Human iNKT Cells Promote Protective Inflammation by Inducing Oscillating Purinergic Signaling in Monocyte-Derived DCs

SUMMARY

Invariant natural killer T (iNKT) cells are innate T lymphocytes that promote host defense against a variety of microbial pathogens. Whether microbial ligands are required for their protective effects remains unclear. Here, we show that iNKT cells stimulate human-monocyte-derived dendritic cells (DCs) to produce inflammatory mediators in a manner that does not require the presence of microbial compounds. Interleukin 2 (IL-2)-exposed iNKT cells selectively induced repeated cytoplasmic Ca²⁺ fluxes in DCs that were dependent on signaling by the P2X₇ purinergic receptor and mediated by ATP released during iNKT-DC interactions. Exposure to iNKT cells led to DC cyclooxygenase 2 (PTGS2) gene transcription, and release of PGE₂ that was associated with vascular permeabilization in vivo. Additionally, soluble factors were released that induced neutrophil recruitment and activation and enhanced control of Candida albicans. These results suggest that sterile interactions between iNKT cells and monocyte-derived DCs lead to the production of non-redundant inflammatory mediators that promote neutrophil responses.

In Brief

Xu et al. show that in the absence of microbial products, autoreactive innate T lymphocytes, called iNKT cells, activate inflammatory dendritic cells to release lipid mediators. This sterile inflammatory interaction promotes neutrophil-mediated control of an opportunistic fungal pathogen.

INTRODUCTION

Inflammation is a multifactorial process that occurs in response to a variety of stimuli. Innate cell types (e.g., mast cells, macrophages, and dendritic cells [DCs]) residing in tissues produce lipid, peptide, and chemical mediators that rapidly induce local vascular changes leading to increased blood flow and edema. Additionally, lipid mediators, chemokines, and cytokines released by innate cells recruit neutrophils and other leukocyte populations to the affected site. Until recently, inflammation was regarded as a response principally initiated by exposure to microbial molecular products. However, it is now becoming clear that inflammation can also arise from endogenous processes and does not require the presence of foreign compounds. While this type of inflammatory response, often termed “sterile” inflammation, is thought to be initiated by endogenous compounds produced in response to cellular stress or damage, the immunological interactions that give rise to sterile inflammatory responses are not yet well characterized. In particular, the role of innate lymphocyte subsets, such as invariant natural killer T (iNKT) cells, is an open question.

iNKT cells express a semi-invariant T cell receptor (TCR) and recognize lipid antigens presented by CD1d glycoproteins, which are non-classical antigen-presenting molecules expressed by most myelomonocytic cell types (Bendelac et al., 2001). However, a signature characteristic of iNKT cells is that they are not dependent on recognition of foreign antigens for activation, as they can respond to CD1d-mediated presentation of self-antigens and are potently co-stimulated by cytokines produced by activated DCs (Brigl et al., 2003). iNKT cells may therefore be particularly well positioned to participate in endogenous pathways of inflammation.

We have recently shown that many human iNKT cells recognize a self-lipid called lysophosphatidylcholine (LPC) (Fox et al., 2009; López-Sagaseta et al., 2012). LPC is generated as a product of the membrane phospholipid cleavage reaction that releases free fatty acids for the biosynthesis of eicosanoid lipid mediators, and thus, it often accumulates to very high levels during inflammatory responses. The recognition of LPC by human iNKT cells suggests that they likely receive specific TCR stimulation from CD1d⁺ antigen-presenting cells (APCs) that are in areas where eicosanoid biosynthesis has been initiated. Additionally, for a period of time after TCR stimulation by self-antigens, iNKT cells can become activated to secrete interferon-γ (IFN-γ) in a TCR-independent manner by exposure to cytokines (e.g., interleukin 12 [IL-12] and IL-18) produced by activated APCs (Wang et al., 2012). Thus, human iNKT cells may become activated by both TCR-dependent and -independent pathways in inflammatory environments.

Circulating human iNKT cells express a pattern of chemokine receptors indicating they are poised to traffic to peripheral sites of inflammation (Kim et al., 2002; Thomas et al., 2003). The observation that their chemokine receptor pattern overlaps with that of human monocytes, suggesting that monocyte-derived cells may represent a major category of APCs for iNKT cells at inflammatory sites. Consistent with this, iNKT cells and CD1d⁺ APCs have been observed in a variety of inflamed human epithelial and endothelial tissues (Amanuma et al., 2006; Bobryshev and Lord, 2005; Chan et al., 2005; Kyriakakis et al., 2010). Primary DCs possessing a genetic signature characteristic of monocyte-derived DCs have also been isolated from inflamed human tissues, supporting the physiological relevance of these DCs for human inflammatory responses (Segura et al., 2013). While it is established that monocyte-derived DCs activate human iNKT cells in a CD1d-dependent manner and that iNKT cells in turn alter the functions of monocyte-derived DCs, the cellular signaling initiated in the DCs as a result of their interactions with iNKT cells and the resulting physiological outcomes from this signaling have not been determined. Here, we have investigated the physiological impact of the pronounced calcium signaling response that is induced in human-monocyte-derived DCs during interactions with iNKT cells.

RESULTS

iNKT Cells Selectively Induce Prolonged Cytoplasmic Ca²⁺ Fluxing in DCs

Primary iNKT cells were enriched from human peripheral blood by magnetically sorting cells labeled with α-GalCer-loaded CD1d tetramer. Concurrently, polyclonal T cells depleted of iNKT cells were isolated from the flow-through fraction. The iNKT-enriched and iNKT-depleted T cells were incubated in medium containing IL-2 for 3 days and then labeled with CellTracker Red dye and mixed with autologous DCs that were labeled with the Ca²⁺ indicator dye Fluo-4. The cell mixtures were briefly centrifuged (20 s) to initiate cell contact and then gently resuspended and placed onto ICAM-1-coated glass slides for analysis by time-lapse microscopy. As shown in Figure 1A, the iNKT-enriched T cells formed clusters or cell-cell conjugates with DCs, whereas T cells in the iNKT-depleted fraction were rarely observed to be closely associated with DCs. Additionally, the DCs in the iNKT-enriched conditions showed significantly brighter Fluo-4 signal than the DCs that were exposed to the iNKT-depleted T cell fraction (Figure 1B). In the presence of iNKT-enriched T cells, individual DCs underwent multiple repeated spikes of Fluo-4 signal intensity over a period of 30–60 min (Figure 1C; Movies S1, S2, S3, and S4). These results suggested that exposure to primary human iNKT cells was selectively associated with cytoplasmic Ca²⁺ signaling in the DCs.

