| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
and DM
Chain Cooperate in the Oxidation and Folding of HLA-DM1
Department of Biological Sciences, University of Durham, Durham, United Kingdom
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
and 
-chains fold, oxidize, and form a complex in the endoplasmic reticulum (ER). In this study, we show that glycosylation promotes, but is not essential for, DM ER exit. However, glycosylation of DM N15 is required for oxidation of the -chain. The DM and -chains direct each others fate: single DM chains cannot fully oxidize without DM, while DM forms disulfide-linked homodimers without DM. Correct oxidation and subsequent ER egress depend on the unique DM C25 and C35 residues. This suggests that the C25-C35 disulfide bond in the peptide-binding domain overcomes the need for stabilizing peptides required by other MHC molecules. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
-2-microglobulin, and engage the peptide-loading complex. The peptide-loading complex consists of TAP, tapasin, calreticulin, and ERp57 (4). Once the MHC class I H chain/-2-microglobulin acquires a peptide it can pass the endoplasmic reticulum (ER)3 quality control checks and is exported to the plasma membrane for scrutiny by CD8+ T cells (5). Classical class II molecules primarily present peptides from exogenous sources, acquired in the endocytic pathway. MHC class II - and -chains form heterodimers in the ER and then dock onto a trimer of invariant chains (Ii) resulting in a nonameric complex (6, 7). Association with the Ii occludes the peptide-binding groove preventing premature binding of peptide Ags (8, 9). The Ii is a prerequisite for export of MHC class II molecules to lysosome-like compartments (10, 11) where most of the Ii is subsequently degraded, leaving a small peptide (termed CLIP) occupying the class II peptide-binding domain (12). CLIP and other weakly binding peptides are exchanged for more stably binding peptides by the nonclassical MHC class II molecule DM (13, 14, 15). The importance of DM is emphasized by the severe Ag-presentation defects in mutant DM cell lines (16, 17) and mice lacking DM (18, 19, 20). MHC class II-bound peptides are subsequently presented at the cell surface to CD4+ T cells.
Like classical MHC class II molecules, DM is a heterodimer consisting of an - and -chain. Together with HLA-DO (DO), DM comprises the nonclassical MHC class II molecules that are unable to bind peptides (21). In contrast to classical MHC class II molecules that rely on the Ii for transport to the MHC class II molecules to lysosome-like compartments, DM is self-sufficient. The cytoplasmic domain of the DM chain contains a lysosomal targeting signal that directs DM into the endosomal/lysosomal pathway, ensuring colocalization of catalyst (DM) and substrate (22, 23). The ability of DM to pass the ER quality control mechanisms without any other apparent accessory proteins makes it unique among MHC molecules.
MHC molecules, as transmembrane heterodimers, present an interesting substrate for ER quality control machinery. During early folding and assembly stages, they engage with several ER resident chaperones. MHC class I molecules interact with calnexin (2) and calreticulin (4), while MHC class II proteins interact with calnexin (24, 25), ERp72, and GRP94 (10). MHC molecules also require disulfide bonds (26, 27, 28). In general, protein disulfide isomerases catalyze the formation of disulfide bonds in various substrates (oxidation) (29). ER oxidoreductins (Ero1p in yeast, Ero1-L, and Ero1-L in mammals) reoxidize protein disulfide isomerases using oxygen as a final electron acceptor (30, 31, 32, 33, 34).
DM has distinct disulfide bond arrangements compared with other MHC class II molecules. Given the unique ability of DM to traffic unaided out of the ER, we investigated the role of the disulfide bonds and glycosylation on oxidative protein folding and pairing of DM molecules. Our results show that glycosylation is not required for DM pairing and ER egress. However, DM N15 is required for the native oxidation pathway. MHC class II and DM -chains are partially endoH sensitive and we now identify the glycan that remains in the high-mannose form as the sugar attached to DM N165. Oxidative folding and quality control is primarily regulated by cross-talk between DM and DM. DM prevents misoxidation of DM into disulfide-linked DM homodimers, whereas the structural integrity of DM, conferred by its unique C25-C35 disulfide bond, is required for correct oxidation of the DM chain. Thus, oxidation of the DM heterodimer relies on the ability of the individual chains to direct the oxidation pathway of the partner chain, which has implications for the oxidation and assembly of other MHC family proteins and heterodimers in the ER.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
HeLa cells were maintained in MEM (Invitrogen Life Technologies) and MelJuSo cells (a gift from Prof. J. Neefjes, The Netherlands Cancer Institute, Amsterdam, The Netherlands) were maintained in DMEM (Invitrogen Life Technologies), supplemented with 8% FCS (Sigma-Aldrich), 2 mM glutamax, 100 U/ml penicillin, and 100 µg/ml streptomycin (Invitrogen Life Technologies). Cells were kept at 5% CO2, 37°C. Transfections with Lipofectamine 2000 (Invitrogen Life Technologies) were done according to manufacturer’s instructions. Subconfluent cells in 6-cm dishes were transfected with 1–2 µg of DNA in Opti-MEM serum-free medium. The medium was replaced after 6 h with normal medium and cells were analyzed 24 h posttransfection.
