Differentiation of human alloreactive CD8⁺ T cells in vitro

Abstract

Expansion and differentiation of alloantigen-reactive CD8⁺ T cells in mixed lymphocyte cultures was followed by measurement of the loss of carboxyfluorescein diacetate succinimidyl ester (CFSE) fluorescence of responder cells. Proliferation of CD8⁺ T cells became detectable on day 4 of culture and, 2 days later, > 60% of the CD8⁺ T cells in culture were dividing alloreactive lymphocytes. In parallel with expansion, CD8⁺ T-cell differentiation was initiated, as evidenced by an increase in the number of CD45RA⁻ and CD27⁻ T cells and acquisition of the ability to produce interferon-γ after restimulation with the specific alloantigen. Finally, although short-term stimulation and measurement of intracellular cytokine production allowed visualization of alloreactive CD8⁺ T cells expanded in vitro, this procedure did not detect circulating alloreactive CD8⁺ T cells activated in vivo in recipients of allogeneic kidney grafts.

Introduction

New assays have emerged to measure antigen-specific T-cell frequencies, specifically ELIspot and intracellular cytokine staining after antigen or peptide stimulation, flow cytometric enumeration of peptide/major histocompatibility complex (MHC) tetramer binding T cells, and loss of carboxyfluorescein diacetate succinimidyl ester (CFSE) fluorescence by proliferation.^1–8 Remarkably, antigen-specific CD8⁺ T-cell frequencies, as obtained by these new assays, are manifold higher than estimates of cytotoxic T lymphocyte precursor (CTLp) frequencies derived from classical, proliferation-based assays.¹ So far, these novel technologies have mainly been used for studying T-cell responses in acute and chronic viral infections and it remains to be resolved whether they will prove helpful in evaluating other clinically relevant immune reactions, such as autoimmune and alloantigen responses. Alloreactive CD8⁺ T cells are present in rejecting grafts and may contribute to the effector mechanisms against the allograft in rejections.^9,10 Quantification of precursor frequencies of alloreactive cytotoxic T lymphocytes (CTLs) may be used to define transplant recipients who can, for instance, safely reduce their immunosuppressive drug medication.¹¹ Novel technologies that allow a more direct enumeration of the alloreactive T-cell repertoire may therefore have implications for clinical practice.

Differentiation of human CD8⁺ T cells is closely linked to expression of specific cell-surface proteins. Heterogeneous, but partially overlapping, cell-surface markers on peripheral blood CD8⁺ T cells are linked to specific functional properties of these cells. Classically, CD45 RA and RO isoforms can be used to differentiate between naive T cells and memory T cells that encountered their cognate antigen in vivo. Recently, several reports refined and redefined the peripheral blood-derived CD8⁺ T-cell subsets.^12–14 After viral infection, CD8⁺ T cells differentiate into CD45RA⁻ CD27⁺ memory T cells.^15,16 These cells produce a heterogeneous set of cytokines, including interlekukin (IL)-4, interferon-γ (IFN-γ) and tumour necrosis factor-α (TNF-α). CD45RA⁻ CD8⁺ T cells may lose CD27 expression and reacquire CD45RA expression. By acquisition of a CD45RA⁺ CD27⁻ phenotype, CD8⁺ T cells gain high expression of cytolytic mediators such as perforin and granzyme B.¹⁴

In the present study, we employed two of the new flow cytometry-based methods, specifically intracellular cytokine staining and loss of CFSE fluorescence by proliferating cells, to analyse alloantigen-induced CD8⁺ T-cell differentiation.^5,6 In addition, we analysed whether alloantigen-induced intracellular cytokine production can be used to detect circulating alloreactive T cells in kidney allograft recipients.

Materials and methods

Subjects

Healthy individuals were asked to participate in the study. In addition, renal transplant recipients who had received a cadaveric renal transplant were investigated, provided that their MHC types displayed at least one mismatch with their donors at a human leucocyte antigen (HLA)-A or -B locus, and donor-derived spleen cells were available. Informed consent was obtained from all subjects, and the medical ethical committee of the Academic Medical Center approved the study. Heparin anticoagulated blood and (autologous) serum were obtained by standard venous puncture. Peripheral blood mononuclear cells (PBMC) were isolated as described previously¹⁷ and used when fresh. Serum was heat inactivated by incubation for 30 min at 56°. Frozen donor-derived spleen cells were thawed rapidly and washed twice in Hanks' balanced salt solution (HBSS) (BioWhittaker, Verviers, Belgium) containing 0·025 m Tris (hydroxymethyl) aminomethane, 5% (vol/vol) heat-inactivated fetal calf serum (FCS) (Integro B.V., Zaandam, The Netherlands), 100 U/ml sodium penicillin G (Brocades Pharma B.V., Leiderdorp, The Netherlands) and 100 µg/ml streptomycin sulphate (Gibco BRL, Paisley, UK) (wash medium), before resuspension in Iscove's modified Dulbecco's medium (IMDM) supplemented with antibiotics, 3·57 × 10⁻⁴ % (vol/vol) β-mercaptoethanol (Merck, Darmstad, Germany) and 5% (vol/vol) heat-inactivated autologous serum from the responder (X) (culture medium).

