Main

CD8-positive CTLs are responsible for the lysis of antigen-bearing target cells. These CTLs recognize peptide antigens presented by class I molecules encoded within the major histocompatibility complex (MHC). Generation of effective CTL responses often requires help from a second subset of T lymphocytes, the CD4-positive helper T (TH) cells1,2,3,4, but this is not always the case6,7. In response to the soluble protein ovalbumin (OVA), both TH-cell-dependent and TH-cell-independent CTL immunity can be induced4,8. When the Kb-restricted OVA peptide determinant spanning residues λ57 to 264 (OVAp) was emulsified in complete Freund's adjuvant (CFA) and injected subcutaneously, CTLs could be generated in mice lacking CD4-positive T cells (Fig.1a)4. In contrast, priming by intravenous injection of irradiated spleen cells loaded with OVA by osmotic shock (OVA-loaded spleen cells) required the presence of CD4-positive T cells (Fig. 1b). This latter form of immunization occurs by cross-priming4,9, requiring re-presentation of antigen by host bone-marrow-derived antigen-presenting cells (APCs). Dissection of the cellular interactions involved in this response4 revealed that, like the induction of CTLs that are specific for the Qa1 antigen1, the TH and CTL populations must recognize OVA on the same APC for effective CTL priming. This could be explained in two ways: either the TH cells need to closely associate with the CTL to deliver short-range signals such as interleukin(IL)-2 (ref. 1), or they are required to modify the APC, converting it into a stimulatory cell for CTL priming5.

Figure 1: OVA-specific CTLs can be generated in CD4-independent and CD4-dependent ways.
figure 1

Normal B6 mice (filled circles) or B6 mice depleted of CD4-positive T cells by twice-weekly intraperitoneal injection of 100 µl GK1.5 ascites4 (open circles) were injected either a, subcutaneously with 20 µg OVAp in 200 µl CFA, or b, intravenously with irradiated B6 OVA-loaded spleen cells as described4. After 8 days, spleen cells from each mouse were restimulated for 6 days in vitro as described4. On the day of assay, effector cells were examined for their ability to lyse 51Cr-labelled EL4 targets that were or were not pulsed with OVAp. Nonspecific EL4 lysis was <10%.

Full size image

There is evidence that CD40 and CD40 ligand (CD154) are important in both the humoral and cellular immune responses (reviewed in refs 10, 11). These molecules have been implicated in the generation of CTL responses to adenovirus12 and in the establishment of CTL memory to lymphocytic choriomeningitis virus13. CD154-deficient mice are unresponsive to adenovirus in that they do not generate either antibody or CTLs14, unless co-injected with a CD40-stimulating monoclonal antibody. However, it was not determined which cellular population was normally responsible for supplying CD154 for CTL induction; CD4-positive TH cells, CD8-positive CTLs, or some other cell type. We reasoned, however, that as this ligand is mainly expressed by activated TH cells15,16, CD154 signalling to CD40 may represent the ‘help’ provided by these cells during CTL generation. To test this, we determined whether a CD40 stimulus could substitute for TH cells in the CTL response to OVA-loaded spleen cells. Mice lacking the MHC locus H–2Ab, which do not express MHC class II molecules and therefore lack class II-restricted CD4-positive TH cells, were primed with OVA-loaded spleen cells in the presence of either a CD40-stimulating monoclonal antibody, FGK45 (ref. 17), or an isotype-control antibody. We then examined OVA-specific CTL generation (Fig. 2). We used spleen cells from bm1 mice as the OVA-loaded priming source to ensure that stimulation could only occur through cross-priming on host APCs, as the H–2bm1 haplotype cannot directly present OVA to CD8-positive T cells. In this case, strong CTL responses were induced in the presence, but not the absence, of the CD40-specific monoclonal antibody. There was no CTL priming when mice were given FGK45 in the absence of OVA-loaded spleen cells (data not shown).

