T-Cell Tolerance for Variability in an HLA Class I-Presented Influenza A Virus Epitope

JOURNAL OF VIROLOGY, Sept. 2009, p. 9206–9214 Vol. 83, No. 180022-538X/09/$08.00�0 doi:10.1128/JVI.00932-09Copyright © 2009, American Society for Microbiology. All Rights Reserved.

T-Cell Tolerance for Variability in an HLA Class I-PresentedInfluenza A Virus Epitope�

Angela Wahl,1 William McCoy,2 Fredda Schafer,1 Wilfried Bardet,1 Rico Buchli,3Daved H. Fremont,2,4 and William H. Hildebrand1*

Department of Microbiology and Immunology, University of Oklahoma Health Sciences Center, 975 Northeast 10th Street,Oklahoma City, Oklahoma, 731041; Department of Pathology and Immunology2 and Department of Biochemistry and

Molecular Biophysics,4 Washington University School of Medicine, 660 South Euclid Avenue, St. Louis, Missouri 63110;and Pure Protein L.L.C., 800 Research Parkway, Suite 340, Oklahoma City, Oklahoma 731043

Received 11 May 2009/Accepted 21 June 2009

To escape immune recognition, viruses acquire amino acid substitutions in class I human leukocyte antigen(HLA)-presented cytotoxic T-lymphocyte (CTL) epitopes. Such viral escape mutations may (i) prevent peptideprocessing, (ii) diminish class I HLA binding, or (iii) alter T-cell recognition. Because residues 418 to 426 ofthe hypervariable influenza A virus nucleoprotein (NP418–426) epitope are consistently bound by class I HLAand presented to CTL, we assessed the impact that intraepitope sequence variability has upon T-cell recog-nition. CTL elicited by intranasal influenza virus infection were tested for their cross-recognition of 20 naturalNP418–426 epitope variants. Six of the variant epitopes, of both H1N1 and H3N2 origin, were cross-recognizedby CTL while the remaining NP418–426 epitope variants escaped targeting. A pattern emerged whereby vari-ability at position 5 (P5) within the epitope reduced T-cell recognition, changes at P4 or P6 enabled CTLescape, and a mutation at P8 enhanced T-cell recognition. These data demonstrate that substitutions at P4and/or P6 facilitate influenza virus escape from T-cell recognition and provide a model for the number, nature,and location of viral mutations that influence T-cell cross-recognition.

Cytotoxic T-lymphocytes (CTL) kill virus-infected cells andrelease antiviral cytokines upon recognition of short viral pep-tides displayed on the cell surface by the class I HLA molecule(36). Virus-derived peptides are processed in the cytoplasm byproteasome degradation of viral proteins (25), shuttled intothe lumen of the endoplasmic reticulum (ER) by the trans-porter-associated protein, and loaded into the basket-likegroove of the class I molecule. Class I HLA molecules awaitpeptide loading in the ER and demonstrate specificity for viralpeptides with particular anchor residues representing a goodfit for the class I HLA binding groove. Once stable class IHLA-peptide complexes are formed, the class I molecule andits peptide cargo are transported via the Golgi apparatus to thecell surface, where the complex is anchored to the plasmamembrane (21, 36–38). CTL then survey class I HLA-pre-sented peptides on the cell surface. Viral peptides must there-fore be processed, specifically bound by class I HLA, andpresented at the plasma membrane for CTL to distinguishinfected cells from uninfected tissue.

A high mutation rate is one of many mechanisms utilized byviruses to escape detection by the immune system. Mutationswithin the genome allow viruses to accumulate and select foramino acid substitutions that (i) inhibit proteasome processingand viral peptide generation (2, 23), (ii) alter anchor residueswithin viral peptides to diminish class I HLA binding specificity(3, 14, 24, 32), or (iii) reduce immune recognition of the class

I HLA-peptide complex by varying amino acids that come incontact with the T-cell receptor (6, 10, 27, 30, 35). While viralmutations might be advantageous for escaping immune detec-tion, such flexibility can cost the virus in terms of replicativefitness. In order to maintain reproductive fitness and structuralintegrity, viruses must temper their use of genetic flexibility asa means of immune escape.

Influenza viruses have the well-documented ability to escapedetection by various immune epitopes (3, 10, 27). A priori,investigators often assume that variable regions of the virusrepresent poor immune targets because such regions will notbe consistently processed, presented, or recognized (15, 20).However, we along with others continue to find that a hyper-variable stretch of the influenza virus nucleoprotein consistingof residues 418 to 426 (NP418–426) is presented to CTL bydifferent HLA-B alleles (B*0702 and B*3501) in spite of ex-tensive viral variability within this epitope (8, 10, 27, 34). More-over, NP418–426 is a dominant immune epitope (8, 10, 27, 34).The consistent processing and presentation of NP418–426 byclass I HLA can be explained by the finding that differentinfluenza virus isolates cannot mutate the proline located atposition 2 (P2) within the epitope because elimination of thisproline reduces viral fitness (4, 5). Little to no variability isfound at the methionine P9 anchor as well. These facts lead tothe unique observation that strain-to-strain variability does notabrogate class I HLA presentation of the influenza virusNP418–426 epitope and that CTL respond to this consistentlypresented viral epitope in an immunodominant fashion.

In this study we took advantage of the anchor residue con-servation that prompts the NP418–426 epitope to be consistentlypresented to CTL by investigating the functional impact thatinfluenza virus intraepitope variability has on CTL recognition.

* Corresponding author. Mailing address: Department of Microbi-ology and Immunology, University of Oklahoma Health Sciences Cen-ter, 975 NE 10th Street, Oklahoma City, OK 73104. Phone: (405)271-1203. Fax: (405) 271-3117. E-mail: [email protected].

� Published ahead of print on 24 June 2009.

9206

The amino acid alignment of human influenza A (H1N1 andH3N2) virus nucleoprotein molecules identifies 20 uniqueNP418–426 peptide sequences which demonstrate amino aciddiversity between the anchors. We infected HLA-transgenicmice intranasally with influenza virus and tested CTL fromthese animals for their ability to recognize each of the 20NP418–426 variants. These 20 NP418–426 sequences represent anatural “recombinant library” of viral epitopes that the im-mune system has and will face. The resulting data demonstratea gradient of viral substitutions whereby CTL recognition di-minishes depending upon the number of viral substitutions andtheir location within the epitope. Understanding how intra-epitope variability impacts CTL recognition is discussed interms of eliciting immune responses to variants of influenza.

