Distinct Functions for the of and Heavy Chains in the Assembly of MHC Class I Molecules

This information is current as Syed Monem Rizvi, Natasha Del Cid, Lonnie Lybarger and of September 28, 2021. Malini Raghavan J Immunol 2011; 186:2309-2320; Prepublished online 24 January 2011; doi: 10.4049/jimmunol.1002959

http://www.jimmunol.org/content/186/4/2309 Downloaded from

References This article cites 44 articles, 23 of which you can access for free at: http://www.jimmunol.org/content/186/4/2309.full#ref-list-1 http://www.jimmunol.org/

Why The JI? Submit online.

• 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 by guest on September 28, 2021

*average

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 © 2011 by The American Association of Immunologists, Inc. All rights reserved. Print ISSN: 0022-1767 Online ISSN: 1550-6606. The Journal of Immunology

Distinct Functions for the Glycans of Tapasin and Heavy Chains in the Assembly of MHC Class I Molecules

Syed Monem Rizvi,* Natasha Del Cid,*,† Lonnie Lybarger,‡ and Malini Raghavan*

Complexes of specific assembly factors and generic (ER) chaperones, collectively called the MHC class I peptide-loading complex (PLC), function in the folding and assembly of MHC class I molecules. The -binding (CRT) and partner oxidoreductase ERp57 are important in MHC class I assembly, but the sequence of assembly events and specific interactions involved remain incompletely understood. We show that the recruitments of CRT and ERp57 to the PLC are codependent and also dependent upon the ERp57 binding site and the glycan of the assembly factor tapasin. Furthermore, the

ERp57 binding site and the glycan of tapasin enhance b2m and MHC class I heavy (H) chain recruitment to the PLC, with the ERp57 binding site having the dominant effect. In contrast, the conserved MHC class I H chain glycan played a minor role in CRT

recruitment into the PLC, but impacted the recruitment of H chains into the PLC, and glycan-deficient H chains were impaired Downloaded from for tapasin-independent and tapasin-assisted assembly. The conserved MHC class I glycan and tapasin facilitated an early step in the assembly of H chain–b2m heterodimers, for which tapasin–ERp57 or tapasin–CRT complexes were not required. Together, these studies provide insights into how PLCs are constructed, demonstrate two distinct mechanisms by which PLCs can be stabilized, and suggest the presence of intermediate H chain-deficient PLCs. The Journal of Immunology, 2011, 186: 2309–2320.

