Early studies of influenza virus-specific CD8+ T cell-mediated immunity indicated that the level of CTL activity associated with H2Db is greatly diminished in mice that also express H2Kk. Such MHC-related immunodominance hierarchies are of some interest, as they could lead to variable outcomes for peptide-based vaccination protocols in human populations. The influence of H2Kk on the H2Db-restricted response profile has thus been looked at again using a contemporary, quantitative, IFN-γ-based flow cytometric assay. The depressive effect of H2Kk was very apparent for the influenza DbPA224 epitope and was also reproduced when CTL activity was measured for H2Db-expressing targets pulsed with the immunodominant NP366 peptide. The secondary CD8+IFN-γ+ DbNP366-specific response was much greater in parental H2b than in H2k×bF1 mice, but the sizes of the CD8+ sets specific for KkNP50 and DbNP366 were essentially equivalent in the F1 animals. Thus, although the immunodominance profile associated with DbNP366 is lost when H2Kk is also present, the response is still substantial. A further, MHC-related effect was also identified for the KkNS1152 epitope, which was consistently associated with a greater CD8+IFN-γ+ response in H2KkDb recombinant than in (H2KkDk × H2KbDb)F1 mice. The diminished DbPA224 response in H2k×bF1 mice was characterized by loss of a prominent Vβ7 TCR responder phenotype, supporting the idea that TCR deletion during ontogeny shapes the available repertoire. The overall conclusion is that these MHC-related immunodominance hierarchies are more subtle than the early CTL assays suggested and, although inherently unpredictable, are unlikely to cause a problem for peptide-based vaccine strategies.
Experiments conducted >20 yr ago uncovered what seemed to be a profound MHC-related regulatory effect for the H2Db-restricted responses to the influenza A viruses and vaccinia virus (1, 2). Potent influenza and vaccinia virus-specific CD8+ CTL activity was consistently apparent for H2KkDb or H2KdDb targets following exposure to in vivo stimulated H2b (KbDb) effectors, while comparable H2KkDb recombinant and H2k×bF1 (KkDk × KbDb) spleen populations were only minimally lytic for H2KdDb cells infected with the homologous virus. Influenza virus is a small, negative strand RNA virus, while vaccinia virus is a large DNA virus, so this is not likely to reflect some inherent quality of the viral pathogens concerned. The overall conclusion was that we were seeing some sort of immunodominance hierarchy, where a concurrent H2Kk-restricted response functioned to minimize that associated with H2Db (1, 2).
Several different types of experiments were performed in efforts to take this observation further. Naive B10.A(4R) T cells (H2KkDb) were depleted of alloreactive potential by in vivo filtration through lethally irradiated KbDb mice, then stimulated with vaccinia virus in an additional set of irradiated KbDb recipients (3). Under these conditions the vaccinia-specific KkDb T cells showed potent, virus-specific, H2Db-restricted CTL activity. Later limiting dilution analysis (LDA)4 to determine influenza-specific CD8+ T cell frequencies showed that significant H2Db-restricted memory populations were generated in mice that also expressed H2Kk (4). Both approaches thus indicated that the concurrent presence of H2Kk throughout ontogeny did not greatly compromise the development of H2Db-restricted CTL precursors (CTLp) specific for vaccinia virus, although there was an effect on the generation of CTL effectors in normal mice.
The opposite conclusion was drawn from other experiments that analyzed response patterns for H2KbDb T cells from mice that had been neonatally tolerized to H2Kk (5). In this case the vaccinia-specific, H2Db-restricted CTL response was profoundly diminished. The favored interpretation was that the presence of H2Kk during development resulted in the deletion of CD8+ T cells that could recognize vaccinia virus associated with H2Db.
