Anandamide-derived Prostamide F_2α Negatively Regulates Adipogenesis

Background: Adipogenesis is the process by which adipocytes are formed to maintain or expand fat depots.

Results: Prostaglandin F_2α ethanolamide (PGF_2αEA) is produced from anandamide in preadipocytes and inhibits adipogenesis.

Conclusion: Conversion of proadipogenic anandamide to antiadipogenic PGF_2αEA is a novel mechanism for the regulation of adipogenesis.

Significance: Discovering a PGF_2αEA-mediated negative regulatory mechanism over adipogenesis may lead to the development of antiobesity therapies.

Abstract

Lipid mediators variedly affect adipocyte differentiation. Anandamide stimulates adipogenesis via CB₁ receptors and peroxisome proliferator-activated receptor γ. Anandamide may be converted by PTGS2 (COX2) and prostaglandin F synthases, such as prostamide/prostaglandin F synthase, to prostaglandin F_2α ethanolamide (PGF_2αEA), of which bimatoprost is a potent synthetic analog. PGF_2αEA/bimatoprost act via prostaglandin F_2αFP receptor/FP alt4 splicing variant heterodimers. We investigated whether prostamide signaling occurs in preadipocytes and controls adipogenesis. Exposure of mouse 3T3-L1 or human preadipocytes to PGF_2αEA/bimatoprost during early differentiation inhibits adipogenesis. PGF_2αEA is produced from anandamide in preadipocytes and much less so in differentiating adipocytes, which express much less PTGS2, FP, and its alt4 splicing variant. Selective antagonism of PGF_2αEA receptors counteracts prostamide effects on adipogenesis, as does inhibition of ERK1/2 phosphorylation. Selective inhibition of PGF_2αEA versus prostaglandin F_2α biosynthesis accelerates adipogenesis. PGF_2αEA levels are reduced in the white adipose tissue of high fat diet-fed mice where there is a high requirement for new adipocytes. Prostamides also inhibit zebrafish larval adipogenesis in vivo. We propose that prostamide signaling in preadipocytes is a novel anandamide-derived antiadipogenic mechanism.

Introduction

Adipocytes, the main components of white adipose tissue, serve as energy stores through the accumulation and storage of triglycerides, which are metabolized upon an increased requirement for higher energy levels (1) and produce an array of factors (adipokines) with physiological roles ranging from immunological responses to the regulation of appetite (2, 3). Excessive adiposity (especially in central depots) is a leading factor for a host of complications, altogether termed metabolic syndrome, that not only lead to a decreased quality of life but may also result in death (4). Adipose tissue mass is determined by the amount of fat stored in each adipocyte as well as the number of adipocytes, both of which can be modified through the interplay between energy input and expenditure (5). Several species are able to produce new fat cells via adipogenesis throughout their entire lifetime, and this process can be increased through high fat or carbohydrate consumption (6). Increased adipocyte number, both during development and in adulthood, is associated with high levels of obesity (6–8). The understanding of the genetic and molecular processes regulating adipogenesis will, therefore, provide insight into the pathological dysregulation of adipose tissue accumulation and aid in the development of therapeutic strategies for the prevention and/or treatment of obesity. The key players in the transcriptional program that regulates adipogenesis have been elucidated. Pparg and Cebpa are key adipogenic regulators across several species, and their expression is up-regulated within the first 2 days of the differentiation process (9–11). Specific overexpression or activation of these proteins is able to promote the differentiation of fibroblasts into adipocytes, and they are critical in the control of adipocyte-specific genes such as Fabp4/AP1, Retn, and Acrp30 (12–14).

The endocannabinoid N-arachidonoylethanolamine (AEA,⁴ anandamide) is a lipid signaling molecule derived from arachidonic acid that is important for adipocyte biology by regulating the adipogenic process and lipid metabolism within mature adipocytes. AEA levels rise and then stabilize or decrease with the progression of adipogenesis (15) and are increased in the plasma and adipose tissue under conditions of central adiposity/obesity (16, 17). AEA stimulates adipogenesis through the G protein-coupled cannabinoid CB₁ receptor or, at higher concentrations, PPARγ (18, 19). In some cells, AEA, instead of being inactivated by fatty acid amide hydrolase (FAAH) (20), can be oxygenated by prostaglandin-endoperoxide synthase 2 (PTGS2) (also known as COX 2), and this reaction, after the subsequent action of either prostamide/prostaglandin F synthase or prostaglandin F synthase (AKR1C3/prostaglandin F synthase), results in the formation of the prostanoid prostaglandin F_2α ethanolamide (prostamide F_2α or PGF_2αEA). These enzymes (as well as COX1, which, however, does not recognize AEA as substrate) are primarily responsible for the conversion of cell membrane-derived arachidonic acid into prostaglandin F_2α (PGF_2α) (21–23). However, PGF_2αEA and PGF_2α, apart from deriving from different biosynthetic pathways, also display distinct pharmacology. They are sensitive to specific antagonists, and this is a result of the fact that the latter mediator acts via the G protein-coupled FP receptor, whereas PGF_2αEA requires for its actions a heterodimer consisting of wild-type and an alternately spliced FP (24–28). Like endocannabinoids, PGF_2α and other prostaglandins also play numerous physiological roles, including the regulation of adipogenesis, generally, but not uniquely, through the activation of G protein-coupled receptors (29–31). Prostaglandin J₂ was the first endogenous PPARγ ligand identified capable of inducing this process (32, 33), whereas PGF_2α inhibits adipocyte differentiation. This activity is dependent on FP activation of the ERK1/2 and calcium-calcineurin signaling pathways, with the former being responsible for the phosphorylation and inactivation of PPARγ (34–36).

To date, it is unknown what, if any, effects PGF_2αEA has on the process of adipogenesis, nor whether this putative lipid mediator is present in differentiating adipocytes. Interestingly, however, bimatoprost (AGN 192024), a stable, synthetic, and potent analog of PGF_2αEA approved for the treatment of ocular hypertension and also capable of activating the FP/FP splice variant heterodimers (28, 37), decreases periorbital fat deposits in glaucoma patients through an unknown mechanism (38–41). This observation may suggest that bimatoprost and, by extension, PGF_2αEA inhibit adipogenesis, perhaps through the stimulation of the aforementioned heterodimers. We have investigated this possibility in vitro and in vivo and report that PGF_2αEA is a potent anandamide-derived antiadipogenic mediator in preadipocytes.

