A subset of functional effector-memory CD8+ T lymphocytes in human immunodeficiency virus-infected patients

Authors


Dr G. Herbein, Department of Virology, Franche-Comte School of Medicine, 2 place Saint-Jacques, F-25030 Besançon, France.
Email: gherbein@chu-besancon.fr
Senior author: Dr G. Herbein

Summary

CD8+ T cells provide protective immune responses via both cytolytic and non-cytolytic mechanisms in subjects infected with human immunodeficiency virus (HIV). In the present study, we investigated the CD28 expression of CD8+ T cells present in the peripheral blood lymphocyte subset isolated from chronically HIV-infected subjects. Using flow cytometric analysis, a continuous spectrum of CD28 intensity ranging from negative to high, which could be separated into CD28-negative, intermediate (int) and high, was seen for CD8+ T cells. Our study focused mostly on the CD28int CD8+ T cells. The CD28int CD8+ T cells are CD57 CD27+ CD45RO+ CD45RA CCR7low CD62Lint. The proliferative capacity of CD28int CD8+ T cells was intermediate between those of CD28 CD8+ T cells and CD28high CD8+ T cells. The CD28int CD8+ T cells are specific for HIV, cytomegalovirus (CMV) and Epstein–Barr virus (EBV) antigens as measured by human leucocyte antigen pentamer binding and produce both intracellular interferon-γ and tumour necrosis factor-α in response to their cognate viral peptides. The CD28int CD8+ T cells have HIV-specific, CMV-specific and EBV-specific cytotoxic activity in response to their cognate viral peptides. These findings indicate that a subset of functional effector-memory CD8+ T cells specific for HIV, CMV and EBV antigens may contribute to an efficient immune response in HIV-infected subjects.

Abbreviations:
BSA

bovine serum albumin

CD

cluster of differentiation

CMV

cytomegalovirus

CTL

cytotoxic T lymphocyte

ELISPOT

enzyme-linked immunospot assay

EBV

Epstein–Barr virus

FITC

fluorescein isothiocyanate

FSC

forward scatter

HIV

human immunodeficiency virus

HLA

human leucocyte antigen

IFN-γ

interferon-γ

IL-2

interleukin-2

int

intermediate

mAb

monoclonal antibody

PFU

particle-forming unit

PBL

peripheral blood lymphocyte

PBMC

peripheral blood mononuclear cells

PBS

phosphate-buffered saline

PE

phycoerythrin

PC5

phycoerythrin–cyanine 5

SSC

side scatter

TCR

T-cell receptor

TNF-α

tumour necrosis factor-α

Introduction

Abundant evidence indicates that CD8+ T cells provide protective immune responses in human immunodeficiency virus (HIV) infection. CD8+ T cells from infected individuals have been shown to suppress HIV replication in autologous CD4+ T cells via both cytolytic and non-cytolytic mechanisms.1,2 HIV-specific cytotoxic T lymphocytes (CTLs) have been identified in exposed but uninfected individuals, and antiviral cytotoxic activity is correlated with the clearance of viraemia in primary infection.3

Recent studies have shown that the expression of not only costimulatory receptor CD28, but also CD27, is associated with different stages of T-cell differentiation.4,5 This has led to a putative model of human CD8+ T-cell differentiation beginning with CD28+ CD27+ T cells, seen as precursors or early differentiated cells, progressing through to CD28 CD27 T cells, thought to be fully differentiated T cells exhibiting shorter telomere length.6 Based on CD28 CD27 expression, CD8+ T cells have been recently separated into three distinct populations: CD28+ CD27+ or early differentiated cells, CD28 CD27+ or intermediately differentiated cells and CD28 CD27 or late-differentiated cells.5–7 CD28+ CD8+ T cells have greater proliferative capacity following T-cell receptor (TCR) stimulation, secrete interleukin-2 (IL-2), produce soluble factors that suppress HIV replication, resist T-cell death, express the chemokine receptor CCR7 and the lymphoid organ-homing marker CD62L, and comprise more than 95% of the CD8+ T cells in normal human lymphoid tissues.8 In contrast, CD28 CD8+ T cells have poor proliferative capacity, secrete low levels of IL-2, express high levels of Fas, are prone to undergo apoptosis, are oligoclonally expanded, express neither CCR7 nor CD62L, and are detected mostly in peripheral blood of HIV-infected subjects.7,9

Although the role of CD8+ T cells in the control of HIV infection has been broadly studied, very little information is available on the CD8+ T-cell subset(s) depleted during HIV infection. Several hypotheses have been advanced to account for the decreased HIV-specific CD8+ T-cell response observed in HIV-infected subjects. First, functional defects and skewed maturation of HIV-specific CD8+ T cells have been reported.7,10–12 Second, the direct depletion of HIV-specific CD8+ T cells by apoptosis or programmed cell death could also contribute to the immune suppression observed in HIV-infected patients.13–16 Third, HIV may adversely affect CTL priming by dendritic cells and/or CD4+ T-cell help in primary infection, resulting in an unfavourable cytokine environment and an altered differentiation of cytotoxic CD8+ T cells.16,17

We observed that a continuous spectrum of CD28 ranging from negative to high, which could be separated into negative, intermediate (int) and high, was seen for CD8+ T cells present in the peripheral blood of chronically HIV-infected subjects. Our data indicate that the CD28int CD8+ T cells are functional effector-memory cells specific for HIV, Epstein–Barr virus (EBV) and cytomegalovirus (CMV) antigens.

