Sensitive methods for the measurement of an adaptive immune response are essential for the development of our understanding of the immune system and the development of immunotherapies. Of particular interest in many immunotherapy protocols is the analysis of CD8+ cytotoxic T lymphocyte (CTL)-mediated killing of virus-infected cells and cancers cells. CTLs recognize target cells through interaction of their T-cell receptors (TCR) with “foreign” peptides presented by major histocompatibility complex class I (MHC-I) molecules on target cells. The engagement of the TCR and the peptide/MHC-I complex results in CTL killing of the antigen-presenting cell, either through exocytosis of lytic granules towards the engaged target cell (1) or via expression of Fas ligand which ligates and multimerizes cognate death receptors on target cells (2).
Several methods have been developed to measure CTL cytotoxicity (3). In a testament to its efficacy, the 51Cr-release assay has remained the “gold standard” method for measuring CTL activity after more than five decades since its inception (4). Despite this, there are several drawbacks to this technique, including its reliance on cell lines that can take up 51Cr; high spontaneous release of 51Cr from cells over time; its semiquantitative nature; interassay variability; the requirement to often restimulate CTLs prior to the assay; its biohazardous nature and its inability to directly assess cell killing in vivo.
One emerging methodology that overcomes the limitations associated with the 51Cr-release assay is the use of flow cytometry-based techniques to detect target cell death. Flow cytometry is a high-throughput platform that has the capacity to simultaneously measure 12 or more parameters based on cell light scatter and fluorescence properties at a single cell level. The premise for CTL assays using this methodology is relatively simple, with target cells being discriminated from other cells by fluorescent labels and target cell death determined using fluorescent viability markers. In this way, the number of target cells actually being killed is directly assessed. Numerous iterations of this technique have been reported and in many instances these assays have compared favorably with the 51Cr-release assay (5–10). However, the true potential of flow cytometry methods for assessing target cell death has not been fully exploited. We have a long history in developing fluorescent dye-labeling methods for use in flow cytometry-based cell tracking and proliferation assays (11–14). Recently, we compared the fluorescent dyes carboxyfluorescein diacetate succinimidyl ester (CFSE), cell trace violet (CTV), and cell proliferation dye (CPD) for their suitability in tracking lymphocyte proliferation in vitro and in vivo (15). These three dyes label lymphocytes to very high levels of fluorescence intensity with exceptionally low fluorescence variability and low cell toxicity. Since these dyes have different fluorescence emission spectra and can be used to label cells with multiple fluorescence intensities it is possible to generate numerous populations of viable cells with unique fluorescence signatures that can be discernable by flow cytometry. In this way, instead of having a limited number of distinguishable target cells in flow cytometry based CTL assays, typically 2–7 using current methods (10), potentially hundreds of different target cell populations could be generated. For example, if cells were labeled with four intensities of the three fluorescent dyes, the number of distinguishable targets would be 4 × 4 × 4 = 64, with five fluorescent intensities of each dye the number would be 5 × 5 × 5 = 125, and with six fluorescent intensities the number would be 216. This is analogous to the fluorescent cell barcoding method used on fixed-permeabilized cells for drug screening (16). With the availability of vital dyes like CFSE, CTV, and CPD this fluorescence multiplexing approach can now be performed on viable cells.
Here we show how CFSE, CTV, and CPD can be used to generate large fluorescence-based arrays of 64, 125, and 216 distinct target cell clusters to assay CTL function in vivo using flow cytometry. We refer to this approach as the fluorescent target array (FTA) killing assay. The assay allows the simultaneous measurement of the in vivo killing of target cells pulsed with numerous peptides at different concentrations and the inclusion of many replicates, with the associated benefit of being able to monitor the phenotype of the CTLs. This not only reduces the number of animals required to test a hypothesis, but also allows previously unfeasible experiments to be performed.
