Qualitative and Quantitative Differences in Peptides Bound to HLA-B27 in the Presence of Mouse versus Human Tapasin Define a Role for Tapasi
http://www.100md.com
免疫学杂志 2005年第12期
Abstract
Tapasin (Tpn) is a chaperone of the endoplasmic reticulum involved in peptide loading to MHC class I proteins. The influence of mouse Tpn (mTpn) on the HLA-B*2705-bound peptide repertoire was analyzed to characterize the species specificity of this chaperone. B*2705 was expressed on Tpn-deficient human 721.220 cells cotransfected with human (hTpn) or mTpn. The heterodimer to 2-microglobulin-free H chain ratio on the cell surface was reduced with mTpn, suggesting lower B*2705 stability. The B*2705-bound peptide repertoires loaded with hTpn or mTpn shared 94–97% identity, although significant differences in peptide amount were observed in 16–17% of the shared ligands. About 3–6% of peptides were bound only with either hTpn or mTpn. Nonamers differentially bound with mTpn had less suitable anchor residues and bound B*2705 less efficiently in vitro than those loaded only with hTpn or shared nonamers. Decamers showed a different pattern: those found only with mTpn had similarly suitable residues as shared decamers and bound B*2705 with high efficiency. Peptides differentially presented by B*2705 on human or mouse cells showed an analogous pattern of residue suitability, suggesting that the effect of mTpn on B*2705 loading is comparable in both cell types. Thus, mTpn has quantitative and qualitative effects on the B*2705-bound peptide repertoire, impairing presentation of some suitable ligands and allowing others with suboptimal anchor residues and lower affinity to be presented. Our results favor a size-dependent peptide editing role of Tpn for HLA-B*2705 that is species-dependent and suboptimally performed, at least for nonamers, by mTpn.
Introduction
The MHC class I molecules bind a large array of peptides, arising mainly from proteasomal degradation of endogenous proteins and present them at the cell surface for recognition by CTL. Because the peptide-binding specificity of MHC I proteins is very broad, the peptide cargo of these molecules requires optimization to ensure selection of ligands with high stability, a feature that is important for immunogenicity (1). This optimization is conducted through a highly organized process of assisted loading, which involves several proteins collectively known as the peptide-loading complex. Besides the nascent MHC I and TAP molecules, this complex is formed by the lectin-like chaperone calreticulin (2), the thiol oxidoreductase ERp57 (3, 4, 5), and the MHC I-dedicated chaperone tapasin (Tpn)3 (2, 6). Numerous studies have demonstrated a pivotal role of Tpn in optimizing peptide binding to MHC I molecules (7, 8), by favoring loading of high affinity ligands (9). Indeed, MHC I molecules synthesized in the absence of Tpn are turned over more quickly and have higher ratios of open to folded conformers (10, 11). This idea was challenged in a recent study (12) reporting that peptide repertoires bound in the absence of Tpn showed comparable or higher overall affinity than those bound with this chaperone. On this basis, it was concluded that Tpn may act as a facilitator of peptide binding, rather than an editor selecting for high affinity ligands.
Tpn bridges MHC I and TAP (13), increases levels of peptide binding to TAP (14, 15), and contributes to the assembly of the peptide-loading complex. The precise mechanism by which Tpn contributes to optimizing the MHC I peptide cargo is unknown, but probably depends on interactions with multiple proteins (16, 17). For instance, the covalent interaction of Tpn with ERp57 is critical for the function of this protein in mediating the establishment of the disulfide bonds in the MHC molecule during folding (18, 19).
MHC I allotypes differ significantly in their Tpn dependency for peptide loading (20, 21). For instance, surface expression of HLA-B*2705 is relatively independent of Tpn, although in its absence B*2705 molecules at the cell surface are less stable (20), suggesting presentation of suboptimal peptides. Indeed, Tpn influences editing (22) and optimization (9) of the B*2705-bound peptide repertoire.
The species specificity of Tpn-mediated interactions is relevant in assessing the suitability of HLA class I expression in mouse cells for T cell Ag presentation and animal disease models. Human Tpn (hTpn) and mouse Tpn (mTpn) share 75% amino acid sequence identity (23, 24). The mouse chaperone is only slightly less efficient than its human counterpart in restoring surface expression of HLA-B5 and HLA-B8 on Tpn-deficient human cells (23), suggesting functional similarity.
The role and species-dependent effects of Tpn in peptide loading are especially relevant in the case of HLA-B27 for reasons related to its strong association with spondyloarthritis (25). First, HLA-B27-mediated peptide presentation might be involved in pathogenesis (26, 27). Second, HLA-B27 transgenic rats (28) and mice (29, 30, 31, 32) are used as animal models for HLA-B27-associated arthritis. Third, HLA-B27 H chain homodimers, which might also play a role in disease, are found at the cell surface following dissociation of MHC-peptide complexes (33, 34, 35). A suboptimal peptide repertoire might favor dissociation of these complexes and formation of homodimers, which might be recognized by autoaggressive T cells (36). This same process may lead to release of 2-microglobulin (2m) that, if trapped in the synovia, might cause inflammation (37).
Despite many studies concerning the role of Tpn in peptide loading, systematic studies that characterize and quantify Tpn-dependent and Tpn-independent peptides are almost lacking. For HLA-B27, one study demonstrated that hTpn had significant quantitative and qualitative effects on the B*2705-bound peptide repertoire (22). More recently, we reported that a natural B*2705 ligand found in human cells but not in mouse B*2705 transfectants was at significantly lower levels in Tpn-deficient human cells reconstituted with mTpn, demonstrating species-dependent modulation of peptide loading for one particular ligand (38).
In the present study, we have addressed the modulation of the B*2705-bound peptide cargo by hTpn and mTpn. To this end, we have comparatively analyzed the B*2705-bound peptide pools from Tpn-deficient human cells reconstituted with hTpn or mTpn, or lacking this chaperone, identified individual ligands differentially expressed in a species-dependent way, and characterized their structural and binding properties as B*2705 ligands.
Materials and Methods
The assay used to measure peptide binding was performed as described (54). Briefly, B*2705-RMA-S transfectant cells were incubated at 26°C for 22 h in RPMI 1640 medium supplemented with 10% heat-inactivated FBS. They were then washed three times in serum-free medium, incubated for 1 h at 26°C with various peptide concentrations without FBS, transferred to 37°C, and collected for flow cytometry after 4 h. HLA-B27 expression was measured using 50 μl of hybridoma culture supernatant containing the mAb ME1. Binding of the natural B*2705 ligand RRYQKSTEL, used as positive control, was expressed as C50, which is the molar concentration of the peptide at 50% of the maximum fluorescence obtained at the concentration range used (10–4 to 10–8 M). Binding of other peptides was assessed as the concentration of peptide required to obtain the fluorescence value at the C50 of the control peptide. This was designated as EC50. Peptides with EC50 < 10 μM were considered to bind with high affinity. EC50 values between 10 and 50 μM were considered to reflect intermediate affinity, and EC50 > 50 μM indicated low affinity.
A previously described (55) cell surface MHC-peptide complex stability assay was used with slight modifications. Briefly, T2-B*2705 cells (5 x 105 cells/well) were incubated overnight at 37°C, in serum-free AIM-V cell culture medium (Invitrogen Life Technologies), in the presence of 100 μM peptide and 100 nM 2m. After washing, cells were incubated for 1 h at 37°C in RPMI 1640, containing 10% FBS (Invitrogen Life Technologies) and brefeldin A (10 μg/ml) to block egress of newly synthesized class I molecules. Cells were washed, and incubation continued in the presence of 0.5 μg/ml brefeldin A at 37°C. Cells were removed at various times (0, 8, 24, and 30 h) and stained with the ME1 mAb, as described above. The decay of B*2705-peptide complexes was determined as follows: percentage of mean linear fluorescence (MLF) remaining = MLFt(+pep) – MLFt(–pep)/MLFt=0(+pep) – MLFt=0(–pep).
