Cross-reactive CD8⁺ T-cell immunity between the pandemic H1N1-2009 and H1N1-1918 influenza A viruses

Abstract

Preexisting T-cell immunity directed at conserved viral regions promotes enhanced recovery from influenza virus infections, with there being some evidence of cross-protection directed at variable peptides. Strikingly, many of the immunogenic peptides derived from the current pandemic A(H1N1)-2009 influenza virus are representative of the catastrophic 1918 “Spanish flu” rather than more recent “seasonal” strains. We present immunological and structural analyses of cross-reactive CD8⁺ T-cell–mediated immunity directed at a variable (although highly cross-reactive) immunodominant NP_418–426 peptide that binds to a large B7 family (HLA-B*3501/03/0702) found throughout human populations. Memory CD8⁺ T-cell specificity was probed for 12 different NP₄₁₈ mutants that emerged over the 9 decades between the 1918 and 2009 pandemics. Although there is evidence of substantial cross-reactivity among seasonal NP₄₁₈ mutants, current memory T-cell profiles show no preexisting immunity to the 2009-NP₄₁₈ variant or the 1918-NP₄₁₈ variant. Natural infection with the A(H1N1)-2009 virus, however, elicits CD8⁺ T cells specific for the 2009-NP₄₁₈ and 1918-NP₄₁₈ epitopes. This analysis points to the potential importance of cross-reactive T-cell populations that cover the possible spectrum of T-cell variants and suggests that the identification of key residues/motifs that elicit cross-reactive T-cell sets could facilitate the evolution of immunization protocols that provide a measure of protection against unpredicted pandemic influenza viruses. Thus, it is worth exploring the potential of vaccines that incorporate peptide variants with a proven potential for broader immunogenicity, especially to those that are not recognized by the current memory T-cell pool generated by exposure to influenza variants that cause successive seasonal epidemics.

The rapid global spread of the A(H1N1)-2009 “swine origin” influenza virus in human populations led the World Health Organization (WHO) to declare this the first influenza pandemic of the 21st century (1). Although A(H1N1)-2009 has not caused uniformly severe disease, the perception that the 1918–1919 H1N1 catastrophe (>40 million people died) began with an earlier milder phase raises the specter that this newly emerged virus could mutate to increased virulence. As it is, women in the third trimester of pregnancy, the very obese, and those living in communities with poor basic health and nutrition seem to be at increased risk, and there have been fatal cases in otherwise healthy children. The A(H1N1)-2009 statistics are also skewed by the fact that although influenza is normally most severe in the elderly, those who were alive before 1950 have some level of preexisting immunity (2).

In the absence of cross-reactive neutralizing antibodies, one way of minimizing the severity of influenza A virus infections is to recall established T-cell memory directed at peptides derived from relatively invariant internal proteins. Evidence from animal models (3) and humans (4–6) indeed suggests a role for CD8⁺ T-cell–mediated protection following heterologous prime/challenge between H1N1, H7N7, H3N2, and H5N1 viruses. Such cross-reactive CD8⁺ T-cell–mediated immunity has, in the face of an emerging influenza pandemic, the potential to ameliorate the disease by promoting enhanced recovery.

Protective CD8⁺ T-cell–mediated immunity can be directed at peptides that are, like M1₅₈ (7), conserved among different influenza subtypes and strains or vary at non-MHC anchor residues in a way that still allows productive T-cell receptor (TCR) binding (8, 9). Thus, although only ∼50% (10) of potentially immunogenic T-cell peptides are shared between the A(H1N1)-2009 and “seasonal” influenza strains, it is possible that individuals with HLA types that present the remaining variable epitopes could have some T-cell immunity to the A(H1N1)-2009 pandemic strain. In this study, we ask whether exposure during successive seasonal epidemics generated at least some memory T cells that also recognize nonconserved (containing at least one amino acid variation) peptides derived from the A(H1N1)-2009 pandemic influenza virus.

