Conservation and Diversity of Influenza A H1N1 HLA-Restricted T Cell Epitope Candidates for Epitope-Based Vaccines

Background The immune-related evolution of influenza viruses is exceedingly complex and current vaccines against influenza must be reformulated for each influenza season because of the high degree of antigenic drift among circulating influenza strains. Delay in vaccine production is a serious problem in responding to a pandemic situation, such as that of the current H1N1 strain. Immune escape is generally attributed to reduced antibody recognition of the viral hemagglutinin and neuraminidase proteins whose rate of mutation is much greater than that of the internal non-structural proteins. As a possible alternative, vaccines directed at T cell epitope domains of internal influenza proteins, that are less susceptible to antigenic variation, have been investigated. Methodology/Principal Findings HLA transgenic mouse strains expressing HLA class I A*0201, A*2402, and B*0702, and class II DRB1*1501, DRB1*0301 and DRB1*0401 were immunized with 196 influenza H1N1 peptides that contained residues of highly conserved proteome sequences of the human H1N1, H3N2, H1N2, H5N1, and avian influenza A strains. Fifty-four (54) peptides that elicited 63 HLA-restricted peptide-specific T cell epitope responses were identified by IFN-γ ELISpot assay. The 54 peptides were compared to the 2007–2009 human H1N1 sequences for selection of sequences in the design of a new candidate H1N1 vaccine, specifically targeted to highly-conserved HLA-restricted T cell epitopes. Conclusions/Significance Seventeen (17) T cell epitopes in PB1, PB2, and M1 were selected as vaccine targets based on sequence conservation over the past 30 years, high functional avidity, non-identity to human peptides, clustered localization, and promiscuity to multiple HLA alleles. These candidate vaccine antigen sequences may be applicable to any avian or human influenza A virus.


Introduction
Influenza A viruses are major pathogens of avian origin with global spread and rapid mutational change, some of which also infect humans and other mammals. Of particular concern are the several ways a human influenza pandemic could emerge. One is through the occurrence of a novel and highly pathogenic zoonotic strain capable of infecting humans, such as the H5N1 avian pathogen. Another possibility is through mutation from a mild to a more pathogenic human transmissible strain, such as mutation to the current H1N1 strain. The most threatening is mutations giving rise to a new highly transmissible-and-pathogenic human strain, as occurred with the original 1918 Spanish influenza. In any event, history teaches us that a vaccine to prevent a new influenza A pandemic must be effective against all future forms of the virus.
Influenza A viruses are single stranded, negative-sense RNA viruses belonging to the family Orthomyxoviridae. The genome is composed of 8 RNA strands of about 13,500 bases, encoding at least ten viral proteins. The viral envelope is a lipid bilayer, consisting of the interior matrix protein 1 (M1) and three exterior transmembrane proteins: hemagglutinin (HA), neuraminidase (NA), and matrix protein 2 (M2). The viral core contains viral ribonucleoprotein complex particles, consisting of viral RNA, nucleoprotein (NP), and three polymerase proteins (PB1, PB2, and PA). Mutation in the viral RNA genome occurs by two mechanisms, known as antigenic drift and antigenic shift.
Antigenic drift is the frequent occurrence of point mutations resulting from defects in RNA replication mechanisms, while antigenic shift is less frequent, involving re-assortment of the RNA segments arising from exchanges between different strains in host cells infected by multiple viruses.
Protection by current human influenza vaccines is achieved by use of inactivated or attenuated forms of the corresponding pathogen and appears to require the function of neutralizing antibodies against the external HA and NA glycoproteins. However, these glycoproteins mutate rapidly through antigenic drift and current vaccines become ineffective as mutational differences accumulate in the circulating strains. In order to overcome the antigenic variability of influenza external glycoproteins, new vaccine strategies are increasingly directed at conserved regions of the viral proteins for production of T cell epitope-based vaccines. The goal is to identify conserved sequences that function as epitopes recognized by human leukocyte antigen (HLA) molecules for presentation to CD8 + and CD4 + T cells that provide immunity against all influenza A virus subtypes and obviate the need for yearly vaccine update. Several animal model studies taking this approach have reported T cell responses that reduce morbidity and promote recovery in mouse models of influenza challenge [1][2][3][4]. Both CD8 + and CD4 + T cell responses are required; CD8 + T cells to kill infected cells [5,6] and CD4 + T cells for the development of an effective immune response and immune memory [7][8][9]. A complication of cellular immunity is that T cell responses are dependent upon antigen presentation by highly polymorphic HLA molecules that vary greatly among human populations. However, the limited population coverage of some HLA alleles may be alleviated by focusing on T cell epitopes recognized by HLA supertypes that bind largely overlapping peptide repertoires on the basis of the specificity for the main anchor positions of the presented peptides [10,11].
We previously reported a detailed study of evolutionarily conserved sequences of all human and avian influenza A viruses that were recorded over the past 30 years (36,343 sequences) [12]. Fifty-four (54) sequences, ranging from 9 to 58 amino acids (aa) of the PB2, PB1, PA, NP, and M1 sequences were conserved in at least 80%, and in most cases 95-100% of all recorded human H1N1, H3N2, H1N2, and H5N1, and avian subtypes. These sequences have remained evolutionarily stable for all recorded human and avian influenza A viruses during the past decades, and are thus prime candidates for the development of T cell epitopebased vaccines against multiple influenza strains. However, the function of these conserved sequences as HLA-restricted T cell epitopes and the incidence of variant sequences in association with the conserved sequences were not known. Herein, we have focused on the identification and characterization of peptides of influenza virus A/New York/348/2003 (H1N1) that contain conserved sequences and elicit HLA-restricted T cell responses. HLA transgenic mice (HLA-A2, -A24, -B7, -DR2, -DR3, and -DR4) were immunized with the selected peptides. The peptides that elicited T cell activation by IFN-c ELISpot assay and thus contained T cell epitopes were selected and analyzed for properties relevant to vaccine development, including evolutionary conservation and diversity, and correspondence to the 2007-2009 human H1N1 sequences.

