Retention of empty MHC class I molecules by tapasin is essential to reconstitute antigen presentation in invertebrate cells

To dissect the role of tapasin in antigen processing, we chose Drosophila melanogaster cells which, like all other invertebrate cells, lack the MHC. This provided a system for reconstitution of MHC class I peptide loading via transfection with various combinations of cDNA clones encoding the individual molecules. We have shown previously that, in contrast to mammalian cells, empty MHC class I molecules are transported to the cell surface in D.melanogaster SC2 cells (Jackson et al., 1992). Thus, when the murine MHC class I heavy chain Kb was co‐transfected with murine β2m into SC2 cells, a high percentage of cells from the resulting cell line KB (for nomenclature of cell lines, see Table I) expressed Kb heterodimers at their surface, as shown by cytofluorometry with the monoclonal antibody (mAb) AF6 (Figure 1A, left panel). Surprisingly, however, almost no surface expression of Kb was detected upon co‐transfection with murine tapasin in KBN cells (Figure 1A, right panel), despite the fact that both cell lines synthesized similar amounts of Kb and β2m, as shown by metabolic labeling and immunoprecipitation (Figure 1B). Since Kb heavy chains failed to acquire EndoH resistance in KBN cells, we conclude that tapasin impaired the intracellular transport of Kb (Figure 1B and C).

Previously it was shown that insect cell‐expressed MHC molecules are empty (Jackson et al., 1992). To confirm that the ER‐retained Kb molecules were also devoid of peptides, we tested the thermostability of the immunoprecipitated complexes. By using the heterodimer‐specific antibody Y3, Kb heterodimers could only be immunoprecipitated from detergent lysates incubated at 37°C when the Kb‐specific epitope Ova8 corresponding to the amino acid sequence 257–264 of ovalbumin (Carbone and Bevan, 1989) was added as synthetic peptide to lysates (Figure 1B). We next examined whether tapasin associated with Kb by immunoprecipitation from metabolically labeled KB and KBN cells lysed in 1% digitonin. As shown in Figure 1C, antisera against Kb co‐precipitated a 48 kDa molecule from tapasin‐expressing KBN cells but not from KB cells. This 48 kDa molecule was identified as tapasin by using two different antisera (anti‐Tpn‐N and anti‐Tpn–C) raised against murine tapasin in parallel precipitation (Figure 1C) and reprecipitation experiments (Figure 5B). Anti‐Tpn‐N did not co‐precipitate Kb molecules, whereas anti‐Tpn‐C co‐precipitated small amounts of Kb, suggesting that the complexes were disrupted or not recognized by anti‐Tpn‐N antibodies. All tapasin molecules synthesized in the insect cells remained EndoH sensitive as shown by immunoblotting (Figure 1D). Thus, we conclude that tapasin associated with and retained empty MHC class I molecules in a pre‐Golgi compartment. ER retention by tapasin is not specific to Kb molecules, since the intracellular transport of other murine MHC class I molecules is also impaired (data not shown).

