Escape from CD8⁺ T Cell Response by Natural Variants of an Immunodominant Epitope from Theileria parva Is Predominantly Due to Loss of TCR Recognition

CD8⁺ T cells play an important role in mediating immunity to a variety of viral, bacterial, and protozoal intracellular pathogens (1). Typically, these CD8⁺ T cell responses focus on a single or limited number of dominant epitopes (2, 3). Sequence polymorphism of dominant Ags that result in amino acid substitutions in epitope residues critical for either binding to the presenting MHC class I (MHC I) molecule (4–6) or recognition by cognate TCR (4–8) can abrogate immune recognition by CD8⁺ T cells, leading to strain-specific immunity.

The importance of epitope polymorphism and strain specificity has been most thoroughly documented in CD8⁺ T cell responses against RNA viruses such as HIV and hepatitis C virus, which replicate rapidly and have a high rate of mutation (7–13). Selective pressure exerted by CD8⁺ T cells promotes the survival of viruses with mutations in the immunodominant epitopes by allowing escape from immune recognition (9, 14). Longitudinal studies of CD8⁺ T cell responses of patients infected with HIV or HCV have provided evidence that limited diversity of the TCR repertoire of CD8⁺ T cells specific for immunodominant epitopes favors emergence of escape mutants (7, 8). Diverse TCR repertoires are not only more likely to contain high-affinity T cell clonotypes that achieve enhanced epitope recognition and pathogen control (15, 16), but also, by virtue of their divergent fine antigenic specificities, convey a greater capacity to recognize and control emergent mutant viruses expressing epitope variants (7, 8, 11). Consequently, TCR diversity is potentially a critical determinant of protective efficacy of CD8⁺ T cell responses against antigenically polymorphic pathogens and is therefore an important qualitative parameter to consider in the development of successful subunit vaccination strategies (17–19).

Strain-specific CD8⁺ T cell responses have also been described for several protozoa including malaria (20, 21), but the precise relationship of antigenic variability and immunity to heterologous parasite strains is less well understood. Unlike RNA viruses, protozoa are genetically relatively stable, and hence most of the antigenic polymorphism represents changes that have occurred over a long period of evolution. One such parasite is Theileria parva, the causal agent of an acute and usually fatal lymphoproliferative disease of cattle known as East Coast Fever, which is a major constraint on livestock production in a large area of southern and eastern Africa. Cattle that naturally recover from infection or are immunized by infection and treatment develop long-lasting immunity to the homologous parasite isolate, but show incomplete immunity to heterologous isolates (22). Observations on the kinetics of the CD8⁺ T cell response, coupled with the demonstration that immunity can be adoptively transferred from immune to naive twin calves by highly enriched populations of CD8⁺ T cells, indicate that immunity is mediated by CD8⁺ T cells specific for the intralymphocytic schizont stage of the parasite (23–25). Moreover, the parasite strain specificity of CD8⁺ T cell responses induced by immunization with a single parasite strain has been shown to correlate with protection against challenge with a heterologous parasite strain (26, 27). Based on these and other experimental observations, it had been proposed that strain specificity of immunity is a consequence of focusing of the CD8⁺ T cell response on a limited number of polymorphic immunodominant epitopes (28, 29). Subsequent studies of responses to defined T. parva Ags recognized by CD8⁺ T cells from immune cattle have confirmed the immunodominant nature of the response (30–32). We have shown that single epitopes in the Tp1 and Tp2 Ags account for >60% of the CD8⁺ T cell response in animals homozygous for the A18 and A10 class I haplotypes, respectively (33). Moreover, recent analyses of field isolates of T. parva confirmed that both of these epitopes are polymorphic. The Tp2 Ag was found to be particularly variable, exhibiting substitutions in >70% of the amino acid residues, resulting in 23 allelic variants of the dominant A10-restricted epitope (Tp2_49–59) among the 80 sequences examined (34). This unique set of naturally occurring parasite epitope variants provides a useful resource both to dissect the basis of strain specificity of the CD8⁺ T cell response and to investigate potential selective forces in generating polymorphism.

