Adipose triglyceride lipase regulates eicosanoid production in activated human mast cells.

Human mast cells (MCs) contain TG-rich cytoplasmic lipid droplets (LDs) with high arachidonic acid (AA) content. Here, we investigated the functional role of adipose TG lipase (ATGL) in TG hydrolysis and the ensuing release of AA as substrate for eicosanoid generation by activated human primary MCs in culture. Silencing of ATGL in MCs by siRNAs induced the accumulation of neutral lipids in LDs. IgE-dependent activation of MCs triggered the secretion of the two major eicosanoids, prostaglandin D2 (PGD2) and leukotriene C4 (LTC4). The immediate release of PGD2 from the activated MCs was solely dependent on cyclooxygenase (COX) 1, while during the delayed phase of lipid mediator production, the inducible COX-2 also contributed to its release. Importantly, when ATGL-silenced MCs were activated, the secretion of both PGD2 and LTC4 was significantly reduced. Interestingly, the inhibitory effect on the release of LTC4 was even more pronounced in ATGL-silenced MCs than in cytosolic phospholipase A2-silenced MCs. These data show that ATGL hydrolyzes AA-containing TGs present in human MC LDs and define ATGL as a novel regulator of the substrate availability of AA for eicosanoid generation upon MC activation.

recombinant human stem cell factor (also termed KITL, KIT ligand) (100 ng/ml; PeproTech, Rocky Hill, NJ). All experiments were carried out with mature MCs (i.e., with cells that had been allowed to differentiate for 6 to 8 weeks) ( 17 ).

MC activation
For immunological activation via IgE receptor cross-linking, MCs were fi rst sensitized for 3 h with human IgE (1 µg/ml, Dia-Tec), washed twice with PBS, resuspended in fresh medium, and then activated by incubation in the presence of polyclonal rabbit anti-human IgE (1 µg/ml, Millipore) for 5, 10, 30, 60, 120, and 240 min.

Western blotting
For the preparation of total cell lysates, MCs were washed twice with PBS, lysed in cell lysis buffer (25 mM Tris/HCl pH 7.4, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 5% glycerol) containing complete protease inhibitor cocktail (Roche). The proteins in total cell lysates were separated by SDS-PAGE under reducing condition and then transferred onto a nitrocellulose membrane (Hybond-C Extra; Amersham Biosciences). Nonspecifi c binding sites were as a major storage site for AA and triggered our interest in the mechanisms regulating the subsequent fate of this particular AA pool.
In this regard, the LDs of MCs are also considered to be important in AA mobilization and subsequent eicosanoid generation ( 10 ). From the liberated AA, the PGs and LTs are formed by the action of cyclooxygenases (COXs) and lipoxygenases (LOs), respectively ( 11 ). The presence of a COX in human lung MC LDs has been visualized by immunogold labeling, demonstrating that MC LDs potentially serve as an active site for PG biosynthesis ( 12 ). For an effi cient utilization of the TG pool in LDs as a source of AA, human MCs need to express appropriate lipolytic enzymes that mediate the hydrolysis of AA from TGs. A candidate for this function is the neutral TG hydrolase, adipose TG lipase (ATGL; also termed patatin-like phospholipase domain-containing protein A2 [PNPLA2]), which mediates the initial step in TG hydrolysis resulting in the formation of an FFA and a diacylglycerol (DAG) molecule ( 13 ). ATGL is expressed in most tissues of the body with the highest expression and activity being found in white and brown adipose tissues ( 14 ). Interestingly, ATGL mRNA is also expressed in macrophages and macrophagederived foam cells of both murine and human origin ( 15 ). Thus, implications of the ATGL in immune response, infl ammation, and atherosclerosis are emerging ( 16 ). However, besides macrophages, no information is available about the presence of ATGL in any other cell types belonging to the immune system.
Based on these considerations and on our previous fi ndings in cultured human MCs, we examined the expression and function of ATGL in MCs. Particularly, we wished to test the hypothesis that ATGL plays a functional role in the hydrolysis of AA present in the TG pool of human MC LDs (i.e., that its activity would be required for providing the substrate for eicosanoid biosynthesis in these cells). We demonstrate that ATGL is expressed in human MCs and that siRNA-mediated silencing of ATGL greatly reduces the amounts of prostaglandin D2 (PGD2) and leukotriene C4 (LTC4) released upon activation of MCs. Thus, ATGL expression in human MCs is necessary for the immediate generation of lipid mediators via both the PG and the LT pathways. These novel observations show that AA-containing TGs in LDs play an important role in providing AA as a substrate for eicosanoid metabolism in human MCs, and that ATGL is an important regulator of this pathway.