Ethics Statement
Mice were maintained in a pathogen-free facility at the Johns Hopkins University according to IACUC guidelines.

Influenza Peptides
Peptide arrays of PB2 (BEI Cat.: NR-2616), PB1 (NR-2617), PA (NR-2618), NP (NR-2611), and M1 (NR-2613) of influenza virus A/New York/348/2003 (H1N1) were obtained through the NIH Biodefense and Emerging Infections Research Resources Repository, NIAID, NIH (BEI). A total of 196 peptides (all 17 aa long except PB2 268-283 and 393-408 ) were selected to fully cover all highly conserved sequences under study. Where these sequences spanned two or more 17 aa peptides, the consecutive peptides were overlapped by 11 aa ( Figure S1). Two immunization peptide pools for immunization were formed: one composed of 84 PB2 and 13 M1 peptides (Table S1), and the second composed of 48 PB1, 23 PA, and 28 NP peptides (Table S2). Each of the 196 peptides was dissolved in 100% DMSO and constituted to 20% with sterile filtered water. The final concentration of each peptide was 2 mg/ml. The dissolved peptides were stored at 220uC.

HLA Transgenic Mice
Six different strains of HLA transgenic mice were used to cover HLA alleles of class I and class II supertypes. These six alleles were selected based on their prevalence in the Caucasian population and the availability of HLA transgenic mouse strains in our laboratory. The HLA class I supertypes studied were HLA-A2 (A*0201) mice expressing a chimeric heavy chain with murine a3 domain and human b2m. Both H-2D b and murine b2m genes were disrupted by homologous recombination [13], HLA-A24 (A*2402) mice express a chimeric heavy chain and human b2m; the H-2K b , H-2D b , and murine b2m genes were disrupted by homologous recombination (Lemonnier et al., unpublished), HLA-B7 (B*0702) mice express a chimeric heavy chain with the HLA-B*0702 a1 and a2 domains and the H-2K d murine a3 domain [14]. The H-2K b and H-2D b genes in HLA-B7 mice were inactivated by homologous recombination. The HLA-A2 and -B7 transgenic mice were kind gifts from Steve Pascolo (Institut Pasteur, France) and Pierre-Simon Rohrlich (Institut Pasteur, France), respectively.

Immunization
Mice were immunized with the selected 196 peptides in 2 pools by use of a protocol which had been validated for T cell studies [18] and adapted for these transgenic mice studies. Peptides were pooled in matrixes as described [19] and injected in groups of 9 mice of each transgenic strain: two for matrix array screening, two for identifying individual peptides, four for characterizing apparent functional avidity of T cells to positive peptides at three titration points: 10, 1, and 0.1 mg/ml peptide concentrations, and one as a control (adjuvant alone). Mice were injected subcutaneously at the base of tail with 100 ml of the immunization peptide pool in TiterMaxH Gold adjuvant (TiterMax, Norcross, GA) (1:1). The amount of each peptide injected was 1 mg/mouse. After two weeks, spleens were harvested for IFN-c ELISpot assay.

