Lipid mediator metabolic profiling demonstrates differences in eicosanoid patterns in two phenotypically distinct mast cell populations

Abstract

Mast cells are inflammatory cells that play key roles in health and disease. They are distributed in all tissues and appear in two main phenotypes, connective tissue and mucosal mast cells, with differing capacities to release inflammatory mediators. A metabolic profiling approach was used to obtain a more comprehensive understanding of the ability of mast cell phenotypes to produce eicosanoids and other lipid mediators. A total of 90 lipid mediators (oxylipins) were characterized using liquid chromatography-tandem mass spectrometry (LC-MS/MS), representing the cyclooxygenase (COX), lipoxygenase (LO), and cytochrome P450 (CYP) metabolic pathways. In vitro-derived murine mucosal-like mast cells (MLMC) and connective tissue-like mast cells (CTLMC) exhibited distinct mRNA expression patterns of enzymes involved in oxylipin biosynthesis. Oxylipins produced by 5-LO and COX pathways were the predominant species in both phenotypes, with 5-LO products constituting 90 ± 2% of the CTLMCs compared with 58 ± 8% in the MLMCs. Multivariate analyses demonstrated that CTLMCs and MLMCs secrete differing oxylipin profiles at baseline and following calcium ionophore stimulation, evidencing specificity in both a time- and biosynthetic pathway-dependent manner. In addition to the COX-regulated prostaglandin PGD₂ and 5-LO-regulated cysteinyl-leukotrienes (e.g., LTC₄), several other mediators evidenced phenotype-specificity, which may have biological implications in mast cell-mediated regulation of inflammatory responses.

Mast cells are important multifunctional inflammatory cells that are particularly well known for their involvement in allergic diseases; however, they are also implicated in many physiological and pathophysiological functions in health and disease (1). Mast cells are distributed in all tissues and are particularly numerous in those tissues that are in close contact with the environment, e.g., the skin, the gastrointestinal tract, and the respiratory system (2). They express a variety of receptors that recognize both endogenous (e.g., cytokines, neuropeptides, complement) and exogenous molecules (e.g., pathogen-derived molecules, allergens through IgE receptor cross-linking). Upon recognition and activation, the cells have the capacity to rapidly release preformed inflammatory mediators stored in granules, produce lipid mediators de novo, and synthesize cytokines, chemokines, and growth factors. The spectra of released mediators are dictated by the stimuli; for example, IgE-receptor aggregation causes degranulation, lipid mediator secretion, and release of cytokines, whereas other stimuli, such as CD153, only lead to a specific release of primarily chemokines (3).

There exists a considerable heterogeneity among tissue mast cells. In humans, they are divided into two main phenotypes: those that contain the granule proteases tryptase and chymase (designated as MC_TC) and those that only express tryptase (MC_T) (4). The corresponding phenotypes in rodents are referred to as “connective tissue-like” and “mucosal-like” mast cells, respectively. Recent findings have identified a more substantial heterogeneity among human lung mast cells assessed by morphology and expression of different receptors, enzymes, and mediators (5). One example is the more pronounced expression of 5-lipoxygenase (5-LO) in mast cells located in the small airways and pulmonary vessels, compared with mast cells in the central airways and parenchyma (5). The consequence of this heterogeneity and differential mediator release is that the effect of mast cell activation can vary substantially, dependent on stimuli and mast cell phenotype.

Oxidized lipid mediators (e.g., eicosanoids, oxylipins) are one major and potent group of mediators released by mast cells. Oxylipins are signaling lipids formed from fatty acids via mono- or dioxygenase-catalyzed oxygenation that play prominent roles in inflammatory diseases (6–8). These compounds include the arachidonic acid (AA)-derived prostaglandins and leukotrienes (LT) as well as analogous compounds produced from an array of polyunsaturated fatty acid substrates [e.g., linoleic acid (LA), α-linolenic acid, docosahexaenoic acid]. The biosynthesis of oxylipins is initiated via three major pathways: cyclooxygenase (COX), lipoxygenase (LO), and cytochrome P450 (CYP) (6–8). The activity of these pathways and the binding of their products to their corresponding receptors are important drug targets for multiple inflammatory diseases (9).

