Broad-Based Influenza-Specific CD8+ T Cell Response without the Typical Immunodominance Hierarchy and Its Potential Implication

Syngeneic murine systems have pre-fixed MHC, making them an imperfect model for investigating the impact of MHC polymorphism on immunodominance in influenza A virus (IAV) infections. To date, there are few studies focusing on MHC allelic differences and its impact on immunodominance even though it is well documented that an individual’s HLA plays a significant role in determining immunodominance hierarchy. Here, we describe a broad-based CD8+ T cell response in a healthy individual to IAV infection rather than a typical immunodominance hierarchy. We used a systematic antigen screen approach combined with epitope prediction to study such a broad CD8+ T cell response to IAV infection. We show CD8+ T cell responses to nine IAV proteins and identify their minimal epitope sequences. These epitopes are restricted to HLA-B*44:03, HLA-A*24:02 and HLA-A*33:03 and seven out of the nine epitopes are novel (NP319–330# (known and demonstrated minimal epitope positions are subscripted; otherwise, amino acid positions are shown as normal text (for example NP 319–330 or NP 313–330)), M1124–134, M27–15, NA337–346, PB239–49, HA445–453 and NS1195–203). Additionally, most of these novel epitopes are highly conserved among H1N1 and H3N2 strains that circulated in Australia and other parts of the world.


Introduction
The 1918 Spanish influenza pandemic resulted in over 50 million deaths worldwide and remains one of the deadliest plagues ever experienced in human history [1]. Annually, influenza infection continues to cause epidemics and is responsible for 3-5 million severe infections and 500,000 deaths worldwide https://www.who.int/en/news-room/ fact-sheets/detail/influenza-(seasonal) (accessed on 27 April 2021), even though influenza vaccines are widely available. These vaccines are designed to elicit neutralising antibodies, however, their efficacy in elderly and young are still debated [2]. Importantly, CD8 + T cells play an essential role in controlling influenza infections [3,4], as they mainly recognise antigens derived from more conserved internal viral proteins, such as NP [5,6] and memory influenza-specific CD8 + T cells are associated with enhanced recovery from novel influenza infection [7].
CD8 + T cells recognise viral peptides bound to HLA-I molecules presented on the surface of infected cells [8]. Many peptides generated from influenza proteins could potentially be presented to CD8 + T cells, however, only a small fraction of such peptides are recognised by virus specific CD8 + T cells with a measurable response. Among these, often only one or two peptides are able to stimulate a large response, a phenomenon known as immunodominance [9,10]. Most studies that investigated the underlying mechanisms of immunodominance often used syngeneic mouse models, such as the well-studied influenza A virus (IAV) infection model in C57BL/6 and BALB/c mice [11,12]. Although such models are well suited for studying factors such as peptide-binding ability to MHC, antigen-processing efficiency, and TCR repertoires and their impact on immunodominance [11][12][13][14][15], they are not the best means to study the impact of MHC polymorphism on immunodominance. Boon et al. studied CD8 + T cell responses in individuals with single HLA-A or -B allele mismatch and clearly demonstrated that some HLA alleles play bigger roles in presenting IAV antigens [16,17]. Similarly, using HLA-A2 positive donors and a systematic CD8 + T cell assessment approach, our laboratory was able to show the immunodominant CD8 + T cell response to HLA-A2/M 58-66 could clearly be affected by other HLA alleles [5]. If such observation is applicable to all HLA-A and -B allele combinations, it is then possible that some individuals may never present immunodominant CD8 + T cell response simply because their HLA alleles do not select such immunodominant epitopes. As a direct result, the CD8 + T cell responses in these individuals will likely be broad and have many subdominant determinants without a typical immunodominance hierarchy; this type of response has not been previously reported for CD8 + T cells specific to IAV infections, although our laboratory has previously reported such an IAV-specific CD4 + T cell response [18].
In this study, we report a broad-based IAV-specific CD8 + T cell response from a healthy donor to nine IAV antigens. Interestingly, the CD8 + T cell responses specific to the epitopes from the nine IAV proteins showed similar intensity. We believe that an in-depth understanding of such CD8 + T cell responses could help vaccine design and may predict severe clinical outcome.

Human Peripheral Blood Mononuclear Cells (PBMCs) and Antigen Presenting Cells (APCs)
PBMCs of healthy donors used in this study were obtained from the Australian Red Cross Blood Service (ARCBS) in Melbourne under the agreement 12-07VIC-17. PBMCs were isolated from blood buffy coats using Ficoll-Hypaque gradient centrifugation, then resuspended in FCS (Hyclone, South Logan, UT, USA) containing 10% dimethyl sulfoxide (DMSO) and stored in liquid nitrogen until use. The autologous B lymphocyte cell line (BLCL) was made from the donor's PBMCs using EBV transformation [5], and cultured in RPMI 1640 supplemented with 10% FCS (Hyclone) (RF-10).

