TAPBPR and tapasin binding to MHC class I is mutually exclusive

Abstract

The loading of peptide antigens onto MHC class I molecules is a highly controlled process in which the MHC class I dedicated chaperone tapasin is a key player. We recently identified a tapasin related molecule, TAPBPR, as an additional component in the MHC class I antigen presentation pathway. Here we show that the amino acid residues important for tapasin to interact with MHC class I are highly conserved on TAPBPR. We identify specific residues in the N-terminal and C-terminal domains of TAPBPR involved in associating with MHC class I. Furthermore, we demonstrate that residues on MHC class I crucial for its association with tapasin, such as T134, are also essential for its interaction with TAPBPR. Taken together, the data indicate that TAPBPR and tapasin bind in a similar orientation to the same face of MHC class I. In the absence of tapasin, the association of MHC class I with TAPBPR is increased. However, in the absence of TAPBPR, the interaction between MHC class I and tapasin does not increase. In light of our findings, previous data determining the function of tapasin in the MHC class I antigen processing and presentation pathway must be re-evaluated.

INTRODUCTION

The MHC class I antigen processing and presentation pathway ensures the efficient and stable presentation of peptide antigen at the cell surface for immunological monitoring, resulting in the elimination of viral infections and tumourigenic cells. Folding and peptide loading of MHC class I heavy chain/ β2m heterodimers in the endoplasmic reticulum (ER) is assisted by the MHC class I specific chaperone tapasin, as well as a number of generic chaperones such as calnexin and calreticulin. Tapasin simultaneously binds to peptide receptive MHC class I heterodimers and to TAP, localising MHC class I to a concentrated source of newly degraded antigenic peptides (1-5). There is also evidence that tapasin optimises or edits the peptides presented on MHC class I by facilitating exchange of sub-optimal peptides for higher affinity cargo (6-10).

The binding interface between tapasin and MHC class I is emerging. The N-terminal domain of tapasin is essential for association and peptide loading of MHC class I (11). By comparing the sequences of tapasin from different species and screening mutant tapasin molecules, Dong et al., identified a region of the N-terminal domain of tapasin that interacts with MHC class I. This cluster of residues on tapasin include E185, R187, Q189, H190, L191, K193, L250 and Q261 defined by the panel of tapasin TN mutants (TN3, TN4, TN5, TN6 and TN7)(12). This region of tapasin is predicted to bind a loop comprising residues 128-136 below the α2-1 helix of the MHC class I heterodimer (12-15). Residues in the predicted contact site in MHC class I, for example T134, are essential for incorporation of MHC class I into the PLC and efficient peptide loading (13-17). A second interaction point between tapasin and the MHC class I heavy chain involves residues 333-342 in the C-terminal Ig-like domain of tapasin (18-21) which are predicted to bind residues 222-229, situated in a beta strand in the α3 domain of the MHC class I heterodimer (20, 22-25).

A tapasin-related protein named TAPBPR is encoded out-with the MHC on chromosome 12 (26). Although the amino acid sequence of TAPBPR is only 22% identical to tapasin, TAPBPR also binds to MHC class I heavy chain/β2m heterodimers in the ER (27). However, in contrast to tapasin, human TAPBPR does not associate with TAP, ERp57 or calreticulin and is not essential for peptide loading onto MHC class I molecules. TAPBPR decreases the rate at which MHC class I molecules mature through the secretory pathway (27). Although it is not a component of the peptide loading complex, TAPBPR is necessary to maintain prolonged contact of MHC class I with the peptide loading complex, a role which might be important for peptide selection by MHC class I molecules.

Given our recent identification of TAPBPR as a second MHC class I specific component in the antigen presentation pathway, our aim was to investigate how TAPBPR interacts with MHC class I.

Materials & Methods

Homology modelling of TAPBPR

A model for the structure of TAPBPR was generated using the Fold and Function Assignment System (FFAS) based on a profile-profile matching algorithm (28, 29). Tapasin was identified as the closest structural homologue available in the Protein Data Bank and its structure (PDB-ID 3F8U, (12)) was used as a template to generate a model for TAPBPR using the program SCWRL4 to predict and optimise side-chain conformations (30). The model was built for only the luminal domains of TAPBPR. Figures were generated with PyMOL (PyMOL Molecular Graphics System, Version 1.3 Schrödinger, LLC).

Cell culture

HEK-293T, HeLa and KBM-7 cells were maintained in DMEM, RMPI 1640 and IMDM media (GIBCO) respectively supplemented with 10% fetal calf serum, 100 U/mL penicillin and 100 μg/mL streptomycin at 37°C and 5% CO₂. To induce expression of endogenous TAPBPR cells were treated with 50 U/ml of IFN-γ (Roche) at 37°C for 48 hours.

