Determinants of in vitro expansion of different human virus-specific FoxP3⁺ regulatory CD8⁺ T cells in chronic hepatitis C virus infection

Abstract

It has been shown previously that suppressive virus-specific FoxP3⁺ regulatory CD8⁺ T cells can be expanded from human peripheral blood mononuclear cells after in vitro antigen-specific stimulation. This study extended this finding by analysing the mechanisms of virus-specific FoxP3⁺ regulatory CD8⁺ T-cell generation during peptide-specific expansion in vitro. It was shown that hepatitis C virus (HCV)-, influenza virus (FLU)-, Epstein–Barr virus (EBV)- and cytomegalovirus (HCMV)-specific FoxP3⁺ regulatory CD8⁺ T cells could be expanded differentially from the blood of chronically HCV-infected patients following in vitro peptide-specific stimulation. The different ability of virus-specific CD8⁺ T-cell populations to express FoxP3 after continuous antigen stimulation in vitro correlated significantly with the ex vivo differentiation status. Indeed, CD27⁺ CD28⁺ CD57⁻ HCV-, FLU- and EBV-specific CD8⁺ T cells displayed a significantly higher ability to give rise to FoxP3⁺ regulatory CD8⁺ T cells compared with CD27⁻ CD28⁻ CD57⁺ HCMV-specific CD8⁺ T cells. Similar T-cell receptor expression patterns of FoxP3⁺ versus FoxP3⁻ CD8⁺ T cells of the same antigen specificity indicated that both cell populations were probably expanded from the same virus-specific CD8⁺ T-cell precursor. In addition, no specific antigen-presenting cell populations were required for the generation of FoxP3⁺ CD8⁺ T cells, as CD8⁺-selected virus-specific FoxP3⁺ CD8⁺ T cells could be expanded by peptide presentation in the absence of antigen-presenting cells. Taken together, these results suggest that the ability to expand FoxP3⁺ regulatory CD8⁺ T cells from virus-specific CD8⁺ T cells differs among distinct virus-specific CD8⁺ T-cell populations depending on the differentiation status.

INTRODUCTION

Several subsets of regulatory T cells have been shown to play important roles in the suppression of antiviral immune responses in humans and mice (Belkaid, 2007; Li et al., 2008). Natural CD25⁺ FoxP3⁺ CD4⁺ regulatory T cells are derived from the thymus and suppress effector T-cell proliferation and cytokine production by direct cell–cell contact. CD25⁺ FoxP3⁺ CD4⁺ regulatory T cells may also be peripherally induced from conventional CD4⁺ T cells during infection (Shevach, 2006; Tang & Bluestone, 2008). Induced CD4⁺ regulatory T cells, like type 1 regulatory T cells and T helper 3 cells, suppress T-cell responses through the secretion of the anti-inflammatory cytokines interleukin (IL)-10 and transforming growth factor (TGF)-β (Belkaid, 2007; Shevach, 2006). A role of CD8⁺ regulatory T cells in the suppression of antiviral immune responses has also been suggested; however, this cell population has not been studied in much detail thus far. In humans, CD8⁺ regulatory T cells with various phenotypes and functional properties have been described (Shevach, 2006). For example, in chronic human immunodeficiency virus (HIV) and hepatitis C virus (HCV) infection, IL-10- or TGF-β-producing regulatory CD8⁺ T cells that suppress cytokine production and proliferation of T cells have been identified (Accapezzato et al., 2004; Alatrakchi et al., 2007; Garba et al., 2002). We have recently shown that the in vitro stimulation of human peripheral blood mononuclear cells (PBMCs) with HCV- and influenza virus (FLU)-specific peptides results in the parallel expansion of two distinct virus-specific CD8⁺ T-cell populations: FoxP3⁻ effector CD8⁺ T cells and FoxP3⁺ regulatory CD8⁺ T cells, which display a suppressive activity (Billerbeck et al., 2007).

In this study, we set out to analyse the presence of virus-specific FoxP3⁺ CD8⁺ T cells, such as HCV-, FLU-, Epstein–Barr virus (EBV)- and human cytomegalovirus (HCMV)-specific CD8⁺ T cells present in the blood of chronically HCV-infected patients ex vivo and after in vitro peptide-specific stimulation. Furthermore, we determined the mechanisms of virus-specific FoxP3⁺ regulatory CD8⁺ T-cell generation during antigen-specific expansion in vitro.

