|Home | About | Journals | Submit | Contact Us | Français|
For their efficient assembly in the endoplasmic reticulum (ER), major histocompatibility complex (MHC) class I molecules require the specific assembly factors transporter associated with antigen processing (TAP) and tapasin, as well generic ER folding factors, including the oxidoreductases ERp57 and PDI, and the chaperone calreticulin. TAP transports peptides from the cytosol into the ER. Tapasin promotes the assembly of major histocompatibility complex (MHC) class I molecules with peptides. The formation of disulfide-linked conjugates of tapasin with ERp57 is suggested to be crucial for tapasin function. Important functional roles are also suggested for the tapasin transmembrane and cytoplasmic domains, sites of tapasin interaction with TAP. We show that interactions of tapasin with both TAP and ERp57 are correlated with strong MHC class I recruitment and assembly enhancement. Presence of the transmembrane/cytosolic regions of tapasin is critical for efficient tapasin-MHC class I binding in interferon-γ-treated cells, and contributes to an ERp57-independent mode of MHC class I assembly enhancement. A second ERp57-dependent mode of tapasin function correlates with enhanced MHC class I binding to tapasin and calreticulin. We also show that PDI binds to TAP in a tapasin-independent manner, but forms disulfide-linked conjugates with soluble tapasin. Thus, full-length tapasin is important for enhancing recruitment of MHC class I molecules and increasing specificity of tapasin-ERp57 conjugation. Furthermore, tapasin or the TAP/tapasin complex has an intrinsic ability to recruit MHC class I molecules and promote assembly, but also uses generic folding factors to enhance MHC class I recruitment and assembly.
MHC class I molecules are activating ligands for T cell receptors of CD8+ T cells, and the down-modulation of MHC class I molecules triggers immune recognition by natural killer cells. Thus, understanding the mechanisms of MHC class I assembly and cell surface expression is fundamental to many immune recognition processes. MHC class I molecules comprise a heavy chain, a light chain, and a short peptide. Assembly of these components occurs in the endoplasmic reticulum (ER) of cells, and involves specific assembly factors TAP and tapasin, and generic ER chaperones. TAP is a critical factor comprising two subunits, TAP1 (ABCB2) and TAP2 (ABCB3), that are required for translocation of peptides from the cytosol into the ER (reviewed in (1)). Tapasin is another critical co-factor required for assembly of MHC class I heavy and light chain heterodimers with peptides (2, 3).
Tapasin binds to TAP and increases steady-state levels of TAP, thereby allowing more peptides to be translocated into the ER (4, 5). Truncated tapasin lacking the transmembrane and cytosolic regions (soluble tapasin) does not bind to TAP, but is able to bind MHC class I molecules and induce MHC class I cell surface expression to some extent (4). A disulfide-linkage is formed between the luminal cysteine 95 of tapasin and cysteine 57 of ERp57 (6, 7). Tapasin-ERp57 binding was shown to enhance tapasin-MHC class I interactions as well as the functional activity of tapasin in reconstituted lysates and in tapasin-deficient cells (8, 9). These studies have led to prevailing models that the transmembrane and cytosolic domains of tapasin are important for TAP stabilization and function, and that the ER luminal domains of tapasin are important for MHC class I recruitment and assembly enhancement. The tapasin-ERp57 conjugate is suggested to be the functional unit for recruiting MHC class I molecules and facilitating their peptide binding (8). However, purified soluble tapasin alone is able to interact with purified soluble MHC class I in a peptide-regulatable manner, but unable to induce MHC class I assembly with peptides (10). Purified soluble tapasin alone when tethered to purified MHC class I molecules via a Fos-Jun linkage is able to impact the assembly of MHC class I molecules with peptides (11). Furthermore, recent studies also suggest a role for the transmembrane/cytosolic region of tapasin for tapasin-MHC class I binding and for the functional activity of tapasin (12).
These different findings make it important to further understand mechanisms of MHC class I recruitment to tapasin and of tapasin-mediated enhancement in assembly of MHC class I molecules. Towards this end, we generated tapasin mutants lacking the ERp57 binding site (tapasin C95A), the TAP binding site (soluble tapasin), or both (soluble tapasin C95A). Functional and interactions studies with these tapasin mutants provided important insights into the distinct modes of tapasin functions.
ERp57 is structurally and functionally related to PDI which was recently found to be associated with the TAP complex, and also suggested to be important for maintaining oxidized forms of TAP-associated MHC class I heavy chains (13). Some recent studies have questioned whether PDI is in fact recruited into the TAP complex ((14) and references cited therein), and previous studies indicated that PDI did not form a disulfide-linked tapasin conjugate, under conditions where tapasin-ERp57 conjugates were observable (15). We also undertook analyses of interactions of PDI with TAP and tapasin in the context of the different tapasin mutants. These studies revealed a tapasin-independent mode of PDI-TAP association, and unexpected insights into the role of TAP-tapasin binding for enhancing specificity of tapasin-ERp57 conjugation.
M553 is a tapasin-deficient human melanoma cell line with HLA-A2-haplotype loss (16). We examined the effects of the tapasin(C95A) mutation upon the abilities of full-length and soluble FLAG-tagged tapasin to induce cell surface expression of endogenous MHC class I of M553 cells (Figure 1A) and exogenously-expressed influenza hemagglutinin-tagged HLA-A2 (M553/A2) (Figure 1B). Tapasin, when expressed exogenously, up-regulated MHC class I cell surface expression in M553 and M555/A2 cells (Figures 1A and 1B), as previously described (16). Tapasin also promoted assembly and ER exit of endogenously- and exogenously-expressed MHC class I molecules in M553 cells (data not shown). The soluble tapasin truncation reduced tapasin’s ability to induce MHC class I surface expression, consistent with previous findings that TAP association is required for full function of tapasin (17, 18) (Figures 1A and 1B). In the context of full-length tapasin, the C95A mutation also partially impacted tapasin’s ability to induce endogenous MHC class I surface expression (Figure 1A) or exogenous HLA-A2 (A2) surface expression (Figure 1B) whereas in the context of soluble tapasin, the C95A mutant had very low activity (Figures 1A and 1B). In M553 cells expressing wild type soluble and full length tapasin, class I expression (total steady state class I) was enhanced compared to M553 alone (lacking tapasin) suggesting that wild type soluble and full length tapasin expression stabilized endogenous MHC class I in M553 cells (Figure 1A, lower panel, class I blot). Similar tapasin-mediated stabilization of exogenous class I (A2) was also observed (Figure 1B, lower panel, A2 blot). Tapasin expression in M553 cells was comparable to the endogenous levels observed in other cell types such as HeLa cells (Figure 1C). The presence of the FLAG tag increased the molecular weight of soluble tapasin causing it to migrate very close to full length tapasin when analyzed following SDS-PAGE and immunoblotting (Figure 1A and 1B, lower panel, tapasin blot). Under steady state conditions, soluble tapasin remains completely Endo-H sensitive, indicating that it is retained in the ER of M553 cells (Figure 1D).