Culture Medium, Synthetic Peptides, Antibodies and Viruses
Unless otherwise stated, antigen presenting cells were cultured in RF-10 containing 50 µM 2-ME, 2 mM L-glutamine and antibiotics (penicillin 100 U/mL and streptomycin 100 µg/mL) (all from Gibco, Grand Island, NE, USA)). Peptides used in this study were synthesized by Mimotopes (Clayton, VIC, Australia). Peptides were reconstituted to a concentration of 1 mM in DMSO. APC-labelled anti-CD8 was purchased from BD Biosciences, and PE-cy7 labelled anti-IFNγ was purchased from Invitrogen (San Diego, CA, USA).

Generation of Poly-Specificity CD8 + T Cell Lines In Vitro
For establishing primary antigen-specific CD8 + T cell lines, 1 × 10 6 PBMCs were resuspended in 200 µL acidic medium and infected with PR8 at an MOI (Multiplicity of Infection) of 10 for 1 h in a 37 • C water-bath. Next, 2 mL of RF-10 was added to the cells and incubated for a further 4 h in a 37 • C 5% CO 2 incubator. The infected PBMCs served as APCs and subsequently co-cultured with 9 × 10 6 PMBCs (responders) for two weeks in T cell culture medium containing 20 IU/mL of hrIL-2.
For CD8 + T cell re-stimulation, BLCLs were used as stimulators; they were infected with PR8 overnight (16 h), then irradiated for 100 Grey before being added to T cells in a T cell culture medium containing 40 IU/mL of hrIL-2. The ratio of T cells to APCs was generally 10:1.

rVV Infection for Identifying Protein-Specific CD8 + T Cells
APCs were infected with a panel of individual IAV protein encoding rVVs (rVV-HA, rVV-NA, rVV-NP, rVV-NS1, rVV-NS2, rVV-M1, rVV-M2, rVV-PB1, rVV-PB2, rVV-PB1F2, rVV-PA) or the empty vector rVV-CR19. Briefly, APCs were washed with 0.1% BSA/PBS twice, then re-suspended in the same buffer and infected with rVV encoding individual IAV proteins at MOI of 10. Cells were then incubated in a 37 • C water-bath for 1 h and were gently vortexed every 15 min. After the first hour of infection, cells were topped up with 2 mL fresh RF-10 and incubated in a 37 • C 5% CO 2 incubator for a further 16 h.
For identifying a specific antigenic protein, IAV-specific CD8 + T cells were screened with the above rVV infected APCs.

Identification of Novel IAV-Specific CD8 + T Cell Epitopes and Its Minimal Epitopes
Once the specific antigenic proteins were identified, polyclonal CD8 + T cells were directly used to work out the specific peptide regions by screening a whole set of synthetic 18mer overlapping peptides with 6 amino acid shifts. For identifying the minimal epitopes, the shorter peptides within the 18mer region of interest were predicted according to the peptide-binding motifs and screened by single peptide-specific CD8 + T cell lines.

HLA Restriction Assay
BLCLs with partial donor HLA match were pulsed with 10 −6 M of relevant peptides for 1 h at room temperature. These cells were then extensively washed to remove free peptides and were used as APCs in a co-culture with antigen-specific CD8 + T cells for 4 h in the presence of Brefeldin A (BFA). HLA-restriction was assessed by intracellular cytokine staining (ICS) for IFN-γ production.

Intracellular Cytokine Staining (ICS) and Antigen Presentation Kinetics Assay
The IFN-γ ICS assay was performed as previously described [5]. For antigen presentation kinetics assay, the detailed method was previously published [19]. Briefly, the antigen-presentation of PR8/rVV-infected autologous BLCLs and homozygous BLCLs were stopped by the addition of BFA (Sigma-Aldrich, St Louis, MO, USA) at different time points. Antigen-specific CD8 + T cells were then added for 4 h to assess the antigenpresentation at that particular time point; the T cells were then harvested and synchronized on ice before being stained together in a standard ICS. Samples were analysed using a BD FACS Canto II machine (BD Biosciences, Franklin Lakes, NJ, USA) and FlowJo software (Tree Star Inc, Ashland, NE, USA).

