Article Text

Download PDFPDF

The Anti-Inflammatory Effects of Prostaglandins
  1. Jose U. Scher, MD*†,
  2. Michael H. Pillinger, MD*†
  1. From the *Division of Rheumatology, Department of Medicine, NYU School of Medicine/NYU Hospital for Joint Diseases; and †Department of Medicine, VA NY Harbor Healthcare System, NY Campus, New York, NY.
  1. Received November 4, 2008, and in revised form December 10, 2008.
  2. Accepted for publication December 10, 2008.
  3. Reprints: Michael H. Pillinger, MD, Division of Rheumatology NYU Hospital for Joint Diseases, 301 E 17th St, New York, NY 10003. E-mail: mhpillinger{at}
  4. This symposium was supported in part by a grant from the National Center for Research Resources (R13 RR023236).


Long regarded as proinflammatory molecules, prostaglandins (PGs) also have anti-inflammatory effects. Both prostaglandin D2 (PGD2) and its dehydration end product 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2) seem to play important roles in regulating inflammation, via both receptor-dependent (DP1 and DP2 receptors) and receptor-independent mechanisms. Intracellular effects of PGD2 and 15d-PGJ2 that may suppress inflammation include inhibition of nuclear factor-κB (NF-κB) by multiple mechanisms (IκB kinase inhibition and blockade of NF-κB nuclear binding) and activation of peroxisome proliferator-activated receptor-γ (PPAR-γ). Prostaglandin F (PGF) may also have important anti-inflammatory effects, although current data are limited. In animal models, expression of both PGD and PGF synthases declines during acute inflammation, only to rise again during the resolution phase, suggesting their possible role in resolving inflammation. Prostaglandin E2 (PGE2), the classic model of a proinflammatory lipid mediator, also has anti-inflammatory effects that are both potent and context dependent. Thus, accumulating data suggest that PGs not only participate in initiation, but may also actively contribute to the resolution of inflammation. Indeed, classic inhibitors of PG synthesis such as nonselective and cyclooxygenase-2 (COX-2) selective inhibitors (nonsteroidal anti-inflammatory drugs) may actually prolong inflammation when administered during the resolution phase. These effects may regulate not only tissue inflammation but also vascular disease, possibly shedding light on the controversy surrounding nonsteroidal anti-inflammatory drug use and its relation to myocardial infarction. In this review, we summarize the current understanding of PGs as dichotomous molecules in the inflammatory process.

Key Words
  • prostaglandin
  • inflammation
  • cardiovascular disease
  • cyclooxygenase

Statistics from

Key Words

Prostaglandins (PGs) are the end products of a series of enzymatic events acting on membrane phospholipids. First, a cytosolic phospholipase A2 (cPLA2) liberates arachidonic acid from the lipid bilayer.1Subsequently, a heterobifunctional cyclooxygenase (endoperoxide synthase)-COX-1 or COX-2-acts on arachidonic acid to generate prostaglandin G2 (PGG2), and to convert PGG2 to prostaglandin H2 (PGH2).2Prostaglandin H2 serves as the substrate for a number of additional reactions, each resulting in a different PG end product. For example, the action of one of several PGE synthases (PGESs) results in the generation of PGE2. To date, 3 different PGESs have been identified (microsomal PGES [mPGES]-1, mPGES-2, and cytosolic PGES [cPGES]), whose individual roles are being studied.3-6Other terminal synthases specifically convert PGH2 into thromboxane A2 (TXA2), PGF, PGD2, or prostacyclin (PGI2).7-11Thus, the specific distribution of PG products in a given cell is determined, not by COX enzymes, but by the PG synthase complement contained.

For many readers, the term PG may be synonymous with the idea of inflammation, and indeed, it has long been appreciated that PGE2 has important proinflammatory effects. The efficacy and popularity of COX inhibitors (nonsteroidal anti-inflammatory drugs [NSAIDs]) testifies to the role of PGEs in pain and inflammation. However, inflammation is not the only role of PGs, and even PGE2 plays important roles in multiple homeostatic effects, including maintenance of the gastric mucosa and regulation of renal function. More recently, it has come to be appreciated that some PGs, including PGE2 itself, have actions that are potently anti-inflammatory. Thus, PGs may be poised to regulate both the onset and resolution of the inflammatory response. In this paper, we review some of the data that inform this new and more nuanced view of PG function.


