BY55/CD160 cannot be considered a cytotoxic marker in cytomegalovirus-specific human CD8⁺ T cells

Abstract

CD160/BY55 is a glucosyl-phosphatidylinositol (GPI)-anchored cell membrane receptor that is expressed primarily in natural killer (NK) cells. Its presence in CD8⁺ T lymphocytes is considered to be a marker of cytotoxic activity, although there are few data in this regard. In the present work, we analysed the expression of CD160 in subpopulations of cytomegalovirus (CMV)-specific CD8⁺ T cells. Subpopulations were defined by CD28 and CD57 expression and exhibited varying degrees of differentiation and cytotoxic potential, as evaluated by the expression of perforin, interferon (IFN)-γ and interleukin (IL)-7Rα/CD127. We included subjects with different intensities of anti-viral immune response. Results showed that the terminally differentiated CD28^– CD57⁺ subset displaying the highest level of perforin expressed CD160 at a level similar to that of memory CD28⁺ CD57^–perforin^– cells. A comparison of the expression of perforin in CD160⁺ cells versus CD160^– cells showed that expression was significantly higher in the absence of CD160. Interestingly, the CMV-specific CD8⁺ T cell subset from a patient with ongoing CMV reactivation did not begin to express CD160 until day +92 of the follow-up period. Taken together, our data show that CD160 cannot be considered a cytotoxic marker in CMV-specific CD8⁺ T cells.

Introduction

During the past few years it has become increasingly clear that T lymphocytes display extensive heterogeneity and include many subsets with distinct effector functions [1–3]. This effector heterogeneity provides the immune system with the plasticity needed to combat a variety of infections. Efforts have been made to use immunophenotyping to identify individual subsets with different functions, but the complexity of the task seems limitless: many phenotypic markers vary with the state of lymphocyte activation and therefore cannot be used as stable markers to define subsets [4]. In addition, different viruses induce the expansion of phenotypically different CD8⁺ T cells [5]. Given such complexity, it is not surprising that very few molecules have been recognized as markers capable of defining T cell subsets.

One of these markers is the glucosyl-phosphatidylinositol (GPI)-anchored cell membrane receptor BY55/CD160. It was identified originally in natural killer (NK) cells [6], and more specifically in the CD56^dim CD16⁺ subset, which exhibits high cytolytic activity [7]. Thereafter, it was shown that CD160 was also present in some T cell subsets, including intestinal intraepithelial lymphocytes, peripheral T cell receptor (TCR)-γδ-bearing cells and a minor subpopulation of CD8^bright TCR-αβ⁺ lymphocytes [7–9]. It is generally accepted that CD160 is a marker of cytotoxic activity on peripheral TCR-αβ⁺ CD8⁺ T cells [6–10]. However, it has not been demonstrated extensively. Studies with antigen-specific cytotoxic CD8⁺ T cells are necessary before this functional significance can be attributed clearly to the presence of CD160 on T lymphocytes.

The present work was undertaken to confirm CD160 as a cytotoxic marker on peripheral TCR-αβ⁺ CD8⁺ T cells. It was performed on cytomegalovirus (CMV)-specific human CD8⁺ T lymphocytes, which provide an excellent tool to study molecules related with cytolytic activity. CMV infects the majority of the world's population and establishes a lifelong infection that is kept under control by a strong virus-specific CD8⁺ T cell response. This response is dominated typically by relatively few individual clones that are greatly expanded, and are present at high frequencies in peripheral blood from healthy CMV carriers. The 495–503 peptide (NLVPMVATV) from the CMV structural protein pp65 is one of the immunodominant epitopes, and is presented by human leucocyte antigen (HLA)-A*0201, which is a frequent HLA allele among the general population. Therefore, it is possible to analyse CMV-specific CD8⁺ T cells in a large number of individuals by tetramer staining [11, 12].

