Main

Analysis with the influenza A virus mouse pneumonia model has established that virus-specific CD8+ CTL activity is central to the normal operation of cell-mediated immunity and optimal virus clearance from the infected lung.1 The classical in vitro correlate of such CTL function is the short-term (4–6 h) 51Cr release assay, an analytical system that is totally dependent on the capacity of the CD8+ effectors to express the effector molecule perforin (pfp).2 Other effector molecules, including the serine proteases granzyme (grz) A and B do not appear to be essential for promoting rapid cell-mediated cytotoxicity,3, 4 although grzB is required for the induction of ‘classical’ apoptosis.4 We have recently shown a possible compensatory role for the much less analysed grzK that features prominently (at least at the mRNA level) in the influenza A virus-specific CTL response.5, 6

Influenza A virus-immune CD8+ T cells specific for the prominent DbNP366 and DbPA224 epitopes mediate comparable levels of rapid 51Cr release from H2b targets pulsed with the cognate peptide,7 and show equivalent pfp and grz mRNA expression profiles when recovered from the pneumonic lung by bronchoalveolar lavage (BAL) at the peak of the inflammatory process.5 However, other qualitative differences exist. For example, DbPA224-specific CTLs showed greater T-cell receptor repertoire diversity compared with DbNP366-specific CTLs.8, 9 In addition, analysis of tetramer elution kinetics showed that the avidity of the transgenic T cells specific for the KbOVA257 epitope (TCR)/pMHCI interaction is higher for DbPA224 than for DbNP366.10, 11 Importantly, the increased avidity observed for DbPA224-specific CTLs correlated with increased cytokine production elicited by short-term, in vitro peptide stimulation. This has been of particular interest because, although the number of specific CTLs are approximately equivalent in magnitude following primary infection, the DbNP366-specific population is (at least in the numerical sense) massively overdominant following secondary challenge.12

The mode of target cell death induced by activated CTLs depends on the spectrum of pfp/grz expression.3, 4 An important implication of earlier observations is that DbPA224 and DbNP366-specific CTLs may differentially interact with target cells resulting in different modes of killing. Such differences in killing are not possible to measure using the standard in vitro 51Cr-release assay. The present experiments use in vitro time-lapse microscopy to probe the issue of TCR/epitope interaction at the level of CTL/target contact time, then the interval to completion of the lytic process for DbNP366- and DbPA224-specific effector populations. The results indicate that both DbNP366- and DbPA224-specific CTLs induce target cell death through a classical apoptosis pathway. Importantly, analysis of activated effectors recovered directly from the virus-infected lung showed that differential pMHCI/TCR avidity correlated with decreased periods of CTL contact with the target cell, although this did not impact the time, or the mode, of target cell death. Interestingly, these differences were masked for very potent, in vitro-stimulated CTL populations. An important implication of this data is that lower pMHCI/TCR avidity and shorter contact times may result in enhanced T-cell killing efficiency during infection.

Results

Characteristics of a single CTL/target interaction

First, we standardized the assay system by activating transgenic OT-I T cells that express a single TCR specific for the ovalbumin KbOVA257 epitope, by culturing them in vitro with peptide, in the presence of IL-2 (10 U/ml) for 5 days. These effector cells were then incubated with adherent MC57G (H2b) fibroblasts that had been pulsed with 1 μM of the OVA257–264 (SIINFEKL) peptide and followed the effector/target interactions by time-lapse microscopy (see Supplementary Movie 1). The analysis (Figure 1) focused on: (i) the duration of CTL/target contact, (ii) the time to target cell rounding and loss of plasma membrane asymmetry, measured by annexin V binding to exposed phosphatidylserine (AV+) and (iii) the interval from AV+ to loss of target cell plasma membrane integrity (measured by propidium iodide (PI) uptake). As illustrated in Figure 1a, motile OT-I CTL and sessile MC57G targets are readily distinguished by both their size and appearance (arrow, panel i). They remain in contact for varying times, although CTL elution does not necessarily precede the rounding-up, detachment (from plastic) and plasma membrane blebbing that signals progression to target cell death (Figure 1a, panel ii). The final stage is the total loss of membrane integrity, evidenced by PI uptake (Figure 1a, panel iii). The fluorescence intensity of both AV and PI staining relative to the maximum fluorescence yielded a mathematical measure of this characteristic progression to apoptotic cell death (Figure 1b).3, 4