IFN-c ELISpot Assay
Harvested spleens from immunized transgenic mice were selectively depleted of CD8 + or CD4 + T cells by use of anti-CD8 or anti-CD4 antibody-coated immunomagnetic beads with LD columns (Miltenyi Biotec, Auburn, CA) according to the manufacturer's protocol. The resulting CD8 + or CD4 + depleted splenocytes were tested individually by IFN-c ELISpot assays against the 196 influenza peptides arranged in two 10610 matrix arrays, resulting in 40 peptide pools, where each peptide was present in two different pools, as described [19]. Peptides identified as immunogenic in the matrix array screen were retested individually in a confirmatory assay and a peptide titration assay. Thus, each ELISpot positive response was detected three times: by matrix array screening, individually by confirmatory assay, and by peptide titration.
The ELISpot assays were performed using mouse IFN-c ELISpot sets from BD Biosciences (San Jose, CA) according to the manufacturer's protocol. Briefly, the ELISpot plates were coated with anti-IFN-c at 5 mg/ml and incubated at 4uC overnight. The plates were blocked with RPMI 1640 containing 10% heat-inactivated fetal calf serum, 2 mM L-glutamine, 100 mg of streptomycin/ml, and 100 U of penicillin for 2 h at room temperature, and either CD8 + -or CD4 + -depleted splenocytes (0.521.0610 6 cells/well) were then added for assays of class II and I T cell epitopes, respectively. The cells were cultured at 37uC in 5% CO 2 in the presence of peptide pools (final concentration of each peptide was 10 mg/ml) or individual peptides at final concentrations of 10 mg/ml, 1 mg/ml, and 0.1 mg/ml. Wells with medium alone served as background; Concanavalin A (2.5 mg/ml; Sigma-Alrich, St. Louis, MO) was used as a polyclonal stimulator; and known HLA-restricted peptides from Dengue serotype 3 were included in each assay as positive controls. After 16 h of culture, the plates were washed and incubated with biotinylated anti-IFN-c for 2 h at room temperature, followed by HRP-conjugated streptavidin for 1 h at room temperature. Reactions were developed with AEC substrate (Calbiochem-Novabiochem, San Diego, CA). Final enumeration of IFN-c spot-forming cells (SFC) was performed using the Immunospot Series 3B Analyzer ELISPOT reader (Cellular Technologies, Shaker Heights, OH) with aid of the Immunospot software version 3.0 (Cellular Technologies), indicating the number of SFC/10 6 cells. The results were considered positive if the number of SFC subtracted by those in the background (culture with medium alone) was above 10 and the number of SFC was higher than the background plus two standard deviations. The results shown are SFC minus background, which was consistently found to be less than 15 spots/10 6 cells throughout the experiments.

Experimentally Identified Human T Cell Epitopes in the Highly Conserved Sequences of Influenza A Viruses
Published influenza T cell epitopes within the highly conserved sequences were identified by use of the Immune Epitope Database and Analysis Resource (IEDB, http://www.immuneepitope.org/) [20]. The published epitope sequences were derived from various T cell assays that included T cell proliferation, IFN-c ELISpot, HLA tetramer staining, and 51 Cr release assays. Only epitope data from unique sequences and containing HLA restriction information were included.

Determination of Human Self-Peptide in Influenza Peptides
The 196 influenza 17 aa peptides were compared using the blastp program against the non-redundant protein sequences database restricted to human (taxid:9606) at NCBI (http://ww. ncbi.nlm.nih.gov/BLAST/) to detect the presence of fragments identical to human peptides. As the default search parameters were not suitable to probe for short peptide sequences of length 30 or less, the following parameters were used: word size of 2, expectation value of 30,000, matrix was PAM30, low complexity filter was disabled, and composition-based statistics was set to 'no adjustment'. We disregarded search results containing predicted sequences and human peptides with fewer than six contiguous identical residues as the probability of matching five or less residues is high and non-significant.

