Early Triggering of Exclusive IFN-γ Responses of Human Vγ9Vδ2 T Cells by TLR-Activated Myeloid and Plasmacytoid Dendritic Cells¹

Sensing of pathogen-associated molecular patterns by the immune system is mediated by an heterogenous set of innate receptors, that includes TLR, Nod-like receptors, and various C-type lectins. TLR are certainly among the best studied pathogen recognition receptors. Up to 12 TLR members have been characterized so far, which are specific for conserved ligands expressed by bacteria (e.g., LPS, flagellin, and CpG), fungi (e.g., peptidoglycans) and viruses (e.g., ssRNA). TLR are broadly expressed by hemopoietic cells of both the myelomonocytic (dendritic cells (DC),³ macrophages, polymorphonuclears) and lymphoid (B, T, and NK) lineages (1, 2). Upon TLR engagement, DC undergo a functional maturation process, referred to as DC activation, that is required for further induction of adaptive immune responses (3, 4). This process is associated with decreased phagocytic function, and up-regulation of surface costimulatory receptors and MHC class I and II molecules. TLR ligands can also trigger/enhance effector functions of other innate players (such as macrophages, neutrophils and NK cells), and can synergize with B and TCR signals to enhance Ab production by B cells and cytotoxic/cytokinic responses by T cells. TLR ligands can act indirectly on T cells, i.e., through activation of TLR-expressing third-party cells, like DCs and macrophages. This indirect effect has been well documented on so-called invariant NKT (iNKT) cells, a philogenetically conserved innate-like T cell subset specific for endogenous or microbial glycolipid Ag presented by CD1d isoforms (5). Indeed, LPS-induced triggering of IFN-γ responses by iNKT cells results from enhanced IL-12 or type I IFN production by DC upon engagement of TLR4 (6). This activation is further enhanced by TLR-induced up-regulation of endogenous iNKT ligands, as suggested by recent studies (7, 8). TLR ligands may also act directly on T cells through engagement of some TLR, such as TLR2, TLR3, TLR4, and TLR5, whose expression is up-regulated on memory T cells. Such a direct effect has been reported in various in vitro models involving conventional (MHC-restricted) αβ T cells as well as non-MHC-restricted αβ and γδ T cells (9, 10, 11, 12). However, the physiological significance of the latter observations remains debated because TLR expression levels on memory T cells is much lower than on B or myelomonocytic cells. Moreover, TLR engagement on T cells enhances a restricted set of functions that require concomitant TCR engagement.

Full DC activation involves not only signaling by pathogen-recognition receptors like TLR, but also endogenous signals generated in particular upon interaction between CD40 expressed by DC and CD40L expressed by activated T cells. Additional tissue-derived signals (e.g., pro- or anti-inflammatory cytokines) induce DC acquisition of specific functional properties required for appropriate polarization of adaptive immune responses, i.e., acquisition of T cell functional properties that are the best suited to eliminate or control the eliciting agent (3, 13, 14). Owing to their broad reactivity against conserved stress-induced ligands, to their high frequency and their preactivated/memory phenotype in most preimmune individuals, both NK and iNKT cells are thought to play a key role in bridging innate and adaptive immune responses, through their ability to provide the above endogenous DC activation/polarization stimuli during primary immune responses (15). Self recognition of immature DC by both NK and iNKT cells, which is presumably enhanced upon TLR engagement, up-regulates CD40L expression on these lymphoid subsets. This process also triggers early production of inflammatory cytokines, such as TNF-α and IFN-γ, which can synergize with CD40/CD40L signaling to induce DC activation and acquisition of Th1 polarizing properties (16, 17, 18). These mechanisms presumably underlie the strong adjuvant properties of iNKT agonists and their synergistic effects with TLR agonists, reported in various in vivo murine models (5, 19, 20). Like their murine counterparts, human iNKT cells can provide potent DC-maturing stimuli in vitro. However, the actual in vivo impact of such a iNKT/DC crosstalk remains debated, owing to the much lower frequency of iNKT cells in humans than in rodents. Other innate-like lymphoid subsets, such as γδ T cells, might also contribute to early DC activation and possibly compensate for the relative defect of iNKT cells in humans (21, 22, 23, 24, 25). This assumption is supported by several studies performed with Vγ9Vδ2 T cells, a major peripheral γδ subset in most adults that recognize isoprenoid pathway metabolites, referred to as phosphoAg (26). Indeed, Vγ9Vδ2 T cells can promote efficient DC activation into IL-12-producing cells in the presence of synthetic phosphoAg or along infection by phosphoAg-producing bacteria (27, 28, 29). However, it is still unclear whether, like iNKT cells, Vγ9Vδ2 T cells can induce full DC activation upon TLR ligand stimulation in the absence of synthetic or microbial TCR stimuli.