Ex vivo, virus-specific FoxP3⁺ CD8⁺ T cells could not be detected in the blood of chronically HCV-infected patients. However, HCV-, FLU-, EBV- and HCMV-specific FoxP3⁺ regulatory CD8⁺ T cells could be expanded differentially by peptide-specific stimulation in vitro depending on their differentiation status.

These findings extend our previous results and give new insight into the determinants of regulatory CD8⁺ T-cell generation in human virus infections that may be useful for the development of novel therapeutic strategies.

METHODS

Subjects.

Blood samples were obtained from 22 chronically HCV-infected patients (S1–S22) after informed consent and in agreement with federal guidelines and the local ethics committee. HIV infection was excluded in all patients. All patients were HLA-A2-positive. The characteristics of the subjects are summarized in Table 1.

Table 1.

Characteristics of the 22 patients with chronic HCV infection analysed in this study

Patient	Age (years)	Gender	HCV genotype	Detectable virus-specific CD8+ T cells
S1	52	Male	1a	HCV, EBV, HCMV, FLU
S2	42	Male	4	EBV, HCMV, FLU
S3	47	Female	2a	HCMV
S4	64	Female	1	EBV, HCMV
S5	54	Male	4	HCMV
S6	31	Male	3a	EBV, HCMV
S7	54	Female	1b	HCV
S8	43	Female	1	HCV, FLU
S9	39	Female	nd	EBV, HCMV
S10	27	Male	1	EBV, HCMV
S11	44	Male	1b	HCMV, FLU
S12	26	Male	1b	HCMV, FLU
S13	31	Female	3a	EBV, HCMV, FLU
S14	54	Female	1b	EBV, HCMV, FLU
S15	55	Male	1a	HCMV, FLU
S16	48	Female	1	HCMV
S17	50	Male	2b	HCV, EBV, HCMV, FLU
S18	26	Female	1	EBV, HCMV, FLU
S19	29	Female	3a	HCMV, FLU
S20	26	Female	1b	HCV, EBV, HCMV, FLU
S21	54	Male	1b	HCV
S22	46	Male	2b	HCV

nd, Not determined.

PBMCs.

PBMCs were isolated from EDTA blood by Ficoll–Histopaque density-gradient centrifugation (Pancoll). Isolated cells were washed twice in PBS (Gibco) and either analysed immediately or cryopreserved in medium containing 80 % fetal calf serum (Gibco), 10 % RPMI 1640 (Gibco) and 10 % DMSO (Sigma-Aldrich).

Peptides and HLA-A2 tetramers.

HCV-, FLU-, EBV- and HCMV-derived peptides, previously shown to be HLA-A2-restricted epitopes, were purchased from Biosynthan. The amino acid sequences of the HLA-A2-restricted HCV-, FLU-, EBV- and HCMV-specific epitopes were as follows: ALYDVVTKL for HCV NS5B protein (aa 2594–2602); CINGVCWTV for HCV NS3 protein (aa 1073–1081); KLVALGINAV for HCV NS3 protein (aa 1406–1415); GILGFVFTL for FLU matrix protein (aa 58–66); GLCTLVAML for EBV BMLF-1 protein (aa 280–288) and NLVPMVATV for HCMV pp65 protein (aa 495–503). HLA-A2 tetramers corresponding to the peptides were provided from the NIH Tetramer Facility, MD, USA, for HCV and from ProImmune for FLU, EBV and HCMV. HCV-specific peptides and tetramers were based on the HCV genotype 1 sequence.

Antibodies.

Peridinin–chlorophyll–protein complex (PerCP)-conjugated anti-CD8, phycoerythrin (PE)-conjugated anti-CD8, anti-CD4–PerCP, fluorescein isothiocyanate (FITC)-conjugated anti-CD38, isotype PE, isotype FITC and isotype allophycocyanin (APC) were obtained from BD Pharmingen. Anti-CCR7–FITC was obtained from R&D Systems. Anti-FoxP3–PE and anti-FoxP3–APC (antibody PCH101) were purchased from eBioscience. Anti-FoxP3–PE (antibody 259D) was purchased from BioLegend. Anti-CD27–FITC was purchased from Hoelzel, anti-CD28–FITC was obtained from Diaclone, anti-CD127–PE and anti-CD57–PE from Beckman Coulter.