CFSE labelling

PBMC or spleen cells were pelleted and resuspended, at a final concentration of 10 × 10⁶ cells/ml, in phosphate-buffered saline (PBS) containing antibiotics. PBMC or spleen cells were labelled with 5 µm (final concentration) of 5-(and 6)-CFSE (Molecular Probes Europe BV, Leiden, The Netherlands) in PBS for 10 min at 37°. Cells were washed in wash medium. Subsequently, cells were resuspended in culture medium.

Determination of allospecific CD8⁺ T-cell frequencies in mixed lymphocyte cultures (MLCs) by measuring the loss of CFSE fluorescence of responder cells

Fifty-thousand CFSE-labelled responder cells/well (X_cfse) were incubated with 50 000 gamma-irradiated (3000 rads) allogeneic (Yr) or autologous (Xr) stimulator PBMC for 1, 3, 4, 5 and 6 days at 37°, 5% CO₂, in a humidified chamber in 96-well round-bottom culture plates (Corning Inc., New York, NY) in a final volume of 170 µl of culture medium/well. Subsequently, the cells were harvested, stained with CD8-phycoerythrin (PE) (Becton Dickinson Immunocytometry Systems, San Jose, CA) and analysed on a flow cytometer (FACSCalibur; Becton Dickinson). CD8^bright T cells, which retained maximal CFSE fluorescence, were presumed to be non-dividing and therefore designated as non-responding cells. CD8^bright T cells that lost any CFSE fluorescence were presumed to have undergone at least one cell division and were therefore designated as alloreactive CD8⁺ T cells.

Phenotypic analysis of alloantigen-driven proliferating cells in MLCs

CFSE-labelled responder PBMC (X_cfse) were incubated with irradiated allogeneic (Yr) or autologous (Xr) stimulator PBMC for 1, 3, 4, 5 and 6 days, as described above. Subsequently, cells were harvested, stained with CD8-peridin chlorophyl protein (perCP) (Becton Dickinson), CD27-biotin (CLB, Amsterdam, The Netherlands), streptavidin-allophycocyanin (APC) (Becton Dickinson) and CD45RA-RD1 (Coulter, Fullerton, CA), and analysed on a flow cytometer, as described above. Based on CFSE fluorescence, cells were gated in non-dividing, few generations of cell divisions or many generations of cell divisions within gated CD8⁺ lymphocytes. Frequencies of CD45RA⁺ CD27⁻ (effector), CD45RA⁺ CD27⁺ (naive), CD45RA⁻CD27⁻ (memory/effector), and CD45RA⁻ CD27⁺ (memory) cells within the respective gates were analysed, using Cellquest software (Becton Dickinson).

Determination of allospecific CD8⁺ T-cell frequencies after 6 days of MLC, followed by restimulation and intracellular cytokine staining

Unlabelled responder PBMC (X; 1 × 10⁶ cells/ml of culture medium) were incubated with CFSE-labelled and irradiated (3000 rads) allogeneic stimulator cells (Yr_cfse; 1 × 10⁶ cells/ml) for 6 days in polystyrene, tissue culture-treated, culture flasks (Corning) at 37°, 5% CO₂, in a humidified chamber. Subsequently, cells were harvested, washed in wash medium, counted and divided over polystyrene culture tubes as 1·0 × 10⁶ cells/tube. Next, 1 × 10⁶ freshly harvested CFSE-labelled allogeneic (Y_cfse) PBMC, CFSE-labelled irradiated autologous (Xr_cfse) restimulator cells from the same donors as used for the primary stimulation, or 1 µg/ml (final concentration) of Staphylococcus aureus Enterotoxin B (SEB; ICN, Costa Mesa, CA), were added to the tubes. The tubes, containing a final volume of 2 ml/tube, were incubated for 6 hr at 37°, 5% CO₂, in a humidified chamber. During the last 5 hr of culture, 10 µg/ml brefeldin A (Sigma Chemical Co., St Louis, MO) was added to the cultures. Subsequently, the cells were fixed, permeabilized and stained for CD8-PerCP, CD69-APC and IFN-γ-PE, as described previously.¹⁸ Cells were analysed on a flow cytometer. CD69⁺ IFN-γ⁺ cells within a CFSE⁻ CD8⁺ allogeneically restimulated culture were designated as allospecific CD8⁺ T lymphocytes.

