Tapasin shapes immunodominance hierarchies according to the kinetic stability of peptide – MHC class I complexes



Peptide loading of MHC class I molecules involves multiple cofactors including tapasin. We showed previously in vitro that tapasin edits the peptide repertoire by favoring the binding of peptides with slow dissociation rates. Here, using tapasin-deficient mice and a DNA vaccine that primes directly, we confirm that tapasin establishes hierarchical responses in vivo according to peptide-MHC stability. In contrast, this hierarchy is lost when the peptides are cross-presented via an alternative DNA vaccine. By regulating transgene expression, we found that the dominant response modifier was antigen persistence. Our findings reveal strategies for activating T cells against low-affinity peptides, of potential importance for patients with repertoires narrowed by deletional tolerance.


N-terminal domain of fragment C from tetanus toxin




Effector CD8+ T cell responses to complex immunogens often focus on a few epitopes compared to the total number of potential epitopes available. One of the most important determinants of this phenomenon, termed immunodominance, is the abundance of peptide – MHC (pMHC) complexes at the surface of the initial priming APC 1. Several factors will contribute to this, including the efficiency of antigen processing, the kinetics and thermodynamics of pMHC interaction, and the efficiency of peptide loading inside APC. Although the molecular mechanism by which peptides are selected and loaded onto MHC class I molecules is becoming clearer, the immunological consequences of these events remain largely uninvestigated.

MHC class I molecules are assembled with peptide cargo in the lumen of the ER while they are part of a peptide-loading complex comprising TAP, tapasin, calreticulin, ERp57, and protein disulfide isomerase 24. Loss of any of these components gives rise to a quantitative loss of MHC class I peptide complexes that are exported to the cell surface with the approximate order TAP > tapasin (depending on the allele) > calreticulin ⩾ ERp57 59. The cofactor tapasin additionally edits the peptide repertoire that is loaded onto class I molecules in favor of those with slow dissociation kinetics 10. This is a similar function to that described for DM in MHC class II loading 11.

We have shown previously in cell lines that the hierarchy of MHC class I-restricted presentation of peptides generated endogenously (from translation of a minigene) is controlled by the kinetics of peptide binding to class I in a tapasin-dependent way 10. Thus, the peptide variants of the ovalbumin-derived peptide SIINFEKL, SIINFEKM and SIINYEKL, represent a hierarchy with decreasing biological half-lives. Transfectants stably expressing minigenes encoding each of these peptides display a level of surface presentation of H2-Kb complexes that corresponds exactly with this half-life but only in the presence of the peptide editor tapasin. When tapasin is absent, the hierarchy distorts in such a way as to lead to preferential presentation of an intermediate half-life peptide. Thus tapasin edits the repertoire of peptides supplied endogenously in a way that is sensitive to half-life 10.

There is evidence from studies investigating MHC class II-restricted antigen presentation that the DM-edited repertoire dictates the intensity of CD4+ T cell responses in mice 1214. However, it is not known whether the tapasin editing we have observed in cell culture is physiologically relevant for the generation of immunodominant CD8+ T cell responses in vivo. To this end we have measured CD8+ T cell priming to the same epitopes we have studied previously in vitro by measuring a single CD8+ T cell reactivity (anti-Kb/SIINFEKL), following the immunization of mice with DNA vectors containing class I- and class II-restricted determinants in two different configurations 15, 16.

In the first design, the CD8+ T cell epitope is expressed as a minimal peptide specifically targeted to the ER by an N-terminal leader sequence. To provide critical “help” for the minigene product, a separate expression cassette is incorporated within the plasmid backbone encoding a hybrid invariant chain molecule, with the CLIP sequence replaced by a Th determinant from tetanus toxin (p30) 15. This vaccine fails to cross-present antigen because of the short half-life of minigene products 17 and induces Th-dependent CD8+ T cell responses via directly transfected APC 15. In the second design, the CD8+ T cell epitope is fused immediately 3′ of the N-terminal domain of fragment C (N.FrC) from tetanus toxin 16. Expression of this fusion protein allows both direct presentation and cross-presentation of the class I-restricted epitope. However, we have shown that priming of CD8+ T cell responses to the C-terminal peptide of this fusion is dependent upon cross presentation, with responses to directly presented determinants aborting due to the absence of help 15.

