Human CD8 T cells of the peripheral blood contain a low CD8 expressing cytotoxic/effector subpopulation

Abstract

Heterogeneity of lymphocyte populations demonstrates the diversity of cellular immune responses and provide a better understanding of the immune system. CD3⁺ CD8⁺ T cells exhibit a low CD8 expressing (CD8^low) population in flow cytometric analysis of peripheral blood T cells. In healthy donors, this population consists of 0·2–7·0% of all CD8 T cells. The majority of the CD8^low T cell population showed an elevated expression of CD25, CD45RA, and CD95L, and low levels of CD28, CD62L and CD45RO. Circulating CD8^low T cells resemble cytotoxic effector cells because they express cytolytic mediators and are able to execute cytotoxicity. A restricted T cell receptor profile with increased Vβ9, Vβ14 and Vβ23 expression was observed and the CD8^low T cell population contain Epstein–Barr virus-specific T cells. Therefore, the CD8^low population represent a subset of activated CD8 effector T cells, resulting most probably from a continous and/or balanced immune response to intracellular pathogens.

Introduction

Specific CD8 T cell-mediated cell lysis is the critical effector mechanism against cells infected with viruses. Most viral infections are cleared successfully by the immune system. However, during infections with the family of herpesviruses, CD8 T cells do not succeed in eliminating the viruses completely from the host. A balance between CD8 effector T cells and the persistent viral infection seems to be established for the entire life of an individual.^1,2 In addition to their classic role in the killing of infected cells CD8 T cells play a role in the regulation and differentiation of CD4 T cells.^3–5

In humans, cell surface marker analysis has been used thoroughly to discern functional distinct subsets of CD8⁺ T cells. The CD8⁺ CD45RA⁻ CD28⁺ subset appears to contain memory-type cells.^6,7 By contrast, in healthy people, effector-type cells were found to express CD45RA in the absence of CD28. These cells exhibit various features related to a cytotoxic effector function, such as expression of perforin, granzyme and the CD95 ligand (CD95L).^6,7 These cells migrate preferentially to sites of inflammation and express a strongly skewed T cell receptor (TCR) Vβ repertoire.^8–10

CD8 expression is believed to be constant on the surface of human peripheral blood T cells. However, during flow cytometric analysis of lymphocyte populations, a low CD8 expressing subpopulation (CD8^low) is visible in the CD3/CD8 dot plot in almost all individuals. Many surface receptors undergo internalization and degradation upon binding their appropriate ligands. It has been demonstrated that triggered TCR–CD3-ζ complexes are degraded in the lysosomes after antigenic stimulation.¹¹ Furthermore, the co-receptors CD4 and CD8 are recruited to TCR and approximately two CD4 and four CD8 molecules are down-regulated for each triggered TCR.¹²

In the present study, the CD8^low population of CD8 T cells demonstrated an activated effector phenotype and oligoclonality. This suggests that chronic or repeated antigen exposure may be responsible for a focused effector response. Regarding antigen specificity, the CD8^low T cell population is not homogeneous, but contains significant numbers of Epstein–Barr virus (EBV)-specific T cells. Therefore, the CD8^low T cell population may represent a long-term, low-level response driven by chronic and/or repeated antigen exposure.

Materials and methods

Donors

Approval for the use of human blood was granted by the Ethical Committee of Davos. The CD8^low population was detected and quantified in 100 donors. In 10 healthy individuals with more than 3·5% CD8^low T cells the CD8^low population was characterized further. Detailed examinations with sorted CD8^low T cells (cytokine content, secretion and proliferative response, cytotoxicity) were carried out using blood samples from four donors. For further characterization (phenotype, TCR profile), cell sorting and culture of the CD8^low and CD8^normal T cells, informed consent of the donors was obtained. Cord blood was obtained from 10 newborns who had undergone full-term, normal deliveries with no evidence of infection or congenital abnormality.