These studies were all performed a very long time ago, before we understood that the primary function of MHC glycoproteins is to present viral peptides to the TCR (6, 7). The assays used (CTL and LDA) were either minimally quantitative (CTL) or at an early stage of development and far from optimized (LDA). We did not know that distinct CD8+ T cell clones could be specific for different peptides from the same virus bound to the same MHC class I glycoprotein (7, 8). Even so, it is clearly important as we move to the use of peptide-based vaccines (9) that we understand whether such MHC-related immunodominance hierarchies are real and, if so, how they operate. The experiment reported here used short term peptide stimulation followed by staining for cytoplasmic IFN-γ (8, 10, 11) to look at the relationship between H2Kk and H2Db (1, 2) for the influenza-specific CD8+ T cell response. This approach gives very similar numbers (8, 10, 11) to those detected by staining with tetrameric complexes of MHC class I glycoprotein plus peptide (tetramers). Tetramer reagents were not available for most of the epitopes analyzed in this study.
Materials and Methods
Mice and tissue sampling
Female C57BL/6J (B6, H2b), C3H/HeJ (C3H, H2k), B6×C3HF1 (B6C3F1, H2k×b), B10.BR (H2k), and B10.A(2R) (H2KkDb) mice were purchased from The Jackson Laboratory (Bar Harbor, ME). They were anesthetized at 8 wk of age by i.p. injection with avertin (2,2,2-tribromoethanol) and were challenged intranasally (i.n.) with 106.8 EID50 of the HK×31 (H3N2) influenza A virus (12). Memory mice for secondary challenge experiments were injected i.p. at least 6 wk previously with 108.5 EID50 of the PR8 influenza virus (11). At the time of sampling, the mice were anesthetized and exsanguinated from the axillary artery. Lymphocytes were obtained from the pneumonic lung by bronchoalveolar lavage (BAL), and adherent cells were removed by incubating them on plastic for 1 h at 37°C (12). Peritoneal exudate lymphocytes (PEL) were processed in the same way. Spleen and mediastinal lymph node (MLN) samples were disrupted and enriched for CD8+ T cells by incubation with mAbs (PharMingen, San Diego, CA) to CD4 (GK1.5) and MHC class II glycoprotein (M5/114.15.2), followed by anti-rat and anti-mouse Ig-coated magnetic beads (Dynal, Oslo, Norway).
Peptides, epitopes, and the Pepγ assay
The influenza virus nucleoproteins ASNENMETM (NP366–374) and SDYGERLI (NP50–57), nonstructural protein EEGAIVGEI (NS1152–160), nuclear export protein RTFSQLI (NS2114–121), polymerase 2 SSLENFRAYV (PA224–233), and matrix protein MGLIYNRM (M128–135) used in this study have all been described previously (6, 8, 13, 14). They were synthesized at the Hartwell Center, St. Jude Children’s Research Hospital using a model 433A peptide synthesizer (Applied Biosystems, Berkeley, CA) and were purified by HPLC. The H2K or H2D epitopes associated with these peptides are identified as DbNP366, DbPA224, KbNS2114, KkNP50, KkNS1152, and KbM1128.
The Pepγ assay (8, 10, 11) used spleen MLN and BAL populations that were enriched for CD8+ T cells and cultured for 5 h in 96-well round-bottom plates (Costar, Corning, NY) at a concentration of 5–8 × 105 cells/well in 200 μl of RPMI 1640 medium containing 10% FCS, 10 U/ml human rIL-2, and 5 μg/ml Brefeldin A (Epicentre Technologies, Madison, WI) in the presence or the absence of 1 μM viral peptide. The T cells were then washed and stained with anti-mouse CD8α-FITC Ab (PharMingen, San Diego, CA). Nonspecific Fc binding was blocked using anti-mouse CD16/32 (PharMingen). The cells were fixed in 1% formaldehyde in PBS for 20 min, then permeabilized in PBS/0.5% saponin for 10 min before staining with a conjugated mAb to mouse IFN-γ (PE-XMG 1.2). The specificity of the staining reaction was checked initially by blocking with excess purified cytokine. Isotype control Ab was also used. The data were acquired on a Becton Dickinson FACScan or FACScalibur flow cytometer, then analyzed using CellQuest software (Becton Dickinson Immunocytometry Systems, San Jose, CA). In each assay, the percentage of CD8+IFN-γ+ without peptide (<0.2%) was subtracted from the percentage of CD8+IFN-γ+ with peptide to give the percentage of specific CD8+ T cells.