In additional experiments, responder cells were CFSE labelled instead of stimulator and restimulator cells that remained unlabelled. After 6 days of primary stimulation and 6 hr of restimulation, cells were harvested and stained with CD8-PerCP, anti IFN-γ-PE and CD69-APC, followed by flow cytometric analysis.

Determination of allospecific CD8⁺ T-cell frequencies after short-term stimulation and intracellular cytokine staining

One million responder PBMC (X) were stimulated with 1·0 × 10⁶ CFSE labelled and (3000 rads) irradiated allogeneic stimulator PBMC or spleen cells (Yr_cfse) for 6 hr at 37°, 5% CO₂, in a humidified chamber in polystyrene round-bottom tubes (Falcon 352054; Becton Dickinson France) in a final culture medium volume of 2 ml. Alternatively, responder PBMC were incubated similarly with autologous CFSE-labelled irradiated stimulator PBMC (Xr_cfse) as a negative control. SEB (ICN) (1 µg/ml, final concentration)-stimulated cultures of responder cells (X) served as positive controls in each experiment. During the last 5 hr of culture, Brefeldin A (10 µg/ml, final concentration; Sigma Chemical Co.) was added to the cultures. Subsequently, cells were fixed, permeabilized and stained with CD8-PerCP (Becton Dickinson, San José, CA, USA) CD69-APC (Caltag Laboratories, Burlingame, CA) and IFN-γ-PE (Becton Dickinson), as described previously.¹⁸ Cells were analysed on a dual-laser flow cytometer (FACSCalibur; Becton Dickinson) using Cellquest software for data acquisition. Instrument set-up was performed as described previously.¹⁷ Cellquest software (Becton Dickinson) was used for data analysis. CD69⁺ IFN-γ⁺ cells within a CFSE⁻ CD8⁺ lymphocyte gate from allogeneically stimulated cultures were designated allospecific CD8⁺ T lymphocytes. Frequencies of allospecific CD8⁺ T cells were counted as significant only if considerably higher than autologously stimulated background frequencies of activated CFSE⁻ CD8⁺ T cells. This background staining did not exceed 0·05% of CFSE⁻ CD8⁺ T cells.

Mixed lymphocyte cultures

Responder PBMC (Yr; 50 000 cells/well), together with irradiated (3000 rads) stimulator (Yr) PBMC (50 000 cells/well) were incubated for 6 days in 96-well round-bottom plates (Corning) in a final culture medium volume of 170 µl/well. On day 6 of culture, 37 MBq/ml [methyl-³H]thymidine (Nycomed Amersham plc, Little Chalfont, Bucks., UK) (20 µl/well) was added to the wells and the plates were incubated for 8 hr at 37° and 5% CO₂ in a humidified chamber. DNA was harvested on a filter (UniFilter^® GF/C™; Packard Bioscience, Groningen, The Netherlands) using the Packard Harvester Filtermate 196 (Packard Instrument Company, Meriden, CT) and dried. Subsequently, scintillation fluid (Packard Bioscience, Groningen, The Netherlands) was added and the filters were sealed (Packard Plate Sealer Micromate 496; Packard Instrument Company). The incorporated [³H]thymidine was measured using a liquid scintillation counter (Packard Topcount microplate scintillation counter; Packard Bioscience, Groningen, The Netherlands). Counts per minute (c.p.m.) were used as a readout for proliferation.

Results

In allogeneic MLCs, alloreactive CD8⁺ T cells can be visualized by loss of CFSE fluorescence

Stimulator–responder combinations were selected that gave a good proliferative response in classical MLCs. To follow alloantigen-induced cell division, responder PBMC were labelled with CFSE (X_cfse) and then incubated with irradiated, allogeneic (Yr) or autologous (Xr) stimulator cells. On day 1, stimulator cells readily appeared within the CD8⁺ lymphocyte gate as the CFSE⁻ peak in the CFSE histograms (Fig. 1a). At that time, stimulator-derived cells constituted ≈50% of the CD8⁺ T cells in these MLCs (Fig. 1b). However, these CFSE⁻ CD8⁺ T cells were lost from the CD8⁺ lymphocyte gate by day 3 of MLC, presumably as a consequence of the radiation-induced cell death (Fig. 1a, 1b).

