Epitope-specific TCRβ repertoire diversity imparts no functional advantage on the CD8⁺ T cell response to cognate viral peptides

Abstract

TCR repertoire diversity has been convincingly shown to facilitate responsiveness of CD8⁺ T cell populations to mutant virus peptides, thereby safeguarding against viral escape. However, the impact of repertoire diversity on the functionality of the CD8⁺ T cell response to cognate peptide-MHC class I complex (pMHC) recognition remains unclear. Here, we have compared TCRβ chain repertoires of three influenza A epitope-specific CD8⁺ T cell responses in C57BL/6 (B6) mice: D^bNP_366–374, D^bPA_224–233, and a recently described epitope derived from the +1 reading frame of the influenza viral polymerase B subunit (residues 62–70) (D^bPB1-F2₆₂). Corresponding to the relative antigenicity of the respective pMHCs, and irrespective of the location of prominent residues, the D^bPA₂₂₄- and D^bPB1-F2₆₂-specific repertoires were similarly diverse, whereas the D^bNP₃₆₆ population was substantially narrower. Importantly, parallel analysis of response magnitude, cytotoxicity, TCR avidity, and cytokine production for the three epitope-specific responses revealed no obvious functional advantage conferred by increased T cell repertoire diversity. Thus, whereas a diverse repertoire may be important for recognition of epitope variants, its effect on the response to cognate pMHC recognition appears minimal.

Defense against a broad range of invading pathogens, while at the same time protecting the host from immune-mediated damage, necessitates a naive T cell compartment exhibiting both exquisite specificity and substantial diversity in TCR usage. Certainly, several studies in which naive T cell diversity has been narrowed, either by means of naturally occurring mutations or introduction of transgenes, have shown deficiencies in the ability of these populations to respond in vivo compared with naturally diverse TCR repertoires (1, 2). Given that a naive repertoire of sufficient diversity is required to generate a detectable T cell response, it is of considerable interest to ascertain whether TCR diversity within epitope-specific populations confers any specific advantage on the T cell response. To date, attempts to address this question have largely focused on the comparative abilities of diverse and narrow CTL repertoires to respond to epitope variants, as occurs during viral escape after chronic infection with viruses such as SIV and Hepatitis C virus (HCV) (3, 4). The findings strongly suggest that enhanced TCR diversity within viral epitope-specific CTL populations prevents the appearance of escape mutants, presumably because of inherent cross-reactivity within CTL recognition (5). Thus, whereas a diverse CTL repertoire seems to be critical for recognition of different, yet closely related epitopes, the effect of TCR diversity on the outcome of cognate epitope recognition is relatively poorly studied, with only one study indicating that a more diverse T cell repertoire may be enriched for higher avidity T cells (6). As such, the functionality of diverse as opposed to narrow epitope-specific CTL populations remains uncertain.

Here, we have used the B6 mouse model of influenza A virus infection to compare CD8⁺ TCRβ repertoires specific for three influenza A epitopes: D^bNP₃₆₆, D^bPA₂₂₄, and a recently described epitope derived from the +1 reading frame of the influenza viral polymerase B subunit (residues 62–70) (D^bPB1-F2₆₂). Multiple features of these epitope-specific responses were analyzed after influenza virus infection, including magnitude, cytotoxicity, cytokine profile, and TCR avidity, and correlated with the degree of CTL repertoire diversity. Interestingly, the vast majority of functional correlates analyzed were able to vary independently of the nature of the repertoire, demonstrating no obvious functional advantage of enhanced TCR repertoire diversity on the response to cognate epitope recognition by CTL.

Results

The Structure of the D^bPB1-F2₆₂ Complex Reveals a High Degree of Antigenicity.

