The mechanisms that maintain memory in T cells are not completely understood. We have investigated the role of antigen and interleukin (IL)-2 in the growth and maintenance of CD8+ T cells using a cytolytic T cell line specific for ovalbumin (OVA)257-264 presented by H-2Kb. This line does not secrete IL-4 or IL-2; hence, stimulation with the OVA-transfected EL4 line (E.G7-OVA) does not induce proliferation without addition of exogenous growth factors. Furthermore, this line can be maintained continuously by weekly addition of irradiated, splenic filler cells and IL-2, with or without E.G7-OVA. Although IL-2 induced proliferation of these cytotoxic T lymphocytes (CTLs), production of interferon γ and tumor necrosis factor α required stimulation of the CTL with E.G7-OVA. The kinetics of lymphokine secretion after stimulation by E.G7-OVA were the same whether the CTL had been maintained with or without antigen (Ag). In addition, both CTL lines killed E.G7-OVA target cells within 4 h. Thus, the effector functions of these CTLs were rapidly induced by T cell receptor (TCR) occupancy. CTLs cultured with or without Ag also served as memory T cells when parked for 100 d in unirradiated, syngeneic recipients without OVA. In the absence of OVA, the precursor frequency was identical in spleens of normal and β2-microglobulin knockout recipients, but significantly less in IL-2 knockout mice. The decline of memory in the absence of IL-2 supports data from other investigators, suggesting that cell cycling is important to the maintenance of CD8+ T cell memory. These data also suggest that stimulation of OVA-specific CTLs by lymphokines seems to be more important to maintaining memory than stimulation of TCRs by cross-reactive peptides complexed to class I molecules.

Memory is a hallmark of the immune response to T cell–dependent antigens. Memory, in both B and T cells, is manifest by stronger and quicker responses upon secondary exposure to antigen. The mechanisms that maintain memory are not completely understood. In part, T cell memory reflects an increase in the frequency of precursor cells. Memory T cells are also qualitatively distinct from naive T cells in that they have less stringent requirements for activation and respond more rapidly than naive T cells (for review see reference 1). Naive and memory T cells are distinct from effector T cells in that the latter are actively engaged in lymphokine secretion and lytic function in the case of CD8+ T cells. By contrast, both naive and memory T cells must be stimulated with antigen to express these effector functions. Naive T cells can be distinguished from effector T cells by differences in the expression of several cell-surface antigens. However, effector T cells cannot be easily distinguished from memory T cells because they both express activation antigens and increased density of adhesion molecules.

Several important questions remain concerning the maintenance of memory T cells. The question of whether antigen is required for memory T cell persistence has been the subject of considerable debate. The majority of studies on the requirement for antigen in maintaining CD8+ memory T cells have been performed using polyclonal T cells activated by viruses such as lymphocytic choriomeningitis virus (LCMV), Sendai virus, or influenza. One of the major obstacles in studying memory in virus-specific, CD8+ T cells is ruling out restimulation by persisting Ags. However, transfer of CD8+ T cells from infected mice into irradiated, syngeneic recipients showed that memory persisted in the absence of virus, as measured by PCR (24). Another important issue is the relationship between effector and memory T cells. Whether memory T cells are derived from effector T cells in a linear fashion or whether they differentiate along separate pathways is not known. Signals that regulate development of memory T cells have not been identified.

We have recently demonstrated that soluble OVA emulsified in CFA (5) or nonionic block copolymer adjuvants primed CD8+ CTL precursors in mice (6). These OVA-specific CTLs (OVA-CTLs) produced IFN-γ and TNF-α upon stimulation with E.G7-OVA or with OVA257-264 in association with H-2Kb (6). However, these OVA-CTL cells do not produce IL-2 or -4. We reasoned that such CD8+ T cells would provide a good model for studying the memory function of CTLs because persistence of Ags would not be a problem. The results presented below compare the requirements for activation of CTLs versus maintenance of memory CTLs in vitro and in vivo. Our results show that CD8+ CTLs specific for the exogenous Ag, OVA, have the same potential to persist as memory T cells specific for viruses and tumors. Moreover, persistence of memory T cells was more dependent upon IL-2 than cross-reactive complexes of peptides and MHC class I antigens.

Materials And Methods

Animals.

C57BL/6 (H-2b), C57BL/10 (H-2b), and BALB/c (H-2d) mice were purchased from the Jackson Laboratory (Bar Harbor, ME). Knockout mice bearing homozygous, targeted mutations for the β2-microglobulin (β2m)1 gene (7), C57BL/6J-β2mtm1Unc, or the IL-2 gene (8), C57BL6J-ll2tm1Hor, also were purchased from the Jackson Laboratory. All of the inbred, knockout mice express the H-2b haplotype and C57BL/6J mice were used as the controls for these experiments as recommended by the Jackson Laboratory. All mice were obtained from pathogen-free colonies and maintained in sterile microisolator cages. Furthermore, no pathogens were detected in sentinel mice housed in the same room with the experimental mice. Female mice, 8–12 wk old, were used in all these experiments. Mice were maintained on standard laboratory chow and water ad libitum in a temperature and light controlled environment. All procedures on animals were conducted according to the principles outlined in the guidelines of the Committee on Care and Use of Laboratory Animals, Institute of Laboratory Animals Resources, National Research Council.

