Abstract
Prostaglandin E2 (PGE2) is the most abundant prostaglandin in the human body. It has a large number of biological actions that it exerts via four types of receptors, EP1–4. PGE2 is formed from arachidonic acid by cyclooxygenase (COX-1 and COX-2)-catalyzed formation of prostaglandin H2 (PGH2) and further transformation by PGE synthases. The isomerization of the endoperoxide PGH2 to PGE2 is catalyzed by three different PGE synthases, viz. cytosolic PGE synthase (cPGES) and two membrane-bound PGE synthases, mPGES-1 and mPGES-2. Of these isomerases, cPGES and mPGES-2 are constitutive enzymes, whereas mPGES-1 is mainly an induced isomerase. cPGES uses PGH2 produced by COX-1 whereas mPGES-1 uses COX-2-derived endoperoxide. mPGES-2 can use both sources of PGH2. mPGES-1 is a member of the membrane associated proteins involved in eicosanoid and glutathione metabolism (MAPEG) superfamily. It requires glutathione as an essential cofactor for its activity. mPGES-1 is up-regulated in response to various proinflammatory stimuli with a concomitant increased expression of COX-2. The coordinate increased expression of COX-2 and mPGES-1 is reversed by glucocorticoids. Differences in the kinetics of the expression of the two enzymes suggest distinct regulatory mechanisms for their expression. Studies, mainly from disruption of the mPGES-1 gene in mice, indicate key roles of mPGES-1-generated PGE2 in female reproduction and in pathological conditions such as inflammation, pain, fever, anorexia, atherosclerosis, stroke, and tumorigenesis. These findings indicate that mPGES-1 is a potential target for the development of therapeutic agents for treatment of several diseases.
I. Introduction
The prostaglandins are biologically active derivatives of arachidonic acid and other polyunsaturated fatty acids that are released from membrane phospholipids by phospholipase A2. The initial transformation of arachidonic acid involves oxygenation and cyclization to an unstable endoperoxide intermediate, prostaglandin (PG1) G2 by cyclooxygenase (COX) enzymes. The same enzymes reduce PGG2 to PGH2 via a separate peroxidase site. Various isomerases and oxidoreductases convert PGH2 to prostaglandins and thromboxane A2 (Samuelsson et al., 1978) (Fig. 1).
The cyclooxygenase exists in two forms, the constitutive COX-1 and the induced form, COX-2. Whereas the traditional NSAIDs inhibit both COX-1 and COX-2, the more recently developed anti-inflammatory coxibs inhibit COX-2 preferentially (Simmons et al., 2004). Prostaglandin E2 exerts its actions via four types of receptors, viz. EP1, EP2, EP3, and EP4. PGE2 plays a role both in normal physiology and in pathology (Kobayashi and Narumiya, 2002). The biological actions include areas such as inflammation, pain, tumorigenesis, vascular regulation, neuronal functions, female reproduction, gastric mucosal protection, and kidney function.
The formation of PGE2 from PGH2, formed by COX-1 or COX-2 is catalyzed by isomerases, PGE synthases. There has been an explosive development of our understanding of these enzymes triggered by the discovery of mPGES-1, which belongs to a recently defined superfamily [membrane-associated proteins in eicosanoid and glutathione metabolism (MAPEG)] (Jakobsson et al., 1999a). This superfamily also contains leukotriene C4 synthase and 5-lipoxygenase-activating protein as well as three glutathione transferases/peroxidases called microsomal glutathione transferase 1–3 (MGST1–3). Microsomal glutathione transferase 1 is the most well characterized for substrate specificity, kinetic mechanism, and three-dimensional structure (Holm et al., 2006) and, being the closest relative to mPGES-1 (38% identity), serves as a good model for generating hypotheses on structure function relationships. mPGES-1 has emerged as a potential target for the development of drugs for treatment of inflammation, pain, cancer, atherosclerosis, and stroke.
II. Discovery and Characterization
A. Identification
Many attempts have been made to purify the protein(s) involved in PGE2 production (Ogino et al., 1977; Moonen et al., 1982). The results indicated that several different PGE2 synthases exist (Tanaka et al., 1987; Watanabe et al., 1997). The first PGE2 synthase was identified in 1999 (Jakobsson et al., 1999b). In this study human MGST1-like-1 (belonging to the MAPEG superfamily) expressed in Escherichia coli was shown to display PGE2 synthase activity. Moreover, the synthase was induced by proinflammatory cytokines. This synthase was named mPGES-1. Subsequently two other PGE2 synthases, cytosolic PGES (cPGES) and mPGES-2 were discovered (Watanabe et al., 1999; Tanioka et al., 2000). The focus of this review is mPGES-1, which seems to play a prominent role in pathophysiology, whereas the other prostaglandin E2 synthases identified, mPGES-2 and cPGES, are constitutively expressed and function in basal PGE2 production.
B. Purification and Structure
The identity of mPGES-1 and the possibility of establishing heterologous expression systems formed the basis for purification from Sf9 insect cells (Ouellet et al., 2002) and E. coli (Thorén et al., 2003). Both systems yield proteins that display apparent homogeneity and comparable activity, allowing for differences based on temperature used and detergents included. Purification from both Sf9 cells and E. coli involves hydroxyapatite fractionation whereas, in addition, immobilized metal ion chromatography is used in the enzyme purification from E. coli for which an N-terminal His-tag was engineered into the expression construct. Apparently the His tag does not change the enzymatic properties to a great extent. The main reason for the successful purification was most likely a high amount of starting material as well as inclusion of glutathione during all steps.
A projection structure of mPGES-1 at 10 Å (Thorén et al., 2003) revealed similar structural properties as had been determined for MGST1 (Holm et al., 2002), suggesting that the enzyme is a trimer of four helix bundles in which the hydrophobic helices traverse the membrane (Fig. 2). The structural resemblance is consistent with the close relation between mPGES-1 and MGST1 (38% identity at the amino acid level) (Jakobsson et al., 1999b). The hydropathy and charge distribution of mPGES-1 suggest that the N terminus (and C terminus) faces the lumen of the endoplasmic reticulum; however, this suggestion needs to be demonstrated experimentally.
Suggestions on active site residues in mPGES-1 have largely been based on phylogenetic information on the MAPEG superfamily, which contains six human members (and another hundred related proteins identified throughout phylogeny, with the notable exception of archaea) (Jakobsson et al., 1999a). Human members are the well characterized 5-lipoxygenase activating protein, leukotriene C4 synthase, and MGST1, as well as the less studied MGST2 and 3 (Sjöström et al., 2001; Bresell et al., 2005). Multiple alignments (Jakobsson et al., 1999a) and earlier mutagenesis experiments (Lam et al., 1997) of related proteins indicated some essential residues. Later it was shown that Arg110 was essential for catalytic function (Ser replacement resulted in inactive enzyme), whereas Tyr117 and Arg70 were not (Murakami et al., 2000). Chemical modification experiments with a thiol reagent (Thorén et al., 2003) showed that cysteine residues are important for catalysis. Computational approaches (Huang et al., 2006) have also been used to build a model of the active site of mPGES-1. From this model a number of predictions were made on active site residues (Gln36, Arg110, Thr114, Tyr130, and Gln134), whose replacement resulted in lowering of activity to between 60 and 15%. In this study Arg110 was replaced with Thr and maintained close to 20% of wild-type activity. On the basis of the recent three-dimensional structure of MGST1 at 3.2 Å resolution (Holm et al., 2006) and the sequence alignment to mPGES-1, none of the above mutated residues (where activity was altered) is implicated in the GSH binding site. No doubt the structure of MGST1 will guide experiments to elucidate the structure-function relationships in mPGES-1. Notwithstanding, the location of the PGH2 binding site still has to be inferred and confirmed by mutagenesis and/or crystallography. The fact that mPGES-1 forms well ordered two-dimensional crystals (Thorén et al., 2003) is promising and offers the prospect of a structural model.
