Structure of the Human MHC-I Peptide-Loading Complex

Letter doi:10.1038/nature24627

Structure of the human MHC-I peptide-loading complex Andreas Blees1*, Dovile Januliene2*, Tommy Hofmann3, Nicole Koller1, Carla Schmidt3, Simon Trowitzsch1, Arne Moeller2 & Robert Tampé1

The peptide-loading complex (PLC) is a transient, multisubunit to the cell surface for T-cell recognition3. The PLC can serve a large pool membrane complex in the endoplasmic reticulum that is essential of MHC-I allomorphs and, therefore, fulfills a central function in the for establishing a hierarchical immune response. The PLC differentiation and priming of T lymphocytes and in controlling viral coordinates peptide translocation into the endoplasmic reticulum infections and tumour development. The compositional heterogeneity with loading and editing of major histocompatibility complex class I and the intrinsic dynamic nature of this ER-resident membrane (MHC-I) molecules. After final proofreading in the PLC, stable complex have hindered detailed structural studies. The overall archi- peptide–MHC-I complexes are released to the cell surface to evoke tecture and the structural elements of the PLC that underlie assembly, a T-cell response against infected or malignant cells1,2. Sampling proofreading, and release of peptide–MHC-I complexes are largely of different MHC-I allomorphs requires the precise coordination unknown. of seven different subunits in a single macromolecular assembly, We isolated endogenous PLC from human Burkitt’s lymphoma including the transporter associated with antigen processing cells using the herpes viral inhibitor ICP47 fused to streptavidin- (TAP1 and TAP2, jointly referred to as TAP), the oxidoreductase binding peptide, ICP47–SBP, as bait (Fig. 1a–c). The affinity tag did ERp57, the MHC-I heterodimer, and the chaperones tapasin and not affect the inhibiting function of ICP47, as shown by a single-cell- calreticulin3,4. The molecular organization of and mechanistic based a ntigen translocation assay (Extended Data Fig. 1a–d). ICP47 events that take place in the PLC are unknown owing to the stabilizes TAP and arrests the PLC in a peptide-depleted state7, thus heterogeneous composition and intrinsically dynamic nature of the blocking the loading and release of MHC-I molecules. Incubation of complex. Here, we isolate human PLC from Burkitt’s lymphoma cell membranes with ICP47–SBP, followed by detergent solubilization cells using an engineered viral inhibitor as bait and determine and affinity purification, resulted in monodisperse PLC containing all the structure of native PLC by electron cryo-microscopy. Two subunits (Fig. 1b, c, Extended Data Fig. 1e and Extended Data Table 2), endoplasmic reticulum-resident editing modules composed of which was further subjected to GraFix8 (Extended Data Fig. 2). In tapasin, calreticulin, ERp57, and MHC-I are centred around TAP in addition to the three polymorphic MHC-I alleles (HLA-A, HLA-B a pseudo-symmetric orientation. A multivalent chaperone network and HLA-C), the non-classical alleles HLA-E, HLA-F, and HLA-G are within and across the editing modules establishes the proofreading incorporated into native PLC (Extended Data Fig. 1e and Extended function at two lateral binding platforms for MHC-I molecules. Data Table 2). Furthermore, we observed two tapasin isoforms (48 and The lectin-like domain of calreticulin senses the MHC-I glycan, 53 kDa) associated with the PLC, both of which were singly N-core whereas the P domain reaches over the MHC-I peptide-binding glycosylated (Fig. 1b and Extended Data Fig. 1e, f). pocket towards ERp57. This arrangement allows tapasin to facilitate We determined the structure of the fully assembled human endo peptide editing by clamping MHC-I. The translocation pathway of genous PLC by single-particle electron cryo-microscopy (cryo-EM) TAP opens out into a large endoplasmic reticulum lumenal cavity, (Fig. 1d, e, Extended Data Fig. 3 and Extended Data Table 1). The confined by the membrane entry points of tapasin and MHC-I. Two macro molecular complex measures 150 Å by 150 Å and has a total height lateral windows channel the antigenic peptides to MHC-I. Structures of 240 Å across the ER membrane. The anisotropy of the consensus of PLC captured at distinct assembly states provide mechanistic structure emphasizes the dynamic nature of the entire co mplex and insight into the recruitment and release of MHC-I. Our work defines limits the attainable resolution (Fig. 1d). The ER-lumenal domains, the molecular symbiosis of an ABC transporter and an endoplasmic arranged in two editing modules, are well-resolved; however, weaker reticulum chaperone network in MHC-I assembly and provides densities are observed for TAP. The translocation unit displays a high insight into the onset of the adaptive immune response. degree of flexibility and the TAP complex is rotated by approximately Nascent MHC-I heavy chains are chaperoned by the calnexin– 30° around the pseudo-C2 symmetry axis (Fig. 1d, e). Classification calreticulin system in the endoplasmic reticulum (ER). Together into multiple models did not improve the density of TAP or the with β2-microglobulin (β2m), MHC-I heavy chains assemble into overall resolution, but revealed distinct structural classes of the PLC heterodimers that act as receptors for antigenic peptides2. Empty (Extended Data Fig. 4). We therefore performed a focused classifica- MHC-I heterodimers are recruited by calreticulin and become part tion and alignment. In the case of TAP, the resolution could not be of the transient macromolecular PLC5,6, where the chaperone tapasin increased. However, the structure of the editing modules was signif- further stabilizes MHC-I molecules. As a disulfide-linked conjugate icantly improved to 7.2 Å (Fig. 2a and Extended Data Fig. 3). Each of with the thiol oxidoreductase ERp57, tapasin is crucial for maintaining the two fully assembled editing modules comprises tapasin, ERp57, the structural stability of the PLC and for facilitating optimal peptide calreticulin, and the MHC-I heterodimer9–12 (Fig. 2a, c, d). Our struc- loading2. After final quality control, in which MHC-I heterodimers ture defines the molecular organization of the ER chaperone network undergo peptide editing, stable peptide–MHC-I complexes are released acting on MHC-I clients. A focused classification on a single editing

1Institute of Biochemistry, Biocenter, Goethe University Frankfurt, Max-von-Laue Strasse 9, 60438 Frankfurt/Main, Germany. 2Department of Structural Biology, Max Planck Institute of Biophysics, Max-von-Laue Strasse 3, 60438 Frankfurt/Main, Germany. 3Interdisciplinary Research Center HALOmen, Martin Luther University Halle-Wittenberg, Kurt-Mothes-Strasse 3, 06120 Halle/Saale, Germany. *These authors contributed equally to this work.

