Parallel T Cell Immunogenic Regions in Influenza B and A Viruses with Distinct Nuclear Export Signal Functions: The Balance between Viral Life Cycle and Immune Escape

Although SARS-CoV-2 has caused an ongoing worldwide pandemic since late 2019, influenza viruses remain a continuous threat to global health (1). Influenza viruses belong to the Orthomyxoviridae family, classified into A, B, C, and D types on the basis of the core proteins (2). Influenza A and B viruses (IAVs and IBVs) can cause epidemic (seasonal or interpandemic) influenza (3, 4). Although both IAV and IBVs can cause severe respiratory diseases in humans, IBVs have received less attention than IAVs (5). From 2000 to 2018, IBV infections accounted for 23.4% of the reported influenza cases in the flu seasons (6), causing a substantial burden in terms of morbidity and mortality (7).

The Ab-induced humoral immune response is critical for controlling acute and persistent influenza viral infections, whereas cellular immunity also plays a crucial role in defending against influenza virus infection (8). As enveloped, negative-sense, ssRNA viruses possessing eight genomic segments, IAVs and IBVs share most of their protein translation profiles, which have a diverse Ag spectrum for humoral and/or cellular immune responses (9). Hemagglutinin (HA) is the major target of the protective Ab response in both IAV and IBV infections. Meanwhile, the M1 protein plays multiple roles in the viral life cycle for both IAVs and IBVs and can shuttle between the nucleus and the cytoplasm (10, 11). A highly conserved immunodominant HLA-A*0201–restricted epitope AM_58–66GL9 (GILGFVFTL) derived from the IAV M1 protein was able to induce a high CD8⁺ T cell response (12–17). Moreover, AM_58–66GL9 peptide has been designed as a vaccine candidate to induce cross-protective T cells against influenza (18). However, to date, major CD8⁺ T cell Ags in IBV infection remain largely unknown. It also remains to be determined whether IBV M1 protein has an immunodominant epitope in the corresponding region of AM_58–66GL9 in IAV.

Specific cellular immune responses are primarily governed by molecules of the MHC (or in humans, HLA). The MHC molecules can bind peptide fragments derived from pathogens and then present them on the cell surface for specific T cell recognition (19). In response to the pathogen infections, the human immune system has evolved specific MHC class I molecules that determine which peptides will be presented and potentially recognized by the immune system (20). Previous structural determination of the HLA-A*0201/AM_58–66GL9 complex shows that the AM_58–66GL9 peptide has a flat (or “featureless”) conformation at the main chain without any longer side chains of the residues protruding out of the peptide binding groove for TCR docking (12, 21). Being able to determine the binding features of a peptide to the MHC and describe the peptide–MHC topology can help us understand the immunodominance of a given peptide and demonstrate the peptide presentation strategy of the host (22–24).

The nuclear localization signal (NLS) and nuclear export signal (NES) can shuttle viral proteins between the nucleus and the cytoplasm in infected cells, which is of importance for the virus replication cycle (5, 25–27). Classic NESs confer CRM1-dependent nuclear export, for which these signal sequences are characterized by leucine and other hydrophobic residues φ-X2-3-φ-X2-3-φ-X-φ, where φ is L, V, I, F, or M, and X is any amino acid (28, 29). Prototypical classic NESs included the NES (LALKLAGLDI) of the protein kinase inhibitors and the HIV-1 Rev NES (LPPLERLTL) (29). Previously, a leucine-rich NES (⁵⁹ILGFVFTLTV⁶⁸) was reported in the M1 protein of IAV, and the nuclear export of M1 contributes to the replication of IAV (27). Colocalization of the NES and the immunodominant CD8⁺ T cell epitope AM_58–66GL9 in M1 explains why escape mutations in this epitope were rarely found in natural IAVs (30). In previous studies, two CRM1-dependent NESs (3–14 aa and 133–124 aa) and one bipartite NLS (76–94 aa) in the IBV M1 protein have been identified and found to participate in the regulation of cytoplasmic-nuclear transport of BM1 (25). However, it is yet to be determined whether an NES exists within IBV as the corresponding region of the leucine-rich NES (⁵⁹ILGFVFTLTV⁶⁸) in the M1 protein of IAV.

In this study, we evaluated the T cell immunogenicity of the M1 protein of IBV in the same region as the immunodominant T cell epitope AM_58–66GL9 in the IAV M1 protein. We identified an HLA-B*1501–restricted epitope BM_58–66AF9, but no HLA-A*02–restricted epitopes. The structure of the HLA-B*1501/BM_58–66AF9 complex showed that the BM_58–66AF9 peptide has a flat and featureless conformation at the main chain, which is similar to AM_58–66GL9 presented by HLA-A*0201. We determined that the sequence surrounding the HLA-B*1501–restricted epitope BM_58–66AF9 does not contain an NES, although an NES exists in the corresponding region of IAV.

