Prostaglandins (PGs) are a family of cellular messengers exerting diverse homeostatic and pathophysiologic effects. Recently, several studies reported significant increases of PGI2 and PGF2α after the inhibition of microsomal PGE synthase-1 (mPGES-1) expression, which indicated that PGH2 metabolism might be redistributed when the PGE2 pathway is blocked. To address the determinants that govern the relative amounts of PGs, we developed an in vitro cell-free method, based on liquid chromatography-tandem mass spectrometry, to measure the exact amounts of these PGs formed in response to the addition of recombinant isomerases and their selective inhibitors. Our in vitro cell-free assay results were confirmed in cells using bone marrow-derived macrophage. Initially, we determined the in vitro stability of PGH2 and noted that there was spontaneous nonenzymatic conversion to PGD2 and PGE2. mPGES-1 markedly increased the conversion to PGE2 and decreased conversion to PGD2. Reciprocally, the addition of hematopoietic or lipocalin PGD synthase resulted in a relative increase of PGD2 and decrease of PGE2. A detailed titration study showed that the ratio of PGE2/PGD2 was closely correlated with the ratio of PGE synthase/PGD synthase. Our redistribution results also provide the foundation for understanding how PGH2 metabolism is redistributed by the presence of distal isomerases or by blocking the major metabolic outlet, which could determine the relative benefits and risks resulting from interdiction in nonrated-limiting components of PG synthesis pathways.
Cyclooxygenase (COX) enzymes, also known as PGH2 synthases, catalyze the oxygenation of arachidonic acid (AA) to PGG2, followed by the reduction of PGG2 to PGH2, which serves as a common substrate for various distal isomerases that generate five distinct primary PGs: PGE2, PGD2, PGF2α, PGI2, and thromboxane A2 (TXA2), of which 6-keto-PGF1α and TXB2 are the main stable nonenzymatic products of PGI2 and TXA2, respectively (Fig. 1). These PGs consist of a series of extracellular and intracellular messengers that produce diverse physiologic effects on pain (Zeilhofer, 2007), inflammation and fever (McAdam et al., 2000), allergy (Pettipher et al., 2007), platelets (FitzGerald, 1991), cardiovascular system (Vane, 1983), cancer growth (Wang et al., 2007), renal function (Hébert et al., 2005), reproduction (Weems et al., 2006), and possibly Alzheimer's disease (McGeer and McGeer, 1999). In many cases, different PGs have counter-regulatory effects. For example, in contrast to PGE2, PGD2 in the brain has a role in promoting sleep (Smyth et al., 2009). Furthermore, various PGs have the potential to both promote and counteract inflammatory processes in the body, especially in acute allergic inflammation. Thus, the exact physiologic or pathophysiologic response depends on the relative amounts of biologically active PG species.
After the enzymatic conversion of PGH2 was reported (Christ-Hazelhof et al., 1976), each PG-specific isomerase was discovered and purified, including PGE synthase, PGD synthase, PGFα synthase, PGI synthase, and TX synthase. Humans express three isoforms of PGE synthase: mPGES-1, mPGES-2, and cytosolic PGE synthase. Whereas mPGES-2 and cytosolic PGE synthase are constitutively expressed in vivo, mPGES-1 is of particular interest because it has been shown to be the most potent (Tanikawa et al., 2002) among PGE synthases and is induced by various stimuli including inflammatory signals in various cells and tissues (Guay et al., 2004). CAY10526 [4-(benzo[b]thiophen-2-yl)-3-bromo-5-hydroxy-dihydro-furan-2(3H)-one; C12H7BrO3S] and CAY10589 (2-[[4-[([1,1′-biphenyl]-4-ylmethyl)amino]-6-chloro-2-pyrimidinyl]thio]-octanoic acid; C25H28ClN3O2S) are synthetic compounds that have been reported to be selective inhibitors of mPGES-1 (Guerrero et al., 2007; Koeberle et al., 2008). PGD synthase activity is also comprised of two isozymes, H-PGDS and L-PGDS. HQL-79 [4-(diphenylmethoxy)-1–3-(1H-tetrazol-5-yl)propyl-piperidine; C22H27N5O] is a selective H-PGDS inhibitor (Matsushita et al., 1998a,b), and AT-56 [4-(5H-dibenzo[a,d]cyclohepten-5-ylidene)-1-[4-(2H-tetrazol-5-yl)butyl]-piperidine; C25H27N5] is reported to be a selective L-PGDS inhibitor (Irikura et al., 2009).