a b c a b Editing module ER lumen 200 1,236 Calreticu li n m 1,048 660 481kDa 146kDa 66kDa kDa ERp57 PLC TAP1TAP2ERp57CalreticulinTapasinMHC-I hcβ 2 130- 150 MHC-I hc 720 PLC 95- 480 75- Tapasin 242 β2m 100 55- 146 90° 45- 66 34- kDa 26- 50 Cytosol 17- * Absorbance at 280 nm TAP1TAP2 10- kDa 0 1 2 3 4 5 ER membrane ER membrane d 150 Å Elution volume (ml) cd

90° 240 Å

180° 90°

Low High

Resolution e 30° 30° Tapasin ERp57 ER membrane Calreticulin Editing module I 40 Å

60 Å

90° 180° Figure 2 | Overall structural organization of the PLC editing modules. β2m MHC-I hc a, Cryo-EM density of the two pseudo-symmetric editing modules at 7.2 Å, 30° 90° shown in side and top views. b, Side view of the single editing module resolved at 5.8 Å. c, d, Ribbon diagram of the two editing modules in top (c) Translocation Editing module II unit and side (d) views. Tapasin is tilted towards the ER membrane, giving Micelle TAP rise to two lateral windows between the individual editing modules. The entrance points of the single transmembrane helices of tapasin (orange) Figure 1 | Composition and architecture of the human PLC. a, Schematic and MHC-I heavy chains (blue) are indicated. All subunits are coloured as overview of the individual components of the PLC showing TAP1, TAP2, in Fig. 1. Symmetry axes are indicated (dashed line). tapasin, MHC-I heavy chain (hc), β2m, calreticulin, and ERp57. b, Endogenous human PLC was affinity purified by ICP47–SBP. The presence of all subunits was verified by SDS–PAGE (Coomassie; immunoglobulin-like domain of the cis tapasin, potentially adding to asterisk, ICP47–SBP) followed by immunoblotting. c, Size-exclusion the stability of the PLC (Fig. 2c, d). Cross-linking mass spectrometry chromatography and blue-native PAGE of purified PLC reveals an apparent (XL-MS) confirmed this unexpected crossover arrangement of the molecular mass of approximately 650 kDa. d, The heat map of the PLC two tapasin–ERp57 modules (Fig. 2a, c, d, Extended Data Fig. 7 and consensus model, filtered to its local resolution (Relion), highlights the Extended Data Tables 2, 3), corroborating the importance of ERp57 for intrinsic dynamic nature of the macromolecular assembly. e, The general stabilizing the PLC, besides its crucial role for MHC-I recruitment13–16. architecture of the fully assembled PLC is displayed as a composite map, Our density map exhibits peptide-deficient MHC-I molecules, which where the threshold for each component was adjusted individually to allow are held in place by two major contact sites on tapasin (Fig. 3). The first visualization of all subunits (Extended Data Fig. 5). The pseudo-symmetry binding site is established by residues in the loop between β-strands axis is indicated (dashed lines). Data shown in b and c are representative of at least eight independent experiments. 17 and 18 of the C-terminal immunoglobulin-like domain of tapasin, which latch onto the CD8 recognition loop of the α3-domain of the MHC-I heavy chains (Fig. 3e). Notably, loop residues S336, H335, and module further improved the map resolution to 5.8 Å and guided H334 of tapasin are positioned close to the conserved MHC-I heavy model b uilding (Fig. 2b and Extended Data Figs 3–5). In the PLC chain residues T225 and Q226, suggesting that MHC-I recognition by editing module, tapasin, ERp57, and even MHC-I heterodimers form tapasin and the CD8 co-receptor co-evolved17 (Extended Data Fig. 6c). a rigid core, whereas calreticulin shows substantial flexibility (Extended Our model excludes the previously described interaction between Data Fig. 5f). residue R333 of tapasin and E222 of MHC-I heavy chains18 and could In the PLC, the two opposing tapasin molecules shape the central therefore explain why molecular dynamics simulations failed to derive scaffold. In our model we find E225 of the N-terminal imm unoglobulin- stable conformations19. At the second binding site, β-strands 6 and 7 like domain of one molecule and R60, located in the short helical motif and the α2-1 -helix of the MHC-I heavy chains are grasped by residues in of the seven-stranded N-terminal β-barrel of the second molecule, in the loops connecting β-s trands 1–2 and 9–10 of tapasin, positioning the salt-bridge distance (Fig. 2c and Extended Data Fig. 6a). These resi- critical residues R187 of tapasin and T134 of the MHC-I heavy chain dues are conserved among jawed vertebrates, but missing in avian PLC, close to one another20 (Fig. 3b). Our structure reveals a high degree of consistent with the existence of a single copy of tapasin in avian PLC plasticity of the N-terminal three-tiered β-sheet sandwich of tapasin and the absence of an N-terminal transmembrane domain (TMD0) in and shows that flexible clamping of MHC-I is essential for tapasin to avian TAP1 (Extended Data Fig. 6b). The ER-lumenal domains of the exert its proofreading function12,16,21. opposing tapasin molecules are tilted by 30° towards each other, posi- The assembly and maturation of MHC-I in the PLC depends strongly tioning the two membrane entry points of the transmembrane helices on calreticulin22. The globular lectin domain of calreticulin harbours 60 Å apart (Fig. 2d). The resulting domain twist excludes binding of a glycan-binding site, which senses monoglucose moieties of N-core calreticulin or calnexin to the glycan of the dimeric tapasin scaffold and glycosylated MHC-I before it associates with tapasin. The monogluco- explains why only mature tapasin can assemble into the PLC. sylated branch of the N-core glycan that emanates from N86 of MHC-I We find ERp57 in its typical U-shaped conformation complexed is pinned to the glycan-binding surface of calreticulin, whereas another to tapasin via the catalytically active a and a′ domains12. The unex- mannose branch is likely to nestle among residues at the edge of the pected orientation of tapasin enables the C-terminal extension of lectin β-sandwich (Fig. 3c). Consistent with its proposed calcium- the a′ domain of the trans ERp57 to interact with the C-terminal dependent lipid-sensing activity23, the C-terminal acidic tail of

a ER lumen ERAP b Calreticulin ERp57

G G Calreticulin P-loop MHC-I hc α2-1 T134T134 β2m Tapasin α1 R187R187 ERp57 b′ Assembly Release Traf c to Golgi Tapasin ADP N-term. Cytosol TAP1 TAP2 ATP α2-2 domain Figure 4 | Model for assembly and disassembly of the PLC. Peptide c receptive MHC-I heterodimers are recruited by calreticulin to the asymmetric PLC and form the fully assembled symmetric PLC, which is composed of two editing modules. TAP transports antigenic precursor Calreticulin N-core glycan peptides from the cytosol into a molecular basket formed by the editing modules. Two lateral windows allow peptides to diffuse into the ER lumen N86 to be edited by ERAP and subsequently loaded onto MHC-I molecules. After tapasin-catalysed proofreading, stable peptide–MHC-I complexes are released from the PLC and traffic via the Golgi to the cell surface. An asymmetric PLC resides in the membrane awaiting MHC-I heterodimers. Coloured as in Fig. 1.