We included in the study seropositive healthy donors (HD), as well as asymptomatic renal transplant recipients (TxR) who were currently undergoing immunosuppressive therapy. It is well known that intermittent virus reactivation occurs in immunosuppressed individuals, and leads to an increase in the numbers of CMV-specific CD8⁺ T cells in an effort to control virus replication [13]. Therefore, we expected that both groups of individuals would display different levels of CMV-specific T cell responses. We also included a transplant recipient who was experiencing ongoing CMV reactivation, in order to carry out a longitudinal study of the cytotoxic specific response over an extended period of time, from its highest degree at the time of diagnosis to its subsequent return to baseline values.

We subdivided the CMV-specific CD8⁺ T cell population into subsets, according to the expression of CD28 and CD57. CD28^– CD57⁺ CD8⁺ T cells are considered to be a terminally differentiated subset and have high cytolytic activity [14]. CD8⁺ T cells down-regulate CD28 expression after multiple rounds of cellular activation, and eventually up-regulate CD57 expression [15, 16]. All peripheral CD8⁺ T cells are CD28⁺ CD57^– at birth, and ageing is accompanied by the progressive loss of CD28 and by a reciprocal enhancement of CD57 expression; this process takes place because of the repeated antigenic stimulation that occurs throughout life [17].

In this study, we analysed different subpopulations of CMV-specific CD8⁺ T cells in different clinical situations with the aim of defining a spectrum of varying degrees of cytotoxic activity, in order to study its correlation with the presence of CD160. Results showed that the terminally differentiated CD28^– CD57⁺ subset showed the highest expression of perforin, but expressed similar levels of CD160 as memory CD28⁺ CD57^– perforin^– cells. Therefore, we concluded that CD160 cannot be considered a cytotoxic marker in CMV-specific human CD8⁺ T cells.

Materials and methods

Subjects

Fifty-four HLA-A*0201 CMV-seropositive asymptomatic renal transplant recipients (average age: 39 years; range 33–77), 32 HLA-A*0201 CMV-seropositive healthy donors (average age: 57 years; range 25–60 years) and one HLA-A*0201 transplant recipient with ongoing CMV reactivation (age: 56 years) were included in the study after approval from the local Ethical Committee. All participants gave their informed consent for the use of blood samples. CMV serostatus was determined by a commercial CMV IgG enzyme-linked immunosorbent assay (ELISA) kit (Roche Diagnostics GmbH, Mannheim, Germany). Detection of CMV antigen in peripheral blood from the patient with CMV reactivation was performed by immunofluorescence and by Shell-Vial viral culture testing. The immunosuppressive regimen of all Tx recipients consisted of administration of a calcineurin inhibitor, mycophenolate mofetil and prednisone. All drugs were given at maintenance dosages. Ethylenediamine tetraacetic acid (EDTA)-anti-coagulated blood samples were collected 12 h after the last dose of immunosuppressants and were processed within 2 h of collection.

HLA-A2 positivity was assessed on peripheral T lymphocytes by flow cytometry using anti-HLA-A2-fluorescein isothiocyanate (FITC) and anti-CD3-phycoerythrin (PE) antibodies (Becton-Dickinson, Erembodegen, Belgium). DNA samples were obtained from HLA-A2⁺ subjects (Qiagen GmbH, Hilden, Germany) and HLA-A*0201 positivity was determined by polymerase chain reaction–sequence-specific primers (PCR–SSP) (Dynal Allset^+TM SSP HLA-A2; Dynal Biotech Ltd, Merseyside, UK).