The aims of the current study were first to examine the diversity and stability of the TCR repertoires of CD8⁺ T cells specific for these two dominant epitopes and second, by analyzing the fine specificity of sets of CD8⁺ T cell clonotypes specific for the highly polymorphic Tp2_49–59 epitope, to determine the relative contribution of altered MHC binding and TCR interaction in facilitating escape from T cell recognition. Our results demonstrate that: 1) both Tp1_212–224 and Tp2_49–59 elicit CD8⁺ T cell responses expressing diverse TCR repertoires that remain relatively stable following repeated parasite challenge; 2) although Tp2_49–59-specific CD8⁺ T cell clonotypes exhibit a broad range of fine specificities, all 17 natural epitope variants examined exhibit complete or partial escape from immune recognition; and 3) immune escape is determined predominantly by disruption of TCR recognition, with mutations in a single residue playing a particularly dominant role in escape.

Four Holstein-Friesian animals homozygous for the A10 or A18 MHC I haplotypes were selected for the study by a combination of serological typing with MHC I-specific mAbs (35) and MHC I allele-specific PCR (36). The animals were aged 18–36 mo at the outset of the study and were maintained indoors on rations of hay and concentrate. Cattle were immunized against the T. parva Muguga stock by infection with cryopreserved sporozoites and simultaneous administration of a long-acting formulation of oxytetracycline as described previously (22). Three of the animals were challenged with a lethal dose of sporozoites on two occasions at ∼18-mo intervals following immunization. All animal experimental work was completed in accordance with the Animal (Scientific Procedures) Act 1986.

Cloned T. parva-specific CD8⁺ T cells were generated as described previously (37). In brief, PBMC from immunized animals were stimulated three times at weekly intervals by coculture with γ-irradiated autologous T. parva-infected cells. Prior to the third stimulation, cell lines were depleted of CD4⁺ T cells and γδ T cells by Ab and complement-mediated lysis. During the third stimulation, the culture medium was supplemented with 100 U/ml recombinant human IL-2 (Chiron, Emeryville, CA). Clones were generated from the CD8⁺ T cell-enriched cell lines by limiting dilution 7 d after the third stimulation and expanded by restimulation as detailed in Goddeeris and Morrison (37). All culture was conducted in RPMI 1640 medium supplemented with 10% FBS, 20 mM HEPES buffer, 5 × 10⁻⁵ M 2-ME, 2mM l-glutamine, and 50 mg/ml gentamicin.

Total RNA was extracted from cloned CD8⁺ T cells using Tri-reagent (Sigma-Aldrich, Dorset, U.K.) and cDNA subsequently synthesized using the Reverse Transcription System (Promega, Madison, WI) with priming by the Oligo (dT)15 primer, according to the manufacturers’ instructions. TCR β-chains (TRB) were then PCR amplified using either Vβ subgroup-specific or Pan-Vβ primers as described previously (38, 39) and the products sequenced. Sequence analysis was performed using the DNAsis Max v2.0 software package under default conditions (Miriabio, Alameda, CA). Throughout the manuscript, we have used the World Health Organization-International Union of Immunological Societies TCR nomenclature system as used by Arden et al. (40) for Vβ genes, whereas Jβ gene nomenclature is based upon their organization in the bovine TRB locus (41).

Standard 4-h [¹¹¹In]-release cytotoxicity assays were used to examine the antigenic-specificity of the CD8⁺ T cell clones, using autologous Theileria annulata-transformed cells that had been incubated with peptide for 1 h prior to the assay as target cells.

All assays were conducted in duplicate, and controls included T. annulata-infected target cells without added peptide and, where appropriate, MHC-mismatched T. parva-infected cells. Percentage specific lysis was calculated as: ([sample release −spontaneous release] × 100%/[maximal release − spontaneous release]) and expressed as the mean of the duplicated assays. Maximal and spontaneous release were derived from triplicates of target cells incubated in 0.2% Tween 20 and RPMI 1640/5% FCS, respectively. All peptides used in this study were supplied by Pepscan Systems (Lelystad, The Netherlands).

In assays to identify Tp2_49–59 and Tp1_212–224-specific T cell clones, target cells were incubated with peptides at a concentration of 1 μg/ml. Because of the large numbers of T cell clones analyzed, effector cells were not counted in these assays but were used in sufficient quantity to ensure the E:T ratio was >1:1. A standard cutoff of >5% specific cytotoxicity was used to define positive clones. In all assays, this was well in excess of 3 SD above the mean spontaneous release value of the respective target cells and was also well in excess of 3 SD above the mean levels of cytotoxicity obtained with MHC-mismatched T. parva-infected and unpulsed T. annulata-infected target cells.