The numbers of memory T cells in the spleens of mice primed with the PR8 virus were determined by the IFN-γ ELISPOT assay (15, 16). Nitrocellulose-bottom 96-well plates (Millipore, Bedford, MA) were coated overnight at 4°C with rat anti-mouse IFN-γ Ab (clone R4-6A2 from PharMingen). Dilutions of responder cells in complete medium were cultured for 48 h with 5 × 105 syngeneic feeders pulsed with (1 μM) or without peptide and 10 U/ml recombinant human IL-2. The plates were then washed and incubated with a biotinylated mAb to IFN-γ (clone XMG 1.2) followed by streptavidin-alkaline phosphatase and developed using 5-bromo-4-chloro-3-indolyl-phosphate/nitro blue tetrazolium alkaline phosphatase substrate (Sigma, St. Louis, MO). Responses were considered positive when there were >10 ELISPOTs/well and the number of peptide-pulsed feeder ELISPOTs was more than two times the number of unpulsed feeder ELISPOTs. The frequency of peptide-specific CD8+ T cells present in the responding population was calculated by subtracting the mean number of spots for feeders with no peptide from the mean number of spots with peptide-pulsed feeders.
The target H2k- or H2Db-transfected L-929 cells (L cells) were labeled with Na51Cr for 1 h, pulsed with viral peptides or infected with the HK×31 influenza A virus for 60 min, washed, then plated at 5000 targets/well (8). The target cells were washed twice and incubated with the effector populations for 5 h before harvesting supernatants for gamma counting. Two-fold lymphocyte dilutions were assayed in triplicate, while untreated and Triton-disrupted controls were measured in quadruplicate. The percent specific lysis was calculated as 100 × (51Cr release from targets with effectors − 51Cr release from targets alone)/(51Cr release from targets with Triton). The level of 51Cr release from targets incubated in the absence of T cells did not exceed 15% of the total Triton-mediated 51Cr release. This background value was subtracted to give the values presented here.
CTL activity following in vivo stimulation
The initial finding (1) that the influenza-specific, H2Db-restricted response was substantially greater in the absence of H2Kk was made using virus-infected L cell (KkDk) and MC57G (KbDb) targets and CTL populations taken directly from PR8-primed mice challenged i.n. with the HK×31 virus. This secondary challenge (HK×31→PR8) experiment (1, 11) was repeated, except that the CTL targets were L cells expressing DbNP366 (Fig. 1, A–D) and KkNP50 (Fig. 1, E–H). The level of DbNP366-specific lysis was greater in every case for the parental B6 (KbDb) than for B6C3F1 (KkDk × KbDb) effectors from the spleen, BAL and MLN (Fig. 1, A–D). An identical result was recorded for the BAL population recovered from mice after primary infection with the HK×31 virus (Fig. 2). The quality of CTL recognition did not, however, differ for the B6 and B6C3F1 responses to the NP366 peptide, as the cut-off point for lysis of the peptide-pulsed L929-Db targets was 10−10 M in each case (Fig. 2).
These later experiments (Fig. 1) could, however, cause us to question the exclusivity of this F1 effect to the H2Db-restricted response (1). Cytotoxicity associated with the recognition of KkNP50 by the secondarily stimulated B6C3F1 CTL was also diminished compared with the level of specific 51Cr release caused by the C3H (KkDk) T cells, with the effect being most obvious for the BAL population (Fig. 1 G). We did not persist further with this, as we clearly now have much better assays (8, 11, 14, 16) for measuring virus-specific CD8+ T cell responses (see subsequent sections). Also, our capacity to analyze the H2k-restricted response is limited by the fact that only two epitopes (13) have been identified (KkNP50 and KkNS1152), and it seems likely from the more detailed analysis that has been performed with the H2b haplotype (6, 8, 14) that there are more to be found.
