Lung-resident memory CD8 T cells (T_RM) are indispensable for optimal cross-protection against pulmonary virus infection

Pulmonary immunization is required for generating lungresident memory CD8 T cells, which confine infection within the bronchial tree during heterosubtypic challenge.

Abstract

Previous studies have shown that some respiratory virus infections leave local populations of tissue T_RM cells in the lungs which disappear as heterosubtypic immunity declines. The location of these T_RM cells and their contribution to the protective CTL response have not been clearly defined. Here, fluorescence microscopy is used to show that some CD103⁺ T_RM cells remain embedded in the walls of the large airways long after pulmonary immunization but are absent from systemically primed mice. Viral clearance from the lungs of the locally immunized mice precedes the development of a robust T_eff response in the lungs. Whereas large numbers of virus-specific CTLs collect around the bronchial tree during viral clearance, there is little involvement of the remaining lung tissue. Much larger numbers of T_EM cells enter the lungs of the systemically immunized animals but do not prevent extensive viral replication or damage to the alveoli. Together, these experiments show that virus-specific antibodies and T_RM cells are both required for optimal heterosubtypic immunity, whereas circulating memory CD8 T cells do not substantially alter the course of disease.

Introduction

CTLs play an important role in clearing infectious influenza virus from the lungs [1]. The major targets of the CTL response include peptides from antigenically conserved viral proteins, such as NP, polymerase acidic protein, and polymerase basic protein 2 [2]. Cross-reactive CTLs that recognize these conserved peptides can provide protection between serologically distinct strains of influenza virus, known as heterosubtypic immunity [3–5]. Cell-mediated immunity declines within ∼6 months after primary infection, in spite of the fact that large numbers of virus-specific memory CD8 T cells circulate through the lymphoid tissues of the infected animals for at least 2 years [6].

Two major populations of virus-specific memory CD8 T cells remain in the circulation after viral clearance [7, 8]. T_CM circulate through the secondary lymphoid tissues and enter encapsulated LNs using CD62L and CCR7 to cross high endothelial venules. T_EM cells lack CD62L and CCR7 expression and preferentially circulate through the red pulp of the spleen and nonlymphoid tissues. A third population of virus-specific memory CD8 T cells also populates some peripheral tissues after local infections, which maintain stable CD69 and CD103 expression [9–12]. These activated memory CD8 T cells do not return to the circulation after viral clearance [11, 13] and are therefore known as tissue T_RM cells [12]. Stable populations of T_RM cells have been detected in the epithelial layers of the skin [14], intestines [15], and selected vital organs, such as the kidneys [16] and brain [17]. Low levels of BrdU incorporation indicate that the T_RM population in the lungs is maintained by extended lifespan and small numbers of new arrivals from the circulation [9, 11, 13, 18].

MHCI tetramers have been used extensively to analyze the kinetics of the CTL response in the lungs after heterosubtypic challenge [19, 20]. Some data indicate that different populations of virus-specific memory CD8 T cells join in the protective CTL response with different kinetics [21] and support a model that suggests that some long-lived, virus-specific CTLs release proinflammatory molecules into the local tissues after reinfection and thus attract other immune cells to the infection. This idea is supported by the finding that nonproliferating CTLs (virus-specific and bystander) can be attracted into the lumen of airways by nonspecific inflammation but do not persist without antigen stimulation [13, 19]. When antigen is carried to the draining nodes, T_CM cells complete several rounds of cell division before entering the lungs a few days later [20]. Whether T_CM, T_EM, and T_RM cells are all required for heterosubtypic immunity is not known. Furthermore, changes in the distribution of virus-specific CTLs inside the lungs after reinfection have not been documented.

In this study, mice were primed with influenza virus by different routes of inoculation and challenged with a serologically distinct strain to determine how protective CTLs provide heterosubtypic immunity in the lungs. Upon reinfection, the lungs were analyzed by fluorescence microscopy to follow changes in the distribution of antigen-specific CTLs as the infection progressed. Substantial numbers of virus-specific T_RM cells remained embedded in the walls of the large airways after priming by pulmonary inoculation. A robust T_eff response in the lungs had little impact on the course of the disease after reinfection, as reduced viral titers were recorded before large numbers T_eff reached the infected tissues. Heterosubtypic immunity was reduced substantially, 7 months after priming, and very few T_RM cells remained in the lungs. Together, our data show that virus-specific T_RM cells play an essential role in optimal heterosubtypic immunity and limit the severity of the pathology in the lungs.

