Natural selection drives diversification of MHC class I proteins, but the mechanism by which selection for polymorphism occurs is not known. New variant class I alleles differ from parental alleles both in the nature of the CD8 T cell repertoire formed and the ability to present pathogen-derived peptides. In the current study, we examined whether T cell repertoire differences, Ag presentation differences, or both account for differential viral resistance between mice bearing variant and parental alleles. We demonstrate that nonresponsive mice have inadequate presentation of viral Ag, but have T cell repertoires capable of mounting Ag-specific responses. Although previous work suggests a correlation between the ability to present an Ag and the ability to generate a repertoire responsive to that Ag, we show that the two functions of MHC class I are independent.
Major histocompatibility complex class I molecules are highly polymorphic in many mammalian populations (1, 2). Variability between MHC class I alleles is focused on peptide-binding residues (3). Unlike new alleles at most loci, MHC class I variants are generated by gene conversion (4, 5). New variants are selectively maintained in populations, leading to the vast diversity of alleles seen in MHC class I loci (6).
MHC class I molecules are critical components in the recognition of intracellular pathogens. Because of this, it has been proposed that selection for MHC class I variants is based on differences in the immune responses of individuals that differ at class I loci (7). MHC class I molecules serve two critical functions in CD8 T cell-mediated immunity to intracellular pathogens. First, during thymocyte development and naive CD8 T cell homeostasis, class I molecules bind and present self-peptides; this mediates positive selection in the thymus and is required for peripheral homeostasis of CD8 T cells (8, 9). In this way, interactions between MHC class I-self-peptide complexes and TCRs shape the immune potential by generating the CD8 T cell repertoire. Second, class I molecules present pathogen-derived peptides during an immune response (10). This allows CD8 T cell activation during the afferent phase of the immune response, as well as targeting of CTLs to infected cells during the efferent phase (11).
Gene conversion variants differ in their ability to respond to infections (12, 13) and immunization (14). Because gene conversion mutations affect binding to both self and foreign peptides, gene conversion variants have the potential to impact both Ag presentation during an immune response and the T cell repertoire that is mediating that response. It has been shown previously that gene conversion mutations can affect presentation of antigenic peptides (15). Recently, we have shown that gene conversion mutations also cause dramatic changes in the T cell repertoire, thereby impacting immune potential (16). T cell repertoire differences have been implicated in the altered ability of gene conversion variants to respond to viral infection (13); however because repertoire selection and Ag presentation are both mediated by class I, it is difficult to distinguish differences in these functions. In the case of the response to the OVA-derived peptide SIINFEKL, the class I variant Kbm8 differs from the parental allele Kb both in the ability to bind SIINFEKL (15) and in the ability to generate a SIINFEKL-responsive repertoire (14).
To study the effects of gene conversion-based changes in immune potential and Ag presentation on immunity to intracellular pathogens, we compared the ability of different MHC class I alleles to direct CD8 T cell responses to Theiler’s murine encephalomyelitis virus (TMEV).3 TMEV is a naturally occurring mouse pathogen; Daniel’s strain is a laboratory strain of TMEV that infects the CNS (17). Susceptible mice are unable to clear the virus after intracranial inoculation, and TMEV infection spreads to the white matter in the spinal cord (18). Persistent infection with TMEV leads to spinal cord white matter lesions consisting of leukocytic infiltrates and regions of demyelination (18). On certain genetic backgrounds, demyelination leads to neurologic impairment similar to that seen in the human disease multiple sclerosis (19). By contrast, mice resistant to TMEV clear the virus after a brief period of encephalitis, and have no long-term impairment or pathology (20).