Figure 1. iNKT Cells Induce Oscillating Spikes of Cytoplasmic Ca²⁺ in Monocyte-Derived DCs.

(A) Human monocyte-derived DCs were labeled with the green Ca²⁺ indicator dye Fluo-4 and combined with iNKT-enriched or iNKT-depleted fractions from autologous PBMCs that were labeled with a red Ca²⁺-insensitive dye (CMTPX). The cell mixtures were centrifuged for 20 s and then gently resuspended and added to slides coated with human ICAM-1-Fc fusion protein. Three examples each are shown to illustrate the clustering of the DCs with the iNKT-enriched T cells and their failure to similarly cluster with the iNKT-depleted T cells.

(B) Plot showing the integrated log fluorescence of Fluo-4 signal from DCs in the presence of iNKT-enriched (n = 24) or iNKT-depleted (n = 24) fractions.

(C) Time-lapse fluorescence microscopic analysis of intracellular Ca²⁺ levels in DCs in the presence of primary iNKT-enriched cells (see also Movies S1, S2, S3, and S4). Microscopic images were taken at 20-s intervals for 40 min. Line plots on the right show the Fluo-4 mean fluorescence intensity (MFI) of the indicated DCs over time. See also Movies S1, S2, and S3.

(D) Plot showing the log fluorescence of Fluo-4 over time as determined from microscopic analysis of slides containing DCs in the presence of a pure iNKT cell clonal line (n = 47, purple lines) or DCs with autologous polyclonal T cells (n = 39, blue lines).

(E) Plot comparing the integrated log fluorescence of Fluo-4 signal for DCs alone (n = 47), DCs with iNKT cells (n = 47), and DCs with autologous polyclonal T cells (n = 39).

(F) Plot showing the Fluo-4 MFI over time of the pictured DC in the presence of highly motile iNKT cells labeled in red. See also Movie S4.

(G) DCs and iNKT cells were centrifuged to generate small clusters of cells that remained stably associated over the course of the experiment. Fluo-4 fluorescence intensity was calculated for DCs that were near iNKT cells (n = 17) and compared to DCs within the same culture that were not in the vicinity of iNKT cells (n = 17).

Since our analysis of the cells in the iNKT-enriched T cell fraction indicated that only ~65% were positive for α-GalCer-loaded CD1d tetramer staining, we could not rule out that contaminating cells might be responsible for the DC signaling effects we observed. To confirm that iNKT cells selectively induce cytoplasmic Ca²⁺ flux in DCs, we performed similar time-lapse microscopy analyses using several clonal iNKT cell lines that had been previously expanded in culture and that showed >99% positive staining by α-GalCer-loaded CD1d tetramer. For comparison, we used polyclonal T cells (generated from the same donor as the DCs) that were cultured under conditions identical to those of the iNKT cell clones. We again observed that DCs exposed to iNKT cells showed elevated Fluo-4 signal intensity that increased over time, whereas DCs exposed to autologous polyclonal T cells showed lower Fluo-4 starting signals and little or no increased intensity over time (Figure 1D). Quantitative analysis of the integrated Fluo-4 fluorescence signal of individual DCs confirmed that intracellular Ca²⁺ concentrations were significantly higher in the presence of iNKT cells compared to DCs alone or DCs with autologous polyclonal T cells (Figure 1E). Similar to what we observed with the iNKT-enriched T cells, in the presence of iNKT cell clonal lines, individual DCs showed repeated spiking of intracellular Ca²⁺ during the 30- to 60-min periods of time that we performed time-lapse microscopic analysis (Figure 1F).

In these analyses, the iNKT cells were highly motile and made dynamic contacts with the DCs, such that individual DCs were contacted repeatedly by different iNKT cells during the course of the time-lapse microscopy analysis (see Movie S4). We hypothesized that the tendency for the Fluo-4 signal intensity of DCs to increase over time might be due to their repeated contacts with iNKT cells. To investigate this further, we centrifuged the iNKT-DC cell mixture for 90 s prior to microscopic analysis, which resulted in the formation of stable iNKT-DC clusters as well as some DCs that did not appear to be in contact with iNKT cells. A comparison of the Fluo-4 signal for DCs that were in close proximity with iNKT cells to DCs that did not appear to be near any iNKT cells showed that iNKT-proximal DCs had significantly higher cytoplasmic Ca²⁺ levels than DCs in the same culture that were not in the vicinity of iNKT cells (Figure 1G). These results confirmed that iNKT cells selectively promoted Ca²⁺ signaling in human-monocyte-derived DCs and suggested that this effect might be due to products generated during iNKT-DC interactions.

Purinergic Signaling Induced in DCs by iNKT Cells

Since extracellular ATP is a major soluble factor that induces Ca²⁺ signaling in monocytic cells, we tested whether the DC Ca²⁺ signaling we observed in the presence of iNKT cells might be due to secreted ATP. Fluo-4-labeled DCs were mixed with unlabeled iNKT cells in the presence or absence of Apyrase enzyme added to rapidly degrade extracellular ATP, and fluorescence intensity was measured in a fluorescence plate reader. Whereas DCs that were co-incubated with iNKT cells showed increased Fluo-4 signal compared to DCs alone or DCs co-incubated with control T cells (Figure S1), DCs co-incubated with iNKT cells in the presence of apyrase enzyme showed little or no elevated Fluo-4 signal (Figure 2A). These results suggested that the Ca²⁺ fluxing we had observed in DCs was due to stimulation by extracellular ATP that was released during their interactions with iNKT cells.

Figure 2. Secreted ATP Is Produced and Rapidly Degraded during iNKT-DC Interactions.

(A) Flow cytometric analysis of Fluo-4 MFI in cultures of DCs alone (green line), DCs and iNKT cells (purple line), or DCs and iNKT in medium containing recombinant apyrase enzyme (blue line). See also Figure S1.

(B) Analysis of extracellular ATP detected in culture supernatants from the indicated conditions. Results are given as arbitrary luminescence units from a luciferase-based ATP assay and are from one representative experiment out of three independent analyses. Bars show the means and SDs of three replicates.