Reagents

PHA was purchased from Sigma (St Louis, MO). Anti-CD2 (4B2 and 6G4) and anti-CD28 (15E4) mAb were from the Red Cross Blood Transfusion Service (Amsterdam, the Netherlands). Anti-CD3 was provided by clone CRL8001 obtained from ATCC (Manassas, VA). mAbs for flow cytometry (fluorescent labelled) and immunocytochemistry were purchased from Beckman Coulter-Immunotech Corp. (Nyon, CH), Dako A/S (Glostrup, DEN), BD PharMingen (San Diego, CA), Caltag Laboratory (Burlingame, CA), R&D Systems (Abingdon, UK) and Santa Cruz Biotech (Santa Cruz, CA). The B8-restricted FLRGRAYGL (EBV latent antigen EBNA3A 325–333) and RAKFKQLL (EBV lytic cycle antigen BZLF1 190–197) peptides were synthesized commercially by Anawa Trading SA (Wangen, CH). The peptides were analysed for > 95% purity by HPLC.

Staining and flow cytometry analysis

For flow cytometry analysis, a lysed whole blood technique and paraformaldehyde fixation was used. Rigorous quality control criteria were applied to exclude non-specific staining or contaminating cells, and to ensure that the relevant populations were included. Cells were stained with FITC-conjugated anti-CD8, ECD-conjugated anti-CD3 and various PE-conjugated mAbs. Three-colour flow cytometry was performed using an EPICS™ XL-MCL flow cytometer (Beckman Coulter). The presence of surface molecules is expressed as the percentage of positive stained cells minus the percentage of isotype control-stained cells. For intracellular cytokine staining cells were permeabilized in a saponin-containing solution before staining with the indicated mAb.

Analysis of the TCR repertoire

The TCR Vβ reportoire in the gated CD8^normal and CD8^low T cells was analysed by the use of PE-labelled anti-Vβ specific mAbs (Beckman Coulter) and three-colour flow cytometry.

Isolation, fluorescence activated cell sorting and culture of T cells

Mononuclear cells were isolated by Ficoll (Biochrom KG, Berlin, GER) density gradient centrifugation of venous blood. CD8^low and CD8^normal CD3 T cells were sorted into different tubes using a MoFlo high-speed cell-sorter (Cytomation Inc., Fort Collins, CO). T cells grown in RPMI-1640 (Life Technologies, Basel, CH), supplemented with 2% fetal calf serum (FCS), 100 U/ml glutamine and 100 µg/ml streptomycin were stimulated with 10 µg/ml PHA or a combination of soluble anti-CD2 (0·5 µg/ml), anti-CD3 (1·0 µg/ml) and anti-CD28 (0·5 µg/ml) mAb.

Quantification of cytokines

Interleukin (IL)-4, IL-10 and interferon (IFN)-γ were determined in culture supernatants after 48 hr by sandwich ELISA (Endogen, Woburn, MA) according to the manufacturer's protocols. The sensitivity of the IL-4 ELISA was < 2 pg/ml, of the IL-10 ELISA <3 pg/ml, and of the IFN-γ ELISA <2 pg/ml.

[³H]-thymidine assay

The proliferative response of CD8^low and CD8^normal T cells was determined after 36 hr by estimating incorporation of [³H]-thymidine into DNA using standard protocols.

Cytotoxicity assay

For the measurement of polyclonal cytotoxicity, EBV-transformed B lymphoblastoid cells were used as target cells. The polyclonal stimulation of CD8^normal and CD8^low effector T cells was achieved by using soluble anti-CD2, anti-CD3 and anti-CD28 mAb. Because of the restricted number of CD8^low T cells available we modified the conditions of the cytotoxicity assay. The effector–target cell mixture (5 × 10⁴−5 × 10⁴; 1 × 10⁵−5 × 10⁴; 1·5 × 10⁵−5 × 10⁴; 2 × 10⁵−5 × 10⁴) was incubated for 24 hr in 96-well flat bottom plates. To determine cytotoxicity, a LDH-based cytotoxicity detection system was used according to the manufacturer's protocol (Roche Molecular Biochemicals, Rotkreuz, CH). Briefly, the LDH activity in the culture supernatant was determined by a coupled enzymatic reaction whereby the tetrazolium salt INT is reduced to formazan which shows an absorption maximum at approximately 500 nm. The absorbance values of the ELISA correlate directly with the increase of dead cells.

