Prostaglandin‐cytokine crosstalk in chronic inflammation

Chronic inflammation underlies various debilitating disorders including autoimmune, neurodegenerative, vascular and metabolic diseases as well as cancer, where aberrant activation of the innate and acquired immune systems is frequently seen. Since non‐steroidal anti‐inflammatory drugs exert their effects by inhibiting COX and suppressing PG biosynthesis, PGs have been traditionally thought to function mostly as mediators of acute inflammation. However, an inducible COX isoform, COX‐2, is often highly expressed in tissues of the chronic disorders, suggesting an as yet unidentified role of PGs in chronic inflammation. Recent studies have shown that in addition to their short‐lived actions in acute inflammation, PGs crosstalk with cytokines and amplify the cytokine actions on various types of inflammatory cells and drive pathogenic conversion of these cells by critically regulating their gene expression. One mode of such PG‐mediated amplification is to induce the expression of relevant cytokine receptors, which is typically observed in Th1 cell differentiation and Th17 cell expansion, events leading to chronic immune inflammation. Another mode of amplification is cooperation of PGs with cytokines at the transcription level. Typically, PGs and cytokines synergistically activate NF‐κB to induce the expression of inflammation‐related genes, one being COX‐2 itself, which makes PG‐mediated positive feedback loops. This signalling consequently enhances the expression of various NF‐κB‐induced genes including chemokines to macrophages and neutrophils, which enables sustained infiltration of these cells and further amplifies chronic inflammation. In addition, PGs are also involved in tissue remodelling such as fibrosis and angiogenesis. In this article, we review these findings and discuss their relevance to human diseases.


Introduction
Upon invasion of foreign pathogens or tissue damage, the innate immune system is immediately activated in response to molecules bearing pathogen-associated molecular patterns (PAMPs) and damage-associated molecular patterns (DAMPs), recruits granulocytes to the injured tissue to clear pathogens, produces inflammatory mediators, including pro-inflammatory cytokines such as TNF-α, IL-1β and IL-6 and lipid mediators such as PGs and leukotrienes (LTs), and evokes an acute inflammatory process (hours to days) to clear the pathogens and damaged tissues. Acute inflammation is resolved and the tissue is repaired when PAMPs, DAMPs, pathogens and damaged tissues are cleared, granulocyte recruitment ceases with a down-regulation and scavenging of chemokines, and recruited granulocytes are subsequently cleared by efferocytosis. However, inflammation often becomes chronic (weeks to months to years), and this underlies various chronic disorders such as autoimmune, neurodegenerative, vascular and metabolic diseases and cancer. Recent studies in various experimental systems have begun to unravel the possible mechanisms through which inflammation is sustained and becomes chronic. They include the generation of positive feedback mechanisms that self-amplify inflammatory responses and the suppression of negative feedback mechanisms that prevent resolution, which leads to the recruitment, activation, phenotypic transformation and synergistic interaction of various types of cells and sustains pro-inflammatory cytokine signalling at inflammatory sites.
PGs including PGD 2 , PGE 2 , PGF 2α , PGI 2 and TXA 2 are produced in most tissue and cells either constitutively by physiological stimuli or in response to noxious stimuli. In either case, C20-unsaturated fatty acids such as arachidonic acid are released from phospholipids in the cell membrane and converted into PGH 2 by cyclooxygenases (COXs including COX-1 and COX-2). PGH 2 is then converted into each PG by respective PG synthases ( Figure 1A). PGs exert their actions through a family of eight types and subtypes of GPCRs, PGD receptor (originally named DP and now called DP 1 ), EP 1 , EP 2 , EP 3 and EP 4 subtypes of PGE receptor, PGF (FP) receptor, PGI (IP) receptor and TXA (TP) receptor and another PGD receptor in a different GPCR family, originally named chemoattractant receptor-homologous molecule expressed on Th2 cells (CRTH2) and now called DP 2 receptor. These PG receptors activate distinct downstream signalling pathways and thus have divergent, sometimes additive and other times opposing, functions in various physiological and pathological processes. For example, while EP 2 , EP 4 , DP 1 and IP receptors activate cAMP signalling, EP 3 and DP 2 receptors inhibit cAMP signalling. EP 1 , FP and TP receptors mainly activate the PKC and Ca 2+ pathways. TP and EP 3 receptors also activate the small G-protein Rho; EP 2 and EP 4 receptors can also activate PI3K and β-arrestin pathways ( Figure 1B). Aspirin-like non-steroidal anti-inflammatory, anti-pyretic and analgesic drugs (NSAIDs) exert their actions by targeting COX and inhibiting PG biosynthesis. PAMPs/DAMPs such as LPS and pro-inflammatory cytokines such as IL-1β and TNF-α induce the expression of inducible isoforms of COX and PGE synthase, COX-2 and microsomal PGE synthase 1 (mPGES1) respectively (Díaz-Muñoz et al., 2010). Therefore, PGs have Figure 1 PG biosynthesis, receptors and signalling pathways. (A) Arachidonic acid is metabolized by COX, either COX-1 or COX-2, to PGH 2 , which is then converted to each PG, PGD 2 , PGE 2 , PGF 2⍺ , PGI 2 or TXA 2 , by respective synthases. NSAIDs inhibit COXs and suppresses PG biosynthesis. Each PG acts on its cognate receptor to exert its actions. PGD 2 acts on PGD receptors, DP 1 and DP 2 (formerly designated as CRTH2, chemoattractant receptor-homologous molecule expressed on Th2 cells); PGE 2 acts on four subtypes of PGE receptors, EP 1 , EP 2 , EP 3 and EP 4 ; PGF 2α acts on PGF receptor, FP; PGI 2 acts on PGI receptor, IP; TXA 2 acts on TXA receptor, TP. (B) All of the nine PG receptors are GPCRs and exert their actions by modulating second messengers and intracellular signal transduction. Dependent on their downstream signalling pathways and structural similarity, they are classified into three groups. The relaxant group consists of DP 1 , EP 2 , EP 4 and IP that are mainly coupled to an elevation of intracellular cAMP levels and activate the cAMP-PKA-CREB pathway. The contractile group consists of EP 1 , FP and TP that are coupled to Ca 2+ signalling and PKC activation. Both EP 3 and DP 2 receptors are coupled to a reduction in the levels of intracellular cAMP and belong to the inhibitory PG receptor group. However, while EP 3 receptors with other PG receptors constitute the family of PG receptors, DP 2 belongs to the chemo-attractant GPCR family. In addition to these main downstream signalling pathways, each PG receptor also activates other signal transduction pathways. For example, EP 2 and EP 4 receptors activate the PI3K pathway and form a complex with β-arrestin to transactivate the EGF receptor; TP and EP 3 receptors activate the small G-protein Rho; IP activates the PKC and Ca 2+ signalling; and DP 2 receptors activate the PI3K pathway. Furthermore, EP 3 receptor has various alternatively spliced variants in its carboxyl terminus, which can couple to various signalling pathways other than the pathways mentioned (Sugimoto and Narumiya, 2007).
been traditionally thought of as inflammatory mediators linking innate immunity to acute inflammation. Indeed, studies using genetically modified mice deficient in each PG receptor identified types of PG and their receptors involved in each of the acute inflammatory responses (Narumiya and Furuyashiki, 2011). However, expression of COX-2 and mPGES1 is not limited to the sites of acute inflammation but widely seen in tissues of chronic inflammation such as the joints of rheumatoid arthritis (RA) patients, the spinal cord of multiple sclerosis (MS) patients, the colon of inflammatory bowel disease (IBD) patients, the cerebral arterial wall of cerebral aneurysm patients and tumours and their microenvironment of many cancers (Ricciotti and FitzGerald, 2011;Wang and DuBois, 2018). These findings suggest that in addition to their actions in acute inflammation, PGs also play important roles in many chronic inflammatory diseases. Recent studies using PG receptor knockout (KO) mice and PG receptor-type selective agonists and antagonists in various animal models of chronic diseases have revealed that PGs intimately crosstalk with cytokines, drive pathogenic conversion, recruit and activate inflammatory cells and contribute to chronic inflammation through various mechanisms. Furthermore, genome-wide association studies (GWASs) and gene signature analyses of disease tissues of patients strongly suggest the relevance of these experimental findings to chronic inflammation in humans. Here, we review these findings and discuss the therapeutic potential of PG-related drugs in chronic human diseases.