In this study, we studied the effect of a large set of TLR ligands on Vγ9Vδ2 T cell activation. Both TLR3 and TLR4 ligands induced strong IFN-γ production by Vγ9Vδ2 T cells. Vγ9Vδ2 but not iNKT PBL were the main IFN-γ-producing T cells in response to LPS- and poly(I:C) stimulation. This phenomenon primarily depended on type I IFN produced by DC upon TLR engagement. Moreover TLR-induced Vγ9Vδ2 IFN-γ response led to efficient DC polarization into IL-12p70-producing cells. The implication of these results, in terms of the possible adjuvant role played by Vγ9Vδ2 T cells along both viral and microbial infections, are discussed.

The following Abs were used: IMMU389-FITC (TCR Vδ2), UCHT1-FITC (CD3), BMA031-PE (TCR pan αβ), IMMU510-PE (TCR pan γδ), 3G8-PE (CD16), and N901-PE (CD56), purchased from Immunotech/Beckman Coulter. For intracellular staining of cytokines, PE- or allophycocyanin-conjugated mAbs specific for human TNF-α, IFN-γ, GM-CSF, and IL-4 were obtained from BD Biosciences. IL-12p70, IFN-γ pairs of mAbs designed for ELISA were purchased from R&D Systems and eBioscience. Detection of IFN-α and IFN-β was performed using ELISA kits purchased from, respectively, Bender MedSystems and ThermoFisher Scientific. Recombinant human GM-CSF, IL-4, TNF-α, IFN-α2a, and IFN-β were obtained from AbCys, CellGenix, or PBL IFN. CD107a-FITC (clone no. H4A3) and CD107b-FITC (clone no. H4B4) were obtained from BD Biosciences. Universal type I IFN was obtained from R&D Systems. Blockades of cytokine activities were performed by using a neutralizing Ab specific for CD118 (IFN α/β receptor chain 2, clone no. MMHAR-2) obtained from PBL IFN. Synthetic BrHPP (bromohydrin pyrophosphate/Phosphostim) was provided by Innate Pharma. Pamidronate, supplied as disodium salts, was obtained from Calbiochem. Ultrapure LPS E.K12, poly (I:C), PGN K12, Pam3CSK4, CpG ODN2216, and polymyxin A were purchased from Cayla-InvivoGen. Flagellin and R848 were provided by, respectively, Dr. P. Jeannin (INSERM U892, Angers, France) and Dr. F. Trottein (Institut Pasteur, Lille, France). Mevastatin, brefeldin A, monensin, PMA, and ionomycin were purchased from Sigma-Aldrich. CFSE was obtained from Molecular Probes/Invitrogen. The yellow fever (YF) 17D (YF-17D) virus strain, a live attenuated vaccine (30), was provided by Dr. G. Barba-Spaeth (Institut Pasteur, Paris). Virus stocks (2 × 10⁸ PFU/ml), produced in supernatants of SW13 cells, were used to infect either PBMC or purified DC. Flow cytometry acquisitions were performed by using LSR or FACSCalibur systems (BD Biosciences). Analysis were performed by using CellQuest Pro (BD Biosciences) or FlowJo (Tree Star) softwares.

