Isopentenyl Pyrophosphate, a Mycobacterial Non-peptidic Antigen, Triggers Delayed and Highly Sustained Signaling in Human γδ T Lymphocytes without Inducing Down-modulation of T Cell Antigen Receptor

The Vγ9Vδ2 T cell subset, which represents up to 90% of the circulating γδ T cells in humans, was shown to be activated, via the T cell receptor (TcR), by non-peptidic phosphorylated small organic molecules. These phosphoantigens, which are not presented by professional antigen-presenting cells, induce production of high amounts of interferon-γ and tumor necrosis factor (TNF-α). To date, the specific signals triggered by these antigens have not been characterized. Here we analyze proximal and later intracellular signals triggered by isopentenyl pyrophosphate (IPP), a mycobacterial antigen that specifically stimulates Vγ9Vδ2 T cells, and compare these to signals induced by the non-physiological model using an anti-CD3 antibody. During antigenic stimulation we noticed that, except for the proximal p56 lck signal, which is triggered early, the signals appear to be delayed and highly sustained. This delay, which likely accounts for the delay observed in TNF-α production, is discussed in terms of the ability of the antigen to cross-link and recruit transducing molecules mostly anchored to lipid rafts. Moreover, we demonstrate that, in contrast to anti-CD3 antibody, IPP does not induce down-modulation of the TcR·CD3 complex, which likely results in the highly sustained signaling and release of high levels of TNF-α.

T cells expressing the ␥␦ T cell receptor (TcR) 1 represent in humans a relatively low T lymphocyte population and, partic-ularly in peripheral blood, these cells account for only 1-5% of the circulating T cells (see Ref. 1 for review). In an adult the majority of these circulating T cells are classified as V␥9V␦2 subset (up to 90%). It has been shown that this ␥␦ T cell subset dramatically increases during infection by intracellular pathogens of bacterial, viral, or parasitic origin (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12). One of the particularity of the V␥9V␦2 T cells is to be activated by components identified as non-peptidic, phosphorylated small organic molecules (13)(14)(15)(16)(17)(18). Some of these components have been purified from bacteria or parasites and are thought to be responsible for the in vivo expansion of V␥9V␦2 T cells during the acute phase of the infection process. There is so far no formal proof that these small molecules do bind to the V␥9V␦2 T cell receptor, however, transfer experiments of the V␥9V␦2 TcR in TcR-negative Jurkat cell mutants have provided strong evidence to suggest that the recognition of the phosphoantigens is mediated by TcR (17).
Stimulation of V␥9V␦2 T cells by phosphoantigens results mainly in the production of high amounts of interferon-␥ (19 -22) and tumor necrosis factor-␣ (TNF-␣) (19,23). However, to date, the specific signals that are triggered in V␥9V␦2 T cells upon stimulation with phosphoantigens have not been characterized. This aspect is of great importance for the possible pharmacological control of TNF-␣ release (24,25) by these cells, because an overproduction of this cytokine could result in immunopathology (26).
In ␣␤ T cells, it is well established that activation occurs as a result of multimolecular interactions between T cells and antigen-presenting cells (27)(28)(29). These interactions include the recognition of the peptide/major histocompatibility (MHC) complex by the T cell receptor and the binding of CD4 coreceptor to non-polymorphic regions on the MHC class II molecules. The cytoplasmic tail of CD4 associates with the Src family tyrosine kinase p56 lck , which plays a key role in the early events of T cell activation. However, in the case of V␥9V␦2 T cell activation, the non-peptidic antigens do not need to be presented in the context of MHC molecules (30,31). Moreover, it has been shown that V␥9V␦2 T cells do not express CD4 and poorly express CD8 (32,33), which normally interact with MHC class II or class I molecules, respectively. Therefore, there is not, as with ␣␤ T cells, recruitment of these coreceptors, which stabilize TcR⅐ligand interaction and are essential for the formation of the "immunological synapse," which determines the extent and qualitative nature of the transduced signal (27,34,35).
