Converse regulation of CCR7-driven human dendritic cell migration by prostaglandin E2 and liver X receptor activation



Migration and homing of DCs to lymphoid organs is pivotal for inducing adaptive immunity and tolerance. DC homing depends on the chemokine receptor CCR7. However, expression of CCR7 alone is not sufficient for effective DC migration. A second signal, mediated by prostaglandin E2 (PGE2), is critical for the development of a migratory DC phenotype. PGE2 is important for inducing efficient immune responses, but, if deregulated, contributes to chronic inflammation, autoimmune diseases through Th17-cell development and tumorigenesis. In contrast, activation of liver X receptor (LXR)α has recently been shown to interfere with CCR7 expression and migration of DCs resulting in a reduced immune response. Here, we demonstrate that PGE2 downregulates LXRα expression in human monocyte derived as well as ex vivo DCs. Moreover, PGE2 stimulation dampens LXR activation, auto-regulation and LXR-mediated gene transcription. Consequently, we show that PGE2 enhances CCR7 expression and migration of LXR-activated DCs. Furthermore, we provide evidence that PGE2 signaling and LXR activation specifically elicit converse effects on CCR7 expression and DC migration. In contrast, production of MMP9, CCL4, COX-2, and IL-23 is solely regulated by PGE2, but not by LXR activation, offering new perspectives for therapeutic interventions.


DCs are professional antigen-presenting cells and critical for inducing adaptive immunity and tolerance [1, 2]. DCs are found in healthy tissues as immature cells and ready to sample the environment for foreign antigens. Upon infection or inflammation, DCs mature and migrate to the draining lymph node, where the peripherally acquired antigens are presented to T cells in the context of MHC molecules. Mature DCs acquire an enhanced capacity to stimulate T cells, which is on the one hand achieved by the upregulation of costimulatory molecules to allow efficient DC-T-cell interactions, and on the other hand by the production of cytokines that T cells need to differentiate and proliferate [3].

DC migration to secondary lymphoid organs fully depends on the two chemokines CCL19 and CCL21, which are constitutively expressed by peripheral lymphatic endothelial cells and lymph node stroma cells [4, 5] and its cognate receptor CCR7, which is upregulated upon DC maturation [2, 5]. The fact that CCR7 and its ligands are mandatory for homing was demonstrated in CCR7 deficient and in plt/plt mice lacking lymphoid CCL21 and CCL19 [5]. The lack of a coordinated CCR7-dependent homing to lymph nodes results in a severely impaired T- and B-cell immunity [5]. However, CCR7 expression on mature DCs alone does not necessarily mean that DCs readily migrate toward CCL19 and CCL21. We and others have discovered that CCR7-expressing DCs migrate poorly unless DCs were exposed to prostaglandin E2 (PGE2) during maturation [6-10]. PGE2, a metabolite of arachidonic acid, plays key roles in initiating and terminating inflammatory responses [11]. Of note, signaling through the PGE2–EP4 axis early during DC maturation not only permits efficient CCR7 signaling and migration in vitro [8, 9] and in vivo [12], but is also critical for the expression of costimulatory molecules and augmented T-cell activation [12-14]. However, sustained PGE2-mediated activation of EP4 in T cells and DCs promotes Th1 and Th17 differentiation and expansion, thereby promoting severe autoimmune diseases [15, 16]. Of note, mice lacking the inflammatory microsomal PGE2 synthase 1 (mPGES1) show reduced symptoms of experimental autoimmune encephalomyelitis, whereas mPGES1 is highly expressed at autoinflammatory sites in multiple sclerosis patients [17].

The nuclear liver X receptors LXRα and LXRβ are oxysterol-activated transcription factors involved in the regulation of lipid metabolism and known to play key roles in innate and adaptive immunity [18]. Human immature monocyte-derived DCs (MoDCs) and ex vivo peripheral blood myeloid DCs (PBDCs) have been shown to express high amounts of LXRα and low, but constant levels of LXRβ [19]. LXR activation in these immature cells affected their T-cell stimulatory capacities either by altering fascin expression required for actin bundling during the formation of an immunological synapse [19], or by modifying NF-κB activation [20]. Recently, activation of LXRs has been shown to prevent TLR-induced CCR7 upregulation in MoDCs as well as HIV-1 capture and trans infection [21] and to interfere with CCR7 expression on mature DCs resulting in a dampened antitumor immune response [22].

As PGE2 and LXR ligands play key roles in inflammation, tumor biology, and autoimmune diseases and as both lipid derivates seem to have opposing effects on CCR7 expression and DC migration, we aimed to elucidate the interplay of PGE2 signaling and LXR activation primarily on CCR7 expression and migration of human DCs.