Following activation (e.g., via the high-affinity IgE-receptor), mast cells rapidly generate prostaglandin D₂ (PGD₂) and the leukotrienes C₄ (LTC₄) and B₄ (LTB₄) (10). Both PGD₂ and the LTs are proinflammatory mediators involved in modulating and attracting immune competent cells, as well as acting on structural cells (10). For instance, LTs are important for the recruitment of neutrophils, eosinophils, monocytes, T-cells, and dendritic cells (11–14), as well as for inducing smooth muscle contraction (15) and affecting development of allergic airway disease, fibrosis, and vascular injury (16–18). Cysteinyl-leukotrienes (cys-LT) are mediators of human asthma, and anti-leukotrienes are effective therapeutics in the clinical management of disease (19). Mast cells are the predominant producer of PGD₂ and its metabolites [e.g., 9α, 11β-prostaglandin F₂ (9α, 11β-PGF₂); 2,3-dinor-9α, 11β-PGF₂; tetranor-PGDM] and can be used as an indicator of mast cell activation (20). Mice overproducing PGD₂ develop increased airway inflammation as well as Th2 cytokine production following sensitization and challenge compared with wild-type mice (21), and targeted disruption of the DP receptor protects mice against allergic asthma (22). Besides products of the COX and 5-LO pathways (Fig. 1), mast cells have been shown to produce products of the 12/15-LO pathway, including eoxins, 15-hydroxy-eicosatetraenoic acid (HETE), and 15-oxo-eicosatetraenoic acid (KETE) (23, 24). However, to date research on mast cell-derived oxylipins has focused on a small number of lipids, and limited information is available regarding the breadth of prospective oxylipins produced or the potential for phenotypic differences.

Fig. 1.

Overview of the arachidonic acid 5-lipoxygenase (5-LO) and cyclooxygenase (COX) pathways. Oxylipin nomenclature is as described in the text and in supplementary Table I.

To perform an in-depth analysis of oxylipin generation in mast cells, we employed targeted quantitative lipid metabolic profiling of supernatants from calcium ionophore-treated murine mast cells of the connective tissue-like mast cell (CTLMC) and mucosal-like mast cell (MLMC) phenotypes. Well-established protocols were used for the development of MLMCs and CTLMCs (25–28), followed by stimulation with the conventional mast cell secretagogue A23187, used since the mid-70s to induce histamine and lipid mediator release from mast cells (29, 30). Multivariate statistical models were then employed to investigate the time dependence of oxylipin production and the relationship between mast cell phenotype and oxylipin biosynthesis.

MATERIALS AND METHODS

Mast cell cultures

CTLMCs and MLMCs were derived by culturing bone marrow cells (BMC) from C57BL/6 mice as previously described (26). All animal experiments were performed according to the guidelines of the Animal Ethics Committee in Stockholm and the PHS policy. In brief, CTLMCs were developed in medium containing recombinant murine stem cell factor (50 ng/ml) and 1 ng/ml recombinant murine interleukin (IL)-4. MLMCs were developed in medium containing 5 ng/ml IL-3, 5 ng/ml recombinant murine IL-9, 1 ng/ml recombinant human TGF-β1, and recombinant murine stem cell factor (50 ng/ml). Media, IL-4, IL-3, IL-9, TGF-β1, and stem cell factor were obtained from PeproTech Nordic, Sweden. All cells were cultured for a minimum of 2 weeks for MLMC and 4 weeks for CTLMC prior to use. The maturity and purity of the cells were examined by toluidine blue staining and flow cytometry analysis for expression of Kit and FcϵRI, using FITC-anti-mouse CD117 (Kit) mAb 2B8 and FITC-conjugated rat IgG2b isotype control (BD Pharmingen, NJ), FITC-conjugated anti-mouse FcϵRI-α mAb MAR-1, and FITC-conjugated Armenian hamster IgG isotype control (eBioscience, CA).