Recently we studied TNF-␣ release by V␥9V␦2 T cells when stimulated either with a monoclonal antibody directed against the CD3 complex or with the mycobacterial phosphoantigen isopentenyl pyrophosphate (IPP) (36). We demonstrated that TNF-␣ production does not involve, as is the case in ␣␤ T cells, CD28 costimulation. Moreover, we noticed that the cytokine production in V␥9V␦2 T cells was highly delayed (ϳ10-h difference) when the cells were activated by a physiological phosphoantigen ligand (IPP) instead of anti-CD3 monoclonal antibody (mAb). This delayed cytokine production could be the result of a delayed triggered signaling or of the recruitment of different signaling molecules according to the stimulating agent used. In the present paper, we therefore studied the signals triggered in V␥9V␦2 T cells upon stimulation with isopentenyl pyrophosphate, a physiological non-peptidic mycobacterial phosphoantigen known to be specifically mitogenic for V␥9V␦2 T cells (14,15), and compared these to the signals induced by anti-CD3 mAb. We show that the kinetics of the signals triggered upon IPP stimulation is quite different from that of the signals induced upon anti-CD3 mAb stimulation; the signals are largely delayed when the cells are stimulated with the nonpeptidic antigen compared with those induced upon anti-CD3 mAb activation except for p56 lck . But this delay cannot be assigned to the synthesis of de novo proteins. Moreover, we show that the majority of the phosphoantigen-induced signals, in contrast to the anti-CD3-triggered ones, are highly sustained and last for several hours. This long-lasting cell signaling observed with IPP stimulation is possibly related to the lack of induction of TcR⅐CD3 down-modulation that we demonstrate herein.
Preparation of Supernatants for Measurement of TNF-␣ Production-␥␦ T cells (2 ϫ 10 6 cells/ml) were cultured in 24-well tissue culture plates in RPMI 1640 supplemented with 5% FCS ϩ 5% human AB serum in a total volume of 0.5 ml per well. When mentioned cells were pretreated with inhibitors (LY 294002, 5 M) for 30 min at 37°C, then stimulated with IPP (50 M) or UCHT1 (2 g). At different times, supernatants were harvested and assayed for TNF-␣ using a human TNF-␣ ELISA kit (OptEIA set: human TNF-␣, PharMingen, San Diego, CA) according to the manufacturer's instructions.
In Vitro Kinase Assay-For p56 lck kinase assay, complexes were resuspended in 50 l of specific kinase buffer (50 mM Pipes, pH 7.5, 10 mM MgCl 2 , 10 mM MnCl 2 , 1 mM PMSF, 100 M NA 3 VO 4 ), and autophosphorylation of p56 lck was determined in the presence of 5 Ci of [␥-32 P]ATP (6000 Ci/mmol, PerkinElmer Life Sciences) and incubated for 10 min at 37°C. p56 lck activity was measured by phosphorylation of the exogenous substrate enolase. Complexes were incubated in 50 l of kinase assay buffer in the presence of 10 g of acid-denatured enolase, 10 Ci of [␥-32 P]ATP (6000 Ci/mmol, PerkinElmer Life Sciences), and 4.5 M unlabeled ATP. For ZAP-70 kinase assay, complexes were incubated in 25 l of kinase buffer (100 mM Tris, pH 7.5, 125 mM MnCl 2 , 25 mM MnCl 2 , 2 mM EGTA, 250 M Na 3 VO 4 , 2 mM dithiothreitol) in the presence of 4 g of recombinant LAT protein, 10 Ci of [␥-32 P]ATP (6000 Ci/mmol, PerkinElmer Life Sciences), and 10 M unlabeled ATP for 10 min at 37°C. The reactions were stopped by addition of 2-mercaptoethanol-containing sample buffer and boiling. Radiolabeled proteins were then resolved on 10% SDS-PAGE, transferred to polyvinylidene difluoride membranes (Millipore), and then detected by autoradiography. Quantification of the phosphorylated bands reported in the results has been performed using a PhosphorImager Storm system driven by ImageQuaNT software (Molecular Dynamics, Sunnyvale, CA).
Flow Cytometry-Cells were stimulated by UCHT1 or IPP at different times, fixed in 1% paraformaldehyde for 15 min, washed in phosphate-buffered saline, and then stained with 1 g of fluorescein isothiocyanate (FITC)-labeled anti-TcR V␥9 mAb in phosphate-buffered saline supplemented with 2% fetal calf serum, 0.02% NaN 3 , on ice in a total volume of 50 l. After 30 min, the cells were washed once, fixed in 1% paraformaldehyde, and analyzed by flow cytometry on a FACSCalibur (Becton Dickinson) with Cell Quest software.