Thus, the two different DNA vaccine designs allowed us to explore whether tapasin impacts on direct presentation and cross-presentation pathways. We find that tapasin establishes a hierarchical CD8+ T cell response when peptides of decreasing half-life are presented directly. However, this hierarchy is lost if peptides are cross-presented over prolonged periods. These observations underline the importance of the nature and persistence of antigen in the priming and breadth of CD8+ T cell responses, and have relevance for vaccine design.

Results and discussion

Direct priming elicits hierarchical CD8+ T cell responses that mirror pMHC stability

To study the relationship between peptide-off rates and immunogenicity, we used a DNA immunogen designed to specifically target nominal MHC class I- and class II-restricted epitopes into the ER of transfected APC. This vaccine, designated pDUO, minimizes the requirement for antigen processing and induces Th-dependent CD8+ T cell responses by direct presentation 15. Naive C57BL/6 mice were immunized intramuscularly with pDUO encoding the peptide variants we have studied previously in vitro10: SIINFEKL (FL), SIINFEKM (FM), and SIINYEKL (YL). After 12 days, the number of SIINFEKL-specific T cells induced was measured directly ex vivo by IFN-γ ELISPOT.

As shown in Fig. 1A, the magnitude of the response to each peptide mirrored the hierarchy of antigen presentation levels measured in cell transfectants 10, which in turn is dictated by the kinetic stability of the pMHC complex. Responses were also measured to the immunizing peptide and we observed no change in the relative number of IFN-γ-producing cells (data not shown). Therefore, as described earlier 17, CD8+ T cells elicited to these variant peptides are fully cross-reactive with the wild-type peptide. This is also consistent with our observation that the T cell hybridoma B3Z, which was raised to the wild-type Kb/SIINFEKL peptide complex, recognized each of the peptide variants equally well when present at saturation (T. Elliott, unpublished data). These data, combined with previous finding 18, 10, demonstrate that kinetic stability is a key parameter in determining the CD8+ T cell response hierarchy to these variant peptides. A similar direct relationship between pMHC stability and immunogenicity was reported by Melief and colleagues 19.

Figure 1.

Immunization with pDUO elicits a hierarchical CD8+ T cell response, mirroring kinetic stability of pMHC complexes that is controlled by tapasin. C57BL/6 mice (A) and tapasin–/– mice on a C57BL/6 background (B) were immunized by intramuscular injection with pDUO encoding the peptide variants indicated. After 12 days, the number of SIINFEKL-specific T cells induced was measured directly ex vivo by IFN-γ ELISPOT. The results shown are combined from at least two separate experiments, with each data point representing an individual mouse. Horizontal bars depict group means.

Hierarchical CD8+ T cell priming is lost in tapasin-deficient mice

The hierarchy of SIINFEKL peptide variant presentation, though underpinned by kinetic stability of the pMHC complex, is nevertheless dependent on the peptide-editing function of tapasin 10. Thus, in tapasin-negative cells transfected with the same variant peptide minigenes, the hierarchy of presentation is different with a peptide of intermediate kinetic stability (SIINFEKM) being presented to higher levels than wild-type peptide. Furthermore, the difference in level of presentation between peptide variants with the slowest half-life (SIINFEKL) and fastest half-life (SIINYEKL) is drastically reduced in tapasin-negative cells 10.

In order to determine the extent to which the response hierarchy that we observed in Fig. 1A is dependent on tapasin-mediated editing, we repeated the pDUO immunizations in tapasin-deficient (tapasin–/–) mice 6. The hierarchy between SIINFEKL and SIINFEKM response intensities seen in wild-type mice was lost (Fig. 1B), with an inversion in the response magnitudes to SIINFEKL and SIINFEKM. Thus, in the absence of tapasin, CD8+ T cell responses to intermediate- rather than high-stability pMHC complexes dominate. Failure to respond to SIINYEKL is consistent with the low levels of this peptide that can be directly presented in the absence of tapasin 10. The correlation observed between the preferential formation of intermediate stability complexes in transfected cells 10, and the level of CD8+ T cell response observed after DNA immunization (Fig. 1B), results in an alteration in the immune-response hierarchy.

Taken together, these data indicate that in vivo, peptide stability is a necessary but insufficient parameter in determining the hierarchy of primary T cell responses and that the additional influence of tapasin editing is essential to establish immunodominance: the mechanism being its ability to select peptide cargo for loading onto MHC class I molecules according to the kinetic stability of the resulting pMHC complex.