MATERIALS AND METHODS

Amino acid alignments. Full-length sequences of human influenza A (H1N1and H3N2) virus NP were initially accessed from the NCBI Influenza VirusResource database on 26 November 26 2007 (1). Following removal of identicalprotein sequences, the NP sequences were aligned, identifying 20 unique aminoacid sequences at positions 418 to 426 of the NP molecule. Synthetic peptideswere generated for the 20 NP418–426 amino acid sequences and single amino acidNP418–426 variants at �95% purity by the Molecular Biology Research Facility atthe University of Oklahoma Health Sciences Center and Mimotopes and utilizedin enzyme-linked immunospot (ELISPOT) and peptide binding assays.

A second amino acid alignment was performed on human influenza A (H1N1and H3N2) virus NP full-length protein sequences accessed from the NCBIInfluenza Virus Resource database on 17 April 2008 revealing an additionalhuman influenza A (H3N2) virus NP418–426 sequence (LPFEKPTVM) (1).

Influenza PR8 virus infection of mice. All animal studies were approved by theInstitutional Animal Care and Use Committee at the University of OklahomaHealth Sciences Center. Six- to eight-week-old male and female HLA-B*0702transgenic H-2KbDb double-knockout C57BL/6 mice (obtained from the labo-ratory of Francois Lemonnier at the Institut Pasteur) were inoculated intrana-sally with a 1,000 50% egg infectious dose (EID50) of mouse-adapted influenzaA/Puerto Rico/8/34 (PR8) virus (kindly provided by the laboratory of DavidWoodland at the Trudeau Institute) or sterile endotoxin-free saline (mock in-fection) at a volume of 10 �l per nostril. Mice were monitored daily for weightloss. Splenocytes were harvested at 12 days postinfection and passed through acell strainer, and lymphocytes were isolated by centrifuging cells over a lym-pholyte-M density gradient (Cedar Lane). Lymphocytes were stored at �180°Cin 90% fetal bovine serum and 10% dimethyl sulfoxide.

Lymphocytes were measured for peptide specific gamma interferon (IFN-�)secretion via ELISPOT assay. Briefly, 105 lymphocytes were incubated with10 �g/ml synthetic influenza virus peptide, medium alone (negative control), 10�g/ml synthetic human immunodeficiency virus Nef peptide RPMTYKAAL(negative control), and 4 �g/ml concanavalin A (positive control) in RPMI 1640culture medium supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin for 24 h at 37°C in triplicate wells of a 96-well membrane (polyvi-nylidene difluoride)-bottomed plate coated with anti-IFN-� antibody (Cell Sci-ences). The assay was performed according to the manufacturer’s instructions.The number of IFN-� spots produced per well was enumerated with a Zeiss KSELISPOT reader. The data generated are illustrated as the number of spot-forming units (SFU) per 105 lymphocytes.

Peptide binding assay. An HLA-B*0702 PolyScreen kit (Pure Protein) wasused to determine peptide concentrations necessary to inhibit 50% polarization(IC50). Briefly, fluorescein isothiocyanate-labeled reference peptide and solubleHLA (sHLA) were incubated with each peptide until equilibrium of peptidereplacement was reached. The fluorescent polarization of the control peptide asread on an Analyst AD plate reader (Molecular Devices) and a dose-responsecurve were used to calculate peptide IC50 values (11–13).

Molecular modeling. HLA-B structures from the Protein Data Bank (PDB)were first restricted to those bound to nonamer peptides. When multiple struc-tures were available for a single HLA-B allele, the highest-resolution structurewas selected. Five structures (1XR8-HLA-B*1501 [26], 1K5N-HLA-B*2709 [16],2CIK-HLA-B*3501 [18], 1A1O-HLA-B*5301 [29], and 2BVP-HLA-B*5703[31]; PDB codes precede the HLA designations) were loaded into InsightII(Accelrys), and each of the selected structures was used to build a model ofHLA-B*0702 bound to the PR8 peptide LPFDRTTVM via the homology and

consensus module. This module threads the unknown structure’s amino acidsequence through the known structure without altering the position of the pep-tide backbone. Side chain clashes were minimized by varying rotamers. Thesemodels then underwent visual inspection of the PR8 peptide to eliminate sidechain clashes, which were then minimized via manual or auto-rotamer adjust-ments of peptide or HLA groove residues. Models were compared to identifygeneral trends in peptide conformation and side chain placement. The modelthat best accommodated the HLA-B*0702 PR8 sequence (1K5N-HLA-B*2709)(16) with a minimum of unfavorable peptide-HLA interactions was then ex-ported as a PDB file. The final figure was generated using this PDB file importedinto MacPyMOL (DeLano Scientific LLC).

RESULTS

The amino acid sequence of NP418–426 is highly variable ininfluenza A H1N1 and H3N2 virus strains. Data demonstrat-ing that an epitope is presented by class I HLA in spite of viraldivergence are scarce. Our direct epitope discovery data (Fig.1), combined with CTL-driven studies in other laboratories,provide the rare demonstration that HLA-B molecules consis-tently present the NP418–426 epitope regardless of influenzavirus strain-to-strain variability (9, 10, 27, 33, 34). Influenzavirus variability does not impact proteolytic processing, trans-location into the ER, or loading into class I HLA. Given steadyepitope presentation, we took advantage of the considerableamino acid variability within NP418–426 of different influenza Avirus subtypes to generate “nature’s” recombinant NP418–426

epitope library for subsequent testing of CTL tolerance tovarious intraepitope substitutions. The data are positioned toprovide fundamental insights for how efficiently CTL cope withviral epitope diversification.

We began with an amino acid alignment of influenza A(H1N1 and H3N2) virus NP molecules from human isolatesfrom 1918 to 2007 to assess the variety and nature of NP418–426

peptide sequences among influenza A virus subtypes (1).Alignments were compared to the PR8 virus. The alignmentrevealed 20 NP418-426 variant sequences (13 of subtype H1N1and 11 of subtype H3N2) (Table 1). Four NP418-426 peptidesequences (LPFDRTTIM, LPFERATVM, LPFDKSTVM,and LPFDKSTIM) are apparent in both H1N1 and H3N2influenza virus strains. The influenza A H1N1 virus isolatesvaried at P4 to P6 and at P8 within the NP418-426 epitope, andsubtype H3N2 isolates had mutated amino acids at P4 to P9. P6exhibited the greatest diversity in amino acids accommodatedat a particular position within the NP418-426 ligand for bothinfluenza A H1N1 and H3N2 virus isolates although H1N1 andH3N2 strains displayed a tendency for different amino acids atthis position (in subtype H1N1, Thr; in subtype H3N2, Ser). Inaddition, influenza A H1N1 virus isolates demonstrated a pro-pensity for Asp at P4 and Ile at P8 while reported influenza AH3N2 virus strains favored Glu at P4 and equally displayed Ileand Val at P8. Based on the frequency of particular aminoacids at P4 to P8 within the NP418-426 epitope, influenza AH1N1 and H3N2 virus isolates exhibit the consensus sequenceLPFDKTTIM (H1N1) and LPFEKST(I/V)M (H3N2) (substi-tutions are underlined). These data demonstrate the following:(i) the breadth of NP418-426 peptide sequences among humaninfluenza A H1N1 and H3N2 virus isolates; (ii) the substantialvariability at internal amino acid positions within the NP418-426

ligand; and (iii) despite 40 years of concurrent circulationwithin the human population, the segregation of the assort-

VOL. 83, 2009 T-CELL TOLERANCE FOR INFLUENZA VIRUS ESCAPE 9207

ment of NP418-426 peptide sequences at P4, P6, and P8 betweenhuman influenza A H1N1 and H3N2 strains.