Resistance to chronic TMEV infection has been mapped to the H-2D locus (21). C57BL mice with Db, Dd, or Dk clear TMEV, preventing persistent infection (21, 22), whereas all other haplotypes tested are susceptible. In B6 mice, which have the allele Db, resistance to TMEV-induced demyelination depends on a CD8 T cell-mediated response to a single immunodominant peptide from the TMEV capsid protein VP2 (23). This CTL response is readily demonstrated by staining brain-infiltrating lymphocytes with tetrameric complexes comprised of the B6 class I allele Db bound to the TMEV-encoded peptide VP2121–130 (22). A large proportion (50–70%) of the infiltrating CD8 T cells are labeled by Db-VP2121–130 tetramers during the peak of the immune response at day 7 postinfection, whereas few cells can be stained with tetramers of Db plus an irrelevant peptide (22).
We studied the immune response to TMEV in B6 mice and the related strain bm14. B6 and bm14 mice differ at the D locus by a single nucleotide (24). Gene conversion mutations were originally described as multiple templated nucleotide substitutions within a single event (4). However, because neither the bm14 mutation nor any other known spontaneous class I mutation introduces a novel nucleotide into the genome (all have potential donor nucleotides), we infer that, as a class, all the single nucleotide spontaneous class I mutants arose by gene conversion (25). We found that although B6 mice resist demyelination and respond to the immunodominant VP2121–130 peptide, bm14 mice are susceptible to chronic infection and myelin loss. Moreover, bm14 CTLs do not respond to VP2121–130. This lack of response could be because 1) bm14 mice have a responsive T cell repertoire but are defective in Ag presentation, 2) bm14 presents viral Ag well but lacks responsive T cells, or 3) both the repertoire and Ag presentation are defective. In this study, we distinguish among these possibilities.
Materials and Methods
Mice and viral infection
C57BL/6 (B6), B6.129S7-Rag1tm1Mom (B6rag−/−), and B6.C-H2bm14/By (bm14) mice were obtained from The Jackson Laboratory and maintained in our colony. To obtain Db/b, Db/bm14, Dbm14/bm14, and bm14rag−/− mice, bm14 mice were intercrossed with B6rag−/− mice; F2 and subsequent offspring were typed at the D and rag1 loci. To distinguish Db and Dbm14, RNA was isolated from blood, and RT-PCR was performed to amplify the D region at primers 5′-AGCCCCGGTACATCTCT-3′ and 5′-CAGGTAGGCCTTGTAATG-3′ (temperature, 54°C). Amplified cDNA was digested with NlaIII, which distinguishes cDNA from Db and Dbm14. Rag−/− mice were identified by the absence of CD3+ cells seen in the blood by flow cytometry. Rag+/+ mice were identified by the absence of the neo marker as determined by PCR of tail DNA at primers 5′-CTTGGGTGGAGAGGCTATTC-3′ and 5′-AGGTGAGATGACAGGAGATC-3′ detect neo, whereas primers 5′-CAAATGTTGCTTGTCTGGTG-3′ and 5′-GTCAGTCGAGTGCACAGTTT-3′ detect an internal standard (temperature touchdown from 64 to 58°C). Male mice between 8 and 16 wk of age were used for all experiments. Mice were infected intracranially with 106 PFU of Daniel’s strain of TMEV. All experiments were performed in compliance with institutional and National Institutes of Health guidelines for animal care and use.
On day 45 or 90 postinoculation, mice were perfused with Trump’s fixative (1.0% glutaraldehyde and 4% formalin in 0.1 M phosphate buffer, pH 7.4). Spinal cords were removed, sectioned, embedded, and stained as previously described (23). Briefly, 15 transverse sections representing all regions of the spinal cord were embedded in glycol methacrylate plastic (JB-4; Polysciences). Each section was stained with erichrome and cresyl violet to reveal myelin (26). Each slide was read with the observer blinded as to the mouse strain; mice were considered susceptible if at least one demyelinated lesion was found in any of the representative sections from that mouse’s spinal cord.