MATERIALS AND METHODS

Mice and reagents

B6 and congenic CD45.1⁺ mice were purchased from Charles River Laboratories (Wilmington, MA, USA) through the National Cancer Institute animal program. The μMT mice were purchased from The Jackson Laboratory (Bar Harbor, ME, USA). The OTI TcR transgenic mice have been described previously [22]. WSN-OVA_I and X31-OVA are recombinant influenza viruses that express a peptide from the chicken OVA gene [23, 24]. Between 8 and 14 weeks of age, female mice were anesthetized with 2,2,2-tribromoethanol (Avertin) before i.n. infection with 10³ PFU of WSN-OVA_I or 10⁴ PFU of WSN-OVA_I, given by i.p. injection. The mice were reinfected with 5 × 10³ PFU of X31-OVA. Virus stocks were grown in fertilized chicken eggs, and titers were determined as described previously [25].

Transfer experiments and flow cytometry

Spleens and peripheral LNs from CD45.1⁺ OTI mice were dissociated and depleted of red blood cells. Recipient mice received 10³ naive OTI cells by tail vein injection. Lymphocytes were released from the chopped lung tissues by digestion with 150 U/ml collagenase (Life Technologies, Rockville, MD, USA) in RPMI medium, supplemented with 1 mM MgCl₂, 1 mM CaCl₂, and 5% FBS at 37°C for 90 min. Nonadherent cells were enriched on 44/67% Percoll gradients at 400 g for 20 min. For flow analysis, washed lymphocytes were stained with MHCI tetramers for 1 h at room temperature. The NP_366–374/D^b tetramer has been described previously [26]. Lymphocytes were stained with PE or allophycocyanin-conjugated tetramers and anti-CD8 (clone 53.6.72). All other markers were stained at 4°C using mAb specific for CD45.1, CD45.2, CD44, CD62L, PD-1, and CD103 (eBioscience, San Diego, CA, USA; or BD PharMingen, San Diego, CA, USA). Samples were analyzed on a Becton Dickinson LSR II flow cytometer and analyzed using FlowJo software (Tree Star, Ashland, OR, USA).

Whole-mount confocal laser microscopy

Fragments of lung and spleen tissues and 350 μm thick vibratome sections of MLNs were fixed in 1% PFA for 1 h at 4°C. Tissues were washed and stained for 6 h at 4°C in round-bottomed, 24-well plates with biotin-conjugated EpCAM antibody (eBioscience), diluted in 2% FBS/PBS solution. The tissues were washed extensively at 4°C in PBS and stained overnight at 4°C with streptavidin-Cy3 antibody (Jackson ImmunoResearch, West Grove, PA, USA) and Alexa Flour 647-conjugated anti-CD31 and Alexa Flour 488-conjugated anti-CD45.1 (BioLegend, San Diego, CA, USA). B cells were detected with anti-B220 conjugated to PE (BioLegend). Stained tissues were washed extensively and then mounted on slides using Shandon Immu-Mount (Thermo Electron, Pittsburgh, PA, USA). Images were collected using a Zeiss LSM 510 Meta confocal microscope, mounted on an Axiovert 100M with automated XYZ control. This was equipped with an argon laser with emissions at 458, 488, and 514 nm and two HeNe lasers with emission wavelengths at 543 and 633 nm. Or images were collected using a Zeiss LSM 780 confocal microscope, mounted on an inverted Axio Observer.Z1 with an argon laser with emissions at 458, 488, and 514 nm, a diode laser with emissions at 405 and 440 nm, a diode-pumped solid-state laser with emission at 561 nm, and a HeNe laser with emission at 633 nm. Image analysis was performed using Imaris suite (Bitplane, South Windsor, CT, USA).

