Ox-40 Ligand: A Potent Costimulatory Molecule for Sustaining Primary CD4 T Cell Responses¹

Productive primary and secondary T cell responses require periods of extensive expansion of Ag-specific cells, which normally exist at low frequency in both the naive and memory pools. Without this expansion, sufficient numbers of Ag-reactive cells are not generated to fully neutralize the Ag, resulting in an immune response that is essentially ineffective. Moreover, T cell activation without appreciable expansion can result in tolerance involving either cell death, hyporesponsiveness, or a combination of both (1, 2), with again the overall result being an inability to mount a protective response.

The factors that regulate T cell expansion are just beginning to be elucidated. It is generally accepted that two signals are required for T cell activation, one provided by TCR interaction with peptide/MHC complexes, and another by coreceptor interaction with APC accessory molecules (3, 4). Ligation of CD28 by B7-1 and/or B7-2 has been proposed to be the predominant costimulatory interaction(s), and the majority of studies have focused on these molecules (5, 6, 7, 8, 9, 10). However, recent data suggest that this simple two-signal model may be an oversimplification and should incorporate multiple signals from various sources, all of which synergize with those through the TCR and perhaps with those through CD28 (see Ref. 11 for review). Thus, in isolated systems, other ligand pairs have been shown to costimulate CD4 responses when combined with TCR signals. These are generally molecules in the Ig superfamily, or TNF/TNFR families, and include the interactions of ICAM (-1,-2) with LFA-1, CD40 with CD40L,³ VCAM-1 (CD106) with VLA-4 (CD49d/CD29), HSA (CD24) with an unknown receptor, and CD70 with CD27 (12, 13, 14, 15, 16, 17).

Because of the plethora of accessory molecules that have been implicated in costimulation, it is unclear which interactions are important and whether only some, or all, are essential for generating a T cell response. It is possible that many accessory molecules play identical roles and are therefore redundant much of the time. An alternative possibility, which is more attractive, is that accessory molecules act in concert with one another, and it is the combined action of several interactions that leads to the most effective response. Thus, multiple interactions may be required to initiate the T cell response, and/or other interactions may be needed at different times after the initial events to allow the response to progress efficiently. In this scenario, accessory molecule-coreceptor ligation would occur either concomitantly, or sequentially in a stepwise fashion, and the type of interaction would be dictated by the expression of the molecules on both the T cell and APC surface, and the nature of the APC encountered. In vitro studies that have assessed two molecules simultaneously suggest that an efficient T cell response does require the combined use of several of these ligand-receptor pairs, and synergy has been described between CD40, ICAM-1, and CD70 when combined separately with B7 (12, 15, 18). In vivo data using CD40- and CD40L-deficient mice have shown that the interaction between these molecules is essential for a productive T cell response (19, 20, 21), and similarly, blockade of ICAM/LFA-1 interactions results in deficient immune responses (22, 23).

While many of the above interactions may be important for initiation of T cell responses, with the coreceptors either being constitutively expressed on T cells or up-regulated very shortly after activation, it is likely that other interactions may be necessary for prolonging or propagating the response. In this regard, two members of the TNFR family, namely 4-1BB and Ox-40, have recently gained prominence as costimulatory receptors (24, 25, 26, 27, 28, 29) and are potential candidates for delivering signals that could function during the later stages of T cell responses. Studies with mitogen stimulation have shown that both receptors are induced on T cells several days after initial activation (26, 30), and that their ligands may only be present on previously activated APCs (31, 32). While 4-1BB is expressed on both CD4 and CD8 cells, and may be more important for CD8 responses (33, 34), Ox-40 appears to be predominantly expressed on CD4 cells (35, 36). Ox-40-positive T cells have been detected in situ in several inflammatory states, including experimental autoimmune encephalomyelitis, rheumatoid arthritis, and graft-vs-host disease (36, 37, 38). The clinical importance of Ox-40-expressing T cells has been demonstrated in experimental autoimmune encephalomyelitis, in which disease prevention was achieved by selective depletion with an immunotoxin-conjugated Ab to Ox-40 (36). In addition to a role in costimulation of CD4 cells, recent data suggest that Ox-40/Ox-40L interactions are intimately involved in effector functions. Ox-40L cross-linking has been shown to promote B cell proliferation and Ab secretion (32), and enhance dendritic cell function (39), and blocking this interaction can inhibit both primary and secondary IgG responses (40). Interestingly, both B cell and dendritic cell responsiveness to Ox-40L appear to involve CD40 signaling, again promoting the concept that Ox-40/Ox-40L interactions may primarily function during later phase responses after engagement of one or more other ligand-receptor pairs.