EXPERIMENTAL PROCEDURES

Cell Culture

3T3-L1 cells (ATCC) were cultured in growth medium (DMEM, Lonza) supplemented with NaHCO₃ (1.5 g/liter), 10% NCS (Invitrogen) and penicillin/streptomycin (Invitrogen). To induce adipogenesis, cells were grown to confluence and, after 1 day, switched to differentiation medium (IDM; growth medium plus 1 μg/ml insulin (Sigma), 250 nm dexamethasone (Sigma), and 500 μm 3-isobutyl-1-methylxanthine (Sigma)) for 2 days, followed by incubation in growth medium with 1 μg/ml insulin for 2 days, and then cultured in growth medium for the rest of the experiment with medium changes every 2 days. Human subcutaneous preadipocytes were obtained from Lonza (catalog no. PT-5020) and were grown and differentiated in the recommended medium (Lonza, catalog no. PT-8002) as described by the manufacturer. The drugs bimatoprost and AGN211335 (Allergan Inc., Irvine, CA), AL8810, PD98059, and R-flurbiprofen (Cayman Chemicals) were added to the medium, and cells were harvested at various times as described below.

Quantitative and Non-quantitative PCR Analysis

Total RNA was isolated from 3T3-L1 cells cultured in 24-well plates using TRIzol (Invitrogen) and treated with DNase I (Ambion) and from human subcutaneous preadipocytes cultured in 96-well plates using TRIzol and Purelink RNA micro scale kits (Invitrogen) and reverse-transcribed with the SuperScript III RT reaction kit (Invitrogen) according to the instructions of the manufacturer. 10 to 20 ng of starting RNA was then used for QPCR analysis using IQ SYBR Green Supermix (Bio-Rad) with primers designed with AlleleID (Premier Biosoft, supplemental Table S1) on a CFX 384 optical thermal cycler (Bio-Rad). Data analysis was performed using CFX Manager software (Bio-Rad) using either Hprt (for 3T3-L1 cells) or actin β (for human cells) as reference genes, and data are expressed as relative mRNA levels with standard errors of the mean of triplicate reactions. Statistical significance was determined with the REST 2009 software. For analysis of PTGFR splice variant expression in undifferentiated and differentiated human preadipocytes, primer combinations for each splice variant were designed by hand (supplemental Table S2) on the basis of sequences available in patent no. US 7,320,871 B2, used to amplify cDNA obtained as described above with Platinum TAQ (Invitrogen) using a C1000 thermal cycler (Bio-Rad), and visualized by agarose gel electrophoresis.

Mouse Feeding and Purification and Quantification of AEA, PGF2αEA, and PGF2α from Adipose Tissues and Cells

Nine-week-old male C57Bl/6J mice (Charles River Laboratories, Inc.) were utilized in all studies. For standard-fat diet (SFD) versus high-fat diet (HFD) studies mice were fed for 1 week with either standard chow or an HFD (Harlan, catalog no. 97366). After treatment, animals were sacrificed, and adipose tissues were removed, snap-frozen, and stored at −80 °C. 3T3-L1 cells were grown in 100-mm plates as described above, washed once in ice-cold PBS, collected, and stored at −80 °C.

For AEA or PGF_2aEA, tissues or cells were Dounce-homogenized and extracted with acetone containing internal deuterated standards for AEA and PGF_2aEA quantification by isotope dilution ([²H]₈ AEA and [²H]₄ PGF_2αEA). The lipid-containing organic phase was dried, weighed, and prepurified by open-bed chromatography on silica gel. Fractions were obtained by eluting the column with 99:1, 90:10, 70:30, and 50:50 (v/v) chloroform/methanol. The 90:10 fraction was used for AEA quantification by LC-APCI-MS and using selected ion monitoring at M+1 values for AEA and its deuterated homologue, as described previously (42). The 70:30 fraction was used for PGF_2aEA quantification by LC-MS-MS using an LC20AB coupled to a hybrid detector IT-TOF (Shimadzu Corp., Kyoto, Japan) equipped with an electrospray ionization (ESI) interface. LC analysis was performed in the isocratic mode using a Discovery HC18 column (15 cm, 62.1 mm, 5 mm) and methanol/water/acetic acid (53:47:0.05 by volume) as mobile phase with a flow rate of 0.15 ml/min. Identification of PFG_2aEA was carried out using ESI ionization in the positive mode with a nebulizing gas flow of 1.5 ml/min and a curved desolvation line temperature of 250 °C.

For PGF_2α, solid-phase extraction cartridges (CHROMABOND® HR-X, 3 ml/200 mg, Macherey-Nagel) were used for extraction. For cellular assays, 2 ml of culture medium was utilized. For tissue extraction, epididymal adipose tissue was weighed, homogenized with pestles in 800 μl of PBS, and centrifuged at 15,000 × g for 15 min at 4 °C. 10 μl of ²[H]₄ PGF_2α (2.5 μm) internal standard was added to the supernatant and the pH was adjusted to 3 with 1 m HCl. Media or tissue samples were then added to a solid-phase extraction cartridge that was preconditioned with 2 ml of methanol, 2 ml of water, and 2 ml of 95:5 water:methanol. PGF_2α was eluted with 3 ml of methanol under vacuum, whereas most (∼95%) of the PGF_2αEA was retained on the column. The eluate was dried under a nitrogen gas stream at 25 °C, and the residue was reconstituted in 30 μl of mobile phase for tissue samples or 2 ml of medium for cellular assays. The recovery of the extraction protocol was between 20 and 50%.

For PGF_2α analysis in tissue, an LC20AB coupled to a hybrid detector IT-TOF (Shimadzu Corp.) equipped with an ESI interface was used, and they were measured via multiple reaction monitoring. LC analysis was performed using a Discovery® C18 column (15 cm × 2.1 mm, 5 μm) at a flow rate of 200 μl/min. The column was equilibrated in solvent A (water-acetonitrile-acetic acid (70:30:0.02, v/v/v)), and 10 μl of sample (100% of medium and 33% of tissue) was injected using a 10-μl injection loop and eluted with 0% solvent B (acetonitrile-isopropyl alcohol (50:50, v/v)) between 0 and 1 min. Solvent B was increased in a linear gradient to 25% solvent B until 3 min, to 45% until 11 min, to 60% until 13 min, to 75% until 18 min, and to 90% until 18.5 min. Solvent B was held at 90% until 20 min, dropped to 0% by 21 min, and held until 25 min (43). Identification of PGF_2α was carried out using ESI ionization in the negative mode with a nebulizing gas flow of 1.5 ml/min and a curved desolvation line temperature of 250 °C. Quantitative prostaglandin determination was performed by the stable isotope dilution method as described previously (44). The limit of detection of this method was 100 fmol/sample. Because of sensitivity issues, PGF_2α analysis in solid-phase extraction cartridge-fractionated cell medium (unlike prostamides, which, being less polar, are retained by cells, prostaglandins are known to be largely released from cells into the culture medium), was instead performed with a PGF_2αEIA kit (Cayman Chemicals) according to the instructions of the manufacturer.