Plaque assay

Lung tissues were homogenized in PBS, supplemented with 1000 U/ml penicillin and 1000 μg/ml streptomycin using MagNA Lyser Green Beads and MagNA Lyser Instrument at 6000 rpm for 1 min (Roche Applied Science, Indianapolis, IN, USA). The amount of infectious influenza virus in lung tissue was measured as described previously [27]. Madin-Darby kidney cells (2×10⁵) were seeded into six-well tissue-culture plates (Corning, Corning, NY, USA) and grown in DMEM (Life Technologies, Grand Island, NY, USA), supplemented with 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. After 24 h of culture at 37°C in a 5% CO₂/96% humidified air atmosphere, the confluent monolayers were washed with HBSS. Lung homogenates were serially diluted in HBSS. Duplicate, 0.75-ml aliquots of serial tenfold dilutions were added to each well. After 1 h of adsorption at 37°C, the wells were washed with HBSS, and 3 ml overlay medium was added to each well. The overlay medium consisted of DMEM with 5% FBS, 1% of nonessential amino acids, 1 mM L-glutamine, and 15 μg/ml trypsin (Worthington Biochemical, Lakewood, NJ, USA) and 1% Bacto agar (BD PharMingen). The plates were placed in a 5% CO₂ atmosphere at 37°C for 72 h. After the incubation period, 2 ml 4% PFA/PBS containing 0.4% w/v crystal violet was added/well. After 2 h at 24°C, the soft agar overlay was decanted gently, and clearly visible plaques on a blue-purple background were enumerated.

BrdU analysis

Mice received 1 mg BrdU by i.p. injection on the days indicated. Two hours later, lymphocytes were harvested from the tissues and processed, as described above. OTI cells were stained with antibodies to CD8 and CD45.1 at 4°C for 20 min. BrdU incorporation was measured using a BrdU flow kit (BD PharMingen), according to the manufacturer's instructions.

FTY720 treatment

Immunized mice were treated four times with 1 mg/kg body weight FTY720 or the vehicle 2% cyclodextrin (Sigma-Aldrich, St. Louis, MO, USA).

Histology

Lungs were inflated with 1 ml Bouin's solution (Sigma-Aldrich) at room temperature for 20 min and fixed in 20 ml Bouin's solution at 4°C for 24 h. Fixed lung tissues were embedded in paraffin, and 4-μm sections were stained with H&E using a standard procedure.

Statistics

Unless stated otherwise, experiments were repeated twice using three to five animals/group. Viral titers in lung tissue were analyzed by Mann-Whitney test. Other statistical significance was determined using paired or unpaired two-tailed Student's t-test. The comparisons that were used to generate P values are indicated by horizontal lines in the figures.

RESULTS

Protective immunity generated by pulmonary immunization declines with time

Cell-mediated immunity lasts 4–6 months after pulmonary influenza virus infection [3] and cannot be induced effectively by other routes of inoculation [4, 28]. In this study, two recombinant influenza viruses that express the SIINFEKL peptide in their neuraminidase stalk [23, 24] have been used to analyze the protective CTL response in immune and nonimmune animals. To verify that this combination of viruses could induce cross-reactive immunity, we primed B6 mice with a sublethal dose of WSN-OVA_I (H1N1 serotype) by i.n. inoculation. Other mice were given a larger dose of WSN-OVA_I by i.p. injection to induce virus-specific memory CD8 T cells without cellular immunity. All of the mice were challenged with a lethal dose of X31-OVA by i.n. inoculation. The second infection was given at IN-1, IN-7, or IP-1. Other mice were infected with X31-OVA without any prior immunization (i.e., UP). Viral titers in the lungs were measured by plaque assays 3 and 5 dpi (Fig. 1).

Figure 1. Cellular immunity requires pulmonary immunization and declines with time.

B6 mice were primed with WSN-OVA_I by pulmonary inoculation (IN-1 and IN-7) or i.p. injection (IP-1). Viral titers in the lungs were measured by plaque assay, 3 dpi (A) and 5 dpi (B; n=5/group; *P<0.05; **P<0.01).

The viral titers in the lungs of the IN-1 mice were ∼20-fold lower than the UP controls 3 dpi (Fig. 1A), and the virus was cleared completely by 5 dpi (Fig. 1B). In contrast, the viral titers in the lungs of the IN-7 and IP-1 mice were similar to UP mice 3 dpi (Fig. 1A), and small quantities of virus persisted until 5 dpi (Fig. 1A). These data confirmed that pulmonary immunization with WSN-OVA_I induced heterosubtypic immunity against reinfection with X31-OVA.