PG-cytokine crosstalk in immune and allergic inflammation
Acute inflammation often becomes chronic, when acquired immunity is raised against pathological antigens generated at inflammatory sites and causes immune inflammation. Antigens generated at inflammatory sites are taken up by antigen-presenting cells (APCs), such as dendritic cells (DCs) and macrophages, which are activated upon this uptake, migrate to draining lymph nodes and present the antigens to T lymphocytes to trigger the adaptive immune system. Activated APCs also produce various cytokines, which prime antigen-activated T-cells to differentiate into specific subsets of helper T (Th) cells that produce the subset-specific cytokines and trigger inflammation. The type 1 subset of Th cells (Th1) produce IFN-γ, Th2 cells produce type 2 cytokines including IL-4, IL-5 and IL-13 and Th17 cells produce IL-17A, IL-17F and IL-22. The generation of Th1, Th2 and Th17 cells is primarily to expel invaded pathogens, but activation and differentiation of these T-cells, if not properly controlled, can trigger immune diseases (Zhu et al., 2010). Th1 and Th17 cells play critical roles in the development of autoimmune inflammatory diseases such as MS, IBD, RA and psoriasis. Indeed, the accumulation of Th1 and Th17 cells and elevation of their signature cytokines have been found in the brain, synovial fluid, gut and skin of patients with MS, RA, Crohn's disease and psoriasis respectively (Zhu et al., 2010). Furthermore, administration of antibodies targeting the cytokines they produce has shown clinical efficacy in some of these diseases (Zhu et al., 2010). In contrast, Th2 cells are involved in allergic inflammation such as atopic dermatitis (AD) and asthma. As described below, a number of studies now suggest that PGs, particularly PGE 2 , contribute critically to immune inflammation by facilitating the differentiation and pathogenic conversion of Th1 and Th17 cells and that various PGs are also involved in different aspects of Th2-mediated allergic inflammation. Studies have also appeared to indicate that PGs are involved in the modulation of innate lymphoid cells (ILCs), a class of lymphoid cells mirroring T-cells but not expressing the T-cell receptor (TCR). Figure 2 summarizes the actions of PGs on different inflammatory responses driven by different subsets of T and ILC cells, which will be discussed in detail below.
PG signalling in immune inflammation mediated by Th1 and Th17 cells PGE 2 -EP 2 /EP 4 receptor signalling in Th1 differentiation. Differentiation of Th1 cells is driven by two critical cytokines, IL-12 and IFN-γ. Upon TCR engagement, T-cells produce a small amount of IFN-γ that binds to the IFN-γ receptor and induces a low level of IL-12 receptor β2 chain (IL-12Rβ2), enabling T-cell responses to IL-12. The IL-12-IL-12 receptor interaction then activates transcription factors T-bet and STAT4 and amplifies the expression of IL-12Rβ2 as well as producing a large amount of IFN-γ for Th1 differentiation (Zhu et al., 2010). PGE 2 has been believed for a long time to act as an immunosuppressant, suppressing Th1 cell development and function in in vitro studies (Betz and Fox, 1991;Harris et al., 2002). However, recent studies indicate that the inhibitory action of PGE 2 depends on the extent of TCR stimulation as well as co-existing cytokines. Substantial in vitro and in vivo evidence now shows that PGs can act as an immune-activator under numerous conditions. Indeed, in the presence of enhanced TCR stimulation, PGE 2 facilitated, rather than inhibited, IL-12-primed Th1 cell differentiation, and this was mediated by EP 2 and EP 4 receptors through the cAMP-PKA pathway (Yao et al., 2009). Furthermore, PGE 2 promotes Th1 cell differentiation by induction of the IL-12Rβ2 and IFN-γ receptor α chain (IFN-γ receptor 1), thus amplifying the IL-12 and IFN-γ signalling pathways (Yao et al., 2013). Mechanistically, PGE 2 -EP 2 /EP 4 -cAMP-PKA signalling activates cAMP-response element binding protein (CREB) by CREB phosphorylation and its cofactor CREB-regulated transcription co-activator 2 (CRTC2) through CRTC2 dephosphorylation (Screaton et al., 2004), both of which then translocate to the nucleus, bind to IL-12Rβ2 and IFN-γ receptor 1 gene loci to initiate their transcription (Yao et al., 2013). These results have thus clearly established that the PGE 2 -EP 2 /EP 4 -cAMP-PKA pathway can facilitate Th1 cell development through crosstalk with IL-12 and IFN-γ ( Figure 3A). In addition to PGE 2 , PGI 2 appears to similarly promote Th1 differentiation possibly through cAMP-PKA signalling (Nakajima et al., 2010).
PGE 2 -EP 2 /EP 4 receptor signalling in expansion and pathogenic conversion of Th17 cells. Th17 cells drive chronic immune inflammation through the pro-inflammatory cytokines they produce, for example, IL-17A, IL-17F and IL-22 (Stockinger and Omenetti, 2017). Mouse Th17 cells are induced from naïve T-cells by TGF-β and IL-6 and then expanded and maturated by IL-23. Primarily differentiated Th17 cells express relatively lower amounts of IL-17A and IL-23 receptors and have less pathogenicity. Once these cells are further stimulated by IL-23, they express high levels of IL-23 receptor, produce large amounts of IL-17A and other cytokines (such as IL-17F, IL-22 and GM-CSF) and become more pathogenic (Lee et al., 2012). Human Th17 cells can be induced by IL-23 and IL-1β either with or without TGF-β. Unlike Th1 cells, Th17 cell differentiation by TGF-β and IL-6 is inhibited by PGE 2 , possibly due to a down-regulation of TCR signalling and TGF-β signalling. However, PGE 2 markedly enhances IL-23-stimulated Th17 cell proliferation and the production of IL-17 and IL-22, an action also mediated by EP 2 and EP 4 receptors through the downstream cAMP-PKA pathway (Yao et al., 2009). Intriguingly, suppression of endogenous cAMP production by a deficiency in the stimulatory Gα subunit of the heterotrimeric G protein on T-cells down-regulated Th17 and Th1 cell differentiation (Li et al., 2012). Recently, Lee et al. examined the molecular mechanisms of PGE 2 -EP 2 /EP 4 receptor signalling-mediated amplification of IL-23-induced Th17 expansion in detail and found that this PGE 2 pathway amplifies IL-23-induced IL-23 receptor expression by not only further activating STAT3 but also activating NF-κB and CREB through cAMP . Interestingly, they found that IL-23 induces the expression of COX-2 in Th17 cells and inhibition of PG synthesis with indomethacin attenuates the IL-23 action, suggesting that IL-23 mobilizes endogenous Th17 cell-intrinsic PGE 2 system to augment its effects. They further showed that the cAMP pathway not only enhances IL-23-induced gene expression but also induces the expression of a variety of genes that are not induced by IL-23 alone and confers more pathogenic signature to Th17 cells . The involvement of CREB and its co-activator CRTC2 in promoting Th17 cells were also reported by other groups (Hernandez et al., 2015;Wang et al., 2017). These studies suggested that the promoting action of PGE 2 on pathogenic Th17 cells is, at least partly, due to direct binding of CREB and CRTC2 to an IL-17A promoter and enhancing its gene transcription ( Figure 3B).
The PGE 2 -EP 2 /EP 4 -cAMP signalling pathway also promotes IL-23/IL-1β-stimulated expansion and activation of human Th17 cells (Chizzolini et al., 2008;Boniface et al., 2009;Napolitani et al., 2009). As seen in mouse T-cells, PGE 2 signalling not only directly enhances the induction of IL-17A gene transcription but also induces receptors for the cytokines involved. Boniface et al. observed an up-regulation of the expression of IL-23 and IL-1 receptors in PGE 2 -stimulated human Th17 cells and proposed that PGE 2 amplifies IL-23-IL-23 receptor and IL-1β-IL-1 receptor signalling to optimize IL-17A production. They further reported that of the two EP receptors, EP 2 is predominantly responsible for Th17 cytokine production, while EP 4 exerts the additional effects such as inhibiting IL-10 and IFN-γ production in Th17 cells (Boniface et al., 2009). Consistent with these findings, blockade of transcriptional co-activators of CREB, CREB binding protein (CBP) and p300, by the bromodomain CBP30 inhibits IL-17A secretion by human Th17 cells (Hammitzsch et al., 2015). Furthermore, a recent study demonstrated that PG-cytokine crosstalk in the regulation of lymphocyte-mediated chronic inflammatory responses. (A) PGE 2 and PGI 2 crosstalk with IL-12 and IL-23 signalling pathways in Th1 and Th17/Th22 cells, respectively, and promote differentiation and expansion of these Th cell subsets both in vitro and in vivo, producing inflammatory cytokines such as IFN-γ, IL-17A, IL-17F and IL-22. These PG actions underlie Th1 and Th17 cell-mediated autoimmune inflammation in mouse models of MS, RA, IBD, atopic dermatitis (AD) and psoriasis. (B) In response to IL-23 stimulation, type 3 innate lymphoid cells (ILC3) produce the cytokine IL-22, critical to epithelial homeostasis and modulation of inflammation. PGE 2 has been shown to augment IL-23-induced IL-22 production from ILC3s through the EP4-cAMP-PKA pathway, facilitating intestinal epithelial homeostasis and control of systemic inflammation. (C) PGE 2 and PGI 2 inhibit antigen-dependent activation of Th2 cells and production of type 2 cytokines (e.g. IL-4, IL-5, IL-13) and thus limit type 2 allergic inflammation. In contrast, PGD 2 promotes allergic inflammation through DP 2 receptor-mediated Th2 cell migration. (D) Epithelial cytokines such as IL-25, IL-33 or thymic stromal lymphopoietin (TSLP) stimulate ILC2 to produce type 2 cytokines, developing type 2 allergic immune diseases such as asthma and AD. While PGD 2 promotes type 2 allergic immune responses through driving ILC2 cell migration via DP 2 receptors, PGI 2 and PGE 2 inhibit allergic immune responses through suppressing epithelial cytokine-induced production of type 2 cytokines from ILC2 cells.
