Frequency, Specificity, and Sites of Expansion of CD8+ T Cells during Primary Pulmonary Influenza Virus Infection
http://www.100md.com
免疫学杂志 2005年第9期
Abstract
We have used intracellular cytokine staining and MHC class I tetramer binding in conjunction with granzyme B protease expression and in vivo BrdU uptake to characterize the primary murine CD8+ T cell response to pulmonary influenza virus infection. We have observed that the majority (>90%) of the CD8+ T cell response to the A/Japan/305/57 virus in the lung at the peak of the response (days 9–11) is directed to four epitopes (three dominant and one subdominant). Using induction of granzyme B as a surrogate to identify specific activated CD8+ T cells, we found that an unexpectedly large fraction (70%) of lung-infiltrating CD8+ T cells expressed granzyme B on day 6 of infection when estimates by MHC tetramer/intracellular cytokine staining yielded substantially lower frequencies (30%). In addition, by using intranasal administration of BrdU during infection, we obtained evidence for proliferative expansion of activated CD8+ T cells in the infected lung early (days 5–7) in the primary response. These results suggest that the frequency and number of specific CTL present in the lung early in infection may be underestimated by standard detection methods, and primary CD8+ T cell expansion may occur in both secondary lymphoid organs and the infected lung.
Introduction
A significant amount of our current understanding of specific CD8+ T cell responses to viral epitopes has been garnered through the study of experimental influenza virus infection of mice. For example, extensive investigations into the basis of immunodominance of CD8+ T cell responses to multiple epitopes have been performed in both BALB/c (H-2d) and C57BL/6 (H-2b) mice after i.p. inoculation (1, 2). Furthermore, because influenza virus is a respiratory tract pathogen, much of our current knowledge about the specific CD8+ T cell response to influenza virus has come from analyzing the specificity of the pulmonary CD8+ T cell response to intranasal (i.n.)3 challenge (3, 4, 5). These important studies have detailed the immunodominance hierarchy to several CD8+ T cell epitopes from the HKx31 (H3N2) and PR/8 (H1N1) influenza viruses in H-2b mice. The most dominant epitopes from these viruses are polymerase A224–233 and nucleoprotein (NP)366–374, which, along with other minor epitopes, are targets for the 40% of pulmonary CD8+ T cells that are identifiable as virus-specific during the primary immune response (3, 4, 5).
By contrast, other respiratory tract viruses have been shown to induce an even more potent specific CD8+ T cell response in mice. For example, i.n. Sendai virus infection elicits a highly restricted CD8+ T cell population in which 70% of pulmonary CD8+ T cells are directed against a single immunodominant epitope (6). Similarly, 50–60% of CD8+ T cells responding to a primary i.n. respiratory syncytial virus infection are directed against two epitopes, the vast majority of which are specific to a single epitope as well (7). Therefore, the frequency of CD8+ T cells responding to respiratory viruses probably parallels the high frequencies of pathogen-specific CD8+ T cells that are prevalent during the response to a potent, systemic viral infection such as acute lymphocytic choriomeningitis virus challenge (8, 9). To examine whether a similarly high frequency of virus-specific CD8+ T cells may be directed against influenza virus, we analyzed the CD8+ T cell response to primary i.n. infection of BALB/c mice with a mouse-adapted A/Japan/305/57 (Japan/57) virus.
CD8+ T cells responding to Japan/57 virus in BALB/c mice recognize four known epitopes restricted by H-2Kd (Table I). Three of these viral epitopes, hemagglutinin (HA)204–212, HA210–219 (10, 11), and HA529–537 (12), are processed from the HA protein, whereas a fourth, NP147–155 (13), is derived from the viral NP. Previous studies have demonstrated that bulk, polyclonal cultures prepared from the splenocytes of mice that were immunized i.p. with Japan/57 possessed high cytolytic activity against the HA204–212, HA529–537, and NP147–155 epitopes, whereas the cytolytic activity against HA210–219 was much weaker (14, 15). However, the frequency of virus-specific CD8+ T cells responding in vivo to i.n. Japan/57 virus infection had not been analyzed.
Table I. Amino acid sequences of CD8+ T cell epitopes from A/Japan/305/57 influenza virus
In this report we have characterized the magnitude and immunodominance hierarchy of the CD8+ T cell response against the four previously described epitopes during pulmonary Japan/57 influenza virus infection. We demonstrate that the frequency of CD8+ T cells that are specific to these influenza virus epitopes is indeed high during the peak of the pulmonary response, as determined by intracellular cytokine staining (ICCS) and MHC class I tetramer labeling. Activated CD8+ T cells directed to these four epitopes represent 90% of the total responding CD8+ T cells in the lungs on day 9 of infection by ICCS using cytokine synthesis by lung-infiltrating CD8+ T cells to influenza virus-infected cells as a measure of total virus-specific CD8+ T cells. The immunodominance hierarchy of the response to these epitopes in vivo followed the predicted pattern based on previous in vitro studies. In addition, we explored the use of granzyme B protease expression by activated CD8+ T cells as a surrogate marker to identify specific CD8+ T cells. We obtained evidence that granzyme B expression may identify virus-specific CD8+ T cells that have migrated to the lungs early after infection, i.e., days 5–7, but are not efficiently detected by ICCS or MHC tetramer binding. Finally, we evaluated the kinetics of CD8+ T cell proliferation in vivo during infection by systemic (i.p.) or local (i.n.) administration of the thymidine analog BrdU. Using this approach, we obtained evidence suggesting the idea that activated virus-specific CD8+ T cells may undergo proliferative expansion in both secondary lymphoid organs (draining lymph nodes) and infected lungs early after infection. The significance of these results for the induction of CD8+ T cell responses to respiratory virus infection is discussed.
Materials and Methods
Mice and experimental infection
Female BALB/c mice (H-2d) were purchased from Taconic Farms and maintained under specific pathogen-free conditions. Mice that were 8–11 wk of age were lightly anesthetized by Metofane (Janssen) inhalation and sublethally infected by i.n. inoculation (in 50 μl) with 3953 half-maximal tissue culture infectious doses (TCID50) of a mouse-adapted A/Japan/305/57 (H2N2) influenza virus.
Preparation of tissue lymphocytes
At multiple days after influenza virus infection, mice were killed by cervical dislocation. Mouse tracheas were then surgically exposed to perform bronchoalveolar lavages (BAL). Briefly, the tracheas were cannulated with a syringe, and five separate 1-ml aliquots of IMEM were used to rinse the lower respiratory tract to collect inflammatory cells from the air spaces. Lavaged lungs were then perfused via the right ventricle of the heart with 5–10 ml of PBS containing 10 U/ml heparin (Sigma-Aldrich) to remove blood lymphocytes from the lung vasculature. The lungs were minced and passed through a steel screen. Tissue debris was removed by centrifugation at 300 x g. Cells were counted and resuspended at appropriate concentrations for each particular experiment.
Peptides
Influenza virus peptides HA204–212, HA210–219, HA529–537, and NP147–155 (Table I) were synthesized by University of Virginia Biomolecular Research Facility. Purity was confirmed by HPLC analysis.
Ab labeling and flow cytometry
For intracellular cytokine staining experiments, harvested cells were incubated for 6 h in 96-well, round-bottom plates in IMEM supplemented with 10% FBS, 10 U/ml penicillin G, 10 μg/ml streptomycin sulfate, 100 U/ml human IL-2, 2 mM L-glutamine, 0.05% 2-ME, and 1 μg/ml brefeldin A (Sigma-Aldrich). This relatively high concentration of human IL-2 was used to maximize the synthesis and accumulation of cytokines achieved in the ICCS. This high IL-2 concentration did not affect the frequency of cytokine-secreting cells (data not shown). Stimulatory conditions included 1 μM of a given synthetic peptide and Japan/57 influenza virus-infected P815 (H-2d) mastocytoma cells. After incubation, cells were washed and incubated for 10 min in FcBlock (BD Biosciences). The allophycocyanin-conjugated anti-CD8 mAb (clone 53-6.7; BD Biosciences) was used to label cells for 30 min at 4°C. After washing with FACS buffer (PBS/1% FBS), cells were fixed, and erythrocytes were simultaneously lysed using FACS Lysing Solution (BD Biosciences). The cells were labeled at room temperature with anti-IFN- (clone XMG1.2) or rat IgG1 (clone R3-34; BD Biosciences) in permeabilization buffer (FACS buffer containing 0.5% saponin). The percentage of background cytokine staining from the appropriate population of unstimulated cells was subtracted from the percentage of cytokine-positive cells under each stimulation condition before reporting. For intracellular granzyme B staining, cell suspensions were prepared from mediastinal lymph nodes (MLN) and lungs at multiple days after influenza virus infection. Cells were first surface-labeled with anti-CD8 (clone 53-5.8; BD Biosciences) for 30 min, then fixed and permeabilized as described above. Permeabilized cells were labeled with anti-granzyme B (clone GB12; Caltag Laboratories) or mouse IgG1 (BD Biosciences) for 45 min at room temperature. Flow cytometry data were acquired for each of the experiments using a BD FACSCalibur (BD Immunocytometry Systems) and were analyzed using CellQuest software (BD Biosciences).
Cell surface marker expression
Lung cell suspensions (see Preparation of tissue lymphocytes above) were prepared from the lungs of four or five mice on day 6 after infection. Lung-infiltrating CD8+ T cells were purified from pooled lung cell suspensions by CD8-conjugated magnetic bead positive selection (Miltenyi Biotec) according to the manufacturer’s instructions. Surface CD107a mobilization was examined using a modification of the protocol described by Bells et al. (16). Briefly lung CD8+ T cells (either in total lung cell homogenates or after purification) were cultured for 6 h at 37°C with P815 cells, which were either untreated or treated with a mixture of the four viral peptide epitopes in the presence of monensin (1 μg/ml) and PE-conjugated anti-CD107 Ab (ID4B; Santa Cruz Biotechnology) at a final concentration of 5 μg/ml throughout the assay period. Cells were then fixed and stained for CD69 and CD8 expression (as above).
MHC class I tetramers
MHC class I tetramers were produced according to published techniques (17). DNA plasmids encoding human 2-microglobulin and the extracellular domain of the murine H-2Kd H chain were provided by J. Altman (Emory University, Atlanta, GA) and E. Pamer (Sloan-Kettering Institute, New York, NY), respectively. MHC class I tetramers were prepared using each of the following influenza virus peptides: HA204–212, HA210–219, HA529–537, and NP147–155. Tetramer labeling of harvested cells was performed for 45 min at 4°C.
Virus titers
Mice were i.n. infected with influenza Japan/57 virus and were killed by cervical dislocation at multiple time points after infection. Excised mouse lungs were weighed and homogenized. Multiple dilutions of clarified homogenates were plated on Mardin-Darby canine kidney cells in 96-well, flat-bottom plates. After 3 days in a 37°C humidified 7% CO2 incubator, 50 μl of supernatant was harvested from each well and mixed with 50 μl of fresh 1% chicken erythrocyte solution in round-bottom, 96-well plates. After 1 h at room temperature, hemagglutination results were determined and used to calculate TCID50 titers.
BrdU labeling
On multiple days after pulmonary influenza virus infection, mice were lightly anesthetized by Metofane (Janssen) inhalation and then were administered 50 μl of 16 mg/ml BrdU (0.8 mg total/mouse) by the i.n. route or 250 μl of 8 mg/ml BrdU (2 mg total/mouse) by i.p. injection. Each mouse was killed 24 h after BrdU treatment. Tissue cells were prepared for cell surface Ab labeling as described above. Anti-CD8-labeled cells were fixed and permeabilized before BrdU staining was performed using the BD FastImmune BrdU kit according to the manufacturer’s instructions (BD Biosciences).
Results
Frequency and magnitude of the pulmonary CD8+ T cell response to defined epitopes after influenza virus infection
We have analyzed the frequency of pulmonary CD8+ T cells directed to four previously identified MHC class I-restricted epitopes of the Japan/57 (H2N2) virus in H-2d haplotype BALB/c mice (Table I) (10, 11, 12, 13) in response to sublethal i.n. Japan/57 infection by ICCS for IFN- and by MHC tetramer binding. Three of these H-2Kd-restricted epitopes, HA204–212, HA529–537, and NP147–155, were characterized in vitro as immunodominant, whereas the fourth epitope, HA210–219, also Kd-restricted, was classified as subdominant based on the magnitude of the in vitro CTL response (14, 15).