Figure 1.

In allogeneic mixed lymphocyte cultures, alloreactive CD8⁺ T cells can be visualized by loss of carboxyfluorescein diacetate succinimidyl ester (CFSE) fluorescence. (a) Histograms of CFSE fluorescence (x-axis, arbitrary units) versus relative cell number (y-axis) of gated CD8^bright lymphocytes from ‘autologous’ (upper panels) or allogeneic mixed lymphocyte cultures (lower panels) after 1, 3, 4, 5 and 6 days of incubation. Responder cells were labelled with CFSE before starting the cultures. (Irradiated stimulator cells were not labelled and therefore appear as the CFSE⁻ peak on day 1 of culture.) (b) Frequencies of CD8⁺ T cells displaying low and negative CFSE fluorescence [see the horizontal marker in the lower right panel of (a)] (y-axis, % of CD8⁺ T cells) versus time (x-axis, days of culture), either stimulated with irradiated autologous peripheral blood mononuclear cells (PBMC) (Xr, ‘autologous’ left panel) or irradiated allogeneic PBMC (Yr, ‘allogeneic’, right panel). Bars represent means; error bars represent standard error of the mean (SEM) of five different allogeneic donor combinations.

From days 4 to 6 of allogeneic MLC, an increasing proportion of CD8⁺ T cells lost CFSE fluorescence, as an apparent consequence of dilution of the dye over dividing cells (Fig. 1a, lower panels; Fig. 1b, right panel). After 6 days, 63% (mean value; range: 41–76%; n=5) of the CD8⁺ T cells were alloreactive CD8⁺ T cells, as measured by loss of CFSE fluorescence. In contrast, no loss of CFSE fluorescence by responder cells was observed in the autologously stimulated cultures (Fig. 1a, upper panels; Fig. 1b, left panel). Thus, for the entire 6-day duration of MLC, responder cells could readily be distinguished from stimulator cells.

A subset of in vitro expanded alloreactive CD8⁺ T cells rapidly produces IFN-γ upon alloantigen-specific restimulation

CFSE labelling of stimulator cells, instead of responder cells, was performed to separate the populations in restimulation experiments (Fig. 2a). When autologous irradiated PBMC were used as stimulators in primary cultures, few CD8⁺ T cells upregulated CD69 expression in conjunction with IFN-γ production (data not shown). Similarly, when autologous PBMC were used for stimulation of alloantigen-expanded CD8⁺ T cells, low numbers of responding cells were detected (Fig. 2c) (mean frequency 0·3%; range: 0·13–0·60%) of CD8⁺ CFSE⁻ responder cells; n=4). In contrast, when the same alloantigen was used for primary and secondary stimulation, frequencies of CD8⁺ T cells were well above autologous background levels (Fig. 2d) (mean frequency 2·40%; range: 0·62–4·64% of CD8⁺ CFSE⁻ responder cells; n=4). Moreover, the alloreactive CD8⁺ T cells were specifically stimulated in an MHC class I restricted manner, as activation could be blocked by addition of the MHC class I-reactive antibody, W6-32 (Fig. 2e) (mean frequency 0·21%; range: 0·09–0·29%). We conclude that, after in vitro expansion and differentiation, allospecific CD8⁺ T cells can be visualized by intracellular cytokine staining and flow cytometry. However, frequencies thus obtained are ≈25-fold lower than the frequencies of allospecific CD8⁺ T cells, as measured by loss of CFSE fluorescence.

Figure 2.