Figure 1
figure 1

MC57 cells targeted by Tg OT-I CD8+ T cells die by apoptosis. (a) Cultured OT-I CD8+ T cells were overlaid on OVA257 peptide-pulsed MC57G targets, and PI and annexin V were added to the media. Images of morphology, annexin V-Fluos and PI were then obtained every 5 min (a). The elongated MC57G target cell maintained a normal morphology (a, arrow) until 100 min, then showed rounding and blebbing by 120 min. Annexin V staining (green) was evident around these cellular protrusions (a, panel ii) by 170 min, and membrane integrity was lost (red, PI+) by 220 min (a, panel iii). (b) The fluorescence profiles for annexin V binding and PI relative to maximum fluorescence are presented for this individual target. The frames (a, panels i–iii) were captured at the intervals indicated by the arrows in b. The asterisk (b) denotes the ‘duration of contact’ to CTL disengagement from the target cell. The interval from target cell rounding to PI uptake is referred to as the ‘execution phase’

Comparison of multiple interactions for in vitro-stimulated CTLs

The criterion adopted throughout these time-lapse microscopy experiments was that only one CTL effector interacted with each peptide-pulsed target. The results presented in Figure 1 trace the kinetics of a single OT-I CTL/KbOVA257+ target interaction, whereas Figure 2a summarizes the profiles for repeated observations. The duration of contact (▪) between individual KbOVA257-specific CTLs and OVA257–264-pulsed MC57G cells ranged from 6 to 60 min (mean 14.61±2.74 S.E.M.), whereas the ‘execution’ (rounding to PI+) phase (□) took 12–210 min (151.56±12.95). There was no correlation between the length of contact and the time to PI uptake (compare ▪ and □, Figure 2a).

Figure 2
figure 2

Profiles for multiple T-cell–target contact and execution times. In vitro-stimulated KbOVA257- (a), DbNP366- (b) and DbPA224- (c) specific CTL lines were added to adherent MC57G fibroblasts as described in the methods and images of morphology. Annexin V-FITC and PI-uptake were recorded every 5 min. The duration of CTL contact (▪) and the ‘execution’ time from target cell rounding to loss of membrane potential and PI uptake (□) was assessed as illustrated in Figure 1

When the analysis was repeated with in vitro-cultured, TCR-diverse, DbNP366- and DbPA224-specific CTL populations (Figure 2b and c), the profiles appeared to be broadly similar to those for the TCR-homogeneous OT-I set that had been stimulated in vitro under comparable conditions. The spectra for duration (min) of contact (DbNP366, 27.83±3.59; DbPA224; 35.25±5.57) and time from rounding to PI+ (DbNP366, 133.44±16.94; DbPA224, 136.31±21.03) also tended to reproduce the situation found for the OT-I analysis (Figure 2a). Even so, relatively more of the virus-specific CTLs remained in contact for >50 min, the values being: OT-I, 4%; DbNP366, 17% and DbPA224 25%. The greater percentage of cells that show longer duration of binding within the DbPA224-specific set may reflect that, unlike the situation for the less TCR-diverse DbNP366-specific response, this T-cell population has been shown13 previously by tetramer elution to partition on the basis of higher (‘best fit’, large clone size) or lower (‘non-canonical’ TCR) avidity.