the EP 2 receptor is overexpressed in T-cells from MS patients, while its expression in T-cells in healthy individuals undergoes retinoic acid-related orphan nuclear hormone receptor C-dependent silencing and that simulation of EP 2 receptors in the MS patient T-cells induces a pathogenic Th17 phenotype characterized by co-expression of both IL-17A and IFN-γ (Kofler et al., 2014), thus providing clinical relevance of the finding on PGE 2 -EP 2 receptor signalling in mouse Th17 cells. Taken together, these studies clearly indicate the important roles for PGE 2 in the development of pro-inflammatory Th17 cells and, as described later, the pathogenesis of chronic inflammation in the mouse and human ( Figure 3B).
The discrepancy between previous observations on PGE 2 suppression of T-cell activation, proliferation and cytokine production (Harris et al., 2002) and recent findings about PGE 2 facilitation of Th1 and Th17 cell responses (Sheibanie et al., 2007a;Chizzolini et al., 2008;Boniface et al., 2009;Yao et al., 2009Yao et al., , 2013Chen et al., 2010;Lee et al., 2018) may be caused by multiple factors. One is that different concentrations of PGE 2 were used in the experiments. In most previous studies with in vitro cell cultures, PGE 2 was used at micromolar levels, but actual concentrations of PGE 2 in vivo, especially in humans, are usually at nanomolar levels, examples being those in cerebral spinal fluid from MS patients (Mattsson et al., 2009) and in joint fluid from RA patients (Hishinuma et al., 1999). PGE 2 was found at similar levels (ng·g À1 wet weight of tissue) in the rectal mucosa from UC patients (Sharon et al., 1978) and in colonic mucosa from rat with dextran sulfate sodium (DSS)-induced colitis (Yamashita, 1993). TCR-dependent T-cell activation and function was inhibited by high rather than low (~nM) concentrations of PGE 2 . In contrast, the low, nanomolar concentrations of PGE 2 facilitated Th1 and Th17 cell responses in both in vitro and in vivo assays (Yao et al., 2009(Yao et al., , 2013Lee et al., 2018). Another factor may be the different T-cell stimulation conditions, for example, the strength of TCR-costimulation and the cytokine milieu present during T-cell differentiation and expansion (Yao et al., 2013; Lee

Figure 3
Molecular mechanisms for PGE 2 -cytokine crosstalk in facilitation of adaptive Th1 and Th17 cell responses. (A) PGE 2 through EP 2 /EP 4 receptors activates the cAMP-PKA-CREB/CRTC2 pathway that in turn drives the expression of IL-12β2 and IFN-γ 1 receptors (R) and amplifies cytokine signalling, leading to facilitation of Th1 cell differentiation. Additionally, activation of PI3K by strengthened CD28 costimulation and that by EP 2 /EP 4 receptors further supports Th1 cell differentiation by antagonizing cAMP/PKA-mediated suppression of TCR activation. (B) IL-23 induces COX-2 expression and low levels of PGE 2 production by activated Th17 cells. T-cell-intrinsic PGE 2 then acts back on Th17 cells for the induction of IL-23 receptor (and possibly also IL-1 receptor) through the positive feedback IL-23-PGE 2 -EP 2 /EP 4 -cAMP-PKA-STAT3/CREB/NF-κB-IL-23 receptor loop, leading to the generation of pathogenic Th17 cells by synergistic actions with IL-23 and/or IL-1. CREB and CRTC2 can also directly drive IL-17A gene expression by binding to its promoter. (C) Activation of NF-κB induced by TLR or CD40 engagement on DCs induces the early phase of IL-23 p19 gene expression. NF-κB also induces DCs to express COX2 and PGE 2 , which then further amplifies NF-κB signalling for the IL-23 p19 expression through the EP 2 /EP 4 -cAMP-PKA-CREB pathways. PGE 2 can also induce IL-23 p19 expression through cAMP-Epac pathway-activated C/EBPβ. et al., 2018). Altogether, PGE 2 can have both pro-and antiinflammatory effects depending on its local concentrations, the disease settings and also the timing of its action. PGE 2 -EP 2 /EP 4 receptor signalling in migration, maturation and IL-23 production of DCs. The involvement of PGs in acquired immunity is not limited to its regulation of T-cells but is also seen in APCs. Kabashima et al. found that, in response to antigen uptake, the migration of Langerhans cells to regional lymph nodes, the expression of costimulatory molecules and their ability to stimulate T-cells were impaired in EP 4 -deficient mice, suggesting a critical role for PGE 2 -EP 4 receptor signalling in DC maturation and migration after antigen uptake (Kabashima et al., 2003b). PGE 2 was also reported to up-regulate OX40 ligand (OX40L) expression on human monocyte-derived DCs, which in turn enhanced the OX40-OX40L costimulation to enhance human antigen-specific T-cell proliferation in in vitro T-cell-DC co-cultures (Krause et al., 2009). PGE 2 is further involved in IL-23 production by activated DCs. Ganea's group showed that various toll-like receptor (TLR) ligands such as LPS, poly-I-C, CpG and proteoglycan induce the expression of IL-23 p19 subunit in DCs and that PGE 2 potently enhances this action via EP 2 and EP 4 receptors in vitro (Sheibanie et al., 2004;Khayrullina et al., 2008). In the animal model of collagen-induced arthritis, administration of PGE 2 analogue, misoprostol, exacerbated joint inflammation, and this was associated with increased mRNA levels of IL-23 p19 and other proinflammatory cytokines such as IL-17, IL-6 and IL-1β (Sheibanie et al., 2007a). This PGE 2 action is exerted by an interaction between NF-κB activated by the TLR pathway and CREB and C/EBP-β activated by the PGE 2 -EP 4 -cAMP-PKA and Epac pathways respectively (Kocieda et al., 2012). The PGE 2 -mediated enhancement of IL-23 production was also seen in DCs stimulated with anti-CD40 antibody and, interestingly, treatment with indomethacin or an EP 4 antagonist almost completely suppressed IL-23 production in vitro (Yao et al., 2009), suggesting that DC-intrinsic PGE 2 amplifies IL-23 production through a positive feedback loop. A further study by Ma et al. showed that anti-CD40 antibody stimulation induced two phases of IL-23 gene expression, the first with a peak at 1 h and the second lasting up to 36 h, while LPS or TNF-α induced only the early response of IL-23 gene expression. PGE 2 or an EP 4 agonist but not an EP 2 agonist amplified both phases of the anti-CD40 response by tens of folds (Ma et al., 2016). Mechanistic analysis revealed that the early phase was mediated by canonical NF-κB signalling and the late phase by non-canonical NF-κB signalling and that PGE 2 -EP 4 signalling exhibited a synergistic action with both signalling pathways ( Figure 3C). One pathological implication of the IL-23 induction by this PGE 2 signalling is in cancer. The IL-23 to IL-17 cascade through the IL-23 receptor is also implicated in various human cancers including those of the colon, ovary, lung, breast, stomach, skin, liver and head and neck (Wang and Karin, 2015). Qian et al. found that mice bearing 4T1 breast tumour cells have an increased number of Th17 cells not only in tumour tissues but also in the spleen and peripheral blood and that this is due to PGE 2 secreted from the tumour, which induces IL-23 in DCs through EP 2 /EP 4 receptors and a CREB-dependent manner (Qian et al., 2013). In addition, Kabashima et al. reported that the engagement of TP receptors in T-cells with TXA 2 produced by DCs resulted in impaired DC-T-cell adhesion and inhibited DC-dependent T-cell proliferation in vitro, while TP-deficient or TP antagonisttreated mice had enhanced acquired immune response to foreign antigens in vivo (Kabashima et al., 2003a). A further study indicates that TXA 2 tonically suppresses the interaction of weak CD4 + T-cells and DCs through TP receptors (Moalli et al., 2014).
