Class II-Independent Generation of CD4 Memory T Cells from Effectors

See allHide authors and affiliations

Science  12 Nov 1999:
Vol. 286, Issue 5443, pp. 1381-1383
DOI: 10.1126/science.286.5443.1381


The factors required for the generation of memory CD4 T cells remain unclear, and whether there is a continuing requirement for antigen stimulation is critical to design of vaccine strategies. CD4 effectors generated in vitro from naı̈ve CD4 T cells of mice efficiently gave rise to small resting memory cells after transfer to class II–deficient hosts, indicating no requirement for further antigen or class II recognition.

Signals through the T cell receptor (TCR), provided by high doses of peptide antigen bound to class II major histocompatibility complex (MHC) molecules on antigen-presenting cells (APCs), are critical for the activation of naı̈ve CD4 T cells and for their transition to effector and memory cells (1). However, transfer of effectors, generated in vitro from naı̈ve CD4 T cells, to adoptive hosts results in the development of long-term CD4 memory even though no further antigen is introduced (2), contradicting some studies stressing the importance of antigen persistence (3), but agreeing with others arguing against such a role (4). Different strengths of TCR interaction with peptide-MHC are needed for T cell response at different stages of T cell differentiation and can induce different outcomes. In the thymus, low-avidity TCR interaction with self peptides bound to self MHC induces positive selection (5). Naı̈ve CD8 T cells in class I–deficient hosts (6) and naı̈ve CD4 T cells in class II–deficient mice (7, 8) have shortened life-spans, suggesting that interactions between TCR and MHC prolong naı̈ve T cell life-span. Naı̈ve CD4 T cells require high avidity/density TCR interactions for induction of cytokine synthesis, whereas effector and memory cells respond efficiently at lower avidity/density (9). Thus, effector and memory cells might be expected to overcome this MHC dependence. However, in two recent studies, activated CD8 T cells did not generate long-term memory after transfer to class I–deficient hosts (10). Using a model where effectors are generated in vitro and then transferred to adoptive hosts, which do or do not express class II, we found that neither antigen recognition nor interaction with class II is necessary for the generation of memory or for its persistence.

T helper cell 1 (TH1) or TH2 cytokine-polarized effectors were generated in vitro from naı̈ve CD4 T cells of AND TCR transgenic (Ig) mice (11) by stimulation with PCCF (a fragment of pigeon cytochrome) and mitomycin-treated I-Ek transfected fibroblast, DCEK-ICAM, or T cell–depleted APC from B10.BR mice (2, 9). The added APCs are no longer present in the cultures after 24 to 48 hours (12). In aged AND mice, Tg+ memory cells do not develop, suggesting a lack of environmental antigens capable of stimulating cross-reacting responses (13). The effectors generated are >99% CD4+, Tg+ cells and contain no detectable APCs or APCs capable of mediating their restimulation (14).

To evaluate the possibility that endogenous TCR chains could contribute to memory generation or persistence, we crossed the AND mice to RAG-2–deficient mice [RAG knockout (KO)] (15). Effector cells were generated from naı̈ve CD4 T cells, and aliquots were transferred to T cell–deficient hosts (16) created by adult thymectomy, lethal irradiation, and bone marrow reconstitution (ATXBM) (17). Both donor and recipient were on a B6 background (17), making allogeneic reactions unlikely. Because the hosts are devoid of T cells, there is ample opportunity for transferred cells to receive other, potentially important, non–TCR mediated signals. Equivalent numbers of Tg+ CD4 T cells were seen in hosts 3, 8, and 13 weeks after transfer. The recovered cells were small, CD44hi with a memory phenotype (17). Equal numbers of Tg+ recovered memory cells were restimulated ex vivo, and cytokine titers in the supernatants were determined (18). The cytokine profiles produced 8 weeks after transfer (Fig. 1A) indicate no differences between the two groups. Thus, the lack of endogenous TCR expression among naı̈ve CD4 T cells had no impact on the generation or persistence of functional CD4 memory cells.