Mice were obtained from the Australian National University (ANU) Bioscience Services, ANU. Mice were housed and handled according to the guidelines of the ANU Animal Experimentation Ethics Committee. Mouse strains used were C57BL/6 (B6), BALB/c and B6.CD45.1 (B6 congenic for CD45.1). Transgenic (Tg) mouse strains were OT-I [TCR-Tg specific for SIINFEKL/Kb on a B6 background (17)], P14 [TCR-Tg specific for GP33/Db on a B6.CD45.1 background (18)] and F5 [TCR-Tg specific for NP68/Db on a B6 background (19)]. Male mice were used at 6–12 weeks of age.
Lymphocytes were obtained from spleen and/or lymph nodes as previously described (20). CD8+ T cells were enriched from lymphocytes by MACS separation (Miltenyi Biotec) as previously described (20).
Peptides and Virus
Peptides were synthesized at the Australian Cancer Research Foundation Biomolecular Resource Facility, JCSMR, ANU. Peptides were SIINFEKL, GP33 (KAVYNFATC), C9M (KAVYNFATM), NP68 (ASNENMDAM), F2L (SPYAAGYDL), F2L mod (SPGAAGYDL), HIV-gag (AMQMLKETI), and HIV mod (AMQMLEKTI). Fowl pox virus (FPV)-HIV and vaccinia virus (VV)-HIV stocks were prepared as described previously (21).
Fluorescent Dye Labeling
CFSE, CTV (both from Molecular Probes, Invitrogen), and CPD (eBioscience) were kept in DMSO and stored at −20°C. TCR-Tg CD8+ T cells were labeled with CFSE for proliferation tracking as detailed previously (15). For dye labeling of target cells, splenocytes were resuspended at 0.5–2 × 108/mL in 20°C RPMI 1640 (Invitrogen) supplemented with 10% fetal calf serum (FCS), and labeled with a final concentration of 0–47000 nM of each dye in 1–2 mL aliquots of cells with immediate vortexing to ensure rapid homogenous staining of cells. Cells were incubated at 20°C for 5 min. After labeling, cells were washed three times with RPMI 1640 supplemented with 10% FCS. This is as described previously (14), using the method videoed by JoVE (22), with the exception that the staining buffer used was always RPMI 1640 supplemented with 10% FCS (the inclusion of high FCS was to ensure cell toxicity was minimized). PKH-26 (Sigma-Aldrich) was used at 12 μM to label cells as described by the manufacturer.
Two milliliter aliquots of equal numbers of lymphocytes were initially labeled with CTV, the concentrations of which were dependent on the type of FTA, as outlined in Table 1. Cells were then split equally and these labeled with the several concentrations of CFSE in 1 mL aliquots as outlined in Table 1. At this point each sample was washed once in 10 mL of RPMI 1640 supplemented with 10% FCS. Each sample was then resuspended in RPMI 1640 supplemented with 10% FCS and split into several 0.25 mL aliquots (based on the type of FTA) and 0.25 mL of a two times concentration of peptide added to each sample for 1 hr at 37°C. Cells were then suspended in 5 mL of 4°C RPMI 1640 supplemented with 5% FCS and an underlay of 3.5 mL of 4°C FCS added. Cells were then spun through the FCS layer at 300 g for 10 min to remove excess peptide and the supernatant removed by aspiration. Cells were then washed once more in 10 mL of 4°C RPMI 1640 supplemented with 5% FCS. At this point, the samples to be labeled with the same concentration of CPD were pooled in 15 mL of RPMI 1640 supplemented with 5% FCS, the cells pelleted and the supernatant removed. Samples were then resuspended in 2 mL of 20°C RPMI 1640 supplemented with 10% FCS and labeled with several concentrations of CPD as outlined in Table 1. Cells were washed twice with 10 mL of RPMI 1640 supplemented with 5% FCS, pooled and washed once more. Where necessary cells were also labeled with PKH-26 (as described by the manufacturer). Cells were then counted and resuspended at up to 25 × 107/mL ready for injection.