Prostaglandin D2 serves multiple functions, of which the best known may be its ability to induce sleep.12-14Two distinct PGD2 synthases have been described. Hematopoietic PGDS (H-PGDS) is most abundant in mast cells, whereas lipocalin-type PGDS (L-PGDS) is found predominantly in brain tissue.8,9Human articular cartilage expresses both forms.15,16Two distinct PGD2 receptors have also been described; DP1 is widespread, whereas DP2 (also known as chemoattractant receptor-homologous molecule expressed on TH2 cells [CTRH2]) seems to be mainly localized to TH2 lymphocytes.17,18Both receptors are G protein coupled but signal via different second messengers. Engagement of DP1 results in protein kinase A activation, the effects of which seem to be largely anti-inflammatory. In contrast, DP2 acts to raise intracellular calcium levels; in T cells, this may lead to activation (Fig. 1).


15-Deoxy-Δ12,14-PGJ2 acts on receptor-dependent and receptor-independent targets to regulate cell responses.

Prostaglandin D2 also serves, via enzymatically independent reactions, as a source for the generation of PGJ2 adducts. Specifically, PGD2 can undergo spontaneous dehydration, leading to the formation of PGJ2. Subsequently, PGJ2 can, via a second spontaneous dehydration, be converted into 15-deoxy-Δ12,14-prostaglandin J2 (15d-PGJ2), the best studied of the PGJ2 products. In the presence of serum albumin, PGJ2 can also undergo conversion to 12d-PGJ2.19-21

Although 15d-PGJ2 can engage both DP1 and DP2, its main targets seem to be intracellular (Fig. 1). Among the most important are NF-κB and the peroxisome proliferator-activated receptor-γ (PPAR-γ). 15d-PGJ2 12,14-PGJ2 inhibits IκB kinase, the activator of the NF-κB system.2215d-PGJ2 12,14-PGJ2 may also directly block the interaction of NF-κB with its nuclear binding sites, reducing the inflammatory response.2315d-PGJ2 also stimulates PPAR-γ, which in turn mediates anti-inflammatory effects.24-28For example, PPAR-γ activation up regulates the NF-κB inhibitor IκB.29Other relevant 15d-PGJ2 effects include activation of the Ras/Erk and PI-3K pathways.30

In vitro, 15d-PGJ2 has been shown to inhibit secretion of interleukin (IL)-6, IL-1β, IL-12, and tumor necrosis factor-α (TNF-α) from macrophages, and to down regulate the expression of inducible nitric oxide synthase (iNOS).24,31,32In an in vitro crystal model of inflammation, 15d-PGJ2 inhibited cytokine generation by macrophages to a greater degree than the PPAR agonist troglitazone, suggesting that 15d-PGJ2 acts on targets in addition to PPAR-γ.33In endothelial cells, 15d-PGJ2 inhibits CXC chemokine production, stimulates apoptosis, and has variable effects on cellular adhesion molecules such as intercellular adhesion molecule (ICAM).26,34,35

Animal studies have confirmed the anti-inflammatory effects of 15d-PGJ2. Administration of 15d-PGJ2 inhibits adjuvant-induced arthritis in rats, in a dose-dependent manner.36In a mouse air pouch model, 15d-PGJ2 inhibited uric acid-induced acute inflammation more effectively than troglitazone, again suggesting that 15d-PGJ2 may have effects beyond the inhibition of PPAR-γ.33

Studies in which the cellular expression of PGDS is altered have provided additional insight into the role of PGD2 and/or PGJ2 in inflammatory responses. Murakami et al.37used a retroviral transfection system to increase PGD2 expression in a mouse air pouch model of inflammation; the result was reduction of exudate volume and inhibition of neutrophil infiltration. Akahoshi et al.33measured wild type expression of PGDS in the mouse air pouch model. With the initiation of inflammation, PGDS expression (measured as messenger RNA levels) declined precipitously, in both infiltrating leukocytes and the soft tissue of the pouch. During the period of peak inflammation, defined by increases in TNF-α and other inflammatory markers, PGDS levels fell to nearly 0. However, as inflammation resolved, PGDS levels rose again, eventually returning to baseline.33Was the rise in late phase PGDS a cause, or a consequence of the end of the inflammatory response?