NP418–426 variants exhibit various degrees of CTL cross-reactivity during PR8 virus infection of HLA-B*0702 trans-genic mice. Upon identification of 20 NP418–426 peptide se-quences among influenza A H1N1 and H3N2 strains, we testedsplenocytes isolated from influenza PR8 (H1N1) virus-infectedHLA-B*0702 transgenic H-2Kb�/�Db�/� mice for NP418–426-specific T-cell cross-reactivity measured by IFN-� production.Ten HLA-B*0702 transgenic mice were inoculated intranasallywith 1,000 EID50 of murine-adapted PR8 and monitored dailyfor weight loss (data not shown), and splenocytes were isolated12 days postinfection. Synthetic peptides were generated for 20human influenza A virus NP418–426 sequences (nine H1N1strain, seven H3N2 strain, and four H1N1/H3N2 strain se-quences) (Fig. 2 and Table 1) and incubated at 10 �g/ml with105 lymphocytes for 24 h in triplicate wells of an anti-IFN-�antibody-coated plate. Lymphocytes incubated with medium

only and with 10 �g/ml HLA-B*0702 human immunodefi-ciency virus Nef peptide RPMTYKAAL (data not shown)served as negative controls, and lymphocytes cultured with4 �g/ml concanavalin A served as a positive control.

The parental PR8 NP418–426 epitope (LPFDRTTVM) andsix NP418–426 variant peptides (LPFDRTTIM, LPFDKTTIM,LPFDRPTIM, LPFDKSTIM, LPFDKSTVM, and LPFDKATIM) were clearly recognized, eliciting an average of at least20 SFU/105 lymphocytes in 10 PR8 virus-infected mice (Fig. 2and Table 1). Of these six NP418–426 variant peptides, threeNP418–426 peptides are derived from H1N1 influenza A virusstrains, and three NP418–426 ligands have been identified inboth influenza A H1N1 and H3N2 strains. One NP418–426 vari-ant, LPFDRTTIM, exhibited greater IFN-� T-cell reactivitythan PR8 NP418–426 LPFDRTTVM, generating an average of168.5 and 103.6 SFU/105 lymphocytes, respectively. ThirteenNP418–426 variant peptides (five H1N1 strain, seven H3N2strain, and one H1N1/H3N2 strain peptides) generated be-

FIG. 1. Mass spectrometric identification of NP418–426 during infection with three influenza A virus strains. Class I HLA peptides eluted fromnaïve and influenza virus-infected HeLa cells were separated by reverse-phase high-pressure liquid chromatography, and fractions were sprayedvia nanospray into a Q-Star Elite mass spectrometer. Naïve (data not shown) and infected mass spectrometric (MS) ion maps were aligned andvisually compared to identify ions unique to the infected MS spectra. Despite tremendous amino acid variability within the influenza virus NP418–426epitope, the NP418–426 peptide was eluted from sHLA-B*0702-transfected HeLa cells and identified during infection with three influenza virus Astrains: PR8 (H1N1) (A), A/Oklahoma/7485/01 (7485; H1N1) (B), and A/Oklahoma/309/06 (309; H3N2) (C). In addition, the NP418–426 peptidewas eluted from sHLA of 309-infected HeLa cells transfected with sHLA-B*3501, a HLA-B7 supertype allele (34) (D).

9208 WAHL ET AL. J. VIROL.

tween 4 and 10 SFU/105 lymphocytes during murine PR8 virusinfection, including the NP418–426 variant found in the newlyemerged H1N1 flu (swine flu) strain (LPFERATVM). AllNP418–426 peptide sequences restricted to H3N2 influenza Avirus strains elicited an average IFN-� T-cell response below10 SFU/105 lymphocytes (Table 1).

In Table 1 H1N1, H1N1/H3N2, and H3N2 NP418–426 peptidesequences were aligned to the PR8 peptide (LPFDRTTVM) inascending order by the average number of IFN-� spots (SFU/105 lymphocytes) for 10 PR8 virus-infected HLA-B*0702transgenic mice, indicating amino acid positions involved in thePR8 NP418–426-specific T-cell response. Substantial NP418–426-specific T-cell cross-reactivity was observed only for NP418–426

variants derived from either influenza A H1N1 strains orH1N1/H3N2 strains. All NP418–426 peptides whose sequenceswere limited to influenza A H3N2 strains failed to generate asizeable IFN-� T-cell response.

Amino acid substitutions at P4 to P6 decreased IFN-� T-cellreactivity during murine PR8 virus infection while exchangingVal at P8 for Ile increased the IFN-� ELISPOT response(Tables 1 and 2). IFN-� ELISPOT responses of PR8 virus-infected HLA-B*0702 transgenic mice to NP418-426 variantpeptides indicate that amino acids at P4 to P6 and at P8 dictateantigen-specific T-cell cross-reactivity. The majority of aminoacid substitutions at these positions decrease IFN-� T-cell re-activity; only a Val 3 Ile substitution at P8 increased theNP418-426-specific IFN-� ELISPOT response. Tremendousamino acid variability is observed at P6 of the NP418-426

epitope. At P6 variant NP418-426 peptides exhibit both conser-vative (Thr 3 Ser) and nonconservative (mutations of Thr toAla, Pro, Gln, and Ile) amino acid substitutions. Considerablyless amino acid variability is observed at P4, P5, and P8 interms of the variety of different amino acids detected at thesepositions and the conservative nature of amino acid substitu-tions (Asp3Glu at P4, Arg3 Lys at P5, and Val3 Ile at P8).

The data generated indicate that both conservative and non-conservative amino acid substitutions within the NP418-426

epitope can impact T-cell recognition and that the influenzavirus exhibits the most flexibility at P6.