Nonsteroidal anti-inflammatory drugs (NSAIDs) competitively interfere with binding of the AA substrate to COX enzymes. Treatment with traditional NSAIDs such as aspirin and ibuprofen decrease PGE2 biosynthesis by nonselectively inhibiting both COX-1 and COX-2 (Garavito and Mulichak, 2003). However, serious gastrointestinal toxicity (Smyth et al., 2009) occurs at ordinary clinical doses because of their nonselective inhibition of both COX enzymes. There are also unintended consequences of selective inhibition of COX pathway enzymes. For example, rofecoxib and celecoxib are relatively selective for COX-2 and were developed under the assumption that the specific inhibition of COX-2 would be anti-inflammatory and analgesic but lack gastrointestinal toxicity (Garavito and Mulichak, 2003). However, these COX-2-selective inhibitors are associated with an increased risk of hypertension, cardiovascular disease, and stroke (Grosser et al., 2006; Timmers et al., 2007). This is, at least in part, because the inhibition of COX-2 also blocks the production of PGI2 and alters the balance between the vasoconstrictive properties of TXA2 and vasodilatory properties of PGI2 (Bombardier et al., 2000; Crofford et al., 2000). The recognition of these unintended consequences of individually blocking COX-1 and COX-2 has redirected efforts to examine the effect of interdiction in the individual downstream isomerases of PGH2, especially on the inhibition of mPGES-1 (Guerrero et al., 2007; Mbalaviele et al., 2010; Rörsch et al., 2010). Because PGH2 serves as a common substrate for at least five distinctive PGs and PGH2 itself is extremely unstable and short-lived, it would be premature to consider mPGES-1 as a promising and safe new therapeutic target without further investigation and understanding of the effect of inhibition of mPGES-1 on the entire PG cascade. To address this important issue, we developed an in vitro cell-free assay system using liquid chromatography-tandem mass spectrometry (LC-MS/MS) to measure the formation of PGE2 and PGD2 and determine the global consequence of the addition of recombinant isomerases and the impact of selective inhibition on the relative production of PGE2 and PGD2, which was also examined in cells using BMDM. We hypothesized that there would be a reciprocal relationship between the production of PGE2 and PGD2 when inhibiting their specific isomerases that could be biologically important.
Materials and Methods
For bone marrow experiments, C57BL/6 mice (6–12 weeks old, 25–35 g) were purchased from Harlan (Indianapolis, IN). Mice were housed in a temperature-controlled room with a 12-h light/dark cycle and given standard chow and tap water. All studies involving mice were approved by the Institutional Animal Care and Use Committee and complied with the Animal Welfare Act.
Ovine COX-1, human recombinant COX-2, human recombinant mPGES-1, human recombinant H-PGDS, human recombinant L-PGDS, GSH, CAY10526, CAY10589, HQL-79, AT-56, AA, PGH2, PGE2, PGD2, PGF2α, 6-keto-PGF1α, TXB2, PGI2 (sodium salt), 15-keto-PGF2α, 13,14-dihydro-15-keto-PGE2, 8-iso PGE2, d4-PGE2, and d4-PGD2 were purchased from Cayman Chemical (Ann Arbor, MI). Hemin, epinephrine, Tris base, hydrogen peroxide, citric acid, EDTA, and lipopolysaccharide (LPS) were purchased from Sigma-Aldrich (St. Louis, MO). Monosodium phosphate, disodium phosphate, hydrochloric acid, and butylated hydroxytoluene were purchased from Thermo Fisher Scientific (Waltham, MA). Formic acid was purchased from EMD Biosciences (San Diego, CA). Purified water was prepared using a Millipore Corporation (Billerica, MA) Milli-Q purification system or an ELGA (Saint Maurice Cedex, France) PURELAB Ultra purification system. Hanks' balanced salt solution (HBSS), fetal bovine serum (FBS), penicillin, and streptomycin were purchased from Invitrogen (Carlsbad, CA). Dulbecco's modified Eagle's medium (DMEM) was purchased from Mediatech (Herndon, VA). All organic solvents were HPLC grade or better and purchased from Thermo Fisher Scientific, and all other chemicals and solvents were American Chemical Society reagent grade, unless stated otherwise.