Human monocyte-derived DC were generated according to two experimental procedures. Monocytes were isolated from peripheral blood samples of healthy donors (Etablissement Français du Sang, Nantes, France). PBMC were collected after Ficoll Hypaque (Eurobio) density centrifugation, resuspended at 7 × 10⁶ cells/ml and allowed to adhere in culture flasks for 90 min at 37°C in RPMI 1640 medium supplemented with 2 mM l-Glutamine, 100 μg/ml streptomycin, and 100 IU/ml penicillin (hereafter referred to as complete medium) supplemented with 5% FCS. Nonadherent cells were removed by gentle pipetting and adherent cells were rinsed five times with PBS. Alternatively, monocytes were enriched by elutriation. Monocytes were further differentiated in immature DC (iDC) for 5 days in the same medium supplemented with IL-4 (1000 IU/ml) and GM-CSF (1000 IU/ml). DC phenotype was routinely checked for cell surface expression of CD80, CD83, CD86, and MHC class I/II markers. Treatments by aminobisphophonates (ABP) or activation by TLR ligands were performed by incubating DC for 2 h in culture medium containing indicated concentrations. DC were extensively washed before use. No significant induction of maturation was detected following ABP treatment. Highly pure plasmacytoid DC (pDC) fractions (>90%) were obtained from ex vivo PBMC by using a BDCA-4 Microbead kit (Miltenyi Biotech) according to the manufacturer’s instruction. Purity of enriched pDC (routinely >90%) was checked by cytometry using CD123 and BDCA-2-specific mAbs (Miltenyi Biotech).

γδ T cell lines and clones were cultivated in complete RPMI 1640 medium supplemented with 300 IU/ml recombinant human IL-2 (rhIL-2; Proleukin; Chiron Therapeutics) and 10% FCS. Vγ9Vδ2 T cell clones (clones no. GR72 and no.GR4) were expanded in vitro in the above described culture medium containing 10% human serum and supplemented with 0.5 μg/ml purified phytohemagglutinin (Sigma-Aldrich) and irradiated (30 Gy) allogeneic feeder cells. For polyclonal Vγ9Vδ2 T cell lines, fresh or frozen PBMC (1 × 10⁶ cells/ml) were specifically activated by BrHPP (3 μM) in RPMI 1640 medium supplemented with rhIL-2 (150 IU/ml) and 10% human serum. After 4 days, cultures were supplemented with IL-2 (300 IU/ml). At day 14, specific expansion of Vδ2⁺ T cells within PBL was confirmed by calculating frequencies and absolute T cell numbers. Ex vivo nonadherent cells from healthy donors blood samples were collected following a standard plastic adherence procedure.

Cytokine producing cells were activated by TLR ligands or ABP stimuli either directly within ex vivo PBMC or in coculture with DC (ratio 1/1, 1.5 × 10⁵ T cells). Two hours after the initial activation, intracellular accumulation of cytokines (IFN-γ, TNF-α, IL-4, and GM-CSF) was induced by adding brefeldin A (10 μg/ml). After 3 h of accumulation, cells were collected, subsequently stained for cell surface markers expression, and then fixed with 2% w/v PFA (paraformaldehyde) in PBS. Cells were further permeated for 20 min at 4°C with BD PhosFlow Perm/Wash buffer (BD Biosciences), incubated with fluorochrome-conjugated mAbs specific for the cytokines described above and analyzed by flow cytometry. Type I IFN (IFN-α and IFN-β), IFN-γ and IL-12p70 levels in the triplicate culture supernatants, collected at the indicated timepoints, were assessed by two mAbs sandwich ELISA, following standard procedures.

