Invariant NK T (iNKT) cells are a subset of innate/memory lymphocytes that recognize lipid Ags presented by CD1d-expressing APCs such as dendritic cells (DCs). Upon primary stimulation through their TCR, iNKT cells promptly produce large amounts of IFN-γ and/or IL-4 that play critical roles in the regulation of innate and adaptive immune responses. To date, the role of environmental factors on iNKT cell functions has been poorly investigated. In this study, we addressed the question of whether PGD2, a potent eicosanoid lipid mediator involved in immune responses and inflammation, could be important in DC/iNKT cell cross-talk. We show that PGD2 dramatically reduced the production of IFN-γ, but not IL-4, by iNKT cells in response to the superagonist α-galactosylceramide (α-GalCer) both in vitro and in vivo. This effect is mediated by the D prostanoid receptor 1 (DP1) expressed by DCs and iNKT cells and requires protein kinase A activation. We also report that PGD2 and BW245C (a selective DP1 agonist) reduce the protective effects of α-GalCer in B16F10-induced melanoma metastasis, an effect that depends on IFN-γ production by iNKT cells. As a whole, these data reveal novel pathways regulating iNKT cell biologic functions and confirm the immunoregulatory roles of PGD2 on the innate response.
Invariant NK T (iNKT)3cells represent an innate/memory cell population that recognizes a limited number of self and exogenous (microbial-derived) glycolipids presented by CD1d-expressing APCs, including dendritic cells (DCs) (reviewed in Ref. 1, 2, 3, 4, 5, 6). The most efficient compound for activating iNKT cells is the synthetic antitumor glycolipid α-galactosylceramide (α-GalCer) (initially isolated from a marine sponge) (7). In vivo stimulation of iNKT cells by α-GalCer promptly induces the production of large amounts of both Th1- and Th2-associated cytokines (especially IFN-γ and IL-4, respectively) that lead to downstream activation of DCs, NK cells, B cells, and conventional T cells (3). This enables iNKT cells to influence the outcome of developing or ongoing immune reactions. Several in vitro and in vivo models demonstrated that, upon natural or intentional activation, iNKT cells are flexible in nature and can either suppress or enhance immune-mediated diseases including inflammation, cancer, and autoimmune diseases (reviewed in Refs. 1, 6). For instance, the production of IFN-γ by iNKT cells enhances the innate and Th1-dependent immune responses of NK cells and CD8+ T cells, ultimately leading to elimination of intracellular pathogens and tumor cells. Conversely, IL-4 production by iNKT cells down-regulates the immune response in several autoimmune diseases, mediates allograft tolerance, and enhances asthma reaction (1, 5, 6). Thus, it is of importance to better understand how IFN-γ and IL-4 production is regulated in iNKT cells. It is known that the type of APCs as well as the nature of the lipid Ag can condition the functional features of iNKT cells (4, 5, 6). In contrast, although still elusive, environmental factors including lipid mediators might play a crucial role in the polarization of iNKT cells. In the present study, we investigated the possibility that PGD2, a potent eicosanoid involved in immune responses and inflammation, could be important in iNKT cell functions.
