A Role for CD40 Expression on CD8+ T Cells in the Generation of CD8+ T Cell Memory

See allHide authors and affiliations

Science  20 Sep 2002:
Vol. 297, Issue 5589, pp. 2060-2063
DOI: 10.1126/science.1072615


The delivery of CD4 help to CD8+ T cell responses requires interactions between CD40 and CD40 ligand and is thought to occur through antigen-presenting cell (APC) activation. Here we show that generation of memory CD8+ T cells displaying an enhanced capacity for cell division and cytokine secretion required CD4 help but not CD40 expression by the APCs. Activated CD4+and CD8+ T cells expressed CD40; and in the absence of this protein, CD8+ T cells were unable to differentiate into memory cells or receive CD4 help. These results suggest that, like B cells, CD8+ T cells receive CD4 help directly through CD40 and that this interaction is fundamental for CD8+ T cell memory generation.

Interactions between CD40 and CD40 ligand (CD40L) play a major role in direct CD4+ T cell–B cell collaboration, and the absence of these molecules results in a failure of germinal center formation, memory B cell activation, immunoglobulin class switching, and somatic hypermutation (1–3). The CD40 and CD40L deficiencies also impair CD8+ T cell responses, suggesting an important role for these molecules in CD4+ and CD8+ T cell collaboration (4–6). This role has been confirmed by experiments showing that monoclonal antibodies (mAbs) to CD40 could substitute for the delivery of CD4 help to CD8+ T cells in vivo (7, 8). Because CD40 is expressed on antigen-presenting cells (APCs) (9), it has been assumed that CD4 help to CD8+ T cells is indirect. Thus, CD4+CD40L+ T cells would first activate APCs by engaging CD40, allowing activated APCs to become competent to drive CD8+ T cell responses (7, 8, 10). Other experimental data, however, have suggested dissociation of APC activation from CD4-CD8+ T cell collaboration. For example, CD8+ T cells can activate APCs without CD4+ T cells (11). In addition, CD4+ T cells can activate APCs through a CD40-independent pathway (12), and CD40-deficient APCs retain their stimulatory capacity (13, 14), even under conditions where CD8+ T cell priming is disrupted (14).

To reconcile these findings, we used an experimental system in which each component of the response (i.e., CD8+ T cells, CD4+ T cells, and APCs) could be independently manipulated (15, 16). Thus, CD8+ T cells from female Rag2-deficient mice expressing a transgenic T cell receptor (TCR) specific for the male histocompatibility-Y (HY) antigen [transgenic (Tg) cells] were stimulated in an in vivo setting with male cells either in the presence or complete absence of CD4+ T cells. To achieve this, we first transferred Tg cells to CD3ɛ-deficient female mice, which lack T cells and express the Ly5.2 congenic marker, with or without CD4+T cells from female mice. These mice were subsequently immunized with cells from Ly5.1 CD3ɛ-deficient male mice (see supporting text S1). In the presence of CD4+ T cells, Tg cells eliminated these male cells, ultimately becoming efficient memory cells. In this respect, they displayed a significantly enhanced capacity for cell division and cytokine secretion, as compared with that of naı̈ve Tg cells (17).

The absence of CD4+ T cells did not prevent the elimination of male cells (Fig. 1A), indicating that high frequencies of CD8+ T cells could kill target cells in the absence of help (18). Consistent with this, Tg cells stimulated by antigen in the presence or absence of CD4+ T cells showed similar changes in the expression of CD69, CD44, L-selectin, interleukin-2R, and CD45 isoforms, as well as TCR and CD8+ T cell levels (19).Additionally, up to day 4 after immunization, the presence of CD4 help did not modify proliferation (Fig. 1B); cell accumulation (Fig. 1C) and cytokine secretion remained at the level seen in naı̈ve cells (19). However, after day 5 after immunization, recovery of Tg cells increased gradually in mice coinjected with CD4+ T cells (Fig. 1C). In the presence of CD4 help, cytokine secretion also increased with time during the primary response, whereas it remained at a naı̈ve cell level in the absence of help (Fig. 1D).

Figure 1

The role of CD4 help in CD8+ T cell responses. Mice carrying either Tg cells or Tg cells and CD4+ T cells were immunized in vivo with male cells expressing the Ly5.1 marker (16). (A) Elimination of Ly5.1 male cells by Tg cells in the presence or absence of CD4+ T cells, 2 months after priming. (B) Tg cells, labeled with 5- (and 6-) carboxyfluorescein diacetate succininyl ester (CFSE), 3 days after immunization with male cells. Dotted lines represent CFSE labeling of Tg cells in mice that were not immunized. By day 5 after immunization, all cells had lost CFSE labeling. (C) The number of Tg cells recovered at different times after immunization in one of three experiments. (D) Interferon-γ (IFN-γ) secretion by purified Tg cells (naı̈ve, or recovered at different days after immunization). Purified Tg cells (naı̈ve, or recovered 2 months after priming) were stimulated in vitro with male APCs: (E) optimal [3H]thymidine incorporation and (F) cytokine secretion. CPM, counts per minute; IL-2, interleukin-2.

