Female C57Bl/6 mice, 6–8 weeks old, were purchased from the National Cancer Institute (Fredericksburg, MD, USA). The study protocol was approved by University of Iowa Animal Care and Use committee. The animals were treated and handled in accordance with guidelines established by the committee.

All IAV infections were performed while mice were under the influence of ketamine/xylazine anesthesia. Virus inoculum was intranasally (IN) delivered in PBS in a final volume of 50 µl.

Secondary memory (2°M) CD8 T cell responses were generated in C57/Bl6 mice in two steps. First, mice were IN infected with 7 × 10³ TCID₅₀ (median tissue culture infective dose) of A/H3N2/X31. Seventy days later (D0), mice were exposed to secondary IN infection with 10⁴ TCID₅₀ of A/H1N1/S12a (SEQ12) (PR8 escape mutant). The S12a mutant virus contains 12 amino acid substitutions in the globular domain of hemagglutinin (HA). It was derived from PR8 virus through 12 sequential selection steps in the presence of anti-HA monoclonal antibodies. This approach allowed for generation of HA drifted virus strain, with no effect on other antigenic determinants (18). Alignment of sequences derived from S13 mutant virus (a descendant of S12a) and wt PR8 showed absence of variation in sequences encoding nucleoprotein (NP) and polymerase acidic (PA) proteins bearing target CD8 T cell epitopes (18).

Primary memory (1°M) CD8 T cell responses were generated on D0 by IN infection with X31 virus. 1°M and 2°M CD8 T cells were analyzed 70–90 days postinfection. For the purpose of longitudinal analysis of the NP₃₆₆-specific response, blood was obtained on days 10, 50, and 100 postinfection and analysis performed as previously described (8, 9).

Protection against pulmonary virus infection was assessed 70–90 days post-last IAV exposure. For this purpose, 1°M and 2°M mice were exposed to IN challenge with 10⁷ or 10⁸ PFU of recombinant vaccinia virus (Vac) expressing full-length influenza NP (Vac-NP) or mock infected with Vac-OVA (both viruses are obtained from Dr. Jack Bennink, Laboratory of Viral Diseases, National Institute of Allergy and Infectious Diseases). Lungs were harvested 3 days post-challenge, homogenized, frozen, and kept at −80°C until used for titration. For determining Vac virus titers, BSC-40 cells were infected with 10× dilutions of lung supernatant in Iscove’s DMEM medium supplemented with 50 µg/ml gentamicin, 100 U/ml penicillin, and 100 µg/ml streptomycin. Cells were incubated with the lung virus samples for 1 h at 37°C, with occasional rocking and afterward overlaid with complete DMEM. After in total 48 h incubation at 37°C, medium was aspirated, cells were fixed with 7% formaldehyde and stained with 0.5% crystal violet in 10% methanol, and plaques were counted (10, 19, 20).

To distinguish cells in the lung vasculature from those imbedded in the lung parenchyma, 3 min prior to sacrifice mice were intravenous (IVS) injected with 2 µg of anti-CD45.2 antibody (clone 104, BioLegend, San Diego, CA, USA) in PBS (21). Harvested lung was cut into small pieces and incubated for 1 h at 37°C in digestion medium composed of 125 U/ml collagenase and 0.1 mg/ml DNAse. After incubation, tissue was processed into single-cell suspension using cell strainers and leukocytes enriched by centrifugation in 35% Percoll in HBSS. Red blood cells were lysed using Vitalyze.

