TLR Signalling Pathways Diverge in Their Ability to Induce PGE2

PGE2 is a lipid mediator abundantly produced in inflamed tissues that exerts relevant immunoregulatory functions. Dendritic cells (DCs) are key players in the onset and shaping of the inflammatory and immune responses and, as such, are well known PGE2 targets. By contrast, the precise role of human DCs in the production of PGE2 is poorly characterized. Here, we asked whether different ligands of Toll-like receptors (TLRs), a relevant family of pathogen-sensing receptors, could induce PGE2 in human DCs. The only active ligands were LPS (TLR4 ligand) and R848 (TLR7-8 ligand) although all TLRs, but TLR9, were expressed and functional. While investigating the molecular mechanisms hindering the release of PGE2, our experiments highlighted so far oversight differences in TLR signalling pathways in terms of MAPK and NF-κB activation. In addition, we identified that the PGE2-limiting checkpoint downstream TLR3, TLR5, and TLR7 was a defect in COX2 induction, while TLR1/2 and TLR2/6 failed to mobilize arachidonic acid, the substrate for the COX2 enzyme. Finally, we demonstrated the in vivo expression of PGE2 by myeloid CD11c+ cells, documenting a role for DCs in the production of PGE2 in human inflamed tissues.


Introduction
PGE 2 is the predominant eicosanoid produced in inflamed tissues and by growing tumors, with a major contribution by infiltrating immune cells [1,2]. Because PGE 2 promotes vasodilatation and accumulation of proinflammatory cells, it is generally recognized as a mediator of active inflammation. However, by suppressing the production of some proinflammatory cytokines, PGE 2 also limits nonspecific inflammation and fosters the immune suppression associated with chronic inflammation and cancer [1,2]. Despite the fact that PGE 2 targeting is easily done by common and effective pharmaceutical agents (i.e., steroids and nonsteroid antiinflammatory drugs), an accurate understanding of PGE 2 regulation and mechanisms of action is crucial to fully deploy the therapeutic potential of these drugs.
The inflammatory synthesis of PGE 2 is regulated by three classes of enzymes: cytosolic phospholipase A 2 (cPLA 2 ) family members that mobilize arachidonic acid (AA) from cellular membranes, cyclooxygenases (COX1 and COX2) that convert AA into PGH 2 , and specific synthases accounting for the final conversion of PGE 2 [2]. While COX1 is housekeeping gene governing homeostatic PGE 2 production, COX2 is potently induced by proinflammatory stimuli [3]. In inflammation, the rate of PGE 2 production largely depends on the expression and activity of COX2, although it can be affected by other factors such as local availability of AA [2].
Dendritic cells (DCs) are professional antigen presenting cells responsible for the activation of the adaptive immune response [4] and also play a crucial role in the regulation of inflammation [5,6]. For doing this, DCs are equipped with the vastest repertoire of pathogen-sensing receptor (pattern recognition receptors, PRR) such as NOD-like receptors, C-type lectin receptors, and Toll-like receptors (TLRs) [7][8][9][10]. 2

Mediators of Inflammation
Human TLRs are a family of type I transmembrane proteins [11]. Upon microbial recognition, TLRs recruit a specific set of adaptor molecules, such as MyD88 and TRIF, to initiate downstream signal transduction pathways. MyD88 is used by all TLRs except TLR3 and activates the transcription factor NF-B and mitogen-activated protein kinases (MAPK) to induce inflammatory cytokines. By contrast, TRIF is used by TLR3 (and TLR4) and induces the secretion of type I interferons and also some NF-B-depending genes [11,12]. The TLR signalling cascades have been described using murine cells from knockout animals or immortalized cell lines of tumor origin. As a result, little is known about the pathways and cellular responses activated by TLRs in human primary cells.
Given the importance of PGE 2 in the orchestration of the immune and inflammatory responses, we set out to dissect the molecular mechanisms underlying its release by TLRspecific ligands in human DCs.

