The Endocannabinoid Metabolite Prostaglandin E₂ (PGE₂)-Glycerol Inhibits Human Neutrophil Functions: Involvement of Its Hydrolysis into PGE₂ and EP Receptors

Acute and chronic inflammation are characterized by leukocyte infiltration, proinflammatory mediator production, and tissue destruction. Non-steroidal anti-inflammatory drugs have long been used to limit pain and inflammatory damage, notably by blocking the production of PGs and thromboxane. Moreover, cyclooxygenase-2 (COX-2) inhibitors, which prevent the biosynthesis of PGE₂, have been helpful at decreasing inflammation and inflammatory pain (1).

Historically, COX-2 was perceived as a proinflammatory enzyme, because its expression is induced by inflammatory stimuli and leads to PGE₂ synthesis. This was eventually challenged by Gilroy et al. (2) who suggested that COX-2 could have anti-inflammatory properties. It was then shown in several murine models that COX-2 blockade worsens inflammation or delays its resolution, at least in mice (3–5). In addition to PGs, COX-2 participates in the synthesis of other lipids that modulate nociception and inflammation, notably PG-glycerol (PG-G) and PG-ethanolamide (PG-EA). These endocannabinoid derivatives are products of the COX-2–dependent oxygenation of 2-arachidonoyl-glycerol (2-AG) and anandamide, respectively. PG-EAs, or prostamides, preferentially activate the prostamide receptor over PG receptors (6). In contrast, PGE₂-G can bind to several PGE₂ receptors to a lesser extent than PGE₂ (7). Moreover, PGE₂-G induces key signaling events that are not mimicked by PGE₂ (8, 9).

PGE₂ is a well-established inhibitor of human neutrophil functions such as leukotriene B₄ (LTB₄) biosynthesis, reactive oxygen species (ROS) production, and migration (10–14). These effects are the consequence of elevated cyclic cAMP and involve the EP₂ receptor. Given that PGE₂-G is produced by COX-2 and binds to some of the EP receptors (7), because PGE₂ inhibits neutrophil functions, and because PG-Gs were shown to regulate inflammation (15), we undertook experiments to assess whether PGE₂-G and PGE₂-EA would also modulate human neutrophil functions.

AH6809, 19-OH-PGB₂, 2-AG, butaprost (free acid), CAY 10598, H-89, JZL184, L-902 688, methyl arachidonoyl fluorophosphonate (MAFP), PGE₂, PGE₂-EA, PGE₂-G, PGE₂-serinol amide (SA), PGD₂-G, PGD₂-SA, PGF_2α-G, sulprostone, tetrahydrolipstatin, WWL70, and the primary Ab for MAG lipase were purchased from Cayman Chemical (Ann Arbor, MI). Thapsigargin was obtained from Tocris Bioscience (Ellisville, MO). DMSO was purchased from Sigma-Aldrich (St Louis, MO). Protease inhibitor mixture tablets and adenosine deaminase were purchased from Roche (Laval, QC, Canada). Aprotinin, leupeptin, and WWL113 were purchased from Sigma-Aldrich. DFP was purchased from BioShop Canada (Burlington, ON, Canada). The HRP-linked anti-mouse IgG and anti-rabbit IgG secondary were obtained from Cell Signaling Technology (Beverly, MA). Primary Abs for ABHD6, ABHD12, and PPT1 were purchased from Abcam (Toronto, ON, Canada). The LYPLA2 primary Ab was purchased from Abnova (Taipei City, Taiwan), the ABHD16A primary Ab was purchased from Thermo Fisher Scientific (Waltham, MA), and the primary Ab for CES1 was purchased from R&D Systems (Minneapolis, MN). PMSF, RO 20-1724, palmostatin B, and the ECL detection kit were purchased from EMD Millipore (Billerica, MA). The magnetic bead–conjugated anti-CD16 mAb and MACS were purchased from Miltenyi Biotec (Auburn, CA). HBSS and RPMI 1640 were obtained from Wisent Laboratories (St-Bruno, QC, Canada). Dextran and HPLC-grade methanol and acetonitrile were purchased from Fisher Scientific. ML349 was a generous contribution from Dr. L. Marnett (Vanderbilt University, Nashville, TN). The HNP1-3 defensin ELISA kit was purchased from Hycult Biotech (Uden, the Netherlands). ONO-AE2-227 was a generous gift from Ono Pharmaceutical (Osaka, Japan). Cytochrome C was obtained from Bio Basic (Amherst, NY).