To clarify the role of Ox-40 in primary CD4 responses, we have assessed the capacity of Ox-40L on Ag-presenting fibroblast transfectants, expressed either alone or in combination with B7, to stimulate responses of T cells derived from TCR transgenic animals. Although Ox-40 was readily expressed 24–48 h after Ag stimulation, APCs transfected with Ox-40L had little ability to induce IL-2 secretion or proliferation of naive T cells. In contrast, strong synergy was observed when B7 was coexpressed, resulting in enhanced proliferation that persisted for several days. More striking was the ability of APC expressing Ox-40L to promote secretion of several cytokines, including IL-2, from previously generated effector T cells. This resulted in further extensive proliferation, which was again augmented if CD28 signals were provided. These data show that Ox-40 signals are a potent source of costimulation for a developing primary CD4 response, and that the major action is during the late phase response promoting continued proliferation and expansion of effector cells.

AND TCR transgenic mice were bred on a B10.BR background, as previously described (41, 42), and used as a source of Vβ3/Vα11 CD4 cells responsive to peptides of pigeon cytochrome c (PCC) presented on I-E^k.

Fibroblast cells transfected with I-E^k (originally generated by Dr. R. Germain, National Institutes of Health, Bethesda, MD) were used as APCs. Two fibroblast lines were used as before (18, 41), which expressed, or lacked, B7-1 (DCEK.Hi7, referred to as B7; DCEK, referred to as −/−). These cells do not express ICAM-1, VCAM-1, VLA-4, B7-2, Ox-40L, 4-1BBL, LFA-1, HSA, or CD48 by FACS analysis. Both lines were transfected with the hygromycin resistance-encoding expression vector pREP4 (Invitrogen, San Diego, CA) into which was inserted cDNA for Ox-40L (a kind gift from Dr. Wayne Godfrey, Stanford University School of Medicine, Stanford, CA) (25). Ox-40L transfectants were selected, and stable lines were cloned by limiting dilution (Fig. 1). APC populations were treated with mitomycin C (75–100 μg/ml; Sigma, St. Louis, MO) for 30 min at 37°C before use.

CD4⁺ T cells were purified from the spleen of TCR transgenic mice (41, 42) by nylon wool depletion, followed by complement treatment with Abs to CD8 (3.155), heat-stable Ag (JIID), class II MHC (M5/114 and CA-4.A12), macrophages (M1/70), and dendritic cells (33D1), cross-linked with mouse anti-rat κ (MAR 18.5). Any residual APC and any in vivo activated T cells were removed by isolating high density cells spun through a Percoll gradient (45, 53, 62, 80%). The resultant cells were resting (low FSc, CD69⁻, CD71⁻, CD25⁻) and >95% CD4⁺, and >95% of these cells possessed a phenotype associated with naive CD4 cells (CD45RB⁺, CD62L⁺, CD44^low) along with expression of the Vβ3/Vα11 TCR (41, 42). Effector T cells were derived as in previous studies by stimulating naive cells for 4 or 5 days in vitro with a peptide of PCC 88–104 presented on B7⁺ICAM⁺DCEK cells. Cultures were supplemented with either IL-2 (10 ng/ml) and IFN-γ (100 ng/ml) to generate Th1-like cells secreting predominantly IL-2 and IFN-γ, or IL-2, IL-4 (20 ng/ml), and anti-IFN-γ (10 μg/ml) to generate Th2-like cells secreting predominantly IL-4 and IL-5.