he MHC class I molecule is a heterotrimer composed of (3). An oligosaccharide structure containing three glucose resides http://www.jimmunol.org/ a heavy (H) chain, a L chain (b2-microglobulin [b2m]), (Glc3Man9GlcNAc2) is initially attached to asparagine residues T and an 8- to 9-residue peptide. MHC class I H chain on newly synthesized glycoproteins. This oligosaccharide struc- assembly with b2m and peptide is facilitated by a multiprotein ture is modified to a monoglucosylated form by glucosidases I and peptide-loading complex (PLC) composed of the TAP, an as- II, which allows for recognition by and CRT. Following sembly factor tapasin, the thiol oxidoreductase ERp57, and the correct folding, the glycoprotein is deglucosylated by glucosidase endoplasmic reticulum (ER) chaperones calreticulin (CRT) and II, resulting in release from calnexin and CRT (4). Previous calnexin (1). TAP is an ATP-binding cassette transporter involved studies have shown that MHC class I binding to CRT and TAP is in translocation of peptides from the cytosol into the ER. Tapasin impacted by blocking glucosidases I and II with the inhibitor acts as a bridge between the PLC components and TAP, and castanospermine, or by inducing MHC class I deglycosylation by by guest on September 28, 2021 facilitates loading of high-affinity peptides onto MHC class I point mutations of a highly conserved N-linked site molecules. Following loading of high-affinity peptides, MHC at asparagine 86 (5–7). class I molecules dissociate from the PLC and are transported to CRT plays a more critical role in MHC class I assembly than the cell surface for recognition by CD8+ T cells (1). calnexin (8, 9), with CRT deficiency resulting in reduced cell The specific function of each protein in MHC class I assembly surface MHC class I (10, 11), enhanced intracellular trafficking and the sequence of assembly events are not completely unders- rates of MHC class I molecules (10, 11), and reduced steady-state tood. MHC class I folding is facilitated by the lectin chaperones levels of MHC class I H chains and tapasin (12). We recently calnexin and CRT. Calnexin and CRT typically interact with mo- showed that the glycan and ERp57 binding sites of CRT are im- noglucosylated glycans on substrate glycoproteins via a carbohy- portant for its recruitment into the PLC and for the assembly- drate-binding site present within a globular domain (2). Calnexin promoting functions of CRT. These studies suggested that CRT and CRT also interact with their partner ER oxidoreductase ERp57 binding to the glycans of MHC class I or tapasin or both mole- via an elongated b-stranded hairpin structure called the P domain cules could be involved in the recruitment of CRT into the PLC (12). Within the PLC, ERp57 is recruited to tapasin via a disulfide- *Department of Microbiology and Immunology, University of Michigan Medical linked interaction between C57 of ERp57 and C95 of tapasin (13), School, Ann Arbor, MI 48109; †Graduate Program in Immunology, University of and mutation of tapasin C95 abrogates tapasin–ERp57 binding Michigan Medical School, Ann Arbor, MI 48109; and ‡Department of Cell Biology and Anatomy, University of Arizona, Tucson, AZ 85724 in human cells (13–16) and reduces the efficiency of CRT re- cruitment to the PLC (14, 16). Because CRT and ERp57 are able Received for publication September 7, 2010. Accepted for publication December 9, 2010. to interact independently of other PLC components (3), P domain– This work was supported by National Institutes of Health Grants AI044115 and dependent binding between CRT and ERp57 could serve as a point AI066131 (to M.R.), by the University of Michigan Rheumatic Diseases Core Center, for the recruitment of CRT into the PLC, in addition to a glycan and by the University of Michigan Diabetes Research and Training Center. within the PLC (MHC class I and/or tapasin). To better understand Address correspondence and reprint requests to Dr. Malini Raghavan, Department of the interactions and functions mediated by the glycans of tapasin Microbiology and Immunology, 5641 Medical Science Building II, University of Mich- igan Medical School, Ann Arbor, MI 48109-5620. E-mail address: [email protected] and MHC class I molecules, in this study we used tapasin and H Abbreviations used in this article: CRT, calreticulin; DTBP, dimethyl 3,39 dithiobis- chain glycan mutants to show that the glycan of tapasin influences propionimidate 2HCl; ER, endoplasmic reticulum; HA, hemagglutinin; LIC, ligation- the recruitment of CRT into the PLC. In contrast, the conserved independent cloning; b2m, b2-microglobulin; MFI, mean fluorescence intensity; MHC class I glycan plays a minor role in CRT recruitment into the PLC, peptide-loading complex; RT, room temperature. PLC, but is important for MHC class I H chain recruitment into Copyright Ó 2011 by The American Association of Immunologists, Inc. 0022-1767/11/$16.00 the PLC and for the assembly of MHC class I H chain–b2m www.jimmunol.org/cgi/doi/10.4049/jimmunol.1002959 2310 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY complexes. These studies provided new insights into intermediate untagged HLA-B*3503 in the MSCV-neo vectors using previously de- complexes of the MHC class I assembly pathway and suggest that scribed procedures (16, 20). A total of 5.5 mg retroviral vectors encoding intermediate complexes can function as a platform for peptide ex- tapasin, HA–HLA-A2, HA–HLA-A2(N86Q), HLA-B*3503, or control empty vectors was mixed with 4 mg pCL-Eco DNA and 0.5 mg VSV-G change. encoding plasmid, and added to a mixture of Opti-MEM (Invitrogen) and FuGENE 6 (Roche). Following incubation for 20 min at RT, the mixtures were added to BOSC cells that had been grown to 70% confluency in a 10- Materials and Methods cm tissue culture dish. Media were changed after 24 h, and after 48 h, Cell lines supernatants containing retroviruses were harvested, filtered, and used to A human melanoma cell line M553 (17) (obtained from N. Bangia, Roswell infect M553 or 721.221 cells. Infected cells were selected by treatment Park Cancer Institute) and 721.221 (18) (obtained from R. DeMars, Uni- with 1 mg/ml puromycin (Sigma-Aldrich) or 1 mg/ml geneticin (Invi- versity of Wisconsin-Madison) were grown in RPMI 1640 (Life Tech- trogen) and maintained in 0.5 mg/ml puromycin or 0.5 mg/ml geneticin. nologies). BOSC cells (obtained from K. Collins, University of Michigan) After verifying HA–HLA-A2 and HA–HLA-A2(N86Q) expression by flow (19) were grown in DMEM (Life Technologies). All growth media were cytometric analyses and by immunoblotting analyses of cell lysates with supplemented with 10% (v/v) FBS (Life Technologies), 100 mg/ml anti-HA (Covance), M553/A2 cells were transduced with the tapasin- streptomycin, and 100 U/ml penicillin (Life Technologies). For 721.221 encoding retroviruses, and selected by treatment with 1 mg/ml puromy- cin (Sigma-Aldrich). The 721.221 cells were infected with retroviruses cells, the media were also supplemented with 13 L-glutamine. encoding HA–HLA-A2, HA–HLA-A2N86Q, or HLA-B*3503 by spin Mice infections, selected, and maintained in geneticin, as above. 2/2 C57BL/6 and b2m mice were purchased from The Jackson Laboratory. FACS analysis to assess MHC class I cell surface expression All mice were maintained in a specific pathogen-free mouse facility. The A total of 1 3 105–106 cells was washed three times with FACS buffer mice were cared for according to the guidelines set by the University of Downloaded from Michigan Committee on Use and Care of Animals. (PBS [pH 7.4] containing 1% FBS) and then incubated with W6/32 or anti- HA (Covance) at 1:250 dilutions for 1 h on ice. Following this incubation, DNA constructs the cells were washed three times with FACS buffer and incubated with The construction of retroviral vectors encoding wild-type and C95A human FITC-conjugated goat anti-mouse IgG (Jackson ImmunoResearch Labo- tapasin has been previously described (16). The N233Q mutants on the ratories) at 1:250 dilutions for 1 h on ice. Following incubation, the cells wild-type and tapasin(C95A) backgrounds were generated using the were washed three times with FACS buffer and analyzed using a FACS- QuikChange site-directed mutagenesis kit (Stratagene). The MSCV-puro Canto cytometer. The FACS data were analyzed with WinMDI software vector expressing wild-type or C95A tapasin was used as template and (J. Trotter, The Scripps Institute, Flow Cytometry Core Facility). http://www.jimmunol.org/ mutated using the following oligonucleotides: 59-GGCCCATGGACCG- GACAGGGGACCTTCTGGCTG-39 and 59-CAGCCAGAAGGTCCCCT- Immunoprecipitation analyses GTCCGGTCCATGGGCC-39. The retroviral vector encoding influenza M553 cells expressing different tapasin constructs were treated with 500 IU hemagglutinin (HA)-tagged HLA-A2 (HA–HLA-A2) was obtained from IFN-g (PeproTech) for 48 h and processed for anti-TAP1 immunoprecip- K. Collins (20). The N86Q mutation of wild-type HA–HLA-A2 was itation. The 721.221 cells were used without IFN-g treatment. The cells generated using QuikChange site-directed mutagenesis kit. The MSCV-neo were washed in PBS (pH 7.4) and lysed in digitonin lysis buffer (1% vector expressing wild-type HA–HLA-A2 was used as template and mu- digitonin [Wako Chemicals USA] in PBS containing EDTA-free protease tated using the following oligonucleotides, 59-CTGCGCGGCTACTAC- inhibitors [Roche] [pH 7.4]) for 1 h on ice. The lysates were centrifuged at CAGCAGAGCGAGGCCGGT-39 and 59-ACCGGCCTCGCTCTGCTGG- 4˚C for 30 min to remove cell debris, and protein concentration in lysates TAGTAGCCGCGCAG-39. HLA-B*3503 was cloned into the pMSCV-neo was determined by a bicinchoninic acid protein assay (Pierce, Thermo by guest on September 28, 2021 vector using ligation-independent cloning (LIC), as previously described Scientific). Equal microgram amounts of cell lysates were incubated with (12), but using the plasmid pMSCV-neo as the parent vector for con- anti-TAP1 antisera (21) (obtained from M. Androlewicz, Lee Moffitt struction of a LIC variant. This was accomplished by inserting a unique Cancer Center) overnight at 4˚C. For anti-tapasin (Pasta-1; obtained from sequence that can be opened with the restriction enzyme Pme I. LIC P. Cresswell, Yale University) immunoprecipitations, M553 cells expres- overhangs were generated by treatment of the linearized vector with T4 sing different tapasin constructs were treated with 500 IU IFN-g, as above, DNA polymerase exonuclease activity. The inserted sequence was con- or not treated with IFN-g, harvested, washed with PBS (pH 7.4), and then structed by annealing two oligos to generate a fragment with overhangs incubated with a membrane-permeable cross-linker, dimethyl 3,39 dithio- complementary to an EcoRI restriction site. The oligo sequences used were bispropionimidate 2HCl (DTBP) in PBS for 1 h at room temperature (RT). 59-AATTAGAGAGTTTAAACTTCCAC-39 and 59-AATTGTGGAAGTT- The cross-linker was quenched with 1 M Tris (pH 7.4) for 15 min at RT, TAAACTCTCT-39. The lyophilized oligos were resuspended in distilled and cells were then washed once with PBS (pH 7.4) and lysed and pro- water at a concentration of 20 mM. Prior to annealing, these were phos- cessed, as above. Equal microgram amounts of lysates were immunopre- phorylated using T4 polynucleotide kinase (New England Biolabs). The cipitated with anti-tapasin (Pasta-1) Ab overnight at 4˚C. For BBM.1 (22) 20-ml reactions contained 2 ml 10 mM ATP, 12.5 ml 20 mM oligo, 2 ml immunoprecipitation analysis, M553 cells expressing different tapasin 3 10 reaction buffer, 1 ml enzyme, and 2.5 ml distilled water. Reactions constructs were harvested, washed with PBS, and then incubated with the were incubated for 1 h at 37˚C, and then shifted to 95˚C for an additional cross-linker and processed for immunoprecipitation analysis, as above. 10 min. The two oligo reactions were then mixed together at 95˚C and The 721.221 cells were processed for BBM.1 immunoprecipitation anal- slowly cooled to 37˚C. A 2.5-ml aliquot was then added to a 20-ml ligation yses without treatment with cross-linker. Following overnight incubation, reaction composed of 1 ml 45 ng/ml EcoRI-cut pMSCV-neo, previously the immunoprecipitation samples were centrifuged at 13,000 rpm for 30 3 treated with calf intestinal phosphatase, 2 ml10 reaction buffer, and 13.5 min at 4˚C to remove nonspecific precipitates, and immunoprecipitates ml distilled water. The reaction was slowly cooled from 37˚C to 16˚C, and were isolated with protein G–Sepharose beads (GE Healthcare) for 1–2 h 1 ml1U/ml T4 DNA ligase (New England Biolabs) was added. Incubation at 4˚C. The beads were washed three to four times with lysis buffer con- ∼ was continued at 16˚C for 16 h. A 2-ml aliquot of the ligation reaction taining 0.1% digitonin, resuspended in reducing SDS-PAGE buffer, and was used to transform 20 ml competent XL1-Blue cells. These were plated boiled. Samples were separated on 10–15% SDS-PAGE and transferred to on Luria broth agar with ampicillin (Amp) selection and incubated at 37˚C Immobilon membranes (Millipore) for immunoblotting. Membranes were overnight. Colonies were picked and grown in 5 ml Luria broth-Amp, and blocked in 5% milk in TBS for 1 h at RT, followed by an overnight in- plasmid DNA was isolated using Qiagen minispin columns. Positive clones cubation with primary Ab in TBS–Tween 20 at 4˚C. Membranes were were identified by restriction digestion with PmeI and ClaI and confirmed washed for 2 h in TBS–Tween 20, incubated for 45–60 min with secondary by DNA sequencing. Oligo sequences for insertion of the HLA-B*3503 Ab, and washed again for 2 h. Chemiluminescence was detected using the gene into LIC–pMSCV-neo were as follows: start, 59-GGAATTAGAGAG- GE Healthcare ECL Plus kit. Western blotting analysis was done using TTTCACCATGCGGGTCACGGC-39 and stop, 59-GAATTGTGGAAG mouse anti-b2m (Biolegend) or rabbit anti-b2m Ab (Sigma-Aldrich), TTTCCTAAGCTGTGAGAGACACATC-3 9. The screening oligos used mouse anti-TAP1 Ab 148.3 (obtained from R. Tampe, Institute of Bio- were as follows: forward, 59-CACCCTAAGCCTCCGCCTCC-39 and re- chemistry, Goethe University), mouse anti-tapasin Ab (23) (obtained from verse, 59-AATGTGTGCGAGGCCAGAGGCC-39. S. Ferrone, University of Pittsburgh) or rabbit anti-tapasin antisera (gen- Viruses and infections erated against an N-terminal peptide of tapasin; obtained from T. Hanson, Washington University), mouse anti-ERp57 (Abcam) or rabbit polyclonal Retroviruses were generated in BOSC cells using the different tapasin anti-ERp57 Abs (Santa Cruz Biotechnology), goat polyclonal anti-CRT Ab constructs in the MSCV-puro vector or different HA–HLA-A2 constructs or (Santa Cruz Biotechnology), HC10 (24), and anti-HA (Covance) Ab. The The Journal of Immunology 2311 secondary Abs used were Peroxidase AffiniPure goat anti-mouse or anti- (N233Q) and tapasin(CANQ) was restored to levels similar to rabbit IgG, Peroxidase AffiniPure bovine anti-goat IgG, and Peroxidase those of the wild-type protein (Fig. 1A, lanes 7–10). The molec- AffiniPure mouse anti-rabbit IgG L chain specific (Jackson Immuno- ular basis for IFN-g–mediated stabilization of deglycosylated Research Laboratories). Immunoprecipitations from mouse cells tapasin remains unclear, but this was a consistent result that could relate to IFN-g–mediated upregulation of TAP and MHC class I, 2/2 Splenocytes from wild-type (C57BL/6) and b2m mice were isolated, and resulting in stabilization of the PLC. As expected based on pre- the RBCs were lysed using red cell lysis buffer (Sigma-Aldrich R7757). The vious studies (13, 15, 16, 25), tapasin(C95A) had reduced func- remaining splenocytes were lysed in lysis buffer (10 mM Na2HPO4,10mM Tris, 130 mM NaCl, 1% digitonin [Wako Chemicals USA], and complete tional activity in inducing MHC class I cell surface expression EDTA free protease inhibitors [Roche] [pH 7.5]). Samples were lysed on ice (Fig. 1B). Despite reduced expression, tapasin(N233Q) had higher for 1 h, followed by a 30-min centrifugation for removal of cell debris. activity than tapasin(C95A), suggesting that the tapasin(N233Q) Lysate supernatants were incubated with or without indicated Abs for 2 h at mutation, in isolation, did not significantly impair the functional 4˚C, followed by centrifugation to remove nonspecific precipitates. Fol- lowing the incubation, the samples were incubated for 1 h with protein G– activity of tapasin (Fig. 1B). However, in IFN-g–treated cells, in Sepharose beads (GE Healthcare), and the beads were washed three times which all MHC-encoded PLC components were significantly up- with 0.1% digitonin lysis buffer. The proteins were separated by SDS-PAGE regulated, tapasin(CANQ) was the only mutant that significantly and analyzed by Western blotting using the ECL Plus kit (GE Healthcare), impacted the functional activity of tapasin (Fig. 1C,1D). as above. A rabbit anti-mouse TAP1 serum (obtained from T. Hansen, To study interactions of the tapasin mutants with PLC compo- Washington University) was used at 1:15 dilution to immunoisolate murine TAP1 and associated proteins. A polyclonal goat anti-TAP Ab (Santa Cruz nents, M553 cells expressing wild-type tapasin or various tapasin Biotechnology; catalogue sc-11465) was used at 1:2000 dilution to detect mutants were treated with IFN-g for48htoequalizetapasin