Conservation of Influenza A (H1N1) T Cell Epitopes
The dataset and methodology for identification of highly conserved influenza protein sequences among pathogenic influenza strains for the past 30 years had been described by Heiny et al. [12]. Briefly as reported, 3763 NP, 3781 M1, 3111 PA, 3175 PB1, and 3144 PB2 sequences were extracted from the NCBI GenBank and GenPept databases (as of September 2006) and multiple sequence alignments of the individual proteins were performed. The Antigenic Variability Analyzer (AVANA) tool [21] was used to identify the most frequent 17 aa sequence present in at least 80% of all the recorded viruses.
The 2007-2009 human H1N1 sequences were compared with the T cell epitopes of A/New York/348/2003 (H1N1), by aligning the protein sequence records of human H1N1 M1, PB1, and PB2 retrieved from the NCBI Influenza Virus Sequence Database (http://www.ncbi.nlm.nih.gov/genomes/FLU/FLU.html, as of Jun 17, 2009) and application of the AVANA tool to identify the most frequent sequence and its variants for each year.

HLA-Restricted Epitopes of Highly Conserved Human and Avian Influenza A Virus Sequences
The previously described 54 highly conserved influenza A sequences of 9 or more contiguous aa of the recorded human and avian influenza strains were represented by a total of 956 aa of the M1, NP, PA, PB1, and PB2 proteins [12]. The majority of the conserved sequences, 650 aa, were in the PB1 and PB2 proteins; there were no conserved sequence in NA, M2, NS1, and NS2. A total of 196 peptides (BEI) of the A/New York/348/2003 (H1N1) M1, NP, PA, PB1, and PB2 proteins were selected based on the presence of the conserved sequences. The immunogenicity of these 196 conserved influenza peptides was studied by immunizing HLA-A2, -A24, -B7, -DR2, -DR3, and -DR4 transgenic mice. Organization of the 54 conserved sequences in the BEI 17 aa peptides depended on their length and position. Conserved sequences that spanned adjacent 17 aa peptides were repeated up to a maximum of 11 aa because of overlapping peptide synthesis (Tables S1 & S2, Figure S1). Peptides with conserved sequences of less than 17 aa contained mixtures of conserved and non-conserved aa. Thirty-two (32) short conserved sequences (9 to 16 aa) were present in various lengths with adjacent non-conserved aa. Conserved sequences of greater length (22 sequences of 17 to 57 aa) were present as complete (65 of the 196 peptides) or partial sequences in the overlapping peptides. The longest conserved sequence was PB1 518-575 which was included as part of a cluster of completely conserved aa of 7 overlapping peptides. Table 1. HLA-A24, -B7, -DR2, -DR3, and -DR4 restriction of 54 peptides of influenza proteins M1, NP, PA, PB1, and PB2 that contain conserved sequences of 9 or more amino acids.

Protein
ELISpot positive 17 aa peptide* A24# Immunization of the HLA transgenic mice with the 196 H1N1 peptides was carried out with 2 pools of about 100 peptides each, with groups of 9 mice of each transgenic strain. Interferon-c (IFNc) ELISpot assays for HLA-restricted class I and class II responses were performed with splenocytes of the immunized mice that were depleted of CD4 + and CD8 + T cells, respectively, to identify the responding T cell subset. The initial assays contained matrix arrays of peptide pools followed by validation assays with individual peptides [19]. Of the 196 peptides, 54 contained T cell epitopes that elicited 63 ELISpot responses (8 A24, 2 B7, 16 DR2, 17 DR3, and 20 DR4) ( Table 1). None of the 196 peptides tested induced T cell responses in mice expressing the HLA-A2 allele. Forty-seven (47) of the 54 epitopes were restricted by one HLA allele; eight by class I HLA-A24 and -B7, and 39 by class II HLA-DR2, -DR3, and -DR4. The remaining 7 peptides contained epitopes that were presented by at least two HLA alleles of distinct supertypes i.e. they contained multiple or promiscuous T cell epitopes. Epitopes of PB1 680-696 and PB2 548-564 were presented by both HLA class I and II alleles. Sixteen (16) pairs of consecutive peptides were restricted by the same HLA allele, possibly because there were identical epitopes in the overlapping 11 aa sequence shared by the 2 adjacent peptides. Clusters of 2 or more T cell epitopes with at least 16 conserved aa were M1 175-197 , PB1 120-142, 340-374, 489-576 , and PB2 42-64, 126-146 (Table 1, Figure 1).
The apparent functional avidity of T cells to each of the 54 peptides was titrated at three peptide concentrations of 10, 1 and 0.1 mg/ml in IFN-c ELISpot assays. Of the 63 positive ELISpot responses, including the responses of peptides restricted by multiple HLA alleles, 52 activated IFN-c secretion at each of the three concentrations used in the ELISpot assay, 9 elicited at concentrations of 10 and 1 mg/ml, and 2 peptides (NP 408-424 and PB1 328-344 ) elicited solely at the highest peptide concentration ( Table 1).