PGE 2 -EP 2 /EP 4 signalling in Th1/Th17-mediated immune inflammation; animal models and human relevance. Given the PGE 2 -EP 2 /EP 4 -mediated facilitation of differentiation of Th1 cells, expansion and pathogenic conversion of Th17 cells and IL-23 production by DCs, several studies have examined the significance of this signalling in Th1-and Th17-driven immune inflammation in various animal models of autoimmune diseases such as MS (Kihara et al., 2009;Yao et al., 2009;Esaki et al., 2010;Hernandez et al., 2015), IBD (Sheibanie et al., 2007b;Wang et al., 2017), arthritis (Chen et al., 2010;Monk et al., 2014), allergic lung inflammation (Li et al., 2011) and atherosclerosis (Kotla et al., 2013) and their relevance to human autoimmune diseases. For example, blockade of PGE 2 -EP 4 receptor signalling by pharmacological (e.g. COX inhibitors and EP 4 antagonists) or genetic approaches (e.g. KOs of mPGES1 or EP 4 genes) suppressed myelin oligodendrocyte glycoprotein peptide-induced experimental autoimmune encephalomyelitis (EAE) and reduced Th1 and Th17 cell development, and these suppressing effects were further enhanced by co-inhibition of EP 2 receptors (Kihara et al., 2009;Yao et al., 2009;Esaki et al., 2010). Pharmacological antagonism of EP 4 receptors also attenuated collagen-induced arthritis progression with concomitant inhibition of the production of IFN-γ and IL-17A (Sheibanie et al., 2007a;Chen et al., 2010). While conditional deletion of EP 4 receptors in T-cells ameliorated T-cell-mediated chronic colitis accompanied by reduction in Th1 accumulation in the colon (Yao et al., 2013), the administration of PGE 2 or misoprostol, an EP 3 /EP 4 receptor agonist, augmented 2,4,6trinitrobenzene sulfonic acid-induced colitis with an increase in IL-17 and IL-23 levels in inflamed colon (Sheibanie et al., 2007b). Recently, Lee et al. showed that T-cell-specific deletion of EP 2 and EP 4 receptors in combination suppressed psoriasislike skin inflammation induced by subcutaneous injection of IL-23 or imiquimod with concomitant suppression of IL-17 + and IL-17 + IFN-γ + T-cell accumulation in the lesion, suggesting the critical role of T-cell-intrinsic EP 2 /EP 4 receptor signalling in the development of immune inflammation . Moreover, deletion of Gαs or CREB in T-cells similarly reduced Th17-induced chronic colitis and EAE (Li et al., 2012;Wang et al., 2017). Although PGE 2 essentially protects against acute epithelial injury (Kabashima et al., 2002;Duffin et al., 2016), these studies together suggested the detrimental action of PGE 2 signalling in Th17 cells in the gut. Similar to PGE 2 , activation of PGI 2 -IP receptor signalling has also been reported to foster Th17 cell responses and exacerbate chronic inflammation mediated by this type of cell in EAE and enhance activation of Th17 cells from systemic sclerosis patients (Truchetet et al., 2012;Zhou et al., 2012). Therefore, blockade of PGE 2 or PGI 2 production by inhibiting COX, mPGES1 or their receptors by small molecular inhibitors would be an effective treatment in the progression of autoimmune diseases.
The clinical relevance of these findings to human autoimmune diseases is indicated by experimental findings that patients with autoimmune inflammatory diseases like IBD and MS have elevated levels of PGE 2 in their serum and the site of inflammation, for example, the cerebral spinal fluid for MS (Mattsson et al., 2009;Prüss et al., 2013), and genetic findings from recent GWAS studies, which have identified Ptger4 (EP 4 receptor) as a susceptible locus in a number of autoimmune diseases including IBD (Glas et al., 2012), MS (International Multiple Sclerosis Genetics Consortium et al., 2011), ankylosing spondylitis (AS) and allergy (Hinds et al., 2013;Parkes et al., 2013). Numerous GWAS studies have suggested that polymorphisms in the 5p13.1 regulatory regions near PTGER4 (e.g. rs9292777, rs7725052, rs77145747, rs6896969 and rs4613763) were significantly associated with PTGER4 gene expression and the susceptibility of MS (De Jager et al., 2009;International Multiple Sclerosis Genetics Consortium et al., 2011;Matesanz et al., 2012). In human IBD, gene polymorphisms in PTGER4 loci (e.g. rs4613763, rs16869977, rs10512739, rs6880934 and rs9292777) were similarly associated with the susceptibility of Crohn's disease (CD) (Libioulle et al., 2007;Barrett et al., 2008;Kenny et al., 2012). Further studies indicated that rs7720838 and rs4495224 were also associated with susceptibility to CD and contributed to increased PTGER4 gene expression by enhancing the binding of NF-κB and XBP1 (Glas et al., 2012) and that the disease behaviour in CD patients was enhanced if mutant alleles in both rs7720838 and NOD2 were present (Prager et al., 2014). PGs (e.g. PGE 2 and PGI 2 ) also promote chronic joint inflammation, and NSAIDs are the first line medications for treating arthritis. Polymorphisms (rs10440635 and rs76523431) in PTGER4 loci were associated with increased PTGER4 gene expression in synovial biopsy samples from patients with spondyloarthritis but also the susceptibility and severity of AS (Evans et al., 2011;Chai et al., 2013). Interestingly, polymorphisms in PTGER4 gene (rs12186979 and rs13354346) were related to susceptibility of AS in Europeans and East Asians respectively (International Genetics of Ankylosing Spondylitis Consortium et al., 2013). In addition, rs76523431 in PTGER4 was also suggested as a risk factor for RA (Rodriguez-Rodriguez et al., 2015). PTGER4 gene polymorphisms (rs7720838 and rs1494558) were associated with allergy and asthma (Kurz et al., 2006;Hinds et al., 2013), while rs4613763 was suggested as a psoriasis susceptible locus (Tsoi et al., 2015). Furthermore, epigenetic changes in the PTGER4 gene enhancer in Th17 cells were also found to be associated with human autoimmune diseases such as MS, CD and allergy (Farh et al., 2015). Consistent with these GWAS findings, the expression of PGE 2 signalling pathway genes (including both PGE 2 synthases and receptors) was found to be positively correlated with IL-23/Th17 signature genes as well as disease severity in biopsy samples of human inflamed tissues with various chronic inflammatory conditions such as MS, CD, psoriasis and atopic dermatitis Robb et al., 2018;and unpublished observations). Collectively, these genetic and epigenetic studies revealed conserved associations of variants in the PGE 2 -EP 4 signalling pathway with various IL-23/Th17-dependent human chronic autoimmune diseases. PGE 2 -EP 4 receptor signalling in the generation of IL-22-producing Th22 cells and Th22mediated inflammation IL-22 that is produced by Th17 cells and other activated T-cells such as Th22 cells is involved in inflammation of the skin, gut, liver, lung as well as infections and tissue remodelling (Sabat et al., 2014). Robb et al. recently reported that PGE 2 promotes IL-22 production from T-cells under Th22-skewing conditions with IL-23 or IL-6 or both, and this effect was mediated by the EP 2 /EP 4 -cAMP-PKA pathway through induction of the aryl hydrocarbon receptor. Consistently, T-cell-specific EP 4 receptor deficiency as well as COX inhibition reduced hapten (e.g. dinitro-fluorobenzene)-induced generation of IL-22 + Tcells in vivo and attenuated chronic allergic contact dermatitis induced by repeated oxazolone challenge (Robb et al., 2018). Interestingly, genes related to PGE 2 and IL-22 pathways were coordinately up-regulated in lesional skin from human atopic dermatitis, and their expression in inflamed skin was also down-regulated after the administration of corticosteroid or UVB treatments (Robb et al., 2018). These results suggest a crucial role for PGE 2 -IL-6/IL-23 crosstalk in the generation of IL-22 + T-cells and the promotion of T-cell-mediated chronic inflammatory skin disease.

PG signalling in Th2 cell-mediated allergic inflammation
Unlike Th1 and Th17 cell differentiation, DCs do not produce IL-4, the key cytokine for Th2 differentiation, and the differentiation of Th2 cells is induced by direct and indirect interaction of T-cells with DCs, basophils, ILCs, epithelial cells and maturing Th2 cells themselves and factors released from these cells, such as the epithelial cell-produced cytokines, IL-25, IL-33 and TSLP, and IL-4 produced by maturing Th2 cells and basophils (Walker and McKenzie, 2017). Th2 cells not only provide protective type 2 immune responses against parasites but also underpin the chronic, allergic inflammatory diseases such as atopic dermatitis and asthma. In contrast to PGE 2 -facilitated Th1 differentiation and Th17 expansion, most PGs appear to suppress Th2 differentiation. Nakajima et al. found that IP receptor stimulation of IP concentration-dependently facilitates Th1 differentiation under the Th1 skewing conditions but suppresses Th2 differentiation of BALB/c CD4 + T-cells under the Th2-skewing condition. This IP receptor-mediated pathway appears to suppress Th2 cell differentiation in vivo (Nakajima et al., 2010). Indeed, mice deficient in IP receptors exhibited significantly higher serum IgE levels compared to wild type mice and showed significantly enhanced allergic inflammation of the lung in an ovalbumin (OVA)-induced allergic asthma model (Nagao et al., 2003). Zhou et al. confirmed these phenotypes of IP-deficient mice and further revealed that these phenotypes caused by IPdeficiency are not dependent on STAT6 (Zhou et al., 2016b). Similar to IP receptor-deficient mice, EP 2 receptordeficient mice also show an exaggerated OVA-induced airway inflammation accompanied by increased IL-13 production, with higher serum IgE levels, and the administration of misoprostol during the sensitization process suppressed the production of IgE and attenuated the inflammatory response (Zasłona et al., 2014). The same phenotype, that is, exaggerated allergic inflammation with higher IgE levels, was also found in mice deficient in either COX-1 or COX-2 subjected to the OVA model. These results suggest that the PGE 2 -EP 2 receptor axis and the PGI 2 -IP receptor axis function as an endogenous brake on allergen sensitization of T-cells, although the detailed mechanism remains unknown.