Analysis of the kinetics of accumulation of IFN--secreting CD8+ T cells in BAL fluid revealed that Japan/57-specific, IFN- secreting CD8+ T cells were first detected on day 5 postinfection (Fig. 1A). The frequency of IFN--secreting T cells to each of the defined epitopes was always <5% at this time, with <10% of BAL CD8+ T cells in aggregate identified as virus-specific by this criterion. As previously reported in other influenza virus infection models (3, 4, 5), over the next 2–5 days the frequency (percentage) of epitope-specific, IFN--producing CD8+ T cells increased significantly and reached a maximum level by days 9–10 of infection, with subsequent contraction of the response during the following week (Fig. 1A). Analysis of the IFN- response of CD8+ T cells infiltrating the lung parenchyma (Fig. 1B) revealed a comparable kinetics of expansion and contraction, although the magnitude of the response, i.e., the frequency of epitope-specific CD8+ T cells in the lung parenchyma, was slightly reduced compared with that in BAL. A companion kinetic analysis of the CD8+ T cell response in BAL by MHC class I tetramer staining demonstrated, as expected, a close correspondence in both the kinetics of accumulation and the frequency of specific T cells between MHC tetramer straining (Fig. 1C) and the ICCS assay (Fig. 1A). We did, however, note a modest delay in the contraction phase of the BAL CD8+ T cell response, as defined by the frequency of epitope-specific MHC class I tetramer staining.
FIGURE 1. Kinetics of influenza virus-specific pulmonary CD8+ T cells, as determined by intracellular cytokine and tetramer staining analyses. BALB/c mice were infected with 3953 TCID50 of Japan/57 influenza virus. BAL (A) and lung parenchyma (B) cells were harvested at multiple days after infection and analyzed in 6-h intracellular cytokine assays. C, BAL cells were collected at multiple days after infection and labeled with specific tetramer. The percentage of specific IFN- expression or tetramer labeling is shown for CD8+ gated T cells at the indicated number of days postinfection. Data are representative of three independent kinetic analyses with pooled tissue (samples) from three to five mice per time point. Values are the mean ± SEM for three or four replicate samples for each time point.
As revealed above (Fig. 1), the hierarchy of immunodominance for the four epitopes, initially suggested by in vitro studies of immune T cell populations, was confirmed by the analysis of epitope-specific responses during the in vivo CD8+ T cell response to primary infection in the respiratory tract. Thus, CD8+ T cells directed to the NP147–155, HA204–212, and HA529–537 epitopes dominated the response in vivo, with the weaker (subdominant) response to HA210–219 likewise demonstrable in the lungs after primary infection. When the kinetics of the total CD8+ T cell response to each epitope was calculated by either tetramer staining (Fig. 2A) or ICCS (Fig. 2B), the rates of expansion and contraction of the response to the three dominant epitopes followed a predictable pattern (3, 4, 5) based on the frequency of responding T cells.
FIGURE 2. Quantitation of influenza virus-specific CD8+ T cell numbers in the lung airways and parenchyma. A, BAL cells were harvested from mice at the indicated number of days after Japan/57 infection and analyzed by tetramer staining. B, Lung parenchyma cells were harvested at multiple days after infection and analyzed using the intracellular cytokine assay. The numbers of individual Ag-specific CD8+ T cells were calculated using total cell count and IFN-+ (A) or tetramer+ (B) frequency data and are displayed in the left graphs. The total number of virus-specific CD8+ T cells at each time point is inset in the right graphs. Data are representative of three independent analyses with values obtained from pools obtained from three to five donor mice per time point.
Contribution of epitope-specific CD8+ T cells to the overall pulmonary CD8+ T cell response
The findings described above suggested that a robust primary pulmonary CD8+ T cell response is mounted to these four epitopes after infection with Japan/57 virus. Although weak/subdominant responses to sites on several viral gene products from various influenza virus strains have been reported (1, 2, 5), previous analyses from our laboratory (14, 15) (unpublished observations) suggested that the response to the Japan/57 virus in H-2d haplotype mice is focused predominately, if not exclusively, on the four described epitopes. We therefore wished to determine the contribution to the overall pulmonary Japan/57-specific CD8+ T cell response by CD8+ T cells that were directed against these four epitopes in vivo after i.n. infection. To this end, we analyzed the frequency of IFN--secreting CD8+ T cells responding to each of the four peptide epitopes in BAL and lung parenchyma by ICCS at the peak of the primary response. In parallel, we also analyzed the IFN- response of pulmonary CD8+ T cells to Japan/57-infected P815 cells (H-2d) as a source of naturally processed viral Ags. The virus-infected cells would probably process and present all potential H-2d-restricted MHC class I epitopes, including, but not limited to, the four epitopes previously identified and analyzed in this study. We then calculated the sum total of the individual CD8+ T cell IFN- responses to each of the four Japan/57 epitopes and compared this value to stimulation by the virus-infected cells (which was normalized to 100%). Strikingly, we observed a very high correlation (93% in lung, 92% in BAL) of IFN- expression when comparing CD8+ T cell stimulation by virus-infected cells to the cumulative frequency of individual peptide-stimulated CD8+ T cells (Fig. 3). These data demonstrate that the specific CD8+ T cell response against influenza virus Japan/57 in H-2d mice is nearly entirely directed against the four viral epitopes presently studied. In this regard, it should be noted that at the peak of the pulmonary CD8+ T cell response (day 10), between 60 and 70% of the total pulmonary CD8+ T cells secreted IFN- in response to infected P815 cells. We also determined that the total complement of lung CD8+ T cells capable of secreting IFN- after nonspecific stimulation with PMA/ionomycin was 70%. Thus, by these criteria, the vast majority of lung CD8+ T cells were virus specific.
FIGURE 3. The total influenza virus-specific CD8+ T cell response is nearly entirely directed against four Japan/57 epitopes. Cell suspensions were prepared from the BAL and lung parenchyma on day 10 postinfection. Cells were stimulated for 6 h with each of the synthetic peptides and Japan/57-infected P815 cells. The frequency of IFN- expression by CD8+ T cells after stimulation by virus-infected cells was set at 100% for comparison with the cumulative response to each of the individual viral epitopes. The data are representative of three independent analyses using data pooled from four mice for infectious virus and individual peptides. Percentages of the total influenza response (reflected in the sum of individual epitopes) ranged from 89 to 95%.
Lung virus replication and the early CD8+ T cell response
Our previously discussed kinetic analysis of influenza virus-specific CD8+ T cell accumulation in the lungs during primary i.n. infection revealed that specific CD8+ T cells (defined by tetramer staining and ICCS) were first detectable on days 5–6 of infection. Furthermore, the response appeared to be directed primarily, if not exclusively, to the four described epitopes. With this information in hand, we next evaluated the time course of virus replication in the lungs to relate the tempo of virus clearance to the kinetics of virus-specific CD8+ T cell trafficking to the lungs. As Fig. 4 demonstrates, virus titers rapidly peaked (106.7 TCID50/g tissue) by 24 h postinfection and were sustained at or near peak titers through day 5 of infection. Between days 5 and 7, virus titers dropped precipitously (1000-fold), and infectious virus was no longer detectable in the lungs by day 9 of infection. This precipitous fall in pulmonary Japan/57 virus titer over days 5–7 of infection is in agreement with previously published results in the murine model using other influenza virus strains (18, 19, 20, 21) and correlates with the early infiltration of specific CD8+ T cells into the lung, i.e., days 5–6 (Fig. 1) (3). These kinetic observations implicate the early CD8+ T cell response in the infected lungs as critical in controlling virus replication and clearance. However, it is also evident from this kinetic analysis that the major control of virus replication and clearance occurred on days 5–7, which is before the peak of pulmonary CD8+ T cell accumulation in the lungs, i.e., days 9–11 (Fig. 2).
FIGURE 4. Kinetics of Japan/57 virus clearance from infected mouse lungs. Mice were i.n. inoculated on day 0 with 3953 TCID50 of Japan/57 influenza virus (inoculum dose denoted by asterisk). At several days after infection, mouse lungs were harvested, weighed, and homogenized. Clarified homogenates were plated on Mardin-Darby canine kidney cells. The presence of infectious influenza virus was analyzed 3 days later by hemagglutination assay to determine TCID50 titers.
Early accumulation of virus-specific CD8+ T cells within the infected lungs
The trafficking and accumulation of activated virus-specific T cells into the infected lungs are believed to be preceded by the activation and proliferative expansion of naive virus-specific CD8+ T cell precursors in organized lymphoid tissues, most notably the MLN draining the infected lung (22, 23), and with subsequent migration of activated effector CD8+ T cells from the lymph nodes to the infected lungs. However, analyses of the frequencies of epitope-specific CD8+ T cells detected in draining MLN by tetramer staining and/or ICCS in the early inductive phase of the CD8+ T cell response to primary i.n. influenza virus infection (i.e., up to day 4 postinfection) have demonstrated epitope-specific CD8+ T cell frequencies of 2% (4, 24, 25). In the case of the four CD8+ T cell epitopes examined in this study, the frequencies of specific CD8+ T cells enumerated by tetramer staining in the draining lymph nodes on day 4 postinfection have a cumulative total of 3–5% (data not shown).
Although the inability to detect significant proliferative expansion of CD8+ T cell precursors in the draining lymph nodes before T cell accumulation in the infected lungs would most likely reflect the inherent low frequency of naive specific CD8+ T cell precursors (22) and the presumed rapid egress of effector CD8+ T cells out of the lymph nodes after proliferative expansion, several recent findings from our laboratory have modified our view of the early events in CD8+ T cell activation in the draining lymph nodes and lungs. First, we have recently demonstrated that early in the activation of influenza virus-specific CD8+ T cells within draining lymph nodes during pulmonary influenza virus infection, activated proliferating CD8+ T cells transiently lose the ability to bind specific tetramers without any loss of cell surface TCR expression.4 This transient loss of tetramer binding by activated T cells (up to days 5–6 postinfection) is associated with a transient diminished ability of the T cells to secrete IFN- in the ICCS assay, but without any deficit in Ag-driven cell proliferation.4 Importantly, a significant fraction (>60%) of activated CD8+ T cells that have migrated to the infected lung early during the T cell response to infection (days 5–7) retain the diminished capacity to bind specific tetramer and secrete IFN- in the ICCS assay. Full tetramer staining and cytokine secretion are, however, restored in CD8+ T cells infiltrating the lung by day 8 postinfection.4 Second, we have recently demonstrated that 24–48 h after naive CD8+ T cell activation and the onset of T cell proliferation in draining MLN in response to pulmonary influenza virus infection, activated virus-specific CD8+ T cells within the draining lymph nodes express the granule-associated lytic molecule granzyme B, and that granzyme B expression is uniformly retained by activated T cells in the lymph nodes and after T cell migration to the infected lungs (23).
Taken together, these previously discussed observations suggested that the frequency of virus-specific CD8+ T cell present in the lungs early in infection (i.e., days 5–7) may be underestimated by tetramer staining and/or ICCS, and that granzyme B expression may serve as a surrogate marker for the detection and enumeration of activated specific CD8+ T cells at early time points during infection. To examine these possibilities, we studied the kinetics of granzyme B expression by CD8+ T cells within draining MLN and lungs after i.n. influenza virus infection using a flow cytometry-based assay (26, 27). As Fig. 5 demonstrates, granzyme B+CD8+ T cells were identified at a detectable frequency on day 3 postinfection. The percentage of granzyme B+ T cells in the MLN increased 10-fold (3%) by day 4, doubled in frequency over the ensuing 24 h (to 6%), and was maintained at this level up to day 9 after infection.
FIGURE 5. High frequencies of granzyme B-expressing CD8+ T cells are present in the draining lymph nodes and lungs during influenza virus infection. The lungs and MLN were harvested from BALB/c mice at multiple days after Japan/57 virus infection. The presence of granzyme B was detected by intracellular staining with specific Ab. The percentage of granzyme B expression by CD8+ gated T cells is shown on the left y-axis (). The absolute numbers of granzyme B+CD8+ T cells are shown on the right y-axis ().
