Quantitative Analysis of Long-Term Virus-Specific CD8⁺-T-Cell Memory in Mice Challenged with Unrelated Pathogens

Mice and viral infection.Female, 8-week-old C57BL/6J (B6) mice (The Jackson Laboratory, Bar Harbor, Maine) were primed intraperitoneally with 10^7.9 50% egg infectious doses of the A/PR8/34 (PR8, H1N1) influenza A virus and then challenged intranasally with 10^6.5 50% egg infectious doses of the HKx31 (H3N2) virus (13) and later given 10⁴ PFU of γHV68 intranasally (19). In a further study, naïve B6 mice were infected intranasally with γHV68, boosted intraperitoneally with 5 × 10⁷ PFU of a recombinant vaccinia virus (Vacc-p56) incorporating the AGPHNDMEI p56 peptide (25), and then challenged intranasally with the HKx31 virus. The interval between sequential infections was 6 weeks in every case. In all experiments, naïve, age-matched controls were challenged with the final virus used.

Tissue sampling.The mice were anesthetized with Avertin (2,2,2-tribromoethanol; Sigma-Aldrich, St. Louis, Mo.) and bled from the right axillar artery (17). The blood was collected into heparinized phosphate-buffered saline (1,000 U/ml; Elkins-Sinn Inc., Cherry Hill, N.J.), and the inflammatory population was recovered from the pneumonic lung by bronchoalveolar lavage. Each mouse was then perfused with 40 to 60 ml of heparinized phosphate-buffered saline, and the lungs, mediastinal lymph nodes, cervical lymph nodes, livers, and spleens were removed. The bone marrow was flushed from the tibias and femurs with 5 to 10 ml of RP-10 medium (RPMI 1640 with 10% fetal calf serum), while the peripheral blood lymphocytes were separated on a 1-Step gradient (Accurate Chemicals, Westbury, N.Y.) by centrifugation for 30 min at 2,000 rpm. The cells were collected from the interface and washed, and the erythrocytes were lysed with ammonium chloride. The bronchoalveolar lavage populations were allowed to adhere to plastic for 1 h at 37°C to remove macrophages and monocytes.

Single-cell suspensions were made from the mediastinal lymph node, cervical lymph node, and spleen by grinding the tissues gently between glass slides. Lungs and livers were minced, forced through a fine stainless-steel mesh, filtered, and digested with collagenase (17). The bronchoalveolar lavage, peripheral blood lymphocyte, and mediastinal lymph node populations were pooled from two mice, while the spleens, lungs, livers, bone marrow, and cervical lymph nodes were analyzed from individual mice.

Flow cytometry.Spleen, cervical lymph node, and mediastinal lymph node populations were enriched for the CD8⁺ set by in vitro depletion with monoclonal antibodies to I-A^b (M5/114.15.2) and CD4 (GK1.5), followed by sheep anti-rat immunoglobulin and sheep anti-mouse immunoglobulin-coated magnetic beads (Dynal, Oslo, Norway). Virus-specific CD8⁺ T cells were stained with phycoerythrin-labeled tetrameric complexes (13, 25) of H-2D^b plus the influenza virus nucleoprotein ASNENMETM (D^bNP₃₆₆), the γHV68 p56 AGPHNDMEI peptide (D^bp56), or H-2K^b plus the γHV68 p79 TSINFKVI peptide (K^bp79) at room temperature, followed by anti-CD8α-Tricolor (Caltag, South San Francisco, Calif.). Unenriched cells were also stained with anti-CD8α-fluorescein isothiocyanate and anti-CD4-phycoerythrin to determine the overall T-cell phenotype in the particular sample. The cells were stained and washed in ice-cold phosphate-buffered saline containing bovine serum albumin (0.1%) and azide (0.01%) and analyzed on a FACScan or FACScalibur with CellQuest software (Becton Dickinson, Mountain View, Calif.). The numbers presented in some of the figures were calculated from the total cell counts, the percent CD8⁺, and the percent CD8⁺ tetramer⁺.

