Cross-Allele Cytotoxic T Lymphocyte Responses against 2009 Pandemic H1N1 Influenza A Virus among HLA-A24 and HLA-A3 Supertype-Positive Individuals

Study subjects and ethics statement.

Two cohorts of subjects were recruited for the research, including a group of 19 subjects from Beijing, China, and a group of 77 subjects from Jiangsu Province, China. The 19 subjects (see Table S1 in the supplemental material) from Beijing consisted of 12 males and 7 females, with an age range of 25 to 48 years, and the 77-subject cohort (see Table S2 in the supplemental material) included 23 females and 54 males (ages, 21 to 76 years). Fifty-three of the subjects were inoculated with the 2009 pH1N1 split-virus vaccine (Hualan Bio, China) 4 months before the collection of a blood sample. None of the subjects had symptoms of influenza virus infection during the sampling period. Written informed consent was obtained from all of the subjects, and the study was approved by the Ethics Review Committee of the Institute of Microbiology, Chinese Academy of Sciences. The study was conducted in accordance with the principles of the Declaration of Helsinki and the standards of good clinical practice (as defined by the International Conference on Harmonization). Venous blood was collected from the subjects in the hospital during the 2009 pH1N1 pandemic. The HLA subtypes at the A and B loci were determined using LABType SSO (One Lambda, Beijing, China) with whole blood from the subjects.

Design and synthesis of HLA-A24-restricted peptide pool.

Proteins derived from the 2009 pH1N1 virus (i.e., HA, NA, M1, M2, NP, PB1, PB2, PA, NS1, and NS2) were chosen for the prediction of potential CD8⁺ T cell-specific epitope peptides with an HLA-A*2402 binding motif. The sequences of the antigens were based on the H1N1 A/California/04/2009 strain. A computer-based program was applied to predict 9-mer and 10-mer peptides through the BioInformatics and Molecular Analysis Section (BIMAS) HLA Peptide Binding Predictions website (56). The 8-mer and 11-mer peptides whose scores of binding to HLA-A*2402 are not available in the computer-based prediction were manually selected according to the typical HLA-A*2402 peptide motif (an aromatic residue at position P2 and an aliphatic or aromatic residue at C-terminal position Pc) (Table 1). The peptides were synthesized and the purity was determined to >90% by high-performance liquid chromatography (HPLC) and mass spectrometry (Scilight-Peptide, Inc., Beijing, China). Peptides were dissolved in dimethyl sulfoxide (DMSO) at 20 mg/ml and diluted in RPMI 1640 medium to a final concentration of 10 μg/ml before being used individually. The predicted peptides were also mixed at a final concentration of 2 μg/ml for each peptide and served as the HLA-A*2402 peptide pool.

As a positive control for binding affinity assessment, two previously determined HLA-A*2402-restricted epitopes, Nef138-10 (derived from the HIV Nef antigen) (34) and N1 (derived from the severe acute respiratory syndrome coronavirus [SARS-CoV] N protein) (44), were also synthesized. The negative control was the HLA-A*1101-restricted peptide Beta309 (RYLTVAAVFR) (67).

T2-A24 cell binding assay.

The binding affinities of the peptides to the HLA-A*2402 molecule were assessed using T2-A24 cells as previously described (44). Briefly, after incubation for 16 h at 26°C, cells were harvested and washed twice with serum-free RPMI 1640 and resuspended in 1 ml culture medium. Peptide (50 μg/ml) and beta 2 microglobulin (β2m) (20 μg/ml) were added to the culture medium and incubated at 26°C for 3 h, followed by 37°C for 3 h. After washing with phosphate-buffered saline (PBS), the cells were incubated with anti-HLA-class I antibody w6/32 at 4°C for 30 min and subsequently incubated with the fluorescein isothiocyanate (FITC)-conjugated F(ab′)2 fragment of anti-mouse immunoglobulin (Dako). The cells were then analyzed using a FACSCalibur (Becton Dickinson). Peptides with a fluorescence index (FI [FI = mean FITC fluorescence of the given peptide/mean FITC fluorescence without the peptide]) of >1.5 were regarded as high-affinity candidate peptides.

Generation of influenza virus-specific T cell lines in vitro.