We next studied the properties of memory T cells recovered 2 months after immunization. In response to male cells or to stimulation with antibody to CD3, Tg cells primed in the absence of help proliferated poorly (Fig. 1E), and cytokine secretion was similar to that seen in naı̈ve Tg cells (Fig. 1F). Thus, whereas CD4 help did not modify the initial response of naı̈ve CD8+ T cells and did not prevent their differentiation into effector cells, it appeared to be fundamental for the generation of efficient memory cells, which was shown by their improved functional capacity relative to naı̈ve cells (see supporting text S2 and S3).

To establish the role of the APCs on CD8+ memory T cells generation, we first examined whether delivery of CD4 help to CD8+ T cells required the formation of a three-cell cluster, in which antigen-specific CD4+ and CD8+ T cells interact with antigen presented by the same APCs (20, 21). Help did require the presence of antigen-specific CD4+ T cells, because depletion of these lymphocytes resulted in a poor response by Tg cells (see supporting text S4) (Fig. 2A). We then examined the requirement for CD4 help, when CD4+ and Tg CD8+T cells were forced to recognize antigen presented by different APCs. CD4+ and Tg CD8+ T cells were transferred into major histocompatibility complex (MHC) class II–/– mice and immunized with two types of male cells: those from MHC class II–/– mice (which could only interact with CD8+ T cells through MHC class I) and those from MHC class I Db–/– mice (which could only interact with CD4+ T cells) (see supporting text S5). In this case, help was absent, demonstrating that efficient CD8+ Tg responses required antigen to be recognized by CD8+ and CD4+ T cells on the same APCs (Fig. 2, A and B). We next tested the potential role of CD40 expression by the APCs in facilitating CD4-CD8+ T cell collaboration. When APCs were derived from CD40–/– mice, the helper effect remained unaffected (Fig. 2, C and D). These results demonstrate that CD4-CD8+ T cell collaboration requires the formation of a CD4–CD8+ T cell–APC cluster but is independent of the CD40 expression by the APCs.

Figure 2

The role of APCs in delivering CD4 help. (A and C) Number of Tg cells per mouse (mean ± SEM of three mice per group, in one of three experiments). (B and D) Optimal cytokine secretion of purified Tg cells (16) 2 months after in vivo immunization with male cells. In (A) and (B), Tg cells were primed without help (open bars) or in the presence of CD4+ T cells in three different regimes: with CD4+ T cells depleted of male-specific cells [hatched bar in (A)], when CD4+ and Tg cells recognized antigen on different APCs (gray bars), or when CD4+ and Tg cells recognized antigen on the same APCs [right-hand black bar in (A) and black bars in (B)]. In (C) and (D), Tg and CD4+cells were transferred to CD40+ female mice and immunized with CD40+ male cells (open bars) or were transferred to CD40–/– female mice and immunized with CD40–/– male cells (solid bars).

It has previously been shown that antibodies to CD40 can substitute for CD4 help in optimizing CD8+ T cell responses. One interpretation for reconciling these findings with our results is that T cells might also express CD40. Consistent with this, CD4+and CD8+ T cells expressed very low levels of CD40 mRNA, and these transcripts were up-regulated after activation (Fig. 3A). Quantification of CD40 mRNA expression by real-time reverse transcriptase–polymerase chain reaction (RT-PCR) in 500 activated CD4+ and CD8+ T cells showed that the number of mRNA molecules was similar in each type of activated T cell and, in these activated cells, was 1/50 to 1/100 times that of B lymphocytes (Fig. 3B). Naı̈ve and memory Tg cells lacked CD40 cell surface expression, although induction of this protein was evident after activation but was again lower than that of B cells (Fig. 3C). These results reveal transient expression of CD40 by both CD4+ and CD8+ T lymphocytes during immune responses. These cells also express CD40L under these conditions (19, 22, 23) [see supporting text S6 (16)].

Figure 3

Expression of CD40 on T cells. C57BL/6 B cells, Tg cells, and CD8+ and CD4+ T cells, either naı̈ve or stimulated in vitro by mAbs to CD3 were sorted. (A) cDNA was amplified with specific CD40 primers. Results are the Southern blots of PCR products hybridized with 33P-labeled CD40-specific probes. CD40 cDNA, 104 cells; hypo- xanthine-guanine phos- phoribosyl-transferase (HPRT), 103 cells. (B) CD40 mRNA quantitation was determined by SYBR-Green incorporation in real-time RT-PCR (16) in 500 naı̈ve B cells and 10 hours in in vitro activated T cells. (C) CD40 cell surface expression on naı̈ve B cells (dotted line) and on naı̈ve Tg cells (dashed line) or primed Tg cells (solid line) at different time points after antigen stimulation. Labeling with an isotype control was at the same level in naı̈ve and activated cells (19).