Tumor Cells and Cell Cultures.

The MHC class II negative, H-2b thymomas, EL4 and E.G7-OVA, were used as target and stimulator cells. E.G7-OVA was produced transfection of EL4 cells with the cDNA gene encoding chicken egg OVA (9; provided by M.J. Bevan, University of Washington, Seattle, WA). All cell cultures were maintained in RPMI 1640 medium supplemented with 10% bovine calf serum, 1 mM l-glutamine, 1 mM Na pyruvate, 50 μM 2-mercaptoethanol (2-ME), and antibiotics, at 37°C in 6% CO2 in air.

Generation of Ag-specific CTLs.

Generation of CTL lines from C57BL/10 (B10) were described previously (5, 6). In brief, spleen cells were obtained from mice primed 10 d earlier with 200 μg OVA in CFA. Spleen cells were incubated with irradiated (20,000 rad) E.G7-OVA for 1 wk as described by Moore et al. (9). Effector T cells resulting from these cultures recognized OVA257-264 (SIINFKEL) peptide in the context of H-2Kb (5, 6). Allogeneic B10 α BALB/c CTLs were initiated by incubating spleen cells of B10 (H-2b) with irradiated (2,000 rad) spleen cells from BALB/c (H-2d) mice. The resulting CTLs recognize allogeneic P815 (H-2d), but not syngeneic EL4 (H-2b) targets (5, 10). To establish long-term CTL lines, OVA-CTLs were stimulated weekly with irradiated E.G7-OVA, irradiated, syngeneic splenic filler cells and 5% supernatant from Con A–stimulated rat spleen cells as a source of IL-2 (9, 11). Similarly, B10 α BALB/c CTLs were maintained by weekly stimulation with irradiated spleen cells from BALB/c mice and IL-2 (10).

Cytofluorometric Analysis.

OVA-CTLs were incubated with filler cells and IL-2 in the presence or absence of E.G7-OVA stimulators for 1 wk. Viable T cells were isolated by Ficoll-Hypaque gradient centrifugation (12) and then stained with the following mAbs: FITC-conjugated M1/9.3.4 (anti-CD45; reference 13), 145-2C11 (anti-CD3; reference 14), or PE-conjugated 3C7 (anti–IL-2R; reference 15). FITC- or PE-conjugated rat IgG or hamster IgG were used as isotype-matched control Abs, respectively. After incubation at 4°C for 30 min, cells were washed three times and then fixed with 0.5% paraformaldehyde. Cell staining was analyzed on a FACScan® cytofluorimeter using LYSIS-II software (Becton Dickinson, San Jose, CA). Forward angle, light scatter was used to exclude dead and aggregated cells. The results are presented as fluorescence histograms with the relative number of cells on a linear scale plotted versus the relative fluorescence intensity on a log scale.

Cytotoxicity Assay.

The 51Cr–release assay used in these experiments has been described in detail elsewhere (10). In brief, target cells were labeled with Na2CrO4 (DuPont, Boston, MA) at 37°C for 60 min in Tris-phosphate buffer, pH 7.4, containing 10% FBS. After washing, 51Cr-labeled target cells were added to effector cells in 96-well round-bottomed plates. After a 4-h incubation at 37°C, supernatants were collected and radioactivity was detected in a gamma counter (Wallac, Turku, Finland). Percent specific lysis was calculated as: 100 × ([release by CTL − spontaneous release] / [maximal release − spontaneous release]). Maximal release was determined by addition of 1% Triton X-100 (EM Science, Gibbstown, NJ). Spontaneous release in the absence of CTLs was generally <15% of maximal release.

Measurement of Cytokine Production.

Cytokine production was tested by two-site sandwich ELISA using a mouse IFN-γ kit (Genzyme, Cambridge, MA) or TNF-α kit (Endogen, Inc., Boston, MA) (16, 17). In brief, supernatants from OVA-CTL cultures were added to microtiter plates precoated with anti–IFN-γ or anti–TNF-α capture Ab and incubated overnight at 4°C. After addition of biotinylated detecting Ab and incubation at room temperature for 45 min, avidin-peroxidase was added and incubated at room temperature for 30 min. Plates were washed extensively with 1% Tween in PBS between each step. Finally, ABTS (2,2′-azino-di-[3-ethylbenzthiazoline sulfonate]) substrate containing H2O2 was added and the colorimetric reaction was read at an absorbance of 450 nm using an automatic microplate reader (Molecular Devices Corp., Menlo Park, CA). The concentrations of IFN-γ (U/ml) or TNF-α (pg/ml) were calculated according to the standard curves produced by various concentrations of recombinant cytokines.

Adoptive Cell Transfer.

OVA-CTL or allogeneic B10 α BALB/c CTLs were harvested 7 d after stimulation as described above. Viable CTLs recovered from Ficoll were washed and resuspended in serum-free HBSS medium. Donor T cells, at 5 × 106 to 1 × 107 per mouse, were injected into the tail veins of naive, syngeneic recipient mice. Recipient mice were killed after various time periods as indicated. Spleen cells were serially titrated in 96-well microtiter plates and stimulated in vitro with 2 × 104 irradiated E.G7-OVA, 2.5 × 105 syngeneic filler cells, and 5% Con A supernatant. For frequency analysis of donor CTLs within recipient spleen cells, each serial dilution of spleen cells from recipient mice were plated in 24 wells of 96-well microtiter plates. Cytolytic activity was tested using 104 of 51Cr–labeled E.G7-OVA targets. The data are displayed as the log of the percentage of negative wells on the ordinate versus number of spleen cells per well on the abscissa. Wells were judged positive if lysis was equal to or greater than the average of the background lysis of 51Cr-labeled E.G7-OVA targets plus 3 SD of the background in the absence of target cells.