We hypothesized that by transfecting tapasin we had reconstituted the retention of empty MHC class I molecules observed in TAP‐deficient mammalian cells (Townsend et al., 1989). Since in TAP mutants class I molecules can be loaded with peptides directed into the ER via fusion to a signal sequence (Anderson et al., 1991), we examined the retention of MHC class I molecules upon co‐transfection of the construct SS‐Ova8 encoding the signal sequence of CD8 followed by Ova8. Indeed, Kb molecules loaded with peptides via this TAP‐independent pathway were released from tapasin retention, as shown by comparing the fluorescence‐activated cell sorting (FACS) profiles of KBNS cells with KBS cells which do not express tapasin (Table I, Figure 2A). Release from tapasin retention was also demonstrated by the finding that some of the HCs became EndoH resistant during overnight labeling in pulse–chase experiments (Figure 2B) which was not observed in the absence of SS‐Ova8 (Figure 1C). To visualize peptide‐loaded MHC complexes directly, cells were co‐stained with mAb 25.D1.16, which specifically recognizes class I–Ova8 complexes (Porgador et al., 1997). KBS and KBNS cells were gated for Kb‐positive cells and analyzed for binding of 25.D1.16. Interestingly, a significantly higher 25.D1.16 fluorescence was observed for KBNS compared with KBS (Figure 2A, lower panel). To ensure that this result was not caused by lower SS–Ova8 expression in KBS cells, we hybridized RNA isolated from KB, KBS and KBNS cells with a radiolabeled SS‐Ova8 DNA probe. The latter two cell lines clearly expressed Ova8, with KBS giving the stronger signal (Figure 2C, lower panel). Kb heavy chain mRNA used for control was expressed in all three cell lines (Figure 2C, upper panel). Therefore, it seemed that the presence of tapasin was responsible for the increased formation of peptide–class I complexes. Moreover, when the two cell lines were used to stimulate the Ova8‐specific, Kb‐restricted T cell hybridoma B3, only the tapasin‐ expressing KBNS cell line was able to stimulate B3 as a result of endogenously processed SS‐Ova8 (Figure 2D). Upon exogenous Ova8 peptide addition, both cell lines stimulated B3 equally well. The lack of T cell stimulation by KBS cells (Figure 2D) despite the presence of Ova8 complexes at the cell surface (Figure 2A) suggests that the number of Ova8 complexes is below the threshold required to activate B3 cells. We conclude that tapasin retains empty Kb molecules until they acquire peptides. By this mechanism, which occurred independently of TAP, tapasin facilitated an increase in the number of peptide–MHC complexes.

Next we examined whether TAP‐dependent import of a peptide product expressed from a minigene would result in intracellular transport of Kb molecules in the presence of tapasin. Two cell lines were generated expressing murine TAP1 and TAP2 together with the minigene for Ova8 lacking the signal sequence. Cells were also transfected with Kb, β2m (KBTM) or, in addition, tapasin (KBTNM, Table I). In agreement with our previous demonstration that murine TAP1 and TAP2 form a functional peptide transporter in D.melanogaster cells (Ahn et al., 1996), labeled reporter peptides were imported into the ER in both cell lines expressing TAP, as shown in a standard transport assay (data not shown; Neefjes et al., 1993). Both cell lines expressed Kb on the cell surface as observed by FACS analysis (Figure 3A), showing that TAP‐dependent import of the Ova8 minigene product released class I molecules from tapasin retention. Binding of TAP‐imported Ova8 to Kb was confirmed by co‐staining with mAb 25.D1.16 (Figure 3A, lower panel). In the presence of tapasin, we observed an increase of Kb–Ova8 complexes. However, the difference between cells with or without tapasin was less pronounced than the difference observed for the SS‐Ova8‐containing cell lines. A significant number of Kb–Ova8 complexes were formed independently of tapasin (compare Figures 2A and 3A). Together with the pulse–chase analysis and T‐cell assay shown below, these results suggest that import by TAP of a pre‐processed high affinity peptide can override the dependence on tapasin for efficient class I peptide loading.

To determine whether processing and presentation of antigens by MHC class I could be reconstituted in insect cells, we co‐transfected cDNAs for Kb, β2m, mTAP1, mTAP2 and ovalbumin into KBTO cells, whereas KBTNO cells were also co‐transfected with tapasin cDNA (Table I). After G418 selection and single cell cloning, two clones (KBTO‐B1 and KBTNO‐D4) were selected for further analysis. With both clones, ovalbumin expression was verified by immunoprecipitation (data not shown). Expression of all other polypeptides is shown in Figure 5. As shown in the upper right panel of Figure 3A, Kb molecules were transported to the cell surface even in the presence of tapasin, thus demonstrating that insect cells generated TAP‐dependent peptides capable of releasing Kb from tapasin retention. To determine whether ovalbumin‐derived Ova8 was among those peptides, we co‐stained with mAb 25.D1.16. A low but significant 25.D1.16 reactivity was observed for KBTN‐O4 whereas no significant staining was observed with KBTO‐B1 cells which lack tapasin (Figure 3A). These results suggested that Ova8 was produced from ovalbumin, translocated into the ER by TAP, and bound to Kb with the help of tapasin.