PGE2 dampens LXRα expression, LXR activation, and LXR-mediated gene transcription in human MoDCs

Stimulation of DCs with PGE2 early during maturation is known to play a key role in efficient CCR7-driven migration and homing to lymphoid organs [6-9, 12]. In contrast, LXRα activation was recently shown to reduce maturation-induced CCR7 expression and migration of DCs resulting in a poor immune response [21, 22]. Whether PGE2 modulates LXR expression and activation in DCs, and how DCs react upon exposure to PGE2 in combination with LXR activation, has not been addressed yet. Therefore, we first investigated whether PGE2 regulates LXR expression in maturing DCs. To this end, we matured terminally differentiated immature human MoDCs by various stimuli in the presence or absence of physiological concentrations of PGE2 found for instance at inflammatory sites ([11] and references therein) and measured LXRα and LXRβ transcripts by quantitative real-time RT-PCR (Fig. 1A). Immature MoDCs expressed high levels of LXRα on mRNA and protein levels (Fig. 1 and [19]). MoDCs matured by either TLR3 or TLR4 ligation, by CD40L triggering or by inflammatory cytokines (IL-1β, IL-6, TNF-α) expressed similar levels of LXRα and LXRβ as compared with those of immature MoDCs (Fig. 1). Interestingly, addition of PGE2 during MoDC maturation provoked a fast, profound, and sustained downregulation of LXRα on both mRNA (Fig. 1A) and protein level (Fig. 1B) and occurred independent of the maturation stimulus used. In contrast, PGE2 did neither downregulate LXRβ (Fig. 1A) nor further decreased the expression of the LXR target gene ABCG1 (Fig. 1C) upon MoDC maturation.

Figure 1.

PGE2 rapidly downregulates LXRα – but not LXRβ or ABCG1 – expression in human MoDCs during maturation. (A, C) Quantitative real-time RT-PCR analysis of (A) LXRα, LXRβ, and (C) ABCG1 mRNA expression during maturation of human MoDCs is shown. Fully differentiated, immature MoDCs were matured by (A, C) cytokine cocktail, (A) LPS, polyI:C or trimeric soluble CD40L in the presence (▴) or absence (▵) of PGE2. Regulation of specific mRNA expression is depicted as relative expression to immature MoDCs (0 h) and shown as mean and SEM from at least four (LXRα) or three (LXRβ, ABCG1) individual healthy donors. (B) Western blot analysis of LXRα protein expression in immature (0 h) and MoDCs matured by cytokine cocktail in presence or absence of PGE2 for 48 h is shown. Levels of β-actin of the same blot were determined and served as loading control. One representative experiment of three with different donors is shown.

In macrophages, LXR activation results in the synthesis of the cholesterol efflux transporter ABCG1, and of LXRα itself through an auto-regulatory mechanism [18]. In order to investigate the influence of PGE2 on LXR activation, we therefore treated MoDCs with the LXR agonist T0901317 during DC maturation. LXR activation strongly induced LXRα as well as ABCG1 expression after 6 h and, more profoundly, after 24 h of maturation (Fig. 2), confirming previous observations [22]. In contrast, PGE2 not only downregulated basal expression of LXRα, but also inhibited T0901317-mediated auto-induction of LXRα (Fig. 2). In addition, T0901317-mediated ABCG1 induction was also significantly reduced in MoDCs matured in the presence of PGE2 (Fig. 2). Comparable results were observed for ABCA1, another LXR target gene (data not shown). Since addition of PGE2 alone during MoDC maturation does not attenuate ABCG1 expression (Fig. 1C), it is tempting to speculate that downregulation of ABCG1 is caused by altered LXR signaling. Our data provide evidence that PGE2 on the one hand downregulates LXRα expression and on the other hand impedes LXR activation, including LXRα autoregulation and LXR-dependent gene transcription in maturing MoDCs.

Figure 2.

PGE2 stimulation of MoDCs rapidly impairs LXR activation. Human blood-derived monocytes were differentiated into immature MoDCs and subsequently matured with cytokine cocktail in the absence (□) or presence (▪) of PGE2 with or without T0901317 for 6 or 24 h. Relative mRNA expression of the LXR target genes (A) LXRα and (B) ABCG1 was determined by quantitative real-time RT-PCR and shown as mean and SEM from at least five individual donors. * p < 0.05;** p < 0.01; *** p < 0.001; Student's paired t test.