Results

Requirements for Growth and Activation of CTLs In Vitro.

OVA-CTLs were primed in vivo with OVA in CFA and stimulated in vitro with E.G7-OVA stimulators, syngeneic filler cells, and IL-2 (5). Some studies have shown that allogeneic CTLs could be maintained by recombinant lymphokines in filler cell–free culture systems (1820). However, the requirement for antigen in the maintenance of OVA-specific CTLs has not been reported. Thus, we tested culture conditions under which the CTL growth would be optimal. OVA-CTLs were cocultured with various combinations of stimulators, filler cells, or IL-2, and their growth was determined 1 wk later. As shown in Fig. 1 A, OVA-CTLs grew well in IL-2 alone. The growth rate of OVA-CTLs in cultures containing IL-2 was somewhat greater if E.G7-OVA cells were present, presumably because antigen increases the expression of IL-2R (21, 22). OVA-CTLs not only grew well in the presence of IL-2, exogenous IL-2 was absolutely for the growth of OVA-specific CTL lines and B10 α BALB/c lines (not shown).

The requirements for activation of the effector functions of CTLs were measured using IFN-γ production because this OVA-CTL line does not secrete detectable IL-2. Although exogenous IL-2 was necessary for growth of the CTL in vitro, neither IL-2 nor filler cells were required for IFN-γ production by CTLs activated with E.G7-OVA (Fig. 1 B). By contrast, E.G7-OVA were absolutely essential for the activation of OVA-CTLs, indicating that stimulation of TCRs by complexes of peptides and MHC class I molecules was required to induce lymphokine secretion. In spite of some reports that exogenous IL-2 may enhance spontaneous production of IFN-γ by CD8+ T cells (23, 24), we did not find any significant production of IFN-γ by OVA-CTLs cultured with IL-2 or filler cells in the absence of E.G7-OVA stimulators. These results indicate that growth of OVA-CTLs can be maintained by IL-2 without receptor-mediated stimulation. Activation of the effector function of CTLs required antigen and this response to antigen could be enhanced by IL-2.

Responses to IL-2 can be enhanced by antigen because antigen induces an increase in the density of IL-2R on CD8+ CTLs (21, 22). To verify that the IL-2–induced enhancement of lymphokine secretion that was observed in the previous experiment resulted from the upregulation of IL-2R by antigen, we cocultured OVA-CTLs with filler cells and IL-2 in the presence or absence of E.G7-OVA stimulators for 1 wk and then stained them with mAbs specific for CD45, CD3, or the α chain of IL-2R. There were no significant changes in the expression of CD45 between OVA-CTLs incubated with or without E.G7-OVA (Fig. 2, A and B). Expression of CD3 was slightly increased by stimulation with E.G7-OVA (Fig. 2, C and D). However, the IL-2R level expressed by stimulated OVA-CTLs (Fig. 2,F) was two to threefold higher than unstimulated CTLs (Fig. 2 E). Thus, triggering TCR by antigen upregulated IL-2R expression permitting IL-2 to enhance growth and lymphokine production.

CTLs Cultured with IL-2, but No Antigen, Retained Antigen-specific Reactivity.

Based on the observation that IL-2 was sufficient for maintaining CTL growth in short-term cultures, we grew the OVA-CTLs by repeated stimulation with filler cells and IL-2, but without E.G7-OVA. After culturing for 50 and 100 d, the cytolytic activity of these unstimulated OVA-CTLs was tested. No significant difference in cytolytic activity of OVA-CTLs maintained without antigen was observed by comparison to OVA-CTLs that were restimulated weekly with E.G7-OVA (Fig. 3). The specificity of the CTL was maintained as shown by killing of E.G7-OVA but not EL4 targets. These data suggest that Ag is not required to maintain T cell memory in vitro.

We also investigated whether there were any differences in cytokine production between these OVA-CTL lines maintained with or without Ags. OVA-CTLs were incubated with or without stimulator cells. Supernatants were collected at various times and tested for IFN-γ or TNF-α production by ELISA. A latent phase of 1–2 h was observed before IFN-γ (Fig. 4,A) or TNF-α (Fig. 4 B) were detected in the supernates of the OVA-CTLs that had been stimulated weekly with E.G7-OVA. Activation of the OVA-CTLs that had been maintained with filler cells and IL-2 occurred a bit later and with slightly slower kinetics than seen with OVA-CTLs maintained with Ag. However, comparable levels of IFN-γ or TNF-α were produced by both cultures of OVA-CTLs after 4 h of incubation with E.G7-OVA. Neither of the cultures produced lymphokines after stimulation by EL4. These results confirm that OVA-CTLs, cultured with filler cells and IL-2 for long periods of time, retained their biological functions as well as their antigenic specificity.