For DTT treatments, cells were incubated in medium with 10 mM DTT for 10 min. After washing with medium, the cells were chased in MEM in the absence of DTT for up to 30 min.
The mAbs against hemagglutinin (HA-7; Sigma-Aldrich), myc (9B11; New England Biolabs), and Ii (Abcam) were commercially available. The 1B5 mAb against DR, HC10 against MHC class I, and the polyclonal serum against DR were a gift from Prof. J. Neefjes.
Immunofluorescence
Cells grown on coverslips were fixed with ice-cold methanol for 10 min, followed by blocking in 0.2% BSA (Sigma-Aldrich) in PBS for 30 min. Coverslips were subsequently incubated with primary Abs, washed, and decorated with fluorescently labeled secondary Abs (Alexa 488; Invitrogen Life Technologies). After washing, coverslips were mounted with Vectashield and imaged with a fluorescence microscope (Axiovert 10; Zeiss). For ER tracker (Invitrogen Life Technologies) and Lysotracker Red (Invitrogen Life Technologies) labelings, cells were incubated with 1 µM ER tracker and 50 nM Lysotracker, respectively, for 30 min, before imaging without fixation.
Constructs and mutagenesis
HA-tagged DM and myc-tagged DM were constructed by PCR with sense primers specific for DM and DM, respectively. The antisense primers contained the 3' end of either chain lacking the stop codon, followed by either a HA tag or a myc tag and a stop codon. Both were cloned into pcDNA3 vectors (Invitrogen Life Technologies) and verified by DNA sequencing. The DR and Ii constructs were a gift from Prof. J. Neefjes.
All single amino acid changes were made using the Quik Change Site-Directed Mutagenesis kit (Stratagene).
Immunoprecipitation
Cells were lysed in 1% Triton X-100, 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM EDTA, and a mixture of 10 µg/ml protease inhibitors (chymostatin, leupeptin, antipain and pepstatin A; Sigma-Aldrich) in the presence of 20 mM iodoacetic acid (IAA) or N-ethylmaleimide (NEM) (Sigma-Aldrich) as a sulfhydryl-alkylating agent. Postnuclear lysates were either directly analyzed by 10% SDS-PAGE, or subjected to immunoprecipitation. For immunoprecipitations, lysates were incubated with Abs immobilized on protein A-Sepharose beads at 4°C for 1 h. The beads were washed three times with lysis buffer before boiling for 5 min in Laemmli buffer with (reducing) or without (nonreducing) 50 mM DTT.
Metabolic labeling
HeLa cells in 6-cm dishes were starved for 20 min in MEM lacking methionine and cysteine (ICN) but supplemented with 2 mM glutamax and 20 mM HEPES. Cells were biosynthetically labeled with 50 µCi [35S]methionine/cysteine (GE Healthcare/Amersham Biosciences) for 10 min. The labeling was either stopped by washing cells twice with ice-cold HBSS (Invitrogen Life Technologies) containing 20 mM NEM or followed by a chase. The cells were washed once with complete growth medium supplemented with 5 mM cysteine and methionine and 1 mM cycloheximide (chase medium) and then maintained in chase medium. The chase was stopped at the indicated times as for the pulse-only cells. After lysis, the lysates were subjected to immunoprecipitation followed by analysis by SDS-PAGE.
Endoglycosidase H (EndoH) treatment
Postnuclear lysates were treated with endoglycosidase H according to the manufacturer’s protocol (New England Biolabs). Briefly, SDS-denatured lysates were incubated in 50 mM sodium citrate (pH 5.0), with or without (mock) endoH at 37°C for 15 h before analysis by SDS-PAGE.