Antigen-specific T-cell proliferation.

PBMCs were stimulated and analysed as described previously (Billerbeck et al., 2007) by using 10 μg synthetic HCV, FLU, EBV or HCMV peptide ml⁻¹, 0.5 μg anti-human CD28 (BD Pharmingen) ml⁻¹, and 0 or 500 U human recombinant IL-2 (Hoffmann La Roche) ml⁻¹ for stimulation. For some experiments, cells were labelled with carboxyfluorescein diacetate succinimidyl ester (CFSE; Molecular Probes) prior to antigen-specific stimulation.

Tetramer and antibody staining, and intracellular FoxP3 staining.

Cells were stained with HLA-A2 tetramers and surface antibodies as described previously (Billerbeck et al., 2007). Intracellular FoxP3 staining was performed by using the eBioscience FoxP3 staining buffer set for antibody PCH101 or the BioLegend FoxP3 Flow kit for antibody 259D according to the manufacturers' instructions. For fluorescence-activated cell sorting (FACS) analysis, all stained cells were fixed in 2 % paraformaldehyde. FACS analysis was performed by using a BD FACSCanto II flow cytometer and FlowJo software (Tree Star).

Isolation of lymphocyte subsets.

A MidiMACS separation system (Miltenyi Biotech) was used for the isolation or depletion of lymphocyte subsets according to the manufacturer's instructions. For the isolation of antigen-specific CD8⁺ T cells, peptide-stimulated PBMCs were stained with the corresponding APC-conjugated HLA-A2 tetramer and then incubated with APC-conjugated microbeads prior to positive selection. CD8⁺ T cells were isolated by using CD8-conjugated microbeads.

Determination of the T-cell receptor (TCR) Vβ repertoire.

Tetramer (Tet⁺) CD8⁺ T cells were isolated from peptide-expanded T-cell lines by positive selection as described above. Subsequently, isolated Tet⁺ CD8⁺ T cells were stained with TCR Vβ antibodies using the IOTest Beta Mark TCR Vβ repertoire kit from Beckman Coulter followed by intracellular FoxP3 staining.

Suppression assays.

Suppression assays were performed as described previously (Billerbeck et al., 2007). Briefly, HCV, FLU or HCMV Tet⁺ CD8⁺ T cells were isolated from PBMCs after 14 days peptide-specific expansion in the presence of 500 U IL-2 ml⁻¹. For some experiments, PBMCs from a single subject were cultured in the absence of exogenous IL-2 or in the presence of 500 U IL-2 ml⁻¹ to obtain Tet⁺ CD8⁺ T cells with different FoxP3 expression. To assess the suppressive capacity of the cells, Tet⁺ CD8⁺ T cells with significant FoxP3 expression after isolation and Tet⁺ CD8⁺ T cells negative for FoxP3 after isolation were co-cultured with autologous PBMCs that were labelled with CFSE and stimulated with 0.04 μg human anti-CD3 (Immunotech) ml⁻¹ at ratios of 1 : 1, 1 : 2 or 1 : 4 (suppressor : effector) for 7 days in a 96-well plate.

Statistical analysis.

An unpaired Student`s t-test and Spearman's rank correlation were performed by using GraphPad Prism version 4 (GraphPad Software).

RESULTS

Analysis of FoxP3 expression in different virus-specific CD8⁺ T cells from chronically HCV-infected patients ex vivo and after in vitro peptide stimulation

We have shown previously that HCV- and FLU-specific memory CD8⁺ T cells can be induced to express FoxP3 after in vitro peptide-specific stimulation in the presence of high doses of IL-2 (500 U ml⁻¹) (Billerbeck et al., 2007). To determine whether FoxP3 expression could also be induced in other virus-specific CD8⁺ T-cell populations, such as herpesviruses, we analysed the expression of FoxP3 in EBV- and HCMV-specific CD8⁺ T cells, as well as in HCV- and FLU-specific CD8⁺ T cells from the PBMCs of 22 chronically HCV-infected patients ex vivo and after 7 and 14 days peptide-specific stimulation in the presence of 500 U IL-2 ml⁻¹. Of the 22 patients with chronic HCV infection, seven showed HCV-specific T-cell responses for further analysis. None of these seven patients showed HCV-specific CD8⁺ T-cell responses to more than one HCV-specific epitope. Chronically HCV-infected patients without detectable HCV-specific CD8⁺ T-cell responses were analysed for FLU-, EBV- and HCMV-specific CD8⁺ T-cell responses only (Table 1).