Figure 2: Treatment with a monoclonal antibody against CD40 replaces CD4-positive T-cell help in generating OVA-specific CTLs.
figure 2

Normal B6 mice (circles) and H–2Ab-deficient B6 mice (squares), were injected intravenously with irradiated, OVA-loaded bm1 spleen cells4. They were then left untreated (circles), or were injected intravenously daily for 4 days with either 0.1 mg of a CD40-specific monoclonal antibody, FGK45 (ref. 17) (open squares) or an IgG2a isotype control, KT50, specific for Vα8 (filled squares). Eight days after priming, the spleen cells from each mouse were restimulated in vitro for six days. We then examined their ability to lyse 51Cr-labelled EL4 targets that were or were not pulsed with OVAp. Nonspecific EL4 lysis was <7%.

Full size image

To exclude the possibility that the few unusual CD4-positive T cells in H–2Ab-deficient mice could substitute for TH cells when a CD40 stimulus was provided, we performed similar experiments in thymectomized B6 mice that were depleted of CD4-positive T cells by administration of monoclonal antibody (Fig. 3). This showed that, even when CD4-positive T cells were depleted by antibody, mice could be primed only in the presence of a CD40-stimulating monoclonal antibody. These data indicate that a CD40 stimulus can substitute for TH cells, suggesting that help is normally mediated through CD40 signalling following engagement by CD154 expressed on TH cells.

Figure 3: A CD40-specific monoclonal antibody can replace CD4-positive T-cell help in the induction of OVA-specific CTLs.
figure 3

Thymectomized B6 mice were depleted of CD4-positive T cells by twice-weekly intraperitoneal injection of GK1.5 acsites4 (squares) or were left untreated (circles). All mice were then immunized with 25 × 106 OVA-loaded B6 spleen cells. CD4-depleted mice were then either treated with 0.1 mg anti-CD40 antibody intravenously daily for 7 days (open squares), or left untreated (filled squares). Spleen cells were then cultured in vitro for 6 days with 107 irradiated E.G7 cells. The resulting effector cells were then included in a chromium-release assay using EL4 target cells that were or were not pulsed with OVAp. Nonspecific EL4 lysis was <5%.

Full size image

An alternative explanation, however, is that signalling by the CD40-specific monoclonal antibody substituted for a completely different signal supplied by TH cells. To exclude this possibility, we examined OVA-specific CTL responses in CD40-deficient or CD154-deficient mice. If CD40 signalling is normally involved in this form of priming, these mice should not generate OVA-specific CTL responses. CTL responses were indeed poor in both CD154-deficient mice (Fig. 4a) and CD40-deficient mice (Fig. 4b). In one of four experiments, using CD40-deficient mice, we detected relatively efficient CTL responses (data not shown). As we performed this experiment soon after introducing the CD40-deficient mice into our colony, we suspect that subclinical infections provided inflammatory signals that substituted for the CD154 signal of TH cells, in much the same way as CFA replaced the need for TH cells in the generation of OVA-specific CTLs4,8 (Fig. 1a). The general lack of CTL responses in mice lacking either CD40 or CD154 support a role for CD40 signalling in the cross-priming of OVA-specific CTLs.

Figure 4: The role of CD154 and CD40 in the generation of OVA-specific CTLs by cross-priming.
figure 4

a, We primed four CD154-deficient mice29 (open squares) and two normal B6 mice (filled squares) with bm1 OVA-loaded spleen cells, and then examined for CTL generation. We performed this experiment twice, with similar results. b, We immunized two CD40-deficient30 mice (open squares) and two normal B6 mice (filled squares) intravenously with bm1 OVA-loaded spleen cells, and then examined for CTL generation. We performed this experiment four times, with three experiments giving similar results. Nonspecific EL4 lysis was <36%.