In addition to these analyses on the role of PGs in allergen sensitization, several studies have also revealed the involvement of PGs in the effector phase of Th2 cell-driven allergic inflammation. The best analysed PG in this phase is PGD 2 , which is abundantly produced by mast cells activated by allergens. PGD 2 acts on DP 1 and DP 2 receptors, and both PGD receptors regulate allergic inflammation positively. Th2 cells highly express DP 2 receptors, and PGD 2 binding to DP 2 receptors on Th2 cells stimulates their migration to inflamed sites and enhances type 2 cytokine production (Hirai et al., 2001;He et al., 2010). Pharmacological inhibition of DP 2 receptors using small molecule inhibitors has been tested in randomized clinical trials for their efficacy to treat asthma and exhibited significant improvements in lung function and inflammation (Walker and McKenzie, 2017). It is unclear whether they only targeted Th2 cells, because eosinophils, mast cells and type 2 ILCs (ILC2s) also express DP 2 receptors. Screening studies using mice deficient in each PG receptor suggest an important role of another PGD receptor, DP 1 , in allergic asthma (Matsuoka et al., 2000). A deficiency in DP 1 receptors did not affect serum IgE levels in OVA-challenged mice but reduced the production of antigen-specific Th2 cytokines, the accumulation of T-cells and eosinophils, epithelial mucus secretion and airway hypersensitivity in the sensitized lung (Matsuoka et al., 2000). This phenotype in DP 1 -KO mice was mimicked by administration of a DP 1selective antagonist, S-5751, to sensitized wild mice (Arimura et al., 2001). Interestingly, the expression of DP 1 receptors in the lung was increased after OVA challenge, and the immune-reactivity was localized in the airway epithelial cells, suggesting that PGD 2 may activate epithelial cells by binding to DP 1 receptor (Arimura et al., 2001). In contrast to these actions of PGD 2 , PGE 2 appears to exert suppressive actions also in the effector phase of type 2 allergic inflammation through various receptors. Kunikata et al. found that, compared with WT mice, EP 3 -deficient mice had increased type 2 cytokine production and T-cell infiltration in the lung and augmented airway inflammation induced by OVA (Kunikata et al., 2005). They further found that OVA challenge induced the expression of a variety of genes related to allergic inflammation and tissue remodelling including the chemokine thymus and activation-regulated chemokine (TARC, also known as CCL17) and eotaxin, in the lung, and the EP 3 agonist treatment suppressed this induction and the above asthmatic symptoms. Intriguingly, EP 3 receptors and CCL17 are both apparently expressed in the airway epithelium and agonist administration suppressed the expression of CCL17 (Kunikata et al., 2005), suggesting that PGE 2 -EP 3 receptor signalling inhibits allergic inflammation by suppressing the allergen-induced gene expression in the epithelium. Because DP 1 receptor is also expressed in the airway epithelium and exerts opposite actions to EP 3 , and EP 3 receptors and DP 1 receptors are coupled oppositely to adenylate cyclase via Gαi and Gαs, respectively, these results suggest that the airway epithelium is the site of action of these PGs. In addition to Th2 cells, T-cells expressing IL-9 named Th9 cells contribute to allergic inflammation by promoting mast cell expansion and IL-13 production. Li et al. examined the role of PGs in Th9 generation by subjecting COX-1-and COX-2-deficient mice to the OVA model and found enhanced generation of Th9 cells in the absence of COX-2. They then found that PGD 2 and PGE 2 suppressed in vitro Th9 differentiation induced by TGF-β and IL-4 through the cAMP pathway, which was due to a down-regulation of IL-17RB that was responsible for the expansion of Th9 induced by IL-25 endogenously produced in T-cells during the differentiation of Th9 cells (Li et al., 2013). These studies collectively suggest that PGs have diverse effects on Th2-cell mediated type 2 immune responses and allergic inflammation, which is likely determined by the balance of different PGs during antigen sensitization and challenge. In addition to the above studies focused on allergic inflammation, Birrell et al. examined EP receptors in PGE2-mediated modulation of anti-inflammatory and bronchodilator activities in the lung and found that EP4 agonists can control airway inflammation in various models induced by endotoxin, OVA or cigarette smoke (Birrell et al., 2015).

PG signalling in inflammation mediated by innate lymphoid cells
ILCs are recently identified new types of innate immune cells with transcriptional, functional and phenotypic similarities to Th subsets but without lymphocyte lineage markers and antigen receptors. ILCs therefore respond not to antigens but to environmental factors such as cytokines and lipid mediators. Similar to Th subsets, ILCs are classified into three subsets, ILC1, ILC2 and ILC3, according to their cytokine production and master transcription factors; ILC1s produce IFN-γ in response to IL-12, ILC2s produce type 2 cytokines such as IL-5 and IL-13 in response to epithelial cytokines (alarmins), IL-33, IL-25 and TSLP, and ILC3s produce IL-22 and/or IL-17 in response to IL-23 and IL-1β. (Klose and Artis, 2016). As indicated from the above properties, ILC2s mediate type 2 immune responses and allergic diseases like Th2 cells in response to alarmins, various cytokines and lipid mediators. The most well-studied PG affecting ILC2 responses is PGD 2 . Like Th2 cells, human ILC2 cells highly express the PGD receptor DP 2 (Mjösberg et al., 2011). The stimulation of DP 2 receptors by PGD 2 enhanced ILC2 migration and induced the expression of not only genes for type 2 cytokines such as IL-4, IL-5 and IL-13 but also genes for IL-33 and Il-25 receptor subunits, ST2 and IL-17B receptors, respectively, thus amplifying IL-33 and IL-25 signalling pathways for the production of type 2 cytokines and promoting allergic inflammation in skin and lung (Chang et al., 2014;Xue et al., 2014). Eastman et al. reported that ILC2 cells were specifically recruited to nasal mucosa in patients with aspirin-exacerbated respiratory disease during aspirin challenge, with a paradoxical increase in urinary PGD 2 metabolite, 11β-PGF 2 (Eastman et al., 2017). This study indicates that PGD 2 may have critical roles in recruiting and activating ILC2s through DP 2 receptors in aspirin-exacerbated respiratory disease. PGD 2 similarly induces ILC2 chemotaxis in a DP 2 receptor-dependent manner in mice. Wild type mice infected with helminth exhibited pulmonary ILC2 accumulation and type 2 inflammation, which was significantly reduced by a deficiency in DP 2 receptors and the administration of a DP 2 antagonist (Wojno et al., 2015), demonstrating that the PGD 2 -DP 2 axis is indeed involved in ILC2 recruitment in vivo. In contrast to PGD 2 , PGE 2 suppresses the up-regulation of GATA3 and CD25 (i.e. IL-2 receptor-α) induced by IL-33/IL-25/TSLP in human ILC2 cells and inhibits ILC2 proliferation and expression of type 2 cytokines such as IL-5 and IL-13 through EP 2 and EP 4 receptors (Maric et al., 2018). In addition, PGI 2 also inhibited IL-33-stimulated ILC2 activation and type 2 cytokine production in vitro. The administration of cicaprost, a synthetic IP agonist, suppressed the ILC2 response and lung inflammation induced by a fungal aeroallergen while a deficiency in IP receptors augmented this ILC2-dependent in vivo allergic immune response, suggesting a critical role for endogenous PGI 2 -IP receptor signalling in the control of ILC2-mediated allergic inflammation (Zhou et al., 2016a).