Antigen-specific CD8 T cells were identified by labeling with D^b/NP₃₆₆ or D^b/PA₂₂₄ in house prepared tetramers for 1 h at 4°C (22). Tetramer staining was followed by surface staining with appropriate antibody cocktails for 20 min at 4°C. Surface markers were stained using following antibodies: anti-CD8 (clone 53-6.7, BioLegend), anti-CD90.2 (clone 30-H12, BioLegend), anti-CD45.2 (clone 104, BioLegend), anti-CD103 (clone 2E7, BioLegend), anti-CD69 (clone H.12F3, BioLegend), anti-KLRG-1 (clone 2F1, eBioscience, San Diego, CA, USA), anti-CD127 (clone A7R34, BioLegend), anti-CX3CR1 (clone SA011F11, BioLegend), anti-CXCR3 (clone CXCR3-173, BioLegend), and anti-CD49a (clone Ha31/8, BD Pharmingen). Intracellular cytokine staining was performed using anti-IFNγ (clone XMG1.2, BioLegend), anti-TNF (clone MP6-XT22, BioLegend), and anti-IL2 (clone JES6-5H4, BioLegend) antibodies. Proliferation of CD8 T cells was assessed by intracellular staining with anti-Ki67 (clone MOPC-21, BD Pharmingen). Flow cytometry data were acquired using LSRFortessa (Becton Dickinson, Rutherford, NY, USA) and analyzed using the FlowJo software (Tree Star Inc., Ashland, OR, USA).

The major aim of this study is to investigate the influence of repeated localized pulmonary infections on shaping the pathogen-specific memory CD8 T cell compartment. For this purpose, we took advantage of a well-established mouse model of IAV infections (23–25) and generated virus-specific 1°M and 2°M CD8 T cells by exposing naive wt C57Bl/6 mice to one or two intranasal IAV infections, respectively. The selected virus strains (H3N2 X31 and H1N1 S12a) share some common gene segments that encode virus core proteins (e.g., NP and PA protein) and thus CD8 T cells epitopes (NP₃₆₆ and PA₂₂₄), enabling successful boosting or primary memory CD8 T cell response by secondary infection (26, 27). This approach allowed us to study and compare the development of endogenous 1°M and 2°M CD8 T responses in an intact, wt host.

To be able to collect samples and perform analysis of both 1°M and 2°M CD8 T cells at the same time and this way minimize the variability between assays, we adopted the infection scheme depicted in the Figure 1A. Namely, 2°M CD8 T cell responses were generated in two steps: primary infection with H3N2 X31 followed 70 days later by secondary infection with H1N1 S12a. At the same time of secondary infection, 1°M CD8 T cell responses were generated in a separate group of mice by exposure to H3N2 X31. Mice harboring 1°M or 2°M CD8 T cell responses were sacrificed in groups of 4–5 mice on days 70–90 after the last infection, and analyses were performed. Longitudinal analysis of NP₃₆₆-specific response was performed in a separate group of mice, and blood for this purpose was collected at days 10, 50, and 100.

To study how repeated exposures to localized pulmonary infection influence the magnitude of memory CD8 T cell response, we focused our analysis on 1°M and 2°M CD8 T cell responses specific for two major epitopes: NP₃₆₆ (Figures 1B–H) and PA₂₂₄ (Figure S1 in Supplementary Material). Antigen-specific CD8 T cells were enumerated using MHC-I tetramer staining. Longitudinal analysis of the NP₃₆₆-specific response in blood shows that secondary localized infection generates CD8 T cell responses that are of superior magnitude (~10× higher) compared to 1°M responses. Importantly, these numerical differences are conserved throughout the memory phase (D50 and D100) (Figure 1B).

Snapshot analysis of NP₃₆₆-specific CD8 T cell responses performed 70–90 days after the last infection in multiple tissues confirmed our observation from the blood compartment (Figure 1C). In both spleen and draining mediastinal lymph nodes, 2°M NP₃₆₆-specific responses exceeded 1°M responses by ~8× fold (Figures 1D,E). Similar results were observed with PA₂₂₄-specific 1°M and 2°M CD8 T cell responses (Figures S1A–C in Supplementary Material).