C. Assays and Kinetic Properties
Historically many different assays have been used to measure PGE2 synthesis including measurement of PGH2 consumption (Basevich et al., 1983), high-performance liquid chromatography and UV/radiometric detection of PGE2 after extraction (Sraer et al., 1981; Thorén and Jakobsson, 2000) as well as immunochemical measurements of PGE2. Interest for mPGES-1 as a drug target has led to the development of novel assays suited for high-throughput formats (David Percival, 2003). These assays involve coupling to a second enzymatic step or fluorescence polarization. The main obstacle in the assay is instability of PGH2, which decomposes rapidly in aqueous solutions. Assays with short incubation times and low temperatures are generally used to optimize sensitivity. mPGES-1 also displays glutathione transferase and peroxidase activity (Thorén et al., 2003) that is comparatively low (another feature linking the enzyme to the MAPEG superfamily) which can be used in alternate, more robust assays (Thorén et al., 2003) to screen for inhibitors; however, inhibitor potency toward different substrates might vary.
In keeping with its prominent role in PGE2 production, mPGES-1 displays high catalytic efficiency toward PGH2 (310 mM-1 s-1) (Thorén et al., 2003). The reported Km values for PGH2 span an order of magnitude (Ouellet et al., 2002; Thorén et al., 2003) apparently resulting from different assay conditions (Table 1). Glutathione is specifically required for catalysis with a Km of 0.71 mM (Thorén et al., 2003), but, as the reaction is an oxidoreduction, there is no net consumption of GSH. Thus, GSH levels in most cell types would be above the measured Km (Meister, 1983). Prostaglandin G2 can substitute for PGH2 with similar kinetics and hence is an alternate substrate whose product, 15-hydroperoxy-PGE2, can be reduced by cyclooxygenase or other peroxidases (Nugteren and Hazelhof, 1973; Arthur, 2000; Thorén et al., 2003), leading to PGE2 via an alternate pathway. As mentioned above, mPGES-1 catalyzes the reduction of lipophilic hydroperoxides of both endogenous and xenobiotic origin (arachidonic acid hydroperoxide, 15-hydroperoxy-PGE2, and cumene hydroperoxide) as well as the conjugation of 1-chloro-2,4-dinitrobenzene, thus displaying GSH peroxidase and transferase activities (Thorén et al., 2003). The activities are lower than those of MGST1. By taking into account the localization and coregulation with COX-2, these activities can be assumed to be remnants of activities of a common ancestor for mPGES-1/MGST1. However the possibility of specific functions with as yet uncharacterized substrates should not be ruled out. The detailed kinetic and chemical mechanism of mPGES-1, including the number of active sites per homotrimer, remains to be determined. The chemical mechanism suggested for mPGES-2 (Yamada et al., 2005) serves as a good starting hypothesis.
D. Inhibition
Some mPGES-1 inhibitors have been characterized. The published inhibitors have been based on COX inhibitors already developed, arachidonic/prostaglandin analogs, and inhibitors developed for related proteins in the MAPEG family. In a screen using NSAIDs, stable prostaglandin H2 analogs, and cysteinyl leukotrienes (Thorén and Jakobsson, 2000), NS-398, sulindac sulfide, and leukotriene C4 elicited inhibition in the low micromolar range. Paracetamol is not an inhibitor (Greco et al., 2003). In another screen that used fatty acids, their analogs, and prostaglandins (Quraishi et al., 2002), 15-deoxy-Δ12,14-prostaglandin J2 and several fatty acids were found to inhibit at 0.3 μM (IC50). It was suggested that production of 15-deoxy-Δ12,14-prostaglandin J2 could aid resolution of inflammation by this inhibitory pathway. The 5-lipoxygenase activating protein (a non-catalytic member of the MAPEG family) indole inhibitor MK-886, which displays an IC50 of 1.6 μM, was used as a lead structure to develop nanomolar inhibitors (Riendeau et al., 2005). The inhibitors were less potent in cellular systems and lacked inhibitory properties in whole blood due to strong protein binding properties. As of yet no information on a good in vivo inhibitor has been published.
In early work (Shalinsky et al., 1989) in which GSH was shown to be required for PGE2 production, allicin inhibition was noted, which is logical considering the thiol-reactive properties of this compound (Rabinkov et al., 2000). However, thiol reagents have also been shown to directly inhibit mPGES-1 (Thorén et al., 2003). Thus, covalent inhibition by suicide substrate analogs could be an alternate strategy for mPGES-1 inhibitor development.
E. Phylogeny
In addition to human mPGES-1 (Jakobsson et al., 1999b), a homolog with similar characteristics (84% amino acid identity) from mouse has been cloned and characterized (Murakami et al., 2000; Lazarus et al., 2002a). Phylogenetic analysis has identified mPGES-1 and MGST1 homologs coexisting in fish, in which the latter has been characterized at the protein level (Bresell et al., 2005). Microsomal prostaglandin E synthase-1 in zebrafish has a conserved gene structure and 63% amino acid identity compared with the human enzyme (Pini et al., 2005). The gene is developmentally coregulated with COX-2 (although anatomically separated, both are predominantly found in the vasculature and aortic arch) indicating a similar role in fish and humans. COX-1, on the other hand, is coregulated with cPGES (Pini et al., 2005). Experiments with antisense morpholinos indicate that suppression of mPGES-1 in zebrafish does not lead to developmental disturbances, in agreement with results from mPGES-1 knockout mice (Trebino et al., 2003). The data support a role for mPGES-1 in regulated PGE2 production in vertebrates.
F. Tissue and Cellular Distribution
High levels of mPGES-1 were seen by Northern blot analysis in many human cancer cell lines, and intermediate levels were observed in placenta, prostate, testis, and mammary gland (Jakobsson et al., 1999b). Reverse transcriptase-polymerase chain reaction detected transcripts in murine brain and lung but not in heart, liver, or colon whereas the mRNA could be detected in all of these organs after lipopolysaccharide (LPS) exposure (Murakami et al., 2003). Western blots revealed constitutive expression in lung, spleen, kidney, and stomach in mice (Boulet et al., 2004) and seminal vesicles in humans (Stark et al., 2005). mPGES-1 was found to be immunolocalized in several structures of the rabbit kidney cortex (e.g., macula densa) and abundantly in the medulla, notably in the collecting duct system (Fuson et al., 2003). The protein was not detected in normal human heart or liver (Murakami et al., 2003) by the same technique. Staining was, however, observed after infarction or hepatitis. In situ hybridization has shown the presence of inducible mPGES-1 in rat brain being an important link in the response to immune-induced pyresis (Ek et al., 2001). In summary, mPGES-1 is constitutively expressed in several tissues as well as being inducible. Experiments with knockout mice indicate that the organism can develop and maintain health despite the lack of mPGES-1.
As a membrane protein, mPGES-1 is found in the microsomal fraction after subcellular fractionation (Jakobsson et al., 1999b; Ouellet et al., 2002). The enzyme displays perinuclear and endoplasmic reticulum staining and colocalizes, at least in part, with COX-2 (Murakami et al., 2000, 2003; Lazarus et al., 2002b; Stark et al., 2005). In human WISH epithelial cells, a study including both biochemical fractionation and immunolabeling of ultrathin cryosections suggested that COX-2 and mPGES-1 are localized in discrete lipid microdomains (Ackerman et al., 2005). The authors showed that the permeabilization method does influence the observed immunocytochemical distribution, which might also be cell type-specific. All in all, further studies are clearly warranted to establish whether physical proximity is the basis for the efficient biosynthetic interaction between COX-2 and mPGES-1. Interestingly COX-2 synthesizes PGH2 in the lumen of the endoplasmic reticulum, whereas mPGES-1 most likely has its active site pointing to the cytosol. The latter supposition has to be experimentally verified, however.