Consistent with our previous results, we did not find virus-specific FoxP3⁺ CD8⁺ T cells ex vivo (Fig. 1a); however, a large fraction (median HCV, 66 %; median FLU, 45 %) of HCV- and FLU-specific CD8⁺ T cells expressed FoxP3 after 7 (Fig. 1a) and 14 (Fig. 1a, b) days peptide-specific stimulation. Interestingly, the ability of EBV-specific CD8⁺ T cells to express FoxP3 after 7 and 14 days EBV-specific stimulation was comparable to those of HCV and FLU-specific CD8⁺ T cells (median, 38 %). In contrast, the induction of FoxP3 expression in HCMV-specific CD8⁺ T cells was significantly lower compared with HCV-, FLU- and EBV-specific CD8⁺ T cells (median, 10 %; Fig. 1a, b).

Fig. 1.

Differential induction of FoxP3 expression in HCV-, FLU-, EBV- and HCMV-specific CD8⁺ T cells after in vitro peptide stimulation. (a) PBMCs were analysed for FoxP3 expression (%) in virus-specific CD8⁺ T cells ex vivo and after 7 or 14 days stimulation with either HCV-, FLU-, EBV- or HCMV-specific peptides in the presence of 500 U IL-2 ml⁻¹. Plots were gated on CD8⁺ T cells and show representative results from the indicated subjects. (b) FoxP3 expression in HCV-, FLU-, EBV- and HCMV-specific CD8⁺ T cells from 22 chronically HCV-infected patients after 14 days antigen-specific expansion in the presence of 500 U IL-2 ml⁻¹. An unpaired Student's t-test was used to analyse statistically significant differences (HCV vs HCMV, P=0.0007; FLU vs HCMV, P=0.03).

These results suggested that the ability to express FoxP3 after in vitro expansion differs among distinct virus-specific CD8⁺ T-cell populations.

Expression of FoxP3 correlates with a suppressive phenotype in virus-specific CD8⁺ T cells

To analyse whether the levels of FoxP3 expression in peptide-expanded HCV-, FLU- and HCMV-specific CD8⁺ T cells were associated with different regulatory functions, we performed suppression assays. Specifically, we stimulated PBMCs of several subjects with HCV-, FLU- or HCMV-specific peptide in the presence of 500 U IL-2 ml⁻¹. After 14 days peptide-specific expansion, Tet⁺ CD8⁺ T cells were isolated from the T-cell lines by using magnetic-bead separation. As shown in Fig. 2(a), HCV- and FLU-specific CD8⁺ T cells largely expressed FoxP3 pre- and post-tetramer positive selection, whilst HCMV-specific CD8⁺ T cells were predominantly negative for FoxP3. To determine the suppressive capacity of these isolated virus-specific CD8⁺ T cells, they were co-cultured at ratios of 1 : 1; 1 : 2 and 1 : 4 for 7 days with autologous CFSE-labelled PBMCs that were stimulated with anti-CD3. As shown in Fig. 2(b), the addition of isolated HCV- and FLU-specific CD8⁺ T cells with FoxP3 expression of 50.5 and 40.4 %, respectively, resulted in a significant dose-dependent suppression of anti-CD3-induced T-cell proliferation. In contrast, the addition of isolated HCMV-specific CD8⁺ T cells with a FoxP3 expression of 12.5 % resulted in minimal suppression of T-cell proliferation. To prove that the suppressive phenotype of virus-specific CD8⁺ T cells was indeed associated with the induction of FoxP3 expression and not limited to HCV- and FLU-specific CD8⁺ T cells, we further determined the suppressive capacity of FoxP3⁺ versus FoxP3⁻ HCMV-specific CD8⁺ T cells. For these experiments, we analysed the PBMCs from subjects with significant FoxP3⁺ expression in HCMV-specific CD8⁺ T cells after peptide-specific expansion in the presence of 500 U IL-2 ml⁻¹. To obtain HCMV-specific FoxP3⁻ CD8⁺ T cells from the same subjects, we expanded cells in the absence of exogenous IL-2. As shown in Fig. 2(c), significantly stronger inhibition of T-cell proliferation was mediated by isolated HCMV-specific CD8⁺ T cells with 31 % FoxP3 expression compared with HCMV-specific CD8⁺ T cells with 12 % FoxP3 expression. These results indicated that, irrespective of virus specificity, the induction of FoxP3 correlated with the suppressive activity of CD8⁺ T cells in a dose-dependent manner.