A Systematic Approach for Identifying IAV-Specific CD8 + T Cell Responses
A two-step systematic approach [5] was used to (i) identify the viral proteins that stimulated CD8 + T cell responses (Figure 1a), and then to, (ii) determine minimum antigenic peptide sequences and their HLA restrictions ( Figure 1b). First, IAV-specific polyclonal CD8 + T cell lines were established by infecting 10% of autologous PBMCs with PR8 to stimulate all potential IAV-specific memory CD8 + T cells among these PBMCs. After two weeks, CD8 + T cells specificities were assessed by infected autologous BLCLs with a panel of rVVs encoding 11 individual IAV proteins followed by synthetic overlapping 18mer peptides via ICS for IFN-γ production. Once the specific antigenic regions were identified, the minimal epitope sequences were determined using either overlapping 13mer peptides and/or shorter predicted peptides within the active 18mer sequence. Partially HLA-matched cell lines were used to determine these epitopes' HLA-restrictions. to stimulate all potential IAV-specific memory CD8 + T cells among these PBMCs. After two weeks, CD8 + T cells specificities were assessed by infected autologous BLCLs with a panel of rVVs encoding 11 individual IAV proteins followed by synthetic overlapping 18mer peptides via ICS for IFN-γ production. Once the specific antigenic regions were identified, the minimal epitope sequences were determined using either overlapping 13mer peptides and/or shorter predicted peptides within the active 18mer sequence. Partially HLA-matched cell lines were used to determine these epitopes' HLA-restrictions. Figure 1. A systematic approach for identifying IAV-specific CD8 + T cell responses. (a) Establishing polyclonal IAV-specific CD8 + T cells. After two weeks culture, CD8 + T cells were screened by a panel of individual IAV proteins encoded by rVV to identify specific antigenic proteins. After knowing the specific antigenic protein, CD8 + T cells were screened with synthetic overlapping 18mer peptides to determine the specific antigenic peptide regions within the protein. (b) Generating individual peptide-specific CD8 + T cell lines using individual peptide-pulsed APCs to expand IAV-specific CD8 + T cells enriched via CD3CD8 downregulation-guided sorting. Two weeks later, CD8 + T cells were tested with shorter peptides to assess the core sequence and HLA restriction of each particular epitope.

Broad-Based IAV-Specific CD8 + T Cell Responses to Multiple IAV Proteins
To identify a donor with broad-based CD8 + T cell responses to IAV antigens, polyclonal IAV-specific CD8 + T cell cultures from PBMCs of 8 HLA-A2 + healthy donors and 7 HLA-A2-healthy donors [5,6] and 6 other donors (no HLA typing, Figure S1) were established as described in Figure 1a and were used to screen rVV encoding individual IAV proteins. Only one of the cultures from the HLA-A2 − healthy donors (donor 4 [6]) showed broad CD8 + T cell responses to nine IAV proteins (PA, NA, HA, M1, M2, NP, NS1, PB1, and PB2) but did not to PB1F2 and NS2 ( Figure 2). Although NP was previously demonstrated to be the most dominant antigen in most healthy donors [5,6], that is not the case in this individual as most of the CD8 + T cell responses specific to the nine IAV proteins showed similar intensity. Each rVV infected the cell line with similar efficiency, demonstrated by their similar rVV antigen expression levels ( Figure S2). A systematic approach for identifying IAV-specific CD8 + T cell responses. (a) Establishing polyclonal IAV-specific CD8 + T cells. After two weeks culture, CD8 + T cells were screened by a panel of individual IAV proteins encoded by rVV to identify specific antigenic proteins. After knowing the specific antigenic protein, CD8 + T cells were screened with synthetic overlapping 18mer peptides to determine the specific antigenic peptide regions within the protein. (b) Generating individual peptide-specific CD8 + T cell lines using individual peptide-pulsed APCs to expand IAV-specific CD8 + T cells enriched via CD3CD8 downregulation-guided sorting. Two weeks later, CD8 + T cells were tested with shorter peptides to assess the core sequence and HLA restriction of each particular epitope.

Broad-Based IAV-Specific CD8 + T Cell Responses to Multiple IAV Proteins
To identify a donor with broad-based CD8 + T cell responses to IAV antigens, polyclonal IAV-specific CD8 + T cell cultures from PBMCs of 8 HLA-A2 + healthy donors and 7 HLA-A2-healthy donors [5,6] and 6 other donors (no HLA typing, Figure S1) were established as described in Figure 1a and were used to screen rVV encoding individual IAV proteins. Only one of the cultures from the HLA-A2 − healthy donors (donor 4 [6]) showed broad CD8 + T cell responses to nine IAV proteins (PA, NA, HA, M1, M2, NP, NS1, PB1, and PB2) but did not to PB1F2 and NS2 ( Figure 2). Although NP was previously demonstrated to be the most dominant antigen in most healthy donors [5,6], that is not the case in this individual as most of the CD8 + T cell responses specific to the nine IAV proteins showed similar intensity. Each rVV infected the cell line with similar efficiency, demonstrated by their similar rVV antigen expression levels ( Figure S2).

