Profound Protection against Respiratory Challenge with a Lethal H7N7 Influenza A Virus by Increasing the Magnitude of CD8⁺ T-Cell Memory

Abstract

The recall of CD8⁺ T-cell memory established by infecting H-2^b mice with an H1N1 influenza A virus provided a measure of protection against an extremely virulent H7N7 virus. The numbers of CD8⁺ effector and memory T cells specific for the shared, immunodominant D^bNP₃₆₆ epitope were greatly increased subsequent to the H7N7 challenge, and though lung titers remained as high as those in naive controls for 5 days or more, the virus was cleared more rapidly. Expanding the CD8⁺ memory T-cell pool (<0.5 to >10%) by sequential priming with two different influenza A viruses (H3N2→H1N1) gave much better protection. Though the H7N7 virus initially grew to equivalent titers in the lungs of naive and double-primed mice, the replicative phase was substantially controlled within 3 days. This tertiary H7N7 challenge caused little increase in the magnitude of the CD8⁺ D^bNP₃₆₆⁺ T-cell pool, and only a portion of the memory population in the lymphoid tissue could be shown to proliferate. The great majority of the CD8⁺ D^bNP₃₆₆⁺ set that localized to the infected respiratory tract had, however, cycled at least once, though recent cell division was shown not to be a prerequisite for T-cell extravasation. The selective induction of CD8⁺ T-cell memory can thus greatly limit the damage caused by a virulent influenza A virus, with the extent of protection being directly related to the number of available responders. Furthermore, a large pool of CD8⁺ memory T cells may be only partially utilized to deal with a potentially lethal influenza infection.

The level of protection conferred by established influenza virus-specific CD8⁺ T-cell memory has tended to be somewhat disappointing, with the maximum effect being generally to enhance virus clearance by 2 to 3 days (2, 5, 10, 13, 21, 22, 30). Recent analysis from this laboratory (10, 30) has utilized prime and challenge experiments with the A/PR8/34 (PR8, H1N1) and A/HK×31 (HK×31, H3N2) viruses in C57BL/6J (B6, H-2^b) mice (10). The relatively avirulent HK×31 virus is a laboratory reassortant of PR8 and A/HK/168 that expresses the surface hemagglutinin (H) and neuraminidase (N) glycoproteins of A/HK/168 and the six internal genes of PR8 (20). There is no cross neutralization with antibodies developed in response to infection with PR8 and HK×31, while the peptides that stimulate the H-2K^b- or H-2D^b-restricted virus-specific CD8⁺ T-cell response are mainly derived from the shared internal proteins (36, 40). The polymerase 2 (PA) and nucleoprotein (NP) genes provide the most prominent epitopes (D^bPA_224–233 and D^bNP_366–374) recognized when immunologically naive mice are infected intranasally (i.n.) with the H3N2 virus, while the secondary CD8⁺ T-cell response generated following HK×31 challenge of PR8-primed mice is dominated by the D^bNP₃₆₆-specific population (4).

A single intraperitoneal (i.p.) exposure to a high dose of the PR8 virus leads to the establishment of long-term memory, with the “resting” NP-specific memory T cells being barely detectable (≤0.5% of the splenic CD8⁺ set) by flow cytometric analysis following staining with tetrameric complexes (tetramers) of D^bNP₃₆₆ (10, 11). Secondary i.n. infection with the H3N2 virus (H3N2→H1N1) induces massive clonal expansion of the D^bNP₃₆₆-specific population, resulting in frequencies of 15 to 25% in the splenic CD8⁺ set within 14 days of challenge. By this time, >70% of the CD8⁺ T cells recovered by bronchoalveolar lavage (BAL) from the pneumonic lung either bind the D^bNP₃₆₆ tetramer or can be induced to synthesize gamma interferon following short-term in vitro stimulation with the NP_366–374 peptide in the presence of brefeldin A. Despite this massive secondary response induced by the H3N2 virus in PR8-primed mice, there is a still a delay of 3 to 5 days or so before virus-specific CD8⁺ T cells can be detected in the BAL population. Furthermore, the maximal H3N2 lung titers achieved at 5 days after i.n. challenge are essentially identical for naive and PR8-primed mice, though the virus is cleared more rapidly by the CD8⁺ recall response (11).