IFN-γ ELISPOT assay

CD8^low T cells secreting IFN-γ in an antigen-specific manner were detected using an ELISPOT assay (R&D Systems). For this purpose, class I tissue typing was performed on lymphocytes using standard serological methods. In brief, both CD8^low T cells from HLA-B8 positive donors and autologus monocytes were added in triplicate at 1·25 × 10⁵, 6·25 × 10⁴ and 3·13 × 10⁴/well in the presence of 2 µm latent or lytic cycle EBV antigen. The plates were incubated 18 hr at 37° in 5% CO₂. The cells were removed and the biotinylated anti-IFN-γ mAb was added for 12 hr at 4°, followed by streptavidin-conjugated alkaline phosphatase (AP) for an additional 2 hr. Individual cytokine-producing cells were detected as dark spots after a 1-hr reaction with substrate.

Cytokine immunocytology

Diff-Quik™ (Dade AG, Düdingen, Switzerland), consisting of buffered eosin and phosphate-buffered Azure B was used for microscopic evaluation of the cell morphology. For immunocytology, 7·5 × 10⁴ cells centrifuged on cell adhesion slides were fixed with 4% formaldehyde. After blocking of endogenous peroxidase and endogenous biotin, slides were incubated with the primary mAb for 1 hr, followed by incubation with biotin-conjugated antigoat IgG and preformed streptAB-complex–peroxidase (Dako A/S) for 30 min. The peroxidase-specific substrate 3-amino-9-ethylcarbazole (Dako A/S) was used for visualization.

Statistical analysis

Results are shown as mean ± SD. The paired Student's t-test was used for comparison of paired conditions.

Results

Low CD8 expressing T cells in humans

The strategy for quantification and sorting of the CD8^low T cell population is shown in Fig. 1a. T cells were defined by the expression of the CD3 molecule and the subsets were identified by CD4 and CD8 expression. In addition to the typical, major CD8 T cell population (CD8^normal), a smaller CD8^low population was identified. Using fluorescence activated cell sorting, CD8^normal and CD8^low populations were sorted as single cells. The proportion of CD8^low T cells was stable if the same individual was tested repeatedly (Fig. 1b). The mean percentage of CD8^low T cells was 2·1 ± 0·7% of the total CD8 T cell population with a range of 0·2–7·0% of all CD8 T cells. There was no significant correlation between age and the proportion of CD8^low T cells among CD8 T cells (Fig. 1c). Using cord blood of healthy newborns we found CD8^low T cells at similar frequencies (Fig. 1c, dots on the x-axis). Microscopically, the CD8^low T cells appeared to be in a more activated state, showing more irregular shape with fingerlike protrusions and increased vacuolization (Fig. 1d).

Figure 1.

Human CD8^low T cells form a distinct population. (a)Dot plot of CD3-ECD fluorescence (x-axis, log scale) versus CD8-FITC fluorescence (y-axis, log scale) of T cells gated on the lymphocyte population. Frequencies of CD8^low T cells within the indicated gate were determined using the System II software (Beckman Coulter). (b)Frequencies of CD8^low T cells (y-axis) in four healthy individuals determined during 1 year. (c)There is no correlation in the frequency of CD8^low T cells (x-axis, percentage of CD8^low T cells) and the age of the individuals tested (y-axis, years; n= 110). (d)Diff-Quik™ staining of CD8^normal and CD8^low T cells after fluorescence activated cell sorting.