Next, we investigated the effects of T0901317 and/or PGE2 stimulation during monocyte to immature MoDC differentiation, on terminally differentiated MoDCs, as well as during MoDC maturation. Addition of T0901317 during monocyte to immature MoDC differentiation induced LXRα and ABCG1 mRNA expression (Supporting Information Fig. 1A). This observation is in line with previous studies describing upregulation of LXRα during MoDC differentiation [19, 20]. Stimulation of myeloid cells with PGE2 was shown to prevent monocyte to MoDC differentiation [23, 24] and stimulation of these differentiating cells with both T0901317 and PGE2 profoundly increases cell death and was concomitant with an absent DC phenotype after 5–6 days of culture (data not shown). Treatment of immature MoDCs with PGE2 alone reduced LXRα levels (data not shown). LXR activation of fully differentiated immature MoDCs 2–24 h prior to DC maturation with polyI:C induced strong LXRα and ABCG1 expression, which was again clearly reduced when PGE2 was present during maturation (Supporting Information Fig. 1B). Finally, LXR activation of immature MoDCs that were differentiated in the presence of T0901317 did not further induce LXRα and ABCG1 mRNA expression upon DC maturation (Supporting Information Fig. 1C). In these cells, addition of PGE2 during maturation impaired LXR-dependent gene transcription (Supporting Information Fig. 1C), however to a lesser extent compared with cells that were stimulated with T0901317 only during maturation (Fig. 2). Hence, our data reveal that the time point of triggering LXR- and PGE2-signaling critically affects the magnitude of LXR-dependent gene transcription. These results also show that PGE2 divergently affects the outcome of LXR activation induced at distinct stages of MoDC development, but occur independent of the MoDC maturation stimulus.

Converse regulation of CCR7 expression and MoDC migration by PGE2 and LXR activation

Induction of the chemokine receptor CCR7 is a hallmark of DC maturation and essential for lymph node homing. LXR activation by T0901317 strongly reduced CCR7 expression (Fig. 3A and B) and CCL21-driven migration of cytokine-matured MoDCs in 2D Trans well assays as well as in 3D matrigel assays (Fig. 3C and D). Similar results in 2D migration were observed for polyI:C matured MoDCs (data not shown), confirming recent observations in LPS- and CD40L-matured DCs [21, 22]. Nonetheless, we still measured marginal-specific migration and residual CCR7 surface expression in MoDCs matured in the presence of T0901317 (Fig. 3). However and more importantly, surface expression levels of CCR7 were shown not to correlate with DC migration efficiency. In fact, PGE2 was identified as crucial factor responsible for efficient DC migration (e.g. ref [9, 12]) and the PGE2 signal was found to be required at very early stages of DC maturation [9]. Although PGE2 is known to augment surface expression of CCR7 or to keep CCR7 expression constant depending on the DC maturation stimulus [7, 9, 10], PGE2 significantly enhances DC migration through an unknown mechanism that does not depend on the magnitude of CCR7 expression. In MoDCs matured by cytokine cocktail, PGE2 significantly enhanced CCR7 expression by two to threefold (Fig. 3B), whereas specific migration was about five times more efficient, both in a 2D Transwell system (Fig. 3C) as well as in a 3D system where MoDCs transmigrated in response to CCL21 through matrigel (Fig. 3D). By contrast, MoDCs matured in the presence of both T0901317 and PGE2 expressed similar levels of CCR7 compared with MoDCs matured with cytokine cocktail alone (Fig. 3A and B), indicating that PGE2 countervails LXR-mediated downregulation of CCR7 in maturing MoDCs. Similar results were observed when T0901317 was replaced with the synthetic LXR agonist GW3965 or the natural LXR agonist 22R-HC, or when fully differentiated, immature MoDCs were preactivated with T0901317 for 2 h prior maturation (data not shown), or during DC differentiation (Supporting Information Fig. 4A), underscoring the ability of PGE2 to preserve migration of mature DCs. Noteworthy, challenging MoDCs simultaneously with a fixed concentration of PGE2 (1 μg/mL) and graded concentrations of T0901317 (10−8–10−6 M) during maturation, or with 10−6 M T0901317 and increasing concentrations of PGE2 (10−1–103 ng/mL) indicate that both PGE2 signaling and LXR activation regulated CCR7 expression in a dose-dependent manner (Supporting Information Fig. 2A and B).

Figure 3.