Full size image

Priming of CTLs by OVA-loaded spleen cells occurs through antigen presentation by host APCs. This priming requires help mediated by a TH-cell interaction with the same APCs as seen by the CTLs4. The requirement for CD40 and CD154 in this response (Fig. 4), and the ability of a CD40-specific monoclonal antibody to replace TH cells (Figs 2, 3), indicates that CD4-positive T-cell help mainly supplies CD154 for CD40 signalling. As CD40 is expressed by APCs such as dendritic cells (reviewed in ref. 10), and as signalling through CD40 is important for maturation of several antigen-presenting functions14,18,19,20,21,22,23,24,25, the cross-priming APC probably represents the target cell that is stimulated through CD40. B cells, which also express CD40, are unlikely to be involved in this response, as presentation of antigen by this population is tolerogenic for naive CD8-positive T cells, even in the presence of TH cells26. A small proportion of CD8-positive T cells has been reported to express CD40 (ref. 27), so a direct interaction between CD154 on the TH cell and CD40 on the CTL may occur. This would, however, require interaction between these cells while associated on the APC, as murine CTLs do not express MHC class II molecules.

How signalling through CD40 enables the cross-priming APC effectively to stimulate CTLs is unclear, although CD40 signalling may lead to the upregulation of co-stimulatory molecules such as B7 (refs 14, 19), CD44H (ref. 20) and ICAM-1 (ref. 21) on various cell types, and to the secretion of cytokines, such as tumour-necrosis factor-α, IL-1β and IL-12 (refs 10, 19,22–25,). Whether these or other events form the critical downstream helper effects provided by TH cells is unknown. Although CD40–CD154 interactions are important for the induction of CTL immunity to adenovirus12, they are not essential for induction of CTL responses to several other viruses13,28. We suggest that, like CFA, which allows the induction of TH-cell-independent CTL responses to OVA, some viruses may provide inflammatory signals that activate APCs directly, circumventing their need for activation through CD40. Perhaps the requirement for CD40 signalling depends on whether the professional APC is directly infected or presents antigen indirectly by cross-presentation. Alternatively, there may be other inherent properties of some viruses that allow CD40-independent activation of APCs.

The ability to replace TH cells with an antibody against CD40 may provide an opportunity for vaccination under circumstances that were previously difficult, either because of a lack of available class II-restricted antigenic determinants (as may occur with some tumour antigens) or because of a deficiency in CD4-positive TH cells (as occurs in AIDS patients). Further dissection of the molecular interactions involved in the generation of immunogenic CTL responses may help in vaccine design and function.

Methods

Generation of CTL responses. CTL responses to OVA were generated as described4. Briefly, mice were primed with 25 × 106 OVA-loaded spleen cells. Seven to eight days later, we restimulated spleen cells from primed mice in vitro for six days with irradiated (1,500 centiGray) OVA-loaded B6 spleen cells or irradiated (20,000 centiGray) E.G7 tumour cells. We then examined the ability of the primed spleen cells to lyse 51Cr-labelled EL4 target cells, pulsed or unpulsed with OVAp as described4. The percentage of OVA-specific lysis was then calculated as the percentage of OVAp-pulsed EL4 lysis minus the percentage of EL4 lysis. To generate CTLs in CD154-deficient29 or CD40-deficient30 mice, we modified the in vitro culture conditions slightly to avoid responses to mouse strain 129 minor histocompatibility antigens. Seven days after immunization with OVA-loaded bm1 spleen cells, we removed ‘responder’ spleens and irradiated half the cells from each spleen (1,500 centiGray). We then loaded these irradiated cells with OVA with OVA by osmotic shock and used these cells as in vitro stimulators for the remaining splenocytes. We combined unirradiated ‘responder’ spleen cells (3 × 106 cells ml−1) with irradiated OVA-loaded cells from the same animal (3 × 106 cells ml−1) and cultured them in vitro for six days in several 2-ml cultures. On day 6, we pooled effector cells from each mouse and used them in a chromium-release assay as described4.