Tapasin-ERp57 interaction was suggested to be critical for tapasin-MHC class I binding in some studies (8, 9), whereas other studies have suggested a critical role for the transmembrane and cytosolic domains of tapasin in MHC class I binding (12). To further examine these possibilities, we undertook various co-immunoprecipitation studies. Efficient observation of tapasin binding to ERp57 and MHC class I in M553 cells required the use of a cross-linker such as Dimethyl 3,3′ dithiobispropionimidate 2HCl (DTBP) or a thiol modifying agent such as Methyl methanethiosulfonate (MMTS). Using DTBP, a cell permeable cross-linker that was quenched prior to cell lysis, tapasin-ERp57 interactions were observed, and were C95-dependent in the context of both full-length and soluble tapasin as expected (Figures 2A and 2D). We believe this is evidence that specific C95-dependent interactions occur in cells and are not an artifact of interactions induced during cell lysis in MMTS. In all our binding studies, interferon-gamma (IFN-γ) treatment of cells was also used to up-regulate TAP and enhance interactions that were potentially TAP-dependent, and also induce MHC class I expression for the interaction studies. Under these conditions, both full-length tapasin-MHC class I complexes were detectable with significantly higher efficiency than either soluble tapasin-MHC class I complex, and the C95A mutation reduced binding to MHC class I in the context of both full-length and soluble tapasin (Figure 2B and 2E). The tapasin-MHC class I binding was observable even in the context of soluble tapasin(C95A), consistent with our previous finding that purified soluble tapasin and MHC class I are able to interact in the absence of all other components of the MHC class I assembly complex (10). However this interaction is essentially functionally sterile in cells (Figure 1), as previously observed with the purified proteins (10).
Calreticulin is a component of MHC class I assembly complex (3), and calreticulin and ERp57 can interact independent of other MHC class I assembly components (19). Additionally, tapasin- and ERp57-deficient cells have reduced abilities to recruit calreticulin into associations with the MHC class I assembly complex (20, 21). Thus, investigations of the ERp57-dependence of tapasin-calreticulin binding was relevant. Calreticulin-tapasin binding was most efficiently detected in the context of full-length wild type tapasin, and the C95A mutation reduced binding efficiency but did not abrogate binding, indicating that calreticulin association with tapasin was linked to ERp57 (Figures 2C and 2F), particularly in the context of full-length tapasin. Similar levels of calreticulin-tapasin interactions were observed when M553 and M553/A2 cells were compared within the same experiment, indicating that over-expression of MHC class I did not significantly impact the efficiency of calreticulin-tapasin binding under the conditions of the analyses (Figure 2G).
Based on previous studies (4, 9), the transmembrane and cytosolic domains of tapasin are expected to be essential for TAP binding, and C95 of tapasin is expected to be essential for TAP-ERp57 binding and critical for TAP-MHC class I and TAP-calreticulin binding. To further examine these requirements for tapasin in bridging different interactions to TAP in IFN-γ treated cells, we undertook co-immunoprecipitation analyses with a TAP-specific antibody. TAP-tapasin binding was indeed strictly dependent on the presence of full-length tapasin (Figure 3A and 3E). No binding was observed between soluble tapasin and TAP1 in the anti-TAP1 IPs (Figure 3A and 3E), demonstrating the specificities of the interactions analyses in these assays. ERp57 binding to TAP1 was also tapasin-dependent, and required the presence of C95 of tapasin (Figure 3B and 3F). Furthermore, calreticulin binding to TAP was ERp57-linked (Figure 3C and 3G). Finally, MHC class I association with TAP was also strongly tapasin-dependent, but at best only partly impacted by the C95A mutation (Figure 3D and 3H), consistent with the findings of Figure 2. TAP1 interactions with indicated proteins were stable and interactions were detectable both in the absence (Figure 3A–3D) and in the presence of cross-linker (3E–3H). Together, these findings indicate that ERp57 association with TAP is dependent on tapasin and requires an intact tapasin C95, that calreticulin association with TAP is ERp57-linked, and that MHC class I association with TAP is strongly tapasin-dependent, but only partly linked to C95 of tapasin.
The binding studies of Figure 2 undertaken with an anti-tapasin antibody revealed that soluble tapasin truncation impaired tapasin-MHC class I binding more significantly than the tapasin(C95A) mutation, indicating that the tapasin transmembrane/cytosolic regions indirectly or directly stabilized tapasin-MHC class I binding. Similar results were obtained when the immunoprecipitation analyses were undertaken with an anti-MHC class I (HC-10) antibody (Figures 4A and B). HC-10 was used as it has previously been shown to be the most efficient MHC class I-specific antibody available in our laboratory to detect tapasin-MHC class I complexes (22). ERp57 binding to MHC class I was readily detectable, in the context of full-length tapasin (Figure 4A and B), and very weakly detectable in the context of soluble tapasin (Figure 4B) likely reflecting the indirect tapasin-mediated association between ERp57 and MHC class I heavy chains. Despite weaker MHC class I binding by soluble tapasin compared to the full-length tapasin(C95A) mutant, soluble tapasin was able to induce MHC class I to a level similar to or greater than that induced by full-length tapasin(C95A), both in the absence of IFN-γ (Figure 1) and in the presence of IFN-γ (Figure 4C). Functional activity of soluble tapasin correlated with its ability to more efficiently recruit calreticulin into association with MHC class I molecules, compared to tapasin(C95A) (Figure 4A and B). Calreticulin binding to MHC class I was detected most efficiently in the presence of full-length tapasin, and additionally, calreticulin binding was enhanced in the context of wild type and soluble tapasin compared with either C95A mutant (Figure 4A and B). Together, these results suggest that the ability to recruit calreticulin into associations with MHC class I molecules in an ERp57-dependent manner may underlie the observed functional activity of soluble tapasin. ERp57 association with tapasin brings calreticulin into proximity with MHC class I, which may be related to the enhancements in steady state MHC class I in cell lysates in the context of full-length or soluble tapasin (Figure 1A and 1B, lower panels).