Mathematical modeling.The two basic variables in the mathematical model are the virus populations and memory CTL specific for each virus. It is assumed that a host can become infected with n different viruses. Virus of type i is denoted by v_i, and the memory CTL response specific to that virus is denoted by z_i, where i = 1, 2, . . ., n. The model is given by the following set of differential equations: $$mathtex$$\[\frac{dv_{i}}{dt}{=}r_{i}v_{i}\left(1{-}\frac{v_{i}}{k_{i}}\right){-}p_{i}v_{i}z_{i}\]$$mathtex$$(1)$$mathtex$$\[\frac{dz_{i}}{dt}{=}c_{i}v_{i}z_{i}{-}b_{i}z_{i}{-}z_{i}{{\sum}_{j{\neq}i}^{j{=}1,2,{\ldots},n}}{\alpha}_{j}v_{j}\]$$mathtex$$(2)

The virus replicates at a rate r_i. Replication is limited by the availability of susceptible host cells, which is captured in the parameter k_i. Virus replication is inhibited by the specific CTL response at a rate p_i. The memory CTL response becomes activated and proliferates in response to its specific antigen at a rate c_i. In the absence of antigen, the CTL response has an average life span of 1/b_i. In addition, the model assumes that heterologous antigenic stimuli, v_j, reduce the memory CTL response to virus i (z_i) at a rate a_j. Thus, the model assumes the following mechanism for the reduction of memory CTL upon heterologous infection: heterologous antigen induces activation of the memory CTL through low-level cross-reactivity or bystander effects. Since the activation is induced by noncognate stimulation, these memory CTL do not go on to expand and increase significantly, but die (22, 31). Hence, the presence of the heterologous antigen is the force driving the reduction of CTL memory. An alternative idea would be that the CTL response to the heterologous infection competes with, and reduces, previously established memory. Such an assumption would, however, lead to the prediction that this new CTL response would also be selectively reduced if the existing memory pool is large.

When plotting the results of the model, it was assumed for simplicity that all infections have identical parameters. The qualitative outcome of the model does not, however, depend on this simplification. For a mathematically detailed analysis of the equations, see reference 32. The use of parameter values specific for the infections in question will require further experiments which would allow us to estimate the relevant parameters (presently unknown).

Challenging influenza virus-primed mice with γHV68.The C57BL/6J (B6) mice were first primed and boosted (13) with two different influenza A viruses (H3N2 and then H1N1) to give large numbers of CD8⁺ D^bNP₃₆₆⁺ T cells, rested for 6 weeks, and then challenged intranasally with γHV68 (6). The influenza viruses cause a transient, lytic infection that is substantially confined to the superficial epithelium of the trachea and lung and is generally resolved within 8 to 12 days (1, 8). There is no cross-neutralization (with antibody) between the H1N1 and H3N2 viruses, though both carry the NP_366-374 peptide that dominates the CD8⁺-T-cell response (15). Respiratory exposure to γHV68 also induces a transient, lytic infection in lung epithelial cells (21, 23). However, this virus establishes latency in B cells and macrophages, and there is indirect evidence of sporadic reactivation to lytic phase for months after the initial encounter with the virus (3, 12, 34). In this respect it resembles other gammaherpesviruses, like Epstein-Barr virus (EBV), which can be detected routinely by PCR in the oropharynx of >30% of the human population (16).