Assays to examine the ability of specific T cell clones to recognize variants of the Tp2_49–59 epitope used autologous T. annulata-infected cells incubated with 100 ng/ml peptide (∼1 × 10⁻⁷ M), prior to addition of effectors at an E:T ratio of 10:1.

The capacity of variants of the Tp2_49–59 peptide to bind MHC I was determined by their ability to compete with Tp2_98–106, a subdominant epitope also presented by the N*01201 MHC I allele. Autologous T. annulata-infected target cells were preincubated with individual Tp2_49–59 peptide variants at serial 3-fold dilutions ranging from 3 μg/ml to 10 ng/ml for 1 h prior to addition of Tp2_98–106 at 100 ng/ml, so that the competitor/target peptide ratio ranged from 30:1 to 0.1:1. After incubation for a further hour, 4-h [¹¹¹In]-release cytotoxicity assays were performed using a Tp2_98–106-specific CD8⁺ T cell clone at an E:T ratio of 10:1. The MHC-binding capacity of the Tp2_49–59 peptide variants was reflected in the inhibition of cytotoxicity as a consequence of competitive blocking of MHC binding and presentation of the Tp2_98–106 peptide.

In previous studies of T. parva-specific CD8⁺ T cell clones from pairs of immune cattle homozygous for the A10 and A18 MHC I haplotypes, ≥60% of the clones were found to be specific for the immunodominant epitopes Tp2_49–59 and Tp1_212–224, respectively (Table I). To examine the diversity of the TCR repertoire expressed by the CD8⁺ T cell response to these dominant epitopes following immunization, TRB cDNAs were sequenced from a panel of CD8⁺ T cell clones generated from each of the four animals. Sequence data for single functional TRB were obtained from between 50 and 69 CD8⁺ T cell clones from each animal (Table II). The response in all four animals exhibited considerable diversity in the expressed TRB. In each animal, the clones used a large number of different TRB variable gene segment (TRBV) genes (range 8–22) representing a range of TRBV subgroups (range 6–10) and had structurally dissimilar CDR3 β-chain (CDR3B) sequences, including different CDR3 sequences associated with the same TRBV gene (e.g., the Vβ14.1⁺ clonotypes in Animal I and Vβ3.1⁺ clonotypes in Animal III).

Despite this high level of TRB diversity, inter- and intra-animal comparisons revealed clonotypes expressing highly homologous TRB in both the Tp2_49–59 and Tp1_212–224 responses. This included multiple examples of clonotypes expressing semiconserved CDR3B sequences (i.e., of the same length and with similar amino acid sequences) in conjunction with identical or similar TRBV and/or TRB joining gene segment (TRBJ) genes (Table II). Additionally, clonotypes expressing completely conserved or public TRB sequences were identified in both the Tp1_212–224 (Vβ1.9⁺-SQDYPTNDPL-Jβ3.3⁺) and Tp2_49–59 (Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺) responses (Table II). Notably, nucleotide sequence analysis revealed that in Animal I the Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ clonotype was composed of at least three unique clonal expansions (Tables II, III).

The TRB sequence data also provided information on the representation of different clonotypes in the responding populations. In three of the animals (II, III, and IV), from which between 10 and 13 clonotypes expressing unique TRB sequences were identified, there was a clear hierarchical structure, with two or three numerically dominant clonotypes accounting for >75% of the clones, whereas the remaining clonotypes were represented at lower levels (Fig. 1). Clones from the fourth animal (Animal I) exhibited a less pronounced hierarchy, with the three most numerically predominant clonotypes accounting for only 28% of the clones and a larger number of clonotypes, 27, represented. These results demonstrate that the Tp1_212–224 and Tp2_49–59-specific CD8⁺ T cell memory pools established by immunization are polyclonal and express diverse TRB, but are generally dominated by a limited number of abundant clonotypes.