(C) Fluo-4 signal from DCs alone (green line) or from DCs that were pretreated with vehicle (purple line) or with the P2X₇ inhibitor KN-62 (blue line) and exposed to iNKT cells. The plot on the right shows aggregated results from six independent experiments, with the Fluo-4 signal of the iNKT+DC condition normalized by the corresponding signal from the DC-alone condition.

(D) Flow cytometric analysis of Fluo-4 MFI by DCs that were exposed to iNKT cells (purple line), placed in medium containing added ATP (orange line), or kept in medium alone (green line). Similar results were obtained in three independent analyses.

(E) Flow cytometric analyses of DCs (top) and iNKT cells (bottom) for expression of the ecto-ATPase CD39. Filled histograms show specific staining for CD39; dotted histograms show isotype control staining. Similar results were obtained in four independent analyses.

(F) Analysis of the stability of extracellular ATP after exposure to DCs or iNKT cells. Synthetic ATP was added to the culture medium at a final concentration of 1 mM and then incubated alone (ATP in medium) or with 5 × 10⁶ iNKT cells or DCs alone or in the presence of 100 μM ARL67156 (CD39 inhibitor). Results represent the means and SDs of three replicates.

We therefore tested culture supernatants from iNKT cells alone, DCs alone, or iNKT cells with DCs for their levels of extracellular ATP. Significantly elevated levels of ATP were present in supernatants from iNKT-DC co-cultures compared to cultures of either cell type alone (Figure 2B). While supernatants from anti-CD3 antibody-stimulated iNKT cells showed slightly higher extracellular ATP levels than those from unstimulated iNKT cells, the amount of extracellular ATP in supernatants, even from activated iNKT cells, was typically at least 10-fold less than that detected in supernatants from iNKT-DC co-cultures (Figure 2B). Supernatants from DCs alone typically showed levels of extra-cellular ATP that were higher than iNKT supernatants but still significantly less than the amounts detected from iNKT-DC co-cultures (Figure 2B). Given that it has been previously shown that human monocytes secrete ATP after lipopolysaccharide (LPS) stimulation (Netea et al., 2009), these results suggested that stimulation from contact with iNKT cells activates the monocyte-derived DCs to secrete ATP.

Multiple different purinergic receptors that bind ATP and induce Ca²⁺ signaling are expressed by human-monocyte-derived DCs, including both P2X (ligand-gated ion channels) and P2Y (G-protein-coupled receptors) subtypes (Marteau et al., 2004). We found that the increase in the DC Fluo-4 signal in the presence of iNKT cells was significantly diminished (mean ± SD, 72% ± 18%) when DCs were pretreated with a selective P2X₇ inhibitor called KN-62 (Figure 2C). The inhibitory effects of KN-62 on DC Ca²⁺ signaling appeared similar to those of a pan-purinergic receptor inhibitor called suramin (data not shown). These results suggested that iNKT-induced DC Ca²⁺ signaling is mainly dependent on stimulation through the P2X₇ receptor. Based on these results, we hypothesized that ATP secreted during iNKT-DC interactions then acts in an autocrine or nearby paracrine manner to induce DC Ca²⁺ signaling.

However, we noted that in contrast to the long-lasting DC Ca²⁺ signaling observed in the presence of iNKT cells, the addition of medium spiked with ATP induced a single Ca²⁺ flux that typically lasted only ~1 min (Figure 2D), suggesting that degradative processes might be rapidly reducing extracellular ATP concentrations. We therefore tested the DCs and iNKT cells for expression of the cell-surface ecto-enzyme CD39, which degrades extracellular ATP. The monocyte-derived DCs consistently expressed high levels of CD39, while the iNKT cells typically showed lower but still clearly detectable CD39 expression (Figure 2D). To test CD39 ecto-enzyme functionality, we incubated the iNKT cells or DCs in medium spiked with ATP in the presence or absence of a specific inhibitor of the CD39 enzyme (ARL67156). Incubation of the ATP-spiked medium with DCs for even 5 min resulted in almost complete abrogation of ATP signal, and this was prevented by the addition of the CD39 inhibitor (Figure 2E). Consistent with their lower CD39 expression levels, exposing the ATP-spiked medium to iNKT cells for 5 min resulted in only an ~50% loss of ATP signal (Figure 2E). These results suggest that extracellular ATP secreted during iNKT-DC interactions is rapidly degraded by cell surface CD39 enzyme activity. However, the recurrent cytoplasmic Ca²⁺ fluxing that we observed microscopically in individual DCs suggests that ATP is repeatedly released and then removed during iNKT-DC interactions.

Induction of Tissue Inflammation

P2X₇ signaling in monocytic cell types has been associated with activation of the NLRP3 inflammasome, leading to caspase-1-mediated cleavage of the IL-1β pro-peptide and release of the inflammatory cytokine IL-1β (Dubyak, 2012), as well as inflammasome-independent biosynthesis of prostaglandin E₂ (PGE₂) (Barberà-Cremades et al., 2012). To investigate whether the interaction between human iNKT cells and monocyte-derived DCs leads to inflammation in vivo, we used a simple bioassay that enables quantitation of a localized inflammatory response (Carrodeguas et al., 1999; Jankowska-Gan et al., 2013). Briefly, purified human cells are injected into the rear footpads of highly immune-deficient mice (non-obese diabetic severe combined immunodeficiency [NOD-SCID] IL-2Rγ _c^−/− or “NSG” mice), and the amount of swelling due to edema (an indicator of local vascular permeabilization) is determined 24 hr after the injection. Using a pressure-sensitive caliper, our footpad thickness measurements are quite consistent, with percent coefficients of variance (%CVs) ranging from 0.5% to 2.0%, allowing us to reliably detect a response when swelling persists that is >0.05 mm. This approach allows us to detect even moderate inflammatory responses and also assess infiltration of the local tissue by murine neutrophils.

Injection of DCs alone produced no significant edema after 24 hr, appearing similar to injection of a sterile PBS solution (Figure 3A). In contrast, co-injection of DCs and iNKT cells consistently produced detectable footpad swelling (mean ± SD, 0.154 ± 0.066 mm) that was significantly greater than that from DCs alone or the modest swelling responses (mean ± SD, 0.053 ± 0.027 mm) resulting from injection of iNKT cells alone (Figure 3A). These experiments were performed with human cells that were sterilely maintained and not exposed to microbial ligands, suggesting that iNKT or DC activation by microbial compounds was not required for the inflammatory response. Co-injection of iNKT cells and LPS-pretreated DCs produced swelling responses (mean ± SD, 0.140 ± 0.102 mm) that were similar in magnitude to those observed from iNKT cells and DCs in the absence of LPS (Figure 3A). Thus, injection of human iNKT cells and DCs was sufficient to induce a local inflammatory response without requiring stimulation from microbial compounds, and treatment of the DCs with a toll-like receptor (TLR) ligand did not clearly enhance the response.