PGD2 represents one of the major products of the cyclooxygenase pathway. It is released by inflammatory and immune cells in response to various physiological and pathological stimuli including mechanical trauma, microbial products, allergens, or inflammatory mediators including cytokines (reviewed in Ref. 8). Moreover, PGD2 can also be produced by transformed cells (9, 10, 11). Recent evidence suggests that PGD2 exerts multiple effects on the promotion as well as on the control of immune and inflammatory responses (reviewed in Ref. 12, 13, 14, 15). These biological roles for PGD2 are mainly effected through two plasma membrane receptors, the D prostanoid receptor (DP) 1 (coupled with Gαs-type G protein) and DP2 (coupled with Gαi-type G protein), also termed the chemoattractant receptor-homologous molecule expressed on Th2 cells (CRTH2) (for reviews, see Refs. 14 and 16). DP1 and DP2 are linked to different signaling pathways and appear to have distinct functions within the immune system (16). For instance, we and others found that PGD2, through DP1 engagement, strongly affects the functions of DCs including their migratory properties as well as their capacities to promote and/or orientate the differentiation of naive T cells, both in the human and mouse systems (17, 18, 19, 20, 21, 22, 23, 24, 25). In contrast, PGD2 enhances the migration and cytokine release of DP2-expressing cells such as eosinophils and Th2 lymphocytes (16). Due to their important roles in the development and resolution of inflammation, DPs thus represent attractive targets for future therapies by means of selective agonists/antagonists (16). In the present study we describe a new function for PGD2 in immune responses by showing that, by targeting DP1 on DCs and iNKT cells, it decreases the α-GalCer-mediated production of IFN-γ (but not IL-4) by the latter, both in vitro and in vivo. We also report that through this mechanism PGD2 is detrimental in the control of B16F10 lung metastasis, a process that requires IFN-γ production by iNKT cells. These results uncover a novel role for PGD2 in iNKT cell functions and might provide a basis for the design of therapeutic agents that can regulate the functions of iNKT cells within the innate immune response.
Materials and Methods
Reagents and Abs
PGD2, DK-PGD2, and BW245C were purchased from Cayman Chemicals. The DP1 antagonist ONO-AE3-237 was kindly provided by Ono Pharmaceutical. The protein kinase A (PKA) activator N6-benzoyladenosine-3′, 5′-cyclic monophosphate (6-Bnz-cAMP) and the Epac protein activator (8-(4-chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate (8-CPT-2′-O-Me-cAMP) were from Biolog Life Science Institute (where cAMP is cyclic AMP and Epac is exchange protein directly activated by cAMP). The PKA inhibitor Rp-adenosine 3′,5′-cyclic monophosphorothioate (Rp-cAMP) was from Sigma-Aldrich. α-GalCer was purchased from Axxora Life Sciences. mAbs against mouse CD3 (unlabeled), CD5 (allophycocyanin-conjugated), NK1.1 (PerCP-Cy5-conjugated), IFN-γ (PE-conjugated), IL-4 (PE-conjugated), and isotype controls were purchased from BD Pharmingen. CD1d monomers were kindly provided by Dr. M. Salio (Oxford University, Oxford, U.K.) and were loaded with α-GalCer (100 ng/ml) as reported (26). ITAC was from R&D Systems.
Mice and cell line
Six- to 8-wk-old male wild-type (WT) C57BL/6 mice were purchased from Janvier. The generation of TCR Jα18−/− mice (backcrossed at least 10 times in C57BL/6) has been described previously (7). The Vα14+ CD1d-restricetd NKT cell hybridomas DN32.D3 and 3C3.N38 were kindly provided by Dr. A. Bendelac (University of Chicago, Chicago, IL) and Dr. K. Hayakawa (Fox Chase Cancer Center, Philadelphia, PA), respectively, and were cultured in IMDM supplemented with 5% FCS as described (27, 28).
Preparation of spleen and liver mononuclear cells and sorted iNKT cells
Spleen mononuclear cells (MNCs) were prepared by classical procedures. To prepare liver MNCs, perfused livers were harvested and homogenized using a 90-μm pore filter. After extensive washes the liver homogenates were resuspended in a 33% Percoll gradient, and after centrifugation the cells in the pellet were recovered. RBCs were removed by lysis in 155 mM NH4Cl (pH 7.4) containing 10 mM NaHCO3 and 0.1 mM EDTA. For the sorting of iNKT cells, liver MNC were labeled with allophycocyanin-conjugated anti-CD5 and PE-conjugated anti-NK1.1 Abs. After cell surface labeling, cells were sorted using a FACSAria flow cytometer (BD Pharmingen). Sorted CD5+NK1.1+ populations were ∼99% pure (∼90% CD1d/α-GalCer tetramer+anti-TCRβ+; not shown).
