H chain and 2m to produce W6/32-reactive complexes was reduced markedly in the mutant cells (Fig. 2A). The levels of free H chain dropped relatively rapidly in 4S8.12 cells, although free H chain was not being converted into 2m-associated, W6/32-reactive H chain. This probably represents proteolytic degradation of misfolded or unassembled H chain, as part of the normal ER quality control system. (It is formally possible that H chain aggregates, takes on a conformation that is not recognized by available mAbs, or is otherwise sequestered from detection. However, we have not been able to find any evidence for such a hidden pool of H chain.) The small amount of assembled MHC class I in 4S8.12 probably represents the small subpopulation of cells in the overall 4S8.12 population that at any time have near-normal surface MHC class I levels.
There were two possible explanations for the reduction in H chain/2m complexes in immunoprecipitates from 4S8.12. H chain and 2m could have failed to assemble altogether. Alternatively, H chain and 2m might have assembled but failed to associate with the peptide. Because peptide-empty complexes are less stable than peptide-loaded complexes, the empty complexes could have dissociated during the immunoprecipitation procedure. To differentiate between these possibilities, we needed to confirm that, under our immunoprecipitation protocol, we were able to detect peptide-empty (unstable) complexes. Therefore, we performed immunoprecipitations in COS-Kb cells expressing the herpes simplex virus protein ICP47 (Fig. 2A), which binds to the TAP peptide transporter and blocks peptide transport into the ER (35, 36), using the adenovirus construct AdICP47. Surface expression of MHC class I is dependent generally on peptide supply to the ER, and ICP47 drastically reduced MHC class I surface expression in the parent COS-Kb cells (Fig. 2B). The reduction in surface expression of MHC class I in COS-Kb cells expressing ICP47 was comparable to that in COS-Kb cells treated with the ER-to-Gogli transport blocker brefeldin A (data not shown), demonstrating that peptide supply to nascent MHC class I complexes was almost completely blocked. Even under these conditions, when the H chain/2m complexes must have been peptide-empty and relatively unstable, we were able to immunoprecipitate W6/32-reactive complexes (Fig. 2A, COS-Kb + ICP47), even though some of the complexes dissociated during the 37°C incubation of the 45-min chase.
Additionally, we confirmed that W6/32 could recognize peptide-empty MHC class I by examining the thermostability of the complexes this Ab recognized (Fig. 2, C and D). Cell lysates were either kept on ice and immunoprecipitated immediately or incubated at 40°C for 1 h before immunoprecipitation. Under these conditions MHC, class I complexes could be immunoprecipitated from COS-Kb cells by W6/32 even after a 15-min pulse. However, these complexes were predominately thermolabile because incubation at 40°C for 1 h abrogated recognition by W6/32. The complexes became progressively more thermostable during the chase period, although at 15 min of chase and even after 45 min of chase a significant proportion of the complexes failed to immunoprecipitate after 40°C incubation. The increasing stability of the complexes presumably corresponds to acquisition of peptide as the complexes mature, and therefore, the dissociating (thermolabile) fraction represents peptide-empty complexes. Lysates from 4S8.12 cells, immunoprecipitated in parallel with those from the parent COS-Kb cells (but without the 40°C incubation), again showed few or no W6/32-reactive complexes. Thus, although W6/32 immunoprecipitated peptide-empty complexes immediately after the pulse, this Ab failed to recognize complexes from 4S8.12 cells.
These experiments confirmed that, with our immunoprecipitation conditions, we could detect assembly of MHC class I complexes even when the complexes were not stabilized by association with high-affinity peptide. Therefore, the inability to detect any W6/32-reactive complexes in 4S8.12 cells, in the same experiments, means that in these mutant cells H chain and 2m never assemble into their initial, peptide-empty configuration. Mutations in Ag presentation genes that prevent peptide transfer into the ER (TAP) or peptide loading (tapasin) are already known, while to our knowledge, no mutant cell lines in which normal H chain and 2m fail to assemble into a peptide-receptive complex have been described previously.