Western Blot Analysis

Cells grown in 6-well plates were differentiated as described above and lysed in 1× TNE buffer (50 mm Tris (pH 7.4), 150 mm NaCl, 1 mm EDTA) with 0.5% Triton X-100 and protease and phosphatase inhibitor mixtures (Sigma, catalog nos. P8340, P5726, and P0044). Proteins were analyzed via a Lowry protein assay (Bio-Rad) and resolved by SDS-PAGE with a routine Tris-glycine buffering system. Proteins transferred onto PVDF membranes were then blocked in 5% skim milk in TBST (20 mm Tris, 137 mm NaCl, 0.1% Tween 20) and probed with anti phospho-ERK, ERK (1:1000, Cell Signaling Technology, catalog nos. 9101 and 9102), PTGS2/Cox2 (1:200, Santa Cruz Biotechnology, catalog no. sc-1747), or actin β (1:30000, Sigma, catalog no. A1978) overnight, followed by HRP-conjugated secondary antibodies (Bio-Rad), and then detected by ECL (Bio-Rad).

Zebrafish Larvae and Adipocyte Staining

Zebrafish (Danio rerio) larvae were obtained from a stable laboratory strain and raised at 28.5 °C on a 14 light:10 dark photoperiod according to the Zebrafish Book. Embryos were raised in filtered egg water until 8 days post-fertilization, and larvae were exposed to drug solutions for a further 8 days. Bimatoprost was dissolved in DMSO, diluted with filtered egg water, and replaced daily. Larvae were then stained with Nile red (Sigma) to a final concentration of 0.5 μg/ml for 30 min. Stained larvae were anesthetized with Tricaine (Sigma), photographed with a Tri-Red filter on a Leica DMI 6000B fluorescence microscope under the same optical conditions, and the signal intensity of adipocytes within the abdominal cavity was analyzed with ImageJ 1.44p (National Institutes of Health). Staining of cultured 3T3-L1 adipocytes was performed using Adipored (Lonza) according to the instructions of the manufacturer and then either photographed as with zebrafish embryos or analyzed with a Genios Pro plate reader (Tecan) to quantify fluorescence intensity.

Ethics Statement

All animal procedures were in conformity with the principles of laboratory animal care (National Institutes of Health publication no. 86-23, revised 1985) and the Italian Decreto Legge no. 116 of 27 January 1992 and associated guidelines in the European Communities Council Directive of 24 November 1986 (86/609/ECC). All efforts were made to minimize animal suffering and the number of animals used.

RESULTS

Prostamides Inhibit Early 3T3-L1 Cell Adipogenesis

Given the well documented effects of prostaglandins on adipogenesis, we set out to determine whether bimatoprost and PGF_2αEA affected the early stages of adipogenesis in vitro. We differentiated 3T3-L1 cells in the presence of various concentrations of either bimatoprost, PGF_2αEA, or PGF_2α and assayed the expression of Pparg and Cebpa after 2 days. All compounds dose-dependently reduced the expression of these targets to levels similar to those of undifferentiated cells, although, as expected from previous pharmacological data (37), bimatoprost was approximately 10 times more potent than PGF_2αEA, yielding maximal effects at about 1 μm as compared with 10 μm for PGF_2αEA (Fig. 1A). Long-term differentiation of 3T3-L1 cells for 8 days in the presence of PGF_2αEA or bimatoprost resulted in the sustained inhibition of Pparg expression as well as the inhibition of the up-regulation of the mature adipocyte markers Fabp4, Retn, and Acrp30 (Fig. 1B). These data show that prostamides are antiadipogenic agents.

FIGURE 1.

Dose-dependent effects of prostamides on 3T3-L1 adipogenesis. A, mRNA levels of Pparg and Cebpa were determined by QPCR in undifferentiated (d0) or differentiating (d2) 3T3-L1 cells in the absence (-) or presence of DMSO (D), bimatoprost, PGF_2αEA, or PGF_2α at the indicated concentrations. B, mRNA levels of Pparg, Fabp4, Acrp30, and Retn were determined by QPCR in 3T3-L1 adipocytes after 8 days of differentiation in the absence (-) or presence of DMSO, bimatoprost, or PGF_2αEA at the indicated concentrations. Relative mRNA values are expressed as mean ± S.E. *, p < 0.05 versus DMSO control.

We then determined whether early inhibition of the adipogenic program alone by bimatoprost was sufficient to inhibit the formation of mature adipocytes. We treated differentiating 3T3-L1 cells at different time points for various periods of time and observed that bimatoprost exerts its antiadipogenic activity when the cells were exposed during the first 2 days of the differentiation protocol only (Fig. 2). This indicated that bimatoprost inhibits the initial stages of differentiation by inhibiting the early up-regulation of the key adipogenic factors Pparg and Cebpa. This inhibition was long-lasting because the subsequent up-regulation of the mature adipocyte markers Fabp4, Retn, and Acrp30 was also repressed. However, Fabp4 expression is similarly inhibited at day 8 by bimatoprost regardless of whether the cells are treated for only the first 2 days of or throughout the entire differentiation program, whereas Retn and Acrp30 gene expression is further reduced by 50% if the bimatoprost treatment is extended beyond 2 days. When we stained triglycerides in 3T3-L1 cells at day 8 of differentiation, we confirmed that treatment during the first 2 days of differentiation is sufficient for the inhibition of triglyceride accumulation (Fig. 2B).

FIGURE 2.

Time course analysis of the effects of prostamide treatment on 3T3-L1 adipogenesis. A, mRNA levels of adipogenic (Pparg, Cebpa) and mature adipocyte markers (Fabp4, Retn, Acrp30) in 3T3-L1 preadipocytes (d0) and mature adipocytes (d8) in the absence or presence of 1 μm bimatoprost for various times (B0–8 days) were determined by QPCR. Relative mRNA values are expressed as mean ± S.E. B, quantification (top panel, mean ± S.D. of 64 independent fluorescence readings) and images (bottom panel) of Adipored-stained undifferentiated (Undiff.) 3T3-L1 cells or after 8 days of differentiation treated in the absence (-) or presence of either DMSO, 1 μm bimatoprost (Bim.), 10 μm PGF_2αEA, or 0.1 μm PGF_2α from day 0–2. *, p < 0.05 versus DMSO control.

