Glycosylation N MHC Proteins With

Distinct Patterns of Folding and Interactions with Calnexin and Calreticulin in Human Class I MHC Proteins with Altered N-Glycosylation

This information is current as Qing Zhang and Russell D. Salter of September 27, 2021. J Immunol 1998; 160:831-837; ; http://www.jimmunol.org/content/160/2/831 Downloaded from References This article cites 34 articles, 24 of which you can access for free at: http://www.jimmunol.org/content/160/2/831.full#ref-list-1

Why The JI? Submit online.

http://www.jimmunol.org/ • Rapid Reviews! 30 days* from submission to initial decision

• No Triage! Every submission reviewed by practicing scientists

• Fast Publication! 4 weeks from acceptance to publication

*average by guest on September 27, 2021 Subscription Information about subscribing to The Journal of Immunology is online at: http://jimmunol.org/subscription Permissions Submit copyright permission requests at: http://www.aai.org/About/Publications/JI/copyright.html Email Alerts Receive free email-alerts when new articles cite this article. Sign up at: http://jimmunol.org/alerts

The Journal of Immunology is published twice each month by The American Association of Immunologists, Inc., 1451 Rockville Pike, Suite 650, Rockville, MD 20852 Copyright © 1998 by The American Association of Immunologists All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. Distinct Patterns of Folding and Interactions with Calnexin and Calreticulin in Human Class I MHC Proteins with Altered N-Glycosylation1

Qing Zhang and Russell D. Salter2

Calnexin is a lectin-like chaperone that binds to class I MHC molecules soon after their synthesis, retaining unassembled heavy ␤ ␤ chains and also assisting their folding. Following association with 2-microglobulin ( 2m) in the endoplasmic reticulum, a large proportion of human class I molecules release from calnexin, whereas mouse class I molecules do not. We asked whether addition of a second N-glycan to the human class I molecule A*0201 at position 176, a site present in mouse, would affect its binding to calnexin. The 176dg mutant with N-glycans at positions 86 and 176, when transfected into CIR cells, demonstrated

␤ Downloaded from increased binding to calnexin, detectable both before and after association with 2m, and reduced interaction with calreticulin and TAP relative to wild-type protein bearing a single N-glycan at position 86. Cell surface levels of the mutant were decreased only slightly relative to the wild type, suggesting that the protein is not misfolded or grossly altered structurally. A subpopulation of mutant molecules was retained in the endoplasmic reticulum, and surprisingly, these molecules reacted with w6/32, which recognizes an epitope present on transport-competent class I HLA complexes. Transfection into Daudi cells demonstrated that ␤ 176dg reacts with w6/32 in the absence of 2m, suggesting that the Ab epitope can be induced by binding of calnexin. These data may explain previously noted differences between mouse and human class I MHC proteins and demonstrate that the http://www.jimmunol.org/ location of N-oligosaccharides within proteins can inﬂuence their folding and interactions with chaperones such as calnexin and calreticulin. The Journal of Immunology, 1998, 160: 831–837.

he molecular chaperone calnexin associates in the endo- ger binding of human class I heavy chains to calnexin before as- plasmic reticulum (ER)3 with many different glycopro- sembly (14, 15). This contrasts with mouse heavy chains, which T teins, including class I MHC molecules in mice and hu- remain strongly bound to calnexin until they leave the ER (1). mans (1–6). In both species, class I heavy chains associate Nossner and Parham (16) determined that the structural basis for cotranslationally via an oligosaccharide-dependent linkage with this difference resides in species-specific characteristics of the by guest on September 27, 2021 calnexin (7–9). Mouse class I molecules remain associated with heavy chain itself, as mouse class I molecules expressed in human ␤ ␤ calnexin during subsequent assembly with 2m and interaction cells remain bound to calnexin after associating with 2m. with TAP, and they dissociate from calnexin only following pep- Calnexin binding to many proteins is dependent on the presence tide loading (10). Human class I heavy chains dissociate from cal- of monoglucosylated N-oligosaccharides generated in the ER by ␤ nexin either immediately before or during assembly with 2m, trimming of glucose residues from the core glycan by glucosidases before association with a second chaperone, calreticulin, which I and II (17–20). Ware et al. (7) and more recent work by Zapun may mediate association with tapasin and TAP as required for et al. (21) demonstrated by direct binding studies that calnexin can efficient peptide loading (11, 12). Although some studies have de- associate with N-oligosaccharides, confirming that calnexin con- ␤ tected complexes of calnexin and human class I heavy chain- 2m tains a lectin-like binding site. We had previously shown that mu- dimers using extremely sensitive assays (13), there is much stron- tagenesis of the carbohydrate attachment site of A*0201 abolished calnexin binding, suggesting that class I heavy chain binding to calnexin is carbohydrate mediated (8). Department of Pathology and University of Pittsburgh Cancer Institute, Univer- In the present study, we asked whether the structural basis for sity of Pittsburgh School of Medicine, Pittsburgh, PA 15213 the observed difference in calnexin binding between mouse and Received for publication June 10, 1997. Accepted for publication September 26, 1997. human class I molecules is due to differing number and location of N-oligosaccharides. Human class I heavy chains contain a single 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 glycosylation site at position 86, whereas mouse heavy chains con- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tain either two or three sites, at positions 86 and 176, and in the 1 This work was supported by Grant IM-668B from the American Cancer Society latter subset, also at position 256 (reviewed in Ref. 22). We hy- and R01-AI39505 from National Institutes of Health. Q.Z. was supported by a student fellowship from the Pathology Education and Research Foundation at the pothesized that additional N-glycans may allow for simultaneous University of Pittsburgh. binding of multiple calnexin molecules or alternatively could 2 Address correspondence and reprint requests to Dr. Russell D. Salter, W957 strengthen binding of a single calnexin molecule by providing ad- Biomedical Science Tower, 200 Lothrop Street, University of Pittsburgh School ditional attachment sites for lectin binding. To test these possibil- ϩ of Medicine, Pittsburgh, PA 15213. E-mail address: rds @pitt.edu ities, we introduced a N-glycan acceptor site at position 176 of 3 ␤ ␤ Abbreviations used in this paper: ER, endoplasmic reticulum; 2m, 2-micro- HLA-A2 by substitution of Asn for Arg. In addition, a mutant globulin; TAP, transporter of antigenic peptides; CHAPS, (3-[(3-cholamidopro- pyl)-dimethylammonio]-1-propanesulfonate), 0.1 mM sodium-p-tosyl-L-lysine molecule bearing a single glycosylation site at position 176 was chloromethyl ketone; TBS, Tris-buffered saline (0.01 M Tris, 0.15 M NaCl, pH generated. The resulting class I heavy chain mutants were then 7.4); Endo H, endoglycosidase H; 176dg, mutant with a second N-glycosylation ␤ site at position 176; 176g, mutant with a single N-glycosylation site at position characterized for intracellular stability, assembly with 2m, and 176. association with calnexin and calreticulin.