Results

We probed the potential for preexisting memory T-cell responses against nonconserved A(H1N1)-2009–derived peptides. Sequence analysis of influenza viruses established that many of the A(H1N1)-2009–derived variable peptides within the main T-cell immunogenic proteins (6, 11), nucleoprotein (NP) and matrix-1 (M1), are more like the pandemic 1918-H1N1 strain rather than the seasonal influenza strains (Table S1). Based on these results, we focused on the variable (although highly cross-reactive between different variants derived from seasonal influenza viruses) immunodominant NP_418–426 peptide that binds a large B7 allelic family (12, 13) found throughout human populations (HLA-B*3501/HLA-B*3503/HLA-B*0702) and probed memory CD8⁺ T-cell specificity for 12 different NP₄₁₈ mutants derived from a spectrum of seasonal and pandemic viruses (8, 12–14). These include the A(H1N1)-2009 virus, the 1918-H1N1 virus, and 10 seasonal viruses from 1933 to 2005 (Table 1).

Table 1.

Naturally occurring NP_418–426 variants used in the study and the stability of the selected HLA-B*3501-NP₄₁₈ complexes

Sequence	Year	Subtype	Representative influenza strain	HLA-B*3501-NP418 thermostability, °C
LPFERATIM	1918	H1N1	A/Brevig Mission/1/1918	73.3 ± 0.7
LPFDRPTIM	1933	H1N1	A/Wilson Smith/1933
LPFDRTTIM	1934	H1N1	A/Puerto Rico/8/1934	75.4 ± 0.5
LPFDKPTIM	1957	H2N2	A/Iran/1/1957
LPFDKSTIM	1972	H3N2	A/Udorn/307/1972	74.7 ± 0.3
LPFDKTTIM	1947	H1N1/H2N2	A/Memphis/13/1978	74.5 ± 0.7
LPFDKSTVM	1977	H1N1/H3N2	A/California/10/1978
LPFEKSTVM	1980	H3N2	A/Memphis/4/1980	75.3 ± 1.7
LPFEKSTIM	2002	H3N2	A/Fujian/411/2002
LPFDKATIM	2005	H1N1	A/Otago/5/2005
LPFDIATIM	2006	H1N1	A/Wellington/75/2006
LPFERATVM	2009	A(H1N1)-09	A/Auckland/1/2009	74.8 ± 1.4

Variable positions between A(H1N1)-09 and other seasonal and pandemic NP₄₁₈ variants are depicted in bold and underlined. Thermostability of HLA-B*3501-NP₄₁₈ complexes was determined by the thermal melting method. Thermostability represents a temperature (°C) at which 50% of the protein is unfolded.

Peripheral blood mononuclear cells (PBMCs) were cultured with pooled peptides for 10 d and then restimulated with all 12 variants (either singly or together) in an IFN-γ intracellular cytokine staining assay. Despite some differences (likely related to the individual infection histories), analysis of PBMCs from healthy HLA-B*3501 (Fig. 1 A–C, E), HLA-B*3503 (Fig. 1 C and D), and HLA-B*0702 (Fig. 1F) donors with no evidence of exposure to A(H1N1)-2009 (Table S2) confirmed a high degree of cross-reactivity for the NP₄₁₈ seasonal variants, particularly 1972-NP₄₁₈-LPFDKSTIM, 1947-NP₄₁₈-LPFDKTTIM, 1977-NP₄₁₈-LPFDKSTVM, and 2002-NP₄₁₈-LPFEKSTIM, and some level of recognition for 1957-NP₄₁₈-LPFDKPTIM and 2005-NP₄₁₈-LPFDKATIM (Fig. 1 A–F). Zero or minimal reactivity was seen for the 2009-NP₄₁₈-LPFERATVM, 1918-NP₄₁₈-LPFERATIM, 1933-NP₄₁₈-LPFDRPTIM, and 1934-NP₄₁₈-LPFDRTTIM variants (Fig. 1 A–F). Thus, there is evidence of shared immunogenicity for recent seasonal NP₄₁₈ variants but not (or minimally) for A(H1N1)-2009 NP₄₁₈ or the early 20th century viruses.