The PB1-F2₆₂ peptide is a recently described influenza epitope (7) and, as yet, is poorly characterized. As an initial characterization, the crystal structure of the D^bPB1-F2₆₂ complex was solved to 2.6 Å resolution and compared with the previously published D^bPA₂₂₄ (8) and D^bNP₃₆₆ structures (9) (Fig. 1 A and B) (supporting information (SI) Table 2). The electron density for the bound PB1-F2₆₂ peptide and the residues within the cleft contacting the peptide was unambiguous. The PB1-F2₆₂ peptide presents a surface that is dominated by the highly solvent exposed, positively charged side chain of the P4-Arg and was observed to display a degree of mobility within the cleft. The importance of this residue for T cell recognition was verified by the failure of PB1-F2₆₂-specific splenocytes to respond to ex vivo stimulation with a mutant peptide in which the Arg at position 4 was mutated to Ala (data not shown). The D^bPA₂₂₄ complex presents a similarly featured surface dominated by the solvent exposed P7-Arg (Fig. 1 A and B). In contrast, the solvent exposed P4-Glu, P6-Met and P7-Glu side-chains of the D^bNP₃₆₆ epitope are orientated toward the MHC helices to present an essentially flat, featureless surface (8). Therefore, the PB1-F2₆₂ peptide can be considered to present a surface with a similar degree of antigenic prominence as PA₂₂₄ although, unlike the PA₂₂₄ peptide, the prominent feature resides in the amino-terminal half of the peptide.

Fig. 1.

D^bPB1-F2₆₂ crystal structural and footprint of TCR recognition. (A) The crystal structure of D^bPB1-F2₆₂ (A and B) was solved to 2.6 Å and compared with the known D^bNP₃₆₆ and D^bPA₂₂₄ structures (8, 9). The pMHC structures of D^bNP₃₆₆, D^bPA₂₂₄, and D^bPB1-F2₆₂ are orientated with the amino terminus on the right, the MHCα1-helix at the back and the MHCα2-helix at the front. The prominent solvent exposed residues of the D^bNP₃₆₆, D^bPA₂₂₄, and D^bPB1-F2₆₂ complexes are highlighted in color (A). The NP₃₆₆ (yellow), PA₂₂₄ (blue), and PB1-F2₆₂ (green) peptides are overlaid as they sit in the groove of H2-D^b and with the amino terminus on the right (B).

The Vβ6⁺ D^bPB1-F2₆₂-Specific CD8⁺ T Cell Repertoire Is Diverse and Private.

Given that the D^bPA₂₂₄ and D^bPB1-F2₆₂-epitope structures both share a prominent arginine residue, albeit at different peptide positions (P7 and P4, respectively) (Fig. 1 A and B), it was of interest to assess the nature of the D^bPB1-F2₆₂ repertoire, in particular whether D^bPB1-F2₆₂ selects the type of diverse TCR repertoire that has been described for the D^bPA₂₂₄-specific response (10). Flow cytometric analysis of Vβ usage in the D^bPB1-F2₆₂-specific CTL population, from splenocytes taken at the acute phase of the primary response to infection, revealed a dominant bias toward Vβ6 that was present on ≈40% of all D^bPB1-F2₆₂-specific T cells (Fig. 2A). Detailed characterization of the D^bPB1-F2₆₂-specific repertoire was thus focused on this Vβ6⁺ population. TCRβ sequencing was performed on individual CD8⁺ Vβ6⁺ D^bPB1-F2₆₂⁺ cells sorted from splenocytes of mice at the acute primary or secondary phase of influenza virus infection.

Fig. 2.

Analysis of TCR Vβ and Jβ usage in the D^bPB1-F2₆₂-specific population after primary virus infection. Immune cells (d10) were stained with antibodies to CD8α and a panel of Vβs. Data are summarized in a pie chart where each slice of the pie represents the mean proportion of CD8⁺ D^bPB1-F2₆₂⁺ cells (n = 4) expressing a particular TCR Vβ ± SD (A). Individual CD8⁺ Vβ6⁺ D^bPB1-F2₆₂⁺ cells were sorted and RT-nested PCR performed with Vβ6-specific oligonucleotide primers, and the CDR3β sequenced as described in Materials and Methods. Shown is the proportion of sequences from mice after both primary (n = 5) and secondary (n = 5) infection expressing given Jβ elements (B).

