Quiescence and functional reprogramming of Epstein-Barr virus (EBV)–specific CD8⁺ T cells during persistent infection

An immune response to antigen induces the activation and expansion of specific T lymphocytes, followed by the clearance of the majority of these cells when the antigen is eliminated.¹  However, the rescue of some antigen-specific T cells from apoptosis is essential for the maintenance of a memory T-cell pool.²  Herpes viruses, such as Epstein-Barr virus (EBV)³  and cytomegalovirus (CMV),⁴  induce persistent latent infection. The memory T-cell pool for such viruses will be subjected to repeated stimulation in response to viral reactivation in vivo. Since the repeated stimulation of T lymphocytes leads to progressive telomere erosion that ultimately leads to growth arrest or replicative senescence,^5,6  factors capable of actively inducing quiescence in memory T cells between episodes of virus reactivation are crucial for long-term maintenance of T-cell memory.^3,7 

A feature of memory T cells that are specific for persistent viruses such as EBV and CMV is the emergence of an antigen-experienced CD8⁺ T-cell population that reexpresses CD45RA (CD45 which predominantly express exon A).⁸  These CD45RA reexpressing T cells also accumulate in elderly individuals.⁹  Some studies have suggested that these cells are close to end-stage differentiation.^10,11  However, more recent reports indicate that they can be reactivated to proliferate,^7,8,12  although to a lesser extent than other T-cell subsets¹³  and represent a stable memory population.^8,14  As CD8⁺ T cells that reexpress CD45RA accumulate during chronic viral infections, they are likely to be involved in the long-term preservation of immunity to agents that induce life-long infection.^3,4  However, the nature of factors that induce quiescence and/or CD45RA reexpression in human virus-specific CD8⁺ T cells and the functional role of these cells have not been defined.

The relative expression of positive and negative regulators of cell cycle in human memory T cells is poorly characterized. Cyclins D (D₁, D₂-D₃), represent positive regulators of proliferation, but only cyclins D₂ and D₃ are found in human T cells.¹⁵  Cyclins are required to form complexes with cyclin-dependent kinases (cdk's) to function. During G₁ phase progression, these cyclin/cdk complexes phosphorylate members of the pocket protein family, leading to the release of the transcription factor E2F.¹⁶  This results in the transcription of genes required for progression from G₁ to S-phase of the cell cycle.¹⁷  Cell-cycle progression is repressed by cdk inhibitors (CKIs) such as p27, which is a member of the cycle inhibitor 21 and kinase inhibitor p27 (Cip/Kip) family of inhibitors.¹⁸  In this study we sought to determine whether the induction of quiescence in activated virus-specific CD8⁺ T cells is due to down-regulation of the cyclins or up-regulation of cyclin inhibitors or both.

To identify factors that induce either quiescence or CD45RA reexpression in human CD8⁺ T cells, we activated and expanded EBV-specific CD8⁺ CD45RO⁺ T cells with viral peptides in vitro as a model of chronic viral infection in humans. Interferon α (IFN-α) and IFN-β belong to the type I IFN family, signal via the same receptor, and, thus, have been shown to share the same antiproliferative, antiapoptotic, and antiviral characteristics.^19-22  We will refer to these cytokines collectively as IFN-I throughout this manuscript. We investigated the ability of IFN-I and also interleukin 15 (IL-15) to promote survival, quiescence, and/or CD45RA reexpression in these cells. We show that, while IFN-I induces quiescence, only IL-15 induces CD45RA reexpression. Furthermore, CD45RA reexpressing EBV-specific CD8⁺ T cells represent a population that has been functionally refocused to mediate effector function without excessive proliferation. This would be crucial for the maintenance of specific cells during life-long stimulation by chronic infective agents.