Flow cytometry analysis of tetramer⁺ lymphocytes

HLA-A*0201- CMVpp65_495−503 tetramer conjugated with allophycocyanin (APC) was purchased from Proimmune (Oxford, UK); 2 × 10⁶ peripheral blood mononuclear cells (PBMC) were incubated with 2 µl of tetramer for 30 min at 4°C. After washing in phosphate-buffered saline (PBS) containing 0·1% bovine serum albumin (BSA) and 0·1% NaN₃, cells were incubated with 10 µl of anti-CD28-biotin (clone CD28·2; BD Pharmingen, San Jose, CA, USA) or 5 µl of anti-CD57-biotin (clone HNK-1; Becton-Dickinson) for 15 min at room temperature in the dark. Samples were then washed again and stained with 0·3 µg of streptavidin-PE-Cy7 (Becton-Dickinson) and with the monoclonal antibodies (MoAb) as indicated (anti-CD57-FITC (clone HNK-1; Becton-Dickinson), anti-CD160-PE (clone BY55; Immunotech, Marseille, France) or anti-CD127-PE (clone R34·34; Immunotech). After 15 min of incubation at room temperature in the dark, cells were washed and fixed in 500 µl of 1% paraformaldehyde. If intracellular perforin expression was to be analysed, cells were permeabilized after cell-surface staining (Fix and Perm; Caltag, Hamburg, Germany) and incubated with 20 µl of anti-perforin-FITC (clone δG9, BD Pharmingen) for 20 min in the dark, washed, and then fixed.

We did not include anti-CD3 or anti-CD8 in the analysis because we had determined previously that the only cells that bound tetramers were CD3⁺ CD8⁺ lymphocytes.

Specificity of tetramer staining was assessed with eight HLA-A*0201 CMV-seronegative healthy donors, six non-HLA-A2 CMV-seropositive healthy donors and three non-HLA-A2 CMV-seronegative healthy donors. None of them showed tetramer⁺ cells.

Sensitivity of tetramer analysis was determined with serial dilutions of two independent samples. We were able to detect tetramer⁺ cells if they were present at a frequency above 0·005% of CD8⁺ T cells.

Samples were acquired on a fluorescence activated cell sorter flow cytometer (FACScalibur; Becton-Dickinson) by following a double-acquisition protocol. In the first acquisition step, 20 000 PBMC were acquired and tetramer⁺ cells were gated. In the second acquisition, all the rest of the sample (at least 1·5 × 10⁶ PBMC) was acquired. Analysis was performed with Paint-A-Gate software (Becton-Dickinson). Samples containing fewer than 200 tetramer⁺ cells were eliminated.

Intracellular IFN-γ staining in tetramer⁺ cells after 6 h of antigenic stimulation

Antigenic stimulation induces internalization of TCR in CD8⁺ T cells and therefore interferes significantly with tetramer staining, whether staining is performed before or after cellular activation [18]. We developed a protocol based on the use of an endocytosis inhibitor (cytochalasin B) after 1 h of antigenic stimulation; this protocol allows tetramer staining while preserving cellular activation (Fig. 1). PBMC (2 × 10⁶) were stimulated with 10 µg/ml of antigenic peptide (CMVpp65_495−503) (Sigma Genosys, The Woodlands, TX, USA) in 400 µl of RPMI-1640/0·1% BSA medium for 1 h at 37°C. Then, 1·6 ml of RPMI-1640 containing 12·5% fetal calf serum (FCS), 10 µg/ml brefeldin A (Sigma, St. Louis, MO, USA) and 100 µM cytochalasin B was added. After another 5 h of culture at 37°C, cells were incubated with 0·02% EDTA for 15 min at 37°C, fixed for 15 min at room temperature in the dark (using Solution A from Fix and Perm; Caltag), washed, permeabilized (using Solution B, Fix and Perm) and stained with anti-IFN-γ-PE (clone 4S.B3, BD Pharmingen) and with the remaining MoAbs.

Fig. 1.

Intracellular interferon (IFN)-γ staining in CD8⁺ T cells from a representative cytomegalovirus (CMV)⁺ healthy donor cultured for 6 h in different experimental conditions. (a) Basal conditions; (b) peripheral blood mononuclear cells (PBMC) were stimulated with antigenic peptide for 1 h. Then, brefeldin A was added for another 5 h of culture; (c) cells were stimulated with antigenic peptide for 1 h in the presence of 100 µM cytochalasin B. Then, brefeldin A was added for another 5 h of culture; (d) cells were stimulated with antigenic peptide for 1 h. Brefeldin A and 100 µM cytochalasin B were then added for another 5 h of culture. Percentages of IFN-γ in CD8⁺ cells show that cellular activation is impaired if cytochalasin B is present from the onset of culture (c). However, cellular activation is preserved if cytochalasin B is added after 1 h of antigenic stimulation (d).