The fact that mPGES-1 and COX-2 are coregulated, colocalized (in the same membrane but perhaps in different membrane domains), and metabolically coupled (Naraba et al., 1998; Murakami et al., 2000) indicates the importance of tight regulation of PGE2 production, especially in the setting of competing pathways. The fact that mPGES-1-deficient macrophages do not produce PGE nonenzymatically (Trebino et al., 2005), as PGH2 is shunted quantitatively to other prostanoids, underscores this view.
G. Regulation and Gene Structure
Microsomal PGES-1 has been shown to respond to inflammatory and mitogenic stimuli and hence is up-regulated in inflammation models (Sampey et al., 2005) and cancer (Kudo and Murakami, 2005). In kidney, mPGES-1 is responsive to low-salt and angiotensin I-converting enzyme inhibition (Fuson et al., 2003). Coinduction of COX-2 and PGE2 synthesis activity (measured in cell lysates) by LPS was first demonstrated in rat peritoneal macrophages (Matsumoto et al., 1997). It was later shown that mPGES-1 is the coregulated enzyme upon interleukin (IL)-1β or tumor necrosis factor (TNF)-α treatment of A549 cells (Jakobsson et al., 1999b; Thorén and Jakobsson, 2000). Dexamethasone completely suppressed the effect of both cytokines (Thorén and Jakobsson, 2000). Numerous studies have since confirmed these findings (e.g., Han et al., 2002; Schaloske et al., 2005) and p38 and extracellular signal-regulated kinase mitogen-activated kinases were shown to be important for induction (Han et al., 2002). TNF-α up-regulation of COX-2 and mPGES-1 was attenuated and augmented by interferon-γ, respectively (Wright et al., 2004). Induction of mPGES-1 by LPS (Devaux et al., 2001; Uematsu et al., 2002) was shown to be mediated by the Toll-like receptor 4/MyD88/NF-IL-6-dependent signaling pathway (Uematsu et al., 2002) and could be augmented by 9-cis-retinoic acid (Tsukamoto et al., 2004).
In inflammatory bowel disease mPGES-1 is up-regulated through the zinc finger-containing transcription factor early growth response gene 1 (EGR-1)-mediated activation of transcription (Subbaramaiah et al., 2004). Here TNF-α was suggested to up-regulate mPGES-1 via stimulation of the phosphatidylcholine-phospholipase C/protein kinase C/nitric oxide/cGMP/protein kinase G signal transduction pathway. Small interfering RNA depression and controlled expression of EGR-1 confirmed its role also in IL-1β-dependent mPGES-1 regulation (Moon et al., 2005). Curcumin, a substance with anti-inflammatory and anticarcinogenic properties, can down-regulate mPGES-1 expression through suppression of EGR-1 as well as nuclear factor-κB and c-Jun NH2-terminal kinase 1/2 (Moon et al., 2005). Constitutive activation of a tyrosine kinase receptor (in papillary thyroid carcinomas) leads to the induction of mPGES-1 through the Shc/RAS/mitogen-activated protein kinase kinase/extracellular signal-regulated kinase pathway (Puxeddu et al., 2003).
Peroxisome proliferator-activated receptor γ (PPARγ) activation can completely inhibit IL-1β-mediated induction of mPGES-1 (Mendez and LaPointe, 2003). Interestingly, 15-deoxy-Δ12,14-prostaglandin J2 is an activator of PPARγ (Forman et al., 1995) (leading to down-regulation of mPGES-1) as well as a direct inhibitor of the enzyme (Quraishi et al., 2002). Thus, this compound can inhibit PGE2 production by alternate modes. It was shown that PPARγ ligands suppress EGR-1 mediated induction (Cheng et al., 2004). mPGES-1 induction by homocysteine was also attenuated by PPARγ activators, but here the enzyme was measured by immunochemistry in cell supernatants, and it is unclear how this relates to cellular levels of the protein (Peng et al., 2005). It has also been reported that 15-deoxy-Δ12,14-prostaglandin J2 can inhibit mPGES-1 expression in a PPARγ-independent fashion by interfering with the NF-κB dependent pathway (Bianchi et al., 2005; Schröder et al., 2006). One mechanism was redox dependent as shown by inhibition with thiol (Schröder et al., 2006). Because 15-deoxy-Δ12,14-prostaglandin J2 can react with thiols, this could be the result of a direct reaction.
The human mPGES-1 gene structure was determined in 2000 (Forsberg et al., 2000); the gene spans 18.3 kilobases, contains three exons, and is situated on chromosome 9q34.3. The gene structure is conserved in fish (Pini et al., 2005). Although the gene is coregulated with COX-2, no obvious similarities exist in the promoter region (Sampey et al., 2005). Reporter constructs from the 5′-flanking region of the first exon displayed promoter activity and transient induction by IL-1β. Later it was shown that two GC boxes are required for basal transcription (Ekström et al., 2003) and that the transcription factors Sp1 and Sp3 bind these regulatory elements. Induction by TNF-α was shown to be mediated by EGR-1 binding to the same two GC boxes (Subbaramaiah et al., 2004). As pointed out earlier (Sampey et al., 2005), a future challenge is to better define the exact cellular regulation of mPGES-1 in relation to the promoter structure.
H. Membrane Prostaglandin E Synthase-2, Cytosolic Prostaglandin E Synthase, and Other Proteins with Prostaglandin E Synthase Activity
Early on it was noted that certain cytosolic glutathione transferases displayed PGES activity; however, the catalytic efficiency seemed modest (Meyer et al., 1996; Beuckmann et al., 2000). A cPGES that is identical to the cochaperone p23 was later identified (Tanioka et al., 2000). This enzyme is broadly expressed with high levels in brain and seems to function in conjunction with COX-1. Similar to the cytosolic glutathione transferase, activity was in the low micromole per minute per milligram range (Table 1).
A second membrane-bound mPGES-2 was first identified and purified from bovine heart (Watanabe et al., 1999). This enzyme did not require GSH, but activity increased when different thiols were added. The human cDNA for mPGES-2 was later cloned and expressed, reveling enzymatic properties similar to those of the bovine enzyme (Tanikawa et al., 2002) (Table 1). The enzyme is widely distributed with high levels in brain, heart, muscle, kidney, and liver. By in vitro mutagenesis Cys110 was shown to be essential for the activity (Watanabe et al., 2003). A peculiar aspect of mPGES-2 is that it can be cleaved to a soluble catalytically competent form, which becomes increasingly present in cells after heterologous expression (Murakami et al., 2003). The enzyme seems to have no preference for COX-1 or COX-2 but can serve as a terminal PGES for both (Murakami et al., 2003). Recently, the structure of mPGES-2 was solved (Yamada et al., 2005) and was shown to contain structural elements similar to those of cytosolic glutathione transferases and cytosolic PGD synthase with additional segments important for dimerization/membrane association, and substrate binding (reviewed in Oakley, 2005). As the enzyme was crystallized in the presence of indomethacin, which contacts the active site residue Cys110, the location of the PGH2 binding site is indicated. The chemical mechanism of the enzyme is discussed in light of the structural information. A general theme for PGE synthases seems to be the requirement of a thiol either as a base on the enzyme or perhaps provided by GSH in mPGES-1 and cPGES.
Much effort has gone into determining the coupling of the various PGES to either COX-1 and COX-2. For instance, tight coupling between mPGES-1 and COX-2 has been suggested to be based on codistribution (Murakami et al., 2000), a concept that has been challenged, however (Ackerman et al., 2005). Perhaps the high catalytic efficiency of mPGES-1 will ensure that temporal regulation (induction) will yield efficient competition against other terminal prostanoid synthases.
III. Inflammation
A. Constitutive Expression of Membrane Prostaglandin E Synthase-1
The constitutive expression of mPGES-1 has been described in urogenital organs (Guan et al., 2001; Lazarus et al., 2002b; Fuson et al., 2003), gastric mucosa, spleen, resident peritoneal macrophages (Boulet et al., 2004), and liver Kupffer cells (Dieter et al., 2000). Thus, in these cells and organs mPGES-1 might have physiological roles.