Table 1. Concentrations of dyes used to construct each FTA
Dye concentrations (nM)
In Vivo Assays
CD8+ T cells were adoptively transferred into the lateral tail vein of host mice at a total cell number of 5 × 107 or less in 100 μL of PBS using a 29-gauge needle and syringe. After 10–60 min postadoptive transfer of CD8+ T cells, host mice were challenged with antigen in the form of a 100 μL bolus of PBS containing 25 μg of OVA (Sigma), 10 μmol of GP33 and/or 50 μmol of NP68 peptide into the lateral tail vein using a 29-gauge needle and syringe. A 1 μg bolus of LPS (Sigma) was co-injected with antigen to act as a danger signal and enhance lymphocyte responses. Injection of antigen and LPS was repeated once (24-hr postfirst injection) to further enhance the lymphocyte response. Primary immunizations with FPV-HIV and VV-HIV were done i.n. with 5 × 106 PFU and 10 × 106 PFU, respectively. Secondary immunizations with FPV-HIV and VV-HIV were performed i.m. with 5 × 106 PFU and 10 × 106 PFU, respectively. FTAs were transferred into the lateral tail vein of host mice at a total cell number of 5 × 107 or less and left in vivo for 18 hr before splenocytes were assayed for target cell killing.
Antibody and MHC-I Tetramer Labeling of Cells
Splenocytes were initially labeled with MHC-I tetramers for 25 min at 37°C and then washed twice, ready for antibody staining. MHC-I tetramers included C9M-peptide-Db-APC (for P14 T-cell detection), NP68-peptide-Db-APC (for F5 T-cell detection), SIINFEKL-peptide-Kb-APC (for OT–I T-cell detection), and HIV-gag-peptide-Kd-APC. Antibodies used to delineate cell populations included anti-CD4-biotin, anti-CD8α-APC-eFluor 780, anti-B220-Alexa Fluor 700, anti-CD45.1-PerCP-Cy5.5, anti-CD45.2-Horizon V500, anti-LAMP-I-PE-Cy7, anti-Vα2-PE (for OT-I and P14 T-cell detection), and anti-Vβ11-bio (for F5 T-cell detection) purchased from either BD Bioscience, eBioscience or Biolegend. Cell viability was assessed with the dye Hoechst 33258 (1 μg/mL, Calbiochem-Behring Corp.). Cells were labeled with antibodies and Hoechst 33258 for 20–30 min on ice then washed twice, as previously described (20). The second step reagent SAV-PE-Alexa Fluor 610 (Invitrogen), which was used to detect biotinylated antibodies, was incubated with cells for 20–30 min on ice and then the cells washed twice.
Flow Cytometry and Data Analysis
Analytical flow cytometry was performed using a Fortessa flow cytometer of standard configuration (Becton Dickinson). Spectral overlap of each fluorochrome was monitored using samples labeled with each fluorescence parameter independent of the other fluorescence parameters. Based on this, compensation was performed to help reduce any spectral “spillover” observed. Postacquisition gating was used to analyze cell subsets using FlowJo software (Tree Star, OR). A detailed example of the gating strategy employed in this study is depicted in Supporting Information Figure 1. Gates were employed based on fluorescence minus one antibody labeled cells and/or cells labeled with a full complement of fluorescent reagents but from a naïve mouse, as appropriate. Data were acquired from January 2011 to February 2012. The flow cytometer had undergone periodic quality control measures using eight channel fluorescent beads throughout experimentation.
% Specific Killing Calculations
The % specific killing was assessed based on the relative ratios of target cells present in the primed animals compared with those in the naïve animals. Assuming that no specific target cell killing occurs in the naïve animal, the ratio of the number of peptide-pulsed target cells to unpulsed target cells in a naïve animal is defined as 0% killing. The value of the ratio between peptide-pulsed and unpulsed target cells should be identical in a primed animal if no killing occurs, but would decrease if killing has occurred. Therefore, comparing these ratios in primed mice relative to naïve mice, the % specific killing can be determined. In this way, the following formula was used to calculate % specific killing:
For samples where n ≥ 10 and data appeared to follow a normal distribution, a normality test was performed using the D'Agostino–Pearson normality test. If samples satisfied this test, a two-tailed t-test was performed to test for statistical significance between two samples. Otherwise the Mann–Whitney nonparametric test was performed. If n = 3 and these data did not come from a larger data set applicable to a normality test (as above), statistical analysis was not performed. Statistic values, including AUC and EC50 (peptide concentrations giving half maximal % Specific killing) were calculated using GraphPad Prism Software.