Freshly isolated nonadherent cells from PBMC (2 × 10⁶ cells/ml) were labeled with CFSE (1 μM in PBS) for 7 min at 37°C, washed, and maintained for 15 min at 37°C in complete medium to allow the release of dye excess, according to the manufacturer’s instructions. Labeled cells were incubated with autologous iDC in 48-wells bottom plates (Nunc), in the presence of IL-2 (60 ng/ml). After 4 days of culture, cells were harvested and stained for TCR Vγ9 chain surface expression. Cell division profile and surface markers expression were measured by flow cytometry. Peaks of cell division and frequencies were calculated by using the Proliferation Platform of the FlowJo analysis software.

Vγ9Vδ2 T cells (2 × 10⁵ cells/well) were activated in the presence of iDC at a γδ T cell to target cell ratio of 1/1 at 37°C in complete RPMI 1640 medium containing monensin (10 μM) and a combination of FITC-conjugated anti-CD107a/b mAbs. After 4 h, cells were harvested, stained with a TCR Vγ9 specific mAb and fixed (0.5% PFA). Double-stained cells were analyzed by flow cytometry.

Several TLR ligands were previously shown to trigger ex vivo proinflammatory cytokine responses of peripheral lymphoid cells after short-term incubation (13). We analyzed in a similar setting the contribution of different human PBL subsets (NK, αβ, and γδ T cells) to the overall PBMC IFN-γ response induced after short-term LPS treatment. In agreement with previous studies (31), NK cells represented the main IFN-γ-producing lymphoid subset within LPS-stimulated PBMC (Fig. 1, A and B, supplemental Table I).⁴ No significant IFN-γ was detected within ex vivo CD3⁺CD56⁺ killer T cells (data not shown). Strikingly, γδ and αβ T cells almost equally contributed to this response (range 5–11% of cells), despite a much lower frequency of the former than the latter subset within whole PBL. Importantly, single cell analysis of IFN-γ accumulation indicated that γδ T cells represented the lymphoid subset that homogeneously produced the highest levels of this proinflammatory cytokine. Moreover, in line with the previously reported Th1 polarization of γδ PBL expressing a Vγ9Vδ2 TCR, Vδ2⁺ T cells represented the vast majority of IFN-γ-producing γδ PBMC after LPS stimulation (Fig. 1 C). Altogether these observations indicate that Vγ9Vδ2 T cells are major T cell contributors to the early ex vivo IFN-γ response induced by short-term LPS treatment of human PBMC.

We compared the ex vivo LPS-induced IFN-γ production of Vδ2, αβ T, NK, and iNKT cell subsets within PBMC vs PBL, to assess the contribution of peripheral myelomonocytic cells to this process. As expected, most Vδ2⁺ and NK cells but a minority only of αβ T cells were IFN-γ⁺ within LPS-treated PBMC (Fig. 2,A, upper panels). iNKT PBMC (stained by both TCR Vα24 and Vβ11-specific mAb) yielded an intermediate pattern. Like αβ and iNKT cells, LPS-induced activation of γδ T cells critically depended on the presence of myelomonocytic cells, because it was no longer detected within PBL. By contrast, almost half of NK cells still responded to LPS within PBL cultures (29 vs 62% of IFN-γ⁺ NK cells) (Fig. 2 A, lower panels). These observations, which are in line with previous reports describing frequent TLR expression on NK cells, suggest that this subset, unlike the vast majority of T cells, can be directly activated by LPS.

Because we recently showed potentiation of γδ T cell activation by DC, we tested whether LPS-treated myeloid DC derived from GM-CSF/IL-4-treated monocytes (moDC), could activate Vγ9Vδ2 T cells. After short-term (30 min) incubation with LPS, iDC, but not B lymphoblastoid cell line, induced potent IFN-γ responses by Vγ9Vδ2 T cells (Fig. 2,B). Similar results were obtained with LPS-treated ex vivo sorted BDCA-1⁺ DC (data not shown). This effect was transient because mature DC, generated after an overnight incubation of iDC with LPS and TNF-α, completely lost their ability to trigger IFN-γ production by γδ T cells in the presence of LPS (Fig. 2 C). Altogether these results indicate that upon LPS exposure, immature myeloid DC rapidly but transiently stimulate IFN-γ production by human Vγ9Vδ2 T cells.