Constructs

PK1-A2 encoding an N-terminally GFP tagged HLA-A2 molecule has been described previously (31). Full length untagged TAPBPR and untagged HLA-A2 were cloned into pCR-Blunt II-TOPO. Site-directed mutagenesis was performed to mutate specific residues in TAPBPR or HLA-A2 using QuikChange site-directed mutagenesis (Stratagene) along with the primers outlined in Table I & Table II. TAPBPR and its variants were subsequently cloned into the lentiviral vector pHRSIN-C56W-UbEM producing TAPBPR under the SFFV promoter and the GFP derivative protein emerald under an ubiquitin promoter. GFP-A2 or untagged HLA-A2 and their variants were cloned into the lentiviral vector pHRSINcPPT-SGW. For RNA interference, lentiviral shRNA plasmid V2LHS_135531 on the pGIPZ backbone was purchased from Open Biosystems. The lentiviral plasmids were transfected into HEK-293T cells using TransIT-293 (Mirus) along with pCMVR8.91 packaging vector and pMD-G envelope vector. These supernatants were used to produce stable transduced HeLa, KBM-7 and 721.221. Cell sorting on a BD Influx cell sorter was performed to generate equally expressing transduced HeLa cell lines on the basis of their GFP expression levels. TAPBPR shRNA depleted transduced cell lines were selected with puromycin. Tapasin deficient KBM-7 cells were produced as described previously (32).

Table I. Panel of TAPBPR mutant molecules.

Name	Residues mutated	Primers used for site directed mutagenesis	Predicted TAPBPR domain
WT
TN4	K22E, D23R	5′GGTCCTAGACTGTTTCCTGGTGGAGCGCGGTGCGCACCGTGGAGCTCTCG3′ 5′CGAGAGCTCCACGGTGCGCACCGCGCTCCACCAGGAAACAGTCTAGGACC3′	N-terminal
TN5	I261K	5′CTATACAGGACGAGGGGACCTACAAGTGCCAGATCACCACCTCTCTG3′ 5′CAGAGAGGTGGTGATCTGGCACTTGTAGGTCCCCTCG TCCTGTATAG3′
TN6	E205K, R207E, Q209S, Q272S	5′GGCTTGGACCTCATCAGTGTGAAGTGGGAACTGTCGCACAAGGGCAGGGGTCAGTTGG3′ 5′CCAACTGACCCCTGCCCTTGTGCGACAGTTCCCACTTCACACTGATGAGGTCCAAGCC3′ 5′CACCACCTCTCTGTACCGAGCTTCGCAGATCATCCAGCTCAACATC3′ 5′GATGTTGAGCTGGATGATCTGCGAAGCTCGGTACAGAGAGGTGGTG3′
TN7	H210S, K211E, R213E	5′CAGTGTGGAGTGGCGACTGCAGTCCGAGGGCGAGGGTCAG TTGGTGTACAGCTGG-3′ 5′CCAGCTGTACACCAACTGACCCTCGCCCTCGGACTGCAGTCGCCACTCCACACTG3′
TC2	R335D	5′GGTGCCTCCTTCTCCAGCCTCGACCAAAGCGTGGCAGGCACCTAC3′ 5′GTAGGTGCCTGCCACGCTTTGGTCGAGGCTGGAGAAGGAGGCACC3′	C-terminal
TC3	Q336D, S337D	5′GGTGCCTCCTTCTCCAGCCTCAGGGATGACGTGGCAGGCACCTACAGCATCTCC3′ 5′GGAGATGCTGTAGGTGCCTGCCACGTCATCCCTGAGGCTGGAGAAGGAGGCACC3′
TC4	A339D	5′CTCCAGCCTCAGGCAAAGCGTGGACGGCACCTACAGCATCTCCTCCTC3′ 5′GAGGAGGAGA TGCTGTAGGT GCCGTCCACG CTTTGCCTGA GGCTGGAG3′
TC5	S343R, I344T	5′CAGGCAAAGCGTGGCAGGCACCTACCGCACCTCCTCCTCTCTCACCGCAGAACC3′ 5′GGTTCTGCGGTGAGAGAGGAGGAGGTGCGGTAGGTGCCTGCCACGCTTTGCCTG3′

Table II. Panel of HLA-A2 mutant molecules.