The answer may reside in the studies by Rajakariar et al.,38who examined the inflammatory response in PGDS knockout mice. The authors used a peritonitis model, in which zymosan was injected into the abdomens of the mice. Compared with wild type littermates, the knockout mice experienced a quicker and larger inflammatory response. Critically, in the PGDS knockout mice, inflammation failed to resolve in a timely manner.38These data suggest that PGDS levels may be programmed to decline early but rise again to promote the resolution of the inflammatory cycle.

One of the more interesting implications of such a model is the possibility that NSAIDs, given early and persistently by physicians for inflammatory diseases, might actually prolong inflammation. More precisely, COX inhibition during the late phase of the inflammatory response might prevent the resolution of inflammation by inhibiting PGDS activity and the production of PGD2/PGJ2. Gilroy et al.39tested this hypothesis using a carrageenin-induced pleurisy model. They confirmed the observation of Akahoshi et al that PGDS expression falls early in acute inflammation but returns to normal as inflammation resolves. The fall and rise of PGDS were accompanied by a concomitant fall and, then, rise in the levels of both PGD2 and PGJ2. When the mice were treated with indomethacin before carrageenin administration, inflammation was significantly abrogated. However, when the mice were treated with indomethacin after the onset of the inflammatory response, inflammation was enhanced and resolution delayed. Similar results were obtained with the COX-2-selective inhibitor NS-398.39These data support an anti-inflammatory role for PGD2/PGJ2 in the late phase of inflammation and suggest that the timing of NSAID administration to patients might have an impact on their efficacy.


Prostaglandins D2 and J2 are not the only PGs that are potentially anti-inflammatory, but the effects of other PGs seem to be mixed. Prostaglandin F is a case in point. In mouse studies by de Menezes et al.,40infusion of PGF was associated with increased cellular infiltration, consistent with a proinflammatory effect. In contrast, Collville-Nash et al.41used a pleural inflammation model to examine the kinetics of PGF2 synthesis, such as PGD2; PGF levels declined during the onset of inflammation and rebounded during the inflammatory phase. Both indomethacin and NS-398 inhibited PGF synthesis but stimulated inflammation as defined by exudate volume and inflammatory cell number. These data do not conclusively define a role for PGF, because the effect of the COX inhibitors could have been because of effects on other PGs such as PGD2 and 15d-PGJ2. However, the proinflammatory effect of COX inhibitors was reversed by the addition of a selective PGF2 analog.41Thus, PGF seems to contribute, at least to a degree, to the late phase of inflammation resolution.


Prostaglandins of the E series are important mediators of inflammation and promote vasodilatation, smooth muscle contraction, and pain. Nonetheless, studies suggest that in some contexts, PGEs can be anti-inflammatory. Seminal investigations by Zurier et al.42-44showed that PGE analogs ameliorate animal models of inflammatory arthritis and reverse nephritis in a New Zealand black/New Zealand white mouse model of lupus. Rossetti et al.45demonstrated that liposomal PGE1 can inhibit uric acid-induced inflammation in the mouse air pouch model. Aronoff et al.46have reported that PGE2 inhibits macrophage phagocytosis in vitro, and our own group has observed that PGE1 inhibits neutrophil adhesion, a critical early feature of acute inflammation.47