NP418–426-specific CTL cross-reactivity is not correlatedwith HLA-B*0702 binding affinity. We next determinedwhether the binding affinity of NP418–426 variant peptides toHLA-B*0702 correlated with the NP418–426-specific IFN-� T-cell responses observed above; it is possible that NP418–426

peptides exhibiting a higher affinity for HLA-B*0702 may bepresented on the surfaces of target cells in greater abundancefor CTL recognition. Therefore, we measured the relativeHLA-B*0702 binding affinity of the 20 naturally occurringNP418–426 synthetic peptides in a competitive binding assay(11–13). Various concentrations of synthetic NP418–426 peptidewere incubated with sHLA-B*0702 and an HLA-B*0702 fluo-rescein isothiocyanate-labeled reference peptide, and theIC50 of synthetic NP418–426 was recorded. Table 1 illustratesthe relative HLA-B*0702 binding affinity of the syntheticNP418–426 variant peptides [high-affinity binders, log(IC50 nM)of �3.7; medium-affinity binders, log(IC50 nM) of 3.7 to 4.7,low-affinity binders, log(IC50 nM) of 4.7 to 5.5; very-low-affinitybinders, log(IC50 nM) of �6.0]. The PR8 NP418–426 peptideLPFDRTTVM had the highest binding affinity for HLA-B*0702 [1.8 log(IC50 nM)]. The average HLA-B*0702 bindingaffinity for variants generating at least 10 SFU/105 lymphocytesduring murine PR8 virus infection is equivalent to NP418–426

variants which elicited, on average, less than 10 SFU/105 lym-phocytes, indicating that the binding affinity of the NP418–426

variant peptides did not correspond to a gain or loss in T-cellcross-reactivity. These data demonstrate that all variants testedhave a high binding affinity and suggest that T-cell cross-reac-tivity to variant NP418–426 peptide sequences is heavily influ-enced by interaction with P4 to P6 and P8 of NP418–426.

TABLE 1. Relative binding affinity of influenza A virus NP418–426 peptides to sHLA-B*0702

Sequence of the NP418–426 peptide at the indicated positionaInfluenza A virus

strainAvg IFN-� T-cell response

(SFU/105 lymphocytes)bBinding affinity�log(IC

50nM)�1 2 3 4 5 6 7 8 9

L P F D R T T V M H1N1 103.69 1.85— — — — K — — I — H1N1 71.9 2.284— — — — — P — I — H1N1 38.89 3.898— — — — K A — I — H1N1 19.93 3.649— — — E K — — I — H1N1 6.13 3.076— — — E — S — I — H1N1 5.8 3.51— — — — K I — I — H1N1 1.43 2.682— — — E — A — I — H1N1 1.23 2.843— — — G K — — I — H1N1 0.33 2.945— — — — — — — I — H1N1/H3N2 168.56 2.257— — — — K S — I — H1N1/H3N2 35.93 3.54— — — — K S — — — H1N1/H3N2 26.83 3.117— — — E — A — — — H1N1/H3N2 0.1 3.335— — — — K P — I — H3N2 9.1 2.525— — — E — S — — — H3N2 6.26 5.174— — — E K S — I — H3N2 4.93 3.891— — — E K S — — — H3N2 4.36 3.053— — — — K Q — I — H3N2 1.8 4.667— — — E K S I — — H3N2 0.13 2.694— — — E — A — I I H3N2 0.06 3.073

a The sequence of the reference NP418–426 peptide from the PR8 virus is shown in boldface. Unchanged residues are indicated by dashes.b Calculated from the number of SFU/105 lymphocytes for 10 PR8-infected mice.


Amino acid residues at P4 to P6 and P8 within the NP418–426

epitope dictate CTL recognition. Naturally occurring variantsof NP418–426 often differ by multiple amino acid substitutions,making it difficult to ascertain how a given amino acid substi-tution or position within the epitope contributes to T-cell rec-ognition. In order to unravel the individual impact that natu-rally occurring NP418–426 amino acid substitutions have upon

T-cell cross-reactivity following PR8 virus infection, we synthe-sized epitopes differing by a single amino acid from the paren-tal NP418–426. Lymphocytes isolated from the spleens of nineHLA-B*0702 transgenic mice inoculated intranasally with PR8virus were assessed for IFN-� ELISPOT reactivity to NP418–426

peptides containing a single amino acid substitution at P4 to P6with respect to the sequence of PR8 NP418–426 (Fig. 3A and B;

FIG. 2. IFN-� ELISPOT reactivity of 10 PR8-infected HLA-B*0702 transgenic mice to 20 NP418–426 variant peptides. IFN-� ELISPOTresponses are illustrated for each mouse in SFU/105 lymphocytes. (A) The parental PR8 NP418–426 peptide (LPFDRTTVM) generated an averageof 103.6 SFU/105 lymphocytes. The T-cell cross-reactive IFN-� response elicited to each H1N1 (B), H1N1/H3N2 (C), and H3N2 (D) variantNP418–426 peptide was compared to the parental PR8 NP418–426 peptide (shaded in gray). Three H1N1 restricted NP418–426 peptides (LPFDKTTIM,LPFDRPTIM, and LPFDKATIM) and three H1N1/H3N2 variants (LPFDRTTIM, and LPFDKSTIM, and LPFDKSTVM) elicited an average ofat least 10 SFU/105 lymphocytes (shown in red).

TABLE 2. Relative binding affinity of PR8 NP418–426 single mutant peptides to sHLA-B*0702

Sequence of the NP418–426 peptide at the indicated positionaAvg IFN-� T-cell response

(SFU/105 lymphocytes)bBinding affinity�log(IC

50nM)�1 2 3 4 5 6 7 8 9

L P F D R T T V M 161.4 1.85— — — E — — — — — 5.1 1.83— — — G — — — — — 3.6 1.6— — — — K — — — — 20.1 2.2— — — — — S — — — 122.4 1.9— — — — — A — — — 68.4 1.96— — — — — P — — — 50.5 1.96— — — — — I — — — 13.9 2.02— — — — — Q — — — 7 1.98— — — — — — I — — 118.4 1.92— — — — — — — — I 46.4 1.98

a The sequence of the reference NP418–426 peptide from the PR8 virus is shown in boldface. Unchanged residues are indicated by dashes.b Average for nine (P4, P5, and P6) or four (P7 and P9) PR8 virus-infected mice.


Table 2). The introduction of these single amino acid substi-tutions into the PR8 NP418–426 epitope had a negligible impacton HLA-B*0702 binding (Table 2). Surprisingly, at P4 bothconservative (Asp 3 Glu) and nonconservative (Asp 3 Gly)amino acid substitutions eliminated PR8 NP418–426 IFN-�ELISPOT cross-reactivity. On average, the LPFERTTVM andLPFGRTTVM (substitutions are underlined) synthetic pep-tides generated fewer than 10 SFU/105 lymphocytes. At P5, theconservative Arg 3 Lys amino acid substitution dramaticallyreduced IFN-� ELISPOT reactivity but did not completelyeliminate CTL recognition. Splenocytes incubated with theLPFDKTTVM synthetic peptide generated an average of 20.1SFU/105 lymphocytes in comparison to PR8 NP418–426, whichgenerated an average of 161.4 SFU/105 lymphocytes. As statedabove, naturally occurring H1N1 and H3N2 NP418–426 variantsexhibit the most variability at P6. The five P6 single amino acidNP418–426 variants varied in IFN-� ELISPOT reactivity. Theconservative Thr 3 Ser substitution did not substantially in-crease or decrease CTL recognition while the nonconservative

substitutions of Ala, Pro, and Ile for Thr drastically decreasedIFN-� ELISPOT reactivity, and the Thr 3 Gln variant virtu-ally eliminated IFN-� ELISPOT reactivity. Substitutions at P4and P6 substantially altered the cross-recognition of naturalNP418–426 variants.