As for the role of PGs in ILC3, Duffin et al. recently reported that PGE 2 promoted IL-23-induced ILC3 activation and IL-22 production in vitro, and this was mediated by EP 2 /EP 4 -cAMP-PKA signalling (Duffin et al., 2016). Given that IL-22-producing ILC3 cells play important roles in mucosal homeostasis and control of acute epithelial damage (Klose and Artis, 2016), they examined the significance of this pathway in LPS-induced systemic inflammation. They found that inhibiting COX with indomethacin markedly exacerbated LPS-induced systemic inflammation, which was caused by enhanced gut bacterial translocation due to reduced number of ILC3 cells and IL-22 production in the intestine and that all of these phenotypes could be prevented by administration of an EP 4 selective agonist or exogenous IL-22. They further showed that EP 4 deletion in ILCs led to a reduction of IL-22 production and augmented systemic inflammation, as exemplified by enhanced TNF-α production. Moreover, PGE 2 also enhanced IL-22 production from IL-23/IL-1β-stimulated human ILC3s. These results together suggest that PGE 2 -EP 4 receptor signalling directly acts on ILC3s to potentiate their homeostasis and function in the intestine, promoting its barrier function to prevent systemic and intestinal inflammation (Duffin et al., 2016). In addition to this cytoprotective effect, IL-22-producing ILC3 cells sustain colon cancer in a colitis-associated cancer model of 129SvEv.Rag À/À mice treated with Helicobacter hepaticus and azoxymethane (AOM) (Kirchberger et al., 2013), and some ILC3s also produce pro-inflammatory cytokines (e.g. IL-17A, IFN-γ and GM-CSF) in response to IL-23 and other cytokines, such as IL-1β and IL-12 (Klose and Artis, 2016), which contribute to intestinal inflammation (Buonocore et al., 2010). Whether PGE 2 or other PGs can also promote such mucosal inflammation through stimulating the pro-inflammatory ILC3s similarly to pathogenic Th17 cells by interacting with IL-23 and/or IL-1β is not known.

PG signalling in the regulation of B cells responses
Previously, PGE 2 was shown to be involved in the modulation of immunoglobulin class switching to IgG1 and IgE synthesis in vitro through EP 2 /EP 4 -cAMP signalling (Harris et al., 2002). Consistent with these findings, Gao et al. found that PGE 2 enhances IL-4-induced STAT6 activation in an EP 2 -dependent manner to prompt IgE switching and that in the OVA-induced asthma model, mice with a deficiency in EP 2 receptors exhibited markedly attenuated IgE antibody responses and airway inflammation (Gao et al., 2016). Furthermore, also consistent with the earlier findings on PGE 2 -mediated inhibition of B cells activation (Roper et al., 1994), Murin et al. found that PGE 2 suppresses B cell receptor-mediated proliferation of B cells lymphomas through EP 4 receptors and that gene knockdown of Ptger4 (encoding EP 4 receptor) in B cells lymphoma markedly accelerates tumour spread in mice, while PTGER4 overexpression yields significant protection (Murn et al., 2008). Because the expression of EP 4 receptors is down-regulated in the lymphoma, the authors suggested a tumour-suppressor role for EP 4 receptors. Prijatelj et al. suggested that PGE 2 -EP 4 receptor signalling-mediated elevation of cAMP and inactivation of NF-κB by EP 4 may be involved in the inhibition of B cell proliferation (Prijatelj et al., 2012). In the germinal centre (GC), B cells are selected through interaction with follicular DCs bearing immune complexes in an antigen affinity-dependent manner in the so-called affinity maturation process, and IL-21 produced by follicular helper T-cells stimulates this process. Magari et al. constructed a co-culture of FL-YB follicular DCs and B cells and found that FL-YB cells produced PGE 2 , which, combined with IL-21, acted on EP 4 receptors and induced B cell death by up-regulation of pro-apoptotic proteins Bim and Foxo1 (Magari et al., 2011). Since treatment with indomethacin and an EP 4 antagonist ONO-AE3-208 significantly decreased the number of apoptotic GC B cell, these authors indicated that the IL-21-PGE 2 crosstalk physiologically regulates β cell death in GC B cell selection. In addition to these actions of PGs on B cell fate in the periphery, in mice deficient in COX-1 early B cell development from pro-B to pre-B stage was arrested, which caused a systematic reduction in total B cells . Mechanistic studies revealed that COX-1-derived TXA 2 regulated JAK3/STAT5 signalling through binding to TP receptors. Treatment of COX-1-deficient mice with a TP agonist restored the defective B cell development and JAK3/STAT5 signalling activity .

PG-cytokine crosstalk in myeloid and stromal cell-mediated chronic inflammation
Chronic inflammation is characterized by the persistent infiltration of mononuclear cells including macrophages, lymphocytes and plasma cells in most cases and, in some cases, polymorphonuclear leukocytes, and consequent tissue destruction largely by the products of these infiltrating inflammatory cells (Kumar et al., 2018). Tissue remodelling such as angiogenesis and fibrosis is also simultaneously seen in chronic inflammation. Substantial evidence has now accumulated indicating that PGs contribute to these processes.

PG-cytokine crosstalk in macrophages
The macrophage is one of the main cell sources for PGs in inflammation and the PGs produced act back on macrophages and amplify their function. For example, while LPS induced the expression of COX-2, IL-1β and IL-6 in cultured macrophages, this induction was attenuated by celecoxib, a COX-2 inhibitor, or RQ-00015986, an EP 4 antagonist, suggesting a positive feedback loop for LPS-activated endogenous PGE 2 -EP 4 receptor signalling in macrophages (Oshima et al., 2011). This PG-mediated positive feedback loop has been shown in various models of chronic inflammation. Intracranial aneurysm (IA) is chronic inflammation of the cerebral artery histologically characterized by degenerative changes in the arterial walls and inflammatory cell infiltration consisting mainly of macrophages (Chyatte et al., 1999). Aoki et al. identified the PGE 2 -EP 2 -NF-κB signalling cascade in macrophages infiltrating the arterial wall as a factor sustaining the pathogenesis of IA and making the inflammation chronic (Aoki et al., 2017). They found that PGE 2 activates NF-κB synergistically with TNF-α via EP 2 receptors in macrophages in vitro to induce proinflammatory genes including COX-2 and the macrophage chemokine CCL2 (also called MCP-1) and that CCL2 mRNA is also stabilized by this pathway. Consistent with these in vitro findings, mice with macrophage-specific EP 2 receptor deletion or with transgenic expression of an IκB mutant that restricts NF-κB activation showed reduced IA incidence with fewer infiltrated macrophages (Aoki et al., 2017). Similarly, administration of an EP 2 antagonist in a rat model of IA reduced macrophage infiltration and suppressed IA formation and progression (Aoki et al., 2011(Aoki et al., , 2017. The authors suggest that PGE 2 -EP 2 receptor signalling not only amplifies inflammation by making a positive feedback loop involving NF-κB and COX-2 but also amplifies macrophage recruitment by increasing and sustaining CCL2 expression at the inflamed site by this loop, contributing to the pathogenesis of IA ( Figure 4A) (Aoki et al., 2017). Indeed, infiltration of macrophage-expressing COX-2 and EP 2 receptors was found in clinical samples of human IA (Aoki et al., 2017). However, Kumei et al., using a diet-induced nonalcoholic steatohepatitis model, found that deficiency in IP receptors accelerates disease progression by augmenting histological derangements including cell infiltration accompanied by increased expression of CCL-2 (also called MCP-1) and TNF-α. Such disease progression in WT mice is significantly suppressed by administration of an IP agonist, beraprost (Kumei et al., 2018). These results indicate that PGI 2 -IP receptor signalling down-regulates CCL-2 and TNF-α and suppresses inflammation, an opposite action to PGE 2 -EP 2 receptor signalling described above and suggest that similar PG signalling exert different actions in different context. Similarly, TXA 2 can also promote the expression of CCL-2 and its receptor CCR2 in macrophages. In the carbon tetrachloride-induced liver injury and repair model, a deficiency in TXA 2 -TP receptor signalling reduced the accumulation of hepatic CD11b + F4/80 + macrophages and hepatic expression of CCL-2, CCR2, IL-6, TNF-α and hepatocyte growth factor, leading to impaired liver regeneration (Minamino et al., 2012).
Other examples of cytokine-PG interactions in macrophages have been reported in inflammation associated with bacterial and viral infection ( Figure 4B). Mayer-Barber et al. found that Mycobacterium tuberculosis infection induced IL-1α and IL-1β production, which in turn induced PGE 2 production by macrophages and enhanced their antimicrobial activity (Mayer-Barber et al., 2014). Type 1 IFNs are known to antagonize the IL-1 receptor pathway during mycobacterium infection and subvert anti-tuberculosis host defence. They found that the IL-1 receptor-PGE 2 pathway inhibits type 1 IFN synthesis in macrophages and counter-regulates their actions. Therefore, blocking IL-1 receptors or COX-2 signalling or enhancing IFN signalling caused uncontrolled M. tuberculosis infection, leading to mortality and necrotic lung inflammation, but this could be prevented by administration of exogenous PGE 2 (Mayer- Barber et al., 2014). This study provided a novel mechanism for how the interaction between IL-1 receptors, PGE 2 and IFN in macrophages determines outcomes of bacterial infection and suggests potential therapeutic strategies for tuberculosis by targeting the lipid-cytokine crosstalk. However, the PGE 2 -IFN interaction can also yield an opposite outcome in viral infection. Influenza virus infection evoked PGE 2 production, which inhibits type 1 IFN production and suppresses apoptosis in macrophages through EP 2 and EP 4 receptor, leading to augmented viral infection and dissemination (Coulombe et al., 2014). Since macrophage apoptosis produces apoptotic vesicles containing viral antigens and type 1 IFNs can enhance antigen cross-presentation, PGE 2 -mediated suppression of apoptosis and inhibition of type 1 IFNs also suppressed macrophage antigen presentation and subsequent T-cell immunity. In influenza virus infection, therefore, blocking COX-2, mPGES1 or EP 2 /EP 4 receptors improves survival against lethal virus infection (Coulombe et al., 2014).