Culture of CD28⁺ lymphocytes with antigen

Immunomagnetically selected CD28⁺ cells (Dynal Biotech Ltd) from peripheral blood mononuclear cells of three HLA-A*0201 CMV-seropositive healthy donors were cultured for 20 days in RPMI-1640 supplemented with 10% FCS, 2 mM glutamine and 40 mg/ml gentamycin, in the presence of 0·5 µg/ml of antigenic peptide (CMVpp65_495−503) (Sigma); 50 U/ml of rhIL-2 (Sigma) was added to the culture every 3 days.

Statistical analysis

Results are expressed as median and range. An ordinary least-squares regression model was used for comparisons between both groups of subjects, including age as a covariate. For comparisons between subpopulations within each group of subjects, we used a generalized linear model for analysis of variance (anova) of repeated measurement, and adjusted for age and for the duration of immunosuppressive treatment. To explore the influence of the length of the period of immunosuppression on the percentages of tetramer⁺ subpopulations we constructed multivariable regression models, with the percentage of each subpopulation as dependent variable and the time of treatment as predictive variable. Analyses were adjusted for age. The Wilcoxon paired data test was used to compare perforin expression between CD160⁺ and CD160^– cells; P < 0·05 was considered to be statistically significant.

Results

The CMV-specific CD8⁺ T cell response was measured by tetramer staining in 32 HLA-A*0201 CMV-seropositive HD and 54 HLA-A*0201 CMV-seropositive asymptomatic Tx recipients (Fig. 2a). Tx recipients showed significantly higher levels of tetramer⁺ cells than healthy controls [1·5% (0·009–15·4%) versus 0·57% (0·034–2·5%), P < 0·05; results are expressed as median and range]. The expression of perforin in tetramer⁺ cells was significantly higher in patients than controls [83% (13–96%) versus 37% (11–92%), P < 0·001] (Fig. 2c).

Fig. 2.

Percentage of tetramer⁺ (TM) cells in CD8⁺ total lymphocytes from human leucocyte antigen (HLA)-A*0201 cytomegalovirus (CMV)-seropositive healthy donors (HD) (n = 32) and asymptomatic transplant (Tx) recipients (TxR) (n = 54) (a) and from one HLA-A*0201 Tx recipient with ongoing CMV reactivation (b). (c) Expression of perforin in tetramer⁺ cells (HD and Tx recipients) and (d) patient with reactivation. (e) Subpopulations defined by CD57 and CD28 expression.

As expected, the percentage of tetramer⁺ cells was very high in the patient suffering from virus reactivation: these cells were present in 33% of peripheral CD8⁺ T cells (Fig. 2b), and all of them were perforin⁺ (Fig. 2d). The percentage of tetramer⁺ cells decreased progressively during the 5-month follow-up period, but their perforin content remained elevated: 93% of the CMV-specific CD8⁺ cells were positive at day +140.

Next, we measured the subpopulations defined by CD28 and CD57 expression in the tetramer⁺ subset (Fig. 2e). Tx recipients had significantly more CD28^– CD57⁺ cells than healthy donors [50% (3–92%) versus 33% (8–77%), P < 0·05], and conversely had significantly fewer CD28⁺ CD57^– cells [6% (1–36%) versus 21% (3–52%), P < 0·01]. No differences were found in the CD28^– CD57^– subpopulation between the HD and TxR groups of subjects.

We also analysed each subpopulation for the expression of perforin and the production of IFN-γ in response to antigenic stimulation, and additionally we examined the expression of CD127/IL-7Rα, because it is well known that the short-lived effector CD8⁺ T cells are characteristically CD127^lo or CD127^– [19–21].