The rapidity with which TNF-α and IFN-γ were secreted into the supernates of the CTLs after stimulation with Ags (Fig. 4) raised a question about whether the secreted lymphokines were newly synthesized or released from preexisting stores. However, cycloheximide inhibited the secretion of IFN-γ (Fig. 5,A) and TNF-α (Fig. 5,B) stimulated by E.G7-OVA, suggesting that these proteins are synthesized in response to antigenic stimulation. The finding that actinomycin D also inhibited secretion of the lymphokines (Fig. 5) suggests that new messenger RNA synthesis was also required for lymphokine secretion. Thus, cultured OVA-specific CTLs upregulate lymphokine gene expression within minutes after stimulation by antigen. By contrast, lysis of targets by OVA-specific CTLs was relatively insensitive to inhibitors of protein and RNA synthesis (Fig. 5 C) as would be expected because the lytic proteins, perforin and granzymes, are stored in granules for release upon target cell recognition.

Persistence of Donor CTLs in Recipients.

CTL lines that are stimulated weekly with antigen and/or IL-2 do not revert to a resting state. Such cell lines should be considered to be effector cells rather than memory T cells. To test whether such CTL lines could persist in vivo in the absence of an antigenic stimulus, OVA-CTLs were harvested 7 d after stimulation with irradiated E.G7-OVA. No viable E.G7-OVA remain in these cultures, as shown by the failure of these cultures to stimulate fresh CTLs (not shown). Dead cells and debris were removed from the CTLs by Ficoll-Hypaque centrifugation. Viable OVA-CTLs and B10 α BALB/c CTLs were adoptively transferred to normal, syngeneic mice without secondary challenge. To provide a physiological environment, the recipient animals were not irradiated. Spleen cells from these recipients were harvested 100 d after transfer and stimulated in vitro with E.G7-OVA. After stimulation for 1 and 2 wk, OVA-specific cytolytic activities were detected in spleen cells from mice receiving OVA-specific CTLs (Fig. 6,A). Spleen cells from mice receiving control B10 α BALB/c CTLs did not develop OVA-specific responses (Fig. 6 B), which is consistent with the previous reports that OVA-specific CTLs cannot be induced by stimulation with E.G7-OVA in vitro without previous priming (9). These results suggest that CTLs persisted in vivo in the absence of viable antigen-bearing tumor cells.

One of the main problems with previous studies on the persistence of memory in CD8+ T cells is that the antigens used frequently were viruses that potentially could be carried over into the recipient animals. Contaminating viruses could activate host CTLs that could be confused with the transferred memory T cells. The OVA-CTLs provide an advantage over virus-specific CTLs in that the antigen used to stimulate the CTLs in vitro is E.G7-OVA that had been irradiated with 20,000 Rad. Furthermore, E.G7-OVA do not survive being cultured with OVA-CTLs for 7 d, as illustrated by the fact that no tumor cells grow from these cultures when unirradiated E.G7-OVA were used as stimulators (not shown). Although no intact antigen-bearing tumor cells were transferred with the CTLs, it is conceivable that some persisting antigenic fragments might have been transferred. Therefore, we sought a method to test the OVA-CTLs for the presence of antigen. One approach would be to irradiate the OVA-CTLs before transfer to prevent their growth in the recipient mice. Development of CTLs by cultures of the recipient spleen cells would then reflect stimulation of host T cells by antigenic carry over. In fact, no CTL activity was primed in such mice (data not shown). However, the interpretation of this experiment is compromised by the fact that irradiated E.G7-OVA also failed to prime CTLs (data not shown and reference 9). Thus, we cannot judge whether there was carry over of antigen by testing the priming of CTL in vivo using irradiated cells.

In contrast to naive animals, OVA-CTL lines can be stimulated to produce IFN-γ by irradiated E.G7-OVA, as illustrated in Fig. 1,B. This observation provides a sensitive method to gauge the amount of antigen remaining in the OVA-CTLs. 7 d after stimulation with E.G7-OVA, the OVA-CTLs were irradiated and added to fresh unirradiated OVA-CTLs. Activation was detected by secretion of IFN-γ. The results verify that OVA-CTLs do not produce IFN-γ without addition of exogenous antigen (Fig. 7), suggesting that little contaminating Ag remains in these cultures. No IFN-γ was produced by irradiated OVA-CTLs, nor did the irradiated OVA-CTLs stimulate IFN-γ production by unirradiated OVA-CTLs. Titration of irradiated E.G7-OVA established a standard curve for stimulation. From this experiment, we estimate that adoptive transfer of 5 × 106 OVA-CTLs would contain <5 × 103 cell equivalents of E.G7-OVA. In other words, <0.1% of the normal immunizing dose of E.G7-OVA might have been transferred as antigenic fragments to adoptive recipients. Hence, we conclude that the CTL activity recovered from B6 recipients of the OVA-CTLs is due to persistence of the CTLs as memory T cells rather than host CTLs primed by contaminating antigen.