After in vitro expansion and differentiation, a fraction of the alloreactive CD8⁺ T cells can be visualized by intracellular cytokine staining and flow cytometry. (a) In this experiment, stimulator and restimulator cells were labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE), whereas responder cells remained unlabelled. The histogram shows CFSE fluorescence (x-axis, arbitrary units) versus relative cell number (y-axis of gated live lymphocytes and lymphoblasts after 6 days of allogeneic stimulation and 6 hr of restimulation). CFSE⁻ (responder) lymphocytes and lymphoblasts were gated for further analysis (horizontal lines). (b) Histogram of CD8-PerCP fluorescence (x-axis, arbitrary units) versus relative cell number of gated CFSE⁻ (responder) lymphocytes and lymphoblasts. CD8⁺ cells were gated for further analysis (horizontal lines). (c), (d) and (e) Dot-plots of anti-interferon-γ-phycoerythrin (anti-IFN-γ-PE) fluorescence (x-axis, arbitrary units) versus CD69-APC fluorescence (y-axis, arbitrary units) of gated CD8⁺ CFSE⁻ lymphocytes and lymphoblasts after restimulation with CFSE-labelled autologous PBMC (c), CFSE-labelled allogeneic PBMC from the same donor as the primary stimulation (d), or CFSE-labelled allogeneic PBMC from the same donor as the primary stimulation in the presence of anti-human leucocyte antigen (HLA) class I antibodies (W6-32). The percentages displayed represent the percentage of CD69⁺ IFN-γ⁺ cells within CD8⁺ CFSE⁻ (responder X) cells. The panels show one representative experiment from four independent experiments. (f) Summary of frequencies of reactive cells (y-axis, percentage of CD69⁺ IFN-γ⁺ cells within CD8⁺ CFSE⁻ (responder X) cells) after 6 days of allogeneic stimulation and 6 hr of autologous restimulation, allogeneic restimulation, allogeneic restimulation in the presence of W6-32, or restimulation with Staphylococcus aureus Enterotoxin B (SEB).

CD69⁺ T-cell frequencies within the CFSE⁻ CD8⁺ responder T-cell population after stimulation and restimulation were, on average, 19·36% (range: 6·92–31·69%) above the background level.

Differentiation of alloreactive CD8⁺ T cells is related to the number of cell divisions

After 6 days of allogeneic mixed lymphocyte reactions, lymphoblasts were observed in forward scatter/side scatter dot-plots (Fig. 3a). Responder CD8⁺ lymphocytes from these allogeneic MLCs could be divided into CFSE^high cells (no cell divisions), CFSE^low cells (few cell divisions) and CFSE⁻ cells (many cell divisions) (Fig. 3d). The non-dividing cells retained a distribution over CD45RA⁺ and CD45RA⁻ cells which was similar to that of the responder cells on day 1 of culture (results not shown). In the CFSE low and negative compartments, increased frequencies of CD45RA⁻ cells were observed. A subset of CD45RA⁻ CD8⁺ T cells showed upregulation of CD27 expression (Fig. 3c). These cells were located in the alloreactive CD8⁺ T-cell compartment, i.e. cells that lost any degree of CFSE fluorescence [compare Fig. 3c (total), Fig. 3e (CFSE⁻) and Fig. 3f (CFSE^low) with Fig. 3g (CFSE^high)]. Additionally, in the CFSE⁻ population a considerable number of CD27⁻ T cells emerged. Therefore, in MLCs, alloreactive CD8⁺ T cells expand and simultaneously differentiate in CD45RA⁻ memory cells. A subset of these cells acquires a CD27⁻ effector phenotype. T cells need a memory or effector phenotype to rapidly produce IFN-γ.^14,16 We therefore measured IFN-γ production of alloreactive CD8⁺ T cells in relation to the number of cell cycles. Most alloreactive CD8⁺ T cells that produce IFN-γ upon specific restimulation with alloantigen are located in the population of cells that undergo the largest number of cell cycles (Fig. 3i), i.e. the population that:

• is alloreactive, as shown by its vigorous proliferation towards the alloantigen; and

• lacks naive T cells (Fig. 3e).

Figure 3.