Pulse chase and Endo-H digestion analyses showed that, in cells lacking tapasin, MHC class I molecules were retained in the ER and did not mature even after 120 minutes. In contrast, in the presence of full-length tapasin(C95A), maturation of MHC class I molecules was accelerated, indicating that tapasin(C95A) actively promoted assembly of MHC class I molecules and their ER exit (Figure 5A). Use of wild type tapasin further enhanced maturation of MHC class I heavy chains (data not shown). The ability of full-length tapasin(C95A) to induce MHC class I assembly and cell surface expression (Figures 1 and and5A)5A) could be related to tapasin(C95A)-mediated induction of TAP steady-state levels (4, 5), an intrinsic assembly promoting function of tapasin(C95A), or tapasin(C95A)-dependent tethering of MHC class I molecules into a physical association with TAP. To further examine the first possibility, we compared the effects of IFN-γ treatment of M553 cells upon the cell surface expression of MHC class I in M553 cells and in M553 cells expressing full length tapasin(C95A) (Figure 5B). IFN-γ up-regulated MHC class I cell surface expression in both cells with a more significant effect being observed in the presence of full length tapasinC95A (Figure 5B). Comparison of steady state levels of TAP and MHC class I in IFN-γ-treated cells or tapasin(C95A)-expressing cells revealed that under the conditions of the analyses, IFN-γ treatment had a much more profound effect on steady state levels of TAP1 and MHC class I, compared to that induced by tapasin(C95A) expression (Figure 5C and 5D). Additionally higher levels of TAP1/TAP2 complexes were induced by IFN-γ-treatment, as assessed by immunoprecipitations with anti-TAP1, followed by immunoblotting analyses for TAP2 (Figure 5D). Despite the stronger induction of active TAP complexes and MHC class I molecules in IFN-γtreated M553 cells, cell surface MHC class I expression was consistently lower than that induced by tapasin(C95A), arguing that the mechanism of tapasin(C95A) function did not relate to its ability to stabilize TAP complexes per se. Rather, the functional activity of full-length tapasin(C95A) relates either to its intrinsic ability to bind MHC class I and promote peptide loading of MHC class I, or to its ability to induce a physical association between MHC class I and the TAP complex.
PDI has previously been shown to be associate with TAP (13), but the tapasin dependence of this interaction is not well-characterized. We examined the interactions of PDI with TAP, in the context of the different tapasin mutations. In contrast to the ERp57-TAP1 interaction which was completely dependent on presence of tapasin (Figure 3B and 3F), PDI interaction with TAP was readily detected in the presence of both soluble tapasin constructs, indicating a tapasin-independent TAP-PDI interaction (Figure 6A and 6B). This result was confirmed in tapasin-deficient M553 cells (Figure 6C). In fact, the absence of tapasin in association with TAP appeared to enhance PDI-TAP associations (Figures 6A and 6C). TAP1-PDI interactions were also verified in other human cell types with an intact MHC class I assembly complex, such as HeLa cells (Figure 6D) and the TAP1-deficient SK-mel-19 cells that express TAP1 exogenously (data not shown).
Both PDI and ERp57 are associated with TAP Figures 3 and and6),6), but tapasin conjugate formation is reported to be specifically induced with ERp57 (15). We re-examined whether PDI is able to form disulfide-linked conjugates with the tapasin constructs used in this study. For these analyses, IFN-γ-treated cells were lysed in the presence of MMTS (without cross-linker), and immunoblotting analyses of lysates or anti-tapasin immunoprecipitates were undertaken following SDS-PAGE under reducing or non-reducing conditions.
Whereas the major ERp57 conjugate observable in cell lysates corresponded to the size of the tapasin-ERp57 conjugate (Figure 7B, lanes 6 and 8), several PDI conjugates were observable (Figure 7C, lanes 6 and 8), suggesting that mixed disulfides between PDI and cellular proteins are more stable or more numerous than those with ERp57. That tapasin is a uniquely-detectable cellular substrate of ERp57 has previously been documented, and the stable disulfide-linkage between tapasin and ERp57 is attributed to an additional non-covalent interaction between the proteins (15). In the absence of chemical cross-linkers (not used to derive the data of Figure 7) or trapping mutants of ERp57, mixed disulfide formation with ERp57 is previously only reported in the context of tapasin (15) or viral glycoproteins (23).
In immunoprecipitations with the tapasin-specific antibody (Pasta-1) (6), as expected, conjugates of 110 kDa were observed under non-reducing conditions with both full-length and soluble tapasin, and were detectable in immunoblotting analyses with anti-tapasin as well as anti-ERp57 (Figures 7A and 7B). These conjugates were C95-dependent (data not shown). Additionally, a PDI conjugate of 110 kDa was observed in immunoprecipitates from cells expressing soluble tapasin (Figure 7C, lane 1). A much weaker signal for PDI was obtained under reducing conditions, indicating that the anti-PDI antibody preferentially detects the conjugated form of PDI (Figure 7C, lane 1 compared to lane 2) (this band is more readily detectable in longer exposures of the PDI blot; data not shown). The PDI conjugate of 110 kDa was undetectable or very weak in the context of full-length tapasin (Figure 7C, lane 1 compared to lane 3), consistent with previous findings that full-length tapasin PDI conjugates are not readily detectable (15). The 110 kDa PDI conjugate in the context of soluble tapasin was not observed in Pasta-1 immunoprecipitates of cells expressing soluble or full length tapasin(C95A) (data not shown).