T cell memory
Other early LDA studies indicated that the numbers of H2Db-restricted memory CTLp were not necessarily significantly different for mice expressing H2Db in the presence or the absence of H2Kk (4). This was confirmed using the IFN-γ ELISPOT assay (Fig. 3), which tends to give lower values than the Pepγ protocol (11), but, because the lymphocytes are diluted, allows virus-specific memory T cells to be detected at lower frequencies. The prevalence of memory T cells specific for DbNP366 was comparable for the B6 and B6C3F1 mice. This was also true for the set reactive to KkNP50 in the C3H and B6C3F1 mice (Fig. 3). The diminished recall of CTL activity for the KkNP50 and DbNP366 epitopes in the F1 group (Fig. 1) is thus not obviously explained by the relative availability of memory T cells (Fig. 3).
CD8+IFN-γ+ T cells in parental and F1 mice
The primary and secondary responses (HK×31→PR8) following i.n. exposure to the HK×31 virus were then analyzed using the Pepγ assay for parental (B6 and C3H) and F1 mice. The results for the DbNP366- and KkNP50-specific sets are given as the percentage of cells staining in the BAL, MLN, and spleen for the H2k, H2b, and H2k×bF1 mice in Fig. 4, while the total cell counts for populations reactive to DbNP366, KkNP50, KkNS1152, and KbNS2114 are shown for the F1 group only in Fig. 5. Cumulating the prevalence data for the different epitopes in the various sites sampled indicates that the magnitudes of the primary F1 and parental responses are essentially equivalent up to 10 days after infection (Fig. 4, A, B, E, F, I, and J). The numbers of CD8+ DbNP366 T cells in the F1 BAL and spleen on day 13 were, however, 2- to 3-fold lower than the parental values (Fig. 4, A, E, and I). The HK×31 virus is generally cleared from the lung within 10 days of primary challenge, so the CD8+ DbNP366+ response seems to resolve more rapidly in the F1 than the B6 mice (17).
The secondary response to DbNP366 was consistently lower in the H2k×bF1 than in the H2b parent (Fig. 4, C, G, and K), an effect that was much less apparent for KkNP50 encountered in the H2k and H2k×b F1 situations (Fig. 4, D, H, and L). However, when we translated the percentages into cell numbers, the magnitudes of the DbNP366- and KkNP50-reactive populations were essentially comparable for the B6C3F1 response (Fig. 5). The divergence in relative prevalence of the DbNP366- and KkNP50-reactive sets is thus seen between the parent and the F1, not within the F1 group. It is important in this regard to note that the scales in Fig. 4 are very different for the DbNP366-specific (Fig. 4, C, G, and K) and KkNP50-specific (Fig. 4, D, H, and L) populations, with the B6 results reflecting the dominance of DbNP366 that we have recognized previously for the secondary influenza-specific response in B6 mice (8, 11, 17).
Responses to the DbPA224 and KkNS1152 epitopes
We recently described (8) a new epitope (DbPA224) that is at least as prominent as DbNP366 in the primary, but not the secondary, response of B6 mice following i.n. challenge with the HKx31 virus. The same relationship was found for virus-specific CD8+ T cells in the spleen and PEL population from naive H2b mice challenged i.p. with the PR8 virus (Fig. 6, A and C), the protocol used to prime for the secondary response (Figs. 1, 2, 4, and 5). However, this equivalence between the DbNP366- and DbPA224-reactive sets (Fig. 6, A and B) was not apparent for H2k×b F1 mice, in which the numbers of T cells specific for DbNP366 were much higher (Fig. 6, B and D). The lower prevalence of DbPA224-specific T cells in F1 mice was also confirmed following primary i.n. challenge with the HK×31 virus (Fig. 7, A, C, and E). The profile for the DbPA224-reactive set in the B10.A(2R) H2KkDb recombinant (Fig. 7, B, D, and F) was intermediate between that for the H2KbDb parent (Fig. 6, B and D) (8) and the (H2KkDk × H2KbDb) F1 (Fig. 6, A and C, and Fig. 7, A, C, and E).