Cells were cultured in RPMI 1640 (Irvine Scientific, Santa Ana, CA) with penicillin, streptomycin, glutamine, 2-ME, sodium pyruvate, and 7% FCS (HyClone, Logan, UT, and Irvine Scientific, Santa Ana, CA). Cultures were generally set up in 0.2-ml vol in 96-well plates (Costar, Cambridge, MA) in triplicate. Naive and effector CD4 cells were plated at a concentration of 1 × 10⁵/ml with varying numbers of fibroblast APCs that had previously been pulsed with an optimal concentration of PCC peptide (20 μM) at 2 × 10⁶/ml for 2 h at 37°C. In the cases in which cytokine secretion was inhibited, blocking Abs to IL-2 and IL-4 (PharMingen, San Diego, CA) were added at the time of stimulation at 20 μg/ml.

The phenotype of the DCEK transfectants was confirmed with Abs to class II MHC (M5/114, rat IgG2b, from American Type Culture Collection (ATCC), Manassas, VA), B7-1 (IG10, rat IgG2a, from PharMingen), and Ox-40L (Ox-40.Fc, a human Ig fusion protein with the extracellular portion of murine Ox-40, from Cantab Pharmaceuticals, Cambridge, U.K.). Staining was visualized with FITC-labeled mouse anti-rat κ (RG7.9), and FITC goat anti-human IgG (Caltag, Burlingame, CA). Isotype controls were rat IgG and human IgG. T cell activation was assessed by staining for CD25 (IL-2R) with FITC-labeled anti-CD25 (rat IgM; PharMingen). Ox-40 expression was measured using an Ig fusion protein with the extracellular portion of Ox-40L (Ox-40L.Fc; Cantab Pharmaceuticals), followed by biotin anti-human IgG (Caltag) and phycoerythrin-streptavidin (PharMingen). CD4⁺ T cells were identified using phycoerythrin- or FITC-conjugated anti-CD4, respectively (PharMingen). FACS analyses were performed on a FACScan flow cytometer (Becton Dickinson, Mountain View, CA) with Cellquest software.

Cell division was assessed by addition of 1 μCi of tritiated thymidine (ICN Biomedicals, Irvine, CA) to 0.2-ml cultures for approximately 18 h at the end of culture. Response was assessed at different times depending on the experiment, between either 24 and 42, 48 and 66, 72 and 90, or 96 and 114 h. Values shown represent mean levels from triplicate cultures, and in each case SDs were less than 15% of the means.

Triplicate supernatants were recovered 20–24 h (effector T cells) and 35–40 h (naive T cells) after stimulation and pooled to assess cytokine content. IL-2 production was determined as before (18, 42) by titrating pooled replicate supernatants onto NK.3 cells, in duplicate, in the presence of anti-IL-4 (purified from the 11B11 cell line, ATCC). IL-4, IL-5, and IFN-γ were measured by ELISA as in previous studies (42), using the Abs 11B11 and biotinylated anti-IL-4 (BVD6; PharMingen), TRFK5 and biotin-TRFK4, and R46A-2 and biotin-XMG1.2, respectively. Standard curves were constructed with purified IL-2, IL-4, IL-5, and IFN-γ (supernatants from the respective X63.Ag. cell lines). All values shown are means of replicates, with SDs in each assay being less than 15% of the means.

The role of Ox-40 during CD4 T cell responses was analyzed using cells isolated from Vβ3/Vα11 TCR transgenic mice that are responsive to peptides of pigeon and moth cytochrome c presented on I-E^k-expressing APCs (42). Ox-40 expression was analyzed on naive CD4 cells stimulated with plate-bound anti-CD3 in the presence or absence of costimulation from anti-CD28, or with PCC peptide presented on I-E^k-bearing fibroblast APCs that do or do not express B7-1 (18). In addition, effector T cells generated from naive T cells over 5 days in vitro were also analyzed for Ox-40 expression in response to Ag. Naive CD4 cells did not express Ox-40, but induction was evident within 24 h after activation, and similar levels were seen regardless of the mode of stimulation (Fig. 2,a). Interestingly, the presence of B7-1, or CD28 signaling (anti-CD28), was not required for induction or maximal expression, even with limiting Ag or APC number (data not shown). This is in direct contrast to our previous data on another member of the TNF/TNFR family, CD40L, whose expression required the presence of B7-1 (41). Ox-40 expression was not evident within 6 h of activation, but a significant number of naive cells were positive by 24 h, with maximal expression between this time and 72 h (Fig. 2,b). Ox-40 expression was lost after 4 to 5 days, but reinduction was rapid on the effector cells after further Ag stimulation, the majority of cells expressing moderate levels by 4 h. As we have previously shown for several other facets of T cell activation (43, 44), effector cells were more responsive to stimulation compared with naive cells, being induced to express Ox-40 with low numbers of APCs (Fig. 2 c).