This work required the use of human cells from healthy volunteers and was approved by our institutional ethics committee. All the experiments were conducted with the understanding and the signed consent of each participant.

Human venous blood was obtained from volunteers and collected in tubes containing K₃EDTA as an anticoagulant. Eosinophil-depleted neutrophils were then isolated as described previously (16). In brief, platelet-rich plasma and erythrocytes were discarded by centrifugation and dextran sedimentation, respectively. Mononuclear cells were then separated from granulocytes by centrifugation on Lymphocyte Separation Medium cushions (Corning), and residual erythrocytes were removed from the granulocyte pellets by hypotonic lysis using sterile water. Neutrophils were separated from eosinophils using anti-CD16–coated magnetic beads, according to the manufacturer’s instructions. The purity and viability of the resulting neutrophil suspensions were always ≥98%, as assessed by Diff-Quik staining and trypan blue exclusion, respectively.

To better mimic the fate of human neutrophils, adenosine deaminase (0.3 U/ml) was prepared and added 10 min before the addition of the stimuli in all experiments (17, 18).

Prewarmed human neutrophil suspensions (37°C, 5 × 10⁶ cells/ml) in HBSS containing 1.6 mM CaCl₂ were incubated 5 min with PGE₂, PGE₂-G, PGE₂-EA, PGE₂-SA, PGD₂-G, PGD₂-SA, PGF_2α-G, or the different EP receptor agonists (see figure legends for concentrations), then stimulated with either 3 μM 2-AG (5 min), 3 μM arachidonic acid (AA) (5 min), or 100 nM thapsigargin (10 min). In experiments where inhibitors were used, they were added 10 min before the addition of the stimulus. Incubations were stopped by adding 0.5 volume of a cold (−20° C) stop solution [MeOH/MeCN, 1/1 (v/v)] containing 12.5 ng of both 19-OH-PGB₂ and PGB₂ as internal standards. Samples were denatured overnight (−20°C), centrifuged (700 × g, 10 min) to eliminate the denatured proteins, then analyzed by reversed phase HPLC using an online extraction procedure (19). LTB₄, 20-COOH-LTB₄, 20-OH-LTB₄, and 5(S)-HETE were quantitated using PGB₂ as internal standard and are referred to as LTs.

Superoxide anion production by human neutrophils was assessed by cytochrome c reduction exactly as described previously (20). PGE₂, PGE₂-EA, PGD₂-G, PGE₂-G, PGF_2α-G, PGD₂-SA, PGE₂-SA, or the EP receptor agonists were added 5 min before fMLF. EP receptor antagonists or serine hydrolase inhibitors were added 10 min before fMLF.

Migration assays were performed using 3 μm pore inserts (Becton Dickinson), as recommended by the manufacturer. In brief, 700 μl of prewarmed (37°C) HBSS containing 1.6 mM CaCl₂ and 100 nM LTB₄ were added in the lower chamber and 200 μl of prewarmed neutrophil suspensions [37°C, 2.5 × 10⁶ cells/ml in HBSS containing 1.6 mM CaCl₂ and 5% (w/v) FBS] were added in the upper chamber of the transmigration apparatus. Neutrophils were allowed to migrate for 2 h at 37°C. The upper chambers were then removed and migrated cells in the lower chamber of the migration apparatus were counted using a Scepter 2.0 handheld automated cell counter. In experiments where PGE₂, PGE₂-G, and the type IV phosphodiesterase RO 20-1724 were used, they were added to both upper and lower chamber for 5 min before the addition of LTB₄ in the lower chamber of the migration assay.

