Anti-human immunodeficiency virus-gag CD8⁺ memory T cells generated in vitro from Listeria-immunized mice

Abstract

The goal of vaccination is the generation of immune memory, an immune state that permits rapid and intense recall responses to a pathogen. Considerable effort is being made to understand the nature of memory T cells. We report here that by extending the length of in vitro culture following a single restimulation with specific peptide, preparations of highly enriched, highly active antigen-specific CD8⁺ memory T cells could be obtained. These cultures were begun with splenocytes from mice primed by infection either with an attenuated strain of Listeria monocytogenes or vaccinia virus, both expressing the human immunodeficiency virus-1-gag gene. In the cultures, antigen-specific cytotoxic T lymphocyte (CTL) activity reached a maximum at about 9 days and thereafter fell to negligible values. Concomitant with the fall of CTL activity, however, we observed enrichment for a subset of CD11a^high antigen-specific gag-tetramer^pos CD8⁺ T cells. The cells showed little or no 4-hr CTL activity, but had high delayed (18-hr) CTL activity, and very high cytolytic activity after restimulation. They rapidly expressed interferon-γ production. Their growth and survival after sorting was completely dependent on interleukin-2 or -15. As few as 5000 of the fluorescence-activated cell sorting-purified cells protected recipients against challenge 3 months after transfer. In response to the challenge, the cells repopulated lymphoid and non-lymphoid organs and showed a sizeable increase in number. The cells therefore demonstrate high protective activity for long periods of time. These cultured cells are thus a potential source of enriched natural memory T cells for reperfusion studies and in which the mechanisms that underlie the generation, differentiation and persistence of memory can be examined.

Introduction

New human immunodeficiency virus (HIV) infections worldwide continue to occur at alarming rates, and an HIV vaccine is the only long-term solution to the epidemic. The goal of a vaccine is to generate a pool of memory cells that retains information about a pathogen for extended time periods and can thus protect against future infections. Immune memory produces a vigorous response to a pathogen because of: (1) the long life and large pool size of memory cells that recognize the pathogen; (2) the rapid and robust response of members of this pool; and (3) the increased avidity of these cells for antigen. Growing evidence suggests that T-cell mediated immunity plays a key role in controlling HIV infection. For example, lack of disease in long-term non-progressors is associated with high levels of HIV-specific CD8⁺ memory cytotoxic T lymphocytes (CTL);¹ exposed but uninfected individuals often have specific antiviral T-cell responses, but no antibody response;^2–4 the viraemia that occurs during primary infection resolves coincident with the development of virus-specific CTL, before the development of specific antibodies;^5,6 and CD8 depletion experiments in rhesus macaques revealed that the level of circulating CD8 T cells was directly correlated with simian immunodeficency virus viraemia and animal survival.^7,8 Thus an effective acquired immune deficiency syndrome (AIDS) vaccine should be able to induce anti-HIV cell mediated immunity. Listeria monocytogenes is an intracellular micro-organism that is a paradigm for the induction of cell-mediated immunity. We have shown that a highly attenuated vaccine strain of Listeria expressing the HIV gag antigen can safely induce long-term systemic and mucosal protective immunity against gag following mouse immunization^9,10 and can activate CTL from HIV-infected humans.¹¹

Because of the important role of memory T cells in vaccine design, considerable effort is being made to understand their nature, generation and maintenance.^12–27 However, specific memory cells usually represent a small fraction of the total population of T cells in an animal. We observed during the in vitro culture of primed lymphocytes in the presence of specific peptide that effector CTL activity reached a peak at about 9 days in culture and then fell to negligible levels thereafter. Nevertheless, the population of cells remaining in extended cultures showed a considerable enrichment for a subset of CD11a^high antigen-specific tetramer^pos CD8⁺ T cells, and we questioned whether these cells had functional activity. Our results demonstrate that they represent a preparation of CD8⁺ memory T cells with very high protective ability following transfer to naïve animals. Their enrichment may result from the fact that memory cells are long lived²⁸ and do not require antigen for survival,²⁹ while most effector T cells that reencounter antigen are lost through activation-induced cell death.³⁰

Materials and methods

Mice

Eight- to 12-week-old BALB/c (H-2^d) female mice (Charles River Laboratories, Raleigh, NC) were used for immunization and adoptive transfer experiments. C.B-17 (H-2^d) severe combined immunodeficient (SCID) mice (kindly provided by Dr J. Cebra, Department of Biology, University of Pennsylvania) were used for some transfer experiments.