Materials and Methods Cell lines and reagents The B cell lines CIR (HLA-A negative, HLA-B*3503 low, HLA-C*0401 ␤ normal) (23) and Daudi ( 2m negative) (24) were grown in RPMI 1640 (Irvine Scientific, Irvine, CA) containing 10% transferrin-supplemented bovine calf serum (HyClone, Logan, UT). Castanospermine was purchased from Genzyme (Cambridge, MA). Protein A-Sepharose beads, formalin- fixed Staphylococcus aureus Cowan I strain (10% suspension), and fluo- FIGURE 1. Acceptor sites for N-glycosylation at positions 86 and rescein isothiocyanate-conjugated goat anti-mouse Ab were purchased 176 in the 176dg mutant are both used. CIR cells transfected with from Sigma Chemical Co. (St. Louis, MO). Trans35S-label ([35S]methi- either wild-type A*0201 (A2) or a mutant with glycan acceptor sites at onine and [35S]cysteine) was from ICN (Costa Mesa, CA). Methionine-free RPMI 1640 medium was obtained from Life Technologies (Gaithersburg, positions 86 and 176 (176dg) were radiolabeled for 20 min. Cells were MD). Endoglycosidase H (Endo H) was purchased from New England lysed and class I heavy chains isolated with Ab UCSF#2. Samples were Biolabs (Beverly, MA). divided and digested with indicated amounts of Endo H. Heavy chains bearing 0 (lanes 2, 4 and 5),1(lanes 1 and 4)and2(lane 3) N-glycans Antibodies are marked by arrowheads. Mock-treated 176dg heavy chains migrate more slowly than wild-type A2 molecules (lanes 1 and 3), and diges- mAb W6/32 (American Type Culture Collection, Rockville, MD) recognizes an epitope on all HLA-A, -B, and -C heavy chains dependent on the tion of 176dg with an intermediate amount of Endo H (lane 4) gener- ␤ ated a mixture of heavy chains with 0 or 1 N-oligosaccharides. presence of 2m, which is not present on free class I heavy chains (25). mAb BB7.2, which binds to HLA-A2 and -A69 complexes (26), and anti- invariant chain Ab PIN1.1 (27) were obtained from Dr. Peter Cresswell Downloaded from (Yale University School of Medicine). Monoclonal Ab AF8 against calnexin (28) was obtained from Dr. Michael Brenner (Harvard Medical 16 h at 37°C. Samples were then precipitated in 70% ethanol and resus- School). Antiserum UCSF#2 reacts with the cytoplasmic tail of class I pended in reducing gel buffer and separated by SDS-PAGE. HLA heavy chains and was provided by Drs. Bruce Koppelman and ␤ Frances Brodsky (both at University of California, San Francisco). Anti- In vitro assembly of class I heavy chains with 2m sera against calreticulin was purchased from Affinity Bioreagents (Golden, Radiolabeled cells were lysed in CHAPS-TBS containing 10 ␮g/ml human CO). Antisera reactive with human ␤ m and human IgG were purchased http://www.jimmunol.org/ 2 ␤ m (Sigma Chemical Co.). Formalin-fixed S. aureus were added to post- from Sigma Chemical Co. 2 nuclear lysates, and samples were incubated at 4°C for 16 h. Abs were then Site-directed mutagenesis and transfections used to isolate specific proteins as described above. A cDNA clone encoding HLA-A*0201 was subcloned into the HindIII and Results SalI sites of M13 mp18 (29). A mutant with substitution of Ser to Ala at Glycosylation and surface expression of class I heavy chains position 88 (S88A) was constructed previously (8). Mutant A*0201 with a second N-glycosylation site at position 176 (176dg) was generated by sub- with N-glycans at positions 86 and 176 stituting Arg with Asn at position 176, using the site-directed mutagenesis Following mutagenesis, mutant and wild-type class I cDNA were method of Kunkel (30). A mutant with a single N-glycosylation site at transfected into CIR cells and surface expression characterized position 176 (176g) was generated by a second round of mutagenesis using by guest on September 27, 2021 the S88A mutant. Sequences of the mutants were confirmed by dideoxy with Abs specific for HLA-A2 by flow cytometry. The mutant sequencing. SalI- and HindIII-purified inserts were then subcloned into the bearing two glycans (176dg) was consistently expressed at be- vector pREP10 (Invitrogen, San Diego, CA). Stable transfectants were gen- tween 70 and 80% of wild-type levels in multiple experiments ␮ erated by electroporation followed by selection of cultures in 300 g/ml (nϾ5) using independently transfected cell lines (data not shown). hygromycin (Sigma Chemical Co.). To examine the ability of the introduced site at position 176 to Metabolic labeling, immunoprecipitation, and gel accept N-glycan, cells were radiolabeled with [35S]methionine, electrophoresis class I proteins isolated with class I-specific Ab UCSF#2 and sub- jected to Endo H digestion, followed by SDS-PAGE. Figure 1 Cells were washed in deficient medium (methionine-free RPMI 1640) and then incubated at 37°C for1hinthepresence or absence of castanosper- shows that glycosylated 176dg migrates more slowly than the mine. Cells were labeled with 150 ␮Ci of Trans35S-label for the times wild-type protein, consistent with the presence of two glycans. indicated. For pulse chase experiments, 10 vol of nonradioactive RPMI Following digestion with 10 mU Endo H, wild-type and mutant 1640 containing 10% bovine serum were added at the end of labeling, and proteins migrate at an identical position, as expected following aliquots were removed at times indicated and stored on ice. After one wash in Dulbecco’s PBS (0.2 g/ml KCl, 0.2 g/ml KH PO , 0.047 g/ml anhydrous removal of glycans. Digestion of the mutant with a lower concen- 2 4 tration of Endo H (6 mU) generated a band that comigrates with MgCl2, 8 g/ml NaCl2, 1.15 g/ml Na2HPO4, pH 7.4), cells were lysed at 4°C in Tris-buffered saline (TBS) containing 2% (w/v) (3-[(3-cholamidopro- undigested wild-type heavy chains and presumably corresponds to pyl)dimethylammonio]-1-propanesulfonate), 0.1 mM sodium-p-tosyl-L-ly- the mutant heavy chain with one or both of the glycans removed. sine chloromethyl ketone (CHAPS), and 1 mM PMSF for 20 min on ice. ϫ Lysates were centrifuged at 13,000 g for 5 min, and postnuclear lysate A*0201 class I heavy chains with N-glycans at positions 86 was precleared at 4°C with 1 ␮l of rabbit ␣-human IgG and 30 ␮lof10% ␤ suspension of formalin-fixed Staph A for 90 min. Aliquots were incubated and 176 bind simultaneously to 2m and calnexin with 1 ␮l of antiserum or 50 ␮l of hybridoma supernatant for 60 min at To test whether the presence of a second glycan strengthens bind- ␮ 4°C. Ag-Ab complexes were isolated with 20 l of 50% suspension of ing of class I heavy chains to calnexin, calnexin-associated pro- protein A-Sepharose beads. After 4 washes in 0.5% CHAPS-TBS, beads were boiled in reducing gel buffer for 10 min at 95°C to elute proteins. In teins were isolated with anti-calnexin Ab (AF8) from radiolabeled some experiments (noted in text), proteins were eluted from beads with 2% cells expressing either mutant or wild-type A*0201 proteins. Cells Triton-TBS for2htodisrupt interactions between calnexin and associated were lysed in CHAPS detergent for this experiment to maintain the proteins. Released proteins were then isolated using specific Abs as before. association between calnexin and ligands, as previously described. Samples were separated on 12% slab gels and fluorography and scanning densitometry performed as described. Following immunoprecipitation with AF8, calnexin-associated material was eluted from protein A beads with 2% Triton X-100 Endoglycosidase H treatment and individual proteins then reisolated with specific Abs reactive Proteins were eluted from beads by boiling in 1% SDS for 10 min. Samples with class I heavy chains (UCSF#2), invariant chain (PIN1), and ␤ ␤ were diluted in PBS that had been adjusted to pH 5.75 with citric acid, 2m (anti- 2m). Figure 2 shows that wild-type and mutant heavy divided into aliquots, and either mock treated or treated with Endo H for chains can be isolated from lysates with Abs reactive with either The Journal of Immunology 833