Figure 2.
Broad-based CD8 + T cell responses to multiple IAV proteins. (a) Autologous BLCLs were infected with rVVs encoding individual IAV proteins for 16 hours before being used as APCs to assess CD8 + T cell responses via a standard ICS for IFN-γ production. (b) FACS profiles from this rVV screen.

Characterising Antigenic Peptide Regions in Multiple IAV Proteins
Following the identification of the IAV proteins responded to by these CD8 + T cells, a whole set of synthetic 18mer overlapping peptides covering the entire length of these proteins were screened by the above bulk T cell line to further identify antigenic regions in each protein ( Figure S3). The polyclonal CD8 + T cell line was subsequently enriched by CD3CD8 downregulation-guided sorting, using APCs infected by a single rVV encoding individual IAV protein to activate the specific CD8 + T cells [20]. The enriched CD8 + T cell lines were then tested with the previously used whole set of synthetic 18mer peptides to further confirm the results obtained from the bulk T cell line used in Figure 2. As shown in Figure 3 (results derived from the enriched CD8 + T cell lines), the antigenic regions from the above-mentioned nine IAV proteins were: . Interestingly, only one antigenic region in each IAV protein was recognised by each enriched specific CD8 + T cell line except for NS1, in which we failed to detect a clear antigenic peptide region although >10% T cells responded to rVV-NS1.

Minimal Epitopes of Differing Lengths Are Presented by Three HLA Alleles
It has been demonstrated that MHC is one of the critical determining factors of immunodominance [21,22] and the MHC-I peptide-binding groove can only accommodate peptides generally of 8-12 amino acid in length [23,24]. Therefore, minimal epitope sequences and their HLA restrictions were investigated to better understand such a broad-based CD8 + T cell response. First, single peptide-specific CD8 + T cell lines were generated either by 18mer peptide stimulation as shown in Figure 1b or by enriching through CD3CD8 downregulation-guided sorting [20]. All the newly identified 18mer peptides were shown to be restricted to HLA-A*33:03, HLA-A*24:02, or HLA-B*44:03 ( Figure S4). Interestingly, not a single peptide was shown to be restricted to HLA-B*46:01 expressed also by this donor. All three HLA molecules have stringent peptide-binding motifs [25][26][27]; for example, HLA-B*44 molecules require a glutamate (E) in position two (P2), and at the C-terminus either a tyrosine (Y) or phenylalanine (F), and less often tryptophan (W) [25]. Using this information, shorter peptides within the 18mer region of interest were predicted according to the peptide-binding motifs and screened by single peptide-specific CD8 + T cell lines. For HA-specific CD8 + T cells, HA 439-456 (AELLVLLENERTLDFHDS) and HA 445-462 (LENERTLDFHDSNVKNLY) peptides (overlapping region shown in bold) revealed a positive CD8 + T cell response among the overlapping 18mer peptides ( Figure 3). Moreover, these two neighbouring peptides stimulated T cell responses to a similar degree, indicating the minimal epitope was likely within the shared sequence HA 445-456 (LENERTLDFHDS). We then worked out these two positive peptides were restricted to HLA-B*44:03 ( Figure S4) using donor HLA-matched APCs from Table 1. As HLA-B*44:03 requires P2 and C-terminal anchor residues E and Y/F respectively, HA 445-453 LENERTLDF was predicted as the minimal epitope. The peptide and its single amino acid truncated versions were synthesised and tested in a peptide titration experiment. As shown in Figure 4a-i, HA 445-453 stimulated a CD8 + T cell response even at a concentration of 10 −9 M. However, HA 446-453 was only able to stimulate a weaker T cell response at much higher peptide concentration (10 −6 M), while HA 445-452 lost all the activity. Therefore HA 445-453 is the minimal epitope; subsequently its HLA-B*44:03 restriction was again confirmed (Figure 4a-ii). According to the Immune Epitope Database (IEDB) (www.immuneepitope.org) (last accessed on 27 April 2021), HLA-B*44:03 restricted HA 445-453 is a novel epitope identified in this study. Similarly, the M2 specific CD8 + T cells recognised both M2 1-18 (MSLLTEVETPIRNE WGCR) and M2 7-24 (VETPIRNEWGCRCNGSSD) to a similar degree ( Figure 3) and showed HLA-B*44:03 restriction ( Figure S4). M2 7-15 (VETPIRNEW) as was predicted as the minimal epitope within the shared M2 7-18 (VETPIRNEWGCR) region. Peptide titration experiments were performed for M2 7-15, M2 8-15 and M2 7-14 and very similar results were obtained (Figure 4b-i) as those acquired above, and HLA-B*44:03 restriction was confirmed (Figure 4b-ii). According to the IEDB, M2 7-15 is a novel epitope identified in this study.
In similar fashion, NP-specific CD8 + T cells recognised 18mer peptides NP 313-330 (YSLIRPNENPAHKSQLVW) and NP 319-336 (NENPAHKSQLVWMACHSA) (Figure 3) and were restricted by HLA-B*44:03 ( Figure S4). NP 319-330 (NENPAHKSQLVW) was predicted as the minimal epitope, which was confirmed by the decreased and total loss of T cell activating capacity of the amino-and carboxyl-terminal single amino acid truncation peptides, respectively (Figure 4c-i). This is a longer minimal epitope containing 12 amino acids and again HLA-B*44:03 restriction was confirmed (Figure 4c-ii). Thus, NP 319-330 is the minimal epitope and is also the longest HLA-B*44:03-restricted novel IAV epitope according to current information available on IEDB.