Figure 1

Recovery of memory CD4 T cells after adoptive transfer of effectors. (A) Development of memory with effectors from RAG KO mice. Naı̈ve CD4 T cells were purified from the spleens of AND TCR transgenic or AND TCR RAG-2 KO mice, and TH1 effectors and TH2 effectors were generated (4, 14). Effectors were transferred to B6 ATXBM mice (15, 16). (B) Memory is proportional to number of transferred cells. TH2 effectors were generated from AND.PL.(Thy1.1) mice. Graded numbers of TH2 effectors were transferred into ATXBM and class II KO mice. Four weeks after transfer, total lymph node and spleen cells of each mouse were recovered and the number donor cell per organ determined (18). (C) Recovery of CD4 memory cells in hosts without class II. TH2 effectors were transferred into ATXBM and class II KO hosts. At various times, hosts were killed and donor cell recovery was determined. Each time point represents the mean and standard error of results from two to three mice, except for class II KO mice at week 5 and 10, for which there was only one mouse.

To evaluate the necessity for TCR-peptide/class II interactions for memory generation from effectors, we transferred graded numbers of TH2-polarized AND.Thy1.1 effectors to ATXBM and class II KO mice (8), and evaluated the number of CD4+, Tg+ memory cells recovered 4 weeks later (16). There was a linear relation between the number of transferred cells and the size of the memory population recovered in both hosts (Fig. 1B). Thus, the recovery of memory is a good measure of the efficiency of memory generation, and both hosts were equivalent in their abilities to support this process. The transferred cells did not expand to repopulate the host to the extent suggested for transferred whole-spleen populations (19). Naı̈ve CD4 T cells that were transferred to class II KO hosts had a shortened life-span (20), as expected from earlier studies (7,8), indicating that contaminating APCs [which are 20 to 30% in naı̈ve populations but undetectable in effectors (14)] are unable to provide necessary signals to promote MHC-dependent survival.

TH2 effectors were transferred to groups of T cell–deficient ATXBM and class II KO mice. The number of donor CD4 memory cells recovered at various times after transfer is indicated (Fig. 1C). In both hosts, the numbers of donor cells recovered initially were high, but fell to somewhat stable levels, which then persisted throughout the 10 weeks of testing. Class II KO mice had smaller lymph node and spleens compared to ATXBM mice, and they contained CD8 T cells, either of which may contribute to the lower absolute numbers recovered in the class II KO hosts. Most CD4 T cells recovered at all times after 2 weeks were small (>95%) cells expressing intermediate to high levels of CD44; of these, most (>80%) were Tg+. Similar results were seen when Th1-polarized effectors were transferred (21).

Recovered cells were purified and restimulated ex vivo (18). The titers of interleukin-2 (IL-2), interferon-γ (IFN-γ), IL-4, and IL-5 in collected supernatants were determined (2). In each case, high production of appropriately polarized cytokines was seen. IL-4 production is shown in Fig. 2A. Over 7 weeks, recovered memory cells produced high levels of IL-4, comparable to those produced by initial effector cells (day 0). At the later times, cytokine production by memory cells recovered from class II KO mice was slightly higher than from ATXBM hosts (Fig. 2A). The cytokine profiles at 6 weeks (Fig. 2B) are representative of the more than 25 mice analyzed.

Figure 2

Cytokine production of recovered memory cells. (A) Maintenance of cytokine production capacity. At different time points after transfer of Thy1.1+ TH2 effectors, Thy1.1 donor T cells were recovered from ATXBM or class II KO recipients. CD4 T cells were purified and recultured at 1.5 × 105 per milliliter with DCEK-ICAM APCs at 0.5 × 105 per milliliter and 5 μM PCCF. Supernatants were collected at 40 hours and assayed for IL-4 production (18). (B) Cytokine production by memory cells. Six weeks after adoptive transfer, Thy1.1 T cells were recovered and restimulated as in Fig. 2A. Supernatants were collected at 40 hours and assayed for IL-2, IL-4, IL-5, and IFN-γ production.