We ask here whether it is possible to improve on this situation if the mice have much larger numbers of influenza-specific CD8⁺ memory T cells prior to virus challenge. The experiments utilize the H3N2→H1N1 priming protocol to expand the virus-specific CD8⁺ memory population, followed by i.n. exposure to the extremely virulent A/equine/London/72 (H7N7) influenza virus (19), which shares the NP_366–374 peptide of PR8 and HK×31 but is not neutralized by antibodies specific for the H1N1 or H3N2 hemagglutinin and neuraminidase glycoproteins. The greatest risk from the influenza A viruses is that a reassortant virus expressing elements of a pathogen like H7N7, or one of the avian influenza virus strains (6), will suddenly enter the human population and spread from person to person (39). In the absence of any preexisting neutralizing antibody, the main immune protective mechanism before a new vaccine could be developed would be cross-reactive CD8⁺ T-cell memory (23).

MATERIALS AND METHODS

Virus infection and mice.

The female B6 mice were purchased from the Jackson Laboratories, Bar Harbor, Maine, and apart from exposure to the different influenza viruses were maintained under specific-pathogen-free conditions. Mice (8 to 10 weeks old) were infected i.p. with 10^7.9 50% egg infective doses (EID₅₀) of the PR8 (H1N1) virus or were maintained as age-matched controls (1, 10). Mice were later anesthetized with Avertin (2,2,2-tribromoethanol) and then infected i.n. with 10^6.8 EID₅₀ of the HK×31 (H3N2) virus or 400 EID₅₀ of the equine (H7N7) virus (19). A few of the influenza A virus-primed mice were challenged later with 10^5.7 EID₅₀ of the unrelated B/Hong Kong/73 (B/HK) influenza B virus to look at possible nonspecific stimulation.

Generation of a mouse-adapted H7N7 virus.

The A/equine/London/72 (H7N7) influenza virus (19) was supplied by R. G. Webster at St. Jude Children's Research Hospital. This was adapted for rapid growth in mice as follows. Diluted virus (1:100; 30 μl) was given i.n. to H3N2-immune B6 mice. The lungs were taken 3 days later, homogenized, freeze-thawed, and passaged again. After four rounds of mouse infection, the virus was grown in embryonated hens' eggs to make the working stock. In a preliminary study, groups of five mice were infected i.n. with dilutions of this stock (100, 200, 400, 1,000, and 5,000 EID₅₀). Those that received 1,000 EID₅₀ were 100% susceptible while approximately 40% of the animals recovered from the 400 EID₅₀ challenge. A dose of 400 EID₅₀ was used for the pathogenesis, and 1,000 EID₅₀ was used in all subsequent experiments.

Tissue sampling and treatment.

Inflammatory cells were obtained from anesthetized infected mice by BAL (1). The BAL cells were absorbed on plastic petri dishes (Falcon, Lincoln Park, N.J.) for 60 min at 37°C to remove macrophages. Lung, brain, spleen, and blood samples were frozen (−70°C) and later homogenized for virus isolation in embryonated hens' eggs. Virus titers are expressed as log₁₀ EID₅₀ per 100 μl of homogenate. Single cell suspensions were prepared from the regional mediastinal lymph node (MLN) and spleen, and the erythrocytes were lysed. The MLN and spleen populations were enriched for CD8⁺ T cells by negative selection (17) with monoclonal antibodies (MAbs) to CD4 (GK1.5) and major histocompatibility complex class II (TIB120), followed by sheep anti-mouse and sheep anti-rat Dynabeads (Dynal A. S., Oslo, Norway). The final preparations from MLN and spleen contained 85 to 95% and 75 to 85% CD8⁺ T cells, respectively. Cerebrospinal fluid (CSF) cells were obtained (8, 25) from the cisterna magna of mice that had been anesthetized and exsanguinated and were pooled from at least three animals.