A DNA-encoded fusion protein breaks the immunodominance response hierarchy

We investigated a second DNA-based immunogen in which the SIINFEKL variant epitope was appended immediately 3′ of N.FrC from tetanus toxin. This DNA vaccine, designated pDOM-peptide, induces Th-dependent CD8+ T cell responses by cross-presentation 15.

We initially established whether tapasin-sufficient cells transfected with pDOM-peptide maintained the hierarchy observed in minigene transfectants 10. H-2b RMA cells were transiently transfected with DNA expression vectors encoding each of the SIINFEKL peptide variants fused to N.FrC. After 6 h in culture, peptide presentation was detected by flow cytometry with the monoclonal antibody 25-D1.16 20, which is able recognize all of the analogues bound to H-2Kb10. The fusion proteins were co-expressed with enhanced GFP from a single bicistronic transcript, enabling normalization for transfection efficiency. As shown in Fig. 2A, their relative presentation matched that of nominal minigene products, giving rise to the hierarchy SIINFEKL > SIINFEKM > SIINYEKL.

Figure 2.

The relationship between kinetic stability and immunodominance in vivo is lost with pDOM-peptide DNA immunization. (A) RMA cells were transfected with pIRES-DOM-peptide constructs, encoding each of the SIINFEKL variants indicated, or pIRES alone (dashed line) and stained 8 h later with H-2Kb/SIINFEKL-specific antibody 25-D1.16 20. Arbitrary gates were introduced into dot plots according to enhanced GFP expression levels, and the 25-D1.16 MFI of cells within the gates calculated using WinMDI 2.8 software. (B) The immunogenicity of pDOM-peptide-encoded SIINFEKL variants in wild-type mice was assessed by IFN-γ ELISPOT analysis 12 days after intramuscular DNA injection. The results shown are combined from two separate experiments, with each data point representing an individual mouse.

In contrast, when we immunized C57BL/6 mice with SIINFEKL variant-encoding pDOM-peptide constructs, we found that the immune response hierarchy we observed with the RMA transfectants and pDUO was absent (Fig. 2B). Instead, responses to all three peptides were roughly equivalent. Most notably, the response to SIINFEKL was significantly lower compared to pDUO immunization (mean 481 SFC/106 splenocytes compared to 1590, p=0.0006) and the relative response to SIINYEKL was significantly higher (mean 491 SFC/106 splenocytes compared to 113, p=0.0027). This flattening of the hierarchy appeared therefore to be specific for DNA delivery (viapDOM-peptide) of cross-presented antigen.

Regulated antigen expression restores hierarchical CD8+ T cell responses induced by cross-priming

We have shown recently that uncontrolled, CMV promoter-driven DOM-FL transgene expression results in prolonged cross-presentation of the SIINFEKL peptide (>20 days) that limits the magnitude of a primary CD8+ T cell response 21, possibly due to supra-optimal stimulation by the high-stability Kb/SIINFEKL complex leading to physical deletion 22. Conversely, using a mifepristone-responsive gene regulation system (GeneSwitch™ 23) to control antigen presentation, we have shown that reducing antigen expression to a short burst (<5 days) results in a significant amplification of the SIINFEKL-specific response.

To examine whether it might be possible to re-establish the response hierarchy by shortening the duration of antigen expression, wild-type C57BL/6 mice were co-injected with the GeneSwitch™ plasmids, encoding each of the peptide variants fused to N.FrC, and antigen expression induced with a single dose of 0.5 mg/kg mifepristone. This dose of the inducer leads to transient antigen expression (<5 days). As shown in Fig. 3A, when expression of DOM-peptide is temporally regulated, a hierarchy is established that is comparable to pDUO immunization. To confirm that this hierarchy is dependent upon the editing function of tapasin, we repeated these experiments in tapasin-deficient mice. Akin to the results obtained with pDUO, shuffling of the response hierarchy was observed, with the peptide of intermediate kinetic stability (SIINFEKM) inducing more robust responses than the wild-type peptide (Fig. 3B). These data suggest that tapasin establishes hierarchical CD8+ T cell responses to cross-presented antigen but such hierarchies are shaped by persistence of high-affinity peptide ligands.