Identifying the Core Sequences of the Epitopes Presented by HLA-A*33:03
Similar to HLA-B*44:03, HLA-A*33:03 also possesses relatively stringent peptidebinding motifs. Although it tolerates alanine (A), isoleucine (I), leucine (L), phenylalanine (F), tyrosine (Y) or valine (V) at P2, it requires almost exclusively an arginine (R) at the C-terminus [26]. The M1-specific CD8 + T cells responded to M1 121-138 (AGALASCMGLIYNRMGAV) and M1 127-144 (CMGLIYNRMGAVTTEVAF) 18mers ( Figure 3) and the response was restricted by HLA-A*33:03 ( Figure S4). As the two 18mer peptides only contained a single arginine (R) at position 134, M1 127-144 still had some T cell activation capacity (Figure 3) likely due to weak stimulation at high concentration. The P2 residue was predicted as 125 A at the amino-terminus rather than 124 L or 123 A because 12mer and 13mer epitopes are rare, therefore, M1 124-134 (LASCMGLIYNR) and its single terminal truncation peptides were synthesised. As P4 is a cysteine, to avoid cysteinylation and peptide dimerization, these peptides were assessed in the presence of 500 µM TCEP [30]. As shown in Figure 4f-i, both M1 124-134 and M1 125-134 were quite potent in the titration experiment, while M1 124-133 was no longer capable of T cell activation due to missing the C-terminal anchor residue arginine. Judging from the titration curves, M1 124-134 is more potent than M1 125-134, therefore, M1 124-134 is the minimal epitope, restricted to HLA-A*33:03 (Figure 4f-ii), and again this is a novel epitope.
The NA-specific CD8 + T cells responded to two neighbouring 18mer peptides NA 331-348 (VKGFSYRYGNGVWIGRTK) and NA 337-354 (RYGNGVWIGRTKSHSSRH) (Figure 3) indicating the minimal epitope should be within the shared region NA 337-348 (RYGNGVWIGRTK). NA 337-346 (RYGNGVWIGR) was therefore predicted as the minimal epitope. This was confirmed by peptide titration experiments showing reduced T cell activation capacity of the two peptides with a single amino acid truncation at either end of NA 337-346 (Figure 4g-i). NA 337-346 was confirmed to be HLA-A*33:03 restricted (Figure 4g-ii) and it is not yet in IEDB.

Epitope Conservancy Among All the IAV Strains Circulated in Australia during 1971-2021 Period
To analyse the conservancy of all the novel immunogenic epitopes discovered in this study, IAV protein sequences were aligned, and amino acid differences were scored for the IAV strains circulated in Australia during the 50 years between 1971 and 2021. According to our analysis, two out of the seven novel IAV epitopes are 100% conserved in these viruses, including NP 319-330 , M1 124-134 in both H1N1 and H3N2. For M2 [7][8][9][10][11][12][13][14][15] , although amino acid differences are found, the HLA anchor residues of these three epitopes are still conserved ( Table 2). The information of circulated IAV strains (Table 3) was obtained from the National Center for Biotechnology Information (NCBI) influenza virus resource database accessed on 13 April 2021. Based on the above analysis, it is highly likely that most IAV-specific CD8 + T cells studied here would be able to fully or partially recognise cells infected by these different IAV strains including the 2009 H1N1 pandemic strain. Taken together, our results revealed multiple novel CD8 + T cell epitopes restricted to the individual's HLA-A and B molecules with 100% conserved epitope sequences among IAV strains.

