We have observed at least four types of CD8+ T cell recognition in the context of these influenza infections. Type 1 was CD8+ T cell responses to epitopes that are unique to each virus infection. Type 2 was CD8+ T cells that recognize both WT and mutated peptides. Type 3 was T cells that recognize related, but mutated peptides from each virus, but in a mutually exclusive manner. Lastly, in influenza rechallenged mice, we observed CD8+ T cells with dual specificity not evident in the primary infections. These observations reveal tremendous plasticity in the ability to develop cross-reactive CD8+ T cells after influenza infection.
In PR8 and X31 infection B6 mouse model 
restricted CD8 epitopes NP366 and PA224 are dominant during primary infection, while NP366 is predominant during secondary infection 
. Although these studies provided convincing data for the dominance of certain influenza CD8 epitopes, characterization and validation of CD8 epitope hierarchy in other influenza virus infections is less well studied, possibly due to the fact that most natural influenza isolates could not easily infect animals. Recently, we have derived a 2009 H1N1 pandemic influenza viral stock, CA/E3/09 that can easily infect mice and cause severe lung disease without further adaption in mice 
. Taking advantage of this new viral isolate, in this study, CD8 epitope profiles for NP and PA proteins of CA/E3/09 virus were analyzed with the aid of bioinformatic prediction and experimental approaches. However, it was not our intent in this study to conduct a full survey for all possible epitopes.
Our data indicate that, although it has valine substituted at position 6 for the methionine, the predicted CANP366 epitope is a dominant CD8 epitope processed and presented during CA/E3/09 infection in B6 mice. It has been shown that M6 residue on NP366 epitope of PR8 or X31 virus is critical for TCR recognition, as single M6A mutations results in complete loss of peptide-induced cytokine production and cytotoxicity 
. Interestingly, the virus carrying M371A mutation maintained NP366 immunodominance during infection 
. Similar to the laboratory M371A mutation, the M371V substitution on CA/E3/09 NP protein also dramatically changed the antigen specificity of the native epitope, yet the CANP366 epitope is still dominant among other epitopes identified. Thus, our data and the previous reports suggested that amino acid of the position 6 on the NP366 CD8 peptide is crucial for peptide specificity, but not essential for the maintenance of the peptide dominance 
Although it was found that PA protein contains peptides that can be recognized by H2-Db
molecule and several peptides were predicted as potential candidates 
, the bona fide CD8 epitopes from this protein were not identified until a report that defined the prominence of PA224 CD8 epitope during primary X31 and PR8 infection 
. In the present study, we found that the predicted PA224 epitope on CA/E3/09 (CAPA224) has one amino acid on position 1 that is different from PA224 of X31 virus (X31PA224). However, this substitution did not affect the ability of the CAPA224 peptide to induce cytokine responses after X31 virus infection as demonstrated by ELISPOT assay. The results support previous conclusion that amino acids of position 5, 6, 7 and 9, but not others on X31PA224 peptide are critical for TCR recognition and activation 
In contrast to the robust anti-PA224 CD8 response during X31 infection, we did not observe any response to either CAPA224 or X31PA224 peptide after CA/E3/09 infection. Further, X31PA224 tetramer staining failed to detect any CD8 cells in various tissue samples from CA/E3/09 infected animals. Thus, S224P substitution does not affect the antigenicity of peptide PA224, but abolishes its immunogenicity during CA/E3/09 infection. The exact reason for the loss of CAPA224 dominance is not known at this moment, but it is possible that CAP224 peptides are not processed and recognized by H2-Db
after CA/E3/09 infection. The fact that the predicted H2-Db
binding for CAPA224 is drastically lower than the X31PA224 supports the possibility that CAP224 is not able to form a stable MHC complex. We have not yet tested this hypothesis. Even though we do not fully understand the molecular basis of the effect of the S to P switch on the PA224 peptide recognition and presentation, further analysis of the first amino acid residue on the MHC complex stability or the established crystal structure of H2-Db
may provide useful hints.
In addition to the CANP366 and CAPA224 epitopes, we examined other predicted epitopes, among which CANP39 elicited a consistent response. Since the frequency of IFN-γ producing cells elicited by peptide CANP39 is slightly over half of that against CANP366, though still at least two-fold higher than the other peptides, we considered CANP39 as subdominant epitope and other peptides including CANP41, CANP55, CANP130, CANP296 and CAPA439 as minor CD8 epitopes for CA/E3/09 virus. These subdominant and minor CD8 epitopes have not been described previously, in spite of the conservation of some of these sequences in X31/PR8 viruses. Thus, our data provided evidence that CA/E3/09 infection results in a diverse usage of the CD8 TCR repertoire and novel hierarchy of CD8 epitopes.
Most importantly, our analysis of X31NP366 and CANP366 specific CD8 T cells using MHC tetramers and ELISPOT showed that a population that is truly cross-reactive to X31NP366 and CANP366 epitopes appeared after CA/E3/09 secondary infection in X31 primed animals, but is not seen in primary infections. Our data suggested that prior priming promotes the generation of cross-reactive NP366 dominant CD8 T cells that can be recognized by both NP366 peptides from X31 and CA/E3/09 viruses. The selective expansion of cross-reactive NP366 cells has also been demonstrated previously using priming and challenge model with X31 or PR8 and NT60, an influenza virus that differs in sequence of the NP366 epitope at position 7 and 8 
. Thus, our results confirmed the previous finding that the rare cross-reactive CD8 T cells generated during primary influenza infection could be robustly expanded and protective for the heterologous variant infection. It is not known whether this type of cross-reactivity is unique to the NP366 epitope, but it seems more likely that other examples of this form of reactivity exist. This phenomenon is particularly important in the context of human influenza immunity, as humans may be exposed and primed to many related and unrelated influenza viruses over a lifetime. Thus, it raises questions about whether CD8+ T cell specificities that may be rare in single infections become more prevalent upon repeated infections in human. Previous published study showed cross-reactive NP418 human CD8 T cells could be generated in-vitro and detected with low frequency in stimulated human PBMCs 
, suggesting similar immunological selection operates in human during repeated infection with influenza variants to provide protection. The number of such examples of immune reactivity is not known, but deserves further investigation.
In summary, we have identified and characterized cross-reactive CD8+ T cells that respond against the 2009 H1N1 pandemic influenza virus during primary infection and secondary challenge. The results provided mechanistic explanations for the reported heterologous protection against CA/E3/09 that is observed in human subjects and stimulated in animal models offered by virus priming. Our data is relevant to the design of vaccination strategies against existing and emerging pandemic influenza viruses.