As the goal of the study was to image virus-specific memory CD8 T cells in the lungs during a protective CTL response, we used naive OTI cells to generate traceable populations of antigen-specific CTLs. Uninfected B6 mice received naive OTI cells (10³/mouse) 48 h before immunization with WSN-OVA_I. Separate groups of primed and UP mice were challenged with X31-OVA and were weighed daily to measure protective immunity (Fig. 2A).

Figure 2. Virus-specific CD8 T cells and antibodies participate in optimal heterosubtypic immunity.

(A) B6 mice received 10³ naive OTI cells, 48 h before i.n. or i.p. infection with WSN-OVA_I. Either 1 (IN-1 and IP-1) or 7 (IN-7) months after priming, the mice were reinfected with X31-OVA, and body weights were recorded daily (mean±sd; n=16–45/group). Data are pooled from five independent experiments. UP mice that lost 25% body weight were given 300 μl saline daily by s.c. injections. (B) μMT and B6 mice were primed with WSN-OVA_I by pulmonary inoculation (IN-1) or i.p. injection (IP-1). Fifty percent of the UP μMT mice died within 5 dpi. (C) Body weights were recorded daily (mean±sd; n=3–5/group).

The UP mice lost >20% body weight by 7 dpi and were given daily, s.c. saline injections until normal body weight was restored. The IN-1 mice experienced very mild weight loss (<5%) and began to recover beyond 4 dpi which indicated the presence of robust, heterosubtypic immunity. In contrast, the IN-7 and IP-1 mice lost >10% of body weight and showed no signs of recovery before 6 dpi which indicated reduced immunity compared with the IN-1 group.

As residual immunity was present in the IN-7 and IP-1 mice, we used μMT mice, which lack mature B cells [29], to investigate whether small quantities of cross-reactive antibodies were responsible for the remaining protection (Fig. 2B and C). Pulmonary infection with WSN-OVA_I was lethal to all of the μMT mice, whereas all of the animals survived i.p. inoculation. Pulmonary infection with X31-OVA was also lethal to 50% of the unimmunized (UP) μMT mice within 5 dpi (Fig. 2B), showing that virus-specific antibodies promote survival during primary viral infection. In contrast, all of the μMT mice that had been immunized with WSN-OVA_I by i.p. injection survived secondary challenge with X31-OVA. These primed μMT mice lost more body weight than IP-1-primed B6 mice, indicating that virus-specific memory CD8 T cells provide weaker immunity in the absence of antibodies (Fig. 2C). Together, these studies show that pulmonary immunization induces cross-reactive CTLs and antibodies that participate in optimal heterosubtypic immunity.

Pulmonary but not systemic infection generates T_RM cells which disappear gradually with time

Although the transferred OTI were not required to induce heterosubtypic immunity between the WSN-OVA_I and X31-OVA viruses (Supplemental Fig. 1A), MHCI tetramer analysis showed slightly reduced numbers of endogenous, NP-specific CTL in the lungs of the reinfected mice when the OTI cells were used (Supplemental Fig. 1B).

We found similar numbers of OTI cells in the spleens of the IN-1 and IP-1 mice before reinfection with X31-OVA, whereas slightly lower numbers were found in the IN-7 group (Supplemental Fig. 2B and D). The OTI cells in the lungs were counted (Fig. 3A) and divided into three subsets using α_eβ₇ integrin (CD103) expression as a marker for T_RM cells [30], with CD62L expression to distinguish T_EM (CD103−CD62L−) from T_CM (CD103−CD62L+) cells (Fig. 3B). The percentages of OTI cells that expressed CD62L varied between the different groups (Fig. 3C); however, the cell counts showed very similar numbers of T_CM cells in all of the animals (Fig. 3D). Additional OTI cells that lacked CD62L expression were found in the lungs of the IN-1 mice, which included T_EM and T_RM cells, whereas only T_EM were found in the IP-1 mice (Fig. 3B). Each subset of OTI cells in the lungs was analyzed for CCR7, CD69, PD-1, and CD44 expression to confirm that they expressed characteristic markers of T_RM, T_EM, and T_CM cells, respectively (Fig. 3E). This analysis confirmed that each method of immunization produced substantial populations of circulating T_CM cells and some T_EM cells, whereas pulmonary immunization (IN-1) induced additional populations of T_RM cells which expressed PD-1 in the lungs (Fig. 3E). Very few T_RM cells remained in the lungs at IN-7, and they were completely absent from the IP-1-primed mice (Fig. 3F and G).