The requirements for persistence of OVA-CTLs were examined next. To test whether memory was maintained by stimulation with cross-reactive peptides complexed to MHC class I, β2m knockout recipients were used. Because OVA-CTLs do not produce IL-2, the potential role of host IL-2 in long-term persistence memory could be addressed using IL-2 knockout mice. 5 wk after transfer, spleen cells from recipient mice were restimulated in vitro with E.G7-OVA, syngeneic filler cells, and IL-2. The frequencies of donor OVA-CTLs in the host spleen cells were then analyzed. As shown in Fig. 8, the frequency of donor OVA-CTLs in spleen cells from β2m−/− mice was in the same order of magnitude as in normal recipients (∼1/105 spleen cells), suggesting that stimulation by MHC class I–bearing cells was not required for the persistence of donor CTLs in vivo. In contrast, the frequency of donor OVA-CTLs in spleen cells of IL-2−/− mice (∼1/3 × 106 spleen cells) was much lower than in normal recipients, indicating that bystander IL-2 provided essential signals for the long-term maintenance of donor CTL in vivo.

Discussion

This study provides evidence that CTLs specific for exogenous protein Ags can persist for long periods both in vitro and in vivo without antigenic stimulation. Unlike some alloreactive, virus- or tumor-specific CD8+ CTLs that produce IL-2 (2528), the OVA-CTLs cultured from mice primed with OVA in CFA are dependent on exogenous IL-2 for their growth. The finding that T cells that have been grown in the presence of exogenous IL-2 develop into IL-2–dependent lines is not surprising. Similar observations been reported by many investigators beginning with the seminal studies of the CTLL line reported by Gillis et al. (9). Moreover, the inability of these CTLs to produce endogenous IL-2 upon activation provided us the opportunity to analyze the requirements for growth and activation separately.

Our in vitro experiments demonstrate that IL-2 was required for CTL growth, but not for the activation of effector functions. This is consistent with the observations from Baker et al. (29) and Cheever et al. (28) that CTLs can be maintained in vitro in an IL-2–dependent proliferative state virtually indefinitely. Proliferation that was induced by IL-2 could be enhanced by the addition of Ag (E.G7-OVA) because Ag upregulated the expression of IL-2Rs. CD8+ CTLs that were maintained for prolonged periods with IL-2 and filler cells but no Ag, retained the ability to be activated by specific antigen to produce lymphokines and to kill targets. After activation by Ag, lymphokines could be detected in the supernatant fluid within 1–2 h and reached a plateau within 4 h. Lymphokine production by the CTLs that had been maintained with IL-2 lagged only slightly behind the cultures that had been maintained with Ags. Lymphokine secretion by the activated CTLs was abrogated by inhibitors of both protein and RNA synthesis. By contrast cytolytic activity, which is largely dependent upon release of lytic factors stored in granules (30), was largely resistant to these inhibitors. These results show that Ag-specific, CD8+ T cells maintained for prolonged periods with IL-2 but not Ag, respond to Ag with rapid kinetics similar to that of CTLs that had been recently stimulated with Ag.

These studies also show that activation of effector functions by the OVA-specific CTLs required exposure to antigen (E.G7-OVA). Antigen was not only necessary, but sufficient to stimulate lymphokine secretion. Neither filler cells, IL-2, or filler cells plus IL-2 induced lymphokine secretion in the absence of Ag, nor did they significantly enhance lymphokine secretion that was induced by Ag. Ag induced messenger RNA synthesis and secretion of lymphokines within 1–2 h of incubation with CTLs. This rapid response is compatible with previous observations that primed T cells are able to respond to antigen much more quickly than naive T cells (for review see reference 1).

Maintenance of memory T cells in vivo has often been examined by adoptive transfer of T cells into irradiated syngeneic recipients. Irradiation of recipients is thought to provide room for transferred lymphocytes to take up residence in the appropriate niches. Indeed, mature peripheral T cells have the capacity to expand to a constant steady state level in irradiated adoptive recipients in the absence of antigen (31). Under physiological conditions, however, memory T cells are maintained in the presence of a vast excess of irrelevant T cells. Therefore, we transferred cultured CTLs into unirradiated, syngeneic recipients to study their capacity to serve as memory T cells. OVA-specific or B10 anti-BALB/c CTLs were injected intravenously into B6 mice where they were parked without stimulation by exogenous antigen. After 100 d, the spleen cells from the recipient mice were cultured for 7 d with E.G7-OVA and assayed for specific lytic activity. It is important to note that OVA-specific CTL activity cannot be stimulated in spleen cells from normal, unprimed mice even after several cycles of stimulation with E.G7-OVA plus IL-2 in vitro (5, 9). Likewise, no OVA-specific CTL activity was detected in cultures of spleen cells from mice that received B10 anti-BALB/c CTLs after stimulation with E.G7-OVA. Low, but significant, levels of cytolytic activity were detected in cultures of spleen cells from mice that received the OVA-specific CTLs. The low level of CTLs detected 1 wk after culture initiation was greatly expanded upon restimulation in vitro. Thus, the donor OVA-CTLs persisted, as memory T cells, in the adoptive recipients for several months in the absence of exogenous antigen.

That OVA-specific, memory CTLs persisted in vivo supports the results of Cheever et al. who demonstrated that long-term cultured T cells survived in adoptive recipients where they were able to kill adoptively transferred tumor cells (28). Our data also confirm the studies of Lau et al. who showed that CTLs specific for lymphocytic choriomeningitis virus persisted for indefinite periods in a secondary, uninfected host (4), and those of Bruno et al. who showed that H-Y–specific memory cells persist in the absence of male cells (32). Thus, CD8+ CTLs specific for a variety of antigens including viruses, tumors, and exogenous proteins persist in the absence of the cognate Ag in vivo and in vitro.