Materials and methods

Patients

Whole peripheral blood was harvested from 67 chronically HIV-infected patients being treated with highly active antiretroviral therapy and 14 healthy HIV-negative donors. HIV-infected donors were recruited from the Department of Dermatology and the Department of Infectious Diseases, University of Franche-Comte School of Medicine (Besançon, France). HIV-infected patients had CD4+ T-cell counts that ranged between 151 and 1 636 cells/ml and their viral load ranged between 1·3 log and 5·25 log. Fifteen HIV-infected donors were human leucocyte antigen (HLA)-A*02 and 10 were HLA-A*03. Control subjects were age- and gender-matched healthy volunteers recruited in the French Blood Centre (EFS, Besançon, France). All subjects gave informed consent before entry into this study, and all studies were approved by the local institutional review board.

Isolation and culture of PBMC and PBL

Peripheral blood mononuclear cells (PBMC) and purified peripheral blood lymphocytes (PBL) were prepared from the peripheral blood of HIV-infected subjects and healthy donors, as previously described.18

Reagents

Phycoerythrin–cyanine-5-conjugated anti-CD8 (anti-CD8-PC5), anti-CD28-PC5 and fluorescein isothiocyanate-conjugated anti-CD28 (anti-CD28-FITC) monoclonal antibodies (mAbs) were purchased from Beckman Coulter (Marseille, France). Phycoerythrin-conjugated anti-CD8 (anti-CD8-PE), anti-CD27-FITC, anti-CD45RA-FITC, anti-CD45RO-PE, anti-perforin-FITC, anti-tumour necrosis factor-α (TNF-α)-FITC, anti-CD38-PE, anti-CD57-FITC, anti-CD62L-FITC and anti-CD107a-PE mAbs were purchased from Becton Dickinson (Franklin Lakes, NJ). The three CD28 CD8 T-cell subsets was distinguished using 0·5 μg anti-human CD28 (huCD28) antibody per 106 cells. Anti-CCR7-FITC and anti-interferon-γ (IFN-γ)-FITC mAbs were purchased from R & D Systems (Minneapolis, MN). Isotype control antibodies were purchased from Becton Dickinson. Anti-Ki67-FITC was purchased from Becton Dickinson and was used as previously described.19 The anti-CD3 mAb was a gift from Dr U. Mahlknecht (University of Heidelberg, Germany) and was used for TCR stimulation as previously reported.19 The following HLA-A*0201 restricted peptides were used: HIV Pol 476–484 ILKEPVHCV, HIV p17 Gag 77–85 SLYNTVATL, HIV Gag 157–165 TLNAWVKVV, HIV Env 120–128 KLTPLCVTL, EBV BMLF-1 GLCTLVAML, CMV pp65 495–503 NLVPMVATV (all from Primm, Milan, Italy). HLA-A*0301 restricted peptides HIV Nef 73–82 QVPLRPMTYK and HIV Gag 20–28 RLRPGGKKK were used (Primm). Molecular HLA class I typing was performed on all study participants using the standard serological methods utilized by the French Blood Centre (EFS, Besançon, France). PE-labelled pentameric peptide-HLA-A*0201 complexes were synthesized by ProImmune (Oxford, UK): PRO5™ Pentamer HIV A*0201/ILKEPVHCV, PRO5™ Pentamer HIV A*0201/SLYNTVATL, PRO5™ Pentamer EBV A*0201/GLCTLVAML and PRO5™ Pentamer CMV A*0201/NLVPMVATV. Constructions used to infect B-EBV cell lines were engineered vaccinia viruses: VVTG 1144 expressing the HIV-1 Gag p55 protein, VVTG 1147 expressing a myristoylated and Thr 15-phosphorylated HIV-1 p27 Nef protein, VVTG 4163 expressing the HIV-1 reverse transcriptase protein and VVTG 5167 expressing the HIV-1 Env gp160 protein (all from Transgène, Strasbourg, France).20