reflect states of MHC-I release in agreement with the highly transient nature of the endogenous PLC. These intermediate states may arrange d into larger super-complexes, consistent with the alterations in the lateral Calreticulin e α3 β2m 30 acidic helix mobility of TAP , and therefore locally increase the peptide concentration in these nanoclusters to serve various MHC-I allomorphs with S336S336 different epitopes. The data also suggest that peptide loading induces T225 Q226 structural rearrangements in MHC-I in one editing module, resulting

Tapasin in the release of peptide–MHC-I complexes (Fig. 4). While tapasin C-term. domain C-term. resides within the PLC, ready to recruit empty MHC-I, peptide loading domain and disassembly take place on the second editing module. Assembly is Figure 3 | Molecular contacts within the editing module. The model much faster than disassembly, as it does not require any peptide proof- of the PLC editing module docked into the cryo-EM density (centre) reading, thus explaining the presence of at least one fully assembled highlights important interactions between the ER chaperone network editing module in all observed states (Fig. 4 and Extended Data Fig. 4). and the MHC-I client. a–e, Interactions of calreticulin and ERp57 (a), Our cryo-EM structure provides an integral framework for understanding MHC-I heavy chains and tapasin (b), calreticulin and MHC-I heavy chains the quality control of antigenic peptides in adaptive immunity and has (N-core glycan) (c), calreticulin (acidic helix) and tapasin (C-terminal unlocked the molecular details that underlie the onset of an adaptive domain) (d), and tapasin (C-terminal domain) and MHC-I heavy chains immune response. (α3) (e). Coloured as in Fig. 1. Online Content Methods, along with any additional Extended Data display items and calreticulin points towards the ER-lumenal membrane leaflet, where Source Data, are available in the online version of the paper; references unique to these sections appear only in the online paper. it is located close to the C-terminal immunoglobulin-like domain of tapasin (Fig. 3d). The arm domain of calreticulin forms a discontinu received 31 August; accepted 12 October 2017. ous elliptic belt around the editing module and connects MHC-I with Published online 6 November 2017. ERp57 (Fig. 2c). The central region of the arm domain hovers over the peptide-binding cleft of MHC-I. The tip of the arm domain rests 1. Neefjes, J., Jongsma, M. L., Paul, P. & Bakke, O. Towards a systems on helix H12 of the b′ domain of ERp57 (Fig. 3a), consistent with understanding of MHC class I and MHC class II antigen presentation. Nat. Rev. NMR studies and mutational analyses24,25. In vivo and in vitro studies Immunol. 11, 823–836 (2011). have highlighted the importance of the intricate network formed by 2. Blum, J. S., Wearsch, P. A. & Cresswell, P. Pathways of antigen processing. 16,26,27 Annu. Rev. Immunol. 31, 443–473 (2013). calreticulin, ERp57, and tapasin . Our cryo-EM structure empha- 3. Hulpke, S. & Tampé, R. The MHC I loading complex: a multitasking machinery sizes the central role of the C-terminal immunoglobulin-like domain of in adaptive immunity. Trends Biochem. Sci. 38, 412–420 (2013). tapasin, which acts as an essential multivalent interaction hub for the 4. Ortmann, B. et al. A critical role for tapasin in the assembly and function of multimeric MHC class I-TAP complexes. Science 277, 1306–1309 remaining PLC subunits (Fig. 2d and Extended Data Fig. 7). A cen tral (1997). cavity that connects the exit of the peptide translocation pathway of 5. Ortmann, B., Androlewicz, M. J. & Cresswell, P. MHC I/β2-microglobulin TAP with the ER lumen via two lateral windows (Fig. 2d) might serve complexes associate with TAP transporters before peptide binding. Nature 368, 864–867 (1994). as a molecular basket for transported peptides before they are edited by 6. Sadasivan, B., Lehner, P. J., Ortmann, B., Spies, T. & Cresswell, P. Roles for ER-resident aminopeptidases. This central cavity is probably targeted calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I by the human cytomegaloviral protein US6, which might dif fuse molecules with TAP. Immunity 5, 103–114 (1996). 7. Herbring, V., Bäucker, A., Trowitzsch, S. & Tampé, R. A dual inhibition laterally through the membrane into the cavity to plug the lumenal mechanism of herpesviral ICP47 arresting a conformationally thermostable 28 gate of TAP . TAP complex. Sci. Rep. 6, 36907 (2016). The electron microscopy dataset contained distinct, asymmetric 8. Kastner, B. et al. GraFix: sample preparation for single-particle electron cryomicroscopy. Nat. Methods 5, 53–55 (2008). subpopulations of PLC that lacked calreticulin, MHC-I, or both exclu- 9. Oldham, M. L. et al. A mechanism of viral immune evasion revealed by cryo-EM sively in one editing module, whereas the second module was always analysis of the TAP transporter. Nature 529, 537–540 (2016). fully assembled (Fig. 4 and Extended Data Fig. 4). This finding is 10. Zhang, S. et al. Structural basis of cross-allele presentation by HLA-A*0301 29 and HLA-A*1101 revealed by two HIV-derived peptide complexes. supported by immunoprecipitation experiments and antibody shift Mol. Immunol. 49, 395–401 (2011). assays (Extended Data Fig. 1g), both of which indicated tapasin:MHC-I 11. Kozlov, G. et al. Structural basis of cyclophilin B binding by the calnexin/ ratios of 2:1 and 2:2 in native PLC complexes. These subpopulations calreticulin P-domain. J. Biol. Chem. 285, 35551–35557 (2010).