Establishing a 64 FTA Killing Assay
To establish the use of CFSE, CTV, and CPD in generating a FTA, lymphocytes where co-labeled with combinations of four concentrations of each dye that were easily discernable by flow cytometry (Fig. 1a). By labeling lymphocytes with all combinations of each of these dyes at each of these fluorescence intensities, a three dimensional fluorescence array of 64 discernable cell clusters could be generated (Fig. 1b). This is more easily visualized by dividing the array into four, based on CPD fluorescence intensity, to generated a two-dimensional array of four target “panels” (A, B, C, and D) each consisting of 16 cell clusters (based on CFSE and CTV fluorescence intensity) (Fig. 1c). In this format, the cell clusters were pulsed with a total of four different peptides (SIINFEKL, GP33, NP68, and C9M), either alone or in different combinations, and at seven different concentrations as outlined in Figure 1d to generate a 64 FTA.
To demonstrate that FTAs can be used to assess CTL killing activity in vivo, the 64 FTA was injected into host mice containing activated CTLs of multiple specificities. To generate CTL activity host mice received by adoptive transfer CD8+ T cells derived from TCR-transgenic mice specific for SIINFEKL/Kb (OT-I mice), GP33/Db (P14 mice), and NP68/Db (F5 mice) and these mice were then primed with specific antigen (referred to as “primed” animals). As a control additional host animals received no antigen (referred to as “naïve” animals). Eighteen hours after FTA transfer, spleen cells from host mice were assessed for target cell death. The FTA was constructed using lymphocytes that were CD45.1+ and transferred into host mice that were CD45.2+, allowing target cells to be easily distinguishable from host cells using CD45 allotype-specific antibodies (Supporting Information Fig. 1). The data obtained from one naïve and one primed mouse are shown in Figure 1. Several target cell clusters present in the primed animal were clearly diminished partially or almost completely relative to the corresponding target cell clusters present in the naïve animal (compare Fig. 1c with 1e). To assess % specific killing the relative ratios of target cells present in the primed animal were compared with those in the naïve animal as described in the Methods section. As an example, the target cell clusters in the first row of panel A and B corresponding to the SIINFEKL titration (clusters 1–8 in Fig. 1d) were analyzed (Fig. 1f). Histogram analysis of these target clusters revealed there was a dramatic reduction in the number of cells pulsed with SIINFEKL (clusters 2–8) in the primed animal compared with the naïve animal, but no reduction of the un-pulsed control target cell cluster (clusters 1) in the primed animal compared with the naïve animal (Fig. 1f). By calculating the number of cells in each target cluster in the primed and naïve animals the % specific killing could be calculated using the formula described in the Methods section (Fig. 1f table).
The 64 FTA Killing Assay can be Combined with the Assessment of Numerous CTL Properties
An advantage of using flow cytometry in CTL killing assays is that in addition to target cell death, the phenotype of CTLs arising from antigen challenge can also be examined and correlated with target cell killing. Thus, before determining the % specific killing of target cells in the 64 FTA assay, we examined the effector status of the adoptively transferred CD8+ T cells and related these parameters to actual target cell death. The transferred CD8+ T cells were CFSE labeled, allowing their proliferation status to be determined (Fig. 2a). At 4 days postinitial antigen exposure, the majority of OT-I-derived CD8+ T cells had divided 5–12 times and, based on antibody and MHC-I tetramer labeling, comprised ∼12.5% of the total CD8+ T-cell pool (Fig. 2b). The P14-derived CD8+ T cells had divided on average slightly less in comparison, and comprised ∼7.5% of the CD8+ T-cell pool and the F5-derived CD8+ T cells had on average proliferated the least, and comprised ∼1.3% of the CD8+ T-cell pool. In addition, the proportion of CTLs exocytosing lytic granules following exposure to peptide-pulsed targets was assessed through measurement of LAMP-I externalization (23) (Figs. 2a and 2c). The extent of degranulation correlated approximately with the number of cell divisions the CTLs had undergone, with 8–10% of F5 and P14 derived CTLs, and ∼29% of OT-I-derived CTLs showing evidence of degranulation. To determine whether these events related to actual target cell killing, the % specific killing was calculated using the 64 FTA assay.