While progressing through a number of cell divisions, alloreactive CD8⁺ T cells display a CD45RA⁻ phenotype and a subset loses CD27 expression. (a) In this experiment, responder cells were labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE), whereas stimulator cells remained unlabelled. A dot-plot is shown of forward scatter (x-axis, arbitrary units) versus side scatter (y-axis, arbitrary units) of cells after 6 days in allogeneic mixed lymphocyte culture. Live resting lymphocytes and lymphoblasts were gated for further analysis. (b) Histogram of CD8-PerCP fluorescence (x-axis arbitrary units) versus relative cell number (y-axis) of gated live lymphocytes and lymphoblasts. CD8⁺ lymphocytes were gated for further analysis (horizontal line). (c) Dot-plot of CD27-APC fluorescence (x-axis, arbitrary units) versus CD45RA-RD1 fluorescence (y-axis, arbitrary units) of gated CD8⁺ lymphocytes and lymphoblasts. (d) Histogram of CFSE fluorescence (x-axis, arbitrary units) versus relative cell number (y-axis) of gated CD8⁺ lymphocytes and lymphoblasts. Based on CFSE fluorescence, the CD8⁺ lymphocytes and lymphoblasts were gated in cells that retained high levels of CFSE (no cell division), cells that lost CFSE to some extent (few cell divisions) and cells that became virtually negative for CFSE (many cell divisions) (horizontal lines). (e), (f) and (g) Dot-plots of CD27 expression (x-axis, arbitrary units) versus CD45RA expression (y-axis, arbitrary units) of gated CD8^bright lymphocytes with negative (many cell divisions) (e), low (few cell divisions) (f), or high (no cell divisions) (g) CFSE fluorescence. (h) and (i) Alloreactive CD8⁺ T cells are more likely to produce interferon-γ (IFN-γ) upon specific restimulation after increasing numbers of cell divisions. The CFSE-labelled responder cells (X_cfse) were incubated for 6 days with unlabelled and irradiated allogeneic stimulator cells (Yr). Thereafter, cells were washed and restimulated for 6 hr with unlabelled and irradiated allogeneic stimulator cells from the same donor used in the primary stimulation. Dot-plots are shown of gated CD8^bright lymphocytes and lymphoblasts of allogeneically restimulated cells with respect to anti-IFN-γ-phycoerythrin (PE) fluorescence and CD69-APC fluorescence (h) or CFSE fluorescence and anti-IFN-γ-PE fluorescence (i). The frequency of IFN-γ-producing cells was as follows: generation 0, 0·84%; generation 1, 5·97%; generation 2, 33·21%; generation 3, 10·93%; generation 4, 9·66%; generation 5, 9·31%. In subsequent generations, cells may be contaminated with unlabelled restimulator cells. Therefore, frequencies will become more meaningless with increased loss of CFSE.

Short-term stimulation and analysis of intracellular cytokine production does not allow visualization of alloreactive CD8⁺ T cells

After autologous stimulation of freshly isolated PBMC, low numbers of CD8⁺ T cells became activated when combined CD69 upregulation and IFN-γ production were used as the readout parameter (Fig. 4d). After 6 hr of allogeneic stimulation, using responder/stimulator combinations that yielded positive MLCs, no activated IFN-γ-producing CD8⁺ T cells in peripheral blood were detected from healthy donors at frequencies above background levels (Fig. 4e). In contrast, in all experiments with SEB-stimulated cultures consistently high frequencies of CD69⁺, IFN-γ-producing CD8⁺ T cells were detected (Fig. 4f).

Figure 4.

In healthy individuals, alloreactive CD8⁺ T cells cannot be visualized after 6 hr of mixed lymphocyte culture followed by intracellular cytokine staining and flow cytometry. (a) In this experiment, stimulator cells were labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE), whereas responder cells remained unlabelled. A dot-plot is shown of forward scatter (x-axis, arbitrary units) versus side scatter (y-axis, arbitrary units) of cells from two healthy donors after a 6-hr mixed lymphocyte culture. Lymphocytes were gated for further analysis. (b) Histogram of CFSE fluorescence (x-axis, arbitrary units) versus relative cell number (y-axis) of lymphocytes from the 6-hr mixed lymphocyte culture, gated as shown in panel (a). CFSE⁻ (responder) cells were gated for further analysis. (c) Histogram of CD8-PerCP fluorescence (x-axis, arbitrary units) versus relative cell number (y-axis) of gated CFSE⁻ (responder) lymphocytes from the 6-hr mixed lymphocyte culture. CD8⁺ CFSE⁻ lymphocytes were gated for further analysis (horizontal line). (d), (e) and (f) Anti-interferon-γ-phycoerythrin (anti-IFN-γ-PE) fluorescence (x-axis, arbitrary units) versus CD69-APC fluorescence (y-axis, arbitrary units) of gated CD8⁺ CFSE⁻ (responder) lymphocytes from autologous (d) or allogeneic (e) mixed lymphocyte cultures or cultures in the presence of Staphylococcus aureus Enterotoxin B (SEB) (f). Frequencies of the cells within the quadrangular gates were positive for both IFN-γ and CD69 expression and therefore defined as autoreactive- (d), alloreactive- (e) or SEB reactive- (f) CD8⁺ T-cell frequency, respectively.

As we were able to detect at least a subset of alloreactive CD8⁺ T cells after in vitro expansion and differentiation in the restimulation and intracellular cytokine staining assay, we analysed whether, after in vivo expansion and differentiation, allospecific CD8⁺ T-cell frequencies would become detectable in peripheral blood. Therefore, four renal transplant recipients, with at least one serological mismatch at one of the HLA class-I loci, were studied (Table 1). Two patients were studied longitudinally, within the first 10 days after transplantation, the time period in which sensitization towards the alloantigen is expected to occur. In these patients, the frequencies of alloreactive cells were never significantly above the autologously stimulated background levels (Fig. 5a).