Fig. 1.

Loss of interepitope cross-reactive T-cell immunity for the NP_418–426 variant with the emergence of the A(H1N1)-2009 strain. Lymphocytes were taken from healthy donors who had no serological or clinical evidence of exposure to the pandemic A(H1N1)-2009 strain [D1, HLA-B*3501⁺ (A); D2, HLA-B*3501⁺ (B); D3, HLA-B*3503/01⁺ (C); D4, HLA-B*3503⁺ (D); D5, HLA-B*3503⁺(8) (E); and D6, HLA-B*0702⁺ (F)] and from patients hospitalized following infection with this virus [D7, HLA-B*0702 (G) and D8, HLA-B*0702 (H)]. Individual PBMC populations were cultured in vitro for 10 d with a pool of 12 variant NP₄₁₈ peptides (Table 1) and then analyzed for CD8⁺ T-cell specificity in a standard IFN-γ intracellular cytokine staining assay. The final 7-h stimulation was with HLA-B*3501⁺CIR-B35 cells (A–E) or with autologous HLA-B*0702⁺ PBMCs (F–H) pulsed with either a pool of 12 peptides (black bars) or single NP₄₁₈ variants (white bars). The percentage of IFN-γ–producing cells in the CD8⁺ T-cell fraction is shown with the no-peptide control background subtracted. Background IFN-γ production for restimulated CD8⁺ T cells was typically 0.046–0.62%. Experiments were repeated with similar results unless limited by cell numbers. CD8⁺ T-cell reactivity of HLA-B*0702⁺ D6, D7, and D8 against the pandemic A(H1N1)-2009 NP₄₁₈ variant in single-peptide PBMC cultures is shown in Fig. 2. Plasma antibody titers are listed in Table S2.

Blood was then obtained within 3 mo of infection from two HLA-B*0702⁺ patients who had been hospitalized with severe A(H1N1)-2009 pneumonia and subsequently recovered (15). Stimulation with the 12 NP₄₁₈ variants showed strong PBMC responses to A(H1N1)-2009-NP₄₁₈, 1918-NP₄₁₈, and 1980-NP₄₁₈ and, to a lesser extent, to 1933-NP₄₁₈ and 1934-NP₄₁₈ (Fig. 1 G and H). This provided formal proof that A(H1N1)-2009 indeed primes CD8⁺ T-cell memory, a conclusion that was further refined by stimulation in culture with the 2009-NP₄₁₈ variant alone (Fig. 2 B and C). In contrast, a healthy HLA-B*0702 donor had no A(H1N1)-2009–specific reactivity (Fig. 2A).

Fig. 2.

Natural infection with A(H1N1)-2009 elicits CD8⁺ T cells specific for 2009-NP₄₁₈ that are cross-reactive with 1918-NP₄₁₈. CD8⁺ T-cell reactivity against the A(H1N1)-2009 NP₄₁₈ variant is shown for single-peptide PBMC cultures obtained from a healthy donor (A) and A(H1N1)-2009–infected HLA-B*0702⁺ donors (B and C). PBMCs were obtained from D6 (A), D7 (B), and D8 (C) and stimulated for 10 d with the 2009-NP₄₁₈ peptide. On day 10, PBMCs were restimulated with autologous PBMCs pulsed with either a pool of 12 or single NP₄₁₈ peptides. CD8⁺ T-cell reactivity was determined following a 7-h restimulation (5 h with Brefeldin A) by IFN-γ production. The percentage of IFN-γ production by CD8⁺ T cells is shown, with background subtracted of no-peptide controls. Background IFN-γ production was 0.2, 0.26, and 0.029% for D6, D7, and D8, respectively. Black bars show the corresponding original restimulation 2009-NP₄₁₈ peptide, and white bars show cross-reactive CD8⁺ T-cell responses.