The CD8^low T cell population displays an activated effector phenotype with oligoclonal TCR expression

To define whether CD8^low T cells represent cells that had undergone in vivo activation, we analysed a number of markers indicative of the naive and effector phenotype. In these experiments, the phenotype of the CD8^low T cells was analysed directly using three-colour flow cytometry. Due to the increased expression of CD16 (Fig. 2a), CD8^low T cells bind more isotype control mAb than the rest of the CD8 T cell population. This was considered carefully by blocking the Fc receptor. The CD8^low T cells expressed increased CD16, CD25, CD45RA, CD95L and HLA-DR but showed down-regulation of CD11a, CD45RO and CD54. Remarkably, essentially all CD8^low T cells lacked CD28, the ligand for CD80 or CD86, which was expressed on 75% of CD8^normal T cells. CD11a, a subunit of the integrin LFA-1, which mediates the adhesion to antigen-presenting cells, was virtually absent on CD8^low T cells. Most of these effector type CD45RA⁺CD28⁻CD8^low T cells lacked CD62L (l-selectin), a lymph node homing receptor. In contrast to T cells, NK cells are defined by the presence of CD16, CD56, CD94, CD158e (KIRp70), and CD161 without the co-expression of CD3. The expression level of various NK cell receptors on CD8^low T cells was relatively low (CD16, CD94, CD158e, CD161) or absent (CD56) (Fig. 2a). CD57, expressed on subsets of NK and T cells, was not expressed on CD8^low T cells. 26·8 ± 12·2% of the CD8^low T cells co-express CD4 in vivo (Fig. 2a). We next analysed CD8 T cells gated as CD8^normal and CD8^low populations by three-colour flow cytometry to determine their TCR Vβ repertoire. We performed direct ex vivo analysis of CD8^low T cells using 21 different TCR Vβ-specific mAb, which usually cover >60% of all T cells within this population. As shown in Fig. 2b, substantial oligoclonal populations could be identified for the TCR Vβ segments Vβ9, 14, 23, whereas for other TCR Vβ segments (Vβ5·2, 5·3, 18) smaller populations could be identified only in some individuals. Expression of all the other Vβ families was barely detected or absent. The CD8^normal T cells expressed a wide Vβ repertoire. The CD8 T cells showed only marginal staining with the anti-CD14 (control), the anti-Vα24 and the antiγ/δ TCR mAb (Fig. 2b).

Figure 2.

Surface phenotype of CD8^low T cells. (a)CD8^low T cells express an effector phenotype. CD8^normal and CD8^low T cells were stained with the indicated panel of cell surface markers. (b)CD8^low T cells express an oligoclonal TCR Vβ repertoire. CD8^low T cells were stained with the indicated panel of Vβ-specific TCR mAb. (a, b) Results of 10 donors are shown.

Cytokine content, secretion and proliferative response of CD8^low and CD8^normal T cells

Further analysis of functional properties of CD8^low T cells including their ability to proliferate and produce cytokines was undertaken. The proliferative response of CD8^low T cells after treatment with soluble anti-CD2, anti-CD3 and anti-CD28 mAb determined by [³H]thymidine incorporation (Fig. 3a) was significantly higher compared to CD8^normal T cells. In parallel, significant more IFN-γ, IL-4 and IL-10 was secreted by CD8^low T cells (Fig. 3a,b). However, the amounts of IL-4 and IL-10 were generally very low compared to IFN-γ. Intracellular cytokine staining revealed more IFN-γ in CD8^low T cells, but no significant differences concerning the IL-4 and IL-10 content (Fig. 3c). CD8^low T cells express increased cytolytic mediators such as CD95L (Fig. 2a), granzyme B and perforin (Fig. 3d).

Figure 3.

Cytokine content and effects of in vitro stimulation. (a)Proliferative response and IFN-γ secretion of resting (US, unstimulated) and stimulated (Stim, with soluble anti-CD2, anti-CD3, and anti-CD28 mAb) CD8^normal and CD8^low T cells cultured for 48 hr, *P < 0·05. (b)IL-4 and IL-10 secretion by resting and stimulated CD8^normal and CD8^low T cells cultured for 48 hr, *P < 0·05. (c)Determination of intracellular IL-4, IL-10 and IFN-γ content of resting CD8^normal and CD8^low T cells. Histogram of fluorescence intensity (x-axis, log scale) versus cell number (y-axis). (d)Frequencies of IFN-γ, granzyme B and perforin positive cells. Immunocytological staining of CD8^normal and CD8^low T cells on cytospins after fluorescence activated cell sorting. (a, b) Mean ± SD of triplicates are shown. (c, d) Results of one representative individual are shown.