Escherichia coli (#25922; American Type Culture Collection) was grown overnight at 37°C in tryptic soy broth. The obtained culture was diluted (1:100) in fresh media and incubated at 37°C until an OD of 0.5 at 600 nm was reached. Then 500 μl of the E. coli culture were washed and suspended in sodium phosphate buffer at a final concentration of 1000 CFU/ml in sodium phosphate buffer. Freshly isolated human neutrophils (37°C, 20 × 10⁶ cells/ml) in HBSS containing 1.6 mM CaCl₂ incubated for 15 min with 10 μM of cytochalasin B then activated with 1 μM 2-AG for 5 min. PGE₂ (1 μM) or PGE₂-G (1 μM) were added 5 min before the addition of 2-AG. Incubations were stopped by transferring the tubes in an ice-water bath. Samples were centrifuged (500 × g; 4°C; 5 min) and the resulting supernatants were mixed with the bacterial suspensions (1:1) and incubated for 4 h on a rotating plate at 37°C. The mixtures then were diluted (1:300) and plated on Luria–Bertani agar plates. Colonies were counted after the incubation of the agar plates overnight at 37°C.

Following cell stimulation, samples were rapidly centrifuged; the supernatants were collected and stored at −80°C until further analysis. Quantitation of α-defensins in the supernatants was performed using a commercially available ELISA kit (Hycult Biotech), according to the manufacturer’s instructions.

Total RNA extracts were prepared using TRIzol according to the manufacturer’s instructions. cDNA was obtained by reverse transcription using the iScript Reverse Transcription Supermix from Bio-Rad, with an RNA input of 1 μg per reaction. For the analysis of EP receptor expression, quantitative PCR (qPCR) assays were performed on a 7900 Fast Real-Time PCR system (Applied Biosystems) using customs RT2 Profiler qPCR Multiplex Array Kit (Qiagen). For the analysis of the different endocannabinoid hydrolases, qPCR was performed with the SsoAdvanced Universal SYBR Green Supermix on a CFX96 thermal cycler (Bio-Rad, Mississauga, ON, Canada) according to the manufacturer’s instructions. Primers for the 18S housekeeping gene were obtained from Qiagen (QT00199367). Primers for all other target genes were designed using the Primer-BLAST tool and synthetized by IDT (Coralville, IA). Primer sequences are as follows: (5′→ 3′): ABHD6 forward 5′-CATCTGGGGGAAACAAGACCA-3′, ABHD6 reverse 5′-TTTCCATCACTACTGAGTGCCC-3′, ABHD12 forward 5′-CGGATACTGAGGGAATTCCTGG-3′, ABHD12 reverse 5′-AGGTCTTCATGCTTCCTTCCC-3′, MAG lipase forward 5′-TGCCTACCATGTTCTCCACA-3′, MAG lipase reverse 5′-CCTCCAGTTATTGCAGTCTGG-3′, CES1 forward 5′-TGCCTTTATCCTGGCCACTC-3′, CES1 reverse 5′-CTTGGGTGCACATAGGAGGG-3′, PPT1 forward 5′-TGGCATGGGATGGGTGTTTT-3′, PPT1 reverse 5′-GGCGTTCCTGAACAACTTTGG-3′, LYPLA2 forward 5′-AAGAAGGCAGCAGAGAACATC-3′, LYPLA2 reverse 5′-CTCCCAGCACGATTCGATTG-3′, ABHD16A forward 5′-CCCCCGGCTCTACAAAATCTAC-3′, ABHD16A reverse 5′-GATAGTACGTATCCCAGGAGCTG-3′. Results are expressed in a relative quantification normalized to the 18S rRNA as reference gene with the 2^−ΔΔCT method.