Bacterial and viral immunizations

Lmdd-gag is an attenuated recombinant strain of L. monocytogenes expressing an HIV-1 gag gene. It was grown as described previously.³¹ Recombinant vaccinia viruses expressing HIV-1 gag (vVK1, or vac-gag) or HIV-1 nef (vTFNef, or vac-nef) were kindly provided by Dr B. Moss (National Institutes of Health, Bethesda, MD). Naïve BALB/c mice were immunized intraperitoneally (i.p.) with 10⁸ Lmdd-gag or 0·5–1 × 10⁷ plaque-forming units (PFU) of vac-gag and boosted once or twice as described previously.¹⁰

Long-term cultures (LTC)

Mice were immunized i.p. with either Lmdd-gag or vaccinia-gag virus, as described.¹⁰ Animals were boosted after 21 days and sometimes again 1 month later. Immune cells were usually isolated 7 days after the last inoculation. Splenocyte suspensions for in vitro stimulation cultures were obtained by pressing the tissue through a nylon mesh screen into RPMI-1640 supplemented with 10% heat-inactivated fetal calf serum, 5 × 10⁻⁵ M 2-mercaptoethanol, 25 mm HEPES, 2 mm l-glutamine and 50 µg/ml gentamicin. Antigen-presenting cells (APC) were splenocytes from naive mice. They were loaded with 5–10 µm of the HIV-1 gag K^d epitope, aa 197–205³² for 60 min, irradiated with a cobalt source at 3000 rad, washed and used at a ratio of three to four lymphocytes per APC. The cultures contained 6 × 10⁷ cells in 8 ml medium and were incubated at 37° in 10% CO₂ for 30–41 days with a single 1 : 1 change of medium at about day 9.

Cell preparation for analysis

Spleens, mesenteric lymph nodes (MLN), Peyer's patches (PP), and fat pads were excised and pressed through nylon mesh bags or teased to release single cells. Peritoneal exudate cells (PEC) were prepared by rinsing the peritoneal cavity with cold Hank's balance salt solution (HBSS). Lamina propria (LP) cells from the small intestine were prepared, with minor modifications, as described elsewhere.³³ Briefly, pieces of small intestine were washed sequentially in phosphate-buffered saline (PBS) and Ca²⁺-Mg²⁺-free PBS, followed by incubation in Ca²⁺-Mg²⁺-free PBS containing 1 mm ethylenediamine tetra-acetic acid (EDTA) and 1 mm dithiothreitol and then in the same buffer containing only 1 mm EDTA. This was followed by digestion in RPMI-1640 containing 10% fetal bovine serum, 1 mg of collagenase (Sigma, St Louis, MO) and 0·5 mg of soybean trypsin inhibitor (Sigma) per ml. Cells were filtered through a cotton wool column and centrifuged over a Percoll gradient (40–70% Percoll in HBBS; 650 × g, 20 min, 4°). LP lymphocytes localized in the interface between the 40 and 70% layers.

Antibodies and major histocompatibility complex (MHC) tetramers

Monoclonal antibodies (mAb) were from BD PharMingen (San Diego, CA). MHC class I HIV-1 gag-specific tetramers were prepared and provided by the MHC Tetramer Core Facility, National Institute of Allergy and Infectious Disease, Atlanta, GA, using the K^d-specific gag-epitope, aa 197–205·³²

Immunofluorescent cell analysis and sorting

Cells were labelled for 60 min on ice with fluoroscein isothiocyanate (FITC)- or phycoerythrin (PE)-labelled mAb against mouse surface molecules and APC-labelled gag tetramers. The fluorescence from 1–3 × 10⁵ cells was analysed using a FACSCalibur flow cytometer and CellQuest software (both from Becton Dickinson, Immunocytometry Systems, Mountain View, CA). LTC cells were sorted for CD11a^high tetramer^pos (double positive, DP) and CD11a^low tetramer^neg (double negative, DN) CD8⁺ T cells using a FACStar Plus cell sorter (Becton Dickinson).

Cytotoxicity assay

Gag-specific CTL activity against gag-peptide (aa 197–205)-labelled P815 target cells was determined in 4 or 18 hr ⁵¹Cr-release assays, as described previously.¹⁰ To compare the results of independent experiments and samples assayed at different effector:target ratios, lytic units^34,35 rather than percent lysis, is plotted in Fig. 1. Lytic units were calculated as the number of effector cells, in a total of 10⁷ lymphocytes, required to produce 15% lysis of 2 × 10⁴ target cells. In this low range of CTL activity the effector : target ratio is linear with CTL activity and is least affected by inhibitory constituents in the lymphocyte preparations.³⁶

Figure 1.