Although the majority of the 20 naturally occurring H1N1and H3N2 NP418–426 variant sequences exhibit variability atamino acids at P4 to P6 and P8, a conservative amino acidsubstitution is found at P7 and P9 in one variant each. TheIFN-� ELISPOT response of lymphocytes isolated from thespleens of four PR8 virus-infected HLA-B*0702 transgenicmice indicates that the Thr 3 Ile substitution at P7 does notsubstantially alter CTL recognition while the Met3 Ile aminoacid substitution results in decreased IFN-� ELISPOT reactiv-ity (Fig. 3C and D; Table 2). The paucity of NP418–426 variantscontaining a P7 Thr3 Ile substitution in nature is most likelydue to the minimal impact of such a substitution on CTLcross-recognition. In contrast, although a Met 3 Ile aminoacid substitution at P9 decreases NP418–426 cross-reactivity

FIG. 3. IFN-� ELISPOT reactivity of PR8 virus-infected HLA-B*0702 transgenic mice to the PR8 NP418–426 variant peptide containing singleamino acid substitutions at P4 to P9. IFN-� ELISPOT responses of splenocytes isolated from mice infected with 1,000 EID50s of PR8 virusgenerated against the P4 to P9 single mutant influenza A virus NP418–426 peptides are illustrated for each mouse in SFU/105 lymphocytes. (A) NinePR8 virus-infected mice generated an average of 161.4 SFU/105 lymphocytes to the parental PR8 NP418–426 peptide (LPFDRTTVM). The T-cellcross-reactive IFN-� response elicited to synthetic PR8 NP418–426 peptides containing a single amino acid substitution at P4 to P6 (B) was comparedto the parental PR8 NP418–426 peptide (shaded in gray). The IFN-� ELISPOT reactivity of lymphocytes isolated from the spleens of fourPR8-infected mice to the PR8 NP418–426 (average, 137.9 SFU/105 lymphocytes) peptide (C) was compared to PR8 NP418–426 synthetic peptidescontaining a single amino acid substitution at P7 and P9 (D). NP418–426 single mutant peptides generating an average of greater than 10 SFU/105

lymphocytes are shown in red, and synthetic peptides that elicited less than 10 SFU/105 lymphocytes are shown in black.


(Fig. 3C and D; Table 2) and has not been shown to alter viralfitness (5), this substitution is not common in influenza H1N1and H3N2 virus NP molecules (Fig. 4).

These data indicate that amino acid residues at P4 to P6 andP8 of the NP418–426 epitope mediate CTL recognition. Aminoacid substitutions located at P4 eliminated IFN-� ELISPOTcross-reactivity while variability at P5 drastically reduced CTLrecognition. Conservative variability at P6 did not substantiallyimpact CTL cross-reactivity while nonconservative amino acidsubstitutions substantially decreased or virtually eliminatedIFN-� ELISPOT reactivity. Influenza virus strains containingthe P8 Val 3 Ile single amino acid substitution have beenisolated in the human population and were shown to haveincreased IFN-� ELISPOT reactivity in the previous section(Fig. 2 and Table 1). CTL cross-recognition is largely mediatedby the position and less by the nature of the amino acid sub-stitution(s).

Molecular modeling of the HLA-B*0702 PR8 NP418–426 pep-tide complex. The data presented thus far indicate that aminoacids at P4 to P6 and P8 within the NP418–426 ligand are keydeterminants of NP418–426-specific T-cell cross-reactivity. It isplausible that amino acids at these positions may influenceT-cell cross-reactivity by directly contacting the T-cell receptoror by altering the conformation of the NP418–426 peptide in theclass I HLA-B*0702 binding groove. To address the orienta-

tion of NP418–426 amino acid residues at P4 to P6 and P8, thePR8 NP418–426 ligand (LPFDRTTVM) was modeled in theHLA-B*0702 binding groove (Fig. 4). The PDB structuresof five HLA-B molecules (HLA-B*2709, -B*1501, -B*3501,B*5301, and -B*5703) (16, 18, 26, 29, 31) were loaded intoInsightII, and each structure was used to model the HLA-B*0702 LPFDRTTVM peptide complex via the homology andconsensus module. Without altering the position of the peptidebackbone, side chain clashes were minimized by manual orauto-rotamer adjustments of the peptide or HLA bindinggroove residues. The model that most favorably accommo-dated the HLA-B*0702 LPFDRTTVM sequence was based onthe structure of HLA-B*2709 (PDB 1K5N) (16). Molecularmodeling indicates that amino acids at P4, P5, and P8 aresolvent exposed and point away from the peptide binding cleftwhile amino acids at P6 are directed toward the alpha-1 helixof the binding groove. Therefore, amino acid substitutions atP4, P5, and P8 are poised to dictate T-cell cross-reactivity bydirectly interacting with the T-cell receptor (TCR). This mayexplain why both nonconservative and conservative amino acidsubstitutions at these positions dramatically impact NP418–426-specific IFN-� ELISPOT cross-reactivity in the PR8 murinemodel. In contrast, a nonconservative substitution at P6 ismore likely to shift the conformation of the NP418–426 peptidein the HLA-B*0702 binding groove by interacting with resi-dues of the alpha-1 helix, indirectly altering T-cell recognition.

DISCUSSION

CTL responses decrease morbidity and mortality and arekey to viral clearance during human influenza A virus infec-tion. As a foil to CTL detection, amino acid substitutions inpeptide anchor positions and at CTL receptor contact residueshinder influenza virus class I HLA presentation and CTL rec-ognition, respectively (3, 10, 27). Over time, influenza A vi-ruses have emerged with anchor position substitutions inepitopes NP380–383/HLA-B*0801 and NP383–391/HLA-B*2705that act to abolish class I HLA binding. Amino acid variabilityat P4 to P9 of the immunodominant HLA-B*0702 and -B*3501influenza A virus NP418–426 epitope has been reported to mod-ulate but not abolish recognition by CTL clones specific forolder/newer influenza virus strains. Our laboratory has sub-stantiated by direct peptide elution that hypervariability doesnot inhibit presentation of the NP418–426 epitope for three viralstrains and two HLA-B alleles (34). Epitope NP418–426 repre-sents a natural recombinant system that maintains processing,binding, and presentation so that the impact of intraepitopevariability on CTL targeting can be examined. To our knowl-edge, no one has identified a consistently presented hypervari-able epitope such that the give and take between CTL andnatural viral mutations can be assessed.