Analysis of lymphocyte proliferation.

Mice were “pulsed” with bromodeoxyuridine (BrdU) (Sigma, St. Louis, Mo.) at 0.8 mg/ml in sterile drinking water, which was given for 8 days at various times after infection (11, 38). The water containing BrdU was protected from light and changed daily. In some experiments the animals were pulsed the first 8 days after infection and then switched to normal drinking water for the “chase” experiments in order to analyze the disappearance of this thymidine analogue.

Flow cytometry.

Single cell suspensions of lymphocytes were blocked with purified anti-mouse CD16/CD32 (Fc-γIII/II receptor; PharMingen, San Diego, Calif.) and then stained for 1 h at room temperature with the D^bNP₃₆₆ tetramer, which is comprised of the influenza NP peptide ASNENMETM complexed with H-2D^b (10). Cells were subsequently stained with the 5.3-6.7 MAb to CD8α for 20 min on ice and were analyzed immediately to determine the number of virus-specific cells. For experiments assaying BrdU content, the cells were further processed as follows: the cells were resuspended in 0.5 ml of ice-cold phosphate-buffered saline, fixed by the addition of 1.2 ml of ice-cold ethanol, and held for 30 min on ice before washing and permeabilization in phosphate-buffered saline–1% paraformaldehyde–0.01% Tween 20 for 1 h at room temperature. The cells were then washed again and incubated with 50 Kunitz units of DNase (Sigma) for 10 min at 37°C. After further washing, the samples were incubated with anti-BrdU antibody (Becton Dickinson) for 30 min at room temperature, washed again, and analyzed on a FACScan using Cell Quest software (Becton Dickinson, Immunocytometry Systems, San Jose, Calif.). The analysis involved gating on total CD8⁺ T cells or CD8⁺ T cells that stained with the D^bNP₃₆₆ tetramer. At least 500 events were collected in each gate for statistical analysis. The two- or three-color flow cytometric analysis utilized fluorescein isothiocyanate-conjugated or biotinylated MAbs (all supplied by PharMingen). The fluorochromes used in the various conjugates for the BrdU analysis (11) were fluorescein isothiocyanate (BrdU), Tricolor (CD8α) and phycoerythrin (tetramer).

RESULTS

Characteristics of infection with the H7N7 virus.

One of the attractions of working with the H7N7 virus was that early mouse experiments showed evidence of substantial systemic spread (19). This was analyzed again for the 400 EID₅₀ of the mouse-passaged H7N7 stock (Fig. 1). Virus was not recovered at any time point from the blood or the spleen of the B6 mice but there was evidence of significant localization to the brain. The reason that none of the mice were sampled after day 9 (Fig. 1) is that many succumbed to the infection. Virus titers in the lung and brain remained at maximum levels on day 9, indicating that the infection was not being effectively controlled. This H7N7 model thus allows us to analyze the effect of preexisting CD8⁺ T cell-mediated immunity in both a mucosal (lung epithelium) and a remote organ (brain) site, an experiment that is not possible with most influenza A virus infections of mice.

FIG. 1.

Virus titers in lung and brain homogenates after respiratory exposure to the H7N7 influenza virus. The B6 mice were infected i.n. with 400 EID₅₀ of the H7N7 virus, and blood, spleen, lung, and brain samples were taken for virus recovery. No virus was detected in the blood or spleen. The data are cumulative from two experiments and show results for individual mice. A further study established that there is indeed an eclipse phase for the H7N7 virus. Essentially no infectious virus was detected in lung homogenates from five mice sampled (10^0.2±0.3) 4 h after i.n. challenge with 400 EID₅₀, though new virus was emerging by 8 h (10^2.0±0.3).