Western blotting
Samples were boiled in Laemmli buffer with (reducing) or without (nonreducing) 50 mM DTT before analysis by SDS-PAGE. For the HLA-DR stability assay, the relevant samples were not boiled before SDS-PAGE. Proteins were transferred to polyvinylidene difluoride membranes (Millipore) at 150 mA for 2 h followed by blocking the membranes in 8% milk in TBST (10 mM Tris-HCl (pH 8.0), 150 mM NaCl, 0.05% Tween 20) for 30 min. The membranes were incubated with primary Abs in TBST for 1 h at the following dilutions: 1/200 anti-HA, 1/2000 anti-myc, 1/200 HC10, 1/1000 anti-Ii, 1/1000 1B5, and 1/1000 anti-DR. After washing three times with TBST, the membranes were incubated with the appropriate secondary Abs (DakoCytomation) for 1 h, washed three times with TBST, and visualized by ECL (GE Healthcare/Amersham Biosciences) and film (Kodak). All experiments were repeated at least three times with similar results.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
DM and DM share only 20–25% identity (
42% similar residues) with DR/DO and DR1/DO, respectively, compared with over 50% identity between DR (and DP/DQ) and DO chains. Despite this divergence, the Ig domains of all class I and II molecules contain a disulfide bond that is shared by all Ig domain-containing proteins. The -chains of MHC class II molecules have an additional disulfide bond in the peptide-binding domain, similar to the disulfide bond in the 2 domain of MHC class I molecules. Both DM chains are unique in that they each have an additional disulfide bond not shared by their MHC class II counterparts, DM C24-C79 and DM C25-C35 (Fig. 1). MHC class II molecules also contain conserved N-linked glycosylation sites. The classical MHC class II and DO chains contain two glycosylation sites at N78/N79 and N118/N119. The DM chain also has two glycosylation sites, but in different positions, N15 and N165. The -chains contain one N-linked glycosylation site. For DR/DO, N19 lies in between the cysteines of the first disulfide bond. For DM, N92 is in the linker region between the peptide-binding groove and the Ig domain, similar to the position of the N78/79 glycan in DR/DO chains (Fig. 1).
|
Before studying the role of the DM glycans and unique cysteine residues on DM folding and assembly, we verified our experimental system. We chose the DM-negative HeLa cell line to enable us to study the folding of wild-type and mutant DM molecules in the absence of competing endogenous partner chains. We reconstituted HeLa cells with HA-tagged DM, myc-tagged DM, or both chains together. At 24 h posttransfection, the transfectants were lysed in the presence of the alkylating agent IAA to preserve disulfide bonds. Cell lysates were subjected to immunoprecipitation with anti-HA or anti-myc Abs, followed by reducing SDS-PAGE and Western blot analysis. Anti-HA efficiently immunoprecipitated DM from single and double transfectants (Fig. 2A, lanes 2 and 3), but did not recognize DM directly, as expected (Fig. 2A, lane 9). Vice versa, anti-myc immunoprecipitated the DM chain (Fig. 2A, lanes 7 and 8), but not DM (Fig. 2A, lane 4). Importantly, anti-HA and anti-myc coimmunoprecipitated DM and DM, respectively, from double transfectants only (Fig. 2A, lanes 5 and 10). Thus, tagged DM and DM efficiently assembled into heterodimers when transfected in HeLa cells. To confirm that tagged DM and DM can also assemble in an APC line, a similar experiment was done using MelJuSo cells, which have been used extensively to characterize MHC class II Ag-presentation pathways (35, 36, 37). DMHA and DMmyc readily assembled in MelJuSo cells (Fig. 2B, lanes 4 and 6), but were less efficiently glycosylated. Therefore, we continued our analysis using HeLa cells.
|
HA/DMmyc transfectants were treated with endoH. EndoH cleaves immature glycans from ER proteins, but not glycans that have been modified in the Golgi. In accordance with previous reports (13, 38), DM or DM chains from single transfectants remained completely endoH sensitive, indicating that they were retained in the ER (Fig. 2C, lanes 1 and 2, 3 and 4). In contrast, part of the DM and DM pool in double transfectants became endoH resistant (Fig. 2C, lanes 5 and 6, 9 and 10). Double transfectants treated with cycloheximide for 6 h had completely endoH-resistant DM chains (Fig. 2C, lanes 7 and 8, 11 and 12), demonstrating that the entire DM pool has matured. Note that DM remained partially sensitive to endoH treatment even in the double transfectant because one of its glycans remains in the high-mannose form (39, 40).
Together, these results show that DMHA and DMmyc chains in HeLa cells formed a complex and passed the ER quality control checkpoints. Immunofluorescence analysis indicated that post-ER tagged DM accumulated in lysosomes (see below, Fig. 5C). In addition, tagged DM is functional in an HLA-DR SDS-stability assay (see below, Fig. 5F). Because the HA and myc tags did not perturb DM assembly, trafficking, or function in HeLa cells, we used this system to study ER folding by both steady state and pulse-chase techniques.
|
Glycosylation and glycan positioning play an important role in glycoprotein folding and quality control mechanisms (41, 42). To assess the role of glycans on oxidation and heterodimer formation of DM and DM, we mutated the asparagines (N) of the glycosylation motifs into glutamines (Q). Expression of DM N15Q and DM N165Q in HeLa cells showed that single glycan mutants increased in electrophoretic mobility, consistent with the absence of one glycan (Fig. 3A, compare lanes 3 and 4 with lane 2). Mutation of both DM glycosylation sites (N15Q/N165Q) resulted in a further increase in mobility consistent with the loss of both glycans (Fig. 3A, lane 5). Similarly, mutation of the DM glycosylation site (N92Q) resulted in increased molecular mobility indicating the loss of the single DM glycan (Fig. 3A, lane 7 vs 8). Mutants lacking all glycan motifs had the same mobility as DM and DM chains from cells when glycosylation was inhibited with tunicamycin, indicating that the mutant chains are not glycosylated (data not shown).