The IFN-� ELISPOT reactivity of 20 natural NP418–426 in-fluenza virus variants demonstrated that amino acid substitu-tions at P4 to P6 and P8 impact T-cell recognition. The con-tribution of individual amino acid substitutions at P4 to P9 ofthe NP418–426 peptide was determined by measuring the IFN-�ELISPOT reactivity of lymphocytes isolated from the spleensof PR8 virus-infected HLA-B*0702 transgenic mice incubatedwith synthetic PR8 NP418–426 peptides containing single aminoacid substitutions. Data revealed that conservative and non-

FIG. 4. Molecular modeling of the HLA-B*0702 PR8 NP418–426peptide complex. The orientation of the PR8 NP418–426 peptide (LPFDRTTVM) in the HLA-B*0702 binding groove was determined bymolecular modeling. The position of LPFDRTTVM amino acid resi-dues in the peptide binding groove are designated as follows: 1, upfrom the peptide binding groove;2, down toward the peptide bindinggroove; 3, toward the alpha-1 helix; 4, toward the alpha-2 helix; ,anchor residue. Molecular modeling indicates that amino acid residuesat P4, P5, and P8 of the peptide point up from the peptide bindinggroove while residues at P6 point toward the alpha-1 helix. Singleamino acid substitutions at P4 to P9 increase, decrease, or produce nochange in IFN-� ELISPOT reactivity during PR8 murine infection.


conservative amino acid variability at P4 eliminated CTL rec-ognition. Likewise, conservative amino acid substitutions at P5and P8 were able to dramatically modulate the NP418–426-specific IFN-� CTL response. The conservative Arg3 Lys P5substitution substantially decreased CTL IFN-� cross-reactiv-ity while the Val 3 Ile P8 substitution increased CTL recog-nition following PR8 infection. In contrast, only nonconserva-tive P6 substitutions impacted CTL cross-recognition. Theconservative Thr3 Ser P6 substitution yielded no substantialin the NP418–426-specific IFN-� CTL response.

We were somewhat surprised that conservative biochemicalsubstitutions such as Asp 3 Glu at P4 or a P5 Arg 3 Lysdiminished recognition by the TCR. One can envision a sce-nario whereby a P4 Asp 3 Glu substitution would not cost apathogen much in terms of viral fitness or protein structure, yetthis conservative substitution would provide the virus with alevel of T-cell escape. The intraepitope location of a substitu-tion is also key, and these data illustrate the dominant role thatP4 plays in NP-specific immunity. Here, we see that subtleamino acid substitutions, when positioned correctly, can dimin-ish immune recognition of a pathogen-derived epitope.

The results obtained here must be interpreted with cautionas they may differ from studies performed with high-affinityT-cell clones. At 12 days postinfection, lymphocytes isolatedfrom the spleens of infected mice represent a polyclonal pop-ulation of T cells that contains few, if any, T cells with high-affinity TCRs. Also, our data using antigen-specific T cellsobtained from the spleen may vary from that obtained from Tcells harvested from draining mediastinal lymph nodes. Finally,these experiments must be repeated using T cells (preferablyfrom the lungs) of HLA-B*0702-positive influenza virus-in-fected patients to confirm the patterns of NP418-426 T-cellcross-reactivity observed in mice.

Molecular modeling of the NP418-426 peptide in context ofthe HLA-B*0702 binding groove provides a model for howimmune receptors interact with this epitope. Amino acids lo-cated at P4, P5, and P8 are solvent exposed and available fordirect interaction with the TCR. In contrast, residues at P6 aredirected toward the alpha-1 helix of the peptide binding grooveand are less accessible to the TCR. Nonconservative aminoacid substitutions at P6 may slightly alter the conformation ofthe peptide in the HLA-B*0702 peptide binding cleft, but it isunlikely that P6 changes directly interact with the TCR. Thesedata suggest that viruses modulate CTL recognition by accu-mulating one or two conservative amino acid substitutions atpeptide positions in direct contact with the TCR. However, itshould be noted that the majority of naturally occurring humanH1N1 and H3N2 NP418-426 peptide sequences that influenceCTL recognition exhibit variability at two or more peptidepositions. Factors that influence viral mutagenesis must extendwell beyond the selective pressures exerted by T-cell recogni-tion.

It is interesting that during 40 years of concurrent circulationin the human population, influenza A H1N1 and H3N2 virusstrains have evolved a propensity for different amino acids atP4 of the NP418-426 epitope (for H1N1, Asp; for H3N2, Glu).Aspartic acid has not been reported at P4 of the NP418-426

sequence in influenza A H3N2 strains isolated in the UnitedStates since 1978, 10 years after the introduction of the H3N2subtype into the human population in 1968. Glutamic acid has

not been absent at P4 in H1N1 isolates from the United States,its appearance in human circulation has been sporadic at best(1). As demonstrated in this study, variability at P4 eliminatesIFN-� CTL cross-reactivity. Based upon these observations, wehypothesize that viral strain specificity at P4 deters CTL cross-reactivity between the H1N1 and H3N2 influenza A virus sub-types and that amino acid variability at P5, P6, and P8 modu-lates CTL recognition within a subtype.

We have so far focused upon viral substitutions within, andCTL recognition of, an influenza virus NP epitope. What canwe learn from HLA molecules of the B7 supertype that inter-act with this epitope? HLA-B*0702 is found in roughly one-fourth of the population (26% and 16% of the Caucasian andAfrican American populations of the United States, respec-tively), and presentation by other members of the B7 supertypefamily would extend presentation of NP418-426 to one-third ormore of the population (22, 28). Indeed, we find that epitopeNP418-426 is presented by HLA-B*3501, a member of the B7supertype and an allele found at a phenotype frequency of 12%and 10% in Caucasian and African American populations ofthe United States, respectively (22). Additional experimentswill need to confirm presentation by all alleles within the B7supertype, yet the extraordinary variability that influenza virusstrains exhibit within this epitope (others have classified thisregion as hypervariable) (8) certainly suggests that the cellularimmune system has focused considerable pressure on this seg-ment of the virus. It is important for vaccine architects torecognize that variable sequences bracketed by conserved an-chors can indicate robust class I HLA-presented immune tar-gets.