The In Vitro System.
All experiments were carried out using an in vitro system based on a COX functional assay (Cao et al., 2010, 2011). In brief, 20 μl of 25 mM GSH for mPGES-1 or 20 μl of 10 mM GSH for H-PGDS and L-PGDS was mixed on ice with 138 μl of 100 mM Tris·HCl buffer (pH 8.0, 37°C). Next, 20 μl of Tris·HCl buffer containing mPGES-1, H-PGDS, or L-PGDS was added and incubated on ice for 2 min. A 2-μl aliquot of enzyme inhibitor in dimethyl sulfoxide was added, and the solution was preincubated at 37°C in an Eppendorf Thermomixer R (Eppendorf North America, New York, NY) for 10 min. Each reaction was initiated by adding 20 μl of 20 μM PGH2 (2 μM final concentration) in Tris·HCl buffer as substrate and terminated after 30 min by adding 50 μl of 2 M HCl. In place of PGH2, 20-μl aliquots of 20 μM PGE2, PGD2, PGF2α, 6-keto-PGF1α, or TXB2 (final concentration 2 μM) were added to separate tubes as quantitative controls, which contained the same solution composition but without enzymes. The quantitative controls represented the maximum amounts of each prostaglandin that could be formed if PGH2 was quantitatively converted to a single product. After measurement using LC-MS/MS, the ratios of the peak areas of each prostaglandin in the experiments to the corresponding quantitative control were determined and expressed as percentages of the maximum theoretical yield.
d4-PGE2 and d4-PGD2 (10 μl; 100 ng/ml each in methanol/water, 50:50, v/v) were added as internal standards to correct for sample losses, degradation, or changes in mass spectrometer response. Each sample was extracted using 800 μl of hexane/ethyl acetate (50:50, v/v), and the organic phase was removed, evaporated to dryness under nitrogen gas, and reconstituted in 100 μl of methanol/water (50:50, v/v) immediately before quantitative analysis using LC-MS/MS. The peak areas were measured by using Applied Biosystems (Foster City, CA) Analyst software and manually inspected. The curves and data fitting were plotted by using Prism 5 (GraphPad Software Inc., San Diego, CA), and other calculations were carried out using Excel (Microsoft, Redmond, WA) or Numbers (Apple Computer, Cupertino, CA).
For the quantitative analysis of all prostaglandins except PGH2, HPLC separations were carried out using a Shimadzu (Columbia, MD) Prominence HPLC system with a Waters (Milford, MA) XTerra MS C18 (2.1 × 50 mm, 3.5 μm) analytical column and a 5-min isocratic mobile phase consisting of acetonitrile/aqueous 0.1% formic acid (37:63, v/v) at a flow rate of 200 μl/min. As shown in Fig. 2, all five derivatives of PGH2 were resolved to baseline in less than 4 min using these chromatographic conditions. The HPLC system was interfaced to an Applied Biosystems API 4000 triple quadrupole mass spectrometer, which was operated using negative ion electrospray. PGH2 was measured using LC-MS/MS with an 11-min linear gradient from 33 to 90% acetonitrile in aqueous 0.1% formic acid. Using gradient LC-MS/MS, PGH2 eluted at a retention time of 7.7 min.