Quantifi cation of MC lipid mediators
The amounts of PGD2 and LTC4 released into culture media were analyzed using commercial enzyme immunoassays (Prostaglandin D 2 -MOX EIA Kit and Leukotriene C 4 EIA Kit, Cayman Chemicals) according to the manufacturer's protocols. For blocking the COX-1-mediated generation of PGD2, MCs were incubated with 200 µM aspirin for 3 h, washed twice with PBS, resuspended in fresh medium, and activated as described previously.

LD visualization and quantifi cation
MC LDs were visualized by Oil Red O staining and quantifi ed by fl ow cytometry as described previously ( 9 ). Briefl y, MCs were sedimented (Cytospin, Shandon Instruments) onto glass slides (15 × 10 3 cells/slide) and fi xed with 10% neutral buffered formalin solution (Sigma). The cells were then stained with Oil Red O for 30 min and counterstained with Mayer's hematoxylin. Coverslips were mounted with aqueous medium to preserve the Oil Red O staining (Aquamount, DAKO). Images were captured with a Nikon Eclipse E600 microscope (original magnifi cation, 20×). For LD quantifi cation, MCs were fi xed with 4% paraformaldehyde and probed with allophycocyanin (APC)-labeled mouse monoclonal anti-human CD117 (4 µg/ml, BD Pharmingen) or APC-labeled mouse IgG1 isotype control (BD Pharmingen). Subsequently, the MCs were stained with the fl uorescent dye Bodipy 493/503 (10 µg/ml, Molecular Probes) to stain intracellular LDs and analyzed (1 × 10 4 cells/measurement) using a FACSAria II fl ow cytometer (BD Biosciences).

Statistical analysis
Statistical analyses were performed using GraphPad Prism 5.0 software. Statistical signifi cance between two groups was determined by one-way or two-way ANOVA, followed by the Bonferroni post hoc test. Data are shown as the means ± SEM. The following levels of statistical signifi cance were used: * P < 0.05, ** P < 0.01, and *** P < 0.001.