Eighteen healthy donors were recruited for this research from Beijing, China, including eight HLA-B*1501–restricted donors (Supplemental Table I). None of the donors had symptoms of influenza virus infection during the sampling period. The study was approved by the Ethics Review Committee of the National Institute for Viral Disease Control and Prevention, Chinese Center for Disease Control and Prevention. The study was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all of the donors. The HLA subtypes of all donors at the A and B loci were determined using LABType SSO (One Lambda) with whole blood. Some experiments were not performed with the samples from all HLA-B*1501–restricted donors because of insufficient PBMCs of donors. The donor numbers in each experiment are indicated in the figure legends.

To identify the potential HLA-A*0201– or HLA-B*1501–binding peptides, we applied a computer-based program via access through a Web site of MHC class I–binding predictions (http://tools.iedb.org/mhci/) using the M1 sequence of IBV [B/Beijing-Chaoyang/12977/2017 Dec (Yamagata)]. Fourteen peptides were synthesized (Table I), and the purity was determined as 95% by analytical HPLC and mass spectrometry (SciLight Biotechnology). The peptides were dissolved in DMSO before use.

PBMCs were isolated from the venous blood of donors using the Lymphocyte Separation Tube (Dakewe). PBMCs from donors were incubated with peptide pools at a final concentration of 2 μg/ml for each peptide or 10 μg/ml of a single peptide in RPMI 1640 (Hyclone) containing 10% FBS (Life Technologies) at 37°C with 5% CO₂ at a density of 2.5 × 10⁶ cells/ml in a 24-well plate. On day 3, 20 U/ml recombinant human IL-2 (PeproTech) was added to the medium. Half of the medium was changed on days 4 and 7 with supplementation by recombinant human IL-2. On day 9, cells were harvested and tested for the presence of peptide-specific CD8⁺ T cells.

The IBV-specific response of T cells induced by peptides was measured using the IFN-γ–secreting ELISpot set (BD Biosciences), as described previously (31, 32). In brief, the ELISpot plate membrane was preincubated, washed, and blocked (17, 22). After cultured with peptide pools (2 μg/ml for each single peptide) or a single peptide (10 μg/ml) for 9 d, PBMCs from donors were incubated in microwells (1 × 10⁵/well), along with stimulating peptides (20 μg/ml) or PMA as a positive control of nonspecific stimulation for 24 h at 37°C with 5% CO₂, and cells incubated without a stimulator were employed as a negative control. Then, the cells were removed and biotinylated detection Abs, streptavidin-HRP conjugate, and substrate 3-amino-9-ethylcarbazole were added to the plate. When the colored spots were intense enough to be visually observed, the development was stopped by thorough rinsing with demineralized water. The results were analyzed using an automatic ELISpot reader (CTL). The response was considered positive when the number of spot-forming cells (SFCs) in the target well was >20 and twice the number in the negative control well without a stimulus (33).

HLA-B*1501 or HLA-A*0201 H chain and β₂-microglobulin (β₂m) were overexpressed as inclusion bodies in Escherichia coli and subsequently in vitro refolded and assembled in the presence of peptide or without peptide as the negative control (22, 33). Generally, 250 ml of refolding buffer was used, and the molar ratio of H chain to β₂m to peptide was 1:1:2. After 12 h of slow stirring at 4°C for protein refolding, the HLA–peptide complex buffer was concentrated and analyzed by Superdex 200 20/50 GL gel-filtration chromatography (GE Healthcare).

The capacity for peptide binding to HLA-A*0201 was evaluated by a T2 cell–based MHC stabilization assay, as described previously (22, 34). In brief, T2 cells were incubated at 26°C for 16 h, then cultured with peptides (50 μg/ml) and supplemented with 1 mM/l human β₂m (Sigma) for 18 h at 37°C with 5% CO₂. Then, the cells were stained with FITC-labeled anti–HLA-A2 mAb. The mean fluorescence intensity was measured using a FACSCalibur (BD Biosciences).

After in vitro culture, T cell lines were stimulated with candidate peptide (10 µg/ml) for 2 h and then incubated with GolgiStop/monensin (BD Biosciences) for an additional 4 h at 37°C in 5% CO₂. Unstimulated or PMA-stimulated cells were included as negative and positive controls, respectively. Then, the cells were harvested and stained with anti–CD3-allophycocyanin and anti–CD8-PE surface markers. Subsequently, the cells were fixed with BD fix/perm buffer on ice for 20 min and then stained with the intracellular markers anti–IFN-γ PE-Cy7 (BD Biosciences), anti–IL-2-FITC (BD Biosciences), and anti–TNF-α-PerCp-5.5 (BD Biosciences). The fluorescent lymphocytes were gated on a FACSCalibur flow cytometer (BD Biosciences). The results were analyzed with FlowJo software (BD Biosciences).

