Differential Requirements for Survival and Proliferation of CD8 Naïve or Memory T Cells

See allHide authors and affiliations

Science  27 Jun 1997:
Vol. 276, Issue 5321, pp. 2057-2062
DOI: 10.1126/science.276.5321.2057


The requisite molecular interactions for CD8 T cell memory were determined by comparison of monoclonal naı̈ve and memory CD8+ T cells bearing the T cell receptor (TCR) for the HY antigen. Naı̈ve T cells required only the right major histocompatibility complex (MHC) class I–restricting molecule to survive; to expand, they also needed antigen. In contrast, for survival, memory cells did not require the restricting MHC allele, but needed only a nonspecific class I; for expansion the correct class I, but not antigen, was required. Thus, maintenance of CD8 T cell memory still required TCR–MHC class I interactions, but memory T cells may have a lower functional activation threshold that facilitates secondary responses.

The molecular basis of T cell memory remains elusive (1, 2). It is not known if memory responses depend exclusively on an increased frequency of antigen-specific T cells (3) or if “memory T cells” with novel biological capacities are generated (4). Memory responses have been reported to depend on continuous antigenic stimulation (5), but others have observed the persistence of increased frequencies of antigen-specific CD8+ T cells in the apparent absence of antigen (6-8).

We have investigated the conditions necessary in vivo for the survival and expansion of naı̈ve and memory antigen-specific CD8+ T cells. Because of the degeneracy and redundancy of T cell receptor (TCR) usage in most immune responses, individual clones of antigen-specific T cells “in vivo” cannot be easily examined. T cells may also coexpress different TCRs, and their behavior may be conditioned by nonspecific antigen effects (1, 2). Thus, to characterize the functional properties and the requirements for persistence of memory T cells, we used monoclonal T cell populations.

Transgenic (Tg) mice bearing a Tg αβ TCR specific for the HY male antigen restricted to major histocompatibility complex (MHC) class I H-2Db and deficient in the recombinase gene RAG2 (Tg RAG2) (9) were used to obtain monoclonal populations of CD8+ T cells. In female TgRAG2 mice, all T cells positively selected in the thymus are CD8+ Tg TCRαβ+ (Fig.1A). These cells represent a pure population of naı̈ve T cells, because cross-reactivity with environmental antigens cannot be detected: All these cells are CD44 and do not divide (10, 11). Studying these cells ex vivo, we could not detect lymphokine mRNAs, but these could be induced after in vitro stimulation with monoclonal antibodies (mAbs) to CD3 (anti-CD3). Virgin T cells constitutively expressed little perforin and FasL mRNAs, which were up-regulated after anti-CD3 stimulation (Fig. 1B) (12).

Figure 1

Interactions required for the survival and expansion of naı̈ve CD8+ T cells. (A) Female TgRAG2 mice LN cells, labeled with T3.70 (anti-Vα Tg) (23) and anti-CD8 mAbs. (B) Lymphokines, perforin, and Fas ligand mRNA production by naı̈ve CD8+ T cell–sorted populations. Southern blots of PCR products (cDNA was from 103 cells) hybridized with 35Plabeled–specific probes are shown. Negative samples were also tested with 10 times the concentration of cDNA, and 35 cycles of amplification. This also failed to reveal mRNA synthesis. (C and D) Irradiated (600 R) host mice were injected with 106 naı̈ve Tg cells. (C) Histograms showing BrdU incorporation of T3.70+CD8+ T cell–sorted populations, recovered at day 7 after transfer from different host mice injected for 3 days with BrdU. Background BrdU labeling (in mice not injected with BrdU) averaged 2%. (D) Absolute numbers of Tg cells recovered at days 1, 2, 7, and 13. The horizontal dashed line shows the number of cells that home to the lymphoid organs studied 1 day after cell transfer.

To study the TCR interactions required for the survival or division of naı̈ve CD8 T cells, we compared their fate after transfer into irradiated hosts (13) that differed in MHC class I and HY antigen expression. These hosts were C57BL/6 CD8-deficient (14) male (HY+H-2b+) and female (HYH-2b+) mice; female H-2Db-deficient mice (15) that lack the MHC class I restriction element of this Tg TCR (16) but express other MHC class I molecules including H-2Kb(HYH-2Db− class I+); and female class I mice deficient in both H-2Db and β2-microglobulin (HYDb−β2M). These mice were used rather than β2M-deficient mice, because the latter mice express H-2Db(17)—enough to induce deletion of male-specific Tg T cells in male mice (18). To correlate cell survival with interactions with MHC class I–restricting element, we used female mice expressing H-2Db but lacking H-2Kb(HYDb+Kb−). These host mice were irradiated (13) and injected 2 days later with 106 naı̈ve cells. Recovery of Tg cells was evaluated at days 1, 2, 7, and 13 after transfer (19). We studied Tg T cell division 1 week after transfer by monitoring bromodeoxyuridine (BrdU) incorporation.