Generation of DCs and coculture
Briefly, bone marrow-derived cells were cultured in IMDM supplemented with 10% FCS and 1% of supernatant from a GM-CSF-expressing cell line (J558-GM-CSF) (29). DCs were used on day 14 of culture. DCs (1 × 106 cells/ml) were stimulated with α-GalCer (100 ng/ml) in the presence or absence of PGD2 or DP agonists (5 μM) for 8 h, extensively washed with PBS, and cultured for 48 h with liver MNCs (105 DCs plus 5 × 105 MNCs per well) or sorted iNKT cells (4 × 104 DCs plus 4 × 104 iNKTs per well) in round-bottom 96-well plates in RPMI 1640 supplemented with 5% FCS. To fix DCs, cells were exposed to glutaraldehyde (0.05% in PBS) for 3 min and to 0.2 M lysine and then extensively washed. PGD2 or DP agonists were added to α-GalCer-pulsed DCs or directly during the DC/liver MNC coculture. Supernatants were collected and IL-4 and IFN-γ concentrations were measured by ELISA (R&D Systems).
RNA extraction, cDNA synthesis, and PCR analysis
Total RNAs from sorted iNKT cells or from the iNKT cell lines DN32.D3 and 3C3 N38 (30, 31) were isolated and cDNAs were synthesized from 1 μg of total RNA with random hexamer primers and SuperScript reverse transcriptase (Invitrogen Life Technologies) using standard procedures. PCR amplification was performed with primers for murine β-actin, DP1, and DP2 (Table I).
|Genesa .||Sense .||Reverse .|
|Genesa .||Sense .||Reverse .|
Generated from mouse genomic databases.
Chemotaxis assays were performed using 48-well Transwell (5 μm pore polycarbonate filter) (Corning Costar; Fisher Scientific). Briefly, liver MNCs were resuspended at 10 × 106 cells/ml in RPMI 1640 supplemented with 0,1% FCS and 100 μl was applied to the upper wells of the chamber. ITAC (100 ng/ml) was added in the lower chamber with or without PGD2, BW245C, or DK-PGD2 (5 μM). After 2 h at 37°C, the number of iNKT (CD5+NK1.1+) cells that had migrated through the filter was determined by FACS staining.
Analysis of iNKT cell activation in vivo
Mice were first injected i.v. with PGD2, BW245C (1 mg/kg), or vehicle. After 30 min, α-GalCer (500 ng/mouse) in 200 μl of PBS was injected. Saline-perfused livers were harvested 2 h later and liver MNCs were prepared as described above. Cell suspensions were stained with anti-CD5 and anti-NK1.1 Abs for 30 min in PBS containing 2% FCS and 0.01% NaN3. The presence of cytokines was assessed by intracellular staining. Briefly, cells were fixed in PBS with 1% paraformaldehyde for 10 min and washed and then resuspended in PBS plus 2% FCS and 0.1% saponin (permeabilization buffer) and incubated with PE-conjugated mAb against IFN-γ, IL-4, or control IgG mAb in permeabilization buffer. Cells were acquired and analyzed on a FACSCalibur (BD Biosciences) cytometer using the CellQuest software.
B16F10 lung metastasis model
B16F10 melanoma cells were maintained as described previously (32). Mice received 5 × 105 B16F10 cells (i.v.) and 4 h later the mice were injected i.v. with PGD2, BW245C (1 mg/kg), or vehicle. Thirty minutes later 500 ng of α-GalCer in 200 μl PBS was injected i.v. On days 4 and 8 the mice received PGD2, BW245C (1 mg/kg), or vehicle, and 30 min later 500 ng of α-GalCer in 200 μl PBS was injected i.v. Mice were killed on day 14 and surface lung metastases were counted.
Results are expressed as the mean ± SD The statistical significance of differences between experimental groups was calculated by ANOVA1 with a Bonferroni posttest or an unpaired Student’s t test (GraphPad Prism 4 Software, San Diego). Results with p < 0.05 were considered significant.