To assess the effects of ligating Ox-40, we transfected two of our existing I-E^k-positive fibroblast lines with cDNA for Ox-40L to generate APCs that expressed only Ox-40L, or Ox-40L with B7 (see Materials and Methods and Fig. 1). These APCs and the parent lines were used to stimulate naive CD4 cells with the cognate Ag, PCC. Ox-40L expression had no enhancing effect on several markers of activation such as IL-2R (Fig. 3,a) and CD69 expression (not shown), which we have previously used as a reliable method for determining the number of T cells activated in culture (2). This was irrespective of Ag concentration, APC number, time of analysis, and coexpression of B7 (Fig. 3,a and unpublished observations). Ox-40L expression also had little effect on IL-2 production when assessed in supernatants taken 40 h after naive T cell stimulation (Fig. 3, b and c), a time that is usually optimal for this facet of response (18), and did not augment the low level of proliferation seen in the absence of Ox-40L (Fig. 3,d). This was in contrast to B7-1, which, as seen in previous studies (18, 41), induced two- to threefold more IL-2 than APCs without either molecule (Fig. 3,b), and resulted in greater proliferation at early times (Fig. 3,d, and not shown). The presence of Ox-40L did result in greater levels of IL-2 recovered in supernatants at 64 h when compared with APCs with B7 or without either molecule (Fig. 3,c), but this did not translate into significantly enhanced proliferation at later times (Fig. 3 d).

In striking contrast to the APCs expressing only one accessory molecule, those bearing both B7 and Ox-40L promoted very high levels of IL-2 secretion and allowed greater proliferation at all times (Fig. 3, b–d). B7 expressed with Ox-40L resulted in an APC that could promote similar levels of IL-2 as an APC with B7 alone, but with ninefold fewer cells, and at optimal numbers resulted in IL-2 secretion that could not be achieved with B7 or Ox-40L alone (Fig. 3,b). Significantly, although proliferation at 60 h was enhanced compared with that with B7 expressing APCs, the main effect of the combination of B7 and Ox-40L was proliferation that continued at high levels for 4 days. In the absence of Ox-40L, the proliferative phase was largely over after 2 days (Fig. 3 d).

Although Ox-40 expression was lost several days after naive T cell activation, it was rapidly reexpressed on the effector cells after stimulation (Fig. 2 b), suggesting that this molecule may also regulate responses of Ag-experienced cells, as well as promoting prolonged naive cell responses. To assess a range of cytokine responses, we generated short-term, partially polarized, Th1-like, and Th2-like effector populations over 5 days using standard protocols (see Materials and Methods). Both populations expressed high levels of Ox-40 shortly after stimulation, although the IL-4-primed Th2-like cells did bear somewhat enhanced levels (data not shown). These cells were restimulated with PCC presented on varying numbers of fibroblast APCs as before with naive T cells.