Quantitative Assessment of HLA-I Peptide Presentation to the Antigen-Specific CD8 + T Cells
It has been demonstrated that the speed of antigen processing and presentation influences the hierarchy of the CD8 + T cell response during viral infections [11,[34][35][36]. To further understand the basis of such a broad-based IAV-specific CD8 + T cell response observed in this study, the antigen presentation kinetics of the above T cell epitopes were investigated using a BFA kinetics assay as slow antigen presentation could potentially explain the observed overall small and lack of immunodominant responses in this individual. BFA blocks pMHC-I egressing the ER [37] and therefore "freezes" pMHC-I complexes on the surface of virus infected APC at the time of addition and allows the assessment of pMHC-I antigen presentation level at any time point after infection [19]. This method also allows the comparison of different T cell populations engaging the same APC which presents various CD8 + T cell epitopes simultaneously. Additionally, the presentation kinetics of a well-studied immunodominant epitope M1 58-66 was used as the "golden standard".
The presentation kinetics results are displayed by their epitopes' HLA restriction (HLA-B*44:03, HLA-A*24:02 and *33:03). For the HLA-A*33:03 restricted M1 124-134 and HLA-B*44:03 restricted NP 319-330 , presentations were detected at 0.5 h after infection (beginning after the first-hour infection), however, at that time the presentation of M1 58-66 was barely detectable. M1 124-134 presentation rapidly activated 50% of its antigen-specific T cells approximately 2 h after rVV-M1 infection (Figure 5c), whereas M1 58-66 reached the same level after an additional 2 h and NP 319-330 after 4 h (Figure 5a). HLA-B*44:03 restricted epitopes, M2 [7][8][9][10][11][12][13][14][15] , showed similar slower presentation kinetics with detectable presentation at 2 h and 50% presentation at approximately 6 h after infection (Figure 5a). The presentation of the HLA-A*24:02-restricted epitopes, PA 130-138 and PB1 498-505 , were faster as they were detected at 2 h and reached 50% at 4 h (Figure 5b). The presentation of HLA-A*33:03-restricted epitope, M1 124-134 was faster than that of M1 58-66 . M1 124-134 became detectable at 0.5 h after infection and reached 50% saturating presentation at 2 h (as previously mentioned), however, NA 337-346 presentation was slightly slower. Taken together, these results indicate the speed of antigen presentation did not seem to correlate with the magnitude of CD8 + T cell response to each epitope albeit these responses were all assessed after in vitro T cell expansion using rVV-infected APCs. Figure 5. Most novel CD8 + T cell epitopes are presented with a fast kinetics. CD8 + T cells were activated by rVV-infected BLCLs at various time points (0, 0.5, 2, 4, 6, and 8 h) after the first-hour infection, then BFA and T cells were added to "freeze" and read out antigen presentation at that time point after rVV infection. The antigen presentation, reflected by the amount of epitope-specific T cell activation was assessed for IFN-γ production by ICS as described in the method. (a) HLA-B*44:03 restricted epitope presentation is compared to that of a well-studied immunodominant epitope M1 58-66 (shown in dotted line). (b) HLA-A*24:02 restricted epitope presentation is compared to that of M1 58-66 (shown in dotted line). (c) HLA-A*33:03 restricted epitope presentation is compared to that of M1 58-66 (shown in dotted line). This experiment was performed and analysed together. For clarity, the results are displayed according to single HLA restriction of the epitopes. Data shown are representative of two independent experiments with similar results. (d) The FACS profiles showing the purities of the CD8 + T cell lines specific for PB1 [498][499][500][501][502][503][504][505] , and HA 445-453 used in (a-c). Nil peptide: APC only control; +peptide: APCs were pulsed with each individual peptide at 10 −6 M for 1 h in RF-10 before being incubated with antigen-specific CD8 + T cells and assessed in a standard ICS.