To further characterize whether PDI did form conjugates with soluble tapasin, and to quantify the extent of conjugate formation, elution profiles of soluble tapasin and soluble tapasin(C95A) were compared following lysis of the corresponding M553 cells in MMTS (without cross-linker), anti-FLAG based purifications of soluble tapasin and associated proteins, and further gel-filtration based separation of free tapasin or its conjugates. Relevant fractions were analyzed by reducing (Figures 8A, left panels) or non-reducing (Figure 8A, right panels) SDS-PAGE and immunoblotting analyses for tapasin, ERp57 and PDI. Soluble tapasin was indeed in conjugate with both ERp57 or PDI in fractions B2-C2, and fractions C3-C5 contained majority of free tapasin (Figure 8A). Bands corresponding to a tri-molecular complexes (Tapasin-ERp57-PDI) were not visualized in the non-reducing gels (Figure 8A), suggesting that only heterodimers of soluble tapasin-ERp57 and soluble tapasin-PDI were present. Although weak signals for a few higher molecular weight band above 110 kDa were detected in the ERp57 blots (Figure 8A, fractions B1-C1) these bands were absent in the PDI blots suggesting that they do not represent a tri-molecular complex between tapasin-ERp57-PDI. The identity of these higher molecular weight bands in the ERp57 blot is unknown. Neither ERp57 nor PDI was visualized in eluates from cells expressing soluble tapasin(C95A) (data not shown), confirming that both conjugates with tapasin were C95-dependent. These findings suggest that PDI-soluble tapasin conjugates are unlikely to be folding intermediates, but rather represent a deviation from the specificity of tapasin conjugation to ERp57.
Soluble tapasin purified from lysates of M553 cells was mainly observed to migrate at the expected position for tapasin-ERp57 or tapasin-PDI conjugate (Figure 8A, fraction B2-C2) or tapasin monomer (Figure 8A, fraction C3-C5) and only a small amount was observed in the void volume of the column (Figure 8A, fraction B12-B11). Soluble tapasin(C95A) migrated at the expected position of monomer, and only a small fraction was present in the column void volume (data not shown). Thus, at the levels expressed in M553 cells, neither soluble tapasin nor soluble tapasin(C95A) derived from cell lysates appeared to be significantly aggregated.
The relative amounts of PDI and ERp57 that were associated with soluble tapasin were quantified by two different methods. Fractions containing conjugates were pooled following the gel filtration chromatography, concentrated, separated by SDS-PAGE and visualized by coomassie staining (Figure 8B). Mass spectrometric analyses of band 1 yielded several proteins including ERp72 and GRp78. Mass spectrometric analysis of band 2 yielded ERp57 and PDI as the two most abundant proteins. The averaged intensity of the ten most abundant ERp57 peptides was 3.2×107, 7.5 fold higher than the averaged intensity of the ten most abundant PDI peptides (4.2×106) (Supplementary Figure 1). This corresponds to PDI representation at a level of 13.3% relative to ERp57. Mass spectrometric quantifications of the relative abundance of ERp57 and PDI were verified by titration of tapasin conjugates against known amounts of purified ERp57 and PDI in immunoblotting analyses (Figure 8C). Band intensities of the purified PDI and ERp57 standards were plotted as a function of protein concentration, linearly fitted, and the relative concentration of each protein in the conjugates calculated. These analyses suggested PDI representation at a level of 12% relative to ERp57. Together, these analyses indicate that although soluble tapasin is preferentially associated with ERp57, conjugation to PDI is observable at a level that is not insignificant.
We previously reported that purification of soluble tapasin from supernatants of CHO cells over-expressing soluble human tapasin and cultured at 37°C yielded protein that was largely aggregated (Figure 9A, dashed chromatogram). Aggregation could be suppressed by culturing the cells at 30°C (10). Surprisingly, soluble tapasin purified from lysates of the CHO cells cultured at 37°C and lysed in the presence of MMTS yielded two major peaks (Figure 9A, solid chromatogram), corresponding to tapasin-oxido-reductase conjugates (peak 2, Figure 9B) or free-tapasin (peak 3, Figure 9B), rather than aggregated protein in the void volume (peak 1, Figure 9A). SDS-PAGE analyses of proteins present in peak 2 revealed the presence of several protein bands in addition to tapasin. Mass spectrometric analyses of bands from peak 2 indicated the presence of peptides corresponding to both ERp57 and PDI, as well as those corresponding to GRP78 and ERp72. A subset of these bands were partially displaced (Figure 9B, peak 2, lane 1 compared to lane 2) when the peak was analyzed following soluble tapasin purification from lysates of CHO cells co-expressing soluble tapasin and human ERp57 lacking an ER-retention motif (Figures 9C and 9D). Regardless of human ERp57 co-expression, cell lysate-derived soluble tapasin was present as conjugates or free tapasin monomers, whereas the cell supernatant-derived soluble tapasin was largely aggregated (Figures 9A and 9D). In the absence of human ERp57 co-expression, soluble tapasin associated with endogenous hamster cell-derived oxido-reductases and chaperones, and some of the interactions persisted despite human ERp57 co-expression. Thus, ER-lumenal interactions were critical for inhibition of soluble tapasin aggregation, and soluble tapasin formed conjugates with several oxido-reductases in hamster cells.
To further study the effect of ERp57 on calreticulin recruitment to tapasin, we isolated tapasin from lysates of CHO cells expressing human tapasin alone or human tapasin and ERp57 (Figure 9E). CHO cells expressing tapasin alone were lysed in the presence and in the absence of MMTS, whereas CHO cells expressing both human tapasin and human ERp57 were lysed in the presence of MMTS. The three sets of purified proteins were analyzed by immunoblotting analyses with antibodies directed against tapasin, ERp57 and PDI. ERp57 representation in eluates from the anti-FLAG column correlated with human ERp57 co-expression, as well as the presence of MMTS during cell lyses (Figure 9E, ERp57 blot, lanes 7–9 compared to lanes 1–6). Conversely, human ERp57 co-expression reduced PDI association with tapasin (Figure 9E, PDI blot, lanes 7–9 compared to lanes 4–6). In the absence of MMTS during lysis, recovery of both ERp57 and PDI were relatively inefficient or not observable (Figure 9E, ERp57 and PDI blots, lanes 1–3 compared to lanes 4–9). The purified protein preparations or BSA were immobilized in 96 well micro-titration plates, and calreticulin binding to each protein or BSA was compared by ELISA (Figure 9F). Calreticulin showed more binding to each tapasin preparation compared to the control protein BSA (Figure 9F). The preparation that had the highest level of ERp57 representation displayed the strongest binding to calreticulin (Figure 9F). Thus, in these in vitro binding assays as in cells, ERp57 binding to tapasin enhances the recruitment of calreticulin. As per the data of Figures 4A and 4B, it appears that the ERp57-associated form of calreticulin must also be able to interact with tapasin-associated MHC class I. It is possible that this binding contributes to the observed enhancement in stability of the tapasin-MHC class complex (Figures 4A and 4B) in the context of tapasin-ERp57 complexes.
