Influenza virus is an immunologically complicated pathogen, particularly in humans, where a long life span allows multiple encounters with genetically variable influenza viruses and vaccines. The first encounter for most humans is through natural infection in early childhood, and therefore, the original memory compartment for influenza virus is based on priming through live infection. After this original contact, individuals encounter influenza virus and its proteins again periodically, perhaps every 2 to 3 years, through natural encounters with different subtypes or strains of influenza virus or through active vaccination (14
). In order to understand how the influenza virus-specific CD4 T-cell memory compartment is shaped over time, it is necessary to first understand the diversity and specificity of the primary response. In this study, we have found that the primary CD4 T-cell response to live influenza virus infection is highly diverse and represented by T cells for each of the viral proteins tested: HA, NA, NP, and NS1. Surprisingly, we found that there was no selective enrichment for CD4 T cells specific for the membrane-associated antigens HA and NA and that of all the virus proteins examined, NS1 appeared to be the most immunogenic, recruiting more than 2,400 CD4 T cells per million CD4 splenic T cells. CD4 T cells specific for NP and HA were also abundant, with less representation of CD4 T cells specific for the membrane-associated NA protein.
Our results for CD4 T-cell responses suggest a much broader epitope distribution than was predicted from a previous study. The primary response of C57BL/6 (H-2b
) mice to a mouse-adapted strain of influenza virus was found to be quite restricted, specific primarily for peptides from HA and NP, with two peptides from these proteins accounting for more than 30% of the total influenza virus-specific CD4 T cells from the lung (17
). We performed a study similar to what we describe here using H-2b
mice with strain A/New Caledonia/20/99 and found results similar to those described previously by Crowe et al. (17
), with a highly restricted repertoire focused on NP (J. Nayak, K. Richards, and A. Sant, unpublished data). In contrast, during the primary response to A/New Caledonia in BALB/c mice, which express I-Ad
, we find a highly diverse repertoire similar to what we have found with HLA-DR1 transgenic mice (A. Sant, unpublished data). We suspect that the pattern seen in the BALB/c and DR1 transgenic mice reflects the overall broad diversity of peptides that can be presented by these host class II molecules (I-Ad
, and HLA-DR1) compared to that of I-Ab
. We hypothesize that under conditions of relatively promiscuous capture of virus-derived peptides by class II molecules, the major force in driving CD4 T-cell specificity will be characteristics of the viral protein itself rather than stringent peptide selection by MHC class II molecules. Because humans express as many as 12 class II proteins, depending on the MHC haplotype expressed, heterozygosity at MHC loci, and options for cross-allelic mixed class II dimers, we expect that the results for the HLA-DR1 transgenic mice described here are more predictive of what will generally be found in the influenza virus-specific CD4 T-cell repertoire in humans. In fact, in a recent study using MHC-derived tetramers to detect influenza virus-specific T cells in human subjects (64
), many of the HLA-DR1-restricted epitopes detected from peripheral blood of healthy human donors, including those from HA, NA, and NP, were identified in our studies. We have also confirmed a subset of these specificities in human CD4 T cells using ELISPOT assays (our unpublished results). The specificities detected for humans are indicated in Table S1 in the supplemental material. Although the diversity of CD4 T cells detected in the current study is surprising, several aspects of our experimental design as well as other ongoing work in our laboratory make us confident that the epitopes which we have defined all belong to the CD4 lineage, are HLA-DR1 restricted, and are not due to nonspecific activation by high concentrations of peptides used in the ELISPOT assays. With regard to the issue of CD4 T-cell specificity, the T-cell population used for all of our assays is rigorously (>99%) depleted of CD8 T cells, and all the peptides identified require HLA-DR1 expression to activate the CD4 T cells. In addition, we have produced CD4 T-cell hybridomas specific for a number of the epitopes identified here, and these T cells all display peptide specificity and HLA-DR1 restriction. Finally, we have no reason to suspect that the cytokine-producing cells are activated nonspecifically by peptides used at high concentrations. Lower concentrations of peptide (2 or 0.2 μM) are potent in activating many of the T cells identified here, and perhaps even more compellingly, in studying the response patterns of different MHC congenic strains of mice, using the same virus and same stock of purified peptides as those used here, we have discovered completely nonoverlapping patterns of epitope specificity of CD4 T cells.