PGD2 inhibits the α-GalCer-induced release of IFN-γ by spleen and liver MNCs
We first aimed to determine the effect of PGD2 on the release of cytokines by MNCs stimulated with α-GalCer, a potent iNKT cell ligand. Because they contain diverse categories and different frequencies of iNKT cells, spleen and liver MNCs were used. As expected, in response to α-GalCer the MNCs from WT mice produced both IFN-γ and IL-4 whereas cells from iNKT cell-deficient mice (Jα18−/− mice) failed to do so (Fig. 1, left column). Having confirmed the requirement of iNKT cells in this setting, the effects of PGD2, BW245C (a selective DP1 agonist), and DK-PGD2 (a selective DP2 agonist) were next studied. PGD2, BW245C, and DK-PGD2 had no effect on the α-GalCer-induced release of IL-4 by MNCs. In stark contrast, PGD2 and BW245C, but not DK-PGD2, dose-dependently reduced IFN-γ secretion by α-GalCer-stimulated spleen (Fig. 1,A) and hepatic (Fig. 1 B) MNCs, with an EC50 value of ∼1 μM (for PGD2) and ∼3 μM (for BW245C), respectively (average of four independent experiments).
PGD2 exerts its inhibitory effect by activating the DP1/cAMP/PKA pathway
To confirm the role of DP1 in this phenomenon, the selective DP1 antagonist ONO-AE3-237 was used. As seen in Fig. 2,A, ONO-AE3-237 reversed the inhibitory effect induced by PGD2 and BW245C on IFN-γ release while it had no role in IL-4 production. Thus, PGD2 targets DP1 to inhibit the production of IFN-γ by α-GalCer-stimulated spleen and liver MNCs. DP1 has been described as enhancing intracellular cAMP to exert its effect. Elevation of cAMP is known to activate a PKA-dependent or PKA-independent pathway, the latter requiring Epac proteins (33, 34). Thus, we investigated whether the activation of PKA or Epac proteins could mimic the effect exerted by DP1 agonists. For this, MNCs were incubated with a selective activator of PKA (6-Bnz-cAMP) or Epac proteins (8-CPT-2′-O-Me-cAMP) just before the addition of α-GalCer. As represented in Fig. 2,B, the activation of PKA, but not Epac proteins, dramatically reduced the α-GalCer-induced production of IFN-γ by spleen and liver MNCs. Finally, as shown in Fig. 2 C, the selective PKA inhibitor Rp-cAMP restored the production of IFN-γ by α-GalCer-stimulated, PGD2-treated (and BW245C-treated) liver and, to a lesser extent, spleen MNCs. Collectively, these data indicate that PGD2 exerts its inhibitory effect on the iNKT cell-mediated IFN-γ production by MNCs by activating a DP1/cAMP/PKA pathway.
PGD2 inhibits the production of IFN-γ by iNKT cells and by NK cells
Presentation of α-GalCer by CD1d-expressing APCs primarily induces the release of IFN-γ by iNKT cells that in turn triggers IFN-γ production by NK cells (35). To investigate whether PGD2 and BW245C act on IFN-γ release by iNKT cells and NK cells, intracellular FACS analysis was performed. To this end, liver cells, which contain a high frequency of iNKT cells, were labeled with anti-CD5 and anti-NK1.1 Abs. Compared with unstimulated cells, stimulation of hepatic MNCs with α-GalCer induced an increased number of CD5+NK1.1+ (iNKT) cells positive for IFN-γ and IL-4 (Fig. 3,A). In the presence of PGD2 or BW245C, the number of CD5+NK1.1+ cells stained positively for IFN-γ was reduced by ∼45% (PGD2) and ∼30% (BW245C), whereas the number positive for IL-4 remained unchanged. Similar results were obtained when liver cells were labeled with the CD1d/α-GalCer tetramer, a probe that exclusively stains iNKT cells (36), plus an anti-TCRβ mAb (not shown). As expected, CD5−NK1.1+ (NK) cells also produced IFN-γ intracellularly but the percentage of NK cells positive for IFN-γ was also reduced (by ∼55 and 30%) in the presence of PGD2 or BW245C, respectively (Fig. 3 B). These data together show that PGD2 reduced, via DP1 activation, the production of IFN-γ by iNKT cells and NK cells while it had no effect on the synthesis of IL-4. Whether PGD2 directly acts on CD1d-expressing APCs or/and on iNKT cells was next investigated.