As shown in Fig. 3, the CD28^– CD57⁺ subpopulation could be considered to be an effector subset, with high levels of perforin (Fig. 3a) and IFN-γ (Fig. 3b) and low expression of IL-7Rα (Fig. 3c). Conversely, CD28⁺ CD57^– cells showed low levels of perforin and IFN-γ and very high expression of IL-7R. Analysis of these three parameters in the CD28^– CD57^– subpopulation showed that it had intermediate levels of expression for the three proteins, between those of the other two subsets.

Fig. 3.

(a) Expression of perforin; (b) interferon (IFN)-γ after 6 h antigenic stimulation; and (c) interleukin (IL)-7Rα/CD127 in subpopulations of cytomegalovirus (CMV)-specific CD8⁺ T lymphocytes from healthy donors (n = 25). Subpopulations are defined by CD28 and CD57 expression. The median of results is represented. Histograms from representative individuals are shown. TM: tetramer.

The observation that CD28^– CD57⁺ cells constituted the effector subpopulation with the highest expression of perforin within the CMV-specific CD8⁺ compartment could explain why significantly higher percentages of this subset were found in Tx recipients versus healthy controls. Interestingly, we found that this percentage increased with increasing length of immunosuppression (months) (B = 0·122; P < 0·05), whereas the percentage of CD28⁺ CD57^– cells decreased reciprocally (B = − 0·1; P < 0·01). The length of immunosuppression did not induce any significant change in the percentage of CD28^– CD57^– cells. We also found that in Tx recipients there was a significant positive correlation between the size of the CMV-specific CD8⁺ T cell subset and the content of CD28^– CD57⁺ cells in this subset (R = 0·415, P = 0·016) and a reciprocal negative correlation with the content of CD28⁺ CD57^– (R = − 0·474, P = 0·005). This correlation was not seen in healthy donors.

Expression of CD160 in subpopulations of CMV-specific CD8⁺ T cells

We compared the expression of CD160 among the subpopulations defined by CD28 and CD57 expression in tetramer⁺ cells. In both the HD and Tx subject groups, the CD28^– CD57^– cells displayed significantly higher expression of CD160 than the other two subpopulations (Fig. 4a,b). It is important to note that no differences were found between the expression of CD160 in cytotoxic effector CD28^– CD57⁺ cells and memory CD28⁺ CD57^– cells with low expression of perforin.

Fig. 4.

Expression of CD160 in subpopulations defined by CD28 and CD57 expression in cytomegalovirus (CMV)-specific CD8⁺ T lymphocytes. (a) Healthy donors, n = 26; (b) transplant (Tx) recipients, n = 33: median of results is shown; (c) histogram from a representative asymptomatic Tx recipient; (d) longitudinal study of CD160 expression in CMV-specific CD8⁺ T cells from a patient with ongoing CMV reactivation. (e) Histogram from a patient with ongoing CMV reactivation at day +92 of the follow-up period. TM: tetramer.

Tetramer⁺ cells from the patient with an ongoing CMV reactivation did not express CD160 at the time of diagnosis. The expression of CD160 appeared very gradually at the beginning of the fourth month of the follow-up period (Fig. 4d,e).

CMV-specific CD8⁺ CD160⁺ cells show lower levels of perforin than CD160^– cells

Given that the expression of CD160 did not seem to correlate with higher cytotoxic potential, we compared the expression of perforin in CD160⁺ and CD160^– tetramer⁺ cells. We found that CD160^– cells had significantly higher levels of perforin, both in healthy donors and immunosuppressed patients (Fig. 5).

Fig. 5.

Expression of perforin in CD8⁺ tetramer⁺ CD160⁺ cells and CD8⁺ tetramer⁺ CD160^– cells from healthy donors (n = 23) (a) and asymptomatic transplant recipients (n = 29) (b). The median of results is represented. TM: tetramer.