TAP by immunoblot. To detect mouse CRT by immunoblot, a goat Ab expression, and cell lysates were immunoprecipitated with anti- Downloaded from specific to the N terminus of mouse CRT (Santa Cruz Biotechnology; cat- TAP1 (Fig. 2A). As expected based on previous studies, ERp57 alogue sc-7431) was used at a 1:2000 dilution. Hamster tapasin-specific Ab recruitment to TAP was C95 dependent (Fig. 2A, ERp57 blot, and rabbit anti-b2m serum were also obtained from T. Hansen and used at 1:3000 dilutions to detect murine tapasin and b2m, respectively, by im- lanes 3 and 4 compared with lanes 1 and 2). Notably, however, munoblotting. A rabbit anti-Kb antiserum (EX8) was used to detect H chains TAP-ERp57 binding was also reduced in cells expressing tapasin by immunoblot (1:7500) and was provided by J. Yewdell, National Institute (N233Q) (Fig. 2A, ERp57 blot, lanes 5 and 6 compared with lanes of Allergy and Infectious Diseases. To detect ERp57, a rabbit polyclonal Ab 1 and 2). CRT interaction with TAP was reduced in cells http://www.jimmunol.org/ (Santa Cruz Biotechnology; catalogue 28823) was used (1:3000). Second- ary Abs (Jackson ImmunoResearch Laboratories) were all conjugated to expressing tapasin(C95A), as previously shown (14, 16). The new HRP: mouse anti-rabbit (L chain specific), bovine anti-goat, and goat anti- result that CRT recruitment was slightly impacted in cells ex- hamster. pressing tapasin(N233Q) (note the slightly higher level of tapasin Metabolic labeling analysis present in the context of this mutant), and essentially undetectable in cells expressing tapasin(CANQ) (Fig. 2A, calreticulin (CRT) Cells expressing HA–HLA-A2 or HA–HLA-A2(N86Q) and tapasin were 35 blot, lanes 1–8) indicated that tapasin’s glycan and tapasin- pulsed with 0.2 mCi S-labeled methionine–cysteine for 10 min and associated ERp57 represented the major modes of CRT recruit- chased in cold media for 0, 30, and 60 min. The cells were then lysed in digitonin lysis buffer (1% digitonin in PBS containing EDTA-free protease ment to the PLC. Similar trends were observed in anti-tapasin inhibitors [pH 7.4]) for 1 h on ice. The lysates were centrifuged at 4˚C for immunoprecipitations of proteins from IFN-g–treated cells (Fig. by guest on September 28, 2021 30 min to remove cell debris. Proteins in lysates were immunoprecipitated 2B, ERp57 and CRT blots), although an impact of the single tapasin with anti-HA or BBM.1 Abs for 1 h at 4˚C and followed by protein G (NQ) mutation on tapasin–CRT binding was not visualized. In anti- beads for 1 h at 4˚C. The beads were washed three to four times with lysis buffer containing 0.1% digitonin, resuspended in reducing SDS-PAGE tapasin immunoprecipitations, efficient observation of PLC com- buffer, and boiled. For secondary immunoprecipitations, the primary ponents required the use of an alkylating agent or a cell-permeable BBM.1 immunoprecipitated protein samples were boiled for 5 min in 13 cross-linker such as DTBP prior to cell lysis (16). The former glycoprotein denaturing buffer (New England Biolabs) and diluted in 1 ml treatment stabilizes disulfide-linked tapasin–ERp57 conjugates and buffer containing 1% Nonidet P-40 (Sigma-Aldrich). The samples were associated PLC components, and the latter treatment stabilizes the processed for immunoprecipitation with anti-HA Ab, as above. The pro- teins were separated by SDS-PAGE and analyzed by phosphorimaging same interactions, but using a cross-linking approach (DTBP is analyses. M553 cells expressing different tapasin constructs were pulsed a reversible cross-linker). b2m recruitment to TAP was essentially with 0.2 mCi 35S-labeled methionine–cysteine for 10 min and treated with undetectable in the context of the tapasin(CANQ) mutant, reduced cross-linker before lysis in digitonin lysis buffer, as above. Proteins in by the tapasin(N233Q) mutation, and strongly reduced in the con- lysates were processed for primary and secondary immunoprecipitations, as above, using BBM.1 and H chain-specific (HC10) Ab (24). text of the tapasin(C95A) mutation (Fig. 2A, b2m blot, lanes 1–8). Similarly, whereas the tapasin(C95A) mutation reduced the effi- Statistical analysis ciency of TAP–MHC class I H chain binding in IFN-g–treated cells Analyses were performed using the GraphPad Prism software (GraphPad (Fig. 2A, H chain blot, lanes 1–8), the tapasin(CANQ) double Software). The p values were calculated using the Student’s paired or un- mutant induced a marked impairment of H chain recruitment to the paired t tests. The p values ,0.05 were considered statistically significant. PLC. Similar results were obtained in the anti-tapasin (Pasta-1) immunoprecipitation analyses (Fig. 2B, H chain and b2m blots). Results In summary, in IFN-g–treated cells, TAP and tapasin binding to The glycan of tapasin facilitates CRT recruitment to the PLC CRT and H chains was abrogated or strongly impacted by the and enhances the functional activity of tapasin tapasin(CANQ) mutation and reduced by the tapasin(C95A) mu- Tapasin-deficient M553 cells (17) were used to characterize tation. Interactions of ERp57 and b2m with TAP and tapasin were interactions and functional activities mediated by tapasin, degly- strongly tapasin(C95) dependent and partially impacted by the cosylated (N233Q) tapasin, the C95A mutant of tapasin that is tapasin(N233Q) single mutant. These findings indicated that the deficient in ERp57 binding, and a double mutant with altered glycan of tapasin and its ERp57 binding site cooperate in the as- glycosylation and ERp57 binding sites [tapasin C95A/N233Q, sembly of the PLC. hereafter referred to as tapasin(CANQ)]. The deglycosylated Despite reduced expression of the tapasin(N233Q) constructs in version of tapasin was expressed at lower levels than the glyco- the absence of IFN-g treatment, Pasta-1 immunoprecipitations sylated version (Fig. 1A, lanes 4 and 5 compared with lanes 2 and recovered comparable amounts of each tapasin protein (Fig. 2C, 3). However, in IFN-g–treated cells, expression of tapasin tapasin panel), thus allowing for assessment of effects of the 2312 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY

FIGURE 1. The glycan of tapasin impacts its functional activity. Tapasin constructs are abbre- viated as wild type (WT), C95A (CA), N233Q (NQ), and C95AN233Q (CANQ). A, Representa- tive immunoblots show expression of tapasin and MHC class I H chains in lysates from M553 cells expressing no tapasin or the indicated tapasin constructs following 48-h IFN-g (500 IU) treat- ment or no treatment. Rabbit anti-tapasin and MHC class I H chain-specific HC10 Abs, re- spectively, were used. Class I HC indicates MHC class I H chains, in this and all subsequent blots. Nonspecific bands are indicated as ns. B and C, Bar graphs show induction of MHC class I cell surface expression in M553 cells by the indicated tapasin constructs following 48-h IFN-g treatment (C) or no treatment (B). The y-axes show ratios of mean fluorescence intensities (MFI) of MHC class I staining in cells that had been infected with retroviral constructs encoding indicated tapasin constructs relative to the parent uninfected M553 Downloaded from cells (+tapasin/2tapasin MFI ratios). The tapasin (CA) and tapasin(CANQ) mutants were signifi- cantly impaired in restoring MHC class I cell surface expression compared with tapasin(WT) and tapasin(NQ), respectively (*p = 0.0275 and 0.0264, respectively; only these comparisons were http://www.jimmunol.org/ made because the proteins were not expression matched in other cases). In IFN-g–treated cells, only the tapasin(CANQ) mutant was significantly impaired in restoring class I surface expression compared with tapasin(WT) (*p = 0.0335). Data are averages of three (B, C) independent FACS analyses. D, Representative histograms show cell surface induction of MHC class I molecules in IFN-g–treated M553 cells in the presence or ab- sence of the indicated tapasin constructs. Filled by guest on September 28, 2021 histogram shows control staining without primary Ab. MHC class I surface expression was analyzed by flow cytometry using the W6/32 Ab. different mutations upon the stabilities of tapasin binding to PLC binding sites of tapasin. Previous studies have described that CRT components in the absence of IFN-g treatment. Under this con- recruitment to the PLC is also reduced in b2m-deficient human dition, tapasin–CRT binding was significantly impacted by the cells (5, 26). To further investigate requirements for CRT and 2/2 tapasin(N233Q) mutation, and appeared to be more strongly im- ERp57 recruitment, b2m mouse splenocytes were analyzed for pacted by tapasin(N233Q) compared with the tapasin(C95A) their PLC compositions (Fig. 3). Steady-state levels of MHC class mutation (Fig. 2C, CRT blots, lanes 5 and 6 compared with lanes I H chains are not reduced in the lysates of b2m-deficient cells 3 and 4). Tapasin–CRT binding was also undetectable in the compared with wild-type cells (Fig. 3, lanes 1 and 2, MHC class I context of tapasin(CANQ). Additionally, the tapasin–ERp57 in- H chain blot). However, the species that exists is entirely ER lo- teraction was significantly impacted by the tapasin(N233Q) mu- calized as assessed by Endo H digestion, whereas a minority of the tation (Fig. 2C, ERp57 blots, lanes 5 and 6 compared with lanes 1 H chains in wild-type lysates are ER localized (data not shown). 2/2 and 2). In the absence of IFN-g treatment, a much stronger effect Splenic cell lysates from wild-type C57BL/6 or b2m mice of the tapasin(C95A) mutation upon tapasin–H chain binding was were immunoprecipitated with anti-TAP1 (Fig. 3). As expected apparent compared with IFN-g–treated cells (Fig. 2B,2C, H chain based on previous results (27, 28), MHC class I H chain re- blots, lanes 1 and 2 compared with lanes 3 and 4), and b2m cruitment was strongly impacted in cells lacking b2m compared binding to tapasin was below the detection threshold in the ab- with wild-type cells (Fig. 3, class I H chain blot, lanes 6 and 7). sence of IFN-g treatment, even in the context of wild-type tapasin Additionally, TAP–CRT association is essentially undetectable in (data not shown). Overall, the interactions analyses of Fig. 2C cells lacking b2m compared with wild-type cells, as previously suggested that tapasin–CRT binding was more strongly impacted shown (5, 26) (Fig. 3, CRT blot, lanes 6 and 7). Notably, ERp57 by tapasin(N233Q) relative to tapasin(C95A) in the absence of recruitment efficiency was reduced in b2m-deficient cells (Fig. 3, IFN-g (Fig. 2C, CRT blot), although the opposite result was ERp57 blot, lanes 6 and 7). ERp57 is thought to be recruited to the obtained from IFN-g–treated cells (Fig. 2B, CRT blot). PLC via heterodimeric interactions with tapasin (13, 29). How- ever, it appears that b m also aids in the recruitment of ERp57 b 2 2m is required for efficient CRT and ERp57 recruitment to the (Fig. 3, ERp57 blot), either via direct interactions, or indirectly. PLC MHC class I H chains have been shown in some studies to interact The findings described above indicated that CRT recruitment to the directly with ERp57 (30, 31). It is thus also possible that b2m PLC was abrogated by mutations of the glycosylation and ERp57 indirectly facilitates recruitment of MHC class I-associated ERp57 The Journal of Immunology 2313 Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 2. The glycan of tapasin impacts interactions within the PLC. For these analyses (A, B), M553 cells expressing wild-type or various tapasin mutants were treated with IFN-g to upregulate TAP and MHC class I or not treated with IFN-g (C). A, Immunoblotting analyses of anti-TAP1 immu- noprecipitates of the indicated cells. B and C, Immunoblotting analysis of anti-tapasin (Pasta-1) immunoprecipitates of the indicated cells. For Pasta-1 immunoprecipitations, cells were treated with the membrane-permeable cross-linker DTBP prior to lysis. Membranes were probed using Abs specific for indicated proteins. Indicated amounts (5–10 ml) of immunoprecipitation samples were used for the analyses. Data are representative of two (B), three (C), or four (A) analyses. The lanes labeled lysate + beads correspond to lysates incubated with protein G beads (without Ab), and the lanes labeled buffer + antibody correspond to Ab incubated with buffer and protein G beads (no cell lysates were used). Ab-derived or other nonspecific bands are indicated as ns. Dotted lines in C, ERp57 blot, indicate lanes that were cut and pasted from same blot to preserve the order of presentation of lanes. by stabilizing a conformation of MHC class I H chain that is better cells expressing HA–HLA-A2 with IFN-g caused a 1.9-fold in- able to interact with ERp57. duction of cell surface expression, surface expression of HA– HLA-A2(N86Q) was much less significantly impacted by the The conserved glycan of MHC class I H chains facilitates H treatment (Fig. 4A, left panel). The HA–HLA-A2(N86Q) mutant chain recruitment into the PLC, and mediates intrinsic and was also highly impaired in tapasin-assisted assembly, with no tapasin-assisted assembly tapasin-mediated induction of cell surface expression observable We next investigated the role of the highly conserved MHC class I either in the presence or absence of IFN-g, in contrast to the wild- H chain glycans (N86) in PLC formation and tapasin-assisted type counterpart, whose cell surface expression was strongly in- assembly. HA–HLA-A2 and deglycosylated HA–HLA-A2 duced by tapasin (Fig. 4A, right panel). (N86Q) were first expressed in M553 cells to assess impacts of To assess interactions mediated or stabilized by HA–HLA-A2 H chain deglycosylation on tapasin-assisted assembly. Flow compared with those mediated by HA–HLA-A2(N86Q), the cytometric analysis showed that HA–HLA-A2(N86Q) was im- constructs were expressed in the MHC class I H chain-deficient paired in cell surface expression compared with wild-type HA– 721.221 cells, which lack HLA-A, -B, and -C expression and HLA-A2, particularly in IFN-g–treated cells (Fig. 4A). This was which express very low levels of HLA-E and -F (18). Cells were despite the slightly stronger intracellular expression of HA–HLA- lysed and proteins were immunoprecipitated with anti-TAP1. As A2(N86Q) (Fig. 4B, lanes 1, 2, 5, and 6). Whereas treatment of previously reported with other deglycosylated class I molecules 2314 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY

FIGURE 3. b m2/2 cells are deficient in recruiting CRT into the PLC.

2 Downloaded from Anti-TAP1–immunoprecipitated proteins (IP) or lysates were analyzed by FIGURE 4. The conserved MHC class I glycan impacts tapasin- immunoblotting with the indicated Abs. Splenic lysates from wild-type 2 2 independent and tapasin-assisted MHC class I assembly. A, Left panel, Bar (WT) or b m / mice were used for the immunoprecipitations. Data are 2 graphs show mean fluorescence of M553 cells expressing HA–HLA-A2 or representative of one analysis for b m, two analyses for class I H chain, 2 HA–HLA-A2(N86Q) as assessed by staining with the anti-HA Ab in un- two analyses for ERp57, three analyses for tapasin, and five analyses for treated and 72-h IFN-g–treated cells (in the absence of tapasin expression). CRT. Two bands are observed for CRT, most likely corresponding to a full- Cell surface expression of HA–HLA-A2(N86Q) was significantly impaired

length version and a degradation product. Two species of MHC class I H http://www.jimmunol.org/ compared with that of HA–HLA-A2 both in IFN-g–treated cells (***p = chains are observed in the cell lysates (labeled as HC1 and HC2), most 0.0005) and in untreated cells (*p = 0.0404). Right panel, Tapasin-medi- likely corresponding to differently glycosylated forms of H chains. Only 2 2 ated induction of HA–HLA-A2 or HA–HLA-A2(N86Q) in untreated and HC2 is detectable in the immunoprecipitations and in b m / cells. No Ab 2 72-h IFN-g–treated M553 cells. The y-axis shows ratio of MFI of MHC controls were performed by incubating indicated lysates with protein G class I staining in M553 cells expressing tapasin and indicated class I beads, and Ab indicates signals obtained with Ab alone (without lysates). relative to M553 cells expressing indicated class I alone (+tapasin/2 Ab-derived or other nonspecific bands are indicated as ns. tapasin MFI ratios). The differences in cell surface expression between HA–HLA-A2 and HA–HLA-A2(N86Q) were also significant in tapasin- expressing IFN-g–treated (**p = 0.0018) or untreated cells (*p = 0.0153).