PGD2 targets DCs to reduce IFN-γ production by iNKT cells
It is well established that, among APCs, myeloid DCs have a strong ability to activate iNKT cells (37, 38). Thus, we assessed whether PGD2 could target DCs to modulate iNKT cell activation. To this end, α-GalCer-stimulated DCs were treated or not treated with PGD2 or DP agonists, washed, and then cocultured with liver MNCs. As depicted in Fig. 4,A, α-GalCer-stimulated DCs induced IL-4 and IFN-γ production by liver MNCs isolated from WT but not from Jα18−/− (not shown) mice. Compared with vehicle, the treatment of DCs with PGD2 or DP agonists just before α-GalCer stimulation had no impact on the synthesis of IL-4 by liver MNCs. In contrast, PGD2 and BW245C, but not DK-PGD2, reduced the synthesis of IFN-γ. To ascertain that PGD2- and BW245C-treated DCs directly act on iNKT cells in this setting, purified iNKT cells were used. As depicted in Fig. 4,B, PGD2- and BW245C-treated DCs strongly reduced IFN-γ production by sorted CD5+NK1.1+ (iNKT) cells. DC-derived IL-12 is known to push iNKT cells to release IFN-γ (39). As observed for IFN-γ, decreased concentration of IL-12 was detected in the supernatants of PGD2- and BW245C-treated (but not DK-PGD2-treated) DCs cocultured with sorted iNKT cells (Fig. 4 B). Thus, targeting of DP1 in DCs abrogated the release of IFN-γ by iNKT cells, probably through diminished IL-12 production by DCs.
DP1 and DP2 are functional in murine iNKT cells
We next aimed to determine whether PGD2 could also act on iNKT cells to modulate their ability to release cytokines. Before this, we first investigated the expression of PGD2 receptors in iNKT cells by RT-PCR, because anti-mouse DP Abs are not yet available. As shown in Fig. 5,A, sorted hepatic CD5+NK1.1+ cells (∼98% pure) express mRNAs for DP1 and DP2, and this finding was confirmed by using the canonical (Vα14Jα281 TCR rearrangement) CD1d-restricted NKT cell hybridomas 3C3.N38 and DN32.D3. We and others have shown that the activation of DP1 inhibits cell migration (18, 20, 21, 25, 40) while that of DP2 exerts an opposing effect (41, 42). Thus, to study DP1 and DP2 functionality in iNKT cells we performed a Transwell chemotaxis assay. As shown in Fig. 5 B, BW245C reduced (by ∼40%) whereas DK-PGD2 enhanced (by ∼60%) the chemotactic response of iNKT cells to ITAC, a CXCR3 ligand. In contrast, PGD2 had no significant effect on iNKT cell migration, a finding explained by the coexpression of DPs in iNKT cells and by their well-known antagonizing functions. These data show that DP1 and DP2 are functional in iNKT cells and that their individual activation impacts the chemokine-driven iNKT cell motility.