PCR array

RNA was isolated from 2 × 10⁶ unstimulated mast cells as pellets in RNAlater by the RNeasy Mini Kit essentially according to the manufacturer's instructions (Qiagen, Sweden). The cells were lysed in a final concentration of 40 mM DTT (Sigma Aldrich, Sweden) and disrupted using a rotor-stator homogenizer. The RNA concentration was measured using a Nanodrop 3300 (Thermo Scientific, DE). For each sample, 160 ng of cDNA was prepared by the QuantiTect Rev. Transcription Kit (Qiagen, Sweden) according to the instructions provided in the kit. The low-density arrays, containing primers for 16 mouse genes involved in the synthesis of eicosanoids and three endogenous controls, were custom made (Applied Biosystems, Sweden). cDNA from the different mast cells were first titrated in a Taqman PCR assay to determine the optimal amount of cDNA to use as a template. A negative control cDNA, synthesized without the reverse transcriptase enzyme, was also included. The primers used for titration were 18S rRNA and JNK. cDNA (160 ng) in Taqman Gene Expression Master Mix (Applied Biosystems, Sweden) was added per port in the low-density array plate and run on an ABI 7900H instrument (Applied Biosystems, Sweden) according to the recommended instructions. The total number of cycles was 42. The final analysis of raw data was performed using the software SDS 2.2 (Applied Biosystems, Sweden).

Mast cell treatment

CTLMCs and MLMCs (1 × 10⁶ cells/ml) were resuspended in RPMI buffer supplemented with 1% penicillin streptomycin and 1% BSA. Cells were treated for 2, 8, and 15 min with 1 μM of calcium ionophore A23187 (Sigma Aldrich, Sweden). After completion of incubation, supernatants were collected for oxylipin analysis.

Oxylipin, cysteinyl-leukotriene, and eoxin extraction and quantification

Calibration and deuterium-labeled oxylipin, cys-LT, and eoxin standards as well as the technical standard N-cyclohexyl-N′-dodecanoic acid urea (CUDA) were obtained from Cayman Chemical (MI) and Larodan Fine Chemicals (Sweden). All standards used, as well as full abbreviations, are provided in supplementary Table I. Oxylipins, cys-LTs, and eoxins were offline SPE extracted from 0.5 ml cell supernatant aliquots using Waters (MA) Oasis-HBL 60 mg cartridge columns. Oxylipins were extracted using the same procedure as previously described (31). Due to the low recoveries of cys-LTs using the standard analytical protocols (31, 32), these compounds were extracted and analyzed using a separate method developed for cys-LT quantification. The recovery of d₅-LTE₄ with the general oxylipin method (31) was 7.1 ± 9.2% averaged across all 96 samples, as opposed to 35.4 ± 14.8% with the cys-LT-specific method (the cys-LT-specific method gave recoveries of 48.4 ± 17.0% and 40.4 ± 12.1% for d₅-LTC₄ and d₅-LTD₄, respectively). Accordingly, cys-LT and eoxin extraction was performed as follows: 4 ml of methanol, 10 µl deuterated standards (200 nM), 10 µl BHT, and EDTA solution (0.2 mg/ml) were added to the samples, which were then placed at −20°C for 1 h, followed by centrifugation (10 min, 4°C <7000 rpm) to remove precipitate. The supernatant was removed, and 4 ml MilliQ water was added, and then the samples were applied to washed and conditioned SPE columns. Samples were washed with 3 ml of MilliQ water by gravity. The aqueous plug was pulled from the cartridges under high vacuum, and cartridges were further dried with low vacuum for ∼30 min. Cys-LT and eoxins were eluted with 1.5 ml of methanol into cryotubes (Corning Inc., NY) containing 6 μl of 30% glycerol in methanol. Volatile solvents were removed using a GMI miVac Concentrator system (MN). Samples were reconstituted in 100 μl of 2:3 methanol:water containing 176 nM of the technical standard (CUDA). The samples were vortexed for 1 min, filtered [Millipore centrifugal filters, Durapore (PVDF) 0.1 μm], transferred to autosampler vials, and stored at −20°C until analysis. A Waters (MA) Acquity UPLC separation module coupled to a Waters Xevo triple quadrupole mass spectrometer equipped with an electrospray source was used for analyses, and separation was performed via a Waters 2.1 × 50 mm Acquity UPLC BEH C18 column with a 1.7 μm particle size. Cys-LT and eoxins were separated using a gradient containing (A) water and (B) acetonitrile (supplementary Table II). Gradient elution was performed at a flow rate of 500 μl/min during a 10 min run. Cys-LT and eoxin transition and LC retention times are provided in supplementary Table III. Oxylipins were separated using a gradient containing (A) MilliQ water and 0.1% acetic acid and (B) acetonitrile:isopropanol 90:10 (supplementary Table IV). Gradient elution was performed at a flow rate of 500 μl/min during a 21 min run. The mass spectrometer was operated in positive SRM mode for cys-LTs and eoxins and negative SRM mode for oxylipins. All compounds were quantified using internal standard methods as previously described (31). Compounds detected above the limit of quantification (LOQ) were quantified and recalculated back to the original sample concentrations. To demonstrate the capability of the method, supplementary Fig. I, A and B, display total ion chromatograms of the oxylipin and cys-LT methods, respectively. The ability of the oxylipin method to distinguish key compounds is shown in supplementary Fig. IC, which displays the extracted ion chromatograms (PGE₂ and PGD₂; 6-trans-LTB₄, LTB₄, and purported isomers; and 5,6-DiHETE and purported isomers).