For the analysis of PGE₂-G, PGE₂, and D₄-PGE₂, incubations were stopped by the addition of one volume of cold (−20°C) MeOH and samples were kept frozen until further processing. Samples were centrifuged (700 × g; 10 min) after addition of 2 ng of the internal standards (D₄-PGE₂ and D₄-PGE₂-G). The supernatants were diluted with water to a final MeOH concentration of 10% and loaded on solid phase extraction cartridges (Strata-X Polymeric Reversed Phase, 60 mg/1 ml; Phenomenex). Cartridges were washed with 2 ml water then lipids were extracted with 1 ml MeOH. The eluates were dried down and reconstituted in 40 μl of HPLC solvent A (8.3 mM acetic acid buffered to pH 5.7 with ammonium hydroxide) and 20 μl of solvent B (MeCN/MeOH, 65/35, v/v). A 25 μl aliquot was injected on to a reversed phase HPLC column (Ascentis C18, 150 × 2.1 mm, 5 μm; Supelco) eluted at a flow rate of 200 μl/min with a linear gradient from 45% solvent B, increased to 75% in 12 min, from 75 to 98% in 2 min, and held for 10 min at 98% B before re-equilibration to 45% B in 10 min. The HPLC system was directly interfaced into the electrospray source of a triple quadrupole mass spectrometer (API 3000; AB Sciex) and mass spectrometric analysis was performed in the negative ion mode using multiple reaction monitoring for the specific mass transitions of PGE₂ (m/z 351.3→271.2) and D₄-PGE₂ (m/z 355.3→275.2) and in the positive ion mode for the ammonium adducts [M+NH₄]⁺ of PGE₂-G (m/z 444.4→391.3) and D₄-PGE₂-G (m/z 448.4→395.3). Quantitation was performed using standard isotope dilution curves, as previously described (21).

In experiments in which concentrations of JZL184 higher than 1 μM were used, LTB₄ biosynthesis was analyzed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) and multiple reaction monitoring. In brief, incubation was stopped by the addition of one volume of cold (−20°C) MeOH and samples were kept frozen until further processing. Samples were centrifuged (700 × g; 10 min) after addition of 2 ng of the internal standard (D₄-PGB₂). The supernatants were diluted with water to a final MeOH concentration of 10% and loaded on solid phase extraction cartridges (Strata-X Polymeric Reversed Phase, 60 mg/1 ml; Phenomenex). Cartridges were washed with 2 ml water then lipids were extracted with 1 ml MeOH containing 2% formic acid. The eluates were dried down and reconstituted in 25 μl of HPLC solvent A (0.05% formic acid in water) and 25 μl of solvent B (0.05% formic acid in acetonitrile). A 25 μl aliquot was injected on to an HPLC column (Kinetec C8, 150 × 2.1 mm, 2.6 μm; Phenomenex) eluted at a flow rate of 400 μl/min with a discontinuous gradient (from 10% solvent B, increased to 25% in 15 min, from 25 to 35% in 5 min, from 35 to 75% in 10 min, from 75 to 95% in 0.1 min and held for 5 min at 95% B before re-equilibration to 10% B in 5 min). The HPLC system was directly interfaced into the electrospray source of a triple quadrupole mass spectrometer (Shimadzu 8050) and mass spectrometric analysis was performed in the negative ion mode using multiple reaction monitoring for the specific mass transitions of D₄-PGB₂ (m/z 337.20→179.05), LTB₄ (m/z 351.2→195.1), 20-OH-LTB₄ (m/z 351.2→195.1), 20-COOH-LTB₄ (m/z 365.3→169.1), and 5-HETE (m/z 319.2→115.1). Quantitation was performed using standard isotope dilution curves, as previously described (21).