Analysis of granzyme B+CD8+ T cells infiltrating the lungs (Fig. 5) was noteworthy in several respects. First, there was minimal accumulation of granzyme B+ T cells in the infected lungs up to day 4 of infection. Thus, despite the high lung virus titer over this period, there was neither significant nonspecific (inflammation induced) induction of granzyme B expression by lung-resident CD8+ T cells nor significant recruitment of specific (or Ag-nonspecific) CD8+ T cells to the infected lungs with an activated granzyme B+ phenotype. Second, by day 5 of infection, >30% of lung-resident CD8+ T cells were granzyme B+. The tempo of granzyme B+CD8+ T cell accumulation in the lungs directly paralleled our recent observations (22) on the time course of activated granzyme B+ T cell migration from the draining lymph nodes to the infected lungs in a TCR transgenic CD8+ T cell adoptive transfer model. More importantly, this frequency of granzyme B+ T cells in the lungs was well above the predicted total frequency of epitope-specific CD8+ T cells detected by ICCS/tetramer staining, i.e., <10% (Fig. 1A), and suggested that early in the CD8+ T cell response to infection in the lungs, the frequency of virus-specific CD8+ T cells may be substantially higher than that detected by the ICCS/tetramer labeling methods. Third, by day 6 postinfection, when pulmonary virus titers began to drop precipitously (Fig. 4), >70% of lung CD8+ T cells were granzyme B+. This T cell frequency was two to three times higher than the cumulative frequency of Japan/57 epitope-specific cells determined by ICCS and/or tetramer staining, again suggesting that the frequency of specific T cells infiltrating the lungs was substantially higher than that estimated by recognition of specific epitopes. Furthermore, cell surface marker analyses of the day 6 CD8+ T cells demonstrated that at least 60% of these cells expressed activation marker phenotypes (CD62L–, CD69+, LFA-1high, and VLA-4high) consistent with recent activation (data not shown). The high frequency of granzyme B+CD8+ T cells (>70%) in the infected lungs was also sustained up to day 9. At this point in time, the frequency (percentage) of granzyme B+CD8+ T cells closely approximated the frequency estimates of virus-specific CD8+ T cells in the lungs, defined in the ICCS assay using specific peptides or infected P815 cells to elicit IFN- synthesis (Fig. 3).
To further explore the Ag responsiveness of the lung-infiltrating granzyme B+CD8+ T cells during the early phase of the adaptive response (i.e., day 6 postinfection), we evaluated the expression (up-regulation) of CD107a (liposome-associated membrane proteins-1a) and CD69 on that cell population after short term in vitro stimulation with a synthetic peptide mixture corresponding to the four A/JAPAN/57 epitopes (see Materials and Methods). We observed that only a small percentage (<10%) of the lung-infiltrating granzyme B+CD8+ T cells up-regulated CD107a expression in response to antigenic stimulation. This finding is consistent with the low frequency of day 6 lung CD8+ T cells responding to peptide stimulation in the ICCS assay. By contrast, >95% of the lung-infiltrating granzyme B+CD8+ T cells up-regulated cell surface CD69 expression in response to in vitro stimulation with the viral peptide mixture (data not shown). The latter finding suggests that these lung-infiltrating, granzyme B+CD8+ T cells, which exhibit diminished tetramer staining and responsiveness in effector assays such as ICCS and CD107a up-regulation (8), represent a population of virus-specific T cells that are responsive to TCR engagement and specific antigenic stimulation.
Proliferation of lung-resident CD8+ T cells
If, as we believe, granzyme B expression is a useful marker for the activated Ag-specific responding CD8+ T cells, then substantial numbers of specific CD8+ T cells should have accumulated in the draining MLN on days 5–6 of infection. Although we did detect a significant expansion of granzyme B+CD8+ T cells in the draining MLN before (day 4) and during (days 5–6) CD8+ T cell accumulation in the lungs, we could not readily account for the elevated numbers of T cells in the lungs based on our calculations of total MLN-resident granzyme B+CD8+ T cells (Fig. 5). One potential explanation is that specific CD8+ T cells were activated in and recruited from other lymphoid tissues (e.g., the spleen) to the lungs. For reasons discussed later (see Discussion) we do not believe that this is the case. Alternatively, either the frequency of granzyme B+ T cells underestimated the true frequency of proliferating CD8+ T cells in the draining MLN, or there was significant proliferation of the responding virus-specific CD8+ T cells in the infected lungs.
To directly estimate the frequency of proliferating CD8+ T cells responding to i.n. infection, we used the strategy of in vivo uptake of the thymidine analog BrdU with a flow cytometry-based analysis to identify BrdU+ proliferating T cells. In a previous report (23), up to 80% of CD8+ T cells infiltrating the lung airways were BrdU+ on day 8 of infection when a continuous in vivo labeling method was used to supply BrdU in the drinking water throughout infection. Our initial attempts to continuously label responding cells by adding BrdU to the drinking water of infected animals were unsuccessful, because mice became moribund shortly after i.n. infection and exhibited a marked decrease in water consumption. To circumvent this problem, we used the strategy of i.p. injection of BrdU at different time points after virus infection and harvested the MLN and infected lungs 24 h later to enumerate BrdU+CD8+ T cells in these sites. This strategy would also allow us to identify the approximate time point of maximum T cell proliferation in the MLN.
As Fig. 6A demonstrates, proliferating (BrdU+) CD8+ T cells were not detectable in the draining MLN at a significant frequency before the 24-h interval between days 3–4 of infection. The rate of CD8+ T cell proliferation reached a maximum in the MLN during the day 5–6 interval, and proliferating T cells were detected at a similar frequency through day 8 before decreasing in frequency later in infection. The frequency of BrdU+ T cells in the draining MLN at the peak of proliferation (12% on day 6) was slightly higher than the corresponding frequency of granzyme B+ cells in the lymph nodes at the same time period (7–8%; Fig. 5). We anticipated that the kinetics of BrdU+ T cell accumulation in the lungs might be slightly delayed compared with those in the MLN, and that the frequency of BrdU+ T cells within the infected lungs might be higher during the day 6–8 interval, reflecting the migration of activated/proliferating CD8+ T cells from the MLN (or other lymphoid tissues) to the lungs over the 24-h labeling interval. As demonstrated in Fig. 6A, this was not the case. BrdU+CD8+ T cells were detected in the lungs at an overall frequency comparable to that in the draining MLN.
FIGURE 6. BrdU incorporation by proliferating CD8+ T cells in MLN and lungs during influenza virus infection. BrdU was either injected i.p. (A) or administered via the i.n. route (B) to detect cells that had synthesized DNA in the previous 24 h. The percentage of BrdU expression is shown for CD8+ gated T cells at the indicated number of days postinfection.
The kinetics of BrdU+ T cell accumulation in the lungs, as determined by 24-h pulse BrdU labeling, was consistent with the concept that activated virus-specific CD8+ T cells proliferated primarily, if not exclusively, in lymphoid tissues (e.g., draining MLN) before migration to the infected lungs. However, the frequency of BrdU+CD8+ T cells in the lungs on days 5–6 (12%) was well below the frequency of granzyme B+ T cells (30% on day 5 and 70% on day 6) detected in the lungs at this time (Fig. 5). Although the division time of activated CD8+ T cells has been reported to be as short as 6 h (28), it is possible that this discrepancy in the frequency of specific CD8+ T cells simply reflected our inability to accurately assess the extent of T cell proliferation in the MLN using the relatively short 24-h interval of BrdU labeling in vivo. An alternative explanation was suggested by the possibility that during the inflammatory response in a site such as the lungs (e.g., during pulmonary influenza virus infection), there is a local accumulation of nucleotide precursors resulting from DNA degradation and the release of precursor pools locally from dead and dying cells (29). This effect could result in dilution of the thymidine analog available for uptake by proliferating CD8+ T cells in the lungs and a decrease in the efficiency of BrdU labeling in the lungs.
We decided to explore this possibility by locally introducing BrdU in the respiratory tract by i.n. administration during virus infection at 24-h intervals before harvesting lungs and MLN for analysis of BrdU uptake in a manner analogous to the i.p. BrdU administration. We anticipated that this i.n. BrdU labeling strategy might result in a decreased efficiency of labeling of T cells in the draining MLN. As shown in Fig. 6B, this was indeed the case. The frequency of BrdU+ cells in the MLN was lower at any time point sampled, but the overall kinetics of BrdU labeling in the lymph nodes after i.n. BrdU administration were similar to the time course determined by i.p. BrdU administration (Fig. 6). In marked contrast, i.n. BrdU administration revealed a substantially higher frequency of BrdU+ T cells in the infected lungs, particularly during the 24-h interval between days 6–7 of infection when 30% of the lung-infiltrating CD8+ T cells were BrdU+. This finding suggests that in addition to undergoing proliferative expansion in the draining lymph nodes, activated T cells may continue to undergo one or more divisions upon entering infected lungs.
Discussion
In this report we examined the contribution of a set of defined MHC class I-restricted CD8+ T cell epitopes to the overall pulmonary CD8+ T cell response during experimental murine type A influenza virus infection (30). We used the complementary techniques of MHC class I tetramer labeling and Ag-dependent cytokine synthesis in the ICCS assay to identify epitope-specific T cells based on structural (TCR binding) and functional criteria. Using this approach, we were able to define the tempo and magnitude of the pulmonary CD8+ T cell response during primary infection and demonstrate that >90% of the total influenza virus-specific pulmonary CD8+ T cell response at its peak (i.e., days 9–11) was directed to four defined epitopes. Based on this information, we went on to examine the early CD8+ T cell response in the draining lymph nodes and lungs using the expression of granzyme B by activated CD8+ T cells as a surrogate marker for specific T cell activation. We observed that early in the primary CD8+ T cell response to infection (i.e., days 4–7), the conventional tetramer/intracellular cytokine methods for enumerating specific T cells appeared to underestimate the frequency of specific CD8+ T cells in the lungs. In addition, we obtained evidence for proliferation of activated CD8+ T cells in the lungs during this early phase of infection by using intranasal administration of the thymidine analog BrdU.
The cumulative frequency of influenza virus epitope-specific CD8+ T cells detected in our study is higher than that previously reported for a primary pulmonary influenza virus infection (3, 4, 25), most likely because the four epitopes represent the vast majority of specific CD8+ T cells directed to this virus strain in H-2d haplotype mice (Fig. 3). However, these frequencies are comparable to the high frequencies of virus-specific CD8+ T cells detectable in the lungs during other respiratory virus infections (6, 7). The kinetics of specific CD8+ T cell appearance, peak accumulation, and contraction in this report also closely parallel previously reported observations of the primary pulmonary CD8+ T cell response to influenza virus infection using different mouse and virus strains (3, 4, 5). This suggests that our findings are not unique to a particular virus/mouse strain combination.
CD8+ T cells have been documented to play an important role in virus clearance during primary influenza virus infection (30). Consistent with previous reports (3, 25, 31), we observed that influenza virus-specific CD8+ T cells were first detected in the BAL and lung parenchyma on days 5–6 postinfection by cytokine synthesis and/or tetramer binding (Fig. 1). This corresponded to the same time frame as the onset of virus clearance from the infected lungs (Fig. 4). However, the drop in lung virus titer was much more dramatic on days 5–7 of infection than might be expected from the relatively low frequency of CD8+ T cells detected in lungs at this time. Furthermore, both the frequency and absolute number of virus-specific lung CD8+ T cells on days 5–7 (as determined by cytokine synthesis/tetramer binding) represented only a small fraction of the specific T cells detected at the peak of the response 4–5 days later when infectious virus was no longer detectable in the lungs. This discrepancy between virus clearance and CD8+ T cell frequency early in infection raised the possibility that techniques to evaluate cytokine synthesis/TCR binding underestimated the actual frequency/number of influenza virus-specific CD8+ T cells in the lungs. This possibility was reinforced by our recent findings in a TCR transgenic model of in vivo CD8+ T cell responses to pulmonary influenza virus infection.4 Using this model, we demonstrated that shortly (1–2 days) after in vivo activation, naive influenza virus-specific CD8+ T cells transiently lost the capacity to bind specific MHC class I tetramers and exhibited diminished cytokine (IFN-) synthesis in the ICCS assay. This loss of tetramer binding/cytokine synthesis was the most pronounced on days 3–4 of infection in the draining lymph nodes and on days 4–6 in the infected lungs, with full restoration of tetramer binding/cytokine synthesis by days 8–10 postinfection.4
Based on these considerations, we explored the possibility of using the expression of the cytolytic effector molecule granzyme B as a surrogate marker to identify virus-specific activated CD8+ T cells responding in the lungs to infection. Granzyme B is not expressed by naive T cells, but is produced and stored in intracellular granules after CD8+ T cell activation (27, 32). In this connection, we have recently reported (23) that granzyme B expression is rapidly induced in the majority of activated virus-specific CD8+ T cells in response to respiratory influenza virus infection shortly after the onset of cell division in draining lymph nodes. Granzyme B expression is also sustained at high levels in lung-infiltrating effector CD8+ T cells after migration from lymph nodes to infected lungs. The frequency of virus-specific CD8+ T cells responding in the draining lymph nodes early (days 5–6) after pulmonary infection has been reported to be extremely low and at the limits of detection by tetramer staining (our unpublished observations) (4, 24, 25). However, we were able to reproducibly detect granzyme B+CD8+ T cells as early as day 3 of infection in the draining MLN, with a maximum frequency of granzyme B+CD8+ T cells of 7% in the draining lymph nodes 3–4 days later.