12. Dong, G., Wearsch, P. A., Peaper, D. R., Cresswell, P. & Reinisch, K. M. Insights 28. Kyritsis, C. et al. Molecular mechanism and structural aspects of transporter into MHC class I peptide loading from the structure of the tapasin–ERp57 thiol associated with antigen processing inhibition by the cytomegalovirus protein oxidoreductase heterodimer. Immunity 30, 21–32 (2009). US6. J. Biol. Chem. 276, 48031–48039 (2001). 13. Wearsch, P. A. & Cresswell, P. Selective loading of high-affinity peptides onto 29. Panter, M. S., Jain, A., Leonhardt, R. M., Ha, T. & Cresswell, P. Dynamics of major major histocompatibility complex class I molecules by the tapasin–ERp57 histocompatibility complex class I association with the human peptide-loading heterodimer. Nat. Immunol. 8, 873–881 (2007). complex. J. Biol. Chem. 287, 31172–31184 (2012). 14. Wearsch, P. A., Peaper, D. R. & Cresswell, P. Essential glycan-dependent 30. Reits, E. A., Vos, J. C., Grommé, M. & Neefjes, J. The major substrates for TAP interactions optimize MHC class I peptide loading. Proc. Natl Acad. Sci. USA in vivo are derived from newly synthesized proteins. Nature 404, 774–778 108, 4950–4955 (2011). (2000). 15. Dick, T. P., Bangia, N., Peaper, D. R. & Cresswell, P. Disulfide bond isomerization and the assembly of MHC class I-peptide complexes. Immunity 16, 87–98 Supplementary Information is available in the online version of the paper. (2002). 16. Peaper, D. R., Wearsch, P. A. & Cresswell, P. Tapasin and ERp57 form a stable disulfide-linked dimer within the MHC class I peptide-loading complex. Acknowledgements This research was supported by the German EMBO J. 24, 3613–3623 (2005). Research Foundation (SFB 807 and GRK 1986 to R.T.). C.S. acknowledges 17. Wang, R., Natarajan, K. & Margulies, D. H. Structural basis of the CD8αβ/MHC funding from the Federal Ministry for Education and Research (BMBF, ZIK class I interaction: focused recognition orients CD8β to a T cell proximal program, 03Z22HN22), the European Regional Development Funds (EFRE, position. J. Immunol. 183, 2554–2564 (2009). ZS/2016/04/78115) and the MLU Halle-Wittenberg. We thank K. Zehl for 18. Simone, L. C., Georgesen, C. J., Simone, P. D., Wang, X. & Solheim, J. C. technical support; all members of the Institute of Biochemistry (Goethe Productive association between MHC class I and tapasin requires the tapasin University Frankfurt) for comments on the manuscript; and especially transmembrane/cytosolic region and the tapasin C-terminal Ig-like domain. W. Kühlbrandt, D. Mills, M. Wilkes, and the staff at the Department of Structural Mol. Immunol. 49, 628–639 (2012). Biology (MPI of Biophysics, Frankfurt/Main) for discussions, cryo-EM 19. Fisette, O., Wingbermühle, S., Tampé, R. & Schäfer, L. V. Molecular mechanism infrastructure and support. E. D’Imprima and R. Sanchez provided the code for of peptide editing in the tapasin-MHC I complex. Sci. Rep. 6, 19085 (2016). RecenterParticles. 20. Lewis, J. W., Neisig, A., Neefjes, J. & Elliott, T. Point mutations in the α2 domain of HLA-A2.1 define a functionally relevant interaction with TAP.Curr. Biol. 6, Author Contributions A.B. isolated the PLC and performed all biochemical 873–883 (1996). experiments. D.J. and A.M. carried out all EM imaging and single-particle 21. Thomas, C. & Tampé, R. Proofreading of peptide–MHC complexes through analyses. T.H. and C.S. performed the mass spectrometry experiments. N.K. dynamic multivalent interactions. Front. Immunol. 8, 65 (2017). implemented the single-cell-based transport analyses. S.T. and A.B. designed 22. Gao, B. et al. Assembly and antigen-presenting function of MHC class I the purification strategy for the PLC. S.T. and D.J. built the PLC model. A.B., molecules in cells lacking the ER chaperone calreticulin. Immunity 16, 99–109 D.J., S.T., A.M., and R.T. interpreted the data and wrote the manuscript with (2002). contributions from all authors. A.M., S.T., and R.T. conceived the study, designed 23. Wijeyesakere, S. J., Bedi, S. K., Huynh, D. & Raghavan, M. The C-terminal acidic the research, and planned the experiments. R.T. initiated and planned the region of calreticulin mediates phosphatidylserine binding and apoptotic cell project. phagocytosis. J. Immunol. 196, 3896–3909 (2016). 24. Zhang, Y. et al. ERp57 does not require interactions with calnexin and Author Information Reprints and permissions information is available at calreticulin to promote assembly of class I histocompatibility molecules, and it www.nature.com/reprints. The authors declare no competing financial enhances peptide loading independently of its redox activity. J. Biol. Chem. interests. Readers are welcome to comment on the online version of the paper. 284, 10160–10173 (2009). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional 25. Frickel, E. M. et al. TROSY-NMR reveals interaction between ERp57 and the tip claims in published maps and institutional affiliations. Correspondence and of the calreticulin P-domain. Proc. Natl Acad. Sci. USA 99, 1954–1959 (2002). requests for materials should be addressed to S.T. (trowitzsch@biochem. 26. Garbi, N. et al. Impaired immune responses and altered peptide repertoire in uni-frankfurt.de), A.M. ([email protected]) or R.T. ([email protected] tapasin-deficient mice.Nat. Immunol. 1, 234–238 (2000). frankfurt.de). 27. Garbi, N., Tanaka, S., Momburg, F. & Hämmerling, G. J. Impaired assembly of the major histocompatibility complex class I peptide-loading complex in mice Reviewer Information Nature thanks P. Cresswell, G. Skiniotis and the other deficient in the oxidoreductase ERp57.Nat. Immunol. 7, 93–102 (2006). anonymous reviewer(s) for their contribution to the peer review of this work.