Groups of at least five γHV68-challenged, H3N2- and H1N1-primed B6 mice (together with the appropriate controls) were sampled at 35, 60, and 100 days after the γHV68 challenge. At every time point tested, the CD8⁺ D^bNP₃₆₆⁺ T-cell frequencies were significantly reduced for the spleen, peripheral blood lymphocytes, and bone marrow recovered from those that had also been exposed to γHV68 (Fig. 1). This was also true for the lymphocytes that were isolated from the perfused, digested pneumonic lung (17) and for the inflammatory population that had first been removed by bronchoalveolar lavage. This pattern held up for the spleen, lung, and bone marrow when we calculated the numbers of CD8⁺ D^bNP₃₆₆⁺ T cells in each site (Fig. 2). Though the difference was not apparent for the draining mediastinal lymph node, cervical lymph node, bronchoalveolar lavage, and liver populations from those sampled on day 100 (Fig. 2), the total CD8⁺ D^bNP₃₆₆⁺ T-cell counts per mouse were approximately halved in the γHV68-infected, influenza virus-primed mice (Fig. 3A and B). The effect was selective, as estimates of the total CD8⁺ lymphocyte count showed that there was no significant difference between the groups of influenza virus-immune mice that were and were not given γHV68 (Fig. 3D and E).

• Open in new tab
• Download powerpoint

FIG. 1.

Changes in the distribution of CD8⁺ D^bNP₃₆₆⁺ T cells at days 35, 60, and 100 after intranasal challenge of influenza virus-primed mice (PR8-X31) with murine γHV68 (PR8-X31-γHV). B6 mice were primed intraperitoneally with the PR8 influenza A virus (H1N1), boosted intranasally with the HKx31 (H3N2) virus, and then challenged intranasally with γHV68 (PR8-X31-γHV) or left unchallenged (PR8-X31) as described in the text. The interval between each virus treatment was 6 weeks, and the 35-, 60-, and 100-day time points are after the final γHV68 challenge. The bronchoalveolar lavage (BAL), peripheral blood lymphocyte (PBL), and mediastinal lymph node (MLN) populations from day 100 from two mice were pooled, and each data point shows the mean ± standard deviation for four samples (eight mice), while the spleens, lungs, livers, bone marrow (BM), and cervical lymph node (CLN) were analyzed from individual mice and each data point shows mean ± standard deviation for five mice. Otherwise, the results from day 35 and day 60 that show error bars are for five individuals, while the absence of error bars (day 35) indicates that the samples were pooled. The asterisk designates groups that were significantly different (P < 0.05).

• Open in new tab
• Download powerpoint

FIG. 2.

Numbers of D^bNP₃₆₆⁺ CD8⁺ T cells per organ were calculated from the total CD8⁺-T-cell counts and the values shown in Fig. 1. The asterisk designates groups that were significantly different (P < 0.05). See Fig. 1 legend for abbreviations.

• Open in new tab
• Download powerpoint

FIG. 3.

Total counts (per mouse) for CD8⁺ K^bp79⁺, CD8⁺ D^bNP₃₆₆⁺ (A and B), CD8⁺ K^bp79⁺CD8⁺ (C), and all CD8⁺ T cells (panels D and E) were calculated for the day 100 (after γHV68) time point for the experiments shown in Fig. 1 to 4. The counts shown here were derived by summing the numbers per mouse for all the organs sampled. The spleens, lungs, livers, bone marrow, and cervical lymph node were analyzed from mice 1 to 5 of the eight sampled. The bronchoalveolar lavage, peripheral blood lymphocyte, and mediastinal lymph node samples were pooled for two mice (1 and 2, 3 and 4, 5 and 6, and 7 and 8). The numbers for the first three pools were halved and then added to the data for the other organs. The asterisk designates groups that were significantly different (P < 0.04).

We also asked whether the presence of large, influenza virus-specific memory T-cell populations at the time of γHV68 challenge would modify the γHV68-specific response. Though the γHV68-specific CD8⁺ D^bp56⁺ set expands first after intranasal exposure to γHV68, this quickly gives way to CD8⁺ K^bp79⁺ T cells, which predominate in the long term (24). The only significant difference in CD8⁺ K^bp79⁺ T-cell numbers was found for the liver on day 100, when the counts were decreased for the mice that had first been exposed to the influenza A viruses (Fig. 4). Calculating the totals (per mouse) for CD8⁺ K^bp79⁺ T cells also showed that the value overall was significantly lower for the mice that had first been primed and boosted with the influenza A viruses (Fig. 3C). However, unlike the effect for the CD8⁺ D^bNP₃₆₆⁺ T cells (Fig. 3A, B, C, and D), this difference (Fig. 3C) was not specific for the CD8⁺ K^bp79⁺ set, as the total counts for all CD8⁺ T cells were significantly lower in the mice that were given both influenza virus and γHV68 (Fig. 3D).