Although antigen-specific CD8 T cell numbers measured in blood and lymphoid compartment can be used as a general assessment of the magnitude of virus-specific CD8 T cell response, characterization of the response in the lung, the target organ of respiratory pathogens, is essential. Furthermore, a major focus has recently been put on the population of tissue resident memory CD8 T cells (T_rm) as the important mediators of protection against localized infections (6, 28, 29). As lung perfusion has shown to be inefficient in removing all the cells trapped in the lung vasculature and has the capacity to destroy lung parenchyma, we administered an intravascular (IV) CD45.2 stain to discriminate between CD8 T cells imbedded in the tissue parenchyma (IV−) from those in the capillaries (IV+) (Figure 1F) (30, 31). Results of NP₃₆₆ tetramer staining of lung cells clearly show a large difference in magnitude of 1°M and 2°M CD8 T cell responses in both the IV+ and IV− lung compartment. Relative abundance of NP₃₆₆-specific CD8 T cells, presented in Figure 1F as % of total IV+ or IV− CD8 T cells, was uniformly higher in lungs of mice exposed to two sequential infections compared to the lungs of mice exposed to a single infection. The difference in absolute numbers of NP₃₆₆-specific 1°M and 2°M CD8 T cells (Figure 1G) was 14× and 5.6× in the lung parenchyma and the vasculature, respectively. Finally, enumeration of the NP₃₆₆-specific IV− CD8 T cells expressing a conventional T_rm phenotype (CD69+CD103+) (32, 33) shows that secondary exposure causes >10× increase in size of T_rm population, relative to a single exposure (Figure 1H). A similar pattern was observed with PA₂₂₄-specific CD8 T cells (Figures S1D–F in Supplementary Material).

In summary, the data show that the size of antigen-specific 2°M CD8 T cell compartment dramatically exceeds the size of 1°M compartment at all time points after infection. Although somewhat enhanced in the lung parenchyma, this trend is well preserved, in all the analyzed tissues. Given the notion that the CD8 T cell-dependent protection against a pathogen infection is mainly determined by the magnitude of the response (34, 35), our results suggest that higher-magnitude responses induced by secondary infection could provide superior protection against subsequent exposure to a pathogen expressing cognate antigen.

Next, we determined whether the substantial increase in numbers of antigen-specific memory CD8 T cells is associated with superior protection of the lung against localized challenge. In order to determine CD8 T cell-mediated protection and avoid the contribution of non-CD8 T cell protective responses potentially induced by previous infections (36–38), we exposed mice to IN challenge with a high dose (10⁷ PFU) of recombinant vaccinia virus expressing full-length influenza PR8 NP (Vac-NP) (Figure 2A).

To formally confirm the role of CD8 T cells as the only mediators of pulmonary protection in this challenge model, mice were intraperitoneally (IP) and IN inoculated with either CD8-depleting antibody or control rat IgG. As shown in Figure 2B, this regime was sufficient to deplete systemic (peripheral blood) and lung-residing CD8 T cells (Figure 2B). Pulmonary protection was assessed by measuring lung virus titers 3 days after infection—time point experimentally determined as the peak of the viral load (data not shown). Both 1°M and 2°M CD8 T cell-bearing mice were able to control the lung Vac-NP infection (Figure 2C). Although we cannot completely rule out the contribution of non-CD8 T cell-mediated cross-protective immunity against vaccinia infection (39), the complete loss of control of lung viral load in both groups of mice after depletion of CD8 T cells strongly suggests that CD8 T cells are the main mediators of the observed pulmonary protection. To our surprise, despite the large difference in magnitude of memory CD8 T cell responses, both 1°M and 2°M CD8 T cell-bearing mice displayed similar control of lung virus burden (Figure 2C). In both groups, we observed a reduction in virus titers by ~100× relative to naive controls. To rule out the possibility that the challenge dose of Vac-NP was not high enough to reveal potential differences in protection between mice containing 1°M and 2°M CD8 T cells, we increased the challenge dose by 10-fold (10⁸ PFU) (Figure 2D). To make the experimental controls more stringent, we infected groups of 1°M and 2°M CD8 T cell-bearing mice with Vac virus expressing chicken ovalbumin (Vac-OVA) as a non-relevant antigen (Figure 2D). Interestingly, in spite of the 10× increase in challenge dose of Vac-NP virus, similar level of pulmonary protection (>500× reduction relative to naive controls) was observed in both 1°M and 2°M CD8 T cell-bearing mice (Figure 2E). Importantly, both groups of mice challenged with Vac-NP showed a statistically significant decrease in lung virus titers relative to the infection- and age-matched controls challenged with Vac-OVA (Figure 2F).