PGF2αEA Is Produced in 3T3-L1 Preadipocytes, and Its Levels and Those of Its Biosynthetic Enzymes Decrease during Differentiation

Recently, we identified PGF_2αEA as an endogenously occurring eicosanoid/neutral lipid in the spinal cord (45). However, no evidence exists for the presence of PGF_2αEA in preadipocytes or adipocytes. Utilizing LC-IT-TOF mass spectrometry, we measured PGF_2αEA levels in lipid extracts from undifferentiated 3T3-L1 cells or cells undergoing adipogenesis. Undifferentiated 3T3-L1 cells produce significant levels of PGF_2αEA and, by day 2 of differentiation, these levels drop to barely detectable levels (Fig. 3A). This may suggest that PGF_2αEA may act to maintain preadipocytes in an undifferentiated state and that its down-regulation, together with other factors, is required for adipogenesis. Given that AEA is the only biosynthetic precursor of PGF_2αEA via PTGS2/COX2, we hypothesized that PGF_2αEA levels in 3T3-L1 cells could be manipulated by incubation with AEA. Incubation of undifferentiated cells with 10 μm AEA resulted in an almost 4-fold increase in PGF_2αEA levels after 4 h, and coincubation with the PTGS2 inhibitor NS398 (10 μm) reduced PGF_2αEA to undetectable levels (Fig. 3A). By day 2 of differentiation, although no endogenous PGF_2αEA was detected, incubation with AEA resulted in 3-fold lower levels of PGF_2αEA compared with undifferentiated cells (Fig. 3A), indicating a marked decrease in the ability of differentiating cells to convert exogenous AEA to this compound. To surmise how levels of PGF_2αEA levels change during differentiation, we assayed the expression of a number of biosynthetic enzymes during this process. We found that both Ptgs2 (COX-2) and Akr1b3 (which was shown to be required for PGF_2α production in 3T3-L1 adipocytes (46)) decrease significantly by the second day of differentiation as expected, whereas Pmpgfs (encoding for prostamide/prostaglandin F synthase) is expressed in these cells at levels that do not change by this time point (Fig. 3B). These data indicate that PTGS2 (COX2) and AKR1B3 may regulate PGF_2αEA levels in adipocytes. However, because the formation of PGH₂EA from PTGS2 action is the rate-limiting step in PGF_2αEA biosynthesis, we cannot rule out a role for prostamide/prostaglandin F synthase. Taken together, these data support the idea that differentiating adipocytes down-regulate the pathway required for AEA-derived PGF_2αEA biosynthesis. Surprisingly, when we measured Ptgs2 expression in differentiating 3T3-L1 cells treated with bimatoprost or PGF_2αEA, we found that they induced a strong up-regulation of Ptgs2 mRNA (Fig. 3C) and protein (see below) expression to levels above those observed in undifferentiated cells. We also observed that PGF_2α similarly up-regulated Ptgs2 expression.

FIGURE 3.

Regulation of prostamide levels during early adipogenesis in 3T3-L1 cells. A, quantification (mean ± S.E.) of PGF_2αEA levels by LC-APCI-MS in undifferentiated (d0) or differentiating (d2) 3T3-L1 cells alone (left panel) or in the presence of DMSO or 10 μm AEA without or with 10 μm NS398 (NS). Total amounts of lipids were similar for all samples (not shown). Note that ∼1 mg of lipid extract is normally obtained from 2–3 million cells. n = 3 for each point. * and #, p < 0.05 versus DMSO and AEA at d0, respectively. B, C, and F) mRNA levels of Pparg, Ptgs2, Akr1b3, Pmpgfs, or Ptgfr were determined by QPCR in undifferentiated (d0) or differentiating (d2) 3T3-L1 cells in the absence (-) or presence of DMSO, 1 μm bimatoprost (Bim.), 10 μm PGF_2αEA or 0.1 μm PGF_2α. D, mRNA levels of PPARG, FABP4, or ADIPOQ were determined by QPCR in undifferentiated (d0) human preadipocytes or cells after 8 days of differentiation in the absence or presence of DMSO or the indicated concentration of bimatoprost. E, mRNA levels of PPARG, PTGFR (FP receptor), and PTGFR alternate variant 4 (PTGFRv4) were determined by QPCR in undifferentiated (d0) or differentiating (d3) human subcutaneous preadipocytes. Relative mRNA values are expressed as mean ± S.E. *, p < 0.05 versus d0 (B and E) or DMSO control (C, D, and F).

Human Preadipocytes Are Sensitive to the Antiadipogenic Activity of Prostamides and Express PTGFR Alternative Splice Transcripts Required for Their Activity

To determine whether the antiadipogenic activity of bimatoprost also occurs in human cells, we differentiated human preadipocytes in the presence of various doses of bimatoprost and assayed the expression of PPARG and the mature adipocyte markers FABP4 and ADIPOQ. We observed dose-dependent inhibition of the up-regulation of all genes after 8 days of differentiation (Fig. 3D).

PGF_2αEA and bimatoprost signal through prostamide receptors, which are splice variants of the FP receptor encoded by the PTGFR gene responsible for PGF_2α signaling (28). To date, there has been no report of the expression of these variants in adipose tissue. We therefore performed RT-PCR on human premature and mature adipocytes using various primers that detect all the known splice variants of PTGFR. In both undifferentiated and differentiated cells we were able to detect all of the known PTGFR splice variants (supplemental Table S2), and subsequent sequencing confirmed the expression of the PTGFR alternate 4 (alt4) splice variant, reported to dimerize with the wild-type FP receptor and mediate bimatoprost/PGF_2αEA signaling. QPCR analysis of differentiating human adipocytes using primers common to all splice variants or just the alt4 variant (PTGFRv4) showed that these receptors are down-regulated during adipogenesis, indicating that the cells would likely be less sensitive to PGF_2α and PGF_2αEA upon differentiation (Fig. 3E). Searches of genome and expression databases failed to identify a potential murine homolog of the PTGFRv4 splice variant. However, we also assayed the expression of the wild type murine Ptgfr gene during 3T3-L1 differentiation. As observed in the human cells, and in contrast to a previous report, we found that, after 2 days of differentiation, Ptgfr expression decreased significantly (Fig. 3F). Further, similar to Ptgs2 expression, bimatoprost, PGF_2αEA, and PGF_2α all caused a very large increase in Ptgfr expression to levels that were 3- to 4-fold higher than those observed in undifferentiated cells. Taken together, these data indicate that prostamides inhibit adipogenesis in both murine and human preadipocytes, possibly through prostamide receptors/PTGFR splice variants, and reverse the down-regulation of prostamide/prostaglandin biosynthesis and action within differentiating adipocytes, thereby creating a self-amplificatory loop for their action.

Prostamide Receptor Antagonism Reverses Prostamide-mediated Inhibition of Adipogenesis

We next tested the ability of bimatoprost to inhibit adipogenesis in 3T3-L1 preadipoctyes in the absence or presence of the prostamide receptor antagonist AGN211335, which specifically blocks bimatoprost/prostamide activity at 1 μm in HEK293 cells and not PGF_2α activity (28), or the prostaglandin FP receptor antagonist AL8810. We found that AGN211335 (5 μm) reversed bimatoprost-mediated (0.1 μm) inhibition of Pparg and Cebpa gene induction in cells differentiated for 2 days back to control levels, whereas AL8810 (5 μm) was much less, if at all, effective (Fig. 4A). Further, bimatoprost-mediated Ptgs2 induction was completely reversed by AGN211335 but only reduced by half by AL8810 (Fig. 4B). In human preadipocytes, we found that bimatoprost-mediated (0.5 μm) inhibition of FABP4 and ADIPOQ expression was completely reversed by the prostamide receptor antagonist (Fig. 4C). These data suggest that the antiadipogenic activity of prostamides is mediated primarily by the prostamide receptors in both murine and human cells.