Table I. Quantitation by scanning densitometry of binding of A*0201 molecules and glycosylation mutants to calnexin in CIR cells

␤ Expt. Cell Line Anti- 2m Anti-Class I H Chain 1 176g 2a 33 2 A*0201 4 31 2 176dg 19 34 3 A*0201 ND 8 3 176dg 4 25 a CIR transfectants were radiolabeled as in Figure 2 and lysed with CHAPS, and Ab AF8 was used to isolate calnexin and associated molecules. Following elution with 2% Triton X-100, individual proteins were reisolated with Abs re- ␤ ␤ active with 2m (anti- 2m), class I heavy chains (UCSF#2), or invariant chain (PIN1.1), as in Figure 2. Values were derived from scanning densitometry and are percentages of the band intensity obtained for invariant chain reisolated from solubilized AF8 complexes from individual cell lines as an internal control. Re- sults from three separate experiments are shown. ND, not detectable.

sition as wild-type class I on SDS-PAGE (data not shown). Figure Downloaded from 2B and Table I show that following AF8 immunoprecipitation, mutant heavy chains could be readily isolated with UCSF#2, but ␤ only in small amounts with anti- 2m. Thus, the mutant with a single glycan at position 176 appears to bind calnexin rather weakly, similar to wild-type A*0201, and in contrast to the doubly

␤ http://www.jimmunol.org/ glycosylated mutant does not include 2m in the heavy chain- calnexin complex.