We then compared the ability of selected TLR ligands to induce IFN-γ production by Vγ9Vδ2 T cells after short-term incubation with iDC. Saturating doses of poly(I:C) (TLR3-specific), flagellin (TLR5-specific) and LPS (TLR4-specific), induced IFN-γ production by Vγ9Vδ2 T cells. By contrast, TLR-1/2 (PGN and Pam3CSK4), TLR7 (Imiquimod), and TLR9 (CpG ODN 2216) ligands had no stimulatory effect (Fig. 3,A). None of these TLR ligands had any direct stimulatory effect on Vγ9Vδ2 T cells (data not shown). Titration experiments performed with iDC sensitized for 2 h with grading doses of LPS and poly (I:C) indicated that γδ T cell activation required much lower doses of LPS than poly (I:C) (ED₅₀ of ∼0.8 ng/ml and ∼800 ng/ml, respectively) (Fig. 3 B).

The overall effects of TLR3–4 ligands on human Vγ9Vδ2 functional and proliferative responses were compared with those of strong Vγ9Vδ2 stimulating compounds, such as phosphoAg (BrHPP) and pharmacological inhibitors of the mevalonate pathway (Pamidronate), known to act in a TCR-dependent fashion. The latter compounds triggered a broad range of γδ T cell functions like: 1) Th1/2 cytokines production (IFN-γ, TNF-α, IL-4, and GM-CSF) by a Vγ9Vδ2 T cell line (Fig. 4,A), 2) efficient IL-2-dependent cell division (79.3% of dividing cells at day 4) within ex vivo PBL-Vγ9Vδ2 T cells (Fig. 4,B) and 3) high CD107a/b surface mobilization (Fig. 4 C). By contrast, TLR-sensitized iDC induced exclusively IFN-γ production by γδ T cells, but had no effect on any other cytokinic or degranulation responses.

Transwell assays indicated that both cell-to-cell contacts and soluble factors were required for induction of Vγ9Vδ2 T cell IFN-γ responses by LPS-treated DC (supplemental data and Figure S1). Because TLR engagement was previously shown to up-regulate production of glycolipid Ags recognized by iNKT cells, and to trigger TCR-dependent IFN-γ responses from this subset, we assessed the TCR dependency of LPS-induced IFN-γ responses from Vγ9Vδ2 T cells. Pharmacological blockade of endogenous phosphoAg production by statins had no effect on LPS- and poly (I:C)- induced Vγ9Vδ2 activation, suggesting lack of TCR involvement in this process (supplemental Figure S2A). Accordingly Vγ9Vδ2 transfectants failed to respond in a similar setting (supplemental Figure S2B). To identify the soluble factors implicated in this process, we assessed the stimulating activity of recombinant proinflammatory cytokines on IFN-γ responses by Vγ9Vδ2 PBL, and the inhibitory effect of blocking anti-cytokine mAbs on LPS-induced responses. Although IL-12, IL-15, IL-18, IL-21, and IL-23 had no or limited effects when used alone, IL-12 and IL-18 synergistically induced strong IFN-γ production by Vγ9Vδ2 T cells. However neither IL-12- nor IL-18-specific blocking mAbs had any inhibitory effect on LPS-induced IFN-γ production by γδ T cells (supplemental Figure S3A and S3B). Moreover Vγ9Vδ2 T cells derived from a CD212-deficient individual yielded IFN-γ responses that were comparable to those of γδ T cells from CD212-sufficient donors when stimulated by LPS or poly (I:C) treated iDC (supplemental Figure S3C). Therefore, this rules out a major involvement of IL-12 in this process.

