Tapasin Is Required for Efficient Peptide Binding to Transporter Associated with Antigen Processing*

The transporter associated with antigen processing (TAP) binds peptides in its cytosolic part and subsequently translocates the peptides into the lumen of the endoplasmic reticulum (ER), where assembly of major histocompatibility complex (MHC) class I and peptide takes place. Tapasin is a subunit of the TAP complex and binds both to TAP1 and MHC class I. In the absence of tapasin, the assembly of MHC class I in the ER is impaired, and the surface expression is reduced. To clarify the function of tapasin in the processing of antigenic peptides, we studied the interaction of peptide and TAP, peptide transport across the membrane of the ER, and association of peptides with MHC class I molecules in the microsomes derived from tapasin mutant cell line 721.220, its sister cell line 721.221 expressing tapasin, and their HLA-A2 transfectants. The binding of peptides to TAP in tapasin mutant 721.220 cells was significantly diminished in comparison with 721.221 cells. Impaired peptide-TAP interaction resulted in a defective peptide transport in tapasin mutant 721.220 cells. Interestingly, despite the diminished peptide binding to TAP, the transport rate of TAP-associated peptides was not significantly altered in 721.220 cells. After transfection of tapasin cDNA into 721.220 cells, efficient peptide-TAP interaction was restored. Thus, we conclude that tapasin is required for efficient peptide-TAP interaction.

hydrophobic domains (5). The feature of TAP1 and TAP2 proteins were revealed as members of the ABC transporter family (5). TAP1 and TAP2 bind short peptides of 7-12 amino acids and have broad specificity (6 -8). The efficient binding of peptides requires expression of both TAP1 and TAP2 (8). The interaction of TAP and peptides is ATP-independent (6 -9). The addition of ATP dissociates peptides from TAP and stimulates the assembly of peptide and MHC class I in the lumen of the ER indicative of translocation of TAP-released peptide across the membrane of the ER (6 -9). Immunoprecipitation with anti-TAP1 antiserum demonstrated that TAP associates with MHC class I heavy chain-␤ 2 -microglobulin dimer (10,11). The importance of the TAP-MHC class I interaction for the assembly of MHC class I and peptides was suggested by the finding in which deficiency in MHC class I surface expression was found in a cell line, 721.220, lacking interaction of MHC class I and TAP (12). With anti-TAP1 antiserum, a 48-kDa protein (tapasin) was coprecipitated, and this protein was missing in 721.220 cells, indicative of requirement of tapasin in the interaction of MHC class I and TAP (13). cDNA cloning of tapasin revealed a type I membrane protein with a cytoplasmic tail containing a double lysine motif known to maintain membrane proteins in the ER (14,15). Immunoprecipitation with anti-TAP1 or anti-tapasin antisera demonstrated a consistent and stoichiometric association of tapasin and TAP1 and TAP2 (14). The importance of tapasin in MHC class I antigen presentation was demonstrated by restored MHC class I surface expression and class I-TAP interaction in 721.220 cells after transfection with tapasin cDNA (15), suggesting that either tapasin directly regulates the assembly of MHC class I and peptide or the association of MHC class I with TAP enhances peptide loading.
The importance of MHC class I-TAP interaction in peptide loading was opposed by recent findings, in which the ER luminal domain of tapasin (soluble tapasin) was expressed in 721.220 cells. Soluble tapasin associated with class I but did not interact with TAP (16). In the absence of interaction with TAP, soluble tapasin restored surface expression of MHC class I and the presentation of viral peptides to CTL (16). The conclusion from this study was that tapasin-MHC class I interaction, but not tapasin-TAP interaction, was required for peptide loading (16). However, there are conflicting reports concerning the importance of the tapasin-MHC class I interaction in promoting peptide loading. It has been reported that HLA-B27 and HLA-A2 could assemble with peptide in tapasin mutant cells, although HLA-B27 and HLA-A2 could interact with tapasin in wild type cells (17,18). In addition, a murine mutant MHC class I Dd (E222K) was discovered having a Glu to Lys mutation at residue 222 of the heavy chain. This mutation caused deficiency in interaction with tapasin (19). Significant peptide loading onto Dd (E222K) was observed (19). These results indicate that the interaction of tapasin and MHC class I is not essential for peptide loading.