Production of TNF-␣ by Anti-CD3 mAb-or IPP-stimulated V␥9V␦2 T Cells-
We previously established that high amounts of TNF-␣ are produced by V␥9V␦2 T cells when stimulated with either anti-CD3 mAb or with the physiological antigen IPP (36). As shown in Fig. 1, upon anti-CD3 mAb stimulation, TNF-␣ is produced very early with its maximum reached after 3 h, whereas maximum production induced by IPP only occurs after 16 h. This delay between the two stimulation processes could reflect either a recruitment of different signals or a difference in the kinetics of the triggered signals. To test these hypotheses, we analyzed the kinetics of the extracellular regulated kinase (ERK) and p38, two mitogen-activated protein kinase (MAPK) pathways directly involved in TNF-␣ production by V␥9V␦2 T cells stimulated either with anti-CD3 mAb or with IPP (36).
Study of ERK Activation-Purified V␥9V␦2 T cells were stimulated with either anti-CD3 mAb or with an optimal dose of IPP over a broad time range, and activation of ERK1/ERK2 was studied. As shown in Fig. 2A, phosphorylation of ERK1/ ERK2 upon anti-CD3 mAb activation is very rapid (maximum reached at 5-min stimulation) and decreases but lasts at a high degree for around 30 min. After 30-min activation, the intensity is reduced to a very low phosphorylation level even though it still remains detectable after 2 h. In contrast, when the cells are stimulated with IPP, the phosphorylation/activation of ERK1/ERK2 begins to be faintly detectable after 1 h and reaches a plateau after 3 h, which lasts for at least one more hour. After 6 h, even though the phosphorylation signal begins to decrease, it still remains high.
Previous studies have shown that IPP induces cell proliferation and cytokine release in ␥9␦2 T cells but not in other subsets of ␥␦ T cells (13-15, 17, 23). However, we could not rule out the possibility that, even if IPP was not able to trigger biological responses in ␥␦ T cells, which do not express the ␥9␦2 TcR complex, it could trigger intracellular signals. To investigate this, we studied ERK2 activity in ␥1␦1 T cells (which are another important subset of ␥␦ T cells in human blood). As shown in Fig. 2B, anti-CD3 stimulation induces a strong and rapid ERK2 activation in ␥1␦1 T cells; however, IPP is not able to trigger any ERK2 activation in these cells following either short or prolonged stimulation.
We also wondered if the observed delay in IPP-induced ERK2 activation could be assigned to a necessity to synthesize de novo proteins. As shown in Fig. 2C, pretreatment with cycloheximide, a protein synthesis inhibitor, does not modify the activation of ERK2 induced by IPP. In addition we confirmed that cycloheximide efficiently blocks protein synthesis at the concentration used in these experiments (10 g/ml; data not shown).
Study of p38 MAPK Activation-We similarly studied activation of p38 kinase in V␥9V␦2 T cells that have been stimulated either with anti-CD3 mAb or with IPP. Fig. 3A shows that, as for ERK activation, p38 MAPK phosphorylation appears to be delayed in IPP stimulation compared with anti-CD3 stimulation. The kinetics are very similar to that of ERK1/ ERK2 with the maximum activation in IPP stimulation occurring 2 h after triggering and lasting as a plateau for at least 4 more hours. In anti-CD3 mAb stimulation, the maximum is already reached within 5 min and then decreases, but the activated form remains elevated for at least 2 h. As we have shown for ERK2 MAPK, IPP does not trigger p38 activation in ␥1␦1 T cells (Fig. 3B) and cycloheximide pretreatment does not modify IPP-induced p38 activation in ␥9␦2 T cells (Fig. 3C).