Lytic activity of immune splenocytes following restimulation in culture. Mice were infected with attenuated gag-recombinant L. monocytogenes (Lmdd-gag) and boosted. Twelve days later, spleen cells were isolated and cultured with gag-peptide-pulsed naïve splenocytes for 41 days. Throughout this period cytolytic activity of the cells was measured using standard 4-hr Cr⁵¹ -release assays with peptide-pulsed P815 target cells. The data is plotted in lytic units (see Material and Methods) to allow comparison of assays performed on different days. As an example of these relative activities, the 28·1 lytic units at day 9 corresponds to a corrected peptide-specific cytolytic activity of 53·4% lysis at an effector : target ratio of 11·1. As shown, at day 41 cytolytic activity was also measured in an 18-hr assay (41/18 hr), as well as after a 6-day restimulation with peptide-pulsed splenocytes in a 4-hr assay (41/+6 days/4 hr). The three assays shown at the termination of culture (41; 41/18 hr; 41/+6 day/4 hr) are representative of more than four independent experiments in which cultures had been initiated with cells from immune mice 7–12 days after boosting and continued for 30–41 days.

Intracellular interferon-γ (IFN-γ) analysis

For cytokine production, 1 × 10⁶ cells were stimulated for 5 hr with 1·3 µg/ml of gag peptide and 66 U/ml of human recombinant interleukin (IL)-2 (BD PharMingen) in the presence of 1·3 µl/ml of monensin (GolgiStop, BD PharMingen). After surface staining with anti-CD8a and gag tetramers, cells were permeabilized with Cytofix-Cytoperm (BD PharMingen) and stained with FITC-labelled anti-IFN-γ (clone XMG1.2, rat immunoglobulin G1; IgG1) or an isotype control mAb (clone R3-34) at 2 µg/ml. Culture without the peptide served as a negative control.

Adoptive transfer

Unsorted LTC cells or fluorescence-activated cell sorting (FACS)-purified CD11a^high tetramer^pos CD8⁺ LTC cells were adoptively transferred into naïve BALB/c mice i.p. At 3 months, recipients were challenged with 1 × 10⁷ PFU of vac-gag i.p. Six days later, the mice were assayed for protection and repopulation of lymphoid and non-lymphoid compartments with gag-specific tetramer^pos CD8⁺ T cells. In some experiments, the mice were lightly irradiated with 550 rad 24 hr before cell transfer. Cells were also transferred to SCID mice. Two days after these transfers, SCID recipients were stimulated i.p. with 5 × 10⁵ PFU of vac-gag. Three months later, some were again infected with 1 × 10⁶ PFU of vac-gag i.p and examined for repopulation of lymphoid and non-lymphoid compartments with gag-specific tetramer^pos CD8⁺ T cells.

Vaccinia virus protection assay

Determination of vaccinia virus titres in mouse ovaries following virus infection was performed as described previously.¹⁰

Cytokine-induced proliferation

Unsorted LTC cells (10⁵) and FACS-purified CD11a^high tetramer^pos (DP) and CD11a^low tetramer^neg (DN) CD8⁺ T cells (10⁴) were cultured in round-bottomed 96-well plates in the presence of 500 ng/ml of either recombinant human IL-2 (BD PharMingen) or IL-15 (R & D Systems) with or without 10⁵ irradiated naïve splenocytes for 48 hr. Cultures were pulsed with [³H]thymidine during the last 18 hr.

Results

Effector activity of cultured lymphocytes

In the following studies, we examined the development and persistence of effector functions of immune lymphocytes that had been restimulated in culture. Mice were immunized and boosted with recombinant L. monocytogenes expressing HIV-gag (Lmdd-gag).^9,10,31 Splenocytes isolated several days after the height of the boosted response were stimulated in culture with irradiated naïve syngeneic splenocytes pulsed with the H-2K^d-specific gag peptide, aa 197–205.^37,38 The cultures were then continued for 30–41 days with one change of medium but with no additional peptide stimulation. Reportedly, peptides in culture medium have a half-life of only 10–100 min.³⁹