CMV-specific CD8⁺ CD28⁺ CD57^– T cells cultured in the presence of antigen gradually acquire CD160 expression and lose CD28 expression

Immunomagnetically selected CD28⁺ cells from healthy donors (n = 3) were cultured for 20 days in the presence of antigenic peptide and IL-2, as described in Materials and methods. Results are shown in Table 1. As expected, in tetramer⁺ CD8⁺ CD28⁺ lymphocytes, CD28 gradually expression decreased and CD57 expression increased concomitantly. The expression of CD160 increased progressively as the expression of CD28 decreased.

Table 1.

Longitudinal study of expression of CD160, CD28 and CD57 on tetramer⁺ cells from immunomagnetically selected CD8⁺ CD28⁺ CD57^– T lymphocytes from a healthy donor. Lymphocytes were stimulated with 0·5 µg/ml of antigenic peptide (CMVpp65_495−503) and were cultured in the presence of 50 U/ml of rhIL-2. Similar results were obtained in three independent trials (n = 3).

	% CD28+CD57–	% CD28–CD57–	% CD28–CD57+	% CD160+
Day 0	100	0	0	42
Day +4	93	0	0	41
Day +6	69	10	8	81
Day +10	10	42	32	80
Day +20	1·3	54	41	85

Discussion

In this study, we measured circulating CMV-specific CD8⁺ T cells by tetramer staining and analysed the expression of perforin in seropositive healthy donors and asymptomatic renal transplant recipients (who were undergoing immunosuppressive therapy). It is known that immunosuppression enhances CMV replication, leading to expansion of virus-specific cells in an effort to control virus replication [13, 22, 23]. Accordingly, Tx recipients showed significantly higher levels of tetramer⁺ cells, with significantly more perforin than was found in healthy donors. As expected, the patient who experienced clinical CMV reactivation showed an extremely high number of virus-specific CD8⁺ perforin⁺ T cells and this level declined gradually throughout the follow-up period.

We assessed different subpopulations within the tetramer⁺ compartment, defined by the expression of CD28 and CD57. As mentioned above, repeated antigenic stimulation induces CD8⁺ T cells to differentiate from CD28⁺ CD57^– cells into terminally differentiated CD28^– CD57⁺ cells [14–17]. Our results showed that cytolytic potential increased as this differentiation process took place: tetramer⁺ CD28⁺ CD57^– cells showed very low levels of perforin and IFN-γ, which are the principal mediators of cytotoxic activity [24]. The high expression of IL-7Rα/CD127 in these cells identified them as memory cells [19–21]. Conversely, tetramer⁺ CD28^– CD57⁺ cells showed significantly higher levels of perforin and IFN-γ and very low expression of IL-7Rα, indicating that this subset consisted of effector cells with high cytotoxic potential. The CD28^– CD57^– subpopulation was in an intermediate position between the other two with regard to the three parameters analysed.

Tetramer⁺ cells from Tx recipients included significantly more CD28^– CD57⁺ cells than were found in healthy donors and, conversely, cells from Tx recipients included significantly fewer CD28⁺ CD57^– cells than were found in healthy donors. This increase in effector cytotoxic cells within the virus-specific compartment in immunosuppressed patients most probably reflects an enhanced response directed at keeping virus replication under control. Interestingly, as the length of the period of immunosuppression increased, the percentage of CD28^– CD57⁺ cells increased significantly in the tetramer⁺ subset from Tx recipients. Consistent with this scenario is a finding that has also been reported by others [25, 26]: a significant positive correlation in Tx recipients between the size of the CMV-specific CD8⁺ T cell subset and the content of CD28^– CD57⁺ cells in this subset. This correlation, which was not observed in healthy donors, indicates that viral replication induced both the expansion of anti-viral CD8⁺ T cells and the differentiation of these cells into cytotoxic effector cells.

The aim of this study was to analyse the expression of BY55/CD160 in subpopulations of antigen-specific CD8⁺ T cells that exhibited different degrees of cellular differentiation. From the results described above, we conclude that CMV-specific CD8⁺ T cells constitute a heterogeneous subset that includes memory and effector cells, with significantly heterogeneous levels of perforin expression. Such heterogeneity is emphasized when this subset is analysed in different clinical conditions, which presented varying intensities of anti-viral immune response.