Peripheral blood mononuclear cells (PBMCs) were isolated from the whole blood of donors by density gradient centrifugation using Ficoll-Hypaque (TBD Science) and washed twice in RPMI 1640 medium with 10% fetal bovine serum (FBS) (Gibco). Freshly isolated PBMCs in RPMI 1640 supplemented with 10% fetal calf serum (FCS) were incubated with the peptide pool at 37°C in 5% CO₂ at a density of 2 × 10⁶ cells/ml in a 24-well culture plate. Recombinant human interleukin 2 (rIL-2) (20 IU/ml) was added to the culture medium on the 3rd day. Half of the medium was replaced with fresh medium supplemented with rIL-2 on days 4 and 7. Peptide-specific T cells were tested via enzyme-linked immunospot (ELISPOT) assays on day 9 or 10.

ELISPOT assay.

Antigen-specific T lymphocyte responses were detected through a gamma interferon (IFN-γ)-secreting ELISPOT assay (BD). Briefly, flat-bottom 96-well ELISPOT plate membranes were precoated with 10 μg/ml anti-IFN-γ monoclonal antibody (MAb) and incubated overnight at 4°C. After washing with PBS, the plates were blocked with culture medium for 1 h at 37°C. A total of 2.5 × 10⁵ freshly isolated PBMCs from donors in 100 μl RPMI 1640 supplemented with 10% FBS were seeded in each well. For the detection of in vitro-cultured T cell lines, 2.5 × 10⁴ cells were added to each well. To stimulate the effector cells, individual peptides or peptide pools diluted in 100 μl RPMI 1640 supplemented with 10% FBS were added to each well and incubated at 37°C in 5% CO₂ for 18 h. Phytohemagglutinin (PHA) was added as a positive control for nonspecific stimulation. Cells incubated with medium alone were employed as a negative control that produced less than five spots in 90% of the experiments. Finally, the cells were removed, and the plates were processed according to the manufacturer's instructions. The colored spots, which represent epitope-specific T cells, were counted and analyzed using an automatic ELISPOT reader. The response was considered positive when the number of spot-forming cells (SFCs) in the target well was more than 5 and twice the number in the negative control well without stimulator.

Intracellular cytokine staining and flow cytometry.

After in vitro culture, T cell lines were rested for 2 h and then stimulated with a specific peptide pool (5 μg/ml for each peptide) or individual peptide (10 μg/ml) for 2 h and incubated with Golgistop/monesin (BD Bioscience) for an additional 4 h at 37°C in 5% CO₂. Unstimulated or PHA-stimulated cells were included as negative and positive controls, respectively. Then, the cells were harvested and stained with anti-CD3 phycoerythrin (PE) and anti-CD8 FITC surface markers. The cells were subsequently fixed and permeabilized in permeabilizing buffer (BD Bioscience) and stained with anti-IFN-γ PE-Cy7 (BD Bioscience). All the fluorescent lymphocytes were gated on a FACSCalibur flow cytometer (BD Bioscence).

Refolding of peptides with HLA heavy chain and β₂m.

HLA-A*2402 or HLA-A*1101 heavy chain and β₂m were overexpressed in Escherichia coli as inclusion bodies and subsequently refolded in vitro in the presence of a high concentration of peptide, as described previously (41, 44, 47, 80). Briefly, the dissolved HLA heavy chain and β₂m inclusion body and peptides were diluted at a molar ratio of 1:1:3, respectively, in a refolding buffer (100 mM Tris-HCl, 400 mM l-arginine, 2 mM EDTA-Na, 5 mM glutathione [GSH] and 0.5 mM l-Glutathione oxidized [GSSG]). After 12 h of slow stirring at 4°C, the peptide–HLA-A complex was then concentrated and purified with Superdex 200 10/300 GL (GE Healthcare) chromatography. If the protein was prepared for crystal screening (HLA-A*2402), it was further purified on an ion-exchange Resource Q (GE Healthcare) column.

X-ray crystallography, structure determination, and refinement.

The concentration of refolded HLA-A*2402–peptide complexes was adjusted to 10 mg/ml in 20 mM Tris-HCl (pH 8.0) and 50 mM NaCl. Crystals were grown by the hanging drop vapor diffusion method at 18°C (45). Single HLA-A*2402–P21 complex crystals appeared in 0.2 M sodium nitrate and 20% (wt/vol) polyethylene glycol 3,350 within 6 days. Crystals of both peptides P27 and P28 complexed to HLA-A*2402 grew in 0.1 M Bis-Tris (pH 5.5) and 10% (wt/vol) polyethylene glycol 3,350. For cryoprotection, crystals were transferred to reservoir solutions containing 20% glycerol. Crystallographic data were collected at 100 K in house on a Rigaku MicroMax007 rotating-anode X-ray generator with a Cu target operated at 40 kV and 20 mA equipped with an R-Axis VII++ image plate detector at a wavelength of 1.5418 Å. The data were indexed and scaled using DENZO and the HKL2000 software package. The structures were determined using molecular replacement with the program CNS (9a). The model used was the structure of Protein Data Bank (PDB) code 3I6L; extensive model building was performed by hand using COOT, and restrained refinement was performed using REFMAC5 (15a, 55a). The stereochemical quality of the final model was assessed with the program PROCHECK (36a). Structure-related figures were generated using PyMOL (http://www.pymol.org/).