B. Induction of Membrane Prostaglandin E Synthase-1 Expression
In the initial study on the distribution of mPGES-1, the enzyme was found only sparsely expressed in tissues but was induced by IL-1β in A549 cells (Jakobsson et al., 1999). Subsequently, a similar induction of mPGES-1 by IL-1β and TNF-α was described along with its suppression by corticosteroids in A549 cells, synovial cells, macrophages, and osteoblasts (Murakami et al., 2000; Thorén and Jakobsson, 2000; Stichtenoth et al., 2001; Kojima et al., 2002, 2003). A tandem induction of COX-2 and mPGES-1 expression was found in most situations except for cancer cells in which these enzymes may be uncoupled. Other cell types in which mPGES-1 has been found to be induced by proinflammatory cytokines include human endothelial cells (Caughey et al., 2001; Uracz et al., 2002), orbital fibroblasts (Han et al., 2002), gingival fibroblasts (Yucel-Lindberg et al., 2004), chondrocytes (Kojima et al., 2004; Masuko-Hongo et al., 2004), smooth muscle cells (Soler et al., 2000; Jaulmes et al., 2006), and cardiac myocytes (Mendez and LaPointe, 2003; Degousee et al., 2006). The bacterial endotoxin LPS was also early described as an inducer of mPGES-1 in rat peritoneal macrophages (Murakami et al., 2000) and smooth muscle cells (Soler et al., 2000). In applying the acute in vivo models of inflammation such as LPS-induced pyresis and adjuvant-induced arthritis, significant induction of mPGES-1 was demonstrated in several mouse organs and the inflamed rat paw (Mancini et al., 2001; Claveau et al., 2003). Together, these findings may suggest a role for mPGES-1 in arthritis, osteoarthritis, and inflammatory processes in the vascular wall and heart. Some of these conditions have been studied more extensively and will be discussed further. In addition, the inflammatory component of the development of atherosclerosis has been identified (Hansson, 2005) as well as part of the reperfusion reactions after an event of ischemia of the brain (Huang et al., 2006).
C. Membrane Prostaglandin E Synthase-1 in Arthritis and Osteoarthritis
In patients with rheumatoid arthritis, the expression of mPGES-1 was markedly up-regulated in the synovial tissue, especially in inflammatory cells (Murakami et al., 2003; Westman et al., 2004). However, the treatment of patients with rheumatoid arthritis with new biological agents, blockers of TNF-α, did not suppress mPGES-1 expression in the synovial tissues, implying that other mechanisms operate in sustaining the inflammation independently of TNF-α (Korotkova et al., 2005). In comparison, local treatment with corticosteroids significantly down-regulated the expressions of COX-1, COX-2, and mPGES-1, suggesting a more profound effect on the inflammatory processes than treatments with TNF blockers. To the best knowledge of the authors, there has been no systematic clinical study performed as yet to address the combination of biological or other so-called disease-modifying antirheumatic drugs with inhibition of the prostaglandin system. Thus, combination therapy with TNF blockers and PGE2 inhibition may provide increased efficacies, but this remains to be elucidated. mPGES-1 has also been demonstrated to be up-regulated in human cartilage and chondrocytes in patients with osteoarthritis or after treatment with proinflammatory cytokines (Kojima et al., 2004; Masuko-Hongo et al., 2004; Farley et al., 2005; Li et al., 2005), and the expression has been demonstrated to be mechanosensitive (Iimoto et al., 2005; Gosset et al., 2006).
D. Membrane Prostaglandin E Synthase-1 in Fever and Pain
NSAIDs and acetaminophen are potent antipyretic and analgesic drugs, resulting in rapid relief of the fever and pain caused by inflammation. Prostaglandin E2 constitutes an important mediator of these symptoms. In a first attempt to address the role of mPGES-1 in the brain, rats were treated with IL-1β, and subsequently brain sections were analyzed by in situ hybridization for the expressions of the IL-1 receptor, COX-2, and mPGES-1 (Ek et al., 2001). Within 1 h COX-2 mRNA expression was increased in endothelial cells of the blood-brain barrier and returned back to baseline after 3 h. mPGES-1 mRNA expression displayed a later onset and declined within 3 to 5 h after treatment with IL-1β. Double staining demonstrated colocalization of mPGES-1, COX-2, and the type 1 IL-1 receptor mRNA in endothelial cells of the blood-brain barrier. Consistent results were reported after treatment of rats with LPS as mPGES-1 and COX-2 were mainly present in the perinuclear region of the endothelial cells (Yamagata et al., 2001). mPGES-1 has also been demonstrated in various areas of the brain after adjuvant-induced arthritis, carrageenan-induced paw edema, intraparenchymal administration of IL-1β, peripheral burn injury, and pilocarpine-induced neurodegeneration as a model for seizures and epilepsia (Engblom et al., 2002; Guay et al., 2004; Moore et al., 2004; Ozaki-Okayama et al., 2004; Turrin and Rivest, 2004). The most abundant cells of the brain expressing mPGES-1 seems to be vascular endothelial cells, but mPGES-1 has also been demonstrated in LPS-activated microglia cells (Ikeda-Matsuo et al., 2005), β-amyloid-treated astrocytes (Satoh et al., 2000), postnatal cortical neurons (Echeverria et al., 2005), and postischemic cortical neurons and microglia cells (Ikeda-Matsuo et al., 2006). mPGES-1 has also been described in the spinal cord and dorsal root ganglia in rats after treatment with LPS (Schuligoi et al., 2003).
E. Membrane Prostaglandin E Synthase-1 in the Cardiovascular System and Kidney
Regarding the prostaglandin pathway in atherosclerosis and myocardial infarction, it has been established that 1) low-dose aspirin is effective as a secondary prophylaxis against myocardial infarction and 2) COX-2 inhibition confers increased risks of severe cardiovascular side effects due to a reduction in prostacyclin production (Grosser et al., 2006). In symptomatic atherosclerotic carotid plaques, there is an up-regulation of mPGES-1, COX-2, and matrix metalloproteinase-2 and -9 (Cipollone et al., 2001), where matrixmetalloproteinase-2 and -9 are induced by PGE2. These data have recently been replicated in patients with carotid atherosclerosis (Gómez-Hernández et al., 2006a). Simvastatin and atorvastatin have been shown to suppress mPGES-1 and the PGE2 pathway in this patient group (Cipollone et al., 2003a; Gómez-Hernández et al., 2006b). In patients with diabetes, there is also increased expression of mPGES-1 in atherosclerotic plaques compared with that in nondiabetic patients (Cipollone et al., 2003b). Together with recent data using mPGES-1 knockout mice (see section V.A.) inhibition of mPGES-1 is likely to bring antiatherogenic effects and plaque stabilization. Renal function is critical for maintaining the body salt and water balance. Before the identification of mPGES-1 or other PGE synthases, PGE synthase activity was demonstrated to be constitutively expressed in the renal medulla and cortex of the rabbit kidney (Sheng et al., 1982). Anatomic distribution analyses of mPGES-1 protein in rodents have demonstrated distinct constitutive expression along the nephron and collecting duct, colocalized with either COX-1 or COX-2 (Câmpean et al., 2003). In rabbit kidneys, a similar distribution was reported. mPGES-1 expression was increased in the macula densa after a low-salt diet and angiotensin-converting enzyme inhibition, suggesting both autocrine regulation of renal salt and water transport by PGE2 and a paracrine effect of PGE2 on the glomerular vasculature (Fuson et al., 2003; Peti-Peterdi et al., 2003). In line with these effects, increased expression of mPGES-1 in macula densa cells has been found in patients suffering from hyperprostaglandin E syndrome or classic Bartter syndrome that presents with an activated renin angiotensin system due to salt and water losses (Kömhoff et al., 2004). Thus, it seems likely that mPGES-1 participates in the mechanisms that compensate for low blood volume through activation of the renin-angiotensin system. In contrast, an acute high salt intake requires rapid urinary sodium excretion and was recently tested in mice lacking mPGES-1 (Cheng et al., 2006; Jia et al., 2006) with conflicting results on blood pressure (see section V.B.). Furthermore, experimental hypertension induced by ligation of one side of the renal artery results in bilateral down-regulation of mPGES-1 paralleled with down-regulation of COX-1 (Theilig et al., 2006). The mechanisms by which the latter findings would affect the hypertension are still unclear. Orchidectomy results in higher mPGES-1 expression in male rats (Sullivan et al., 2005), suggesting a role for testosterone in regulation of mPGES-1, and the PPARγ agonist troglitazone may prevent induction of mPGES-1 in cultured human glomerulus cells treated with glucose or LPS (Ling et al., 2004). In summary, mPGES-1 seems to play a pathophysiological role in the development of atherosclerosis but a physiological role in the kidneys.