Thus, despite the superior magnitude of 2°M CD8 T cell response observed in all the tissue compartments, the level of protection against localized pulmonary challenge provided by 1°M and 2°M responses is unexpectedly similar when assessed 70–90 days postinfection. This finding suggests that, in addition to sole numbers of memory CD8 T cells, other cell intrinsic factors play an important role in shaping the protection against localized pulmonary challenge.

To address the possibility that secondary infection induces critical phenotypic and/or functional changes of memory CD8 T cells, this way diminishing their protective capacity, we performed detailed analysis of the memory CD8 T cell pool isolated from lungs of 1°M and 2°M mice 70–90 days post-last infection. Phenotypic characterization evaluating effector (T_em) vs central memory (T_cm) properties (KLRG-1, CD127), trafficking (CXCR3), response to inflammatory chemokines (CX3CR1), and tissue localization (CD49a) (8, 40, 41), revealed no major difference between memory CD8 T cells isolated from lungs (IV+ or IV−) of 1°M and 2°M CD8 T cell-bearing mice (Figure 3A). Of note, secondary IAV infection, similar to systemic infections, induced slower acquisition of CD62L expression (T_cm properties) in memory spleen- and blood-derived CD8 T cells (Figure 3B). At the same time, both 1°M and 2°M lung-residing CD8 T cells remained CD62L^lo (data not shown).

Furthermore, the functional characteristics of lung-derived memory CD8 T cells were assessed by measuring production of IFNγ, TNF, and IL2 after ex vivo NP₃₆₆ peptide stimulation. As depicted in Figure 3C, we observed no major difference in functionality of 1°M and 2°M cells, as they were equally able to produce IFNγ, TNF, and IL2 upon peptide restimulation. Importantly, normalizing the numbers of IFNγ-producing CD8 T cells to total number of NP₃₆₆-specific CD8 T cells enumerated by tetramer staining revealed that similar percentage of antigen-specific 1°M and 2°M CD8 T cells produce IFNγ after ex vivo peptide restimulation (Figure 3D). Finally, a similar fraction of CD8 T cells capable of producing 1, 2, or 3 cytokines upon ex vivo peptide restimulation was observed in both 1°M and 2°M CD8 T cell-bearing mice (Figure 3E). Thus, the similar level of localized pulmonary protection observed in mice harboring 1°M and 2°M CD8 T cells was not due to altered phenotype or compromised functionality of 2°M CD8 T cells.