Kb surface expression was lower in KBTNO‐D4 cells compared with KBTO‐B1 (Figure 3A). This was also observed for other clones as well as the parental cell line (not shown). To examine whether this lower surface expression was due to tapasin retention, we studied the intracellular transport of Kb by pulse–chase analysis and immunoprecipitation. Kb heavy chains rapidly acquired EndoH resistance in KBTO‐B1 cells so that some EndoH resistance was already observed at the end of the 30 min labeling period (Figure 3B, time point 0). By contrast, in KBTNO‐D4 cells, most Kb molecules remained Endo H sensitive even after 6 h of chase, indicating continuous tapasin retention. Since overexpression of the minigene Ova8 induced a more rapid release from tapasin retention (compare KBTNM with KBTNO‐D4), it seems that insect cells were unable to provide sufficient peptides to release all of the Kb molecules from tapasin retention. However, TAP‐dependent peptide transport seemed to provide more peptides than signal sequence‐dependent translocation since intracellular transport of Kb in KBTNO‐D4 cells was faster compared with KBNS cells (Figure 2B).

Peptides which bind with high affinity to MHC class I molecules stabilize solubilized MHC heterodimers at 37°C (Townsend et al., 1990). Furthermore, acquisition of such thermostability correlates with intracellular transport of MHC molecules in mammalian cells. To examine whether Kb molecules had acquired stabilizing peptides in insect cells upon co‐transfection with tapasin, we incubated the respective lysates obtained after 6 h of chase for 1 h at 37°C with or without addition of Ova8 before immunoprecipitation (Figure 3B). A high proportion of Kb molecules were thermostable in both minigene‐containing cell lines (Figure 3B), consistent with the mAb 25.D1.16 staining shown in Figure 3A. By contrast, KBTO‐B1 cells expressed mainly empty Kb molecules, similar to results obtained in the absence of TAP (Figure 1B). Thus it seems that, despite the presence of a functional transporter in these cells, no stabilizing peptides were acquired by the Kb molecules in the absence of tapasin. By contrast, the transported EndoH‐resistant fraction of Kb expressed by the tapasin‐containing clone KBTNO‐D4 was thermostable while the majority of the EndoH‐sensitive heterodimers dissociated at 37°C. Thus, the Kb molecules which were exported to the cell surface of KBTNO‐D4 cells contained peptides which were of sufficient affinity to confer thermostability. The tapasin‐retained fraction, however, was unstable, clearly demonstrating that tapasin binds specifically to MHC class I molecules which are either empty or do not contain stabilizing peptides. These data also show that Kb molecules were released from tapasin retention during the chase period after they acquired stabilizing peptides.

In an independent experimental approach, we tested the thermostability of surface‐expressed Kb molecules by subjecting whole cells to temperature challenge prior to analyzing the presence of Kb heterodimers by cytofluorometry (Figure 3C). The mean fluorescence of KB cells incubated at 37°C was ∼20% of the mean fluorescence of cells incubated at 37°C in the presence of synthetic Ova8, indicating that most molecules did not contain stabilizing peptides. Also, KBTO‐B1 cells displayed predominantly empty Kb molecules at their cell surface. By contrast, 50–70% of Kb molecules at the cell surface of the two minigene‐expressing cells were thermostable. Moreover, co‐expression of tapasin rendered the majority of surface‐expressed Kb molecules thermostable in KBTNO‐D4 cells, thus confirming the thermostability observed by pulse–chase experiments. These results confirm that in the presence of tapasin, the surface‐expressed fraction of the Kb molecules obtained stabilizing insect cell peptides imported by TAP.