Cells were lysed by sonication at 4°C in sucrose buffer containing 10 μg/ml leupeptin, 10 μg/ml aprotinin, 1 mM PMSF, 3 mM DFP, and one tablet protease inhibitor mixture (for 10 ml of buffer). Laemmli sample buffer [62.5 mM TRIS-HCl (pH 6.8), 2% SDS, 10% glycerol, 0.01% bromophenol blue] was added to sonicated cell lysates and samples were boiled for 10 min. Buffer volumes were adjusted to obtain a final concentration of 2 × 10⁶ cells/50 μl of lysate for all samples. Proteins were separated by SDS-PAGE on 12% polyacrylamide gels and transferred on to polyvinylidene difluoride membranes. Transfer efficiency and equal protein loading were verified by Ponceau Red staining. Membranes were placed in TBS-Tween buffer (25 mM Tris-HCl [pH 7.6], 0.2 M NaCl, 0.15% Tween 20) containing 5% non-fat dried milk (w/v) for 30 min at room temperature, then probed with the primary Ab (4°C, overnight). The membranes were revealed by chemiluminescence using a HRP-coupled secondary Ab and an ECL detection kit.

The effect of the lipase inhibitors on PGE₂-G hydrolysis was analyzed with GraphPad Prism 6. The software was used to perform one-way ANOVA with Dunnett’s test. The p values <0.05 were considered significant.

We first performed a series of experiments in which we assessed the impact of PGE₂, PGE₂-G, and PGE₂-EA on LT biosynthesis. A 5-min pretreatment of neutrophils with PGE₂-G, but not PGE₂-EA, prevented LTB₄ biosynthesis by thapsigargin-stimulated neutrophils in a concentration-dependent manner (Fig. 1A). Under that experimental setting, the inhibitory effect of PGE₂-G on LTB₄ biosynthesis was less potent than that observed with PGE₂ by one order of magnitude (IC₅₀ of 3 and 30 nM for PGE₂ and PGE₂-G, respectively). A similar pattern was obtained when we assessed ROS production induced by fMLF (Fig. 1B). PGE₂-G also inhibited the migration of human neutrophils ex vivo, although this required a higher concentration (10 μM) and the use of the type IV phosphodiesterase inhibitor RO 20-1724 (Fig. 1C). Finally, PGE₂-G and PGE₂ also inhibited the ability of 2-AG–activated neutrophil supernatants to kill E. coli (Fig. 1D). This inhibitory effect of PGE₂ and PGE₂-G correlated with an inhibition of antimicrobial peptide release by 2-AG– or fMLF-stimulated neutrophils (Fig. 1E). Altogether, the results presented in Fig. 1 indicate that in contrast to PGE₂-EA, PGE₂-G exerts potent inhibitory effects on human neutrophil functions. Of note, neither PGE₂-G or PGE₂-EA stimulated any of the functional responses investigated above (data not shown).

PGE₂ mediates its effects by activating the EP receptors 1 to 4. We thus analyzed EP receptor expression in human neutrophils by qPCR array and confirmed that they mainly express the EP₂ and EP₄ receptors (Fig. 2A). To establish the contribution of the EP₂ and EP₄ receptors in the inhibitory effect of PGE₂-G and PGE₂ on human neutrophils, we performed experiments with the EP₁/EP₂ receptor antagonist AH-6809 and the EP₄ antagonist ONO-A2E-227. The inhibitory effect of PGE₂ and PGE₂-G on the fMLF-induced ROS production was completely prevented by the EP₁/EP₂ receptor antagonist AH-6809 (Fig. 2B). In contrast, the EP₄ antagonist ONO-A2E-227 did not prevent the inhibitory effects of PGE₂ or PGE₂-G. Of note, the EP₂ receptor agonist butaprost mimicked the effects of PGE₂ and PGE₂-G on neutrophil activation as previously described (22), whereas the EP_1/3 agonist sulprostone had no effect (data not shown), supporting the involvement of the EP₂ receptor in the effect we observed. Finally, we confirmed that like those of PGE₂, the inhibitory effects of PGE₂-G were prevented by the cAMP-dependent protein kinase (PKA) inhibitor H-89, underscoring similar downstream signaling events for both lipids (Fig. 2C).