IV. Disruption of the Membrane Prostaglandin E Synthase-1 Gene
A. Endotoxin-Induced Shock
The first report of the critical role of mPGES-1 for induced PGE2 biosynthesis in vivo was published in 2002 (Uematsu et al., 2002). The authors hypothesized that mPGES-1 could play a role in LPS-induced shock and therefore developed mPGES-1-deficient mice on the C57BL/6 × 129/SvJ background. Although there was no effect on the shock in mPGES-1-deficient mice, the authors demonstrated that the massive induction of PGE2 biosynthesis in wild-type peritoneal macrophages, treated with LPS for 24 h, was abolished in knockout cells. The authors also demonstrated that the 3-fold increase in serum PGE2 concentration, 6 h after intraperitoneal administration of LPS, was suppressed to background levels in knockout animals. Furthermore, the effect of LPS was shown to be mediated through a Toll-like receptor 4/MyD88-dependent pathway (in analogy with COX-2, as shown by the same laboratory) and also with contribution of the transcription factor NF-IL-6. The mPGES-1 knockout animals showed no phenotypic changes compared with wild-type animals, and there were no differences in TNF-α, IL-12p40, or IL-6 production in response to LPS.
B. Experimental Models of Arthritis
In 2003, the pathophysiological role of mPGES-1 in experimental arthritis was described using mPGES-1 knockout mice on the DBA/1LacJ background (Trebino et al., 2003). Arthritis was induced by subcutaneous administration of native chicken collagen type II, and symptoms of arthritis were assessed by clinical evaluation and histopathology. The results demonstrated significant reduction in the incidence of collagen-induced arthritis (CIA). In wild-type mice approximately 60% of the animals developed CIA compared with approximately 10% of the knockout animals. The severity of inflammation was also significantly reduced in knockout animals (clinical score 0–1) compared with wild-type animals (clinical score of approximately 4). The reduction in clinical signs of inflammation correlated with the reduction of several histopathological parameters in knockout mice, such as grade of hyperplasia, loss of proteoglycan, pannus formation, bone erosion, and destruction of surface cartilage. When severe joint arthritis was defined by combining the scoring system, the authors reported that severe arthritis was found in only 0.04% of joints from mPGES-1 knockout mice compared with 36.4% of joints isolated from mPGES-1 wild-type mice. In analogy, mPGES-1 knockout mice presented with considerably less inflammation in a model of local inflammation (delayed hypersensitivity).
In a subsequent report, the role of mPGES-1 was also demonstrated in the collagen antibody-induced arthritis (CAIA) model (Kamei et al., 2004). Here, the authors used mPGES-1 knockout mice raised on a C57BL/6 × 129/SvJ background (as described above). There are important differences between these two models of arthritis induction. In the CIA model the joint inflammation depends on intact immunological events starting from antigen recognition at the site of collagen administration, T and B cell activation, and anti-collagen antibody production followed by the local events in the joints eventually causing arthritis. In the CAIA model, antibodies directed to collagen are injected intraperitoneally, thus surpassing activation of the immune system. To achieve sufficient engagement of the joints, LPS was injected every 3rd day repeatedly. The results demonstrated 100% incidence of CAIA in both wild-type and knockout animals, but the knockout animals displayed significant reductions in the clinical signs of inflammation, with an approximately 40 to 50% reduction after day 7. At the time of onset of clinical arthritis, there was a 50% reduction of PGE2 concentrations in homogenates of affected joints from knockout animals. Although COX-2 and mPGES-1 as assessed by immunoblotting were both induced in wild-type animals; knockout animals presented the same levels of COX-2 protein but naturally no expression of mPGES-1.
C. Role of Membrane Prostaglandin E Synthase-1 in Fever, Pain, and Anorexia
Previous studies have demonstrated the brain distribution of mPGES-1 and COX enzymes after induction of peripheral inflammation (section III.D.). In addition, certain central responses to inflammation have been evaluated using mPGES-1 knockout mice.
Bacterial wall LPS injection represents a common model of inducing fever in experimental animals. In response to intraperitoneal administration of 2 μg of LPS, wild-type DBA/1LacJ mice displayed a fever response that started at approximately 1.5 h after injection and persisted for approximately 5 to 6 h (Engblom et al., 2003). In contrast, the fever response in mPGES-1 knockout mice was abolished. Analysis of the PGE2 concentration in the cerebrospinal fluid, 3 h after LPS injection, demonstrated a significant increase of PGE2 levels in wild-type mice, whereas mPGES-1 knockout mice presented with normal levels of PGE2. This increase in PGE2 concentration also correlated with a significant induction of PGE synthase activity in the brain microsomal fraction. Injection of PGE2 into the brain ventricles elicited a fever response in mPGES-1 knockout mice, confirming the intact downstream signaling cascade in these mice and also the fact that PGE2-induced fever was mediated through the EP3 receptor. To extend these findings to encompass the role of mPGES-1 in cytokine-mediated fever, an aseptic model of peripheral inflammation was used (Saha et al., 2005). In this model, a small amount of turpentine was injected subcutaneously, which elicited a biphasic febrile response in wild-type but not the mPGES-1 knockout mice. Although the endotoxin model could not rule out a direct effect of LPS on the brain through Toll-like receptor activation, the turpentine model strongly supports a cytokine-mediated PGE2 effect on the brain. Together with the results demonstrating the massive up-regulation of the PGE2 biosynthetic machinery in the blood-brain barrier (see section III.B.), these reports suggest that among the mechanisms behind fever as a consequence of peripheral inflammation, there is a cytokine-dependent effect on the blood-brain barrier, which in turn leads to the release of PGE2 acting on specific thermoregulatory cells in the brain.
The inflammatory pain response has been assessed using diluted acetic acid injected into the peritoneal cavity of mice. The pain response as recorded by counting the number of writhings was significantly reduced in mPGES-1 knockout mice compared with wild-type mice and comparable with the group of wild-type mice receiving the NSAID piroxicam or indomethacin (Trebino et al., 2003; Kamei et al., 2004). Both reports address the biosynthesis of intraperitoneal PGE2 and PGI2 after noxious stimulation, although with some inconsistencies regarding the formation of PGI2. The first article (Trebino et al., 2003) reported no significant changes in the levels of 6-keto-PGF1α, whereas the second study (Kamei et al., 2004) in fact reported diminished formation of PGI2 in the knockout mice compared with wild-type mice after noxious stimulation. For clarity, these effects on PG biosynthesis in wild-type or knockout mice relates to the concentrations determined in peritoneal lavage fluid before and after noxious stimulation. Redirection of prostaglandins in the absence of mPGES-1 has also been addressed in, e.g., LPS-treated peritoneal macrophages, where, after LPS stimulation, an increase of PGI2 has been observed. Thus, in vitro, cell-specific redirection that not necessarily reflects the in vivo situation will occur. In both cases, the reduction of PGE2 in knockout mice correlated with reduction in pain responses, however, the authors (Kamei et al., 2004) observed that the reduction of PGE2 biosynthesis was less than expected to solely explain the marked reduction in pain responses. Therefore, they speculated that mPGES-1 may also play a role at the spinal or supraspinal levels. To test for such events, the authors (Trebino et al., 2003) also applied the hot-plate assay, measuring the withdrawal latencies, but did not report any differences among either genotype. In a following investigation, these mice were used in a model of neuropathic pain (Mabuchi et al., 2004). The left L5 spinal nerve was transected, and after recovery the mice were subjected for testing of thermal and mechanical sensitivity. The results demonstrated that wild-type mice developed allodynia/hyperalgesia on the ipsilateral side of injury as shown by the significant reduction in their thresholds for mechanic or thermal responses. Knockout mice did not develop such responses and interestingly could not be separated from the wild-type mice population before transaction of the nerve; thus mPGES-1 seems not to play a role in the physiological responses to these stimuli but does play a major role in the pathophysiological mechanisms for the development of central pain.