Converse regulation of CCR7 expression and MoDC migration by PGE2 and LXR activation. Human MoDCs were matured by cytokine cocktail for 2 days. PGE2 and/or T0901317 were added simultaneously as indicated during maturation. (A, B, E) Surface expression of (A, B) CCR7 and (E) CD83 was determined by flow cytometry. Results from one representative experiment showing CCR7 expression (black lines) and isotype controls (gray dotted lines) in (A) or CCR7/CD83 expression with indicated percentages of mature MoDCs in (E) are depicted. (C, D) Relative-specific cell migration toward CCL21 was determined in (C) Transwell chemotaxis assays or (D) 3D matrigel assays. Relative CCR7 surface expression (B) and relative CCL21-mediated migration (C, D) of MoDCs matured by cytokines and simultaneously stimulated with or without T0901317 in the absence (□) or presence (▪) of PGE2 are shown as mean and SEM of at least 14 different donors (B, C) or five different donors (D). *p < 0.05; ** p < 0.01; Friedman with Dunn post test (B, C); repeated measures ANOVA with Bonferroni post test (D). (F) Morphological changes of MoDCs matured by cytokine cocktail in the presence or absence of PGE2 and T0901317 were documented by phase contrast imaging using a 10× objective. One representative experiment of four individual donors is depicted.

As controversial observations on the influence of LXR activation on MoDC maturation were described [19, 21, 22], we determined the expression of various surface markers by flow cytometry. We did not detect major alterations in CD80, CD86, and HLA-DR expression on MoDCs matured in the presence of T0901317 (Supporting Information Fig. 2C). However, in the majority but not all donors tested, we found a reduced CD83 expression in MoDCs matured in the presence of T0901317 (Fig. 3E). Importantly, CCR7 expression levels were reduced by T0901317 in MoDCs that still expressed high levels of CD83 (Fig. 3E) suggesting that the observed effects on CCR7 expression and migration are not simply due to inefficient DC maturation. As noted previously [6], MoDCs matured in the presence of PGE2 were nonadherent and formed clusters, whereas many MoDCs matured in the absence of PGE2 showed a needle-like adherent morphology (Fig. 3F). Of note, both adherent and nonadherent MoDCs matured in the absence of PGE2 migrated poorly in response to CCL21 (data not shown) whereas PGE2 rendered almost the entire mature MoDC population into a migratory phenotype, lacking an adherent morphology (Fig. 3F). In contrast, LXR-activated mature MoDCs showed a less migratory, more adherent phenotype (Fig. 3F). As DC migration on a 2D surface was shown to depend on cell adhesion, whereas in a 3D environment DCs can also migrate in an integrin-/adhesion-independent manner [25], we investigated MoDC migration in 3D through matrigel. CCL21-driven migration in 3D was reduced if MoDCs were treated with T0901317, but significantly increased if matured in the presence of PGE2 (Fig. 3D). MoDC migration to the lower compartment in the absence of chemokine was very inefficient in 2D and 3D (below 1.25% of input cells) in all conditions (not shown). Thus, LXR activation during MoDC maturation leads to downregulation of CCR7 expression and reduced DC migration in 2D and 3D, which can be rescued by PGE2. Highest CCR7 expression and most efficient MoDC migration, however, are achieved if PGE2 alone is added to the maturation stimulus.

Modulation of LXR and EP2/EP4 signaling pathways by T090137 and PGE2

So far, mechanisms that regulate CCR7 expression are poorly described. It has been noted previously that MoDCs do not express EP1 and EP3 [7] and that in humans EP2/EP4-signaling, as shown using receptor-specific agonists and antagonists, is responsible for the development of a migratory DC phenotype [9]. In order to gain first mechanistic insights in the converse modulation on CCR7 expression by T0901317 and PGE2, we addressed whether increased levels of cAMP, a second messenger produced upon EP2/EP4-mediated adenylate cyclase activation, contribute to PGE2-mediated attenuation of LXRα and induction of CCR7 expression. To this end, we matured MoDCs in the presence of forskolin, a pharmacological activator of adenylate cyclase, or PGE2, and measured LXRα and CCR7 mRNA expression after 6 h. However, forskolin — in contrast to PGE2 — did not affect LXRα and CCR7 expression (Supporting Information Fig. 3A). Second, we addressed whether alternative EP2/EP4 downstream signaling molecules, namely PKA, MEK, and PI3K, modulate this process. Inhibition of PKA by H-89 did not alter LXRα and CCR7 expression (Supporting Information Fig. 3A). Interestingly, inhibition of MEK by PD98059 further enhanced PGE2-mediated CCR7 induction as reported previously in a myeloid cell line [26], but did not affect LXRα expression (Supporting Information Fig. 3A). Inhibiting PI3Ks using Ly294002 did not regulate CCR7 expression induced by forskolin or PGE2 (Supporting Information Fig. 3A). Our data indicate that PGE2-mediated suppression of LXRα and induction of CCR7 expression depends on an alternative EP2/EP4 signaling pathway than the cAMP/PKA or PI3K branches of EP2/EP4 signaling.