Peters-Golden and collaborators examined signalling mechanisms downstream of cAMP in PGE 2 -mediated inhibition of alveolar macrophages. They found that, in alveolar macrophages, the inhibition of the generation of proinflammatory mediators (such as TNF-α, MIP-1α and LTB 4 ) and increase in IL-10 and IL-6 production are mediated by PKA, and suppressing FcR-mediated phagocytosis is mediated by Epac, while inhibition of bacterial killing by ROS is mediated by both PKA and Epac (Aronoff et al., 2005). They then showed that PGE 2 -EP 2 -PKA/Epac signalling interferes with the translocation of p47phox to phagosomal membrane and thus limits bacterial killing by reducing the generation of ROS (Serezani et al., 2007). Wall et al. further found that differential effects of cAMP on the production of pro-and anti-inflammatory cytokines were mediated by different classes of A kinase-anchoring proteins (AKAPs) that complex with PKA and determine the cellular localization of PKA specific to each action. Specifically, they showed that cAMP-dependent suppression of LPS-induced expression of TNF-α is carried out by the AKAP95-PKA complex that targets PKA to NF-κB p105 (Wall et al., 2009). The formation of various targeting multimer complexes may help us to understand the context-and cell-dependent pro-and anti-inflammatory actions of PGE 2 -EP 2 /EP 4 receptor signalling. Peter-Golden's group also showed that PGE 2 inhibits macrophage maturation. Macrophages deficient in EP 2 receptors exhibited enhanced in vitro maturation and EP 2 À/À mice had a higher percentage of F4/80 high /CD11b high cells and greater expression of macrophage colony-stimulating factor receptor.
This inhibitory effect was also mediated through the EP 2 receptor -PKA signalling pathway (Zaslona et al., 2012).

PG-cytokine crosstalk in neutrophils
Neutrophils have been traditionally regarded as innate immune cells that combat invaders in acute infection but with restricted pro-inflammatory functions. Recent studies, however, indicate that neutrophils are capable of a vast array of pro-inflammatory functions and also contribute to chronic inflammation (Kolaczkowska and Kubes, 2013). Indeed, in IBDs, neutrophils extensively infiltrate the mucosa and crypts of the intestine, which correlates with mucosal injury and patient symptoms (Brazil et al., 2013). Recent studies have shown that PGs are involved in various proinflammatory functions of neutrophils and modulate inflammation. Ma et al. found extensive neutrophil infiltration in association with tumour development in the colon in the AOM/DSS-induced colitis-associated colorectal cancer model and reported that PGE 2 was involved in this sustained neutrophil accumulation . Infiltrating neutrophils highly express EP 2 receptors and an EP 2 receptors agonist and TNF-α synergistically induced the expression of

Figure 4
Molecular mechanisms for PG-cytokine crosstalk in the amplification of innate immune cell responses. (A) TNF-α and LPS activate NF-κB in macrophages to express COX-2 and PGE 2 , which acts back on macrophages via EP 2 receptors to further amplify NF-κB signalling for the induction and stabilization of CCL2 mRNA, recruiting macrophages to inflamed sites and promoting chronic inflammation. (B) Mycobacterium tuberculosis (Mtb) infection induces IL-1 production that in turn drives PGE 2 production by macrophages. PGE 2 then suppresses macrophage production of type 1 IFNs via EP 2 /EP 4 receptors, leading to an exacerbation of Mtb infection. Similarly, influenza A virus infection also activates macrophage production of PGE 2 , which then suppresses type 1 IFNs to control virus replication. (C) TNF-α activates NF-κB in neutrophils to express COX-2 and PGE 2 , which acts back on neutrophils via EP 2 receptors to amplify NF-κB signalling for induction of CXCL1 and inflammatory cytokines, leading to the recruitment of neutrophils to chronic inflammation sites. TNF-α-activated NF-κB also induces inflammatory cytokines that promote tumour growth. (D) TNF-α and IL-1β activate NF-κB in stromal cells (e.g. fibroblasts) to express COX-2 and produce PGs (e.g. PGE 2 and PGI 2 ). PGE 2 and PGI 2 act back on stromal cells through EP 2 /EP 4 and IP receptors, respectively, to induce IL-6 (likely also IL-1β) production, supporting activation of residential Th17 cells and leading to chronic inflammation such as RA and tumour progression.
various inflammation-related genes in vitro in primary cultured neutrophils, which included COX-2, IL-6 and CXCL1, through NF-κB. A deficiency in EP 2 receptors or EP 2 antagonism abolished neutrophil infiltration, suppressed the expression of inflammatory genes including CXCL1 and decreased the number of colon tumours . These findings are consistent with the preceding study by Katoh et al. who used the same model and found that inflammatory cells accumulate in the colon in a manner dependent on CXCR2, a receptor for CXCL1, although they suggested that these cells are neutrophilic myeloid-derived suppressor cells (Katoh et al., 2013). As neutrophils produce PGE 2 via COX-2, these results together indicate that neutrophils selfamplify their recruitment through the TNF-α-primed PGE 2 -EP 2 -NF-κB-CXCL1 pathway and recruited neutrophils amplify inflammation for tumour development ( Figure 4C). These results have thus provided new mechanistic interpretation for epidemiological findings showing that NSAIDs prevented colorectal cancer development and progression (Thun et al., 1991;Rothwell et al., 2010) and experimental work showing that genetic deletion of COX-2, mPGES1 or EP 2 receptors reduced the number and size of adenomas in AOM/DSS-or APC Min/+ -induced CRC models (Oshima et al., 1996;Sonoshita et al., 2001;Nakanishi et al., 2008). The involvement of PGs in the above scenario may not be limited to PGE 2 . Wallace et al.
found that PGF 2α -FP receptor signalling is elevated in endometrial adenocarcinoma cells and up-regulates tumorigenic and angiogenic genes including COX-2, FGF2 and VEGF (Wallace et al., 2009). They further showed that CXCL1 and CXCR2 expression was elevated in the cancer tissue and that PGF 2α -FP receptors signalling promotes CXCL1 expression on endometrial adenocarcinoma cells and attracts CXCR2-expressing neutrophils (Wallace et al., 2009). Neutrophil-produced TXA 2 has recently been reported to modulate neutrophil-dependent control of lymphocyte egress after adjuvant administration during immunization and the neutrophil spread outside the draining lymph nodes (Yang and Unanue, 2013). The pro-inflammatory actions of PGE 2 -EP 2 receptor signalling on neutrophils described above appear to be contradictory to the classic findings that PGE 2 inhibits human neutrophil function and migration through EP 2 receptors (Wheeldon and Vardey, 1993;Armstrong, 1995) and the recent findings that PGE 2 -EP 4 receptor signalling inhibits neutrophil migration in vivo (Mizuno et al., 2014). There is also a study reporting that during acute mucosal infection with Toxoplasma gondii, intestinal monocyte-produced PGE 2 controls neutrophil activation to commensal bacteria, limiting mucosal damage during acute intestinal infection (Grainger et al., 2013). This apparent contradiction suggests that whether PGE 2 exerts a pro-or anti-inflammatory action depends on the context of each pathology and experimental setting such as whether PGE 2 acts alone or with cytokines and acts in which cellular background. In contrast to PGE 2 and PGF 2α , PGD 2 -DP 2 receptor signalling is implicated as another PG signalling oppositely controlling neutrophil recruitment. DP 2 receptor -deficient mice were more resistant to caecal ligation and, in a puncture-induced sepsis model, the CXCR2-positive neutrophil accumulation was enhanced in the infections focus, the pro-inflammatory cytokine production reduced and bacterial clearance improved (Ishii et al., 2012). While this finding suggests that PGD 2 -DP 2 receptor signalling negatively regulates neutrophil recruitment, its mechanism remains unknown.