Study of p56 lck Activation-We questioned whether the delay observed for ERK and p38 kinase activation, which are later signals in the transduction cascade, could be due to a difference in the triggering of a signal directly related to TcR⅐CD3 ligation. For that purpose, we studied activation of the Src family kinase p56 lck , which represents one of the earliest events in ␣␤ T cell stimulation. Fig. 4A shows that the immunoprecipitated p56 lck , in the presence of [␥-32 P]ATP, displays a 60-kDa-shifted band in anti-CD3 stimulation, which corresponds to the conversion of p56 lck to an Lck form phosphorylated on Ser-59 (37,38). This hyperphosphorylation was already observed in ␣␤ T cells stimulated with anti-CD3 mAb or phorbol 12-myristate 13-acetate (39) and was shown not to be concomitant with an increase of the kinase activity but rather to be accompanied by a decrease in the kinase activity (40 -43). Similarly, in our experiments, we could not detect, through phosphorylation of enolase used as exogenous substrate, any activity of p56 lck in anti-CD3 stimulated ␥␦ T cells.
In IPP stimulation (Fig. 4B) conversion of p56 lck to a slower migrating form also exists but occurred along with an increased intensity of the p56 band (as a control we checked that the amounts of immunoprecipitated p56 lck loaded on the gel were similar in each sample; data not shown). This increased inten- sity of the p56 lck band reflects autophosphorylation of the kinase and its activation. Indeed, kinase activity was detectable through phosphorylation of enolase. It has to be noted that activation of p56 lck in IPP-stimulated cells is rapid (detectable at 5 min) and peaked at 30 -45 min. Therefore, because p56 lck activation in IPP stimulation is high and rapid, it can hardly be accountable for the delay observed in the MAPK late signals.
Study of ZAP-70 Kinase Activity-It is generally accepted that immunoreceptor tyrosine-based activation motifs (ITAM) of the signal-transducing subunits of CD3, as well as the chain, are phosphorylated by Lck, making them competent to associate with zeta-associated protein (ZAP)-70 (reviewed in Ref. 28). Once recruited, ZAP-70 is activated through its phosphorylation by Lck and then is able to recruit and phosphorylate its own substrates, SLP76 and a linker for activation of T cells (LAT) (28,29,44). Because we observed a rapid activation of p56 lck upon IPP stimulation, we questioned whether the chain and ZAP-70 are also rapidly activated. First, we immunoprecipitated the chain proteins and studied their phosphorylation in samples from unstimulated or stimulated cells. We were unable to detect by Western blot, using an anti-phosphotyrosine Ab, phosphorylation of this protein in either unstimulated or stimulated samples (data not shown). This is probably due to the low rate of expression and phosphorylation of this protein in non-transformed cells such as ␥9␦2 T cells. However, we showed that in the immunocomplex, ZAP-70, is coprecipitated with the chain and the amount of coprecipitated ZAP-70 is higher in the immunocomplex from anti-CD3-or IPP-stimulated cells than from unstimulated-cells. This indicates that the chain may also be more phosphorylated in these samples (Fig. 5A). Following stimulation with IPP, the maximum amount of coprecipitated ZAP-70 protein is observed after 30min IPP stimulation. Moreover, we studied ZAP-70-induced phosphorylation using recombinant LAT as an exogenous substrate. As shown in Fig. 5B, recombinant LAT is phosphorylated by immunoprecipitated (IP)-ZAP-70 from cells stimulated by IPP. However, as for ERK and p38, phosphorylation of LAT is largely delayed compared to activation of p56 lck , indeed phosphorylation is detectable only with IP-ZAP-70 from 30-minactivated cells and lasts as a plateau with IP-ZAP-70 from at least 2-h-activated cells. As a control, LAT appears to be highly phosphorylated with IP-ZAP-70 from cells activated for 5 min with anti-CD3 mAb. The delayed phosphorylation of recombinant LAT thus demonstrates that ZAP-70 is activated tardily in IPP stimulation. Study of PKB Phosphorylation-Several papers have shown that TNF-␣ production in several cell types, including T cells, is dependent on phosphoinositide 3-kinase (PI3K) activation (45,46). Moreover, in ␣␤ T cells, it was demonstrated that TcR engagement results in rapid phosphorylation of Tyr-685 in the p85 subunit of PI3K (47), and that this phosphorylation and the consequent activation of PI3K have been attributed to Lck. We therefore studied whether, in V␥9V␦2 T cells, TNF-␣ is also dependent on PI3K activation. As shown in Fig. 6A, TNF-␣ production induced with either IPP or anti-CD3 mAb is inhib-

FIG. 3. Kinetics of p38 MAPK activation in human peripheral blood derived ␥␦ T cells. ␥9␦2 T cells (A) or ␥1␦1 T cells (B)
were stimulated for the indicated times by UCHT1 (10 g/ml) or by IPP (100 M). When indicated, ␥9␦2 T cells were pretreated 30 min with cycloheximide (10 g/ml) before performing stimulation (C). Total cellular proteins were separated on 10% SDS-PAGE and revealed by Western blot analysis using an anti-phospho-p38 MAPK Ab (which specifically reveals the phosphorylated and active form of p38) and reprobed with an anti-p38 MAPK Ab after Ab stripping. This experiment is representative of three experiments.