In addition to the 25.D1.16‐staining, we wanted to use specific T cells to prove that the peptide Ova8 was excised correctly from ovalbumin and loaded onto Kb molecules. However, both KBTNO‐D4 and KBTO‐B1 cells were unable to stimulate the Ova8‐specific T cell hybridoma B3 unless synthetic Ova8 was added to the cells (Figure 4A). By contrast, the Ova8 minigene‐expressing cell lines KBTNM and KBTM stimulated B3 hybridomas in the absence of the synthetic peptide (Figure 4B). A possible explanation for this lack of T cell reactivity was that the numbers of Ova8–Kb complexes on the surface of KBTNO‐D4 cells were insufficient to trigger T cell stimulation, as observed above for KBS cells (Figure 2D). To decrease the threshold for T cell triggering, we introduced ICAM‐1 and B7.1, since it has been shown that expression of these adhesion and co‐stimulatory molecules in Ld‐transfected SC2 cells strongly increased T cell stimulation (Cai et al., 1996). Expression of the transfected molecules was verified by FACS (Kb, β2m, ICAM‐1 and B7.1), immunoprecipitation (TAP1, TAP2, tapasin and ovalbumin) and TAP peptide transport assay (data not shown). The resulting cell lines KBTOI7 and KBTNOI7 (Table I) were tested for stimulation of B3 cells in the presence or absence of exogenous Ova8. As shown in Figure 4C, both cell lines were equally effective in presenting the synthetic peptide. Without synthetic peptide added, however, the tapasin‐containing cell line KBTNOI7 elicited a much stronger T cell response as compared with KBTOI7 cells. Thus, tapasin increased the efficiency of peptide presentation to T cells. These data are consistent with the 25.D1.16 staining and support the above notion that insect cells had faithfully generated the predominant epitope from ovalbumin.

In mammalian cells, it has been shown that ovalbumin is cleaved by proteasomes to yield Ova8 (Craiu et al., 1997). However, other proteases have been implicated in the generation of MHC class I‐binding peptides as well (Beninga et al., 1998; Glas et al., 1998). To analyze whether proteasomal degradation was involved in the generation of Ova8 in the SC2 transfectants, we treated KBTNOI7 cells with the proteasome inhibitor lactacystin (Fenteany et al., 1995). In order to compare the effect of lactacystin on exogenous versus endogenous presentation of Ova8, we induced expression of the transfected genes 24 h prior to co‐treatment of the cells with lactacystin, which was carried out for an additional 16 h. During the initial 24 h, sufficient Kb molecules were expressed to allow efficient T cell stimulation by exogenous peptide loading but not enough to trigger a T cell response from endogenously processed Ova8 (not shown). Therefore, presentation of synthetic Ova8 was not inhibited by additional incubation with lactacystin (Figure 4D). In contrast, presentation of SIINFEKL peptide from ovalbumin was markedly reduced by lactacystin treatment. We conclude that, as in mammalian cells, proteasomes are involved in ovalbumin degradation in D.melanogaster cells. The production of Kb‐binding peptides therefore seems a result of the constitutive cytosolic protein turnover which is largely mediated by proteasomes.

Tapasin is essential for the formation of a complex between MHC class I molecules and TAP (Ortmann et al., 1997). In addition, the chaperones calnexin, calreticulin and ER‐60 are found in this complex (Hughes and Cresswell, 1998; Lindquist et al., 1998; Morrice and Powis, 1998). To investigate whether tapasin alone would be sufficient for MHC class I association with TAP in the heterologous ER folding compartment of D.melanogaster cells, we immunoprecipitated the complexes from digitonin lysates of KBTO‐B1 and KBTNO‐D4 cells. As shown in Figure 5A, the TAP heterodimer was co‐precipitated with Kb and tapasin in KBTNO‐D4 cells. Either TAP‐, tapasin‐ or Kb‐specific antibodies co‐precipitated the other molecules, although the amounts varied depending on the antibodies used. The individual proteins were identified by reprecipitation (Figure 5B). By contrast, TAP and Kb did not co‐precipitate in the absence of tapasin in KBTO‐B1 cells. These data show that tapasin‐mediated association of empty MHC class I molecules with TAP does not depend on the co‐expression of mammalian calreticulin, calnexin or ER‐60. Whether or not tapasin recruits the Drosophila homologs of these molecules (Christodoulou et al., 1997) to the complex remains to be investigated.