Fluorescence values at the various time points were adjusted by linear regression analysis, and only those experiments with R2 0.9 were taken into account. Stability was measured as DC50. This is the time required to obtain 50% of the fluorescence value at t = 0.
Statistical analysis
The 2 test with Yates correction or, for small data sets, the Fisher’s exact test, were used.
Results
Expression and quantification of hTpn and mTpn in .220 transfectant cells
The expression levels of hTpn and mTpn in the B*2705.220 transfectants used in this study were determined by quantitative RT-PCR. The amount of mTpn relative to hTpn, as assessed by this method, was somewhat higher (1.67 ± 0.37) (Fig. 1A).
Expression of the hTpn and mTpn proteins was confirmed by Western blot (Fig. 1B). Two high affinity polyclonal antisera, raised against hTpn or mTpn, respectively, were used. The former antiserum cross-reacted more weakly with mTpn, precluding a quantitative comparison. Titration of cell equivalents using separate antisera for each protein suggested high expression levels of hTpn and mTpn in the corresponding transfectants. Although titration of mTpn was higher, consistent with results from RT-PCR, the blots cannot be formally compared because different antisera were used.
Cell surface stability of HLA-B*2705 is Tpn species dependent
The surface expression of HLA-B*2705 was analyzed on .220 cells transfected either with only B*2705 or with this allotype plus hTpn or mTpn. Expression of the B*2705 heterodimer was clearly but moderately increased in the presence of Tpn, relative to Tpn-deficient cells, as determined with the ME1 and W6/32 mAbs. The effects of hTpn and mTpn were indistinguishable with these antibodies. (Fig. 2A, Table I).
The reliability of assigning species-specific peptide differences depends on the intensity of the corresponding ion peaks. It is possible that ion peaks found only in one peptide pool might escape detection in the other if the corresponding peptides were below the detection levels of our MALDI-TOF analysis. In our comparison, 61 and 64% of the peptides assigned as hTpn- or mTpn-specific, respectively, showed strong signal (50% of the maximal intensity of the MALDI-TOF MS spectrum). Examples of major and minor ion peaks assigned as species-specific differences are shown in Fig. 3.
Quantitative differences in the expression of shared ligands with either hTpn or mTpn was estimated as follows. Shared ion peaks showing 50% or more of the maximal intensity value in each MALDI-TOF MS spectrum were selected. Their intensity was compared with those of their counterparts in the other peptide pool. Ion peaks showing 10-fold or more intensity difference were considered as quantitative differences predominant in the corresponding peptide pool. This procedure provides only an approximate estimation, because MALDI-TOF MS is not quantitative. Nevertheless, control experiments showed that MALDI-TOF MS spectra of equivalent HPLC fractions were largely reproducible (data not shown; also see Refs. 22 and 61). A total of 226 major ion peaks corresponding to shared ligands were compared in this way. Of these, 36 (16%) and 38 (17%) were assigned as peptides expressed at significantly higher levels with hTpn or mTpn, respectively (Table II). Thus, in addition to determining species-specific expression of a limited number of peptides, hTpn and mTpn modulate peptide levels for a significant number of B*2705 ligands in a species-dependent way.
Structural features of shared ligands
The sequence of 46 shared ligands, including 26 nonamers, was determined (Fig. 4). A systematic comparison of residue frequencies among these nonamers with a series of 108 B*2705-bound nonamers previously sequenced from human lymphoid cells (52), showed no statistically significant differences, except for a marginal increase of M8 (p = 0.045). These results are in agreement with the very high sharing between the B*2705-bound peptide repertoires loaded with mTpn or hTpn.
Suboptimal features of B*2705 ligands differentially bound with mTpn
The sequences of 8 and 12 peptides found only with hTpn or mTpn, respectively, were determined (Fig. 5A). Residue usage at the P1, P3, and P anchor positions was compared using the Fisher’s exact test. A few statistically significant differences (p < 0.05) were observed at P1 and P3 between both peptide sets. The most conspicuous one concerned A3. This residue was found in 6 (50%) mTpn-specific and none of the hTpn-specific peptides. A3 has low frequency (5.2%) among natural B*2705 ligands sequenced from human cells (52), so that the high frequency of this residue among mTpn-specific peptides was also statistically significant when compared with this series. In a previous study (62), in which we scanned poly-Ala analogues for binding to B*2705, A3 was less favored than more bulky aliphatic or aromatic residues. In addition, 4 other mTpn-specific peptides had P3 residues less favored than A (E, Q, R). Thus, 10 of 12 (83%) mTpn-specific peptides had a suboptimal P3 residue, relative to 3 (R, T, E) of 8 (37.5%) hTpn-specific peptides. These results indicate that mTpn allows presentation of peptides with weak P3 anchor residues that are less frequently found among B*2705 ligands in the presence of hTpn.
At P, 5 (41.7%) mTpn-specific and none of the hTpn-specific peptides showed a basic residue. The percentage of peptides with C-terminal R or K residues among B*2705-bound peptides is 26.6% (52). Thus, although the difference did not reach statistical significance, our results suggest that peptides with C-terminal basic residues might be under-represented among hTpn-specific ligands. Two mTpn-specific peptides (16.7%) had C-terminal A. This is significantly high, relative to the very low frequency of C-terminal A (2 of 174: 1.1%) among B*2705-bound peptides (52), suggesting that this weak C-terminal anchor is more easily allowed in the presence of mTpn.
The previous comparison suggested that mTpn might allow peptides with suboptimal anchor residues to bind B*2705. To assess this possibility, we compared residue usage among hTpn-specific, mTpn-specific, and shared peptides, using the RF and DMP scores, as defined in Materials and Methods. The results (Table III) showed that both scores were consistently higher for hTpn than for mTpn-specific peptides. Shared nonamers showed global frequency scores (P1-P9 and anchor) higher than mTpn-specific nonamers and similar or slightly lower than the hTpn-specific counterparts. The strongest score differences were mainly at anchor positions. Mouse Tpn-specific nonamers used, on an average, P1 and P3 residues with lower RF than those used by hTpn-specific or shared nonamers, and with mean DMP values <1. In contrast, score differences at P9 were smaller between hTpn- and mTpn-specific nonamers and similar between mTpn-specific and shared nonamers. These results suggest that, in the presence of mTpn, suboptimal nonamers are allowed that are not presented with hTpn. Conversely, some highly suitable nonamers that are presented by B*2705 with hTpn, are not allowed in the presence of mTpn.
The only sequenced hTpn-specific decamer scored higher than the mTpn-specific counterparts, but the difference was less pronounced than for nonamers. Moreover, in contrast to nonamers, shared decamers showed global scores that were either lower (P1-P10 and non-anchor) or only slightly higher (anchor) than for mTpn-specific decamers, suggesting that both mTpn-specific and shared decamers were similarly suitable.
In conclusion, hTpn-specific peptides tend to include highly suitable B*2705 ligands. Mouse Tpn allows loading of a majority of the natural B*2705 ligands, but also of some suboptimal nonamers not found in the constitutive B*2705-bound peptide repertoire of human cells. The influence of mTpn on selection of nonamers and decamers does not seem to be identical.