Cell surface markers and HLA-peptide pentamer staining

For cell surface phenotyping of CD28+ CD8+ T-cell subsets, fresh PBL and autologous PBMC were stained with FITC-conjugated or PE-conjugated anti-huCD28, with FITC-conjugated anti-huCD57 and anti-huCD45RA mAbs, with PE-conjugated anti-huCD38 and anti-huCD45RO mAbs, and with PC5-conjugated anti-huCD8 mAbs. Triple-colour flow cytometric analysis was performed using a FACScan flow cytometer (Beckton Dickinson). Gates corresponding to CD28/CD28int/CD28high CD8+ T cells were delineated according to the limits of CD28 CD8 and CD28+ CD8 populations for each HIV-infected patient tested. For pentamer complex staining, 5 × 105 frozen PBMC were washed twice in phosphate-buffered saline (PBS) supplemented with 0·5% bovine serum albumin (BSA, Sigma) and resuspended in 50 μl cold buffer. Ten microlitres of HIV, EBV or CMV pentamer was added, and cells were incubated for 30 min at room temperature. Ten microlitres anti-CD8-FITC and 10 μl anti-CD28-PC5 were added for another 30 min of incubation at 4°. Cells were analysed on a FACSCalibur (Becton Dickinson) using CellQuest software. Pentamers were titrated against appropriate CTL clones to determine the dose that induces maximal staining. Controls for the pentamer assays included A*0201- or HLA-B7-negative individuals and A*0201- or HLA-B7-positive uninfected donors. The limit of detection was 0·02% CD8+ T cells.

ELISPOT assay

The IFN-γ enzyme-linked immunospot (ELISPOT) assay was performed in triplicate according to the manufacturer's instruction (Diaclone, Besançon, France). Assays were performed with the HLA-A*0201-restricted HIV, EBV and CMV peptides, or with the two HLA-A0301-restricted HIV peptides, according to the HLA type of the samples. The frequency of peptide-specific CD8+ cells was calculated by subtracting the mean number of non-specific IFN-γ spots in the control sample from the mean number of specific IFN-γ spots in the peptide-stimulated sample. This number was extrapolated for 1 × 106 PBMC. Results are expressed as spot-forming cells/106 PBMC. The limit of detection per well was 100/106 PBMC.

Detection of cytokine-secreting CD8+ T cells following stimulation with viral peptides

PBMC (5 × 105 cells) were incubated for 90 min at 37° with the 40 μg/ml HIV-1, CMV or EBV peptides. Phytohaemagglutinin-A (Sigma, Saint-Louis, MO)-stimulated cells at 1 μg/ml were used as a positive control and a negative control was performed without peptide stimulation. Brefeldin A (5 μg/ml, Sigma) was added to the cells and they were incubated for an additional 15 hr. Cells were then washed twice with PBS/0·5% BSA and stained for 30 min at 4° for the surface markers CD8 and CD28. After a fixation step with PBS/1% paraformaldehyde for 30 min at 4°, the cells were permeabilized using two washes in saponin solution (PBS, 1% fetal calf serum, 0·1% saponin, Sigma). Cells were then incubated at room temperature for 30 min with FITC-labelled anti-IFN-γ mAb, anti-TNF-α, or anti-perforin mAbs. Samples were then analysed by three-colour flow cytometry using a Becton Dickinson FACSscan machine. Data from 5 × 104 cells were collected, stored and analysed using CellQuest software (Becton Dickinson). Isotype antibodies were carried out in each condition (activated or non-activated) and were used to define the limits for cytokine production.

Magnetic cell sorting for enrichment in CD28+ CD8+ T-cell subsets

Fresh PBMC were isolated from HIV-infected whole blood, washed with RPMI medium and resuspended in cold sorting buffer PBS containing 0·5% BSA and 2 mm ethylenediaminetetraacetic acid. PBMC were first sorted with the CD8+ T-cell isolation kit on magnetic LS columns according to the manufacturer's instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). The CD8+ T-cell fraction obtained was stained for 20 min at 4° with a limited concentration (0·375 μg/106 cells) of purified mouse anti-huCD28 antibody (clone CD28.2, Immunotech, Beckman Coulter). Cells were washed twice, resuspended in 80 μl/10 × 106 cells of cold sorting buffer, and coupled with 20 μl of anti-immunoglobulin microbeads and sorted on magnetic MS columns (Miltenyi Biotec). The positive fraction obtained was enriched in CD8+ CD28high cells. The negative fraction, enriched in CD8+ CD28 and CD8+ CD28int cells, was sorted a second time, with 0·5 μg/106 cells of purified mouse anti-huCD28 and goat anti-mouse immunoglobulin G magnetic microbeads. After the second CD28 sorting, the positive fraction obtained was CD8+ CD28int enriched and the negative fraction was CD8+ CD28 enriched. Aliquots of each cellular fraction were stained for 20 min at 4° with 10 μl purified anti-CD28 (Immunotech, Beckman-Coulter), washed, stained for 20 min at 4° with 10 μl goat anti-mouse immunoglobulin G1-PE (Beckman Coulter), washed and finally stained for 20 min at 4° with 10 μl mouse anti-CD8-PC5 (Beckman Coulter). Samples were then analysed by flow cytometry as reported above.