Two of the healthy HLA-B*3501⁺ donors showed low-level reactivity to the A(H1N1)-2009-NP₄₁₈ variant following culture with the NP₄₁₈ peptide pool (Fig. 1 A and B); thus, the protocol was repeated using the individual NP₄₁₈ variants (Fig. 3). These cultures were split on day 10 and restimulated separately with pooled or single peptides. The analysis confirmed that at the population level, there is a high level of cross-reactive immunity for a number of the mutants that emerged between the 1950s and 2005 (Fig. 3). All the cross-reactive NP₄₁₈ variants (1957-NP₄₁₈-LPFDKPTIM, 1972-NP₄₁₈-LPFDKSTIM, 1947-NP₄₁₈-LPFDKTTIM, 1977-NP₄₁₈-LPFDKSTVM, and 2005-NP₄₁₈-LPFDKATIM) are characterized by a conserved aspartic acid (D) at position (P) 4 and a lysine (K) at P5 (Fig. 1 A–F). There was no evidence for recognition of A(H1N1)-2009-NP₄₁₈ or 1918-NP₄₁₈, with minimal reactivity for 1934-NP₄₁₈ following single-peptide cultures.

Fig. 3.

Patterns of NP₄₁₈-specific cross-reactive T-cell immunity in the healthy HLA-B*3501⁺ donor in single-peptide cultures. Individual PBMC populations from D1 HLA-B*3501 were stimulated for 10 d with either the 12 pooled (Table 1) or individual NP₄₁₈ peptides (shown above each panel). On day 10, PBMCs were restimulated with peptide-pulsed CIR-B35 cells and assayed for IFN-γ production, as described in the legend to Fig. 1. The black bars give the results for the homologous restimulation, whereas the white bars show the cross-reactive CD8⁺ T-cell responses. Background IFN-γ production was 0.34–0.82% (subtracted from the samples).

Given that point mutations within NP₄₁₈ may influence the stability of immunogenic peptide–MHC-I (pMHC-I) complexes, and thus TCR recognition (16), profiles of pMHC-I thermal stability were then analyzed for six NP₄₁₈ variants from seasonal (1934, 1947, 1972, and 1980) and pandemic (1918 and 2009) viruses. All the NP₄₁₈ variants appeared equally effective at stabilizing HLA-B*3501, with the pMHC-I complexes showing (Table 1) comparable levels of thermostability (∼74 °C; range: 73.3 °C ± 0.7–75.3 °C ± 1.7). Overall, the stability of the HLA-B*3501-NP₄₁₈ complexes was extremely high, an effect that can lead to prolonged half-life on the cell surface and, in turn, increase the immune-mediated selective pressure favoring the emergence of NP₄₁₈ variants.

To gain further insight into the nature of these pMHC-I epitopes, we determined the high-resolution structures (Fig. 4 and Table S3) of HLA-B*3501 complexed to six NP₄₁₈ peptides (from NP₄₁₈ variants emerging between 1918 and 2009). The six NP₄₁₈ peptides adopt a similar conformation in the HLA-B*3501 binding cleft, with an average rmsd of 0.13 Ų on the peptide when compared with the 1980-NP₄₁₈ structure (as a reference). In addition, subtle changes in MHC-I residues were observed in the specific NP₄₁₈ variants (Fig. S1). The NP₄₁₈ peptides use P2-P and P9-M as anchor residues buried inside the cleft. There was no secondary anchor residue in any of the HLA-B*3501-NP₄₁₈ structures. This is in contrast to other peptide-HLA-B*3501 structures solved to date, which typically use P5-D or P5-L as the secondary anchor residues (17–19). Instead, the central residue in the NP₄₁₈ variants [P5-R in 1918-NP₄₁₈, 1934-NP₄₁₈, and A(H1N1)-2009-NP₄₁₈ and P5-K in 1947-NP₄₁₈, 1972-NP₄₁₈, and 1980-NP₄₁₈] (Fig. 4) bulges out of the binding cleft, exposing its side chain to the solvent for potential contact by the TCR. The variation at this most exposed residue may, at least in part, explain the mutually exclusive cross-reactivity for A(H1N1)-2009-NP₄₁₈ and 1918-NP₄₁₈ (prominent R) vs. the alternative recognition profile for the seasonal NP₄₁₈ variants (1947-NP₄₁₈ and 1972-NP₄₁₈) that display the DK motif (Fig. 1). These results suggest that distinct cross-reactive CD8⁺ T-cell populations recognize HLA-B*3501/3503/0701 NP₄₁₈ epitopes displaying either the prominent R or DK motif.