Statistical analysis.

The null hypothesis proposed that tested parameters in different HLA supertype-restricted individuals would not differ significantly from each other. Differences in mean values were evaluated for statistical significance (P <0.05 or <0.01) by Student's two-tailed t test. Data were assembled and statistically calculated on a computerized spreadsheet program (Excel; Microsoft Corp.).

Protein structure accession numbers.

The coordinates and structure factors of peptides P21, P27, and P28 complexed to HLA-A*2402 have been deposited in the PDB under accession numbers 4F7M, 4F7P, and 4F7T, respectively.

Selection of potential HLA-A24-restricted epitope peptides from 2009 pH1N1 virus and the establishment of an HLA-A24 peptide pool.

To evaluate the CTL responses specific to influenza A virus in the HLA-A24⁺ population, we performed genome-wide screening of HLA-A*2402-restricted epitopes derived from the 2009 pH1N1. Based on software prediction, 34 8-mer to 11-mer candidate peptides with high estimated half-time dissociation were selected and synthesized for further identification (Table 1). Compared to the seasonal H1N1 influenza A virus (A/Brisbane/59/2007), half of the candidate peptides (17/34) were completely conserved (Table 1). This conservation rate corresponds to a previous analysis of the T cell epitopes contained in 2009 pH1N1 (15). To evaluate the overall influenza A virus-specific CTL response among the HLA-A24⁺ population during the pandemic in 2009 and 2010, all of the candidate peptides were mixed together at equal concentrations to construct a peptide pool specific to 2009 pH1N1 and restricted by HLA-A24.

Influenza virus peptide pool-specific CTL responses among HLA-A24⁺ donors and high-level cross-responses in HLA-A11⁺ donors.

Freshly isolated PBMCs from 19 donors were tested for CTL responses against the HLA-A24 peptide pool by using ELISPOT assays. The results revealed that the peptide pool could be recognized in six out of eight HLA-A24⁺ subjects (Fig. 1) (with a mean response of 145 SFCs/10⁶ PBMCs among the six donors). The highest response was detected in donor a-8 (551 SFCs/10⁶ PBMCs), which corresponded to the donor also having the highest 2009 pH1N1-specific antibody level among all of the subjects.

Additionally, we found that the responses against the peptide pool in two of the HLA-A24-negative (HLA-A24⁻) individuals were also very strong. Further analysis of the HLA alleles of these donors revealed that both of them were HLA-A11 positive (see Table S1 in the supplemental material). One of them was HLA-A11 homozygous (donor a-5), and the other was an HLA-A01/HLA-A11 heterozygote (donor a-13). Further, there were no HLA alleles in common at the HLA-B loci for these donors. Thus, we speculated that the peptides in the pool were likely cross-recognized by the HLA-A11 allele. Further analysis revealed that the peptide pool-specific T cells among all the HLA-A11 donors (209 ± 150 SFCs/10⁶ PBMCs) were much more numerous than those among the HLA-A24⁻ and HLA-11⁻ donors (16 ± 15 SFCs/10⁶ PBMCs) (P < 0.01) (Fig. 1).

Identification of the antigenic peptides in the influenza virus peptide pool.

To further identify the individual peptides in the pool that could prime CD8⁺ T cell responses in HLA-A24⁺ donors and define the epitopes cross-recognized by HLA-A11⁺ donors, we evaluated the HLA-binding affinity and the CTL-specific antigenicity of the individual peptides in the pool. First, we performed an HLA stabilization assay to evaluate the binding affinities of the candidate peptides for HLA-A*2402 in vitro utilizing T2-A24 cells (47). Increased expression of HLA-A*2402 on the T2-A24 cell surface was detected when the cells were loaded with the high-affinity peptide. Fifteen of the pooled peptides enhanced cell surface HLA-A*2402 expression (FI > 1.5) compared to the other peptides, indicating higher binding affinity of these 15 peptides for HLA-A*2402 (Table 1).