Cytotoxicity assays

Fresh PBMC (107 cells) isolated from HIV-infected donors were infected with 2·5 ml EBV-supernatant (UTCG, EFS, Besançon, France) in the presence of 1 μg/ml cyclosporin A and 500 μl fresh RPMI medium. The cellular suspension was distributed as 600 μl per well in five wells of a 24-well plate. Twenty-four hours later, 1·5 ml complete RPMI medium was added to the culture. After 10 days, B-EBV-transformed cells were transferred in a culture flask and expanded in complete medium to be used as autologous antigen-presenting cells. Caspase 3-based killing assays were performed using a Cytoxilus Kit (OncoImmune, Gaithersburg, MD), as previously described.21 Briefly, 106 autologous B-EBV target cells per ml were infected with 5 plaque-forming units (PFU)/cell of HIV Gag p55, Nef p27, reverse transcriptase and Env gp160 vaccinia constructs or pulsed with 40 μg/ml EBV BMLF-1 GLCTLVAML or CMV pp65 NLVPMVATV peptides in complete RPMI-1640 medium for 2 hr at 37° and 5% CO2. Target cells were washed and resuspended at 2 × 106 cell/ml in culture medium containing 0·1% of reconstituted FLT4 kit reagent and incubated for 1 hr at 37° with 5% CO2. Effector cells were added to HIV, EBV, CMV or control target cells for a final effector to target ratio of 10 : 1 and 4 : 1. Control samples were target cells not pulsed with peptides, non-infected by vaccinia viruses, or non-treated with cell-permeable fluorogenic caspase-3 substrate. Samples were analysed using flow cytometry on an EPIC ultra (Beckman Coulter) with Expo 32. A CD107a staining assay was performed as previously described.22 Briefly, fresh isolated or cryopreserved PBMC were washed in RPMI-1640 and resuspended at 3 × 106 cells/ml. PBMC were stimulated in 24-well plates either by autologous B-EBV cells infected for 2 hr at 37° in 5% CO2, with 5 PFU/ml HIV vaccinia constructs (ratios ranged between 1 : 1 and 1 : 20), or by 100 μg/ml HIV, EBV, CMV peptides, at a final concentration of 106 cells/ml. Negative controls were uninfected target cells or samples tested in the absence of peptide stimulation. Positive controls were performed following stimulation with 10 ng/ml phorbol 12-myristate 13-acetate (Sigma) and 500 ng/ml ionomycin (Sigma). Cell samples were analysed using flow cytometry with CellQuest software (FACScalibur, Beckton-Dickinson).

Statistical analysis

The Mann–Whitney U-test was used and P-values were considered significant when P < 0·05.

Results

Appearance of a continuous spectrum of CD28 expression among CD8+ T cells in chronic HIV infection

Expression of CD8/CD28 was investigated using flow cytometry on fresh PBL isolated from chronically HIV-infected patients. All samples obtained from HIV-infected patients showed a continuous spectrum of CD28 intensity, ranging from negative to high, in CD8+, but not CD4+ T cells (Fig. 1a). CD8+ T cells from healthy donors were mostly CD28 highly positive (Fig. 1). The reason why the continuous figures are seen in HIV-infected patients is not clear but our findings suggest that a novel, third population of CD8+ T cells, a CD28int subset, whose level of CD28 intensity is between those of highly positive and negative cells, might have emerged during HIV infection. We identified the third population as a CD28int CD8+ T-cell subset.

Figure 1.

 A continuous spectrum of CD28 intensity ranging from negative to high is detected in PBL isolated from HIV-infected subjects, but not in those from HIV-negative healthy individuals. After PBL were separated using Ficoll–Hypaque and adherence to plastic, the expression of CD28 and CD8 was examined in the CD3+ cell population by flow cytometry. CD28 expression was analysed on CD8+ T-cell subsets present in PBL freshly isolated from the peripheral blood of HIV-negative healthy subjects and from chronically HIV-infected patients using two-colour flow cytometric analysis. (a) PBL were selectively gated with forward/side scatter, CD3 and CD45/CD14 analysis, and the percentages of CD28 CD8+ T-cell subsets in the gates (CD28, CD28int, CD28+) are indicated. An experiment representative of 20 is shown for HIV-positive subjects and an experiment representative of 14 is shown for HIV-negative subjects. Gates corresponding to CD28/CD28int/CD28high CD8+ T cells were delineated according to the limits of CD28 CD8-negative and CD28+ CD8-negative populations for each subject tested. (b) Distribution of CD28 CD8+ T-cell subsets in PBL isolated from HIV-infected subjects and HIV-negative controls. Results represent mean values (± SD) of 20 and 14 independent experiments on PBL isolated from the peripheral blood of HIV-infected patients and HIV-negative subjects, respectively. ***P < 0·01.