LXR activation was shown to interfere with the expression of the actin-bundling protein fascin in LPS (but not in sCD40L)-matured MoDCs [19]. As fascin is required for podosome dissolution and efficient DC migration [27], we determined the expression levels of fascin in MoDCs matured in the presence or absence of PGE2 and/or T0901317. However, we found no obvious changes in fascin expression under these conditions (Supporting Information Fig. 3B).

So far, it is unknown how CCR7 expression is regulated in MoDCs. However, in classical Hodgkin disease-derived cells, CCR7 expression was shown to involve constitutive NF-κB activation [28]. We therefore measured the activation of distinct NF-κB subunits in MoDCs matured by TLR3-ligation in the presence or absence of PGE2 and/or T0901317. Pilot experiments revealed that differences in NF-κB activation mediated by PGE2 were detectable after 6 h of maturation. MoDCs matured in the presence of PGE2 exhibit slightly, but statistically insignificant, enhanced cRel activity, whereas MoDCs matured in the presence of T0901317 showed significantly reduced p52, p65, and RelB activity (Supporting Information Fig. 3C). Interestingly, in MoDCs matured in the presence of both T0901317 and PGE2, p52, and RelB activity was largely unaffected, indicating that PGE2 preserved the activation of distinct NF-κB subunits in LXR-activated MoDCs.

PGE2 solely regulates expression of CCL4, COX-2 MMP9, and IL-23 independently of LXR activation

PGE2 was described to induce metalloproteinase (MMP)9 expression in mouse BMDCs that is essential for their homing to lymphoid organs [29]. As LXR activation in macrophages was shown to inhibit the expression of MMP9 and enzymes for PGE2 synthesis [18, 30], we investigated on the expression of MMP9 and COX-2 in cytokine cocktail-matured MoDCs exposed simultaneously to PGE2 and/or T0901317. As depicted in Figure 4, PGE2 induced MMP9 and COX-2 expression in human MoDCs upon maturation. However, LXR activation had no influence on MMP9 and COX-2 expression in MoDCs matured neither in the presence nor in the absence of PGE2 (Fig. 4A, C). In addition, PGE2 stimulation inhibited the expression of the proinflammatory chemokine CCL4 [31] independently of LXR activation (Fig. 4B). Furthermore, PGE2 stimulation of DCs is required for IL-23 production (Fig. 4D) and hence Th17 development [15, 16]. However, LXRα activation did not interfere with PGE2-induced IL-23p19 mRNA expression (not shown) and IL-23 secretion (Fig. 4D). Similar results were obtained in MoDCs activated with T0901317 during differentiation from monocytes to immature DCs (Supporting Information Fig. 4). These data provide evidence that CCR7 expression and DC migration are conversely regulated by PGE2 and LXR activation, whereas tissue remodeling and changes in cytokine production are regulated by PGE2 alone.

Figure 4.

Expression of MMP9, COX-2, CCL4, and IL-23 is solely regulated by PGE2 and not by LXR activation. Human monocytes were differentiated into MoDCs in the presence of (A, B, D) IL-4 and GM-CSF or (C) IL-13 and GM-CSF. Immature MoDCs were treated with T0901317 or solvent and simultaneously matured for 2 days with a cocktail of cytokines in the absence (□) or presence (▪) of PGE2. Relative (A) MMP9, (B) CCL4, and (C) COX-2 mRNA expression and (D) IL-23 secretion were determined by quantitative real-time RT-PCR and ELISA, respectively, and shown as mean and SEM of 8 (A, B), 3 (C), and 4 (D) different donors. n.d.: not detectable.

LXR activation dampens, whereas PGE2 enhances, CCR7 expression, and ex vivo PBDCs migration

Finally, we investigated whether CCR7 expression and migration of human ex vivo peripheral blood (PB)DCs is also conversely regulated by PGE2 and LXR activation. Ex vivo PBDCs from healthy individuals expressed substantial amounts of CCR7 (Fig. 5A and B) as noted previously [9] and resemble steady state/semi-mature DCs [5]. CCR7 expression levels were significantly reduced by T0901317 on PBDCs (Fig. 5A, B and D). Again, PGE2 counter-balanced LXR-mediated downregulation of CCR7 expression and, more efficiently, DC migration (Fig. 5C), substantiating our findings in MoDCs.

Figure 5.