Cell isolation and tetramer staining
To isolate brain-infiltrating lymphocytes for flow cytometry, we dissected and homogenized brains from mice that had been infected with TMEV for 7 days. Homogenates were mixed with 30% Percoll (Sigma-Aldrich) in RPMI 1640 (Cambrex Bioscience), and cells were separated by centrifugation as previously described (23). Erythrocytes were lysed by treatment for 2 min with ACK (8.29 g/L ammonium chloride, 1 g/L potassium bicarbonate, and 0.037 g/L EDTA). R-PE-labeled tetramers were made as previously described (22). Peptides used to make the tetramers include FHAGSLLVFM (TMEV-VP2121–130) (27), as well as the irrelevant control peptides RAHYNIVTF (human papillomavirus E749–57) (28), and KAVYNFATC (lymphocytic choriomeningitis virus) (29). Peptides were synthesized in the Mayo Protein Core Facility. Abs to CD4, CD8, and CD45 were purchased from BD Pharmingen. Cells were incubated on ice with tetramers for 40 min, washed in FACS medium (HBSS containing 10 g/L BSA and 0.2 g/L sodium azide), incubated with Abs for 20 min, washed in FACS medium, and fixed with 2% paraformaldehyde. FACS analyses were performed by the Mayo Flow Cytometry Core Facility on a FACSCalibur (BD Biosciences), and data collected as log10 fluorescence were analyzed using CellQuest (BD Biosciences). CD45highCD8+ cells were analyzed for tetramer staining, and were reported as a percentage of tetramer-positive CD45highCD8+ cells. The total number of brain-infiltrating lymphocytes was similar in each group within an experiment; therefore, differences in the percentage of tetramer-positive cells reflect differences in the absolute number of tetramer-positive cells.
MHC class I stabilization assay
Ltk fibroblasts were transfected with Db, Dbm14, Kb, or Kbm8 as previously described (30), and were maintained in HAT medium (RPMI 1640 complete medium containing 100 μM hypoxanthine, 0.4 mM aminopterin, and 16 μM thymidine). The following peptides were generated in the Mayo Protein Core Facility: FHAGSLLVFM (VP2121–130), SIINFEKL (OVA257–264), RKKRRQRRRAAAFHAGSLLVFM (HIVtat-VP2121–130), and RKKRRQRRRAAASIINFEKL (HIVtat-OVA257–264). For the class I stabilization assay, cells were cultured overnight in serum-free RPMI 1640 (Cambrex Bioscience). Either no peptide or 10 mg/ml VP2121–130, OVA257–264, HIVtat-VP2121–130, or HIVtat-OVA257–264 were added. After peptide incubation, cells were washed and incubated with KH95 mAb (BD Pharmingen) to detect D alleles or 28-14-8S mAb (31) to detect K alleles. (K alleles were comprised of Kb or Kbm8α1α2 domains linked to the α3 domain of Ld, which is recognized by 28-14-8S (31)). Alternatively, cells were incubated overnight with 20 μg/ml VP2121–130, washed, and incubated in serum-free RPMI 1640 for 1, 7, or 24 h before labeling with 28-14-8S mAb. Cells were analyzed in the Mayo Flow Cytometry Core Facility as earlier described. Data are reported as the fold change in class I expression produced by a given peptide compared with expression in the absence of exogenous peptide.
Monomeric class I stability assay
Soluble monomeric complexes of Db-VP2121–130 and Dbm14-VP2121–130 were prepared as previously described (22), except that mouse β2-microglobulin (β2m) was used rather than human β2m. Monomers were incubated at 37°C for 0, 8, 24, or 48 h. After incubation, stable complexes were detected using 28-14-8S mAb and immunoprecipitated with protein G-Agarose beads (Calbiochem). Immunoprecipitates were separated by SDS-PAGE on a 12.5% gel (Bio-Rad). The amount of stable class I remaining at each time point was quantitated using a Storm 840 system (Molecular Dynamics) and was normalized to the amount of precipitating Ab loaded.