Methods on a DionexUltiMate 3000 RSLCnano System (Thermo Scientific) coupled with No statistical methods were used to predetermine sample size. The experiments a Q Exactive plus hybrid mass spectrometer (Thermo Scientific) using mobile were not randomized. The investigators were not blinded to allocation during phase A, 0.1% FA and mobile phase B, 80% ACN/0.1% FA. Peptides were separated experiments and outcome assessment. on an Acclaim PepMap 100 C-18 column (Thermo Scientific) with a gradient of ICP47–SBP preparation. The ICP47-coding sequence was subcloned into 4–90% solvent B and eluted into the Q Exactive plus hybrid mass spectrometer the pETM-30 vector (European Molecular Biology Laboratory, EMBL) via the (Thermo Scientific), operating in data-dependent mode. MS conditions were restriction sites NcoI and BamHI to yield the pETM-30_ICP47 plasmid. An SBP set to spray voltage of 2.0 kV, capillary temperature of 275 °C and a normalized tag- co ding sequence was inserted via restriction sites BamHI and HindIII to collision energy of 30. Each cycle was initiated with an MS scan (350−1600 m/z), obtain the plasmid pETM-30_ICP47–SBP. Escherichia coli BL21(DE3)-RIL cells acquired at a resolution of 7 × 104 at m/z 400 and an automatic gain control (ACG) were transformed with pETM-30_ICP47–SBP and grown in lysogeny broth (LB) target of 3 × 106. The ten most intense ions were selected for HCD fragmentation 4 medium at 37 °C to an OD600 of 0.4. Expression of the ICP47–SBP construct was (ACG target, 1.75 × 10 ). Previously selected ions were dynamically excluded for induced with 0.2 mM isopropyl-β- d-thiogalactoside (IPTG) for 4 h at 20 °C. Cells 30 s. Single charged ions were excluded. Internal calibration of the Orbitrap was were lysed by sonication in 50 mM Tris/HCl pH 7.5, 500 mM NaCl, 1 mM dithio- performed using the lock mass option39. The LC–MS/MS data were processed and threitol (DTT) supplemented with protease inhibitor mix HP (Serva). To capture analysed by pLink software40. Raw data were converted to .mgf files using pXtract ICP47–SBP, the soluble fraction was incubated with glutathione sepharose 4 Fast (http://pfind.ict.ac.cn/software/pXtract/index.html). For protein identification, Flow resin (GE Healthcare) for 1 h at 4 °C. Beads were diluted 1:1 with 50 mM .mgf files were searched against the SwissProt database using Mascot search engine Tris/HCl pH 7.5 before TEV protease was added for 2 h at 25 °C. Subsequently, 2.5.1.1 (Matrix Science). For identification of cross-linked peptides, .mgf files were the NaCl concentration was readjusted to 500 mM and GST-free ICP47–SBP was searched against a reduced database containing sequences of PLC using pLink eluted. ICP47–SBP was further purified on a Superdex 200 10/300 GL column software. Potential cross-linked di-peptides were evaluated by their spectral quality. (GE Healthcare) pre-equilibrated with 25 mM Tris/HCl pH 7.5 and 500 mM NaCl. Cryo-EM sample preparation and data acquisition. PLC preparations were ICP47–SBP-containing fractions were pooled and concentrated to 1 mg ml−1. routinely screened by negative-stain EM before cryo-EM analyses. Onto freshly The inhibitory function of ICP47–SBP was tested with a peptide translocation glow-discharged carbon-coated G400-C3 grids (Gilder Grids), 3 μl of sample at a assay as described31. protein concentration of 0.1 mg ml−1 was applied, blotted from the side and stained PLC purification. PLC was obtained from 250 l Burkitt’s lymphoma cells (German using 2% uranyl formate solution as previously described41. EM micrographs were Collection of Microorganisms and Cell Culture, Leibniz Institute DSMZ) cultured recorded on a Tecnai-Spirit transmission electron microscope (Thermo Fisher, in RPMI 1640 medium (Gibco) supplemented with 10% fetal calf serum (FCS, former FEI), operating at 120 kV and equipped with Gatan 4 × 4K CCD camera. Capricorn), 25 mM HEPES pH 7.5 and 1% penicillin/streptomycin (Gibco) at Images were collected automatically using the Leginon software package42. −1 37 °C and 8% CO2 in a shaking incubator (Eppendorf). Cells were harvested in For cryo-EM grid preparation, 3 μl PLC sample (2 mg ml ) supplemented 20 mM HEPES pH 7.5, 500 mM NaCl supplemented with protease inhibitor mix with 1.5 mM fluorinated Fos-Choline-8 (Anatrace) was applied onto freshly glow- HP (Serva) and homogenized with a glass homogenizer. Membrane fraction was discharged CF 2/2 grids (Protochips) and plunge-frozen in liquid ethane using a collected by centrifugation at 100,000g for 30 min. Membranes were resuspended Vitrobot (Thermo Fisher, former FEI). Micrographs (Extended Data Fig. 3) were in 20 mM HEPES pH 7.5, 500 mM NaCl, 15% glycerol supplemented with protease recorded automatically (EPU, Thermo Fisher, former FEI), using a Titan Krios inhibitor mix HP (Serva) and incubated with ICP47–SBP for 1 h before extrac- microscope operated at 300 kV (Thermo Fisher, former FEI), equipped with a tion with 2% glyco-diosgenin (GDN) for 2 h. The soluble fraction was isolated by BioQuantum energy filter and a K2 camera (Gatan) at a nominal magnification of centrifugation at 100,000g for 30 min and Streptavidin High Capacity Agarose 130,000×, corresponding to a pixel size of 1.08 Å. Dose-fractionated movies were (Pierce) was added for 3 h. PLC was eluted in 20 mM HEPES pH 7.5, 250 mM NaCl, acquired at an electron flux of 8 e− per pixel per s over 8 s with 0.2 s exposures per 0.05% GDN, 2.5 mM biotin supplemented with protease inhibitor mix HP (Serva) frame (40 frames in total), corresponding to a total electron dose of ∼55 e− Å−2. and stabilized by chemical cross-linking using GraFix8. Purification was finalized Images were recorded in the defocus range from −1.5 to −4 μm. by size-exclusion chromatography on a KW404-4F (Shodex) pre-equilibrated with Cryo-EM image processing. Frame-based motion correction was performed 20 mM HEPES pH 7.5, 150 mM NaCl and 0.01% GDN. The peak corresponding using MotionCor243 with a dose filter of 1.37 e− per Å2 per frame. The contrast to PLC was collected for analysis by negative stain and cryo-EM. transfer function (CTF) was estimated from non-dose-weighted images, using Analysis of PLC. For SDS–PAGE and native PAGE analyses, precast NuPAGE Gctf44. Initially, particles were picked using the DoG-picker45 as implemented gradient gels (Novex) or NativePAGE gels (Invitrogen) were used. Gels were either in Appion46 and subjected to likelihood-based 2D classification in Relion47. The stained by Instant Blue (Expedeon) or directly transferred to a PVDF membrane best 2D classes were used to generate templates for automated particle selection (Bio-Rad). The subunit composition of the PLC was verified by using the anti- using Gautomatch (unpublished, developed by K. Zhang, MRC; http://www.mrc- bodies anti-TAP1 (clone 148.3)32, anti-TAP2 (clone 438.3)33, anti-tapasin (clone lmb.cam.ac.uk/kzhang/Gautomatch/). Particles belonging to those best 2D class 7F6)34, anti-HLA-A/B/C (clone W6/32, BioLegend, catalogue number 311402), averages were subjected to 3D classification in Relion, using a low-pass filtered anti-HLA-A (Abcam, catalogue number ab52922), anti-HLA-B (Abcam, c atalogue global average as a starting model. The best map already showed the overall number ab76795), anti-HLA-C (Abcam, catalogue number ab126722), anti- domain architecture of the PLC and was used as an initial model for subsequent 3D ERp57 (Abcam, catalogue number ab10287), anti-calreticulin (Abcam, catalogue classification (Extended Data Fig. 4). number ab2907), and anti-β2m (Novo Antibodies, catalogue number HPA006361). In total, 620,487 particles were extracted at a box size of 92 pixels with 4.3 Å per To determine the stoichiometry, PLC was subjected to an antibody shift assay pixel and directly subjected to a single round of multi-model 3D classification into as described previously35,36 using the antibodies anti-tapasin (PaSta1, gift from six classes (Relion) to eliminate particles of poor quality, resulting in a dataset of P. Cresswell)15, anti-HLA-A/B/C (clone W6/32, BioLegend, c atalogue number 185,076 particles (Extended Data Fig. 4). The selected particles were re-centred 311402), anti-HLA-B/C (clone HC10, Acris Antibodies, c atalogue number and re-extracted at full pixel size. This cleaned dataset was subjected to the next AM33035PU-N), and anti-TAP1 (1p2)37. The N-linked glycans of PLC compo- round of 3D classification (Relion) using four classes to separate the particles into nents were removed using endoglycosidase H (EndoH, NEB) or N-glycosidase F different conformations. Here, 29% of particles corresponded to the fully assem- (PNGase F, NEB) according to the manufacturer’s protocol. bled complex comprising two editing modules and the translocation unit, whereas Chemical cross-linking. In order to identify distance restraints for protein sub the rest of the data represented partially assembled PLC (Extended Data Fig. 4). units within the PLC, 10–30 μg of purified PLC was incubated with varying The local resolution map was determined in Relion 2.1 on the symmetric PLC, amounts of bis(sulfosuccinimidyl)suberate (BS3) for 1 h at 25 °C. Crosslinked which was individually refined using ‘gold standard’ refinement (Fig. 1). The proteins were either precipitated with ethanol followed by trypsin digestion in- same 3D classification procedure starting with 185,076 particles was repeated in solution or separated by SDS–PAGE before proteolysis and liquid chromatography Frealign48 and CryoSparc49, with similar results. with tandem mass spectrometry (LC–MS/MS) analysis. For in-gel digests, bands In a focused classification procedure, selected areas were extracted into smaller of interest were cut out from NuPAGE gradient gels and treated as described38. boxes using RecenterParticles (developed by E. D’Imprima and R. Sanchez, MPI for Prior to in-solution digestion, cross-linked proteins were ethanol-precipitated, Biophysics; code can be obtained upon request from [email protected]). washed, and dried. Dried pellets were dissolved in 25 mM ammonium bicarbonate Here, the area of interest is centred and extracted from the raw single particle and 1% RapiGest surfactant (Waters). Disulfide bridges were reduced with 50 mM stack into a smaller box, using shifts and Euler angles determined on the full DTT and cysteine residues were alkylated with 100 mM iodoacetamide. Finally, particle dataset. The re-centred particles were locally refined in Relion and taken proteins were digested with trypsin overnight at 37 °C in 0.1% RapiGest, dried, and into Frealign. The overall resolution was estimated to be 7.2 Å for the pseudo-C2 subjected to LC–MS/MS analysis. symmetrical editing modules and 5.8 Å for the singe editing module, using the LC–MS/MS analyses. Tryptic peptides were dissolved in 2% acetonitrile 0.143 cut-off cr iteria (Extended Data Fig. 3). The final maps were sharpened with (ACN)/0.1% formic acid (FA) and separated by nanoflow liquid chromatography EM-BFACTOR50 using B factor value of −250 Å2.