Figure 3. Induction of Tissue Edema by Human iNKT Cells and DCs.

(A) Compiled results from footpad inflammation bioassay experiments. Human iNKT cells and DCs were suspended in sterile PBS and injected alone or in combination into rear footpads of NSG mice. The plots show net swelling (post-injection footpad thickness minus pre-injection thickness) measured 24 hr after the injection; each symbol represents the result from an independent experiment. Where indicated, the human cells were maintained under sterile conditions and not exposed to LPS, or the DCs were pretreated with 250 ng/ml LPS for 2 hr and then washed prior to the injection.

(B) Compiled results from experiments performed in parallel where the DCs were co-injected with iNKT cells or with an autologous polyclonal T cell line.

(C) Compiled results from experiments where iNKT cells and DCs were co-injected in the presence or absence of a blocking antibody against human CD1d.

(D) iNKT cells were co-injected with DCs that were either mock treated or pretreated with the P2X₇ inhibitor KN-62.

Consistent with our observations that exposure to autologous polyclonal T cells did not activate Ca²⁺ signaling in the DCs, we found that co-injection of polyclonal T cells and autologous DCs produced little detectable swelling (Figure 3B). We tested whether iNKT+DC-induced swelling responses were inhibited by inclusion of a monoclonal antibody that is specific for human CD1d and does not cross-react with murine CD1d. We consistently observed diminished swelling responses when the iNKT cells and DCs were co-injected in a solution containing an anti-human CD1d blocking antibody (Figure 3C), suggesting that direct iNKT interactions with the human DCs were required. Pre-treating the DCs with KN-62 and then washing them before co-injecting DCs with iNKT cells resulted in significantly diminished swelling compared to iNKT co-injection with vehicle pretreated DCs (Figure 3D). Thus, the tissue swelling responses induced by injection of iNKT cells and DCs appeared to result from purinergic signaling induced in the human DCs.

Mechanism of iNKT-Induced DC Inflammatory Effects

To investigate what factors might be responsible for the swelling responses, we tested whether exposure of DCs to iNKT cells is associated with release of IL-1β protein. Compared to DCs incubated alone, which did not typically produce detectable secreted IL-1β, >2-fold increased levels of IL-1β protein were detectable in culture supernatants of DCs co-incubated with iNKT cells in ~50% of our experiments (Figure 4A, right plot). There was no detectable IL-1β in supernatants from iNKT cells activated by an anti-CD3 antibody, indicating that the IL-1β was produced by the DCs (Figure 4B). Pretreating the DCs with the YVAD peptide inhibitor of caspase-1 activity (which prevents release of mature IL-1β protein) did not cause reduced swelling responses compared to control DCs (Figure 4B), although this treatment did significantly (p < 0.02) reduce the amount of secreted IL-1β detected by ELISA. These results suggested that release of IL-1β by the DCs was not required for their inflammatory effects in vivo.

Figure 4. Role of IL-1β.

(A) Left plot shows compiled results from multiple independent analyses of secreted IL-1b in culture supernatants of DCs alone or co-incubated with iNKT cells. Right plot shows fold increase in secreted IL-1β in the presence of iNKT cells compared to DCs alone.

(B) Analysis of secreted IL-1β in supernatants of iNKT cells stimulated with plate-bound anti-CD3 or an isotype control mAb. The plot shows the means and SDs of three replicates.

(C) Analysis of footpad swelling after co-injection of iNKT cells with DCs that were either mock treated or pretreated with YVAD peptide (an inhibitor of caspase-1).

In contrast, pretreatment of the DCs with a glucocorticoid drug (dexamethasone) completely abrogated the swelling response (Figure 5A). Similarly, pretreatment of the DCs with an inhibitor of NADPH oxidase (VAS2870) prevented the response (Figure 5A). These results suggested that the tissue edema resulting from iNKT-DC interactions results from release of non-protein inflammatory mediators by the DCs. We therefore tested iNKT cells and DCs for production of prostaglandin E₂ (PGE₂), which is a central lipid mediator of inflammation. Analysis of supernatants from iNKT-DC co-cultures consistently revealed elevated PGE₂ compared to either cell type alone (Figure 5B). When we pretreated the DCs with a specific cyclooxygenase-2 (COX-2) (PTGS2) inhibitor and then washed them prior to co-incubation with iNKT cells, the PGE₂ production was abrogated (Figure 5B). This indicated that the PGE₂ was produced by the DCs (and not the iNKT cells) and that its biosynthesis was mainly due to COX-2 (PTGS2), an inducible enzyme that is responsible for sustained production of prostaglandins such as PGE₂. Consistent with this, we found that co-incubation of DCs with iNKT cells led to the upregulation of transcript for COX-2 (PTGS2), whereas anti-CD3 monoclonal antibody (mAb)-activated iNKT cells showed no significant upregulation of COX-2 (PTGS2) transcript (Figure 5C). DCs that were activated by an anti-CD40 mAb showed stronger COX-2 (PTGS2) PCR signal than untreated DCs; however, the signal from iNKT-DC co-cultures was significantly stronger than that from anti-CD40 mAb-treated DCs (Figure 5C). These results suggested that signals from iNKT cells turn on COX-2 (PTGS2) gene transcription in DCs and drive their secretion of PGE₂.

Figure 5. Role of Non-protein Inflammatory Mediators.

(A) Analysis of footpad swelling after co-injection of iNKT cells with DCs that were either mock treated or pretreated with dexamethasone or an inhibitor of NADPH oxidase (VAS2870).

(B) Detection of prostaglandin E₂ (PGE₂) from culture supernatants of iNKT cells and DCs, as quantitated by ELISA. Left plot shows compiled results from multiple independent experiments, with each symbol showing the mean PGE₂ detected from three replicate samples. Right plots shows secreted PGE₂ from co-cultures of iNKT cells that were co-incubated with mock treaded or cyclooygenase-2 (COX-2) (PTGS2) inhibitor (COX inhib)-pretreated DCs. Bars represent means and SDs of three replicates.