Binding of calnexin to 176dg depends on trimming of glucose residues from N-glycans ␤ FIGURE 2. A, Mutant 176dg forms a complex with 2m and calnexin. CIR transfectants expressing A2 or 176dg were radiolabeled for It seemed possible that the presence of two glycans attached to ␤ ␣ ␤ 176dg could cause it to misfold, perhaps resulting in strong caln- 20 min and lysed as in Figure 1. Abs reactive with 2m( - 2m), class I heavy chains (UCSF#2), and calnexin (AF8) were then used to isolate exin binding through determinants exposed in protein portions of proteins from lysates (lanes marked total lysate or AF8 total). Mole- the heavy chain. There is a relatively small group of nonglycosy- by guest on September 27, 2021 cules bound to calnexin were eluted with 2% Triton X-100 and reisolated proteins that associate with calnexin, and it has been shown ␣ ␤ lated with - 2m, UCSF#2, and anti-invariant chain-specific Ab that calnexin binding to some class I MHC proteins does not re- PIN1.1 (lanes marked AF8 elute). A longer exposure of the fluorogram quire glycan (31). It was thus important to determine whether cal- is shown for the AF8 and reisolated samples. B, Calnexin binds weakly nexin binding to 176dg is dependent on glycan, as is the wild-type to a mutant A*0201 heavy chain (HC) with a single N-oligosaccharide A*0201 molecule. This was tested using castanospermine, an in- at position 176. CIR cells transfected with a mutant bearing a single hibitor of glucosidases I and II, which blocks trimming of the two N-glycan acceptor site at position 176 were analyzed as in A. IC, invariant chain. outermost glucose residues from the core N-glycan and prevents generation of the monoglucosylated carbohydrate ligand recognized by calnexin. Figure 3A demonstrates that binding of 176dg to calnexin is ␤ Ͼ ␮ UCSF#2 or anti- 2m. Following elution from AF8-protein A blocked by concentrations of castanospermine 100 g/ml, con- beads, both wild-type and mutant class I heavy chains were iso- firming the involvement of N-glycan. At a concentration of 50 lated with UCSF#2, whereas only the mutant could be isolated ␮g/ml, however, a calnexin-associated band of intermediate mo- ␤ with anti- 2m. The amount of mutant heavy chain bound by bility is seen which represents a population of heavy chains bear- UCSF#2 was greater than for wild-type in most experiments, when ing one trimmed and one untrimmed glycan. Relatively strong binding of the invariant chain molecule to calnexin was used as an binding of calnexin to invariant chain following treatment is also internal standard for scanning densitometry (Table I). This sug- seen, consistent with other reports (15, 32, 33). In Figure 3B, cal- gests that the presence of the second glycan strengthens binding to nexin binding to the mutant bearing a single glycan at position 176 calnexin. This was not due to differences in radiolabeling effi- is shown also to be dependent on glucose trimming, as previously ciency between cell lines (data not shown). In addition, the amount demonstrated for wild-type class I HLA molecules. Together, of invariant chain associated with calnexin was similar amounts in these results suggest that calnexin binding can be directly mediated these samples (Fig. 2A). by glycans attached to either position 86 or position 176 and are inconsistent with misfolding of either the 176dg or the 176g pro- Residues 86 and 176 must both be glycosylated to teins promoting calnexin association. strengthen binding of calnexin relative to wild-type class I heavy chains Assembly and transport of 176dg are heterogeneous with We considered whether the presence of a glycan at position 176 evidence for a nontransported subpopulation of molecules could by itself increase calnexin binding, or whether both glycans Pulse-chase analyses were conducted to determine the efficiency of ␤ needed to be present. To distinguish between these possibilities, a 176dg assembly with 2m and subsequent transport through the mutant was generated with a single glycan acceptor site at position Golgi (Fig. 4). Following a short period of radiolabeling, cells 176. This mutant was glycosylated and migrated at the same po- were incubated for the indicated times in nonradioactive medium 834 EFFECT OF N-GLYCAN ON CLASS I MHC FOLDING AND CHAPERONE BINDING Downloaded from