Control of the infection in mice primed with heterologous influenza A viruses.

The H7N7 challenge experiment (Fig. 1) was then repeated in naive mice (Fig. 2, 1°), mice that had been exposed i.p. to an H1N1 virus (Fig. 2, 2°), and H1N1-primed mice that had been further infected i.n. with an H3N2 virus (H3N2→H1N1; Fig. 2, 3°). At least 1 month elapsed between each virus dose. The H7N7 virus established as well in the respiratory tract of the double-primed animals as in the naive group (Fig. 2, lung, day 1). By day 3, however, the H3N2→H1N1-immune mice had largely controlled the H7N7 infection in the lung and only minimal amounts of virus were detected in the brain (Fig. 2, day 3, 3°). Virus was still present at substantial titers in both the lung and brain samples from the H1N1-primed mice on day 7, though the lungs (and most of the brains) were clear by day 10 (Fig. 2, 2°). The virus titers were also lower in the naive group on day 10, indicating that the immune response was beginning to limit the primary H7N7 infection (Fig. 2, day 10, 1°) in the few surviving mice.

FIG. 2.

Comparison of H7N7 virus replication in naive and single- and double-primed mice. Age-matched B6 female mice were infected i.n. with 1,000 EID₅₀ of the H7N7 virus after either no prior experience with an influenza A virus (1°), i.p. infection with an H1N1 virus (2°), or further i.n. challenge of the H1N1-primed mice with an H3N2 virus (3°). All mice were rested for at least 1 month between each infection. The lungs and brains were removed at intervals for virus titration and the analysis of virus-specific CD8⁺ T-cell numbers (Fig. 3). The data are expressed as the means ± the standard deviations for five mice per group. The remaining mice in the 1° group had succumbed by day 13 after infection.

FIG. 3.

Total number of CD8⁺ D^bNP₃₆₆⁺ T cells after primary, secondary, and tertiary influenza infection. This analysis utilized the same mice that were assayed for Fig. 2. The MLN and spleen (SPL) samples were analyzed from five individuals, while the BAL samples were pooled. The tetramer staining results are expressed as mean values (BAL) or as the means ± the standard deviations.

Quantitation of the D^bNP₃₆₆-specific CD8⁺ T-cell response.

The CD8⁺ D^bNP₃₆₆⁺ T-cell response was analyzed for MLN, spleen, and BAL populations (Fig. 3) for the same kinetic study that also quantified virus titers in the lung and brain (Fig. 2). The localization of virus-specific CD8⁺ T cells to the brain was not measured sequentially, though we did show that a substantial CD8⁺ D^bNP₃₆₆⁺ set could be detected in the CSF (Fig. 4). The numbers of D^bNP₃₆₆-specific memory T cells present in lymphoid tissue prior to challenge with the H7N7 virus reached levels that were barely detectable by flow cytometric analysis of MLN and spleen from the H1N1-immune mice (Fig. 3, day 0, 2°). These PR8-primed mice were partially protected from the consequences of challenge with the H7N7 virus (Fig. 2, day 10, 2°), though the expansion of the CD8⁺ population was neither as rapid nor as massive as that seen previously following the comparable challenge with the less virulent H3N2 virus (Fig. 3, MLN and spleen, 2°). Even so, greatly increased numbers of D^bNP₃₆₆-specific CD8⁺ T cells were recovered from the BAL by day 7 after the H7N7 infection (Fig. 3, BAL, 2°, days 7 and 10).

FIG. 4.

Visualization of CD8⁺ D^bNP₃₆₆⁺-specific populations recovered from different anatomical sites 7 days after i.n. challenge of H3N2→H1N1-primed mice with the H7N7 virus. The CSF samples were obtained from the cisterna magna of anesthetized, exsanguinated mice.