To determine whether repeated challenge with the same parasite strain alters the memory CD8⁺ T cell repertoire and hence, potentially, the fine specificity of the response, three of the animals were subjected to two homologous parasite challenges over a period of 36 mo. CD8⁺ T cell clones isolated 30 d after the second challenge were analyzed for antigenic specificity and expressed TRB sequences. Screening of clones demonstrated that a majority of clones (>61%) from the A10⁺ and A18⁺ animals remained focused on the Tp2_49–59 and Tp1_212–224 epitopes, respectively (Table I). Comparison of the TRB sequences of these clones (postchallenge [PC]) with those analyzed following initial immunization (postimmunization [PI]) in the same animal revealed a slight reduction in clonotypic complexity in each animal, but an overall preservation of the clonotypic structure of the Ag-specific populations (Fig. 1). There was also a high level of stability in clonotype content: 82–99% of the clones in the postchallenge populations were identical to clonotypes detected postimmunization. This stability was largely attributable to preservation of clonotypes that were abundant postimmunization (Table II). Some of these clonotypes maintained a dominant position in the hierarchy (e.g., the Vβ14.1⁺-SVGNSNYEQ-3s7⁺ clonotype in Animal II), whereas others showed marked changes in their relative abundance; for example, the Vβ1.7⁺- SHEWYSTDTQ-3s4⁺ clonotype in Animal III showed a decrease in frequency, whereas Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ clonotypes became dominant in both Animals I and II following challenge. Detection of clonotypes present at low frequency (<5%) postimmunization was more variable, some being detected in both postimmunization and postchallenge populations whereas others were restricted to either population. Although the latter might reflect genuine changes due to loss/gain of clonotypes during evolution of the Ag-specific memory pools, at least some of the discrepancies probably reflect an inherent limitation of small sample sizes to reproducibly detect low frequency clonotypes, thus underestimating the overall stability of the memory pools.

The recent identification of a large number of allelic variants of the Tp2_49–59 epitope in field isolates of T. parva (34) provided an opportunity to analyze how polymorphism in the epitope influences T cell recognition and strain specificity of the T cell response.

We first sought to identify the residues in Tp2_49–59 responsible for MHC binding, using an MHC I-binding competition assay. This assay used a CD8⁺ T cell clone specific for a second, subdominant epitope in Tp2 (Tp2_98–106) presented by the same class I allele (N*01201) and tested the ability of a series of Tp2_49–59 synthetic peptides containing single alanine substitutions (peptides A1–11) to competitively inhibit recognition of Tp2_98–106 peptide by this clone in a cytotoxicity assay. The reference Tp2_49–59 peptide (natural variant 1 [NV1]) inhibited recognition of Tp2_98–106 in a dose-dependent manner, resulting in >80% inhibition when added at 30-fold excess of the Tp2_98–106 peptide concentration. All of the alanine-substituted peptides, with the exception of A11, showed levels of competitive inhibition similar to that observed with the wild-type Tp2_49–59 peptide. By contrast, P11 showed no inhibition of Tp2_98–106 recognition, identifying the lysine at position 11 (K₁₁) as a critical MHC-binding residue (Fig. 2).

To determine whether the TCRs expressed by Tp2_49–59-specific CD8⁺ T cell clonotypes influenced their fine antigenic specificity, we examined the ability of different T cell clones to recognize the Tp2_49–59 single-alanine substituted peptides. A panel of 23 T cell clones expressing 12 different TRB was used; these included representatives from four different clonal expansions expressing the Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ TRB. For each clone, the cytotoxic activity against the alanine-substituted peptides (A1–A10) was expressed as a percentage of that obtained with the reference (NV1) Tp2_49–59 peptide (Table IV). The relative cytotoxic activity recorded ranged from 160 to 0%, demonstrating that alanine substitution could in some cases enhance recognition of the Tp2_49–59 peptide (≥110%) or in others have a negligible effect (∼90–110%), but in the majority of cases reduced recognition to some extent (<90%). Assigning relative cytotoxic activity ≤10% as complete abrogation of recognition provides a simplified, binary, and informative parameter by which to describe the fine specificity profile of the individual clones. The results demonstrated that nearly all of the different TRB were associated with a unique profile. Thus, among the non–Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ clones, 10 distinct fine specificity patterns were observed for the 11 TRB clonotypes represented; clones expressing the same TRB amino acid sequence exhibited similar recognition profiles. In contrast, clones expressing the shared Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ TRB, which represented a large component of the postchallenge response in both A10⁺ animals, had divergent fine specificities manifest as differences in recognition of the A5, A9, and A10 peptides. This divergence in fine specificity was observed between clones that expressed the same TRB nucleotide sequence. Different fine specificities most probably reflect heterogeneity in the paired TCR α-chain (TRA) expressed in distinct clonal expansions, indicating that the clonotypic composition of the Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ population is more complex than TRB sequencing alone indicates.