Despite the data showing that memory T cells persist in the absence of cognate antigen, the precise requirements for the maintenance of these T cells are still not clear. A number of possibilities could account for the long-term survival of memory T cells, the most straight forward being that they are long-lived cells. Indeed, studies by Tough and Sprent demonstrated that a sizable proportion of T cells with a memory phenotype remained in interphase for prolonged periods (33). However, these authors also showed that a large portion of the T cells with a memory phenotype were rapidly dividing. The nature of the stimulus that drives proliferation of the memory T cells, in the absence of Ag, has not been defined. It is possible that they are stimulated via their TCR by cross-reactive peptides associated with MHC class I, as discussed by Beverley (34). Alternatively, they could be stimulated by ligands, such as costimulatory molecules (for review see reference 1) or cytokines (35, 36), whose receptors are upregulated on memory T cells.

The mice used in our adoptive transfer experiments were not exposed to OVA, experimentally or environmentally. To address the possibility that memory was maintained by cross-reacting peptides presented by Kb, we compared the recovery of OVA-specific memory T cells from normal or β2m−/− adoptive recipients. The precursor frequencies of CTLs from these mice were identical. Similar results were reported by Hou et al. (3) for persistence of virus-specific CD8+ T cells in β2m−/− mice. However, in the latter experiments, memory T cells were transferred into irradiated recipients where the lack of homeostatic mechanisms (31) could have contributed to expansion and hence the persistence of the memory pool. We minimized this problem by adoptively transferring the OVA-specific CTLs into unirradiated hosts.

Studies of T cell memory in β2m−/− mice by Hou et al. (3) were previously criticized by Matzinger (37) on the grounds that β2m−/− mice express low levels of free MHC class I antigens on cell surfaces that are sufficient to prime MHC class I–specific, CD8+ CTLs (38). Theoretically, such peptide-loaded, MHC class I molecules could stimulate cross-reactive memory T cells. However, in the absence of β2m, free H-2Db molecules are transported to the cell surface but H-2Kb molecules are not (39). Moreover, Zugel et al. have shown that H-2Db–restricted peptides, but not H-2Kb–restricted peptides, stabilize MHC class I expression on the surface of β2m−/− macrophages and that the H-2Db peptide complexes were recognized by CD8+ CTLs (40). Since OVA-specific, CD8+ CTLs used in our experiments recognize OVA257-264 presented by H-2Kb (5), the expression of H-2Db molecules in these mice is irrelevant.

Another potential complication in the adoptive transfer of Sendai virus–specific CTLs into β2m−/− mice (3) is that unfractionated spleen cells were transferred. Since the spleen cells contain stem cells, it was reasoned that they could have repopulated the recipient β2m−/− mice with sufficient MHC class I–positive cells to stimulate cross-reactive memory (37). In our experiments, OVA-specific CTL lines that had been maintained for >2 yr in vitro by culturing with irradiated splenic filler cells were transferred into the β2m−/− mice. Thus, no contaminating stem cells were transferred in our experiments. In addition, the OVA-specific CTLs were transferred intravenously into unirradiated, β2m−/− recipient mice where the transferred cells would be diluted by a vast excess of viable host cells. Thus, we consider the possibility that the OVA-specific CTLs bearing MHC class I could have presented cross-reacting peptides to themselves to be highly unlikely. We conclude that maintenance of CD8+ memory T cells does not require stimulation of TCRs by cross-reactive, peptides complexed to MHC class I.

The possibility that some ligands interacting with other receptors that are upregulated on memory T cells may play a role in the maintenance of memory has also been addressed by our studies. The frequency of OVA-specific CTLs recovered from adoptively transferred IL-2−/− mice was significantly lower than that obtained from normal mice. We conclude that bystander IL-2 provides one mechanism of support for memory T cells.

The observation that IL-2 helps to sustain memory in OVA-CTLs suggests that the OVA-CTL memory cells express IL-2R. Expression of IL-2R generally reflects activation of T cells via TCR–CD3, which raises the question of how IL-2R were stimulated in the absence of complexes of peptide and MHC class I in the β2m−/− mice. Low levels of IL-2Rs have been reported to be expressed by some CD8+ memory T cells (for review see reference 34). IL-2R are expressed by OVA-CTLs maintained by weekly stimulation in vitro (Fig. 2). Thus, expression of IL-2R may have been maintained in the adoptive recipients. Unutmaz et al. have recently reported that resting T cells could be stimulated to proliferate by IL-2 in the absence of TCR occupancy if they were stimulated with a combination of TNF-α and IL-6, which upregulated the IL-2R (36). Hence, cytokines produced during bystander T cell activation could activate IL-2R expression in the OVA-CTL in the absence of β2m and TCR stimulation. In addition, Siefken et al. recently reported that certain antibodies to CD28 can induce expression of IL-2 and IL-2R by resting T cells in the absence of ligation of the TCR complex (41). Thus, stimulation through other cell surface receptors might have activated IL-2R gene expression in the β2m−/− adoptive recipients.