Name	Residues mutated	Primers used for site directed mutagenesis	Predicted TAPBPR domain
WT	-		-
D122	D122R	5′CCAGTACGCCTACGACGGCAAGCGTTACATCGCCCTGAAAGAGGACC3′ 5′GGTCCTCTTTCAGGGCGATGTAACGCTTGCCGTCGTAGGCGTACTGG3′	Loop below α2-1
E128	E128R	5′CAAGGATTACATCGCCCTGAAAAGGGACCTGCGCTCTTGGACCGCGG3′ 5′CCGCGGTCCAAGAGCGCAGGTCCCTTTTCAGGGCGATGTAATCCTTG3′
W133	W133T	5′AAAGAGGACCTGCGCTCTACGACCGCGGCGGACATGGCA3′ 5′TGCCATGTCCGCCGCGGTCGTAGAGCGCAGGTCCTCTTT3′
T134	T134K	5′GAAAGAGGACCTGCGCTCTTGGAAAGCGGCGGACATGGCAGCTCAGAC3′ 5′GTCTGAGCTGCCATGTCCGCCGCTTTCCAAGAGCGCAGGTCCTCTTTC3′
K144	K144D	5′GACATGGCAGCTCAGACCACCGACCACAAGTGGGAGGCGGCCCATG3′ 5′CATGGGCCGCCTCCCACTTGTGGTCGGTGGTCTGAGCTGCCATGTC3′	α2-1 helix
K146	K146D	5′GCAGCTCAGACCACCAAGCACGACTGGGAGGCGGCCCATGTGGCG3′ 5′CGCCACATGGGCCGCCTCCCAGTCGTGCTTGGTGGTCTGAGCTGC3′
α3-1	E222K, D223R, D227R, E229K	5′GGGATGGGGAGGACCAGACCCAGAGGACGAAGCTCGTGGAGACCAGGCCTGCAGG3′ 5′CCTGCAGGCCTGGTCTCCACGAGCTTCGTCCTCTGGGTCTGGTCCTCCCCATCCC3′ 5′CACACTGACCTGGCAGCGGGATGGGAAGAGGCAGACCCAGGACACGGAGCTCGTG3′ 5′CACGAGCTCCGTGTCCTGGGTCTGCCTCTTCCCATCCCGCTGCCAGGTCAGTGTG3′	α3
α3-2	D227R, E229K	5′GGGATGGGGAGGACCAGACCCAGAGGACGAAGCTCGTGGAGACCAGGCCTGCAGG3′ 5′CCTGCAGGCCTGGTCTCCACGAGCTTCGTCCTCTGGGTCTGGTCCTCCCCATCCC3′

Antibodies

The following antibodies were used: rabbit anti-TAPBPR R014 raised against amino acids 22-406 of human TAPBPR (27), rabbit anti-TAPBPR R021 raised against the cytoplasmic tail of human TAPBPR, a conformational specific monoclonal antibody raised against amino acid 22-406 of human TAPBPR named PeTe4 (27), a mouse anti-TAPBPR raised against amino acids 23-122 of full-length human TAPBPR (ab57411, Abcam), the tapasin specific mAb Pasta1 & Rgp48N (both kind gifts from Peter Cresswell, Yale University School of Medicine), rabbit anti-GFP (Ab290, Abcam), mouse anti-GFP (Roche), rabbit anti-calnexin (Enzo life sciences), mAb HC10 which recognises HLA-A, -B and -C containing a PxxWDR motif at amino acid 57-62 in the α1 domain (33, 34), mAb specific for conformational HLA-A2 and -A68 (One Lambda), HLA-A2 specific mAb BB7.2 (35). IgG1 and IgG2a isotype control antibodies were also used (Dako). Species-specific fluorescent and HRP conjugated secondary antibodies were from Molecular Probes and Dako, respectively.

Radiolabeling & pulse chase

Cells were starved in methionine and cysteine free RPMI for 30 min at 37°C, then labelled with [³⁵S] methionine/cysteine Promix (Amersham Pharmacia) at 37°C for the indicated time. In TAPBPR half-life experiments, samples were chased at 37°C for 0-72 hours in medium containing excess methionine/cysteine.

Immunoprecipitation & Gel electrophoresis

Cells were lysed in either 1% Triton X-100 (Sigma) or 1% digitonin (Calbiochem) in Tris-buffered saline (TBS) (20 mM Tris-HCl, 150 mM NaCl, 5 mM MgCl2, 1 mM EDTA) containing 10 mM N-Ethylmaleimide (Sigma), 1 mM PMSF and protease inhibitors (Roche) at 4°C. Post-nuclear supernatants were subsequently made by centrifugation. Samples were precleared on IgG- and protein A-Sepharose beads (GE Healthcare). Immunoprecipitations were performed using the indicated antibody and protein A-sepharose. Beads were washed thoroughly in 0.1% detergent in TBS. For TAPBPR half life, primary TAPBPR immunoprecipitates were eluted in 1% SDS-TBS with 10 mM DTT, followed by quenching in a 10-fold dilution of 1% Triton X-100 in TBS with 20 mM iodoactetamide. TAPBPR was re-immunoprecipitated using mouse anti-TAPBPR (Abcam). All samples were heated at 80°C for 10 min in reducing sample buffer prior to separation by SDS-PAGE. For immunoblotting, proteins were transferred onto an Immobolin transfer membrane (Millipore) then blotted with the indicated antibodies. For radiolabelled samples, gels were fixed and dried, then images were obtained using a phosphor screen (Perkin-Elmer) or on film. PhosphorImager analysis was performed using Typhoon Trio variable mode imager (GE Healthcare) together with ImageQuantTL software. Graphs were generated using GraphPad Prism 6.

Flow Cytometry

Following trypsinisation, cells were incubated at 37°C in RMPI supplemented with 10% HIFCS to allow membrane recovery from trypsinisation. Cells were stained at 4°C with MHC I specific antibodies anti-A68 or BB7.2. Isotype control antibodies were used as negative controls. Antibodies were subsequently detected with species-specific Alexa Fluor 647 secondary antibodies (Molecular probes). Cells were analysed on a BD Bioscience FACS Calibur 4-colour analyser.