Assessing the capacity of the different CTL populations to kill their cognate peptide-presenting targets (Fig. 2d, left panel) revealed that the majority of SIINFEKL-pulsed targets had been killed (65–92% killing) across the entire range of peptide concentrations used (pM to μM). The majority (>80%) of targets pulsed with 2.5 μM of GP33 were also killed, but the killing of targets pulsed with lower concentrations of GP33 steadily diminished across the nanomolar concentration range. Except at the highest peptide concentration used (2.5 μM), NP68 pulsed targets were also killed to a similar extent as GP33 pulsed targets, despite there being ∼6-fold fewer CD8+ T cells being derived from NP68-specific F5 donor cells than from GP33-specific P14 donor cells (Fig. 2b). It should be noted that killing of SIINFEKL, GP33, and NP68-pulsed target cells was dependent on the presence of their cognate T cell “clone,” since host mice adoptively transferred with each clone separately could only kill targets bearing their cognate peptide (data not shown). It should also be noted that peptide-pulsed B cell, CD4+ T cell and CD8+ T-cell targets were killed to a similar extent by all three CTL populations (Fig. 2d, right panel and data not shown).
This assay also revealed the dominance of the OT-I-derived CTLs over the other two T cell clones since target cells presenting both the OT-I specific peptide, SIINFEKL, and either or both of the other two cognate peptides were not killed any more efficiently than target cells pulsed with SIINFEKL alone (Supporting Information Fig. 2). Targets simultaneously pulsed with midrange nM amounts of GP33 and NP68 appeared to be killed more efficiently (in an additive manner) compared with cells pulsed with either of these peptides alone, but this effect disappeared at higher peptide concentrations (Supporting Information Fig. 2).
In addition to these basic titrations of the loading concentration of cognate peptides, the FTA killing assay also allowed the simultaneous measurement of the lysis of target cells pulsed with modified versions of cognate peptides. For example, several target cell clusters were also pulsed with C9M, a modified version of the GP33 peptide that has a single amino acid substitution (C→M) at position 9, resulting in a much higher affinity interaction of the GP33 peptide/MHC-I complex with the GP33-specific P14-derived TCR (24). In contrast to the wild type GP33 peptide, essentially 100% of target cells pulsed with the entire picomolar to micromolar concentration range of the C9M peptide were killed (Fig. 2c), representing at least a 1,000-fold reduction in the concentration of GP33 peptide required to render target cells susceptible to P14 CTL killing.
From these examples it is clear that the FTA killing assay can be used in parallel with standard tetramer binding and degranulation assays to generate much more information about the killing efficiency of CTLs in vivo.