The vast majority (79%) of D^bPB1-F2₆₂-specific clonotypes (making up 85% of clones) identified across all mice from both time points exhibited a highly conserved glycine residue at position 3 of the CDR3 region (SI Table 3). Furthermore, unlike the D^bNP₃₆₆-and D^bPA₂₂₄-specific repertoires, in which the Jβ2.2 and Jβ1.1/2.6 biases constitute ≈95% and 80%, respectively (11), there did not seem to be a comparable dominance of particular Jβ segments in the D^bPB1-F2₆₂ repertoire (Fig. 2B).

Analysis of clonotype diversity and sharing between individuals in the Vβ6+ D^bPB1-F2₆₂-specific repertoires is summarized and compared with that of the D^bNP₃₆₆- and D^bPA₂₂₄-specific sets in Table 1. Examining total sequences from all individuals as well as within individuals, both the D^bPA₂₂₄- and D^bPB1-F2₆₂-specific repertoires are significantly more diverse than the D^bNP₃₆₆-specific set. The D^bPB1-F2₆₂-specific repertoire is significantly, although not to a great extent, narrower than D^bPA₂₂₄-specific set overall, but diversity is similar within individuals. Thus, the diversity of the D^bPB1-F2₆₂- and D^bPA₂₂₄-specific sets seems similarly broad and is significantly greater than in the limited D^bNP₃₆₆-specific repertoire (Table 1).

Table 1.

Analyses of diversity and sharing in D^bPB1-F2₆₂-, D^bNP₃₆₆-, and D^bPA₂₂₄-specific CDR3β repertoires

Epitope	DbPB1-F262	DbNP366	DbPA224
No. of mice analyzed	10	20	29
No. of TCR clones sequenced	990	1896	2615
Modal CDR3β length	8	9	6
Diversity
Total no. of clonotypes	182	71	394
No. of clonotypes per individual (mean ± SD)	22.4 ± 7.1	7.95 ± 0.8	22.3 ± 1.36
1 − D total	0.978*,**	0.837**,#	0.984*,#
1 − D individual average	0.859*	0.548**,#	0.897*
Sharing
Proportion of clones ″repeated″ in ≥40% of sampled mice
from total	0.17*	0.675**,#	0.14*
from individuals	0.17*	0.68**,#	0.13*
No. of public sequences (in ≥75% of sampled mice)	0	3	0
Chao–Jaccard sharing estimator (avg of individual comparisons)	0.058*	0.410r**,#	0.078*

Clone refers to a single sequenced sample (therefore abundance is taken into account); clonotype refers to a unique amino acid sequence. *, P < 0.05, compared with D^bNP₃₆₆; **, P < 0.05, compared with D^bPA₂₂₄; #, P < 0.05, compared with D^bPB1-F2₆₂. Statistical analyses were performed using variations of the Kruskal–Wallis test (see SI Text).

Analysis of clonotype sharing between individuals was performed to determine the presence of selective pressures on the repertoires exerted either at the level of the individual or on the overall population. Interestingly, whereas the proportion of shared clonotypes in the D^bPB1-F2₆₂ repertoire was intermediate between D^bNP₃₆₆ and D^bPA₂₂₄ (2.2%, compared with 5.6% and 1.0%, respectively), the proportion of the total response (clones) that was shared for D^bPB1-F2₆₂ was similar to D^bPA₂₂₄ and significantly less than D^bNP₃₆₆ (Table 1). Thus, presumably because of enhanced diversity, D^bPB1-F2₆₂ and D^bPA₂₂₄ repertoires exhibit significantly reduced clonotype sharing, and those clonotypes that are shared are not necessarily heavily represented in all mice (as they are for D^bNP₃₆₆).

Repertoire Diversity Does Not Confer a Numerical Advantage on Epitope-Specific CTL Responses.