RESULTS

MHC class I binding sites defined on tapasin are conserved on TAPBPR

A series of tapasin mutants (TN3, TN4, TN5, TN6 and TN7) identified an MHC class I binding site on tapasin (12). To determine if a similar MHC class I binding site is conserved on TAPBPR we compared amino acid sequence alignments of human tapasin with human TAPBPR (Fig 1A). Although tapasin and TAPBPR are only ~22% identical, a number of the key residues on tapasin critical for binding to MHC class I are well conserved on TAPBPR (Fig 1A). MHC I binding residues defined by the tapasin TN6 mutant (E185, R187, Q189, Q261), are completely conserved on TAPBPR, while those identified by the tapasin TN7 mutant (H190, L191, K193) are relatively conserved, sharing the histidine and a charged residue. A leucine at residue 250, characterised by the tapasin TN5 mutant, also contributes to MHC class I binding. This is an isoleucine in TAPBPR, and is localised in a region with high sequence identity with the two molecules sharing the preceding residues EGTY.

Figure 1. MHC class I binding sites defined on tapasin are conserved on TAPBPR.

A) Amino acid sequence of human TAPBPR (NP_060479.3) and human tapasin (NP_003181.3) were aligned using Clustal W. Residues constituting a binding site for MHC class I on the N-terminal domain of tapasin defined by Dong et al., (12) are highlighted: TN4 (yellow), TN5 (green), TN6 (red), TN7 (dark blue). Residues 333-342 of tapasin form a binding site for MHC class I on the C-terminal domain. The corresponding region on TAPBPR, further subdivided into TC patches, is highlighted: TC2 (light blue), TC3 (orange), TC4 (lime green), TC5 (purple). B &C) A model of human TAPBPR was generated based on the human tapasin crystal structure as a template (PDB ID: 3F8U). B) Tapasin is shown without the ERp57 chaperone present in the original pdb file. Localisation of the residues contributing to MHC class I binding are highlighted on the structure of tapasin using colouring as in A. C) The equivalent regions are also highlighted on the TAPBPR model indicating their structural conservation.

An additional MHC class I interaction site has been suggested on the membrane proximal domain of tapasin comprising of residues 333-342 (18-21). The amino acid alignment suggests that while R333 and S341 are conserved between the two proteins there is considerable variation between the two molecules in this region. However, it is possible that Q334 and S335 in TAPBPR form similar hydrogen bonds with MHC class I and could therefore contribute to the binding.

We used the crystal structure of human tapasin obtained by Dong et al., (12) to create a homology model for TAPBPR. The residues that have been shown to be important for MHC class I binding by tapasin lie on a well conserved and highly ordered face of tapasin. Our homology model indicates that this face is structurally conserved in TAPBPR allowing us to predict that MHC class I will bind TAPBPR in the same manner as it does to tapasin (Fig 1B & C). Many of these residues are highly conserved in TAPBPR across different species, supporting the possibility that they comprise a functionally important region of TAPBPR (supplementary Fig 1).

Residues in the N-terminal and C-terminal domains of TAPBPR are involved in the association with MHC class I

To determine whether the conserved residues described above form a MHC class I binding site on TAPBPR, a panel of eight mutant TAPBPR molecules was made in which charge alterations were applied following the same concept as Dong and colleagues (see Table I) (12). This panel of TAPBPR molecules was cloned into a lentiviral expression vector with a bicistronic GFP reporter and transduced into HeLa cells. Since endogenous TAPBPR expression is undetectable in HeLa (27) this cell line provides an ideal system for testing the effect of alterations to TAPBPR on its ability to bind to MHC class I. To produce stable HeLa cells expressing comparable TAPBPR levels, cell lines were sorted based on emerald expression encoded bistronically from the TAPBPR protein. Western blotting for transduced TAPBPR revealed that all TAPBPR mutant molecules were stably expressed (Fig 2A).

Figure 2. Residues in the N-terminal and C-terminal domains of TAPBPR are involved in the association with MHC class I.

A) TAPBPR was isolated by immunoprecipitation (using polyclonal antisera R014) from a panel of HeLa cells stably transduced with either WT-TAPBPR, TAPBPR with mutations in the N-terminal domain (TN4-7) or TAPBPR with mutations in the C-terminal domain (TC2-5) lysed in 1% Triton X-100 TBS. As a negative control non-transfected HeLa cells were included (−). Western blot analysis was performed for TAPBPR (using mouse anti-TAPBPR), the MHC class I heavy chain (using HC10) or calnexin on lysates and TAPBPR immunoprecipitates as indicated. B) The half-life of TAPBPR mutants which exhibited reduced binding to MHC class I was compared to WT TAPBPR expressed in HeLa via pulse chase analysis. HeLa cells were labelled for 60 min with [³⁵S] methionine/cysteine then chased for 0-72 H. After solubilisation in 1% Triton X-100 TBS, TAPBPR was isolated by immunoprecipitation using mAb PeTe4 from precleared post-nuclear supernatants. Following elution and denaturation, TAPBPR was re-immunoprecipitated with TAPBPR mAb raised to aa 23-122 of TAPBPR. Densitometry on the TAPBPR band was performed and the amount of TAPBPR remaining at each time point was plotted as a percentage of the signal intensity at the 0 H time point. Symbols indicate specific data points for each of the TAPBPR variants as indicated on figure label. Lines represents the best fit through the data points for each variant: WT (black solid line), TN5, TN7, and TC3 (all grey solid lines), TN6 (grey dotted line), and TC2 (black dotted line). C & D) Cytofluorometric analysis of HLA-A68 on non-transduced HeLa (black line histogram) and HeLa transduced with WT-TAPBPR (grey filled histogram). Staining with an isotype control on HeLa is included (grey dotted histogram). HLA-A68 on HeLa cell tranduced with two of the mutant TAPBPR molecules C) TN5 and D) TC2 is shown with the black dotted line. E) Bar graph showing mean fluorescence intensity for surface HLA-A68 expression on the full panel of mutant TAPBPR molecules from three independent experiments as performed in C & D. Error bars show −/+ SEM. All experiments were repeated independently at least three times.