Use of a 125 FTA to Obtain Many Replicates of CTL Killing In Vivo and Compare CTL Target Cell Avidities
Using the 64 FTA killing assay single data points were obtained for each of the different peptide loaded target cell populations. To obtain replicate assays for each specific target cell, a 125 FTA was constructed. This larger array was also used to further assess how well the FTA killing assay could be used to monitor TCR-peptide/MHC-I avidity differences. Thus the efficacy of the altered GP33 peptide, C9M, in mediating target cell killing by P14-derived CTLs was re-investigated using a wider range of peptide loading concentrations to that depicted in Figures 1 and 2. The 125 FTA was composed of target cell populations loaded with five concentrations of each the fluorescent dyes CFSE, CTV, and CPD. Labeling lymphocytes with all combinations of each of these three dyes at each of these five fluorescence intensities, generated a three-dimensional fluorescence array of 125 cell clusters or a two-dimensional array of five panels (based on CPD intensity), each consisting of 25 cell clusters (based on CFSE and CTV intensity) (Fig. 3a). Twelve different concentrations of either the GP33 or C9M peptide were loaded onto the 25 cell clusters, together with a no peptide (NIL) control, within each panel and this replicated five times across the panels to generate a 125 FTA of peptide loaded target cells (Fig. 3a, right panel). To generate antigen-specific CTLs, various numbers (1.5 × 105 to 1 × 107) of CFSE-labeled CD8+ T cells derived from GP33-specific P14 mice were adoptively transferred into host mice, and the recipient mice primed with specific antigen. At 3 days postpriming, the peptide loaded 125 FTA was transferred into host mice and spleen cells harvested 18 hr later. At this time point the P14-derived CD8+ T cells showed a strong proliferative response, with the majority dividing 5–12 times and representing between 1 and 12% of the CD8+ T-cell pool, this percentage roughly correlating with the number of P14-derived CD8+ T cells initially injected (Fig. 3b). The % specific killing of all target cell clusters of the FTA was calculated for each of the mice given different numbers of P14-derived CD8+ T cells (Fig. 3c). With % killing calculated for all five intra-animal replicates, the mean and variance of target cell killing could be calculated at each peptide loading concentration (Fig. 3c) and individual mean and variance plotted across all experimental groups for comparison (Fig. 3d). The remarkable killing efficiency of target cells pulsed with C9M was highlighted in this analysis, most targets (70–100%) pulsed with 4 nM to 4 μM of this peptide being killed, even when P14 CD8+ T cells represented only ∼1% of the CD8+ T-cell pool. Compared with cells pulsed with GP33, target cells were killed with similar efficiency when pulsed with 4,000-fold less C9M. Thus, the FTA killing assay allows the measurement of TCR-peptide/MHC-I avidity differences in vivo. It should be noted that the killing activity measured by FTAs was not only highly reproducible within an animal (i.e., intra-animal variability), but also between different animals (i.e., interanimal variability) (Supporting Information Fig. 3).
Use of a 216 FTA to Assess the Killing Efficiency of CTLs Induced by HIV Vaccines
Having established the use of the FTA killing assay in TCR transgenic models, we then examined the ability of the FTA killing assay to assess CTL responses generated from a polyclonal CD8+ T-cell population following different vaccination strategies. A focus of emerging vaccines against chronic viral infections, like human immunodeficiency virus (HIV), is the generation of strong CTL responses. Heterologous prime–boost vaccination has been proven to generate strong CTL responses against many vaccine antigens (25). It consists of using two different vectors expressing the same target “immunogen” in the primary (“prime”) and secondary (“boost”) immunization. To establish the ability of the FTA killing assay to assess the efficacy of such vaccine strategies, a heterologous virus prime–boost regime was compared with a homologous virus prime–boost regime against HIV-gag (a model epitope for HIV vaccines) in a mouse (BALB/c) model (21). The virus used for the priming immunization was a recombinant fowl pox virus expressing HIV-gag (FPV-HIV) administered intranasally (i.n) and the virus used in the boost immunization was either FPV-HIV (i.e., homologous prime–boost regime) or recombinant VV expressing HIV-gag (VV-HIV) (i.e., heterologous prime–boost regime) administered intramuscularly (i.m.) as previously described (21). To further test the limits of constructing a FTA using CFSE, CTV, and CPD, the FTA consisted of lymphocytes labeled with combinations of six concentrations of each