The PGs formed abundant [M-H]− carboxylate ions during negative ion electrospray, which were fragmented using collision-induced dissociation with nitrogen as a collision gas. The collision energy (−24 to −30 V) was optimized for each PG to maximize the formation of product ions for detection using selected reaction monitoring (SRM). Isomeric PGE2, PGD2 (Cao et al., 2008), and PGH2 were measured using a SRM transition of m/z 351 to m/z 271, and the SRM transition of m/z 353 to m/z 193 was selected for PGF2α (Dahl and van Breemen, 2010). The SRM transition of m/z 369 to m/z 163 was used for 6-keto-PGF1α, and the transition of m/z 369 to m/z 169 was used for the measurement of TXB2. Likewise, the SRM of the transition of m/z 355 to m/z 275 was selected for the internal standards d4-PGE2 and d4-PGD2 (Cao et al., 2008).
High-resolution negative ion electrospray tandem mass spectra of PGH2 and its metabolites were acquired using a Waters Synapt G1 quadrupole time of flight (TOF) hybrid tandem mass spectrometer with a Waters Alliance 2690 HPLC system or a Shimadzu ion trap-TOF mass spectrometer with a Prominence HPLC system. HPLC separations were carried out as described above except that the mobile phase consisted of an 11-min linear gradient from 33 to 90% acetonitrile in aqueous 0.1% formic acid.
Cell Culture Assay.
Although the in vitro assay provided information regarding biological mechanisms of action, the results might not necessarily reflect in vivo processes or even the situation within a cell. Therefore, the BMDM was used in which mPGES-1 and H-PGDS (L-PGDS) could be selectively inhibited to observe the redistribution of PGH2 metabolism.
BMDM was isolated from the rear legs of sacrificed C57BL/6 mice. The harvested rear legs were soaked in HBSS containing 2% heat-inactivated FBS under aseptic conditions. The bone marrow cells were obtained by flushing the tibias and femurs using HBSS and cultured in DMEM supplemented with 10% FBS, 10% L929 cell-conditioned medium, 100 U/ml penicillin, and 100 μg/ml streptomycin. After 72 h of cultivation, the nonadherent cells were removed by changing the medium. Adherent cells were subsequently propagated in culture. Cells were split at day 7 by EDTA (1 mM) and plated at a density of 5 × 105/ml into six-well plates with L929 cell-conditioned medium. On day 8, the cell culture medium was changed to DMEM containing 1% FBS, penicillin, and streptomycin for 1.5 h before the addition of inhibitors CAY10526, CAY10589, HQL-79, and AT-56 2 h before the LPS (1 μg/ml) treatment. Cells were incubated with 5% CO2 humidified air at 37°C. The cell supernatants were collected and stored at −80°C until analysis after 16-h treatment with LPS and different inhibitors. LPS was used to stimulate BMDM to activate COX-2 production and prostaglandin synthesis (Xiao et al., 2008).
Each supernatant was spiked with d4-PGE2 and d4-PGD2 as internal standards, and citric acid and butylated hydroxytoluene were then added to prevent free radical-catalyzed peroxidation. PGs were extracted using hexane/ethyl acetate. After centrifugation, the upper organic phase was collected and evaporated to dryness. Immediately before analysis using LC-MS/MS, each extract was reconstituted in methanol/water (Cao et al., 2008). Standards for calibration curves and quality-control measurements were prepared by spiking cell culture medium with measured amounts of PGE2 and PGD2. These standards were then processed as described above. The concentrations of PGE2 and PGD2 in these standards ranged from 0.1 to 1000 ng/ml (Cao et al., 2008).
Samples were run in triplicate, and values are expressed as mean ± S.D. Statistical significance was assessed using one-way analysis of variance, and p values <0.05 were considered to indicate significant differences for all statistical tests.
Validation of Enzyme and PGH2 Purity.
To ensure the purity of each enzyme, ovine COX-1, human COX-2, mPGES-1, H-PGDS, and L-PGDS were incubated with AA and PGH2 at 37°C for 10 min, and the production of PGs was then determined. Ovine COX-1 and human COX-2 did not increase the formation of PGE2 or PGD2 from PGH2 compared with the control (Fig. 3A), suggesting that neither ovine COX-1 nor human COX-2 was contaminated with PG isomerases. In addition, the isomerases mPGES-1, H-PGD, and L-PGDS had no effect on the metabolism of AA in the absence of ovine COX-1 or human COX-2 (Fig. 3B), suggesting that these isomerases were not contaminated with either ovine COX-1 or human COX-2. The 2 μM PGH2 solution used in these experiments was also analyzed using LC-MS/MS and was found to already contain 6.1 ± 1.1% PGE2, 1.8 ± 0.7% PGD2, and 0.2 ± 0.1% PGF2α because of nonenzymatic rearrangement.