In this study, we used reporter peptides in which the ⑀-amino group of lysine was covalently modified by coupling a chemical cross-linker (ANB-NOS), and the tyrosine was labeled by iodination ( 125 I). These modifications allowed photocross-linking of the reporter peptides to TAP and MHC class I and enabled us to monitor the peptide binding to TAP, peptide translocation, and assembly of peptide and MHC class I in purified microsomes derived from tapasin mutant cells, 721.220, and its sister cells, 721.221, as well as their HLA-A2 transfectants.
Our results clearly indicate that tapasin is required for efficient peptide-TAP interaction.

MATERIALS AND METHODS
Cells-721.220 and 721.221 cell lines and 721.220-HLA-A2 and 721.221-HLA-A2 cell lines were kindly provided by Drs. J. C. Solheim and T. Elliott, respectively. The 721.174 cell line was a gift from Dr. S. Kvist The cell lines were cultured in RPMI 1640 medium (Life Technologies, Inc.), supplemented with 10% heat-inactivated fetal bovine serum, 100 IU/ml penicillin, 100 mg/ml streptomycin, and 2 mM glutamine at 37°C in a 5% CO 2 atmosphere.
The cDNA encoding wild type human tapasin was constructed into pCEP4 expression vector (CLONTECH). The constructs were transfected into 721.220 cells by electroporation. Transfected cells were selected in hygromycin-containing medium (Life Technologies, Inc.) and then cloned by limiting dilution. Expression of transfected tapasin was measured by immunoblotting with anti-human tapasin antiserum.
Antibodies-The monoclonal antibody BB7.2, specific for HLA-A2, and rabbit antiserum to human MHC class I (R425) were kindly provided by Dr. S. Kvist. Rabbit antisera against human TAP1 and tapasin were described previously (9,14). The polyclonal antibodies were affinity-purified before use.
Metabolic Labeling, Immunoprecipitation, and Immunoblotting-Cells were washed twice with phosphate-buffered saline and incubated for 15 min at 37°C in methionine-free RPMI 1640 medium containing 3% dialyzed fetal bovine serum. Then 0.2 mCi/ml [ 35 S]methionine (Amersham Pharmacia Biotech) was added, and the incubation was continued for 60 min. At the end of labeling, cells were washed three times with ice-cold phosphate-buffered saline and lysed in 1% digitonin (Sigma) or 1% Nonidet P-40 lysis buffer containing 0.15 M NaCl, 25 mM Tris-HCl, pH 7.5, 1.5 mg/ml iodoacetamide, and a mixture of protease inhibitors (2 mM phenylmethylsulfonyl fluoride, 10 mg/ml leupeptin, 30 mg/ml aprotinin, 10 mg/ml pepstatin). The cleared lysates were added to antibodies previously bound to protein A-Sepharose beads (Amersham Pharmacia Biotech, Uppsala, Sweden). After washing, the immunoprecipitates were analyzed by SDS-PAGE. Western blotting and FACS analysis were performed as described previously (9).
Preparation of Microsomes and Photocross-linking-Microsomes from cell lines were prepared and purified according to Saraste et al. (20). For photocross-linking, 125 I-labeled and ANB-NOS-modified peptide was mixed with 20 l of microsomes (concentration of 60 A 280 /ml) to a final concentration of 100 nM, in RM buffer (250 mM sucrose, 50 mM triethanolamine-HCl, 50 mM KOAc, 2 mM MgOAc 2 , 1 mM dithiothreitol). This mixture was then kept at 26°C for 5 min UV irradiation was subsequently carried out for 5 min on ice at 366 nm. Microsomal membranes were recovered by centrifugation through a 0.5 M sucrose cushion in RM buffer containing 1 mM cold peptide (unlabeled peptide without ANB-NOS modification). The microsomal membranes were washed once with cold RM buffer, lysed by 1% digitonin or 1% Nonidet P-40, and subjected to immunoprecipitation. Cross-linked microsomal proteins were immunoprecipitated with specific antiserum. The precipitates were analyzed by SDS-PAGE or quantitated by a ␥-counter. Cross-linking reactions with transport buffer containing 1 mM ATP were performed as described previously (7,14). For peptide competition, 100 nM of the 125 I-labeled and ANB-NOS-modified peptide was mixed with a 10-fold molar excess or with the concentrations indicated in Fig.  3 of unlabeled and unmodified peptide before the cross-linking reaction.