Fig. 2.

FoxP3 expression correlates with a suppressive phenotype in virus-specific CD8⁺ T cells. (a) To assess the suppressive activity of FoxP3⁺ virus-specific CD8⁺ T cells, HCV-, FLU- and HCMV-specific CD8⁺ T cells with different levels of FoxP3 expression were isolated from peptide-specific T-cell lines. Plots from representative experiments and the indicated subjects show FoxP3 expression (%) in HCV-, FLU- and HCMV-specific CD8⁺ T cells pre- and post-selection. (b) Isolated HCV (▪)-, FLU (⧫)- and HCMV (▴)-specific CD8⁺ T cells (50.5, 40.4 and 12.5 % FoxP3 expression, respectively) were co-cultured at ratios of 1 : 1, 1 : 2 and 1 : 4 with autologous CFSE-labelled PBMCs that had been stimulated with anti-CD3. After 7 days culture, the inhibition of T-cell proliferation was analysed. (c) Isolated HCMV-specific CD8⁺ T cells with significant FoxP3 expression (31 %; FoxP3⁺, ⧫) and HCMV-specific CD8⁺ T cells with low FoxP3 expression (12 %; FoxP3⁻, ▪) from the same representative subject (S15) were co-cultured at ratios of 1 : 1, 1 : 2 and 1 : 4 with autologous CFSE-labelled PBMCs that had been stimulated with anti-CD3. After 7 days culture, the inhibition of T-cell proliferation was analysed. The results from one representative experiment are shown.

The ex vivo phenotypic differentiation stage of virus-specific CD8⁺ T cells is associated with their ability to express FoxP3 after in vitro peptide-specific stimulation

Next, we investigated whether the apparent difference between virus-specific in vitro expansions of FoxP3⁺ regulatory CD8⁺ T cells was defined by the memory phenotype of virus-specific CD8⁺ T cells ex vivo. It has been described previously that antigen-specific memory CD8⁺ T cells in different human virus infections exhibit distinct phenotypic differentiation patterns (Appay & Rowland-Jones, 2004; Joshi & Kaech, 2008; Radziewicz et al., 2007; Wherry & Ahmed, 2004). To determine whether different phenotypic profiles may indeed correlate with the ability of virus-specific CD8⁺ T cells to express FoxP3 after in vitro stimulation, we performed an ex vivo surface-marker expression analysis of all virus-specific CD8⁺ T cells from our 22 chronically HCV-infected patients prior to peptide-specific stimulation. In agreement with several studies (Bengsch et al., 2007; Radziewicz et al., 2007; van Leeuwen et al., 2005), HCV- and FLU-specific CD8⁺ T cells from most subjects displayed an early differentiation phenotype that was characterized by expression of the co-stimulatory molecules CD28 and CD27 (Fig. 3a), the chemokine receptor CCR7 and the IL-7 receptor chain α (CD127) (data not shown). HCV- and FLU-specific CD8⁺ T cells did not express CD57 (Fig. 3a), a marker of replicative senescence that has been reported to be expressed on virus-specific CD8⁺ T cells with a late differentiation phenotype (Brenchley et al., 2008). EBV-specific CD8⁺ T cells expressed similar levels of CD27, but reduced levels of CD28, CCR7 and CD127 and higher levels of CD57 compared with HCV- and FLU-specific CD8⁺ T cells (Fig. 3a and data not shown) (Hislop et al., 2007), indicating that these cells displayed an intermediate differentiation stage. HCMV-specific CD8⁺ T cells, in contrast, were characterized by low expression of CD27, CD28, CCR7 and CD127 but high expression of CD57 (Fig. 3a and data not shown) (Brenchley et al., 2008; Gillespie et al., 2000; Rufer et al., 2003). Thus, the reduced ability of expanded HCMV-specific CD8⁺ T cells to express FoxP3 correlated with a more terminally differentiated status of these cells. Indeed, as shown in Fig. 3(b), a significant positive correlation was observed between the ex vivo expression of CD28 (r=0.37, P=0.014) and CD27 (r=0.44, P=0.0031) on virus-specific CD8⁺ T cells and FoxP3 expression in the same cells after 7 days antigen-specific stimulation. In contrast, a statistically significant negative correlation (r=−0.5, P=0.0013) was observed between the expression of CD57 and the expression of FoxP3 after 7 days in vitro peptide-specific expansion (Fig. 3b). The expression of CD57 on virus-specific CD8⁺ T cells has been associated with a low proliferative capacity of these cells. To examine whether the inability of HCMV-specific CD8⁺ T cells to express FoxP3 was due to their inability to divide rather than to their differentiation stage, we analysed the proliferation of virus-specific CD8⁺ T cells from several patients by labelling PBMCs with CSFE prior to peptide-specific stimulation for 7 days. As shown in Fig. 3(c) both HCV- and HCMV-specific CD8⁺ T cells showed antigen-specific expansion with CFSE dilution. However, whilst a subpopulation of dividing HCV-specific CD8⁺ T cells expressed FoxP3, dividing HCMV-specific CD8⁺ T cells were largely negative for FoxP3. Taken together, these results suggested that the ex vivo differentiation stage rather than the proliferative capacity of virus-specific CD8⁺ T cells was indeed associated with their ability to express FoxP3 after in vitro peptide-specific stimulation.