To analyse the interaction between the TAPBPR molecules and MHC class I, TAPBPR was immunoprecipitated from cell lysates using a TAPBPR polyclonal antiserum, followed by western blotting for the MHC class I heavy chain using HC10. A strong association between TAPBPR and the MHC class I heavy chain was observed with WT TAPBPR (Fig 2A). The TN4 mutant also bound strongly to MHC class I (Fig 2A). However no association was observed between TAPBPR and MHC class I using the TN5 or TN6 TAPBPR mutants (Fig 2A). The TN7 mutant exhibited a reduced capacity to interact with MHC class I. Therefore like tapasin, a conserved patch on the N-terminal domain of TAPBPR constitutes a major binding domain for MHC class I with residues I261 (TN5), E205, R207, Q209, Q272 (TN6) and H210, K211, R213 (TN7) on TAPBPR contributing to the interaction.

The C-terminal domain of TAPBPR also appeared to contribute to the interaction with MHC class I. No association was observed between TAPBPR and MHC class I using the TC2 or TC3 mutant TAPBPR molecules in which residues R335 or Q336/S337 were altered (Fig 2A). In contrast, MHC class I could bind to TAPBPR in which nearby residues A339 (TC4) or S343/I344 (TC5) were mutated (Fig 2A).

The half life of the five TAPBPR mutants which showed a reduction in binding to MHC class I was determined by pulse chase analysis in order to determine if the loss of association was a consequence of TAPBPR instability. Pulse chase analysis revealed that the TAPBPR mutants TN5, TN7, and TC3 had a similar half life to WT TAPBPR when expressed in HeLa cells (Fig 2B). A difference was observed in radiolabelling efficiency for TN6. However the labelled TAPBPR TN6 protein also appeared stable over-time. The only mutant that appeared to be less stable than WT TAPBPR was the TC2 mutant (Fig 2B). To further examine protein stability of the TAPBPR mutants we determined the melting temperature for purified WT TAPBPR and mutant TAPBPR molecules by differential scanning fluorimetry. The melting temperatures for all TAPBPR variants were found to be between 50°C and 53°C suggesting that all variants are stable at 37°C. The melting temperature for TC2 (50.4°C) and TC3 (50.15°C) was slightly lower than WT TAPBPR (52.56°C) (supplementary Fig 2). Taken together, our data suggests that the loss of interaction of the TAPBPR mutant with MHC class I is not a direct consequence of TAPBPR protein instability.

Over-expression of TAPBPR in HeLa cells results in down-regulation of MHC class I expression on the cell surface (27). This phenotypic effect was used as an assay to determine the effect of the TAPBPR mutants on MHC class I surface expression. WT TAPBPR resulted in a significant down-regulation of HLA-A68 expression in HeLa cells (Fig 2C-E). However, the TAPBPR mutants that cannot bind to MHC class I (TN5, TN6, TC2, TC3) did not down-regulate HLA-A68 expression (Fig 2C-E). The other TAPBPR mutants that bind to MHC class I (TN4, TN7, TC4 and TC5) down-regulated MHC class I surface expression (Fig 2E). However, they were not quite as efficient as WT-TAPBPR (Fig 2E). Surprisingly the TN4 mutant, which appeared to show a strong association with MHC class I as measured by coimmunoprecipitation with TAPBPR (Fig 2A), only resulted in a 50% reduction of HLA-A68 surface expression compared to a 90% reduction observed with WT-TAPBPR. Therefore this mutant may have a more complex effect on TAPBPR function than originally suggested.