The companion analysis of CD8+ T cells in the infected lungs revealed that >30% of lung-infiltrating CD8+ T cells expressed granzyme B on day 5 of infection, and >70% of the CD8+ T cells in the lungs were granzyme B+ on day 6 postinfection. This frequency was 3 times higher than that of virus epitope-specific CD8+ T cells detected by tetramer/cytokine at the corresponding day 6 point. It is also noteworthy that granzyme B+CD8+ T cells were detected at a similar frequency (70%) on day 9 of infection when the overall frequency of influenza-specific CD8+ T cells in infected lungs was determined by ICCS using infected P815 cells to likewise correspond to 70%. These findings support the view that effector CD8+ T cells accumulate in the lungs at significant frequencies and in substantial numbers early in the course of infection, i.e., days 5–6, when substantial virus clearance occurs. However, because the majority of these lung-infiltrating granzyme B+CD8+ T cells present early (i.e., days 5–6) in the host response do not bind specific tetramer and do not respond in either the ICCS or CD107a mobilization assay, we cannot directly document that these T cells are influenza specific.
Because granzyme B expression merely marks CD8+ T cells as having been previously activated (and therefore not necessarily virus specific), it is possible that the high frequency of granzyme B+CD8+ T cells infiltrating infected lungs early after infection (days 5–6) could reflect nonspecific recruitment of irrelevant CD8+ T cells to a site of inflammation. Indeed, Ag-nonspecific memory CD8+ T cells have been shown to migrate to the lungs after respiratory infection with RSV, influenza virus, and Sendai virus (33, 34, 35). However, nonspecific recruitment of irrelevant memory CD8+ T cells to the inflamed lungs is an unlikely explanation for our findings. This phenomenon has been reported to peak early after infection, i.e., days 4–5 (33, 36) and to not occur at the sustained high frequency exhibited by the lung-infiltrating granzyme B+ cells demonstrated in this study. Furthermore, the level of granzyme B protein expression in memory cells has been previously demonstrated to be very low (26, 27, 36), whereas we detected a uniformly high level of granzyme B in the CD8+ T cells infiltrating the lungs on days 5–6 after infection (data not shown). In this regard, it is noteworthy that differential granzyme B gene expression has been previously used in a reporter-based transgenic strategy to distinguish effector T cells from memory T cells in vivo (37). Finally, the extremely close correspondence between the frequency of granzyme B+CD8+ T cells and the predicted frequency of influenza virus-specific CD8+ T cells on day 9 of infection supports the view that granzyme B may serve as a useful marker in identifying recently activated Ag-specific CD8+ T cells throughout the course of infection.
Our analysis of the frequency and total number of granzyme B+CD8+ T cells in draining MLN and infected lungs suggested that the observed numbers of granzyme B+ cells detected in the lungs early in infection could not readily be accounted for by the accumulation of migrant CD8+ T cells derived from the draining lymph nodes. We believe this is unlikely to be due to the recruitment of activated virus-specific T cells from other secondary lymphoid organs (e.g., spleen or nondraining lymph nodes). The reasons for this are 2-fold. First, during influenza virus infection, Ag-bearing respiratory dendritic cells were found to preferentially, if not exclusively, migrate to the lung-draining lymph nodes (38). Second, in a recent analysis of CD8+ T cell activation in vivo in response to influenza infection, initial/early CD8+ T cell activation was restricted to those cells present in draining MLN (23).
We therefore investigated the possibility that after migration from draining lymph nodes, activated CD8+ T cells proliferated in the lungs and thereby contributed to the elevated T cell numbers detected. We were able to demonstrate proliferation of activated CD8+ T cells directly in situ in the infected lungs by i.n. administration of the thymidine analog BrdU. Delivery of BrdU by the i.n. route proved superior to i.p. BrdU administration in the detection of proliferating CD8+ T cells in the lungs. It is unlikely that the efficient labeling of lung CD8+ T cells with BrdU after i.n. administration was due to transport of the nucleoside analog to the draining lymph nodes and labeling of proliferating T cells there before their migration to the infected lungs. In fact, the efficiency of BrdU uptake by CD8+ T cells in the draining lymph nodes after i.n. administration was lower than the efficiency of labeling after i.p. administration. We believe that the high local concentration of DNA nucleotide/nucleoside at sites of inflammation and/or associated cell death (39), such as influenza virus-infected lungs, results in dilution of BrdU after parenteral administration and short term (24-h) in vivo labeling. Because in vivo BrdU uptake by T cells after i.n. administration was monitored at 24-h intervals, and transfer of BrdU from the respiratory tract to the draining MLN was readily detected in our analysis, it remains a formal possibility that the high frequency of BrdU+CD8+ T cells detected in the lung resulted from uptake of a label by cells at a distant sites (e.g., the secondary lymphoid organs). As yet, we cannot rigorously exclude this possibility. However, the finding that i.p. BrdU administration does not result in the preferential accumulation of BrdU+CD8+ T cells in the infected lungs does not readily support this view.
In conclusion, we have characterized the CD8+ T cell responses to a series of defined virus-specific CD8+ T cell epitopes. These sites account for >90% of the overall CD8+ T cell response directed against influenza virus in the lungs. With this information available, we were able to explore the use of a surrogate marker, i.e., granzyme B expression, to demonstrate that this protein may be useful in identifying specifically activated effector CD8+ T cells. By using granzyme B expression, we demonstrated that virus-specific CD8+ T cells may accumulate in the lungs at higher frequencies earlier in the course of the antiviral CD8+ T cell response than predicted by the use of conventional tetramer/intracellular cytokine methods. Finally, we provided evidence for the first time that activated effector CD8+ T cells may continue to undergo one or more rounds of proliferation in situ in the infected lungs after migrating from the draining lymph nodes. Thus, in situ CD8+ T cell proliferation at a site of infection/inflammation such as the lungs as well as the proliferative expansion of naive CD8+ T cells in secondary lymphoid organs such as the draining MLN contribute to the overall magnitude of the CD8+ T cell response during pulmonary influenza virus infection.
Acknowledgments
We thank Dr. Donald R. Drake III for advice and assistance with the design of experiments.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by U.S. Public Health Service Grants AI15608, HL33391, and HL71875 and U.S. Public Health Service Training Grant AI07496.
2 Address correspondence and reprint requests to Dr. Thomas J. Braciale, Beirne B. Carter Center for Immunology Research, University of Virginia Health Sciences Center, MR-4 Building, HSC Box 801386, Charlottesville, VA 22908. E-mail address: tjb2r@virginia.edu
3 Abbreviations used in this paper: i.n., intranasal; BAL, bronchoalveolar lavage; HA, hemagglutinin; ICCS, intracellular cytokine staining; MLN, mediastinal lymph node; NP, nucleoprotein; TCID50, half-maximal tissue culture infectious dose.
4 D. R. Drake, R. M. Ream, C. W. Lawrence, and T. J. Braciale. Transient loss of MHC class I tetramer binding after CD8+ T cell activation reflects T cell effector function. Submitted for publication.
Received for publication October 13, 2004. Accepted for publication February 18, 2005.
References
Chen, W., L. C. Anton, J. R. Bennink, J. W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12:83.
Chen, W., J. R. Bennink, P. A. Morton, J. W. Yewdell. 2002. Mice deficient in perforin, CD4+ T cells, or CD28-mediated signaling maintain the typical immunodominance hierarchies of CD8+ T-cell responses to influenza virus. J. Virol. 76:10332.
Flynn, K. J., G. T. Belz, J. D. Altman, R. Ahmed, D. L. Woodland, P. C. Doherty. 1998. Virus-specific CD8+ T cells in primary and secondary influenza pneumonia. Immunity 8:683.
Belz, G. T., W. Xie, J. D. Altman, P. C. Doherty. 2000. A previously unrecognized H-2Db-restricted peptide prominent in the primary influenza A virus-specific CD8+ T-cell response is much less apparent following secondary challenge. J. Virol. 74:3486.
Belz, G. T., W. Xie, P. C. Doherty. 2001. Diversity of epitope and cytokine profiles for primary and secondary influenza a virus-specific CD8+ T cell responses. J. Immunol. 166:4627.
Usherwood, E. J., R. J. Hogan, G. Crowther, S. L. Surman, T. L. Hogg, J. D. Altman, D. L. Woodland. 1999. Functionally heterogeneous CD8+ T-cell memory is induced by Sendai virus infection of mice. J. Virol. 73:7278.
Chang, J., A. Srikiatkhachorn, T. J. Braciale. 2001. Visualization and characterization of respiratory syncytial virus F-specific CD8+ T cells during experimental virus infection. J. Immunol. 167:4254.
Butz, E. A., M. J. Bevan. 1998. Massive expansion of antigen-specific CD8+ T cells during an acute virus infection. Immunity 8:167
Murali-Krishna, K., J. D. Altman, M. Suresh, D. J. Sourdive, A. J. Zajac, J. D. Miller, J. Slansky, R. Ahmed. 1998. Counting antigen-specific CD8 T cells: a reevaluation of bystander activation during viral infection. Immunity 8:177.
Sweetser, M. T., V. L. Braciale, T. J. Braciale. 1989. Class I major histocompatibility complex-restricted T lymphocyte recognition of the influenza hemagglutinin: overlap between class I cytotoxic T lymphocytes and antibody sites. J. Exp. Med. 170:1357.
Hahn, Y. S., C. S. Hahn, V. L. Braciale, T. J. Braciale, C. M. Rice. 1992. CD8+ T cell recognition of an endogenously processed epitope is regulated primarily by residues within the epitope. J. Exp. Med. 176:1335.
Milligan, G. N., L. A. Morrison, J. Gorka, V. L. Braciale, T. J. Braciale. 1990. The recognition of a viral antigenic moiety by class I MHC-restricted cytolytic T lymphocytes is limited by the availability of the endogenously processed antigen. J. Immunol. 145:3188.
Bodmer, H. C., R. M. Pemberton, J. B. Rothbard, B. A. Askonas. 1988. Enhanced recognition of a modified peptide antigen by cytotoxic T cells specific for influenza nucleoprotein. Cell 52:253.
Braciale, T. J., M. T. Sweetser, L. A. Morrison, D. J. Kittlesen, V. L. Braciale. 1989. Class I major histocompatibility complex-restricted cytolytic T lymphocytes recognize a limited number of sites on the influenza hemagglutinin. Proc. Natl. Acad. Sci. USA 86:277.
Cao, W., B. A. Myers-Powell, T. J. Braciale. 1996. The weak CD8+ CTL response to an influenza hemagglutinin epitope reflects limited T cell availability. J. Immunol. 157:505.
Betts, M. R., J. M. Brenchley, D. A. Price, S. C. De Rosa, D. C. Douek, M. Roederer, R. A. Koup. 2003. Sensitive and viable identification of antigen-specific CD8+ T cells by a flow cytometric assay for degranulation. J. Immunol. Methods 281:65.
Altman, J. D., P. A. Moss, P. J. Goulder, D. H. Barouch, M. G. McHeyzer-Williams, J. I. Bell, A. J. McMichael, M. M. Davis. 1996. Phenotypic analysis of antigen-specific T lymphocytes. Science 274:94.
Belz, G. T., D. Wodarz, G. Diaz, M. A. Nowak, P. C. Doherty. 2002. Compromised influenza virus-specific CD8+-T-cell memory in CD4+-T-cell-deficient mice. J. Virol. 76:12388.
Eichelberger, M., W. Allan, M. Zijlstra, R. Jaenisch, P. C. Doherty. 1991. Clearance of influenza virus respiratory infection in mice lacking class I major histocompatibility complex-restricted CD8+ T cells. J. Exp. Med. 174:875.