The BAL counts for the CD8⁺ D^bNP₃₆₆⁺-specific set were higher earlier for the H7N7→H3N2→H1N1 (Fig. 3, 3°) than for the H7N7→H1N1 challenge (Fig. 3, 2°) but never reached the massive levels recorded on days 7 and 10 for the secondary response (Fig. 3, compare BAL, 3°, and BAL, 2°). This presumably reflects the lesser antigenic load resulting from the more rapid control of virus growth in the lungs of the double-primed (H3N2→H1N1) mice. However, the tertiary response to the H7N7 virus was associated with little change in the magnitude of the CD8⁺ D^bNP₃₆₆⁺ population in the MLN and spleen (Fig. 3, 3°), with the values being no greater than those found for these sites immediately prior to challenge (Fig. 3, day 0, 3°).

The CD8⁺ T-cell response is protective.

The tetramer analysis (Fig. 3) provided indirect evidence that the H3N2→H1N1 immunization regimen confers a high level of protection (Fig. 2) against challenge with the virulent H7N7 virus but did not formally show that this was mediated by the recall of virus-specific CD8⁺ T-cell memory. Mice that had been double-primed (H3N2→H1N1) for 8 months were thus treated with the 2.43.1 MAb to CD8 (or with a rat immunoglobulin control) every second day, beginning 6 days prior to i.n. challenge with the H7N7 virus (18). This procedure removes >99.0% of the CD8⁺ T cells. The lung virus titers at 5 days after infection were 10^5.8±0.7 for the CD8-depleted mice and 10^3.0±0.7 for the controls. Other mechanisms of protection, such as immunoglobulin-mediated effects, either specific for shared components like the matrix 2 transmembrane protein or nonspecific (27, 28), do not seem to be playing a critical role during the initial infection, as the levels of virus in the lung are the same 24 h after i.n. infection with the H7N7 virus in naive, primary, and secondary mice (Fig. 2). Furthermore, our previous studies with the less virulent H3N2 virus showed that CD4⁺ memory T cells are inefficient at protecting mice in the absence of antibody and CD8⁺ T cells (34). Thus, the increased viral titers in the CD8-depleted mice clearly suggest that the recall CD8⁺ response is the major effector of protective immunity in these mice primed with heterologous influenza A viruses.

Extent of clonal expansion when CD8⁺ memory T-cell numbers are high.

The size of the CD8⁺ D^bNP₃₆₆⁺-specific sets in the MLN and spleen increased greatly following the H7N7 challenge of the single (H1N1)- but not the double (H3N2→H1N1)-primed mice (Fig. 3). This raised the possibility, as suggested many years ago (14), that there is little clonal expansion following antigenic stimulation when memory CD8⁺ T-cell numbers are high. Though most, if not all, D^bNP₃₆₆⁺-specific CD8⁺ T cells incorporated the BrdU thymidine analogue (supplied in drinking water) when H1N1-immune mice were exposed i.n. to the H3N2 virus (Fig. 5, Pulse, day 10), the turnover rate of the persisting D^bNP₃₆₆⁺-specific memory CD8⁺ population was low (Fig. 5, Pulse, days 15 to 90, and Chase, days 10 to 90). To what extent does i.n. challenge with the H7N7 virus modify the cycling characteristics of this substantial set of CD8⁺ D^bNP₃₆₆⁺-specific memory T cells in H3N2→H1N1-immune mice?

FIG. 5.

Acute proliferation and long-term cycling of CD8⁺ D^bNP₃₆₆⁺ T cells in secondarily stimulated (H3N2→H1N1) mice. The mice in the pulse analysis were given BrdU in drinking water for 8 days prior to sampling. Those in the chase study were fed BrdU from day 0 to 8 following secondary challenge. The results, which show only the values for the BrdU^hi subset, are the means ± the standard deviations for groups of five mice.