CD28int CD8+ T cells are intermediately differentiated cells

To characterize the phenotype of the CD28int CD8+ T cells, we analysed the expression of a panel of cell surface markers by flow cytometric analysis (Fig. 2). On CD28+ CD8+ T-cell subsets from purified PBL isolated from chronically HIV-infected patients we measured the expression levels of CD27, CCR7, CD62L, CD57, CD45RA, CD45RO and CD38. CD27 is a costimulatory receptor involved in the generation of antigen-primed cells and therefore is particularly useful in distinguishing between subsets of differentiated CD8+ T cells.5,23 While CD28high CD8+ and CD28 CD8+ T cells expressed high and diminished levels of CD27 on their cell surface, respectively, the CD28int CD8+ T cells expressed the CD27 marker in an intermediate manner (Fig. 2). These results indicate that CD28 CD8+ T cells are mostly late-differentiated cells and the CD28high CD8+ T cells are early-differentiated cells, whereas CD28int CD8+ T cells could represent a subset of intermediately differentiated cells. Naive CD8+ T cells express high levels of the chemokine receptor CCR7, a secondary lymphoid organ-homing marker, associated with distinct CD4+ and CD8+ T-cell functional subsets.7,24 In contrast to CD28high CD8+ T cells that express high levels of CCR7, low levels of CCR7 were present on the cell surface of CD28int CD8+ T cells, and CCR7 was not present on CD28 CD8+ T cells (Fig. 2). However, both CD28int CD8+ T cells and CD28high CD8+ T cells expressed intermediate and high levels of the adhesion molecule l-selectin CD62L, respectively (Fig. 2).7,24 Theses results are in agreement with a previous study that demonstrated that CCR7–/low memory cells expressed CD62L at a lower and variable extent.24 CD57, a 110 000 molecular weight glycoprotein first identified on natural killer cells, is expressed on most CD28-negative cells, but barely on CD28-positive cells. Most of the CD28int and CD28high CD8+ T cells did not express CD57, whereas the majority of CD28 CD8+ T cells were CD57-positive (Fig. 2). CD8+ T cells expressed high levels of CD3 on the cell surface independently of CD28 expression levels (data not shown). These results indicate that CD28int CD8+ T cells are a subset of CD8+ T cells that express the TCR/CD8 complex, and therefore are distinct from natural killer cells that are CD8+ CD3 CD57+. In contrast to CD28 CD8+ T cells that express low levels of CD45RO, CD28int and CD28high CD8+ T cells expressed intermediate levels of CD45RO, indicating that both CD28int and CD28high CD8+ T-cell subsets, isolated from chronically HIV-infected patients, might represent memory cells (Fig. 2). These results are consistent with those from a previous study that indicated that CD45RA did not correlate with the distinction between early, intermediate and late phenotypes.5 We did not find any significant difference between the three CD28 CD8 T-cell subsets with regard to the expression of the CD45RA marker. Thus, the bulk of CD28int CD8+ T cells are CD27+ CCR7low CD62Lint CD57 and therefore might represent a subset of intermediately differentiated CD8+ T cells in the process of differentiating from CD28high CD8+ T cells to CD28 CD8+ T cells. We did not find any correlation between the levels of CD28 expression and the activation status of CD8+ T cells, assessed by the expression of the activation marker CD38, which is closely related to the extent of virus replication in HIV infection (P = NS) (Fig. 2).25 Using the measurement of Ki67 expression, we observed that the CD28int CD8+ T cells had a proliferative capacity intermediate between those of CD28 CD8+ T cells and CD28high CD8+ T cells (Fig. 3).

Figure 2.

 Expression of cell surface markers on CD28 CD8 T-cell subsets present in PBL isolated from HIV-infected subjects. Following PBL purification by Ficoll–Hypaque and an adherence step to plastic, the expression of CD8, costimulatory molecules (CD27, CD28), activation markers (CD45RO, CD38), a secondary lymphoid organ-homing marker (CCR7), an adhesion molecule (CD62L) and CD57 were assessed by flow cytometry. (a) Flow cytometric data representative of 10 independent experiments are shown. (b) The histogram represents mean values (± SD) of 10 independent experiments performed on PBL isolated from HIV-infected patients. ***P < 0·01; **P < 0·05.

Figure 3.

 Proliferative capacity of CD28 CD8+ T-cell subsets present in PBL isolated from HIV-infected subjects. Following PBL purification by Ficoll–Hypaque and a step of adherence to plastic, the expression of CD8, CD28 and intracellular Ki67 was assessed by flow cytometric analysis. (a) Spontaneous proliferative capacity of CD28 CD8+ T-cell subsets present in PBL isolated from HIV-infected subjects. Upper panel: flow cytometric data representative of 10 independent experiments are shown. Lower panel: the histogram represents mean values (± SD) of 10 independent experiments performed on PBL isolated from HIV-infected patients. (b) Proliferative capacity of CD28 CD8+ T-cell subsets present in PBL isolated from HIV-infected subjects following anti-CD3 stimulation. PBL were left unstimulated or were stimulated overnight with anti-CD3 (1 μg/ml). The results are expressed as: [Ki67 expression in stimulated cells] − [Ki67 expression in unstimulated cells]. The histogram represents mean values (± SD) of 10 independent experiments.