PGE 2 and CXCL8
Determination. DCs were incubated for 24 h with the indicated treatments. Cell-free supernatants were harvested and PGE 2 production was measured by EIA (Cayman Chemical) kit. Secreted CXCL8 was measured by ELISA assay according to the manufacturer instructions (R&D Systems, Minneapolis, MN, USA).

SDS-PAGE and Western Blot.
Following the designated treatments, DCs were washed twice with PBS and lysed in L1 buffer (50 mM Tris-HCl, pH 8.0; 2 mM EDTA; 0.1% NP-40 and 10% glycerol) with inhibitors to separate cytoplasmic proteins. Nuclear pellets were washed twice with L1 buffer with inhibitors and then lysed in NP-40 Lysis buffer (50 mM Tris-HCl, pH 8.0; 250 mM NaCl; 1 mM EDTA; 0.1% NP-40; and 10% glycerol) with inhibitors. Total cell extracts were obtained with NP-40 Lysis buffer. Equal amounts of cytoplasmic, nuclear, or total extracts were analysed through 8-12% SDS-PAGE followed by Western blotting with antibodies against COX2 (mouse monoclonal, Cat. 160112, Cayman Chemical), phospho-ERK1/2 (rabbit polyclonal, Cat. 9101, Cell Signalling Technologies, Massachusetts, USA), phospho-p38 (rabbit polyclonal, Cat. 9211, Cell Signalling), phospho-cPLA 2 (rabbit polyclonal, Cat. 2831, Cell Signalling), phospho-MSK1 (rabbit polyclonal, Cat. 9595, Cell Signalling), NF-B p65 (rabbit polyclonal, C-20 Cat. sc-372, Santa Cruz Biotechnology),actin (mouse monoclonal, C-4 Cat. sc-44478, Santa Cruz Biotechnology), and Lamin B (goat polyclonal, C-20 Cat. sc-6216, Santa Cruz Biotechnology). Protein bands were detected with SuperSignal West Pico Chemiluminescent Substrate (Pierce, Rockford, USA). Densitometric analysis was performed using ImageJ (version 1.48) software package from National Institutes of Health. Immunoblots were scanned as JPEG images and the areas under the curves were measured for each band and quantified. Data were normalized based on -actin or Lamin B content.  , TLR2/6 (FSL-1), TLR3 (Poly I:C), TLR5 (Flagellin), TLR7 (Imiquimod), and TLR9 (CpG) were by contrast ineffective. Thus, we asked whether these receptors were expressed and functional in DCs. Figure 1(b) shows that DCs express all TLR mRNAs, exception made for TLR9. While the absence of TLR9 in myeloid DCs is generally recognized, the expression of TLR7 is controversial [14][15][16]. However, since TLR1-8 ligands activated DCs to produce CXCL8 (Figure 1(c)), we concluded that these receptors were indeed expressed and functional in our experimental setting. CpG was excluded from further analysis because of the lack of its cognate receptor. Previous works have compared TLR ligands for their capability to induce differential cytokine production by human DCs [17][18][19][20][21][22]. To our knowledge, our study is the first to investigate eicosanoid production induced by different TLRs and to highlight so far oversight differential ability of TLR ligands to induce the release of PGE 2 in human DCs.

MAPKs and NF-B Are Key Downstream Signalling
Molecules for PGE 2 Production. LPS, a ligand inducing robust PGE 2 secretion over a vast range of concentrations (Figure 2(a)), was used to further investigate the signalling pathways responsible for PGE 2 production in DCs. Since in other experimental settings the regulation of PGE 2 involves the activation of MAPKs and NF-B [3,23,24], DCs were stimulated in the presence of specific MAPK inhibitors such as U0126 (inhibitor of MAPK kinase), PD98059 (inhibitor of extracellular-signal-regulated kinase 1/2-ERK1/2), SB203580 (inhibitor of MAPK p38), and, of the NF-B inhibitor, BAY11-7082, which all significantly reduced the release of PGE 2 (Figure 2(b)) when used at the lower concentration, without affecting cell viability (not shown). Of note, the same signalling pathways also regulated the induction of COX2, the rate-limiting enzyme for PGE 2 synthesis (Figure 2(c)).
These results confirm that, in human DCs, the release of PGE 2 depends on the activation of the MAPK and NF-B pathways.