HLA-B*1501–peptide or HLA-A*0201–peptide tetramers were produced as described previously (17). In brief, expression of the HLA-B*1501 H chain was limited to the extracellular domain (aa residues 1–274) and modified by the addition of a substrate sequence for the biotinylating enzyme BirA at the C terminus of the α3 domain. In vitro–renatured HLA/peptide complexes were purified and biotinylated by incubation with the biotin protein ligase BirA (recombinant expressed), another d-biotin, and ATP (Avidity). The biotinylated HLA was purified again through gel filtration before multimerization, achieved by mixing with streptavidin conjugated with PE or allophycocyanin (Sigma-Aldrich, St. Louis, MO). PBMCs from the donors were cultured with the peptide for 9 d, then stained with HLA-B*1501/BM_58–66AF9-PE, HLA-B*1501/BM_58–66AF9-allophycocyanin, HLA-A*0201/AM_58–66GL9-PE, or HLA-A*0201/BM_58–66AF9-allophycocyanin tetramers (0.05 µg/ml) and V500-conjugated anti-CD8 Ab. The cells were analyzed by FACSCalibur flow cytometry (BD Biosciences) after staining.

The HLA-B*1501/peptide complex was produced by the gradual-dilution refolding method, as described earlier (35). Subsequently, the remaining soluble portion of the complex was concentrated and purified by Superdex 200 20/50 GL gel filtration chromatography (GE Healthcare). HLA-B*1501/BM_58–66AF9 crystals were grown at 4°C and obtained in 0.1 M MES monohydrate (pH 6.0) and 20% w/v polyethylene glycol monomethyl ether 2000. For cryoprotection, the crystals were transferred to reservoir solutions containing 20% glycerol and were then flash frozen in a stream of gaseous nitrogen at 100 K. Diffraction data for the crystals were collected on Beamline 19U of the Shanghai Synchrotron Radiation Facility (Shanghai, China) and were processed using HKL2000 software.

The structure of HLA-B*1501/BM_58–66AF9 was determined by molecular replacement by CCP4 software. The HLA-B*1501 crystal structure (Protein Data Bank [PDB] code: 5TXS) was used as the search model. Extensive model building was performed manually using Wincoot, and constraint refinement and stereochemical quality assessment were performed using Phenix (23). Structural figures were generated using PyMOL (http://www.pymol.org/).

Atomic coordinates and structure factors have been deposited in the PDB (https://www.rcsb.org/structure/7XF3) under accession number 7XF3 for HLA-B*1501 in complex with BM_58–66AF9. They are publicly available as of the date of publication of this article.

A series of truncated IBV M1 protein sequences (B/Lee/40 strain of IBV, NP_056664.1) containing previously determined NES and/or NLS motifs or currently predicted NES were synthesized (25) (Genewiz) and then inserted between the XhoI and BamHI sites, respectively, into the pEGFP-C1 vector. The amino acid sequence of the M1 protein sequence of IBV [B/Beijing-Chaoyang/12977/2017 Dec (Yamagata)] at position 55–70 aa is identical with that of the M1 protein (B/Lee/40 strain of IBV). For transfection with plasmids, HEK293T cells were seeded into 24-well plates with a cover slide in each well at 50% confluency and then cultured within 24 h before being transfected with plasmids using transfection reagent Lipofectamine 3000 (Invitrogen). After transfection for 20 h, the cells were stained with Hoechst 33258, and then the intracellular location of the truncated M1–EGFP fusion protein was recorded using the laser scanning confocal microscopes (Leica SP8).

Two-tailed paired or unpaired t tests were used to compare data. Asterisks in each figure indicate statistical significance (*p < 0.05 and **p < 0.01). Analyses were performed with GraphPad Prism 8 software (GraphPad Software) and OriginPro software (OriginPro Lab Corporation).

Peptide AM_58–66GL9 (GILGFVFTL) is a dominant T cell epitope in the M1 protein of IAV (12, 14, 17). It plays an important role in the immune response to IAV infection among HLA-A*0201–restricted patients. To investigate the immunogenicity of this region covering the AM_58–66GL9–located position in the M1 protein of IBV, we first performed amino acid alignment of the M1 protein sequence of IBV [B/Beijing-Chaoyang/12977/2017 Dec (Yamagata)] with that of the M1 protein of IAV [A/California/7/2009(H1N1)]. According to the alignment, we designed a long peptide (BM_55–70IK16 IQKALIGASICFLKPK) with three or four additional amino acids at each end of BM_58–67AL10 peptide with HLA-A*0201–restricted characteristics (Fig. 1A). PBMCs from five donors were cultured with BM_55–70IK16 peptide for 9 d, and then ELISpot assays were performed to detect Ag-specific T cell responses against this long peptide. The results indicated that BM_55–70IK16 peptide could stimulate PBMCs to produce IFN-γ in some donors, but not all HLA-A*02 donors were able to induce T cell immune responses (Fig. 1B).