CTL effectors recovered directly ex vivo

Our long-term analysis of the influenza virus-specific CD8+ CTL response14 has concentrated on minimally manipulated, whole animal models. Thus, having established time-lapse profiles for TCR homogeneous (Figure 2a) versus polyclonal (Figure 2b and c) CTL populations that had been stimulated in vitro under comparable conditions, the next step was to repeat the microscopy analysis for potent CTL effectors15 recovered directly by BAL of the virus-infected lung (Figure 3). We first examined the length of time it took for the CTLs to form a stable contact after addition to targets. Although there was a trend for the higher avidity DbPA224-specific cells to find and attach to a target faster (DbPA224 14±4 min, n=11) than DbNP366 CTLs (25±5 min, n=13), this was not significant (Figure 3c). We observed that over 60% of DbPA224-specific CTLs remained in contact with their targets for longer than 50 min (Figure 3b), but DbNP366-specific CTLs did not, showing target cell contact after 50 min (Figure 3a). Interestingly, despite the higher avidity and longer duration of contacts, the DbPA224-specific CTLs took significantly more time to induce target cell rounding (117±10 min, n=11) than the DbNP366 CTLs (52±7 min, n=11) (Figure 3d and e), but there was no apparent correlation between duration of contact time and the duration of the execution phase of death duration, that is, rounding to loss of membrane integrity (Figure 3). Induction of the classical apoptosis mode of killing by activated CTLs is dependent on the pattern of pfp/grz expression.4 To determine if mode of killing by DbNP366 and DbPA224-specific CTLs was through classical apoptosis, DbNP366- and DbPA224-specific CTLs were cultured with MC57G targets pulsed with 1 μM of appropriate peptide (Figure 4, Supplementary Movies 2 and 3). The morphological characteristics of target cell death (Figure 4, arrows) were measured by time-lapse microscopy as described earlier. A representative montage is shown. Target cells killed by both DbNP366- and DbPA224-specific CTLs showed morphological changes characteristic of classical apoptosis. After contact with a single CTL effector, both NP366- and PA224-pulsed MC57G target cells showed membrane blebbing, AV+ and rounding after approximately 2 h 40 min (Figure 4). Loss of membrane integrity (measure by PI uptake) occurred after 3 h from contact for both DbNP366- and DbPA224-specific CTLs. Therefore, despite observed differences in contact time (Figure 3), DbNP366- and DbPA224-specific CTLs kill target cells through classical apoptosis.

Figure 3
figure 3

Interaction characteristics of the ex vivo-isolated CTL. Effector CD8+ T cells were obtained ex vivo by BAL of influenza virus-infected mouse lung, and added to MC57G targets pulsed previously with the NP336 or PA224 peptides. Images of morphology, annexin V-FLUOS and PI were obtained every 5 min to measure the duration of CTL contact (▪) and the time from target cell rounding to PI uptake (□) for individual DbNP366- (a) and DbPA224- (b) specific CTL. The duration of ‘searching time’, presented as the average time (minutes±S.E.M.) it took for the CTL to form stable contacts with targets after their addition to the assay (c), is shown. n.s.d (no statistical difference) and the time from CTL contact until target cell rounding (d) or PI+ (e) (mean minutes±S.E.M., *P<0.05) is shown. Data is pooled from three independent experiments

Figure 4
figure 4

Montage of time-lapse microscopy of DbNP366- and DbPA224-specific CTL killing. Influenza-specific CD8+ T cells were isolated from the spleen of a PR8-primed mouse and further stimulated in vitro before adding onto MC57 target cells pulsed with NP336 or PA224 peptide. Images of morphology, annexin V-FLUOS and PI were obtained every 5 min. A montage of NP366- and PA244-specific CD8+ T cells mediating classical apoptosis of targets is shown here. The arrows indicate the target cell

Figures 2 and 3 indicated that differences in target cell contact time between DbNP366- and DbPA224-specific CTLs was only observed directly ex vivo. This was confirmed by enumerating the average contact times both in vitro-stimulated (Figure 5a) or ex vivo-isolated (Figure 5b) DbNP366- and DbPA224-specific CTLs. The mean±S.E.M. duration of target cell contact and time from initial cell rounding until death (PI+) are shown. The average duration of contact was not statistically different for in vitro-activated DbNP366- and DbPA224-specific CTLs (Figure 5a; average 30 versus 40 min, respectively). In contrast, ex vivo-isolated DbNP366-specific CTLs showed significantly shorter contact time compared with DbPA224-specific CTLs (Figure 5b; 11 versus 90 min, P<0.05). Although in vitro-activated CTLs induced cell death more quickly compared with ex vivo CTLs, there was no difference in the time to cell death induced by either DbNP366- and DbPA224-specific CTLs (Figure 5a and b).