We recently examined the role of PGE2 in synovial fibroblasts, key cells in the inflammatory responses of rheumatoid arthritis.48In some of these investigations, we took as our readout the secretion of matrix metalloproteins (MMPs), important mediators of rheumatoid tissue destruction. Prostaglandins E1 and E2, but not PGF, inhibited MMP-1 secretion in response to IL-1β and TNF-α. Conversely, the nonselective COX inhibitors indomethacin, ibuprofen, and 6-methoxy-2-naphthyl acetic acid (6MNA; active form of the prodrug nabumetone), and the COX-2 selective inhibitors celecoxib, NS-398, and SC-299 enhanced MMP-1 secretion. The mechanism of action of PGEs on MMP-1 secretion seemed to result from their ability to inhibit Erk, an established regulator of MMP-1 secretion. Prostaglandins inhibited, but COX inhibitors enhanced Erk activation (Fig. 2).49,50


Erk and cyclooxygenase interactions in synoviocyte signaling for MMP-1 release. UO126 and PD098059 are specific inhibitors of the proximal Erk activator, Mek.

As noted earlier, the NF-κB signaling pathway is an important regulator of inflammation, mediating transcription of more than 40 inflammation-related proteins.48Nuclear factor-κB is typically a heterodimer composed of the subunits p50 and p65 (RelA), which are locked into the cytosol by IκB. IκB kinase phosphorylates IκB and induces its degradation, freeing p50/p65 to translocate to the nucleus and carry out its effects.51

We tested the effects of PGE2 on NF-κB signaling and were surprised to observe that PGE2 inhibited NF-κB activity. When we examined the translocation of the individual NF-κB subunits to the nucleus, the results were equally unexpected: PGE2 inhibited the nuclear translocation of p65, but enhanced that of p50. The explanation for this dichotomy seems to reside in the fact that p50/p65 heterodimers are not the only form that NF-κB molecules can take. In particular, p50/p50 homodimers may assemble, translocate to the nucleus, and bind to the same DNA sites as the heterodimers. Because p50/p50 homodimers lack intrinsic signaling capacity, they competitively inhibit p50/p65. Our data suggest that PGE2 can inhibit p50/p65, while stimulating p50/p50 translocation. Consistent with these observations, both celecoxib and ibuprofen enhanced p65, while partially inhibiting p50 nuclear translocation. In addition to p50/p50-dependent effects, PGE2 also enhanced the expression of IκB, by an as-yet undetermined mechanism. Given that NF-κB induces COX-2 synthesis and subsequent PGE2 generation, we hypothesize that PGE2 provides negative feedback for the NF-κB pathway (Fig. 3).49,52


Downregulation of NF-κB activation by PGEs.

How can PGE2 have anti-inflammatory effects? In the case of NF-κB, one possibility is that the cell relies on kinetics to separate proinflammatory and anti-inflammatory processes. In this model, PGE2 generation, delayed until NF-κB activation has led to COX-2 expression, sets an alarm clock for NF-κB inhibition. A second possibility is that the effects of PGE2 on inflammatory responses are context dependent. In studies with gastric epithelial cells, PGE2 inhibited Erk activation and MMP-1 secretion in the presence of cytokines but had the opposite effect when cytokines were absent.53A third possibility is that the effects of PGE2 are concentration-dependent. In support of this model, Tchetina et al.54have reported that very low PGE2 concentrations inhibit, whereas higher concentrations enhance, chondrocyte-dependent collagen cleavage in osteoarthritis cartilage. Yet another possibility is that the effects of PGE2 may depend upon intracellular localization and/or targeting (Pillinger et al, unpublished observation). These models are not mutually exclusive, and the net result of PGE2 on inflammatory processes may well depend upon a complex interaction of time, concentration, context, and locale.