Paired t tests were used for the statistical analyses. B, Immunoblotting by guest on September 28, 2021 (6, 7, 32), HA–HLA-A2(N86Q) showed reduced binding to TAP analyses of indicated cell lysates using Abs specific for tapasin (Tpn) and compared with HA–HLA-A2 (Fig. 5A). However, despite im- HA–HLA-A2 (anti-HA). Data are averages of three (A) independent FACS paired recruitment of HA–HLA-A2(N86Q) H chains, the HA– analysis, and the blots (B) are representative of two independent analyses. HLA-A2(N86Q) mutation had no significant impact on the re- Nonspecific bands are indicated as ns. cruitment of b2m, tapasin, CRT, and ERp57 into the PLC (Fig. 5A, lanes 3 and 4 compared with lanes 5 and 6 of b2m, tapasin, CRT, and ERp57 blots). Slightly more TAP1 was immunoprecipitated H chain recruitment. Furthermore, b2m recruitment was also not from HA–HLA-A2(N86Q) cells compared with HA–HLA-A2 strongly linked to that of the H chain. Two interpretations of these cells, and correspondingly more b2m, tapasin, CRT, and ERp57 data are as follows: 1) H chains are recruited into preassembled were coimmunoprecipitated with TAP1 (Fig. 5A). PLCs containing b2m, with heterodimer assembly occurring The finding that enhanced recruitment of HA–HLA-A2 H chains within the PLC; or 2) the PLC may function as a platform for into the PLC did not cause the additional recruitment of b2m, peptide and b2m exchange. tapasin, CRT, and ERp57 suggested the possibility that the latter interactions were independent of H chain recruitment. This finding Tapasin and the MHC class I glycan facilitate an early step in b was confirmed by overexpression of another MHC class I H chain, the formation of 2m–H chain heterodimers HLA-B*3503, in 721.221 cells (Fig. 5B). In immunoblotting To further investigate the possibility that the initial assembly of analyses, HLA-B*3503 is efficiently recognized by the H chain- heterodimeric MHC class I molecules occurs within the PLC, we specific (HC10) Ab (which does not efficiently recognize HA– compared early assembly of b2m–H chain complexes in cells HLA-A2). Using this Ab, we estimated that there was an 8-fold expressing different tapasin mutants by metabolic labeling analyses increase in the recruitment of HC10-reactive H chains into the (10-min pulse). The tapasin(C95A) mutation was a good tool for PLC of 721.221-B3503 cells compared with the parent 721.221 these analyses, as the tapasin(C95A) mutation strongly impacted cells (Fig. 5B, HC10 blot, compare lanes 5–8 with lanes 1–4). A the recruitment of b2m into the PLC (Fig. 2). Immunoprecipitations ,2-fold increase in ERp57 association with TAP was observed in were undertaken with BBM.1, an Ab that can recognize both free the context of HLA-B*3503 that had not been observed in the b2m and heterodimeric forms of b2m (22), followed by a secondary context of HA–HLA-A2 (Fig. 5B,5A). HLA-B*3503–induced immunoprecipitation with a H chain-specific Ab (HC10) (24). ERp57 recruitment to the PLC could be an allele-specific effect Whereas tapasin deficiency impacted the formation of H chain– or could reflect different levels of expression of HLA-B*3503 b2m complexes (relative to cells expressing wild-type tapasin), compared with HA–HLA-A2 in 721.221 cells. Notably, however, none of the tapasin mutants displayed a significant impairment in the 8-fold enhancement in the recruitment of H chains into the early steps of heterodimer assembly (Fig. 6A, phosphorimaging PLC did not correlate with significant increases in PLC-associated analyses; 6B, quantification of phosphorimaging analyses). Im- CRT (Fig. 5B). Thus, CRT recruitment was largely independent of paired binding of b2m to PLC components in the context of tapasin The Journal of Immunology 2315 Downloaded from

FIGURE 5. PLCs of MHC class I-deficient 721.221 cells display defective recruitment of deglycosylated H chains and do not significantly alter their compositions under conditions in which more HC10-reactive H chains are recruited. A, Upper panel, Immunoblotting analyses of anti-TAP1 immuno- precipitates from 721.221 cells expressing HA–HLA-A2 or HA–HLA-A2(N86Q) using Abs specific for indicated proteins. Lower panel, Immunoblotting analyses of lysates from 721.221 cells expressing HA–HLA-A2 or HA–HLA-A2(N86Q) using Ab specific for HA–HLA-A2 (anti-HA). B, Immunoblotting http://www.jimmunol.org/ analysis of anti-TAP1 immunoprecipitates or lysates from 721.221 cells expressing or lacking exogenous HLA-B*3503 using Abs specific for indicated proteins. Indicated amounts (5–10 ml for A, and 2–16 ml for B) of immunoprecipitation samples were used for the analyses. Data are representative of two (B) or three (A) independent analyses. The lanes labeled buffer + Ab correspond to Ab and lysis buffer (no lysates) incubated with protein G beads. Ab- derived or other nonspecific bands are indicated as ns.

(C95A) mutants was verified by immunoblotting analyses of glycan mutant, by MHC class I and BBM.1 immunoprecipitation BBM.1 immunoprecipitates with various Abs (Fig. 6C). The im- analyses (7). However, no significant effects of HA–HLA-A2 by guest on September 28, 2021 munoprecipitation/immunoblotting analyses (Fig. 6C) were under- glycan mutation or HA–HLA-A2 expression upon b2minter- taken without IFN-g treatment of cells, as were the pulse–chase actions with tapasin, CRT, or ERp57 were detectable (Fig. 6F), analyses (Fig. 6A). Under these conditions, tapasin(N233Q) and again indicating an absence of coupling of b2m binding to TAP, tapasin(CANQ) display reduced expression relative to the wild-type tapasin, or CRT to the b2m–H chain interaction. and tapasin(C95A) constructs, as shown in Figs. 1 and 6C (tapasin blots, lysates). The tapasin(C95A) mutation strongly reduced or Discussion impaired b2m–tapasin, b2m–ERp57, and b2m–CRT binding effi- Several studies have addressed the role of MHC class I glycans in ciencies, and the tapasin(CANQ) mutation rendered those inter- MHC class I folding and assembly (5–7, 33–35), but the role of the actions essentially undetectable (Fig. 6C, tapasin, ERp57, and CRT glycan of tapasin in the functional activity of tapasin has not been blots). Recovery of H chains was slightly reduced in cells investigated. We previously observed that glycan and ERp57- expressing tapasin(C95A) and more strongly reduced in cells dependent interactions contribute to CRT recruitment to the expressing tapasin(CANQ) compared with those expressing wild- PLCs of murine fibroblasts (12). The current findings (Fig. 2) type tapasin (Fig. 6C, H chain blot), consistent with the reductions support the model that tapasin-associated ERp57 and tapasin’s in functional activities of the CANQ mutants (Fig. 1B). These glycan are the relevant sites that mediate CRT recruitment. PLCs results indicated that the steady-state assembly of heterodimers was from cells expressing the tapasin(C95A) mutant showed the suboptimal in the context of the tapasin(C95A) mutants (Fig. 6C), presence of CRT, H chains, and b2m, although at reduced levels as expected. However, the C95A mutation did not impact the initial compared with those present in the context of wild-type tapasin assembly efficiency of MHC class I heterodimers (Fig. 6B). Thus, (Fig. 2A, lanes 3 and 4). Mutation to tapasin(CANQ) further re- whereas the presence of tapasin did significantly influence an early duced the efficiency of H chain recruitment and rendered b2m and step in heterodimer assembly (Fig. 6A,6B), an intact PLC is not CRT binding undetectable (Fig. 2A, lanes 3 and 4 compared with required for efficient early assembly of MHC class I heterodimers. lanes 7 and 8). These findings indicate the presence of PLCs in Instead, an intact PLC appears to be required for optimization of which MHC class I recruitment to TAP is stabilized by tapasin’s heterodimer assembly. glycan, most likely via the binding of CRT (Fig. 7A). The ob- The H chain glycan mutation strongly impaired the recovery of servation that the tapasin(N233Q) mutation has a stronger impact heterodimers (Fig. 6D [phosphorimaging analyses], 6E [quantifi- upon PLC formation in the context of tapasin(C95A) compared cation of phosphorimaging analyses]), indicating that the MHC with the wild-type tapasin context (Fig. 2A,2B) is important, as it class I glycan is also critical for the efficient early assembly of suggests that tapasin’s glycan may in fact be relevant for stabi- MHC class I heterodimers. Additionally, there was a reduction in lizing PLCs containing tapasin molecules that are not in hetero- steady-state level of heterodimer formation in the context of HA– dimeric association with ERp57. HLA-A2(N86Q) compared with HA–HLA-A2 in the 721.221 The tapasin(N233Q) mutation reduced the extent of ERp57 cells (Fig. 6F), as previously reported for the HLA-B*2705 N86Q recruitment to tapasin and TAP compared with wild-type tapasin 2316 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 6. Tapasin and the conserved MHC class I glycan impact an early step in the assembly of MHC class I heterodimers. A, M553 cells expressing indicated tapasin constructs or M553 cells lacking tapasin were pulsed with [35S]methionine–cysteine, treated with the cross-linker (DTBP), followed by cell lysis and immunoprecipitations. Phosphorimaging analyses show relevant bands from primary immunoprecipitation with anti-b2m (BBM.1; top panel), anti- MHC class I (HC10; middle panel), and secondary immunoprecipitation with HC10 following primary immunoprecipitation with BBM.1 (bottom panel). B, Quantification of A, plotting the ratios of band intensity quantifications (bottom panel/top panel). The y-axis shows the ratio of intensities of the H chain band in the bottom panel relative to the corresponding b2m band intensity in the top panel. Averaged ratios from two independent analyses are presented. Ratios derived from M553 cells were significantly reduced compared to those obtained from M553 cells expressing wild-type tapasin (p = 0.0179 based on an unpaired t test), whereas the ratios derived from M553 cells expressing mutant tapasins were not significantly reduced. C, Upper panel, Immunoblotting analysis of anti-b2m (BBM.1) immunoprecipitates from M553 cells expressing indicated tapasin constructs using Abs specific for indicated proteins. Cells were treated with DTBP prior to lysis. Lower panel, Immunoblotting analysis of lysates from M553 cells expressing indicated tapasin constructs using Ab specific for tapasin (Tpn). D, M553 cells expressing tapasin and indicated HA–HLA-A2 constructs were pulsed with [35S]methionine–cysteine and chased in cold media for indicated time points, followed by cell lysis and immunoprecipitations. Phosphorimaging analyses show primary immunoprecipitation with anti-b2m (BBM.1; top panel)or anti-HA (middle panel) and secondary immunoprecipitation with anti-HA following primary immunoprecipitation with BBM.1 (bottom panel). Samples were processed, as indicated in A, but without treatment with DTBP. E, Quantification of D, plotting the ratios of band intensity quantifications (bottom panel/top panel). The y-axis shows the ratio of intensities of the HA–HLA-A2 or HA–HLA-A2(N86Q) H chain band in the bottom panel relative to the corresponding b2m band intensity in the top panel. Ratios are plotted rather than the actual bottom panel HA–HLA-A2 and HA–HLA-A2(N86Q) intensities, to normalize for potential differences in the immunoprecipitation efficiencies of the different samples. Significant differences between HA–HLA-A2 and HA–HLA-A2(N86Q) were observed in two of three experiments (*p = 0.0037 and 0.0002) based on a paired t test. The differences between HA–HLA-A2 and HA–HLA-A2(N86Q) in a third experiment, although reduced at all three time points for HA–HLA-A2(N86Q) compared with HA–HLA-A2, did not achieve statistical significance. F,