Heparinized peripheral blood was collected from patients with acute infectious mononucleosis (AIM), from patients who recovered from AIM at least 10 years previously (chronic phase), and from healthy donors with no history of AIM. Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll density centrifugation and cryopreserved in liquid nitrogen. Purified CD8⁺ T-cell subsets were separated from PBMCs by negative selection using the VARIO MACS (magnetic-activated cell sorting) system (Miltenyi Biotec, Surrey, United Kingdom) as described in detail elsewhere.²³  Approval was obtained from the Royal Free and University College Medical Ethics Group, University College London institutional review board for these studies. Informed consent was provided according to the Declaration of Helsinki.

Soluble phycoerythrin (PE)–labeled peptide-HLA class I tetramers (Proimmune, Oxford, United Kingdom) containing the HLA-B8–restricted peptide ligand RAKFKQLL from the EBV protein BZLF1 (abbreviated to RAK) were used to identify EBV lytic epitope-specific CD8⁺ T cells.²⁴ 

Specific CD8⁺ T-cell lines were generated by adding RAK peptide (Alta Bioscience, The University of Birmingham, United Kingdom) to freshly isolated PBMCs of individuals with chronic infection and who were HLA-B8 positive. Cells were incubated for 3 days at 37°C at which point 20 U/mL exogenous human recombinant IL-2 (R&D Systems, Abingdon, United Kingdom) was added. Cells received fresh complete medium (RPMI, 100 IU penicillin, 100 μg/mL streptomycin, 2 mM l-glutamine, and 10% fetal calf serum [Life Technologies, Paisley, United Kingdom]) and 20 U/mL exogenous IL-2 every 3 days. RAK-specific CD8⁺ T cells were restimulated every 14 days.

Freshly isolated PBMC samples were analyzed by flow cytometry using PE-conjugated HLA class I-peptide tetrameric complexes (Pro-immune) as previously described.⁸  Specific antibodies used included anti-CD3 and anti-CD8 (Dako, Cambridgeshire, United Kingdom), anti-CD45RA and anti-CD45RO (BD Biosciences, Oxford, United Kingdom). Staining of intracellular proteins has been described elsewhere.⁸  Immunoglobulin G₁ (IgG₁) isotype controls were used in all experiments.

Cells were stained with anti-CD8 and RAK tetramer (Pro-immune), permeabilized with Fix and Perm reagent (CALTAG-Medsystems, Silverstone, United Kingdom), and then stained with anti-Ki67 FITC (Dako). Appropriate IgG₁ isotype antibodies as well as resting PBMCs were used as negative controls.

Cell-cycle analysis was also determined by 7-aminoactinomycin D (7AAD) staining of DNA; freshly isolated PBMCs were labeled with RAK tetramer and other antibodies to identify particular T-cell subsets. Cell suspensions were permeabilized using Cytofix/Cytoperm (BD Biosciences). After incubation at 4°C for 15 minutes in Cytofix/Cytoperm and 2 further washes, 7AAD (40 μg/mL) was added to cell pellets. Noncycling, resting PBMCs were used as negative controls.

Loss of carboxy-fluorescein diacetate, succinimidyl ester (CFSE; Sigma-Aldrich, Gillingham, United Kingdom) staining was also used as a marker of proliferation. T cells were washed twice and resuspended in 1 mL phosphate-buffered saline. CFSE (0.5 μM/mL) was added, and cells were incubated for 5 minutes at room temperature then washed in PBS.

The rescue of T cells from cytokine-dependent death has been described in detail previously.⁸  Fibroblast-conditioned medium, a physiologic source of IFN-β, was used to prevent apoptosis.^19,20  Viable cell recovery (VCR) was determined by trypan blue dye exclusion after 24 hours.