During the culture period, the cells were periodically assayed for direct cytolytic activity on peptide-pulsed P815 target cells in a standard 4-hr chromium release assay. Because assays were performed at multiple different times, the assay results were normalized by conversion to lytic units (see Materials and methods). Initial cytolytic activity was low, but gradually rose to a peak at 9 days and then became negligible by 29 and 41 days Fig. 1. Thus, cells taken at 41 days had virtually no 4-hr cytolytic activity. Nevertheless, during an 18-hr assay, these same cells showed high peptide-specific cytolytic activity Fig. 1. The 18-hr activity depended on granule release because over 80% of the activity was inhibited by 3 mm egtazic acid and inhibition was blocked by addition of Ca²⁺ (data not shown). Additional evidence that the 41-day cells were inherently capable of high cytolytic effector function was shown following restimulation with peptide-pulsed cells for 6 additional days and standard 4-hr cytolytic assay Fig. 1. This resulted in lytic activity that was extremely high, threefold higher than seen at the peak of the in vitro stimulation culture. The highly active 41-day cells are thus unlike normal effector cells since they possess potential for high CTL activity but show minimal or no activity during a standard 4-hr assay.¹⁰ This is characteristic of CD8⁺ memory T cells.^12,21

The induction of IFN-γ by these cells was examined. Fig. 2 shows an assay for the accumulation of intracellular IFN-γ in LTC lymphocytes compared with the activity in immune splenocytes 7 days after a second boost, and with naïve splenocytes. Following 5 hr culture in the presence of gag peptide and IL-2, 14·7% of the 41-day-cultured lymphocytes expressed IFN-γ (Fig. 2a), while only 2·3% of the ex vivo immune CD8⁺ splenocytes did so (Fig. 2b). Among CD8⁺ T cells, these fractions were 52% and 11%, respectively. When the analysis was gated on gag-tetramer^pos CD8⁺ T cells, 92% of the cultured cells were IFN-γ^high (Fig. 2e). In these assays neither naïve T cells (Fig. 2d), nor immune T cells in the absence of gag peptide (Fig. 2c) expressed IFN-γ. The LTC cells thus rapidly recalled effector function, expressing IFN-γ immediately (5 hr) and CTL activity after a brief delay (18 hr).

Figure 2.

IFN-γ production by LTC cells. Three-colour FACS analysis of intracellular IFN-γ production after 5 hr stimulation with gag peptide plus IL-2 by (a) LTC cells; (b) splenocytes from 3× i.p. Lmdd-gag immunized mice; (c) the culture shown in (a) but in the absence of gag peptide; (d) naïve splenocytes. The numbers shown are the percentages of all lymphocytes that produced intracellular IFN-γ. (e) Intracellular IFN-γ production by the cells shown in (a) after gating on CD8⁺ gag-tetramer^pos T cells. (f) Intracellular IFN-γ production by the cells shown in (b) after gating on CD8⁺ gag-tetramer^pos T cells. In this case (but not in (e)) the gag-tetramer-labelled TCR was strongly down regulated. The data are typical and have been reproduced not less than twice.

Phenotype of long-term cultures (LTC)

The phenotypic properties of non-adherent cells from long-term cultures were compared with those of naïve and immune splenocytes. The ex vivo immune lymphocytes (and those from naïve mice, not shown) contained two heterogeneous populations of different sized cells by forward scattering. These two populations were more distinct in the LTC cells, with the smaller cells predominating, suggesting that most were in the G₀ phase of cell cycle (Fig. 3a). The phenotypic properties described below were found in both size populations. The LTC cells contained 34% CD8⁺ T cells, and among these 45% were CD11a^high gag-tetramer^pos (Fig. 3b, right), whereas lymphocytes isolated ex vivo from immune mice contained 13% CD8⁺ T cells and only 11% of these were CD11a^high gag-tetramer^pos (Fig. 3b, left).

Figure 3.

LTC results in enrichment of activated gag-specific tetramer^pos CD8⁺ T cells. Mice were infected with Lmdd-gag and boosted twice. Spleen cells were isolated and cultured 30–41 days to generate LTC cells. Uncultured immune splenocytes and LTC cells were analysed by three-colour flow cytometry for the presence of total CD8⁺ T cells and CD11a^high tetramer^pos cells (double positive, DP) within the CD8⁺ T-cell subset. (a) Representative FACS analysis. The percentages of DP cells in ex vivo immune splenocytes and LTC cells are shown. (b) The proportion of total lymphocytes as CD8⁺ T cells and CD11a^high tetramer^pos (DP) CD8⁺ T cells in 3× i.p. ex vivo preparations and in LTC cells. Data shown represent the mean ± the standard error of the mean (SEM). The analyses were repeated not less than 11 times.