Adoptive transfer of T cells
Donor splenocytes were isolated, and erythrocytes were lysed with ACK as previously described. Splenocytes were then passed through T cell selection columns (R&D Systems) according to the manufacturer’s instructions. Each adoptive transfer recipient received 2 × 107 T cells i.v. One day after adoptive transfer, mice were infected with TMEV.
Bone marrow transplants
Bone marrow transplant recipients were given 900 rad whole body irradiation using a 137Cs irradiator (J. L. Shepherd and Associates) 4–8 h before transplantation. Donor bone marrow cells were isolated as previously described (32). Isolated bone marrow cells were depleted of T cells using anti-CD4 and anti-CD8 magnetic beads as recommended by the manufacturer (Miltenyi Biotec). A total of 5 × 106 T cell-depleted bone marrow cells were transplanted i.v. into each recipient. Recipient mice were given 1 g/L tetracycline (Fort Dodge Animal Health, Fort Dodge, IA) in their drinking water for 1 wk before and 2 wk following bone marrow transplantation.
Paired data were analyzed using Student’s paired t test. Unpaired data were analyzed using Student’s unpaired t test. Values of p < 0.05 were considered statistically significant.
Mice bearing the bm14 mutation are susceptible to demyelination caused by persistent TMEV infection
The bm14 mutant mouse is derived from the C57BL/6 (B6) strain, but bm14 differs from B6 by a single amino acid at the class I locus H-2D (Q70H) (24). To test whether the subtle bm14 mutation alters resistance to persistent TMEV infection and spinal cord demyelination, we infected B6 and bm14 mice intracranially with the Daniel’s strain of TMEV. Forty-five days after being inoculated with TMEV, the mice were sacrificed and perfused with Trump’s fixative. This time point has been used previously to determine whether mice are resistant or susceptible to demyelination (21). Spinal cord sections were cut, embedded in plastic, and stained with erichrome and cresyl violet to reveal myelin. As expected, spinal cord sections from B6 mice displayed normal spinal cord architecture, with few inflammatory infiltrates and no demyelination (Fig. 1, A and C). In contrast, bm14 spinal cords contained both infiltrates and demyelinating lesions (Fig. 1, B and D). Demyelination was seen in 50% of bm14 mice (three of six) at day 45 postinfection, whereas 0% (none of six) B6 mice had lesions at this time point. In a similar experiment in which mice were infected for 90 days, the penetrance of demyelination was 100% (five of five) for bm14 (data not shown). B6 mice do not show any demyelination even after 90 days of infection (20). Thus, bm14 mice, unlike parental B6 mice, are susceptible to demyelination caused by persistent TMEV infection.
Mice that lack the Db allele fail to generate a CD8 T cell response to the TMEV-derived peptide VP2121–130
Because resistance to TMEV-induced demyelination is dependent on a CTL response to the viral peptide VP2121–130, we tested whether the Dbm14 molecule can generate a CD8 T cell response to VP2121–130. To do this, we intercrossed B6 and bm14 mice; then we typed the F2 offspring for Db vs Dbm14. Intercross offspring that had Db/b, Db/bm14, or Dbm14/bm14 were infected with TMEV. Seven days after infection, brain-infiltrating lymphocytes were isolated and stained with tetramers. To determine whether cells were specific for Db and/or Dbm14, we used tetramers of Db-VP2121–130, Dbm14-VP2121–130, and Db-E7 (an irrelevant peptide). As shown in Fig. 2, mice with Db (Db/b and Db/bm14) had significantly more CD8 T cells reactive with Db-VP2121–130 than with Db-E7 (p < 0.0001); they also had more CD8 T cells reactive with Dbm14-VP2121–130 than with Db-E7 (p < 0.005). Both Db/b and Db/bm14 mice had more Db-VP2121–130-reactive than Dbm14-VP2121–130-reactive CD8 T cells (p < 0.0005). The CD8 T cell response to Dbm14-VP2121–130 was cross-reactive rather than Dbm14-restricted in nature, because all of the Dbm14-VP2121–130-reactive cells were also reactive with Db-VP2121–130 (data not shown). Mice with only Dbm14 did not mount a CD8 T cell response reactive with either allele (Fig. 2). Thus, although T cells can react with Dbm14-VP2121–130, mice without Db cannot support activation and expansion of these T cells. Furthermore, the number of CD8 T cells in the brains of these mice is comparable to that seen in other TMEV-susceptible strains, whereas VP2121–130-responsive mice have more intracranial CD8 T cells (data not shown). Therefore, it is unlikely that bm14 mice generate a significant CD8 T cell response to any TMEV Ag. Importantly, because the mice used in this experiment were offspring of an intercross between B6 and bm14, we conclude that the failure of Dbm14/bm14 mice to respond to VP2 maps was determined by the region of chromosome 17 harboring the mutation in the D locus rather than to another genetic region.