LXR activation dampens, whereas PGE2 enhances, CCR7 expression and subsequent ex vivo PBDC migration. (A, B) Surface CCR7 expression, (C) CCL21-mediated migration, and (D) CCR7/CD83 expression with indicated percentages of human cytokine cocktail-matured ex vivo PBDCs are shown. Results from one representative experiment showing CCR7 expression in presence (gray-filled areas) or in absence (black lines) of T0901317 and controls (gray-dotted lines) is depicted in (A). (B, C) PBDCs were stimulated with T0901317 in the absence (□) or presence (▪) of PGE2 during maturation and results are shown as mean + SEM of seven (B) and five (C) different donors. (D) One representative experiment of at least five individual donors is presented. * p < 0.05; ** p < 0.01; Student's paired t test.


Deregulated PGE2 production can contribute to severe immune inflammation resulting in tissue damage, autoimmune diseases through Th1 and Th17 cell differentiation, or tumorigenesis [11, 15-17]. However, under physiological conditions, PGE2 is also known to be necessary for initiating immune responses, e.g. by promoting CCR7-driven DC migration and homing to lymph nodes [9, 12], for efficient T-cell priming [12, 14], for keeping the gut mucosal barrier intact preventing colitis [32], and for homeostasis [33]. Recently, other lipid derivatives, namely LXRα ligands, were found to dampen DC migration [21, 22]. Noteworthy, certain tumors were shown to produce LXR ligands that interfered with CCR7 expression on DCs and hence with their migration to lymph nodes resulting in a reduced antitumor immune response [22]. LXRs play critical roles in maintaining lipid homeostasis, particularly of whole-body cholesterol and control for instance cholesterol reabsorption in the intestine and are involved in atherosclerosis and dyslipidaemia [18]. Upon inflammation, LXR activation in macrophages regulates the expression of inflammatory mediators [30]. Because PGE2 and oxysterols were both shown to influence CCR7 expression and DC migration — although in an opposing way — and because both lipid derivatives can be produced by tumors and at sites of (chronic) inflammation, we aimed to elucidate the interplay of PGE2 signaling and LXR activation on CCR7 expression and CCR7-driven human DC migration. Here, we demonstrate that PGE2 counteracts LXR-mediated dampening of CCR7 expression and DC migration, suggesting that PGE2 and LXR ligands conversely regulate CCR7 expression and DC migration. Importantly, CCR7 expression on DCs does not correlate with the cell migration potential. Human MoDCs, as well as ex vivo PBDCs migrate only efficiently in response to chemoattractants if DCs were confronted with PGE2 [7, 9]. In fact, PGE2 stimulation of DCs must occur simultaneously with maturation to render DCs motile as stimulation of DCs with PGE2 shortly after maturation has been initiated did not lead to the development of a migratory DC pheno-type [9]. Moreover, endogenous production of PGE2 by MoDCs was shown not to be sufficient for developing a migratory phenotype [9], explaining why mPGES-1-deficient BMDCs produce less PGE2 but do not show migration deficits [34], whereas EP4-deficient DCs show severely impaired migration [12]. Although it is currently not fully clear how PGE2 regulates DC migration on a molecular level, PGE2 was shown to facilitate and enhance CCR7 signaling leading to migration [8]. Here, we show that PGE2 upregulates MMP9, known to be critical for DC migration [29, 35], in human MoDCs and that LXR activation has no effect on MMP9 transcription (Fig. 4). Moreover, we show that LXR activation does neither perturb downregulation of CCL4 and IL-23 (Fig. 4) known to be regulated by PGE2 [36, 37], nor alter the expression of the PGE2 synthesizing enzyme COX-2 in DCs (Fig. 4) formerly described to be reduced by LXRα activation in macrophages [30].