Statistical methods

Univariate statistical analysis was performed using the Student t-test and Bonferroni's multiple comparison test (33). Principal component analysis (PCA) and orthogonal projections to latent structures (OPLS) were performed using SIMCA-P+12 (Umetrics, Umeå, Sweden) following log base 10 transformation, mean centering, and UV scaling (34). Model performance was reported as cumulative correlation coefficients for the model (R²), and predictive performance was based on cross validation calculations (Q²) and CV-ANOVA.

RESULTS

PCR array analysis of both mast cell phenotypes demonstrates the presence of transcripts for the required enzymes in oxylipin biosynthesis

Oxylipins are formed via two major metabolic pathways involving a suite of enzymes that oxidize the fatty acid substrate to form a bioactive product (Fig. 1). To confirm the presence of the necessary message of the enzymes responsible for oxylipin formation in both CTLMCs and MLMCs, we performed a PCR array on mRNA purified from resting CTLMCs and MLMCs as well as on the BMCs that are the source for differentiation. Overall, both cell types demonstrated the presence of the necessary mRNA for oxylipin biosynthesis (supplementary Fig. II). CTLMCs expressed the highest mRNA levels of COX-1, cytosolic PGE synthase (cPGES), and microsomal PGE synthase (mPGES), as well as high levels of thromboxane A synthase (TXAS). MLMCs expressed greater levels of TXAS mRNA and low levels for COX-1 and COX-2. Compared with BMCs, both mast cell types showed increased mRNA expression of the hematopoetic PGD synthase PTGDS2. For LT synthesis, CTLMCs expressed the highest levels of LTA₄ hydrolase (LTA4H), whereas MLMCs expressed the highest levels of 5-LO and LTC₄ synthase (LTC4S). Both CTLMCs and MLMCs expressed high levels of five-lipoxygenase activating protein (FLAP), but the levels were not significantly greater than those of BMCs. For the 12/15-LO pathway, both CTLMCs and MLMCs evidenced higher mRNA levels than BMCs. Overall, the transcript data illustrated an induction of critical enzymes for oxylipin formation during mast cell differentiation. The data evidenced both expression similarities as well as distinct patterns between CTLMCs and MLMCs.