The reduced assembly of endogenous MHC class I and 2m could be due to a mutation in one of these molecules; for example, a dominant-negative mutant of 2m has been described which disrupts H chain/2m association (37). Levels of 2m were similar in 4S8.12 and COS-Kb cells (see below, Fig. 5B). We cloned and sequenced 2m from 4S8.12 and parent COS-Kb cells and found that the sequences were identical (GenBank accession no. AY570381). Both HLA-A-like and HLA-B-like molecules showed reduced expression in 4S8.12 cells (Fig. 1, B and C), showing that the effect was not a mutation in a single H chain allele. As well, transiently transfected human MHC class I molecules (e.g., HLA-A*0302; Fig. 1E) showed reduced assembly in 4S8.12 compared with parent COS-Kb cells. To further confirm that the reduction in assembly was not due to a primary mutation in endogenous 2m or MHC class I genes, we transfected the parent and mutant cells with a plasmid expressing human 2m covalently attached through a flexible linker to the N terminus of HLA-A*0302 (single-chain HLA-A3, "SC-A3"). In these transfectants, although both members of the complex were of wild-type sequence, assembly and surface expression of the HLA-A3/human 2m molecule (assessed with the conformation-sensitive Ab GAP-A3) was markedly lower in the mutant 4S8.12 than the parent COS-Kb cell line (Fig. 3A). Therefore, the mutation in these cells must affect an accessory factor that facilitates assembly of 2m with human and monkey MHC class I H chain.
Discussion
In this article, we describe a mutant cell line, 4S8.12, with a defect in quality control that manifests differently for primate and mouse MHC class I molecules. In 4S8.12 cells, monkey and human MHC class I H chains fail to assemble with 2m so that few peptide-receptive complexes form in the ER. In contrast, mouse MHC class I H chains apparently assemble normally with 2m and can bind to peptide as well as in normal cells. 4S8.12 cells show no growth abnormalities, and other glycoproteins (e.g., influenza hemagglutinin; Fig. 1D) are apparently unaffected by these defects, reaching the cell surface at wild-type levels and in a mature conformation.
Assembly and quality control in the ER is the function of chaperones, and we suggest that the phenotype of 4S8.12 cells is the result of a defect in a chaperone that facilitates assembly of MHC class I complexes. This putative chaperone has not yet been identified but does not appear to be one of the chaperones known to associate with MHC class I during its maturation: those chaperones are all present at normal levels and do not rescue the defect when transfected.
Mutations in several genes known to be involved in MHC class I Ag presentation (e.g., TAP and tapasin) prevent peptide from associating with H chain/2m complexes in the ER. These peptide-empty complexes, which are retained in the ER, are unstable and consequently can be difficult to detect under some conditions. However, we confirmed that in our experiments we were able to detect peptide-empty H chain/2m complexes in the ER. Therefore, the failure to detect such complexes in 4S8.12 cells was because H chain does not assemble with 2m in these cells.
Free HC in 4S8.12 cells is degraded more rapidly than in COS-Kb cells (Fig. 2, B and C; compare HC10 lanes). ER proteins that are misfolded or that fail to assemble with their appropriate binding partner are translocated normally from the ER and destroyed by ER-associated degradation pathways. The rapid degradation seen in 4S8.12 is probably because H chain fails to fold properly in these cells and is treated like any other misfolded protein. In fact, at very early time points (e.g., <10 min; data not shown), H chain in normal COS-Kb cells is degraded rapidly, at a rate very similar to that in 4S8.12, but in the normal cells, this rapid degradation quickly slows down, probably as H chain folds into a more native conformation. It is also conceivable that rapid degradation is the primary defect in 4S8.12 cells, analogous to the situation in human CMV-infected cells expressing the viral immune evasion proteins US2 and US11, which induce rapid ubiquitin-proteasome-mediated degradation of MHC class I H chain (47, 48). However, because a significant amount of free H chain is still present in 4S8.12 cells even after 45 min of chase (e.g., Fig. 2A), yet very few H chain/2m complexes are formed at that time, it is unlikely that rapid degradation is the sole abnormality in these cells.