RESULTS
Because the intracellular lipolytic enzyme ATGL is a prime candidate for hydrolyzing AA from TGs, we fi rst analyzed its expression and regulation in cultured mature human MCs derived from blood CD34 + progenitors. In addition, we also investigated the expression of cPLA 2 , which has been traditionally considered to be the major enzyme mediating the release of AA from membrane PLs. For this purpose, the cultured MCs were immunologically activated by IgE-triggered cross-linking of the cell surface IgE , and LTC4S transcript levels in immunologically activated human MCs. For immunological activation, MCs were fi rst sensitized in the presence of IgE (1 g/ml) for 3 h at 37°C and subsequently activated by adding anti-IgE (1 g/ml). Untreated MCs were used as a control. The activated cells were collected at 5, 10, 60, and 120 min, and RNA was isolated and analyzed for the mRNA expression of ATGL and cPLA 2 (A); COX-1 , COX-2 , and HPGDS (B); and ALOX5 and LTC4S (C) by quantitative PCR (qPCR). For data normalization, GAPDH was used as an endogenous control. Transcript expression levels are shown as fold change and represent the mean ± SEM of three to four different donors. * P < 0.05 . a transient surge with a peak at 120 min. Thus, the responses of the protein levels of ATGL, COX-1, COX-2, and HPGDS to immunological activation of the MCs agreed well with those of their respective transcript levels. It is important to note that, although the protein and transcript levels of ATGL remained stable throughout the MC activation period, this fi nding does not exclude the possibility that changes in its enzymatic activity had occurred. Thus, as shown in human adipose tissue, the protein and mRNA levels of ATGL do not necessarily correlate with its TGhydrolyzing activity, which is primarily regulated by the LD-associated proteins of the PAT family (perilipins) and its activating/inhibitory cofactors comparative gene identifi cation-58 (CGI-58)/G0S2 ( 20,21 ).
Next, we investigated the release of PGD2 upon activation of MCs. As shown in Fig. 3A , MC activation resulted in a rapid increase of the concentration of PGD2 in the incubation medium, which reached the maximal level 60 min after activation and thereafter slowly declined. We also incubated MCs with the COX inhibitor aspirin before their activation and measured the release of PGD2 by the activated MCs over 120 min in the absence of aspirin. As shown in Fig. 3B , during the initial 30 min, when only COX-1 (but not COX-2; see Fig. 1 ) was expressed by the cells, PGD2 secretion was fully inhibited and started to rise only after this period when the COX-2 mRNA levels also began to rise. Taken together, the data show that the immediate generation of PGD2 by activated human MCs solely depends on COX-1.
To test whether the action of ATGL is critical for eicosanoid biosynthesis in activated MCs, we silenced its expression using an siRNA-mediated approach. In addition, we also performed knockdown experiments targeting cPLA 2 . The ATGL and cPLA 2 mRNA levels in siRNA-transfected MCs were examined by qPCR analysis ( Fig. 3C ). The siRNA treatment in the single knockdown experiments reduced ATGL expression by 83% and cPLA 2 expression by 77%, while control siRNA did not show any effect. In the double knockdown, the mRNA expression levels of ATGL and cPLA 2 were reduced by 81% and 71%, respectively ( Fig. 3C ). The release of the major MC eicos anoids, PGD2 and LTC4, was analyzed in the siRNA-silenced MCs at 5, 10, and 30 min after their activation. The siRNA control cells showed normal release responses of PGD2 and LTC4 to the IgE-mediated activation, which were similar to those shown in Fig. 3A . Thus, the release of PGD2 was maximally increased, on average, by 89-fold, and that of LTC4 by 79fold. At all time points, the release of PGD2 was reduced in ATGL and cPLA 2 single-silenced and in double-silenced MCs by 30% to 40% ( Fig. 3D ), while the release of LTC4 in the single-silenced MCs was reduced by ‫ف‬ 60% and in the double-silenced cells, on average, by 80% ( Fig. 3E ). Taken together, besides confi rming the well-known function of cPLA 2 as an enzyme capable of releasing AA from PLs for eicosanoid generation, these data reveal a novel role for ATGL in providing TG-derived AA as a substrate for eicosanoid production.
Based on the observation that ATGL silencing clearly impairs the availability of AA for the generation of PGD2 which catalyzes PGD2 synthesis in MCs. The COX-1 and HPGDS mRNA levels did not change signifi cantly in the stimulated MCs ( Fig. 1B ). Interestingly, COX-2 mRNA levels were low in unstimulated MCs but dramatically increased at 60 min postactivation and then sharply declined ( Fig. 1B ). These results agree with the well-established fact that COX-1, as a constitutively expressed enzyme of the PG pathway, is responsible for the basal PG generation ( 11 ), whereas COX-2 acts as the inducible isoform under infl ammatory conditions ( 19 ). Moreover, we analyzed the gene expression of the major enzymes of the LTC4 pathway, arachidonate-lipoxygenase 5 (ALOX5) and LTC4 synthase (LTC4S). As demonstrated in Fig. 1C , the mRNA expression levels of neither ALOX5 nor LTC4 changed signifi cantly. Taken together, these fi ndings demonstrate that human MCs express ATGL and cPLA 2 , as well as all the major enzymes of the PGD2 and LTC4 pathways, and that upon immunological activation of these cells, only the expression of COX-2 is signifi cantly increased.
To substantiate these fi ndings at the protein level, we next analyzed ATGL and major enzymes of the PGD2 pathway in unstimulated and stimulated MCs by immunoblotting. The 54 kDa immunoreactive band representing ATGL was clearly present in unstimulated cells and remained essentially unchanged upon MC activation ( Fig. 2A ). COX-1 (72 kDa) and HGPDS (25 kDa) displayed considerable protein levels in unstimulated cells with rather small changes over the time period analyzed after MC activation (up to 8 h) ( Fig. 2B ). In sharp contrast, the protein levels of COX-2 (70 kDa), which were undetectable in nonstimulated cells and up to 60 min in the stimulated cells, showed neutral lipid-accumulating phenotype, ATGL appears to play a quantitatively important role in TG hydrolysis of LDs in these immune cells.