Interaction of Peptides with TAP in Microsomes of Tapasin
Mutant 721.220 Cells and Wild Type Cells-It was previously reported that the peptide transport across the ER membrane is a stepwise process including an ATP-independent interaction of peptide and TAP and a subsequent ATP-dependent translocation of peptide (6,9). To investigate the involvement of tapasin in TAP function, three reporter peptides (see "Materials and Methods") were modified with cross-linker and labeled with 125 I as described previously (7) and incubated with microsomes derived from the tapasin mutant cell 721.220, its sister cell line 721.221 with wild type tapasin, and a TAP mutant cell line 721.174 in the absence of ATP. After incubation, the microsomes were irradiated by UV to induce cross-linking. The peptide-bound TAP molecules were precipitated by anti-TAP1 antiserum. Analysis of precipitates revealed efficient binding of all three reporter peptides to TAP in microsomes from 721.221 (Fig. 1A, lanes 2, 4, and 6). Significantly diminished peptide-TAP interaction was observed in tapasin mutant microsomes (Fig. 1A, lanes 3, 5, and 7). The expression level of TAP1 and TAP2 in 721.221 and 721.220 was very similar, as indicated by immunoprecipitation with anti-TAP1 antiserum (Fig. 1B, lanes  2-3). TAP mutant 721.174 served as a negative control in both cross-linking and immunoprecipitation experiments (Fig. 1,  lane 1, in A and B). A similar experiment performed with microsomes of tapasin-transfected 721.220 cells demonstrated restoration of an efficient peptide binding to TAP (Fig. 2). These results indicate that tapasin is required for an efficient peptide-TAP interaction.
Low Affinity Peptide-TAP Interaction in Tapasin Mutant 721.220 Cells-To assess the binding affinity of peptide to TAP in 721.221 or 721.220 cells, 100 nM of cross-linker-modified reporter peptide, 125 I-OVA-ANB-NOS, was incubated with 721.221 or 721.220 microsomes in the presence of native and unlabeled OVA peptide at different concentrations. In order to dissect the peptide binding from the peptide translocation, the assay was done in the absence of ATP as previously reported (9). Again a significant deficiency of peptide-TAP interaction was detected in 721.220 microsomes in the absence of competing peptide (Fig. 3, lane 1). The peptide binding to TAP in 721.220 microsomes was completely competed by a lower concentration (400 nM) of native peptide (Fig. 3), whereas in 721.221 microsomes, more than 1.6 M concentration was required for completely competing away reporter peptide binding (Fig. 3). These data indicate that the affinity for peptide binding to TAP in tapasin mutant cells is much lower than in wild type cells.
Peptide Transport Efficiency Is Not Reduced in Tapasin Mutant Cells-After having demonstrated a lower peptide binding to TAP in tapasin mutant cells, we examined the transport efficiency by measuring the time for translocation of 50% TAPbound peptides into the microsomes of wild type, 721.221A2, or tapasin mutant cells, 721.220A2, in the presence of ATP. After incubation of 125 I-MP-ANB-NOS peptide with microsomes from 721.220A2 or 721.221A2 cells, the excess 125 I-MP-ANB-NOS peptides were washed off. Peptide-loaded microsomes were then resuspended in transport buffer with 100 M ATP and incubated for different periods of time. After incubation, the microsomes were lysed in 1% Nonidet P-40 lysis buffer and subsequently precipitated with anti-TAP1 or anti-HLA-A2 antibodies. The peptide-bound TAP or HLA-A2 molecules were quantitated by a ␥-counter. In the presence of ATP, peptide rapidly dissociated from TAP in both 721.220A2 and 721.221A2 microsomes, despite the fact that the amount of peptide-bound TAP1 in 721.220A2 cells was lower (Fig. 4, upper panel). Fifty percent dissociation of peptides from TAP was detected at an early time point in both 721.220A2 and 721.221A2 microsomes (Fig. 4, upper panel). Moreover, 50% of translocated peptides were recovered by anti-HLA-A2 antibody at the same time point in both 721.220A2 and 721.221A2 microsomes, although a much lower amount of peptide-bound A2 was obtained in 721.220A2 microsomes (Fig. 4, lower panel). These results indicated that mutation of tapasin greatly affected peptide interaction with TAP but affected the peptide translocation much less or not at all.