The expression of several markers on these cells was examined. The LTC cells retained their activation status and were CD11a^high and CD62L^int (Fig. 4a, b). CD43 is a putative marker for effector lymphocytes.¹⁴ In vivo, the expression of this marker was increased at day 3 after a boost and then returned to the naïve level at day 7 (Fig. 4c and data not shown). In the restimulated culture, CD43 was highest at day 7 and then decreased gradually to day 41 (Fig. 4d and data not shown). In vivo, CD69, an immediate activation marker, was high on day 1 after a boost, was intermediate at day 3, and returned to its naïve level by day 7 (Fig. 4e and data not shown); in the culture it remained low during the entire culture period (Fig. 4f). Among other markers assayed on LTC cells (data not shown), CD44 and receptors for IL-15 (CD122) and IL-2 (CD25) were somewhat increased over naïve cells; and CD95L was low (it was up regulated in vivo only at day 3 after a boost).

Figure 4.

Surface marker expression of LTC cells. (a, b, d, f) Mice were infected with Lmdd-gag and boosted two times. Spleen cells were then cultured to generate LTC cells. The cultured, uncultured immune splenocytes, and naive cells were analysed by three-colour flow cytometry. (a) CD11a expression and (b) CD62L expression on LTC cells, bold line; on immune uncultured spleen cells, fine line; on naïve spleen cells, dashed line. (d) CD43 expression and (f) CD69 expression on LTC cells, bold line; or immune splenocytes cultured for only 7 days, fine line. (c) CD43 expression on ex vivo immune spleen cells at different days after a third boost; day 3, fine line, day 7, bold line. (e) CD69 expression on ex vivo immune spleen cells at different days after a third boost; day 1, fine line; day 7, bold line. Most histograms are gated on CD8⁺ gag-tetramer^pos T cells. For naïve spleen cells (a, b) and on day one after boost (e), histograms are gated on total CD8⁺ T cells. The data for LTC cells are typical and reproduced not less than five times.

LTC cells protect for long periods against virus challenge

Authentic memory cells can persist and confer protection against challenge after very long time periods in a host. However it has been reported that some preparations of Listeria-induced CD8⁺ T cells have strictly limited effector life.^40,41 LTC lymphocytes were therefore transferred to naïve mice and three months later the mice were challenged by infection with gag-recombinant vaccinia virus.^10,42,43 A control group of mice was challenged with vaccinia-nef virus. As shown in Fig. 5(a), the transfer of 1 × 10⁶ cells resulted in complete protection (a 6-log reduction of virus titre) in vaccinia-gag-challenged animals as long as 3 months after transfer. As expected, there was no effect of the transferred cells on virus titre in animals challenged with vaccinia-nef virus, demonstrating in vivo antigen specificity. Protection was associated with the appearance of an expanded subset of gag-specific tetramer-positive CD8⁺ T cells that was detected after the vac-gag challenge but not after the vac-nef challenge (Fig. 5b). Titration of the LTC cells showed that transfer of 2 × 10⁵ cells into intact animals reduced the challenge virus titre 4 logs 3 months after transfer; 5 × 10⁴ of these unsorted cultured cells failed to protect (Fig. 5a).

Figure 5.

Adoptive transfer of LTC memory cells to naïve BALB/c mice confers protection against subsequent challenge with vaccinia-gag (Vac-gag) virus. (a) Recipient mice received either unsorted LTC cells or FACS-purified CD11a^high tetramer^pos CD8⁺ LTC cells, and at 3 months were challenged with 1 × 10⁷ PFU vac-gag (or vac-nef where indicated), killed 6 days later and assayed for virus protection and outgrowth of CD11a^high tetramer^pos CD8⁺ T cells. Vac-gag challenged naïve mice served as positive controls for vaccinia infection. Some recipient mice were lightly irradiated (indicated by asterisks). (b) Recipient mice received 1 × 10⁶ of unsorted LTC cells i.p and then challenged with 1 × 10⁷ PFU of either vac-gag or vac-nef. Six days later spleen cells were examined for the percentage of CD11a^high tetramer^pos CD8⁺ T cells using three-colour FACS analysis. The plots are gated on CD8⁺ T cells. (c) Lightly irradiated mice received 5 × 10³ of FACS-purified CD11a^high tetramer^pos CD8⁺ T cells i.p and then challenged with 1 × 10⁷ PFU vac-gag. Six days later cells from lymphoid and non-lymphoid tissues were isolated and examined for the percentage of CD11a^high tetramer^pos CD8⁺ T cells using three-colour FACS analysis. Control mice received no transfer cells but were challenged with vac-gag. The plots are gated on CD8⁺ T cells. All data shown are means ± SEM (at least five mice per group were examined).