The studies described here demonstrate that weak intrinsic interactions between the lumenal domains of tapasin and MHC class I are strongly enhanced by the presence of the transmembrane and cytosolic domains of tapasin, and also enhanced by tapasin-ERp57 binding (Figures 2 and and4).4). The inability to recruit ERp57 strongly impairs the functional activity of soluble tapasin(C95A), which is a weak inducer of MHC class I assembly (Figure 1). While the C95A mutation also reduces the functional activity of full-length tapasin, significant ERp57 binding-independent functional activity is measurable with full-length tapasin(C95A), which is not strongly correlated with tapasin-mediated stabilization of TAP (Figures 1 and and5).5). The tapasin(C95A) mutation also impacts the efficiency of calreticulin recruitment to MHC class I molecules (Figure 4). While ERp57 and calreticulin binding to TAP are strongly dependent on or linked to tapasin(C95), PDI association with TAP is tapasin-independent, and in fact appears to be enhanced by the absence of tapasin or a fully-functional tapasin molecule. These findings indicate distinct modes of binding and distinct functions for ERp57 and PDI within MHC class I assembly complexes. Significant C95-dependent conjugation of PDI to tapasin is only detectable in the context of soluble tapasin, studies that point to the importance of full-length tapasin for enhancing specificity of interactions within MHC class I assembly complexes.
The panel of mutants described here were also characterized in a recent report from the Cresswell laboratory (24), published while this manuscript was under revision. While the two studies come to similar conclusions about the importance of both tapasin-ERp57 binding and the transmembrane/cytosolic regions of tapasin in its interactions with MHC class I and functional activities, there are some notable differences between the studies. Our studies in M553 cells suggest that the deletion of the transmembrane/cytosolic domains of tapasin had a more significant impact on the efficiency of tapasin-MHC class I binding than did ERp57 conjugation to tapasin (Figures 2 and and4).4). Immunoprecipitations with the anti-tapasin antibody by the Cresswell laboratory suggested a more critical role for tapasin-ERp57 binding in HLA-B*4402 recruitment to tapasin, following analyses of various tapasin mutants in 721.220/HLA-B*4402 cells (24). Additionally, despite the lower efficiency of soluble tapasin-class I complex formation, we were able to detect complexes between soluble tapasin(C95A) and MHC class I heavy chains (Figures 2 and and4),4), whereas these complexes were not detectable in the other study (24). The observed differences may relate to the use of IFN-γ in our assays, which upregulates TAP as well as MHC class I subunits, promoting several of the interactions that are dependent on one or both of these proteins. TAP expression levels are very relevant to interactions mediated by full-length tapasin, and thus analyses under conditions where TAP expression is not limiting are important. It is alternatively possible that there are allele-specific or other cell type-specific differences in requirements of ERp57 associations vs. the tapasin transmembrane/cytoplasmic domains for efficient tapasin-MHC class I binding. Our analyses examined tapasin binding to endogenously-expressed MHC class I allotypes of M553 cells, which include HLA-A28, HLA-B*5701 and HLA-B*5001 (16). The Cresswell laboratory study examined interactions of exogenously expressed HLA-B*4402 in 721.220 cells which lack endogenous HLA-A and HLA-B, but express an endogenous HLA-C. It is possible that binding of tapasin to HLA-B*4402, a highly tapasin-dependent allotype (25), has more stringent requirements for accessory ER chaperones.
Our data indicate that tapasin promotes MHC class I assembly by two major mechanisms, involving (i) an ERp57-independent mode of tapasin function that requires the transmembrane and cytosolic domains of tapasin and involves associations with TAP (Figure 10A) and (ii) an ERp57-dependent mode of tapasin function requiring the ER luminal domains of tapasin and associations with calreticulin (Figure 10B). The ERp57-independent mode of tapasin function may involve the editing functions of tapasin which optimizes its peptide repertoire, as suggested using fos-jun tethered tapasin-MHC class I complexes (11). Furthermore, localization of MHC class I in the vicinity of the peptide source, TAP, could account for the ERp57-independent activity of tapasin. In some species such as chicken, which lack the counterpart of C95, the ERp57-independent mode of tapasin function may be the sole mechanism of tapasin function (26). Importantly, while the ERp57-independent mode of function requires the transmembrane and cytosolic domains of tapasin (Figure 1), the role of these domains extends beyond the previously suggested functions for these domains in TAP stabilization (4) (Figures 2, ,4,4, ,55).
Tapasin-dependent recruitment of ERp57 enhanced the ability of human tapasin to mediate MHC class I assembly (Figure 1), even under conditions where tapasin-MHC class I interactions were relatively weak (soluble tapasin; Figures 2 and and4).4). ERp57 and calreticulin interact outside of the MHC class I assembly complex (19), and higher representation of ERp57 in conjugation with purified tapasin correlated with increased calreticulin binding in vitro (Figure 9F). Tapasin-C95-mediated recruitment of ERp57 and enhanced recruitment of calreticulin contributed to enhanced calreticulin-MHC class I binding and increased stabilities of MHC class I heavy chains (Figure 1A, 1B, ,22 and and4).4). Thus calreticulin is likely an important player in the ERp57-dependent functional activity of tapasin, but further studies are needed to understand how calreticulin interacts with MHC class I and tapasin, and the extent of its contribution to the functional activity of tapasin. It is widely believed that binding monoglucosylated glycans is sufficient to recruit calreticulin to substrates. However, the findings reported here indicate that multiple tapasin-dependent interactions rather than glycan binding alone contribute to efficient recruitment of calreticulin to the MHC class I assembly complex. Further understanding of the mechanisms of calreticulin recruitment into the MHC class I assembly complex will allow insights into whether similar mechanisms could be used promote recruitment of ER chaperones to enhance assembly of other multi-protein complexes, particularly of viruses.