In Vitro System Optimization.
PBS buffer (pH 6.0–8.0, 37°C) and Tris·HCl buffer (pH 7.0–8.0, 37°C) produced similar results during incubations with PGH2, but Tris·HCl buffer (pH 8.0, 37°C) produced slightly larger enzymatic yields and is recommended by Cayman Chemical. Therefore, Tris·HCl buffer (pH 8.0, 37°C) was used for all subsequent experiments. GSH has been reported to function as an essential cofactor for three isomerases (mPGES-1, H-PGDS, and L-PGDS) (Ouellet et al., 2002; Herlong and Scott, 2006; Hohwy et al., 2008). In our hands, GSH enhanced PGs formation only slightly for mPGES-1 and L-PGDS, and it was not essential for the function of these enzymes. For H-PGDS, however, GSH was indispensable (Table 1). GSH concentrations (0.1–12.5 mM) were varied among three isomerases, but did not affect reaction rates significantly. Thus, according to recommendations of the supplier (Cayman Chemical), 2.5 mM GSH was used for mPGES-1, and 1 mM GSH was used for H-PGDS and L-PGDS. The optimal levels of mPGES-1, H-PGDS, and L-PGDS were approximately 3, 0.1, and 1 unit/μl, which was at the upper part of the linear range of the dose-response curve. A preincubation time of 10 min was used before adding the substrate to allow the inhibitors to interact fully with the enzyme and reach maximal inhibition potency (Cao et al., 2010, 2011). A reaction time of 30 min was used for enzymatic conversion of PGH2 to PGs. A substrate dose-response curve was generated that indicated the production of each PG product after 30 min reached a plateau at approximately 4 μM substrate. A substrate concentration of 2 μM corresponded to 50% response and was determined to be the Km of the substrate.
Identification of PGH2 and Characterization of Unknown Derivatives.
During the LC-MS/MS analysis of a PGH2 standard (monitoring the SRM transition of m/z 351 to m/z 271), five peaks were observed as shown in Fig. 4. The first two peaks were identified according to comparison with standards as PGE2 and PGD2, and the last peak was confirmed to be PGH2 because it continued decreasing, whereas other peaks increased over 160 min (Fig. 5). The two remaining peaks (retention times 5.7 and 6.3 min) remain unidentified and are not (by comparison with standards) PGI2, 15-keto PGF2α, 13,14-dihydro-15-keto PGE2, or 8-iso PGE2. In a separate experiment, the half-life of PGH2 in Tris·HCl buffer at pH 8 and 37°C was determined to be 5 min (Fig. 6).
For additional characterization of the unknown peaks, LC-MS/MS analysis was carried out using high-resolution quadrupole TOF and ion trap-TOF mass spectrometers. All five compounds separated as shown in the chromatogram in Fig. 4 were isomeric with identical elemental compositions of C20H32O5. During tandem mass spectrometry on these mass spectrometers using different collision energies, all five compounds produced similar tandem mass spectra without any characteristic fragmentation that could distinguish them.
In Vitro Nonenzymatic Conversions of PGH2.
Because PGH2 is extremely unstable and has a short half-life in aqueous solution, we used LC-MS/MS with SRM to determine the main degradation products of PGH2 and their relative yields. In the absence of any enzyme, PGH2 spontaneously converted to PGE2, PGD2, and a small amount of PGF2α but no 6-keto-PGF1α or TXB2. Therefore, neither PGI2 nor TXA2 was spontaneously produced from PGH2. The yields of PGE2, PGD2, and PGF2α were approximately 44, 15, and 1.6%, respectively, with a constant ratio among these species of ∼3:1:0.1. We observed an unexpected phenomenon after optimizing the reaction conditions. There was nearly a 100% increase in PGD2 formation from PGH2 after the addition of 1 or 2.5 mM GSH, and for PGF2α, there was a 600% increase. However, PGE2 formation was slightly decreased or not affected. Moreover, by adding an equivalent amount of H2O2, the effect of GSH addition on PGD2 and PGF2α was totally eliminated (Fig. 7A), indicating that reactive oxygen species might have a regulatory role in governing the relative amounts of these prostaglandins.