Type I IFN was recently shown to be essential for the IFN-γ response of murine iNKT cells induced by CpG ODN-treated DC (8). We therefore asked whether IFN-α/β, which is released by maturing moDC during the first hours following LPS or poly (I:C) stimulation (Fig. 5,A), was similarly required for activation of the Vγ9Vδ2 T cell IFN-γ response. Blockade by a neutralizing mAb against CD118 (IFNAR2), one of the two subunits of the ubiquitously expressed receptor for IFN-α/β, reduced by ∼80% the LPS-induced production of IFN-γ by human Vγ9Vδ2 T cells (Fig. 5,B). Accordingly, recombinant type I IFN, like LPS, induced IFN-γ responses by ex vivo Vγ9Vδ2 PBL (but not Vδ2⁻ PBL) when cocultured with moDC. Moreover this response was inhibited by a neutralizing CD118-specific mAb (Fig. 5 C). This confirms that type I IFN, released by maturing DC, plays a major role during LPS- and poly (I:C)-induced activation of Vγ9Vδ2 T cells.

We next analyzed whether exogenous type I IFN could mimic the functional effects of LPS ligands on activated Vγ9Vδ2 T cells. In line with the selective triggering of IFN-γ production by TLR ligands, recombinant human type I IFN potentiated the IFN-γ response of both ex vivo and cultured Vγ9Vδ2 T cells, either unstimulated or following antigenic activation. Furthermore, it dramatically inhibited phosphoAg-induced TNF-α production (Fig. 6,A). To assess the biological impact of this finding, we studied the ability of Vγ9Vδ2 T cells to induce efficient polarization of DC into IL-12-producing cells after short-term LPS incubation. IL-12p70 production was greatly enhanced when LPS-sensitized DC were incubated with Vγ9Vδ2 T cells. Interestingly, although Vγ9Vδ2 T cells yielded strong IFN-γ responses when stimulated by pamidronate-treated iDC, they failed to induce IL-12p70 production by DC. This suggests that optimal DC maturation toward IL-12 secretion requires not only IFN-γ and membrane-bound signals provided by activated Vγ9Vδ2 T cells, but also some key signals generated upon TLR engagement by LPS (Fig. 6 B).

pDC belong to a distinct subset of DC able to quickly secrete large amounts of type I IFN in response to viral infections. Accordingly, strong IFN-α responses (Fig. 7,A), but not IL-12 (data not shown), where observed upon incubation of purified peripheral blood pDC with ligand specific for two TLR known to be expressed by this DC subset (R848 for TLR7/8 and CpG for TLR9). In agreement with lack of TLR4 on pDC, LPS did not induce any IFN-α production. In parallel, relevant TLR ligands triggered strong IFN-γ production by Vγ9Vδ2 T cells when cocultured with pDC. We then switched to a more physiological setting using as microbial stimuli a vaccinal YF virus strain (YF-17D). Preferential activation of Vγ9Vδ2 PBL was observed after short-term incubation of PBMC with YF-17D (Fig. 7,B). Moreover, in line with results obtained with TLR ligands, YF-17D induced concomitant production of type I and type II IFN by pDC and Vγ9Vδ2 T cells, respectively, within γδ T/DC cocultures (Fig. 7 C).