Presence of Reported Human T Cell Epitopes in the Conserved Sequences of Influenza A
The conserved peptides of this study were compared with reported T cell epitope sequences of humans infected with influenza A viruses extracted from the IEDB [20]. Twenty-one (21) of about 800 reported human T cell epitopes of PB2, PB1, PA, NP, and M1 were found to contain sequences of 9 or more conserved amino acids of all recorded 1977-2006 influenza A viruses ( Table 2). These HLA-restricted T cell epitopes were mainly from H1N1, H3N2, and H5N1 infections and included sequences restricted by a broad range of HLA class I and II alleles, including many alleles not covered by the transgenic mice of this study. For example, the same T cell epitope ''RMVLASTTAK'' in M1 178-187 was reported to be restricted by HLA-A3 and -A11 [22,23]. Clusters of overlapping epitopes were also observed within the conserved sequences, for example, M1 123-137 had three overlapping epitopes ( 123 ALASCMGLIY 132 was restricted by A1; 125 ASCMGLIY 132 by B35; and 129 GLIYNRMGA 137 by A2) [22,24]. Thus, the highly conserved sequences contained common epitopes shared by pathogenic influenza strains and could be restricted by a broad range of HLA alleles.

Analysis of the Presence of Human Peptide Sequences in Influenza Peptides
Each of the 196 influenza 17 aa peptides used in this study was compared with human proteome sequences to investigate the   possibility of identity to human self-antigens that could trigger an autoimmune response to immunization. Specifically, we screened for exactly identical sequences of at least 8 continuous aa, the minimum binding peptide length for MHC class I [25]. Many of the conserved sequences of the influenza peptides contained sequences of 6 aa found in human proteins such as voltage-gated sodium channel, dystrophin etc. The longest influenza A sequence with an identical human counterpart was 7 aa of PA 131-137 but none contained sequences of 8 or more aa identical to the human proteome ( Table 3).

Variants of the Conserved T Cell Epitope Sequences
The 54 HLA-restricted T cell epitopes of A/New York/348/ 2003 (H1N1) strain were analyzed by the Antigenic Variability Analyzer (AVANA) tool [21] for identification of (a) the consensus sequence (most frequent sequence) in the context of influenza A conserved sequences over the past   These data demonstrated that when T cell epitopes contain mixtures of conserved and non-conserved aa, the occurrences of mutated sequences in a subsequent influenza A strain are greatly enhanced.