Table 1. Numbers of serologically typed mismatches in the renal transplant recipients studied.

	HLA-A	HLA-B	HLA-DR
pt A	0	1	2
pt B	1	1	1
pt 1	0	1	1
pt 2	1	0	1

HLA, human leucocyte antigen; pt, patient.

Figure 5.

In renal transplant recipients soon or long after transplantation, donor-specific alloreactive CD8⁺ T cells cannot be visualized after 6 hr of mixed lymphocyte culture followed by intracellular cytokine staining and flow cytometry. (a) Experiments similar to those described in Fig. 4 were performed. However, donor-derived, carboxyfluorescein diacetate succinimidyl ester (CFSE)-labelled spleen cells were used as stimulator cells instead of allogeneic CFSE-labelled peripheral blood mononuclear cells (PBMC). Results are expressed as time (x-axis, days after transplantation) versus frequency of reactive cells [y-axis, percentage of CD69⁺ interferon-γ (IFN-γ)⁺ cells within CD8⁺ CFSE⁻ cells] specific for alloantigen (closed circles) or SEB (closed squares). Frequencies of CD69⁺ IFN-γ⁺ cells after autologous negative-control stimulations are depicted in open circles. Data are from two patients (pts) shortly after transplantation: pt A (left panel) and pt B (right panel). Patients A and B were diagnosed as having acute cellular rejection on days 8 and 10, respectively, after transplantation (indicated by arrows). (b) The frequency is shown of reactive cells (y-axis: percentage of CD69⁺ IFNγ⁺ cells within CD8⁺ CFSE⁻ cells) specific for alloantigen (grey bars) or Staphylococcus aureus Enterotoxin B (SEB) (black bars). Frequencies of CD69⁺ IFN-γ⁺ cells after autologous negative-control stimulations are depicted by white bars. Data are from two patients long after transplantation: pt 1 (left) 1165 days after transplantation; and pt 2 (right) 1238 days after transplantation. Neither patient experienced complications and both received basic immunosuppressive therapy. (c) and (d) Representative histogram plots of the frequency of CD8⁺ T cells responsive to autologous PBMC (c) or allogeneic, donor-derived spleen cells (d). Cells were derived from a renal transplant recipient and his kidney donor.

As both patients suffered from acute cellular rejection on days 8 and 10, respectively, after transplantation, we believe that expansion and differentiation of alloantigen-specific T cells should have occurred in vivo. In addition, peripheral blood-derived cells were analysed from two patients long after transplantation, who were receiving only basic immunosuppression and had no history of complications. No alloactivated cells could be detected in the peripheral blood of these two patients, 1163 days (patient 1) and 1238 days (patient 2) after transplantation (Fig. 5b).

To control for proliferation, proliferation assays were performed for CFSE after both autologous and allogeneic proliferation. Similarly to the data shown in Fig. 1, CD8⁺ T cells did proliferate in response to donor-derived spleen cells, but not in response to autologously irradiated PBMC (Figs 5c, 5d).

Discussion

In this study, CFSE labelling was used to document a number of differentiation events occurring in human CD8⁺ T cells upon stimulation with alloantigens in vitro. Although after 6 days of culture the majority of CD8⁺ T cells were alloantigen-reactive T cells that had undergone many cell divisions, we found that only a small fraction of these cells were able to produce IFN-γ after restimulation with the specific alloantigen. The inability to secrete IFN-γ was not caused by hypo- or non-responsiveness of the T-cell receptor (TCR), as most cells upregulated the early activation gene, CD69, after contact with alloantigen-presenting cells. In this respect, it is interesting to note that high availability of IL-2 for CD8⁺ T cells results in strong proliferative allo-induced responses, but concomitantly in poor IFN-γ production upon restimulation.¹⁹ Possibly, the proportion of IFN-γ-producing T cells may increase at later time-points when IL-2 becomes limiting in the culture. In addition, it has been shown that in acute viral infection in mice, IFN-γ production by virus-specific CD8 T-cell responses are rapidly turned off and on in, respectively, the absence or presence of a high-affinity virus-derived peptide in the context of the proper MHC on APCs. In our stimulation system no information is available on this parameter but it is probable that the expanded population contains a mixture of TCRs with high and low affinities for the alloantigenic ligands. The observation that only a minority of the SEB-stimulated, alloantigen-expanded T cells produce IFN-γ argues against a strong influence of the affinity of the TCR on cytokine production in our system. Rather, the SEB response supports the notion that only a fraction of the alloantigen-activated cells acquire the ability to secrete IFN-γ in short-term stimulation assays. In a previous report we demonstrated that alloreactive CD4⁺ T cells are also detectable by IFN-γ production in a similar assay after in vitro expansion.²⁰