Using this long peptide BM_55–70IK16 as a target sequence, 14 truncated short candidates were synthesized and predicted the potential binding to HLA-A*0201 (Table I, Fig. 1A). Among these peptides, a predicted peptide (BM_58–67AL10 ALIGASICFL) that shared the characteristics of an HLA-A*0201–restricted epitope was found in the IBV M1 protein (Fig. 1A). Then, we investigated whether the 14 candidate peptides could induce T cell immune responses in the HLA-A*02–restricted donors. The candidate peptides were mixed at equal concentrations to generate a peptide pool, and PBMCs from 10 donors were cultured with the peptide pool for 9 d; then ELISpot assays were performed to assess CTL responses against the 14 candidate peptides. The 14 candidate epitopes did not stimulate PBMCs to produce IFN-γ in five of the seven HLA-A*02 donors (Fig. 1C). However, candidate peptides BM_57–66KF10, BM_58–66AF9, or BM_58–68AK11 did stimulate an immune response in all four HLA-B*1501–restricted donors, including the two donors with HLA-A*02 at the HLA-A loci (Fig. 1C, Supplemental Table I). These results indicated that HLA-B*1501–restricted epitopes may exist in the candidate peptides. Moreover, the number of SFCs stimulated by BM_57–66KF10, BM_58–66AF9, or BM_58–68AK11 in all of the HLA-B*1501 donors was more than twice the number in the negative control well without stimulator (Fig. 1C, left). Furthermore, the number of SFCs stimulated by BM_57–66KF10, BM_58–66AF9, or BM_58–68AK11 was >20 in the D09 donor, compared with <10 in the negative control (Fig. 1C, right). However, although the number of SFCs stimulated by BM_55–64II10 and BM_57–67KL11 in the D04 donor was more than twice the number in the negative control, the number in the negative control well was <10 and the number of SFCs stimulated with BM_55–64IL10 and BM_57–67KL11 in the D04 donor was also <20 (Fig. 1C, right). These data showed no T cell immune responses can be induced by the candidates among HLA-A*02 donors without HLA-B*1501 dependence, and candidate peptides BM_57–66KF10, BM_58–66AF9, or BM_58–68AK11 could be recognized as antigenic T cell epitopes within HLA-B*1501–restricted donors.

Then, we also evaluated the potential HLA-B*1501 binding of these truncated peptides derived from the long peptide BM_55–70IK16 based on software prediction (Table I). Peptides BM_57–66KF10 and BM_58–66AF9 showed high binding indexes to HLA-B*1501. Thus, BM_57–66KF10, BM_58–66AF9, and BM_58–68AK11 are candidate HLA-B*1501–restricted T cell epitopes of matrix protein 1 of IBV for the following investigations. Although BM_58–68AK11 does not have the typical anchor residues for HLA-B*1501–binding peptides, the T cell function of this peptide may depend on the truncated peptide BM_58–66AF9.

Then, we explored the refolding experiment to investigate the potential HLA-B*1501 restriction of the candidate peptides BM_57–66KF10, BM_58–66AF9, and BM_58–68AK11, which could stimulate immune responses in all four HLA-B*1501–restricted donors. The refolding of HLA-B*1501 H chain and β₂m in the presence of the candidate peptide revealed that BM_58–66AF9 peptide could form the stable MHC class I complex, whereas the refolding efficiency of BM_57–66KF10 was relatively low, and BM_58–68AK11 could not bind to HLA-B1501 (Fig. 2A–C). Thus, BM_58–68AK11 was not further investigated in the following experiments. As the positive control, AM_58–66GL9 possesses high capacity for binding to HLA-A*0201 molecules (Fig. 2D), and the refolding without any peptide as the negative control (No pep) showed no peak for the HLA I monomer. We also found that the candidate peptide BM_58–66AF9 could form a complex not only with HLA-B*1501 but also with HLA-A*0201 in vitro (Fig. 2E). Meanwhile, BM_58–67AL10, a predicted peptide with HLA-A*0201–restricted characteristic, could not bind to HLA-A2, as determined by the T2-binding assay (Fig. 2F).