All of the clonotypes retained the ability to recognize the A2 and A3 peptide variants to some extent, indicating that either the serine residues at positions 2 and 3 are generically of minor significance in TCR recognition or alternatively that the relatively conservative serine to alanine substitutions at P2 and P3 are well tolerated. Recognition of the other alanine-substituted peptides, with the notable exception of A8, varied between the different clones, demonstrating that the substituted residues serve as determinants of fine specificity of some clones dependent on their expressed TCR. However, the most striking result was that none of the clones demonstrated cytotoxicity against the A8 peptide, identifying the residue at position 8 as critical for recognition by all of the Tp2_49–59-specific clones, irrespective of the expressed TCR. In separate experiments, uncloned T. parva-specific CD8⁺ T cell lines from Animals I and II also did not recognize A8 peptide but retained some recognition of peptides substituted at all other positions, indicating that the diversity of TCR could accommodate most single-alanine substitution variants but A8 substitution was sufficient to abrogate recognition by the entire Tp2_49–59-specific response.

Seventeen of the recently identified unique NV of Tp2_49–59 (NV2–18) were examined for MHC binding activity, using the MHC I-binding competition assay described above. These included variants that have amino acid substitutions at between one and eight of the amino acid residues relative to the reference NV1 sequence (Table V). The majority of these NV were found to have inhibitory activity, but at varying levels, indicating that they are able to bind to the presenting N*01201 MHC I allele to some extent. All seven variants that maintained the K₁₁ residue were capable of sufficient competitive inhibition of Tp2_98–106 binding to reduce cytotoxicity by >50% when present at an excess of 30:1 relative to the target peptide (Fig. 3A). The remaining variants all had an arginine substitution at position 11 (R₁₁) and exhibited a variable degree of reduced efficiency in MHC binding: when present at an excess of 30:1. NV4, 5, 7, and 8 inhibited cytotoxicity by ∼50%, whereas NV13, 14, 11, and 12 exhibited progressively less inhibition, and NV15 and 16 did not have any inhibitory effect (Fig. 3B). Notably, positions 4 and 7 were substituted in NV11–16 but not NV4, 5, 7, and 8, suggesting that either one or both of these positions may serve as subsidiary MHC-binding residues. Thus, most of the NV variants (15 out of 17) had the potential to be presented by MHC I molecules for recognition by cognate TCRs, although for several of these (i.e., NV13–16), presentation is likely to be constrained by weak MHC I binding.

Peptides representing the allelic variants of Tp2_49–59 were screened for recognition by the 23 specific CD8⁺ T cell clones described previously. As anticipated, NV15 and 16, which do not bind to the presenting MHC I allele, did not elicit a response from the T cell clones (data not shown). In addition, 9 of the remaining 15 variants were not recognized by any of the T cell clones (Table VI). The five variants (NV3, 4, 5, 7, and 17) that retained recognition were recognized by different subsets of the Tp2_49–59-specific T cell clones, and in the case of NV17 only by a single clone. Most strikingly, the peptides that were not recognized included NV2, which differs from NV1 solely by a lysine to glutamine substitution at position 8. Conversely, the five variants that retained recognition were distinguished by being the only ones in which K₈ is conserved (NV16 also has lysine at position 8 but is a non-MHC binder). Uncloned T. parva-specific CD8⁺ T cell lines from Animals I and II exhibited a similarly limited cross-recognition of four of these variants (NV3, 4, 5, and 7; lack of detectable recognition of NV17 most likely reflects limited representation of the single clonotype recognizing this variant), indicating the results obtained for the representative clonotypes broadly reflect those for the Tp2_49–59-specific responses. Together, these results demonstrate that substitution of K₈ in Tp2_49–59 variants is a dominant determinant in their ability of escape recognition of CD8⁺ T cell responses using a diverse TRB repertoire.