When we analysed the expression of CD160 in the previously described subpopulations of CMV-specific CD8⁺ T cells, we found that CD160 expression did not correlate with the expression of perforin. Effector cytotoxic CD28^– CD57⁺ cells expressed similar levels of CD160 as memory CD28⁺ CD57^– cells which are devoid of perforin. CD160 expression was maximal in the CD28^– CD57^– intermediate subpopulation. In addition, CD160 was absent from the surface of effector CMV-specific CD8⁺ T cells, which showed maximal levels of perforin expression during virus reactivation. These cells did not begin to express CD160 until the beginning of the fourth month of the follow-up period.

When we compared the expression of perforin between CD160⁺ and CD160^– tetramer⁺ cells we found that CD160^– cells had significantly higher levels of perforin, both in healthy donors and immunosuppressed patients. Nikolova et al. have reported higher levels of granzyme B in CD160⁺ CD8^high+ cells than in CD160^– CD8^high+ cells, both in HIV⁺ and HIV^– subjects. However, this analysis was performed in total mononuclear cells and from only a few subjects (n = 5) [27]. To our knowledge, our study is the first to deal with CD160 expression in antigen-specific CD8⁺ T cells.

Our results show that cell surface expression of CD160 cannot be used as a cytotoxic marker in CMV-specific CD8⁺ T cells. Most of the available data that attribute this functional significance to the presence of this molecule in the surface of CD8⁺ T cells are based on the observation that this molecule is expressed primarily by CD28^– T cells [7, 28], and it is well known that this subset contains terminally differentiated effector-memory cells [29–31]. However, our results demonstrated that this subset also includes cells with a lesser degree of differentiation. The terminally differentiated effector subpopulation, with highest expression of perforin and lowest expression of CD127/IL-7Rα, was CD28^– CD57⁺ and these cells expressed significantly less CD160 than did the CD28^– CD57^– subpopulation, which displayed a minor degree of differentiation and perforin expression. Interestingly, the expression of CD160 was maximal in CD28^– CD57^– cells, which have lost CD28 expression but have not yet acquired CD57 expression. Cultures of immunomagnetically selected CD28⁺ cells that were grown in the presence of antigenic peptide and IL-2 for 20 days showed that when CMV-specific CD8⁺ T lymphocytes began to lose CD28 expression, they acquired reciprocally CD160 expression. CD160 could be a co-stimulatory molecule that acts in the absence of CD28. In this sense, Nikolova et al. have demonstrated that CD160 functions as a co-receptor upon T cell activation, and provides co-stimulatory signals that lead to the expansion of CD28^– T lymphocytes [27, 28]. It is well known that there exist co-stimulatory molecules that function as alternatives to CD28, and presumably act when CD28 is not present. CD160 is a potential candidate for this function, as are molecules such as signaling lymphocyte activation molecule (SLAM) [32], among other candidates. Based on our observations, we suggest the following scenario: in response to repeated antigenic stimulation, CD28⁺ T lymphocytes would lose CD28 expression and would enhance CD160 expression concomitantly. As the process of differentiation proceeded further, CD8⁺ T lymphocytes would acquire CD57 expression and would modulate CD160 to lower levels of surface expression.

This down-modulation might be performed through a process of shedding, as described for this molecule in activated NK cells [33], as well as for many other GPI-anchored molecules [34]. The fact that terminally differentiated CD8⁺ CD28^– CD57⁺ T cells express significantly less CD160 than CD28^– CD57^– cells can be interpreted in the context of a general down-regulation of co-stimulatory molecules occurring in a subset of cells that possess very high cytotoxic activity.

In summary, we report here that the cell surface expression of CD160 cannot be considered a cytotoxic marker in CMV-specific CD8⁺ T lymphocytes.