The VP2121–130 peptide binds and stabilizes both Db and Dbm14
The failure of mice with the bm14 mutation to respond to VP2121–130 could be due to several reasons. Either Dbm14 is incapable of presenting the VP2121–130 peptide, and/or Dbm14 selects and maintains a CD8 T cell repertoire that lacks cells with TCRs recognizing VP2121–130. It is possible that both Ag presentation and repertoire formation are adversely affected by the bm14 mutation. Because the ability to present a peptide requires peptide binding, we sought to determine whether VP2121–130 could bind to Dbm14. As demonstrated in Fig. 2, tetramers of Dbm14 and VP2121–130 could be made and were able to bind to T cells. However, our tetramers use human β2m, so they cannot be used to rigorously assess whether the peptide Ag is normally bound by Dbm14 when mouse β2m is present. To test this, we used a peptide stabilization assay. Cells cultured overnight in serum-free media do not express maximal levels of class I molecules; expression of a class I allele can be increased by adding a high concentration of a peptide that binds to that allele. To test for binding of VP2121–130 to Dbm14 and Db, we incubated cells transfected with either allele with the peptide in serum-free medium. Both Db and Dbm14 were stabilized by VP2121–130, as indicated by their increased expression in the presence of peptide compared with when no peptide is added (Fig. 3,A; p < 0.05). Stabilization by VP2121–130 was only seen with H-2D alleles, as control cells transfected with either Kb or Kbm8 did not exhibit increased expression upon incubation with VP2121–130. Conversely, Db and Dbm14 were not stabilized by the Kb-binding peptide OVA257–264 (Fig. 3 A).
Although the data discussed were obtained using free peptides, viral peptides must be generated by degradation of viral proteins. One important difference between free and viral peptides is that although free peptides can bind to cell surface class I molecules, viral peptides must be loaded onto class I molecules intracellularly. To mimic true viral peptides, we generated chimeric peptides in which a peptide portion of HIVtat is linked to VP2121–130. These chimeric peptides are able to cross the plasma membrane into cells (33), but they must be cleaved by peptidases to yield class I-binding fragments. In this way, chimeric peptides are processed more similarly to actual Ags than are free peptides. We performed the class I stabilization assay using HIVtat-VP2121–130 and, as a control, HIVtat-OVA257–264. As with the free peptides, HIVtat-VP2121–130 stabilized both Db and Dbm14, but not Kb or Kbm8 (Fig. 3,A). Surprisingly, although HIVtat-OVA257–264 stabilized Kb (but not Kbm8) as expected, it also stabilized Db, even though free OVA257–264 did not (Fig. 3 A). One possible explanation is that this stabilization could occur because the chimeric peptide harbors a novel fragment, ASIINFEKL, which contains appropriate anchor residues for binding Db (34, 35). Nonetheless, we can conclude from this analysis that both Db and Dbm14 bind in vitro to VP2121–130.