Figure 3. Pulmonary inoculation is required to induce PD-1 and CD103 expression on T_RM OTI cells in the lungs.

B6 mice received 10³ naive OTI cells 48 h before i.n. or i.p. infection with WSN-OVA_I. Either 1 (IN-1 and IP-1) or 7 (IN-7) months after priming, the lungs were harvested for quantitative and phenotypic analysis of the OTI cells. (A) Total numbers of OTI cells in the lungs. (B) CD62L and CD103 expression was used to gate subsets of T_RM, T_EM, and T_CM OTI cells in the lungs. (C) Overlaid histograms show CD62L expression on OTI in the lungs (mean±sd; n=3–6 animals). (D) Total numbers of CD62L+ and CD62L− OTI cells in the lungs. (E) Subsets of OTI cells in the lungs of IN-1 mice were analyzed for CCR7, CD69, PD-1, and CD44 expression. Shaded histograms are naive CD8 T cells in the spleen. (F) Percentages of OTI cells in the lungs of IN-1, IP-1, and IN-7 mice expressing PD-1, CD103, or both. Labels show percentages of cells in each quadrant (mean±sd; n=4–9 animals). (G) Total numbers of OTI cells expressing PD-1, CD103, or both (DP). Data are pooled from three independent experiments (*P<0.05; **P<0.01; ***P<0.001).

As PD-1 is often expressed on CD8 T cells during chronic antigen stimulation [31], these data support the idea that prolonged antigen presentation plays a role in the maintenance of T_RM cells in the lungs, as indicated by our previous studies [11, 13]. Small numbers of OTI cells expressed CD103 without PD-1, which may reflect extended exposure to the lung environment. Although PD-1 was expressed on small numbers of OTI cells in the lungs of the IP-1 and IN-7 mice, they did not express CD103. Similar PD-1⁺ OTI cells were also found in the lymphoid tissues, which suggested that some residual antigen may be present outside of the lungs in these animals (not shown). These data show that there were many T_EM cells in the lungs of the IP-1 and IN-7 mice, whereas T_RM cells were absent. Thus, the pulmonary, but not the systemic, route of inoculation was necessary to establish T_RM cells in the lungs, but the size of the T_RM population shrank with time and became undetectable within ∼7 months.

T_RM cells are embedded in the walls of the large airways after pulmonary infection

Next, we used confocal microscopy to examine the distribution of OVA-specific memory CD8 T cells in the lungs after priming. The lungs, MLNs, and spleens were harvested on different days after WSN-OVA_I infection, as shown in Fig. 4. Fixed lung tissue was stained with anti-CD45.1 to identify the transferred OTI cells, as well as anti-CD31 to identify blood vessels and anti-EpCAM to identify epithelial cells in the airways. Sections of spleen and MLN were stained with anti-B220 to identify the B cell zones and anti-CD31 for blood vessels (Supplemental Figs. 2 and 3).

Figure 4. OVA-specific T_RM cells line the walls of the large airways after pulmonary inoculation.

B6 mice received 10³ naive OTI cells, 48 h before i.n. or i.p. infection with WSN-OVA_I. The lungs were harvested 1 or 7 months after priming to visualize OTI cells in situ. (A) Fixed lung tissue was analyzed with anti-CD45.1 (green) to identify transferred OTI cells, anti-EpCAM (red) for epithelial cells, and anti-CD31 (blue) for blood vessels (BV). Z-Stack images were acquired at normal magnification, ×20. Representative images show sections across large airways (AW; upper) and areas of lung parenchyma (P; lower) from groups of four to 10 animals. (B) OTI cells (green) in the lungs of IN-1 mice were analyzed for CD103 expression (blue). Cross-sections of large airways from groups of three animals. b, Imaris software was used to identify overlapping blue and green staining shown in pink.