Although the frequency of memory cells recovered from IL-2−/− mice was lower than from normal mice, significant numbers were obtained. This raises the possibility that other cytokines that activate memory T cells, such as IL-15 (42), may also provide signals that support memory T cells. Alternatively, the memory CTLs recovered from IL-2−/− mice could represent the long-lived, nonproliferating population that remains in interphase for prolonged periods (33). Experiments to examine these possibilities and to determine whether IL-2 is required for maintenance of memory T cells specific for other antigens are in progress.

Acknowledgments

We thank Dr. Michael J. Bevan (University of Washington, Seattle, WA) for providing the E.G7-OVA cell line. We particularly thank Ms. Linda M. Kapp for excellent technical assistance.

This work was supported by grant CA 70372 from the National Cancer Institute, National Institutes of Health.

References

References
1
Ahmed
R
,
Gray
D
Immunological memory and protective immunity: understanding their relation
Science
1996
272
54
60
[PubMed]
2
Mullbacher
A
The long-term maintenance of cytotoxic T cell memory does not require persistence of antigen
J Exp Med
1994
179
317
321
[PubMed]
3
Hou
S
,
Hyland
L
,
Ryan
KW
,
Portner
A
,
Doherty
PC
Virus-specific CD8+T-cell memory determined by clonal burst size
Nature
1994
369
652
654
[PubMed]
4
Lau
LL
,
Jamieson
BD
,
Somasundaram
T
,
Ahmed
R
Cytotoxic T-cell memory without antigen
Nature
1994
369
648
652
[PubMed]
5
Ke
Y
,
Li
Y
,
Kapp
JA
Ovalbumin injected with complete Freund's adjuvant stimulates cytolytic responses
Eur J Immunol
1995
25
549
553
[PubMed]
6
Ke
Y
,
Hunter
RL
,
Kapp
JA
Induction of humoral and cytolytic responses by ovalbumin in TiterMax and a new synthetic copolymer adjuvant
Vaccine Res
1995
4
29
45
7
Koller
BH
,
Marrack
P
,
Kappler
JW
,
Smithies
O
Normal development of mice deficient in β2M, MHC class I proteins, and CD8+ T cells
Science
1990
248
1227
1230
[PubMed]
8
Schorle
H
,
Holtschke
T
,
Hunig
T
,
Schimpl
A
,
Horak
I
Development and function of T cells in mice rendered interleukin-2 deficient by gene targeting
Nature
1991
352
621
624
[PubMed]
9
Moore
MW
,
Carbone
FR
,
Bevan
MJ
Introduction of soluble protein in to the class I pathway of antigen processing and presentation
Cell
1988
54
777
785
[PubMed]
10
Dombrowski
KE
,
Ke
Y
,
Thompson
LF
,
Kapp
JA
Antigen recognition by cytotoxic T lymphocytes is dependent upon ectoATPase activity
J Immunol
1995
154
6227
6237
[PubMed]
11
Gillis
S
,
Ferm
MM
,
Ou
W
,
Smith
KA
T-cell growth factor: parameters of production and a quantitative microassay for activity
J Immunol
1978
120
2027
2032
[PubMed]
12
Davidson
WF
,
Parish
CR
A procedure for removing red cells and dead cells from lymphoid cell suspensions
J Immunol Methods
1975
7
291
300
[PubMed]
13
Springer
T
,
Galfre
G
,
Secher
DS
,
Milstein
C
Monoclonal xenogeneic antibodies to murine cell surface antigens: identification of novel leukocyte differentiation antigens
Eur J Immunol
1978
8
539
551
[PubMed]
14
Leo
O
,
Foo
M
,
Sachs
DH
,
Samelson
LE
,
Bluestone
JA
Identification of a monoclonal antibody specific for murine T3 polypeptide
Proc Natl Acad Sci USA
1987
84
1374
1378
[PubMed]
15
Ortega
G
,
Robb
R
,
Shevach
E
,
Malek
T
The murine IL-2 receptor
J Immunol
1984
133
1970
1975
[PubMed]
16
Meager
A
,
Leung
H
,
Woolley
J
Assays for tumor necrosis factor and related cytokines
J Immunol Methods
1989
116
1
17
[PubMed]
17
Mosmann
TR
,
Fong
TAT
Specific assays for cytokine production by T cells
J Immunol Methods
1989
116
151
158
[PubMed]
18
Maraskovsky
E
,
Chen
W-F
,
Shortman
K
Interleukin-2 and interferon-γ are two necessary lymphokines in the development of cytolytic T cells
J Immunol
1989
143
1210
1214
[PubMed]
19
Rogers
LA
,
Zlotnik
A
,
Lee
F
,
Shortman
K
Lymphokine requirements for the development of specific cytotoxic T cells from single precursors
Eur J Immunol
1991
21
1069
1072
[PubMed]
20
Rogers
LA
,
Zlotnik
A
,
Lee
F
,
Shortman
K
The maintenance of lytic specificity during the development of clones of cytotoxic T lymphocytes from single precursor cells
J Immunol Methods
1991
143
241
250
[PubMed]
21
Meuer
SC
,