FIGURE 3. Binding of calnexin to A*0201 mutants requires trim- FIGURE 4. Heterogeneity in intracellular transport of 176dg in CIR ming of glucose residues. A, CIR cells transfected with 176dg were cells. A, CIR cells transfected with 176dg were radiolabeled for 3 min treated with castanospermine (CST) for3hattheindicated concen- and then incubated in nonradioactive medium for the indicated times. http://www.jimmunol.org/ tration before radiolabeling for 10 min. Cells were lysed and class I Abs reactive with class I HLA complexes (w6/32), calnexin (CNX) heavy chains and calnexin were isolated with Abs UCSF#2 and AF8, (AF8), and class I HLA heavy chains (HC) (UCSF#2) were used to iso- respectively. A slight decrease in migration evident for heavy chains late specific proteins for analysis on SDS-PAGE. IC, invariant chains. B, (HC) in lanes 2–5 and 7 results from altered N-glycan structure. Class Samples were digested with Endo H before analysis on SDS-PAGE. II MHC-associated invariant chain (IC) associates strongly with cal- Endo H-sensitive (superscript s) and Endo H-resistant (superscript r) nexin and is also evident in lanes 6 to 10. B, CIR cells expressing the bands are indicated. Addition of sialic acid residues to the glycan de- A2 mutant with a single N-glycan at position 176 (176g) were treated creases migration and results in the heterogeneous population of mol- as in A. Calnexin (CNX) is also visible. ecules seen in lanes 7 to 10 and 17 to 20, which migrate above the undigested sample shown in lane 11 of B. by guest on September 27, 2021 before lysis and with use of specific Abs to isolate total class I ␤ heavy chains (UCSF#2) or 2m-associated class I heavy chains molecules described above is that the presence of the second gly- (w6/32). In Figure 4A, it is apparent that a proportion of the 176dg can on 176dg imposes a conformation that allows binding of Ab ␤ has decreased mobility and is heterogeneous, characteristic of ad- w6/32 independently of 2m interaction. This might be a direct dition of sialic acid residues to N-glycans. Molecules within this consequence of mutation at position 176, resulting in structural subpopulation are evident in increasing amounts 30 min after syn- alteration of the protein, or may be more indirectly due to strength- thesis (lane 3, Fig. 4A), and are Endo H resistant (lanes 7–10 and ened interaction of 176dg with calnexin. To test whether w6/32 ␤ 15-20, Fig. 4B), indicating that they have passed through the binding to the mutant occurs in the absence of 2m, the 176dg Golgi. UCSF#2 and w6/32, but not AF8, react with these mole- construct was transfected into Daudi cells. Following radiolabel- ␤ cules, suggesting that they are associated with 2m, and not cal- ing, cells were lysed using CHAPS and incubated for 16 h with or ␤ nexin. A second subpopulation, which remained sensitive to Endo without exogenously added 2m; then class I proteins were iso- H for at least 2 h (Fig. 4B), was also observed and represented a lated with w6/32 or UCSF#2, and SDS-PAGE analysis was per- ␤ major proportion of the total heavy chain even at later time points formed (Fig. 5). In samples to which no 2m has been added, there (Fig. 4A). Failure of these molecules to become resistant to Endo is significant reactivity of w6/32 with 176dg, which contrasts with H suggests that they are not transported through the Golgi, which the endogenous class I heavy chains found in Daudi. Addition of ␤ ␤ is typically due to a failure in assembly with 2m. It was unex- 2m during the incubation period results in increased binding of pected that this latter subpopulation reacted with w6/32 and thus w6/32 to the endogenous Daudi heavy chains, but there is no ad- ␤ appeared to be associated with 2m (Fig. 4B). In contrast, wild- ditional increase in binding observed for 176dg. Pretreatment of ␤ type A*0201 molecules in CIR cells pass through the Golgi within cells with castanospermine prevented the 2m-induced increase in ␤ 15 to 30 min after assembly with 2m and do not remain sensitive w6/32 binding, suggesting that calnexin is required to maintain a ␤ to Endo H for long periods. Together, these experiments demon- conformation receptive to 2m (data not shown). These results strate that 176dg molecules are heterogeneous in phenotype, with demonstrate that the epitope recognized by w6/32 is not absolutely ␤ a proportion assembling normally and passing rapidly through the dependent on 2m, but can be achieved by 176dg following asso- Golgi, while the rest remain in the ER in a transport-incompetent ciation with calnexin. state. Reactivity of 176dg from Daudi cells with w6/32 suggests that ␤ it obtains a more fully folded state without 2m than do other class Reactivity of w6/32 with 176dg heavy chains in the absence I molecules. It seemed possible, therefore, that this mutant might ␤ of 2m be able to bind peptides and/or be transported to the cell surface, ␤ One potential explanation for the apparent uncoupling of 2m as- in contrast to other class I HLA proteins, which are retained in the sembly and transport competence in a subpopulation of class I ER in Daudi cells. However, no Endo H-resistant 176dg protein The Journal of Immunology 835

FIGURE 6. Mutant 176dg binds to calreticulin in CIR cells. CIR cells transfected with wild-type A2 or 176dg were radiolabeled for 20 min ␤ FIGURE 5. Binding of w6/32 to 176dg in the absence of 2m. Daudi before lysis. Class I molecules and calreticulin were isolated with cells transfected with 176dg were radiolabeled for 20 min and then UCSF#2 (U) and anti-calreticulin (CR), respectively. Bands corre- ␮ ␤ lysed in the presence or absence of 10 g/ml exogenous human 2m. sponding to 176dg and A2 heavy chains are indicated by arrowheads. After 16 h, class I proteins were isolated with UCSF#2 and W6/32 and A third band visible particularly in the 176dg sample may represent then analyzed by SDS-PAGE. Migration positions of 176dg heavy tapasin and is indicated by an unlabeled arrowhead. chains (176) and endogenous Daudi class I heavy chains (HC) are marked. Downloaded from Table II. Quantitation by scanning densitometry of binding of A*0201 molecules and glycosylation mutants expressed in CIR cells to calreticulin and ␤ m was seen in lysates of 176dg-Daudi transfectants, nor was expres- 2 sion at the cell surface detectable using speciﬁc Abs (data not Ab A*0201 176dg 176g shown). In addition, 176dg did not bind calreticulin in the Daudi ␤ a transfectants (data not shown). These results reveal a strict require- Anti- 2m 118 47 36 http://www.jimmunol.org/ ␤ Anti-calreticulin 45 5 11 ment for 2m in allowing class I HLA heavy chains to bind cal- a ␤ ␤ reticulin, followed by TAP-mediated transfer of peptides and The percentage of class I heavy chains associated with 2m (anti- 2m) or transport of mature class I molecules to the cell surface. with calreticulin (anti-calreticulin) are shown relative to total class I heavy chains isolated with UCSF#2 from each cell line, as determined by scanning densitometry. CIR cell transfectants were radiolabeled and processed as described in Fig- Calreticulin binding to 176dg is reduced relative to wild- ure 5, except that radiolabeling was for 30 min. Incorporation of radiolabel into type A*0201 protein class I heavy chains were comparable for each of the three cell lines. ␤ After assembly with 2m, class I HLA heavy chains associate with calreticulin, a chaperone with structural and functional similarities