(C) iNKT cells and DCs were cultured as indicated for 24 hr, and total cDNA was then prepared and subject to PCR amplification using primers for human COX-2 (PTGS2). PCR product band intensities were normalized by the intensity of the GAPDH PCR product band amplified in parallel from the same cDNA preps. Similar results were observed in four independent experiments.

(D) Results from three independent, paired footpad swelling experiments where iNKT cells were co-injected with DCs that were either mock treated or pretreated with COX-2 (PTGS2) inhibitor.

(E) Analysis of footpad swelling detected at the indicated times after injection of synthetic PGE2 (left plot) or iNKT cells and DCs (right plot). Plots show the mean and SD of swelling of two to seven independently injected footpads.

To investigate the impact of this pathway on the in vivo inflam-matory response resulting from iNKT-DC interactions, we carried out a paired analysis in which DCs were pretreated with COX-2 (PTGS2) inhibitor or vehicle and then washed and co-injected with iNKT cells into NSG mouse footpads in a side-by-side manner (i.e., one footpad received COX-inhibitor-treated DCs, and the other received vehicle control DCs). This analysis revealed a partial, but reproducible, reduction in swelling at 24 hr for footpads that received DCs pretreated with the COX-2 (PTGS2) inhibitor (Figure 5D). We also noted that injection of sterile PBS containing synthetic PGE₂ into NSG mouse foot-pads was sufficient to cause a detectable tissue swelling response but that this response was comparatively short-lived, and by 24 hr post-injection, there was consistently no detectable swelling (Figure 5E, left plot). In contrast, edema resulting from the iNKT-DC mixture typically remained detectable for 48–72 hr (Figure 5E, right plot), suggesting a sustained inflammatory interaction.

Recruitment and Activation of Neutrophils

To further investigate the inflammation resulting from co-injection of iNKT cells and DCs, we used immunohistochemistry to assess tissue infiltration by murine neutrophils. Tissue sections from footpads that were injected with PBS alone showed no elevated staining for Ly-6G, a marker of murine neutrophils (Figure 6A, left image). Tissue sections from footpads that were injected with DCs or iNKT cells alone typically showed only slightly elevated Ly-6G staining (Figure 6A, middle image). However, in footpads that had been injected with a combination of iNKT cells and DCs, areas of the subdermal tissues that showed extensive neutrophilic infiltration were typically observed (Figure 6A, right image). Quantitation of the amount of Ly-6G staining in whole-footpad sections revealed significantly greater signal in footpads that were injected with the iNKT+DC combination compared to those injected with either cell type alone (Figure 6B). Moreover, Ly-6G staining of iNKT+DC-injected footpad sections was significantly diminished when the DCs were pretreated with the KN-62 inhibitor of purinergic signaling, but not by pretreatment with the YVAD inhibitor of caspase-1 (Figure 6C). These results suggested that the inflammatory pathway resulting from iNKT-mediated induction of purinergic signaling in DCs leads to the secretion of factors that promote neutrophil recruitment.

Figure 6. Neutrophil Recruitment.

(A) Immunohistochemical analyses of tissue sections from representative iNKT or DC injected footpads stained with an antibody against Ly-6G to detect murine neutrophils. Photographs were taken by light microscopy at 10× magnification.

(B) Compiled results showing quantitation of Ly-6G staining detected in longitudinal sections of whole footpads injected with the indicated human cells. Each symbol represents the amount of staining quantitated from a tissue section from one independent injection experiment.

(C) Compiled results from quantitation of Ly-6G staining of tissue sections of whole footpads injected with iNKT cells and mock-treated DCs or DCs pretreated with the YVAD caspase-1 inhibitor peptide or with the P2X₇ inhibitor KN-62.

(D) Plots showing results from a representative microfluidics assay of directional migration of human neutrophils in response to sample ports containing medium alone, culture supernatant from iNKT cells alone, DCs alone, iNKT+DC co-culture, or culture medium with added leukotriene B₄ (LTB₄). Migration of neutrophils was assessed by time-lapse microscopy, and tracks of individual cells were determined using Je’Xperiment software (Berthier et al., 2010). The origins of individual neutrophil migration tracks were superimposed at the center of the plots. Tracks where the neutrophils migrated toward the sample port are colored red, and those where the neutrophils migrated away are shown in yellow.

(E) Compiled results from multiple independent microfluidic migration experiments performed as described in (D), showing the chemotactic indices of human neutrophils in response to the indicated samples.

We therefore tested whether secreted factors are produced during iNKT-DC interactions that promote directional neutrophil chemotaxis. Purified human neutrophils were placed into fibrinogen-coated microchannels of a molded microfluidics device that allows for the formation of chemoattractant gradients (Berthier and Beebe, 2014). Reservoirs connected to the microchannels were filled with filtered culture supernatants from iNKT cells alone, DCs alone, or iNKT-DC co-cultures or with culture medium alone as a negative control or medium containing leukotriene B₄ (LTB₄) as a positive control. Time-lapse microscopy was used to evaluate directional migration of the neutrophils along the microchannels toward the reservoirs containing the culture supernatants or control media. As shown in Figure 6D, supernatant from iNKT-DC co-cultures induced robust directional migration of the neutrophils, whereas little or no migration was observed in response to supernatants from iNKT cells or DCs alone. Calculation of the chemotactic indexes from multiple independent experiments confirmed that iNKT-DC co-culture supernatants induced significantly more directional migration than supernatants from either cell type alone (Figure 6E).

To investigate whether the iNKT-DC interaction leads to the activation of murine neutrophils recruited to injected footpads, we made use of a reagent (680 FAST) that is a specific substrate for neutrophil elastase and that releases a fluorescent product upon cleavage by this enzyme. We consistently observed greater fluorescent signal when this reagent was injected into murine footpads that were co-injected with iNKT cells and DCs than for footpads injected with either cell type alone (Figure 7A). Because this observation suggested that the interaction of iNKT cells and DCs leads to local neutrophil activation, we investigated whether the presence of iNKT cells and DCs can enhance host defense against Candida albicans, an opportunistic pathogen that is initially controlled mainly by neutrophils. Mice were injected in a rear footpad with 5 × 10⁶ C. albicans cells alone, or the Candida cells were co-injected with iNKT cells, DCs, or iNKTs plus DCs. After 72 hr, tissues were harvested, homogenized, and plated out in a series of dilutions to determine the Candida colony-forming units. Compared to Candida injected alone, or Candida injected with only iNKT cells or only DCs, injection of Candida with both iNKT cells and DCs resulted in significantly reduced colony-forming units in footpads (Figure 7B). A similar trend was observed with Candida colony counts from the murine kidneys, but the results from this analysis were more variable and did not always reach statistical significance (data not shown). We also noted that despite the reduced fungal burden, footpad swelling was significantly higher when iNKT cells and DCs were present in addition to Candida (Figure 7C), suggesting that the inflammatory pathway induced by the human cells is not redundant to that induced by C. albicans.