The lungs of the IN-1 mice contained substantial numbers of OTI cells, 30 dpi (Fig. 4A), which included some CD103⁺ cells that were located along the walls of large airways (Fig. 4B, b). Other OTI cells were found in the tissue surrounding the alveoli and mostly lacked CD103 expression (i.e., green cells). Very few OTI cells were found in the airways of the other primed mice (Fig. 4A). These data confirm that the pulmonary route of infection was required to populate the airways with T_RM cells, whereas systemic inoculation produced predominantly circulating T_EM cells. Most of the T_RM cells disappeared from the airways of the locally primed mice within 7 months after infection.

The protective CTL response is focused around the large airways after heterosubtypic challenge

Next, confocal microscopy was used to follow temporal changes in the size and anatomical distribution of the T_eff response in the lungs during heterosubtypic challenge (Fig. 5). Tissues from groups of three to six mice were analyzed on different days after reinfection. Only small changes in the numbers (Fig. 6A) and distribution of OTI cells could be detected in the lungs before 3 dpi (Fig. 5A, upper). Additional OTI cells began to accumulate around selected large airways (Fig. 5A, lower) and blood vessels (Fig. 5B) by 6 dpi, whereas other airways remained unchanged. Some OTI cells were also found near the arterial blood supply, which nourishes the walls of the airways [32] and may be an important access point for new cells to enter the lungs from the circulation (Supplemental Fig. 4). In addition, large numbers of OTI cells were found in the lung parenchyma of the IN-7 and IP-1 groups, 6 dpi, which indicated widespread infection (Fig. 5C).

Figure 5. The protective CTL response is focused around the large airways early after reinfection.

B6 mice received 10³ naive OTI cells, 48 h before i.n. or i.p. infection with WSN-OVA_I. One (IN-1 and IP-1) or 7 (IN-7) months later, the mice were challenged with X31-OVA. Fragments of fixed lung tissue were stained 3 and 6 dpi using antibodies to EpCAM (red), CD45.1 (green), and CD31 (blue). Z-Stack images were taken at normal magnification, ×20. Representative fields from groups of three to six animals are shown. (A) Cross-sections of large airways. (B) Cross-sections of large blood vessels. (C) Lung parenchyma.

Figure 6. Large accumulations of OTI cells appear in the lungs ∼5 dpi after reinfection.

B6 mice received 10³ naive OTI cells, 48 h before i.n. or i.p. infection with WSN-OVA_I. One (IN-1 and IP-1) or 7 (IN-7) months later, the mice were challenged with X31-OVA. (A) Total numbers of OTI cells in the lungs. (B) Frequencies of OTI cells with a T_RM phenotype. (C) Total numbers of T_RM cells in the lungs. Graphs show means ± sd from three to 12 animals/group. Data were pooled from four independent experiments. Symbols indicate significance among different groups (*P<0.05, **P<0.01, ***P<0.001 IN-1 vs. IN-7; §P<0.05, §§P<0.01 IN-1 vs. IP-1; †P<0.05, ††P<0.01, †††P<0.001 IP-1 vs. IN-7; ‡P<0.05 IN-1 Day 0 vs. IN-1 Day 3).

Parallel flow cytometry data confirmed that the total numbers of OTI cells inside the lungs remained stable until 5 dpi (Fig. 6A), even though there were lower viral titers in the lungs of IN-1 mice than in the other primed animals (Fig. 1). Consequently, the early accumulations of OTI cells may have included cells that redistributed inside the lungs or new arrivals that replaced some dying OTI cells. The numbers of T_RM cells in the lungs of IN-1 mice transiently declined 3 dpi, but larger numbers of CD103⁺ CTLs began to reappear in the lungs by 5 dpi, suggesting that a new generation of T_RM cells had begun to develop (Fig. 6B and C). Together, these data showed that lung-resident CTLs were participating in early viral clearance from the locally immunized animals (IN-1) and that the protective CTL response was focused around large airways. In contrast, widely dispersed OTI cells in the lungs of the systemically immunized IP-1 mice and IN-7 group (Fig. 5C) indicated that infection reached alveoli, where there was more immune pathology (Supplemental Fig. 5A) and edema in the lungs of the IP-1 mice compared with the IN-1 mice (Supplemental Fig. 5B).