The question was addressed in two ways. Mice that had been double-primed 6 months previously were given drinking water containing bromodeoxyuridine (BrdU) throughout the course of i.n. infection with the H7N7 virus or with an unrelated influenza B virus (left half of Fig. 6, Pulse). In addition, the “chase” mice (Fig. 5) that had been exposed to BrdU at the time of the H3N2→H1N1 challenge were further infected with the H7N7 virus (right half of Fig. 6, Chase). The two approaches gave similar results. In both cases almost all the CD8⁺ D^bNP₃₆₆⁺-specific T cells recovered from the BAL had clearly gone through one or more cycles of cell division. The “pulsed” CD8⁺ set (left side of Fig. 6, Pulse) had incorporated BrdU (BrdU^hi), while the BrdU had been lost (BrdU^lo) from the majority of the memory T cells that had retained this thymidine analogue (BrdU^hi) after incorporation (Fig. 5) more than 6 months previously (Fig. 6, Chase). The BrdU-staining characteristics of these CD8⁺ D^bNP₃₆₆⁺-specific T cells in the absence of antigen challenge are shown for the “resting” MLN and spleen populations in Fig. 6. The influenza B virus clearly caused some “bystander” proliferation of the CD8⁺ D^bNP₃₆₆⁺-specific population, though this was less than that associated with the H7N7 challenge in all sites sampled (Fig. 6, Pulse, FluB, BrdU^hi).

FIG. 6.

Cycling characteristics of the CD8⁺ D^bNP₃₆₆⁺ set recovered 8 days after i.n. challenge of double-primed (H3N2→H1N1) mice with the H7N7 virus or with an influenza B virus. All mice were injected i.p. with the H1N1 virus, given the H3N2 virus i.n. 1 month later, and rested for a further 6 months before i.n. challenge with the H7N7 virus or the B/HK (FluB) virus. The mice in the pulse experiment (left half) were given BrdU in the drinking water for the 8 days after i.n. exposure to the H7N7 or B/HK virus, while those in the chase study (right half) had been fed BrdU 6 months previously, at the time of the secondary H3N2 stimulation. Cycling through the pulse analysis is thus characterized by BrdU incorporation (BrdU^hi) and through the chase study by the loss of BrdU (BrdU^lo).

Only a portion of the D^bNP₃₆₆⁺-specific CD8⁺ T cells in the MLN and spleen were, however, induced to cycle following the H7N7 challenge. Some in the pulse experiment remained BrdU^lo (Fig. 6, Pulse), a pattern that was not seen for the H3N2→H1N1 challenge (Fig. 5, day 10), while a portion in the chase study remained BrdU^hi (Fig. 6, Chase). The latter set in the chase experiment will, of course, tend to be diluted out by the dividing BrdU^lo population. The converse obviously applies for the pulse experiment.

DISCUSSION

The present analysis shows that a greatly expanded pool of virus-specific CD8⁺ memory T cells provides substantial protection against respiratory challenge with an extremely virulent influenza A virus. The mechanism by which the CD8⁺ memory T cells are acting could be either direct cytotoxicity of infected cells, a cytokine-mediated process, or some combination of mechanisms. The H7N7 infection still became established in the respiratory tract of the H3N2→H1N1-primed mice but was rapidly controlled, and the mice remained clinically normal. The much smaller secondary response (H7N7→H1N1) also provided some measure of protection. It thus seems reasonable to think that immunizing people with a live, attenuated heterologous influenza A virus (7) or by some other protocol that introduces shared peptides into the major histocompatibility complex class I processing pathway (12) might have some positive effect in the face of a pandemic caused by a novel influenza A virus. The case for heterologous priming with live virus is also strengthened by recent evidence that antibody to components common to diverse influenza A viruses can have some protective effect (27).