Fig. 3.

The ex vivo differentiation stage of virus-specific CD8⁺ T cells is associated with their ability to express FoxP3 after in vitro peptide-specific stimulation. (a) The expression of CD28, CD27 and CD57 on HCV-, FLU-, EBV- and HCMV-specific CD8⁺ T cells from the PBMCs of 22 subjects was analysed ex vivo. (b) The ex vivo data shown in (a) were correlated with FoxP3 expression in the same cells after 7 days peptide-specific stimulation in the presence of 500 U IL- 2 ml⁻¹. Spearman's rank correlation coefficient was used to determine statistically significant correlations. (c) PBMCs were labelled with 5 μM CFSE and stimulated with HCV- or HCMV-specific peptides in the presence of 500 U IL-2 ml⁻¹. After 1 week culture, virus-specific cells were analysed for their proliferative capacity and for FoxP3 expression (%). Representative dot plots for HCV- and HCMV-specific proliferation of the indicated subjects are shown. The upper plots were gated on CD8⁺ T cells and the lower plots on Tet⁺ CD8⁺ T cells.

Virus-specific FoxP3⁻ and FoxP3⁺ CD8⁺ T cells express the same TCR Vβ repertoire and can be expanded in the absence of antigen-presenting cells

The strong correlation between the in vitro induction of FoxP3 expression in CD8⁺ T cells and the ex vivo differentiation phenotype of virus-specific CD8⁺ T cells raised the possibility that FoxP3⁺ and FoxP3⁻ CD8⁺ T cells are generated from the same virus-specific CD8⁺ T cells rather than from separate phenotypic and functional CD8⁺ T-cell lineages. To address this issue, we determined the clonal relationship of virus-specific FoxP3⁻ and FoxP3⁺ CD8⁺ T cells targeting the same antigen by comparing the TCR Vβ repertoire of both cell subsets. For these experiments, we stimulated PBMCs of several subjects with HCV-, FLU-, EBV- or HCMV-specific peptide in the presence of 500 U IL-2 ml⁻¹. After 14 days peptide-specific expansion, Tet⁺ CD8⁺ T cells were isolated from the T-cell lines and analysed for FoxP3 and TCR Vβ repertoire expression. As shown in Fig. 4(a), peptide-expanded HCV-, FLU-, EBV- and HCMV-specific FoxP3⁻ and FoxP3⁺ CD8⁺ T cells from the same subjects expressed a largely similar TCR Vβ repertoire. These findings indicated a great clonal homology between both CD8⁺ T-cell subsets.

Fig. 4.

FoxP3⁻ and FoxP3⁺ virus-specific CD8⁺ T cells express the same TCR Vβ repertoire and can be expanded in the absence of antigen-presenting cells. (a) Virus-specific CD8⁺ T cells isolated from peptide-expanded T-cell lines were stained with anti-TCR Vβ antibodies and anti-FoxP3 antibody. The TCR Vβ repertoire usage of virus-specific FoxP3⁺ (open bars) and FoxP3⁻ (shaded bars) CD8⁺ T cells from the indicated subjects are shown. (b) Isolated CD8⁺ T cells (open bars) or whole PBMCs (filled bars) of several subjects were stimulated with HCV-, FLU- or HCMV-specific peptides in the presence of 100 U IL-2 ml⁻¹. FoxP3 expression in virus-specific CD8⁺ T cells was analysed after 7 days culture. Results from the indicated subjects are shown. (c) Representative plots from the indicated subjects show FoxP3 expression (%) in HCV- and HCMV-specific CD8⁺ T cells expanded from PBMCs or from isolated CD8⁺ T cells.