TLR Ligands Differentially Activate MAPKs and NF-B in
Human DCs. In order to clarify the molecular mechanisms hindering the release of PGE 2 by inactive TLR ligands, we analysed how different TLR stimulation impacted the activation of MAPKs and NF-B. Figure 3 shows that all ligands induced ERK1/2 phosphorylation, although at different extent. However only LPS, R848, PAM 3 CSK 4 , and FSL-1 also induced p38 phosphorylation and NF-B p65 nuclear translocation, while Poly I:C, Flagellin, and Imiquimod did not. Finally, TLR2 ligands failed to phosphorylate MSK1, a 4 kinase downstream ERK and p38 MAPK that was described to play a role in PGE 2 production [25,26]. Similar activation patterns were also detected at 15 and 60 minutes after stimulation (not shown). Such striking differences in the activation of MAPKs and NF-B are interesting because, according to the literature, all TLR agonists are expected to converge on these pathways to exert their biological effects [27,28]. Our results underline the importance to confirm and refine previous findings, obtained in model cell lines and often by transfection, in primary cells expressing TLRs at physiological levels.

Mediators of Inflammation
Of particular interest was the difference in the activation induced by Imiquimod and R848, both in terms of intracellular signalling and in terms of PGE 2 secretion. TLR7 and TLR8 both recognize ssRNA, are similar in sequence and localization, and, together with TLR9, form an evolutionary related TLR subfamily sharing common signalling pathways responsible for antiviral responses [28]. The differences we have observed may thus merely depend on the limited expression of TLR7, which would explain the lower PGE 2 secretion induced by Imiquimod. However, Imiquimod phosphorylated ERK1/2 at levels that were comparable to, if not exceeding, those induced by R848, despite the fact that it failed to activate other signalling molecules. This may unveil a qualitative rather than a quantitative difference between the signalling pathways activated by TLR7 and TLR8, as suggested by other authors [20,29]. In addition, we hypothesized that R848, by concomitantly triggering TLR7 and TLR8, may activate a synergy between the two signalling pathways. This issue represent an interesting area of investigation that will be intensively pursued.  pathways could differentially affect COX2 activation. Because COX2 activity directly correlates with protein levels [3], Western blot analysis was used to address this issue. Figure 4(a) clearly shows that Poly I:C, Flagellin, and Imiquimod failed to induce COX2 accumulation, which fully explains the lack of PGE 2 secretion and also confirms that NF-B activation is critical for COX2 expression [23,24]. By contrast, PAM 3 CSK 4 and FSL-1 were as effective as LPS or R848 in COX2 induction, suggesting that these ligands lack in downstream steps of PGE 2 synthesis.