CD28int CD8+ T cells are virus-specific and produce TNF-α and IFN-γ

Since virus-specific populations during chronic infections are present in each of the CD28+ CD27+ and CD28 CD27+ phenotype subsets,5 we assessed the association between phenotype and viral specificity, using pentamers representing different epitopes from the same virus, HIV, CMV and EBV. We observed that all three CD28 CD8+ T-cell subsets contain pentamer-positive cells specific for HIV, EBV and CMV (Fig. 4a and data not shown). Interestingly, pentamer-positive cells specific for HIV, EBV and CMV were mostly present among CD28int CD8+ T cells, and to a lesser extent among CD28 CD8+ or CD28high CD8+ T cells (Fig. 4a). Although virus-specific populations were represented in each of the three different phenotypic subsets, EBV-specific CD8+ T cells expressed higher levels of cell surface CD28 than HIV-specific and CMV-specific CD8+ T cells, with mean fluorescences of 84·9, 67·1 and 49·8, respectively, in agreement with the phenotype/specificity differentiation model proposed by Appay et al. (Fig. 4b).5 We then analysed the CD28 CD8+ T-cell subsets for the production of TNF-α and IFN-γ using flow cytometric analysis. The intracellular IFN-γ and TNF-α production was measured in response to HIV, CMV and EBV peptide stimulation in PBLs isolated from chronically HIV-infected subjects (Fig. 5a,b).10,11 In response to HIV peptide stimulation, the CD28int and CD28 CD8+ T-cell subsets were the main subsets that preferentially produced IFN-γ and TNF-α, respectively (Fig. 5a). Interestingly, the CD28int CD8+ T-cell subset produced both IFN-γ and TNF-α in response to HIV peptide stimulation (Fig. 5a). By contrast, CD28 CD8+ T cells produced mostly TNF-α in response to HIV peptide stimulation (Fig. 5a). Moreover, CD28int CD8+ T cells produced both IFN-γ and TNF-α in response to EBV and CMV peptide stimulation (Fig. 5b). We also studied the expression of perforin, which is closely linked with the cytolytic activity in virus-specific CD8+ T cells. Following stimulation with HIV, CMV or EBV peptides, the CD28int CD8+ T-cell subset showed the most perforin production (Fig. 5c). In agreement with previous observations10 CMV peptide stimulation was the most efficient for perforin production. CD28high CD8+ T cells did not produce significant amounts of perforin, independently of the stimulus used (P = NS) (Fig. 5c). Thus, the CD28int CD8+ T-cell subset is the main CD8+ T-cell subset that produces large amounts of IFN-γ, TNF-α and perforin in response to stimulation with HIV, CMV or EBV peptides.

Figure 4.

 CD28 expression on HIV-, EBV- and CMV-specific CD8+ T cells in PBL isolated from HIV-infected subjects. The distribution of CD28 CD8+ T-cell subsets (CD28, CD28int, CD28+) among pentamer-positive cells, investigated using flow cytometric analysis, is shown. (a) The histogram represents the relative percentages of HIV-, EBV- and CMV-specific cells in the CD28 CD8+ T-cell subsets in PBL isolated from HIV-infected subjects. The mean percentage of HIV-, EBV- and CMV-pentamer-positive cells among CD3+ CD8+ T cells from HIV-negative EBV-negative CMV-negative donors is shown as a negative control. Results represent mean values (± SD) of six independent experiments. ***P < 0·01; **P < 0·05. (b) CD28, CD8 and HIV-, EBV-, and CMV-pentamer labelling was performed on PBL isolated from HIV-infected subjects. Pentamer-positive CD8+ T cells were gated and CD28 expression (MFI, mean fluorescence intensity) was measured using flow cytometry. Results represent mean values (± SD) of six independent experiments.

Figure 5.

 Cytokine and perforin production by CD28 CD8+ T-cell subsets in response to HIV, EBV and CMV peptide stimulation. (a) Fresh PBMC from chronically HIV-infected individuals were left unstimulated or were stimulated with HIV-specific peptides. Following stimulation, PBMC were treated 6 hr later with brefeldin A and intracellular expression of IFN-γ and TNF-α was measured using flow cytometric analysis. The distribution of CD28 expression on IFN-γ-producing and TNF-α-producing CD8+ T cells was measured in PBL from HIV-infected subjects. Results representative of 10 independent experiments are shown. (b) The histogram represents intracellular production of cytokines (IFN-γ, TNF-α) in CD28 CD8+ T-cell subsets from HIV-infected patients after stimulation with HIV, EBV and CMV peptides, or in the absence of peptide stimulation. Data obtained from two HIV-infected subjects are shown. (c) Perforin production by CD28 CD8+ T-cell subsets in response to HIV, EBV and CMV peptide stimulation. Upper panel: fresh PBMC from chronically HIV-infected individuals were stimulated with HIV-specific peptides or left unstimulated. Following stimulation, PBMC were treated 6 hr later with brefeldin A and the intracellular expression of perforin was measured using flow cytometry. Distribution of CD28 expression on perforin-positive CD8+ T cells is shown. Results representative of five independent experiments are shown. Lower panel: Fresh PBMC from chronically HIV-infected individuals (positive for HIV-, CMV-, and EBV-specific CD8+ T cells as measured by ELISPOT) were stimulated with HIV-specific, CMV-specific or EBV-specific peptides. The histogram shows a representative experiment (n = 10) for the intracellular production of perforin in the CD28 CD8+ T-cell subsets following stimulation with EBV-, HIV-, and CMV-specific peptides.