The defect in 4S8.12 cells is partial, because a low level (5% of wild type) of apparently properly assembled monkey MHC class I complexes are present at the cell surface. As well, a minority of cells in the population express normal levels of endogenous MHC class I on their surface and some environmental conditions, such as growth in glucose-depleted medium or treatment with deoxyglucose or the calcium ionophore A23187 (data not shown), all treatments that strongly induce many ER chaperones, at least partially reverse the defect. We do not know whether these normal-appearing cells are genuinely completely normal or whether they may still have subtle defects in MHC conformation or function. As well, the reversal of the phenotype under these conditions may be due to correction of the underlying defect or may involve overexpression of other chaperones that compensate for the defect: redundancy in chaperone function is common.
MHC class I molecules undergo a precisely orchestrated program of maturation in the ER that results in expression of a mature complex at the cell surface. If any of the three components of the complex (H chain, 2m, or peptide) is missing, the incomplete complex is normally retained within the ER and ultimately degraded. Assembly of H chain and 2m in vivo is presumably facilitated by chaperones, and indeed, the MHC class I H chain associates with a series of ER chaperones during its maturation: GRP78, CNX, CRT, and ERp57. These chaperones associate with many other nascent ER proteins and glycoproteins (49, 50) and generally facilitate proper protein folding (51, 52). However, while the general function of these chaperones is clear, their precise roles in MHC class I maturation are not well understood. CNX has been shown to enhance folding and assembly of MHC class I (19) and may also help retain misfolded complexes in the ER. However, even in a cell line that lacks CNX, MHC maturation proceeds apparently normally (53). After H chain binds to 2m, CNX dissociates from human H chain, and the H chain/2m dimer associates with several other proteins to form a peptide-binding complex, containing at least CRT, ERp57, tapasin, and TAP. CRT plays a role in loading MHC complexes with optimal peptide (11). ERp57, which has PDI motifs, is required for the production of a disulfide bond between tapasin and the MHC class I peptide-binding groove (54); however, the physiological importance of this disulfide link is not clear. ERp57 has also been implicated in the production of intrachain disulfide bonds in the H chain (55), but again, the importance of this is not clear.
We tested 4S8.12 cells for the presence and function of multiple chaperones. All those tested showed equal levels by Western blots (Fig. 5A), and transfection of 4S8.12 cells with wild-type chaperones did not restore expression of MHC class I (Fig. 5, C–F). In addition, we confirmed that the primary sequence of several genes critical to MHC class I folding and assembly (ERp57, CRT, and 2m) were all wild type.
Tapasin is involved in loading of H chain/2m complexes with peptide and in retention of peptide-empty complexes in the ER (13, 14, 18, 45, 46). Because tapasin deficiency leads to a marked reduction in peptide binding to newly synthesized MHC class I, it was a strong candidate as the mutant gene in 4S8.12 cells. However, several lines of evidence rule this possibility out. First, presentation of SIINFEKL on H-2Kb in 4S8.12 cells was similar to, or higher than, that in parent cells (Fig. 3, D and E), whether the SIINFEKL was delivered directly into the ER (bypassing TAP) or generated by proteasome-mediated degradation of full-length OVA. Efficient generation of H-2Kb-SIINFEKL complexes has been shown previously to be dependent on tapasin (45, 46). Second, MHC class I complexes are associated physically with TAP in parent and mutant cells (Fig. 5B), which requires tapasin (13, 14). Third, sequencing of cDNA from parent and mutant cells showed no change in the primary sequencing of tapasin. Finally, transfection of 4S8.12 cells with wild-type tapasin did not restore surface MHC class I to wild-type levels (Fig. 5G).