Distinct behavior of differentially presented nonamers and decamers on binding to B*2705 in vitro
To test the suitability of peptides differentially presented by B*2705 with hTpn or mTpn their binding efficiency was compared in an epitope stabilization assay using B*2705-RMA-S transfectant cells (Table IV). Of four hTpn-specific nonamers tested, three bound with a high efficiency (EC50 < 10 μM), and one bound with intermediate efficiency (10 μM < EC50 < 50 μM). In contrast, the four peptides differentially presented with mTpn showed intermediate or low (EC50 > 50 μM) binding. Cell surface stability, as assessed with a variant of the epitope stabilization assay that measured the decay of B*2705 surface expression on T2 transfectant cells at 37°C as a function of time, showed similar differences. The results correlated with the RF/DMP scores of these peptides, supporting the validity of this scoring approach to assess the suitability of nonamers as B*2705 ligands and demonstrated that nonamers specifically presented by B*2705 with mTpn tend to be suboptimal B*2705 binders.
In contrast, five decamers specifically presented with mTpn bound B*2705 with high efficiency. The surface stability of the mTpn-specific decamers tested in complex with B*2705 was also similar to the hTpn-specific decamer sequenced. This ruled out the possibility that stability differences between the hTpn- and mTpn-specific decamers were masked in the epitope stabilization assay.
Tpn species-dependent peptide levels are modulated based on suitability as B*2705 ligands
The sequences of eight peptides assigned as quantitative differences, six predominant with hTpn and two predominant with mTpn, were determined (Fig. 5B). The nonamers predominant with hTpn had mean RF and DMP values (P1–P9) similar to those of hTpn-specific nonamers and higher than the only sequenced nonamer predominant with mTpn (Table V). This peptide had RF and DMP scores (P1–P9) only slightly higher than those of mTpn-specific nonamers (Table III). Despite a suitable P1 residue, it used an unusual P9 anchor, which had a strong negative effect on scoring. The global RF of the anchor positions for the only sequenced decamer predominant with hTpn was similar to the corresponding value for the hTpn-specific decamer (Table III), due to very suitable anchor residues.
Thus, peptides showing Tpn species-dependent expression levels have global residue frequencies comparable to those of hTpn or mTpn-specific peptides. This suggests that both qualitative and quantitative species-dependent modulation of peptide loading by Tpn are based on structural suitability for binding to B*2705.
Mouse Tpn incompletely restores B*2705 loading in human cells
We next addressed whether mTpn had an intermediate effect between absence and presence of hTpn on HLA-B27 loading or had a distinct editing function. To this end, we compared the B*2705-bound peptide repertoires from B*2705.220 transfectants lacking Tpn or expressing the human or mouse chaperone. The total peptide yield obtained from similar cell numbers (3.5 x 109 cells) was 3-fold higher with hTpn or mTpn than in the absence of this chaperone (Table VI). Systematic peptide comparisons revealed that only 2% of the Tpn-independent B*2705 ligands were not found in the presence of hTpn or mTpn (Fig. 6A, Table VI). In contrast, 26–27% of the B*2705 ligands bound with hTpn or mTpn were not detected in the absence of this chaperone (Table VI). It is unlikely that the observed differences in peptide sharing are due to the lower peptide yields from Tpn-deficient cells, because many differential ion peaks had significant intensity and did not correspond to minor signals that could be missed in a less abundant peptide pool (Fig. 6B).
Discussion
Species-dependent effects of Tpn for HLA class I loading have been reported in several studies concerning mainly B*4402, a strongly Tpn-dependent allotype that is poorly expressed on mouse cells (17, 20). In particular, Tan et al. (17) reported that the B*4402-bound peptide repertoires from .220 transfectants expressing either hTpn or mTpn were different, although this was analyzed only in a qualitative way. In contrast, B*2705 is relatively independent of Tpn (20) and is expressed at high levels on mouse cells. In a recent study (38), a peptide was more efficiently presented on .220 cells expressing hTpn than mTpn. This peptide was not found in B*2705-transfected mouse cells despite being normally produced by the murine 20S proteasome, suggesting that species-dependent effects of Tpn also influenced B*2705 loading.
In this study, we have 1) quantified the effect of the heterologous expression of mTpn in human cells on the cell surface stability and peptide repertoire of B*2705; 2) identified peptides whose presentation was specifically dependent or quantitatively modulated by Tpn in a species-dependent way; 3) demonstrated that residue usage among hTpn-specific peptides is biased toward residues that are frequent among the constitutive B*2705-bound peptide repertoire, whereas among mTpn-specific peptides is biased toward less frequent residues; 4) shown that mTpn-specific nonamers, but not mTpn-specific decamers, are suboptimal B*2705 binders; 5) determined that the effect of mTpn on peptide selection for B*2705 was intermediate between absence and presence of hTpn; and 6) shown that residue usage among peptides specifically presented by B*2705 on mouse cells is similarly suboptimal as for peptides specifically presented on human cells with mTpn, so that homologous interactions of this chaperone with non-MHC proteins in the loading complex do not improve the B*2705 cargo. This suggests a peptide-modulating role of Tpn that is dependent on its direct interaction with B*2705.
Surface expression of B*2705 on Tpn-deficient .220 cells was only slightly lower than with hTpn. Although with mTpn B*2705 levels were indistinguishable from those with hTpn, slightly increased dissociation was compatible with the possibility that B*2705 on the surface of cells containing mTpn presented peptides with suboptimal stability and that these were a small fraction of the B*2705-bound repertoire. An alternative possibility was hypothesized by Zarling et al. (12): Tpn would influence MHC I stability not on the basis of peptide affinity, but by conditioning the conformational state of the class I molecule after peptide binding in a way that remained undefined.
Identification of the peptides bound to B*2705 in the presence of hTpn or mTpn helped us to distinguish between both alternatives and to characterize the effect of heterologous interactions involving Tpn on the B*2705-bound peptide repertoire. First, mTpn allowed binding of 95% of the native B*2705 ligands, which is significantly more than the already large percentage (75%) bound in the absence of Tpn. However, 5% of each peptide repertoire was differentially presented by B*2705 in the presence of hTpn or mTpn, and 30% of shared ligands were expressed at significantly different levels. Thus, there was a clear Tpn species-dependent modulation of B*2705 loading.
The molecular mechanism by which Tpn influences B*2705 loading and cell surface stability may be inferred from the molecular features of differentially presented peptides. We took advantage of our knowledge of many B*2705 ligand sequences, and of the most suitable anchor residues for this allotype, to compare the suitability of structural features between differentially bound peptides. This approach was based on a previous analysis of residue frequencies at every peptide position among nearly 200 natural B*2705 ligands (52). Thus, any amino acid residue at a given peptide position could be scored on the basis of its frequency among natural B*2705 ligands, giving an unbiased indication of the suitability of a peptide for B*2705 presentation in human cells with an intact Ag-processing-loading system.
When this analysis was applied to individual positions of differentially bound peptides, the subset found only with hTpn scored significantly higher than the mTpn-specific subset, particularly at anchor positions. Shared nonamers scored similarly or slightly lower than the hTpn-specific ones, but higher than the mTpn-specific counterparts. However, this was not the case for shared decamers. These results suggested that differential presentation of nonamers is largely based on affinity. Low binding of mTpn-specific nonamers to B*2705 in vitro, confirmed the conclusions derived from residue scoring. The results explain the small but clear decrease in cell surface stability of B*2705 in .220 cells transfected with mTpn, because suboptimal peptides are a small fraction of the total B*2705-bound repertoire in these transfectants.
In contrast, decamers presented with mTpn scored similarly as shared ligands and bound in vitro with high efficiency. Two aspects of this result should be discussed. First, it is possible that the RF and/or DMP scores might be less reliable for decamers than for nonamers, because these are based on >100 peptides, relative to only 39 decamers (52). Moreover, the epitope stabilization assays may not fully reflect peptide binding in vivo. Although we cannot rule out these possibilities, the fact remains that for both nonamers and decamers there is a correlation between residue scoring and binding efficiency in vitro; whereas mTpn-specific nonamers scored lower than shared nonamers and bound poorly, the mTpn-specific decamers scored similarly as shared decamers and bound with high efficiency. Thus, our results suggest that the role of Tpn in mediating peptide loading may not be identical for nonamers and decamers and that optimization of the peptide cargo is more strictly performed for nonamers. The mechanism for this differential effect is currently unclear to us.