As seen in previous studies with effector cells (44, 45), similar responses were induced with APCs expressing or lacking B7 (Figs. 4 and 5). In contrast, Ox-40L expression resulted in enhanced IL-2, IL-4, and IL-5 production compared with the control APCs lacking this molecule, but had little effect on IFN-γ production. At least twofold more IL-4 and IL-5 were induced regardless of APC number, with fewer Ox-40L-expressing cells being required to produce equivalent levels of these cytokines (Fig. 5). The most striking effect of Ox-40 engagement was on IL-2 secretion, which was produced from the effector cells at far higher levels than with Ag presented with B7, and was induced under limiting APC conditions, in which little or no IL-2 was detected in the absence of Ox-40L (Fig. 4). Interestingly, even the IL-4-primed effectors, which did not produce detectable IL-2 with B7 alone, were induced to secrete some IL-2 with high numbers of Ox-40L-positive APCs (Fig. 5). Since these effectors are not fully polarized at this stage, they may contain both Th1- and Th2-like cells, and therefore the results do not necessarily imply that Ox-40L induces IL-2 production from IL-4- or IL-5-secreting cells. B7 collaborated with Ox-40L with all cytokines, but because of the pronounced response to Ox-40L in the absence of B7, this was not as dramatic as seen with naive T cells. Again, IL-2 was the cytokine that was augmented to the greatest degree with both effector populations (Figs. 4 and 5).

Because of the enhanced secretion of IL-2 and IL-4, cytokines that can be involved in T cell expansion, we analyzed the ability of Ox-40L to promote proliferation of short-term Th1- and Th2-like effector cells. Five days after naive T cell activation, the effectors were restimulated as before with varying numbers of fibroblast APC-presenting PCC peptide, and proliferation was assessed over time (Fig. 6). As seen in previous studies (44), little difference was observed at the effector stage between responses in the presence or absence of B7. In contrast, APCs expressing Ox-40L promoted higher levels of proliferation at low APC:T cell ratios (1:9, 1:3) in both Th1- and Th2-like effector populations, and appreciably more proliferation when equivalent APC numbers were used (Fig. 6, a and b). Strong cooperation was observed between B7 and Ox-40L when fewer APCs were in culture, resulting in extremely high levels of proliferation, but similar responses were seen regardless of the presence of B7 with greater numbers of APCs. The action of Ox-40L was seen most dramatically when assessing proliferation over time. Interestingly, very similar levels of response were observed with all APCs when early proliferation was measured at 42 h (Fig. 6, b and c). In contrast, both Ox-40L-expressing APCs promoted continued proliferation at later times (66 and 90 h), while the APCs that expressed B7 or neither molecule were unable to sustain cell expansion at the level seen at 42 h. Similar response patterns were seen for both Th1- and Th2-like effector populations. Blocking studies with naive and Th1 effector T cell populations showed that proliferation was primarily dependent on IL-2, whereas both IL-2 and IL-4 were responsible for the extended proliferation seen with Th2 effectors (Fig. 7).

In this work, we have demonstrated that Ox-40L represents a costimulatory molecule for CD4 T cells that may be preferentially utilized during the later phases of T cell responses to promote continued proliferation and clonal expansion. Thus, Ox-40 ligation enhanced responses of naive T cells to Ag, and was strongly stimulatory for both cytokine secretion and proliferation of short-term effector cells generated several days after naive T cell activation. Ox-40 signaling on effector T cells was particularly powerful for promoting the secretion of the growth factors IL-2 and IL-4, and resulted in proliferation that was sustained at high levels for several days, a phenomenon not observed in the absence of Ox-40/Ox-40L interaction. Cooperation between Ox-40L and B7 was observed with naive cells as well as with effector cells, and the overall result was to generate cytokine and proliferative responses that persisted over time. The data therefore suggest that Ox-40/Ox-40L interactions represent a critical pathway for generating a productive, potentially nontolerizing, and long-lived, T cell response, and may promote efficient memory cell generation.