Mutation of T134 on MHC class I inhibits TAPBPR binding

The central region of the N-terminal domain of tapasin is predicted to bind to residues 128-136 of the MHC class I heavy chain. Mutation of these residues on MHC class I inhibits tapasin binding (12-15, 20). Given the importance of the central region of the N-terminal domain of TAPBPR in associating with MHC class I, we asked whether amino acids in the loop under the α2-1 domain of MHC class I were also required to associate with TAPBPR. Using an N-terminally tagged GFP HLA-A2 construct, we created a panel of mutant HLA-A2 molecules (Table II) which were expressed in HeLa cells. The GFP tag was used to differentiate between the endogenous HLA alleles in HeLa and the transduced HLA-A2 which runs at ~70 kDa. We previously reported that this N-terminal GFP tag did not significantly alter export rates or surface expression of HLA-A2 compared to untagged HLA-A2 (31). To induce TAPBPR expression, the HeLa cells were treated with IFN-γ (27). In TAPBPR immunoprecipitation experiments, an association was observed between TAPBPR and the GFP-A2 WT molecule (Fig 3A). In contrast, no association was observed between TAPBPR and GFP-A2-T134K in which a threonine at position 134 was mutated to a lysine. In agreement with previously reported findings (13, 14), mutation of T134K reduced the expression of conformation HLA-A2 detected by BB7.2 to approximately 30% of WT-A2 in IFN-γ induced HeLa (Fig 3B). Other single point mutations (D122R, E128R, and W133T) in the loop under the α2-1 similarly affected the ability of TAPBPR to associate with HLA-A2 (supplementary Fig 3). The lysine residue at position 144, which points outwards in the α2-1 helix, was also important for the association between TAPBPR and the MHC class I heavy chain (supplementary Fig 3), whilst the lysine residue at position 146 which points in towards the peptide binding groove was not critical in the association between TAPBPR and MHC class I (Fig 3A). As controls, immunoprecipitation experiments were performed in parallel for tapasin in the IFN-γ treated HeLa panel. As expected, GFP-A2-WT bound to tapasin, but GFP-A2-T134K failed to associate (Fig 3A). These findings were confirmed in reciprocal immunoprecipitation experiments in which the GFP-A2 was immunoprecipitated, followed by blotting for tapasin and TAPBPR (Fig 3C).

Figure 3. residues in the α2 and α3 domain of MHC class I are crucial for associating with TAPBPR.

A) TAPBPR or tapasin were isolated by immunoprecipitation using PeTe4 and Pasta1 respectively from IFN-γ treated HeLa cells stably expressing a panel of GFP tagged HLA-A2 mutant molecules lysed in 1 % digitonin TBS. Western blot analysis was performed for TAPBPR (R021), tapasin (Rgp48N), or GFP-A2 (using anti-GFP) on lysates and immunoprecipitates as indicated. The last lane on TAPBPR and tapasin immunoprecipitation gels is a lysate from GFP-A2 WT cells. B) Cytofluorometric analysis of HLA-A2 on HeLa cells stably transduced with selected GFP-A2 mutant molecules. All cell lines were treated with IFN-γ for 48 hours before staining with the HLA-A2 conformational specific antibody BB7.2. HeLa (grey dashed line), GFP-A2-WT (grey filled histogram), GFP-A2-T134K (black solid line), GFP-A2-α3-2 (grey solid line). C) GFP-A2 was isolated by immunoprecipitation using a GFP specific antibody from IFN-γ treated HeLa cells stably expressing the panel of GFP tagged HLA-A2 mutant molecules. Western blot analysis was performed for TAPBPR, tapasin or GFP-A2 (using an anti-GFP antibody) as indicated. The last lane on the gel is a lysate from GFP-A2 WT cells. All experiments were repeated independently at least three times. D) Untagged HLA-A2 WT, A2-T134K or A2 α3-2 were expressed in the classical MHC class I negative cell line 721.221. After solubilisation in 1% digitonin TBS, immunoprecipitation was preformed for TAPBPR (using mAb PeTe4) or β2m from precleared post-nuclear lysates. Western blot analysis was performed for HLA-A2 using HCA2. Non-transfected 721.221 was included as a negative control (−).

Mutation of residues in MHC class I α3 domain inhibit TAPBPR binding

The C-terminal domain of tapasin is predicted to bind to residues 222-229 in the α3 domain of the MHC class I heavy chain (20, 22-25). To determine if these residues on MHC class I were required for association with TAPBPR, two GFP-A2 constructs were made in which amino acids in this region were mutated (Table II). The GFP-A2 α3-1 mutant in which four residues in the α3 domain were mutated (E222K, D223R, D227R, E229K) no longer bound to TAPBPR, demonstrating that this region is crucial for MHC class I to associate with TAPBPR (Fig 3A & C). The GFP-A2 α3-2 mutant, in which only residue D227 and E229 were mutated, also did not bind to TAPBPR, refining the binding site further to involve these two amino acids. As expected, mutation to this region of the α3 domain also inhibited binding of tapasin to the MHC class I heavy chain (Fig 3A & C). Again, cell surface expression of conformational HLA-A2 detected with BB7.2 was severely reduced in these mutants, which were no longer able to bind to tapasin or TAPBPR (Fig 3B).

Mutant MHC class I molecules associate with β2m

To further verify the importance of T134 and residues in the α3 domain of MHC class I in the association with TAPBPR, untagged WT-A2, T134K, or α3-2 A2 were expressed in the classical MHC class I negative cell line 721.221. In TAPBPR immunoprecipitation experiments only untagged WT-A2 was found to associate with TAPBPR (Fig 3D). Neither untagged T143K nor untagged α3-2 were able to associate with TAPBPR (Fig 3D). These results using untagged HLA-A2 molecules expressed in 721.221 confirm the finding observed with the GFP-tagged HLA-A2 molecules expressed in HeLa cells.