Given that analysis of the D^bPB1-F2₆₂-specific CD8+ T cell response has, to date, been limited to the intracellular cytokine staining (ICS) assay for peptide-specific IFN-γ production (7, 12), we directly assessed the magnitude of this response by specific tetramer binding. Tetramer staining of splenic D^bPB1-F2₆₂-specific cells at acute and memory phases of the primary response revealed a subdominant response magnitude that was substantially reduced compared with D^bNP₃₆₆- and D^bPA₂₂₄-specific responses (Fig. 3 A and B). In the acute population recovered from the pneumonic lung by bronchoalveolar lavage (BAL), whereas the magnitude of the D^bPB1-F2₆₂-specific population was smallest, it was not significantly different from the other responses analyzed (Fig. 3D). After secondary challenge, the splenic D^bPB1-F2₆₂-specific population was similar in size to the D^bPA₂₂₄-specific response, in both spleen and BAL, where the D^bNP₃₆₆-specific response dominates (13, 14) (Fig. 3 C and E). Thus, after both primary and secondary influenza infection, and in contrast to what might be expected from a TCRβ repertoire comprising multiple clonotypes, the D^bPB1-F2₆₂-specific response is subdominant, whereas the heavily restricted D^bNP₃₆₆-specific repertoire is significantly larger.

Fig. 3.

Analysis of epitope-specific CD8⁺ T cell magnitudes and cytokine profiles after primary and secondary influenza virus infection. Naive (primary) and PR8-immune (secondary) B6 mice were infected, and analyzed at various phases of immune response. Enriched splenocytes and BAL cells were stained with specific tetramers, followed by anti-CD8α-FITC. Shown are the total numbers of CD8⁺ tetramer⁺ cells (A–E). Cells were stained for CD8α and intracellular IFN-γ, TNF-α, and IL-2 (F–M). Shown is the mean proportion of IFN-γ⁺ cells also producing either TNF-α (F–I), or IL-2 (J–M) ± SD for five mice. *, P < 0.05 using a two-tailed Student's t test, compared with the D^bPB1-F2₆₂ response.

Correlation Between TCRβ Repertoire Diversity and TNF-α, but Not IL-2, Production.

Cytokines produced by virus-specific CTL are important in the establishment and maintenance of inflammation and for their direct anti-viral properties. We were interested in whether TCR repertoire diversity impacted on the ability of populations to produce multiple cytokines. Parallel T cell populations from the sets used for the tetramer staining analysis (Fig. 3 A–E) were restimulated in vitro with peptide for analysis by the multiparameter (IFN-γ, TNF-α, IL-2) ICS assay. The frequency of IFN-γ+ cells were found to be generally comparable to that determined by tetramer staining (data not shown). Analysis of TNF-α production by D^bPB1-F2₆₂-specific cells revealed, in both spleen and BAL at all time points, a profile similar to that seen in the D^bPA₂₂₄-specific population, with a significantly larger percentage of TNF-α producers relative to the D^bNP₃₆₆-specific set (Fig. 3 F–I). Conversely, the proportion of D^bPB1-F2₆₂-specific cells producing IL-2 was, broadly speaking, lower than in the D^bPA₂₂₄-specific set, and more characteristic of the D^bNP₃₆₆-specific population (Fig. 3 J–M).

Repertoire Diversity Does Not Correlate with Cytotoxicity or TCR Avidity.

The ability of CTLs to kill virally infected cells is the hallmark of an efficient CD8+ T cell response. The cytotoxicity of D^bPB1-F2₆₂-specific cells for peptide pulsed EL4 targets was compared with D^bNP₃₆₆-and D^bPA₂₂₄-specific cells directly ex vivo 10 days after primary intranasal (i.n.) infection. Importantly, equivalent numbers of D^bPB1-F2₆₂, D^bNP₃₆₆-, and D^bPA₂₂₄-specific effectors were used in the assay after enumeration by tetramer staining combined with splenocyte counts. Interestingly, D^bPB1-F2₆₂-specific cells were significantly less cytotoxic than either D^bNP₃₆₆- or D^bPA₂₂₄-specific cells (Fig. 4A), which showed similar cytotoxic potential at an individual cell level. Thus, whereas the cytotoxicity of D^bPB1-F2₆₂-specific cells correlated with their position in the immunodominance hierarchy, there was no apparent effect of repertoire diversity on cytotoxic potential.