In contrast to PGE₂, which binds to all EP receptors, PGE₂-G only binds to the EP₁, EP₃, and EP₄ receptors (7). Given that PGE₂-G inhibits neutrophil functions in an EP₂-dependent manner, we investigated whether PGE₂-G was hydrolyzed into PGE₂ by human neutrophils. As shown in Fig. 3A, the incubation of human neutrophils with 300 nM PGE₂-G resulted in a time-dependent decrease in its levels (half-life of ∼60 min) and a concomitant buildup of PGE₂. To confirm that this PGE₂ buildup originated from PGE₂-G hydrolysis rather than de novo biosynthesis, we incubated neutrophils with D₄-PGE₂-G. We observed a time-dependent increase of D₄-PGE₂ levels, whereas those of PGE₂ did not change (Fig. 3B), demonstrating that PGE₂-G is hydrolyzed into PGE₂ by neutrophils. In agreement with the hydrolysis of PGE₂-G into PGE₂ in our neutrophil suspensions, the inhibitory effect of 300 nM PGE₂-G occurred over time, and reached its maximal effect after 5 min of preincubation (Fig. 3C), which corresponds to the buildup of midnanomolar range concentrations of PGE₂ (Fig. 3A). In contrast, PGE₂ and the EP₂ receptor agonist butaprost did not require any preincubation time to inhibit the AA-induced LT biosynthesis (Fig. 3C). Altogether, this suggests that PGE₂-G is hydrolyzed into PGE₂ and that PGE₂ mediates the inhibitory effects of PGE₂-G on human neutrophil functions. Of note, PGE₂, PGE₂-G, and PGE₂-EA did not modulate the expression of COX-2 or mPGES-1, the main proteins involved in PGE₂ biosynthesis in neutrophils [data not shown (23)].

PG-Gs can by hydrolyzed to some extent by several lipases that are more or less sensitive to lipase inhibitors (Table I). In this respect, by qPCR and immunoblot we assessed which of these hydrolases are expressed by freshly isolated neutrophils. We were able to detect, by qPCR, the mRNA for all of these lipases, with the exception of CES1. In contrast, our immunoblot data, which includes a positive control for each target, indicates that only ABHD12 and ABHD16A are expressed at a level that allows their detection with this technique (Fig. 4). As for LYPLA2, we found mRNA levels comparable to those of the documented positive control, the MDA-231 cell line (24). However, we were unable to detect the protein in our neutrophil lysates, despite getting a strong band for MDA-231 cells. Altogether, these experiments underscore that the expression of PG-G–hydrolyzing lipases is limited to ABHD12 and ABHD16A in neutrophils.

In an attempt to pinpoint whether ABHD12 and ABHD16A are involved in the hydrolysis of PGE₂-G into PGE₂, we next undertook experiments using numerous hydrolase inhibitors and assessed if they could prevent the inhibitory effect of PGE₂-G on human neutrophils. Knowing that neutrophils display a strong 2-AG–hydrolyzing activity that is involved in the 2-AG–induced LTB₄ biosynthesis (25), we also assessed the impact of these inhibitors on that biosynthetic pathway, to compare the pharmacological profiles of both hydrolytic reactions. Palmostatin B, JZL184, and MAFP were the only compounds we tested that completely inhibited the 2-AG–induced LTB₄ biosynthesis in a concentration-dependent fashion (Fig. 5A, 5B). In contrast, although the effect PGE₂-G was sensitive to MAFP and palmostatin B, it was only partially sensitive to WWL113 and JZL184 at higher concentrations (Fig. 5C). We obtained similar results when we attempted to prevent the inhibitory effect of PGE₂-G on thapsigargin-induced LTB₄ biosynthesis (data not shown). In addition to showing these effects on neutrophil functions, we sought to confirm that the inhibitors indeed prevented PGE₂-G hydrolysis into PGE₂. We thus incubated neutrophils with PGE₂-G for 10 min, in the presence of 10 μM MAFP, palmostatin B, MAFP, or WWL113 and measured the accumulation of PGE₂ in supernatants by LC-MS/MS (Fig. 5E). We found that MAFP, JZL184, and palmostatin B almost completely blocked the buildup of PGE₂, whereas WWL113 had a partial, statistically significant inhibitory effect (p < 0.001). Of note, none of the inhibitors prevented the inhibitory effects of PGE₂ on the fMLF-induced ROS production or the AA-induced LTB₄ biosynthesis (data not shown).