Reactivated T_CM are not necessary for heterosubtypic protection

The imaging studies emphasized a central role of T_RM cells during early viral clearance, but they did not address the critical question of whether additional memory CD8 T cells also made a delayed contribution to the protective CTL response after proliferating in the local lymphoid tissues [20]. To identify the location of proliferating OTI cells, mice were primed with WSN-OVA_I and challenged with X31-OVA as before. Each mouse received a single dose of BrdU by i.p. injection, 3 or 4 dpi with X31-OVA, and CD45.1⁺ OTI cells were analyzed 2 h later (Fig. 7). All of the mice had substantial numbers of BrdU⁺ OTI cells in the MLN, 3 dpi, which corresponded with the appearance of small clusters of OTI cells in the paracortex area of the MLN (Supplemental Fig. 3). Relatively few OTI cells incorporated BrdU in the lungs by 3 dpi, but the numbers increased substantially by 4 dpi (Fig. 7C and D). These results confirmed that some OVA-specific memory cells were reactivated in the MLN and proliferated before migrating to the lungs. These proliferating CTLs began to accumulate after reduced viral titers were recorded in the lungs of the locally immunized mice (Fig. 1).

Figure 7. Nascent T_eff cells proliferate in the MLNs and lungs after reinfection.

B6 mice received 10³ naive OTI cells, 48 h before i.n. or i.p. infection with WSN-OVA_I. One (IN-1 and IP-1) or 7 (IN-7) months later, the mice were challenged with X31-OVA. BrdU was given by i.p. injection, 3 or 4 dpi, and OTI cells were analyzed 2 h later. The data are means ± sd from four animals/group (*P<0.05; **P<0.01). A) Histograms show gated OTI cells with percentages BrdU+ cells in the MLN. B) Total numbers BrdU+ OTI cells in the MLN. C) Histograms show gated OTI cells with percentages BrdU+ cells in the lungs. D) Total numbers BrdU+ OTI cells in the lungs.

The delayed kinetics of BrdU incorporation suggested that newly generated T_eff cells did not play a major role in heterosubtypic protection. To test this hypothesis more rigorously, we treated additional animals from the IN-1 and IP-1 groups with FTY720, which is an analog of S1P that induces the irreversible internalization of the S1P1 receptor from the cell surface and prevents circulating T cells (both CD8 and CD4) from leaving the LNs [33]. Flow analysis confirmed that the FTY720 treatments caused peripheral lymphopenia in locally (IN-1) and systemically immunized (IP-1) mice (Fig. 8A). Increased numbers of NP tetramer⁺ CD8 T cells were detected in the MLNs of the FTY720-treated mice compared with the vehicle-treated controls (Fig. 8B). In contrast, there were no significant differences in the numbers of NP-specific CTLs in the lungs of the two groups of mice (Fig. 8C). The body weights of the mice were not altered by the FTY720 treatments (Fig. 8D), and thus, this demonstrates that circulating T cells, including nascent T_eff cells that were recruited from the MLN to the lungs, 4 dpi [20], were not required for protection against heterosubtypic challenge.

Figure 8. Nascent T_eff cells do not participate in protection.

B6mice were primed with WSN-OVA_I and challenged 1 (IN-1 and IP-1) or 7 (IN-7) months later with X31-OVA. Mice were treated four times (48-h intervals) with FTY720 or vehicle control, beginning 24 h before X31-OVA infection. (A) Percentages of CD8 T cells in the blood, 8 dpi. (B) Numbers of NP-specific CD8 T cells in the MLN, 5 dpi (IN-1) or 8 dpi (IP-1). (C) Numbers of NP-specific CD8 T cells in the lungs, 5 dpi (IN-1) or 8 dpi (IP-1). (D) Mean body weights (±sd) from four to five mice/group (*P<0.05; **P<0.01; ***P<0.001).

Together, our data support the idea that local memory CD8 T cells are necessary and sufficient for T cell-mediated heterosubtypic immunity in the lungs, whereas circulating memory cells are unable to accelerate viral clearance.

DISCUSSION

Long-lived T_RM cells have been detected in the lungs after a variety of different respiratory virus infections [9, 10, 34]; however, their role in heterosubtypic immunity has not been analyzed directly. In this study, a heterosubtypic reinfection model is used to follow the CTL response in situ. We find that the rate of viral clearance after reinfection is tightly linked to the numbers of virus-specific T_RM cells in the large airways of the primed mice at the time of reinoculation. Surprisingly, nascent T_eff cells played very little role in immunity, as reduced viral titers were found in the lungs of the locally immunized mice before additional T_eff cells began to proliferate in the local tissues. As the infection progressed, most T_eff cells remained localized around the large airways of the locally immunized mice, whereas many T_eff cells dispersed to remote parts of the lungs in the other animals.