To further assess the direct cytotoxic activity of the CD28int CD8+ T-cell subset, we enriched the CD8+ T-cell fractions from different HIV-infected donors for either CD28 CD8+, CD28int CD8+ or CD28high CD8+ T-cell populations using the technique of differential magnetic sorting. The direct cytolytic potential of these cellular fractions from HIV-infected patients were tested in the presence of HLA-matched B-EBV cell lines presenting HIV peptides. We observed that when the proportion of CD28int CD8+ T cells increased six-fold in the total CD8+ T-cell fraction tested, whereas the proportion of CD28 CD8+ T cells was stable, the specific lysis of HIV target cells was increased eight-fold (Fig. 6a). In agreement with these data, we observed that cellular fractions enriched with CD28int CD8+ T cells triggered specific lysis of HLA-matched HIV-positive target cells, as measured by the percentage of caspase 3-positive target cells (Fig. 6b). We observed a positive correlation between HIV-specific lysis, as measured by the percentage of caspase 3-positive target cells, and the ratio of (CD28int CD8+ T cells) to (CD28 CD8+ + CD28high CD8+ T cells) (r2 = 0·52) (Fig. 6c). We obtained similar results for CMV-specific and EBV-specific lysis (Fig. 6c), suggesting that CD28int CD8+ T cells mediate cytolytic activity in an antigen-specific manner.

Figure 6.

 CD28int CD8+ T cells have cytotoxic activity specific for autologous target cells infected with recombinant vaccinia viruses expressing HIV-peptide or directly loaded with EBV- or CMV-specific peptides. (a) Target cells were selectively gated with forward scatter (FSC)/side scatter (SSC) analysis and FTL4-positive staining and the percentage of caspase-3-positive cells were measured in the presence of different enrichments in the CD28 CD8+ T-cell subsets. The direct cytotoxic activity on HIV-presenting target cells from the CD28 CD8+ T-cell fractions enriched from an HIV-infected donor is shown. (b) Target cells were selectively gated with FSC/SSC analysis and FTL4-positive staining. The percentage of caspase-3-positive cells was measured in the absence or in the presence of sorted CD8+ T cells enriched in CD28int CD8+ T cells, and for target cells infected or not by vaccinia viruses expressing different HIV peptides. Target cells presented peptides derived from HIV Env gp160, peptides derived from HIV-1 Gag p55, or pooled peptides derived from HIV Env gp160 + Gag p55 + Nef p27 + reverse transcriptase, as described in the Materials and methods. Data representative of five independent experiments are shown. (c) Correlation between the percentage of caspase-3-positive target cells and the CD28int to (CD28 + CD28+) ratio in the CD28 CD8+ T-cell enrichments tested. Target cells were infected by vaccinia viruses expressing HIV peptides or loaded with either EBV or CMV peptides. Results represent combined data from independent experiments. (d) Cell surface expression of CD107a on CD28 CD8+ T-cell subsets after 5 hr of stimulation in the presence of increasing concentrations of HLA-matched B-EBV cells expressing HIV peptides from the four proteins Gag p55, Nef p27, reverse transcriptase and Env gp160 as described in the Materials and methods. The results represent mean values of five independent experiments. (e) Distribution (percentage among total CD8+ T cells) of cell surface CD107a expression on the different CD28 CD8+ T-cell subsets isolated from chronically HIV-infected individuals, following HIV-, EBV- or CMV-specific peptide stimulation in vitro. Results represent mean values of seven independent experiments.

To provide a more complete assessment of the functionality of CD28 CD8+ T-cell subsets, we measured CD107a expression, as previously reported.22 This assay allows the measurement of the exposure of CD107a, which is present in the membrane of cytotoxic granules, on the cell surface as a result of degranulation.22 CD107a-expressing CD8+ T cells have been shown to mediate cytolytic activity in an antigen-specific manner.22 We observed that, in contrast to CD28high CD8+ T cells, both CD28int CD8+ and CD28 CD8+ T-cell subsets expressed high levels of CD107a in response to HIV peptide stimulation (Fig. 6d). Moreover both CD28int CD8+ and CD28 CD8+ T cells displayed cytolytic activity, as measured by CD107a expression, in a dose-dependent manner (ratio E : T) (Fig. 6d). In response to CMV and EBV peptide stimulation, both CD28int CD8+ and CD28 CD8+ T cells expressed increased CD107a levels on the cell surface (Fig. 6e), indicating that the cytolytic activity specific for CMV and EBV was predominantly present in these two CD8+ T-cell subsets.