DISCUSSION
The current investigation into AA metabolism in MCs is based on our key observation that silencing of ATGL expression leads to a marked accumulation of neutral lipids in cytoplasmic LDs and, concomitantly, to a reduced production of eicosanoids by these cells. As human MCs can store substantial amounts of esterifi ed AA within their large LD TG pool ( 7,8 ), with a signifi cant fraction of the TG molecules containing two or three arachidonyl chains and LTC4 upon MC activation, we wanted to investigate its impact on TG hydrolysis. Therefore, neutral lipids of LDs were stained with Oil Red O ( Fig. 4 A-C ). Compared with mock-transfected and control siRNA-transfected MCs, silencing of ATGL showed a clear increase in number and size of LDs. To quantify the accumulation of neutral lipids, LDs were stained with the fl uorescent dye Bodipy 493/503 and analyzed by fl ow cytometry ( Fig. 4D ). Reduced expression of ATGL resulted in a signifi cant increase of mean fl uorescence intensity of LDs compared with mock-transfected cells, refl ecting intracellular accumulation of neutral lipids in their LDs. This fi nding is in good agreement with the results obtained by Oil Red O imaging of the LDs, which showed an increase in their number and size ( Fig. 4  A-C ). Because ATGL-silenced human MCs developed a Fig. 3. COX-1-dependent immediate release of eicosanoids is reduced in ATGL-and cPLA 2 -silenced human MCs. Human MCs were immunologically activated as described in Fig. 1 (A, B, D, E). A: Media were collected after the indicated periods of activation, and PGD2 concentrations (pg/ml) were measured with a PGD2mox enzyme immunoassay. Media of untreated MCs were used as a control (time 0). Representative data from one human donor are shown. B: The COX-1-mediated release of PGD2 was analyzed in MCs that had been incubated with or without aspirin (200 µM) for 3 h prior to MC activation. PGD2 concentrations (pg/ml) were measured at the indicated time points after MC activation, as described previously. C-E: Human MCs were transfected with siRNAs targeting ATGL (100 nM) and/or cPLA 2 (25 nM) for 20 h at 37°C. Both mock-transfected cells and control siRNA-transfected cells were used as controls. C: ATGL and cPLA 2 mRNA levels were analyzed by qPCR. For data normalization, GAPDH was used as an endogenous control. Transcript levels are shown as fold changes, representing the mean ± SEM of four donors. D, E: PGD2 and LTC4 concentrations (pg/ml) were measured from the collected media with a PGD2mox or LTC4 enzyme immunoassay, respectively. Data are shown as the percentage of control cells and are expressed as the mean ± SEM of data obtained from three and fi ve different donors (for PGD2 and LTC4, respectively). *** P < 0.001, ** P < 0.01, * P < 0.05. and protein expression profi les in human MCs indeed showed that COX-1, as a constitutive enzyme, was expressed already prior to stimulation, while COX-2 expression was observed in the delayed phase of MC stimulation, suggesting that the generation of PGD2 during the immediate phase of eicosanoid release after MC stimulation was solely COX-1 dependent. This presumption was confi rmed by incubating unstimulated MCs in the presence of aspirin to irreversibly inhibit COX-1, followed by stimulation of the cells in the absence of aspirin. Under these specifi c conditions, the release of PGD2 was fully abrogated during the fi rst 30 min after MC activation, indicating that the MCs lacked suffi cient activities of COX-1 or of the subsequently induced COX-2 ( Fig. 3B ) for the generation of detectable amounts of PGD2 within this early activation phase. Our data strongly suggest that the acute generation of PGD2 in activated human MCs is solely mediated by COX-1, emphasizing that under resting conditions, human MCs constitutively express suffi cient amounts of COX-1 and HPGDS required for the immediate generation of PGD2 in response to stimuli that trigger eicosanoid release.
To gain further insight into the role of ATGL in eicosanoid metabolism in human MCs, we measured the acute phase release of PGD2 and LTC4 in ATGL and cPLA 2 single-silenced and in double-silenced MCs after their activation. Importantly, the acute release of PGD2 by the ATGL and cPLA 2 single-silenced MCs was clearly, but equally reduced compared with that by control siRNA (siCTRL)silenced MCs ( Fig. 3D ). Surprisingly, the same effect was even more pronounced for the release of LTC4. Based on the current literature, ATGL is a single-compartmentacting enzyme that exerts its activity on TGs in LDs only, ( 9 ), these stores could serve as a quantitatively major source of AA for eicosanoid biosynthesis. In adipocytes under basal conditions, ATGL mediates the initial step in TG hydrolysis by producing 1,3-DAG, while under hormonal stimulation of lipolysis the enzyme becomes activated by its coactivating factor CGI-58, and the selectivity of ATGL broadens to the sn -1 position leading to the production of 2,3-DAG as well ( 22 ). Therefore, because CGI-58 is also expressed in MCs (unpublished observation), MC activation could further enhance the TG deacylation in LDs by ATGL via hydrolysis of two arachidonyl chains from their AA-rich TG species. The conclusion that ATGL has a functional role in TG metabolism of MCs is supported by the observed intracellular accumulation of TGs in various tissues and peripheral blood leukocytes in patients with neutral lipid storage disease, which is caused by a loss-of-function mutation in ATGL (23)(24)(25). To our knowledge, however, no reports of eicosanoid production by the blood leukocytes of such patients are available.