CD8^low T cells display increased cytotoxicity and EBV-specific cells

Activation-induced T cell death is shown in Fig. 4a. Immediately after cell sorting, CD8^low T cells showed significantly increased activation as well as cellular cytotoxicity compared to CD8^normal T cells, pointing to their preactivated state in vivo. Cytotoxic T cell function was assessed by LDH release from target cells (Fig. 4b). We observed significantly more cytotoxicity in cultures of the CD8^low population compared with the CD8^normal population. Interestingly, at the highest number of CD8^low T cells (i.e. 2 × 10⁵ effector cells), the activation-induced T cell death results in diminished target cell lysis. The oligoclonality of the TCR in vivo suggested that this skewed TCR repertoire could result from chronic antigen stimulation. After primary infection, EBV establishes a latent, usually asymptomatic, chronic infection. The frequency of cells within the CD8^low population reactive with lytic and latent cycle epitopes of EBV was therefore analysed by ELISPOT assays in donors, who had no history of infectious mononucleosis, but were tested positive for serum IgG antibody against EBV. Approximately 10% of the CD8^low T cells were able to secrete IFN-γ in 18-hr stimulation assays. HLA-B8 restricted RAKFKQLL and FLRGRAYGL specific CD8^low T cells could be detected (Fig. 4c). No significant differences between the CD8^normal and the CD8^low T cells was found regarding the secretion of IFN-γ in response to RAKFKQLL or FLRGRAYGL.

Figure 4.

CD8^low T cells display increased cytotoxic activity and EBV-specificity. (a)Activation-induced cell death of CD8^low and CD8^normal T cells 24 hr after fluorescence-activated cell sorting and stimulation by soluble anti-CD2, anti-CD3 and anti-CD23 mAb during the assay. (b)Cytotoxicity assay showing target cell death 24 hr after fluorescence activated cell sorting and stimulation. Target cells were incubated in the presence or absence (controls) of effector cells for 24 hr. Culture supernatant samples for controls and the effector–target cell mix were assayed for LDH activity. The curves shown were generated when the background control values are substracted from the effector–target cell values. (c)IFN-γ ELISPOT assay showing EBV-antigen-specific CD8^low T cells 18 hr after fluorescence activated cell sorting and exposure to the MHC class I compatible peptide. (a-c) Results of one representative individual are shown.

Discussion

We have characterized a population of peripheral blood T cells expressing lower amounts of CD8 than the major CD8 population. These CD8^low T cells display a distinct flow cytometric pattern and form a small distinct population. The CD8^low T cells represent activated effector CD8 T cells comprising a limited number of TCR Vβ. It was shown that triggered TCR and CD4 and CD8 co-receptors are down-regulated and degraded with identical kinetics.^11,12 Recently, a phenotypically distinct subpopulation of human peripheral blood T cells expressing lowered levels of CD4 was described.¹³ It was suggested that these cells are activated, temporarily apoptosis-resistant CD4 T cells which accumulate in the elderly.¹³