Species-dependent effects of mTpn in human cells have been ascribed, in studies with B*4402, not only to bridging MHC and TAP, but also to defective interactions with other components of the peptide-loading complex (16, 17). Mouse Tpn was reported to be as efficient as its human counterpart in stabilizing TAP levels in B*4402.220 transfectants (17). It is possible that alterations in the stability, composition, or stoichiometry of the loading complex, resulting from heterologous interactions of mTpn with calreticulin and ERp57(17), may influence suboptimal B*2705 loading on human cells transfected with the mTpn. However, the fact that suboptimal residue scoring was also observed among B*2705 ligands specifically loaded in mouse cells argues against a dominant influence of heterologous interactions between Tpn and non-MHC I proteins on suboptimal B*2705 loading and supports a direct role of Tpn in this process.
It is generally accepted that Tpn contributes to stabilize a peptide-receptive conformation of the HLA class I molecule, facilitating peptide binding. After this is accomplished, the HLA molecule dissociates from the loading complex and migrates to the cell surface. On the basis of our results, we would like to hypothesize that dissociation of the HLA class I molecule from the loading complex requires a minimum affinity threshold. Because of suboptimal stabilization of the human loading complex by mTpn, the peptide affinity threshold for dissociation of the MHC molecule from this complex might be lower, allowing for suboptimal ligands to reach the cell surface. The fact that some highly suitable peptides were presented only with hTpn is compatible with this model, because their selective expression might result from lower competition among peptides for productive binding to B*2705 when the affinity threshold is higher and, therefore, more restrictive. However, given the complexity and incomplete knowledge of the interactions among components of the peptide-loading complex other alternatives cannot be ruled out. For instance, hTpn and mTpn might stabilize empty class I molecules in slightly different conformations, one more receptive to only high affinity peptides and the other somewhat more permissive.
Our results argue against the view (12) that Tpn functions as a facilitator of peptide binding, rather than as a peptide editor favoring loading of high affinity MHC ligands. This conclusion was largely based on the observation that the average affinity, as measured in vitro with whole peptide pools, of HLA-B8 or HLA-A2 ligands from Tpn-deficient cells was higher or similar, respectively, than the corresponding peptide pools from Tpn-proficient cells. In addition, HLA-B8 ligands from Tpn-deficient or Tpn-proficient cells had similar anchor residues. In contrast, our results with B*2705 show that individual mTpn-specific nonamers tend to have both suboptimal residues and low binding efficiency, fully consistent with an editor role of Tpn. Our results also suggest that this function may be peptide size dependent, because mTpn-specific decamers were good B*2705 binders.
There could be various explanations for the apparent disparity between our results and those of Zarling et al. (12). For instance, the effect of Tpn on shaping MHC class I-bound peptide repertoires might be variable among allotypes. Another possibility is that, in the binding assay using whole peptide pools, the presence of individual ligands with lower affinity might be masked by a dominant effect of peptides with high affinity in the same pool. For instance, many of the peptides bound in the absence of Tpn are also bound in its presence and may have an affinity that is comparable to the Tpn-dependent ligands, as we have previously shown for B*2705 (22). To our knowledge, affinity measurements of individual HLA class I-bound ligands presented in the absence, but not in the presence of Tpn, have not been reported. Whatever the explanation, the differences we observed between nonamers and decamers point out to unsuspected complexities in the way Tpn modulates peptide loading.
In conclusion, our study provides evidence for suboptimal editing of the B*2705-bound peptide repertoire by mTpn, leading to presentation of peptides with less suitable residues and lower affinity, lack of presentation of some ligands that bind with hTpn, and quantitative alterations in the expression level of shared ligands. As a result, surface expressed B*2705 is slightly less stable in the presence of mTpn. These results have obvious implications for animal models of HLA-B27-associated disease. We previously reported that more than half of the peptides differentially presented by B*2705 on human and mouse cells do not arise from species-related protein polymorphism, but from differences in Ag processing (38). Now we have quantified the effect of mTpn as the only heterologous component in the loading complex on peptide presentation by B*2705. Although it remains to be proven, it is likely that the effects on peptide selection will be similar or larger on mouse cells, in which the only human component is HLA-B27. This was suggested by results concerning a single peptide in our previous study (38). Now we have shown that B*2705 ligands arising from conserved sequences between mouse and man, but differentially expressed on human cells have more suitable residues than those differentially expressed on mouse cells, in a pattern very similar to that found for human or mTpn-specific peptides from .220 cells. This suggests that the function of mTpn for B*2705 peptide loading is not improved when other components of the loading complex are also of mouse origin. Therefore, suboptimal peptide presentation, relative to humans, is to be expected when B*2705 is expressed on mouse cells and transgenic mice.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by Grants SAF2002/00125 and SAF2003/02213 from the Ministry of Science and Technology, 08.3/0005/2001.1 from the Comunidad Autónoma de Madrid, and an institutional grant of the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa. A.W.P. is a recipient of the Russell Grimwade Fellowship.
2 Address correspondence and reprint requests to Dr. José A. López de Castro, Centro de Biología Molecular, Severo Ochoa, Facultad de Ciencias, Universidad Autónoma, 28049 Madrid, Spain. E-mail address: aldecastro{at}cbm.uam.es
3 Abbreviations used in this paper: Tpn, tapasin; hTpn, human Tpn; mTpn, mouse Tpn; .220, 721.220; 2m, 2-microblobulin; MS, mass spectrometry; MS/MS, tandem MS; RF, residue frequency; DMP, deviation from mean frequency in the human proteome; m/z, mass-to-charge.
Received for publication April 29, 2004. Accepted for publication April 12, 2005.
References
van der Burg, S. H., M. J. Visseren, R. M. Brandt, W. M. Kast, C. J. Melief. 1996. Immunogenicity of peptides bound to MHC class I molecules depends on the MHC-peptide complex stability. J. Immunol. 156: 3308-3314.
Sadasivan, B., P. J. Lehner, B. Ortmann, T. Spies, P. Cresswell. 1996. Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I molecules with TAP. Immunity 5: 103-114.
Hughes, E. A., P. Cresswell. 1998. The thiol oxidoreductase ERp57 is a component of the MHC class I peptide-loading complex. Curr. Biol. 8: 709-712.
Lindquist, J. A., O. N. Jensen, M. Mann, G. J. Hammerling. 1998. ER-60, a chaperone with thiol-dependent reductase activity involved in MHC class I assembly. EMBO J. 17: 2186-2195.
Morrice, N. A., S. J. Powis. 1998. A role for the thiol-dependent reductase ERp57 in the assembly of MHC class I molecules. Curr. Biol. 8: 713-716.
Ortmann, B., M. J. Androlewicz, P. Cresswell. 1994. MHC class I/2-microglobulin complexes associate with TAP transporters before peptide binding. Nature 368: 864-867.
Momburg, F., P. Tan. 2002. Tapasin—the keystone of the loading complex optimizing peptide binding by MHC class I molecules in the endoplasmic reticulum. Mol. Immunol. 39: 217-233.
Paulsson, K., P. Wang. 2003. Chaperones and folding of MHC class I molecules in the endoplasmic reticulum. Biochim. Biophys. Acta. 1641: 1-12.
Williams, A. P., C. A. Peh, A. W. Purcell, J. McCluskey, T. Elliott. 2002. Optimization of the MHC class I peptide cargo is dependent on tapasin. Immunity 16: 509-520.