The recent crystal structure analysis of a soluble tapasin-ERp57 heterodimer showed that both catalytic domains of ERp57 (a and a′ domain) are involved in tapasin binding (7). ERp57 residues involved in binding to tapasin are conserved in PDI (7). Specificity of tapasin-ERp57 heterodimer formation was attributed to differences in inter-domain orientations between different PDI family members. A shorter distance was noted between Cα atoms of C57 and C406 of ERp57 in the tapasin-ERp57 heterodimer (7), compared to counterpart PDI residues in the yeast PDI structure which was crystallized as free protein rather than in a substrate-conjugated form (27). Our studies indicate that soluble tapasin per se is not fully specific for ERp57 conjugation as conjugates between soluble tapasin and PDI were also readily detectable, although at a level approximately 7.5 fold lower than that of ERp57 (Figures 7, ,88 and Supplementary Figure 1).
How might TAP binding inhibit tapasin-PDI binding and promote tapasin-ERp57 interaction specificity? Cooperative interactions that stabilize the heptameric complexes involving TAP1/TAP2/tapasin/MHC class I/calreticulin/ERp57 via multimodal binding between the different components, must impose a strong structural constraint on the complete complex, that is absent in the corresponding soluble tapasin complexes. The lack of complex stability in soluble tapasin complexes is likely directly responsible for the reduction in oxido-reductase conjugation specificity. Thus, structural constraints of the complete MHC class I assembly complex contribute to the specificity of tapasin-ERp57 binding.
We show that PDI is likely a bonafide component of the TAP complex not by its ability to conjugate with tapasin, but rather via its ability to bind to TAP (Figure 6). TAP-associated PDI has been implicated in the maintenance of oxidized forms of MHC class I heavy chains, as higher levels of reduced heavy chains were found in PDI-depleted cells (13). However, the majority of TAP-associated heavy chains were oxidized in PDI-depleted cells (13), which suggests that PDI did not have a major influence on maintaining oxidized forms of TAP-associated heavy chains. The possibility of a function for TAP-associated PDI in mediating tapasin oxidation is suggested by the findings that TAP-associated forms of tapasin are more oxidized compared to Pasta-1 associated forms of tapasin (data not shown); however, the oxidized form of tapasin found in complex with TAP may also simply result from steric constraints of reductase access to TAP-associated forms of tapasin, rather than PDI-dependent catalytic oxidation. The relevant role of TAP-associated PDI may relate to the sequestration of TAP-translocated peptides, to prevent their premature retro-translocation, as previously suggested (13, 28). Alternatively, since the presence of functional forms of TAP-associated tapasin reduced the extent of PDI-TAP binding (Figures 6A and C), it is also possible that PDI functions in the retro-translocation and degradation of components of the TAP complex, including TAP itself, when not stabilized by a complete repertoire of associated factors. PDI binding to TAP may involve interactions with surface hydrophobic patches on PDI (27). Further functional studies will be needed to assess whether TAP-associated PDI represents a specific mechanism for mediating peptide retention in the ER, or PLC component degradation. It is a formal possibility that the observed TAP-PDI complex represents a more non-specific hydrophobic interactions-based binding.
Our studies also allowed a direct assessment of the effects of ERp57 and TAP binding on tapasin folding. Abrogation of C95-mediated conjugate formation did not induce significant soluble tapasin(C95A) aggregation in cell lysates, and soluble tapasin (C95A) migrated predominantly as monomers (data not shown). However, soluble wild type tapasin was aggregated when purified from cell supernatants, even when co-expressed with ERp57 (Figures 9A and 9D), suggesting that in addition to oxido-reductase binding, interactions with ER lumenal factors such as calreticulin and GRP78 may be important for maintaining tapasin stability and inhibiting its aggregation (Figure 9B).
In summary, we show that tapasin functions by two major mechanisms, involving (i) binding of MHC class I molecules via interactions involving its ER lumenal and transmembrane/cytosolic domains and (ii) recruitment of ERp57 and calreticulin into the MHC class I assembly complex (Figure 10). Additional studies are required to understand whether the tapasin transmembrane and cytosolic domains directly impact tapasin-MHC class I binding, or whether TAP/tapasin binding impacts the efficiency of tapasin-MHC class I binding. Alternatively, the transmembrane domain of tapasin could induce binding merely by increasing the local concentrations of tapasin in membranes containing MHC class I. Each of these possibilities remains to be further investigated. Additionally, whether the ERp57-independent functional activity of tapasin results in part from its ability to promote physical interactions between MHC class I and TAP (thus achieving a higher local concentration of peptides) also remain to be investigated. Furthermore, modes of calreticulin interaction and function remain to be defined, as also the precise function of TAP-associated PDI.
M553 cells (obtained from Dr. Naveen Bangia (16) and HeLa cells were grown in RPMI 1640 and DMEM (Invitrogen) respectively, supplemented with 10% FBS and penicillin-streptomycin. Chinese hamster ovary (CHO) cells expressing tapasin or tapasin and ERp57 were grown in alpha-MEM media (Invitrogen) supplemented with 5% dialyzed FBS and antibiotics-antimycotics (Invitrogen).
The construction of retroviral vectors encoding tapasin and influenza hemagglutinin(HA)-tagged HLA-A2 has been previously described (22, 29). The C95A mutant of full-length tapasin was generated using the QuikChange site-directed mutagenesis kit and full length tapasin in the pAcUW51 vector using the following oligonucleotides, 5-ACCCCCGCGCAGAACGCCCCGCGGGCCCTG GAT-3 and5-ATCCAGGGCCCGCGGGGCGTTCTGCGCGGG GGT-3. Mutant clones were sequenced and ligated into pMSCV-puro. The C95A mutant of soluble FLAG-tagged tapasin was generated by site-directed mutagenesis of soluble tapasin in the pBJ5-GS vector (10) using the oligonucleotides described above. Mutant or wild type soluble tapasin were sequenced and ligated into pMSCV-puro. Viruses were generated as previously described (30, 31) using BOSC cells, and used to infect M553 cell. Retrovirus infectivity to human cells was achieved by pseudotyping the retroviral envelope with the vesicular stomatitis virus G protein. M553 cells were transduced with retroviruses encoding HA-HLA-A2, and selected by treatment with 1 mg/ml Geneticin (Invitrogen), and maintained in 0.5 mg/ml Geneticin (this cell lines is designated as M553/A2). After verifying HA-HLA-A2 expression by flow cytometric analyses and by immunoblotting analyses of cell lysates, M553 or M553/A2 cells were transduced with the different tapasin-encoding retroviruses, and selected by treatment with 1 μg/ml puromycin (Sigma). Cells were maintained in 0.5 μg/ml puromycin.