One day after cell transfer, the fraction of donor cells homing to the pool of lymphoid organs studied was the same in all groups of host mice (about half of the donor cell population) (20). Naı̈ve T cells could survive in a resting state in female CD8-deficient mice (Fig. 1, C and D); they did not incorporate BrdU (21), and the number recovered was constant from day 1 up to 2 weeks after injection. Expansion of naı̈ve cells required stimulation with male antigen (22) because they divided only after transfer into male CD8-deficient hosts. Survival of naı̈ve T cells required the right MHC-restricting element. In mice lacking H-2Db or expressing no class I (H-2Db−β2M), naı̈ve cells did not survive, but decayed to an average of 3% of the injected cohort at 1 week, 1% at 13 days (Fig. 1D and Table1), and were undetectable at 2 weeks. This decay correlated with the absence of interactions with the MHC restriction element, because naı̈ve H-2Db–restricted Tg cells persisted after transfer into H-2Kb–deficient mice expressing H-2Db (below). Thus, as described during thymus positive selection, a minimal state of cell activation may allow survival in the absence of cell division (23).

Table 1

Recovery of Tg cells, 1 week after transfer into various irradiated hosts, that differed in MHC class I or antigen presentation. Mice were injected with 1 × 106 Tg cells. Results represent the absolute number of Tg cells (×10−5) recovered in host mice and are the mean of two experiments (male mice) or the mean ± SE of three to five independent experiments in other mice (two mice per experiment). Mice injected with naı̈ve and memory cells were studied simultaneously. KO, knockout.

View this table:

We next studied the TCR interactions required to maintain CD8+ T cell memory. To obtain memory cells, we stimulated female naı̈ve Tg T cells with relatively low doses of male antigen in vivo in male→ female bone marrow (BM) chimeras (24). We produced male→female B mice by injecting a mixture of 90% female and 10% male BM cells from CD3ɛ-deficient mice into RAG2-deficient female mice (25). The hosts did not have endogenous T cells, and 10% of BM-derived cells were of male origin (Fig. 2A, left).

Figure 2

Tg cells eliminated male antigen-expressing cells in vivo. (A) The spleen of male→female chimeras, labeled with anti-B220 and anti-Ly5.1 mAbs. The fraction of B220+Ly5.1+ indicates the percentage of male B cells present in host mice. Labeling in the absence of T cells (left), 1 month after injection of 5 × 105 LN naı̈ve Tg T cells (middle), and 5 × 105 memory Tg T cells, obtained from the LN of male→female chimeras injected 6 months previously with naı̈ve monoclonal T cells. (B) Memory and naı̈ve cells in the same antigenic environment. Male→female chimeras were injected with Thy1.2+ naı̈ve Tg cells. Five to 6 months later a new set of naı̈ve Thy1.1+ Tg cells was parked during 2 weeks in the same mice. Hosts were then injected with BrdU for 3 days, and cell suspensions were labeled with T3.70, anti-CD44, anti-CD8, and anti-Thy1.1 mAbs. Four-color analysis and cell sorting were done in a FACS Vintage (Becton Dickinson). Contour graphs show CD44 and Thy1.1 expression of CD8+T3.70+–gated T cells. Histograms show the percentage of BrdU+ cells in naı̈ve, newly injected Thy1.1+ (left), and resident memory-sorted Thy1.1 Tg cells (right). Similar results were obtained in six mice.

Naı̈ve T cells transferred into these chimeras expanded. In chimeras injected with 0.5 × 106 naı̈ve T cells, 15 × 106 to 20 × 106 T cells could be recovered 1 month later and for at least 7 months. These cells were able to eliminate the antigen in vivo, as shown by the disappearance of male BM-derived cells (Fig. 2A, middle). These antigen-experienced Tg T cells proliferated in response to the male antigen “in vitro” (8) and maintained the capacity to mediate effector functions in vivo, because they eliminated male BM-derived cells when transferred into new set of male→female BM chimeras (Fig. 2A, right). As described in other systems (6,7), memory cells in these chimeras appeared to survive in the absence of antigen. Because no male cells were detected in the spleen (Fig. 2A), lymph nodes (LNs), or BM (8), when a new set of naı̈ve cells were parked for 2 weeks in these mice, the cells retained the CD44 naı̈ve phenotype and did not divide (Fig. 2B). Six months after the apparent elimination of antigen, resident memory cells were still cycling (26) because 20 to 30% incorporated BrdU after a 3-day discontinuous pulse (Fig. 2B) (19). Therefore, naı̈ve and antigen-experienced cells behaved differently in the same host antigenic environment, suggesting that they may differ in their requirements for survival and stimulation. As evaluated by CD44, CD25, and CD69 expression (Fig.3A), antigen-experienced Tg cells have the memory T cell phenotype (26). They constitutively expressed mRNA encoding interleukin-2 (IL-2) and interferon-γ (IFN-γ), FasL, and perforin, the latter two to a larger extent than naı̈ve cells. This pattern of mRNA expression was stable and identical to that detected in CD44+ non-Tg CD8+T cells (Fig. 3B).