Figure 7. Neutrophil Activation.

(A) Imaging analyses showing fluorescence signal from a dye that is activated by elastase enzyme secreted by activated neutrophils. Footpads were injected with the indicated human cell types in the presence of the 680 FAST dye and fluorescent imaging was performed after 24 hr using an IVIS pre-clinical imager. Similar results were obtained in nine independent experiments.

(B) Murine footpads were injected with 5 × 10⁶ C. albicans yeast alone or in the presence of the indicated human cells. Harvested footpads were homogenized and plated out to determine fungal colony counts. Each symbol represents the results from analysis of an individual footpad.

(C) Murine footpads were injected as described in (B), and footpad swelling was measured after 24 hr. Each symbol represents the swelling measured for one footpad.

(D) Neutrophils were isolated from bone marrow of C57Bl/6 mice and co-incubated with C. albicans yeast in the presence of filtered supernatants from iNKT cells alone, DCs alone, or an iNKT-DC co-culture or with the culture medium used for the human cells. The plot shows compiled results from five independent experiments performed at neutrophil:Candida ratios of 1:1 or 2:1. Each symbol represents the mean of six to ten replicate samples.

Finally, to confirm that the effects we observed were due to secreted factors produced during the iNKT-DC interaction and not simply an artifact of the injected human cells compensating for a deficiency in the ability of the neutrophils of NSG mice to control the Candida infection, we tested the impact of filtered supernatants from iNKT-DC co-cultures on killing of Candida by wild-type (C57Bl/6) neutrophils in vitro. Although baseline Candida lysis by the wild-type neutrophils in this assay was quite robust, we did reproducibly observe a slight enhancement of killing in the presence of iNKT-DC supernatants (Figure 7D). These results underscore the ability of secreted factors produced during sterile iNKT-DC interactions to promote protective anti-fungal responses by neutrophils.

DISCUSSION

Our results demonstrate that in the absence of microbial ligands, IL-2-activated human iNKT cells induce sustained purinergic signaling in human-monocyte-derived DCs that results in DC production of inflammatory mediators and produces a local inflammatory response. While previous studies have focused on cytokine secretion by iNKT cells in response to lipid antigens presented by CD1d molecules, these findings suggest that iNKT cells also stimulate DCs in a way that induces their de novo biosynthesis of inflammatory lipid mediators such as PGE₂. Prior studies have indicated that activated iNKT cells express cell-surface CD40L and that they thus stimulate DC cytokine secretion by ligating CD40 (Fujii et al., 2004; Kitamura et al., 1999). We hypothesize that in our system, the IL-2-activated iNKT cells stimulated the DCs to secrete ATP, which was in turn responsible for activating DC purinergic signaling, leading to their cytoplasmic Ca²⁺ fluxing. Interestingly, we have found that blocking CD40L during iNKT cell interactions with DCs does not inhibit the DC Ca²⁺ signaling response (Figure S2), suggesting that the ATP secreted during iNKT-DC interactions is due to a distinct receptor-ligand pathway. While the molecules responsible for this pathway remain unknown, it is intriguing to speculate that TCR-mediated ligation of CD1d might play an important role, since this has been shown to directly activate human monocytic cells (Yue et al., 2005).

Consistent with iNKT cells engaging a unique DC signaling pathway, we observed that in the absence of added antigens, autologous T cells failed to activate similar DC inflammatory responses. In contrast, when cognate peptide antigens were supplied (e.g., tetanus toxoid peptides), we did consistently observe detectable footpad swelling responses (Figure S3). However, whereas iNKT cells injected with an anti-CD3 mAb failed to induce a swelling response, anti-CD3 stimulation of polyclonal T cells produced a robust swelling response (Figure S4). Thus, while the ability to induce tissue inflammation is not unique to iNKT cells and DCs, their interaction is distinctive in that it does not require the addition of a foreign antigen, and the inflammation is mediated by factors produced by the DCs and not the iNKT cells.

Also worth noting is that the inflammatory outcome we observed here is likely dependent on the specific characteristics of monocyte-derived DCs, which are considered to be highly inflammatory APCs. We speculate that similar interactions between iNKT cells and other types of APCs may result in the production of distinct lipid mediator profiles and that in some cases these may be anti-inflammatory. For example, we have previously observed that iNKT cells induce freshly isolated human monocytes to differentiate into APCs that have non-inflammatory properties in a similar bioassay of tissue inflammation (Hegde et al., 2009). Additionally, it has recently been observed by others that murine iNKT cells can interact with peritoneal exudate macrophages to promote the resolution of a sterile inflammatory response (Zeng et al., 2013). Thus, iNKT cells may contribute to either the induction or resolution of tissue inflammation, depending on the nature of the APCs with which they interact and the lipid mediator profile synthesized by that type of APC.

Our results also suggest that the pathway of inflammation induced by iNKT-DC interactions is not replaced by the inflam-matory response resulting from exposure to a microbial challenge (C. albicans). C. albicans is an opportunistic fungal pathogen that is often found on our epithelial surfaces. Candida infections are initially combatted mainly by phagocytic cells, particularly neutrophils (Miramón et al., 2013). In our analysis, the NSG mice showed a robust local inflammatory response after injection of Candida into the footpad. Nevertheless, co-injection of human iNKT cells and DCs led to an additive footpad swelling response and to reduced fungal colony counts. While we cannot rule out that the co-injected human cells had a direct effect on fungal viability in vivo, addition of iNKT-DC supernatants to in vitro analyses of Candida killing by murine neutrophils indicate that secreted products produced during sterile iNKT-DC interactions enhance neutrophil cytotoxic activity. Thus, we hypothesize that the inflammatory response resulting from sterile iNKT-DC interactions provides signals beyond those induced by the infection itself, and this results in enhanced anti-fungal protection. Important questions for further research include determining how inflammation initiated by innate immune cells differs from pathogen-induced inflammation in regards to the nature of the inflammatory mediators produced and understanding the extent to which iNKT-DC-mediated pathways contribute to pathogenic or protective inflammation in vivo.