Interestingly, the anti‐mTAP1 antiserum co‐precipitated less mTAP2 in the absence of tapasin (Figure 5A and B) despite the fact that both clones expressed similar amounts of mTAP2, as determined by immunoblotting with anti‐mTAP2 antiserum (Figure 5C). This suggested that tapasin influenced the stability of the TAP1–TAP2 heterodimer. This increased stability might also account for our observation that most KBTNO clones showed a higher TAP peptide transport activity than the KBTO clones (data not shown), consistent with previous observations that peptide transport activity of TAP is higher in the presence of tapasin (Lehner et al., 1998). Increased heterodimer stability might also be responsible for the higher steady‐state level of TAP observed upon tapasin transfection of .220 cells (Lehner et al., 1998). Thus, tapasin has several functions in the peptide loading process. In addition to controlling class I peptide loading, tapasin is responsible for facilitating formation of the Kb–TAP complex. Furthermore, tapasin stabilized heterodimeric complexes of TAP1 and TAP2.

Rabbit antisera to murine tapasin (Grandea et al., 1998) were raised against bovine serum albumin (BSA)‐conjugated peptides corresponding to either the 20 amino acids (anti‐Tpn‐N) following the predicted signal sequence or the C‐terminal 20 amino acids (anti‐Tpn‐C). mAb Y3 recognizes heterodimeric Kb molecules (Hämmerling et al., 1982). Phycoerythrin‐conjugated Kb‐specific antibody AF6‐88.5 was purchased from Pharmingen. Polyclonal anti‐Kb antiserum 270 as well as β2m‐specific antiserum K355 was described in Andersson et al. (1985). Anti‐Kb HC antiserum 193 was generated against a synthetic peptide corresponding to the cytoplasmic tail of Kb (Song et al., 1994). Antisera against recombinant ATP‐binding domains of murine TAP1 and murine TAP2 were described previously (Früh et al., 1992; Wang et al., 1994).

All cDNAs were inserted into expression plasmid pRMHa.3 under control of the Cu²⁺‐inducible metallothionein promoter (Jackson et al., 1992). Mouse tapasin cDNA (Grandea et al., 1998) was subcloned into pRMHa3 using SmaI and BamHI restriction sites. Construction of vectors carrying Kb, β2m and ovalbumin was published previously (Jackson et al., 1992); murine TAP1 and TAP2 constructs were described in Ahn et al. (1996). ICAM‐1 as well as B7 expression plasmids were described in Cai et al. (1996). Oligonucleotides encoding SS‐Ova8 (sequence: MALPVTALLLPLALLLHAARPSIINFEKL) or M‐Ova8 (sequence: MSIINFEKL) were inserted into the SacI and BamHI restriction sites of pRMHa3.

SC2 cells were grown at 28°C in Schneider's medium with 10% fetal calf serum (FCS). A total of 1.2×10⁷ cells were transfected with 24 μg of DNA by the CaPO₄ method using phshsneo as a marker selectable by 500 μg/ml of G418 (Jackson et al., 1992). Individual cell lines were transfected with the respective DNA constructs. Stable cell lines were obtained after 4 weeks of selection in G418. Single cell clones were generated by limiting dilution using irradiated SC2 cells as feeders. Clones were grown initially in Schneider's medium containing 20% FCS. G418 was added as soon as individual colonies became visible.

Cell pellets were resuspended either directly in SDS sample buffer or solubilized in 1% NP‐40 prior to adding sample buffer. Samples were separated on a 10–20% minigel (Bio‐Rad) and electrophoretically transferred onto Hybond membranes (Amersham). Non‐specific binding sites were blocked using 5% skim milk powder in Tris‐buffered saline, 0.3% Tween‐20 (TBST). Bound antibodies were detected using horseradish peroxidase‐conjugated secondary antibody and chemiluminescence (Amersham).