|
or the various DM glycan mutants (Fig. 3B, lanes 1–5). Similarly, DMHA coimmunoprecipitated both wild-type and glycan mutant DM chains (Fig. 3B, lanes 6–10). Thus, these results show that glycosylation of DM or DM is not a prerequisite for DM interaction.
DM chains resolved as a doublet (Fig. 3B, lane 1), corresponding to two differently modified glycans of DM. The glycan modification occurred only on DM glycan N15, as the doublet was absent from DM lacking this glycan (Fig. 3B, lanes 2 and 4). Note that the presence of two bands in DMHA N15Q/DM was due to less efficient glycosylation of the second DM glycan in this mutant, resulting in a nonglycosylated species in addition to glycosylated DM N165 (Fig. 3B, lane 2, species 4 and 5). The different glycosylated DM species are explained by the diagram on the left of Fig. 3B, with the glycans represented by the vertical lines on a horizontal line (representing the DM chain) and the modified glycan indicated by the filled circle (Fig. 3B, see legend).
Although the DM glycans were not required for pairing, they may still be required for DM egress from the ER. Therefore, we investigated whether the DM glycan mutants acquired endoH resistance. Lysates from wild-type DM and glycan mutant transfectants were treated with endoH or mock treated and analyzed by reducing SDS-PAGE and Western blotting. As in Fig. 2B, wild-type DM was partially resistant to endoH treatment (Fig. 3C, lanes 3 and 4). The DM chain in the DMN15Q/DM transfectant was completely sensitive (Fig. 3C, lanes 5 and 6, top panel), while part of the DM pool in this transfectant became resistant to endoH treatment (Fig. 3C, lanes 5 and 6, bottom panel). Thus, despite progression through the Golgi, the DM N165 glycan was endoH sensitive, suggesting that this glycan always remains in the high-mannose form. In contrast, both DM and DM chains in the DMN165Q/DM transfectant became partly resistant to endoH treatment (Fig. 3C, lanes 7 and 8). However, only the top band of the DM doublet was endoH resistant, showing that the DM doublet is composed of a post-Golgi species (top) and a ER species (bottom). The partial endoH sensitivity of DM chains in wild-type DM (Fig. 3C, lanes 3 and 4) can now be explained by conversion of species 1 and 2 of the doublet into species 3 and 5, respectively (diagram in Fig. 3B). Note that the endoH treatment of wild-type DM and DMN165Q resulted in a similar pattern (Fig. 3C, compare lanes 4 and 8, top panel).
Mutation of both DM glycan sequons does not preclude ER exit, as part of the DM pool gained endoH resistance (Fig. 3C, lanes 9 and 10). Likewise, DM lacking its glycan does not preclude ER exit of the DM heterodimer, as the DM chain became partially endoH resistant (Fig. 3C, lanes 11 and 12). The amount of endoH resistant material, however, was reduced in all of the glycan mutants, compared with wild-type DM (Fig. 3C, bottom panel, compare lanes 6, 8, and 10 with lane 4), indicating that glycosylation was beneficial, but not essential for DM folding and ER exit. To see whether the endoH sensitivity is caused by slower ER egress of glycan mutant chains, we pulse-labeled wild-type DM or DMHA glycan mutant/DMmyc transfectants for 10 min. Cells were either immediately lysed, or chased for 2 h in normal medium. Lysates were subjected to endoH treatment before analysis by SDS-PAGE. Both DM and DM chains were completely endoH sensitive immediately after the pulse labeling (Fig. 3D, lanes 1 and 2, 5 and 6, 9 and 10, 13 and 14). After 2 h of chase, a proportion of DM and DM chains became resistant to endoH treatment for wild-type DM (Fig. 3D, lanes 3 and 4), but to a lesser extent for DMHA glycan mutant/DMmyc (Fig. 3D, lanes 7 and 8, 11 and 12, 15 and 16). This suggests that the lack of glycans does not prevent ER egress, but slows down the release of glycan mutant DM from the ER. Thus, glycosylation promotes, but is not an absolute requirement for, ER exit of DM.