EXPERIMENTAL PROCEDURES

Monocyte-Derived DCs

Human monocytes were isolated from peripheral blood mononuclear cells (PBMCs) of healthy subjects by magnetic sorting using anti-CD14 beads (Miltenyi Biotech). Monocyte-derived DCs were prepared by culturing the monocytes for 3–4 days at 37°C and 5% C0₂ in medium (RPMI 1640, 2 mM L-glutamine, 100 μg/ml penicillin and streptomycin, and 10% fetal bovine serum [FBS]) containing 300 U/ml granulocyte-macrophage colony-stimulating factor (GM-CSF) (Berlex Labs) and 200 U/ml IL-4 (PeproTech). All protocols involving the collection and use of human tissues were approved by the University of Wisconsin Minimal Risk institutional review board (IRB), and written informed consent was obtained from all blood donors.

Primary T Cells

Monocyte-depleted PBMCs from a healthy donor (400 × 10⁶ cells) were labeled with PBS-57-loaded human CD1d tetramer from the NIH tetramer facility and subjected to two rounds of magnetic sorting using anti-PE (phycoerithrin) beads (Miltenyi Biotech), yielding a total of 0.5 × 10⁶ cells. Cells that passed over the column without being removed were collected as source of iNKT-depleted T cells. The purified cells were incubated for 3 days at 37°C and 5% CO₂ in IL-2-containing medium (RPMI 1640, 10% bovine calf serum from HyClone, 3% human AB serum from Gemini, 2 mM L-glutamine, and 100 μg/ml penicillin and streptomycin). Flow cytometric analysis showed that ~65% of the cells in the iNKT-enriched fraction stained positively by PBS-57 loaded human CD1d tetramer.

Cultured Human T Cells

Human CD1d-restricted iNKT cell clones were generated as previously described (Brigl et al., 2006). Briefly, iNKT cells were sorted from blood of healthy subjects using α-GalCer-loaded CD1d-Fc tetramers and then expanded using phytohemagglutinin (PHA) and irradiated PBMCs in IL-2-containing medium (RPMI 1640, 10% bovine calf serum from HyClone, 3% human AB serum from Gemini, 2 mM L-glutamine, and 100 μg/ml penicillin and streptomycin). Results presented here are from experiments using the previously established iNKT clones J3N.5, J24N.22, J24L.17, and JC2.8; similar results were obtained using two different polyclonal iNKT cell cultures that each have >90% purity. Polyclonal T cells used as negative controls were isolated from PBMCs of healthy donors by magnetic sorting using pan-T cell negative selection beads (Miltenyi Biotech), which remove most of the iNKT cells. The polyclonal T cells were expanded in culture in a manner identical to that used to grow the iNKT cell clones (i.e., using PHA, irradiated PBMCs, and IL-2).

Analyses of Intracellular Ca²⁺

Cytoplasmic Ca²⁺ levels in DCs were assessed using three different approaches. (1) Time-lapse fluorescence microscopy was performed using a 1:1 ratio of Fluo-4-labeled DCs and CMTPX-labeled iNKT cells that were allowed to settle onto poly-L-lysine-coated glass slides. Microscopic analysis was performed in a 37°C and 5% CO₂ chamber, with images taken by a Nikon Ti-Eclipse inverted wide-field microscope every 30 s. Data analysis was carried out using NIS Elements software (Nikon) version 4.13.04. (2) DCs were labeled with Fluo-4 and mixed with a 1:1 ratio of unlabeled iNKT cells, and 2 × 10⁵ cells were added per well of a 96-well black-walled plate (Costar) and analyzed for 30–60 min at 37°C using a Synergy HT fluorescence plate reader (BioTek Instruments). (3) DCs were labeled with Fluo-4 and FuraRed (Invitrogen) and mixed at a 1:1 ratio with unlabeled iNKT cells. The cells were spun down for 2–3 min to initiate contact, incubated at room temperature for 2 min, and then vortexed and analyzed using a BD LSRII flow cytometer.

Determination of Secreted ATP

Culture supernatants from iNKT cells alone, DCs alone, or iNKT-DC co-cultures were harvested and tested for ATP using the Kinase-Glo luminescence assay kit (Promega) according to the manufacturer’s instructions. Where indicated, 100 μM ARL67156 CD39 inhibitor was included in the cell cultures. Medium alone and medium containing 1 mM ATP were used as negative or positive controls, respectively.

Induction of Tissue Inflammation

Tissue inflammation studies were performed using NOD.Cg-Prkdc^scid Il2rg^tm1Wjl/SzJ mice (The Jackson Laboratory). Footpads were measured prior to injection using a dial thickness caliper gauge (Swiss Precision Instruments). iNKT cells and DCs (1–3 × 10⁵ of each) were mixed at a 1:1 ratio and pelleted by centrifugation, re-suspended in 50 μl sterile PBS, and injected subcutaneously into rear footpads using a 0.5-cc insulin syringe (28G × 0.5 in). Injected footpads were re-measured after 24 hr. The thickness of each footpad prior to injection was subtracted from the post-injection value to obtain the net swelling. Where indicated, the DCs were pretreated with the following inhibitor drugs: 20 μM NS-398 (COX-2 (PTGS2) inhibitor, Cayman Chemical), 100 μg/ml dexamethasone (glucocorticoid, Sigma-Aldrich), 20 μM VAS2870 (NAPDH oxidase inhibitor, Enzo Life Sciences), 100 μM Suramin (P2 receptor inhibitor, Cayman Chemical), and 30 μM KN-62 (P2X₇ inhibitor, Sigma-Aldrich).

Detection of PGE₂

iNKT cells and DCs were cultured at 37°C and 5% CO₂ either alone or together at a 1:1 ratio (1 × 10⁵ cells of each) for 24 hr. PGE₂ content in culture supernatants was quantified using the DetectX PGE₂ EIA kit (Arbor Assays). Additionally, supernatants were analyzed by liquid chromatography-mass spectrometry (LC-MS) as follows. Freeze-dried supernatants were extracted using chloroform:methanol followed by an acidified methanol step. Extracts were purified on a SepPak Classic C18 cartridge (Waters Corp) with methyl formate elution. Eluates were chromatographically resolved on an Agilent ZORBAX 300SB-C18 column run on an Agilent 1200 high-performance liquid chromatography (HPLC), using a linear gradient of 70% water:30% acetonitrile:0.1% formic acid to 95% acetonitrile:5% water:0.1% formic acid. Mass spectrometry was performed using an Agilent liquid chromatography/mass spectrometry time-of-flight (LC/MSD TOF) with electrospray ionization used in negative-ion mode. Internal calibration was achieved with assisted spray of two reference masses, 112.9856 m/z and 1,033.9881 m/z. Data were processed using the Analyst QS 1.1 build:9865 software (Agilent) to extract parent masses observed in the range from 100 to 3,200 amu, and corresponding molecular formulas were generated with an integrated molecular mass calculator and the Agilent Mass Hunter Qualitative Analysis software (version B.01.03, build: 1.3.157.0).