Next, we detected the BM_55–70IK16–, BM_57–66KF10–, and BM_58–66AF9–specific CD8⁺ T cell responses after stimulation of the donor PBMCs with the long peptide (BM_55–70IK16) using the intracellular cytokine staining (ICS) assay (Fig. 3A, Supplemental Fig. 1A). First, PBMCs from HLA-B*1501⁺ or HLA-A*02 donors were cultured with the long peptide (BM_55–70IK16) for 9 d; then cells were stimulated with the BM_55–70IK16, BM_57–66KF10, or BM_58–66AF9 peptides; and the percentages of IFN-γ–, IL-2–, and TNF-α–secreting cells among the CD8⁺ T cells were determined by flow cytometry. IFN-γ expression was clearly increased in PBMCs from the HLA-B*1501⁺ donor (D06) stimulated with BM_55–70IK16, BM_57–66KF10, and BM_58–66AF9 peptide, respectively, but IFN-γ expression was not elevated in cells from the HLA-A*02⁺/B*1501⁻ donor (D04) (Fig. 3A, Supplemental Fig. 1A). The high level of BM_57–66KF10– and BM_58–66AF9–elicited IFN-γ–secreting cells among the CD8⁺ T cells of the HLA-B*1501⁺ donor demonstrated the role of both peptides in stimulating an immune response. The BM_55–70IK16, BM_57–66KF10, and BM_58–66AF9 peptides did not induce the expression of IL-2 or TNF-α in the CD8⁺ T cells of the HLA-B*1501⁺ donor (D06) or the HLA-A*02⁺/B*1501⁻ donor (D04) (Fig. 3B).

In addition, PBMCs from HLA-B*1501⁺ or HLA-A*02⁺/B*1501⁻ donors were cultured with BM_57–66KF10 or BM_58–66AF9 peptides for 9 d and then stimulated with BM_57–66KF10 and BM_58–66AF9 peptide, respectively, and the percentage of IFN-γ–secreting cells among CD8⁺ T cells was detected by flow cytometry. The results showed that both BM_58–66AF9 and BM_57–66KF10 could induce the expression of IFN-γ in the CD8⁺ T cells of the HLA-B*1501⁺ donors after stimulation with the BM_57–66KF10 and BM_58–66AF9 peptides, respectively (Fig. 3C–F, Supplemental Fig. 1B, 1C). The percentage of BM_57–66KF10–induced IFN-γ increased significantly in PBMCs from HLA-B*1501⁺ donors compared with the mock control group (p < 0.05) (Fig. 3D). The percentage of BM_58–66AF9–induced IFN-γ also increased in PBMCs from HLA-B*1501⁺ donors compared with the mock control group, although without statistical significance (p = 0.051) (Fig. 3F). By contrast, BM_58–66AF9 could not stimulate the expression of IFN-γ in the CD8⁺ T cells of the HLA-A*02⁺/B*1501⁻ donors after stimulation with BM_58–66AF9 peptides (Fig. 3E, 3F, Supplemental Fig. 1C). The percentage of BM_58–66AF9–induced IFN-γ was lower in PBMCs from HLA-A*02⁺/B*1501⁻ donors compared with HLA-B*1501⁺ donors (p = 0.052) (Fig. 3F).

We also detected the percentage of candidate peptide-specific CD8⁺ T cells circulating in the peripheral blood of HLA-B*1501⁺ and HLA-A*02⁺/B*1501⁻ donors using ex vivo staining of the tetramers. The HLA-B*1501/BM_57–66KF10 and HLA-B*1501/BM_58–66AF9 tetramers were prepared and analyzed to confirm which candidate peptide was an immunodominant epitope of IBV M1 protein in the HLA-B*1501–restricted population. First, PBMCs from HLA-B*1501⁺ donors and HLA-A*02⁺/B*1501⁻ donors were stained with HLA-B*1501/BM_58–66AF9 or HLA-B*1501/BM_57–66KF10 tetramer after a 9-d incubation with the long peptide (BM_55–70IK16). The results showed that BM_58–66AF9-specific CD8⁺ T cells from PBMCs of the HLA-B*1501⁺ donors (D06 and D08) displayed percentages of 9.94% and 8.03%, respectively. However, the percentages of BM_57–66KF10–specific CD8⁺ T cells from HLA-B*1501⁺ donors (D06 and D08) were only 0.05% and 0.09%, respectively. The percentages of BM_58–66AF9– and BM_57–66KF10–specific CD8⁺ T cells in PBMCs from HLA-A*02⁺/B*1501⁻ donors and other donors (A02⁻/B*1501⁻) were all low as the negative controls (Fig. 4A).

Next, PBMCs from HLA-B*1501⁺ and HLA-A*02⁺/B*1501⁻ donors were stained with HLA-B*1501/BM_57–66KF10 or HLA-B*1501/BM_58–66AF9 tetramer after a 9-d incubation with BM_57–66KF10 and BM_58–66AF9, respectively. As observed after long peptide incubation, the frequencies of BM_57–66KF10–specific CD8⁺ T cells in PBMCs from HLA-B*1501⁺ and HLA-A*02⁺/B*1501⁻ donors were low as the negative controls (Fig. 4B), whereas the frequencies of BM_58–66AF9–specific CD8⁺ T cells were obviously increased in PBMCs from the HLA-B*1501⁺ donor compared with the mock control group (p < 0.05), the irrelevant control group (H2-K^d-HBV) (p < 0.05), and the HLA-A*02⁺/B*1501⁻ donor group (p < 0.01) (Fig. 4C, 4D, Supplemental Fig. 2A).