Yu, Y. Y., H. R. Turnquist, N. B. Myers, G. K. Balendiran, T. H. Hansen, J. C. Solheim. 1999. An extensive region of an MHC class I 2 domain loop influences interaction with the assembly complex. J. Immunol. 163: 4427-4433.(Laura Sesma*, Bego?a Galo)
Tapasin (Tpn) is a chaperone of the endoplasmic reticulum involved in peptide loading to MHC class I proteins. The influence of mouse Tpn (mTpn) on the HLA-B*2705-bound peptide repertoire was analyzed to characterize the species specificity of this chaperone. B*2705 was expressed on Tpn-deficient human 721.220 cells cotransfected with human (hTpn) or mTpn. The heterodimer to 2-microglobulin-free H chain ratio on the cell surface was reduced with mTpn, suggesting lower B*2705 stability. The B*2705-bound peptide repertoires loaded with hTpn or mTpn shared 94–97% identity, although significant differences in peptide amount were observed in 16–17% of the shared ligands. About 3–6% of peptides were bound only with either hTpn or mTpn. Nonamers differentially bound with mTpn had less suitable anchor residues and bound B*2705 less efficiently in vitro than those loaded only with hTpn or shared nonamers. Decamers showed a different pattern: those found only with mTpn had similarly suitable residues as shared decamers and bound B*2705 with high efficiency. Peptides differentially presented by B*2705 on human or mouse cells showed an analogous pattern of residue suitability, suggesting that the effect of mTpn on B*2705 loading is comparable in both cell types. Thus, mTpn has quantitative and qualitative effects on the B*2705-bound peptide repertoire, impairing presentation of some suitable ligands and allowing others with suboptimal anchor residues and lower affinity to be presented. Our results favor a size-dependent peptide editing role of Tpn for HLA-B*2705 that is species-dependent and suboptimally performed, at least for nonamers, by mTpn.
Introduction
The MHC class I molecules bind a large array of peptides, arising mainly from proteasomal degradation of endogenous proteins and present them at the cell surface for recognition by CTL. Because the peptide-binding specificity of MHC I proteins is very broad, the peptide cargo of these molecules requires optimization to ensure selection of ligands with high stability, a feature that is important for immunogenicity (1). This optimization is conducted through a highly organized process of assisted loading, which involves several proteins collectively known as the peptide-loading complex. Besides the nascent MHC I and TAP molecules, this complex is formed by the lectin-like chaperone calreticulin (2), the thiol oxidoreductase ERp57 (3, 4, 5), and the MHC I-dedicated chaperone tapasin (Tpn)3 (2, 6). Numerous studies have demonstrated a pivotal role of Tpn in optimizing peptide binding to MHC I molecules (7, 8), by favoring loading of high affinity ligands (9). Indeed, MHC I molecules synthesized in the absence of Tpn are turned over more quickly and have higher ratios of open to folded conformers (10, 11). This idea was challenged in a recent study (12) reporting that peptide repertoires bound in the absence of Tpn showed comparable or higher overall affinity than those bound with this chaperone. On this basis, it was concluded that Tpn may act as a facilitator of peptide binding, rather than an editor selecting for high affinity ligands.
Tpn bridges MHC I and TAP (13), increases levels of peptide binding to TAP (14, 15), and contributes to the assembly of the peptide-loading complex. The precise mechanism by which Tpn contributes to optimizing the MHC I peptide cargo is unknown, but probably depends on interactions with multiple proteins (16, 17). For instance, the covalent interaction of Tpn with ERp57 is critical for the function of this protein in mediating the establishment of the disulfide bonds in the MHC molecule during folding (18, 19).
MHC I allotypes differ significantly in their Tpn dependency for peptide loading (20, 21). For instance, surface expression of HLA-B*2705 is relatively independent of Tpn, although in its absence B*2705 molecules at the cell surface are less stable (20), suggesting presentation of suboptimal peptides. Indeed, Tpn influences editing (22) and optimization (9) of the B*2705-bound peptide repertoire.
The species specificity of Tpn-mediated interactions is relevant in assessing the suitability of HLA class I expression in mouse cells for T cell Ag presentation and animal disease models. Human Tpn (hTpn) and mouse Tpn (mTpn) share 75% amino acid sequence identity (23, 24). The mouse chaperone is only slightly less efficient than its human counterpart in restoring surface expression of HLA-B5 and HLA-B8 on Tpn-deficient human cells (23), suggesting functional similarity.
The role and species-dependent effects of Tpn in peptide loading are especially relevant in the case of HLA-B27 for reasons related to its strong association with spondyloarthritis (25). First, HLA-B27-mediated peptide presentation might be involved in pathogenesis (26, 27). Second, HLA-B27 transgenic rats (28) and mice (29, 30, 31, 32) are used as animal models for HLA-B27-associated arthritis. Third, HLA-B27 H chain homodimers, which might also play a role in disease, are found at the cell surface following dissociation of MHC-peptide complexes (33, 34, 35). A suboptimal peptide repertoire might favor dissociation of these complexes and formation of homodimers, which might be recognized by autoaggressive T cells (36). This same process may lead to release of 2-microglobulin (2m) that, if trapped in the synovia, might cause inflammation (37).
Despite many studies concerning the role of Tpn in peptide loading, systematic studies that characterize and quantify Tpn-dependent and Tpn-independent peptides are almost lacking. For HLA-B27, one study demonstrated that hTpn had significant quantitative and qualitative effects on the B*2705-bound peptide repertoire (22). More recently, we reported that a natural B*2705 ligand found in human cells but not in mouse B*2705 transfectants was at significantly lower levels in Tpn-deficient human cells reconstituted with mTpn, demonstrating species-dependent modulation of peptide loading for one particular ligand (38).
In the present study, we have addressed the modulation of the B*2705-bound peptide cargo by hTpn and mTpn. To this end, we have comparatively analyzed the B*2705-bound peptide pools from Tpn-deficient human cells reconstituted with hTpn or mTpn, or lacking this chaperone, identified individual ligands differentially expressed in a species-dependent way, and characterized their structural and binding properties as B*2705 ligands.
Materials and Methods
The assay used to measure peptide binding was performed as described (54). Briefly, B*2705-RMA-S transfectant cells were incubated at 26°C for 22 h in RPMI 1640 medium supplemented with 10% heat-inactivated FBS. They were then washed three times in serum-free medium, incubated for 1 h at 26°C with various peptide concentrations without FBS, transferred to 37°C, and collected for flow cytometry after 4 h. HLA-B27 expression was measured using 50 μl of hybridoma culture supernatant containing the mAb ME1. Binding of the natural B*2705 ligand RRYQKSTEL, used as positive control, was expressed as C50, which is the molar concentration of the peptide at 50% of the maximum fluorescence obtained at the concentration range used (10–4 to 10–8 M). Binding of other peptides was assessed as the concentration of peptide required to obtain the fluorescence value at the C50 of the control peptide. This was designated as EC50. Peptides with EC50 < 10 μM were considered to bind with high affinity. EC50 values between 10 and 50 μM were considered to reflect intermediate affinity, and EC50 > 50 μM indicated low affinity.
A previously described (55) cell surface MHC-peptide complex stability assay was used with slight modifications. Briefly, T2-B*2705 cells (5 x 105 cells/well) were incubated overnight at 37°C, in serum-free AIM-V cell culture medium (Invitrogen Life Technologies), in the presence of 100 μM peptide and 100 nM 2m. After washing, cells were incubated for 1 h at 37°C in RPMI 1640, containing 10% FBS (Invitrogen Life Technologies) and brefeldin A (10 μg/ml) to block egress of newly synthesized class I molecules. Cells were washed, and incubation continued in the presence of 0.5 μg/ml brefeldin A at 37°C. Cells were removed at various times (0, 8, 24, and 30 h) and stained with the ME1 mAb, as described above. The decay of B*2705-peptide complexes was determined as follows: percentage of mean linear fluorescence (MLF) remaining = MLFt(+pep) – MLFt(–pep)/MLFt=0(+pep) – MLFt=0(–pep).