Turner, S. J., R. Cross, W. Xie, P. C. Doherty. 2001. Concurrent naive and memory CD8+ T cell responses to an influenza A virus. J. Immunol. 167:2753.(Christopher W. Lawrence, )
We have used intracellular cytokine staining and MHC class I tetramer binding in conjunction with granzyme B protease expression and in vivo BrdU uptake to characterize the primary murine CD8+ T cell response to pulmonary influenza virus infection. We have observed that the majority (>90%) of the CD8+ T cell response to the A/Japan/305/57 virus in the lung at the peak of the response (days 9–11) is directed to four epitopes (three dominant and one subdominant). Using induction of granzyme B as a surrogate to identify specific activated CD8+ T cells, we found that an unexpectedly large fraction (70%) of lung-infiltrating CD8+ T cells expressed granzyme B on day 6 of infection when estimates by MHC tetramer/intracellular cytokine staining yielded substantially lower frequencies (30%). In addition, by using intranasal administration of BrdU during infection, we obtained evidence for proliferative expansion of activated CD8+ T cells in the infected lung early (days 5–7) in the primary response. These results suggest that the frequency and number of specific CTL present in the lung early in infection may be underestimated by standard detection methods, and primary CD8+ T cell expansion may occur in both secondary lymphoid organs and the infected lung.
Introduction
A significant amount of our current understanding of specific CD8+ T cell responses to viral epitopes has been garnered through the study of experimental influenza virus infection of mice. For example, extensive investigations into the basis of immunodominance of CD8+ T cell responses to multiple epitopes have been performed in both BALB/c (H-2d) and C57BL/6 (H-2b) mice after i.p. inoculation (1, 2). Furthermore, because influenza virus is a respiratory tract pathogen, much of our current knowledge about the specific CD8+ T cell response to influenza virus has come from analyzing the specificity of the pulmonary CD8+ T cell response to intranasal (i.n.)3 challenge (3, 4, 5). These important studies have detailed the immunodominance hierarchy to several CD8+ T cell epitopes from the HKx31 (H3N2) and PR/8 (H1N1) influenza viruses in H-2b mice. The most dominant epitopes from these viruses are polymerase A224–233 and nucleoprotein (NP)366–374, which, along with other minor epitopes, are targets for the 40% of pulmonary CD8+ T cells that are identifiable as virus-specific during the primary immune response (3, 4, 5).
By contrast, other respiratory tract viruses have been shown to induce an even more potent specific CD8+ T cell response in mice. For example, i.n. Sendai virus infection elicits a highly restricted CD8+ T cell population in which 70% of pulmonary CD8+ T cells are directed against a single immunodominant epitope (6). Similarly, 50–60% of CD8+ T cells responding to a primary i.n. respiratory syncytial virus infection are directed against two epitopes, the vast majority of which are specific to a single epitope as well (7). Therefore, the frequency of CD8+ T cells responding to respiratory viruses probably parallels the high frequencies of pathogen-specific CD8+ T cells that are prevalent during the response to a potent, systemic viral infection such as acute lymphocytic choriomeningitis virus challenge (8, 9). To examine whether a similarly high frequency of virus-specific CD8+ T cells may be directed against influenza virus, we analyzed the CD8+ T cell response to primary i.n. infection of BALB/c mice with a mouse-adapted A/Japan/305/57 (Japan/57) virus.
CD8+ T cells responding to Japan/57 virus in BALB/c mice recognize four known epitopes restricted by H-2Kd (Table I). Three of these viral epitopes, hemagglutinin (HA)204–212, HA210–219 (10, 11), and HA529–537 (12), are processed from the HA protein, whereas a fourth, NP147–155 (13), is derived from the viral NP. Previous studies have demonstrated that bulk, polyclonal cultures prepared from the splenocytes of mice that were immunized i.p. with Japan/57 possessed high cytolytic activity against the HA204–212, HA529–537, and NP147–155 epitopes, whereas the cytolytic activity against HA210–219 was much weaker (14, 15). However, the frequency of virus-specific CD8+ T cells responding in vivo to i.n. Japan/57 virus infection had not been analyzed.
Table I. Amino acid sequences of CD8+ T cell epitopes from A/Japan/305/57 influenza virus
In this report we have characterized the magnitude and immunodominance hierarchy of the CD8+ T cell response against the four previously described epitopes during pulmonary Japan/57 influenza virus infection. We demonstrate that the frequency of CD8+ T cells that are specific to these influenza virus epitopes is indeed high during the peak of the pulmonary response, as determined by intracellular cytokine staining (ICCS) and MHC class I tetramer labeling. Activated CD8+ T cells directed to these four epitopes represent 90% of the total responding CD8+ T cells in the lungs on day 9 of infection by ICCS using cytokine synthesis by lung-infiltrating CD8+ T cells to influenza virus-infected cells as a measure of total virus-specific CD8+ T cells. The immunodominance hierarchy of the response to these epitopes in vivo followed the predicted pattern based on previous in vitro studies. In addition, we explored the use of granzyme B protease expression by activated CD8+ T cells as a surrogate marker to identify specific CD8+ T cells. We obtained evidence that granzyme B expression may identify virus-specific CD8+ T cells that have migrated to the lungs early after infection, i.e., days 5–7, but are not efficiently detected by ICCS or MHC tetramer binding. Finally, we evaluated the kinetics of CD8+ T cell proliferation in vivo during infection by systemic (i.p.) or local (i.n.) administration of the thymidine analog BrdU. Using this approach, we obtained evidence suggesting the idea that activated virus-specific CD8+ T cells may undergo proliferative expansion in both secondary lymphoid organs (draining lymph nodes) and infected lungs early after infection. The significance of these results for the induction of CD8+ T cell responses to respiratory virus infection is discussed.
Materials and Methods
Mice and experimental infection
Female BALB/c mice (H-2d) were purchased from Taconic Farms and maintained under specific pathogen-free conditions. Mice that were 8–11 wk of age were lightly anesthetized by Metofane (Janssen) inhalation and sublethally infected by i.n. inoculation (in 50 μl) with 3953 half-maximal tissue culture infectious doses (TCID50) of a mouse-adapted A/Japan/305/57 (H2N2) influenza virus.
Preparation of tissue lymphocytes
At multiple days after influenza virus infection, mice were killed by cervical dislocation. Mouse tracheas were then surgically exposed to perform bronchoalveolar lavages (BAL). Briefly, the tracheas were cannulated with a syringe, and five separate 1-ml aliquots of IMEM were used to rinse the lower respiratory tract to collect inflammatory cells from the air spaces. Lavaged lungs were then perfused via the right ventricle of the heart with 5–10 ml of PBS containing 10 U/ml heparin (Sigma-Aldrich) to remove blood lymphocytes from the lung vasculature. The lungs were minced and passed through a steel screen. Tissue debris was removed by centrifugation at 300 x g. Cells were counted and resuspended at appropriate concentrations for each particular experiment.
Peptides
Influenza virus peptides HA204–212, HA210–219, HA529–537, and NP147–155 (Table I) were synthesized by University of Virginia Biomolecular Research Facility. Purity was confirmed by HPLC analysis.
Ab labeling and flow cytometry
For intracellular cytokine staining experiments, harvested cells were incubated for 6 h in 96-well, round-bottom plates in IMEM supplemented with 10% FBS, 10 U/ml penicillin G, 10 μg/ml streptomycin sulfate, 100 U/ml human IL-2, 2 mM L-glutamine, 0.05% 2-ME, and 1 μg/ml brefeldin A (Sigma-Aldrich). This relatively high concentration of human IL-2 was used to maximize the synthesis and accumulation of cytokines achieved in the ICCS. This high IL-2 concentration did not affect the frequency of cytokine-secreting cells (data not shown). Stimulatory conditions included 1 μM of a given synthetic peptide and Japan/57 influenza virus-infected P815 (H-2d) mastocytoma cells. After incubation, cells were washed and incubated for 10 min in FcBlock (BD Biosciences). The allophycocyanin-conjugated anti-CD8 mAb (clone 53-6.7; BD Biosciences) was used to label cells for 30 min at 4°C. After washing with FACS buffer (PBS/1% FBS), cells were fixed, and erythrocytes were simultaneously lysed using FACS Lysing Solution (BD Biosciences). The cells were labeled at room temperature with anti-IFN- (clone XMG1.2) or rat IgG1 (clone R3-34; BD Biosciences) in permeabilization buffer (FACS buffer containing 0.5% saponin). The percentage of background cytokine staining from the appropriate population of unstimulated cells was subtracted from the percentage of cytokine-positive cells under each stimulation condition before reporting. For intracellular granzyme B staining, cell suspensions were prepared from mediastinal lymph nodes (MLN) and lungs at multiple days after influenza virus infection. Cells were first surface-labeled with anti-CD8 (clone 53-5.8; BD Biosciences) for 30 min, then fixed and permeabilized as described above. Permeabilized cells were labeled with anti-granzyme B (clone GB12; Caltag Laboratories) or mouse IgG1 (BD Biosciences) for 45 min at room temperature. Flow cytometry data were acquired for each of the experiments using a BD FACSCalibur (BD Immunocytometry Systems) and were analyzed using CellQuest software (BD Biosciences).
Cell surface marker expression
Lung cell suspensions (see Preparation of tissue lymphocytes above) were prepared from the lungs of four or five mice on day 6 after infection. Lung-infiltrating CD8+ T cells were purified from pooled lung cell suspensions by CD8-conjugated magnetic bead positive selection (Miltenyi Biotec) according to the manufacturer’s instructions. Surface CD107a mobilization was examined using a modification of the protocol described by Bells et al. (16). Briefly lung CD8+ T cells (either in total lung cell homogenates or after purification) were cultured for 6 h at 37°C with P815 cells, which were either untreated or treated with a mixture of the four viral peptide epitopes in the presence of monensin (1 μg/ml) and PE-conjugated anti-CD107 Ab (ID4B; Santa Cruz Biotechnology) at a final concentration of 5 μg/ml throughout the assay period. Cells were then fixed and stained for CD69 and CD8 expression (as above).
MHC class I tetramers
MHC class I tetramers were produced according to published techniques (17). DNA plasmids encoding human 2-microglobulin and the extracellular domain of the murine H-2Kd H chain were provided by J. Altman (Emory University, Atlanta, GA) and E. Pamer (Sloan-Kettering Institute, New York, NY), respectively. MHC class I tetramers were prepared using each of the following influenza virus peptides: HA204–212, HA210–219, HA529–537, and NP147–155. Tetramer labeling of harvested cells was performed for 45 min at 4°C.
Virus titers
Mice were i.n. infected with influenza Japan/57 virus and were killed by cervical dislocation at multiple time points after infection. Excised mouse lungs were weighed and homogenized. Multiple dilutions of clarified homogenates were plated on Mardin-Darby canine kidney cells in 96-well, flat-bottom plates. After 3 days in a 37°C humidified 7% CO2 incubator, 50 μl of supernatant was harvested from each well and mixed with 50 μl of fresh 1% chicken erythrocyte solution in round-bottom, 96-well plates. After 1 h at room temperature, hemagglutination results were determined and used to calculate TCID50 titers.
BrdU labeling
On multiple days after pulmonary influenza virus infection, mice were lightly anesthetized by Metofane (Janssen) inhalation and then were administered 50 μl of 16 mg/ml BrdU (0.8 mg total/mouse) by the i.n. route or 250 μl of 8 mg/ml BrdU (2 mg total/mouse) by i.p. injection. Each mouse was killed 24 h after BrdU treatment. Tissue cells were prepared for cell surface Ab labeling as described above. Anti-CD8-labeled cells were fixed and permeabilized before BrdU staining was performed using the BD FastImmune BrdU kit according to the manufacturer’s instructions (BD Biosciences).
Results
Frequency and magnitude of the pulmonary CD8+ T cell response to defined epitopes after influenza virus infection
We have analyzed the frequency of pulmonary CD8+ T cells directed to four previously identified MHC class I-restricted epitopes of the Japan/57 (H2N2) virus in H-2d haplotype BALB/c mice (Table I) (10, 11, 12, 13) in response to sublethal i.n. Japan/57 infection by ICCS for IFN- and by MHC tetramer binding. Three of these H-2Kd-restricted epitopes, HA204–212, HA529–537, and NP147–155, were characterized in vitro as immunodominant, whereas the fourth epitope, HA210–219, also Kd-restricted, was classified as subdominant based on the magnitude of the in vitro CTL response (14, 15).