2×106 cells were washed three times with FACS buffer (PBS pH 7.4 containing 1% FBS). Cells were then incubated with anti-HA ascites (Covance Scientific) or W6/32 ascites fluid (a human MHC class I heterodimer-specific antibody) (32) at a 1:250 dilution for 1h on ice. Following this incubation, cells were washed three times with FACS buffer. Cells were then incubated for 1h on ice with FITC-conjugated goat anti-mouse IgG (Jackson Laboratories) at 1:250 dilutions, followed by three washes with FACS buffer. Cells were then analyzed using a FACS-Canto cytometer and WinMDI software (Joe Trotter, The Scripps Institute, Flow Cytometry Core Facility).
Three-six confluent 100 × 20 mm plates (4x×06 cells per plate) of M553/A2 or M553 cells expressing different tapasin constructs were treated with 500 IU IFN-γ (PeproTech) for 48h and processed for different immunoprecipitation as described below. IFN-γ treatment of cells was used to up-regulate TAP and promote TAP-mediated binding interactions.
To study tapasin and MHC class I binding to various proteins (Figure 2 and and4)4) the IFN-γ treated M553/A2 or M553 cells were harvested and washed in PBS pH 7.4. The cells were then incubated with membrane permeable cross-linker (4mM DTBP) in PBS pH 8.2 for 1h at RT followed by incubation with 1 M Tris pH 7.4 for 15 min at RT to quench the free cross-linker. The cells were then washed once with PBS pH 7.4 and lysed in lysis buffer (25 mM Tris, 150 mM NaCl, 0.2 mM CaCl2, 1% digitonin and protease inhibitors) for 1h on ice. The lysates were immunoprecipitated with Pasta-1 or HC10 (33), overnight at 4°C and processed as described above. Immunoblotting analyses of the Pasta-1 and HC10-based immunoprecipitations were undertaken with the following antibodies; rabbit anti-tapasin antisera (antisera generated against an N-terminal peptide of tapasin) obtained from Dr. Ted Hanson (Washington University, St. Louis, Mo) the HC10 antibody (33) for MHC class I heavy chains, rabbit polyclonal anti-ERp57 antibody (Santa Cruz Biotehnology, Inc.) and rabbit polyclonal anti-calreticulin (Stressgen). The secondary antibodies used were Peroxidase AffiniPure goat anti-rabbit IgG (heavy and light chain) or Peroxidase AffiniPure goat anti-mouse IgG (heavy and light chain) (Jackson ImmunoResearch Laboratories, Inc.).
TAP binding was analyzed both in the absence (M553 cells) and in the presence (M553-A2 cells) of cross-linking agent. For these analyzes, IFN-γ treated M553 cells expressing different tapasin constructs were harvested and washed in PBS pH 7.4. The cells were then lysed in lysis buffer (25 mM Tris, 150 mM NaCl, 0.2 mM CaCl2, 1% digitonin and protease inhibitors) for 1h on ice. M553-A2 cells were first incubated with 4 mM DTBP in PBS pH 8.2 for 1h at RT. The cells were then washed once with PBS pH 7.4 after quenching the free cross-linker and lysed as described for the M553 cells. Lysates were immunoprecipitated with anti-TAP1 antisera (34) obtained from Dr. M. J. Androlewicz (Lee Moffit Cancer Center, Tampa, Fl) and processed as above. HeLa cells were also processed for TAP1 immunoprecipitation as above. For anti-TAP1 immunoprecipitations, the following antibodies were used: mouse anti-TAP1 (SB05 212) or mouse anti-tapasin (35) obtained from Dr. Soldano Ferrone, HC10 (33) to detect MHC class I heavy chains, mouse anti-PDI monoclonal antibody (Abcam), mouse anti-ERp57 monoclonal antibody (Abcam), goat polyclonal anti-calreticulin (Santa Cruz). The secondary antibodies used were Peroxidase AffiniPure goat anti-mouse IgG (heavy and light chain) or Peroxidase AffiniPure bovine anti-goat IgG (heavy and light chain) (Jackson ImmunoResearch Laboratories, Inc.).
To study rates of intracellular trafficking of MHC class I in M553 or M553 cells expressing full length tapasinC95A, cells were pulsed with 0.1 mCi [S35]methionine-cysteine and chased for 0, 30, 60 and 120 minutes. The cells were then lysed in 1% digitonin, pre-cleared for 2h with protein G beads and lysates were immunoprecipitated with W6/32 antibody. The IPs were processed as described for other IPs and divided into two equal aliquots. One set of aliquots were treated with Endo-Hfor 1h at 37°C. Following Endo-H digestion the samples were resolved by SDS-PAGE and developed by phosphorimaging analysis.
(Figure 7): For these analyses, IFN-γ treated cells were harvested and incubated with 2 mM MMTS (Methyl methanethiosulfonate, Pierce) in PBS pH 7.4 for 10 min on ice. MMTS was used alone and not used in combination with DTBP. The cells were then pelleted and lysed in lysis buffer containing 1% digitonin and 10 mM MMTS. The lysates were centrifuged at 4°C for 30 min and protein concentration in lysates was determined by a BCA protein assay (Pierce, Thermo Scientific). Equal microgram amounts of lysates were incubated with anti-tapasin (Pasta-1; (6) obtained from Dr. Peter Cresswell) overnight at 4°C. Following overnight incubation, immunoprecipitates were isolated with protein G beads for 2h at 4°C. The beads were washed three times with lysis buffer containing 0.1% digitonin, resuspended in reducing or non-reducing SDS-PAGE buffer and boiled. The proteins were separated by SDS-PAGE and analyzed by western blotting using the ECL Plus Kit (Amersham, GE Healthcare).