We further addressed the mode of antigen presentation necessary for the expansion of virus-specific FoxP3⁻ and FoxP3⁺ CD8⁺ T cells. To test whether antigen-presenting cells played a role in the induction of virus-specific FoxP3⁺ CD8⁺ T cells, we isolated CD8⁺ T cells from PBMCs of several subjects, stimulated them with virus-specific peptides and 100 U IL-2 ml⁻¹ in the absence of antigen-presenting cells and assessed the expression of FoxP3 in virus-specific CD8⁺ T cells after 7 days culture. As a control, we analysed FoxP3 expression of whole PBMCs after peptide-specific stimulation. As shown in Fig. 4(b, c), HCV-, FLU- or HCMV-specific stimulation of isolated CD8⁺ T cells alone resulted in significant induction of FoxP3 expression in HCV-, FLU- and HCMV-specific CD8⁺ T cells (44, 41 and 40 %, respectively). These results showed that peptide presentation on MHC class I molecules on the CD8⁺ T cells in the presence of IL-2 was sufficient to expand virus-specific FoxP3⁺ CD8⁺ T cells.

DISCUSSION

In this study, we showed that FoxP3⁺ regulatory CD8⁺ T cells can be differentially expanded from distinct virus-specific memory CD8⁺ T cells, such as HCV-, FLU-, EBV- and HCMV-specific CD8⁺ T cells derived from the blood of chronically HCV-infected patients following peptide-specific stimulation in vitro.

Importantly, the in vitro induction of FoxP3 expression in virus-specific CD8⁺ T cells was associated with a suppressive phenotype. In contrast to our results, some recent studies have indicated that FoxP3 expression and regulatory functions do not necessarily correlate in human CD4⁺ and CD8⁺ T cells (Campbell & Ziegler, 2007; Gavin et al., 2006; Roncarolo & Gregori, 2008; Tran et al., 2007). However, the T cells in these studies were stimulated in an antigen-non-specific manner with anti-CD3 and anti-CD28, which led to only a transient upregulation of FoxP3 (Gavin et al., 2006; Wang et al., 2007). Thus, antigen-specific expansion of virus-specific T cells may induce FoxP3 expression and suppressive functions in a mechanism different from that of transient FoxP3 upregulation induced by anti-CD3/anti-CD28 stimulation. In line with this and in agreement with our results, Mahic et al. (2008) recently showed that continuous antigen stimulation with Staphylococcus enterotoxin B can generate CD25⁺ FoxP3⁺ CD8⁺ T cells with a suppressive capacity. In this context, it is also important to note that the in vitro induction of suppressive virus-specific T cells was not limited to the CD8⁺ T-cell subset. Indeed, a recent study by Ebinuma et al. (2008) reported the expansion of HCV- and FLU-specific regulatory CD25⁺ FoxP3⁺ CD4⁺ T cells from the blood of chronically HCV-infected patients after in vitro antigen-specific stimulation.

Our study showed differential expression of FoxP3 in distinct virus-specific CD8⁺ T cells after in vitro peptide-specific stimulation. Indeed, HCV- and FLU- as well as EBV-specific CD8⁺ T cells had a significant higher ability to give rise to FoxP3⁺ regulatory CD8⁺ T cells compared with HCMV-specific CD8⁺ T cells. This finding may reflect specific conditions associated with HCV, FLU, EBV or HCMV infection in vivo, such as differences in compartmentalization of virus replication, virus tropism, antigen presentation and antigen load. However, clearance, persistence and latency of these viruses do not seem to be major determinants for the expansion of virus-specific regulatory FoxP3⁺ CD8⁺ T cells. Indeed, T cells derived from patients with resolved FLU infection, as well as chronic HCV infection or latent EBV infection, all gave rise to virus-specific FoxP3⁺ CD8⁺ T cells.