It is widely accepted that the environment in which DCs develop and undergo maturation, shapes the outcome of the immune response [11, 24]. There is a growing body of evidence that the time point when DCs encounter PGE2 and/or LXR agonists is crucial for their functions [19, 20, 23, 24, 38]. For instance, monocyte to DC differentiation is accompanied by the induction of LXRα expression, which can be further increased by its concomitant activation (Supporting Information Fig. 1A; [19, 20]). Maturation of MoDCs per se does not modulate LXRα expression (Fig. 1; [19]). However, the presence of PGE2 during MoDC maturation significantly reduces LXR expression and activation (Fig. 1 and 2) along with a preserved CCR7 expression and migratory ability of DCs (Fig. 3 and 5). In contrast, LXR activation in the absence of PGE2 of mature MoDCs strongly diminishes CCR7 expression (Fig. 3; [21, 22, 38]). Interestingly, stimulation of fully differentiated immature MoDCs with LXR agonists in combination with 9-cis-retinoic acid (9cRA), but in the absence of a maturation stimulus, was shown to induce CCR7 expression [38]. Thus, it is tempting to speculate that LXR agonists in combination with 9cRA, known to induce tolerogenic DCs [39], promotes tolerance in the absence of canonical “danger” signals by inducing CCR7 expression in non-mature/tolerogenic DCs [5]. In contrast, under inflammatory conditions, PGE2 is present during immunogenic DC maturation resulting in CCR7 expression and efficient homing [9, 12]. Addition of PGE2 alone to monocytes is known to prevent human monocyte to DC differentiation [23] and results in the development of immuno-suppressive myeloid-derived suppressor cells [24]. Here, we demonstrate that PGE2 counterbalances LXR-dependent gene transcription and rescues LXR-mediated downregulation of CCR7 expression as well as migratory abilities of DCs independent of prior LXR activation.

Depending on the maturation stimulus, PGE2 and LXRα agonists have both been shown to regulate CCR7 at the mRNA level [7, 22]. How CCR7 expression is transcriptionally regulated remains largely elusive. However, inhibition of NF-κB activation was shown to impair CCR7 expression in Hodgkin disease-derived cells [28]. Therefore, we investigated on the activation of distinct NF-κB subunits in MoDCs matured in the presence or absence of PGE2 and/or LXR activation. MoDCs matured in the presence of PGE2 show enhanced cRel activity (Supporting Information Fig. 3C). In contrast, MoDCs matured in the presence of T0901317 show reduced p52, p65, and RelB activities, whereas, MoDCs matured in the presence of both T0901317 and PGE2 show normal p52 and RelB activities (Supporting Information Fig. 3C). Our data indicate that PGE2 preserves the activation of distinct NF-κB subunits in LXR-activated MoDCs that are presumably involved in the regulation of CCR7 expression.

Taken together we provide evidence that PGE2 interferes with LXR activation, downregulates LXRα expression and rescues the migratory ability of DCs to migrate toward CCR7 ligands. Our findings add to the complexity of how the immune system can be regulated by lipid derivates and provide a new perspective how to modulate immune responses in health and diseases by pharmacologically targeting either the PGE2 or the LXR signaling pathway.

Materials and methods

Generation of human MoDCs

PBMCs from healthy donors were enriched by density gradient centrifugation on Ficoll Paque Plus (Amersham Biosciences, Uppsala Sweden) and monocytes were isolated using anti-CD14-conjugated microbeads (Miltenyi Biotec, Bergisch-Gladbach, Germany). Monocytes (2 × 106 cells/mL) were differentiated to immature MoDCs in serum-free AIM-V medium (Gibco, Paisley, UK) supplemented with 50 ng/mL GM-CSF (PeproTech, London, UK) and 50 ng/mL IL-4 (PeproTech) or 10 ng/mL IL-13 (PromoCell, Heidelberg, Germany) to assess COX-2 expression. At day 2 of differentiation, the same amount of medium (including IL-4 and GM-CSF) was added. After 5–6 days, immature DCs were harvested and matured for 2 days by adding 0.5 μg/mL soluble trimeric CD40L (sCD40L; PromoCell), 20 μg/mL polyI:C (Sigma, Saint Louis, MO), 10 μg/mL LPS (Salmonella abortus equi; Sigma) or a cocktail of cytokines including 20 ng/mL TNF-α, 20 ng/mL IL-6, and 10 ng/mL IL-1β (PeproTech and PromoCell). MoDCs were matured in the presence or absence of 1 μg/mL PGE2 (Minprostin® E2, Pharmacia, Uppsala, Sweden). Where indicated, cells were treated with 1 μM T0901317 (Sigma). Cell morphology was assessed using a Zeiss Axiovert M200 microscope equipped with an AxioCam MRm Rev.3 camera and a 10× objective and analyzed using the AxioVision Software. The study using human blood was approved by the cantonal ethics committee and all participants gave written informed consent.

Peripheral blood DCs

Ex vivo myeloid CD1c+ DCs were isolated from PBMCs using the CD1c (BDCA-1) DC Isolation Kit (Miltenyi Biotec). PBDCs were directly matured with the cytokine cocktail for 18–24 h as described for MoDCs.