After in vitro expansion allospecific CD8⁺ T cells appeared to lose CD45RA expression. With respect to CD27 expression a more complex picture emerged: alloreactive cells had either increased or normal CD27 expression, whereas a subset of cells had lost CD27 expression. The CD45RA⁺ CD27⁻ effector CD8⁺ T-cell phenotype was not encountered in these alloreactive CD8⁺ T-cell populations, which is in line with the notion that the CD45RA⁺ CD27⁻ phenotype emerges when proliferation of in vivo expanded and differentiated cells ceases. It has previously been demonstrated that, in healthy individuals, alloreactive CD8⁺ T cells with proliferative capacity were primarily derived from the naive, CD45RA⁺ T-cell compartment.²¹ Alloreactive CD8⁺ T cells that appeared after 3 days of culture, as detected by their cell division-induced loss of CFSE fluorescence, were virtually all CD45RA⁻, similar to the SEB-reactive IFN-γ- or TNF-α-producing cells after short-term stimulations.²² After 6 days of culture, an average of 63% of the CD8⁺ T cells were allospecific, as defined by their loss of cellular CFSE content.

If PBMC or spleen cells are used as stimulator cells (Y) in flow cytometry-based assays, cells derived from the stimulator will appear in the CD8⁺ T-cell gate. Therefore, the frequency of cells responding with, e.g. IFN-γ production, will be obtained as a frequency both of responder cells (X) and of stimulator cells (Y). Moreover, the stimulator cells may contribute to the IFN-γ-producing cell fraction. Several strategies are available to circumvent this problem and thereby obtain the true frequency of IFN-γ-producing cells as a percentage of responder cells only. To this end, some have employed the use of T-cell-depleted PBMC or spleen cells, or generated Epstein–Barr virus-transformed B-cell lines as stimulator cells,²³ while others argued that this is a virtual problem in their experiments.²⁴ In the present work we chose to employ CFSE-labelled PBMC as stimulator cells. In a stimulation period of up to 48 hr, these cells do not lose CFSE fluorescence and therefore stimulator cells (CFSE⁺) can be distinguished from responder cells (CFSE⁻) in flow cytometric analyses. After short-term (less than 48 hr) allogeneic stimulation, no allospecific CD8⁺ T cells were detected in healthy individuals. This might be a result of the fact that alloantigen-specific precursor cells reside within the CD45RA⁺ population,²¹ which requires a number of cell divisions before being able to secrete IFN-γ (Fig. 3i). However, in sensitized renal transplant recipients alloreactive T cells also remained undetectable in the circulation. These results may be explained in several ways. First, alloreactive CD8⁺ T-cell frequencies may in fact be too low in peripheral blood to be detected by intracellular cytokine staining and flow cytometry. Indeed, published CTLp frequencies specific for, e.g. allogeneic HLA-A or -B antigens after limiting-dilution analyses remain well below 600 per 1 × 10⁶ PBMC.²⁵ Second, as demonstrated using in vitro-expanded alloreactive T cells, only a fraction of the allospecific T cells may produce IFN-γ, which will further limit the detection of these cells. Third, organ-specific alloantigens in the transplant (e.g. kidney) may be absent in spleen cells. Therefore, allospecific CD8⁺ T cells towards kidney-derived alloantigens may be unresponsive to spleen cells from the same donor.¹⁰ Finally, cells may sequester in the allograft, thereby precluding their detection in peripheral blood. In biopsy specimens taken from patients suffering from renal allograft rejections, CD8⁺ T cells were observed by histological examination of the biopsy specimens. A substantial proportion of the infiltrating CD8⁺ T cells were shown to proliferate and to be double positive for both CD45RA and CD45RO isoforms, indicative of a recently activated, ‘transitional’ state.²⁶ Needless to say, the availability of tetrameric complexes specifically binding to alloreactive T cells will settle a number of these points. Alternatively, if immunostimulatory allogeneic peptides or directly activating HLA conformations for a substantial number of human mismatches are revealed, usage of ex vivo IFN-γ production may be revisited, as a relatively simple method for detecting allospecific T-cell frequencies.²⁷