Fig. 4.

Cytotoxicity and avidity of viral epitope-specific CD8⁺ T cell populations. The CD8⁺D^bNP₃₆₆⁺, CD8⁺D^bPA₂₂₄⁺, and CD8⁺D^bPB1-F2₆₂⁺ cell numbers were equalized (based on tetramer staining) and titrated before incubation with ⁵¹Cr-labeled NP₃₆₆, PA₂₂₄, or PB1-F2₆₂ peptide-pulsed EL4 target cells for 4 h in round-bottom 96-well plates. The extent of lysis was then determined as specific ⁵¹Cr release, and the results are expressed as mean ± SD (A). The kinetics of tetramer dissociation for splenic CD8⁺ T cells were analyzed directly ex vivo on day 10 after primary i.n. infection (n = 5). Enriched CD8⁺ T cells were stained with specific tetramer for 1 h at room temperature. Cells were washed and incubated for designated times at 37°C in the presence of mAb to H-2D^b, before costaining with anti-CD8α-FITC. Shown are CD8⁺ tetramer⁺ cells expressed as a percentage of the maximum binding observed at time 0 (B). *, P < 0.05, comparing D^bPB1-F2₆₂ with D^bNP₃₆₆; #, P < 0.05 comparing D^bPB1-F2₆₂ with D^bPA₂₂₄.

One might anticipate that, generally, a broad TCR repertoire is better able to bind specific pMHC complexes with higher avidity than a restricted repertoire (6). We were particularly interested in determining whether the similarly diverse repertoire characteristics of the D^bPB1-F2₆₂ and D^bPA₂₂₄-specific populations translated into similarities in TCR avidity. Dissociation of the D^bPB1-F2₆₂ tetramer, on populations isolated after primary virus infection, occurred at a rate that was significantly faster than the D^bPA₂₂₄-specific set, and similar to that observed for the D^bNP₃₆₆-specific population (Fig. 4B). Virtually identical results were obtained after secondary viral infection (data not shown). These results suggest that the D^bPB1-F2₆₂-specific population is relatively low avidity, indicating that broad repertoire diversity is not sufficient to confer increased avidity on the responding population.

Discussion

TCR diversity within T cell repertoires is generally considered to be advantageous, facilitating responsiveness to multiple antigenically distinct pathogens (15). At the level of epitope-specific CD8+ T cell responses however, investigation into TCR repertoire diversity has largely focused on the value of diversity in mediating recognition of escape mutants (3, 4). Whereas this work is clearly of great importance, it provides no information on how TCR diversity influences the quality of the cognate response to pMHC. The aim of this study, therefore, was to ascertain whether the diversity of an epitope-specific T cell repertoire is linked to its functionality in the context of cognate pMHC recognition. To this end, we have characterized TCRβ diversity and functionality in the D^bPB1-F2₆₂-specific CD8⁺ T cell population, and compared features of this response to those of the previously characterized D^bNP₃₆₆ and D^bPA₂₂₄-specific populations. We demonstrate that the D^bPB1-F2₆₂ complex exhibits a relatively featured topology and thus induces a dominant TCR repertoire characterized by extensive diversity and relatively little clonotype sharing (8, 10). Whereas similar to D^bPA₂₂₄ in this regard, the D^bPB1-F2₆₂-specific population exhibited distinct functional properties, including reductions in response magnitude, cytotoxicity, IL-2 production, and TCR avidity. Thus, whereas the relationship between pMHC structure and TCR repertoire diversity was confirmed in this study, the functionality of such repertoires seems variable and unrelated to the degree of TCR diversity.