Model building. Structures of tapasin-ERp57 conjugate (PDB accession number 38. Shevchenko, A. et al. A strategy for identifying gel-separated proteins in 12 10 sequence databases by MS alone. Biochem. Soc. Trans. 24, 893–896 (1996). 3F8U) , HLA-A*0301 and β2m (PDB accession number 3RL1) were automatically docked into our final cryo-EM maps of the single and pseudo-C2 symmetric 39. Olsen, J. V. et al. Parts per million mass accuracy on an Orbitrap mass spectrometer via lock mass injection into a C-trap. Mol. Cell. Proteomics 4, editing modules using Chimera. Docking of the individual subunits was improved 2010–2021 (2005). 51 by rigid body fitting in Coot . Calreticulin was modelled based on the structures 40. Yang, B. et al. Identification of cross-linked peptides from complex samples. of its lectin domain in complex with a tetrasaccharide (PDB accession number Nat. Methods 9, 904–906 (2012). 3O0W)52, the tip of the P-domain (PDB accession number 1HHN)53, and the 41. Tao, H. et al. Engineered nanostructured β-sheet peptides protect membrane homologous region of the P-domain of calnexin (PDB accession number 1JHN)54. proteins. Nat. Methods 10, 759–761 (2013). 42. Suloway, C. et al. Automated molecular microscopy: the new Leginon system. A C-terminal poly-alanine helical extension in calreticulin was modelled de novo. J. Struct. Biol. 151, 41–60 (2005). A disaccharide of N-acetylglucosamine molecules was placed close to Asn86 of 43. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced motion HLA-A*0301. The model of the editing module was adjusted by flexible fitting for improved cryo-electron microscopy. Nat. Methods 14, 331–332 (2017). of individual subunits in FlexEM55, using two iterations of molecular dynamics 44. Zhang, K. Gctf: Real-time CTF determination and correction. J. Struct. Biol. 193, optimization with secondary structure elements as rigid bodies. Side-chains were 1–12 (2016). 45. Voss, N. R., Yoshioka, C. K., Radermacher, M., Potter, C. S. & Carragher, B. DoG removed from the final model and geometry minimization was performed in 56 Picker and TiltPicker: software tools to facilitate particle selection in single Phenix . particle electron microscopy. J. Struct. Biol. 166, 205–213 (2009). Sequence alignments. Alignments were generated using the ClustalOmega57 46. Lander, G. C. et al. Appion: an integrated, database-driven pipeline to facilitate server and plotted with Geneious. Protein sequences were retrieved from the EM image processing. J. Struct. Biol. 166, 95–102 (2009). UniProt database (www.uniprot.org). 47. Scheres, S. H. RELION: implementation of a Bayesian approach to cryo-EM Figure preparation. Figures were prepared using PyMOL58, Chimera59, FlowJo structure determination. J. Struct. Biol. 180, 519–530 (2012). 60 48. Lyumkis, D., Brilot, A. F., Theobald, D. L. & Grigorieff, N. Likelihood-based 10.2, and xVis . classification of cryo-EM images using FREALIGN.J. Struct. Biol. 183, 377–388 Data availability. The cryo-EM density maps and the corresponding model were (2013). deposited in the Electron Microscopy Data Bank under accession numbers EMD- 49. Punjani, A., Rubinstein, J. L., Fleet, D. J. & Brubaker, M. A. cryoSPARC: 3905, EMD-3904 and EMD-3906 and PDB ID 6ENY. The MS data were deposited algorithms for rapid unsupervised cryo-EM structure determination. to the ProteomeXchange Consortium (http://www.proteomexchange.org) via the Nat. Methods 14, 290–296 (2017). 50. Fernández, J. J., Luque, D., Castón, J. R. & Carrascosa, J. L. Sharpening high PRIDE partner repository with the dataset identifier PXD007928. All other data resolution information in single particle electron cryomicroscopy. J. Struct. Biol. are available from the corresponding authors upon reasonable request. 164, 170–175 (2008). 51. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of 31. Fischbach, H. et al. Ultrasensitive quantification of TAP-dependent antigen Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010). compartmentalization in scarce primary immune cell subsets. Nat. Commun. 52. Kozlov, G. et al. Structural basis of carbohydrate recognition by calreticulin. 6, 6199 (2015). J. Biol. Chem. 285, 38612–38620 (2010). 32. Meyer, T. H., van Endert, P. M., Uebel, S., Ehring, B. & Tampé, R. Functional 53. Ellgaard, L. et al. NMR structure of the calreticulin P-domain. Proc. Natl Acad. expression and purification of the ABC transporter complex associated with Sci. USA 98, 3133–3138 (2001). antigen processing (TAP) in insect cells. FEBS Lett. 351, 443–447 (1994). 54. Schrag, J. D. et al. The structure of calnexin, an ER chaperone involved in 33. van Endert, P. M. et al. A sequential model for peptide binding and transport by quality control of protein folding. Mol. Cell 8, 633–644 (2001). the transporters associated with antigen processing. Immunity 1, 491–500 55. Topf, M. et al. Protein structure fitting and refinement guided by cryo-EM (1994). density. Structure 16, 295–307 (2008). 34. Hulpke, S., Baldauf, C. & Tampé, R. Molecular architecture of the MHC I 56. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for peptide-loading complex: one tapasin molecule is essential and sufficient for macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. 66, antigen processing. FASEB J. 26, 5071–5080 (2012). 213–221 (2010). 35. Rufer, E., Leonhardt, R. M. & Knittler, M. R. Molecular architecture of the 57. Sievers, F. et al. Fast, scalable generation of high-quality protein multiple TAP-associated MHC class I peptide-loading complex. J. Immunol. 179, sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011). 5717–5727 (2007). 58. DeLano, W. L. The PyMOL Molecular Graphics System (DeLano Scientific, 2002). 36. Kim, J. Y., Kwak, P. B., Gebert, M., Duong, H. A. & Weitz, C. J. Purification and 59. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory analysis of PERIOD protein complexes of the mammalian circadian clock. research and analysis. J. Comput. Chem. 25, 1605–1612 (2004). Methods Enzymol. 551, 197–210 (2015). 60. Grimm, M., Zimniak, T., Kahraman, A. & Herzog, F. xVis: a web server for the 37. Lin, J. et al. A negative feedback modulator of antigen processing evolved from schematic visualization and interpretation of crosslink-derived spatial a frameshift in the cowpox virus genome. PLoS Pathog. 10, e1004554 (2014). restraints. Nucleic Acids Res. 43, W362–W369 (2015).