Fluorescence values at the various time points were adjusted by linear regression analysis, and only those experiments with R2 0.9 were taken into account. Stability was measured as DC50. This is the time required to obtain 50% of the fluorescence value at t = 0.
Statistical analysis
The 2 test with Yates correction or, for small data sets, the Fisher’s exact test, were used.
Results
Expression and quantification of hTpn and mTpn in .220 transfectant cells
The expression levels of hTpn and mTpn in the B*2705.220 transfectants used in this study were determined by quantitative RT-PCR. The amount of mTpn relative to hTpn, as assessed by this method, was somewhat higher (1.67 ± 0.37) (Fig. 1A).
Expression of the hTpn and mTpn proteins was confirmed by Western blot (Fig. 1B). Two high affinity polyclonal antisera, raised against hTpn or mTpn, respectively, were used. The former antiserum cross-reacted more weakly with mTpn, precluding a quantitative comparison. Titration of cell equivalents using separate antisera for each protein suggested high expression levels of hTpn and mTpn in the corresponding transfectants. Although titration of mTpn was higher, consistent with results from RT-PCR, the blots cannot be formally compared because different antisera were used.
Cell surface stability of HLA-B*2705 is Tpn species dependent
The surface expression of HLA-B*2705 was analyzed on .220 cells transfected either with only B*2705 or with this allotype plus hTpn or mTpn. Expression of the B*2705 heterodimer was clearly but moderately increased in the presence of Tpn, relative to Tpn-deficient cells, as determined with the ME1 and W6/32 mAbs. The effects of hTpn and mTpn were indistinguishable with these antibodies. (Fig. 2A, Table I).
The reliability of assigning species-specific peptide differences depends on the intensity of the corresponding ion peaks. It is possible that ion peaks found only in one peptide pool might escape detection in the other if the corresponding peptides were below the detection levels of our MALDI-TOF analysis. In our comparison, 61 and 64% of the peptides assigned as hTpn- or mTpn-specific, respectively, showed strong signal (50% of the maximal intensity of the MALDI-TOF MS spectrum). Examples of major and minor ion peaks assigned as species-specific differences are shown in Fig. 3.
Quantitative differences in the expression of shared ligands with either hTpn or mTpn was estimated as follows. Shared ion peaks showing 50% or more of the maximal intensity value in each MALDI-TOF MS spectrum were selected. Their intensity was compared with those of their counterparts in the other peptide pool. Ion peaks showing 10-fold or more intensity difference were considered as quantitative differences predominant in the corresponding peptide pool. This procedure provides only an approximate estimation, because MALDI-TOF MS is not quantitative. Nevertheless, control experiments showed that MALDI-TOF MS spectra of equivalent HPLC fractions were largely reproducible (data not shown; also see Refs. 22 and 61). A total of 226 major ion peaks corresponding to shared ligands were compared in this way. Of these, 36 (16%) and 38 (17%) were assigned as peptides expressed at significantly higher levels with hTpn or mTpn, respectively (Table II). Thus, in addition to determining species-specific expression of a limited number of peptides, hTpn and mTpn modulate peptide levels for a significant number of B*2705 ligands in a species-dependent way.
Structural features of shared ligands
The sequence of 46 shared ligands, including 26 nonamers, was determined (Fig. 4). A systematic comparison of residue frequencies among these nonamers with a series of 108 B*2705-bound nonamers previously sequenced from human lymphoid cells (52), showed no statistically significant differences, except for a marginal increase of M8 (p = 0.045). These results are in agreement with the very high sharing between the B*2705-bound peptide repertoires loaded with mTpn or hTpn.
Suboptimal features of B*2705 ligands differentially bound with mTpn
The sequences of 8 and 12 peptides found only with hTpn or mTpn, respectively, were determined (Fig. 5A). Residue usage at the P1, P3, and P anchor positions was compared using the Fisher’s exact test. A few statistically significant differences (p < 0.05) were observed at P1 and P3 between both peptide sets. The most conspicuous one concerned A3. This residue was found in 6 (50%) mTpn-specific and none of the hTpn-specific peptides. A3 has low frequency (5.2%) among natural B*2705 ligands sequenced from human cells (52), so that the high frequency of this residue among mTpn-specific peptides was also statistically significant when compared with this series. In a previous study (62), in which we scanned poly-Ala analogues for binding to B*2705, A3 was less favored than more bulky aliphatic or aromatic residues. In addition, 4 other mTpn-specific peptides had P3 residues less favored than A (E, Q, R). Thus, 10 of 12 (83%) mTpn-specific peptides had a suboptimal P3 residue, relative to 3 (R, T, E) of 8 (37.5%) hTpn-specific peptides. These results indicate that mTpn allows presentation of peptides with weak P3 anchor residues that are less frequently found among B*2705 ligands in the presence of hTpn.
At P, 5 (41.7%) mTpn-specific and none of the hTpn-specific peptides showed a basic residue. The percentage of peptides with C-terminal R or K residues among B*2705-bound peptides is 26.6% (52). Thus, although the difference did not reach statistical significance, our results suggest that peptides with C-terminal basic residues might be under-represented among hTpn-specific ligands. Two mTpn-specific peptides (16.7%) had C-terminal A. This is significantly high, relative to the very low frequency of C-terminal A (2 of 174: 1.1%) among B*2705-bound peptides (52), suggesting that this weak C-terminal anchor is more easily allowed in the presence of mTpn.
The previous comparison suggested that mTpn might allow peptides with suboptimal anchor residues to bind B*2705. To assess this possibility, we compared residue usage among hTpn-specific, mTpn-specific, and shared peptides, using the RF and DMP scores, as defined in Materials and Methods. The results (Table III) showed that both scores were consistently higher for hTpn than for mTpn-specific peptides. Shared nonamers showed global frequency scores (P1-P9 and anchor) higher than mTpn-specific nonamers and similar or slightly lower than the hTpn-specific counterparts. The strongest score differences were mainly at anchor positions. Mouse Tpn-specific nonamers used, on an average, P1 and P3 residues with lower RF than those used by hTpn-specific or shared nonamers, and with mean DMP values <1. In contrast, score differences at P9 were smaller between hTpn- and mTpn-specific nonamers and similar between mTpn-specific and shared nonamers. These results suggest that, in the presence of mTpn, suboptimal nonamers are allowed that are not presented with hTpn. Conversely, some highly suitable nonamers that are presented by B*2705 with hTpn, are not allowed in the presence of mTpn.
The only sequenced hTpn-specific decamer scored higher than the mTpn-specific counterparts, but the difference was less pronounced than for nonamers. Moreover, in contrast to nonamers, shared decamers showed global scores that were either lower (P1-P10 and non-anchor) or only slightly higher (anchor) than for mTpn-specific decamers, suggesting that both mTpn-specific and shared decamers were similarly suitable.
In conclusion, hTpn-specific peptides tend to include highly suitable B*2705 ligands. Mouse Tpn allows loading of a majority of the natural B*2705 ligands, but also of some suboptimal nonamers not found in the constitutive B*2705-bound peptide repertoire of human cells. The influence of mTpn on selection of nonamers and decamers does not seem to be identical.
Distinct behavior of differentially presented nonamers and decamers on binding to B*2705 in vitro
To test the suitability of peptides differentially presented by B*2705 with hTpn or mTpn their binding efficiency was compared in an epitope stabilization assay using B*2705-RMA-S transfectant cells (Table IV). Of four hTpn-specific nonamers tested, three bound with a high efficiency (EC50 < 10 μM), and one bound with intermediate efficiency (10 μM < EC50 < 50 μM). In contrast, the four peptides differentially presented with mTpn showed intermediate or low (EC50 > 50 μM) binding. Cell surface stability, as assessed with a variant of the epitope stabilization assay that measured the decay of B*2705 surface expression on T2 transfectant cells at 37°C as a function of time, showed similar differences. The results correlated with the RF/DMP scores of these peptides, supporting the validity of this scoring approach to assess the suitability of nonamers as B*2705 ligands and demonstrated that nonamers specifically presented by B*2705 with mTpn tend to be suboptimal B*2705 binders.