To determine whether the CD11a^high gag-tetramer^pos cells in the cultures were responsible for the protection, cultured cells were sorted (see Fig. 3a) and 5 × 10³ double positive cells were used for transfer. Too few negative cells could be isolated to test in this way. Because a low number of cells were transferred, the recipient mice were lightly irradiated to promote their homeostatic proliferation. The recipients were challenged 3 months later. These cells reduced virus titres by 3·5 logs (Fig. 5a). As indicated above, lightly irradiated mice that had received 10⁶ unsorted cells were fully protected against vac-gag challenge but showed no protection against vac-nef challenge (Fig. 5a).

In the challenged animals that had received 5 × 10³ sorted cells, CD11a^high gag-tetramer^pos cells were found in a variety of lymphoid and non-lymphoid tissues. In the spleen and peritoneal exudates of these animals, the gag-specific T cells represented 43-fold and 23-fold amplifications, respectively, over the initial number of cells transferred (data not shown). Therefore, their amplification in the whole animal was considerable. Their abundance was far lower in the control mice after vaccinia infection alone, showing that they were derived predominantly from the transferred LTC cells (Fig. 5c). Too few cells were transferred to reliably detect them in the various tissues prior to challenge.

In another experiment, a group of SCID mice received 4 × 10⁵ unsorted LTC cells. Because of some indication that CD8⁺ T cells may fail to proliferate in SCID mice⁴⁴ these recipient animals were stimulated at 2 days with 1/10 dose of vaccinia-gag virus to promote the survival and expansion of the transferred cells. When the mice were challenged 3 months later with vaccinia-gag, all animals survived and CD11a^high gag-tetramer^pos cells represented the vast majority of cells found in spleen (71%), mesenteric lymph nodes (68%), peritoneal exudates (92%), lamina propria (76%) and abdominal fat pads (88%) (data not shown). Conversely, in the control group of mice that had not received transfer cells but were infected with vaccinia, 60% of the animals expired. The surviving mice showed less than 0·01–10% CD11a^high gag-tetramer^pos CD8⁺ T cells in the various tissues examined. Virgin SCID mice contained no gag-tetramer^pos T cells in any of the tissues. Therefore, in both normal and SCID mice, the large increase of gag-specific T cells in all tissues examined represented the massive amplification of the small number of cultured gag-specific T cells initially transferred.

IL-2 and IL-15 promote the proliferation of purified LTC cells

To determine whether gag-specific CD8⁺ T-cell survival and proliferation were controlled by cytokines, preparations of cultured lymphocytes derived either from spleens of Lmdd-gag- (Fig. 3a) or vaccinia-gag-immunized mice (Fig. 6a) were sorted into CD11a^high gag-tetramer^pos and CD11a^low gag-tetramer^neg CD8⁺ T subsets. These, and unsorted cultured cells, were incubated in the presence or absence of IL-2 or IL-15 for 48 hr, in the absence of gag antigen. As shown in Fig. 6(b), both cytokines induced [³H]TdR incorporation by the unsorted populations and by the CD11a^high gag-tetramer^pos double positive (DP), but not the CD11a^low gag-tetramer^neg double negative (DN) CD8⁺ T cells. Furthermore, either cytokine was found to be both necessary and sufficient for proliferation to occur; in their absence no thymidine uptake was seen. To determine whether the thymidine incorporation reflected true cell growth, cell number in cultures of the purified tetramer-positive and tetramer-negative cells was followed. As shown in Fig. 6(c), the double positive CD11a^high gag-tetramer^pos cells continued multiplying for over 30 days in the presence of either cytokine, but failed to survive in their absence. The double negative CD8⁺ T cells failed to survive even in the presence of cytokine. Thus the double positive cells from the cultures absolutely required either IL-2 or IL-15 for survival and proliferation. It is likely that their survival in culture also depended on a supply of these or other cytokines. IL-2 could be provided by the lymphocytes and IL-15 by fibroblasts and macrophages. The survival of the double negative cells in the cultures must have depended on other factors.

Figure 6.