Similar level of pulmonary protection observed in 1°M and 2°M CD8 T cell-bearing mice could alternatively be explained by a substantial increase in numbers of 1°M CD8 T cells in the lung during the course of the challenge infection. This would potentially allow 1°M CD8 T cells to compensate for the initially low numbers. To directly test this hypothesis, we followed the kinetics of the antigen-specific CD8 T cell response upon pulmonary Vac-NP challenge. To this end, 1°M and 2°M CD8 T cell-bearing mice were exposed to IN challenge with 10⁷ PFU of Vac-NP. Small group of mice was sacrificed before infection (D0) to establish a “baseline” NP₃₆₆-specific CD8 T cell response in the lung and peripheral blood. Infected mice were sacrificed either on day 2 or 3 postinfection, and a snapshot analysis of NP₃₆₆-specific CD8 T cell responses in the lung and blood was performed. On day of infection (D0), numbers of total or IV− NP₃₆₆-specific CD8 T cells in lungs of 2°M CD8 T cell-bearing mice were, respectively, 7.5× and 11× higher compared to the lungs of 1°M CD8 T cell-bearing mice (Figures 4A,B). Furthermore, steep increase in numbers of antigen-specific CD8 T cells was observed after the onset of the infection in both groups of mice. Interestingly, while numbers of antigen-specific 1°M cells continued to rise by day 3 postinfection, numbers of 2°M cells reached a plateau by day 2 postinfection. Due to this non-proportional expansion in the population size of 1°M and 2°M CD8 T cells in the lung, initial numerical differences were completely gone (total NP₃₆₆-specific CD8 T cells) or substantially diminished (IV− NP₃₆₆-specific CD8 T cells) by day 3 postinfection (Figures 4A–C). Analyses of NP₃₆₆-specific CD8 T cells in blood revealed that while the size of the 2°M population seemed to be stable, the size of the 1°M population decreased by day 3 postinfection (Figures 4D,E), suggesting possible relocation of 1°M cells from the circulation into the lung compartment.

The major increase of 1°M cells in the infected lung could occur as a consequence of the substantial relocation of cells from the blood to the lung and/or due to a superior proliferative capacity of 1°M cells in the lung. To formally test this, 70–90 days after the last IAV infection, 1°M and 2°M CD8 T cell-bearing mice were IN challenged with 10⁷ PFU Vac-NP, and half of them were injected IP with pertussis toxin (Ptx) before (D0) and 2 days after infection to block the access of circulating memory CD8 T cells into the infected lung (Figure 5A) (20). Mice were sacrificed 3 days post-Vac-NP infection, and lung Vac titers and NP₃₆₆-specific CD8 T cell responses were assessed. Proliferation of antigen-specific CD8 T cells was assessed in both blood and lung by staining for Ki67, a cellular marker of proliferation (42). As shown in Figures 5B,C, a substantially higher proportion of 1°M relative to 2°M CD8 T cells had entered one of the active phases of the cell cycle (based on Ki67 expression) in the inflamed lung but not in the blood on day 3 postinfection. Additionally, Ptx treatment caused >80× decrease in numbers of IV− 1°M NP₃₆₆-specific CD8 T cells in the lung, compared to <15× decrease of 2°M. At the same time, Ptx-induced retention of antigen-specific CD8 T cells in blood of 1°M CD8 T cell-bearing mice was more pronounced (16× increase in numbers after Ptx treatment) than in blood of 2°M CD8 T cell-bearing mice (4.5× increase) (Figure 5D).

Importantly, and consistent with the differential impact of Ptx treatment on relocation of 1°M and 2°M CD8 T cells from circulatory to the lung compartment, Ptx-induced block had a greater impact on the control of infection in mice harboring 1°M CD8 T cells relative to those harboring 2°M CD8 T cells. The substantial Ptx-induced decrease in numbers of 1°M CD8 T cells in the lung parenchyma was accompanied by a substantial decrease in protection. As depicted in Figure 5E, ~7× higher virus titers were measured in lungs of 1°M CD8 T cell-bearing mice treated with Ptx as compared to 2°M CD8 T cell-bearing mice. Taken together, the presented data suggest that the superior proliferative capacity of 1°M CD8 T cells in combination with increased recruitment into the infected lung efficiently compensates for the initial lower numbers of antigen-specific 1°M relative to 2°M CD8 T cells, which eventually allows for similar control of localized pulmonary infection in both groups of mice.

Thus, CD8 T cell-mediated protection against localized pathogen infections is not exclusively determined by the magnitude of the response but is also influenced by multiple intrinsic properties (e.g., proliferation and migration to infected tissue) of memory CD8 T cells. History of exposure to localized infections, by shaping such cell properties, determines the protection against subsequent pathogen challenge.