Activated mast cells demonstrate phenotype-specific oxylipin profiles

Ninety oxylipins representing three metabolic pathways (COX, LO, and CYP) were screened using LC-MS/MS analysis. Oxylipins produced by 5-LO and COX pathways (Fig. 1) were the predominant species in both phenotypes (95–98%), but 8-LO, 12/15-LO, and CYP products were also observed. The distributions of COX- and 5-LO-regulated compounds were distinct, with the portion of 5-LO products constituting 90 ± 2% of the CTLMCs compared with 58 ± 8% in the MLMCs at 15 min (Fig. 2A). In total, 28 oxylipins were detected above the method limit of quantification (LOQ) in the CTLMCs (COX, n = 4; 5-LO, n = 12; 8-LO, n = 4; 12/15-LO, n = 5; and CYP, n = 4), and 27 oxylipins were detected above the LOQ in the MLMCs (COX, n = 7; 5-LO, n = 13; 8-LO, n = 3; 12/15-LO, n = 4; and CYP, n = 0) (supplementary Table V). An additional four oxylipins (PGE₁, PGE₂, PGF₂α, and LTE₄) were present above the method LOD in the CTLMCs and an additional three oxylipins [12-HETE, 9(10)-epoxy-octadecenoic acid (EpOME), and 12(13)-EpOME] were detected above the LOD in the MLMCs (supplementary Table V). As shown in Fig. 3, oxylipin levels increased greatly in both phenotypes upon activation with the ionophore A23187. However, CTLMC levels increased linearly over time in contrast to MLMC levels, which increased exponentially with minor changes after 2 min. The average concentrations (nM) and the standard deviations for both cell types are shown in Table 1, with individual values given in supplementary Table VI. Total ion chromatograms (TIC) from one of the MLMC samples at control and 15 min (oxylipin method) and one of the CTLMC samples at control and 15 min (cys-LT method) are shown in supplementary Fig. III.

Fig. 2.

Mast cell-derived oxylipin percentage composition based upon enzymatic pathway of formation (CTLMC and MLMC). (A) Oxylipins grouped into overall biosynthetic pathways: COX, 5-LO, 8-LO, 12/15-LO, and CYP. (B) Distribution of COX-derived oxylipins. Colorings are based upon enzymatic pathways downstream of COX (PGES, PGDS, and TXAS). (C) Distribution of 5-LO-regulated oxylipins. Colorings are based upon direct 5-lipoxygenase products (direct 5-LO), nonenzymatic products of LTA₄ (nonenzymatic from LTA₄), and enzymatic pathways downstream of LTA4H, LTC4S, and 12/15-LO. Oxylipin nomenclature is as indicated in the text and in supplementary Table I.

Fig. 3.

Sum of the 5-LO and COX products from CTLMCs and MLMCs. (A) Concentration of the individual oxylipin mediators in cell supernatant (nM). COX products are marked in yellow and 5-LO products are marked in red. (B) Sum of the total 5-LO and COX products produced from differentiated BMCs taken from an individual mouse (nM) for CTLMCs (n = 6) and MLMCs (n = 6). Oxylipin nomenclature is as described in the text and in supplementary Table I.

TABLE 1.

Oxylipin concentrations (nM) in CTLMC and MLMC supernatants prior (control) and following activation at 2, 8, and 15 min