COX-2 PCR Analysis

cDNA was prepared from RNA isolated from DCs, iNKT cells, or iNKT-DC co-cultures. PCR was performed using the primer pairs (COX-2) (PTGS2) 5′-CGAGGTGTATGTATGAGTGTG-3′ and 5′-TCTAGCCAGAGTTT CACCGTA-3′, yielding a 540-bp product. Reaction conditions were 95′ C for 2 min, 35 cycles of 94°C for 45 s, 62°C for 45 s, 72°C for 1 min, and 72°C for 10 min. Equal amounts of cDNA were used for each reaction, and PCR for GAPDH was also performed on each cDNA sample in parallel. PCR products were separated on a 1.5% agarose gel and visualized using ethidium bromide, and band intensities of the appropriate-sized products were determined using ImageJ software.

Immunohistochemical Analysis

Whole footpads were harvested and fixed in 10% neutral buffered formalin for ≥24 hr. Tissue was decalcified using formic acid and paraffin embedded. Sections (5 μm thick) were de-paraffinized and rehydrated, and antigen retrieval was performed using Rodent Decloaker solution (Biocare Medical) at 80°C for 3 hr. Sections were stained with H&E, blocked with Rodent M Block (Biocare Medical), and labeled with anti-Ly-6G (clone 1A8-Ly6g, eBioscience).

Endogenous peroxidases were quenched with 3% H₂O₂, and antibody labeling was detected using horseradish peroxidase (HRP)-conjugated reagents and then visualized with diaminobenzidine chromogen (DakoCytomation). Quantitation of neutrophil staining was done using ImageJ software to integrate the total Ly-6G staining intensity in microscopic images (2× magnification) of whole-footpad sections.

Neutrophil Chemotaxis

Human neutrophils were isolated from heparinized peripheral blood by Polymorphoprep density-gradient centrifugation (Axis-Shield), and red blood cells were lysed. Washed neutrophils were resuspended in Hank’s balanced salt solution (HBSS) containing 10% BSA at a final concentration of 2 × 10⁶ cells/ml. Neutrophil directional chemotaxis was assayed using a fibrinogen-coated molded polydimethylsiloxane (PDMS) microfluidic device, as previously described (Cavnar et al., 2012). Briefly, 5 μl of the purified neutrophil suspension (10 × 10⁴ cells) was added to a cell chamber, and 3 μl test solution was added to a chemoattractant reservoir connected to the cell chamber by a microchannel. Test solutions included filtered iNKT and DC culture supernatants, medium alone (negative control), or medium containing 2 μM synthetic LTB4 (Cayman Chemical). Neutrophil chemotaxis was assessed by time-lapse microscopy using a Nikon Eclipse TE300 microscope, with images taken every 30 s for 1 hr at 10× magnification using phase-contrast illumination. Cell migration was tracked using Je’Xperiment software (Berthier et al., 2010), and the average chemotactic index for cells in the channel was calculated as the ratio of the linear distance between the first and last point of the cellular track and the total distance traveled by the cell.

Neutrophil Elastase Imaging

Human iNKT cells alone, DCs alone, or a 1:1 ratio of iNKT cells and DCs (3 × 10⁵ of each cell type) were resuspended in 50 μl sterile PBS, and 3 μl of the neutrophil elastase 680 FAST fluorescent imaging agent (Perkin Elmer) was added. The samples were injected subcutaneously into the rear footpad of an NSG mouse, and fluorescent imaging was performed after 24 hr using an IVIS series preclinical imager (PerkinElmer).

Candida Challenge

C. albicans yeast form (strain SC5314) was grown in YPD medium (1% yeast extract, 1% peptone, and 2% dextrose) at 25°C with rotation at 20 rpm overnight. The yeast cells were washed, counted, and re-suspended in sterile PBS. Yeast cells (5 × 10⁶) were injected in the presence or absence of 3 × 10⁵ of the indicated human cells into one footpad of each mouse. Footpad swelling was measured every 24 hr. After 72 hr, the mice were sacrificed and kidneys were harvested, placed in 2 ml sterile PBS, and homogenized. Titrated dilutions of homogenized kidney tissue were plated on YPD-agar plates. The plates were incubated for 24 hr at 37°C, individual colonies were counted, and the total colony-forming unit burden in the kidney was back-calculated according to the dilution factor.

Neutrophil Killing of Candida

Neutrophils were enriched from bone marrow of C57Bl/6 mice using Percoll density-gradient centrifugation as previously described (Sterkel et al., 2016). Neutrophils and C. albicans yeast were suspended at ratios of 1:1 or 1:2 in assay culture medium (RPMI 1640, 2% FBS, and 100 μg/ml penicillin and streptomycin) containing 1/4- to 1/8-fold dilutions of filtered supernatants from cultures of iNKT cells alone, DCs alone, or iNKT cells and DCs or of T cell medium lacking cells as a negative control. After 4 hr, neutrophils were lysed with sterile distilled water, and C. albicans viability was quantified using (2,3-Bis-(2-Methoxy-4-Nitro-5-Sulfophenyl)-2H-Tetrazolium-5-Carboxanilide) (XTT) to assess redox potential of living cells, as previously described (Nett et al., 2011).

Statistical Analyses

Experimental results were analyzed using a two-tailed non-parametric t test (Mann-Whitney analysis). In cases where results are shown with individual symbol pairs connected by dashed gray lines, the data were analyzed using a two-tailed non-parametric paired t test (Wilcoxon matched-pairs test).

Highlights.

• iNKT cells induce prolonged Ca²⁺ flux in monocyte-derived DCs via P2X₇ signaling

• This leads to DC release of inflammatory mediators such as PGE₂

• Factors produced during sterile iNKT-DC interactions promote neutrophil responses