fluorescent dye to generate a three-dimensional fluorescence array of 216 cell clusters or a two-dimensional array of six panels (based on CPD intensity), each consisting of 36 cell clusters (based on CFSE and CTV intensity) (Fig. 4a). In addition, since BALB/c mice with a CD45 allotype difference are not readily available, to allow the FTA to be readily detected in syngeneic hosts we used a fourth dye, PKH-26, to label all targets cells for their detection in vivo (data not shown). At 6 days post-mouse vaccination, the 216 FTA was constructed and included 36 target cell clusters pulsed with seven different concentrations of the following BALB/c-specific peptides: HIV-gag (the HIV-gag immunodominant epitope), F2L (the immunodominant VV epitope), F2L mod (a lower avidity version of F2L) or, as negative controls, SIINFEKL (nonimmunogenic in BALB/c) and HIV mod (modified HIV-gag epitope that is not immunogenic). This panel of target cells was repeated six times within the FTA based on CPD fluorescence intensity (Fig. 4a, right panel). The 216 FTA was injected into host mice and after 18 hr in vivo target cell clusters were assessed for % specific killing in mice given either just the primary vaccination of each virus (as added controls) (Fig. 4b) or the prime–boost vaccination regimes (Fig. 4c). An advantage of using numerous peptide loading concentrations in the FTA killing assay is that it allows the generation of data that quantify the potency and avidity of the CTL response induced by each vaccination regime (Fig. 5). Thus, a measure of the overall killing efficiency of the CTLs generated by each vaccine regime was the area under the curve (AUC) of each % specific killing response (Fig. 5d). As expected the mice immunized with VV-HIV generated CTLs that were more efficient at killing targets pulsed with F2L than targets pulsed with the lower avidity version of F2L (i.e., F2L mod) (Figs. 5a and 5d). None of the negative control targets, pulsed with either SIINFEKL or HIV mod, were recognized by the CTLs (Fig. 4b). Generally, it appeared that vaccination with VV-HIV generated a stronger CTL killing response to the HIV-gag peptide than FPV-HIV vaccination (Figs. 5b and 5d). However, HIV-gag peptide specific killing was evident following a primary FPV-HIV vaccination when there were negligible HIV-gag-peptide/Kd tetramer reactive clones detected (<0.05%) (Figs. 5b–5d). Furthermore, following homologous prime–boost immunization the secondary CTL killing response against the HIV-gag peptide was greater than the primary response but overall the heterologous prime–boost regime elicited the best killing response to this epitope (Figs. 5b and 5d). Importantly, killing of target cells did not appear to correlate linearly with CTL clone size as measured by HIV-gag-peptide/Kd tetramer reactivity (Figs. 5c and 5d), suggesting that assays based on the enumeration of MHC-I tetramer binding T cells cannot be assumed to directly correlate with actual target cell killing activity.
Another useful statistic for detecting differences between CTL responses is the effective peptide concentration that gives half the maximal % specific killing (EC50) (Fig. 5e) and the associated statistic of fold decrease in EC50 from the primary response to the secondary response (Fig. 5f). The EC50 should reflect the functional avidity of CTLs for their respective targets. Functional avidity is the ability of T-cell populations to respond to a measured amount of antigen (MHC-peptide). High avidity CTL recognize low concentrations of antigen, whilst low-avidity CTL are functionally ineffective at low antigen concentrations (26, 27). Thus, high avidity CTL have greater capacity to clear an infection compared with low-avidity T cells (28). EC50 analysis revealed that the heterologous prime–boost vaccination regime (2° FPV-HIV/VV-HIV) gave a much lower EC50 for the HIV-gag peptide (mean of 0.9 nM) than the homologous prime–boost vaccination regime (2° FPV-HIV/FPV-HIV, mean of 8.6 nM) (Fig. 5e). The heterologous prime–boost vaccination regime (2° FPV-HIV/VV-HIV) was also the only vaccination strategy that generated a CTL response against the HIV-gag peptide that resulted in a significant reduction (∼3.4-fold) in the EC50 compared with the CTL response generated by the primary vaccination alone (1° VV-HIV) (Figs. 5e and 5f). Thus, overall it appears that the heterologous prime–boost vaccination regime generates the best HIV-gag peptide CTL response, both in terms of magnitude and avidity, a finding consistent with that previously reported (29).
From this analysis, it is evident that the FTA killing assay is a robust technique that is highly sensitive, detecting even minor CTL populations, and can measure both quantitative and qualitative differences in polyclonal CD8+ CTL responses against pathogens.