We also investigated whether IAV- and IBV-specific CD8⁺ T cells could cross-recognize AM_58-66GL9 and BM_58-66AF9 from IAV and IBV. PBMCs from the donors were cultured with the AM_58-66GL9 and BM_58-66AF9 for 9 d, respectively, then stained with HLA-A*0201/AM_58-66GL9-PE and HLA-B*1501/BM_58-66AF9-allophycocyanin tetramers. The results showed that HLA-B*1501/BM_58–66AF9 was able to recognize the specific T cells from only the HLA-B*1501⁺ donors, and HLA-A*0201/AM_58–66GL9 could bind the specific T cells from the HLA-A*0201⁺ donors (Fig. 4E, Supplemental Fig. 2B). These results indicate that the BM_58–66AF9 peptide from the M1 protein of IBV is a dominant HLA-B*1501–restricted T cell epitope, and there was no CD8⁺ T cell cross-reactivity between peptides AM_58-66GL9 and BM_58-66AF9.

A previous study showed that the immunodominant epitope AM_58–66GL9 of AIV has a flat (or “featureless”) conformation within the structure of the HLA-A*0201/AM_58–66GL9 complex. To elucidate the structural conformation of the newly identified peptide BM_58–66AF9, we determined the crystal structure of the HLA-B*1501/BM_58–66AF9 complex to a resolution of 1.9 Å (Table II). An overview of the HLA-B*1501/BM_58–66AF9 complex presents a typical conformation of HLA class I molecules (Supplemental Fig. 3). The unambiguous electron density of the peptide ligand BM_58–66AF9 clearly shows the main-chain conformation of the peptide and the orientations of the residue side chains (Fig. 5A). In the HLA-B*1501 structure, peptide positions 2 and 9 are primary anchors, with P2-Leu deeply buried in the B pocket and P9-Phe in the F pocket (Fig. 5A). Residues P3-Ile and P6-Ser also protrude the side chains into the peptide binding groove, as secondary anchors for the peptide. The side chains of residues P5-Ala, P7-Ile, and P8-Cys in the middle part of peptide BM_58–66AF9 point to the α1 or α2 helices of HLA-B*1501 with their Cα solvent exposed, together with P4-Gly. Peptide BM_58–66AF9 in the HLA-B*1501/BM_58–66AF9 complex displays a similar conformation to AIV AM_58–66GL9 in the HLA-A*0201/AM_58–66GL9 structure (Fig. 5B). The backbone of peptide BM_58–66AF9 from P4 to P8 is solvent exposed as for peptide AM_58–66GL9 (Fig. 5C–F). This indicated that both peptides possess a flat and featureless conformation in the main chain.

A leucine-rich NES (⁵⁹ILGFVFTLTV⁶⁸) almost completely overlaps the epitope AM_58–66GL9 (GILGFVFTL) in the M1 protein of IAV. Meanwhile, within the corresponding position in IBV, peptide BM_58–67AL10(⁵⁸ALIGASICFL⁶⁷) partially meets the characteristics of the NES motif and overlaps with the identified HLA-B*1501–restricted epitope BM_58–66AF9. Thus, we investigated whether an NES exists at this position. Previously, two leucine-rich NESs ³LFGDTIAYLLSL¹⁴ and ¹²⁴LLYCLMVMYL¹³³ and one bipartite NLS ⁷⁶RRFITEPLSGMGTTATKKK⁹⁴ were identified in the amino acid sequence of BM1 (Fig. 6A). According to the locations of the currently predicted NES and the previously identified NESs and NLS in the amino acid sequence of BM1, plasmids encoding various EGFP-tagged truncations of BM1 were constructed (Fig. 6B).

The nucleus was stained with Hoechst 33258, and then the intracellular distribution of the EGFP-tagged proteins was determined by confocal fluorescence microscopy (Fig. 6C). The results showed that EGFP was distributed across both the cytoplasm and nucleus when only EGFP was expressed, but EGFP-AM1 59–68 and EGFP-BM1 106–133, which contained the one identified NES, were predominantly located in the cytoplasm, whereas EGFP-BM1 71–123, which contained one identified NLS but none of the predicted NESs, was predominantly located in the nucleus. These results were consistent with previous research (25). However, EGFP-BM1 15–123, which contained one of the identified NLSs and the peptide BM_58–67AL10 as a predicted NES, were still predominantly located in the nucleus, which was consistent with previous research (25). Furthermore, the nuclear export activities of peptide BM_58–67AL10 were tested by fusing the EGFP to the truncated BM1 15–75 peptides, which contain only predicted signal peptides. The result suggested that no NES existed in the BM1 15–75. This was also confirmed by the even distribution of EGFP in the nucleus and cytoplasm, detected by the constructions of EGFP-BM1 15–75, EGFP-BM1 55–70, and EGFP-BM1 58–67, respectively (Fig. 6B, 6C). Based on these results, it was concluded that peptide BM_58–67AL10 was not an NES, and the sequence around the HLA-B*1501–restricted epitope does not overlap with an NES.