We previously showed that Ag presentation to naive T cells in the absence of costimulation led to partial activation, i.e., blastogenesis and expression of several cell surface Ags such as CD69 and IL-2R, but failed to induce CD40L expression, efficient production of IL-2, and strong proliferation (2). This resulted in poor long-term cell survival and hyporesponsiveness, and suggested that induction of CD40L and production of IL-2 represented key areas that were highly regulated during naive T cell activation (2). CD40L expression has been shown to be required for up-regulating B7 expression and costimulatory activity on all APCs (46, 47, 48), and IL-2 is required for growth and cell survival; and both in turn are controlled by the interaction of other accessory molecule-coreceptor pairs (e.g., ICAM-LFA), and at least in part by CD28 signaling (8, 41, 49). Ox-40 was not present early after naive T cell activation, only being expressed on a majority of cells between 24–48 h (Fig. 2). This contrasts with CD28 and LFA-1, which are constitutively expressed on naive T cells, and CD40L, the expression of which after Ag stimulation is seen by 3 h, peaks from 5–8 h, and is generally absent by 15 h (41). Surprisingly, even though Ox-40 was expressed, Ox-40L in the absence of B7 resulted in elevated IL-2 levels only at later times, and was ineffective at inducing proliferation above that induced with the control APCs lacking both molecules (Fig. 3). In combination with B7, Ox-40L promoted greatly elevated levels of IL-2 at both early and late times, enhanced proliferation at early times, and resulted in proliferation that continued at high levels for several days. Thus, Ox-40L appears to be a molecule that can promote naive T cell expansion only after signals through CD28 are received. The interaction between Ox-40 and Ox-40L may therefore represent an additional event during the primary CD4 response that is essential for optimal clonal expansion, acting after the initial interactions between ICAM-LFA, B7-CD28, and CD40-CD40L have taken place.

The effect of B7 in combination with Ox-40L was reminiscent of our earlier data with B7 in combination with ICAM-1 (18), again showing that cooperation/synergy between accessory molecules is required for an optimal naive T cell response. However, several differences exist between the action of ICAM and Ox-40L. Ox-40L coexpressed with B7 did not significantly affect the ability of the fibroblast APCs to activate naive T cells, as assessed by IL-2R induction (Fig. 3 a) and CD69 induction (not shown), with similar dose-response curves being seen regardless of the presence of Ox-40L. In contrast, ICAM-1 coexpressed with B7 results in activation of many more naive T cells under conditions of limiting Ag or APC number, such that the dose-response curve for IL-2R/CD69 is shifted 10–100-fold lower (50). Thus, part of the action of ICAM can be attributed to enhanced numbers of T cells responding, whereas Ox-40L functioned largely by enhancing production of IL-2 in the absence of an increased number of responding cells. In addition, Ox-40L under optimal conditions of stimulation induced sustained IL-2 secretion and proliferation that are also not as apparent with ICAM in combination with B7, again promoting the concept that Ox-40L is a molecule that primarily enhances late phase naive T cell responses.

More striking were the results with effector T cells. As shown previously (44), and in Figs. 4–6, these T cells respond fairly well to Ag presented on APCs lacking molecules such as B7-1 and ICAM-1, and the presence of B7-1 and ICAM-1 expressed separately has little effect on the responses induced. Cytokines such as IFN-γ, IL-4, and IL-5 are induced efficiently, whereas IL-2 is only weakly produced, which translates to an early burst of proliferation over 1 to 2 days that dies quickly (44). Stimulation through Ox-40, particularly under optimal Ag and APC conditions, resulted in enhanced secretion of IL-2, IL-4, and IL-5, but not IFN-γ, and again collaboration with B7 was also observed. The reason that Ox-40L alone was active on effector T cells, but not on naive T cells, is unclear. This action was only apparent with high APC numbers and high dose Ag, with the presence of B7 being required for enhancing responses under conditions of limited stimulation. We have shown previously that effector T cells are more responsive to TCR signals than naive T cells (44, 45), and it is possible that with sufficient signaling through the TCR, effector cells do not require additional signals through CD28 to respond to signals through Ox-40. In contrast, IL-2 secretion from naive T cells is highly dependent on CD28 signals, regardless of the extent of TCR signaling, possibly accounting for the lack of effect of Ox-40L expressed alone.

Although IL-4 and IL-5 were up-regulated efficiently, the major effect of Ox-40L appeared to be on IL-2 production by the effector cells. Enhanced IL-2 was evident with those partially polarized to be Th1 like, and was also observed with IL-4-primed cells, a phenomenon that generally does not occur with these Th2-like populations. Correlating with the extent of growth factor production, proliferation was augmented dramatically at late times with Ox-40 ligation, with high levels being maintained up to 4 days after effector T cell restimulation. This contrasts strikingly with the other fibroblast APCs that we have used (including those expressing ICAM-1; see 44 , which are unable to sustain proliferation of either Th1- or Th2-like effector populations. These data therefore support previous observations that showed that an Ab to Ox-40, in combination with anti-CD3, could promote high levels of IL-2 and strong proliferation from a CD4 clone (29).