Anorexia, cachexia, and fatigue can be severe symptoms of chronic inflammatory diseases or cancer. The exact mechanisms for how the brain receives signals from the peripheral inflammation are still being investigated. The prostaglandin pathway and PGE2 in particular have been implicated in the anorexic behavior through the use of NSAIDs and observations with COX-2 knockout mice, as discussed (Pecchi et al., 2006; Elander et al., 2007). An anorectic dose of IL-1β injected intraperitoneally caused anorectic behavior with significant reduced food intake in wild-type mice but not in mPGES-1 knockout mice (Pecchi et al., 2006; Elander et al., 2007). Interestingly, intraperitoneal administration of LPS to normal-fed, wild-type and mPGES-1 knockout animals resulted in significant loss of food intake and body weight in both genotypes (Elander et al., 2007). In contrast, prestarved mPGES-1 knockout animals responded with less reduction in food intake and body weight compared with wild-type mice after the LPS challenge. The latter findings suggest a role for mPGES-1 in the regulation of food intake that could be more related to endogenous metabolism rather than to a pathophysiological response to LPS in the context of anorexia. On the other hand, there was no difference between wild-type and knockout feeding behavior without the LPS or IL-1β challenge. Interestingly, it was recently demonstrated that mPGES-1 is predominantly expressed in mice white adipose tissue and that obese animals contain significantly less mPGES-1 (Hetu and Riendeau, 2007). In addition, deletion of the EP3 receptor in mice has been associated with night eating behavior and obesity (Sanchez-Alavez et al., 2007).
V. Cardiovascular Diseases
A. Atherosclerosis
NSAIDs are commonly used drugs for treating symptoms of inflammation. In hope of reducing the number of severe gastric ulcerations by NSAIDs, specific COX-2 inhibitors were developed [e.g., rofecoxib (VIOXX) and celecoxib (Celebrex)]. However, in 2004, VIOXX was withdrawn from the market because of increased incidence of severe cardiovascular side effects. These are believed to be the result of depression of prostacyclin, removal of a restraint on thromboxane and other endogenous agonists of platelet aggregation, vasoconstriction, and vascular proliferation and are most likely a class effect that is proportional to the degree of COX-2 inhibition (Grosser et al., 2006). In contrast, patients at risk for developing myocardial infarction receive low-dose aspirin that inhibits platelet COX-1 and thereby thromboxane formation while not significantly influencing prostacyclin biosynthesis in endothelial cells. Recent analyses have shown that mice lacking COX-2, mice expressing modified COX-2 (with no cyclooxygenase activity but retained peroxidase activity), or mice treated with a COX-2 inhibitor produce significantly less prostacyclin with unaltered thromboxane formation and develop thrombosis and hypertension (Cheng et al., 2006). Interestingly, in COX-1 knockdown mice (a genetic simulation model of low-dose aspirin) these effects were rescued to some extent. Evidence was also presented demonstrating that both COX-1 and COX-2 contribute significantly to urinary PGE2 excretion (11α-hydroxy-9,15-dioxo-2,3,4,5-tetranor-prostane-1,20-dioic acid) and that mPGES-1 is the major source of this PGE2 metabolite (75–80% reduction in mPGES-1 knockout mice). mPGES-1 knockout mice were also found to excrete more (60% increase) of the prostacyclin metabolite (2,3-dinor-6-keto PGF1α), but similar amounts of the thromboxane metabolite (2,3-dinor-thromboxane B2). Most importantly, mPGES-1 deletion affected neither blood pressure nor thrombogenesis (Cheng et al., 2006). In the follow-up study, mPGES-1 knockout mice were crossed with low-density lipoprotein receptor (LDLR) knockout mice, and the resulting mice were mPGES-1-/- LDLR-/- and their littermate controls were mPGES-1+/+ LDLR-/- on a mixed background (Wang et al., 2006). In the control population receiving high-fat diet, mPGES-1 protein expression was demonstrated in atherosclerotic plaques after 3 and 6 months. Mice devoid of mPGES-1 showed significantly less atherosclerotic lesions; in females the reductions were 41 and 35% and in males were 15 and 50% after 3 and 6 months, respectively. Analysis of the cellular composition and histopathological appearance showed that the number of macrophage foam cells was reduced by 30% in mPGES-1 knockout mice, and the atherosclerotic plaques displayed less necrosis. Interestingly, both prostacyclin and thromboxane synthases were expressed at higher levels in the endothelium of the mPGES-1 knockout mice, suggesting a regulatory function, a cross-talk among these terminal prostaglandin synthases (Wang et al., 2006). As in the normolipidemic mice, mice receiving the highfat diet and lacking mPGES-1 excreted increased amounts of prostacyclin metabolite (50% increases after 6 months) and similar amounts of thromboxane, compared with the single LDLR knockout mice.
B. Blood Pressure
The distribution of mPGES-1 along the nephron suggests a physiological role regulating blood pressure through the renin-angiotensin system and sodium balance (see section III.E.). The potential impact of mPGES-1 on blood pressure was examined by comparing mPGES-1 knockout mice with wild-type mice after normal or high salt intake. In one study, no segregating effects was observed (Cheng et al., 2006), whereas celecoxib significantly raised the blood pressure in control animals. On the other hand, in another study a marked difference was reported after high salt intake in mPGES-1 knockout mice compared with controls (Jia et al., 2006). The apparent differences among these two reports call for further investigations. The discrepancies may be explained by different strains used and different ways of administering food. At this point the impact of future mPGES-1 inhibitors on blood pressure cannot be predicted. Speaking in favor of mPGES-1 inhibition over COX-2 is the fact that COX-2 knockout mice develop heart fibrosis and renal and reproductive deficiencies (Dinchuk et al., 1995), whereas mPGES-1 knockout mice have no reported inborn changes of the normal phenotype.