EBV-specific CD8⁺ T cells were activated with peptide-pulsed autologous antigen-presenting cells (APCs), and maintained in culture for 3 weeks. T cells were washed and assessed for CD45 expression by flow cytometric analysis. T cells (2 × 10⁶) were subsequently added to 24-well inserts with a 0.4-μm pore diameter (BD Biosciences; Discovery Labware, Oxford, United Kingdom), which were then placed into wells of a 24-well plate containing confluent human embryonic lung fibroblasts as a source of IFN-β.^19,20  The cultures were replenished with complete medium every 2 days. CD45 expression and VCR were assessed up to 19 days later.

CD8⁺ CD45RO⁺ T cells were purified as already described and cultured with 20 U/mL exogenous IL-15 (R&D Systems). CD45RA reexpression and cell-cycle progression were determined by flow cytometry 7 and 13 days later.

EBV-specific CD8⁺ T cells that were cultured with either medium, IL-2, IFN-2α, or IFN-β were restimulated as described in maintenance of T-cell lines. After 2 hours, 5 μg/mL Brefeldin A (Sigma-Aldrich) was added, and cells were incubated overnight at 37°C. These cells were subsequently washed and stained for surface antigens and then for intracellular IFN-γ as described.²⁵  Perforin expression in these cells was also determined 5 days after restimulation. Five days after reactivation and after staining with surface antigen, T cells were permeabilized, stained with anti-Perforin–fluorescein isothiocyanate (FITC; BD Biosciences), and fixed.

CD8⁺ T-cell subsets were sorted using a MoFlo cell sorter on the basis of CD27 and CD45RA expression. All CD8⁺ T-cell subsets were activated with a concentration range of anti-CD3 and irradiated autologous APCs. Proliferation was determined by tritiated thymidine incorporation 72 hours later. To determine the cytotoxic potential of CD8⁺ T cells, sorted T cells were stimulated with a concentration range of anti-CD3 plus exogenous APCs in conjunction with antibodies directed against CD107 A and B.²⁶  Cells were incubated for 1 hour at 37°C after which Monensin A 100 μM (Sigma-Aldrich) was added, and cells were incubated for a further 4 hours at 37°C. Cells were stained for other surface markers at this point.

Whole-cell extracts were prepared by lysing cells with 4 times the packed cell volume of lysis buffer (1% Nonidet P-40 ([Octylphenoxy]polyethoxyethanol), 100 mM NaCl, 20 mM Tris (tris(hydroxymethyl)aminomethane)–HCl pH 7.4, 10 mM NaF, 1 mM Na₃VO₄, and protease inhibitors; Roche, Lewes, United Kingdom) on ice for 15 minutes. Protein yield was quantified by Bio-Rad Dc protein assay kit (Bio-Rad, Hemel Hempstead, United Kingdom). Lysate (50 μg) was separated by sodium dodecyl sulfate (SDS)–polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes (Technopath, Limerick, Ireland), and specific proteins were detected using antibodies. The monoclonal antibodies against cyclin D₂ and cyclin D₃ antibody were acquired from NeoMarkers (Union City, CA). The actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and p27 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). The antibodies were detected using horseradish peroxidase–linked goat anti–mouse or anti–rabbit IgG (Dako) and visualized by the enhanced chemiluminescent (ECL) detection system (Amersham Biosciences, Chalfont St. Giles, United Kingdom). Normalization ratios were calculated by dividing the optical density value of the antibody-positive protein band by that of the associated β-actin control band. Optical density values were calculated using Quantity One software (Bio-Rad)

The Student t test was used to determine the significance of the results.

There is considerable expansion of EBV lytic epitope-specific CD8⁺ T cells during infectious mononucleosis (AIM; Figure 1A). However, once the infection resolves, the virus-specific T-cell population contracts (Figure 1B). While the majority of EBV-specific CD8⁺ T cells are in cycle during AIM, the same EBV-specific memory populations are largely quiescent (Figure 1C). To investigate factors that could induce the quiescence in these populations, we stimulated EBV-specific CD8⁺ T cells from individuals with chronic infection (Figure 1C) with specific peptide (RAK)–pulsed autologous APCs (Figure 1D). After 2 stimulations over 4 weeks of culture, the percentage of CD8⁺ T cells that were specific for RAK increased from 0.6% (Figure 1D) to 96% (Figure 1E). In addition, before stimulation in vitro less than 3% of these cells were Ki67⁺, while greater than 80% of cells that were maintained in culture for 4 weeks expressed this marker (Figure 1F).