Here, in focusing on the generation of PGD2 and LTC4 by activated MCs, we analyzed the expression of the major enzymes along the synthetic pathways of these two eicosanoids ( 11 ). No signifi cant changes were observed in mRNA expression levels for the two major enzymes of the LTC4 pathway, ALOX5 and LTC4S, upon MC activation. The expression patterns of COX-1 and COX-2, the initializing enzymes of the PG pathway, found in resting and immunologically activated human MCs are in accordance with those observed in a murine MC cell line (MMC-34), in which IgE-dependent activation did not have any effect on COX-1 expression levels but strongly induced the expression of COX-2 ( 26 ). The COX-1 and COX-2 transcript whereas cPLA 2 is a multicompartment enzyme that acts on PLs in a multitude of cellular organelles ( 21,27 ). Considering this fact, it is quite surprising that silencing ATGL in human MCs reduces the release of eicosanoids to the same extent as silencing cPLA 2 . This observation suggests that the potential contribution for eicosanoid production of AA present in the large TG pool of MC LDs can be maximally of the same magnitude as the amount of AA derived from various PL pools via the action of cPLA 2 .
The observed dominance of the TG-derived AA as a precursor of lipid mediators is a highly challenging fi nding because, historically, PLs of the cellular plasma membrane, and more recently also the PLs of various cellular compartments, have been described as the only source of AA for eicosanoid biosynthesis ( 28 ). In particular, cPLA 2 has been considered to be the major enzyme mediating the release of AA from the sn -2 position of membrane PLs ( 6 ). Also, the formation of PGD2, LTC4, and leukotriene B4 (LTB4) in murine MCs is highly dependent on cPLA 2 , as demonstrated using bone marrow-derived MCs generated from cPLA 2 Ϫ / Ϫ mice ( 29 ). Although our studies support the hypothesis that, by providing AA for eicosanoid biosynthesis in human MCs, ATGL plays an important role as a TG hydrolase, it is not yet ascertained whether AA released from the TGs can be used directly or indirectly for eicosanoid generation. Direct utilization of TG-derived AA (i.e., bypassing the transfer of AA from TGs to PLs and its subsequent liberation by cPLA 2 ) would enable an accelerated provision of the substrate for eicosanoid-forming enzymes, which would be important for the ultrarapid release of eicosanoids from MCs upon their activation. However, the results of the above-mentioned study ( 29 ) on eicosanoid formation in bone marrow-derived murine MCs generated from cPLA 2 -defi cient animals point to an indirect pathway, at least in those cells.
It is becoming obvious that major functions of LDs in infl ammatory cells differ fundamentally from the classical energy-providing role of LDs in adipocytes. Thus, LDs accumulate in various immune cells during infl ammatory and infectious conditions (reviewed in Ref. 28 ), and, moreover, LDs have been identifi ed as sites of eicosanoid biosynthesis in certain immune cells ( 30,31 ). The early speculations that AA-containing TGs might also serve as a source of AA for the generation of lipid mediators (discussed in Refs. 7, 8, 28 ) have now been strengthened by a mechanistic revelation implicating that functional ATGL in activated human MCs is required for the generation of PGD2 and LTC4. The identifi cation of the regulatory role of ATGL in eicosanoid production by MCs is an important step toward defi ning and specifying the unique role of LDs in infl ammatory cells. The establishment of cultured human MCs as a well-defi ned cellular system for studies on the regulation of eicosanoid production during all phases of mediator release will facilitate our current studies aimed at directly quantifying the relative contributions of the substrate AA to eicosanoid production along the three candidate pathways (i.e., derived directly from TGs, indirectly from TGs, or directly from PLs; Fig. 5 ).

Fig. 5.
Schematic model for the proposed role of ATGL in providing triglyceride-derived AA for eicosanoid biosynthesis in human MCs. In human MCs, ATGL mediates the hydrolysis of AA from triglycerides (TGs) stored in intracellular LDs. The liberated AA may contribute to eicosanoid production via two different routes, either directly from TGs (1) or indirectly from TGs by transient incorporation into LD PLs and subsequent release by the LDassociated phospholipase cPLA 2 (2 ). This enzyme can also provide AA directly from the PLs present in the LDs (3). Modulation of ATGL activity could alter the contribution of TG-derived AA to eicosanoid formation by enzymes of the PG pathway (the COXs) and the LT pathway (the LOs), thereby affecting the generation and release of eicosanoids from human MCs, as demonstrated in this study by knockdown of ATGL.