Despite conservation K₈ in NV3, 4, 5, 7, and 17, the lack of recognition of these variants by a variable proportion of the T cell clones indicated that other residues are also critical in determining recognition by particular TCR. For example, NV3, which differs from the reference Tp2_49–59 sequence only by one amino acid (lysine to arginine substitution at P1), is not recognized by several clonotypes (Table VI); this is consistent with the absence of recognition by these clones of Tp2_49–59 when the P1 residue has been substituted with alanine (Table IV), demonstrating the importance of the P1 residue in defining the specificity of this subset of clonotypes. Similarly, NV4, which differed from the reference sequence only by a lysine to arginine substitution at P11, was not recognized by clones expressing Vβ14.1-SVGNSNYEQ-3s7, indicating that as well as serving as an MHC binding residue, this residue also functions as a critical determinant of TCR recognition for this clonotype. Thus, substitutions at amino acid residues other than P8 can result in loss of recognition of the Tp2_49–59 epitope by T cells expressing particular TCRs.

Finally, we sought to examine the association between the observed recognition specificities of Tp2_49–59-specific clonotypes and sequence variation in the epitope among the NV identified in the field. Comparisons for each amino acid residue in the Tp2_49–59 epitope between the percentage heterozygosity in the 80 field isolate dataset and the percentage of CD8⁺ T cell clonotypes that had recognition abrogated by single-alanine substitution revealed a marked association (Fig. 4). The overall correlation between the two values (Pearson product moment correlation = 0.67, p = 0.017) indicates that amino acid substitutions in NV have occurred most frequently in positions that would preferentially facilitate escape from Tp2_49–59-specific CD8⁺ T cell clonotypes, suggesting that CD8⁺ T cell-mediated immunological selection may have played a role in the evolution of epitope variation.

In the current study, we have used a combination of TRB sequencing and in vitro analysis of recognition of natural and alanine-substituted variant epitope peptides by specific CD8⁺ T cell clones to investigate how polymorphisms in immunodominant T. parva epitopes affect the strain specificity of immunity to this parasite.

As induction of narrow TCR repertoires has been identified as a contributor to the development of strain-specific immunity to some viruses, we first evaluated the diversity of TCR expressed in Tp2_49–59- and Tp1_212–224-specific CD8⁺ T cell populations. Analysis of postimmunization memory pools in individual animals showed that both epitopes elicited responses characterized by highly diverse TCR repertoires, expressing a range of TRBV genes with diverse CDR3B sequences Additionally, substantial differences in TRB repertoires of the epitope-specific CD8⁺ T cells between the pairs of A10⁺ and A18⁺ animals illustrate that there is little structural constraint on the TCRs used to recognize either epitope. However, the presence of clonotypes using semiconserved and public TRB suggests TRB with particular TRBV-CDR3B-TRBJ characteristics may be preferentially selected in both Tp2_49–59 and Tp1_212–224 responses. We identified between 10 and 27 unique clonotypes in individual epitope-specific postimmunization memory populations, with each population typically containing a limited number of highly abundant clonotypes. Overall, the composition and diversity of these populations were comparable to that described for human and murine CD8⁺ T cell responses to viral epitopes using equivalent in vitro cloning-based methodologies (42, 43). Comparison with results of direct ex vivo clonotypic analysis performed for these viral epitopes using tetramer isolation have shown that in vitro cloning provides an accurate representation of the abundant clonotypes but underrepresents low frequency clonotypes and so underestimates the diversity of the clonotypic composition and TCR repertoire (42, 44). Detailed analysis of the clonotypic composition of CD8⁺ T cell responses to various viral epitopes based on TRB sequencing have estimated that they contain >1000 clonotypes that conform to a log-series and/or power law distribution, with a few abundant clonotypes and an extensive low-frequency clonotype tail (45–47). The clonotypic structures of the parasite epitope-specific CD8⁺ T cell populations in this study generally concurred with a log-series distribution but with truncated tails, suggesting that the component of the response constituted by low prevalence clonotypes is underappreciated in our results. Thus, the true diversity of the TCR repertoires used by CD8⁺ T cell responses to Tp2_49–59 and Tp1_212–224 is likely to be much greater than documented in this study.