nexin attached at positions 176 or 256, which are relatively distant by guest on September 27, 2021 to calnexin (11, 12, 34). Like calnexin, binding of calreticulin to from position 86. After binding of ␤ m, class I heavy chains bear- many ligands requires glucose trimming to generate the appropri- 2 ing one or more N-oligosaccharides would associate with calreti- ate carbohydrate structure. We tested whether the presence of two culin, possibly via the position 86 glycan. This model would pre- N-glycans on 176dg affected binding to calreticulin. Figure 6 dict that the binding stoichiometry of calnexin to class I heavy shows that 176dg associates with calreticulin, but to a lesser extent chains bearing more than one glycan could change following in- than the wild-type A*0201 molecule. Quantitation of data from a teraction with ␤ m and that calreticulin and calnexin might jointly separate experiment is shown in Table II, demonstrating a marked 2 associate with such a class I molecule. We are currently testing reduction in the percentage of ␤ m-associated 176dg heavy chains 2 these possibilities. bound to calreticulin. In contrast to 176dg, A*0201 molecules with In addition to an altered pattern of calnexin binding, several a single N-glycan at position 86 (wild-type) or 176 (176g mutant) other changes in the biosynthesis of the 176dg mutant were seen. bound more strongly to calreticulin and, as shown earlier, inter- Cell surface expression was impaired by 20 to 30%, and about acted weakly with calnexin. Furthermore, interaction of 176dg one-half of the 176dg synthesized was retained in the ER. Binding with TAP was reduced, consistent with previous suggestions that to calreticulin and TAP was impaired, suggesting that stronger calreticulin is required to mediate binding of TAP and tapasin binding of 176dg to calnexin may either lessen the requirement for (data not shown). calreticulin or directly compete for its binding. This latter observation could potentially explain the observed inefﬁciency in export Discussion of 176dg from the ER, since calreticulin may be essential for me- A number of previous reports have suggested that mouse and hu- diating TAP association and peptide loading. Alternatively, the man class I molecules differ in their biosynthesis, particularly in 176dg subpopulation that is retained in the ER may misfold and be ␤ the strength of class I heavy chain binding to calnexin following unable to bind 2m, thereby strengthening and prolonging binding ␤ association with 2m. Our study suggests that this is due to the of calnexin while not directly interfering with calreticulin binding. number and position of N-oligosaccharides, with the additional It is not possible currently to distinguish between these possibili- glycan present at position 176 on mouse heavy chains increasing ties, although the presence of nearly normal amounts of 176dg on stability of class I-calnexin complexes. A model to explain these the surface of transfectants argues against a gross folding problem. observations is suggested as follows (Fig. 7). Calnexin associates Structural requirements for recognition of ligands by calnexin with human class I heavy chains via the single N-glycan attached and calreticulin are presently unclear, with widely disparate results ␤ at position 86 and is then displaced when 2m binds to a sterically reported in the literature. Several groups have reported that human overlapping site. In mouse class I molecules, the presence of two class I heavy chains dissociate from calnexin either before or dur- ␤ or three N-oligosaccharides allows binding of multiple calnexin ing binding to 2m (8, 14–16), while others have detected asso- ␤ ␤ molecules, possibly one to each glycan. 2m displaces only cal- ciation of calnexin with human class I heavy chain- 2m dimers (9, nexin bound to the glycan at position 86, while leaving intact cal- 13). These differences may be due to the type of assay or Abs used 836 EFFECT OF N-GLYCAN ON CLASS I MHC FOLDING AND CHAPERONE BINDING