We confirmed that the in vitro–expanded EBV-specific cells were in cycle by tritiated thymidine incorporation (Figure 2A). Peptide was removed by washing the cells, and the ability of either IL-15 or IFN-I to induce quiescence and or survival after 4 days in culture was determined. The virus-specific CD8⁺ T cells that were cultured in IL-15 were highly proliferative (Figure 2B) and showed good viable cell recovery, due in part to the cell expansion (Figure 2C). In contrast, virus-specific CD8⁺ T cells that were incubated in medium alone or those that were cultured in IFN-I were not actively cycling (Figure 2B). However, there was poor survival in cells that were cultured in medium alone compared with those cultured in the presence of IFN-I (Figure 2C). These results show that, while survival of virus-specific CD8⁺ T cells can be promoted by both IL-15 and IFN-I, only the latter cytokines were capable of inducing quiescence. These findings are in agreement with our previous studies in which both IL-15 and IFN-I were shown to inhibit apoptosis in EBV-specific CD8⁺ T cells isolated from patients with AIM.^27,28  However, in accordance with the results described in this report, previous reports have shown that only IFN-I was capable of inducing quiescence in activated T cells.^27,28 

It was not clear whether virus-specific CD8⁺ T cells that were maintained in IL-15 or IFN-I retain their ability to be reactivated by antigen. We, therefore, restimulated EBV-specific CD8⁺ T cells that had been incubated in medium alone or fresh medium supplemented with IL-15 or IFN-I for 4 days with RAK-pulsed irradiated APCs plus exogenous IL-2. We found that cells that had been cultured in all the different conditions could be induced to proliferate (Figure 3A) and up-regulated both IFN-γ (Figure 3B) and perforin (Figure 3C). These results indicate that irrespective of whether virus-specific CD8⁺ T cells are maintained in a resting or an activated state, they retain their capacity to react functionally to antigenic rechallenge.

It was not clear whether the induction of quiescence in activated virus-specific CD8⁺ T cells was a result of changes in the expression levels of cyclins or CKIs. To address this, we first investigated the expression of 2 key cyclins, namely cyclin D₂ and D₃, and also the CKI p27 in freshly isolated resting CD8⁺ T cells and activated CD8⁺ T cells from a patient with AIM that had been activated in vivo⁸  (Figure 4A). Freshly isolated CD8⁺CD45RA⁺ and CD8⁺CD45RO⁺ T cells expressed high levels of p27 (Figure 4A). While CD8⁺CD45RA⁺ T cells did not express cyclins D₂ or D₃, CD8⁺CD45RO⁺ T cells showed low cyclin D₂ expression. In contrast, CD8⁺ T-cell populations from patients with AIM showed high levels of cyclin D₃, low levels of cyclin D₂, and considerably decreased expression of p27 compared with resting populations. The cyclin signature of activated EBV-specific CD8⁺ T cells that were expanded by specific peptide in vitro (Figure 4A, lane 3) resembled that of the same populations activated in patients with AIM in vivo (Figure 4B, lane 1). There was up-regulation of cyclin D₃, low to negative cyclin D₂, and considerably down-regulated p27 in these cells.