Fig. 4.

Chronological mutations in the NP_418–426 variants occur at the most solvent-exposed residues. Crystal structures of the binary HLA-B*3501 complexes with the NP_418–426 variants emerging chronologically between 1918 and 2009 were resolved: 1918-NP₄₁₈ LPFERATIM (A), 1934-NP₄₁₈ LPFDRTTIM (B), 1947-NP₄₁₈ LPFDKTTIM (C), 1972-NP₄₁₈ LPFDKSTIM (D), 1980-NP₄₁₈ LPFEKSTVM (E), and 2009-NP₄₁₈ LPFERATVM (F). Further characteristics of pMHC-I interactions are presented in Fig. S1.

Discussion

Overall, our study indicates how and why NP₄₁₈ changed over the decades to escape this immunodominant CD8⁺ T-cell response. The 1918-NP₄₁₈ variant displays the characteristic ER motif at the most solvent exposed P4–P5 region of the peptide and, given the high level of cross-reactivity with A(H1N1)-2009-NP₄₁₈, it is likely that the 1918-NP₄₁₈ would have elicited potent CD8⁺ T-cell immunity. The 1934-NP₄₁₈ point mutation (E→D at P4) led to a partial loss of recognition for A(H1N1)-2009-NP₄₁₈; then, from the 1950s onward, the virus mutated further to escape any cross-reactive immunity (driven by P5-R) and introduced the DK motif seen in the 1947-NP₄₁₈ and 1972-NP₄₁₈ variants. Cross-reactive seasonal variants displaying the DK motif then dominated, although other mutants have emerged more recently (Table 1). Thus, most cross-reactive immunity at the population level is directed at variants displaying the P4-D and P5-K solvent-exposed residues. The 1980-NP₄₁₈ P4-E→D mutant does not cross-react with the “DK” variants, although the P4-E substitution likely explains reciprocity for A(H1N1)-2009-NP₄₁₈. Interestingly, the 2002 NP_418–426 peptide also exhibits cross-reactivity with the 1980, 2009, and 1918 NP_418–426 peptides, most probably because of the shared P4-E.

The emergence of the A(H1N1)-2009-NP₄₁₈ variant indicates that the virus (which was being maintained in pigs) avoided preexisting cross-reactive T-cell immunity established over the past 60–90 y, maintaining a unique structure that shows little cross-reactivity with seasonal strains. The finding that the conserved DK motif associated with cross-reactive T-cell responses for the recent seasonal viruses was replaced by ER in A(H1N1)-2009-NP₄₁₈ suggests that P4 and P5 are crucial for TCR recognition. The NP gene of the 2009 pandemic virus is one of three genes (together with HA and NS) derived from the classic H1N1-swine lineage maintained in swine since its introduction from avian sources around 1918. It is therefore not surprising to find high sequence conservation in the NP gene, including the NP_418–426 sequence. This provides a fortuitous advantage to the 2009-H1N1 pandemic virus to emerge in humans that have memory CD8⁺ T-cell responses directed against evolved variant seasonal peptides rather than against the 1918 and 2009 variants. Furthermore, the fact that the A(H1N1)-2009-NP₄₁₈-LPFERATVM variant appeared sporadically in both H1N1 and H3N2 influenza viruses isolated since 1988 (A/Wisconsin/3523/1988 H1N1, A/MD/12/1991 H1N1, A/Iowa/ceid23/2005 H1N1, and A/Ontario/rv1273/2005 H3N2) suggests that the virus was already mutating naturally to escape DK-directed immunity. Because the majority of immunogenic T-cell peptides derived from A(H1N1)-2009 NP and M1 proteins highly resemble those derived from the 1918-H1N1 strain rather than the seasonal influenza strains (Table S1), it seems that our results can be generalized to at least some of the remaining 1918-H1N1–like variable T-cell peptides.