Since β2m is crucial for the association of both TAPBPR and tapasin with the MHC class I heavy chain, the ability of the T134K and α3-2 mutant HLA-A2 molecules to associate with β2m was determined. Like WT-A2, both T134K and α3-2 mutant HLA-A2 were found to associate with β2m when expressed in 721.221 (Fig 3D). Therefore, the loss of interaction of the mutant HLA-A2 molecules with TAPBPR is not an indirect consequence of failure to associate with β2m.

The association of MHC class I with TAPBPR is increased in the absence of tapasin

Since TAPBPR and tapasin bind in a similar orientation to the same face of MHC class I, it is possible that they compete with each other for MHC class I binding. To investigate if tapasin competes with TAPBPR for MHC class I, the association of MHC class I with TAPBPR was compared in wild-type and tapasin deficient KBM-7 cells. Immunoprecipitation of total cellular TAPBPR revealed a significant increase in the association of MHC class I with TAPBPR in tapasin deficient cells compared to WT cells (Fig 4A). This increased association was further quantified in radiolabelled cells which revealed an approximate 2-fold increase (1.97 fold increase +/− SEM 0.065) in the association of MHC class I with TAPBPR in the absence of tapasin (Fig 4B). Therefore, tapasin can compete with TAPBPR for MHC class I binding.

Figure 4. The MHC I:TAPBPR association increases in the absence of tapasin, but the MHC I:tapasin association does not increase in the absence of TAPBPR.

A) Total cellular levels of TAPBPR and tapasin were isolated by immunoprecipitation with PeTe4 and Pasta1 respectively from IFN-γ treated WT, tapasin gene trap knockout, or KBM-7 cells depleted of TAPBPR by stable transduction with TAPBPR specific shRNA lysed in 1% digitonin-TBS. Western blot analysis was performed for TAPBPR (R021), tapasin (Rgp48N), MHC I (HC10 and HCA2), or calnexin (rabbit anti-calnexin) as indicated. B) IFN-γ treated WT, tapasin gene trap knockout, or TAPBPR depleted KBM-7 cells were radiolabelled with [³⁵S] cysteine / methionine for 30 mins, followed by immunoprecipitation of TAPBPR (using mAb PeTe4) or tapasin (using Pasta1). Signal intensity of the MHC I heavy chain bands were determined by densitometry. To calculate the fold change in MHC I association with tapasin or TAPBPR the following calculation was used: MHC I band signal intensity for test cell line / MHC I band signal intensity for WT cell line. This was then normalised to the fold change in signal intensity of immunoprecipitated protein i.e tapasin or TAPBPR. C) Total cellular levels of TAPBPR and tapasin were isolated by immunoprecipitation with PeTe4 and Pasta1 respectively from IFN-γ treated HeLa-S, HeLa-S shTAPBPR, HeLa-M or HeLa-M over-expressing WT TAPBPR cells lysed in 1% digitonin. Western blot analysis was performed for TAPBPR (R021), tapasin (Rgp48N), MHC I (HC10 and HCA2), and calnexin (rabbit anti-calnexin) as indicated. All experiments were repeated independently at least three times.

The association of MHC class I with tapasin does not increase in the absence of TAPBPR

Next we determined if TAPBPR can compete with tapasin for MHC class I binding. If TAPBPR competes with tapasin for MHC class I binding, then the association of MHC class I with tapasin should be negatively regulated by TAPBPR expression i.e. depletion of TAPBPR should increase the association of MHC I with tapasin while over-expression of TAPBPR should decrease the association of MHC I with tapasin. However, tapasin immunoprecipitation experiments did not reveal an increase in MHC class I binding to tapasin upon TAPBPR depletion in KBM-7 (Fig 4A) or in HeLa cells (Fig 4C). In contrast, densitometry in radiolabelled cells revealed the association between tapasin and MHC I was slightly reduced (by 6.5% +/− SEM 0.02) in the absence of TAPBPR (Fig 4B). Supporting this finding, a slight decrease in the association between tapasin and MHC class I was observed in the absence of TAPBPR in steady state immunoprecipitation experiments in HeLa (Fig 4C). Finally, the inability of TAPBPR to compete with tapasin for MHC class I was confirmed in HeLa cells over-expressing TAPBPR, in which a slight, but consistent, increased association was observed between tapasin and MHC class I when compared to WT HeLa cells (Fig 4C). Taken together, these results suggest that TAPBPR does not compete with tapasin for MHC class I.

DISCUSSION

Here we show that TAPBPR binds in a similar orientation to the same face of MHC class I as tapasin. Therefore, a single MHC class I molecule cannot bind tapasin and TAPBPR at the same time. Thus, TAPBPR and tapasin binding to MHC class I is mutually exclusive. In agreement with this, we previously showed there was no association between tapasin and TAPBPR in immunoprecipitation experiments. Furthermore, tapasin bound to MHC class I in the absence of TAPBPR, and TAPBPR bound to MHC class I in the absence of tapasin (27).