Extended Data Figure 1 | Functional arrest of endogenous PLC by (solid line, ATP; grey filled, ADP). Data are representative of two ICP47–SBP and biochemical analyses of purified human PLC. a, Design independent experiments. d, Summary of single-cell events indicated by of ICP47–SBP. Full-length ICP47 was C-terminally fused to a glycine/ mean fluorescence intensity (MFI), mean ± s.d. (n = 4). e, PLC subunits serine (G/S) linker and a streptavidin-binding peptide (SBP). b, Single- were analysed by SDS–PAGE and subsequent immunoblotting of the cell-based peptide translocation assay. Burkitt’s lymphoma cells were indicated proteins. f, PLC was treated by EndoH or PNGaseF and analysed semi-permeabilized with streptolysin O and incubated with fluorescent by SDS–PAGE (Coomassie). Band shifts are indicated by short lines. peptide RRYQNSTC(AF647)L (NST-AF647, 30 nM). Peptide translocation Asterisk, nonspecific band. g, PLC was incubated with increasing amounts was carried out at 37 °C for 15 min in the presence of ATP or ADP (10 mM of the respective antibodies and antibody shifts were visualized by native each) and with or without ICP47–SBP (5 μM). Peptide translocation was PAGE and immunoblotting. The asterisk indicates formation of higher stopped by EDTA (20 mM) and analysed by flow cytometry. c, Histograms oligomeric states of the PLC observed only with the anti-tapasin antibody. of the single-cell events (660-nm channel) demonstrate the inhibition Images shown are representative of five independent experiments. of ATP-dependent translocation of antigenic peptides by ICP47–SBP

Extended Data Figure 2 | Direct comparison of negatively stained PLC that the cross-linking procedure did not affect the organization of the particles. The individual particles (expanded views of the representative whole particle. The scale bar is 100 nm in the micrograph and 25 nm in the micrographs) and 2D class averages of native PLC without any treatment 2D averages; inset is magnified 4×. (a) and prepared by GraFix (b) display the same architecture, confirming

Extended Data Figure 3 | Cryo-EM analysis of the human PLC. the sample. Individual cylinder bars are proportional in height to the a, Representative micrograph and 2D class averages. Scale bars, 50 nm number of particles in each view. The most frequent views are coloured in the micrograph and 10 nm in 2D class averages. Various views of the red and the least common ones blue. c, Fourier shell correlation (FSC) individual particles are readily discernable in the raw image. In the 2D curve of unfiltered reconstructions from two independently refined half class averages, note the clear densities for the ER-lumenal domain and datasets of the full PLC, generated by post-processing in Relion, displays the blurred signal for TAP. b, Angular assignment of the final dataset. 9.9 Å resolution as judged from the 0.143 threshold. d, FSC curves for the The occurrence of multiple individual views, which are well spread over pseudo-C2 symmetric editing modules and the single module generated in the entire sphere, displays an almost random orientation of the PLC in Frealign show resolutions of 7.2 Å and 5.8 Å, respectively.

Extended Data Figure 4 | Processing work flow. The full dataset was assemblies. For high-resolution structure determination, a single editing directly submitted to multimodel classification in 3D. Particles from the module was extracted computationally from the consensus map and two best classes were merged into a single stack and refined together further refined individually (dotted lines indicate the use of the Frealign as a consensus map. This dataset was further subjected to multimodel software package, whereas the solid lines represent processing in Relion). refinement (left branch), which led to the identification of different PLC

Extended Data Figure 5 | Individual subunits of the PLC editing consistency of the domain-based docking. f, The multi-model 3D module. a–e, Individual segments of the single PLC editing module classification, focused on a single editing module, emphasizes the highlight the quality of the fit in the EM density. For each segment, the flexibility of calreticulin (yellow). Whereas tapasin, ERp57, and the MHC-I experimental map, the corresponding low-pass filtered version of the heterodimer display the same relative position in all classes, calreticulin atomic model, the actual atomic model, and its fit into the experimental shows a substantial shift in its position, indicating high flexibility. map are shown side by side in two different views to emphasize the