To test the longevity of Db and Dbm14 stabilization by VP2121–130, we incubated cells with the VP2121–130 peptide, then washed the cells and reincubated them at 37°C in the absence of peptide. Although both Db and Dbm14 levels decreased after 24 h without VP2121–130, cells bearing either allele still had increased surface expression of class I over baseline after 24 h (Fig. 3 B). In fact, the majority of the peptide-mediated increase in cell surface class I expression remains after 24 h, indicating that both Db-VP2121–130 and Dbm14-VP2121–130 complexes have a surface half-life of at least 24 h.
In a complementary experiment, we generated complexes of Db or Dbm14, mouse β2m, and VP2121–130. We then allowed these complexes to decay at 37°C for up to 48 h. We then immunoprecipitated stable complexes using the 28-14-8S mAb, which binds only to class I molecules that are complexed with peptide and β2m. Using this assay, we found that 80 ± 23% of Db-VP2121–130 complexes and 33 ± 5% of Dbm14-VP2121–130 complexes are stable for 48 h. Although both of these kinetic studies demonstrate a slightly lower half-life of Dbm14-VP2121–130 compared with Db-VP2121–130, Dbm14-VP2121–130, both alleles demonstrate significant binding over a period of at least 48 h. Therefore, the difference in stability between Db-VP2121–130 and Dbm14-VP2121–130 does not by itself account for their difference in antigenicity.
T cells from resistant mice are unable to respond to VP2121–130 when it is presented by Dbm14
Although Dbm14 can bind to VP2121–130 in vitro, we wanted to determine whether Dbm14 could present VP2121–130 in vivo with enough efficiency to prime a CD8 T cell response. Accordingly, we transferred purified T cells from B6, F1, or bm14 donors into rag−/− recipients expressing either Db/b (B6rag−/−) or Dbm14/bm14 (bm14rag−/−). Because a portion of the VP2121–130-reactive cells bind to Dbm14-VP2121–130, we would predict that if Dbm14 can prime Ag-reactive cells, transferred Dbm14-VP2121–130-specific F1 T cells should clonally expand and migrate to the site of infection. One day after the T cell adoptive transfer, mice were infected with TMEV. Seven days postinfection, brain-infiltrating lymphocytes were isolated and analyzed for tetramer staining. F1 T cells did not respond to VP2121–130 when primed by Dbm14, even though the F1 cells did respond when primed by Db, and even though that response contained Dbm14-VP2121–130-reactive cells (Fig. 4). This result indicates that Dbm14 cannot effectively present VP2121–130 as an Ag in vivo during CD8 T cell priming, despite binding to the peptide in vitro.
Although the T cell repertoire generated by Dbm14 differs from the repertoire generated by Db, it can mount a response to the VP2121–130 peptide when Db is the Ag-presenting molecule
To determine whether Dbm14 is defective in generating T cells responsive to VP2121–130, we wanted to transfer T cells from bm14 donors to Db-bearing recipients. However, donor cells that have not been exposed to Db will initiate a graft-vs-host response upon adoptive transfer to a Db-positive host. To overcome this, we performed reciprocal bone marrow transplants in which B6 or bm14 bone marrow cells were transplanted into irradiated B6 or bm14 recipients. In parallel work, we have shown that transplantation of bone marrow from H-2q mice transgenic for Db into Db-negative H-2q recipients fails to produce Db-VP2121–130-specific T cells capable of expanding in the setting of TMEV infection. However, adoptive transfer of T cells from Db-positive donors into the described bone marrow chimeras leads to an Ag-specific response, indicating that the Db-positive donor cells are capable of Ag presentation (36). Collectively, this demonstrates that in D-mismatched bone marrow chimeras, the T cell repertoire is restricted by the D allele of the recipient, and the presence of Db in the donor genome is not sufficient to generate a Db-restricted repertoire.