Previous reports have demonstrated that TCR repertoire diversity is influenced by the structural complexity of the pMHC structure being recognized (8, 16, 17). This finding has been comprehensively demonstrated in the influenza system where the antigenic structures of two D^b-restricted peptides, D^bNP₃₆₆-and D^bPA₂₂₄, were found to differ markedly, corresponding to substantial differences in the relative diversities of the responding T cell repertoires (8, 10, 16, 18). Consistent with these reports, the D^bPB1-F2₆₂ structure revealed a relatively prominent P4 arginine, and this antigenic profile was reflected in a D^bPB1-F2₆₂-specific TCR repertoire with a similarly broad TCR repertoire diversity to that observed for D^bPA₂₂₄-specific CTL responses. This observation was particularly interesting as the prominent residue in the D^bPB1-F2₆₂ complex was situated further toward the amino terminus than that of D^bPA₂₂₄. Thus, it seems that the degree, rather than the location, of antigenic prominence is critical in determining the diversity of the TCR repertoire. It should be noted, however, despite similarities in diversity, that the D^bPB1-F2₆₂-specific TCR repertoire was qualitatively distinct from the D^bPA₂₂₄ repertoire, exhibiting a unique Vβ bias, a conserved glycine motif at position 3 of the CDR3, and no obvious Jβ preference. Thus, the effects of antigenic prominence in this system seem limited to repertoire diversity.

Diverse TCR repertoires have been shown to be advantageous in the cross-reactive CTL response to heterologous infections (5), as well as in preventing the emergence of viral escape mutants (3, 4). It seems intuitive that a more diverse T cell repertoire might be advantageous for recognition of different, yet closely related, epitopes. But does a diverse repertoire confer any functional advantage in the response to cognate antigen?

The clonotypic diversity of an epitope-specific T cell repertoire may be expected to reflect the number of available T cell precursors in the naive repertoire (19). Despite this observation, the D^bPB1-F2₆₂-specific response, while comprising a highly diverse repertoire compared with D^bNP₃₆₆, is subdominant after primary infection. Furthermore, the extremely restricted D^bNP₃₆₆-specific population is codominant in the primary response. Thus, T cell diversity does not seem to be a good predictor of response magnitude. It remains a possibility, however, that repertoire diversity does accurately reflect precursor frequency, but that after virus infection, epitope abundance may compensate for, or overwhelm, the effects of precursor frequency on response magnitude (14). Indeed, unlike NP and PB1, PB1-F2 is not incorporated in the forming virions and it is not required for virus production (7, 20). Furthermore, the PB1-F2 protein has a relatively short half-life (7). It is possible that these factors may combine to result in the later and more limited cell-surface expression of D^bPB1-F2₆₂, which could in turn drive less cycling in the responding T cells. Indeed, early viral clearance after secondary infection results in the failure of PB1-F2-specific cells to maximally expand (21). This possibility is conjecture however, and the extent and kinetics of D^bPB1-F2₆₂ presentation are yet to be tested.

Ex vivo cytotoxicity, a definitive measure of CTL function, also showed no correlation with repertoire diversity for the epitope-specific responses analyzed. Moreover, the cytotoxic potential of the three populations was not reflected in their respective TCR dissociation rates, suggesting that factors other than T cell diversity and TCR avidity are influencing cytotoxicity in this system. The subdominant nature of the D^bPB1-F2₆₂-specific response and its reduced cytotoxicity may both be a consequence of reduced expansion of D^bPB1-F2₆₂-specific cells. We know from other experiments that the acquisition of cytotoxic potential, measured by perforin and granzyme mRNA profiles, is directly related to cell division (M. R. Jenkins, P.C.D., and S.J.T., unpublished results).