C. Stroke
Stroke is caused by occlusion of arterial vessels in the brain resulting in ischemia in the region supplied with blood from the specific vessels engaged. The most common reasons are rupture of carotid atherosclerotic plaques or emboli from an irregularly beating heart, e.g., paroxysmal atrial fibrillation. Treatment includes reperfusion within 3 to 4 h after symptoms, which can only be managed if the patient is admitted to hospital care within this time frame; otherwise, the patient is treated conservatively. The role of inflammation after ischemic stroke is being increasingly discussed (Nilupul Perera et al., 2006), and COX-2 inhibition has shown beneficial effects in animal models (Iadecola and Gorelick, 2005). In a recent publication (Ikeda-Matsuo et al., 2006), the significance of mPGES-1 on the outcome after ischemic stroke was investigated. In rats, after 2 h of ischemia by occlusion of the right middle cerebral artery, followed by 24 h of reperfusion, both mPGES-1 and COX-2 were induced in the ipsilateral cortex. mPGES-1 expression was limited to the ipsilateral cortex region and striatum, whereas COX-2 was also found in the cortex and hippocampus. PGE2 biosynthesis was mainly induced in the cortex with a maximum concentration 1 day after ischemia. No induction was shown for mPGES-2, cPGES, or COX-1. In line with other data, COX-2 protein expression was more transiently expressed and earlier in time, whereas mPGES-1 expression persisted with a maximal expression between days 1 and 3 before returning to baseline at day 7. mPGES-1 was mainly expressed in neurons in the peri-infarct regions, whereas in the lesion core it was mostly found in microglia and endothelial cells. Mice devoid of mPGES-1 were used to investigate the role of this enzyme for the development of infarction. First, the postischemic and regional specific PGE2 production observed in wild-type mice was abolished in mPGES-1 knockout mice, whereas the expressions of mPGES-2, cPGES, and COX-1 and induced COX-2 proteins were similar in both genotypes. There were also no significant differences between mPGES-1 knockout and wild-type mice in mean arterial pressure, pH, pCO2, or pO2 levels and changes of cerebral blood flow before, during, or after middle cerebral artery occlusion. Mice devoid of mPGES-1 presented with significantly less infarct volume and edema. The neurological deficits were also significantly reduced in mPGES-1 knockout mice compared with controls. On the level of mechanisms by which the lack of mPGES-1 may affect the outcome, it was shown that mPGES-1 knockout mice presented with significantly less apoptosis. Finally, it was demonstrated that substitution of PGE2 before arterial occlusion worsened the infarct outcome as a proof of intact downstream signaling in these mice lacking mPGES-1.
VI. Cancer
It has been known for a long time that there is a significant reduction in risk of colorectal cancer and other cancers in persons who take NSAIDs on a regular basis (Marnett, 1992). It was also demonstrated that COX-2 levels are elevated in colorectal cancers and some other cancers (Eberhart et al., 1994; Kargman et al., 1995; Sano et al., 1995; Kutchera et al., 1996; Chapple et al., 2000; Anderson et al., 2002). Additional studies have provided strong support for a role of COX-2 in cancer (DuBois, 2003). Of the PGH2-derived products, PGE2 is of particular interest because PGE2 levels are increased in a variety of human tumors (Jung et al., 1985; Vanderveen et al., 1986; McLemore et al., 1988; LeFever and Funahashi, 1990; Rigas et al., 1993). Moreover, PGE2 stimulates tumor growth, mediates immune suppression, inhibits apoptosis, and stimulates angiogenesis (Plescia et al., 1975; Ben-Av et al., 1995; Sheng et al., 2001).
Support for a role of mPGES-1 in carcinogenesis was recently obtained by measuring mPGES-1 in non-small cell lung cancer (NSCLC) (Yoshimatsu et al., 2001a). The expression of the inducible prostaglandin E synthase (mPGES-1) in 19 paired samples (tumor and adjacent normal tissue of NSCLC) was compared using immunoblot analysis. mPGES-1 was overexpressed in 80% of NSCLCs and was localized to neoplastic epithelial cells. COX-2 was frequently up-regulated in theses tumors, however, there were marked differences in the extent of up-regulation of mPGES-1 and COX-2 in individual tumors, indicating that the regulation of the two enzymes is not identical. Levels of mPGES-1 mRNA and protein were increased in NSCLC cell lines containing mutant Ras. A similar up-regulation of COX-2 was reported earlier. TNF-α induced mPGES-1 and COX-2 in NSCLC cell lines, but there was no effect on the expression of either enzyme in a nontumorigenic bronchial epithelial cell line. These data indicated that both cellular transformation and cytokines contribute to the up-regulation of mPGES-1 in NSCLC (Yoshimatsu et al., 2001a). It is of interest in this context that normal lung fibroblasts, when exposed to cigarette smoke, induce COX-2 and mPGES-1 expression (Martey et al., 2004). However, targeted overexpression of mPGES-1 and elevated PGE2 production in alveolar and airway epithelial cells is not sufficient for lung tumorigenesis in transgenic mice (Blaine et al., 2005).
In a study of prostaglandin biosynthetic pathway markers in lung cancer, it was proposed that these markers may act synergistically and enhance tumor angiogenesis, the expression of angiogenic factors, and metastases in patients with NSCLC (Yoshimoto et al., 2005). In another study designed to analyze the expression of mPGES-1 in invasive breast cancer using immunohistochemistry, mPGES-1 was overexpressed in 79% of the cases. It was undetectable in normal breast epithelial cells. The overexpression of mPGES-1 was more frequent than that of COX-2 and did not correlate with it. Neither did it correlate with prognostic markers of breast cancer such as estrogen receptor expression or HER-2/neu status (Mehrotra et al., 2006). A difference in the regulation of COX-2 and mPGES-1 was recently reported (Cui et al., 2006). In NSCLC cells IL-4 inhibited the formation of PGE2 predominantly via a decrease in COX-2 mRNA transcription whereas the expression of mPGES-1 was unaffected.
In colorectal adenomas and cancer both COX-2 and mPGES-1 are overexpressed compared with adjacent normal tissue. Marked differences in the extent of up-regulation of COX-2 and mPGES-1 were observed in individual tumors. Chenodeoxycholic acid induced COX-2 but not mPGES-1 in colorectal cancer cells. TNF-α induced both COX-2 and mPGES-1. Overexpression of Ras caused an increase in mPGES-1 promotor activity as was previously described for COX-2 (Sheng et al., 2000; Yoshimatsu et al., 2001b).
Studies of intestinal-type gastric adenocarcinomas showed increased expression of mPGES-1. Moreover, in gastric cancer cell lines the regulation of expression of mPGES-1 and COX-2 differ (van Rees et al., 2003). It has been proposed that COX-2 and mPGES-1 affect both histogenesis and carcinogenesis of human gastric cancer (Jang, 2004).
Helicobacter infection induces COX-2 and mPGES-1 expression in the gastric mucosa (Oshima et al., 2004). A mouse model of Helicobacter infection (K19-C2mE) also expresses COX-2 and mPGES-1 in the gastric mucosa and develops hyperplastic tumors in the proximal glandular stomach (Oshima et al., 2004). It has been proposed that TNF-α-dependent inflammation is responsible for the gastric hyperplasia in the transgenic mouse model (Oshima et al., 2005). Helicobacter pylori-dependent induction of COX-2 and mPGES-1 is associated with enhanced production of P-gp (product of multidrug resistance-1) and Bcl-XL (the antiapoptic protein) that may contribute to gastric tumorigenesis and resistance to therapy (Nardone et al., 2004). Transgenic mice simultaneously expressing COX-2 and mPGES-1 in gastric epithelial cells developed metaplasia, hyperplasia, and tumorous growths in the glandular stomach. Helicobacter infection up-regulated epithelial PGE2 production (Oshima et al., 2004). COX-2 and mPGES-1 are also induced in gastric hamartoma tissues of various genetically modified mice (Takeda et al., 2004).
Several other studies on various cancers have shown increased expression of COX-2 and mPGES-1. This indicates a role for PGE2 in tumorigenesis (Jabbour et al., 2001; Cohen et al., 2003; Kamei et al., 2003; Golijanin et al., 2004; Jang et al., 2004; Kawata et al., 2006; Shi et al., 2006).
VII. Reproductive Endocrinology
Ovulation is an essential step in mammalian reproduction and leads to the release of the maternal germ cell. Prostaglandins have been shown to be important mediators of ovulation (Sirois et al., 2000), and several lines of evidence indicate that PGE2 is the key compound (Davis et al., 1999). It was also found earlier that COX-2 is induced in granulosa cells by gonadotropins before ovulation (Sirois, 1994).The same pattern of expression of mPGES-1 was seen in bovine granulosa cells of ovarian follicles in vivo, 18 to 24 h after hCG treatment or onset of estrus (Filion et al., 2001). The expression of mPGES-1 mRNA and protein is also increased in primates in response to hCG administration, showing a peak just before the expected time of ovulation. The mRNA levels of mPGES-2 and cPGES did not change in response to hCG administration. Hence, mPGES-1 may be the primary PGE synthase responsible for the increased follicular PGE2 levels necessary for primate ovulation (Duffy et al., 2005).