The DM N15 glycan is required for wild-type oxidation of the DM chain
Because DM pairing and ER egress occurred in DM glycosylation mutants, we examined the role of glycosylation on the oxidation of DM. We analyzed the lysates from Fig. 3C by nonreducing SDS-PAGE. These conditions keep disulfide bonds intact and cause proteins to migrate faster in the gel. Wild-type DM in the presence of wild-type DM resolved as a discrete lower band and a more fuzzy upper band (Fig. 3E, lane 8, NR) that migrated faster than under reducing conditions (Fig. 3E, lane 8, R). Wild-type DM in combination with either one of the DM glycan mutants gave essentially the same DM oxidation pattern (Fig. 3E, lanes 9–11, NR). The DM N92 glycan mutant in combination with wild-type DM resulted in a decreased molecular mass of DM consistent with the lack of one glycan, but the oxidation pattern was similar to that of wild-type DM (Fig. 3E, lane 12, NR). Thus, none of the glycans on either DM or DM influence the final oxidation state of the DM chain.
In contrast, the DM chain in wild-type DM complexes resolved as three discrete bands (Fig. 3E, lane 2, NR) with the upper band corresponding to reduced DM. The DM N92Q glycan mutant had no effect on the wild-type DM oxidation pattern (Fig. 3E, compare lanes 2 and 6, NR). Likewise, oxidation of DM N165Q was not affected, taking into account the decrease in molecular mass because of the lack of a glycan (Fig. 3E, lane 4, NR). The oxidation pattern of the DM chain, however, was altered when glycan DM N15 was absent in either DM N15Q or in the glycan double mutant DM N15Q/N165Q (Fig. 3E, lanes 3 and 5). Note that the absence of glycan DM N15 had no effect on the oxidation pattern of the DM chain (Fig. 3E, lanes 9 and 11). Thus, of the three glycans in wild-type DM, the DM N15 glycan is the only sugar required for native oxidative folding of the DM chain.
DM prevents oxidative misfolding of DM
We next examined the oxidation of the DM chains. HeLa cells transfected with DMmyc alone or together with increasing amounts of DMHA were lysed in the presence of IAA. The lysates were analyzed by reducing and nonreducing SDS-PAGE followed by Western blotting. Analysis under reducing conditions showed similar levels of DM (Fig. 4A, lanes 1–4, myc R) with increasing levels of DM (Fig. 4A, lanes 2–4, HA R). When the lysate from the DM transfectant was run on nonreducing gels, hardly any monomeric DM could be detected. Instead DM migrated entirely as a 50 kDa complex (Fig. 4A, lane 1, myc NR). These complexes are likely disulfide-bonded DM dimers, as has been demonstrated in an EBV-transformed B cell line 2.2.93 lacking DM (38). Titrating in increasing amounts of DM prevented the formation of DM dimers, restoring the wild-type monomeric oxidation state (Fig. 4A, lanes 2–4, myc NR). In contrast, DM does not form such disulfide-bonded homodimers in the absence of DM (data not shown). Thus, DM prevents misfolding of its partner and is required for the proper oxidation of DM.
|
homodimers form cotranslationally or posttranslationally, we metabolically labeled HeLa cells transfected with DMmyc for 10 min followed by a chase for up to 60 min. Lysates were subjected to immunoprecipitation with anti-myc Abs followed by nonreducing and reducing SDS-PAGE. DM homodimers were present immediately after the pulse labeling (Fig. 4B, lane 1) and persisted for up to 60 min chase (Fig. 4B, lanes 2–4), demonstrating that they rapidly formed after biosynthesis. To see whether DM dimers could also form posttranslationally, living DMmyc or DMHA/DMmyc transfected HeLa cells were treated with 10 mM DTT for 10 min to reduce any disulfide linked dimers. After washing out DTT, DM was allowed to reoxidize for up to 30 min in the absence of DTT. Reducing SDS-PAGE showed that all DM chains were of the expected molecular mass (Fig. 4C, lanes 1–10, myc R). When the disulfide bonds were left intact by analyzing samples by nonreducing SDS-PAGE, DM from single transfectants formed homodimers, that were efficiently disrupted by DTT (Fig. 4C, lanes 1 and 2, myc NR). Upon replacing the DTT-containing medium with normal medium, the DM homodimer appeared within 5 min (Fig. 4C, lane 3, myc NR). Thus, the entire DM pool at steady state is able to form homodimers from reduced monomers. In the presence of DM, DM migrated faster in nonreducing gels than reduced single chain DM immediately after DTT treatment (Fig. 4C, lanes 6–10 vs lane 2, myc NR), indicating that DM became more oxidized in the double transfectant. In addition, DM is protected from complete reduction when DM is present (Fig. 4C, lanes 6–10). Thus, the formation of DM homodimers is rapid, reversible, and occurs both cotranslationally and posttranslationally in the absence of DM.