Extended Data Figure 6 | Two opposing tapasin molecules shape the rat, mouse, fish, and chicken. Conserved Arg and Glu residues forming central scaffold. a, E225 of the N-terminal immunoglobulin-like domain the salt bridges are indicated (asterisks). c, Multiple sequence alignment of one molecule and R60 located in the short helical motif of the seven- (MAFFT) of different human HLA-A/B/C allomorphs. Conserved Thr stranded N-terminal β barrel of the second molecule are in salt-bridge and Gln residues in the α3 domain are indicated (asterisks). Numbering is distance. b, Multiple sequence alignment (MAFFT; https://mafft.cbrc. according to UniProt, including signal sequences. jp/alignment/software/) of tapasin orthologues from human, bovine,

Extended Data Figure 7 | Cross-linking network. a, ICP47–SBP-purified single in-solution). Bands used for in-gel digestion are indicated. PLC was cross-linked with BS3 and applied to in-gel or in-solution A representative gel is shown. b, Intra- and inter-cross link network digestion before LC–MS/MS analysis (duplicates for in-gel digestion and (xVis60).

Extended Data Table 1 | EM data collection statistics

full PLC Pseudo-C2 Single editing module symmetric editing (EMD-3905) (EMD-3906) module (PDB ID 6ENY) (EMD-3904) Data collection and processing Magnification 130,000130,000 130,000 Voltage(kV)300 300300 Electron exposure (e–/Å2)555555 Defocus range (μm) -1.5 to -4 -1.5 to -4 -1.5 to -4 Pixel size (Å)1.077 1.0771.077 Symmetry imposedC1C1C1 Initial particle images (no.)620,487 620,487 620,487 (185,076)* (185,076)* (185,076)* Final particle images (no.)53,66622,915141,078 Map resolution (Å) 9.9 7.2 5.8 FSC threshold 0.143 0.143 0.143

*Number of automatically picked particles and (in brackets) number of particles after a single round of 3D classification to remove false positives.

Extended Data Table 2 | Overview of protein identification

in-solution digest in-gel digest sequence sequence Mass # peptide # # peptide # ProteinUniProtKB coverage coverage (Da) sequences spectra sequences spectra (%) (%) β2mP61769 137066 30 52 4 133 28 Calreticulin P27797 4811238218 72 20 100 38 ERp57P30101 5674745332 68 54 103565 ICP47P03170 9793 9* 140* 67* HLA-A03P04439 4081528271 70 15 458 44 HLA-B15P30464 4036317157 57 HLA-C03P04222 4083517153 56 HLA-C04P30504 4069625188 60 HLA-C16Q2996040727 18 18650

HLA-E P13747 40123148743 HLA-F P30511 39037 10 26 34 HLA-G P17693 382006 37 24 TapasinO1553347596 10 167279 239 22 TAP1 Q03518 8716325150 29 22 436 22 TAP2 Q03519-175616 30 1873425 419 30

*Obtained from MaxQuant search against a reduced database.

Extended Data Table 3 | Inter- and intramolecular cross-links in the PLC

Protein I Protein II Pos.1 Pos.2 Peptide Sequence 1 Peptide Sequence2 MHC-I β m 26 267 TPKIQVYSR PAGDGTFQKWAAVVVPSGEEQR 2 (A03/C03/C04) MHC-I β m 26 267 TPKIQVYSR TFQKWAAVVVPSGEEQR 2 (A03/B15/C03/C04) ICP47 TAP1 31 530 TYADVRDEINKR KAVGSSEKIFEYLDR ICP47 TAP1 31 537 TYADVRDEINKR AVGSSEKIFEYLDR MHC-I (A03) Calreticulin 92 209 NVKAQSQTDR IKDPDASKPEDWDER MHC-I (A03) Calreticulin 92 224 NVKAQSQTDR AKIDDPTDSKPEDWDKPEHIPDPDAK MHC-I (A03) Calreticulin 92 232 NVKAQSQTDR AKIDDPTDSKPEDWDKPEHIPDPDAK MHC-I (A03) Tapasin 92 104 NVKAQSQTDR FVPLPASAKWASGLTPAQNCPR Tapasin ERp57 104 305 FVPLPASAKWASGLTPAQNCPR KTFSHELSDFGLESTAGEIPVVAIR Tapasin ERp57 104 366 FVPLPASAKWASGLTPAQNCPR YLKSEPIPESNDGPVK Tapasin ERp57 342 494 SQKAEGQR VIQEEKPKKKKK Tapasin ERp57 342 366 SQKAEGQR YLKSEPIPESNDGPVK Tapasin Calreticulin 342 374 SQKAEGQR KKRKEEEEAEDK Tapasin Calreticulin 342 377 SQKAEGQR KEEEEAEDK

β2m β2m 26 111 TPKIQVYSR VNHVTLSQPKIVK Calreticulin Calreticulin 41 48 WIESKHKSDFGKFVLSSGK Calreticulin Calreticulin 55 360 FVLSSGKFYGDEEKDK DKQDEEQR Calreticulin Calreticulin 62 360 FYGDEEKDKGLQTSQDAR DKQDEEQR Calreticulin Calreticulin 153 209 GKNVLINK IKDPDASKPEDWDER Calreticulin Calreticulin 360 368 DKQDEEQR LKEEEEDK EEEEAEDKEDDEDKDEDEEDEEDKEEDE Calreticulin Calreticulin 375 385 LKEEEEDKKRK EEDVPGQAKDEL Calreticulin Calreticulin 385 401 RKEEEEAEDKEDDEDK DEDEEDEEDKEEDEEEDVPGQAKDEL ERp57 ERp57 129 147 TADGIVSHLKK KFISDK ERp57 ERp57 274 366 NAKGSNYWR YLKSEPIPESNDGPVK ERp57 ERp57 289 347 KFLDAGHKLNFAVASR DGKALER ERp57 ERp57 296 332 FLDAGHKLNFAVASR TAKGEK ERp57 ERp57 332 366 TAKGEK YLKSEPIPESNDGPVK ERp57 ERp57 335 366 GEKFVMQEEFSR YLKSEPIPESNDGPVK ERp57 ERp57 362 366 FLQDYFDGNLKR YLKSEPIPESNDGPVK ERp57 ERp57 425 494 LSKDPNIVIAK EATNPPVIQEEKPK ERp57 ERp57 460 466 GFPTIYFSPANKK KYEGGR MHC-I (A03) MHC-I (A03) 92 170 NVKAQSQTDR KWEAAHEAEQLR MHC-I (A03) MHC-I (A03) 145 170 QDAYDGKDYIALNEDLR KWEAAHEAEQLR Tapasin Tapasin 104 342 FVPLPASAKWASGLTPAQNCPR SQKAEGQR TAP1 TAP1 449 453 SFANEEGEAQKFR EKLQEIK