A herpes simplex virus model was used previously to show that T_RM cells control local skin infections more efficiently than T_EM cells [35, 36]. The exact mechanism by which T_RM cells promote viral clearance from the lungs is not known; however, some studies suggest that T_RM cells may promote viral control by secreting inflammatory cytokines and chemokines into the airways, which recruit additional T_EM and other immune cells to the site of viral replication [37]. Similarly, during hepatitis B virus infection, activated CD8 T cells produce CXCL8 in the liver, which drives neutrophil infiltration into the local tissues and causes local pathology [38]. Virus-specific T_RM cells that remained in the brain after murine CMV infection also produced IFN-γ which caused long-term microglia activation [39].

Substantial numbers of CD69+/CD103+ T_RM cells can be recovered from the lungs in BALF, whereas others cannot be released without mechanical disruption or collagenase digestion of the lung tissues [9, 13]. Earlier studies have shown that the cells in the BALF exhibit weak lytic activity in ex vivo assays [9], which was attributed to the loss of CD11a expression soon after their arrival in the airways [18] as a result of high concentrations of proteases in the mucus [40]. The T_RM cells in the lung parenchyma express CD11a at high levels which suggests that they are capable of some lytic activity [18]. Our images show substantial numbers of CD103+ T_RM cells embedded in the walls of the airways, 30 dpi, which may be part of the T_RM population that maintains CD11a expression. Alternatively, interactions between CD103⁺ CTL and E-cadherin were sufficient to promote the polarization and exocytosis of lytic granules in a tumor model [41]. Similarly, the CD103+ CTL may be capable of killing epithelial cells that express E-cadherin in the lungs without the assistance of CD11a expression. In either case, the T_RM cells that were embedded in the walls of the airways were in an excellent position to kill newly infected epithelial cells after reinfection, and thus eliminate an important source of new virus particles.

The signals that are required for the long-term maintenance of pathogen-specific T_RM cells in peripheral tissues are poorly defined. After pulmonary infection, the CD103⁺ T_RM cells in the lungs expressed PD-1 at low levels, which is a marker of chronic antigen stimulation in other models [31]. In contrast, quantitative PCR analysis revealed little evidence of sustained inflammation (data not shown). Similarly other investigators found that recent exposure to peptide antigens was required for T_RM cells to maintain CD103 expression in the brain [17]. In contrast, the prolonged presence of antigen was not necessary for stable CD103 expression on T_RM cells in the small intestine, salivary gland, or skin [15, 16, 34, 42]. Whereas one study found that nonspecific inflammation was enough to generate stable populations of T_RM cells in the skin [35], others found that the highest concentrations of T_RM cells were located near the site of a recent infection [17, 43]. Locally administered chemokines were used recently to augment vaccine-induced immunity in the female reproductive tract; however, pathogen-specific CTLs were not analyzed for markers of T_RM cells in the local tissues [44].

In summary, our combined studies show that the cumulative effects of virus-specific antibodies and T_RM cells are required for optimal heterosubtypic immunity, whereas circulating memory CD8 T cells play a very little role. The efficacy of the immunity correlates with the numbers of T_RM cells in the lungs before reinfection rather than the speed or magnitude of the response by the new generation of T_eff cells. Importantly, direct pulmonary inoculation is required to establish T_RM cells in the airways, which disappear as cellular immunity declines. Early after reinfection, the numbers of T_RM cells decrease and are replaced by a new generation of CTLs that express CD103. These data indicate that vaccines that are designed to establish optimal heterosubtypic immunity in the lungs must induce sufficient numbers of T_RM cells in the airways to prevent severe damage outside of the bronchial tree.

AUTHORSHIP

T.W. performed most experiments and analyzed the data. Y.H. and K.R.B. assisted with tissue preparation. Y-T.L. and K.K. assisted with microscopy. A.B. assisted with the FTY720 experiments. T.W. and L.S.C. wrote the manuscript, with revisions by K.K.

DISCLOSURES

The authors declare no competing financial interests.