The DM C25-C35 disulfide bond is essential for ER egress of DM
It has been shown that mutations disrupting the Ig domain disulfide bond (26) or the MHC class II 1/class I 2 domain disulfide bond (27, 38) result in misfolding. Because DM extensively misoxidizes in the absence of DM, we mutated the unique DM cysteines, C25, C35, and C46 (Fig. 1) into alanine to determine whether these residues were involved in oxidative misfolding. HeLa cells transfected with the relevant constructs were analyzed as before by reducing and nonreducing SDS-PAGE. Like wild-type DM, each DM cysteine mutant still formed disulfide-linked homodimers (Fig. 5A, lanes 5–8). Thus, homodimer formation cannot be attributed to any of these single cysteines. However, unlike wild-type DM, some monomers could be detected in all of the DM cysteine mutants. This indicates that the lack of C25, C35, or C46 improved the efficiency of oxidation of the individual DM chains. These results suggest that the unique DM cysteines may be required for another function that comes at the expense of efficient oxidative folding.
We reasoned that these residues may be necessary for ER exit in the absence of stabilizing peptides or Ii that are required by classical MHC class I and II molecules, respectively. To see whether DM C25, C35, and C46 indeed had a stabilizing function, wild-type DM was cotransfected with wild-type DM or the C25A, C35A, and C46A DM mutants into HeLa cells. The lysates were subjected to mock or endoH treatment to monitor ER egress. Wild-type DM heterodimers became resistant to endoH treatment, consistent with exit from the ER (Fig. 5B, lanes 1–2). Similar to wild-type DM, DM with DM C46A became endoH resistant, indicating that heterodimers lacking the free cysteine folded properly and were transported through the Golgi (Fig. 5B, lanes 7–8). In contrast, mutation of DM C25 and C35 was detrimental for ER exit for DM (Fig. 5B, lanes 3 and 4, 5 and 6).
To demonstrate that gaining endoH resistance not only indicates ER egress, but also results in lysosomal deposition of DM, we used immunofluorescence to determine the intracellular localization of DM. HeLa cells transfected with DMmyc, DMHA and DMmyc, or DMHA and DMmyc C35A were methanol fixed and stained with an anti-myc Ab. As expected from Fig. 2C, single chain DMmyc gave a reticular staining pattern (Fig. 5C, top left panel), similar to that of the ER tracker (Fig. 5C, bottom right panel), indicating that it was retained in the ER. Immunofluorescence on wild-type DM transfectants showed a perinuclear punctate staining (Fig. 5C, top middle panel), comparable to the Lysotracker pattern (Fig. 5C, bottom left panel). Thus, wild-type DM is not only able to exit the ER, it is also routed to lysosomes. In contrast, DMHA/DMmyc C35A, which by biochemical analysis had the most severe retention phenotype (Fig. 5B), had a reticular, ER-like staining pattern (Fig. 5C, top right panel). The failure of this DM/DM C35A mutant to exit the ER is not due to an inability to heterodimerize, as both wild-type and the mutant DM chains could still be coimmunoprecipitated with DM (Fig. 5D, lanes 1–4). Thus, disrupting the highly conserved disulfide bond unique to DM increased ER retention of DM, but did not prevent DM-DM complex formation. We conclude that the unique DM cysteines make it prone to misoxidation and/or misfolding in the absence of DM, but in the proper environment (the presence of DM) help to stabilize the DM complex to allow ER exit.
DM C35A is required for normal DM oxidation
Disruption of the DM C35 residue did not prevent dimerization, but minor folding defects caused by a lack of the conserved disulfide bond might be sufficient to cause ER retention of the DM complex. Therefore, we compared the oxidation of the DM chain in HeLa cells expressing wild-type DM or DM with the DM C25A, C35A, and C46A cysteine mutants. Under nonreducing conditions, DMHA in wild-type DM transfectants resolved as three discrete bands (Fig. 5E, lane 5), as seen in Fig. 3E. The top band corresponds to the most reduced form (r), as it migrated at the position of DMHA analyzed by reducing SDS-PAGE (Fig. 5E, lanes 7–12). When cells transfected with wild-type DM were treated with cycloheximide for 6 h, only the middle band remained (termed ox1; Fig. 5E, lane 6), indicating that this was the final DM oxidation state. DMmyc C25A and DMmyc C46A both permitted oxidation of wild-type DM (Fig. 5E, lanes 2 and 4). The oxidation pattern of DMHA in the presence of the "free" DMmyc C46A cysteine mutant was expected, as it behaves as wild-type DM when comparing endoH patterns (Fig. 5B, lanes 1 and 2 vs 7 and 8). DMmyc C25A did facilitate the oxidation of DM, although this complex was still impaired in ER exit. The small amounts of DMHA/DMmyc C25A that are able to exit the ER (Fig. 5B, bottom panel, lanes 3 and 4) may account for the presence of the final oxidation state in this mutant DM. Alternatively, attaining the ox1 oxidation state of DM may not be sufficient to completely satisfy the requirements for ER exit. DMHA alone and DMHA in combination with DMmyc C35A migrated as the most reduced form (Fig. 5E, lane 1 and 3), indicating that the C35A mutant could not support oxidation of DM and ER exit. Although in some experiments DMHA (alone or in combination with DMmyc C35A) ran as a combination of r and ox2, it never reached the final oxidation state, ox1. Thus, although the conserved DM disulfide C25-C35 makes DM more prone to misoxidation, it is required for facilitating the oxidation of the partner chain, DM, and is essential for ER egress of the DM heterodimer.