ence of two N-oligosaccharides has a signiﬁcant effect on the conformation of the class I heavy chain. We know of no report of ␤ w6/32 binding to class I heavy chains without 2m, which was previously interpreted as evidence that the epitope is formed only after assembly of class I dimers (25). Our data suggest that stronger binding of calnexin to class I heavy chains bearing a second glycan results in a conformation recognized by w6/32, which is ␤ usually obtained only after association with 2m. Further matura- tion of these class I molecules is not seen, however, probably due ␤ to an absolute requirement for 2m in providing longer term stability in folding or in facilitating binding to TAP. In support of the ␤ latter possibility, 2m has been shown to bind directly to TAP and perhaps to directly mediate association of unassembled class I heavy chains with TAP (13). Acknowledgments We thank M. Brenner for providing AF8 Ab, P. Cresswell for PIN1.1, and FIGURE 7. A hypothetical model of calnexin binding to class I MHC B. Koppelman and F. Brodsky for UCSF#2. proteins. Calnexin attached via position 86 glycan is displaced by Downloaded from ␤ binding of 2m, while a second calnexin molecule, attached via po- References sition 176 glycan, is unaffected. The ␣1, ␣2, ␣3 domains of a class I 1. Degen, E., and D. B. Williams. 1991. Participation of a novel 88 kd protein in the heavy chain with two N-glycans are shown. The binding surface for biogenesis of murine class I histocompatibility molecules. J. Cell Biol. 112:1099. ␤ 2m is below 86 and relatively distant from position 176. 2. David, V., F. Hochstenbach, S. Rajagapolan, and M. B. Brenner. 1993. Interac- tion with newly synthesized and retained proteins in the endoplasmic reticulum suggests a chaperone function for human integral membrane protein IP90 (calnexin). J. Biol. Chem. 268:9585. http://www.jimmunol.org/ 3. Degen, E., M. F. Cohen-Doyle, and D. B. Williams. 1992. Efﬁcient dissociation to detect this association, as previously suggested. The use of mAb of the p88 chaperone from major histocompatibility complex class I molecules ␤ AF8 to isolate calnexin-class I complexes has been criticized as requires both 2-microglobulin and peptide. J. Exp. Med. 175:1653. potentially disrupting a pre-existing complex of calnexin, class I 4. Rajagopalan, S., and M. B. Brenner. 1994. Calnexin retains unassembled major ␤ ␤ histocompatibility complex class I free heavy chains in the endoplasmic reticu- HLA heavy chain, and 2m, causing the release of 2m from the lum. J. Exp. Med. 180:407. ␤ complex. This would presumably occur only with human class I 5. Sugita, M., and M. B. Brenner. 1994. An unstable 2-microglobulin: major histocompatibility complex class I heavy chain intermediate dissociates from caln- molecules, since we and others have shown that complexes includ- exin and then is stabilized by binding peptide. J. Exp. Med. 180:2163. ␤ ing human calnexin, human 2m, and mouse class I heavy chains 6. Jackson, M. R., M. F. Cohen-Doyle, P. Peterson, and D. B. Williams. 1994. Regulation of MHC class I transport by the molecular chaperone, calnexin (p88,

can be readily detected using this assay (16) (our unpublished by guest on September 27, 2021 IP90). Science 263:384. data). The present results argue against this possibility, since com- 7. Ware, F. E., A. Vassilakos, P. A. Peterson, M. R. Jackson, M. A. Lehrman, and plexes containing human 176dg heavy chains in association with D. B. Williams. 1995. The molecular chaperone calnexin binds glc1 man9glcnac2 ␤ m and calnexin were detected using Ab AF8. oligosaccharide as an initial step in recognizing unfolded glycoproteins. J. Biol. 2 Chem. 270:4697. In addition, we have demonstrated a requirement for N-glycans 8. Zhang, Q., M. Tector, and R. D. Salter. 1995. Calnexin recognizes carbohydrate present on class I heavy chains to mediate calnexin association (8), and protein determinants of class I major histocompatibility complex molecules. J. Biol. Chem. 270:3944. while others have seen no requirement for glycosylation (32). It is 9. Vassilakos, A., M. F. Cohen-Doyle, P. Peterson, M. R. Jackson, and not clear how to explain these differences. The results in this study D. B. Williams. 1996. The molecular chaperone calnexin facilitates folding and clearly show that N-glycans can influence interactions of class I assembly of class I histocompatibility molecules. EMBO J. 15:1495. 10. Suh, W.-K., E. K. Mitchell, Y. Yang, P. Peterson, G. L. Waneck, and molecules with calnexin and calreticulin and are consistent with D. B. Williams. 1996. MHC class I molecules form ternary complexes with our previous demonstration that N-glycan of A*0201 is required calnexin and TAP and undergo peptide-regulated interaction with TAP via their for calnexin binding. We have also shown that calreticulin does not extracellular domains. J. Exp. Med. 184:337. 11. Sadasivan, B. K., P. J. Lehner, B. Ortmann, T. Spies, and P. Cresswell. 1996. bind to aglycosylated A*0201, suggesting that the lectin-like com- Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC ponents of the two chaperones function similarly (our unpublished class I proteins with TAP. Immunity 5:103. 12. Solheim, J. C., M. R. Harris, C. S. Kindle, and T. H. Hansen. 1997. Prominence data). of ␤ -microglobulin, class I heavy chain conformation, and tapasin in the inter- ␤ 2 Data obtained by comparison of 2m-deficient and positive cell actions of class I heavy chain with calreticulin and the transporter associated with lines suggests that calnexin and calreticulin bind mutually exclu- antigen processing. J. Immunol. 158:2236. 13. Carreno, B. M., J. C. Solheim, M. Harris, I. Stroynowski, J. M. Connolly, and sive subsets of class I HLA proteins (11, 34), and other studies T. H. Hansen. 1995. TAP associates with a unique class I conformation, whereas suggest that both chaperones require the presence of monoglucosy- calnexin associates with multiple class I forms in mouse and man. J. Immunol. 155:4726. lated N-oligosaccharide (7–10, 17–21). This can be interpreted as ␤ 14. Ortmann, B., M. J. Androlewicz, and P. Cresswell. 1994. MHC class I/ 2-mi- evidence that distinct peptide determinants are recognized by the croglobulin complexes associate with TAP transporters before peptide binding. two chaperones in addition to the shared carbohydrate ligand. Al- Nature 368:864. 15. Tector, M., and R. D. Salter. 1995. Calnexin influences folding of human class I ternatively, it may be that accessibility of N-oligosaccharides ␤ histocompatibility proteins but not their assembly with 2m. J. Biol. Chem. 270: change as proteins fold and assemble with subunits and that fol- 19638. lowing ␤ m binding only calreticulin has access to N-oligosaccha- 16. Nossner, E., and P. Parham. 1995. Species-specific differences in chaperone in- 2 teraction of human and mouse major histocompatibility complex class I mole- rides on class I heavy chains, as suggested for other proteins (35). cules. J. Exp. Med. 181:327. The model for class I MHC biosynthesis proposed in Figure 7 17. Hammond, C., I. Braakman, and A. Helenius. 1994. Role of N-linked oligosac- suggesting that ␤ m sterically hinders binding of calnexin, but not charide recognition, glucose trimming, and calnexin in glycoprotein folding and 2 quality control. Proc. Natl. Acad. Sci. USA 91:913. calreticulin, to position 86 glycan is consistent with either of these 18. Hammond, C., and A. Helenius. 1994. Folding of VSV G protein: sequential possibilities. interaction with BIP and calnexin. Science 266:456. ␤ 19. Hebert, D. N., B. Foellmer, and A. Helenius. 1996. Calnexin and calreticulin The observation that Ab w6/32 can recognize non- 2m-associ- promote folding,delay oligomerization and suppress degradation of influenza ated 176dg molecules in Daudi cells demonstrates that the pres- hemagglutinin in microsomes. EMBO J. 15:2961. The Journal of Immunology 837