Figure 3

Interactions required for the survival and expansion of memory CD8+ T cells. (A) Memory LN Tg population, labeled with T3.70 (anti-Vα Tg) and anti-CD8 mAbs. Expression of activation markers in Tg naı̈ve (solid lines) and memory Tg T cells (dotted lines). (B) Lymphokines, perforin, and Fas ligand mRNA production by sorted memory Tg CD8+ T cells or CD8+CD44+ and CD8+CD44 T cells from C57BL/6 mice. Southern blots were done as described in Fig. 1B. (C andD) Irradiated (600 R) host mice were injected with 106 memory Tg T cells. (C) Histograms showing BrdU incorporation of T3.70+ CD8+ T cell–sorted populations 7 days after transfer (see Fig. 1). (D) Absolute numbers of Tg cells recovered in different hosts at days 1, 7, and 11 after injection. The horizontal dashed line shows the number of cells that home to the lymphoid organs studied 1 day after transfer.

To identify the TCR interactions required for the survival or division of memory CD8 T cells, we injected purified (>97%) memory populations [depleted of B cells and other class II–positive antigen-presenting cells (APCs)] into irradiated hosts (13) that differed in MHC class I or antigen presentation (16). Transgenic T cell recovery was determined at days 1, 2, 7, and 11 after transfer, whereas T cell division was evaluated by BrdU incorporation at day 7 (Fig. 3, C and D, and Table 1).

At 24 hours after transfer, homing of memory Tg cells was the same in all groups of host mice and similar to that of naı̈ve Tg cells (Fig. 1). When memory Tg cells were stimulated after transfer into male CD8 mice, their rate of division (>90% BrdU+ cells) was higher than that of naı̈ve cells. In contrast to naı̈ve cells, memory cells transferred into female CD8 hosts also divided extensively (70% BrdU+) (27) and survived and divided in mice lacking the H-2Db–restricting element (42% BrdU+). In mice lacking class I (H2-Db−β2M mice), about 30% of memory cells still incorporated BrdU, indicating a response to autocrine or environmental growth factors, even in the absence of T cell stimulation (26). However, this response was not sufficient to maintain memory T cells, which disappeared progressively (Fig. 3, C and D, and Table 1). Two weeks after T cell transfer, donor cells in class I–deficient (H-2Db−β2M) host mice were barely detectable.

The disappearance of Tg T cells after transfer into MHC-deficient irradiated hosts was not due to natural killer cell activity or T cell–mediated graft rejection by the irradiated hosts (Fig.4). CD4+ cells from H-2b mice (that could interact with host MHC class II) survived and expanded after transfer into all types of MHC class I–deficient hosts (Fig. 4A and Table 2). When CD8+ T cells from H-2b normal mice (which presumably contain mixtures of H-2Kb– and H-2Db–restricted naı̈ve and memory cells) were transferred, they survived and expanded in both H-2Db− and H-2Kb− hosts and decayed only after transfer into class I (Db−β2M) hosts (Fig. 4A and Table 2) (28). These experiments demonstrate that the survival of transferred CD8+ T cells correlated with their requirement for TCR–MHC class I interactions, because if cell decay were due to cell rejection, all donor cells (that express both H-2Db and H-2Kb) should be recognized and eliminated in all types of MHC-deficient host mice. These results also suggested that normal CD8+ T cells, like Tg lymphocytes, also require TCR–MHC class I interactions to survive in the periphery.

Figure 4

Survival of CD8+ T cells after transfer requires TCR-MHC interactions. (A) Irradiated (600 R) Ly5.2 host mice were injected simultaneously with 1 × 106 CD8+ and 2 × 106CD4+ LN T cells from Ly5.1 B6 donors. Histograms represent the percentage of CD8+ T cells among donor T cell populations recovered in the LN of host mice 1 week after T cell transfer. (B) Recovery of naı̈ve Tg cells in Kb− and Db− mice. Both types of recipient mice were irradiated, injected with the same suspension of 1 × 106 naı̈ve Tg cells, and studied 1 week later. Contour plots show the Tg populations recovered in the LNs of one Kb− and one Db− mouse. (C) (Left) Absolute number of naı̈ve Tg cells recovered in individual mice. Kb− mice (open symbols) and Db− mice (closed symbols). In Kb− mice (a and b), Tg cell recovery was lower than in class I+ mice. These mice down-regulated Db, as shown in the histograms displaying spleen cells from these mice (solid lines) and a normal mouse (dotted lines) labeled with anti-Db mAb. All other Kb− mice used in these studies expressed normal levels of Db. After intravenous transfer, only a fraction of injected cells migrates to the spleen and LN (27). The fraction of naı̈ve T cells recovered in female H-2Kb–deficient mice corresponds to the fraction of donor cells homing to the pool of lymphoid organs studied (27) (Fig. 1).