Discussion
An enigma of the immunobiology of influenza A is that vaccines fail to provide long term protection against infection and natural infection does not prevent reinfection. The rapid mutation of the viral proteins, particularly the external HA and NA proteins that are targets for neutralizing antibodies, is credited with a significant role in this loss of immunity. Defective adaptive immunity is also observed with several RNA viruses (including HIV-1 and dengue viruses) with high rates of mutation that result in multiple genetic variants bearing mutated T cell epitope sequences. This has resulted in widespread attention to the use of conserved T cell epitopes of non-structural viral internal proteins [26][27][28][29]. However, the occurrence of reinfection, despite the human T cell response to conserved sequences after natural infection, suggests the function of a viral mechanism that negatively effects in the host immune response to influenza infection. One possibility is the dual immunosuppressor roles of the influenza A NS1 protein that inhibit innate immunity by preventing type I IFN release, as well as adaptive immunity by attenuating human dendritic cell maturation and the capacity of dendritic cells to induce T cell responses [30]. There is also the concepts of immunological ''original sin'' and altered epitope ligands where mutations in or adjacent to T cell epitopes preserve binding to MHC molecules but present an altered surface to the original T cell antigen receptor, resulting in an impaired or modified T cell response, including T cell immunosuppression [31][32][33][34][35][36][37]. Thus, variant viruses bearing mutated T cell epitopes escape memory T cells recognizing the initial epitopes, leading to impaired clearance of variant viruses reinfecting the same individuals.
In this study, HLA transgenic mice, HLA-A2, -A24, -B7, -DR2, -DR3, and DR4, were immunized with 196 overlapping H1N1 peptides of the A/New York/348/2003 strain that contained the highly conserved sequences of the M1, NP, PB1, PB2, and PA proteins of all recorded human and avian influenza A viruses of the past 30 years [12]. Fifty-four (54) of these peptides (22 PB1, 16 PB2, 9 NP, 4 PA, and 3 M1) elicited 63 HLA-restricted T cell responses by IFN-c ELISpot assay, where 7 peptides were restricted by multiple alleles. It should be recognised that the HLA-restricted T cell activation elicited by these peptides in the transgenic mice identified peptides that contain T cell epitopes, but not the specific identity of the T cell epitopes. While current convention refers to antigenic sequences recognized by T cells in association with an HLA molecule as ''epitopes,'' this term was introduced by Niels K. Jerne in 1960 in relation to defined structures recognized by antibodies. In contrast, he used the term ''cryptotopes'' to define the processed peptides that are recognized by T cells in association with an HLA molecule. There are very few sequences called T cell epitopes for which the actual structures of the epitopes are known. Moreover, as is shown in the text, 7 of the peptides elicit responses in 2 or 3 HLA transgenic mice, indicative of multiple or promiscuous epitopes in each peptide.
Many of the epitope peptide sequences reported herein also contain sequences reported by others as T cell epitopes of individuals infected by H1N1, H3N2, and H5N1 viral strains and restricted by a broad range of HLA class I and II alleles. Thus, it is reasonable to expect that the conserved peptides identified here can elicit human T cell epitope responses in the context of several HLA alleles and HLA-supertypes [38] and that the memory T cells can cross-react with epitopes from H1N1, H3N2, and H5N1 [27,39,40]. The class I alleles described herein HLA-A*0201, -A*2402, and -B*0702 belong to the distinct supertypes A2, A24, and B7, respectively [10,11]. HLA class II supertypes are not as well documented but the 3 alleles of the transgenic mice of this study are assigned to supertypes DR1, DR3, and DR4 [41] based on similar protein and three-dimensional structures. Thus, fewer HLA alleles within the supertypes could be selected to protect a broad population in contrast to selection of individual alleles. Analysis of the conservation and mutational variants of these H1N1 HLA-restricted epitopes was revealing of the marked effect that single aa mutations may have on the representation of T cell epitopes in evolving virus populations. Over the 3 years interval (2007 to 2009) between the database records analyzed by Heiny et al. (2006) to the current 2009 H1N1 sequence analysis, only 8 of the 54 highly conserved T cell epitope sequences were without mutational change. These 8 peptides (M1 175-191 , 181-197 , PB1 31-47 , 120-136 , 126-142 , 489-505 , 495-511 , and 548-564 ) were representative of almost complete conservation, 95-100%, during the previous recorded history of human H1N1 virus sequences. All others of the identified HLA-restricted T cell epitopes contained at least 1 aa substitution, primarily but not exclusively, of the nonconserved aa of the H1N1 peptides. Our data suggest that the most favorable sequences for a T cell epitope-based vaccine are the 17 H1N1 T cell epitopes of the PB1, PB2, and M1 proteins (Table 4). We are presently investigating the protective effect of these sequences against live influenza challenge. These 17 T cell epitopes were highly conserved over the 33 years (1977-2009) of the examined database records, representing 88 to 100% of all   recorded avian and human influenza A viruses, including the H1N1 isolates. Further, they are clustered and have distinct advantages in the design of an epitope-based genetic vaccine, including the retention of native sequences for the function of transporters associated with antigen processing (TAPs) [42] and for the flanking sequences that are reported to modulate epitope processing and function in the selection of immunodominant epitopes [43]. Each of these 17 sequences, except M1 181-197 and PB1 537-553 , was also characterized by high apparent functional avidity at the lowest peptide concentration of 0.1 mg/ml in the IFN-c ELISpot assay. Several studies showed that high avidity CD8 + T cells were more effective in limiting viral replication in vitro [44][45][46]. Further, the 17 T cell epitopes had no identity of 8 or more continuous aa to human peptides that might trigger onset of human autoimmune diseases by priming autoreactive T cells. It is also noteworthy that several of the epitopes are located in described functional domains: PB1 518-575 in the interacting domain of PB1 with PB2 (PB1 506-659 ) [47]; and the overlapping PB2 126-142 and PB2 132-148 in the PB1-and NP-binding domain of PB2 1-269 [48]. T cell epitopes within funtional domains would remain conserved over time as viral mutations useful towards immune escape may disrupt the function of the domains and affect viral fitness unless compensated functionally by multiple comutations [49,50]. Thus, a T cell-based vaccine including these 17 highly conserved T cell epitopes, as described in this study, may possibly greatly reduce, if not eliminate, the incidence of variant amino acids of the corresponding T cell epitopes of any future influenza A pathogen.