To determine whether CD40 expression by CD8+ T cells was required to receive CD4+ T cell help, we studied monoclonal CD40-deficient TCR-Tg mice. The CD40 deficiency did not modify the naı̈ve Tg cell response to antigen in vitro, because proliferation (Fig. 4A) and cytokine secretion (19) were analogous to those of naı̈ve CD40+ Tg cells. Early cell division, measured in vivo during the primary immune response, was also not affected (Fig. 4B). In contrast, CD40 deficiency had a major impact on the properties of primed T cells generated during the primary immune response. CD40+ Tg cells in mice primed in the presence of CD4+ T cells and challenged again by antigen in the presence of CD4+ T cells showed stronger proliferation than did naı̈ve cells (compare Fig. 4B, top, with Fig. 4C, top) and secreted higher levels of cytokines (Fig. 4D). In contrast, CD40–/– Tg cells, stimulated in vivo in the presence of help, proliferated poorly (Fig. 4C). Cytokine secretion after antigen restimulation remained similar to that of naı̈ve cells throughout the primary immune responses (19) and in the memory phase (Fig. 4D). We conclude that without CD40 expression on Tg cells, these cells could not acquire the capacity for rapid proliferation or high levels of cytokine secretion, even in the presence of CD4 help.

Figure 4

Role of CD40 expression on CD8+T cell function. (A) Dose curve of in vitro proliferation measured by [3H]thymidine incorporation of naı̈ve Tg cells from CD40+ (open symbols) or CD40–/–(solid symbols) in response to increasing doses of HY peptide. (B) In vivo proliferation of naı̈ve Tg cells: CD40+ or CD40–/– naı̈ve Tg cells were labeled with CFSE, immunized in vivo with male cells in the presence of CD4 cells, and followed from 1 to 7 days after immunization. Data show CFSE labeling on day 3. (C and D) CD40+ or CD40–/– Tg cells, along with CD4+ T cells, were immunized with male cells and isolated 3 months after immunization. In (C), these Tg cells were labeled with CFSE and restimulated in vivo with antigen in the presence of CD4+ T cells. Data show CFSE-labeled cells on day 3 after immunization. CD40+ memory cells lost CFSE labeling by day 4 after immunization, whereas CD40–/–-primed cells were CFSE-negative by day 7. In (D), these Tg cells were restimulated in vitro with male cells, and optimal cytokine secretion was determined. (E) CD4 help to Tg CD40–/– cells. CD40–/– Tg cells were primed in vivo with male cells (with or without CD4 cells) and recovered 3 months after immunization. Results show optimal cytokine secretion after antigen restimulation.

To determine whether CD40 expression by CD8+ T cells was necessary and sufficient to allow CD8+ T cells to receive helper signals specifically from CD4+ T cells, we primed CD40–/– Tg cells in the presence or absence of CD4+ T cells and tested them after antigen restimulation, in vitro. The presence of CD4 help did not modify the behavior of CD40–/– Tg cells (Fig. 4E), demonstrating that CD4 helper effects were mediated through interactions with CD40 expressed by CD8+ T cells.

Our data indicate that CD4+ help to CD8+responses does not require CD40 expression by the APCs but depends instead on the expression of CD40 by CD8+ T cells. These data suggest a reevaluation of the cellular interactions involved in CD4-CD8+ T cell collaboration. The CD40L signals required by CD8+ T cells do not appear to originate from an autologous source, as CD40L-deficient CD8+ T cells maintain the capacity to receive CD40 signals (24). Because murine APCs do not express CD40L (24–26), it is likely that CD40L signals to CD8+ T cells originate from activated CD4+ T cells, with help involving a direct CD4-CD8+ T cell interaction. This direct interaction explains previous data suggesting a dispensable role of CD40 interaction on APC activation in CD4 helper activity (11–14, 24). It is also compatible with the capacity of mAbs to CD40 to substitute for CD4 help in vivo (7, 8) but suggests that the mAb in these experiments acted directly on CD8+ T cells. Other reports have shown that activated APCs result in the detection of cytolytic T lymphocyte (CTL) activity in the absence of CD4 help (10). These activated APCs, however, could have influenced CD8+ T cell responses by an independent mechanism, different from that used by CD4+ T cells, because it has been demonstrated that differentiation of CD8+ T cells into CTL does not require CD4 help (18), (see supporting text S7).

Our data suggest that the differentiation of CD8+ T lymphocytes into efficient primed cells is analogous to isotype switching and somatic hypermutation during memory B cell generation. Both phenomena are apparent relatively late in the primary immune response, requiring CD4 help and CD40 expression by the target cell. The requirement for this type of direct CD4-CD8+ T cell interaction suggests that a reevaluation of strategies of vaccination to induce optimal CD8+ T cell response might be appropriate. Although activated APCs may not be sufficient to substitute for CD4 help, our data suggest that mAbs to CD40, which operate directly on CD8+ T cells, might have therapeutic value in correcting CD8+ T cell dysfunction associated with CD4 lymphopenia, as found in HIV infection. The expression of both CD40 and CD40L on activated CD4+ and CD8+ T cells and the expression of CD40 on the APCs suggest that although help can be received by CD8+ T cells directly, complex cross talk between these three populations may nevertheless exist.

Supporting Online Material

Materials and Methods

Supporting Text S1 to S7


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


View Abstract

Navigate This Article