The present study indicates that both NK and Vγ9Vδ2 T cells yielded strong IFN-γ responses when exposed to TLR stimuli. Although both TLR3 and TLR4 agonists efficiently activated NK cells in the absence of myelomonocytic cells, the latter cells were strictly required for TLR-induced γδ T cell activation. Although direct NK cell activation by LPS and poly(I:C) is fully consistent with previous reports (31), the ability of TLR ligands to directly activate human γδ T cells has remained debated, owing to very low expression levels of surface or intracellular TLR (frequently below flow cytometry detection thresholds). On the one hand, Wesch and colleagues (11) described a direct costimulatory effect of poly(I:C) on phosphoAg-stimulated Vγ9Vδ2 T cells. In contrast, Kunzmann and colleagues (32) described selective induction of CD69 on Vγ9Vδ2 PBL upon poly(I:C) exposure, which required myelomonocytic cells. In line with this latter study, we show in this study that a wide set of TLR ligands (specific for not only TLR3, but also TLR4, TLR5, TLR8, and TLR9) activated Vγ9Vδ2 T cells through a process that strictly required the presence of either myeloid or plasmacytoid DC expressing the relevant TLR.

Whereas Vγ9Vδ2 TCR signaling upon phosphoAg activation triggered proliferation, degranulation, and various Th1 and Th2 cytokine responses (33, 34), TLR ligands induced exclusively IFN-γ production. This extreme focalization of functional responses is reminiscent with the effect of several proinflammatory cytokines, such as IL-12, IL-15, IL-18, and type I IFN. In particular, iNKT cells, which share with γδ T cells several innate-like features, were previously shown to produce IFN-γ in an IL-12-dependent fashion upon coculture with moDC infected by Salmonella typhimurium (6) or following stimulation by various TLR (7). Along this line, Ag-stimulated Vγ9Vδ2 T cells were previously shown to strongly up-regulate IL-12p70 release by suboptimally stimulated moDC, which could in turn up-regulate Vγ9Vδ2-derived IFN-γ responses (28). Although IL-12 might contribute to late up-regulation of Vγ9Vδ2 IFN-γ responses induced by TLR, we could formally ruled out a mandatory role played by this cytokine in this process on the basis of the following observations. Kinetics studies failed to detect IL-12p70 production by TLR-stimulated DC at time-points when strong IFN-γ responses were induced within γδ T cells. Accordingly, blocking IL-12-specific mAb had no effect on this process and finally TLR-induced IFN-γ responses were still observed using Vγ9Vδ2 T cells from PBL of patients deficient for both IL-12 and IL-23 receptors. Further analysis revealed a major role played by type I IFN. Indeed, 1) TLR-induced Vγ9Vδ2 IFN-γ responses were strongly inhibited by mAb against type I IFN receptors and 2) recombinant type I IFN could readily trigger IFN-γ responses by Vγ9Vδ2 PBMC or Vγ9Vδ2 T cell lines incubated with moDC. Accordingly, moDC produced type I IFN upon TLR3 and TLR4 stimulation, though at much lower levels than TLR8/TLR9-stimulated pDC. These observations are consistent with a recent report, describing type I IFN-dependent IFN-γ production by iNKT cells upon incubation with TLR-stimulated moDC (8). Both type I IFN and TCR engagement were required for optimal induction of IFN-γ production by iNKT cells. In the present study, while Transwell experiments suggested contribution of additional membrane-associated stimuli to TLR-induced Vγ9Vδ2 IFN-γ responses, we failed to evidence implication of Vγ9Vδ2 TCR using either Vγ9Vδ2 transfectants or pharmacological inhibitors of endogenous phosphoAg production. Nevertheless, these negative results do not formally rule out weak γδ TCR interactions that could lead to cryptic IFN-γ responses potentiated by type I IFN.