• Open in new tab
• Download powerpoint

FIG. 4.

Numbers of CD8⁺ K^bp79⁺ T cells are shown for the experiment described in the legend to Fig. 1. See Fig. 1 legend for abbreviations.

Over the first 20 days or so after intranasal challenge with γHV68, the proliferation rate (measured by 6-day bromodeoxyuridine incorporation) of the CD8⁺ D^bp56⁺ and CD8⁺ K^bp79⁺ sets reached peak values of >85%, with evidence of cycling falling off more rapidly for the CD8⁺ D^bp56⁺ population (3). The CD8⁺ D^bp56⁺ T cells also turned over more slowly with elapsed time, supporting other indications that the lytic-phase K^bp79 epitope is present at higher levels in the long term. In fact, the CD8⁺ K^bp79⁺ set cycled at levels substantially above background for several months. During the acute phase of γHV68 infection (until day 16), an established CD8⁺ D^bNP₃₆₆⁺ T-cell population was also found to cycle at levels (18 to 20%) that were significantly above background (4 to 6%). The present findings establish beyond any doubt that this increased replication does not lead to augmentation of the CD8⁺ D^bNP₃₆₆⁺ memory T-cell pool and are in accord with findings from other experimental systems that the consequence of such bystander activation may be partial depletion of the memory T-cell pool (22).

Influenza virus pneumonia in mice with large numbers of CD8⁺ D^bp56⁺ T cells.Respiratory exposure to the HKx31 influenza A virus causes a substantial, nonfatal pneumonia, which generally starts to resolve within 14 days of infection (1). Memory T cells specific for unrelated antigens are readily detected in the population recovered by bronchoalveolar lavage (7, 29), and it is not clear whether such lymphocytes return to the recirculating pool or are lost to the host. Some increase in cycling is found for these “irrelevant” T cells (7).

Influenza virus challenge of mice that had been infected with γHV68 and then boosted with Vacc-p56 (4, 25) did not significantly modify the numbers of CD8⁺ D^bp56⁺ T cells recovered from the spleen 14 or 200 days later (Fig. 5A). Significantly more CD8⁺ D^bNP₃₆₆⁺ T cells were detected in the spleens of the γHV68-Vacc-p56 mice that were sampled on day 14, but the converse was true on day 200 (Fig. 5B). However, the total numbers of CD8⁺ T cells in the spleen were also significantly reduced on day 200, though not on day 14 (Fig. 5C). In the present experiments, this did not lead to any decrease in the size of the established CD8⁺ D^bp56⁺ T-cell population in the spleen, though there is the caveat that the sporadic reactivation of γHV68 to lytic phase may have been associated with some degree of further antigenic stimulation. The effect of persistent if low-level γHV68 infection on T-cell turnover rates may also have had some influence on the significantly decreased numbers of both total CD8⁺ and CD8⁺ D^bNP₃₆₆⁺ T cells on day 200 (Fig. 5).

• Open in new tab
• Download powerpoint

FIG. 5.

Effect of influenza virus infection analyzed for spleen populations from mice primed with γHV68 and boosted to expand the CD8⁺ K^bp56⁺ population. The mice were given γHV68 intranasally, boosted intraperitoneally with Vacc-p56, and then (with age-matched, naïve controls) challenged intranasally with the HKx31 virus. The interval between each treatment was 6 weeks, and the mice were sampled 14 and 200 days after the influenza virus challenge. The number of CD8⁺ D^bp56 (A), CD8⁺ D^bNP₃₆₆⁺ (B), and total CD8⁺ (C) T cells was calculated and compared (mean ± standard deviation) for groups of five mice. The asterisk designates groups that were significantly different (P < 0.05).