In summary, observations including direct HLA-B*0702epitope elution following infection by divergent viral strainsjustifies a detailed characterization of the consistently pre-sented influenza A virus epitope NP418-426. This particularepitope cannot avoid class I HLA-B7 presentation, and CTLcross-recognition experiments demonstrate that influenza virusutilizes substitutions at P4 to P6 and P8 within this epitope todiminish and/or abrogate immune targeting. The natural vari-ants tested indicate that the virus is constrained in its ability tosubstitute amino acids, yet by accumulating one or two con-servative intraepitope substitutions, the influenza virus is ableto thwart cross-targeting of NP418-426. These data additionallyshow that particular substitutions do not disrupt CTL recog-nition, which is useful information for the design of CTL-eliciting therapeutics. Finally, it is conceivable that other classI HLA-B alleles present immunodominant epitopes from re-gions of variability; the HLA-B locus plays a critical role incontrolling several viruses (7, 9, 17, 19), and future testing isneeded to determine the presentation of epitopes spanningmore variable regions to CTL.

ACKNOWLEDGMENTS

We thank Gillian Air and Sherry Crowe for their insight on thehuman influenza A virus and PR8 mouse model of influenza virusinfection, respectively.

This work was supported by National Institutes of Health ContractHHSN266200400027C (W.H.H.) and National Institute of Allergy andInfectious Disease institutional training grant A1007633-006 (A.W.).


REFERENCES

1. Bao, Y., P. Bolotov, D. Dernovoy, B. Kiryutin, L. Zaslavsky, T. Tatusova, J.Ostell, and D. Lipman. 2008. The influenza virus resource at the NationalCenter for Biotechnology Information. J. Virol. 82:596–601.

2. Beekman, N. J., P. A. van Veelen, T. van Hall, A. Neisig, A. Sijts, M. Camps,P. M. Kloetzel, J. J. Neefjes, C. J. Melief, and F. Ossendorp. 2000. Abroga-tion of CTL epitope processing by single amino acid substitution flanking theC-terminal proteasome cleavage site. J. Immunol. 164:1898–1905.

3. Berkhoff, E. G., A. C. Boon, N. J. Nieuwkoop, R. A. Fouchier, K. Sintnicolaas,A. D. Osterhaus, and G. F. Rimmelzwaan. 2004. A mutation in the HLA-B*2705-restricted NP383-391 epitope affects the human influenza A virus-spe-cific cytotoxic T-lymphocyte response in vitro. J. Virol. 78:5216–5222.

4. Berkhoff, E. G., E. de Wit, M. M. Geelhoed-Mieras, A. C. Boon, J. Symons,R. A. Fouchier, A. D. Osterhaus, and G. F. Rimmelzwaan. 2005. Functionalconstraints of influenza A virus epitopes limit escape from cytotoxic T lym-phocytes. J. Virol. 79:11239–11246.

5. Berkhoff, E. G., E. de Wit, M. M. Geelhoed-Mieras, A. C. Boon, J. Symons,R. A. Fouchier, A. D. Osterhaus, and G. F. Rimmelzwaan. 2006. Fitness costslimit escape from cytotoxic T lymphocytes by influenza A viruses. Vaccine24:6594–6596.

6. Bertoletti, A., A. Costanzo, F. V. Chisari, M. Levrero, M. Artini, A. Sette, A.Penna, T. Giuberti, F. Fiaccadori, and C. Ferrari. 1994. Cytotoxic T lym-phocyte response to a wild type hepatitis B virus epitope in patients chron-ically infected by variant viruses carrying substitutions within the epitope. J.Exp. Med. 180:933–943.

7. Bihl, F., N. Frahm, L. Di Giammarino, J. Sidney, M. John, K. Yusim, T.Woodberry, K. Sango, H. S. Hewitt, L. Henry, C. H. Linde, J. V. Chisholm,3rd, T. M. Zaman, E. Pae, S. Mallal, B. D. Walker, A. Sette, B. T. Korber, D.Heckerman, and C. Brander. 2006. Impact of HLA-B alleles, epitope bind-ing affinity, functional avidity, and viral coinfection on the immunodomi-nance of virus-specific CTL responses. J. Immunol. 176:4094–4101.

8. Boon, A. C., G. de Mutsert, R. A. Fouchier, A. D. Osterhaus, and G. F.Rimmelzwaan. 2006. The hypervariable immunodominant NP418-426epitope from the influenza A virus nucleoprotein is recognized by cytotoxicT lymphocytes with high functional avidity. J. Virol. 80:6024–6032.

9. Boon, A. C., G. de Mutsert, Y. M. Graus, R. A. Fouchier, K. Sintnicolaas,A. D. Osterhaus, and G. F. Rimmelzwaan. 2002. The magnitude and speci-ficity of influenza A virus-specific cytotoxic T-lymphocyte responses in hu-mans is related to HLA-A and -B phenotype. J. Virol. 76:582–590.

10. Boon, A. C., G. de Mutsert, Y. M. Graus, R. A. Fouchier, K. Sintnicolaas,A. D. Osterhaus, and G. F. Rimmelzwaan. 2002. Sequence variation in anewly identified HLA-B35-restricted epitope in the influenza A virus nu-cleoprotein associated with escape from cytotoxic T lymphocytes. J. Virol.76:2567–2572.

11. Buchli, R., R. S. Vangundy, C. F. Giberson, and W. H. Hildebrand. 2006.Critical factors in the development of fluorescence polarization-based pep-tide binding assays: an equilibrium study monitoring specific peptide bindingto soluble HLA-A*0201. J. Immunol. Methods 314:38–53.

12. Buchli, R., R. S. VanGundy, H. D. Hickman-Miller, C. F. Giberson, W.Bardet, and W. H. Hildebrand. 2004. Real-time measurement of in vitropeptide binding to soluble HLA-A*0201 by fluorescence polarization. Bio-chemistry 43:14852–14863.

13. Buchli, R., R. S. VanGundy, H. D. Hickman-Miller, C. F. Giberson, W.Bardet, and W. H. Hildebrand. 2005. Development and validation of afluorescence polarization-based competitive peptide-binding assay for HLA-A*0201—a new tool for epitope discovery. Biochemistry 44:12491–12507.

14. de Campos-Lima, P. O., V. Levitsky, J. Brooks, S. P. Lee, L. F. Hu, A. B.Rickinson, and M. G. Masucci. 1994. T cell responses and virus evolution:loss of HLA A11-restricted CTL epitopes in Epstein-Barr virus isolates fromhighly A11-positive populations by selective mutation of anchor residues. J.Exp. Med. 179:1297–1305.

15. De Groot, A. S., A. Bosma, N. Chinai, J. Frost, B. M. Jesdale, M. A.Gonzalez, W. Martin, and C. Saint-Aubin. 2001. From genome to vaccine: insilico predictions, ex vivo verification. Vaccine 19:4385–4395.

16. Hillig, R. C., M. Hulsmeyer, W. Saenger, K. Welfle, R. Misselwitz, H. Welfle,C. Kozerski, A. Volz, B. Uchanska-Ziegler, and A. Ziegler. 2004. Thermo-dynamic and structural analysis of peptide- and allele-dependent propertiesof two HLA-B27 subtypes exhibiting differential disease association. J. Biol.Chem. 279:652–663.