A comparable double-priming protocol that used recombinant vectors expressing the same viral peptide provided almost complete protection against the respiratory growth phase of a murine γ-herpesvirus, though there was no long-term effect on the establishment of viral latency (33). The massive numbers of influenza-specific CD8⁺ memory T cells available in the H3N2→H1N1-immune mice failed to completely prevent the establishment of respiratory infection, with the H7N7 virus titers being as high as those in the naive group 24 h after i.n. challenge. The absence of early control also indicates that antibody (27) to shared viral components (such as the transmembrane matrix 2 protein) did not play a significant protective role in the early phase of these H7N7 challenge experiments.

Whether or not priming only the CD8⁺ T-cell compartment is a useful vaccine strategy for protection against persistent infectious agents is thus likely to depend on the nature of the pathogen. Malaria, for example, is gradually controlled by the immune response in those that survive, though the parasite may persist and the process can take years. Providing a “jump start” for this process with appropriate vectors carrying CD8⁺ T-cell epitopes could thus speed recovery (26). The relatively low growth rates of tumor cells (3, 18, 39) when compared with those of viruses also favor control by primed CD8⁺ T cells. Are we likely to see the same benefit for the immunodeficiency viruses (9, 31, 32) which, once established, progressively subvert and destroy the immune system?

The analysis of cycling characteristics for the D^bNP₃₆₆⁺-specific CD8⁺ set indicated that only a portion of the greatly expanded memory population in the lymphoid tissue of the double-primed (H3N2→H1N1) mice was stimulated to divide as a consequence of the H7N7 challenge. This probably reflects limited exposure to antigen, as the H7N7 virus was not shown to replicate in either the regional lymph nodes or the spleen. The great majority of the CD8⁺ D^bNP₃₆₆⁺ T cells in the BAL population recovered from the site of maximal virus growth in the respiratory tract had, however, cycled at least once. Even so, prior cell division was not a prerequisite for localization to the pneumonic lung. Infection with an unrelated influenza B virus induced some “bystander” proliferation (29, 35, 38) of the CD8⁺ D^bNP₃₆₆⁺ memory set, but many of the D^bNP₃₆₆⁺-specific CD8⁺ T cells in the BAL population recovered from the influenza B virus-infected mice had not incorporated BrdU. This comparison between the consequences of challenge with a cross-reactive (H7N7) and an irrelevant (FluB) virus raises the possibility that antigen-driven proliferation of the inflammatory CD8⁺ D^bNP₃₆₆⁺ T cells continues after extravasation into the H7N7-infected lung environment.

Once the virus-specific CD8⁺ T-cell compartment is primed, the degree to which any memory T cell is involved in the recall response is thus directly correlated with the magnitude of further antigen challenge and inversely related to the number of available precursors. Practically all the virus-specific effectors that localize to the site of virus-induced pathology are induced to divide, either in the lymphoid tissue prior to exit into the blood and extravasation into the infected tissue or following exposure to antigen-presenting stimulator cells in the target organ. Whether such stimulation can result from direct contact with infected epithelial cells or requires an encounter with an antigen-presenting dendritic cell (15, 24) is not clear. What is apparent for this influenza mouse model is that the magnitude of virus-specific CD8⁺ T-cell recruitment to the respiratory tract is a direct function of the extent of virus growth and the consequent lung damage.

In conclusion, there can now be no doubt that the recall of virus-specific CD8⁺ T-cell memory has the potential to prevent fatal influenza, though it is questionable that memory T-cell populations equivalent in size to the D^bNP₃₆₆⁺-specific CD8⁺ set detected in double-primed (H3N2→H1N1) B6 mice (10, 11) could ever be achieved in humans. The preferred vaccine strategy is clearly to use an attenuated (or inactivated) version of the pandemic virus or a closely related virus (30). However, if such a product is not immediately available, giving a live, attenuated influenza A virus vaccine that is not likely to be neutralized by high levels of preexisting antibody (30) may be of some protective value in the face of a major outbreak.