Finally, we investigated if the inhibitory effect of PGE₂-G is mimicked by other PG-Gs. We found that like PGE₂-G, PGD₂-G also inhibits the AA-induced LT biosynthesis as well as the fMLF-induced ROS production, whereas PGF_2α-G was ineffective (Fig. 6). Importantly, the structurally similar but non hydrolysable PGE₂-SA and PGD₂-SA had no effect, again underscoring the importance of PG-G hydrolysis for their inhibitory effect to occur.

Endocannabinoids have been classified as anti-inflammatory lipids, mainly because of the proinflammatory state that CB₁/CB₂ deficient mice usually display in experimental models of inflammatory disease (26, 27). However, arachidonoyl-ethanolamide and 2-AG do not only modulate leukocyte functions by activating the CB receptors, but also through their numerous metabolites, notably eicosanoids, prostamides, and PG-Gs (26, 28). The cellular and molecular mechanisms involved in the immunomodulatory effects of prostamides and PG-Gs remain ill defined. In this study, we provide evidence that: 1) the 2-AG metabolites PGE₂-G and PGD₂-G inhibit human neutrophil functions; 2) the effects of PGE₂-G and PGD₂-G require their hydrolysis into PGE₂ and PGD₂, respectively; 3) the PGE₂-G hydrolases ABHD12 and ABHD16A were detected by qPCR and immunoblot in human neutrophils; 4) the hydrolysis of PGE₂-G into PGE₂ likely involves more than one hydrolase; and 5) the effects of PGE₂-G are blocked by the EP₂ receptor antagonist AH-6809 and the PKA inhibitor H-89.

A limited number of studies have previously evaluated the bioactivity of COX-2 metabolites of endocannabinoids. Hu et al. (29) observed that in rats, PGE₂-G and PGE₂ had similar effects on NF-κB activity, mechanical allodynia and thermal hyperalgesia. Interestingly, the mixture of EP receptor antagonists they used completely blocked the effects of PGE₂, but only partially blocked the effects of PGE₂-G, suggesting different mechanisms of action. In this regard, Nirodi et al. (7) observed that PGE₂-G (but not PGE₂) induces a quick, dose-dependent Ca²⁺ mobilization in RAW264.7 cells. These results indicate that PGE₂ and PGE₂-G can exert either similar or opposite effects, a phenomenon that is likely attributable to differential patterns of EP_1–4 and PG-G receptor expression and activation.

Our data show that PGE₂-G, but not PGE₂-EA, inhibits every neutrophil function we tested in a similar fashion to PGE₂ (Fig. 1). The EP₂ antagonist AH-6809 (10 μM) and the PKA inhibitor H-89 prevented the inhibitory effects of both PGE₂ and PGE₂-G, suggesting the involvement of EP₂ and PKA (Fig. 2). Given that PGE₂-G has some binding affinity for EP₄ but practically none for EP₂ (7), this indicates that the EP₂-dependent effects we observed are not caused by PGE₂-G but are rather the result of PGE₂-G hydrolysis into PGE₂. This is supported by the facts that: 1) PGE₂-G is hydrolyzed into PGE₂ by neutrophils (Fig. 2); 2) the non-hydrolysable version of PGE₂-G, PGE₂-SA, does not inhibit human neutrophils (Fig. 5); and 3) the effect of PGE₂-G is mediated by the only EP receptor not activated by PGE₂-G (Fig. 2).