20. Peterson, J. R., A. Ora, P. Van Nguyen, and A. Helenius. 1995. Transient lectin- antigen receptors and major histocompatibility complex antigens during their like association of calreticulin with folding intermediates of cellular and viral assembly. Proc. Natl. Acad. Sci. USA 89:4734. glycoproteins. Mol. Biol. Cell 6:1173. 29. Ennis, P. D., J. Zemmour, R. D. Salter, and P. Parham. 1990. Rapid cloning of 21. Zapun, A., S. F. Petrescu, P. M. Rudd, R. A. Dwek, D. Y. Thomas, and HLA-A,B cDNA using the polymerase chain reaction. Proc. Natl. Acad. Sci. USA J. J. M. Bergeron. 1997. Conformation-independent binding of monoglucosylated 87:2833. ribonuclease B to calnexin. Cell 88:29. 22. Maloy, W. L. 1987. Comparison of the primary structure of class I molecules. 30. Kunkel, T. A. Rapid and efficient site-specific mutagenesis without phenotypic Immunol. Res. 6:11. selection. Proc. Natl. Acad. Sci. USA 82:488. 23. Zemmour, J., A.-M. Little, D. J. Schendel, and P. Parham. 1992. The HLA-B 31. Carreno, B. M., K. L. Schreiber, D. J. McKean, I. Stroynowski, and T. H. Hansen. “negative” mutant cell line C1R expresses a novel HLA-B35 allele, which also 1995. Aglycosylated and phosphatidylinositol-anchored MHC class I molecules has a point mutation in the translation initiation codon. J. Immunol. 148:1941. are associated with calnexin: evidence implicating the class I-connecting peptide ␤ 24. De Preval, C., and B. Mach. 1983. The absence of 2-microglobulin in Daudi segment in calnexin association. J. Immunol. 154:5173. cells: active gene but inactive messenger RNA. Immunogenetics 17:133. 32. Arunachalam, B., and P. Cresswell. 1995. Molecular requirements for the inter- 25. Parham, P., C. J. Barnstable, and W. F. Bodmer. 1979. Use of monoclonal an- action of class II major histocompatibility complex molecules and invariant chain tibody (W6/32) in structural studies of HLA-A,B,C antigens. J. Immunol. 123: with calnexin. J. Biol. Chem. 270:2784. 342. 33. Romagnoli, P., and R. N. Germain. 1995. Inhibition of invariant chain (Ii)-cal- 26. Parham, P., and F. M. Brodsky. 1981. Partial purification and some properties of nexin interaction results in enhanced degradation of Ii but does not prevent the BB7.2: a cytotoxic monoclonal antibody with specificity for HLA-A2 and a vari- assembly of alpha beta Ii complexes. J. Exp. Med. 182:2027. ant of HLA-A28. Hum. Immunol. 3:277. ␤ 27. Lamb, C. A., and P. Cresswell. 1992. Assembly and transport properties of in- 34. Tector, M., Q. Zhang, and R. D. Salter. 1997. 2-Microglobulin and calnexin can variant chain trimers and HLA-DR-invariant chain complexes. J. Immunol. 148: independently promote folding and disulfide bond formation of class I histocom- 3478. patibility proteins. Mol. Immunol. 34:401. 28. Hochstenbach, F., V. David, S. Watkins, and M. B. Brenner. 1992. Endoplasmic 35. Helenius, A., E. S. Trombetta, D. N. Hebert, and J. F. Simons. 1997. Calnexin, reticulum resident protein of 90 kilodaltons associates with the T and B-cell calreticulin and the folding of glycoproteins. Trends Cell Biol. 7:193. Downloaded from http://www.jimmunol.org/ by guest on September 27, 2021