To test whether Dbm14 can form a VP2121–130-responsive repertoire, we made the following bone marrow transplants: B6 into B6, bm14 into B6, B6 into bm14, and bm14 into bm14. After allowing the recipients to recover and to regenerate T cells, we infected them with TMEV. One week after inoculation with virus, brain-infiltrating lymphocytes were isolated and stained with tetramers as before. As expected, B6 into B6 recipients responded well to the TMEV infection. The resulting T cells bound to both Db-VP2121–130 and Dbm14-VP2121–130 tetramers, whereas bm14 into bm14 recipients failed to respond to the virus (Fig. 5). Consistent with data from the adoptive transfer experiments (Fig. 4), mice with a Db-generated repertoire cannot respond to VP2121–130 when it is presented by Dbm14 (bm14 into B6 recipients). However, B6-to-bm14 recipients responded well to VP2121–130, indicating that although Dbm14 cannot adequately present the viral peptide during T cell priming, Dbm14 can generate a repertoire responsive to VP2121–130.
It is evident from Fig. 2 that T cells selected in the context of Db (Db/b and Db/bm14) and responding to Db-positive APCs preferentially recognize Ag in the context of Db. In the bone marrow chimera studies shown in Fig. 5, T cells selected in the context of Db had a similar preference for the Db tetramer. In contrast, T cells selected in the context of Dbm14 responding to Ag presented by Db did not show this preference. A direct comparison of the binding preferences revealed a statistically significant difference in the T cell preference for the two alleles (p = 0.0249). This indicates that the repertoires selected by Db and Dbm14 are different.
We have shown that the gene conversion mutant strain bm14 is susceptible to chronic TMEV infection and demyelination, whereas the parental strain B6 is resistant (Fig. 1). Mutant mice lack the ability to respond to the viral peptide VP2121–130, the single Ag driving the protective CD8 T cell response in wild-type B6 mice. Although TMEV has multiple proteins, each of which may generate a large number of potential peptide Ags, no other viral Ag is able to replace VP2121–130 in eliciting an effective response leading to viral clearance. The deficit in the bm14 response is due to the Dbm14 mutation itself, as determined by intercrossing B6 and bm14, then typing F2 offspring. Offspring with the genotype Dbm14/bm14 failed to mount an anti-VP2121–130 response, whereas offspring with one or two copies of Db responded to the virus (Fig. 2). All CTL responses observed in B6 mice were cross-reactive with Db and Dbm14; tetramers of either allele, complexed to the VP2121–130 TMEV peptide, labeled an expanded population of T cells isolated from the brains of infected animals. Dbm14 can bind to VP2121–130 in vitro, as shown by a class I stabilization assay (Fig. 3) and by immunoprecipitation of stable complexes after 48 h. However, in vivo, Dbm14 cannot present VP2121–130 in a manner adequate for CTL activation and expansion, despite the fact that the T cell repertoire generated by Dbm14 is capable of responding to the viral peptide (Figs. 4 and 5). Although both alleles generated T cell repertoires capable of responding to VP2121–130, the responses were different in allelic preference, indicating that the T cell repertoires generated by Db and Dbm14 were, in fact, different.
It is important that although Ag-specific T cells from both B6 and bm14 T cell repertoires recognize VP2121–130 in the context of Dbm14 (as demonstrated by tetramer staining), naive T cells cannot be primed to expand by Dbm14-VP2121–130. Clearly, priming of naive T cells and Ag recognition by activated T cells have different requirements for either Dbm14-VP2121–130 interactions or interactions between the TCR and Dbm14-VP2121–130 complexes. Furthermore, VP2-reactive CD8 T cells are only cytotoxic in the context of Db, despite the fact that they recognize tetramers of both Db and Dbm14 (our unpublished observation). To investigate the possibility that Dbm14 fails to present VP2121–130 due to instability of the class I-peptide complex, we measured stability using two separate techniques. First, we demonstrated that addition of exogenous VP2121–130 to cells bearing either Db or Dbm14 stabilizes cell surface class I for at least 24 h. Second, we measured the stability of class I-peptide complexes by immunoprecipitation, which demonstrated that a significant proportion of Dbm14-VP2121–130 complexes were stable after 48 h. This response suggests that the inability of Dbm14 to adequately present VP2121–130 is not due to unstable peptide binding. Although we have demonstrated similar stability of Db-VP2121–130 and Dbm14-VP2121–130 complexes, we detect a lower level of cell surface stabilization of Dbm14 than Db by the peptide (Fig. 3 A). This may represent less efficient peptide-class I complex formation by Dbm14 than Db, which might reduce the immunogenicity of Dbm14-VP2121–130 complexes. Alternatively, it is possible that the interaction between TCRs and the Dbm14-VP2121–130 complex, although detectable by tetramer staining, is insufficient to trigger T cell activation.