FIG. 4. Autophosphorylation and kinase activity of p56 lck in human peripheral blood derived ␥␦ T cells.
Human peripheral blood-derived ␥␦ T cells were stimulated for the indicated times by UCHT1 (10 g/ml) or by IPP (100 M). Immunoprecipitations were performed using a total lysates from 2 ϫ 10 7 cells with anti-Lck Ab. Activation of p56 lck was estimated by its autophosphorylation (A) and by phosphorylation of enolase used as exogenous substrate (B); these experiments were carried out in the presence of [␥-32 P]ATP as described under "Experimental Procedures." The amount of p56 lck in each lane was evaluated by Western blot using mouse anti-p56 lck Ab (data not shown). The phosphorylation of enolase was quantified by PhosphorImager analysis. This experiment was repeated twice. ited by LY294002, an inhibitor of PI3K, suggesting that TNF-␣ release is dependent on activation of this kinase. We also studied activation of PI3K through phosphorylation of protein kinase B (PKB), one of its secondary substrates (48,49), upon stimulation with anti-CD3 mAb or IPP. Even though activation of PI3K is directly dependent on Lck activation, its response is largely delayed in IPP activation (maximum after 2-h stimulation) compared to anti-CD3 mAb activation (maximum after 5 min stimulation) (Fig. 6B).
Study of TcR⅐CD3 Down-modulation upon IPP or Anti-CD3 mAb Stimulation-One of the most striking characteristics of the signals triggered by IPP compared with those induced by anti-CD3 mAb, aside from the fact that they are delayed, is that they last for a long period of time, i.e. several hours. Receptor internalization following ligand binding is generally considered to be an important mechanism that limits both the quantity of signals received by the cell and the duration of the triggered signals (50). Concerning TcR, several papers have shown that sustained signaling results from prolonged T cell receptor occupancy (51)(52)(53). We therefore questioned whether the long-lasting signaling triggered in IPP stimulation compared with that induced during anti-CD3 stimulation could be parallel to a difference in the rate of TcR down-modulation. We therefore stimulated ␥␦ T cells with either anti-CD3 mAb or with IPP. The cells were paraformaldehyde-fixed at different times after stimulation and analyzed for TcR expression using FITC-conjugated anti-V␥9 mAb. As shown in Fig. 7, TcR is down-modulated in a time-dependent manner upon anti-CD3 stimulation whereas, with IPP, it remains unmodified. In parallel, as a control of stimulation efficacy, we analyzed ERK phosphorylation, which showed the same kinetics as that presented above (data not shown).