Figure 5
figure 5

Summary and TCR/pMHCI avidity profiles for the different CTLs. (a and b) The time intervals for: CTL/target contact, initial contact to execution (PI+) and target rounding to PI+ are presented as mean±S.E.M. values for the in vitro- (a) and in vivo (b)-stimulated CTL populations (n=13 for ex vivo, n=25–59 for in vitro). *Denotes significant differences (P<0.05) between the comparable DbNP366- and DbPA224-specific sets. (c and d) Tetramer elution characteristics are shown for in vitro (c) and ex vivo (d) CTLs stained with the DbNP366 and DbPA224 tetramers (see Materials and Methods). Lymphocytes were stained with DbNP366 and DbPA224 tetramers for 1 h in the presence of H-2 Db antibody, to minimize rebinding of tetramer. The results show the % of maximum tetramer binding for: (c) the average of duplicates or (d) mean±S.D. (n=5, *P<0.05)

We have showed previously that in vitro culture of DbPA224-specific CTLs results in loss of high avidity T cells due to antigen-induced cell death.10, 16 The loss of high-avidity DbPA224 CTLs after in vitro stimulation was reflected in the lower proportion of CTLs that were in contact with target cells for longer than 50 min (25%, Figure 2c) when compared with ex vivo-isolated CTLs (61%, Figure 3b). The changes in CTL avidity were confirmed by measuring TCR/epitope avidity by tetramer elution. Although there was no obvious difference for the in vitro-stimulated DbNP366 and DbPA224-specific sets (Figure 5c), the DbPA224-specific CTLs reproduced the profile of higher avidity (Figure 5d) that has been found repeatedly11 for this T-cell/epitope combination. Therefore, no difference in CTL contact time after in vitro peptide stimulation is most likely due to loss of high-avidity DbPA224-specific CTLs.

Discussion

We have showed recently that CTL lacking grzA and B induce cell death of targets that is qualitatively different to the classical mode of apoptotic cell death typical of intact CTL.4 Although DbNP366 and DbPA224-specific CTL show similar cytotoxic capacity on a per cell basis,7 single cell analysis showed that there is heterogeneity of effector gene expression.5 Furthermore, these CTL populations show qualitative differences in TCR repertoire diversity,17 pMHC/TCR avidity and cytokine production.10, 11 Therefore, it remained a formal possibility that different CTL populations may use different modes of cytotoxic killing. This study clearly shows that both in vitro and ex vivo isolated DbNP366- and DbPA224-specific CTLs induce classical apoptosis of peptide-pulsed targets, indicated by sequential annexin V+ and PI+ staining. The same mode of killing was observed also for in vitro-activated OT-I CTL, indicating that target cell killing through classical apoptosis is independent of TCR usage and an intrinsic property of CTLs.

The findings presented here indicate that any differences between effector CTL populations are diminished as a consequence of maximal activation following in vitro culture with peptide-pulsed stimulators. In agreement with previous studies,10, 11 ex vivo-derived DbPA224-specific CTLs displayed higher avidity compared with DbNP366-specific CTLs, whereas no difference in avidity was observed between in vitro-stimulated DbNP366- and DbPA224-specific CTLs. This most likely reflects loss of high-avidity DbPA224-specific CTLs after in vitro peptide stimulation due to antigen-induced cell death.16 Therefore, TCR avidity is most likely to be a major contributor to the duration of the CTL–target contact.

Interestingly, increased contact time had no significant impact on the kinetics of CTL induced apoptosis of target cells. It has recently been showed that as few as three specific peptide-MHC complexes need to be recognized to mediate CTL killing.18 Importantly, the recognition of so few pMHC complexes did not result in the establishment of a stable immunological synapse suggesting that only partial signalling is required for cytotoxic function CTL killing function.18 Given the low threshold of antigen required for eliciting cytotoxic function, it is no surprise that target cell death is independent from the duration of CTL contact.

While CTL mediated cytotoxicity only requires brief, low-density interactions,18 cytokine secretion requires sustained contact and TCR signalling.19 Therefore, while TCR avidity may not determine kinetics of target cell death, it is still possible that variation in pMHC/TCR avidity may still impact other functional differences. The high avidity DbPA224-specific CTL population recovered directly ex vivo shows higher levels of IFN-γ, TNF-α and IL-2 production following short-term in vitro stimulation with peptide.11 Furthermore, differential levels of antigen availability may be a key factor in determining the role of specific CTL during in vivo influenza virus infection. Earlier experiments20 indicated that after infection, DbNP366 is expressed on the surface of a variety of cell types while DbPA224 is found only on dendritic cells. As such, differences in contact time observed between DbNP366- and DbPA224-specific CTL may reflect different roles during virus infection. It is tempting to suggest that while DbPA224-specific CTL are capable of mediating cytotoxicity,7 maybe their dominant role is helping establish an appropriate inflammatory environment by extended production of IFN-γ, TNF-α and IL-2 from prolonged contact with pMHC on dendritic cells.