Analysis of the kinetics of accumulation of IFN--secreting CD8+ T cells in BAL fluid revealed that Japan/57-specific, IFN- secreting CD8+ T cells were first detected on day 5 postinfection (Fig. 1A). The frequency of IFN--secreting T cells to each of the defined epitopes was always <5% at this time, with <10% of BAL CD8+ T cells in aggregate identified as virus-specific by this criterion. As previously reported in other influenza virus infection models (3, 4, 5), over the next 2–5 days the frequency (percentage) of epitope-specific, IFN--producing CD8+ T cells increased significantly and reached a maximum level by days 9–10 of infection, with subsequent contraction of the response during the following week (Fig. 1A). Analysis of the IFN- response of CD8+ T cells infiltrating the lung parenchyma (Fig. 1B) revealed a comparable kinetics of expansion and contraction, although the magnitude of the response, i.e., the frequency of epitope-specific CD8+ T cells in the lung parenchyma, was slightly reduced compared with that in BAL. A companion kinetic analysis of the CD8+ T cell response in BAL by MHC class I tetramer staining demonstrated, as expected, a close correspondence in both the kinetics of accumulation and the frequency of specific T cells between MHC tetramer straining (Fig. 1C) and the ICCS assay (Fig. 1A). We did, however, note a modest delay in the contraction phase of the BAL CD8+ T cell response, as defined by the frequency of epitope-specific MHC class I tetramer staining.
FIGURE 1. Kinetics of influenza virus-specific pulmonary CD8+ T cells, as determined by intracellular cytokine and tetramer staining analyses. BALB/c mice were infected with 3953 TCID50 of Japan/57 influenza virus. BAL (A) and lung parenchyma (B) cells were harvested at multiple days after infection and analyzed in 6-h intracellular cytokine assays. C, BAL cells were collected at multiple days after infection and labeled with specific tetramer. The percentage of specific IFN- expression or tetramer labeling is shown for CD8+ gated T cells at the indicated number of days postinfection. Data are representative of three independent kinetic analyses with pooled tissue (samples) from three to five mice per time point. Values are the mean ± SEM for three or four replicate samples for each time point.
As revealed above (Fig. 1), the hierarchy of immunodominance for the four epitopes, initially suggested by in vitro studies of immune T cell populations, was confirmed by the analysis of epitope-specific responses during the in vivo CD8+ T cell response to primary infection in the respiratory tract. Thus, CD8+ T cells directed to the NP147–155, HA204–212, and HA529–537 epitopes dominated the response in vivo, with the weaker (subdominant) response to HA210–219 likewise demonstrable in the lungs after primary infection. When the kinetics of the total CD8+ T cell response to each epitope was calculated by either tetramer staining (Fig. 2A) or ICCS (Fig. 2B), the rates of expansion and contraction of the response to the three dominant epitopes followed a predictable pattern (3, 4, 5) based on the frequency of responding T cells.
FIGURE 2. Quantitation of influenza virus-specific CD8+ T cell numbers in the lung airways and parenchyma. A, BAL cells were harvested from mice at the indicated number of days after Japan/57 infection and analyzed by tetramer staining. B, Lung parenchyma cells were harvested at multiple days after infection and analyzed using the intracellular cytokine assay. The numbers of individual Ag-specific CD8+ T cells were calculated using total cell count and IFN-+ (A) or tetramer+ (B) frequency data and are displayed in the left graphs. The total number of virus-specific CD8+ T cells at each time point is inset in the right graphs. Data are representative of three independent analyses with values obtained from pools obtained from three to five donor mice per time point.
Contribution of epitope-specific CD8+ T cells to the overall pulmonary CD8+ T cell response
The findings described above suggested that a robust primary pulmonary CD8+ T cell response is mounted to these four epitopes after infection with Japan/57 virus. Although weak/subdominant responses to sites on several viral gene products from various influenza virus strains have been reported (1, 2, 5), previous analyses from our laboratory (14, 15) (unpublished observations) suggested that the response to the Japan/57 virus in H-2d haplotype mice is focused predominately, if not exclusively, on the four described epitopes. We therefore wished to determine the contribution to the overall pulmonary Japan/57-specific CD8+ T cell response by CD8+ T cells that were directed against these four epitopes in vivo after i.n. infection. To this end, we analyzed the frequency of IFN--secreting CD8+ T cells responding to each of the four peptide epitopes in BAL and lung parenchyma by ICCS at the peak of the primary response. In parallel, we also analyzed the IFN- response of pulmonary CD8+ T cells to Japan/57-infected P815 cells (H-2d) as a source of naturally processed viral Ags. The virus-infected cells would probably process and present all potential H-2d-restricted MHC class I epitopes, including, but not limited to, the four epitopes previously identified and analyzed in this study. We then calculated the sum total of the individual CD8+ T cell IFN- responses to each of the four Japan/57 epitopes and compared this value to stimulation by the virus-infected cells (which was normalized to 100%). Strikingly, we observed a very high correlation (93% in lung, 92% in BAL) of IFN- expression when comparing CD8+ T cell stimulation by virus-infected cells to the cumulative frequency of individual peptide-stimulated CD8+ T cells (Fig. 3). These data demonstrate that the specific CD8+ T cell response against influenza virus Japan/57 in H-2d mice is nearly entirely directed against the four viral epitopes presently studied. In this regard, it should be noted that at the peak of the pulmonary CD8+ T cell response (day 10), between 60 and 70% of the total pulmonary CD8+ T cells secreted IFN- in response to infected P815 cells. We also determined that the total complement of lung CD8+ T cells capable of secreting IFN- after nonspecific stimulation with PMA/ionomycin was 70%. Thus, by these criteria, the vast majority of lung CD8+ T cells were virus specific.
FIGURE 3. The total influenza virus-specific CD8+ T cell response is nearly entirely directed against four Japan/57 epitopes. Cell suspensions were prepared from the BAL and lung parenchyma on day 10 postinfection. Cells were stimulated for 6 h with each of the synthetic peptides and Japan/57-infected P815 cells. The frequency of IFN- expression by CD8+ T cells after stimulation by virus-infected cells was set at 100% for comparison with the cumulative response to each of the individual viral epitopes. The data are representative of three independent analyses using data pooled from four mice for infectious virus and individual peptides. Percentages of the total influenza response (reflected in the sum of individual epitopes) ranged from 89 to 95%.
Lung virus replication and the early CD8+ T cell response
Our previously discussed kinetic analysis of influenza virus-specific CD8+ T cell accumulation in the lungs during primary i.n. infection revealed that specific CD8+ T cells (defined by tetramer staining and ICCS) were first detectable on days 5–6 of infection. Furthermore, the response appeared to be directed primarily, if not exclusively, to the four described epitopes. With this information in hand, we next evaluated the time course of virus replication in the lungs to relate the tempo of virus clearance to the kinetics of virus-specific CD8+ T cell trafficking to the lungs. As Fig. 4 demonstrates, virus titers rapidly peaked (106.7 TCID50/g tissue) by 24 h postinfection and were sustained at or near peak titers through day 5 of infection. Between days 5 and 7, virus titers dropped precipitously (1000-fold), and infectious virus was no longer detectable in the lungs by day 9 of infection. This precipitous fall in pulmonary Japan/57 virus titer over days 5–7 of infection is in agreement with previously published results in the murine model using other influenza virus strains (18, 19, 20, 21) and correlates with the early infiltration of specific CD8+ T cells into the lung, i.e., days 5–6 (Fig. 1) (3). These kinetic observations implicate the early CD8+ T cell response in the infected lungs as critical in controlling virus replication and clearance. However, it is also evident from this kinetic analysis that the major control of virus replication and clearance occurred on days 5–7, which is before the peak of pulmonary CD8+ T cell accumulation in the lungs, i.e., days 9–11 (Fig. 2).
FIGURE 4. Kinetics of Japan/57 virus clearance from infected mouse lungs. Mice were i.n. inoculated on day 0 with 3953 TCID50 of Japan/57 influenza virus (inoculum dose denoted by asterisk). At several days after infection, mouse lungs were harvested, weighed, and homogenized. Clarified homogenates were plated on Mardin-Darby canine kidney cells. The presence of infectious influenza virus was analyzed 3 days later by hemagglutination assay to determine TCID50 titers.
Early accumulation of virus-specific CD8+ T cells within the infected lungs
The trafficking and accumulation of activated virus-specific T cells into the infected lungs are believed to be preceded by the activation and proliferative expansion of naive virus-specific CD8+ T cell precursors in organized lymphoid tissues, most notably the MLN draining the infected lung (22, 23), and with subsequent migration of activated effector CD8+ T cells from the lymph nodes to the infected lungs. However, analyses of the frequencies of epitope-specific CD8+ T cells detected in draining MLN by tetramer staining and/or ICCS in the early inductive phase of the CD8+ T cell response to primary i.n. influenza virus infection (i.e., up to day 4 postinfection) have demonstrated epitope-specific CD8+ T cell frequencies of 2% (4, 24, 25). In the case of the four CD8+ T cell epitopes examined in this study, the frequencies of specific CD8+ T cells enumerated by tetramer staining in the draining lymph nodes on day 4 postinfection have a cumulative total of 3–5% (data not shown).
Although the inability to detect significant proliferative expansion of CD8+ T cell precursors in the draining lymph nodes before T cell accumulation in the infected lungs would most likely reflect the inherent low frequency of naive specific CD8+ T cell precursors (22) and the presumed rapid egress of effector CD8+ T cells out of the lymph nodes after proliferative expansion, several recent findings from our laboratory have modified our view of the early events in CD8+ T cell activation in the draining lymph nodes and lungs. First, we have recently demonstrated that early in the activation of influenza virus-specific CD8+ T cells within draining lymph nodes during pulmonary influenza virus infection, activated proliferating CD8+ T cells transiently lose the ability to bind specific tetramers without any loss of cell surface TCR expression.4 This transient loss of tetramer binding by activated T cells (up to days 5–6 postinfection) is associated with a transient diminished ability of the T cells to secrete IFN- in the ICCS assay, but without any deficit in Ag-driven cell proliferation.4 Importantly, a significant fraction (>60%) of activated CD8+ T cells that have migrated to the infected lung early during the T cell response to infection (days 5–7) retain the diminished capacity to bind specific tetramer and secrete IFN- in the ICCS assay. Full tetramer staining and cytokine secretion are, however, restored in CD8+ T cells infiltrating the lung by day 8 postinfection.4 Second, we have recently demonstrated that 24–48 h after naive CD8+ T cell activation and the onset of T cell proliferation in draining MLN in response to pulmonary influenza virus infection, activated virus-specific CD8+ T cells within the draining lymph nodes express the granule-associated lytic molecule granzyme B, and that granzyme B expression is uniformly retained by activated T cells in the lymph nodes and after T cell migration to the infected lungs (23).
Taken together, these previously discussed observations suggested that the frequency of virus-specific CD8+ T cell present in the lungs early in infection (i.e., days 5–7) may be underestimated by tetramer staining and/or ICCS, and that granzyme B expression may serve as a surrogate marker for the detection and enumeration of activated specific CD8+ T cells at early time points during infection. To examine these possibilities, we studied the kinetics of granzyme B expression by CD8+ T cells within draining MLN and lungs after i.n. influenza virus infection using a flow cytometry-based assay (26, 27). As Fig. 5 demonstrates, granzyme B+CD8+ T cells were identified at a detectable frequency on day 3 postinfection. The percentage of granzyme B+ T cells in the MLN increased 10-fold (3%) by day 4, doubled in frequency over the ensuing 24 h (to 6%), and was maintained at this level up to day 9 after infection.
FIGURE 5. High frequencies of granzyme B-expressing CD8+ T cells are present in the draining lymph nodes and lungs during influenza virus infection. The lungs and MLN were harvested from BALB/c mice at multiple days after Japan/57 virus infection. The presence of granzyme B was detected by intracellular staining with specific Ab. The percentage of granzyme B expression by CD8+ gated T cells is shown on the left y-axis (). The absolute numbers of granzyme B+CD8+ T cells are shown on the right y-axis ().