IL-2 and IL-15 stimulate proliferation of unsorted LTC memory cells and FACS-purified CD11a^high tetramer^pos LTC memory cells. (a) FACS analysis of uncultured spleen cells from 3× i.p. vac-gag immunized mice and LTC cells derived from them. The plots shown are gated on CD8⁺ T cells and the percentages of CD11a^high tetramer^pos (DP, double positive) and CD11a^low tetramer^neg (DN, double negative) CD8⁺ T cells are indicated. (b) [³H]Thymidine proliferation assay. Unsorted LTC and FACS-purified DP and DN LTC cells were cultured with 500 ng/ml of either IL-2 or IL-15 for 48 hr and proliferation was assessed by [³H]thymidine incorporation. Irradiated APC (10⁵) were included in the cultures of FACS-purified cells. Exp.1 and 2, Lmdd-gag-derived LTC cells; Exp.3, vac-gag-derived LTC cells. Scale = 10⁻³ c.p.m. (c) 10⁴ of FACS-purified DP and DN LTC cells derived from vac-gag-immune mice were cultured in the presence of 500 ng/ml of either IL-2 or IL-15 and 10⁵ irradiated APC. Periodically the cultures were counted, split and additional APC and cytokines provided.

To determine whether these purified, antigen-free sorted cells continued to express effector function, their cytolytic activity was tested. Cells taken from either cytokine culture up to 30 days after sorting showed high direct cytolytic activity at 18 hr but not at 4 hr (data not shown), like the unsorted cells from which they were derived.

Discussion

The goal of vaccination is the generation of immune memory, a state of the immune system that produces a more rapid and intense response to an infectious agent than can occur after primary exposure. This results from a sizeable pool of memory cells that recognize the pathogen – at the time of initial infection such cells are rare. These cells respond more quickly than naïve cells and can possess receptors with higher affinity for the pathogen.²⁸ Stimulation of T cells with antigen delivered in an appropriate context commits the cells to enter cell cycle and proliferate, but ultimately many of the activated cells die.³⁰ Antigen-induced cell death is seen both in vivo and in vitro. Antigen-specific T cells that avoid death can progress to become memory cells. It has been shown that fully differentiated effector T cells can give rise to memory cells,^15,22,25 although only a particular subset of these may achieve that status.¹⁸ However, cells that express no obvious effector function following a pulse with antigen and growth in IL-15 or low IL-2 can also give rise to memory.¹⁹ Also, in the unique circumstance of lymphopenia, memory cells can be generated in the complete absence of antigen.^13,20,45 Apparently, only cells that avoid some terminal stage of differentiation survive to become memory. Such cells may be ones that have experienced a suitable level of receptor engagement, costimulation and/or cytokine exposure during commitment and activation.^{16,17,23,26,46,47}

We examined cells that had been restimulated in culture but survived subsequent antigen-induced cell death. A large fraction of the surviving cells exhibited properties of memory cells. One week after the start of such cultures, the population of T cells expressed a high level of direct 4-hr CTL activity. Over the next several weeks, however, this activity became undetectable (Fig. 1), while at the same time a subset of CD11a^high gag-tetramer^pos CD8⁺ T cells increased from less than 2% to 45% or more (Fig. 3). According to prior reports, peptide in culture medium has a half-life between 10 and 100 min³⁹ and therefore the initially added peptide would be degraded within the first days of culture. The CD11a^high gag-tetramer^pos CD8⁺ T cells, while expressing very low immediate CTL activity in a 4-hr assay showed high 18-hr CTL activity (Fig. 1). The expression of delayed cytolytic function is a characteristic feature of memory T cells and distinguishes them from effector cells.^12,21,27 Virtually all of the CD8⁺ gag-tetramer^pos T cells also expressed IFN-γ after 5-hr stimulation with gag peptide, while fewer cells from immune spleen did so (Figs 2e, f). Thus by allowing a large fraction of effector T cells to succumb to death as a result of antigen-induced activation, we effectively concentrated the subset of long-lived death-resistant memory T cells in these long-term cultures. In accord with our results, studies of CD8⁺ and CD4⁺ memory T-cell generation during in vitro stimulation of lymphocytes from transgenic animals have shown that acquisition of effector function can occur within one or a few cell divisions after stimulation, and that the appearance of a memory phenotype (characterized by delayed CTL activity or protection of recipients) required additional cell divisions^15,21,22.