Labeled probes for SS‐Ova8 or Kb‐HC were prepared by PCR amplification followed by ³²P‐labeling using an oligonucleotide labeling kit (Pharmacia). Unincorporated nucleotides were removed by column purification (Qiagen). Probes were denatured and added to the hybridization solution at a concentration of ∼3×10⁶ c.p.m./ml.

FastTrack 2.0 mRNA isolation kit (Invitrogen) was used according to the manufacturer's instructions to isolate mRNA from ∼10⁸ cells. mRNA was separated on a 1.5% (w/v) agarose gel containing formaldehyde and transferred to a nylon membrane in 20× SSC. The mRNA was cross‐linked to the nylon membrane by two cycles of irradiation at 1200 μjoules×100. Pre‐hybridization was performed for 4 h in Northern pre‐hybridization buffer (5′→3′) containing 50% (v/v) formamide and denatured salmon sperm DNA at 42°C. Hybridization was performed overnight in Northern hybridization buffer (5′→3′) containing 50% (v/v) formamide and denatured salmon sperm and probe DNA. The blots were washed twice in 2× SSC, 0.1% SDS and twice in 0.1× SSC, 0.1% SDS at 50°C prior to autoradiography.

B3.1 hybridoma cells (Carbone et al., 1992) were grown in Dulbecco's modified Eagle's medium (DMEM) with glutamine (Gibco‐BRL), 10% FCS, 10% NCTC‐109 (BioWhittaker) and 700 μg/ml G418 (Calbiochem). SC2 cell transfectants induced for 48 h with 1 mM CuSO₄ were washed once in defined Insect Xpress medium (BioWhittaker) and incubated for 4 h with or without synthetic peptide Ova8 (SIINFEKL). Cells were washed twice in phosphate‐buffered saline (PBS) and fixed with 1% paraformaldehyde in PBS for 15 min at room temperature. Free aldehyde was quenched using NH₄Cl, 50 mM in PBS. After an additional two washes in PBS, cells were serially diluted in 96‐well plates in a volume of 100 μl. A total of 2×10⁵ B3 cells/well were co‐incubated for 20 h and the supernatant was harvested. CTLL‐2 cells were washed five times in CTLL2 medium [RPMI (Gibco‐BRL), 10% FCS, 50 μM β‐mercaptoethanol, 10% NCTC‐109] and adjusted to 5000 cells/well. At 24 h after adding 100 μl of B3 supernatant, 1 μCi/well of [³H]thymidine (NEN) was added. After 16 h, cells were harvested and counted by scintillation.

Approximately 10⁷ transfected cells induced for 24 h in 1 mM CuSO₄ were used for each immunoprecipitation. Prior to labeling, cells were starved for 30 min in Grace's insect medium without methionine (Gibco‐BRL). Cells were labeled with 0.5 mCi/ml of [³⁵S]methionine for 0.5–16 h. For chase experiments, labeled cells were washed once in PBS and incubated in complete Grace's medium (Gibco‐BRL) for the respective time periods. Cells were lysed at 4°C in PBS containing either 1% NP‐40 (Sigma) or 1% digitonin (Calbiochem) for 30 min. Lysates were cleared of nuclei and debris by 30 min centrifugation at 14 000 r.p.m. in the microcentrifuge. For temperature challenge experiments, lysates were incubated for 1 h at 37°C. After pre‐clearing with protein A–Sepharose beads (Pharmacia), the cleared lysate was incubated for 4 h with the respective antibodies which subsequently were bound to protein A–Sepharose beads. Bound antigen complexes were washed five times in PBS, 0.1% Triton X‐100 (Sigma) or 0.1% digitonin and, when needed, digested with endoglycosidase H (Boehringer Mannheim) according to the manufacturer's specifications.