Overall, the data therefore suggest that costimulation through Ox-40 is a major pathway that acts after the initial activation events and results in prolonged naive and primary effector cell responses. These results complement the recent observations from Stuber et al. (32, 40) and Ohshima et al. (39), which showed that Ox-40L signaling to both B cells and dendritic cells could synergize with CD40 signals to enhance activity of these APCs. All studies promote the concept that Ox-40/Ox-40L interactions transduce critical signals for downstream T cell-APC responses, and may contribute to the massive expansion that is seen during a primary CD4 response in vivo (1). Attention has recently been focused on the need for an inflammatory event to elicit a productive, nontolerizing, CD4 response, as has been demonstrated many times by priming with protein Ags in the presence or absence of adjuvant. The reasons for this requirement are not clear, but it has been proposed that inflammatory signals may help ensure that either costimulatory molecules such as B7-1 and B7-2 are induced efficiently, or that cytokines such as IL-6 and IL-1 are produced, all of which may contribute to cell expansion and/or survival (51, 52). We suggest that the interaction between Ox-40 and Ox-40L represents a major signaling pathway that occurs after the initial T cell-APC activation events, and will ensure that weak transient responses are not induced, and that tolerance does not result. We additionally propose that Ox-40L and Ox-40 may be targets for adjuvant action, with their efficient expression being necessary to generate sufficient secondary signals to keep the primary T cell response going.

The results presented in this work and elsewhere also substantiate the notion that Ox-40/Ox-40L interactions are pivotal to productive T cell/B cell collaboration. Generally, it is thought that naive T cell activation occurs on dendritic cells, which are present in the T cell zones of spleen and lymph nodes, and that subsequent interactions take place with B cells in the outer T zones and in the follicles (53, 54). Studies with B cell-deficient mice (55, 56, 57) suggest that these later interactions may play a major role in promoting efficient T cell expansion as well as effector function. As Ox-40L is predominantly expressed on activated B cells, and appears to require CD40 signaling for induction, the absence of Ox-40L and signaling through Ox-40 represents one component that may contribute to the weak T cell priming seen when animals are depleted of B cells. As well as the generation of primary T cell responses, an additional aspect that may be highly relevant to the findings in this study is the generation of memory T cell responses. It is largely accepted that efficient primary T cell responses, and especially a high level of expansion of naive cells specific for the particular Ag are necessary for a strong memory response. It may therefore be predicted that the Ox-40/Ox-40L interaction will contribute to the development of memory, and in support of this, recent data have shown that Ox-40L.Fc, as a T cell costimulus, can enhance tumor immunity in a model involving CD4 cells (A. D. Weinberg, unpublished observations).

Although Ox-40L is obviously a potent costimulatory molecule, its mechanism of action is largely unknown. As Ox-40 is a member of the TNFR family, it utilizes several of the TRAF family of molecules (2 and 3) and nuclear factor-κB (58), and thus should produce signals distinct from those generated through CD28. We have demonstrated that a dominant action is in driving cytokine secretion and proliferation (this article and 29 , with clonal expansion being largely dependent on IL-2, with IL-4 also playing a role with Th2-like cells. Additionally, it is possible that Ox-40 signals are antiapoptotic and promote short-term cell survival, although assessment of cell recoveries over an extended period suggests that other signals may be required for long-term survival (unpublished observations). Potential mechanisms of action, other than promoting cytokine transcription, could include up-regulation of members of the bcl family of proteins, inhibition of Fas or FasL expression or function, or overcoming down-regulatory signals through CTLA4, and we are initiating investigations of these aspects of T cell regulation by Ox-40L. Regardless of the mechanisms involved, the studies reported in this work suggest that Ox-40 and Ox-40L may be a major ligand-receptor pair that is not only crucial for induction of efficient APC activity, but is also intimately involved in development of this T cell effector function and in initiating long-lived T cell responses.

We thank Dr. Wayne Godfrey for providing cDNA for Ox-40L.