During implantation, the embryo makes contact with the endometrium for the establishment of pregnancy. PGE2 and prostacyclin (PGI2) have been considered to play important roles in implantation and decidualization in rodents (Lim et al., 1999), and COX-2-deficient mice show multiple reproductive failures including implantation and decidualization (Lim et al., 1997). In the same species mPGES-1 mRNA and protein were highly expressed in the stroma immediately surrounding the blastocyst but not in the luminal epithelium on day 5 of pregnancy. From days 6 to 8 of pregnancy, mPGES-1 was strongly expressed in the decidualized cells (Ni et al., 2002). Additional studies in hamsters showed that PGE2 but not PGI2 is the major prostaglandin at implantation sites where COX-2 and mPGES-1 but not PGI synthase are expressed (Wang et al., 2004). A detailed study on the PGE2-forming enzymes driving the development of the blastocyst in mice has also been reported (Tan et al., 2005). Moreover, a study in rats showed that, as in mice, both PGE2 and PGI2 have important roles in implantation (Cong et al., 2006).
The corpus luteum, formed after ovulation, contributes to maintenance of early pregnancy by secretion of progesterone. The development of a functional corpus luteum depends on vascularization, a process in which both vascular endothelial growth factor (VEGF) and PGE2 play important roles. It was recently reported that VEGF enhances PGE2 formation by stimulating COX-2 and mPGES-1 expression in rat corpus luteum and that the effect of VEGF on luteal cells may be partially mediated by this stimulation of PGE2 production (Sakurai et al., 2004). A detailed study of the regulation of the expression of mPGES-1 in relation to other prostaglandin-synthesizing enzymes and receptors in the bovine corpus luteum has been reported (Arosh et al., 2004). In mice, mPGES-1 mRNA levels are induced by hCG in granulosa cells and in the corpus luteum (Sun et al., 2006).
Endometrial synthesis of prostaglandins plays an important role in the regulation of the estrous cycle, pregnancy recognition, pregnancy, and parturition in ruminants. In a study of bovine endometrial tissue, COX-2 mRNA and protein were expressed at low and high levels on days 1 to 12 and 13 to 21 of the estrous cycle, respectively. The level of expression of mPGES-1 was moderate, low, and high on days 1 to 3, 4 to 12, and 13 to 21 of the estrous cycle, respectively. COX-1 mRNA and protein were not expressed on any day of the estrous cycle (Arosh et al., 2002). Using a bovine endometrial cell line all three PGE synthases (cPGES, mPGES-1, and mPGES-2) were detected; however, mPGES-1 is the main enzyme associated with increased PGE2 production in vitro (Parent and Fortier, 2005). A regulatory effect on mPGES-1 expression of IFN-τ has also been reported (Parent et al., 2002; Guzeloglu et al., 2004). In a study of equine endometrium, expressions of mPGES-1 and PGFS were not found to change during late diestrus or in pregnant animals. This finding was interpreted to suggest that the equine conceptus may modulate endometrial prostaglandin synthesis mainly by regulating the expression of the COX-2 enzyme (Boerboom et al., 2004).
The endometrial PGE2/PGF2α ratio is important for regulating the estrous cycle and establishment of pregnancy. In pigs, there was a modest modulation of mPGES-1 throughout the estrous cycle. However, PGF synthase expression was highly up-regulated in the endometrium around the time of luteolysis (Waclawik et al., 2006). A study of the expression of mPGES-1 in the endometrium of the rhesus monkey during the menstrual cycle has also been reported (Sun et al., 2004).
A number of studies of the expression of mPGES-1 in fetal membranes, placenta, and myometrium during pregnancy and labor in humans and other species have been reported (Giannoulias et al., 2002; Martin et al., 2002; Alfaidy et al., 2003; Meadows et al., 2003, 2004; Sun et al., 2003; Palliser et al., 2004). PGE2 is the major vasodilator eicosanoid of the ductus arteriosus and has been suggested to be of importance for maintenance of a patent ductus arteriosus perinatally. In piglets, COX-2, mPGES-1, and PGE2 concentrations were approximately 7-fold higher in the immediately postnatal newborn than in fetus, whereas there was no change in cPGES. Platelet-activating factor was reported to contribute to up-regulation of COX-2 and mPGES-1 expression (Bouayad et al., 2004). In male reproductive organs significant species differences in the expression of mPGES-1 were noted (Lazarus et al., 2002b, 2004).
VIII. Therapeutic Implications
In inflammation and related pathological conditions, the formation of PGE2 is increased through stimulation of arachidonic acid release and induction of COX-2. The discovery of mPGES-1 and its induction by proinflammatory cytokines explains the selective formation of PGE2 among the multitude of endoperoxide-derived products, during inflammation. Hence, mPGES-1 provides a novel molecular target for inhibition of the induced formation of PGE2.
Inhibition of PGE2 formation by COX-2 inhibitors is effective in ameliorating symptoms of inflammation. However, the cardiovascular side effects associated with COX-2 inhibitors have limited their use. The adverse reactions are probably linked to the significant expression of COX-2 in vascular endothelial cells under normal conditions and depression of prostacyclin biosynthesis by COX-2 inhibitors. mPGES-1 is normally only expressed to a limited extent. Moreover, mPGES-1 knockout animals, unlike COX-2 knockouts, have a normal phenotype. This difference might be due to the fact that disruption of the mPGES-1 gene with decreased formation of PGE2, in contrast to COX-2 inhibition, results in augmented prostacyclin expression and only marginal effects on the formation of the other cyclooxygenase-derived products. The concomitant increased cardiovascular safety observed in mPGES-1 knockout animals compared with COX-2 inhibition under similar conditions makes mPGES-1 an attractive target for development of a new class of therapeutic agents.
Pathological conditions in which increased formation of PGE2 is linked to augmented expression of mPGES-1 are inflammation, pain, fever, anorexia, atherosclerosis, stroke, and cancer. The efficacy of mPGES-1 inhibition in relieving symptoms of inflammation and pain has been demonstrated in models of arthritis in knockout mice. A similar approach has provided support for a role of mPGES-1-generated PGE2 in anorexia, atherosclerosis, and stroke. Future clinical studies will address the important question of the efficacy and safety of mPGES-1 inhibition in human diseases.
Footnotes
-
↵1 Abbreviations: PG, prostaglandin; COX, cyclooxygenase; NSAID, nonsteroidal anti-inflammatory drug; mPGES, membrane prostaglandin E synthase; MAPEG, membrane-associated proteins in eicosanoid and glutathione metabolism; MGST1, microsomal glutathione transferase 1; cPGES, cytosolic prostaglandin E synthase; GSH, glutathione; NS-398, N-[2-(cyclohexyloxyl)-4-nitrophenyl]-methane sulfonamide; MK-886, 3-[1-(p-chlorobenzyl)-5-(isopropyl)-3-tert-butylthioindol-2-yl]-2, 2-dimethylpropanoic acid; LPS, lipopolysaccharide; IL, interleukin; TNF, tumor necrosis factor; NF, nuclear factor; EGR-1, early growth response gene 1; PPARγ, peroxisome proliferator-activated receptor γ; CIA, collagen-induced arthritis; CAIA, collagen antibody-induced arthritis; LDLR, low-density lipoprotein receptor; NSCLC, non-small cell lung cancer; hCG, human chorionic gonadotropin; VEGF, vascular endothelial growth factor.
-
Studies in the authors' laboratories were supported by grants from the Swedish Cancer Society, the Swedish Research Council, and funds from Karolinska Institutet.
-
This article is available online at http://pharmrev.aspetjournals.org.
-
doi:10.1124/pr.59.3.1.
- The American Society for Pharmacology and Experimental Therapeutics
References
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵
- ↵