Upper panel, Immunoblotting analysis of anti-b2m (BBM.1) immunoprecipitates from 721.221 cells expressing HA–HLA-A2 or HA–HLA-A2(N86Q), using Abs directed against indicated proteins. Lower panel, Immunoblotting analysis of lysates from 721.221 cells expressing HA–HLA-A2 or HA–HLA-A2(N86Q) using anti-HA Ab. In C and F, the lanes labeled buffer + Ab correspond to Ab incubated with lysis buffer and protein G beads (no cell lysates were used). Ab- derived or other nonspecific bands are indicated as ns. Indicated amounts (7–14 ml for A, 5–10 ml for C, and 2.5–10 ml for F) of immunoprecipitation samples were used for the analyses. Data are representative of two (A–C, F) or three (D, E) independent analysis. The Journal of Immunology 2317

(Fig. 2, ERp57 blots, lanes 1 and 2 compared with lanes 5 and 6). observations). Thus, we postulate that conjugation between C57 of Furthermore, b2m deficiency, which strongly impaired CRT re- ERp57 and C95 of tapasin may induce a conformational change cruitment to the PLC (Fig. 3), or CRT deficiency itself, also im- within the complex depicted in Fig. 7A, which disengages CRT pacted the recruitment efficiencies of ERp57 into the PLC (data from tapasin’s glycan, repositioning CRT for binding to tapasin- not shown). CRT bound to tapasin’s glycan could transiently conjugated ERp57, via a P domain–based interaction (Fig. 7B). initiate ERp57 recruitment via a CRT P domain–ERp57 in- Thus, CRT may equilibrate between conformational states in teraction. This interaction could position ERp57 in appropriate which its recruitment to the PLC is stabilized by tapasin’s glycan proximity for conjugation to C95 of tapasin. Molecular modeling and tapasin-associated ERp57 (Fig. 7A 7B). In turn, CRT could studies [using calnexin’s structure (2) as a model for CRT struc- stabilize the recruitment of both ERp57 and MHC class I mole- ture and the tapasin–ERp57 complex structure (36)] suggest that cules via cooperative binding interactions. a CRT bound to tapasin’s glycan cannot simultaneously contact Within the same cell type, differences were noted in the com- the ERp57 molecule present within the same tapasin–ERp57 positions and functional activities of tapasin complexes in IFN-g– conjugate (S.J. Wijeyesakare and M. Raghavan, unpublished treated cells compared with untreated cells (Fig. 2B compared Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021

FIGURE 7. Within the PLC, CRT equilibrates between tapasin glycan- and tapasin(C95)-associated forms, and intermediate H chain-depleted PLCs are detectable. A, Tapasin’s glycan contributes to the stabilization of CRT, H chain, and b2m recruitment to tapasin (Tpn). Such a complex or its subcomplex may facilitate ERp57 recruitment by transient interaction of ERp57 with the tip of CRT’s P domain, thus bringing ERp57 into proximity to tapasin(C95). B, ERp57 becomes conjugated to tapasin(C95). This is shown to displace CRT from tapasin’s glycan, and link CRT to the PLC via P domain-mediated interactions with tapasin(C95)-associated ERp57. In this study, recruitments of CRT and ERp57 to the PLC are shown to be codependent and also de- pendent upon tapasin(C95) and tapasin’s glycan. In IFN-g–treated M553 cells, where there is strong upregulation of MHC class I expression, CRT re- cruitment to tapasin was largely tapasin(C95) dependent. Tapasin’s glycan and tapasin(C95) both enhanced MHC class I recruitment to the PLC, with dominant stability conferred by tapasin(C95). Thus, the equilibrium is shifted toward B in IFN-g–treated M553 cells, under conditions in which ERp57 is not limiting. In M553 cells not treated with IFN-g, conditions under which MHC class I H chain and b2m expression levels are low, CRT recruitment to tapasin was largely tapasin(N233) dependent. Thus, the equilibrium is shifted toward A. MHC class I recruitment was nevertheless largely C95 dependent, suggesting that the A→B transition is favored by the presence of both MHC class I heterodimers and ERp57. Thus, cellular conditions are expected to determine the steady-state prevalence of each complex and their subcomplexes. C and D, In proposed models of the MHC class I assembly pathways based on the data from Figs. 5 and 6, intermediate H chain-depleted complexes are predicted to form. Suboptimally assembled MHC class I heterodimers are predicted to associate with intermediate PLC, resulting in b2m exchange and peptide dissociation from incoming heterodimers, and resulting in the as- sembly of a mature PLC that has incorporated H chains. Intermediate H chain-depleted complexes are depicted as being tapasin(N233) and tapasin(C95) linked (C and D, respectively). However, further investigations will be needed to verify each possibility. 2318 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY with 2C). CRT recruitment to tapasin was more strongly tapasin of classical H chains in 721.221 cells (37, 38), and the lack of (N233) dependent in cells that were not treated with IFN-g (Fig. significant induction of b2m or CRT recruitment into the TAP 2B compared with 2C, CRT blots, lanes 1–6). Nonetheless, tapasin complex by overexpression of HLA-B27 H chains (38). The (N233)-mediated stabilization of MHC class I recruitment was analyses described in this study confirm that associations of b2m not readily detectable in the absence of IFN-g treatment (Fig. 2C, and CRT with the PLC are relatively unaffected by an 8-fold or MHC class I blot, lanes 3 and 4 compared with lanes 7 and 8). higher enhancement in H chain incorporation. ERp57 incor- Because MHC class I H chain and b2m expression levels are both poration into the PLC was slightly induced by the presence of low in M553 cells in the absence of IFN-g treatment, these find- HLA-B*3503 H chains. Although it is possible that nonclassical ings are consistent with the possibility that binding of MHC class I MHC class I H chains facilitate b2m, CRT, and ERp57 re- molecules to tapasin in an interaction stabilized by CRT shifts the cruitment, nonclassical H chains are estimated to be present at low Fig. 7A 7B binding equilibrium toward Fig. 7B. Thus, sub- levels (#1%) in 721.221 cells compared with classical MHC class complexes of Fig. 7A (e.g., those involving just tapasin and CRT) I H chain (39). Together, these findings support the existence of may accumulate under conditions in which MHC class I expres- intermediate PLCs containing tapasin, ERp57, b2m, and CRT sion is low (Fig. 2C). Furthermore, complexes of the Fig. 7A type (Fig. 7C,7D). It is presently unclear whether the observed binding might also accumulate under conditions in which ERp57 levels of CRT to TAP in the 721.221 cells is dominated by N233- become limiting relative to those of tapasin and MHC class I dependent interactions, C95-dependent interactions, or both. molecules, a scenario possible in IFN-g–treated cells, when Thus, whereas intermediate H chain-depleted complexes are tapasin is under endogenous IFN-g control. Together, the findings depicted as being tapasin(N233) and tapasin(C95) linked (Fig. 7C, of Fig. 2 suggest a coupling between tapasin(N233)-mediated 7D), further investigations will be needed to verify both possi- Downloaded from complexes and tapasin(C95)-mediated complexes (Fig. 7A 7B), bilities. and that cellular conditions determine the particular complex that Despite recruitment of b2m into intermediate PLCs in a H predominates in the steady state. The data also indicate that, in chain-independent manner, the PLC does not appear to be the site human cells, tapasin(C95)-mediated complexes are more stabi- for initial H chain–b2m assembly, as PLCs deficient in b2m re- lizing for MHC class I recruitment. Tapasin from species that lack cruitment [via the tapasin(C95A) mutation] were competent for