Interestingly, we found that the ability of virus-specific CD8⁺ T cells to express FoxP3 was statistically significantly associated with the ex vivo phenotypic differentiation stage of these cells. Indeed, our results showed that early- and intermediate-differentiated HCV-, FLU- and EBV-specific CD27⁺ CD28⁺ CD57⁻ CD8⁺ T cells gave rise to virus-specific FoxP3⁺ CD8⁺ T cells, whereas late-differentiated HCMV-specific CD27⁻ CD28⁻ CD57⁺ CD8⁺ T cells were largely FoxP3-negative after in vitro stimulation. It has been shown that CD27⁺ CD28⁺ CD57⁻ CD8⁺ T cells still have a significant capacity to differentiate and proliferate compared with CD27⁻ CD28⁻ CD57⁺ CD8⁺ T cells, which exhibit only a reduced proliferative capacity (Wherry & Ahmed, 2004; Wherry et al., 2003). We have shown previously that proliferation is a prerequisite for the development of FoxP3⁺ CD8⁺ T cells (Billerbeck et al., 2007). However, here we found that even those CD27⁻ CD28⁻ CD57⁺ HCMV-specific CD8⁺ T cells that proliferated significantly following in vitro stimulation failed to express FoxP3, indicating that the inability of HCMV-specific CD8⁺ T cells to express FoxP3 is not necessarily due to their low proliferative capacity. Taken together, our data suggest that the distinct differentiation stage of memory CD8⁺ T cells might be one major determinant for the differential capacity to generate FoxP3⁺ CD8⁺ T cells from virus-specific CD8⁺ T cells. Nevertheless, some patients exhibited significant FoxP3 expression in HCMV-specific CD8⁺ T cells, regardless of their terminal differentiation stage. This observation indicates that T-cell differentiation status is not an exclusive factor for the ability of virus-specific CD8⁺ T cells to express FoxP3.

An important question arising from our study is the origin of the virus-specific FoxP3⁺ regulatory CD8⁺ T cells. These cells might be generated in parallel with FoxP3⁻ cells from the same virus-specific CD8⁺ T cell, or they might derive from a distinct virus-specific CD8⁺ T-cell lineage. Of note, we found very similar TCR Vβ expression patterns of FoxP3⁻ and FoxP3⁺ CD8⁺ T cells generated from virus-specific CD8⁺ T cells after in vitro stimulation. These data suggested that FoxP3⁺ and FoxP3⁻ CD8⁺ T cells might rather be generated from the same virus-specific CD8⁺ T cells rather than from separate virus-specific CD8⁺ T-cell lineages. Consistent with our results, a recent study demonstrated the antigen-specific expansion of human CD25⁺ FoxP3⁺ regulatory CD4⁺ T cells as well as CD25⁻ FoxP3⁻ effector CD4⁺ T cells with close TCR clonal homology from the memory CD4⁺ T-cell population (Akbar et al., 2007).

The biological relevance of our findings remains largely unknown. However, it is possible that the generation of virus-specific FoxP3⁺ regulatory T cells might be a natural process to limit virus-specific effector T-cell responses and overwhelming tissue damage. Of note, we did not find virus-specific FoxP3⁺ CD8⁺ T cells in the peripheral blood of chronically HCV-infected patients ex vivo. Possible explanations for this finding include a preferential induction and accumulation of these cells at the site of infection during the chronic phase of infection, for example the liver in HCV infection or the lymph nodes in EBV or HCMV infection. It might also be possible that virus-specific FoxP3⁺ regulatory CD8⁺ T cells are primarily generated during the acute phase of infection. It is noteworthy that two recent studies showing the expansion of CD25⁺ FoxP3⁺ CD8⁺ T cells during acute simian immunodeficiency virus infection in cynomolgus macaques support the hypothesis of preferential induction of virus-specific FoxP3⁺ CD8⁺ T cells during acute virus infections (Malleret et al., 2008). Whether virus-specific FoxP3⁺ regulatory CD8⁺ T cells might also be induced during acute HCV infection, however, remains to be determined in future studies.

In summary, the results of our study extend our previous findings by giving important new insights into the mechanisms of virus-specific FoxP3⁺ regulatory CD8⁺ T-cell generation during in vitro expansion. Knowledge about the determinants of the induction of virus-specific FoxP3⁺ regulatory CD8⁺ T cells is a prerequisite for the potential application of these cells for future therapeutic interventions, such as adoptive cell transfer, during virus infections.