In contrast, five decamers specifically presented with mTpn bound B*2705 with high efficiency. The surface stability of the mTpn-specific decamers tested in complex with B*2705 was also similar to the hTpn-specific decamer sequenced. This ruled out the possibility that stability differences between the hTpn- and mTpn-specific decamers were masked in the epitope stabilization assay.
Tpn species-dependent peptide levels are modulated based on suitability as B*2705 ligands
The sequences of eight peptides assigned as quantitative differences, six predominant with hTpn and two predominant with mTpn, were determined (Fig. 5B). The nonamers predominant with hTpn had mean RF and DMP values (P1–P9) similar to those of hTpn-specific nonamers and higher than the only sequenced nonamer predominant with mTpn (Table V). This peptide had RF and DMP scores (P1–P9) only slightly higher than those of mTpn-specific nonamers (Table III). Despite a suitable P1 residue, it used an unusual P9 anchor, which had a strong negative effect on scoring. The global RF of the anchor positions for the only sequenced decamer predominant with hTpn was similar to the corresponding value for the hTpn-specific decamer (Table III), due to very suitable anchor residues.
Thus, peptides showing Tpn species-dependent expression levels have global residue frequencies comparable to those of hTpn or mTpn-specific peptides. This suggests that both qualitative and quantitative species-dependent modulation of peptide loading by Tpn are based on structural suitability for binding to B*2705.
Mouse Tpn incompletely restores B*2705 loading in human cells
We next addressed whether mTpn had an intermediate effect between absence and presence of hTpn on HLA-B27 loading or had a distinct editing function. To this end, we compared the B*2705-bound peptide repertoires from B*2705.220 transfectants lacking Tpn or expressing the human or mouse chaperone. The total peptide yield obtained from similar cell numbers (3.5 x 109 cells) was 3-fold higher with hTpn or mTpn than in the absence of this chaperone (Table VI). Systematic peptide comparisons revealed that only 2% of the Tpn-independent B*2705 ligands were not found in the presence of hTpn or mTpn (Fig. 6A, Table VI). In contrast, 26–27% of the B*2705 ligands bound with hTpn or mTpn were not detected in the absence of this chaperone (Table VI). It is unlikely that the observed differences in peptide sharing are due to the lower peptide yields from Tpn-deficient cells, because many differential ion peaks had significant intensity and did not correspond to minor signals that could be missed in a less abundant peptide pool (Fig. 6B).
Discussion
Species-dependent effects of Tpn for HLA class I loading have been reported in several studies concerning mainly B*4402, a strongly Tpn-dependent allotype that is poorly expressed on mouse cells (17, 20). In particular, Tan et al. (17) reported that the B*4402-bound peptide repertoires from .220 transfectants expressing either hTpn or mTpn were different, although this was analyzed only in a qualitative way. In contrast, B*2705 is relatively independent of Tpn (20) and is expressed at high levels on mouse cells. In a recent study (38), a peptide was more efficiently presented on .220 cells expressing hTpn than mTpn. This peptide was not found in B*2705-transfected mouse cells despite being normally produced by the murine 20S proteasome, suggesting that species-dependent effects of Tpn also influenced B*2705 loading.
In this study, we have 1) quantified the effect of the heterologous expression of mTpn in human cells on the cell surface stability and peptide repertoire of B*2705; 2) identified peptides whose presentation was specifically dependent or quantitatively modulated by Tpn in a species-dependent way; 3) demonstrated that residue usage among hTpn-specific peptides is biased toward residues that are frequent among the constitutive B*2705-bound peptide repertoire, whereas among mTpn-specific peptides is biased toward less frequent residues; 4) shown that mTpn-specific nonamers, but not mTpn-specific decamers, are suboptimal B*2705 binders; 5) determined that the effect of mTpn on peptide selection for B*2705 was intermediate between absence and presence of hTpn; and 6) shown that residue usage among peptides specifically presented by B*2705 on mouse cells is similarly suboptimal as for peptides specifically presented on human cells with mTpn, so that homologous interactions of this chaperone with non-MHC proteins in the loading complex do not improve the B*2705 cargo. This suggests a peptide-modulating role of Tpn that is dependent on its direct interaction with B*2705.
Surface expression of B*2705 on Tpn-deficient .220 cells was only slightly lower than with hTpn. Although with mTpn B*2705 levels were indistinguishable from those with hTpn, slightly increased dissociation was compatible with the possibility that B*2705 on the surface of cells containing mTpn presented peptides with suboptimal stability and that these were a small fraction of the B*2705-bound repertoire. An alternative possibility was hypothesized by Zarling et al. (12): Tpn would influence MHC I stability not on the basis of peptide affinity, but by conditioning the conformational state of the class I molecule after peptide binding in a way that remained undefined.
Identification of the peptides bound to B*2705 in the presence of hTpn or mTpn helped us to distinguish between both alternatives and to characterize the effect of heterologous interactions involving Tpn on the B*2705-bound peptide repertoire. First, mTpn allowed binding of 95% of the native B*2705 ligands, which is significantly more than the already large percentage (75%) bound in the absence of Tpn. However, 5% of each peptide repertoire was differentially presented by B*2705 in the presence of hTpn or mTpn, and 30% of shared ligands were expressed at significantly different levels. Thus, there was a clear Tpn species-dependent modulation of B*2705 loading.
The molecular mechanism by which Tpn influences B*2705 loading and cell surface stability may be inferred from the molecular features of differentially presented peptides. We took advantage of our knowledge of many B*2705 ligand sequences, and of the most suitable anchor residues for this allotype, to compare the suitability of structural features between differentially bound peptides. This approach was based on a previous analysis of residue frequencies at every peptide position among nearly 200 natural B*2705 ligands (52). Thus, any amino acid residue at a given peptide position could be scored on the basis of its frequency among natural B*2705 ligands, giving an unbiased indication of the suitability of a peptide for B*2705 presentation in human cells with an intact Ag-processing-loading system.
When this analysis was applied to individual positions of differentially bound peptides, the subset found only with hTpn scored significantly higher than the mTpn-specific subset, particularly at anchor positions. Shared nonamers scored similarly or slightly lower than the hTpn-specific ones, but higher than the mTpn-specific counterparts. However, this was not the case for shared decamers. These results suggested that differential presentation of nonamers is largely based on affinity. Low binding of mTpn-specific nonamers to B*2705 in vitro, confirmed the conclusions derived from residue scoring. The results explain the small but clear decrease in cell surface stability of B*2705 in .220 cells transfected with mTpn, because suboptimal peptides are a small fraction of the total B*2705-bound repertoire in these transfectants.
In contrast, decamers presented with mTpn scored similarly as shared ligands and bound in vitro with high efficiency. Two aspects of this result should be discussed. First, it is possible that the RF and/or DMP scores might be less reliable for decamers than for nonamers, because these are based on >100 peptides, relative to only 39 decamers (52). Moreover, the epitope stabilization assays may not fully reflect peptide binding in vivo. Although we cannot rule out these possibilities, the fact remains that for both nonamers and decamers there is a correlation between residue scoring and binding efficiency in vitro; whereas mTpn-specific nonamers scored lower than shared nonamers and bound poorly, the mTpn-specific decamers scored similarly as shared decamers and bound with high efficiency. Thus, our results suggest that the role of Tpn in mediating peptide loading may not be identical for nonamers and decamers and that optimization of the peptide cargo is more strictly performed for nonamers. The mechanism for this differential effect is currently unclear to us.