FIGURE 4.

Effect of prostamide receptor antagonism on prostamide antiadipogenic activity. A and B, mRNA levels of Pparg, Cebpa, and Ptgs2 were determined by QPCR in differentiating 3T3-L1 cells at day 2 in the presence of DMSO, 0.1 μm bimatoprost, 5 μm AGN211335 (AGN), or 5 μm AL8810 (AL) as indicated. C, mRNA levels of FABP4 and ADIPOQ were determined by QPCR in human preadipocytes undifferentiated (Undif.) or differentiated (d4) in the absence or presence of DMSO, 0.5 μm bimatoprost, or 5 μm AGN211335 as indicated. Relative mRNA values are expressed as mean ± S.E. * and #, p < 0.05 versus DMSO control and bimatoprost-cotreated samples, respectively.

Activation of MAPK Signaling Is Required for Prostamide Antiadipogenic Activity

As activation of the FP receptor by PGF_2α activates MAPK signaling and ERK phosphorylation/activation, which negatively regulates adipogenesis, we next determined whether the antiadipogenic activity of bimatoprost and PGF_2αEA required the activation of MAPK signaling in 3T3-L1 cells through the use of the MEK inhibitor PD98059. PD98059 alone resulted in a small up-regulation of Pparg and Cebpa expression after 2 days of differentiation per se and completely abolished the bimatoprost- and PGF_2αEA-mediated inhibition of the expression of these genes (Fig. 5A). Interestingly, the same concentration of PD98059 was only able to partially revert the PGF_2α-mediated inhibition of Cebpa expression while fully restoring Pparg expression. As above (Fig. 2B), bimatoprost, PGF_2αEA or PGF_2α treatment for the first 2 days of differentiation results in significantly decreased accumulation of triglyceride levels at day 8, completely reversed by coincubation with PD98059 (Fig. 5C). We then tested the requirement of MAPK signaling for the bimatoprost-mediated inhibition of adipogenesis in primary human preadipocytes. As with 3T3-L1 cells, the ability of bimatoprost to inhibit adipogenesis, as measured by expression of the mature adipocyte markers FABP4 or ADIPOQ, was completely dependent on MAPK signaling (Fig. 5B).

FIGURE 5.

Effect of MEK inhibition on prostamide antiadipogenic activity. A and D, mRNA levels of Pparg, Cebpa, or Ptgs2 were determined by QPCR in differentiating (d2) 3T3-L1 cells in the absence or presence of DMSO, 1 μm bimatoprost (Bim.), 10 μm PGF_2αEA, 0.1 μm PGF_2α, and/or 25 μm PD98059 (PD) as indicated. B, mRNA levels of FABP4 and ADIPOQ were determined by QPCR in human preadipocytes undifferentiated (Undif.) or differentiated (d4) in the absence or presence of DMSO (D), 1 μm bimatoprost, and/or 25 μm PD98059 as indicated. Relative mRNA values are expressed as mean ± S.E. C, quantification (left panel, mean ± S.D. of 16 independent fluorescence readings) and images (right panel) of Adipored-stained 3T3-L1 cells after 8 days of differentiation treated in the absence (-) or presence of either DMSO, 1 μm bimatoprost, 10 μm PGF_2αEA, or 0.1 μm PGF_2α from day 0–2. E, Western blot analysis of phospho-ERK, ERK, PTGS2, and actin β (ACTB) in lysates from undifferentiated (d0) or differentiating (d2) 3T3-L1 cells in the absence (-) or presence (+) of DMSO, 1 μm bimatoprost, 10 μm PGF_2αEA, 0.1 μm PGF_2α, or 25 μm PD98059 in the indicated combinations. * and #, p < 0.05 versus DMSO control and bimatoprost or PGF_2αEA or PGF_2α-cotreated samples, respectively.

Given that bimatoprost and PGF_2αEA induce a large up-regulation of Ptgs2 in differentiating 3T3-L1 cells at day 2, we tested whether this was also dependent on MAPK signaling. PD98059 was able to significantly reverse the up-regulation of Ptgs2 by bimatoprost, PGF_2αEA, and PGF_2α (Fig. 5D). Western blot analysis confirmed that treatment of 3T3-L1 cells with bimatoprost or PGF_2αEA during the first 2 days of differentiation increased ERK phosphorylation to levels similar to those observed in undifferentiated 3T3-L1 cells and that both compounds similarly result in the up-regulation of PTGS2 protein levels in a MEK-dependent manner (Fig. 5E). Taken together, these data indicate that bimatoprost and PGF_2αEA signal through the MAPK pathway in preadipocytes to inhibit adipogenesis and up-regulate PTGS2 expression.

Bimatoprost Inhibits the Early Development of Adipose Tissue in D. rerio in Vivo

Fish rely on the same transcriptional cascade to regulate adipogenesis and lipid metabolism as mammals (47, 48). In zebrafish (D. rerio), adipogenesis initiates in larvae not sooner than 8 days post-fertilization. Adipocytes form as clusters of cells initially located along the pancreas and posterior to the swim bladder along the peritoneal cavity (47). This process is extremely labile and dependent not only on the age but the size and nutritive state of the fish. As such, food deprivation inhibits the early development of adipocytes (48). To determine whether bimatoprost was able to inhibit fat formation in vivo, we utilized zebrafish as a developmental model of adipogenesis. Although bimatoprost-treated fish had identifiable adipocytes, the adipose mass was not only smaller but much less intensely stained as well, indicating lower levels of triglyceride accumulation (Fig. 6A, arrows). Accordingly, quantification of pixel intensity within the abdominal region where adipocytes developed showed that bimatoprost resulted in significantly less average fluorescence in this region (Fig. 6B). Staining of the gall bladder was observed periodically, as reported previously. No toxicity was observed within the fish throughout the course of the experiment, nor were there any observable differences in size between control and treated specimens. Control unfed fish were found to not have any positively stained adipocytes (Fig. 6A).

FIGURE 6.