The DM C35A mutation caused the most severe phenotype for oxidation and ER exit of DM molecules. To see whether this DM C35A mutation resulted in impaired Ag presentation, we tested the ability of DM to induce an SDS-stable (peptide-loaded) conformation in HLA-DR molecules. HeLa cells reconstituted with DR and the Ii were either mock transfected or transfected with wild-type DMHA/DMmyc or DMHA/DMmyc C35A. Lysates from the transfectants were boiled or left at room temperature in reducing sample buffer and analyzed by SDS-PAGE and Western blotting. As expected, part of the DR pool from the DM-positive MelJuSo cells migrated as dimers when analyzed under nonboiling conditions (Fig. 5F, lane 2). DR molecules in HeLa cells in the absence of DM are unstable and ran as monomers (Fig. 5F, lane 4). When the cells were transfected with wild-type DMHA/DMmyc, a significant proportion of SDS-stable DR dimers could be detected (Fig. 5F, lane 6), demonstrating that tagged DM is functional. In contrast, cells expressing DMHA/DMmyc C35A do not form SDS-stable DR dimers, indicating that this mutant DM is not able to support Ag presentation. Reconstitution of the Ag-presentation components in the HeLa transfectants was confirmed by Western blotting (Fig. 5F, right panel). Taken together, disruption of the DM C25-C35 disulfide bond by mutating DM C35 results in impaired oxidation, ER egress, and consequently leads to loss of function.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
and DM to understand how heterodimers of the Ig superfamily fold and assemble in the ER. By analyzing DM glycosylation mutants, we can now explain several previous unresolved observations. It has been demonstrated for Igs (43) and the -chains of DR and DM (39, 40) that not all N-linked glycans are processed in the Golgi. Here, we have identified glycan DM N165 as the glycan that remains in the high-mannose form (Fig. 3, B and C). It has been suggested for Igs that the lack of glycan processing in the Golgi is due to steric hindrance (43). However, the DM N165 residue is located on an exposed strand of the Ig domain (44), making it unlikely that it is inaccessible to glycan-modifying enzymes. The location of the Ig domain glycan in DR (and DO) is different from that of the DM glycan, yet one DR glycan also remains in the high-mannose form (39). In addition, part of the DM N15Q pool is not glycosylated (Fig. 3B, lane 2 and Fig. 3C, lane 5). Taken together, this suggests that inefficient glycosylation and subsequent inefficient glycan modification is an inherent characteristic of MHC -chains.
The glycosylated wild-type DM chain appears as a doublet, but only when expressed with wild-type DM (or DM C46A). This doublet has been observed previously (15, 40, 45) and we now demonstrate that the top band in the doublet arises from a modification of glycan DM N15. The modified glycan is resistant to endoH treatment suggesting that it is present on mature DM heterodimers. This is supported by the observations that 1) the top band of the doublet is enriched in DM immunoprecipitations (Figs. 2A and 3B), 2) the top band is absent in DM C25A/C35A mutants that fail to exit the ER (Fig. 5B), and 3) the bottom band disappears upon cycloheximide treatment (Fig. 2C).
Glycosylation is not an absolute requirement for heterodimer formation, ER egress (Fig. 3), or function (46). However, the amount of endoH-resistant DM chains decreased when glycan sequons were mutated, suggesting that glycosylation still promotes the efficiency of folding and ER egress. The absence of glycan DMN15 has the most clear folding defect of the DM chain, as it is required for full oxidation of the DM chain (Fig. 3E). Most glycosylated proteins interact with the lectin-binding proteins calnexin and calreticulin during the early stages of protein folding (47). The position, or removal, of glycans within a protein sequence has been shown to affect protein interaction with chaperones (41, 42, 48). Thus, the absence of the DMN15 glycan may change the chaperone set it interacts with and affect the oxidation of the DM chain. Also, not all glycans within a substrate are equally important for protein folding (49), which may explain the efficient folding, oxidation, and ER egress of the DMN165Q/DM and DM/DMN92Q mutants (Fig. 3E, lanes 4 and 6). The fact that the DMN92 glycosylation site is less well-conserved than the DM glycosylation sites implies that it has a minor role in protein folding and stability.