Species-dependent effects of mTpn in human cells have been ascribed, in studies with B*4402, not only to bridging MHC and TAP, but also to defective interactions with other components of the peptide-loading complex (16, 17). Mouse Tpn was reported to be as efficient as its human counterpart in stabilizing TAP levels in B*4402.220 transfectants (17). It is possible that alterations in the stability, composition, or stoichiometry of the loading complex, resulting from heterologous interactions of mTpn with calreticulin and ERp57(17), may influence suboptimal B*2705 loading on human cells transfected with the mTpn. However, the fact that suboptimal residue scoring was also observed among B*2705 ligands specifically loaded in mouse cells argues against a dominant influence of heterologous interactions between Tpn and non-MHC I proteins on suboptimal B*2705 loading and supports a direct role of Tpn in this process.
It is generally accepted that Tpn contributes to stabilize a peptide-receptive conformation of the HLA class I molecule, facilitating peptide binding. After this is accomplished, the HLA molecule dissociates from the loading complex and migrates to the cell surface. On the basis of our results, we would like to hypothesize that dissociation of the HLA class I molecule from the loading complex requires a minimum affinity threshold. Because of suboptimal stabilization of the human loading complex by mTpn, the peptide affinity threshold for dissociation of the MHC molecule from this complex might be lower, allowing for suboptimal ligands to reach the cell surface. The fact that some highly suitable peptides were presented only with hTpn is compatible with this model, because their selective expression might result from lower competition among peptides for productive binding to B*2705 when the affinity threshold is higher and, therefore, more restrictive. However, given the complexity and incomplete knowledge of the interactions among components of the peptide-loading complex other alternatives cannot be ruled out. For instance, hTpn and mTpn might stabilize empty class I molecules in slightly different conformations, one more receptive to only high affinity peptides and the other somewhat more permissive.
Our results argue against the view (12) that Tpn functions as a facilitator of peptide binding, rather than as a peptide editor favoring loading of high affinity MHC ligands. This conclusion was largely based on the observation that the average affinity, as measured in vitro with whole peptide pools, of HLA-B8 or HLA-A2 ligands from Tpn-deficient cells was higher or similar, respectively, than the corresponding peptide pools from Tpn-proficient cells. In addition, HLA-B8 ligands from Tpn-deficient or Tpn-proficient cells had similar anchor residues. In contrast, our results with B*2705 show that individual mTpn-specific nonamers tend to have both suboptimal residues and low binding efficiency, fully consistent with an editor role of Tpn. Our results also suggest that this function may be peptide size dependent, because mTpn-specific decamers were good B*2705 binders.
There could be various explanations for the apparent disparity between our results and those of Zarling et al. (12). For instance, the effect of Tpn on shaping MHC class I-bound peptide repertoires might be variable among allotypes. Another possibility is that, in the binding assay using whole peptide pools, the presence of individual ligands with lower affinity might be masked by a dominant effect of peptides with high affinity in the same pool. For instance, many of the peptides bound in the absence of Tpn are also bound in its presence and may have an affinity that is comparable to the Tpn-dependent ligands, as we have previously shown for B*2705 (22). To our knowledge, affinity measurements of individual HLA class I-bound ligands presented in the absence, but not in the presence of Tpn, have not been reported. Whatever the explanation, the differences we observed between nonamers and decamers point out to unsuspected complexities in the way Tpn modulates peptide loading.
In conclusion, our study provides evidence for suboptimal editing of the B*2705-bound peptide repertoire by mTpn, leading to presentation of peptides with less suitable residues and lower affinity, lack of presentation of some ligands that bind with hTpn, and quantitative alterations in the expression level of shared ligands. As a result, surface expressed B*2705 is slightly less stable in the presence of mTpn. These results have obvious implications for animal models of HLA-B27-associated disease. We previously reported that more than half of the peptides differentially presented by B*2705 on human and mouse cells do not arise from species-related protein polymorphism, but from differences in Ag processing (38). Now we have quantified the effect of mTpn as the only heterologous component in the loading complex on peptide presentation by B*2705. Although it remains to be proven, it is likely that the effects on peptide selection will be similar or larger on mouse cells, in which the only human component is HLA-B27. This was suggested by results concerning a single peptide in our previous study (38). Now we have shown that B*2705 ligands arising from conserved sequences between mouse and man, but differentially expressed on human cells have more suitable residues than those differentially expressed on mouse cells, in a pattern very similar to that found for human or mTpn-specific peptides from .220 cells. This suggests that the function of mTpn for B*2705 peptide loading is not improved when other components of the loading complex are also of mouse origin. Therefore, suboptimal peptide presentation, relative to humans, is to be expected when B*2705 is expressed on mouse cells and transgenic mice.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by Grants SAF2002/00125 and SAF2003/02213 from the Ministry of Science and Technology, 08.3/0005/2001.1 from the Comunidad Autónoma de Madrid, and an institutional grant of the Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa. A.W.P. is a recipient of the Russell Grimwade Fellowship.
2 Address correspondence and reprint requests to Dr. José A. López de Castro, Centro de Biología Molecular, Severo Ochoa, Facultad de Ciencias, Universidad Autónoma, 28049 Madrid, Spain. E-mail address: aldecastro{at}cbm.uam.es
3 Abbreviations used in this paper: Tpn, tapasin; hTpn, human Tpn; mTpn, mouse Tpn; .220, 721.220; 2m, 2-microblobulin; MS, mass spectrometry; MS/MS, tandem MS; RF, residue frequency; DMP, deviation from mean frequency in the human proteome; m/z, mass-to-charge.
Received for publication April 29, 2004. Accepted for publication April 12, 2005.
References
van der Burg, S. H., M. J. Visseren, R. M. Brandt, W. M. Kast, C. J. Melief. 1996. Immunogenicity of peptides bound to MHC class I molecules depends on the MHC-peptide complex stability. J. Immunol. 156: 3308-3314.
Sadasivan, B., P. J. Lehner, B. Ortmann, T. Spies, P. Cresswell. 1996. Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I molecules with TAP. Immunity 5: 103-114.
Hughes, E. A., P. Cresswell. 1998. The thiol oxidoreductase ERp57 is a component of the MHC class I peptide-loading complex. Curr. Biol. 8: 709-712.
Lindquist, J. A., O. N. Jensen, M. Mann, G. J. Hammerling. 1998. ER-60, a chaperone with thiol-dependent reductase activity involved in MHC class I assembly. EMBO J. 17: 2186-2195.
Morrice, N. A., S. J. Powis. 1998. A role for the thiol-dependent reductase ERp57 in the assembly of MHC class I molecules. Curr. Biol. 8: 713-716.
Ortmann, B., M. J. Androlewicz, P. Cresswell. 1994. MHC class I/2-microglobulin complexes associate with TAP transporters before peptide binding. Nature 368: 864-867.
Momburg, F., P. Tan. 2002. Tapasin—the keystone of the loading complex optimizing peptide binding by MHC class I molecules in the endoplasmic reticulum. Mol. Immunol. 39: 217-233.
Paulsson, K., P. Wang. 2003. Chaperones and folding of MHC class I molecules in the endoplasmic reticulum. Biochim. Biophys. Acta. 1641: 1-12.
Williams, A. P., C. A. Peh, A. W. Purcell, J. McCluskey, T. Elliott. 2002. Optimization of the MHC class I peptide cargo is dependent on tapasin. Immunity 16: 509-520.
Yu, Y. Y., H. R. Turnquist, N. B. Myers, G. K. Balendiran, T. H. Hansen, J. C. Solheim. 1999. An extensive region of an MHC class I 2 domain loop influences interaction with the assembly complex. J. Immunol. 163: 4427-4433.(Laura Sesma*, Bego?a Galo)