To further evaluate the antigenicities of the peptides that had high HLA-A*2402 binding affinity, their capacity to stimulate IFN-γ secretion in the PBMCs from the HLA-A24⁺ donors was tested in vitro using the peptide pool-specific T cell lines. PBMCs collected from HLA-A24⁺ subjects were stimulated in the presence of the peptide pool and IL-2. On the 10th day, individual peptide-specific induction of IFN-γ was detected by ELISPOT assays. Generally, five (P4, P6, P14, P21, and P28) of the 15 high HLA-A*2402 binding affinity peptides could be recognized by the peptide pool-stimulated T cell lines (Fig. 2). The results demonstrated that distinct levels of IFN-γ production were detected in cell lines from different HLA-A24⁺ donors. The variability of the response levels under simulation by the peptides may reflect the different proliferation rates of the epitope-specific T cells in different individuals or the differences in the memory precursors and phenotypes of the influenza virus-specific T cells in different donors.

Among the five newly identified antigenic peptides, peptide P28 (RYGFVANF) was found to overlap the longer peptide P27 (FYRYGFVANF) in the peptide pool, which also has a typical HLA-A24 peptide motif. Thus, P27 was also used in the following experiments. In a parallel report recently, peptide P27 was determined to have CTL-specific immunogenicity with HLA-A24 restriction (2).

Conservation of the newly identified influenza virus-specific CTL epitopes.

Recent studies (3, 37) have focused on the conserved epitopes that can induce a broader T cell response against different subtypes of human influenza viruses. By comparing the corresponding sequences of the six peptides defined here among different subtypes of influenza A viruses (Table 2), we analyzed their conservation. Four peptides, P6, P21, P27, and P28, are highly conserved among all influenza A virus subtypes known to infect humans. Epitopes P4 and P14 are relatively variable compared to the other four epitopes. However, no variable site was observed in the positions of the anchor residues (P2 and Pc) of P4 and P14. With the same binding motifs, these two peptides from different viral subtypes may still be able to bind to the HLA-A24 heavy chain, as well. In conclusion, the conservation of these six epitopes may indicate their potential significance in influenza virus-specific T cell immunity evaluation and vaccine development for the HLA-A24⁺ population.

Individual peptide-specific cross-responses among HLA-A24⁺ and HLA-A11⁺ donors by ex vivo evaluation.

The individual peptide-specific CD8⁺ T cells in the peripheral blood were also analyzed by utilizing freshly isolated PBMCs from HLA-A24⁺ donors. Different levels of specific CD8⁺ T cell responses against the six peptides (P4, P6, P14, P21, and P28, as well as P27) could be detected in the HLA-A24⁺ individuals, although none of the subjects could recognize all six peptides (Fig. 3A). The peptide P4-, P6-, P14-, and P21-specific CTL response ratio was 1 to 2 out of 4 detected HLA-A24⁺ subjects. In contrast, CTL responses to P27 and P28 were simultaneously detected in all four of the tested subjects, which may indicate the immunodominance of P27 and P28. Utilizing the intracellular cytokine staining assay, we further confirmed that P27- and P28-specific CD8⁺ T cells could be detected after in vitro expansion with an A24 peptide pool (Fig. 4). The high level of both P27- and P28-elicited IFN-γ-secreting cells among the CD8⁺ T cells of the HLA-A24⁺ donor also demonstrates the immunodominant roles of peptides P27 and P28.

Considering the high level of T cell responses of the peptide pool among the HLA-A11⁺ donors, we attempted to identify the individual peptides among the six newly defined HLA-A24-restricted peptides that could be cross-recognized by HLA-A11⁺ donors. Ex vivo ELISPOT assays using the freshly isolated PBMCs from two HLA-A11⁺ donors (donor a-5 and donor a-13) revealed that peptide P21 could specifically stimulate IFN-γ production in these two donors (Fig. 3B). To investigate the small population of influenza virus-specific memory CD8⁺ T cells, PBMCs collected from donor a-5 were cultured for 10 days in the presence of the A24 peptide pool. The subsequent ELISPOT assays demonstrated that the expanded-peptide-pool-specific T cell lines could still recognize P21 (Fig. 3C). Additionally, P6-specific CD8⁺ T cells were also detected in the expanded CTL lines (Fig. 3C). This result indicated that P21 and P6 may be the major elements in the HLA-A24 peptide pool that trigger the cross-responses among the HLA-A11⁺ donors.