These results suggest that the reduced surface expression and translocated peptides in the ER of 721.220A2 cells result from deficient peptide-TAP interaction. Moreover, despite the deficiency of peptide-TAP interaction in 721.220A2 cells, TAP still can transport peptides in the presence of ATP, which is compatible with previous findings (12).
Interaction of Peptide-loaded HLA-A2 with Tapasin-Previously, it was reported that the assembly of peptide and MHC class I in tapasin mutant cells was defective (8). We have previously demonstrated that murine tapasin associated with peptide-bound H-2K b (21). To further confirm the interaction of peptide-bound MHC class I with human tapasin, purified microsomes from 721.220A2 and 721.221A2 were incubated with 125 I-MP-ANB-NOS peptide in the presence of ATP. Anti-tapasin antiserum precipitated peptide-bound HLA-A2 and TAP in 721.221A2 but not in 721.220A2 microsomes (Fig. 7, lanes 1  and 2). This result clearly indicates that tapasin does not exclusively bind to peptide free MHC class I. DISCUSSION The function of TAP to mediate peptide transport into the ER is well established (2,5,8). TAP interacts with peptides at its cytosolic part and with MHC class I at its luminal part (6 -9). The importance of a direct interaction between peptide and TAP for peptide translocation was clearly demonstrated by the findings in which the herpes simplex virus ICP47 protein inhibits the MHC class I antigen presentation pathway by occu-pying the peptide binding site on TAP (22)(23)(24). The association of MHC class I and TAP is mediated by tapasin (13-15, 18, 21). Tapasin is a type I ER membrane protein and bridges the association of MHC class I with TAP (13,14). Cells with mutated tapasin have a defective surface expression of MHC class I (12), and this defect can be corrected by transfection of wild type tapasin cDNA (13). Tapasin was suggested to function by promoting the peptide loading onto MHC class I (8). In the present study, we systematically examined the peptide interaction with TAP, the peptide translocation into the ER, and the peptide assembly with MHC class I in microsomes purified from tapasin mutant cells and wild type cells as well. A severe defect in peptide-TAP interaction was revealed in the tapasin mutated cell 721.220. In contrast, the peptide translocation across the membrane of the ER was intact, as indicated by the same off rate of TAP-associated peptides in tapasin mutated and wild type microsomes in the presence of ATP. The translocated HLA-A2-binding peptide could bind to HLA-A2 molecules in both tapasin mutated and wild type microsomes, although the amount of peptide-associated HLA-A2 in 721.220-A2 cells was much less than that in its sister cell line 721.221-A2 expressing wild type tapasin. This reduction was due to the deficient interaction between peptide and TAP.
Tapasin is a subunit of the TAP complex as indicated by the consistent and stoichiometric association of tapasin with TAP1 and TAP2 (14). Previously, it was reported that peptide transport in tapasin mutant cells was not altered (12). In that study, translocation of peptides in 721.220 and wild type cells was analyzed by recovery of a glycosylated reporter peptide in streptolysin O-permeabilized cells. Since TAP-dependent peptide translocation was intact in tapasin mutant cells, it was conceivable that accumulation of translocated peptides in the ER might result from the decay kinetics of transported peptides in tapasin mutant cells. Restored peptide-TAP binding and peptide translocation in tapasin-transfected 721.220 cells clearly demonstrated that tapasin is required for efficient peptide binding to TAP. In agreement with our results from tapasin-transfected 721.220, a study of 721.220B8 tapasin transfectant also revealed a more than 4-fold increase in peptide translocation after expressing tapasin in 721.220B8 cells (16). In addition, it has recently been reported that the C-terminal region of tapasin was identified as a binding site of TAP (25). Transfection of the C-terminal region of tapasin enhanced the function of TAP in tapasin mutant cells (25).
In previous studies (14,21) as well as the present studies (Fig. 5, lane 4), we found a weak interaction of peptides and tapasin in the presence of ATP but not in the absence of ATP. Since the peptide-TAP interaction is independent of ATP, it is not likely that tapasin is involved in the peptide binding site on the TAP complex. Both TAP1 and TAP2 are required for efficient peptide binding, certain peptides binding preferentially to TAP1 and others to TAP2 (10). Although peptide binding to human TAP is relatively promiscuous, the difference of binding affinity of various peptides can be as great as 3 orders of magnitude, depending on both the terminal and the internal sequence of the peptides (7,26). Among three peptides tested in this study, similar deficiency in their ability to bind to TAP of 721.220 cells suggested that the requirement of tapasin for peptide-TAP interaction is not based on the sequence of the peptides.