DISCUSSION
The present paper studies signals triggered in V␥9V␦2 T cells by IPP, a physiological antigen specific for this T cell subpopulation. It has first to be noted that, in contrast to in vitro studies of ␣␤ T cell activation by physiological antigen (51), those of V␥9V␦2 T cell stimulation do not require that the antigen be presented by antigen-presenting cells. The IPPinduced signals were compared with those triggered by the non-physiological model, using an anti-CD3 mAb. It appears that signals triggered by IPP leading to TNF-␣ release were delayed compared with those induced by anti-CD3 mAb, and this delay cannot be assigned to the synthesis of de novo proteins as shown in the experiments in the presence of cycloheximide. In contrast, the Src family kinase p56 lck , the activation of which represents one of the earliest events in T cell stimulation (28,54), appears to be triggered very early in IPPstimulated ␥9␦2 T cells. Moreover, its enzyme activity can be detected through its autophosphorylation and by phosphorylation of enolase used as exogenous substrate. In ␣␤ T cells, p56 lck enzyme activation was shown to occur upon TcR⅐CD3 ligation, but this was demonstrated in cell lines (mostly Jurkat cells) (41,55). In primary ␣␤ T cells, engagement of the TcR⅐CD3 complex by anti-CD3 mAb leads to hyperphosphorylation (on Ser-59) of p56 lck , observed in SDS-PAGE as a slower migrating band (60 kDa) (37,38), with no increase but even a decrease in kinase activity (40 -43). Autophosphorylation activation of the kinase in primary ␣␤ T cells is, however, detectable when co-receptors CD4 or CD8 are engaged by interacting components such as anti-CD4 mAb or human immunodeficiency virus external glycoprotein gp120 in CD4 ϩ cells (56). In this case, p56 lck activation is detectable early (5 min after activation), as is the case in IPP stimulation of ␥9␦2 T cells. In V␥9V␦2 T cells, similarly to what happens in ␣␤ T cells, anti-CD3 stimulation leads to the appearance of a 60-kDa band, but there is no visible increase in enzymatic activity. Signal triggering in anti-CD3 stimulation, leading to activation of downstream signaling pathways, can involve, as has been postulated in ␣␤ T cells, another Src family kinase, p59 fyn , which has been shown to be directly associated with the T cell receptor complex (57,58). The fact that V␥9V␦2 T cells do not express CD4 and express CD8 poorly (the cells we used in the present study were CD4 Ϫ ,CD8 Ϫ ; data not shown) and that IPP stimulation leads to p56 lck activation could suggest that this antigen, in addition to engaging the TcR⅐CD3 complex, could also engage another cell surface molecule as a co-receptor. Such an hypothesis that responsiveness of V␥9V␦2 T cells is modulated by the expression of a (unknown) molecule with a co-receptor-like function is similar to that described for CD4 and CD8 co-receptors in ␣␤ T cells recently put forward by Bü rk and co-workers (59). p56 lck has been described in ␣␤ T cells to phosphorylate ITAM on the CD3 ⑀ chain and TcR (60) chain, rendering them competent for recruitment of the Syk family kinase ZAP-70. Human peripheral blood-derived ␥␦ T cells (5 ϫ 10 7 cells) were stimulated for the indicated times by UCHT1 (10 g/ml) or by IPP (100 M). A, after cell lysis, chain was immunoprecipitated from the clarified supernatants with an anti-chain Ab, and the amount of co-precipitated ZAP-70 was evaluated by Western blot using an anti-ZAP Ab and reprobed with an anti-chain Ab after Ab stripping. B, after cell lysis, ZAP-70 was immunoprecipitated from the clarified supernatants with an anti-ZAP Ab. The amount of ZAP-70 was evaluated on Western blot using an anti-ZAP-70 Ab. Activation of ZAP-70 was estimated by phosphorylation of recombinant LAT proteins used as exogenous substrate; these experiments were carried out in the presence of [␥-32 P]ATP as described under "Experimental Procedures." The phosphorylation of LAT was quantified by PhosphorImager analysis. This experiment was repeated twice. Subsequent phosphorylation activation of this kinase triggers the downstream signaling pathways regulating the transcription of genes essential for cytokine production. In IPP stimulation, the events that are directly dependent on p56 lck activation, i.e. recruitment and activation of ZAP-70 and PI3K as well as later signals like ERK and p38 kinase, appeared delayed in reference to the kinetics of p56 lck activation, or in comparison to that induced in anti-CD3 stimulation. This delay is likely not to be attributable to a slow rate in IPP⅐TcR interaction, because p56 lck is activated early upon IPP stimulation. On the other hand, it is now generally accepted that in T cells, microdomains of the plasma membrane, commonly referred to as lipid rafts, play an important role in TcR signaling (61). They are mostly involved through the recruitment of transducing molecules like