Specific role of prostamides in adipogenesis in vitro and in vivo. A, zebrafish larvae were treated from 8 to 16 days post-fertilization with DMSO or bimatoprost (Bim.) and stained with Adipored in the final 24 h of treatment before analysis by fluorescent microscopy. Anterior is left, dorsal is up. Arrow, adipocytes; asterisk, swim bladder; g, gall bladder. B, quantification of pixel intensity from stained adipocytes in the abdominal cavity. n = 16 for each treatment. *, p < 8⁻⁷. C, quantification (mean ± S.E.) of PGF_2αEA (n = 6) and PGF_2α (n = 3) levels by LC-APCI-MS in epididymal fat of mice on an SFD or HFD for 1 week for each treatment. *, p < 1⁻⁶. D, quantification (mean ± S.E.) of PGF_2αEA by LC-APCI-MS in undifferentiated 3T3-L1 cells in the presence of DMSO or 10 μm AEA without or with 10 μm R-flurbiprofen (RF) for 4 h. Total amounts of lipids were similar for all samples (not shown). n = 3 for each point. * and #, p < 0.05 versus DMSO and AEA, respectively. E, quantification (mean ± S.D.) of PGF_2α levels by EIA in medium from undifferentiated (d0) or differentiating (d2) 3T3-L1 cells in the absence or presence of 100 μm AA and 10 μm R-flurbiprofen alone or in combination for 4 h. n = 3 for each point. * and #, p < 0.05 versus DMSO of the same day and similar treatment at d0, respectively. Note that 10 pg of PGF_2α corresponds to ∼28.5 fmol and that 2 ml of medium from 2–3 million cells were used for each data point. Therefore, the base-line amounts of PGF_2α released from 2–3 million cells (i.e. ∼57 fmol) was below the limit of detection of our LC-IT-TOF MS method. Hence, the use of EIA. F, QPCR quantification of Pparg, Cebpa (both at day 1), Acrp30 (at day 3), and Fabp4 (at day 4) mRNA levels in differentiating 3T3-L1 cells (relative to undifferentiated controls) treated with either DMSO (D) or the indicated concentration of R-flurbiprofen for 24 h prior to the onset of the differentiation protocol. *, p < 0.05 versus DMSO control. G, quantification (mean ± S.D.) of Adipored-stained 3T3-L1 cells after 4 days of differentiation as in F. n = 6. *, p < 0.05 versus DMSO control.

PGF2αEA Levels Are Reduced in WAT of Mice under Dietary Conditions That Favor Adipocyte Mass Expansion

Exposure to an HFD is able to induce adipose hyperplasia (6, 49, 50). We therefore tested whether, as compared with an ad libitum SFD, PGF_2αEA levels were altered in the WAT of mice on a short (5-day) period of ad libitum HFD, insufficient to increase body weight more than the SFD. We were able to detect levels of PGF_2αEA in the epididymal WAT of SFD mice. However these were reduced to undetectable levels in HFD mice (Fig. 6C). By contrast, epididymal WAT PGF_2α levels_, which were lower than PGF_2αEA levels, were not reduced following the short period of HFD (Fig. 6C). These data suggest that PGF_2αEA is down-regulated in adipose tissue under dietary conditions that favor adipocyte mass expansion, implicating PGF_2αEA as a novel, endogenous regulator of adipogenesis and bimatoprost as a potential tool for studying this process.

Selective Inhibition of Prostamide versus Prostaglandin F_2α Biosynthesis Enhances Adipogenesis

To distinguish the endogenous role of PGF_2αEA from PGF_2α in adipogenesis, we utilized R-flurbiprofen, a substrate-specific inhibitor of PTGS2 targeting AEA and 2-arachidonoylglycerol but not arachidonic acid oxygenation (51). We found that pretreatment of 3T3-L1 preadipocytes at day 0 with 10 μm R-flurbiprofen for 15 min reduced basal levels of PGF_2αEA to below detectable levels after 4 h and totally blocked the up-regulation of PGF_2αEA production induced by exposure of the cells to AEA (Fig. 6D). Analysis of PGF_2α production from the media of similarly treated undifferentiated 3T3-L1 cells revealed that R-flurbiprofen treatment did not modulate basal PGF_2α levels nor prevent its increase in response to treatment with 100 μm arachidonic acid (Fig. 6E). At day 2 of the differentiation protocol, the addition of arachidonic acid resulted in significantly less PGF_2α production, as expected from PTGS2 down-regulation at this time point, and R-flurbiprofen did not reduce this production and only slightly affected basal PGF_2α levels (Fig. 6E).

We then went on to determine whether the selective inhibition of PGF_2αEA with R-flurbiprofen (1, 3, and 10 μm) at day 0 in 3T3-L1 cells could affect the subsequent adipogenesis. When we assayed the expression of adipogenic and mature adipocyte markers at 1, 2, and 4 days of the differentiation protocol, we found that, although R-flurbiprofen did not change Pparg levels, it did result in significantly increased Cebpa expression at day 1. Furthermore, we observed that the onset of the mature adipocyte marker Acrp30 expression was enhanced markedly at day 2 of the differentiation protocol, whereas Fapb4 expression was increased significantly by day 4 (Fig. 6F), indicating that R-flurbiprofen-mediated inhibition of PGF_2αEA production positively modulates adipogenesis. Indeed, triglyceride levels in 3T3-L1 cells differentiated in the presence of R-flurbiprofen were increased significantly by about 10% as compared with controls after 4 days of differentiation (Fig. 6G). These data point to PGF_2αEA as an anandamide-derived, endogenous mediator of adipogenesis that is distinct from PGF_2α.

DISCUSSION

Prostaglandin ethanolamides (or prostamides) are a distinct class of lipid signaling molecules related to prostaglandins and derived from the oxygenation of the endocannabinoid anandamide to PGH₂-EA by PTGS2 (COX-2, encoded by Ptgs2) and the subsequent reduction of PGH₂-EA by prostaglandin synthases. Either AKR1C3/prostaglandin F synthase or prostamide/prostaglandin F synthase (PMPGFS) catalyze the conversion of PGH₂-EA to PGF_2αEA (21–23). Likewise, PGF_2α is produced from the PTGS2-catalyzed oxidation of AA, followed by reduction catalyzed by AKR1C3 or AKR1B3 (46, 53, 54). However, although the biosynthesis of AEA, and, hence, ultimately, PGF_2αEA, relies on AA from the sn-1 position of phospholipids, AA serving as a PGF_2α precursor is released from the sn-2 position of phospholipids (55). Thus, divergent biosynthetic pathways lead to PGF_2αEA and PGF_2α. Furthermore, despite their structural similarity, significant differences in the pharmacology of PGF_2αEA and PGF_2α exist in various cell types, including the dependence of the biological responses to the former on the formation of heterodimers between the wild-type FP receptor and one of its splice variants (both encoded by Ptgfr). Although PGF_2α does not activate such heterodimers, PGF_2αEA is nearly inactive at the wild-type FP receptor homodimers (24–28). PGF_2αEA pharmacology has been mostly investigated through the use of its synthetic structural analog bimatoprost (37), which has been approved for the treatment of glaucoma and eyelash hypotrichosis (56, 57), and of antagonists specific for the FP/FP splice variant heterodimer versus FP homodimers, such as AGN211335 (28).