During culture the Listeria-induced gag-immune cells replicated slowly, if at all, accounting for their more uniform smaller size than T cells taken ex vivo from hyperimmune spleen. A gradually decreasing size of CD4⁺ memory T cells was also noted during in vitro culture.¹⁵ When supplemented with IL-2 or IL-15, the cultured cells efficiently incorporated ³H-thymidine (Fig. 6b). Cell sorting demonstrated that this ability resided exclusively in the gag-specific activated subset of cells. After sorting, these cells multiplied for several weeks, but only if supplemented with either IL-2 or IL-15 (Fig. 6c). In contrast, the cells in the non-activated, gag-tetramer^neg fraction failed to survive even in the presence of cytokine. The effect of IL-15 is not surprising. The cultured memory cells express IL-15 receptor and IL-15 has been shown in many studies to be important for the basal proliferation of CD8⁺ memory T cells.^26,48 IL-2 promotes the survival of naive T cells and stimulates proliferation of activated lymphocytes. However, a role for IL-2 in the maintenance of CD8 memory is ambiguous. Both a requirement for⁴⁹ as well as inhibition by⁵⁰ this cytokine in vivo have been reported. The cultured cells apparently fail to produce endogenous IL-2, but require and can utilize exogenous cytokine through their expressed IL-2R as an alternative to IL-15. We infer that the viability of the memory T cells in the long term cultures for at least 40 days depended on the production of these cytokines and other factors by stromal cells as well as T cells. Stromal cells from haematopoietic and lymphopoietic tissues have been shown to support the long-term survival of various stem cells cultured with them.^51–54 Memory cells share properties of stem cells.⁵⁵

A defining characteristic of memory is its ability, even after long periods of quiescence, to provide protection to an animal upon reinfection. We found that modest numbers of the LTC cells (10⁶) conferred total protection (6 log virus reduction) against a vaccinia challenge at least 3 months after their transfer to a naïve recipient host. More strikingly, just 5000 of the FACS-purified cells were sufficient to produce 3·5 log of protection in this model (Fig. 5a). These results can be compared with an earlier report in which CTL efficacy against vaccinia challenge was examined immediately after the adoptive transfer to SCID mice of 10⁷ CTL especially selected in culture for high T-cell receptor (TCR) avidity.⁵⁶ Only cells selected for highest avidity were able to reduce virus titres by 100–1000-fold in ovaries of those challenged animals. These data suggest that the conditions used in our experiments to generate memory CTL may therefore produce highly active high avidity cells, since we observed greater virus titre reductions (3·5–6 log) with the transfer of fewer cells (5 × 10³−10⁶) as long as 30 days after transfer to recipient mice. The memory cells we describe may also differ from those examined by some other investigators^40,41 who found that protection by Listeria-induced CD8⁺ T cells did not survive longer than 48 hr in recipient animals. These differing results need to be explored, but could reflect the different culture techniques used, the nature of the challenge organism, and perhaps most likely, the effector/memory status of the transferred cells.

Naïve, effector and memory T cells possess different functional and migratory capabilities resulting from their different states of differentiation. The relationship between these states may be linear but continues to be examined. Studies in humans indicate that at least two subsets of memory T cells with different properties and functional capacities can be distinguished.^21,24 One subset, called central memory, home like naïve T cells to secondary lymphoid organs as a result in part of high expression of CD62L and CCR7 and was found to express low levels of effector activities. A second, tissue-homing subset, called effector memory, was CD62L^low and CCR7^low and capable of immediate effector function. Both subsets are found in spleen. A recent study in mice⁵⁷ confirmed the existence of these classes of memory cells but found that both exhibited similar functional activities – immediate IFN-γ induction and delayed (18-hr) but not immediate (4-hr) CTL activity. It was also shown that effector memory cells converted with time to central memory (and that this conversion was not accelerated by exposure to vaccinia challenge). Finally, central memory cells were more effective than effector memory cells for long-term protection, perhaps because of their greater potential for replication.⁵⁷

Our cultured cells appear to share properties of both subsets. They showed rapid IFN-γ induction and slightly delayed CTL activity. They were CD62L^int, expressing somewhat higher levels of l-selectin than effector cells, suggesting they may be like effector memory cells. This property would be expected to allow their migration only to non-lymphoid tissues and spleen. However, after virus challenge they were found both in non-lymphoid tissues, like lamina propria, peritoneal exudate and abdominal fat, as well as in mesenteric lymph nodes and spleen. While this cell distribution could reflect the effect of virus challenge and a redifferentiation of their native surface and migratory properties, conversion of CD62L^low to CD62L^high memory cells as a result of vaccinia infection was not seen in the previous study.⁵⁷ However, during the 90-day period of residence in the recipient, a natural conversion of effector memory to central memory could have occurred. Also, the low numbers of cultured cells that conferred high protection after transfer may indicate either the transfer of, or the conversion to, central memory.⁵⁷ Additional markers like CCR7 and CD27 must be examined to help further define the nature of the cultured cells.

These long-lived in vitro-generated cells may be a useful population in which to explore further aspects of memory cell behaviour and development. In addition, several laboratories are investigating the use of adoptive transfer of in vitro expanded autologous CTL for the treatment of viral infections, cancer and AIDS.^58–61 The methods used here to obtain highly active CTL may therefore be of clinical value.