By observing T cells in the time-lapse experiments, it was clear that, subsequent to elution from one target, virus-specific CTLs go on to find, and destroy additional cells expressing the appropriate pMHCI complex (data not shown). Using both in vitro and in vivo cytotoxic assays, we have shown previously that DbNP366- and DbPA224-specific CTLs show the same level of cytotoxic capacity on a per cell basis.7 This analysis establishes that the lower-avidity DbNP366-specific CTL population can achieve the same level of target cell elimination as the DbPA224-specific set, but with shorter contact times; therefore, it is tempting to speculate that DbNP366-specific CTL may play a more significant role in ensuring viral clearance. This supports the observation that mutant influenza A viruses, lacking the NP366 epitope, are more virulent in B6 mice.17 Furthermore, the ability of the DbNP366-specific effectors to elute more rapidly, and to potentially receive multiple stimulations, may be one of the several10 reasons why CTLs with this specificity are so prominent following secondary in vivo challenge.12 As such, an overall profile of low, but sufficient, TCR/epitope avidity may thus be functionally advantageous and help to ensure efficient viral clearance after infection.

Materials and Methods

Generating CTLs by in vitro culture

Influenza-immune cells were recovered from the spleens of mice primed i.p. with the A/PR/8/34 H1N1 influenza A virus (PR8) influenza A virus 6 weeks earlier. Splenocytes from naive B6 mice were pulsed with 1 μM of the OVA257 or NP366 peptide, or 10 nM PA224 peptide for 1 h at 37 °C, irradiated (3000 rad from 60Co source), and mixed with an equal number of B6-OT-I TCR transgenic or PR8-immune cells. These cultures were then maintained in c-RPMI (RPMI supplemented with 10% FCS/penicillin-streptomycin/200 mM L-glutamine/5 × 10−5 M β2-mercaptoethanol) for 7 days at 37 °C, 5% CO2, stimulated again with peptide-pulsed irradiated splenocytes (as above), resuspended in c-RPMI, with 5 U/ml of recombinant human IL-2, and cultured for a further 5 days. Before time lapse imaging (see below), these in vitro-stimulated CTLs were Ficoll-purified using lymphocyte separation medium (Cappel Laboratory, ICN Biomedicals, Seven Hills, New South Wales, Australia).

The ex vivo effectors

Secondary influenza-specific CTLs were generated by priming B6 mice i.p. with 1.5 × 107 PFU of the PR8 (H1N1) virus 6 weeks before intranasal challenge with 104 PFU of the HK (H3N2) virus.15 The CTL effectors were then harvested by BAL after a further 8 days, and incubated on plastic six-well plates for 1 h at 37 °C, 5% CO2, to remove adherent macrophages and monocytes.

Time-lapse microscopy

MC57 (H2b) fibroblasts were plated at 1 × 104 cells per well in a 96-well plate and coated with 1 μM of the appropriate peptide for 1 h at 37 °C before washing and adding 50 ng/ml PI and annexin V-FLUOS (2 μg/ml). The plates were then maintained at 37 °C on the temperature-controlled stage (Prior Proscan; GT Vision) of a light fluorescence microscope (IX-81; Olympus). Equal numbers of effector CTLs were added to the targets, and images were captured at specified intervals using a CCD camera (model ORCA-ER; Hamamatsu) controlled by MetaMorph software (Universal Imaging Corp.).4 To avoid the complication of individual targets receiving multiple hits from several effectors, only those encountering one CTL are shown in the analysis. Statistical significance was determined using an unpaired Student's t-test.

Tetramer elution to measure avidity

Lymphocyte populations (0.5–2 × 106 cells) were stained with the DbNP366-PE and DbPA224-PE tetramers for 1 h at room temperature, incubated with an mAb to H2Db (50 μg/ml) to prevent tetramer rebinding, and analysed by flow cytometry for tetramer loss over time.11