However, we found that the cross-reactive peptides P21 and P6 induce T cell responses among only a portion of HLA-A24⁺ individuals. This may indicate a subdominant role of the cross-reactive peptides. Moreover, we can conclude that cross-reactivity does not occur among all the identified peptides, including the immunodominant epitopes P27 and P28, which may reveal that the cross-reactivity is related to the intrinsic characteristics of different peptides.

Structural evidence of peptide presentation by HLA-A*2402 and HLA-A*1101.

The in vitro refolding of HLA-A*2402 heavy chain and β₂m in the presence of peptide revealed that all six of the analyzed peptides aided in the formation of the HLA-A*2402–peptide complex (Fig. 5A) (47). We also performed the T2-A24 cell binding assay to evaluate the binding affinities of these six peptides for HLA-A*2402. The peptides enhanced cell surface HLA-A*2402 expression compared to the negative-control peptide, indicating a higher binding affinity for HLA-A*2402 (Fig. 5B). Peptide P27 also showed a moderate binding affinity to HLA-A*2402 in these further tests.

Subsequently, we determined the structures of P21, P27, and P28 complexed to HLA-A*2402 (Fig. 6A to C and Table 3), by which we clearly confirmed the typical HLA-A24-restricted presentation characteristics of these newly identified antigenic peptides. In these structures, the three peptides adopt the classical conformations of HLA-A*2402-restricted peptides. They use similar primary anchors (Tyr and Phe), which insert deeply into the B and F pockets of HLA-A*2402, respectively (Fig. 6D). The P21 peptide assumes a moderately bulged conformation, similar to previously observed peptide binding (47), with the side chains of residues Ser4, Pro5, Leu7, and Glu8 protruding into the solvent, which may play a potential role in T cell receptor (TCR) docking. The long side chain of Gln6 of peptide P21 points into the peptide binding groove of HLA-A*2402, which acts as a secondary anchor.

The structural determination of the two closely related peptides P27 and P28 complexed to HLA-A*2402 made it possible to elucidate the relatedness of these two overlapping immunodominant peptides. The structure of the HLA-A*2402–P27 complex demonstrated that the middle region of P27 may adopt a flexible conformation, because the structures of Gly5, Phe6, and Val7 of the peptide could not be determined due to their poor electron densities (Fig. 6B). The well-defined electron densities of peptide P28 complexed to HLA-A*2402 clearly show that the peptide assumes a flat conformation (Fig. 6C). Peptide P28 is a truncated form of P27, with 8 residues derived from the C terminus of P27. In P27, the N-terminal position 2 residue, Tyr, is embedded in the B pocket, and the position 4 residue, Tyr (position 2 in the P28 peptide), protrudes above the binding surface and would interact with the TCR. However, the three residues (Ala, Asn, and Phe) at the C termini of the two peptides adopt similar conformations.

Despite the different lengths of the overlapping peptides P27 and P28, they had correlated immunodominant antigenicities (i.e., robust responses in similar individuals) in the ex vivo detection of specific CTLs of HLA-A24⁺ donors (Fig. 3A). The structural analysis of the two peptide complexes with HLA-A*2402 also indicated that the structures of the two peptides correlate with each other. First, the conformations of the solvent-exposed residues with well-defined electron densities of P27 are similar to the exposed residues of P28, especially Tyr4 of P27 versus Phe4 of P28 and Asn9 of P27 versus Asn7 of P27 (Fig. 6E). Second, the extremely flexible conformation of P27 may allow the peptide to induce a diverse TCR repertoire that covers the P28-specific TCRs.

The cross-recognized peptides P6 and P21 also displayed some mild capability to bind to HLA-A*1101 in the refolding assay (P21 in Fig. 7A). Based on the HLA-A*1101 structure determined by previous studies and the HLA-A*2402–P21 structure determined here, we generated a structural model of the cross-recognized peptide P21 complexed with HLA-A*1101. The peptide binding groove of HLA-A*1101 accommodates the P21 peptide comfortably (Fig. 7B). Subsequently, we analyzed in detail the F pocket of the HLA-A*1101/P21 model, as the major distinct part of the peptide binding groove of HLA-A*2402 and HLA-A*1101 is located in the F pocket, which determined the different peptide motifs of the two alleles. Although the F pocket of HLA-A*1101 presents negatively charged static electricity due to residues D74, D77, and D116, the bulky side chain of Phe9 of P21 can still fit the large space in the F pocket of HLA-A*1101 well (Fig. 7C). The fit of the P21 peptide with the peptide binding groove, especially the F pocket of HLA-A*1101, might explain the moderate binding of P21 to HLA-A*1101.