Assessing functionality by means of cytokine production showed that D^bPB1-F2₆₂ had a similarly high proportion of TNF-α producers as the D^bPA₂₂₄-specific set, yet a low proportion of IL-2 producers, more like the D^bNP₃₆₆-specific population. This previously uncharacterized finding of an influenza epitope-specific response showing disparate relative abundance of TNF-α- and IL-2-producing cells suggests a possible link between TNF-α production and repertoire diversity. The lack of any other functional correlate of repertoire diversity, however, suggests that this link may be coincidental. The production of IL-2, however, seems to remain a correlate of avidity for all five of the epitope-specific responses analyzed to date (22, 23). The mechanism by which highly avid epitopes may specifically stimulate IL-2 production versus other cytokines remains unknown and is a promising topic for future investigation. There is evidence that the ability to produce multiple cytokines, and IL-2 in particular, correlates with the persistence of epitope-specific cells into the memory phase (23, 24), as well as with protection against Leishmania major infection in mice (25). Similarly, high avidity T cell populations have also been shown to be enriched in some memory populations (26, 27) and to exhibit enhanced functionality (28–30).

This work leads us to consider the feature of the CTL response thought most likely to correlate with TCR repertoire diversity, TCR avidity. It has been suggested that diverse, rather than restricted, populations are more efficient at clearance of herpes simplex virus (6). It was proposed that such efficacy was a consequence of diverse repertoires being more likely than narrow repertoires to contain high avidity CTL. Interestingly, our studies revealed no consistent correlation between TCR avidity and diversity, with the diverse D^bPB1-F2₆₂-specific population exhibiting a similarly low avidity as the narrow D^bNP₃₆₆ set. Thus, if high avidity T cell populations are indeed functionally desirable, as has been indicated in studies of tumor and virus clearance (28–30), then our results suggest that increasing TCR diversity is not necessarily an efficient mode of generating such a population.

In conclusion, the global analysis performed here has strengthened the link between pMHC structure and TCRβ repertoire diversity. However, whereas other studies have demonstrated the benefits of a broad TCR repertoire for preventing immune escape or for promoting cross-reactivity during heterologous infection, there is no evidence from this study that such repertoires are functionally advantageous in response to the cognate pMHC in the context of viral infection. Therefore, vaccine strategies to enhance functional responsiveness to virus peptides should avoid simply focusing on diversification of the TCR repertoire. Furthermore, with few exceptions, most features of the CD8⁺ T cell response to viral infection studied here are able to vary independently of each other, and as such may be primarily determined by the environment encountered during T cell stimulation.

Materials and Methods

Mice and Tissue Harvesting.

The female B6 (H2^b) mice used in this study were bred and housed in the animal facility of the Department of Microbiology and Immunology at the University of Melbourne. All experimental procedures were reviewed and approved by the University of Melbourne Animal Experimentation Ethics Committee. Naive 6- to 8-week-old mice were either infected i.n. with 1 × 10⁴ pfu of the HKx31 (x31) influenza A virus for analysis of the primary response, or primed i.p. with 1.5 × 10⁷ pfu of the A/PR8/34 (PR8) influenza A virus at least 6 weeks before secondary i.n. HKx31 challenge. The PR8 and HKx31 viruses express different surface hemagglutinin (H) and neuraminidase (N) glycoproteins (H1N1 and H3N2, respectively), but share the PR8 internal components. Single cell preparations of spleen were enriched for CD8⁺ cells by panning for 1h at 37°C on plates coated with a mixture of anti-mouse IgG/IgM (The Jackson Laboratory). Lymphocytes were obtained from the lung by bronchoalveolar lavage (BAL) and adherent cells were removed by incubating on plastic for 1 h at 37°C.

Tetramer and Antibody Staining.

Epitope-specific CD8⁺ T cells were identified by using tetrameric complexes of the influenza virus H-2D^b MHC class I glycoprotein and the NP_366–374 (ASNENMETM), PA_224–233 (SSLENFRAYV), or PB1-F2_62–70 (LSLRNPILV) peptides. Enriched cells were incubated with D^bNP_366--PE, D^bPA_224--PE, or D^bPB1-F2₆₂-PE tetramers for 1 h at room temperature, washed, and stained with anti-CD8α-FITC. Alternatively, tetramer stained cells were stained with anti-CD8α-APC and one of a panel of FITC-conjugated TCR Vβ-specific antibodies (all antibodies purchased from PharMingen unless otherwise stated). Cells were washed, and the staining profiles were analyzed by using a BD FACSCalibur.