To simulate the effects of immune resolution, we removed IL-2 and peptide from the EBV-specific CD8⁺ T cells that were generated in vitro and examined the expression of cell-cycle regulators (Figure 4B, lane2). After 4 days of culture, cells were not in cycle and showed down-regulation of cyclin D₃ and up-regulation of p27. However, in the absence of exogenous cytokines there was considerable apoptosis that eventually led to the loss of all cells (see Figure 2). This indicates that additional signals are required for the persistence of these cells. As shown in Figure 2, both IL-15 and IFN-I promote survival of EBV-specific CD8⁺ T cells; however, the cyclin profiles of cells rescued by either group of mediators were quite distinct. Maintenance in IFNα–induced quiescence associated with down-regulation of cyclin D₃ and up-regulation of p27 (Figure 4B, lane 4), while maintenance in IL-15 induced cell cycle associated with high cyclin D₃ levels and low p27 expression (Figure 4B, lane 4). Normalization ratios (based on optical densities) of antibody-positive protein bands versus β-actin controls were calculated to normalize the differences in β-actin expression between samples examined (Tables 1 and 2) support the results described here. It is possible that up-regulation of p27 is due only to growth factor removal, since elevated p27 levels were also found in cells maintained in fresh medium alone (Figure 4B, lane 2). However, previous studies have shown that IFN-α can also directly induce p27.^21,29 

Since virus-specific memory CD8⁺ T cells that reexpress CD45RA are quiescent,⁸  we investigated whether induction of cell-cycle exit in these cells by IFN-I was linked to CD45RA reexpression (Figure 5). We removed both peptide and IL-2 from EBV-specific CD8⁺ CD45RO⁺ cells that were expanded in vitro and cocultured these cells in a dual chamber system with fibroblasts. We have performed extensive studies to show that fibroblast-derived IFN-I promotes activated T-cell survival without proliferation in a manner identical to that of exogenous IFN-I.^20,30  Furthermore, blocking anti–IFN-I receptor (IFNR) antibodies inhibit the survival and quiescence-promoting effects of fibroblast-conditioned medium.^20,25  The advantage of using fibroblasts in this coculture system is that we are investigating a physiologic source of this cytokine that is continuously secreted during the culture period and, therefore, does not need to be replenished during the experiment.

CD45RO⁺ expression, as expected, was found to be enhanced in both total CD8⁺ T cells (Figure 5A) and in EBV tetramer⁺ T cells (Figure 5B) isolated from the same patient with AIM, when compared with CD45RO⁺ expression in freshly isolated EBV-specific CD8⁺ T cells isolated from a patient with chronic infection (Figure 5C). The proportion of freshly isolated tetramer⁺ cells, isolated from a donor with chronic EBV infection, expressing CD45RO⁺, both CD45RO⁺ and CD45RA⁺, and CD45RA⁺ alone was 34%, 20%, and 46%, respectively (Figure 5C). After 28 days of culture with peptide and IL-2, 98% of these virus-specific T cells were CD45RO⁺ (Figure 5D), and 35% of these cells were in active cell cycle (Figure 5E). When these cells were washed and cocultured with fibroblasts for 7 days, 99% of cells retained CD45RO expression (Figure 5F), but only 0.4% (Figure 5G) was in the active phases of cell cycle. This showed that the induction of quiescence could be dissociated from CD45RA reexpression in virus-specific CD8⁺ T cells.

It has been shown previously that IL-15 can induce CD45RA reexpression in polyclonal populations of central memory CD8⁺ T cells (CD45RO⁺ CCR7⁺) but not from the more differentiated effector memory (CD45RO⁺, CCR7^–) populations.¹³  The EBV-specific populations that we expanded in vitro and used in the previous experiments were highly differentiated and of the effector memory type (not shown). It was, therefore, possible that the lack of CD45RA reexpression observed in the previous experiments was due to extensive differentiation beyond the point where reexpression of this molecule could be induced. We, therefore, investigated the ability of different cytokines to mediate CD45RO to CD45RA reversion in freshly isolated CD8⁺ T cells that consisted of a large proportion of cells of the central memory type (data not shown).