The response to KkNS1152 in B10.A (2R) mice was, if anything, greater than that to KkNP50 (Fig. 7, B, D, and F). However, KkNS1152 was clearly less prominent following both primary (Fig. 5, A–C, and Fig. 7, A, C, and E) and secondary (Fig. 5, D–F) challenge of the B6C3F1 mice. Comparison of primary BAL populations from congenic B10.BR (KkDk) and B10.A (2R) mice also showed a relatively greater response to KkNS1152 in the recombinant (Table I) than in the parental strain, an impression that was confirmed by further analysis using C3H (H2KkDk) mice (Fig. 8). The response to KkNS1152 thus seems to be lower when H2Dk is also present.
|Source of Cells .||Mouse Strain .||Epitope .||.||.|
|.||.||KbNP366 .||KkNP50 .||KkNS1152 .|
|Source of Cells .||Mouse Strain .||Epitope .||.||.|
|.||.||KbNP366 .||KkNP50 .||KkNS1152 .|
The B6, B10.BR, and B10.A(2R) mice were infected i.n. with 106.8 EID50 of the HKx31 influenza A virus, and the percentage of CD8+ IFNγ+ T cells was determined 8 days later.
Loss of a prominent TCRβ phenotype in the H2k×bF1 response to DbPA224
The CD8+DbNP366+ response is characterized by prominent usage of a spectrum of Vβ8.3+ TCRs associated with a variety of TCR α-chains (18). Recent analysis indicates that Vβ7 dominates the CD8+DbPA224-specific set generated in B6 mice (G.T.B. and P.C.D., unpublished observations). We thus asked whether the low CD8+DbPA224 response in the H2k×bF1 mice might reflect some change in the pattern of TCR involvement. This was indeed found to be the case. When HK×31-immune H2b and H2k×b F1 spleen populations were stimulated separately in vitro with the PA224 (Fig. 9,A) or NP366 (Fig. 9,B) peptides, the diminished response in the F1 animals was associated with a complete absence of Vβ7 in cultures established from all but one of five mice (Fig. 9,A). There was, however, no difference in the Vβ8 staining profile for the B6 and B6C3F1 mice (Fig. 9,B). The effect for DbPA224 could reflect clonal deletion of cross-reactive Vβ7+ T cells specific for self peptide(s) presented in the context of H2Kk or H2Dk (5). The higher DbPA224-specific response (spleen and MLN; Fig. 7) in the H2KkDb (compared with the H2KkDk × KbDb F1) mice might indicate that the defect for DbPA224 is more likely to be associated with H2Dk than with the H2Kk allele implicated in the early analysis of immunodominance hierarchies (1, 2). However, it is also possible that the Ag involved in the putative deletion of the Vβ7+ DbPA224+-specific set is a peptide from the C3H background presented by H2Kk. This will be analyzed further.
These experiments use a contemporary, quantitative approach (8, 11, 14) to confirm that there are indeed MHC-related immunodominance hierarchies in the CD8+ T cell responses to various peptides derived from the influenza A viruses. The spectrum of MHC alleles expressed in a particular mouse strain can clearly modify the relative prevalence of CD8+ T cells specific for an individual epitope. However, although we did confirm that the level of virus-specific and DbNP366-specific 51Cr release caused by CD8+ CTL effectors is lower when H2Kk is also present in the responding mouse strain, the effects are generally subtle and far from the all-or-none situation described in earlier studies (1, 2) in which the only in vitro measure of the CD8+ T cell response was cytotoxicity.