Flow cytometry

DCs were stained for 1 h at 4°C in PBS containing 1% FCS and 0.02% sodium azide with phycoerythrin or allophycocyanin conjugated anti-CD80, anti-CD83, anti-CD86, (BD Biosciences, Erembodegen, Belgium), anti-CCR7 (R&D Systems, Minneapolis, MN), and isotype-matched control antibodies and analyzed on a LSRII flow cytometer (BD Biosciences). Relative CCR7 expression was calculated by subtraction of isotype mean fluorescence intensity (MFI) from specific CCR7 MFI and related to DCs matured in presence of PGE2. Data were assessed using FlowJo Software (TreeStar, Inc., OR).

Chemotaxis assays

For 2D-Transwell migration, 1 × 105 DCs in AIM-V medium were placed on a polycarbonate filter with a pore size of 5 μm in a 24-well Transwell plate (Corning Costar) and allowed to migrate toward the lower chamber containing 250 ng/mL CCL21 (PeproTech) for 3 h. For 3D matrigel assays, polycarbonate filters were precoated with 60 μg/filter of Matrigel (BD Biosciences) solved in AIM-V medium according to the manufacturer's instructions. DCs (2 × 105) were allowed to transmigrate in response to CCL21 for 22 h. Input and migrated MoDCs were counted by flow cytometry for 60 s. The number of spontaneously migrated cells toward medium without chemokine was subtracted and related to DCs matured in the presence of PGE2.

Quantitative real-time RT-PCR

Total RNA of MoDCs was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) and transcribed into cDNA using random hexamer primers and the Hi Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Rotkreuz, Switzerland). Amplification of human NR1H3 (LXRα) and MMP9 transcripts was performed using the Fast SYBR Green PCR Master Mix on a 7900HT Fast Real-Time PCR System (Applied Biosystems) with an initial denaturation step at 95° for 20 s followed by 40 cycles of 1 s at 95°C and 20 s at 60°C. Used primers (200 nM each): CCL4: 5′-CGCCTGCTGCTTTTCTTACAC, 5′-GGTTTGGAATACCACAGCTGG; COX-2: 5′-GCCTTCTCTAACCTCTCC, 5′-CTGATGCGTGAAGTGC; MMP9: 5′-CCAGTCCACCCTTGTGCTC, 5′-TTTCGACTCTCCACGCATCTC; NR1H3 (LXRα): 5′-GCTGCAAGTGGAATTCATCAACC, 5′-ATATGTGTGCTGCAGCCTCTCCA; ß2M (ß-2 microglobulin): 5′-GCTATCCAGCGTACTCCAAAGATTC, 5′-CAACTTCAATGTCGGATGGATGA; UBC (ubiquitin C): 5′-ATTTGGGTCGCGGTTCTTG, 5′-TGCCTTGACATTCTCGATGGT. For the amplification of ABCG1 and NR1H2 (LXRβ) mRNA the following Taqman® Gene Expression Arrays in combination with the appropriate Fast Universal Master Mix (Applied Biosystems) with the same thermal conditions were used: ABCG1: Hs00245154_m1, Hs01555190_g1; NR1H2: Hs00190268_m1. mRNA expression was normalized to ß2M and UBC. Relative mRNA expression was calculated by the ΔΔCt-method.

Cytokine production

Immature MoDCs were stimulated with the cytokine cocktail in the presence or absence of PGE2 and/or T0901317. After 2 days, culture supernatants were collected and analyzed using the human IL-23 Ready-Set-Go! ELISA Set (eBioscience, San Diego, CA).

Cell lysis and immunodetection

MoDCs were harvested and solubilized in lysis buffer containing 1% NP40 and protease inhibitor cocktail (Roche). Samples were separated under reducing conditions on SDS-PAGE and analyzed by western blot using an anti-human LXRα-specific mAb (Abcam, Cambridge, UK) and a mAb against β-actin (Cell Signaling, Danvers, MA).

Statistical evaluation

Statistical analysis was assessed by Student's two-tailed paired t test, Friedman with Dunn post test, or repeated measures ANOVA with Bonferroni post test in data sets fulfilling the criteria for normality testing (n ≥ 5) using GraphPad InStat (GraphPad Software, Inc., San Diego, CA).


We are grateful to Marc Müller for assistance in blood collections, and Dr. Farhan, Groettrup, and Dr. Basler for helpful discussions. This study was supported in parts by research funding from the Swiss National Science Foundation (SNF 31003A-127474/1), the Pablo Frolich Stiftung, the Thurgauische Stiftung für Wissenschaft und Forschung, the Swiss State Secretariat for Education and Research, and the Thurgauische Krebsliga to DFL.

Conflict of interest

The authors declare no financial and commercial conflict of interest.


liver X receptor


monocyte-derived DC


prostaglandin E2