PGD2 acts on iNKT cells to reduce IFN-γ release
To study the effects of PGD2 on iNKT cells, we took advantage of the fact that, after stimulation with α-GalCer, glutaraldehyde-fixed DCs still maintain their ability to activate hepatic cells to release cytokines. Therefore, to target liver cells and not α-GalCer-sensitized DCs, glutaraldehyde-fixed DCs were cocultured with hepatic MNCs, in the presence or absence of PGD2, BW245C, or DK-PGD2. As Fig. 6,A shows, the addition of PGD2 or BW245C, but not DK-PGD2, inhibited IFN-γ (but not IL-4) releases by liver MNCs. These data indicate that iNKT cells could be targeted by PGD2 via DP1 to decrease IFN-γ synthesis. To investigate whether this effect is direct, sorted iNKT cells were stimulated with anti-CD3 Ab or with α-GalCer-loaded CD1d in the presence of DP agonists. As observed in Fig. 6 B, PGD2 and BW245C abrogated IFN-γ synthesis by iNKT cells in the context of (APC-free) CD3- or α-GalCer/CD1d -stimulated cytokine production.
PGD2 inhibits IFN-γ production by iNKT cells in vivo
Having established that PGD2, by targeting DP1 on DCs and on iNKT cells, affects the production of IFN-γ by the latter in vitro, we next intended to determine whether this effect also occurs in vivo. For this, PGD2, BW245C, or vehicle alone was injected 30 min before α-GalCer, and the activation state of iNKT cells was next determined by FACS staining. As represented in Fig. 7,A, α-GalCer induced an increased number of IFN-γ positive iNKT (CD5+NK1.1+) cells in the liver 2 h postinjection. Compared with vehicle-treated mice, PGD2 and BW245C treatment significantly decreased the frequency of IFN-γ-positive iNKT cells by ∼55 and 65%, respectively. Of note, the number of IL-4-positive iNKT cells in α-GalCer-injected mice was not affected by PGD2 or BW245C (not shown). Similarly, the frequency of IFN-γ-positive NK cells (CD5−NK1.1+) was reduced after PGD2 or BW245C treatment (by ∼50%; Fig. 7 B). As a whole, PGD2 reduced the α-GalCer-induced production of IFN-γ by iNKT cells as well as the bystander activation of NK cells in vivo.
PGD2 reversed the iNKT cell-mediated protection against lung metastasis
α-GalCer has been shown to be important to enhance the antitumor effect against B16F10 experimental pulmonary metastases, an effect that depends on IFN-γ production by iNKT and NK cells (32, 43). To study the role of DP1 in iNKT cell functions in vivo, PGD2 or BW245C was injected with α-GalCer and the number of lung metastases in B16F10-inoculated mice was determined. As seen in Fig. 8, B16F10-inoculated mice developed large number of metastatic lung tumor nodules and, in agreement with previous reports (32, 44), this number was dramatically decreased in mice injected with α-GalCer. Interestingly, the protective effect of α-GalCer was blocked in mice treated with PGD2 or BW245C. As a whole, these data show that exogenous activation of DP1 by PGD2 or BW245C decreases the iNKT cell-mediated anticancer immunity, at least in this model. It is likely that NK cells are also involved in this phenomenon, owing to their transactivation by iNKT cell-released IFN-γ (35, 43).
One important issue in the field of iNKT cells is to understand how these cells can become polarized toward Th1-type or Th2-type cytokine producers during pathological conditions. Several studies have shown that environmental cytokines, either produced or not produced by APCs, can drive iNKT cells toward a Th1 or Th2 direction (45, 46, 47). More recently, the role of other potentially important factors has been proposed. Through an unknown mechanism, the estrogen class of sexual hormones enhances the α-GalCer-induced release of IFN-γ by iNKT cells (48), while agonists of the adenosine A2A receptor exert an opposing effect (49). In this study, we investigated the role of PGD2, an important lipid mediator released during inflammation, on the CD1d-dependent production of cytokines by iNKT cells. Our data uncover a novel role for PGD2 in innate immunity.