Hussey
R
,
Cantrell
D
,
Hodgdon
JC
,
Schlossman
SF
,
Smith
K
,
Reinherz
E
Triggering of the T3–Ti antigen-receptor complex results in clonal T-cell proliferation through an interleukin-2 dependent autocrine pathway
Proc Natl Acad Sci USA
1984
81
1509
1513
[PubMed]
22
Cantrell
DA
,
Smith
KA
The interleukin-2 T-cell system: a new cell growth model
Science
1984
224
1312
1316
[PubMed]
23
Kelso
A
,
MacDonald
HR
,
Smith
KA
,
Cerottini
JC
,
Brunner
KT
Interleukin 2 enhancement of lymphokine secretion by T lymphocytes: analysis of established clones and primary limiting dilution microcultures
J Immunol
1984
132
2932
2938
[PubMed]
24
Dunn
DE
,
Herold
KC
,
Otten
GR
,
Lancki
DW
,
Gajewski
T
,
Vogel
SN
,
Fitch
FW
Interleukin 2 and concanavalin A stimulate interferon-γ production in a murine cytolytic T cell clone by different pathways
J Immunol
1987
139
3942
3948
[PubMed]
25
Mizouchi
T
,
Hugin
AW
,
Morse
HC
III
,
Singer
A
,
Buller
RML
Role of lymphokine-secreting CD8+T cells in cytotoxic T lymphocyte responses against vaccinia virus
J Immunol
1989
142
270
273
[PubMed]
26
Kasaian
MT
,
Biron
CA
The activation of IL-2 transcription in L3T4+ and Lyt-2+ lymphocytes during virus infection in vivo.
J Immunol
1989
142
1287
1292
[PubMed]
27
Fong
TAT
,
Mosmann
TR
Alloreactive murine CD8+T cell clones secrete the Th1 pattern of cytokines
J Immunol
1990
144
1744
1752
[PubMed]
28
Cheever
MT
,
Thompson
DB
,
Klarnet
JP
,
Greenberg
PD
Antigen-driven long term–cultured T cells proliferate in vivo, distribute widely, mediate specific tumor therapy, and persist long-term as functional memory T cells
J Exp Med
1986
163
1100
1112
[PubMed]
29
Baker
PE
,
Gillis
S
,
Smith
KA
Monoclonal cytolytic T-cell lines
J Exp Med
1979
149
273
278
[PubMed]
30
Isaaz
S
,
Baetz
K
,
Olsen
K
,
Podack
E
,
Griffiths
GM
Serial killing by cytotoxic T lymphocytes: T cell receptor triggers degranulation, re-filling of the lytic granules and secretion of lytic proteins via a non-granule pathway
Eur J Immunol
1995
25
1071
1079
[PubMed]
31
Rocha
B
,
Dautigny
N
,
Pereira
P
Peripheral T lymphocytes: expansion potential and homeostatic regulation of pool sizes and CD4/CD8 ratios in vivo
Eur J Immunol
1989
19
905
911
[PubMed]
32
Bruno
L
,
Kirberg
J
,
von Boehmer
H
On the cellular basis of immunological T cell memory
Immunity
1995
2
37
43
[PubMed]
33
Tough
DF
,
Sprent
J
Turnover of naive- and memory-phenotype T cells
J Exp Med
1994
179
1127
1135
[PubMed]
34
Beverley
PCL
Is T-cell memory maintained by crossreactive stimulation?
Immunol Today
1990
11
203
205
[PubMed]
35
Tough
DF
,
Borrow
P
,
Sprent
J
Induction of bystander T cell proliferation by viruses and type I interferon in vivo
Science
1996
272
1947
1950
[PubMed]
36
Unutmaz
D
,
Pileri
P
,
Abrignani
S
Antigen-independent activation of naive and memory resting T cells by a cytokine combination
J Exp Med
1994
180
1159
1164
[PubMed]
37
Matzinger
P
Memories are made of this?
Nature
1994
369
605
606
[PubMed]
38
Bix
M
,
Raulet
D
Functionally conformed free class I heavy chains exist on the surface of β2 microglobulin negative cells
J Exp Med
1992
176
829
834
[PubMed]
39
Williams
DB
,
Barber
BH
,
Flavell
RA
,
Allen
PM
Role of β2-microglobulin in the intracellular transport and surface expression of murine class I histocompatibility molecules
J Immunol
1989
142
2796
2806
[PubMed]
40
Zugel
U
,
Schoel
B
,
Kaufmann
SHE
β2-microglobulin independent presentation of exogenously added foreign peptide and endogenous self-epitope by MHC class I α–chain to a cross-reactive CD8+ CTL clone
J Immunol
1994
153
4070
4080
[PubMed]
41
Siefken
R
,
Kurrle
R
,
Schwinzer
R
CD28-mediated activation of resting human T cells without costimulation of the CD3/TCR complex
Cell Immunol
1997
176
59
65
[PubMed]
42
Kanegane
H
,
Tosato
G
Activation of naive and memory T cells by interleukin-15
Blood
1996
88
230
235
[PubMed]
1

Abbreviation used in this paper: β2m, β2-microglobulin.

Author notes

Address correspondence to Judith A. Kapp, Jules and Doris Stein RPB Professor of Ophthalmology, Emory University School of Medicine, Bldg. B, Rm. 2623, 1365 Clifton Rd., NE, Atlanta, GA 30322. Phone: 404-778-4754; Fax: 404-778-2109; E-mail: jkapp@emory.edu