Tetramer Dissociation Kinetics.

An established protocol was used to measure tetramer dissociation kinetics (23, 31). Briefly, after tetramer staining, cells were incubated at 37°C with the 28-14-8 mAb to H2D^b (50 μg/ml) for various times to prevent tetramer rebinding. Cells were then stained with anti-CD8α-FITC, and residual tetramer staining was analyzed on a BD FACSCalibur using CellQuest software.

Stimulation and Intracellular Cytokine Staining.

Stimulation and intracellular cytokine staining of lymphocyte populations has been described (23). Briefly, enriched lymphocytes from spleen and BAL were stimulated in vitro for 5 h with 10 units/ml IL-2 in the presence or absence of 1 μM peptide. They were then stained for cell surface expression of CD8α, and intracellular expression of IFN-γ, TNF-α, and IL-2, and analyzed by flow cytometry.

Ex Vivo ⁵¹Cr Release Cytotoxicity Assay.

Analysis of ex vivo cytotoxicity of splenocytes harvested from mice 10 days after primary i.n. infection was performed as described in ref. 32. Briefly, ⁵¹Cr-labeled target cells pulsed with 1 μM NP₃₆₆, PA₂₂₄, or PB1-F2₆₂ peptides (or left unpulsed) were incubated with effectors at 37°C, 5% CO₂ for 4 h. Supernatants (50 μl) were harvested, and the percentage of specific ⁵¹Cr release was calculated.

Analysis of D^bPB1-F2₆₂-Specific Vβ T Cell Repertoires.

Individual CD8⁺Vβ6⁺D^bPB1-F2₆₂⁺ cells were sorted from splenocytes harvested at the acute primary (day 10) or secondary (day 8) time point, and total RNA was extracted by using TRIzol according to the manufacturer's instructions. Reverse transcription was performed as described in ref. 10, and a nested PCR strategy (8, 10, 18) was used to amplify Vβ6 cDNA using the following oligonucleotide primers: first round-Vβ6 ext, 5′-CAGACACCCAAATTCCTGATTGGTC-3′ and Cβa, 5′-CCAGAAGGTAGCAGAGACCC-3′; second round-Vβ6 int, 5′-GCTATGATGCGTCTCGAGAGAAGAAGTC-3′ and Cβb, 5′-CTTGGGTGGAGTCACATTTCTC-3′. Second round Vβ PCR products were then purified by using the QIAquick PCR Purification Kit (Qiagen), sequenced by using 3.2 pmol of the Vβ6 int primer, and analyzed on an ABI Prism 3700 sequence analyzer.

Repertoire Analysis Statistics.

As detailed in SI Text, a variety of statistical approaches have been used to describe species diversity and the amount of shared sequences between different mice (33–36).

Protein Purification, Crystallization, and Structure Determination.

The recombinant D^bPB1-F2₆₂ complex was refolded and purified as described in ref. 37. Crystals of the D^bPB1-F2₆₂ epitope, which belong to space group P4₃2₁2, were obtained essentially as described in ref. 8. A 2.6-Å dataset was collected and processed and scaled by using the HKL package (HKL Research, Charlottesville, NC) (38). A summary of the data collection is found in SI Table 2. The structure of D^bPB1-F2₆₂ was solved by using the coordinates of D^bPA₂₂₄ (minus the peptide and water molecules) as a starting model for rigid body refinement as implemented in CNS. The progress of refinement was monitored by the R_free value (7% of the data) with neither a sigma, nor a low-resolution cut-off being applied to the data. The structure was refined initially by using the simulated annealing protocol implemented in CNS (version 1.0) (39), followed by TLS-refinement in REFMAC. Tightly restrained individual B-factor refinement was used, and bulk solvent corrections were applied to the dataset (SI Table 2).