Gene conversion mutations have been shown to alter both Ag presentation (15) and the T cell repertoire (16, 37). However, neither of these changes is likely the direct substrate for selection of class I diversity. Rather, changes in repertoire or Ag presentation should affect selection only if they alter the ability of the organism to respond to a pathogenic challenge. In previous studies, the ability to present a nominal Ag has correlated with the ability to generate a T cell repertoire capable of responding to that Ag (14). Alleles that failed to present an Ag also failed to generate a repertoire that could respond to the Ag in the context of a related presenting allele (14). Our data, however, show that the ability to present and the ability to generate a responsive repertoire are independent; although Dbm14 was an inadequate presenter of VP2121–130, it did positively select T cells capable of responding to the peptide. Furthermore, we and others have demonstrated a case in which T cells specific for the SIINFEKL peptide are not positively selected by a class I allele, yet they respond to SIINFEKL in the context of the nonselecting allele (Ref. 38 and our unpublished observation). Thus, the ability to select Ag-responsive T cells does not depend on the ability to present that Ag, nor does being able to present an Ag mean that T cells capable of recognizing the Ag will be positively selected.
Previously, we demonstrated that class I polymorphisms cause significant changes in the T cell repertoire (16). We showed that when repertoires generated by closely related alleles respond to an Ag presented by one of the alleles, the repertoires respond differently, as indicated by their tetramer-binding preferences; however, these differences did not dictate whether a recipient could respond to TMEV. Rather, the failure of some animals to respond was due to defective presentation of the TMEV-derived peptide VP2121–130. It is possible that although Ag presentation is the limiting factor in an Ag response in some cases of polymorphism-related differences in Ag responsiveness, including Dbm14-VP2121–130 complex, other cases may exist in which the T cell repertoire is limiting. This is supported by work using the HSV-8p Ag from the herpesvirus HVH-1. C57BL/6 (Kb) and B6.C-H-2bm8 (Kbm8) mice differ in their ability to resist HVH-1 infection; this discrepancy is attributed to a difference in the clonal diversity and avidity of the CTL response to the HVH-1 Ag HSV-8p (13). The authors conclude that the difference in the CTL response is due to the fact that Kbm8 positively selects a more diverse, more avid CD8 T cell repertoire to that selected by Kb. Supporting this conclusion, CD8 T cells developing in a Kbm8 thymus show broader cross-reactivity between Kb/HSV-8p and Kbm8/HSV-8p than do cells developing in a Kb thymus, even when the HSV-8p Ag is presented by Kb/bm8 APCs (37). Thus, T cell repertoire differences can, in some cases, account for differences in the nature of the CD8 T cell response to an Ag. It is unclear, however, whether repertoire differences ever dictate whether or not an individual will respond at all to an Ag, given the ability to present the Ag.
Newly arising class I variants are selectively maintained in populations, and their ability to alter the immune response is a mechanistically plausible means by which selection may occur. Although both the T cell repertoire and Ag presentation are affected by class I polymorphisms, we have shown that differences in presentation can alter the ability to respond to an infectious challenge.
The authors have no financial conflict of interest.
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.
This work was supported by National Institutes of Health Grant P01 NS38468.
Abbreviations used in this paper: TMEV, Theiler’s murine encephalomyelitis virus; β2m, β2-microglobulin.