Similar to our findings, a recent report revealed the structural basis for cross-reactivity between the pandemic 2009-H1N1 and 1918-H1N1 antibody responses (20). Comparison of the antibody antigenic sites within the viral HA showed high antigenic conservation in the Sa, Sb, and Cb epitopes between H1N1 A/California/04/2009 (>99% similarity to the A/Auckland/1/2009 strain used in our study) and A/South Carolina/1/1918. Conversely, the antibody antigenic sites within the H1N1-2009 pandemic strain showed only 50% conservation when compared with the current seasonal A/Brisbane/59/2009 strain. Taken together, the structural and immunological data for both antibody (20) and T-cell immunity (our present study) show extremely high antigenic conservation between the 2009-H1N1 and 1918-H1N1 pandemic strains (and partial cross-reactivity with the seasonal strain from 1930s), which may explain the low infection rate with the 2009-H1N1 influenza strain in the elderly.

Our study builds on previous experiments with HLA-B*3501/HLA-B*3503/HLA-B*0702-NP₄₁₈ that used mainly T-cell clones generated against NP_418–426 peptides from the 1957, 1972, 1977, and 1980 influenza strains (8) or responding T cells from transgenic HLA-B*0702 mice infected with an H1N1 virus (13). The present analysis mapped the spectrum of current ex vivo CD8⁺ T-cell cross-reactivity for NP₄₁₈ variants using human PBMCs from eight donors, including those from individuals infected with the 2009-H1N1 pandemic strain. The availability of NP₄₁₈ variants reflecting 9 decades of natural selection has further allowed the identification of two distinct cross-reactive T-cell sets (specific for ER/EK or DK) that potentially cover the possible spectrum of TCR recognition. Thus, although ∼20 different NP₄₁₈ variants are known for seasonal and pandemic influenza strains (13), it may only be necessary to prime T-cell memory to two distinct sets of peptides with core critical motifs. In general, identification of such key solvent-exposed residues/motifs that elicit cross-reactive T-cell sets could allow the evolution of vaccination regimens that provide a measure of protection against unpredicted pandemic and seasonal influenza strains. The overall strategy might also be extended to other highly variable pathogens and to at least some cancers.

Materials and Methods

Donors and PBMC Isolation.

PBMCs were obtained from HLA-B*3501⁺, B*3503⁺, and B*0702 healthy donors (D1–D6) and B*0702⁺ patients hospitalized with A(H1N1)-2009 influenza virus infection confirmed by PCR (15) for patients D7 and D8. Cells were isolated from heparinized blood by Ficoll–Paque density centrifugation. HLA class I genotyping was performed by the Victorian Transplant and Immunogenetics Service (Parkville, Australia). Donor age was 47 (D1), 24 (D2), 27 (D3), 37 (D4), 59 (D6), 17 (D7), and 51 (D8) y at the time of blood collection. All the experiments were approved by the University of Melbourne and the Royal Melbourne Hospital Research Human Ethics Committees.

NP_418–426 Peptide Sequence Selection.

A panel of 12 unique NP₄₁₈ epitope variants was selected based on the alignment of 65 H1N1, H2N2, and H3N2 NP sequences obtained from the WHO Collaborating Centre for Reference and Research on Influenza and previously published (8) sequences (Table 1). The year and the strain name denote when the particular NP₄₁₈ variant first appeared according to the alignment (8).

T-Cell Restimulation and Intracellular Cytokine Assay.