Searching for peptide-binding sites, the regions of the TAP subunits to which the photoactive peptides bind were mapped by proteolysis of the TAP proteins after photocross-linking and immunoprecipitation with antisera specific for distinct hydrophilic regions (27,28). The results suggest that the binding site is composed of multiple regions of TAP1 and TAP2. Interaction of the same photoreactive peptide to multiple regions of TAP1 and TAP2 may indicate a large binding pocket (27,28). This study also suggested that the binding regions are very close to or part of the transmembrane domains. The interaction of tapasin with TAP may largely regulate the peptide binding site. Recently, the ER luminal domain of tapasin was expressed as a soluble form of tapasin in 721.220 cells (16). Soluble tapasin bound to MHC class I, but not to TAP. The peptide transport was measured in 721.220, 721.220 transfected with soluble tapasin, and 721.220 with wild type tapasin. Results showed that wild type tapasin increased the efficient peptide translocation in 721.220 cells, but soluble tapasin transfectants retained their deficient peptide transport (16). Taken together, these results and our own indicate that the interaction of tapasin and TAP is critical for efficient peptide transport. Tapasin may function to stabilize the peptide-binding site of TAP.
Since tapasin directly interacts with MHC class I heavy chain and ␤ 2 -microglobulin dimer, it was suggested that this interaction enhances peptide loading onto class I molecules, possibly due to a high local concentration of peptides or to the requirement of tapasin itself for peptide-MHC class I assembly. Support for the importance of MHC class I-TAP interaction was evidenced by the finding of a deficient TAP-dependent peptide assembly of a mutant HLA-A2.1 (29,30). A point mutation of theronine 134 to lysine resulted in HLA-A2.1 incapable of interacting with TAP (29,30). Moreover, the deficient assembly of HLA-A2.1 (T134K) was corrected by a direct delivery of peptide to the ER in a TAP-independent manner. Therefore, it was concluded that the interaction of TAP and MHC class I is essential for peptide loading. Recently, this notion has been called into question by the finding that a soluble tapasin restored the surface expression of HLA-B8 in 721.220B8 cells, despite the lack of interaction between MHC class I and TAP (16). Notably, soluble tapasin-MHC class I interaction was unstable in detergent lysates and required chemical cross-linking for their detection. Since soluble tapasin lacks the motif of the ER retrieval signal present in the wild type tapasin, the increased surface expression can be explained as a lack of control of MHC class I retention in the ER. Tapasin may regulate the MHC class I release from the ER rather than directly load peptides on MHC class I. Recently, it was found that HLA-B27 and HLA-A2 presented peptides to cytotoxic T cells in tapasin mutant 721.220B27 or 721.220A2 cells (17,31), although both alleles were found to associate with tapasin in wild type cells. Furthermore, a mutant H2-Dd with a Glu to Lys mutation at residue 222 (Dd E222K) was shown not to be substantially impaired in its ability to present peptides (18) despite its loss of ability for interaction with tapasin. Although there is a lack of direct evidence for the involvement of tapasin in peptide loading onto MHC class I, a recent study of the HLA-A2 T134K mutant indicates that the interaction of MHC class I with the TAP complex may be important for the optimization of MHC class I assembly (32). The T134K mutant did not bind to tapasin or calreticulin, and it failed to present endogenous viral peptides to T cells (32). However, the mutant class I had the ability to bind peptides in vitro (32). Therefore, it was suggested that the TAP complex is involved in a quality control stage during MHC class I assembly. This view is supported by the study of green fluorescence protein-tagged MHC class I (33), which shows that peptide-loaded MHC class I can also be retained in the ER for optimizing peptide loading. In the presented study, most of the peptide-loaded HLA-A2 was found to be associated with tapasin in 721.221A2, which is in line with our previous finding of an association of peptide-loaded mouse MHC class I with tapasin (21) and indicates that tapasin may also regulate the transport of assembled MHC class I to the cell surface.