Discussion

Our results indicate that a subset of functional effector-memory CD8+ T cells specific for HIV, CMV and EBV antigens is present in the peripheral blood of HIV-1-infected subjects. Recently, it has been shown that during chronic infection in humans, distinct virus infections (HIV, EBV, CMV) are characterized by enrichments of CD8+ T cells with different phenotypes, consistent with distinct stages of differentiation.5 We detected not only intermediately differentiated CD28 CD27+ CD8+ T cells but also CD28 CD27 CD8+ T cells in the peripheral blood of chronically HIV-infected subjects. This latter subset was detected in the peripheral blood isolated from HIV-infected patients with CMV-specific immune responses, which matches previous reports.5,26 Based on phenotypic expression, our results suggest the existence of a novel intermediate population of effector-memory CD8+ T cells present during the differentiation of antiviral CD8+ T cells: CD28int CD27+ CCR7low CD62Lint CD45RO+ T cells (Fig. 7). Thus on a CD8+ T-cell differentiation pathway, CD28int CD8+ T cells could be more committed than CD28high CD8+ T cells along with sequential down-regulation of CCR7.5,27 Our data support a model of CD8+ T-cell differentiation in chronically HIV-infected subjects, beginning with early-differentiated CD28high CD8+ T cells, which express high levels of the homing receptor CCR7 and adhesion molecule CD62L, through more differentiated effector-memory CD28int CD8+ T cells, which express lower CCR7 levels but still express intermediate levels of CD62L, to CD28 CD8+ T cells, which express neither CCR7 nor CD62L and are intermediately differentiated (CD28 CD27low) or late-differentiated (CD28 CD27 CD45RA and CD28 CD27 CD45RA+) cells.

Figure 7.

 Proposed differentiation pathway for CD8+ T cells. CD28int CD8+ T cells could represent effector-memory cells with an intermediately differentiated phenotype that are more committed than the CD28high CD8+ T cells, but less committed than the CD28 CD8+ T cells, according to differentiation models presented in previous studies.5,7,33

Most virus-specific CD8+ T cells, whatever their differentiation phenotype, exhibit a range of effector functions, such as expression of cytotoxic factors and antiviral cytokines such as IFN-γ and TNF-α.7,10,28–30 Our data indicate that, although both CD28 CD8+ and CD28int CD8+ T cells produced IFN-γ and TNF-α in response to HIV peptide, the CD28int CD8+ T-cell subset exhibited both stronger antiviral cytokine production (IFN-γ and TNF-α) and more pronounced direct ex vivo HIV-1-specific cytotoxicity. Consistent with our data, a subset of CD8+ T cells secreting both IFN-γ and TNF-α has been reported recently to be substantially more strongly associated with cytotoxicity.31 Since the bulk of CD28 CD8+ T-cell subsets could contain not only distinct subsets specific for HIV but also those for other viruses such as CMV and EBV, we observed the production of antiviral cytokines and perforin in CD28int CD8+ T cells, specific not only for HIV-1 but also for CMV and EBV following viral peptide stimulation. CD8+ T cells are dependent on IL-15, which mediates proliferation, and on IL-7, which enhances survival.32 We observed that CD28int CD8+ T cells proliferate in response to IL-7 and IL-15 (data not shown).

In summary, we have characterized the phenotype and function of a CD28int CD8+ T-cell subset present in the peripheral blood of HIV-infected patients. We observed that CD28int CD8+ T cells are intermediately differentiated effector-memory cells specific for HIV, EBV and CMV that produce antiviral cytokines and perforin, and mediate cytolytic activity in an antigen-specific manner. Our results indicate that a subset of functional effector-memory CD8+ T lymphocytes could contribute to the antiviral response observed in HIV-infected subjects. A better understanding of the mechanisms underlying the differentiation and death of CD8+ T cells is likely to lead to new therapeutic approaches, which could help to improve immune defence in HIV-infected subjects.

Acknowledgements

This work was supported by grants from the National Institutes of Health of the US Public Health Service NIAID 1R21 AI45335-01A2, the American Foundation for AIDS Research 02631-26-RGI, the Agence Nationale de Recherche sur le SIDA (ANRS), the Franche-Comté University, and the Cellule SIDA-CHU Besancon to G.H. A.Z.D. is a recipient of an ANRS graduate student grant.

We thank Patrick Hervé, Pierre Tiberghien, Eric Robinet and Jean-Marie Certoux (EFS, Besançon, France) for material assistance and Transgène (Strasbourg, France) for vaccinia constructs.

Ancillary