Peptides representing naturally occurring NP₄₁₈ epitope variants (8) (Table 1) were purchased from Auspep. Autologous PBMCs (∼3 × 10⁶ cells) were pulsed with 10 μM peptide in 1 mL of serum-free medium RPMI 1640 (Gibco, Invitrogen) for 90 min at 37 °C and then washed with RPMI. Peptide-pulsed PBMCs were then incubated with autologous nonpeptide-pulsed PBMCs (6 × 10⁶ cells) for 10 d in cRPMI (RPMI supplemented with 2 mM l-glutamine, 1 mM sodium pyruvate, 100 mM nonessential amino acids, 5 mM Hepes buffer, 55 mM 2-mercaptoethanol, 100 U/mL penicillin, 100 mg/mL streptomycin) plus 10% vol/vol heat-inactivated FCS (all from Gibco), with rHuman IL-2 at 10 U/mL (Apollo) on day 4, and with a half-medium change every 2 d. At day 10 after restimulation, C1R cells transfected with HLA-B*3501 (kindly provided by Mandvi Bharadwaj, University of Melbourne, Melbourne, Australia) or autologous HLA-B*0702 PBMCs were pulsed with 10 μM peptide. Peptide-pulsed C1R-B35 cells (1 × 10⁵) were then incubated with restimulated PBMC samples (2 × 10⁵) for 2 h in the presence of 10 U/mL IL-2 in cRPMI. Diluted GolgiPlug (1:1,000; BD Biosciences) was then added and incubated for a further 5 h at 37 °C. Cells were washed with PBS/1% BSA/0.02% sodium azide and stained with anti-CD8α-allophycocyanin (BD Biosciences) for 30 min on ice and then washed. Using the BD Cytofix/Cytoperm Plus Fixation/Permeabilization Kit (BD Biosciences), cells were fixed and permeabilized and then stained with anti-IFN-γ–FITC (BD Biosciences) for 30 min on ice. Cells were analyzed by flow cytometry on a BD FACSCalibur (BD Biosciences) using FlowJo software (Tree Star, Inc., Ashland, OR). Background fluorescence of “no-peptide” controls was subtracted for the analysis.

Protein Expression, Purification, and Crystallization.

Soluble class I heterodimers containing the NP peptides were prepared as described previously (21). Crystals of the HLA-B*3501-NP₄₁₈ complexes were grown by the hanging-drop vapor-diffusion method at 20 °C with a protein/reservoir drop ratio of 1:1 at a concentration of 5 mg/mL in 10 mM Tris (pH 8) and 150 mM NaCl. Crystals grew using 15–25% PEG 4000, 0.2 M NH₄ acetate, and 0.1 M Na-citrate (pH 5.6).

Data Collection and Structure Determination.

The peptide HLA (pHLA) crystals were soaked in a cryoprotectant solution containing mother liquor solution with a PEG concentration increased to 30% (wt/vol) and then flash-frozen in liquid nitrogen. The data were collected on the 3BM1 and 3ID1 beamlines at the Australian Synchrotron facility (Clayton, Australia) using the ADSC Quantum 210 and 315 CCD detectors (at 100 K). Data were processed using XDS software and scaled using XSCALE software (22). The HLA-B*3501-NP₄₁₈ crystals (for all peptides) belonged to the space group P2₁2₁2₁ with unit cell dimensions (Table S3), consistent with one pHLA complex in the asymmetrical unit. The structure was determined by molecular replacement using the PHASER program (23) with the HLA-B*3501-HPVG for the MHC model without the peptide (Protein Data Bank ID code 2FYY) (24). Manual model building was conducted using Coot software (25), followed by maximum-likelihood refinement with the REFMAC 5 program (26). “Translation, liberation, and screw-rotation” displacement refinement was also used during the refinement process to model anisotropic displacements of defined domains. The final model has been validated using the Protein Database validation Web site, and the final refinement statistics are summarized in Table S3. All molecular graphics representations were created using PyMol (27).

ID Codes.

The coordinates of the HLA-B*3501-NP₄₁₈ complexes have been deposited in the Protein Data Bank under ID codes 3LKR (NP-2009), 3LKS (NP-1980), 3LKQ (NP-1977), 3LKP (NP-1972), 3LKO (NP-1934), and 3LKN (NP-1918).