During continuous or repeated Ag exposure, multiple forces including stochastic recall, avidity selection, and clonotype exhaustion and deletion can operate to modify the CD8⁺ T cell repertoire (42, 44, 48–52). Understanding how repeated challenge modulates the TCR repertoire and clonal composition of the CD8⁺ T cell response to Tp2_49–59 and Tp1_212–224 had obvious relevance to our studies of parasite strain specificity. Comparison of the postimmunization memory CD8⁺ T cell response with that following two homologous parasite challenges over a 3-y period demonstrated that the response remained focused on the Tp2_49–59 and Tp1_212–224 epitopes and revealed a remarkable level of stability in the clonal composition of the epitope-specific CD8⁺ T cell populations in all three animals examined. Over 80% of clonotypes detected following challenge were represented in the initial postimmunization populations. Furthermore, as the majority of the observed discrepancies between postimmunization and postchallenge populations were associated with low-frequency clonotypes, which are most liable to inconsistent detection during analysis (50, 53), it is likely that the stability of the populations has been underestimated. Nevertheless, the clonal composition of the Ag-specific populations was not static; a few clonotypes showed dramatic alterations in their position within the clonotypic hierarchies, either increasing (e.g., the Vβ28.1⁺-AEYGGENTQPL-Jβ3.2⁺ TRB clonotype in both A10⁺ animals) or decreasing (e.g., Vβ1.7⁺-SHEWYSTDTQ-Jβ3.4⁺ TRB clonotype in Animal III) in abundance or absent from (e.g., the VβX⁺-SKAAAEDGYEQ-Jβ3.7⁺ clonotype in Animal II) postchallenge responses. There was also some evidence of immunofocusing (i.e., increased numerical dominance by larger clonotypes) in the responses, reflected in a left shift of the postchallenge compared with postimmunization responses in Fig. 1. Selection on the basis of TCR avidity, which has been implicated in preferential expansion of particular clonotypes within CD8⁺ T cell responses to viral infections (52, 54), may have been responsible for the expansion of certain clonotypes in the current study. Similarly, the failure to detect the VβX⁺-SKAAAEDGYEQ-Jβ3.7⁺ clonotype following challenge in Animal II, in which this clonotype was initially abundant, may have been due to failure to sustain the response because of low avidity or to other mechanisms such as clonal exhaustion. Importantly, despite these changes in composition of the repertoire, we were able to show in one of the animals that the polyclonal postimmunization and postchallenge CD8⁺ populations had similar recognition profiles for Tp2_49–59 variants, demonstrating that the changes did not substantively alter the pattern of strain specificity.

The Tp1 and Tp2 Ags are proteins of unknown function, which possess a predicted signal peptide and hence are believed to be secreted from the parasite into the host cell cytosol. Sequencing of the Tp1 and Tp2 genes from field isolates of T. parva has demonstrated polymorphism in both Ags, including the epitopes analyzed in the current study. We had previously reported differential recognition of one Tp1 allelic variant and two Tp2 variants by CD8⁺ T cell clones specific for the respective dominant epitopes (33). The availability of a large panel of NV of the Tp2_49–59 epitope, together with specific T cell clones of defined TRB clonotypes, provided an opportunity to investigate the basis of escape from immune recognition of this epitope by the specific cell response. Twelve of the variant epitopes were not recognized by any of the CD8⁺ T cell clones examined; the remaining five nonreference variants were recognized only by some TCR clonotypes, and with the exception of NV3, none of the individual clones demonstrated the same level of cytotoxic activity elicited by the reference Tp2_49–59 peptide. Hence, all variants, including three that had only single amino acid substitutions, exhibited the ability to completely or partially escape from immune recognition. These results, together with data from experiments using peptides containing single alanine substitutions, were used to investigate the amino acid residues that determine specificity and the relative contributions of loss of MHC binding and TCR recognition to variant specificity. Using the alanine-substituted peptides in an MHC I-binding competition assay, we identified the C-terminal lysine (K₁₁) as a dominant MHC-anchor residue. The failure to detect additional anchor residues toward the N terminus of the epitope may reflect weaker binding of such residues that is not detected by the assay used. Alternatively, serine (P2 and P3) to alanine may represent a substitution that is sufficiently conservative to not disrupt MHC binding, or either of the P2 or P3 serine residues may serve as an anchor, and loss of MHC binding may require the simultaneous substitution of both residues. In this regard, it is notable that the Tp2_98–106 epitope, which is also presented by the N*01201 MHC I gene product and has a lysine residue at the C terminus, has a serine residue at P2. Further analysis using substitutions with residues that are more physicochemically distinct from serine are required to address this question. Analysis of the natural Tp2_49–59 variants from field isolates examined in this study showed that more than half (10 out of 18) had arginine at P11. This relatively conserved substitution maintained MHC-binding capacity, although with a reduced efficiency; in some Tp2_49–59 variants, additional substitutions reduced MHC binding further but only two of the peptides were incapable of MHC binding. Given that CD8⁺ T cells can recognize very low numbers of peptide–MHC class I ligands (55, 56), these results indicate that polymorphism of MHC I binding residues makes a relatively minor contribution to strain specificity of Tp2_49–59 recognition. By inference, therefore, strain specificity is determined largely by changes that disrupt TCR recognition of the peptide–MHC I ligand.