Analysis of granzyme B+CD8+ T cells infiltrating the lungs (Fig. 5) was noteworthy in several respects. First, there was minimal accumulation of granzyme B+ T cells in the infected lungs up to day 4 of infection. Thus, despite the high lung virus titer over this period, there was neither significant nonspecific (inflammation induced) induction of granzyme B expression by lung-resident CD8+ T cells nor significant recruitment of specific (or Ag-nonspecific) CD8+ T cells to the infected lungs with an activated granzyme B+ phenotype. Second, by day 5 of infection, >30% of lung-resident CD8+ T cells were granzyme B+. The tempo of granzyme B+CD8+ T cell accumulation in the lungs directly paralleled our recent observations (22) on the time course of activated granzyme B+ T cell migration from the draining lymph nodes to the infected lungs in a TCR transgenic CD8+ T cell adoptive transfer model. More importantly, this frequency of granzyme B+ T cells in the lungs was well above the predicted total frequency of epitope-specific CD8+ T cells detected by ICCS/tetramer staining, i.e., <10% (Fig. 1A), and suggested that early in the CD8+ T cell response to infection in the lungs, the frequency of virus-specific CD8+ T cells may be substantially higher than that detected by the ICCS/tetramer labeling methods. Third, by day 6 postinfection, when pulmonary virus titers began to drop precipitously (Fig. 4), >70% of lung CD8+ T cells were granzyme B+. This T cell frequency was two to three times higher than the cumulative frequency of Japan/57 epitope-specific cells determined by ICCS and/or tetramer staining, again suggesting that the frequency of specific T cells infiltrating the lungs was substantially higher than that estimated by recognition of specific epitopes. Furthermore, cell surface marker analyses of the day 6 CD8+ T cells demonstrated that at least 60% of these cells expressed activation marker phenotypes (CD62L–, CD69+, LFA-1high, and VLA-4high) consistent with recent activation (data not shown). The high frequency of granzyme B+CD8+ T cells (>70%) in the infected lungs was also sustained up to day 9. At this point in time, the frequency (percentage) of granzyme B+CD8+ T cells closely approximated the frequency estimates of virus-specific CD8+ T cells in the lungs, defined in the ICCS assay using specific peptides or infected P815 cells to elicit IFN- synthesis (Fig. 3).
To further explore the Ag responsiveness of the lung-infiltrating granzyme B+CD8+ T cells during the early phase of the adaptive response (i.e., day 6 postinfection), we evaluated the expression (up-regulation) of CD107a (liposome-associated membrane proteins-1a) and CD69 on that cell population after short term in vitro stimulation with a synthetic peptide mixture corresponding to the four A/JAPAN/57 epitopes (see Materials and Methods). We observed that only a small percentage (<10%) of the lung-infiltrating granzyme B+CD8+ T cells up-regulated CD107a expression in response to antigenic stimulation. This finding is consistent with the low frequency of day 6 lung CD8+ T cells responding to peptide stimulation in the ICCS assay. By contrast, >95% of the lung-infiltrating granzyme B+CD8+ T cells up-regulated cell surface CD69 expression in response to in vitro stimulation with the viral peptide mixture (data not shown). The latter finding suggests that these lung-infiltrating, granzyme B+CD8+ T cells, which exhibit diminished tetramer staining and responsiveness in effector assays such as ICCS and CD107a up-regulation (8), represent a population of virus-specific T cells that are responsive to TCR engagement and specific antigenic stimulation.
Proliferation of lung-resident CD8+ T cells
If, as we believe, granzyme B expression is a useful marker for the activated Ag-specific responding CD8+ T cells, then substantial numbers of specific CD8+ T cells should have accumulated in the draining MLN on days 5–6 of infection. Although we did detect a significant expansion of granzyme B+CD8+ T cells in the draining MLN before (day 4) and during (days 5–6) CD8+ T cell accumulation in the lungs, we could not readily account for the elevated numbers of T cells in the lungs based on our calculations of total MLN-resident granzyme B+CD8+ T cells (Fig. 5). One potential explanation is that specific CD8+ T cells were activated in and recruited from other lymphoid tissues (e.g., the spleen) to the lungs. For reasons discussed later (see Discussion) we do not believe that this is the case. Alternatively, either the frequency of granzyme B+ T cells underestimated the true frequency of proliferating CD8+ T cells in the draining MLN, or there was significant proliferation of the responding virus-specific CD8+ T cells in the infected lungs.
To directly estimate the frequency of proliferating CD8+ T cells responding to i.n. infection, we used the strategy of in vivo uptake of the thymidine analog BrdU with a flow cytometry-based analysis to identify BrdU+ proliferating T cells. In a previous report (23), up to 80% of CD8+ T cells infiltrating the lung airways were BrdU+ on day 8 of infection when a continuous in vivo labeling method was used to supply BrdU in the drinking water throughout infection. Our initial attempts to continuously label responding cells by adding BrdU to the drinking water of infected animals were unsuccessful, because mice became moribund shortly after i.n. infection and exhibited a marked decrease in water consumption. To circumvent this problem, we used the strategy of i.p. injection of BrdU at different time points after virus infection and harvested the MLN and infected lungs 24 h later to enumerate BrdU+CD8+ T cells in these sites. This strategy would also allow us to identify the approximate time point of maximum T cell proliferation in the MLN.
As Fig. 6A demonstrates, proliferating (BrdU+) CD8+ T cells were not detectable in the draining MLN at a significant frequency before the 24-h interval between days 3–4 of infection. The rate of CD8+ T cell proliferation reached a maximum in the MLN during the day 5–6 interval, and proliferating T cells were detected at a similar frequency through day 8 before decreasing in frequency later in infection. The frequency of BrdU+ T cells in the draining MLN at the peak of proliferation (12% on day 6) was slightly higher than the corresponding frequency of granzyme B+ cells in the lymph nodes at the same time period (7–8%; Fig. 5). We anticipated that the kinetics of BrdU+ T cell accumulation in the lungs might be slightly delayed compared with those in the MLN, and that the frequency of BrdU+ T cells within the infected lungs might be higher during the day 6–8 interval, reflecting the migration of activated/proliferating CD8+ T cells from the MLN (or other lymphoid tissues) to the lungs over the 24-h labeling interval. As demonstrated in Fig. 6A, this was not the case. BrdU+CD8+ T cells were detected in the lungs at an overall frequency comparable to that in the draining MLN.
FIGURE 6. BrdU incorporation by proliferating CD8+ T cells in MLN and lungs during influenza virus infection. BrdU was either injected i.p. (A) or administered via the i.n. route (B) to detect cells that had synthesized DNA in the previous 24 h. The percentage of BrdU expression is shown for CD8+ gated T cells at the indicated number of days postinfection.
The kinetics of BrdU+ T cell accumulation in the lungs, as determined by 24-h pulse BrdU labeling, was consistent with the concept that activated virus-specific CD8+ T cells proliferated primarily, if not exclusively, in lymphoid tissues (e.g., draining MLN) before migration to the infected lungs. However, the frequency of BrdU+CD8+ T cells in the lungs on days 5–6 (12%) was well below the frequency of granzyme B+ T cells (30% on day 5 and 70% on day 6) detected in the lungs at this time (Fig. 5). Although the division time of activated CD8+ T cells has been reported to be as short as 6 h (28), it is possible that this discrepancy in the frequency of specific CD8+ T cells simply reflected our inability to accurately assess the extent of T cell proliferation in the MLN using the relatively short 24-h interval of BrdU labeling in vivo. An alternative explanation was suggested by the possibility that during the inflammatory response in a site such as the lungs (e.g., during pulmonary influenza virus infection), there is a local accumulation of nucleotide precursors resulting from DNA degradation and the release of precursor pools locally from dead and dying cells (29). This effect could result in dilution of the thymidine analog available for uptake by proliferating CD8+ T cells in the lungs and a decrease in the efficiency of BrdU labeling in the lungs.
We decided to explore this possibility by locally introducing BrdU in the respiratory tract by i.n. administration during virus infection at 24-h intervals before harvesting lungs and MLN for analysis of BrdU uptake in a manner analogous to the i.p. BrdU administration. We anticipated that this i.n. BrdU labeling strategy might result in a decreased efficiency of labeling of T cells in the draining MLN. As shown in Fig. 6B, this was indeed the case. The frequency of BrdU+ cells in the MLN was lower at any time point sampled, but the overall kinetics of BrdU labeling in the lymph nodes after i.n. BrdU administration were similar to the time course determined by i.p. BrdU administration (Fig. 6). In marked contrast, i.n. BrdU administration revealed a substantially higher frequency of BrdU+ T cells in the infected lungs, particularly during the 24-h interval between days 6–7 of infection when 30% of the lung-infiltrating CD8+ T cells were BrdU+. This finding suggests that in addition to undergoing proliferative expansion in the draining lymph nodes, activated T cells may continue to undergo one or more divisions upon entering infected lungs.
Discussion
In this report we examined the contribution of a set of defined MHC class I-restricted CD8+ T cell epitopes to the overall pulmonary CD8+ T cell response during experimental murine type A influenza virus infection (30). We used the complementary techniques of MHC class I tetramer labeling and Ag-dependent cytokine synthesis in the ICCS assay to identify epitope-specific T cells based on structural (TCR binding) and functional criteria. Using this approach, we were able to define the tempo and magnitude of the pulmonary CD8+ T cell response during primary infection and demonstrate that >90% of the total influenza virus-specific pulmonary CD8+ T cell response at its peak (i.e., days 9–11) was directed to four defined epitopes. Based on this information, we went on to examine the early CD8+ T cell response in the draining lymph nodes and lungs using the expression of granzyme B by activated CD8+ T cells as a surrogate marker for specific T cell activation. We observed that early in the primary CD8+ T cell response to infection (i.e., days 4–7), the conventional tetramer/intracellular cytokine methods for enumerating specific T cells appeared to underestimate the frequency of specific CD8+ T cells in the lungs. In addition, we obtained evidence for proliferation of activated CD8+ T cells in the lungs during this early phase of infection by using intranasal administration of the thymidine analog BrdU.
The cumulative frequency of influenza virus epitope-specific CD8+ T cells detected in our study is higher than that previously reported for a primary pulmonary influenza virus infection (3, 4, 25), most likely because the four epitopes represent the vast majority of specific CD8+ T cells directed to this virus strain in H-2d haplotype mice (Fig. 3). However, these frequencies are comparable to the high frequencies of virus-specific CD8+ T cells detectable in the lungs during other respiratory virus infections (6, 7). The kinetics of specific CD8+ T cell appearance, peak accumulation, and contraction in this report also closely parallel previously reported observations of the primary pulmonary CD8+ T cell response to influenza virus infection using different mouse and virus strains (3, 4, 5). This suggests that our findings are not unique to a particular virus/mouse strain combination.
CD8+ T cells have been documented to play an important role in virus clearance during primary influenza virus infection (30). Consistent with previous reports (3, 25, 31), we observed that influenza virus-specific CD8+ T cells were first detected in the BAL and lung parenchyma on days 5–6 postinfection by cytokine synthesis and/or tetramer binding (Fig. 1). This corresponded to the same time frame as the onset of virus clearance from the infected lungs (Fig. 4). However, the drop in lung virus titer was much more dramatic on days 5–7 of infection than might be expected from the relatively low frequency of CD8+ T cells detected in lungs at this time. Furthermore, both the frequency and absolute number of virus-specific lung CD8+ T cells on days 5–7 (as determined by cytokine synthesis/tetramer binding) represented only a small fraction of the specific T cells detected at the peak of the response 4–5 days later when infectious virus was no longer detectable in the lungs. This discrepancy between virus clearance and CD8+ T cell frequency early in infection raised the possibility that techniques to evaluate cytokine synthesis/TCR binding underestimated the actual frequency/number of influenza virus-specific CD8+ T cells in the lungs. This possibility was reinforced by our recent findings in a TCR transgenic model of in vivo CD8+ T cell responses to pulmonary influenza virus infection.4 Using this model, we demonstrated that shortly (1–2 days) after in vivo activation, naive influenza virus-specific CD8+ T cells transiently lost the capacity to bind specific MHC class I tetramers and exhibited diminished cytokine (IFN-) synthesis in the ICCS assay. This loss of tetramer binding/cytokine synthesis was the most pronounced on days 3–4 of infection in the draining lymph nodes and on days 4–6 in the infected lungs, with full restoration of tetramer binding/cytokine synthesis by days 8–10 postinfection.4
Based on these considerations, we explored the possibility of using the expression of the cytolytic effector molecule granzyme B as a surrogate marker to identify virus-specific activated CD8+ T cells responding in the lungs to infection. Granzyme B is not expressed by naive T cells, but is produced and stored in intracellular granules after CD8+ T cell activation (27, 32). In this connection, we have recently reported (23) that granzyme B expression is rapidly induced in the majority of activated virus-specific CD8+ T cells in response to respiratory influenza virus infection shortly after the onset of cell division in draining lymph nodes. Granzyme B expression is also sustained at high levels in lung-infiltrating effector CD8+ T cells after migration from lymph nodes to infected lungs. The frequency of virus-specific CD8+ T cells responding in the draining lymph nodes early (days 5–6) after pulmonary infection has been reported to be extremely low and at the limits of detection by tetramer staining (our unpublished observations) (4, 24, 25). However, we were able to reproducibly detect granzyme B+CD8+ T cells as early as day 3 of infection in the draining MLN, with a maximum frequency of granzyme B+CD8+ T cells of 7% in the draining lymph nodes 3–4 days later.