In Vitro Enzymatic Conversions of PGH2.
Because PGE2 and PGD2 play significant physiological roles in humans and are predominant products of PGH2, we examined the roles in their formation of three commercially available human recombinant distal enzymes, mPGES-1, H-PGDS, and L-PGDS. When the highest concentration of mPGES-1 (300 units/μl) was incubated with PGH2, the yield of PGE2 increased to 62%, whereas PGD2 decreased to 3.6% and PGF2α decreased to basal level (Fig. 7A). Likewise, when either the highest concentration of H-PGDS (120 units/μl) or L-PGDS (10 units/μl) was incubated with 2 μM PGH2, the yield of PGD2 increased to 78%, whereas the PGE2 formation decreased to 10.8% and PGF2α decreased to basal level (Fig. 7B).
We next examined the impact of the addition of pharmacologic inhibitors of these downstream enzymes by using the commercially available mPGES-1-selective inhibitors CAY10526 and CAY10589, H-PGDS-selective inhibitor HQL-79, and L-PGDS-selective inhibitor AT-56 in our in vitro system. After incubating PGH2 with mPGES-1 (3 units/μl) and 300 μM CAY10526 or 100 μM CAY10589, PGE2 formation decreased to the basal level by 45%, whereas PGD2 formation increased to 25%, and PGF2α increased only slightly (Fig. 7A). Incubation with the H-PGDS-specific inhibitor HQL-79 (50 μM) inhibited approximately 90% of PGD2 formation (decreased to 26%) in the presence of H-PGDS (0.1 unit/μl), whereas PGE2 formation increased to 38%, and PGF2α decreased to basal level (Fig. 7B). Likewise, the L-PGDS-specific inhibitor AT-56 (100 μM) completely prevented PGD2 formation (decreased to 24%) when incubating PGH2 with L-PGDS (1 unit/μl), whereas PGE2 formation increased to 40% and PGF2α decreased to basal level (Fig. 7B). The effects of these inhibitors on other isomerases were also tested. We were surprised to find that CAY10526 and CAY10589 were less specific than expected and also inhibited both H-PGDS and L-PGDS in vitro. In contrast, HQL-79 had no inhibitory effect on either L-PGDS or mPGES-1, and AT-56 inhibited neither H-PGDS nor mPGES-1 (Table 2).
In Vitro Redistribution Assay.
Because obvious shifts of PGH2 to other PGs were observed using inhibitors to block their corresponding isomerases, we predicted that there would be redistribution in products by inhibition of isomerases that compete for the substrate PGH2. We first mixed different volumes of mPGES-1 (3 units/μl) and H-PGDS (0.1 unit/μl) or L-PGDS (1 unit/μl) in the same system. As a result, with more mPGES-1 activity, more PGE2 and less PGD2 were produced and vice versa (Fig. 8) in a predictable fashion according to the ratio of PGES to PGDS activity. We also examined the impact of pharmacologic inhibition with HQL-79 and AT-56 to further determine the consequence of redistribution by inhibiting H-PGDS and L-PGDS, respectively (Fig. 9). As expected, by increasing the concentrations of HQL-79 or AT-56, conversions of PGH2 moved toward the production of PGE2, whereas the production of PGD2 decreased. Because our data showed that the purported mPGES-1 inhibitors CAY10526 and CAY10589 also strongly inhibited H-PGDS and L-PGDS, redistribution tests by inhibiting mPGES-1 with these agents did not produce redistribution of PGH2 metabolism in vitro.
Cell Culture Redistribution Assay.