We found that only IL-15 could generate CD45RA “revertants” from CD45RO⁺ EBV^–-specific T cells that were isolated from fresh PBMCs obtained from donors who had AIM at least 10 years previously (Figure 6). The freshly isolated EBV-specific CD8⁺ CD45RO⁺ population contained greater than 0.01% of cells that expressed CD45RA (Figure 6A, far left). These cells were cultured with either FCM (a recognized physiologic source of IFN-I^19,20  which will be referred to as IFN-I for the rest of this manuscript), exogenous IL-2, IL-7, or IL-15 for 7 and 13 days, and the extent of cell cycling was determined by Ki67 expression (Figure 6A). As expected, there was no proliferation in the presence of IFN-I. In contrast, 55% of EBV tetramer⁺ cells that were cultured for 7 days in IL-15 were found to be Ki67 positive (Figure 6A). However, a lower percentage of EBV-specific CD8⁺ T cells exposed to IL-2 and IL-7 were found to be proliferating (12% and 14%, respectively) as determined by Ki67 expression (Figure 6A). The extent of cycling cells was reduced in all conditions by day 13 (Figure 6A). IL-15, but not IFN-I, IL-2, or IL-7, induced significant CD45RO to CD45RA reexpression in EBV tetramer⁺ T cells (Figure 6B), and by day 13 up to 16% of the cells showed complete loss of CD45RO. When examined on day 7, 27% and 13% of cells previously exposed to either IL-2 or IL-7, respectively, were CD45RA⁺. However, on day 13 very few CD45RA single-positive antigen-specific T cells were found in conditions in which either IL-2 (2%) or IL-7 (0%) were added (Figure 6B). By day 13, the number of cells found in conditions exposed to either IL-2 or IL-7 was much less than in those in which IL-15 was added (Figure 6B). This might be due to the fact that cells exposed to either IL-2 or IL-7 proliferated to a lesser degree, as demonstrated by a reduced Ki67 expression on day 7 (Figure 6A).

In these experiments, IFN-I was added at the beginning of culture and not replenished. In other similar experiments, where a dual chamber coculture system was used (confluent human embryonic lung fibroblasts and CD8⁺ T cells), no proliferation or CD45RA reexpression by freshly isolated CD8⁺CD45RO⁺ T cells was observed even after 3 weeks of coculture (data not shown). Finally, similar results were achieved in experiments in which total CD8⁺ T cells were examined (data not shown).

To rule out the possibility that CD45RA⁺ reexpression on EBV-specific CD8⁺ T cells was due to the expansion of small numbers (< 0.01%) of CD45RA⁺ contaminating cells, the CD45RO⁺ T cells were stained with CFSE prior to exposure to IL-15, and loss of staining was determined by FACS analysis 7 and 13 days later. We calculated the extent of expansion compared with the number of cell divisions that had occurred. These results showed that expansion of a contaminant population could not account for the frequency of EBV-specific CD8⁺ T cells observed (data not shown).

We next examined the effect of IL-15–induced “homeostatic” proliferation on cyclin expression in freshly isolated CD8⁺ CD45RO⁺ T cells (Table 1). These cells are not in cycle and express elevated levels of p27, no cyclin D₃, and low levels of cyclin D₂ (Figure 6C, lane 1). IL-15 induced these cells to proliferate and caused up-regulation of cyclin D₃ after 7 days (Table 1). Cyclin D₂ expression was also reduced, suggesting that this cyclin was not required for IL-15–induced homeostatic proliferation. At day 13, when the number of cells cycle in these cultures was much reduced (Figure 6A), cyclin D₃ levels were reduced significantly (Figure 6C, lane 3). The p27 expression remained relatively constant throughout days 0 to 13 (Table 3). Finally, freshly isolated resting CD8⁺ CD45RA⁺ revertants express significant amounts of p27 as well as low levels of cyclins D₂ and D₃ (Table 3). Collectively, these results indicate that IL-15–induced homeostatic proliferation can induce CD45RA reexpression in EBV-specific CD8⁺ CD45RO⁺ T cells and that this proliferation is associated with a transient elevation in cyclin D₃. Normalization ratios (based on optical densities) of antibody-positive protein bands versus β-actin controls were calculated to normalize the differences in β-actin expression between samples examined (Table 3). These ratios support the results described here.