It has been estimated that the potential repertoire of α/β TCR exceeds 10¹⁵, which presumably allows peripheral T cells to respond to a wide variety of foreign antigens.¹⁴ The present study describes an unprecedented degree of conservation in TCR among unrelated individuals towards Vβ9, Vβ14, and Vβ23 selected in the CD8^low population. It was shown that identical TCR protein sequences are used by clones from each of four healthy unrelated EBV carriers.¹⁵ The response to a subdominant EBV epitope (EBNA4399–408) was highly restricted with conserved Vβ usage and identical length and amino acid motifs in the third complementary-determining regions (CDR3).¹⁶ Similar findings were observed in other experimental systems and natural infections to viruses and bacteria.^17–20 There are several advantages of mAb usage for TCR repertoire studies. Foremost, antibodies detect the TCR proteins rather than measuring RNA levels. A standardization of the nomenclature of TCR V gene specific mAb was achieved.^21,22 The human CD8 T cell repertoire is often dramatically skewed by predominant clones, which can arise after antigenic challenge.^10,23,24 It is becoming increasingly evident that a degree of flexibility in peptide recognition is an inherent property of many TCR. Studies using synthetic peptide analogues of T cell epitopes have shown that most TCR can undoubtely recognize multiple peptide ligands, demonstrating an ability to accomodate amino acid substitutions at selected positions.^25–27 It is tempting to speculate that the bias in the production of specific TCR rearrangements has co-evolved with viruses and the MHC to facilitate a balanced virus–host relationship. We demonstrate here that the CD8^low T cell population includes circulating EBV-specific CD8 effector T cells.²⁸ However, it is obvious that the CD8^low population is not homogeneous and may include T cell specificities against other intracellular pathogens.²⁹ We have demonstrated that the CD8^low T cells circulate in a state wherein they can display rapid effector function after antigen contact.³⁰ Their continued presence in a relatively activated state may reflect ongoing low-level stimulation by persisting antigen.

Naive CD8 T cells express CD45RA and CD62L and the co-stimulatory receptors CD27 and CD28. Memory-type cells express CD45RO, CD27 and CD28 molecules and low levels of CD62L. Using flow cytometric staining with mAb, approximately 50% of the CD8^low T cells expressed neither CD45RO nor CD45RA. It is possible that T cells which express CD45RB and CD45RC isoforms are contained within the CD8^low population.^31,32 Effector cells present a CD45RA, CD27-negative and CD28-negative phenotype and contain perforin, granzyme B and CD95L to execute cytolysis directly.^6,7,33,34 CD8 T cells can mediate antiviral protection via two pathways, either via perforin-dependent cytotoxicity or via Fas ligand (CD95L)-Fas (CD95) interaction.³⁵ Perforin-dependent CD8 T cell activity controls virus at low levels of persistence. Recent studies suggest that granzymes may be co-secreted with perforin and enter the target cell through the polyperforin pore to trigger an internal pathway that results in DNA fragmentation.³⁶ Our results show abundant perforin and granzyme B in the CD8^low T cells pointing to their role as cytotoxic/effector T cells. The CD8^low effector T cells also produce high amounts of IFN-γ. IFN-γ induces the increased expression of MHC class I and other molecules involved in peptide loading. This increases the chance that infected cells will be recognized as target cells. In addition, IFN-γ up-regulates Fas expression on target cells and renders them susceptible to Fas ligand-mediated apoptosis.³⁷ Numerous factors have been demonstrated that regulate IFN-γ expression of T cells. Recently, even histamine was demonstrated to play a role.³⁸ CD8^low T cells produce significant amounts of the immunoregulatory cytokines IL-4 and IL-10 and it is probable that they may also play a role in immune regulation.^3–5

It is proposed that when CD8 T cells progress through the naive towards effector stages, expression of NK cell receptors increases.³⁹ Our results demonstrate that some of these molecules are indeed expressed on the CD8^low T cell population. Interestingly, we observed an increased CD4 expression on the CD8^low T cells. Frequencies up to 39% of CD4 expression on activated CD8 T cells during infection by HIV-1 were reported.^40,41 This may facilitate the entry of the HIV into CD8 T cells; however, the biological function of CD4 expression on CD8 T cells is not known. It is probable that the CD8^low T cells that are detectable in umbilical cord blood T cells in our study may represent the intrauterine initiation of antigen-specific responses.⁴² However, further investigations are necessary to clarify the function of CD8^low T cells in the newborn.

Taken together, the present study demonstrates that the CD8^low T cell population represents highly activated cytotoxic CD8 T cells. Such cells may have a critical role in controlling the replication of persistent viruses. Separating these CD8^low T cells offers the possibility to directly investigate in vivo activated CD8 T cells.