CAY10526 and CAY10589 did not produce redistribution of PGH2 metabolism in BMDM because these agents inhibited not only mPGES-1 but also H-PGDS and L-PGDS. AT-56 had no effect on decreasing PGD2 production, which was not surprising because L-PGDS protein expression was not detectable in BMDM (L. Xiao, unpublished data). Therefore, only H-PGDS was available for inhibition by pharmacologic intervention in BMDM. Similar to the in vitro redistribution assay, treatment of BMDM with the H-PGDS inhibitor HQL-79 altered PGH2 conversion toward the production of PGE2, whereas the production of PGD2 decreased (Fig. 10).
Previously, we reported that a COX functional assay (Cao et al., 2011) used to screen COX-2-selective inhibitors results in the production of both PGE2 and PGD2, which raised the obvious question as to how PGE2 and PGD2 are produced in the absence of downstream isomerases. We initially suspected that the COX-2 enzyme was contaminated with certain isomerases or it might have some unknown functions that could catalyze conversion of PGH2 to PGs. When incubating COX-1 or COX-2 with PGH2 produced no increase in the production of the PGs, we then redirected our focus to investigating the stability of the intermediate PGH2. These studies showed that PGH2 was extremely unstable in aqueous solution in vitro and spontaneously and nonenzymatically converted to PGE2, PGD2, and PGF2α, with a constant ratio of ∼3:1:0.1 (Fig. 7). This not only provided an explanation for the results in our COX functional assay but also clarified the identities and yields of PGH2 products. Based on our in vitro findings, it is possible that PGH2 would also convert to PGs in vivo without any downstream isomerases, which further brings into question the efficacy of inhibiting PGE2 formation using mPGES-1 inhibitors.
The pivotal finding of our study is the PGH2 metabolic redistribution to other forms of PGs by the inhibition of either PGE synthase or PGD synthase (Fig. 7). A possible explanation is that mPGES-1 and H-PGDS compete for the substrate PGH2 so that inhibition of one enzyme would shunt PGH2 to the other. To prove this explanation, several different approaches were used. Titration studies (Fig. 8) showed how the conversion of PGH2 was closely correlated with the relative levels of PGE synthase and PGD synthase. A similar redistribution phenomenon (from PGD2 to PGE2) was also observed in vitro (Fig. 9) when using the H-PGDS-selective inhibitor HQL-79 or the L-PGDS-selective inhibitor AT-56, because the major outlet PGD2 was blocked. Moreover, the BMDM cell culture study (Fig. 10) confirmed the PGH2 metabolism redistribution in cells by inhibiting H-PGDS. Purported to be mPGES-1-selective inhibitors, CAY10526 and CAY10589 also inhibited PGD synthase and could not be used to study redistribution toward PGD2. PGH2 metabolism redistribution also explains the reported increases of PGI2 and PGF2α by inhibition of mPGES-1 in mice (Guay et al., 2004; Trebino et al., 2005; Mbalaviele et al., 2010; Rörsch et al., 2010).
In the near future, our observations could probably be used to provide therapeutic guidance for diseases involving PGs, because the redistribution and accumulation of certain types of PGs might be beneficial and useful for treatment. For example, higher levels of PGI2 and PGF2α would be favorable because they have anti-inflammatory effects, and more PGD2 could have anti-inflammatory effects (Smyth et al., 2009). Because various PGs can produce opposing effects, imbalance among them might cause unexpected problems or risks that must not be ignored such as with the use of highly selective COX-2 inhibitors. For instance, an increase of PGI2 might cause side effects such as low blood pressure in certain patients and even result in certain chronic and fatal diseases that are not easy to observe (Smyth et al., 2009).
We observed moderate inhibition of PGE2 formation by CAY10526 in vitro; however, CAY10526 was reported to be a strong mPGES-1-selective inhibitor (Guerrero et al., 2007). The probable reason (Guerrero et al., 2007) is that CAY10526, which caused a significant reduction in PGE2 production in the induction phase, did not affect PGE2 formation in the postinduction phase, which means that CAY10526 selectively inhibited mPGES-1 protein expression.
During the identification of PGH2 (Fig. 4), two peaks (retention times 5.7 and 6.3 min) remained unidentified but were not (by comparison with standards) PGI2, 15-keto PGF2α, 13,14-dihydro-15-keto PGE2, or 8-iso PGE2. Some possible structures include the reactive γ-keto aldehydes and levuglandins E2 and D2 (Boutaud et al., 1999). All of these compounds are isomers of PGH2.