We found that the magnitude of the primary DbPA224-specific response is much lower in H2k×bF1 than in H2b mice when measured quantitatively by the Pepγ assay. However, the minimal CTL activity associated with DbPA224 (8) indicates that this epitope is poorly presented on virus-infected cell lines, so the diminished numbers of DbPA224-specific T cells in H2k×bF1 mice fails to explain the H2Db-restricted difference between F1 and parent detected earlier by cytotoxicity (1). On the other hand, DbNP366 is a potent CTL target. The primary response to DbNP366 measured as numbers of CD8+IFN-γ+ cells was at an equivalent level (at least to day 10) in the H2k×bF1 and H2b mice, although the level of specific 51Cr release caused by freshly isolated BAL populations was lower in the F1 group. It thus seems that CD8+ T cell frequency does not necessarily predict the magnitude of differentiated CTL function, at least in the primary response to the influenza A viruses. Unfortunately, we do not currently have an assay available that allows CTL activity to be measured at the single-cell level.
While the numbers of DbNP366-specific CD8+ memory T cells were comparable for PR8-primed H2k×bF1 and H2b mice immediately before i.n. challenge with the HK×31 virus, the extent of further clonal expansion (17) showed the hierarchy for the CD8+IFN-γ+ set that would have been predicted from the CTL assays. The dominance by the DbNP366-specific population during the secondary influenza-specific CD8+ T cell response in H2b mice had led to the impression that DbNP366 is some sort of “superepitope” (8, 11). However, this is not the case in the H2k×bF1, where the responses to DbNP366 and KkNP50 are essentially equivalent. The relative prevalence of a particular epitope-specific CD8+ population is clearly a function of the spectrum of MHC glycoproteins that are expressed in the responder environment. This effect is also seen for KkNS1152, which, compared with KkNP50, shows the hierarchy H2KkDb>H2KkDk>(H2KkDk × H2KbDb)F1.
The primary and recall responses to different epitopes were similar in magnitude for the H2k×bF1, but not the H2b, mice. The major difference is the prominence of the DbNP366-specific set in the parental strain following secondary challenge. The situation for the H2k×bF1 is much more comparable to that described previously for epitopes derived from Listeria monocytogenes (19). Immunodominance hierarchies apparently become unpredictable with the addition or removal of other MHC glycoproteins.
The idea that H2Db-restricted CD8+ T cells are deleted during thymic development as a consequence of exposure to self peptides presented by H2k glycoproteins throughout ontogeny could explain the much lower response to DbPA224 in H2k×bF1 than in H2b mice (5). This cross-tolerance concept was developed before it was known that MHC-restricted CD8+ T cells are specific for viral peptides. The observation that such effects are apparent for one (DbPA224), but not another (DbNP366), epitope makes sense in the context of established models of self tolerance (20, 21, 22). It is certainly the case that the absence of key TCRs in the mature repertoire can diminish the magnitude of a CD8+ T cell response (reviewed in Ref. 7).
The obvious question is whether we should be concerned about MHC-related immunodominance hierarchies as we move to develop vaccines that incorporate peptides expressed by a spectrum of MHC molecules. A case in point is the polytope approach that uses linked viral peptides that bind a range of HLA glycoproteins to protect, for instance, against EBV infection (9, 23). The results presented here indicate that any MHC-related hierarchies are generally much less absolute than suggested by the early CTL assays and are not likely to cause a problem for a vaccine incorporating multiple peptides. Even so, it is appropriate to assure that the peptides used are recognized widely by people that express the particular HLA glycoprotein.
This work was supported by Public Health Service Grants AI29579, AI38359, and CA21765, and the American Lebanese Syrian Associated Charities. G.T.B. is a C. J. Martin fellow of the Australian National Health and Medical Research Council (Reg. Key. 977 309).
Abbreviations used in this paper: LDA, limiting dilution analysis; i.n., intranasally; BAL, bronchoalveolar lavage; ELISPOT, enzyme-linked immunospot; MLN, mediastinal lymph node; H, influenza hemagglutinin; M, matrix protein; N, neuraminidase; PA, polymerase 2 protein; PEL, peritoneal exudate lymphocytes; NP, nucleoprotein; NS1, nonstructural protein; NS2, nuclear export protein.