C95, but that have N233, such as Grass carp (Ctenopharyngodon early assembly of heterodimers (Fig. 6A,6B). Rather, the in- http://www.jimmunol.org/ idellus), Zebra fish (Danio rerio), and Atlantic salmon (Salmo termediate PLCs that are enriched in b2m (relative to H chains) salar), may stabilize their interactions with MHC class I mole- may serve as a platform for peptide exchange that is coupled to b2m cules solely via the tapasin glycan-dependent mechanism. exchange from unstable H chain–b2m heterodimers that are Tapasin–MHC class I interactions were generally stronger in recruited into the PLC. Indeed, the presence of excess exogenous IFN-g–treated cells compared with untreated cells. For example, b2m has been shown to facilitate peptide exchange from MHC whereas b2m recruitment to tapasin was readily detectable in IFN- class I molecules in vitro (40). Interactions of suboptimally as- g–treated cells, this interaction was not detectable in untreated sembled H chain–b2m heterodimers with intermediate b2m- cells even in the context of wild-type tapasin, and H chain re- enriched PLC could result in peptide and b2m exchange from the cruitment was strongly diminished by the tapasin(C95A) mutation preassembled heterodimers, simultaneously recruiting H chains by guest on September 28, 2021 alone in cells that were not treated with IFN-g (Fig. 2B,2C). IFN- into the PLC. Previous in vitro studies have demonstrated inter- g treatment markedly enhances steady-state levels of TAP, tapasin, actions between H chains and tapasin with substoichiometric levels MHC class I H chains, and b2m, and the increase in these protein of b2m [unless b2m is added in excess (41)]. Additionally, b2m- concentrations could directly contribute to stronger PLC assem- deficient cells are able to recruit H chains with low efficiency (Fig. bly in IFN-g–treated M553 cells compared with untreated cells. 3) (42). These findings, taken together with the findings that b2m Somewhat paradoxically, the assembly-promoting functions of can be recruited to the PLC independently of H chains (Fig. 5), lead tapasin were much less significant in IFN-g–treated M553 cells to the postulate that ERp57- and/or CRT-dependent recruitment of than in untreated cells (Fig. 1B compared with 1C). However, it is b2m into the PLC can stabilize weaker binding between H chains to be noted that under the experimental conditions used in this and tapasin, and serve as a point of contact for stable H chain re- study, tapasin expression is not under endogenous IFN-g control. cruitment to the PLC. Thus, distinct interactions may be important Thus, tapasin’s activity may become limited (saturated) under for stabilizing H chain and b2m recruitment to the PLC. conditions in which MHC class I H chains or b2m are expressed in The findings described in this work also show that the conserved stoichiometric excess relative to tapasin. Further investigations are H chain glycan is not central to CRT recruitment into the PLC. The needed to understand whether, in tapasin-sufficient cells, IFN-g finding that the H chain glycan impacts the efficiency with which H treatments generally induce stoichiometric increments in TAP, chain is recruited into the PLC (Fig. 5) raises the possibility that tapasin, and MHC class I H chain components. Regardless of tapasin the H chain glycan is engaged by the glycan binding site of CRT expression, when TAP activity is high following IFN-g treatment, within the tapasin(C95)-dependent mode of CRT recruitment (Fig. there may be increased probability for tapasin-independent assem- 7B). However, if b2m is an important point of contact between H bly arising from increased peptide availability, resulting in a re- chain and the PLC, as suggested above, an alternative possibility duction in the stringency of quality control. Such a scenario may be is that the H chain glycan mutant has disrupted interactions with advantageous to the host in an infectious setting, to present reper- b2m within the PLC, also suggested by the finding of reduced toires of both optimal and suboptimal peptide–MHC complexes, efficiency of early heterodimer assembly in cells in the context of and increase the diversity of pathogenic peptides that can be pre- HA–HLA-A2(N86Q) compared with HA–HLA-A2 (Fig. 6E). sented. Further studies are needed to resolve these two possibilities. Thus, PLCs of 721.221 cells are able to incorporate b2m, CRT, and whereas structural studies of the tapasin–ERp57 heterodimers (36) ERp57 under conditions of H chain deficiency, and enhanced re- provide a valuable starting point for the construction of model for cruitment of H chains into the PLC of 721.221 cells does not result a mature PLC, whether CRT in fact interacts with the MHC class I in a parallel increase in the recruitment of b2m, CRT, or ERp57 glycan within the PLC as suggested (36) remains to be estab- (Fig. 5). The latter findings are supported by previous studies that lished, as well as the nature of interactions among tapasin–ERp57, have also shown binding of b2m and CRT to TAP in the absence b2m, and CRT. The Journal of Immunology 2319