Cross-recognition of the HLA-A24 motif-specific peptide pool among A3 supertype-positive subjects.

Previous studies showed that a number of CTL epitopes derived from other pathogens (or tumors) can be cross-presented by different HLA alleles assigned to the same A3 supertype that share similar binding motifs, including HLA-A11, -A31, -A33, and -A30 (only A*3001 in our cohort) (42, 52, 70). Therefore, we hypothesized that the HLA-A24-restricted influenza A virus-specific epitopes recognized by HLA-A11⁺ subjects would also be cross-recognized in the context of whole A3 supertype alleles. A larger cohort including 77 healthy subjects from Jiangsu Province, China (see Table S2 in the supplemental material), was therefore recruited during the 2009 H1N1 pandemic to analyze their immune responses against the influenza virus peptide pool. The subjects were assigned to four major groups (Fig. 8): A24⁻/A3st⁻ (A24 negative and A3 supertype allele negative), A24⁺/A3st⁻, A24⁺/A3st⁺, and A24⁻/A3st⁺. Four subgroups of the A3 supertype—HLA-A11-, -A33-, -A31-, and -A30-positive subjects in the A24⁻/A3st⁺ group—were also independently analyzed. In the four A24⁻/A3st⁺ subgroups, A3 supertype allele heterozygote subjects were assigned according to a priority order of HLA-A11, -A33, -A31, and -A*3001 to make sure that none of the subjects were assigned to different groups simultaneously and analyzed repeatedly.

As expected, the influenza virus-specific T cell responses in the HLA-A11⁺ group were much higher than the responses in the A24⁻/A3st⁻ subjects (average, 75 SFCs/10⁶ PBMCs compared to 8 SFCs/10⁶ PBMCs; P < 0.05) (Fig. 8). The cross-responses against the influenza viruses were also observed in other alleles of the A3 supertype (104 SFCs/10⁶ PBMCs for A31, 120 SFCs/10⁶ PBMCs for A33, and 31 SFCs/10⁶ PBMCs for A30; P < 0.05 or < 0.01). Statistical analysis of the T cell responses in the entire supertype revealed that the responses against the peptide pool in A24⁻/A3st⁺ (whole A3 supertype; average, 88 SFCs/10⁶ PBMCs), A24⁺/A3st⁻ (average, 144 SFCs/10⁶ PBMCs), and A24⁺/A3st⁺ (average, 171 SFCs/10⁶ PBMCs) donors were much higher than that in A24⁻/A3st⁻ subjects (average, 8 SFCs/10⁶ PBMCs; P < 0.05 or <0.01). This is strong evidence that influenza virus-specific cross-responses occur in the context of all A3 supertype alleles, including HLA-A11⁺ subjects. Although only two donors showed positive responses in the HLA-A31⁺ group (with one just over the cutoff of 20 SFCs/PBMC), the average T cell response of the group was significantly higher than that of the A24⁻/A3st⁻ group. However, more donors with HLA-A31 and other uncommon A3 supertype alleles (e.g., A*6801) may be needed to confirm the conclusion.

Moreover, the ratios for the A24⁺ or A3st⁺ donors with positive responses within the cohort are higher than for the A24⁻/A3st⁻ donors (9/14 for A24⁺/Ast3⁻, 11/13 for A24⁺/A3st⁺, 10/17 for A11⁺, 6/9 for A33⁺, 2/4 for A31⁺, 5/9 for the A30⁺ group, and 1/11 for A24⁻/A3st⁻). This is concordant with the analysis of the average response levels.

We also generated structural models of the cross-recognized peptide P21 complexed with HLA-A*0301 and HLA-A*6801 based on the previously determined structures of these two molecules and HLA-A*2402–P21 (82a). Similar to HLA-A*1101, the peptide binding grooves of HLA-A*0301 and HLA-A*6801 accommodate the P21 peptide well (see Fig. S1 in the supplemental material). Alignment of HLA-A*2402-presented P21 with the different peptides presented by A3 supertype alleles indicated similar conformations of P21 with HLA-A*1101- and HLA-A*6801-presented peptides (see Fig. S1). This may indicate a structural basis of cross-supertype presentation of the peptides by A24 and A3 supertype alleles.