The functional role of CD8⁺ T cells that reexpress CD45RA is not known. Previous reports suggest that, although these cells can be reactivated,^7,8,12  the extent of proliferation elicited is less than that observed in other CD8 subsets.¹³  To address this we investigated the dose dependence for proliferation of freshly isolated naive (CD45RA⁺ CD27⁺), central memory (CD45RA^– CD27⁺), effector memory (CD45RA^– CD27^–), and revertant memory (CD45RA⁺ CD27^–) T cells after anti-CD3 and autologous APC stimulation (Figure 7A). Revertant T cells proliferated the least in response to T-cell receptor (TCR) stimulation, whereas the nave T-cell subset showed the best response (Figure 7A, left). However, the revertant CD8⁺ T cells exhibited the greatest cytotoxic activity (Figure 7A, right) even at low concentrations of anti-CD3 antibody, showing that they can mediate efficient effector function without proliferation. Furthermore, these cells were able to synthesize greater amounts of IFN-γ in response to stimulation compared with naive and central memory CD8⁺ T cells (Figure 7B). Finally, in Figure 7C we show that the CD8⁺ CD45RA⁺ revertant T cells isolated from healthy donors show almost complete loss of the costimulatory markers CD27 and CD28 that are crucial for stimulation, suggesting that this population has been functionally reprogrammed to mediate effector function instead of proliferation. It was not possible to carry out similar experiments on EBV-specific CD8⁺ T cells, as all 4 T-cell subsets, defined by the relative expression of CD27 and CD45RA, are not represented in these antigen-specific populations. In addition, there are too few EBV-specific CD8⁺ T cells in patients with chronic EBV infection to examine cytotoxic and proliferative function.

Persistent infection imposes a significant strain on the proliferative capacity of CD8⁺ virus-specific T cells, and repeated episodes of T-cell activation may lead to the loss of specific T cells through telomere erosion and replicative senescence.^5,6  Mechanisms that can keep memory T cells in a resting state, between episodes of stimulation, are, therefore, crucial to optimize their persistence in vivo. In addition, the ability to induce functional activity in memory T cells, without the need for proliferation, would minimize telomere erosion and prevent the loss of cells by replicative senescence. This would be particularly important for the maintenance of T-cell memory to chronic infective agents, such as EBV and CMV, whereby memory T cells may have to persist for more than 8 decades without significant input of new cells from the thymus.⁶  In this regard, the 3 key observations in this study are, first, that quiescence in activated EBV-specific CD8⁺ T cells can be induced by IFN-I. Second, CD45RA-reexpressing CD8⁺ T cells that accumulate during chronic viral infections and in elderly individuals represent an effector population that mediates potent functional activity without the need for proliferation. And finally, the reexpression of CD45RA by these EBV-specific CD8⁺ T cells can be induced during IL-15–mediated homeostatic proliferation.

We investigated the effects of IL-15 and IFN-I on EBV-specific T cells because these cytokines have been shown to be crucial for the maintenance of CD8⁺ T-cell memory in vivo; however, their mechanism of action is unclear.^31-34  Both these groups of cytokines have been shown to prevent T-cell apoptosis in vitro and in vivo, which may be one way by which memory T cells are preserved.^{2,19,27,33,35,36}  However, these cytokines may also have additional effects on memory CD8⁺ T-cell persistence through the induction of quiescence and CD45RA reexpression. Interestingly, these phenomena were found to be distinct; IFN-I induces quiescence, while IL-15 induces CD45RA reexpression. It must be noted that the timing and concentration levels of these 2 cytokines are likely to be crucial as to the effect on local T cells. IFN-α–induced IL-15 has been shown to be important for driving immune responses,³⁰  whereas it is an established fact that IFN-α exhibits potent antiviral and immunomodulatory roles at the start of an immune response.³³  However, once local supplies of γ-chain cytokines and antigen-presenting cells become exhausted, it is probable that IFN-I, secreted by stromal cells in the microenvironment,^6,28  can promote the survival of any remaining activated T cells. We (data not shown) and others¹³  have suggested that IL-15 is not capable of generating CD45RA revertants from TCR-activated T cells. It is likely, therefore, that IL-15–mediated reversion occurs in noninflamed tissues as part of the homeostatic maintenance of CD8⁺ memory T cells in persistent viral infection.