The FTA killing assay described here allows the simultaneous measurement of numerous CTL-mediated target cell killing events in vivo. The development of this assay has become possible with the availability of fluorescent dyes like CFSE, CTV, and CPD, which can label live cells with multiple, stable fluorescence intensities at discrete emission wavelengths. These dyes also label lymphocytes with fluorescence of low variance, typically with a fluorescence standard deviation of <25% of the mean fluorescence intensity (MFI) (15), allowing the delineation of several discretely labeled populations. While live cell multiplexing has been investigated previously, the paucity of compatible vital dyes with fluorescence of low variance has meant only a maximum of 8–12 spectrally distinct cell clusters has been achieved (30). We have demonstrated that with CFSE, CTV, and CPD that up to six intensities of each fluorescent dye can be used in combination to generate an array of up to 216 discernable cell clusters allowing the simultaneous measurement of killing of target cells pulsed with numerous peptides at different concentrations and the inclusion of many replicates. The FTA killing assay therefore is able to measure the fine antigen specificity and avidity of polyclonal CTL responses that have been previously not feasible to monitor in vivo. Since the robustness of this assay relies on the fidelity of the fluorescence signature of each target cell cluster, consideration should be given to any parameters that could affect this. We have noted three such parameters.
First, we have observed that B cells and T cells labeled with the same amount of each of the three dyes have different fluorescence intensities (15). The impact of this in discerning target cell clusters is relatively minor and is more apparent in larger/“tighter” arrays like the 125 and 216 FTAs (data not shown). Discriminating B cell and T-cell subsets through the inclusion of appropriate antibody labels can easily compensate for this effect.
Second, transfer of dye, particularly CPD, between cells was noted in our earlier study (15). This is more obvious when cells are cultured in vitro and negligible with labeled cells in vivo (15). Thus, while a FTA can be used to assess killing in vitro (data not shown), it is important that excess dye is washed away from cells and incubation times to assess killing activity kept to a minimum, particularly for larger/“tighter” FTAs, to reduce the effects of dye transfer in in vitro assays.
Finally, we have also noticed that B cells within a FTA can dilute their fluorescence intensity when left in vivo for more than 48 hr (data not shown). This suggests that B cells were proliferating, probably a result of them presenting cognate antigen. This makes it difficult to use B cells as markers for target cell killing in longer assays. Therefore, it would be better to assay T-cell subset targets for killing in these situations. A side effect of this “antigen presentation” is that cells in a FTA can generate an immune response in their own right (data not shown). This is a consideration of any assay that introduces a “foreign” peptide in vivo.
Since a FTA killing assay monitors cell death, it is also important to consider the toxicity of the different dye concentrations used in generating the target cell array. We have previously observed a toxic effect when cells are labeled with CFSE, however, this can be buffered against by incorporating protein into the labeling solution (15, 31). Indeed, more recently we have shown that CFSE, CTV, and CPD can be used at a combined concentration of up to 120 μM to label lymphocytes, without any effect on viability or function, provided labeling is undertaken in a buffer containing a high protein content (15). In the FTA assays generated in the current study, the combined concentration of the dyes used was <120 μM. Furthermore, we have not observed any differences in the killing efficiency of cells labeled with different dye concentrations (data not shown).
In conclusion, application of the vital dyes CFSE, CTV, and CPD to generate numerous fluorescent signatures of live cells has allowed the development of a robust and versatile killing assay. This assay has the potential to be applied to monitor the complexities of a polyclonal immune response against pathogens in vivo that have previously been unfeasible. Data indicate that the FTA killing assay is a robust, highly sensitive technique, which is able to detect quantitative and qualitative differences in polyclonal CTL responses, against pathogens, including minor CTL subsets. Thus, this assay has great potential to be used as a tool to evaluate the functional capacity of CTLs, following vaccination. Finally, there is great potential for the application of these live-cell fluorescent arrays in other areas of biology that might benefit from simultaneous screening of multiple live-cell events (16, 30).
The authors thank Harpreet Vohra and Michael Devoy for their excellent maintenance of the JCSMR FACS laboratory and Kerong Zhang, Cameron McCrae, and Kerry McAndrew of the Australian Cancer Research Foundation Biomolecular Resource Facility, JCSMR, ANU for the synthesis of peptides and MHC-I tetramers. They also thank Dr. Hilary Warren for critical reading of the manuscript.