Short-term TLR stimulation of ex vivo human PBMC not only confirmed that NK cells represent the predominant PBL subset that produces IFN-γ (31, 35, 36) but also highlighted, for the first time, the significant contribution of γδ T cells, and more specifically of the main Vδ2⁺ subset, to this early functional immune response. Although PBL-iNKT also produced IFN-γ upon stimulation by TLR-stimulated DC, they represented very minor players in this process as their frequency within PBMC was ∼10 to 100 times lower than that of Vγ9Vδ2 T cells. Most peripheral Vγ9Vδ2 T cells, unlike those expressing the Vδ1 TCR chain, exhibit a memory phenotype in human adults (37). In depth analysis indicated that Vγ9Vδ2 T cells responding to TLR-stimuli belonged to both central (CD27⁺CD45RA⁻) and effector (CD27⁻CD45RA⁻) memory subsets (M.C. Devilder, data not shown). Interestingly, we could also detect a minor but significant fraction (∼10%) of peripheral αβ T cells which exhibited a central or effector memory phenotype that also responded to TLR3 and TLR4 stimuli. These observations could therefore suggest a link between the memory status of peripheral T cells and their ability to rapidly respond to short-term TLR stimuli. One trivial explanation is that the functional readout used in our study (short-term cytokine production) is restricted to memory cells. However previous studies have shown that TLR ligands also induced selective up-regulation of CD69 on memory T cells (32), although CD69 is an early and sensitive activation marker that can be rapidly up-regulated on naive cells as well upon antigenic stimulation. Therefore, this suggests that naive T cells are unable to respond to soluble/membrane-bound factors provided by short-term TLR-stimulated DC.

In line with our recent study describing DC licensing by Vγ9Vδ2 T cells following mycobacterial infections (28), a strong potentiation of TLR-induced DC-derived IL-12 responses by Vγ9Vδ2 T cells was observed. This would be consistent with an adjuvant role played by this peripheral γδ subset in Th1 polarization, even in situation where pathogens, such as viruses, do not express phosphoAg (38). These observations could also provide clues for previously reported in vitro activation of Vγ9Vδ2 T cells along some viral infections (e.g., HSV; Ref. 39) and functional defects within Vγ9Vδ2 T cells in patients chronically infected by HIV and hepatitis C virus (40, 41). Owing to expression of CCR5 and CXCR3, peripheral effector Vγ9Vδ2 T cells are likely to be recruited to inflamed tissues during early stages of microbial infections (42, 43, 44, 45). Therefore, following contact with short-term pathogen-associated molecular patterns-sensitized iDC and pDC, infiltrating effector γδ T cells might contribute to the local innate immune response by rapidly secreting high amounts of IFN-γ and thus play a role in the containment of invading pathogens. Through this exclusive IFN-γ response, γδ T cells might also contribute to the adaptative immune response by promoting efficient DC maturation, their migration to T cell zones of lymphatic tissue and participate to T cell priming (29).

Quite strikingly, type I IFN not only induced restricted production of type II IFN from TLR-stimulated Vγ9Vδ2 T cells and DC cocultures, but also downmodulated TNF-α production from Ag-stimulated γδ T cells. To our knowledge, this is the first evidence for opposite effects of a given cytokine on induction of these two proinflammatory cytokines. The fact that type I IFN had no or minor effects on Ag-induced TNF-α production from other proinflammatory subsets (such as human iNKT cells, data not shown) suggests a specific feature of Vγ9Vδ2 T cells, although this assumption needs to be confirmed by further studies. Irrespective of this issue, the possibility to induce a highly focused IFN-γ response, without concomitant induction of any other proinflammatory cytokines (most notably TNF-α) or cytolytic responses, might enhance the efficacy of DC maturation mediated by γδ T cells and might limit pathogen spreading. Furthermore, the ability of type I IFN to potentiate IFN-γ while inhibiting TNF-α by Vγ9Vδ2 T cells could be of therapeutic interest because this cytokine could enhance antimicrobial and antitumoral efficacy of Vγ9Vδ2 agonists while limiting their side effect, mainly associated with systemic TNF-α release (46).

We thank Dr. François Trottein for fruitful discussions, the plateform “Développement et Transfert à la Clinique” for providing elutriated PBMC subsets, Dr. Jean-Laurent Casanova, and Dr. Shen-Ying Zhang for providing PBMC samples from CD212⁻ patients and Dr. Giovanna Barba-Spaeth for providing YF-17D virus stocks.

The authors have no financial conflict of interest.