Discussion
In this study, we report a broad-based CD8 + T cell response, rather than one with a typical immunodominance hierarchy, in an IAV-experienced healthy individual. Combining a systematic antigen screen approach with epitope prediction, CD8 + T cell responses to nine IAV proteins (M1, M2, NP, NS1, PA, PB1, PB2, HA and NA) except for NS2 and PB1-F2 and their minimal epitope sequences were identified. These epitopes are restricted to HLA-B*44:03, HLA-A*24:02 and HLA-A*33:03 and seven out of the nine epitopes are novel  ). Most of these novel epitopes are highly conserved among strains of H1N1 and H3N2 that circulated in Australia and other parts of the world.
Several HLA molecules such as HLA-A*02:01, HLA-A*03:01, HLA-B*57:01, HLA-B*18:01 and HLA-B*08:01 present IAV peptides to stimulate robust CD8 + T cell responses against IAV infection, whereas HLA-A*24:02, HLA-A*01:01, HLA-A*68:01, and HLA-B*15:01 often stimulate more limited CD8 + T cell responses specific to IAV [38]. In addition, HLA-A*24:02 was reported to be associated with severe pH1N1 disease and is an allele that is highly expressed in Indigenous Australians and Alaskans [39,40]. Due to the lack of immunodominant epitope presentation by this HLA, especially in the absence of other HLAs that present immunodominant epitopes, such as HLA-A*02:01, HLA-B*08:01, HLA-B*15:01, and HLA-B*35:03 [5,6,21,41], may indicate HLA-A*24:02 does not present an immunodominant IAV epitope and therefore plays a limited role during an anti-IAV cellular immune response which would result in severe pH1N1 disease [42]. Additionally, Quinones-Parra et al. [38] investigated pre-existing memory CD8 + T cell pools specific for the conserved H7N9 epitopes and found eight conserved NP epitopes including an HLA-A*24:02-restricted epitope NP [39][40][41][42][43][44][45][46][47]. They then showed CD8 + T cell responses to six of the eight conserved epitopes, however, no CD8 + T cell specific for A*24:02-NP 39 was detected ex vivo, suggesting although NP 39-47 is a conserved epitope, it is a weak one that does not elicit a significant CD8 + T cell memory response. Such CD8 + T cell responses are believed to be less protective. Our results might at least partially explain why Indigenous Australians were more susceptible to severe IAV infection and had higher mortality rates than other Australian populations during the 2009 H1N1 pandemic [43,44]. To date, researchers have mainly focused on assessing immunogenic epitopes for common HLA molecules, such as HLA-A*02:01, with much less attention paid to the HLA molecules associated with Indigenous Australians.
Immunodominance has been studied for decades. In many cases, removing the most dominant T cell epitope enables the T cells specific to the subdominant epitopes to expand more completely [45,46]. Although the exact mechanisms associated with this phenomenon remain poorly understood, several factors have been shown to influence it, such as peptide-binding ability, T cell competition for particular pMHC-I on the APC surface, dominant T cells actively suppressing other T cell development, as well as other factors [46]. For example, Kedl et al. generated CD8 + T cells to dominant epitope SIINFEKL (ova8) and subdominant epitope KRVVFDKL from Ovalbumin in H-2 b mice; when highaffinity TCR transgenic T cells (OT1 T cells) for ova8 were transferred, they completely inhibited the response of host OVA-specific T cell response to either epitope due to APC level competition [47]. However, in our case, no functional immunodominant CD8 + T cell responses were discovered, indicating that other mechanisms rather than APC level competition are more likely.
Early evidence demonstrated that T cell precursor frequency and the efficiency of peptide generation could influence immunodominance hierarchy [48]. For example, Day et al. found that co-expression of H-2K k and H-2D b greatly diminished CD8 + T cell response to the H-2D b -restricted immunodominant epitope PA 224-233 , as part of the T cell repertoire was deleted by having H-2K k , while T cell responses to H-2D b restricted NP 366-374 remained unaltered [49]. However, more recently a study from Neller et al. demonstrated that T cell precursor frequency might not correlate closely with the immunodominance hierarchy after being pathogen challenged [50]. It is possible none of the epitopes we identified in this study could stimulate a typical immunodominant response.
Even though MHC restriction should be the most significant determining factor, it is often under-considered and sometimes not considered, especially in syngeneic murine systems in which MHC are pre-fixed. In humans, since HLA molecules are extremely polymorphic, the overall immunodominance hierarchy is shaped up by the CD8 + T cell responses restricted to all the involved HLA alleles expressed by the individual. For example, HLA-A*02:01-M1 58-66 complexes stimulate an immunodominant CD8 + T cell response in many HLA-A*02:01 donors, but not in those who also express HLA-A*01:01, HLA-B*08:01, HLA-B*15:01, HLA-B*18:01, and HLA-B*35:03 [5]. Additionally, some epitopes can be restricted to a few different HLA alleles. For example, M1 125-134 has been reported as an HLA-A*33:01 epitope [33]. Interestingly, in this study, we show this same epitope can be restricted to HLA-A*33:03, therefore, it is envisaged if one individual expressed both HLA-A*33:03 and HLA-A*33:01, the response to M1 125-134 might be boosted into an immunodominant status.