Primate and mouse MHC class I H chains behaved very differently in the 4S8.12 cell line. Whereas human and monkey MHC class I complexes failed to associate with 2m and were retained in the ER until they were degraded, mouse MHC class I H chains reached the cell surface at least as well as in the wild-type COS-Kb cells; in fact, the mouse alleles tested reached higher levels at the surface of 4S8.12 than on COS-Kb cells (Fig. 4, A–C). As well, although human MHC class I failed to interact with binding peptide (Fig. 3), mouse H-2Kb in 4S8.12 cells presented peptide at least as well as did wild-type cells (Fig. 4, D and E). Therefore, the putative chaperone that is defective in 4S8.12 cells is essential for folding and assembly of human MHC class I but is not required (or is redundant) for the assembly of mouse MHC class I. Mouse and human MHC class I alleles are known to interact differently with ER chaperones: in particular, mouse alleles remain associated with CNX even after 2m association, whereas CNX dissociates from human alleles at this stage (6). However, CNX is present in 4S8.12 at normal levels (Fig. 5A), and transfection of wild-type CNX into 4S8.12 does not alter expression of human MHC class I (Fig. 5C) or of H-2Kb (data not shown); as well, the phenotype of a mutant human cell line lacking CNX is quite different from that of 4S8.12 (53), and disruption of CNX association with human MHC class I does not inhibit association with 2m (56) as occurs in 4S8.12.
Although most if not all glycoproteins interact with chaperones during their maturation, MHC class I complexes may be unusual in their chaperone requirements because for their normal function the complexes must achieve a nearly fully folded conformation to bind peptides in the ER. The difference in conformation between peptide-empty and peptide-loaded H chain/2m complexes is very subtle; for example, almost all mAbs that recognize conformational determinants on MHC class I recognize both the peptide-empty and loaded complexes, although not always with equal efficiency. Yet the peptide-empty complex is normally efficiently retained within the ER, whereas the peptide-loaded form is released. Presumably, there is some subtle conformational change that permits MHC class I to exit the ER.
The gene that is affected in 4S8.12 cells is not yet identified. Most known chaperones in the MHC class I pathway were identified based on finding a physical association with MHC class I, but in immunoprecipitations, we have not detected differences in molecules that coprecipitate with MHC class I in 4S8.12 and wild-type COS-Kb cells (data not shown). However, such biochemical analyses are limited by the ability to preserve and detect stable intermolecular associations. In contrast, this is not a limitation of mutational analysis, which allows genetic screens to identify essential new steps in complex pathways. The 4S8.12 cells provide strong genetic evidence for a new component in the MHC class I pathway. This novel gene is important in the folding and assembly of primate but not mouse MHC class I.
Footnotes
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.
1 These studies were supported by grants from the National Institutes of Health (to K.L.R.).
2 Address correspondence and reprint requests to Dr. Ian A. York, Department of Pathology, University of Massachusetts Medical Center, 5 Lake Avenue, Worcester MA 01655. E-mail address: Ian.York{at}umassmed.edu
3 Current address: Millennium Pharmaceuticals, Cambridge, MA 02139.
4 Current address: Ipsen, London, U.K.
5 Abbreviations used in this paper: 2m, 2-microglobulin; ER, endoplasmic reticulum; PDI, protein disulfide isomerase; CNX, calnexin; CRT, calreticulin.
Received for publication April 9, 2004. Accepted for publication March 29, 2005.
References
]
York, I. A., A. L. Goldberg, X. Y. Mo, K. L. Rock. 1999. Proteolysis and class I major histocompatibility complex antigen presentation. Immunol. Rev. 172: 49-66.
Shastri, N., S. Schwab, T. Serwold. 2002. Producing nature’s gene-chips: the generation of peptides for display by MHC class I molecules. Annu. Rev. Immunol. 20: 463-493.
Solheim, J. C.. 1999. Class I MHC molecules: assembly and antigen presentation. Immunol. Rev. 172: 11-19.
Paulsson, K., P. Wang. 2003. Chaperones and folding of MHC class I molecules in the endoplasmic reticulum. Biochim. Biophys. Acta 1641: 1-12.
Kahn-Perles, B., J. Salamero, O. Jouans. 1994. Biogenesis of MHC class I antigens: involvement of multiple chaperone molecules. Eur. J. Cell Biol. 64: 176-185.
Nossner, E., P. Parham. 1995. Species-specific differences in chaperone interaction of human and mouse major histocompatibility complex class I molecules. J. Exp. Med. 181: 327-337.(Ian A. York2,*, Ethan P. )