Although class II KO mice do not express class II or contain class II–restricted CD4 T cells (11), they do express CD-1 and class I, so the TCR in the AND mice could interact with these alternate MHC molecules. To examine this possibility, wild-type and class II KO mice were lethally irradiated and injected with bone marrow from AND.Thy1.1 mice. The extent of positive selection was evaluated in each host at week 6 by examining the appearance of single positive CD4+ donor T cells in the thymus and periphery (Fig. 3A). In control hosts, a large population of single positive cells developed (43% of the lymphocyte gate), indicating efficient positive selection, and in the periphery, 6% of the cells were Thy1.1+ (donor) CD4+ cells. In contrast, there were virtually no single positive CD4 T cells in the thymus of class II KO hosts and no detectable Thy1.1+, CD4+ cells (less than in the staining control) in the periphery. Thus, even the weak signals required for positive selection are not available in the class II KO recipients.

Figure 3

Positive selection and CD4 T cell turnover in class II KO hosts. (A) Transfer of AND.B6.Thy1.1 bone marrow to irradiated hosts. Thymus and peripheral staining of class II KO and B6 mice reconstituted with bone marrow cells from AND-B6.PL.(Thy1.1) mice. Class II KO and B6 mice were irradiated with 475 rads (4.75 Gy) twice with a time interval of 4 hours. The irradiated mice were reconstituted with T cell–depleted bone marrow cells from AND-B6.PL mice. After 6 weeks, the thymocytes and total lymph node and spleen cells of reconstituted mice were stained with FITC-CD8α [or CD8β; (24)], PE-Vβ3, cychrome-CD4, and strepavidin-APC-Thy1.1. The thymus staining was gated on live Thy1.1+ cells and peripheral staining were gated on live cells. (B) Memory cell turnover is very slow. BrdU staining of Th2 effectors after adoptive transfer in class II KO mice. Eight weeks after transfer, mice were fed with BrdU-containing water (0.8 mg/ml) for 2 or 4 days. BrdU staining was analyzed on Vβ3-, CD4-, and Thy1.1-positive cells of total lymph node or spleen cells.

In adoptive transfer experiments, minor histocompatibility differences may exist, which could lead to allogeneic reactions. In TCR Tg mice, genes in regions around the Tg may be selectively retained. Moreover, KO mice may themselves recognize on donor cells products of the gene they lack. This should not be a problem in our transfers to class II KO mice, because the hosts are T deficient and the transferred T cells do not express class II. If there are any unexpected responses of donor T cells, we would expect them to undergo some division. The CD4 T cells we recovered were small and resting, but it has been suggested that in normal animals, memory cells divide occasionally as detected by the incorporation of bromodeoxyuridine (BrdU) (22). To evaluate the extent of turnover, BrdU was added to the drinking water of class II KO recipients, which had received TH2 effectors 8 weeks previously. The fraction of donor BrdU+ CD4 T cells in spleen and pooled lymph nodes was determined after 0, 2, and 4 days of labeling. Only 1 to 3% of the CD4 T cells were labeled, a very low rate (Fig. 3B). This lack of donor cell division in the class II KO mice indicates antigen stimulation is indeed negligible and that the cells in the memory population had an intermitotic time on the order of 100 days. Thus, individual memory cells, as well as the population, were long lived in the absence of both antigen and class II.

The antigen independence of the generation of memory from effectors is consistent with properties of T cell effectors. Effector cells restimulated with antigen rapidly divide and produce cytokines in large quantities (2), and most effectors undergo rapid activation-induced cell death (12). Antigen thus promotes continued activation, division, and death, none of which are compatible with development of small resting memory cells. Recent studies of CD8 T cell memory persistence, in a model where donor class I expression is minimized, suggest that CD8 memory cells, like the CD4 memory cells studied here, are antigen- and MHC-independent (23).

The transition from naı̈ve to effector and memory cells thus includes a loss of dependence on interaction with self MHC for prolonged survival. With this transition, removal of antigen—not persistence in an immunogenic form—fosters memory generation and longevity.

  • * To whom correspondence should be addressed. E-mail: sswain{at}


Stay Connected to Science

Navigate This Article