We showed that PGD2, by targeting DP1 in both DCs and iNKT cells, selectively reduced the production of IFN-γ by iNKT cells upon stimulation with the canonical ligand α-GalCer. This effect is mediated by an increase cAMP elevation that culminates in PKA activation, but not Epac protein activation. Notably, we also found similar data for other PG members known to elevate intracellular cAMP content and be functionally relevant in many pathological situations (12, 13), such as PGE2 and PGI2 (not shown). Thus, it is likely that the production of cAMP-elevating PGs could bias iNKT cells toward a less Th1 type-producing cell in vivo. Interestingly, this property contrasts with that of agonists of some TLRs, a family of innate receptors sensing microbial components (for review, see Ref. 50), which rather polarize iNKT cells toward a Th1 direction (5, 6). Thus, according to the stress-promoting agents (PGs vs microbial components), iNKT cells could be biased toward different directions. The mechanism by which PGD2-treated DCs act on iNKT cells to reduce IFN-γ production is still elusive. We tend to eliminate the possibility that it is via decreased CD1d synthesis, because equal levels of CD1d expression were detected in PGD2-treated DCs even after DC/MNC coculture (24 h; data not shown). It is admitted that following the primary stimulation of iNKT cells through their TCRs, costimulatory surface molecules as well as cytokines provided by DCs are important to shape the outcome of the cytokine production by iNKT cells. For instance, TCR/CD1d-dependent iNKT cell stimulation induces CD40/CD40L-dependent as well as CD86/CD28-dependent IL-12 production by DCs, which finally pushes iNKT cells to release IFN-γ (39, 51). Our data indicate that PGD2 does not modulate the expression of CD40 and CD86 on DCs during DC/MNC coculture (24 h; data not shown). In contrast, we observed decreased IL-12 production in DC/iNKT cell coculture, which paralleled that of IFN-γ. This supports the contention that IL-12 may be implicated in this regulation system and extends the previous finding showing that PGD2 abrogates IL-12 production by maturing DCs (19, 20, 24). Interestingly enough, our data suggested that PGD2 also directly targets iNKT cells to reduce IFN-γ production. Here too, the mechanism of the selective inhibitory action of PGD2 on IFN-γ release is as yet unknown. It is likely that, early after TCR engagement, PGD2 decreases the synthesis and/or DNA binding activity of some transcription factors known to be important in IFN-γ production by iNKT cells. In this context, c-Rel, one component of the NF-κB factors, appears to be important for the effective production of IFN-γ by activated iNKT cells (52). Finally, although PGD2 is sufficient to inhibit IFN-γ production by iNKT cells in the absence of APCs (Fig. 6 B), it could also exert its inhibitory effect during DC/iNKT cell contact by affecting the expression of surface molecules necessary for optimal IFN-γ release by iNKT cells, such as receptors for costimulatory molecules or cytokines (i.e., IL-12 receptor). We report for the first time PG receptor expression and functionality in iNKT cells, thus suggesting that PG members could act on these cells in physiological and pathological situations. Interestingly, PGD2 is known to favor Th2 functions (through DP2) and to restrain Th1 functions (via DP1) of conventional T cells (53, 54). Although both receptors are functional in iNKT cells, the use of selective agonist and antagonist revealed that DP1 is responsible for the reduced production of IFN-γ by iNKT cells, whereas DP2 has no role on cytokine release. These data are in line with those from a recent report showing that PGD2 directly inhibits IFN-γ production of human NK cells via signaling through DP1 (55).