17. Hollsberg, P. 2002. Contribution of HLA class I allele expression to CD8�

T-cell responses against Epstein-Barr virus. Scand. J. Immunol. 55:189–195.18. Hourigan, C. S., M. Harkiolaki, N. A. Peterson, J. I. Bell, E. Y. Jones, and

C. A. O’Callaghan. 2006. The structure of the human allo-ligand HLA-

B*3501 in complex with a cytochrome p450 peptide: steric hindrance influ-ences TCR allo-recognition. Eur. J. Immunol. 36:3288–3293.

19. Kiepiela, P., A. J. Leslie, I. Honeyborne, D. Ramduth, C. Thobakgale, S.Chetty, P. Rathnavalu, C. Moore, K. J. Pfafferott, L. Hilton, P. Zimbwa, S.Moore, T. Allen, C. Brander, M. M. Addo, M. Altfeld, I. James, S. Mallal, M.Bunce, L. D. Barber, J. Szinger, C. Day, P. Klenerman, J. Mullins, B.Korber, H. M. Coovadia, B. D. Walker, and P. J. Goulder. 2004. Dominantinfluence of HLA-B in mediating the potential co-evolution of HIV andHLA. Nature 432:769–775.

20. Martin, W., H. Sbai, and A. S. De Groot. 2003. Bioinformatics tools foridentifying class I-restricted epitopes. Methods 29:289–298.

21. McMichael, A. J., and C. A. O’Callaghan. 1998. A new look at T cells. J. Exp.Med. 187:1367–1371.

22. Middleton, D., L. Menchaca, H. Rood, and R. Komerofsky. 2003. New allele fre-quency database: http://www.allelefrequencies.net. Tissue Antigens 61:403–407.

23. Milicic, A., D. A. Price, P. Zimbwa, B. L. Booth, H. L. Brown, P. J. Easter-brook, K. Olsen, N. Robinson, U. Gileadi, A. K. Sewell, V. Cerundolo, andR. E. Phillips. 2005. CD8� T cell epitope-flanking mutations disrupt pro-teasomal processing of HIV-1 Nef. J. Immunol. 175:4618–4626.

24. Rimmelzwaan, G. F., A. C. Boon, J. T. Voeten, E. G. Berkhoff, R. A. Fouchier,and A. D. Osterhaus. 2004. Sequence variation in the influenza A virusnucleoprotein associated with escape from cytotoxic T lymphocytes. VirusRes. 103:97–100.

25. Rock, K. L., C. Gramm, L. Rothstein, K. Clark, R. Stein, L. Dick, D. Hwang,and A. L. Goldberg. 1994. Inhibitors of the proteasome block the degrada-tion of most cell proteins and the generation of peptides presented on MHCclass I molecules. Cell 78:761–771.

26. Roder, G., T. Blicher, S. Justesen, B. Johannesen, O. Kristensen, J. Kastrup,S. Buus, and M. Gajhede. 2006. Crystal structures of two peptide-HLA-B*1501 complexes; structural characterization of the HLA-B62 supertype.Acta Crystallogr. D 62:1300–1310.

27. Rohrlich, P. S., S. Cardinaud, H. Firat, M. Lamari, P. Briand, N. Escriou,and F. A. Lemonnier. 2003. HLA-B*0702 transgenic, H-2KbDb double-knockout mice: phenotypical and functional characterization in response toinfluenza virus. Int. Immunol. 15:765–772.

28. Sette, A., and J. Sidney. 1999. Nine major HLA class I supertypes account forthe vast preponderance of HLA-A and -B polymorphism. Immunogenetics50:201–212.

29. Smith, K. D., B. E. Mace, A. Valenzuela, J. L. Vigna, J. A. McCrutcheon, J. A.Barbosa, E. Huczko, V. H. Engelhard, and C. T. Lutz. 1996. Probing HLA-B7conformational shifts induced by peptide-binding groove mutations and bound pep-tide with anti-HLA monoclonal antibodies. J. Immunol. 157:2470–2478.

30. Soudeyns, H., S. Paolucci, C. Chappey, M. B. Daucher, C. Graziosi, M.Vaccarezza, O. J. Cohen, A. S. Fauci, and G. Pantaleo. 1999. Selectivepressure exerted by immunodominant HIV-1-specific cytotoxic T lympho-cyte responses during primary infection drives genetic variation restricted tothe cognate epitope. Eur. J. Immunol. 29:3629–3635.

31. Stewart-Jones, G. B., A. J. McMichael, J. I. Bell, D. I. Stuart, and E. Y.Jones. 2003. A structural basis for immunodominant human T cell receptorrecognition. Nat. Immunol. 4:657–663.

32. Voeten, J. T., T. M. Bestebroer, N. J. Nieuwkoop, R. A. Fouchier, A. D.Osterhaus, and G. F. Rimmelzwaan. 2000. Antigenic drift in the influenza Avirus (H3N2) nucleoprotein and escape from recognition by cytotoxic Tlymphocytes. J. Virol. 74:6800–6807.

33. Voeten, J. T., G. F. Rimmelzwaan, N. J. Nieuwkoop, R. A. Fouchier, and A. D.Osterhaus. 2001. Antigen processing for MHC class I restricted presentationof exogenous influenza A virus nucleoprotein by B-lymphoblastoid cells.Clin. Exp. Immunol. 125:423–431.

34. Wahl, A., F. Schafer, W. Bardet, R. Buchli, G. M. Air, and W. H. Hildebrand.2009. HLA class I molecules consistently present internal influenza epitopes.Proc. Natl. Acad. Sci. USA 106:540–545.

35. Weiner, A., A. L. Erickson, J. Kansopon, K. Crawford, E. Muchmore, A. L.Hughes, M. Houghton, and C. M. Walker. 1995. Persistent hepatitis C virusinfection in a chimpanzee is associated with emergence of a cytotoxic Tlymphocyte escape variant. Proc. Natl. Acad. Sci. USA 92:2755–2759.

36. Yewdell, J., J. R. Bennink, and Y. Hosaka. 1988. Cells process exogenousproteins for recognition by cytotoxic T lymphocytes. Science 239:637–640.

37. York, I. A., S. C. Chang, T. Saric, J. A. Keys, J. M. Favreau, A. L. Goldberg,and K. L. Rock. 2002. The ER aminopeptidase ERAP1 enhances or limitsantigen presentation by trimming epitopes to 8–9 residues. Nat. Immunol.3:1177–1184.

38. York, I. A., and K. L. Rock. 1996. Antigen processing and presentation by theclass I major histocompatibility complex. Annu. Rev. Immunol. 14:369–396.


T-Cell Tolerance for Variability in an HLA Class I-Presented Influenza A Virus Epitope

Documents