Patients treated with PGF_2α prodrugs such as travoprost may develop a reversible reduction in periorbital fat pads, leading to the theory that these compounds result in the atrophy of fat in the vicinity of topical administration (38–41, 58). These results are not surprising because PGF_2α is a well documented potent antiadipogenic agent (59, 60). Several hypotheses have been put forward in the search of the mechanism of action of PGF_2α on adipogenesis, including activation of the MEK-ERK MAPK cascade, hypoxia-inducible factor 1, Ca²⁺/calmodulin-dependent protein kinase, and calcineurin via increases in intracellular calcium (34–36, 61). Differentiation assays of human orbital adipose precursors showed that, like PGF_2α, commercially available prodrugs of this mediator inhibit adipogenesis in vitro (62). Surprisingly, however, bimatoprost, which is not hydrolyzed to PGF_2α, inhibits periorbital fat formation, thus suggesting that PGF_2αEA too might inhibit adipogenesis. To date, no known role for PGF_2αEA in the regulation of adipogenesis exists, nor is there any information on the mechanisms behind the antiadipogenic activity of bimatoprost. We have addressed these open questions by utilizing well characterized in vitro model systems and have found that PGF_2αEA and bimatoprost (prostamides) dose-dependently inhibit the up-regulation of the early adipogenic program, resulting in a near complete blockade of the maturation of 3T3-L1 adipocytes and human adipocytes, as measured by the expression of mature adipocyte markers and the production of triglyceride stores. Bimatoprost appeared to be about 10 times more potent than PGF_2αEA, which may be due to its higher stability (24). This property furthers the usefulness of bimatoprost as a tool for the study of prostamide-mediated regulation of adipogenesis, which we have taken advantage of in these studies.

We also showed here that preadipocytes produce significant levels of PGF_2αEA and that these levels quickly decrease with the onset of differentiation, as has been observed with PGF_2α (46), along with the concomitant down-regulation of Ptgs2 expression. Furthermore, in mice under a regime of high-fat feeding, which eventually results in increased adiposity, PGF_2αEA levels in the white adipose tissue are reduced to undetectable levels. We, like others, were unable to detect the expression of Akr1c18 in 3T3-L1 cells (data not shown and Ref. 63)), the homolog of human AKR1C3 that regulates prostamide production (21), but found that these cells do express the prostamide synthase Pmpgfs. However, the expression of this enzyme did not decrease during early adipogenesis. In contrast, we confirm the results of Fujimori et al. (46) by finding that the PGF_2α synthase Akr1b3 does decrease during adipogenesis, indicating that either the decrease in Ptgs2 expression alone results in the observed decrease in PGF_2αEA levels or that AKR1B3, like AKR1C3, may have prostamide as well as prostaglandin synthase activity, a possibility that remains to be examined. We also show that the expression of the FP receptor (Ptgfr) is decreased during adipogenesis in 3T3-L1 cells and human subcutaneous preadipocytes. In the latter cells, the PTGFRv4 (alt4) splice variant responsible for PGF_2αEA/bimatoprost signaling is also down-regulated. Taken together, these data indicate that not only PGF_2αEA production decreases in differentiating adipocytes but that the ability of these cells to respond to this compound also diminishes, pointing to a potential endogenous role for PGF_2αEA as a negative regulator of adipogenesis.

Although we were unable to identify through Basic Local Alignment Search Tool analysis mouse splice variants similar to those of the human Ptgfr gene, the fact that 3T3-L1 cells responded to bimatoprost in a manner and concentration similar to those that inhibit adipogenesis in human cells, which do express FP receptor splice variants, suggests that murine preadipocytes also express prostamide receptors. This notion is supported by evidence that, like in human cells, pharmacological antagonism of PGF_2αEA receptors with AGN211335 reversed PGF_2αEA/bimatoprost-mediated inhibition of adipogenesis to a much greater degree than the FP receptor antagonist AL8810. These data are in line with ocular hypertension studies indicating that bimatoprost and PGF_2αEA signal through receptors different from FP. However, as has been shown for PGF_2α (35), we observed that PGF_2αEA and bimatoprost treatment stimulate the phosphorylation of ERK1/2 in 3T3-L1 cells downstream of MEK activation and that this is required for the antiadipogenic activity of PGF_2αEA and bimatoprost, indicating some convergence of the signaling pathways downstream of PGF_2αEA and PGF_2α in preadipocytes. Interestingly, the MEK antagonist PD98059 we utilized appeared to be more efficacious at inhibiting the antiadipogenic activities of PGF_2αEA and bimatoprost than of PGF_2α, which was also shown to require the calcium-calcineurin cascade for its antiadipogenic activity (36). The requirement for this latter signaling cascade feature was not observed with bimatoprost (data not shown), indicating, in this case, a divergence of the bimatoprost/PGF_2αEA and PGF_2α signaling cascades.

To further examine the potential of prostamides to regulate the process of adipogenesis in vivo, we opted to utilize a developmental model of adipogenesis within juvenile zebrafish (D. rerio) (47). Zebrafish produce PGF_2α and express homologues of the FP receptor (64), PTGS2 and prostamide/prostaglandin F synthase (65 and Zebrafish Information Network, curation of protein database links, automated data submission), as well as producing AEA and expressing biosynthetic and catabolic enzymes for this endocannabinoid,⁵ indicating that they may also be capable of producing prostamides. We found that exposing larvae to bimatoprost dissolved in water resulted in a marked decrease in the number of adipocytes that were observable by microscopy. These data indicate that prostamides are able to inhibit developmental adipogenesis in vivo and point to an evolutionarily conserved role for PGF_2αEA in the regulation of adipogenesis.

Bimatoprost, PGF_2αEA, and PGF_2α all induced a large increase in Ptgs2 expression in 3T3-L1 cells treated with IDM adipogenic medium, as has been observed also with the FP receptor agonist fluprostenol in undifferentiated 3T3-L1 cells, resulting in increased de novo PGF_2α production (52). The fluprostenol-mediated induction, however, was transient, being observed for only 1 h, after which Ptgs2 transcript levels decreased significantly below base line. These results differ significantly from ours in that we not only demonstrated a reversal of the decrease in Ptgs2 expression normally associated with adipogenesis (46, 67) by bimatoprost, PGF_2αEA, and PGF_2α but also observed that these compounds greatly increase Ptgs2 expression beyond the levels observed in undifferentiated cells and do so in a long-lasting manner, the effect still being present after 2 days. We suspect, therefore, that bimatoprost and PGF_2αEA also elevate PTGS2 activity and, thus, prostamide/prostaglandin levels, indicating that PGF_2αEA induces a feed-forward loop that may work to keep prostamide/prostaglandin signaling high within preadipocytes. Self-propagation of prostaglandin signaling is not limited to cells of the adipocyte lineage because a positive feedback loop for PGF_2α production/signaling via Ptgs2 up-regulation was reported in endometrial adenocarcinoma cells as well as in the corpus luteum (68, 69).