In inflammatory conditions, neutrophils release intracellular structures composed of chromatin DNA, histones and granular proteins, which make a net-like structure. This structure, that is, neutrophil extracellular trap (NET), functions to trap and clear bacteria. NET formation also occurs during various chronic inflammatory diseases (Kolaczkowska and Kubes, 2013). PGE 2 inhibited NET formation in isolated neutrophils induced by PMA or rapamycin through EP 2 /EP 4 -cAMP receptor signalling (Domingo-Gonzalez et al., 2016;Shishikura et al., 2016), although the extent to which this mechanism operates in vivo remains unknown.

PG-cytokine crosstalk in inflammatory stroma
The stroma of chronic inflammation is not only the site where active inflammation, tissue degradation and remodelling occur but also contributes itself to inflammation, as its cell components actively evoke and perpetuate inflammatory responses, as exemplified by fibroblast-like synoviocytes (FLSs) in RA (Bartok and Firestein, 2010) and tumourassociated fibroblasts (TAFs) in many cancers (Kalluri, 2016). The involvement of PGs in this active process of stroma has also been reported. For example, Honda et al. used collageninduced arthritis, a model of RA, and found that deficiency in IP receptors or EP 2 /EP 4 receptors significantly reduced the severity of arthritis, as assessed by synovial cell proliferation, inflammatory cell infiltration and joint destruction, which was accompanied by a significant reduction in the content of IL-6 in arthritic paws (Honda et al., 2006). They further used cultured synovial fibroblasts and found PGI 2 synergized with IL-1β to amplify IL-6 production, which was reduced by a COX inhibitor, indomethacin. Microarray analysis revealed that, in addition to IL-6, PGI 2 -IP receptor signalling amplified the expression of genes related to inflammation (e.g. IL-11 and CXCL7), cell proliferation (e.g. FGF and vascular and endothelial growth factor), tissue remodelling (e.g. RANKL and ADAM8) as well as IL-1 receptor itself. Given that PGI 2 alone did not induce the expression of these genes, these findings indicate that PGI 2 -IP receptor signalling functions as an amplifier of IL-1β signalling by expression of its receptor in synovial fibroblasts to significantly augment inflammation (Honda et al., 2006). Consistently, Kunisch et al. found, by using FLSs from RA patients, that TNF-α stimulation drove COX-2 expression and the production of PGE 2 , which acts back on FLSs to induce IL-6 production via EP 2 receptors (Kunisch et al., 2009). Paulissen et al. found that co-culture of primary human Th17 cells with FLSs from RA patients led to more IL-17A production, and this autocrine IL-17 production as well as the induction of IL-6, IL-8 and MMP-1 and MMP-3 was effectively prevented by co-blockade of COX-2 and TNF-α (Paulissen et al., 2013).
Similar to FLSs in RA, TAFs exert pleiotropic functions in the tumour micro-environment for progression of cancer (Kalluri, 2016). In an aforementioned work on colitisassociated cancer, Ma et al. used bone marrow chimera and found that not only EP 2 receptor-expressing neutrophils but also EP 2 receptor-expressing TAFs made a significant contribution to tumour progression. They found that the positive feedback loop of PGE 2 -EP 2 -NF-κB-COX-2 signalling also functioned in TAFs and in synergy with TNF-α to amplify the expression of pro-inflammatory genes such as IL-6, tumour promoting genes such as various WNT molecules and genes for tissue remodelling such as BDNF, MMP12 and osteopontin . These studies suggest that fibroblasts play an active role in chronic inflammation and that PG signalling contributes to their function by a synergistic interaction with cytokines ( Figure 4D).
Fibrosis and angiogenesis are ultimate hallmarks of chronic inflammation. PGs also play critical roles in these processes. In the bleomycin-induced pulmonary fibrosis model, Oga et al. found that deficiency in FP receptors attenuated the fibrosis (i.e. reduced collagen synthesis) in this model, which was independent of TGF-β and without influence on inflammatory responses (Oga et al., 2009). Consistently, PGF 2α enhanced collagen synthesis in lung fibroblasts in vitro in an additive way to TGF-β, indicating that PGF 2α -FP receptor signalling exerts a pro-fibrotic action on its own in fibrosis (Oga et al., 2009). In contrast, other PGs were reported to be anti-fibrotic. For example, mice without IP signalling had augmented bleomycin-induced pulmonary fibrosis (Lovgren et al., 2006). IP KO mice developed cardiac fibrosis, which was suppressed completely by coincidental deletion of TP receptors (Francois et al., 2005), suggesting that IP and TP receptor signalling pathways inhibit cardiac fibrosis. PGE 2 -EP 4 receptor signalling has been reported to prevent tubulointerstitial fibrosis in the kidney of mice subjected to unilateral ureteral obstruction (Nakagawa et al., 2012). These studies indicate critical roles of PGs in the activation and function of fibroblasts, but whether these effects can be modulated by the immune system and cytokines remains for further studies. Given that both type 2 cytokines (e.g. IL-4/IL-5/IL-13)-producing Th2/ILC2 cells and IL-17/ IL-22-expressing Th17/ILC3 cells have been demonstrated to regulate fibrosis (Barron and Wynn, 2011;Hams et al., 2015) and that PGs play essential roles in the regulation of immune responses as described above, it will be interesting to examine whether PGs control fibrosis through the immune system. Likewise, PG signalling functions in angiogenesis. Amano et al. implanted a Matrigel sponge or tumour cells in mice and found PGE 2 -EP 3 receptor signalling facilitates the angiogenesis associated with chronic inflammation in the sponge and tumours (Amano et al., 2003). A bone marrow transfer experiment indicates that EP 3 -bearing bone marrow-derived cells mediate VEGF expression in the stroma around the implants and the recruitment of VEGFR-1 + / VEGFR-2 + cells to the site of angiogenesis (Ogawa et al., 2009).

Conclusion
As reviewed here, substantial evidence derived from in vitro and in vivo animal model experiments now suggest that PGs are significantly involved in chronic inflammation by regulating both innate and adaptive immune cells, and gene signature analysis of clinical samples as well as GWAS of patients appears to support these experimental findings. In the processes of chronic inflammation, PGs crosstalk intimately with cytokines in various ways. One is to induce or enhance the expression of receptor(s) for the cytokines involved, which is seen in PGE 2 -EP 2 /EP 4 receptor signalling-mediated induction of IL-12β2 and IFNγ1 receptors in Th1 cells, IL-23 receptors in Th17 cells, PGD 2 -DP 2 signallingmediated induction of ST2 and IL-17B receptors in ILC2 cells, and PGI 2 -IP receptor signalling-mediated IL-1 receptor 1 induction in synoviocytes. Another is to collaborate with cytokines, particularly, TNF-α, to enhance NF-κB activation and the induction of COX-2 to self-amplify this signalling, which is seen in inflammation driven by macrophages such as in IA and neutrophil-driven inflammation in colitis-associated cancer. It may not be overstated that this COX-2-mediated amplification mechanism operates in various chronic inflammatory diseases that exhibit a high expression of COX-2 together with activated NF-κB. It is clearly noted, however, that the actions of PGs and their signalling in chronic inflammation are strictly dependent on the context, for example, the types of inflammation, cytokines, cells and so on, the timing of the disease process and the sites of inflammation, which cannot easily be generalized as seen in acute inflammation. For example, while PGE 2 -EP 2 /EP 4 receptor signalling facilitates Th1/Th17-mediated immune inflammation and sustains macrophage-or neutrophil-driven inflammation, the same signalling pathway dampens Th2-mediated allergic inflammation. The same is true for PGI 2 -IP receptor signalling. Although PGD 2 -DP 2 receptor signalling recruits Th2 cells, eosinophils and ILC2 cells, this signalling appears to function differently during neutrophil recruitment in the caecal ligation punture model. It should be mentioned, therefore, that although we believe in great potential of PG receptor type-selective agonists and antagonists, the application of these drugs should be carefully thought out by analysing the clinical and pathological context of each disease in the background of experimental studies. One successful example among this line may be the application of DP 2 antagonists to asthmatic patients (Erpenbeck et al., 2016;Kuna et al., 2016). As seen there, experimental studies to date have now accumulated enough amount of information as the base and framework for identifying a clinical indication. Given the adverse effects of traditional NSAIDs, the time is right to examine the potential of such receptor-selective molecules to manipulate chronic inflammatory diseases. In this respect, one aspect of PG studies we cannot cover in this review due to the space limit is the action of PGs in immune evasion in cancer, which is currently an active and expanding area of research. Interested readers are recommended to read a recent review (Wang and DuBois, 2018).

Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www. guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander et al., 2017a,b,c). A25246). S.N. is supported by Core Research for Evolutional Science and Technology Program on Chronic Inflammation from the Japan Agency for Medical Research and Development, by a coordination fund from the Japanese Ministry for Education, Culture, Sports, Science and Technology (MEXT) and Astellas Pharma Inc. to Kyoto University and a collaborative grant to the Kyoto University from Ono Pharmaceuticals.