Figure 3.

Transient antigen expression restores a hierarchical CD8+ T cell response that is controlled by tapasin. Wild-type (A) and tapasin–/– mice (B) were immunized with pSwitch and pGene/DOM-peptide, and transgene expression induced with 0.5 mg/kg mifepristone. After 12 days, splenocytes were prepared and the number of SIINFEKL-specific T cells measured by IFN-γ ELISPOT. The results shown are combined from two separate experiments.

It is not clear why a response was seen to the low-stability peptide SIINYEKL only when it was cross presented – either as a bolus or by sustained delivery. One possibility is that its level of presentation simply breeches a threshold required for T cell activation that is not reached following immunization with pDUO. Consistent with this idea is the frequent observation that antigen processing by cross-presentation pathways is more efficient than via the endogenous pathway 24, and our own observation that responses to all peptides in the hierarchy, including SIINYEKL, are elevated following delivery of a bolus of cross-presented antigen (compare Fig. 1A and 3A), resulting in a detectable anti-SIINYEKL response.

Concluding remarks

We have shown that tapasin establishes a hierarchical CD8+ T cell response that reflects peptide half-life when peptides are targeted directly to the ER of the priming APC. This is consistent with our previous findings that the relative level of presentation of each of these peptides in transfected cells is predicted by their half-life when tapasin is present, this relationship breaking down in its absence 10, and with other studies that correlate immunogenicity with kinetic stability of the pMHC complex 19, 25. Taken together, therefore, these data indicate that tapasin controls the hierarchical response to diverse peptides by regulating their level of expression at the surface of APC. This was true for both direct presentation and cross-presentation pathways, although for the latter, hierarchical responses to the tapasin-groomed peptide repertoire could be altered by antigen persistence.

Materials and methods


pDUO, pDOM-peptide and pIRES-DOM-peptide, encoding each of the SIINFEKL peptide variants, were constructed as previously described 15. To construct the GeneSwitch™ vectors, open reading frames flanked by Hind III and Not I restriction sites were amplified by PCR using pDOM-peptide as template and subcloned into pGene/V5 (Invitrogen, Carlsbad, CA). Plasmid DNA was purified for immunization using a QIAfilter Giga kit (Qiagen, Hilden, Germany). All constructs were sequenced and checked for expression in vitro using the TNT T7 Coupled Reticulocyte Lysate System (Promega, Southampton, UK).

In vitro transfection

RMA cells (ATCC) were transfected by electroporation (320 V, 750 μF in 400 μL serum-free RPMI) using a Gene Pulser II (BioRad, Richmond, CA).

Mice and in vivo experiments

Tapasin–/– mice 6 used in these studies were backcrossed for at least seven generations onto the C57BL/6 background. Wild-type C57BL/6 and tapasin–/– mice, both bred in-house, were vaccinated at 8-10 wk of age with a total of 50 μg of plasmid DNA in normal saline injected into two sites in the quadriceps. For the GeneSwitch(tm) experiments, mice were injected intramuscularly with 50 μg pGene/DOM-peptide and 25 μg pSwitch (Invitrogen). Four hours after plasmid DNA injection, mifepristone (Sigma, Poole, UK) was given to the mice intraperitoneally at the indicated dosage. Animal experiments were conducted according to the UK Home Office license guidelines and approved by the University of Southampton's ethical committee.

Evaluation of peptide-specific T cell responses

ELISPOT analysis was performed using the BD™ ELISPOT Set for murine IFN-γ. Spots were developed using 5-bromo-4-chloro-3-indolyl phosphate (Zymed Laboratories, San Francisco, CA) and counted with a Transtec 1300 ELISPOT reader (AID Diagnostika GmbH, Strassberg, Germany).

Statistical analysis

Statistical significance between vaccination groups was determined by the nonparametric Mann–Whitney test (two-tailed) using Prism 4 (GraphPad) software.


We thank R. N. Germain for providing the 25-D1.16 antibody, L. van Kaer for kindly providing tapasin–/– mice, and members of the biomedical facility for their technical assistance. This work was supported by grants from the Leukaemia Research Fund UK and Cancer Research UK, awarded to S. Thirdborough and T. Elliott, respectively.


Conflict of interest: The authors declare no financial or commercial conflict of interest.


  1. 1


  2. 2


  3. 3