Adaptive humoral and cellular responses can provide immune protection against influenza (36). Influenza virus–specific Abs mainly target the variable head domain of the surface glycoprotein HA, but the constant antigenic drift of HA results in poor cross-protection to different virus strains or subtypes. T cell epitopes could engineer immune responses by the presentation of MHC class I or MHC class II molecules (37, 38). Recent studies showed that the PB1_413–421 peptide as a universal HLA-A*0201 peptide and the HLA-A24–restricted PB2_550–558 peptide have the ability to confer cross-reactivity across IAV and IBV (39, 40). These peptides providing cross-protection across seasonal IAVs and/or IBVs may act as the immunogen in the development of universal vaccines based on peptides (39–43).

As a highly conserved immunodominant MHC class I–restricted epitope, peptide AM1_58–66GL9 derived from IAV is used in peptide-based vaccine development to induce influenza-specific CTLs in the HLA-A2 population (18, 44). In this study, we noticed that a potential peptide with HLA-B*0201–restricted characteristic, BM_58–67AL10 ALIGASICFL, displayed the characteristics of HLA-A*0201 binding at a similar position to AM_58–66GL9 in the IBV M1 protein. We therefore carried out screening and obtained three candidate peptides of IBV by ELISpot, namely, BM_57–66KF10 (KALIGASICF), BM_58–66AF9 (ALIGASICF), and BM_58–68AK11 (ALIGASICFLK). These three candidate peptides could induce IFN-γ in the PBMCs of an HLA-B*1501–restricted population, but not an HLA-A*02–restricted population, although these results were inconsistent with our initial hypothesis that there was an HLA-A*02–restricted GL9-like epitope around a similar position to AM_58–66GL9 in M1 of IBV. However, we identified a dominant HLA-B*1501–restricted epitope in this position by its binding capacity in vitro, ex vivo staining of the tetramers, and the crystal structure of the HLA-B*1501/BM_58–66AF9 complex. Previous studies determined that HLA-B*1501 prefers hydrophobic residues at the B pocket (45–47) and the large, bulky aromatic residues Tyr (Y) and Phe (F) and, in some cases, Trp (W) in the F pocket (46, 48, 49). The immunodominant peptide BM_58–66AF9 (ALIGASICF) identified in this study possesses the characteristic feature of HLA-B*1501–restricted peptide with Leu (L) in position 2 and Phe (F) at position 9.

Although peptide BM_58–68AK11 could induce IFN-γ in the PBMCs of an HLA-B*1501–restricted population, BM_58–68AK11 peptide could not assist the renature of HLA-B*1501 in vitro. There is a K at the C terminus of BM_58–68AK11 peptide, as well as two more amino acids at the C terminus than in the functional epitope BM_58–66AF9, which leads to failure of BM_58–68AK11 refolding with HLA-B*1501 in vitro. However, peptide BM_58–68AK11 may be cleaved into BM_58–66AF9 in cells to stimulate CTLs to produce IFN-γ, or the minute amount of peptide BM_58–68AK11 loaded on the HLA-B*1501 in vivo can stimulate the function of T cells. Moreover, we also found that another candidate peptide BM_57–66KF10 (KALIGASICF) possesses only one more K at the N terminus compared with BM_58–66AF9. Because the A and B pockets of HLA-B*1501 display high variation in the N-terminal amino acids of the HLA-B*1501–restricted epitopes, BM_57–66KF10 peptide could refold with HLA-B*1501 in vitro, but this complex cannot perform a correct conformation in vitro in the tetramers and cannot bind to BM_58–66AF9–specific TCRs. Our previous functional and structural studies on the HLA-A*2402–restricted T cell epitopes derived from 2009 pandemic H1N1 virus showed that both peptides P27 (FYRYGFVANF) and P28 (RYGFVANF) could induce robust T cell responses among HLA-A*2402⁺ healthy adults (33). Although 8-mer peptide P27 has two more residues longer than P38, the second residue is still the preferred anchor Y for HLA-A*2402–binding peptides. Thus, both peptides P27 and P28 could act as independent immunodominant epitopes with different conformations, as shown in the structures of HLA-A*2402.