Table 2

Recovery of B6 Ly5.1 T cells, 1 week after transfer into various irradiated hosts. Host mice were injected simultaneously with 1 × 106 CD8+ and 2 × 106 CD4+ LN T cells from Ly5.1+ B6 donors. Results represent the absolute numbers (×10−5) of CD8+ and CD4+ donor T cells recovered in the different hosts in a single experiment. Similar results were obtained in three other experiments (37).

View this table:

Because in the above experiments all transferred B6 populations may have undergone expansion whereas naı̈ve cells remained resting, the results could be biased if host mice would preferentially eliminate resting donor cells. To exclude this possibility, we compared the survival of naı̈ve Tg T cells after transfer into female Db− and Kb− mice, in which naı̈ve cells do not divide (Fig. 4, B and C). Kb− hosts were also selected according to the levels of Db expression (29). In Kb− hosts expressing normal amounts of Db, the recovery of naı̈ve donor cells was the same as in class I–bearing recipients. In Kb− mice with a reduced expression of Db, cell recovery was lower, but still higher than that observed in Db− mice (Fig. 4C). Therefore, the frequency of surviving naı̈ve T cells recovered in the female recipient mice was related to the level of class I MHC expression, as described during thymus positive selection (30).

The disappearance of memory CD8+ cells in a class I environment disagrees with previous findings describing the survival of memory T cells in β2M hosts (6). These differences may be due to the expression of MHC class I in β2M mice (17), which interferes with T cell repertoire selection (18), or to the high number of T cells injected, or to both factors: in these instances it is likely that donor cells expressing class I interact among themselves in the host environment.

The fate of memory cells and their level of activation varied markedly in different hosts (Ag+ right MHC class I > Ag right MHC > Ag wrong MHC > no MHC) (Fig. 3, C and D). Because all mice received the same cohort of cells simultaneously, these results indicate that the fate of donor cells is conditioned by their interaction with the host environment, and not by interactions with peptides or MHC class I molecules present in the donor cell inoculum. The different kinetics of cell growth also suggests that the affinity of the TCR–MHC peptide interactions may determine a different expansion rate of memory T cells.

The properties of primed CD8+ cells we describe may explain why injection of a virus can sometimes elicit a secondary response to a previously injected but unrelated virus (2), mimicking the “original antigenic sin” phenomenon (31). They may also explain the kinetics underlying the editing of CD8+ T cell memory. Early after antigen stimulation, the increase in the local production of environment growth factors, as well as the promiscuous stimulation of memory cells, may be responsible for the bystander activation of memory T cells of unrelated specificities (26,32). The different expansion rate of memory cells, conditioned by the type of TCR-MHC peptide interaction (Fig. 3, C and D), will result in the progressive selection of antigen-specific T cells that will compete out (33) T cells bearing unrelated specificities later in the immune response (34).

Because our Tg cells express a single TCR, we can exclude nonspecific antigen effects caused by endogenous receptor rearrangements. Thus, in vivo antigen stimulation induced permanent changes in the physiological status of antigen-specific T cells, generating memory cells with unique characteristics. These results are similar to those obtained with cells from normal mice (11), so the requirements for survival and expansion may be generalized.

Thus, survival and expansion requirements of CD8+ T cells in the peripheral pools differed for naı̈ve and memory cells and depended on TCR-MHC peptide interactions. Survival of naı̈ve T cells required interactions with H-2Db, similar to thymic positive selection (15). Activation and expansion of naı̈ve cells required the presence of the male antigen. In contrast, memory cells required a nonspecific class I interaction for survival and expanded in the presence of the right MHC class I, but in the absence of antigen. Because survival of memory T cells still required interactions with MHC class I, TCR engagement is involved in the survival of memory cells. We do not know what type of peptide recognition is implicated, but cross-reactive peptides may be involved. We cannot exclude, however, promiscuous recognition as in thymic positive selection. Memory T cells may have a lower functional threshold that allows their expansion in the absence of the nominal male peptide, or their survival in the absence of the selecting H-2Db–restricting element. These properties may facilitate both maintenance of memory and secondary immune responses.


Stay Connected to Science

Navigate This Article