Tapasin facilitated an early step in the assembly of heterodimers, function of MHC class I molecules in cells lacking the ER chaperone calreti- culin. Immunity 16: 99–109. and this step did not require tapasin-associated ERp57 or CRT (Fig. 11. Ireland, B. S., U. Brockmeier, C. M. Howe, T. Elliott, and D. B. Williams. 2008. 6A,6B). Tapasin stabilizes TAP and increases peptide transport by Lectin-deficient calreticulin retains full functionality as a chaperone for class I TAP (43, 44), and, by this mechanism, tapasin could impact an histocompatibility molecules. Mol. Biol. Cell 19: 2413–2423. 12. Del Cid, N., E. Jeffery, S. M. Rizvi, E. Stamper, L. R. Peters, W. C. Brown, early step in heterodimer assembly. However, we have previously C. Provoda, and M. Raghavan. 2010. Modes of calreticulin recruitment to the shown that the tapasin(C95A) mutant induces MHC class I as- major histocompatibility complex class I assembly pathway. J. Biol. Chem. 285: sembly via a mechanism independent of its effect on TAP stabi- 4520–4535. 13. Dick, T. P., N. Bangia, D. R. Peaper, and P. Cresswell. 2002. Disulfide bond lization and peptide transport (16). It is possible that tapasin isomerization and the assembly of MHC class I-peptide complexes. Immunity facilitates heterodimer formation simply by localizing H chains in 16: 87–98. the vicinity of the peptide source (TAP). Thus, there are at least 14. Peaper, D. R., and P. Cresswell. 2008. The redox activity of ERp57 is not es- sential for its functions in MHC class I peptide loading. Proc. Natl. Acad. Sci. two distinct levels of tapasin-assisted assembly, an early step that USA 105: 10477–10482. impacts heterodimer formation and a later step involving CRT and 15. Vigneron, N., D. R. Peaper, R. M. Leonhardt, and P. Cresswell. 2009. Functional significance of tapasin membrane association and disulfide linkage to ERp57 in ERp57, which could impact the generation of optimally assembled MHC class I presentation. Eur. J. Immunol. 39: 2371–2376. heterodimers. The conserved MHC class I glycan also facilitates 16. Rizvi, S. M., and M. Raghavan. 2010. Mechanisms of function of tapasin, an early step in the assembly of heterodimers (Fig. 6D–F). The a critical major histocompatibility complex class I assembly factor. Traffic 11: 332–347. MHC class I glycan most likely facilitates heterodimer formation 17. Belicha-Villanueva, A., S. McEvoy, K. Cycon, S. Ferrone, S. O. Gollnick, and by directly stabilizing H chain–b2m complexes, and/or by facili- N. Bangia. 2008. Differential contribution of TAP and tapasin to HLA class I tating CRT recruitment, which could enhance the formation or antigen expression. Immunology 124: 112–120. 18. Shimizu, Y., D. E. Geraghty, B. H. Koller, H. T. Orr, and R. DeMars. 1988. stability of heterodimers. Transfer and expression of three cloned human non-HLA-A,B,C class I major Downloaded from In summary, the studies described in this work provide evidence histocompatibility complex genes in mutant lymphoblastoid cells. Proc. Natl. for distinct and important roles for different glycans within the Acad. Sci. USA 85: 227–231. 19. Pear, W. S., G. P. Nolan, M. L. Scott, and D. Baltimore. 1993. Production of assembly complex. Furthermore, evidence is presented for the high-titer helper-free retroviruses by transient transfection. Proc. Natl. Acad. Sci. presence of intermediate complexes that are H chain depleted. USA 90: 8392–8396. These studies provide new insights into the sequence and nature of 20. Roeth, J. F., M. Williams, M. R. Kasper, T. M. Filzen, and K. L. Collins. 2004. HIV-1 Nef disrupts MHC-I trafficking by recruiting AP-1 to the MHC-I cyto- assembly events within the MHC class I assembly pathway. plasmic tail. J. Cell Biol. 167: 903–913. http://www.jimmunol.org/ 21. Androlewicz, M. J., B. Ortmann, P. M. van Endert, T. Spies, and P. Cresswell. 1994. Characteristics of peptide and major histocompatibility complex class I/ Acknowledgments beta 2-microglobulin binding to the transporters associated with antigen pro- cessing (TAP1 and TAP2). Proc. Natl. Acad. Sci. USA 91: 12716–12720. We thank Dr. Sanjeeva Joseph Wijeyesakere for docking studies of putative 22. Brodsky, F. M., W. F. Bodmer, and P. Parham. 1979. Characterization of CRT–tapasin–ERp57 complex structures that assisted with the develop- a monoclonal anti-beta 2-microglobulin antibody and its use in the genetic and ment of the model proposed in Fig. 7, Dr. Naveen Bangia for the M553 biochemical analysis of major histocompatibility antigens. Eur. J. Immunol. 9: cell line, Dr. Robert DeMars for 721.221 cells, and Dr. Kathleen Collins 536–545. 23. Ogino, T., X. Wang, S. Kato, N. Miyokawa, Y. Harabuchi, and S. Ferrone. 2003. for the HA–HLA-A2 retroviral construct and BOSC cells. We thank Dr. Endoplasmic reticulum chaperone-specific monoclonal antibodies for flow Clay Brown (University of Michigan High Throughput Protein Labora- cytometry and immunohistochemical staining. Tissue Antigens 62: 385–393. tory) and Dr. Nasir Salam for assistance with the retroviral HLA- 24. Stam, N. J., H. Spits, and H. L. Ploegh. 1986. Monoclonal antibodies raised by guest on September 28, 2021 B*3503 construct. We thank Dr. Matt Androlewicz, Dr. Peter Cresswell, against denatured HLA-B locus heavy chains permit biochemical characteriza- tion of certain HLA-C locus products. J. Immunol. 137: 2299–2306. Dr. Soldano Ferrone, Dr. Ted Hansen, Dr. Robert Tampe, and Dr. Jonathan 25. Simone, L. C., X. Wang, A. Tuli, and J. C. Solheim. 2010. Effect of a tapasin Yewdell for the Ab reagents that were critical for this work. We thank the mutant on the assembly of the mouse MHC class I molecule H2-K(d). Immunol. University of Michigan DNA sequencing core for sequencing analyses, Cell Biol. 88: 57–62. and the University of Michigan hybridoma core for Ab production. 26. Diedrich, G., N. Bangia, M. Pan, and P. Cresswell. 2001. A role for calnexin in the assembly of the MHC class I loading complex in the endoplasmic reticulum. J. Immunol. 166: 1703–1709. 27. Ortmann, B., M. J. Androlewicz, and P. Cresswell. 1994. MHC class I/beta 2- Disclosures microglobulin complexes associate with TAP transporters before peptide bind- The authors have no financial conflicts of interest. ing. Nature 368: 864–867. 28. Suh, W. K., M. F. Cohen-Doyle, K. Fruh, K. Wang, P. A. Peterson, and D. B. Williams. 1994. Interaction of MHC class I molecules with the transporter associated with . Science 264: 1322–1326. References 29. Peaper, D. R., P. A. Wearsch, and P. Cresswell. 2005. Tapasin and ERp57 form 1. Wearsch, P. A., and P. Cresswell. 2008. The quality control of MHC class I a stable disulfide-linked dimer within the MHC class I peptide-loading complex. peptide loading. Curr. Opin. Cell Biol. 20: 624–631. EMBO J. 24: 3613–3623. 2. Schrag, J. D., J. J. Bergeron, Y. Li, S. Borisova, M. Hahn, D. Y. Thomas, and 30. Lindquist, J. A., G. J. Ha¨mmerling, and J. Trowsdale. 2001. ER60/ERp57 forms M. Cygler. 2001. The structure of calnexin, an ER chaperone involved in quality disulfide-bonded intermediates with MHC class I heavy chain. FASEB J. 15: control of protein folding. Mol. Cell 8: 633–644. 1448–1450. 3. Frickel, E. M., R. Riek, I. Jelesarov, A. Helenius, K. Wuthrich, and L. Ellgaard. 31. Kienast, A., M. Preuss, M. Winkler, and T. P. Dick. 2007. Redox regulation of 2002. TROSY-NMR reveals interaction between ERp57 and the tip of the cal- peptide receptivity of major histocompatibility complex class I molecules by reticulin P-domain. Proc. Natl. Acad. Sci. USA 99: 1954–1959. ERp57 and tapasin. Nat. Immunol. 8: 864–872. 4. Helenius, A., and M. Aebi. 2004. Roles of N-linked glycans in the endoplasmic 32. Martayan, A., L. Sibilio, A. Setini, E. Lo Monaco, E. Tremante, D. Fruci, reticulum. Annu. Rev. Biochem. 73: 1019–1049. M. Colonna, and P. Giacomini. 2008. N-linked glycosylation selectively reg- 5. Sadasivan, B., P. J. Lehner, B. Ortmann, T. Spies, and P. Cresswell. 1996. Roles ulates the generic folding of HLA-Cw1. J. Biol. Chem. 283: 16469–16476. for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class 33. Zhang, Q., M. Tector, and R. D. Salter. 1995. Calnexin recognizes carbohydrate I molecules with TAP. Immunity 5: 103–114. and protein determinants of class I major histocompatibility complex molecules. 6. Harris, M. R., Y. Y. Yu, C. S. Kindle, T. H. Hansen, and J. C. Solheim. 1998. J. Biol. Chem. 270: 3944–3948. Calreticulin and calnexin interact with different protein and glycan determinants 34. Radcliffe, C. M., G. Diedrich, D. J. Harvey, R. A. Dwek, P. Cresswell, and during the assembly of MHC class I. J. Immunol. 160: 5404–5409. P. M. Rudd. 2002. Identification of specific glycoforms of major histocompati- 7. Harris, M. R., L. Lybarger, N. B. Myers, C. Hilbert, J. C. Solheim, T. H. Hansen, bility complex class I heavy chains suggests that class I peptide loading is an and Y. Y. L. Yu. 2001. Interactions of HLA-B27 with the peptide loading adaptation of the quality control pathway involving calreticulin and ERp57. J. complex as revealed by heavy chain mutations. Int. Immunol. 13: 1275–1282. Biol. Chem. 277: 46415–46423. 8. Sadasivan, B. K., A. Cariappa, G. L. Waneck, and P. Cresswell. 1995. Assembly, 35. Wearsch, P. A., C. A. Jakob, A. Vallin, R. A. Dwek, P. M. Rudd, and peptide loading, and transport of MHC class I molecules in a calnexin-negative P. Cresswell. 2004. Major histocompatibility complex class I molecules cell line. Cold Spring Harb. Symp. Quant. Biol. 60: 267–275. expressed with monoglucosylated N-linked glycans bind calreticulin in- 9. Scott, J. E., and J. R. Dawson. 1995. MHC class I expression and transport in dependently of their assembly status. J. Biol. Chem. 279: 25112–25121. a calnexin-deficient cell line. J. Immunol. 155: 143–148. 36. Dong, G., P. A. Wearsch, D. R. Peaper, P. Cresswell, and K. M. Reinisch. 2009. 10. Gao, B., R. Adhikari, M. Howarth, K. Nakamura, M. C. Gold, A. B. Hill, Insights into MHC class I peptide loading from the structure of the tapasin- R. Knee, M. Michalak, and T. Elliott. 2002. Assembly and antigen-presenting ERp57 thiol oxidoreductase heterodimer. Immunity 30: 21–32. 2320 TAPASIN’S GLYCAN FACILITATES MHC CLASS I ASSEMBLY

37. Carreno, B. M., J. C. Solheim, M. Harris, I. Stroynowski, J. M. Connolly, and purified class-I major histocompatibility (MHC) antigen. Scand. J. Immunol. 39: T. H. Hansen. 1995. TAP associates with a unique class I conformation, whereas 64–72. calnexin associates with multiple class I forms in mouse and man. J. Immunol. 41. Rizvi, S. M., and M. Raghavan. 2006. Direct peptide-regulatable interactions 155: 4726–4733. between MHC class I molecules and tapasin. Proc. Natl. Acad. Sci. USA 103: 38. Solheim, J. C., M. R. Harris, C. S. Kindle, and T. H. Hansen. 1997. Prominence 18220–18225. 42. Bangia, N., P. J. Lehner, E. A. Hughes, M. Surman, and P. Cresswell. 1999. The of beta 2-microglobulin, class I heavy chain conformation, and tapasin in the N-terminal region of tapasin is required to stabilize the MHC class I loading interactions of class I heavy chain with calreticulin and the transporter associated complex. Eur. J. Immunol. 29: 1858–1870. with antigen processing. J. Immunol. 158: 2236–2241. 43. Lehner, P. J., M. J. Surman, and P. Cresswell. 1998. Soluble tapasin restores 39. Shimizu, Y., and R. DeMars. 1989. Production of human cells expressing in- MHC class I expression and function in the tapasin-negative cell line .220. dividual transferred HLA-A,-B,-C genes using an HLA-A,-B,-C null human cell Immunity 8: 221–231. line. J. Immunol. 142: 3320–3328. 44. Garbi, N., N. Tiwari, F. Momburg, and G. J. Ha¨mmerling. 2003. A major role for 40. Pedersen, L. O., A. S. Hansen, A. C. Olsen, J. Gerwien, M. H. Nissen, and tapasin as a stabilizer of the TAP peptide transporter and consequences for MHC S. Buus. 1994. The interaction between beta 2-microglobulin (beta 2m) and class I expression. Eur. J. Immunol. 33: 264–273. Downloaded from http://www.jimmunol.org/ by guest on September 28, 2021