We previously demonstrated that over-expression of TAPBPR severely reduces surface levels of peptide loaded MHC class I and increases MHC class I free heavy chain expression (27). As over-expression of TAPBPR resembles the phenotype of a tapasin deficient cell, superficially TAPBPR appears to oppose the function of tapasin in the MHC class I antigen presentation system. This is highly reminiscent of the opposing effect of HLA-DO on the function of HLA-DM in the MHC class II antigen presentation system (36-38), raising the possibility that TAPBPR is the equivalent of HLA-DO for the MHC class I system. It has recently been shown that HLA-DO is a MHC class II mimic and binds tightly to HLA-DM, directly suppressing HLA-DM from interacting with MHC class II (39). Since there is no direct association of TAPBPR with tapasin, mechanistically TAPBPR does not appear to work in the same manner as HLA-DO. However, with TAPBPR and tapasin being orientated on MHC class I in a similar manner, an opposing role of TAPBPR could be envisaged by it directly competiting with tapasin for MHC class I binding. However, this also does not appear to be the case as TAPBPR expression does not decrease the interaction between MHC class I and tapasin. Surprisingly TAPBPR expression slightly increases the tapasin:MHC class I association. Thus, TAPBPR does not compete with tapasin for MHC class I binding, but may actually cooperate with tapasin. However, tapasin competes with TAPBPR for MHC class I. Therefore, how can we explain the fact that MHC class I surface expression in HeLa over-expressing TAPBPR resembles tapasin-deficiency? In both situations the ratio of tapasin:TAPBPR is altered in favour for TAPBPR implicating an increased association of MHC class I with TAPBPR in some of the phenotypic effects observed in tapasin deficient cells.

In light of the discovery of TAPBPR as an additional chaperone orientated on MHC class I in the same way as tapasin, the precise function of tapasin in the antigen presentation pathway now needs to be re-evaluated. Much of the work examining tapasin function has understandably been performed either using tapasin deficient cells (6, 8, 40-45) or by using mutant MHC class I molecules which no longer associate with tapasin (13, 14, 16, 23, 25). However, in tapasin-deficient cells, it is now apparent that TAPBPR is present, and is still capable of binding to MHC class I. In fact, MHC class I molecules exhibit increased binding to TAPBPR in the absence of tapasin. Therefore, the findings observed in cells lacking tapasin are likely to be highly reflective of TAPBPR function. More alarming is the issue of experiments in which MHC class I mutants were used, such as T134K, as it is clear from our data that it is not only the functional effect of tapasin on MHC class I which is lost, but also that of TAPBPR.

Our discovery of the interaction between TAPBPR and MHC class I will help explain some of the conflicting data generated regarding the function of tapasin over the past two decades. There is no doubt about the critical role played by tapasin in the loading of peptide onto MHC class I given the severe reduction of MHC class I surface expression in the absence of tapasin (2, 8, 40, 45, 46). However, there have been some discrepancies regarding the function of tapasin as a peptide editor, a process in which low affinity peptides are replaced for those with higher-affinity (3, 6, 8-10, 17, 43, 47-49). It is likely that some of the conflicting data generated is due to the influence of TAPBPR on MHC class I. For example, the T134K mutation has a more severe effect on surface expression of conformational HLA-A2 compared to the absence of tapasin (T134K-A2 is expressed at only 20% of WT-A2 in C1R cells, while HLA-A2 in 721.220 cells is expressed at 50-60% of HLA-A2 in 721.221) (13, 14, 41). Such differences are likely to be due to a lack of binding of TAPBPR to HLA-A2-T134K compared with an increased association between MHC class I and TAPBPR in the experiments using 721.220.

Our findings also raise the question of why are there two MHC class I specific chaperones in the antigen presentation pathway orientated on MHC class I in a similar manner? Are there two alternative pathways of peptide loading, or a single pathway which involves sequential engagement of the two related proteins? Is one protein primarily involved in peptide loading and the other in peptide editing? Is one the ER resident MHC class I chaperone, while the other performs a similar function out with the peptide loading complex? We favour the idea of a single pathway for MHC class I peptide loading in which tapasin and TAPBPR represent sequential steps in the same pathway. We speculate that MHC class I first associates with tapasin in order to be loaded with peptide, then interacts with TAPBPR. An initial interaction of MHC class I with tapasin instead of TAPBPR could be influenced by subtle alterations in the form of MHC class I, accessory proteins or spatial separation. If MHC class I sequentially engages with the two related proteins in this manner, then TAPBPR could monitor the stability of the MHC class I as it dissociates from the peptide loading complex. This is consistent with the increased binding of TAPBPR to MHC class I in the absence of tapasin, a condition which produces sub-optimally loaded MHC class I. In this way, TAPBPR would act as a second quality control checkpoint or post-peptide loading complex gatekeeper for MHC class I. Given the orientation of TAPBPR on MHC class I, it is conceivable that TAPBPR is capable of peptide editing. In support of an influence of TAPBPR in peptide selection, TAPBPR slows down the anterograde transport of MHC class I, a property which has been shown to allow MHC class I to efficiently optimise its peptide cargo (16, 27). Furthermore, TAPBPR prolongs the contact between the MHC class I and the peptide loading complex, an event which is highly likely to alter peptide selection by MHC class I (27). Another question under investigation is whether tapasin and TAPBPR exhibit differential preference for specific HLA gene products. Separation of the function of tapasin from TAPBPR is now needed to further elucidate the molecular mechanism governing peptide selection by MHC class I molecules.