HLA combinations were shown to be one of the most important elements contributing to CD8 + T cell immunodominance hierarchy [35]. Akram et al. demonstrated during IAV infection in HLA transgenic mice that only certain combinations of HLA allele co-expression enabled immunodominant CD8 + T cell responses. For instance, the IAV immunodominant CD8 + T cell response could only be detected in IAV-infected HLA-B*7/HLA-B*27 transgenic mice but not in HLA-A*2/ HLA-B*7 or HLA-A*2/HLA-B*27 mice [51]. Moreover, Boon et al. reported HLA alleles were not used equally and there was a hierarchy that existed between HLA alleles; for example, HLA-B*27:05 and HLA-B*35:01 were preferentially used in IAV-specific CD8 + T cell response, and HLA-B*08:01 and HLA-A*01:01 seemed to contribute less [17]. According to the above results, co-expression of certain alleles may determine not only which HLA may present an immunodominant CD8 + T cell response but also whether there is immunodominant CD8 + T cell response at all. In HIV studies, epitopes restricted by HLA-B molecules were significantly more frequently recognised than those presented by HLA-A or HLA-C molecules [52]. Interestingly in our case, the study subject expresses HLA-A*24:02, HLA-A*33:03, HLA-B*44:03, HLA-B*46:01, HLA-Cw*01:02, and HLA-Cw*07:01, yet the HLA-B*46:01 and the HLA-C alleles contributed little to the overall anti-IAV response. Furthermore, this set of HLA was not able to support an immunodominant CD8 + T cell response during IAV infection.
It is also worth considering infection exposure and vaccination history, as a specific viral strain might affect the outcome of CD8 + T cell response and the immunodominance hierarchy. Although little is known about the precise IAV exposure history of this donor, we have compared the conservancy of novel epitopes among all the IAV strains circulated in Australia in the past 50 years and found two out of the seven identified novel epitopes are 100% conserved among these strains ( Table 2). These results suggest that our assessment is likely valid even though only one IAV strain (PR8) was used for conducting the experiments in this study.
Chronic virus infections, such as HIV and HCMV are characterised by the accumulation of antigen-specific CD8 + T cells specific for immunodominant epitopes [53]. However, IAV usually causes acute infection which might not allow CD8 + T cells with a particular specificity to accumulate significantly. Therefore, if one does not possess a set of HLAs that is able to elicit a robust IAV-specific CD8 + T cell response, such individual may have a higher risk for developing severe IAV-related diseases. Our present results provide better understanding of IAV-specific CD8 + T cell responses in humans, especially in high-risk groups with HLAs that do not present major CD8 + T cell epitopes.
In this study, the novel peptides discovered are presented by HLA-B*44:03 and HLA-A*33:03. According to the IEDB, there is no HLA-B*44:03 restricted IAV immunodominant CD8 + T cell epitope published previously; moreover, no HLA-A*33:03 restricted IAV epitope has been identified so far. It is possible the set of HLA alleles this individual expresses failed to present any immunodominant epitope resulting in a broad-based, relatively low-magnitude CD8 + T cell response during IAV infection. Although we have only studied one such case on the broad-based CD8 + T cell responses without the typical immunodominance hierarchy, we believe this phenomenon might be wider spread than previously expected as published studies often only focused on a relatively small numbers of high frequency HLA alleles. It is possible many other HLA alleles, just like the alleles HLA-A*33:03 and HLA-B*44:03 investigated in this study, might not present IAV epitopes that are capable of stimulating immunodominant response partly because their peptide motifs are too stringent. As both the HLA haplotypes encoding HLA-A*24:02-HLA-B*44:03 and the ones encoding HLA-A*33:03-HLA-B*44:03 are very rare in Caucasian populations, quite rare in African and indigenous populations and generally less than 1% in most Asian groups except for China Jingpo minority and South Koreans (among these populations HLA-A*33:03-HLA-B*44:03 associated haplotypes can reach 5% according to http://www.allelefrequencies.net/, accessed on 25 April 2021), we have not been able to confirm our observation in another donor. It would be interesting and worthwhile to either confirm our observation using donor samples collected from the above mentioned Jingpo minority and/or South Korean populations or to carry out a large-scale screen to identify the HLAs associated with such responses in the future. Alternatively, it might be possible to HLA-type all the IAV-infected severe individuals to see whether there is any special association between their HLA alleles and their disease severity. Such knowledge may allow us to predict high-risk individuals during IAV infection.