GSH has been reported to be an essential cofactor for mPGES-1 (Ouellet et al., 2002), H-PGDS (Hohwy et al., 2008), and L-PGDS (Herlong and Scott, 2006). Our studies confirmed the indispensability of GSH for H-PGDS, but it was not necessary for either mPGES-1 or L-PGDS in our hands (Table 1). This indicated that GSH was not an essential cofactor for mPGES-1 and L-PGDS in our in vitro cell-free assay if these two isomerases were not contaminated with GSH. During the nonenzymatic conversion study of PGH2, a 600% increase of PGF2α production and an almost 100% increase of PGD2 production were found when adding GSH into PGH2 standard in aqueous solution (Fig. 7A). It has been reported that glutathione transferase increases PGF2α formation, which results mainly from the glutathione transferase enzymatic catalysis (Burgess et al., 1987). However, other reducing agents, like GSH, can also increase PGF2α production (Christ-Hazelhof et al., 1976; Nugteren and Christ-Hazelhof, 1980; Burgess et al., 1987), which indicates that this is a nonenzymatic process, possibly by serving as a specific electron donor promoting the conversion of PGH2 to PGF2α (Keeting et al., 1987). However, little is known about the increase of PGD2 by adding GSH, which raised our interest in the undiscovered differences between PGE2 and PGD2. Because both PGE2 and PGD2 have the same elemental composition and similar structure, it is very curious that they behave so differently in the presence of GSH or other reagents. These issues are being addressed by our ongoing studies.
In summary, although mPGES-1 inhibitors could be used as alternatives to COX-2 inhibitors to specifically decrease the amount of PGE2, they could also cause PGH2 metabolism redistribution. Thus, it is necessary to have a full understanding of the mechanisms of action and physiological effects of all PGs. The consequences of selective inhibition of one or more prostaglandin isomerases determine possible subsequent risks. mPGES-1 was found to be functionally coupled with COX-2 (Murakami et al., 2000), which means that mPGES-1 would normally catalyze the transformation of PGH2 produced by COX-2. Then using one or more inhibitors to partially inhibit COX-2 and strongly inhibit mPGES-1 might be a more reasonable therapeutic approach. Alternatives might include PGE2 receptor antagonists or high-affinity PGH2 analogs.
Participated in research design: Yu, Xiao, Zhao, Christman, and van Breemen.
Conducted experiments: Yu and Zhao.
Contributed new reagents or analytic tools: Xiao, Christman, and van Breemen.
Performed data analysis: Yu.
Wrote or contributed to the writing of the manuscript: Yu, Xiao, Zhao, Christman, and van Breemen.
We thank Dr. Shunyan Mo and Linlin Dong for assistance with high-resolution mass spectrometry.
This work was supported by the National Institutes of Health Division of Intramural Research [Grants R01-HL075557, HL103643a, 5R01-HL083218, 3R01-HL083218-01A2S1]; the National Institutes of Health National Cancer Institute [Grant P01-CA048112]; the National Institutes of Health National Center for Complementary and Alternative Medicine [Grant P50-AT00155]; the Department of Veterans Affairs Merit Review [Grant 1I01BX000108]; and the University of Illinois, Chicago Faculty Scholarship Support Program.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- microsomal PGE synthase
- nonsteroidal anti-inflammatory drug
- liquid chromatography-tandem mass spectrometry
- bone marrow-derived macrophage
- hematopoietic PGD synthase
- lipocalin PGD synthase
- arachidonic acid
- thromboxane A2
- thromboxane B2
- Hanks' balanced salt solution
- fetal bovine serum
- Dulbecco's modified Eagle's medium
- selected reaction monitoring
- time of flight
- high-performance liquid chromatography
- 2-[[4-[([1,1′-biphenyl]-4-ylmethyl)amino]-6-chloro-2-pyrimidinyl]thio]-octanoic acid
- Received June 22, 2011.
- Accepted August 22, 2011.
- U.S. Government work not protected by U.S. copyright