In considering the CD4 T-cell specificities reported here, it is intriguing that NS1 and NP appear to be highly immunogenic during the primary response to influenza virus despite their localization in the cytosol and nucleus, sites that are thought to have relatively inefficient access to MHC class II molecules. This finding requires the consideration of alternative factors that may control epitope dominance in the response to influenza virus. The absolute abundance of the viral protein within APC may play a critical role in determining its access to MHC class II molecules, as might the kinetics of viral protein expression. In most cells studied, influenza virus infection essentially shuts down host cell gene expression within a few hours of infection (41
). Because the MHC class II-restricted presentation of antigen utilizes primarily newly synthesized class II molecules (reviewed in references 11
, and 68
), it is possible that the earliest-synthesized and most-abundant proteins will most efficiently access MHC class II molecules in endosomal compartments. Although the kinetics of influenza virus protein synthesis in cultured cell lines suggest that NS1 is among the earliest proteins synthesized (66
), little is known about the kinetics of individual influenza virus proteins in dendritic cells, the relevant cell type in CD4 T-cell priming known to have distinctive characteristics of influenza virus infection compared to those of other cell types (5
). With regard to the restrictions imposed by subcellular localization, it is possible that the cytosolic and nuclear antigens NP and NS1 gain access to endosomal compartments of APC via the process of autophagy, an intracellular process that allows the engulfment and ultimate delivery of cytosolic and nuclear proteins to lysosomal compartments by the fusion of membrane vesicles derived from the endoplasmic reticulum (16
). Autophagy rates increase upon virus infection (47
), and it is possible that this alternative mechanism allows the needed access of viral NS1, NP, and other cytosolic/nuclear proteins to the MHC class II loading compartments.
Independently of the mechanisms that control antigen presentation of influenza virus peptides by MHC class II molecules and, thus, the repertoire of elicited CD4 T cells, it is important to consider the implications of our findings for host defense. The most important function of CD4 T cells for protection from influenza virus is thought to be the provision of “help” for the production of high-affinity neutralizing antibodies to HA and NA (reviewed in reference 38
). Recent data suggest that for some viruses, there may be an obligate link between the specificity of CD4 T cells and the antigen-specific B cells (74
). If this model is correct, then the most useful CD4 T cells may be those that are specific for HA and NA. CD4 T cells that are specific for intracellular NP, polymerase, and NS1 proteins may be of more-limited value in facilitating antibody responses. If true, then the implications for heterosubtypic immunity are profound. Because of the high degree of genetic drift within influenza viruses (6
), intermittent encounters with influenza virus strains in humans may preferentially boost T cells that are specific for conserved peptides enriched in nucleoprotein and polymerase proteins (2
). However, if these CD4 T cells have limited contributions in facilitating antibody responses, they may not be particularly valuable for protection. Instead, heterosubtypic immunity will depend on the cross-reactivity of CD4 T cells specific for the HA and NA epitopes that are shared among seasonal viruses and newly emerging viral strains. Because of the importance of this issue, we evaluated a subset of these HA- and NA-derived epitopes, localized primarily to genetically conserved regions of these proteins, for cross-reactivity with homologous sequences from a human isolate of an H5N1 virus. Our study revealed a substantial degree of cross-reactivity between H1N1 and H5N1 sequences. It is likely that for those peptides that have amino acid differences within a peptide epitope, the amino acid substitutions may be located at MHC anchor sites rather than T-cell contact sites. Anchor substitutions typically have very little negative impact on T-cell reactivity (44
). Our results are encouraging and suggest that although substantial numbers of CD4 T cells that are specific for internal viral proteins are elicited in the primary response to live influenza virus infection, the CD4 T cells primed during seasonal encounters with influenza virus or through vaccination with seasonal vaccines may prime CD4 T cells that are reactive with HA and NA which have the potential to be reelicited and expanded upon challenge with a heterosubtypic strain of influenza virus. CD4 T cells of this specificity may be effective in promoting more-rapid and more-robust antibody responses to heterosubtypic challenge than would occur in a nonvaccinated individual.