Similar to the conformation of peptide AM_58–66GL9 in the HLA-A*0201/AM_58–66GL9, the BM_58–66AF9 peptide also has a flat, featureless conformation. Previous studies indicated that GL9-specific public TCRs carrying the Vβ17⁺ broadly exist among populations with different ethnicities (17). In the TCR recognition of the AM_58–66GL9, the featureless conformation of AM_58–66GL9 leads to a direct contact of the TCR to the main chain of AM_58–66GL9 (12–17). Thus, the featureless conformation of peptide BM_58–66AF9 may also indicate a similar TCR-docking mode, which needs further investigation. Furthermore, it is known that low levels of variation exist at the B and F pockets of HLA-A; however, it was reported that the B pocket of HLA-B showed a particularly high level of variation between supertypes (18). HLA-A*0201 and HLA-B*1501 prefer aliphatic residues such as Leu in the B pocket (47, 48, 50). Although the F pockets of HLA-A*0201 and HLA-B*1501 are predominantly hydrophobic, the F pocket of HLA-A*0201 is narrower and shallower compared with HLA-B*1501 (48, 51). Thus, the peptide P9 anchor of HLA-A*0201 prefers smaller, aliphatic amino acids such as Val and Leu (47, 50), whereas HLA-B*1501 has a strong preference for Tyr and Phe (F) at P9 (48). This explains why the BM_58–66AF9 peptide could refold with HLA-A*02 but could not induce an immune response in the HLA-A*02–restricted population.

Protein nuclear transport plays a critical role in facilitating viral replication and evading the host antiviral response by subversion of the nucleocytoplasmic transport systems during viral infection (52–54). Some viral proteins contain both NLSs and NESs to mediate nuclear transport (53). As we know, IAV replicates RNA within the nucleus of the host cells (11). The M1 protein of IAV was found to enter the nucleus and be indispensable for the viral ribonucleoproteins nuclear export, and the M1 protein nuclear export was dependent on its NES and critical for IAV replication (27, 55). Also, the N protein–derived NLS motifs promoting viral ribonucleoproteins nuclear import could cause immune evasion of IAV (56). Similarly, West Nile virus NS5 is a definitive example of the importance of nucleocytoplasmic shuttling for productive infection, because mutation of its NES impairs viral replication (57). In general, this shuttling phenomenon is likely important to the completion of each respective viral life cycle (53).

In addition, an almost overlapping sequence between the NES (59–68 aa) and the CD8⁺ T cell epitope, AM_58–66GL9, was found in M1 protein of IAV (27). Actually, this phenomenon of the overlapping of NES sequences and T cell epitopes was also observed in other viruses, such as LLYKISLTT (194–202 aa) in NES (188–202 aa) of HTLV-1 Tax protein (58), TLTCGFADL (125–133 aa) in NES (109–133 aa) HCV core protein (59), and MLIDLGLDL (7–15 aa) in NES (5–15 aa) HSV-1 ICP27 protein (60). The colocalization of the immunodominant AM_58–66GL9 epitope and the NES as a conservative sequence in M1 of IAV limited the immune escape of this peptide (27, 30). Although immune escape after acute infection of IAV and IBV may not occur in individuals because of the short duration and the relatively mild course in the vast majority of infections for most current strains of influenza virus, immune escape can also occur after circulation among the populations (61, 62). The selection of the conserved regions as viral NES adjacent sequences as T cell epitopes may be one of the antiviral strategies of the hosts (63) and better for the patient to eradicate the viruses. In this study, we found that the peptide of BM_55–70IK16 (IQKALIGASICFLKPK) is close to the characteristics of NES sequences whereby φ-X2-3-φ-X2-3-φ-X-φ, within which φ is L, V, I, F, or M, and X is any amino acid (28, 29). So, we performed a comprehensive investigation to investigate whether an NES exists in proximity to the HLA-B*1501–restricted epitope. Our results showed the absence of an NES in the BM_55–70IK16 peptide, which was consistent with previous findings (25). Actually, peptide BM_55–70IK16 does not exactly load the NES feature and has four residues, “IGAS,” between the second and the third hydrophobic residues (L and I).

In conclusion, in the similar location as the immunodominant peptide AM_58–66GL9 of IAV, we determined an immunogenic region within the M1 protein of IBV. A series of T cell immunity assays showed that the HLA-B*1501–restricted epitope BM_58–66AF9 (ALIGASICF) within this region could induce immunodominant T cell responses. Interestingly, the crystal structure of the HLA-B*1501/BM_58–66AF9 complex revealed that the BM_58–66AF9 peptide is a flat, featureless peptide as AM_58–66GL9 presented by HLA-A*0201. The parallel existence of immunodominant T cell epitopes within the corresponding position of IAV and IBV may be the result of the long-term interaction of the influenza viruses and the human host. But the absence of the NES within this region of the IBV may indicate different balance for viral life cycle and immune escape between IAV and IBV. Our study provides further insights into influenza virus–specific T cell immunity and may shed light on the development of universal vaccines for influenza viruses.

The authors have no financial conflicts of interest.

We appreciate the great help of Prof. Jianxun Qi, who participated in the collection of the diffraction data for the crystals. We thank Xiaolan Zhang at the Institute of Microbiology, Chinese Academy of Sciences, for her technical support in confocal microscopy analysis.