Twelve confluent 100 × 20 mm plates of M553 cells expressing wild type soluble tapasin or soluble tapasin(C95A) were harvested and washed with PBS pH 7.0. The cells were lysed in lysis buffer (50 mM sodium phosphate, pH 7.0, 10 mM MMTS, 1% Triton X-100 and protease inhibitors) for 1h on ice. MMTS was used alone and not used in combination with DTBP. For CHO cell experiments, 20 confluent plates of CHO cells (100 X 20 mm) expressing tapasin or tapasin and ERp57 were harvested and washed with PBS pH 7.0. The cells were lysed as above. The lysates were centrifuged at 4°C for 30 min and loaded onto an anti-FLAG agarose column (Sigma). The column was washed with 50 mM sodium phosphate buffer containing 1% Triton X-100 and 2 mM MMTS, then with 50 mM sodium phosphate buffer alone, and proteins eluted using FLAG peptide. Proteins were concentrated and analyzed by gel filtration chromatography. Different fractions from gel filtration chromatography were analyzed for the presence of tapasin, ERp57 and PDI by immunoblotting analyses. To determine the relative amounts of PDI and ERp57 that were associated with soluble tapasin following gel filtration analysis, two different methods were used. The soluble tapasin conjugates or different amounts of purified ERp57 and PDI proteins (ProSpec) were separated by SDS-PAGE and immunoblotting analyses undertaken with anti-PDI (Santa Cruz Biotechnology, Inc.) and anti-ERp57 (Santa Cruz Biotechnology, Inc).The blots were scanned and relative band intensities were quantified using ImageQuant software. Signals for the purified ERp57 and PDI were plotted as a function of protein concentration, and analyzed by Prism to estimate the amounts of ERp57 and PDI present in the conjugate fractions. In a second analysis the complexes were resolved by SDS-PAGE, coomassie blue stained, and indicated bands excised and submitted for mass spectrometric analyses. Relative amounts of each protein were estimated by comparing average intensities of ten most abundant peptides for each protein (Supplementary Figure 2).
For proteins purified from CHO cells, eluted proteins were separated by SDS-PAGE, coomassie blue stained, and indicated bands were subjected to mass spectrometric analysis. All mass spectrometric analyses were conducted at the Taplin Biological Mass Spectrometry Facility (Harvard Medical School). Protein purifications from CHO cells supernatants were undertaken as previously described (10). Full-length human calreticulin (hCRT) lacking the signal sequence was generated using ligation-independent cloning (LIC). The sequence was amplified by PCR using primers that allowed for subsequent LIC into the pMCSG7 vector. The following oligonucleotide primers were used; Forward: 5′-TAC TTC CAA TCC AAT GCT GCC GCA GAC CCT GCC ATC-3′ and Reverse: 5′-TTA TCC ACT TCC AAT GTT ACA GCT CAT CCT TGG CTT-3′. The purified constructs contain an N-terminal MHHHHHHSSGVDLGTENLYFQSNA fusion sequence. The protein was induced at room temperature overnight and purified using nickel beads according to the standard protocol. The protein was further purified by gel filtration column and stored at −20°C in 10% glycerol.
ERp57 lacking the ER retention sequence (KDEL) was amplified using following oligonucleotides : 5-AAGCGGCCGCCAGCTGCCACCATGCGCCTC CGCCGC-3 and ATGCGGCCGCGGTACCTTATGCCTTCTTCTT CTT-3 and cloned into the pBJ5GS vector using BamHI and Not1 sites. CHO cells were transfected with pBJ5-GS vector expressing tapasin and ERp57 by the lipofectin protocol and cells were selected with methionine sulfoximine as previously described (10). Clones surviving selection were screened for expression of tapasin and ERp57 by immunoprecipitations of cell supernatants with anti-FLAG (Sigma) and anti-ERp57 monoclonal antibody (Abcam) antibodies, following metabolic labeling of cells (10). Cells expressing both proteins were clonally amplified, and clones re-tested for expression of both proteins by metabolic labeling and co-immunoprecipitation analyses, prior to large-scale amplification and protein purifications as described above.
For these assays, tapasin was purified from lysates of CHO cells under three different conditions (a) CHO-tapasin cells lysed in the absence of MMTS (b) CHO-tapasin cells lysed in the presence of MMTS or (c) CHO-tapasin-ERp57 cells lysed in the presence of MMTS. An anti-FLAG column was used for purifying protein under each condition, as described above. The protein concentrations were estimated by a BCA assay. The ELISA plates (BD Falcon) were coated by overnight incubation at 4°C with 1–2 μg of each type of tapasin, or BSA in PBS pH 7.4. Unbound proteins were removed by washing the wells three times with ELISA buffer (50 mM Tris, 150 mM NaCl, 0.5 mM CaCl2) containing 0.5 % Tween 20 (Sigma). The wells were blocked with 10% FBS in ELISA buffer at 37°C for 1h and excess FBS was removed. To study calreticulin binding to tapasin, 1–2 μg of human calreticulin in ELISA buffer was added and plates were incubated at 37°C for 3h. Unbound calreticulin was removed, anti-calreticulin antibody (goat anti-calreticulin) was added, and plates incubated at 37°C for 1h. The primary antibody was washed, secondary antibody (bovine anti-goat HRP) added, prior to further incubation at 37°C for 1h. After washing the wells three times with ELISA buffer containing 0.5% Tween 20, developer solution (BD Falcon) was added. The developer reaction was stopped by adding 2N H2SO4 and absorbance read at 450 nm.
4.8×107 M553/soluble tapasin cells were lysed in the presence of MMTS (no cross-linker) and immunoprecipitated with anti-FLAG beads. The anti-FLAG eluates were run on a gel filtration column to isolate soluble tapasin and associated proteins. Gel filtration fractions containing ERp57/PDI and soluble tapasin were concentrated, protein separated by SDS-PAGE and coomassie blue staining and indicated bands (Figure 8B) were subjected to mass spectroscopic analysis. The most abundant proteins identified in band 2 (Figure 8B) were ERp57 and PDI. The bar graphs show ten of the most abundant peptides identified for ERp57 (A) and PDI (B). The NL (relative peptide abundance) values of top ten peptides were used to estimate the relative amounts of each protein associated with soluble tapasin.
We thank Christopher Perria for early contributions to the project, and Dr. Naveen Bangia for the M553 cell line and Dr. Kathleen Collins for a retroviral construct encoding HA-tagged HLA-A2. We sincerely thank our colleagues, Dr. Matt Androlewicz, Dr, Michael Brenner, Dr. Peter Cresswell, Dr. Soldano Ferrone and Dr. Ted Hansen for the antibody reagents that were critical for this work. We thank the University of Michigan DNA sequencing core for sequencing analyses, and the hybridoma core for antibody production. This work was supported by a National Institutes of Health grant AI-044115 (to MR) and by financial support from the Michigan Diabetes Research and Training Center and the Rheumatic Diseases Core Center of the University of Michigan.