The companion analysis of CD8+ T cells in the infected lungs revealed that >30% of lung-infiltrating CD8+ T cells expressed granzyme B on day 5 of infection, and >70% of the CD8+ T cells in the lungs were granzyme B+ on day 6 postinfection. This frequency was 3 times higher than that of virus epitope-specific CD8+ T cells detected by tetramer/cytokine at the corresponding day 6 point. It is also noteworthy that granzyme B+CD8+ T cells were detected at a similar frequency (70%) on day 9 of infection when the overall frequency of influenza-specific CD8+ T cells in infected lungs was determined by ICCS using infected P815 cells to likewise correspond to 70%. These findings support the view that effector CD8+ T cells accumulate in the lungs at significant frequencies and in substantial numbers early in the course of infection, i.e., days 5–6, when substantial virus clearance occurs. However, because the majority of these lung-infiltrating granzyme B+CD8+ T cells present early (i.e., days 5–6) in the host response do not bind specific tetramer and do not respond in either the ICCS or CD107a mobilization assay, we cannot directly document that these T cells are influenza specific.
Because granzyme B expression merely marks CD8+ T cells as having been previously activated (and therefore not necessarily virus specific), it is possible that the high frequency of granzyme B+CD8+ T cells infiltrating infected lungs early after infection (days 5–6) could reflect nonspecific recruitment of irrelevant CD8+ T cells to a site of inflammation. Indeed, Ag-nonspecific memory CD8+ T cells have been shown to migrate to the lungs after respiratory infection with RSV, influenza virus, and Sendai virus (33, 34, 35). However, nonspecific recruitment of irrelevant memory CD8+ T cells to the inflamed lungs is an unlikely explanation for our findings. This phenomenon has been reported to peak early after infection, i.e., days 4–5 (33, 36) and to not occur at the sustained high frequency exhibited by the lung-infiltrating granzyme B+ cells demonstrated in this study. Furthermore, the level of granzyme B protein expression in memory cells has been previously demonstrated to be very low (26, 27, 36), whereas we detected a uniformly high level of granzyme B in the CD8+ T cells infiltrating the lungs on days 5–6 after infection (data not shown). In this regard, it is noteworthy that differential granzyme B gene expression has been previously used in a reporter-based transgenic strategy to distinguish effector T cells from memory T cells in vivo (37). Finally, the extremely close correspondence between the frequency of granzyme B+CD8+ T cells and the predicted frequency of influenza virus-specific CD8+ T cells on day 9 of infection supports the view that granzyme B may serve as a useful marker in identifying recently activated Ag-specific CD8+ T cells throughout the course of infection.
Our analysis of the frequency and total number of granzyme B+CD8+ T cells in draining MLN and infected lungs suggested that the observed numbers of granzyme B+ cells detected in the lungs early in infection could not readily be accounted for by the accumulation of migrant CD8+ T cells derived from the draining lymph nodes. We believe this is unlikely to be due to the recruitment of activated virus-specific T cells from other secondary lymphoid organs (e.g., spleen or nondraining lymph nodes). The reasons for this are 2-fold. First, during influenza virus infection, Ag-bearing respiratory dendritic cells were found to preferentially, if not exclusively, migrate to the lung-draining lymph nodes (38). Second, in a recent analysis of CD8+ T cell activation in vivo in response to influenza infection, initial/early CD8+ T cell activation was restricted to those cells present in draining MLN (23).
We therefore investigated the possibility that after migration from draining lymph nodes, activated CD8+ T cells proliferated in the lungs and thereby contributed to the elevated T cell numbers detected. We were able to demonstrate proliferation of activated CD8+ T cells directly in situ in the infected lungs by i.n. administration of the thymidine analog BrdU. Delivery of BrdU by the i.n. route proved superior to i.p. BrdU administration in the detection of proliferating CD8+ T cells in the lungs. It is unlikely that the efficient labeling of lung CD8+ T cells with BrdU after i.n. administration was due to transport of the nucleoside analog to the draining lymph nodes and labeling of proliferating T cells there before their migration to the infected lungs. In fact, the efficiency of BrdU uptake by CD8+ T cells in the draining lymph nodes after i.n. administration was lower than the efficiency of labeling after i.p. administration. We believe that the high local concentration of DNA nucleotide/nucleoside at sites of inflammation and/or associated cell death (39), such as influenza virus-infected lungs, results in dilution of BrdU after parenteral administration and short term (24-h) in vivo labeling. Because in vivo BrdU uptake by T cells after i.n. administration was monitored at 24-h intervals, and transfer of BrdU from the respiratory tract to the draining MLN was readily detected in our analysis, it remains a formal possibility that the high frequency of BrdU+CD8+ T cells detected in the lung resulted from uptake of a label by cells at a distant sites (e.g., the secondary lymphoid organs). As yet, we cannot rigorously exclude this possibility. However, the finding that i.p. BrdU administration does not result in the preferential accumulation of BrdU+CD8+ T cells in the infected lungs does not readily support this view.
In conclusion, we have characterized the CD8+ T cell responses to a series of defined virus-specific CD8+ T cell epitopes. These sites account for >90% of the overall CD8+ T cell response directed against influenza virus in the lungs. With this information available, we were able to explore the use of a surrogate marker, i.e., granzyme B expression, to demonstrate that this protein may be useful in identifying specifically activated effector CD8+ T cells. By using granzyme B expression, we demonstrated that virus-specific CD8+ T cells may accumulate in the lungs at higher frequencies earlier in the course of the antiviral CD8+ T cell response than predicted by the use of conventional tetramer/intracellular cytokine methods. Finally, we provided evidence for the first time that activated effector CD8+ T cells may continue to undergo one or more rounds of proliferation in situ in the infected lungs after migrating from the draining lymph nodes. Thus, in situ CD8+ T cell proliferation at a site of infection/inflammation such as the lungs as well as the proliferative expansion of naive CD8+ T cells in secondary lymphoid organs such as the draining MLN contribute to the overall magnitude of the CD8+ T cell response during pulmonary influenza virus infection.
Acknowledgments
We thank Dr. Donald R. Drake III for advice and assistance with the design of experiments.
Disclosures
The authors have no financial conflict of interest.
Footnotes
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
1 This work was supported by U.S. Public Health Service Grants AI15608, HL33391, and HL71875 and U.S. Public Health Service Training Grant AI07496.
2 Address correspondence and reprint requests to Dr. Thomas J. Braciale, Beirne B. Carter Center for Immunology Research, University of Virginia Health Sciences Center, MR-4 Building, HSC Box 801386, Charlottesville, VA 22908. E-mail address: tjb2r@virginia.edu
3 Abbreviations used in this paper: i.n., intranasal; BAL, bronchoalveolar lavage; HA, hemagglutinin; ICCS, intracellular cytokine staining; MLN, mediastinal lymph node; NP, nucleoprotein; TCID50, half-maximal tissue culture infectious dose.
4 D. R. Drake, R. M. Ream, C. W. Lawrence, and T. J. Braciale. Transient loss of MHC class I tetramer binding after CD8+ T cell activation reflects T cell effector function. Submitted for publication.
Received for publication October 13, 2004. Accepted for publication February 18, 2005.
References
Chen, W., L. C. Anton, J. R. Bennink, J. W. Yewdell. 2000. Dissecting the multifactorial causes of immunodominance in class I-restricted T cell responses to viruses. Immunity 12:83.
Chen, W., J. R. Bennink, P. A. Morton, J. W. Yewdell. 2002. Mice deficient in perforin, CD4+ T cells, or CD28-mediated signaling maintain the typical immunodominance hierarchies of CD8+ T-cell responses to influenza virus. J. Virol. 76:10332.
Flynn, K. J., G. T. Belz, J. D. Altman, R. Ahmed, D. L. Woodland, P. C. Doherty. 1998. Virus-specific CD8+ T cells in primary and secondary influenza pneumonia. Immunity 8:683.
Belz, G. T., W. Xie, J. D. Altman, P. C. Doherty. 2000. A previously unrecognized H-2Db-restricted peptide prominent in the primary influenza A virus-specific CD8+ T-cell response is much less apparent following secondary challenge. J. Virol. 74:3486.
Belz, G. T., W. Xie, P. C. Doherty. 2001. Diversity of epitope and cytokine profiles for primary and secondary influenza a virus-specific CD8+ T cell responses. J. Immunol. 166:4627.
Usherwood, E. J., R. J. Hogan, G. Crowther, S. L. Surman, T. L. Hogg, J. D. Altman, D. L. Woodland. 1999. Functionally heterogeneous CD8+ T-cell memory is induced by Sendai virus infection of mice. J. Virol. 73:7278.
Chang, J., A. Srikiatkhachorn, T. J. Braciale. 2001. Visualization and characterization of respiratory syncytial virus F-specific CD8+ T cells during experimental virus infection. J. Immunol. 167:4254.
Butz, E. A., M. J. Bevan. 1998. Massive expansion of antigen-specific CD8+ T cells during an acute virus infection. Immunity 8:167
Murali-Krishna, K., J. D. Altman, M. Suresh, D. J. Sourdive, A. J. Zajac, J. D. Miller, J. Slansky, R. Ahmed. 1998. Counting antigen-specific CD8 T cells: a reevaluation of bystander activation during viral infection. Immunity 8:177.
Sweetser, M. T., V. L. Braciale, T. J. Braciale. 1989. Class I major histocompatibility complex-restricted T lymphocyte recognition of the influenza hemagglutinin: overlap between class I cytotoxic T lymphocytes and antibody sites. J. Exp. Med. 170:1357.
Hahn, Y. S., C. S. Hahn, V. L. Braciale, T. J. Braciale, C. M. Rice. 1992. CD8+ T cell recognition of an endogenously processed epitope is regulated primarily by residues within the epitope. J. Exp. Med. 176:1335.
Milligan, G. N., L. A. Morrison, J. Gorka, V. L. Braciale, T. J. Braciale. 1990. The recognition of a viral antigenic moiety by class I MHC-restricted cytolytic T lymphocytes is limited by the availability of the endogenously processed antigen. J. Immunol. 145:3188.
Bodmer, H. C., R. M. Pemberton, J. B. Rothbard, B. A. Askonas. 1988. Enhanced recognition of a modified peptide antigen by cytotoxic T cells specific for influenza nucleoprotein. Cell 52:253.
Braciale, T. J., M. T. Sweetser, L. A. Morrison, D. J. Kittlesen, V. L. Braciale. 1989. Class I major histocompatibility complex-restricted cytolytic T lymphocytes recognize a limited number of sites on the influenza hemagglutinin. Proc. Natl. Acad. Sci. USA 86:277.
Cao, W., B. A. Myers-Powell, T. J. Braciale. 1996. The weak CD8+ CTL response to an influenza hemagglutinin epitope reflects limited T cell availability. J. Immunol. 157:505.
Betts, M. R., J. M. Brenchley, D. A. Price, S. C. De Rosa, D. C. Douek, M. Roederer, R. A. Koup. 2003. Sensitive and viable identification of antigen-specific CD8+ T cells by a flow cytometric assay for degranulation. J. Immunol. Methods 281:65.
Altman, J. D., P. A. Moss, P. J. Goulder, D. H. Barouch, M. G. McHeyzer-Williams, J. I. Bell, A. J. McMichael, M. M. Davis. 1996. Phenotypic analysis of antigen-specific T lymphocytes. Science 274:94.
Belz, G. T., D. Wodarz, G. Diaz, M. A. Nowak, P. C. Doherty. 2002. Compromised influenza virus-specific CD8+-T-cell memory in CD4+-T-cell-deficient mice. J. Virol. 76:12388.
Eichelberger, M., W. Allan, M. Zijlstra, R. Jaenisch, P. C. Doherty. 1991. Clearance of influenza virus respiratory infection in mice lacking class I major histocompatibility complex-restricted CD8+ T cells. J. Exp. Med. 174:875.
Turner, S. J., R. Cross, W. Xie, P. C. Doherty. 2001. Concurrent naive and memory CD8+ T cell responses to an influenza A virus. J. Immunol. 167:2753.(Christopher W. Lawrence, )