Our data also indicate that this feed-forward loop is not limited to Ptgs2 up-regulation alone because we also found that the expression of FP (PTGFR) and FP alternative transcript 4 (PTGFRv4), the proposed prostaglandin and PGF_2αEA receptors, respectively, are down-regulated during adipogenesis in 3T3-L1 and human preadipocytes and that PGF_2αEA and PGF_2α, along with bimatoprost, not only inhibit this down-regulation but also significantly stimulate the expression of these mRNAs. These data differ from results published previously in which it was shown that FP receptor expression increases with 3T3-L1 adipogenesis (46) and that FP receptor stimulation by fluprostenol decreases FP receptor expression in preadipocytes (52). These differences may be due to clonal differences in the 3T3-L1 cells used and the fact that activation of FP by different agonists might produce different effects. However, we must point out that we observed receptor down-regulation not only in 3T3-L1 adipocytes but also in human primary adipocytes and that this phenomenon is consistent with that observed in murine mesenchymal stem cells with adipogenic capacity (63). We speculate that activation of the FP prostaglandin and prostamide receptors by PGF_2α and PGF_2αEA or bimatoprost, respectively, during the initial phases of adipogenesis, inhibits the entry of preadipocytes into the differentiation pathway in vivo in part by up-regulating not only the production of prostamides/prostaglandins within preadipocytes but also by sensitizing them to paracrine and/or autocrine PGF_2α and PGF_2αEA signaling within a microenvironment such as that existing around the adipose vasculature, which provides a niche for adipocyte progenitors and adipogenesis (70, 71). In support of this concept, it has been shown that human microvascular adipose endothelial cells promote the proliferation of preadipocytes in vitro (72) and that adipose capillary endothelial cells produce PGF_2α, which is able to increase the rate at which Swiss 3T3 cells enter the S phase (73, 74). Thus, prostamide/prostaglandin signaling may be part of an endogenous mechanism to regulate preadipocyte number and adipocyte turnover. The loss of periorbital fat observed in patients taking bimatoprost for glaucoma treatment may be a result of the bimatoprost-mediated down-regulation of this process rather than an induction of atrophy of adipose tissue, as has been speculated (39). Given the conservation of the adipogenic program between different fat depots, bimatoprost and other synthetic PGF_2αEA analogs may provide a unique opportunity for the treatment of localized hyperadiposity. Our in vivo data in a vertebrate species phylogenetically distant from mammals but still capable of producing prostaglandins, endocannabinoids, and their respective receptors suggest that the phenomenon described here might, in fact, be relevant not only to adipogenesis in adults but also during development.

Perhaps the most novel and intriguing implication of the data described here (and, in particular, of our finding that PGF_2αEA, via its own receptor, can exert similar, albeit not identical, effects as PGF_2α, while at the same time amplifying not only its own production and action but also that of the more studied cognate compound) lies in the fact that these two prostanoids originate from two divergent biosynthetic pathways and that PGF_2αEA, unlike PGF_2α, is produced at the expense of a well established proadipogenic mediator, i.e. AEA. In fact, we have shown here that PGF_2αEA is the product of AEA metabolism by PTGS2 in preadipocytes and, much less so, in differentiating adipocytes. Further, the pattern of both AEA production and expression of its CB₁ receptor during adipogenesis is the opposite to that observed here for PGF_2αEA because AEA levels and CB₁ expression rise with differentiation (15, 75). PGF_2αEA formation, as mentioned above, may represent an endogenous “stop” signal to adipogenesis that keeps under control the “go” signal of this process afforded by AEA activation of CB₁ and PPAR-γ (18, 19). This stop signal may be terminated not only at the onset of WAT development but also when the adult WAT needs to adapt to changes in dietary energy intake and its storage as fat, as we have shown here for the WAT of HFD- versus STD-fed mice. During adipocyte differentiation, adipocytes start producing more and more AEA that, rather than being hydrolyzed in these cells, may be converted by neighboring undifferentiated preadipocytes into PGF_2αEA to terminate a prolipogenic signal and at the same time keep the formation of further adipocytes on “standby” (Fig. 7). In support of this hypothesis, it was shown that FAAH, the main hydrolytic enzyme for AEA, is significantly less expressed in differentiating 3T3-L1 adipocytes than in preadipocytes (17), thus creating the ideal conditions to convert AEA into PGF_2αEA because tissue prostamide levels are significantly higher when FAAH expression is inactivated (as in Faah-null mice) (76). However, both leptin and insulin up-regulate the expression of FAAH (15, 17, 66). Therefore, the above mechanism for adipogenesis suppression should be reversed following food deprivation/refeeding or when the organism needs to cope with high energy intake, such as during a HFD but before the development of insulin or leptin resistance. Under these conditions, the levels of the two hormones and, hence, those of adipocyte FAAH, would be highest, and subsequently AEA would be processed in adipocytes and not in neighboring preadipocytes, thereby stopping the production of PGF_2αEA formation (as suggested by our data in mice) and disinhibiting adipogenesis. Clearly, such a mechanism is not a mere replica of the antiadipogenic action of PGF_2α that, in fact, unlike PGF_2αEA, was shown not to be down-regulated in vivo following a short period of HFD. Accordingly, we also report that the selective inhibition of the biosynthesis of PGF_2αEA versus that of PGF_2α with R-flurbiprofen (51) in preadipocytes enhances their subsequent differentiation into adipocytes.

FIGURE 7.

Model of AEA- and PGF_2αEA- mediated adipocyte-preadipocyte communication during adipogenesis or adaptation to increased energy intake. During developmental adipogenesis, adipocyte differentiation starts and self-amplifies when the levels of PTGS2, FAAH, PGF_2αEA, and prostamide receptors (as well as PGF_2α and FP receptors) start decreasing and those of AEA, CB₁, and PPARG (as well as of other proadipogenic signals) start increasing. Adipogenesis and excessive lipogenesis (red dots) are then kept under control when, in the presence of down-regulated FAAH, adipocyte-produced AEA is oxidized by COX-2 in neighboring preadipocytes. This negative control is strengthened by PGF_2αEA via further up-regulation of PTGS2 and FP expression in preadipocytes and early differentiating adipocytes but might be blunted in the presence of high FAAH in adipocytes, such as during short term high-fat diet-induced hyperinsulinemia and hyperleptinemia.

In conclusion, we have provided evidence for the presence, biosynthesis from AEA, and mechanism of action in preadipocytes of PGF_2αEA, a novel antiadipogenic mediator. Prostamide signaling or its inactivation might represent a novel “switch” mechanism from inhibition to stimulation of adipogenesis or vice versa that becomes necessary during development and/or to comply with alterations in the nutritional state.