PGE₂-EA and PGE₂-G are relatively stable in biological systems. Indeed, in human plasma, PGE₂-EA undergoes slow dehydration/isomerization into PGB₂-EA, whereas PGE₂-G is hydrolyzed with a half-life of over 10 min (30). However, the enzymatic pathways involved in the hydrolysis of PGE₂-G into PGE₂ remain unclear. Of the seven candidate lipases that were reported to hydrolyze PGE₂-G (Table I), six were detected by qPCR, but only two were detected by qPCR and immunoblot, namely ABHD12 and ABHD16A (Fig. 4). Of note, we were unable to detect ABHD6, MAG lipase, PPT1, and LYPLA2 in neutrophils by immunoblot, despite the presence of mRNA. Given that our experiments included documented positive controls that did yield a band at the expected size, we can conclude that the Abs we used are reliable in this setting. It is possible, however, that our failure to detect these lipases at the protein level was caused by artifacts during cell lysis and protein denaturation, even though we tried three different lysis methods (sonication, immediate solubilization in boiling Laemmli buffer, and hypotonic lysis), which yielded similar results. Although we also found LYPLA2 mRNA in cells that contaminate our neutrophil suspensions (eosinophils and monocytes), we could not detect any protein from those two cell types either (data not shown). This raises the interesting possibility that perhaps a miRNA or another posttranscriptional modification prevents the translation of the LYPLA2 mRNA we detected.

We also used numerous inhibitors (summarized in Table I) to establish the pharmacological profile linked to the hydrolase(s) involved. MAFP, palmostatin B, and JZL184 were potent inhibitors of 2-AG hydrolysis with IC₅₀ values in the midnanomolar range, as assessed by 2-AG-induced LTB₄ biosynthesis (25). Furthermore, WWL70, WWL113, tetrahydrolipstatin, and ML349 did not prevent the hydrolysis of 2-AG. In contrast, the inhibitory effect of PGE₂-G was completely prevented by MAFP and palmostatin B, and partially prevented by micromolar concentrations of JZL184 and WWL113. WWL70, tetrahydrolipstatin, and ML349 did not have any effect. This suggests that the lipases involved in 2-AG and PGE₂-G hydrolysis are likely to be different. In addition, none of the documented PGE₂-G hydrolases match the pharmacological profile we found. Thus, the data we gathered did not allow us to confirm that one enzyme among our candidates (Table I) is responsible for the hydrolysis of PGE₂-G by human neutrophils. Therefore, it is likely that the PGE₂-G hydrolytic activity that we observed is catalyzed by more than one lipase, and/or by a lipase that has yet to be characterized.

A recent study showed that the inhibition of 2-AG hydrolysis led to decreased inflammation in mice (15). This was due, at least in part, to the endocannabinoid metabolite PGD₂-G, which prevented the production of IL-1β. Of note, PGE₂-G and PGF_2α-G did the opposite. Thus, it is possible that PG-Gs regulate inflammation through multiple mechanisms of actions and thus participate in the inflammatory process in a coordinated and timely manner. Our data indicate that human neutrophils have the ability to hydrolyze PG-Gs, notably PGE₂-G and PGD₂-G. Although our data support that PG-G hydrolysis leads to PG production and dampens neutrophil functions, it remains possible that neutrophils prevent the anti-inflammatory effects of PG-Gs by eliminating those putative proresolving mediators in vivo.

In conclusion, we provide clear evidence that PGE₂-G, but not PGE₂-EA, inhibits numerous functions of human neutrophils in a concentration-dependent manner similar to PGE₂. This inhibitory effect requires the hydrolysis of PGE₂-G into PGE₂, the subsequent activation of the EP₂ receptor and PKA. These results also support the view that the anti-inflammatory effects related to the endocannabinoid 2-AG might be due, at least in part, to its metabolism by the COX pathway.

We thank Dr. Lawrence Marnett for generously providing the selective LYPLA2 inhibitor ML349.

The authors have no financial conflicts of interest.