We examined cyclin expression in freshly isolated virus-specific T cells to determine the molecular signatures of quiescence and activation in these cells. Recent reports in mice showed that resting CD8⁺ memory T cells possess high cyclin D₃,^37,38  which may permit a more rapid response to recall antigen.³⁸  In contrast, we found high cyclin D₂ expression and the complete absence of cyclin D₃ in resting human CD8⁺CD45RO⁺ memory T cells. These discrepancies may reflect species-specific differences.³⁹  EBV-specific CD8⁺ T cells that were freshly isolated from patients with AIM or activated in vitro showed very similar cyclin signatures with up-regulated cyclin D₂ and D₃ but down-regulation of p27. The rescue of activated EBV-specific T cells from apoptosis with either IFN-I or IL-15 induced different cyclin profiles, the addition of IFN-I down-regulated cyclin D₃, whereas IL-15 induced cyclin D₃ expression. While we cannot rule out the possibility that the increase in p27 by IFN-I in our experiments is simply due to removal of the stimulus, it is likely that there is also a component of active p27 induction by these cytokines.^21,29,40 

Primed CD8⁺ T cells that reexpress CD45RA are found after a variety of different viral infections, including EBV, CMV, and HIV in humans ^{4,7,8,13,24,41-43} , and a similar subset is also found in animals.⁴⁴  Our data suggest that the induction of CD45RA reexpression in CD8⁺ T cells is part of a functional reprogramming of effector memory cells away from proliferation that also includes the loss of both CD27 and CD28 costimulatory molecules. The loss of these costimulatory molecules also occurs in EBV-specific effector memory CD8⁺ T cells that do not reexpress CD45RA.^43,45-47  This suggests that loss of costimulatory molecule expression alone cannot account for the poor proliferative activity of this population. Nevertheless, the potent cytotoxic activity and capacity for IFN-γ secretion by these cells,^8,13,24  together with their capacity to migrate to nonlymphoid tissue,⁴⁵  show that they are ideally suited for a role in immunosurveillance. Signals associated with the generation of CD45RA revertants by IL-15 in virus-specific CD8⁺ T cells are unclear; however, since the IL-2R shares both the β- and γ-chains with the IL-15R⁴⁸  but does not induce reversion (data not shown), IL-15R α-chain–specific signals are likely to be involved.

In summary, memory T cells have to be maintained for many decades in humans, and cells that are specific for persistent infective agents will be repeatedly stimulated during life-long infection. This predisposes these populations to loss by telomere-dependent replicative senescence.^5,6  Quiescence-inducing factors such as IFN-I and those that induce CD45RA reexpression like IL-15 are mediators that may, therefore, be crucial for the long-term maintenance of these cells. It is not clear whether there is an equivalent murine CD8⁺ memory T-cell population that is polarized away from proliferation but toward effector function. However, the much shorter mouse compared with human life span suggests that these cells would be more important in the latter species. A closer understanding of the timing and location of endogenous production of these cytokines in vivo will give a clearer rationale for vaccine therapy, especially in elderly individuals.

We thank Professor Peter C.L. Beverley for valuable discussions and Mr Andrew Worth for excellent assistance with cell sorting.