TLR3, TLR5, and TLR7 Stimulation
Thus, we analysed the activation of cPLA 2 , the other PGE 2 key-producing enzyme. Within minutes, cPLA 2 is regulated by phosphorylation [30]. We found that only LPS and R848 induced significant cPLA 2 phosphorylation at 30 minutes after stimulation (Figure 4(b)). Based on the observations in Figure 3, it is tempting to speculate that, in our system, cPLA 2 phosphorylation may depend on MSK1 activation, as previously demonstrated in human fibroblasts stimulated with IL-1 [25,26].
Consistent with inefficient cPLA 2 phosphorylation, PAM 3 CSK 4 and FSL-1 did not induce AA release as compared to LPS and R848 (Figure 4(c)), suggesting that TLR1/2 and TLR2/6 stimulation may fail to induce PGE 2 secretion because of the unavailability of AA, the substrate for COX2 enzyme. According to this hypothesis, the administration of exogenous AA (Figure 4(d), black bars) restored the production of PGE 2 by PAM 3 CSK 4 and FSL-1, but not by Poly I:C, Flagellin, and Imiquimod due to their inability to accumulate COX2.
Altogether, these results identify AA and COX2 accumulation as the PGE 2 -limiting checkpoints downstream TLR1/2-2/6 and TLR3-5-7 stimulation, respectively.  2 In Vivo. Despite the fact that DCs are very well known PGE 2 targets [2], their potential as prostaglandin sources in humans is less investigated and remains under debate. In fact, using in vitro differentiated DCs as a model, it was described that human DCs either could [31,32] or could not produce PGE 2 [33]. In addition, the strict ligand selectivity we have demonstrated so far raises the question of how primary DCs may respond to real pathogens in terms of PGE 2 production.

Human DCs Produce PGE
To shed light on the possible role of DCs as PGE 2producing cells in vivo, we stimulated primary, circulating mDCs with TLR4 ligands. Figure 5 cells respond to TLR4 stimulation releasing amounts of PGE 2 that are much higher than those observed for their in vitro-derived counterparts. This is in line with previous observations that IL-4 used to generate DCs may hinder the activity of cPLA 2 [33]. Finally, we performed double immunohistochemistry stainings on human inflamed lymph nodes, showing a fraction of CD11c + cells expressing PGE 2 in their cytoplasm ( Figure 5(b)). These results conform that primary DCs can actively secrete PGE 2 in inflammatory conditions in vivo.
The net effect of the simultaneous stimulation of different TLRs and also of other classes of innate immune receptors by whole microorganisms remains to be elucidated. However, it is plausible that pathogens expressing molecular patterns activating TLR4 and TLR7-8 may be stronger PGE 2 inducers. Thus, the pathogen-dictated modulation of the release of PGE 2 may represent a novel mechanism through which DCs shape the immune and inflammatory responses.

Conclusion
In the present paper, we demonstrate a differential ability of TLR ligands to induce the release of PGE 2 and provide a detailed description of the mechanisms governing TLR-mediated eicosanoid production in human DCs. A schematic representation of our findings is outlined in Figure 6. Briefly, only the stimulation of TLR4 and TLR7-8 could activate ERK1/2, p38, MSK1, and NF-B and induce PGE 2 . By contrast, the PGE 2 -limiting checkpoints downstream TLR1/2-2/6 and TLR3-5-7 stimulation were identified in AA and COX2 accumulation, respectively.
Our results also highlighted so far oversight differences in MAPK and NF-B activation by TLR ligands. These divergences may have come to light because, contrary to works performed in transfected cell lines, our system consisted of primary cells expressing a physiologic repertoire of receptors and intracellular adaptor molecules.
Finally, by demonstrating the expression of PGE 2 by CD11c + cells in human inflamed lymph nodes, this study further expands our knowledge on the complex role of DCs in the regulation of immune responses.  Figure 6: Mechanisms of PGE 2 production by TLR family members. TLR4 triggering induces the transcription of COX2 via NF-B and MAPK p38 and ERK1/2 as well as cPLA 2 phosphorylation and AA mobilization, presumably via MAPK and/or MSK1, which is in turn converted into PGE 2 by COX2 and released into the medium. Similar mechanisms of action can be envisaged when TLR7 and 8 are concomitantly activated by R848. TLR3, TLR5, and TLR7 fail to activate NF-B and selected MAPKs, thus hindering the transcription of COX2. By contrast, TLR1/2 and TLR2/6 cause no PGE 2 release because of inefficient cPLA 2 phosphorylation and AA mobilization, which may correlate with their inability to phosphorylate MSK1.