One of the more interesting advances in the past decade is the insight that vascular disease is primarily an inflammatory process. With that in mind, several observers have asked whether PGD2, 15d-PGJ2, or even PGE2 may play a salutary role in controlling coronary artery and other forms of vascular disease. Kaplan et al approached this question in a rat model of ischemia-reperfusion injury. Ischemia-reperfusion injury resulted in cardiac tissue damage, inflammatory infiltration, and a rise in CPK and myeloperoxidase. Administration of 15d-PGJ2 before injury protected against both tissue damage and inflammation. In contrast, the PPAR-γ agonist ciglitazone protected against damage but not inflammation, suggesting yet again that 15d-PGJ2 has both PPAR-γ-dependent and PPAR-γ-independent effects. Consistent with this interpretation, the PPAR-γ antagonist GW-9662 reversed the effect of ciglitazone but not 15d-PGJ2.55

A cardinal feature of atherosclerosis is the formation of foam cells (lipid-laden macrophages) in the lining of arteries. Chan et al.56examined the possible anti-inflammatory role of PGs on foam cell formation in vitro. Treatment of THP-1 human macrophages with an NSAID or selective COX-2 inhibitor caused lipid accumulation and a foam cell phenotype. The mechanism of this effect was related to down-regulation of 27-hydroxylase and adenosine 5′-triphosphate-binding cassette, subfamily A, member 1 (ABCA1), enzymes that play critical roles in reverse cholesterol transport. The effects of COX inhibition could be reversed by either PGD2 or PGE2. These data underline the importance of PGs to vascular biology, and suggest that the ability of COX inhibitors to increase cardiac risk may relate not only to inhibition of prostacyclin generation57but also to direct effects on the inflammatory responses of the atheromatous plaque.

Could anti-inflammatory PGs be protective against vascular disease in humans? A single study has addressed this question. Cipollone et al.58examined carotid artery specimens from patients undergoing endarterectomy for atherosclerotic disease. Patients were categorized as symptomatic versus asymptomatic, on the supposition that symptomatic patients represented a more severely affected cohort. Symptomatic patients expressed less PGDS, and less PPAR-γ, than their asymptomatic counterparts. The authors concluded that a loss of endovascular PGDS and/or PPAR-γ may contribute to the atherosclerotic phenotype.58


Because many PG studies involve the use of COX inhibitors, it is important to note that some COX inhibitors may also have COX-independent anti-inflammatory properties. For example, nonacetylated salicylates are relatively poor COX inhibitors but have a plethora of other effects, including the inhibition of NF-κB59and Erk.60Aspirin, or acetylsalicylic acid, was first reported to inhibit COX enzymes by Vane in 1970.61However, Serhan has recently shown that whereas aspirin inhibits COX-1, it may actually divert the activity of COX-2 into the generation of lipoxins, potent anti-inflammatory lipids that participate in the resolution of inflammation.62,63Nabumetone, a prodrug that is converted by the liver into the COX inhibitor 6MNA, actually has its own COX-independent ability to inhibit NF-κB.49These, and other COX-independent effects, may need to be considered both in the clinic and when interpreting laboratory studies using these agents.


It is increasingly clear that any paradigm identifying PGs as exclusively proinflammatory is both limited and in need of revision. Even PGE2, a classic promoter of inflammation, can be potently anti-inflammatory in specific contexts. Other PGs, particularly PGD2 and 15d-PGJ2, exert predominantly anti-inflammatory effects, via targets such as PPAR-γ and NF-κB. These effects may be segregated in time and contribute to the resolution of acute inflammation. The so-called NSAIDs, by inhibiting PG production, may actually have mixed anti-inflammatory and proinflammatory effects, a phenomenon deserving consideration when deciding how best to use NSAIDS and selective COX-2 inhibitors. Indeed, in animal models, administration of an NSAID early in the course of inflammation seems to abrogate inflammation, whereas administration of an NSAID after the onset of inflammation resolution may actually prolong the inflammatory response. These mixed effects may have implications for coronary artery disease, including the effects on NSAIDs on the risk of myocardial infarction. Future studies will further elucidate the role(s) of PGs in the yin/yang of inflammation and permit a more directed and effective exploitation of PG inhibitors.


The authors thank Edwin Chan and Allison Reiss for encouraging the writing of this manuscript, Steven Abramson for supporting the authors' research described herein, and Jean Park, Nada Marjanovic, Aryeh Abeles, Matthew Axelrod, Mark Fisher, Mara Pillinger, Victoria Dinsell, and Sonia Tolani for their contributions to our investigations.


View Abstract

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.