Our observation may have important functional implications. First, the release of PG members, including PGD2, by inflamed tissues might avoid the overreacting stimuli of iNKT cells during stress conditions (such as infection) and might thus represent a mechanism to retro-control type 1 responses. Moreover, accumulated experimental evidence has supported the role of iNKT cells in promoting innate antitumor immunity (3, 6, 32, 56). The demonstration that PGD2 exerts a potent immune-suppressive effect in the lung metastasis model used in this study confirms and extends the emerging concept that PGs are important in tumor immunity in general. Indeed, one mechanism of tumor escape is the release of PG members by tumor cells into their microenvironment, leading to immunosuppression and tumor cell dissemination to distant metastatic sites (57, 58, 59, 60). There is abundant documented evidence of the elevated expression of cyclooxygenase-2 in tumors, and selective inhibitors of cyclooxygenases have the potential to inhibit tumorigenesis (reviewed in Ref. 61). Furthermore, it is anticipated that the development of specific PG receptor antagonists will provide new advantageous tools for chemoprevention. It is possible that, by targeting APCs and/or iNKT cells, tumor-derived PGs could display, at least in part, their immunosuppressive effects, a hypothesis presently investigated in our laboratory. In contrast, activation of the iNKT cell/NK cell axis has a great potential for the treatment of patients carrying tumors, and the use of DCs pulsed with α-GalCer or α-GalCer-derived analogues has provided promising results in overcoming tumor progression (62, 63, 64, 65). For human purposes, PGE2 is currently used to prepare DCs from monocytes. It remains to be seen whether the omission of PGE2 during differentiation/maturation steps could improve the antitumor properties of DCs by increasing iNKT cell functions in this setting. The herein demonstrated role of PGD2 in decreasing IFN-γ production by iNKT cells might also be relevant in other situations. iNKT cells play an important role in protection against pathogens and tumors but also in the enhancement of some autoimmune and/or inflammatory diseases, including asthma (66, 67). Whether or not PG production from injured/inflamed tissues might intervene in pathological situations by biasing iNKT cells toward a less Th1 phenotype is unknown and worthy of future investigations. For instance, during acute asthmatic episodes PGD2 is massively released by mast cells into the lungs (68) and might serve as an endogenous regulator of iNKT cell activity to enhance the pathology. If true, pharmacologic approaches to control iNKT cell functions by means of PGD2 receptor deactivation might be of interest. In contrast, the use of some α-GalCer analogues that bias the Th2 activity of iNKT cells has been shown to suppress experimental autoimmune encephalomyelitis and protect NOD mice against diabetes (69, 70). Thus, the activation of some PG receptors by selective agonists that control the Th1 activity of iNKT cells (and thus enhance their Th2-promoting activities) might also be of great value in controlling Th1-mediated autoimmune diseases.
To conclude, we show for the first time that PGD2 represents, along with glycolipid Ags, another class of lipids able to control cytokine release by iNKT cells and, through this mechanism, exert potent immune-regulatory effects. Because PGD2 might serve as an endogenous regulator of iNKT cell activity in vivo, our data might be relevant in several physiological and pathological situations.
We thank Drs. T. Nakayama and M. Taniguchi (Chiba University, Chiba, Japan) for the gift of Jα18−/− C57BL/6 mice. We thank Drs. A. Bendelac (University of Chicago, Chicago, IL) and K. Hayakawa (Fox Chase Cancer Center, Philadelphia, PA) for the gift of the NKT cell hybridomas. We also thank C. Vendeville (Institut Pasteur de Lille, Lille, France) for her technical assistance.
The authors have no financial conflict of interest.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by the Institut National de la Santé et de la Recherche Médicale, the Pasteur Institute of Lille, and the University of Lille 2. D.T. and C.P. are recipients of a postdoctoral or a doctoral fellowship from the Conseil Régional Nord-Pas-de-Calais/ Institut National de la Santé et de la Recherche Médicale. C.F. and P.G. are supported by Institut National de la Santé et de la Recherche Médicale, and F.T. by the Centre National de la Recherche Scientifique.
Abbreviations used in this paper: iNKT, invariant NK T cell; 6-Bnz-cAMP, N6-benzoyladenosine-3′,5′-cyclic monophosphate; 8-CPT-2′-O-Me-cAMP, (8-(4-chlorophenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate; α-GalCer, α- galactosylceramide; cAMP, cyclic AMP; DC, dendritic cell; DP, D prostanoid receptor; Epac, exchange proteins directly activated by cAMP; ITAC, IFN-inducible T cell α chemoattractant; MNC, mononuclear cell; PKA, protein kinase A; WT, wild type.