The panel of Tp2_49–59-specific TRB clonotypes exhibited marked plasticity in their recognition profile when tested against the single alanine-substituted peptides. This plasticity manifest as differential abilities of individual clonotypes to cross-recognize the A1, A4, A5, A6, A7, A9, and A10 alanine-substituted variants and as the capacity to recognize variable numbers (5–9) of the variant peptides. Four CD8⁺ T cell TRB clonotypes expressing the same Vβ28.1⁺-AEYGGENTQPL-Jβ3s2⁺ chain also exhibited distinct fine specificities, which we hypothesize is due to heterogeneity in the coexpressed TRA. Pairing of diverse TRA with highly conserved TRB has been demonstrated in the CD8⁺ T cell response to other immunodominant epitopes including NP_366–374 of influenza A, FL8 from HIV, and NY-ESO-1_157–165 of a testicular cancer Ag (57–59). Intriguingly, the differences in specificity for Tp2_49–59 were primarily at the C terminus of the peptide, suggesting that the influence of the TRA on the fine specificity of the Vβ28.1⁺-AEYGGENTQPL-Jβ3s2⁺ clonotypes is acting indirectly by modifying the conformation of the CDRs of the shared TRB (59).

Despite the breadth of fine specificities observed among the Tp2_49–59-specific clonotypes, alanine substitution at P8 completely abrogated recognition by all of the clonotypes analyzed. Moreover, none of the natural Tp2_49–59 variants containing amino acid substitutions at P8 were recognized by any of the T cell clonotypes. Because the alanine-substituted A8 peptide and an NV (NV2) containing a single lysine to glutamine substitution at P8 both retained MHC binding, these findings indicate that the lysine at P8 in Tp2_49–59 plays a key role in TCR recognition of the epitope. Structural studies of other dominant peptides bound to MHC have demonstrated that such key residues can either form a prominent component of the exposed peptide feature with which the TCR interacts (57) or adopt a position within the groove that is critical for stabilizing the conformation of the bound peptide. We have recently reported that the Tp1_212–224 epitope represents an example of the latter; alanine substitution of the glutamine residue at P9 abrogates recognition of this epitope by a range of TRB clonotypes, and the structural data demonstrate that this residue is critical in stabilizing the prominent P6–P8 bulge on which the TCR response is focused (60).

Unlike viral pathogens such as HIV and influenza that undergo rapid antigenic changes, often within individual hosts, the polymorphism in T. parva Ags is believed to reflect historical evolution of the pathogens resulting in allelic variants that are stable within current parasite populations. This is consistent with Tp1 and Tp2 sequencing data from laboratory-maintained parasite isolates, which have shown that sequences remain identical following tick passage of the parasites (R. Pelle, unpublished observations). Recent analyses of Tp2 nucleotide sequences have revealed evidence of positive selection at the codon level over the length of the gene, but the failure to detect obvious enrichment for positively selected residues within six defined epitopes compared with the remainder of the sequence suggested that the changes may not be due to immune selection (34). However, immune selection may not be detectable by this approach if the Tp2 protein contains a large number of additional undefined CD8⁺ T cell epitopes. The present study demonstrated a significant correlation between heterozygosity of individual residues in the panel of natural Tp2_49–59 variants and the proportion of TRB clonotypes that were affected by alanine substitution of the respective residues. Of particular note were the findings that the P8 residue showed the highest level of heterozygosity and that alanine substitutions in P3 and P6, which are invariant in NV of Tp2_49–59, resulted in little or no loss of recognition of the epitope. These results provide supportive evidence that the Tp2_49–59 epitope has been subject to immune selection, but further studies are required to determine whether this is a general feature of dominant polymorphic T. parva epitopes.

We thank the staff at the University of Edinburgh Veterinary School for the care given to the animals and the technical advice and support provided by Dr. F. Katzer, A. Burrells, and Dr. D. Shaw.

The authors have no financial conflicts of interest.