DM is a noncovalent heterodimer with only intrachain disulfide bonds. The DM chain, in the absence of DM, is prone to complete misoxidation into disulfide-linked homodimers (Fig. 4) (13, 38). We show that these DM homodimers form rapidly upon biosynthesis (Fig. 4B). Hardly any monomeric DM is detected immediately after pulse labeling, suggesting that homodimers form cotranslationally. However, upon reduction of DMmyc dimers in living cells, these rapidly reform from the entire ER-resident pool (Fig. 4C). Thus, disulfide-linked DM dimers efficiently form both co- and posttranslationally. The DM dimer formation can be prevented by introducing increasing amounts of DM, indicating that a quantitative association of DM with DM is required (Fig. 4A). These data imply that noncovalent DM heterodimerization has allowed the error-prone individual cysteine chemistry of the DM chain to be tolerated.
Conserved disulfide bonds are essential for proper folding and function of MHC class I (26, 27, 28). Busch et al. (38) have previously analyzed a DM mutant, C79Y, which results in a disrupted C11-C79 disulfide bond. This DM mutant forms disulfide-linked homodimers. Misfolding due to the disruption of disulfide bonds that are conserved in all MHC proteins may be expected, but the role of the three additional cysteines unique to DM was previously unknown. These cysteines are apparently not required for any of the other MHC molecules (Fig. 1). We show that DM without the free cysteine DM C46 has a wild-type phenotype (e.g., Fig. 5B, compare lanes 1 and 2 with lanes 7 and 8). It acquires the DM doublet, gains endoH resistance, and oxidizes like wild-type DM. This DM-free cysteine is not completely conserved across mammalian species i.e., not in rabbit, cattle, and dog. Together, this suggests that it is not required for oxidative folding of DM. In contrast, DM molecules with mutations that disrupt the DM C25-C35 disulfide bond are impaired in ER exit and affect oxidation of DM (Fig. 5). They also fail to acquire the glycan modifications resulting in appearance of the DM doublet, probably because of failure to exit the ER. Despite the ability to dimerize with DM, the DM C35A mutant affects oxidation of DM and abolishes DM activity in the DR stability assay (Fig. 5F). Thus, heterodimer formation on its own is not sufficient to impose the correct conformation of both chains upon the DM heterodimer. In addition, the Ii cannot compensate for the loss of the C25-C35 disulfide bond (data not shown), probably because other features are required for the Ii to escort classical class II molecules.
Our results show that oxidation (or misoxidation) is determined by the intrinsic properties of the DM and DM chains, rather than by the ER oxidation machinery. In light of the current interests regarding how different soluble (42), polytopic (50, 51), or multisubunit (48, 52) proteins fold in the ER, our work on DM demonstrates that two subunits of a heterodimeric complex can direct each others oxidative folding. We have shown that DM chains require DM chains to prevent misoxidation and conversely, DM chains need DM with the right disulfide bonds for proper oxidation of the DM complex. To our knowledge, this is the first example of such a cooperative, interdependent folding strategy. Whereas default ER retention of MHC class I and II molecules can be overcome by peptide binding, related proteins, like DM, have escaped this requirement by evolving additional disulfide bonds. It would be interesting to see whether other heterodimeric and Ig domain proteins use similar cooperative mechanisms to avoid misfolding in the ER.
|
Acknowledgments |
|---|
|
Disclosures |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
|
Footnotes |
|---|
1 This work was supported by a Marie Curie fellowship and the Biotechnology and Biological Sciences Research Council. 
2 Address correspondence and reprint requests to Dr. Adam M. Benham, Department of Biological Sciences, University of Durham, South Road, Durham, DH1 3LE, U.K. E-mail address: adam.benham{at}durham.ac.uk
3 Abbreviations used in this paper: ER, endoplasmic reticulum; Ii, invariant chain; HA, hemagglutinin; IAA, iodoacetic acid; endoH, endoglycosidase H; NEM, N-ethylmaleimide.
Received for publication April 20, 2006. Accepted for publication July 31, 2006.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionDisclosuresReferences |
|---|
dimers and facilitates peptide loading. Cell 82: 155-165. [Medline] chain directs HLA-DM to MHC class II compartments. J. Cell Biol. 131: 351-369. 2-microglobulin. Mol. Immunol. 32: 531-540. [Medline]2 domain disulfide bridge of the class I molecule HLA-A*0201: effect on maturation and peptide presentation. Hum. Immunol. 39: 261-271. [Medline] (ERO1-L), a human gene induced in the course of the unfolded protein response. J. Biol. Chem. 275: 23685-23692. . EMBO J. 19: 4493-4502. [Medline] disrupts antigen presentation without altering endosomal localization. J. Immunol. 160: 4289-4297.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
www.anthonynolan.com/HIG/data.html
.