Lck and LAT, which are anchored to them. Aggregation of lipid raft-associated proteins and TcR⅐CD3 complex can be induced in T cell stimulation through cross-linking with anti-CD3 mAb, thus taking part in the formation of the immunological synapse (35). It has recently been shown in an elegant study that TcR is naturally associated with lipid rafts even though this association is sensitive to nonionic detergent (61). However, aggregation of the TcR by anti-CD3 cross-linking causes aggregation of raft-associated proteins, which leads to triggering of tyrosine phosphorylation of chain, ZAP-70 activation, and downstream signal transduction. It thus appears that triggering of the signaling cascade must occur when the TcR⅐CD3 proximal sig-naling molecules are brought into close contact through crosslinking. A possibility therefore exists that the small non-peptidic molecule IPP, which likely does not have multivalent TcR-interacting sites and is not presented to TcR by MHC or MHC-like molecules, is probably not able to rapidly and efficiently cross-link with the TcR⅐CD3 complex and the lipid rafts. As a possible consequence, the rate for the physical recruitment of a sufficient number of engaged TcR⅐CD3 complexes to colocalize with the rafts anchored transducing molecules is slower than that occurring in anti-CD3 stimulation, leading to a delay in cell signaling (ZAP-70, PI3K, MAPKs) and TNF-␣ release. Of course, the results we present herein are in vitro results obtained with purified V␥9V␦2 T cells, and it cannot be totally ruled out that in vivo the antigens could be presented by other cell types through cell surface molecules not yet determined. According to such an hypothesis, the kinetics of the triggered signals could be faster. To investigate this point, we studied TNF-␣ production by V␥9V␦2 T cells stimulated by IPP in the presence of syngeneic paraformaldehyde-fixed PBMC. This experiment was done to allow IPP to bind to putative cell surface molecules involved in its possible presentation to ␥␦ TcR. We did not notice any difference either in the kinetics or in the amounts of TNF-␣ produced upon stimulation with IPP alone or with IPP in the presence of fixed PBMC (data not shown). Moreover, we cannot totally rule out that when non-peptidic antigens are expressed on the surface of pathogens, they do not FIG. 6. Effect of LY294002 inhibitor on TNF-␣ production and kinetics of PKB activation in Human peripheral blood derived ␥␦ T cells. A, human peripheral blood-derived ␥␦ T cells were preincubated or not 30 min with LY 294002 inhibitor (5 M) and then stimulated with UCHT1 (2 g/ml) for 3 h or with IPP (50 M) for 16 h. TNF-␣ production was then measured in the culture supernatants using an ELISA kit. This experiment is representative out of four. B, human peripheral blood-derived ␥␦ T cells were stimulated for the indicated times by UCHT1 (10 g/ml) or IPP (100 M). Total cellular proteins were separated on 10% SDS-PAGE and revealed by Western blot analysis using an anti-phospho-(Ser-473) PKB Ab (which specifically reveals the phosphorylated form of PKB) and reprobed with an anti-PKB Ab after Ab striping. This experiment is representative of three experiments. behave as a monovalent antigen and thus could engage several TcR⅐CD3 complexes together, in this case the kinetics of the triggered signals could be faster. To test this hypothesis, we compared the kinetics of TNF-␣ production by V␥9V␦2 T cells induced by a non-peptidic antigen such as IPP and by a whole pathogen. We chose to use as a pathogen a strain of Brucella, which we have shown produces a non-peptidic antigen that is able to stimulate ␥9␦2 T cells (62). TNF-␣ production that we measured was lower in supernatants from cells stimulated with gentamicin-killed bacteria than from those stimulated with IPP (probably due to the lower concentration of nonpeptidic antigen present on the surface of the bacteria compared with the IPP concentration that we used), but the kinetics of TNF-␣ production was identical (data not shown).
One of the striking features concerning IPP-induced signals, is that they are highly sustained compared with those induced by anti-CD3 mAb or with those described in ␣␤ T cells stimulated either with anti-CD3 mAb or physiological antigens presented in the context of MHC molecules. It has been shown that sustained signaling can be related to TcR occupancy (51). Here we demonstrated that, in contrast to anti-CD3 mAb, IPP does not induce down-modulation of the TcR⅐CD3 complex. Therefore, a possibility exists that the sustained signals observed in V␥9V␦2 T cells stimulated by IPP results from a long lasting interaction between the antigen and the T cell antigen receptor. This possibility could account for the high and durable production of TNF-␣ detected in activated V␥9V␦2 T cells and which has been shown in several cases to result in immunopathology (26).