			Average ± SD (nM)								Significance (P-value)
			Control		2 min		8 min		15 min
Enzyme	PUFAa	Oxylipin	CTLMC	MLMC	CTLMC	MLMC	CTLMC	MLMC	CTLMC	MLMC	Control	2 min	8 min	15 min
COX	DGLA	PGE1	ND	ND	ND/LOD	0.17 ± 0.14b	ND/LOD	0.19 ± 0.10b	ND/LOD	0.33 ± 0.13	—	3.9E-02	8.1E-03	1.9E-03
		PGD1	ND	0.11 ± 0.15b	0.13 ± 0.09b	2.8 ± 0.6	0.65 ± 0.25	3.9 ± 1.2	1.4 ± 0.4	4.3 ± 1.3	3.6E-01	1.0E-04	9.3E-04	2.1E-03
	AA	PGE2	LOD	ND/LOD	LOD	2.5 ± 1.0	LOD	3.1 ± 1.1	LOD	4.0 ± 1.3	4.1E-03	2.9E-03	1.6E-03	1.1E-03
		PGD2	ND/LOD	2.8 ± 3.1	2.2 ± 1.7b	29 ± 4	9.7 ± 2.2	35 ± 8	17 ± 4	38 ± 7	8.4E-02	4.1E-08	3.7E-04	3.2E-04
		PGF2α	ND	ND	ND	0.21 ± 0.03	ND/LOD	0.32 ± 0.07	ND/LOD	0.40 ± 0.15	—	1.5E-05	8.1E-05	1.4E-03
		TXB2	0.09 ± 0.10b	2.1 ± 2.7	0.33 ± 0.10	33 ± 5	1.2 ± 0.5	42 ± 9	2.2 ± 1.0	46 ± 10	1.2E-01	1.5E-05	7.8E-05	1.3E-04
		12-HHTrE	1.98 ± 1.63b	1.8 ± 2.8b	2.6 ± 1.3b	60 ± 12	4.3 ± 1.3	68 ± 19	6.2 ± 2.8	66 ± 13	9.0E-01	7.7E-05	4.0E-04	5.6E-05
		Sum (n = 7)	2.5 ± 1.7	7.1 ± 8.8	5.6 ± 0.8	128 ± 21	16 ± 3	154 ± 36	28 ± 7	160 ± 33	2.7E-01	3.1E-05	2.3E-04	1.2E-04
5-LO	DGLA	LTB3	ND	ND	ND	1.9 ± 1.8b	ND	2.2 ± 2.1b	ND	2.1 ± 1.7b	1.0E+00	6.6E-02	5.9E-02	3.2E-02
		5-HETrE	ND	ND	1.2 ± 1.1b	30 ± 22	9.2 ± 4.4	28 ± 15	16 ± 5	27 ± 19	1.0E+00	2.5E-02	2.7E-02	2.1E-01
	AA	LTB4	ND	1.5 ± 1.3b	1.3 ± 1.2b	14 ± 5	11 ± 7	19 ± 5	20 ± 9	21 ± 6	4.5E-02	6.7E-04	5.6E-02	7.4E-01
		6-trans-LTB4	ND	0.364 ± 0.43b	2.7 ± 2.1b	38 ± 25	19 ± 12	49 ± 22	26 ± 12	57 ± 22	2.0E-01	1.9E-02	1.3E-02	1.3E-02
		LTC4	0.47 ± 0.39b	5.5 ± 4.3	8.6 ± 6.7b	41 ± 12	85 ± 47	71 ± 26	161 ± 47	79 ± 44	3.5E-02	2.0E-04	5.4E-01	1.1E-02
		LTD4	ND/LOD	ND/LOD	0.17 ± 0.14b	0.19 ± 0.14b	1.2 ± 0.6	0.38 ± 0.30b	2.3 ± 0.7	0.41 ± 0.32b	9.2E-02	8.2E-01	8.1E-03	8.8E-05
		LXA4	ND/LOD	ND/LOD	ND/LOD	0.17 ± 0.22b	0.041 ± 0.043b	0.36 ± 0.40b	0.054 ± 0.07b	0.39 ± 0.44b	6.0E-01	1.4E-01	1.1E-01	1.2E-01
		5-HETE	0.18 ± 0.17b	ND	21 ± 17	31 ± 15	144 ± 77	29 ± 8	223 ± 85	28 ± 10	7.9E-02	2.8E-01	1.4E-02	2.4E-03
		5-KETE	ND/LOD	ND	0.31 ± 0.20b	0.83 ± 0.38	1.6 ± 0.6	0.75 ± 0.16	2.5 ± 0.8	0.92 ± 0.49	3.6E-01	1.5E-02	1.5E-02	2.4E-03
		5,6-DiHETE	ND	ND/LOD	1.1 ± 1.1b	7.4 ± 4.2	11 ± 12b	11 ± 6	16 ± 14	12 ± 6	3.6E-01	1.2E-02	9.9E-01	4.9E-01
		5,15-DiHETE	ND	ND	ND/LOD	1.7 ± 1.7b	ND/LOD	3.1 ± 1.9b	0.90 ± 0.56b	3.5 ± 1.1	1.0E+00	1.0E-01	2.0E-02	5.6E-04
	EPA	LTB5	ND	ND/LOD	ND/LOD	0.80 ± 0.41	0.092 ± 0.086b	1.1 ± 0.3	0.17 ± 0.17b	1.3 ± 0.4	3.6E-01	5.9E-03	3.3E-04	8.6E-05
		5-HEPE	ND	ND	0.45 ± 0.40b	7.2 ± 5.2	3.6 ± 3.3	10 ± 4	5.6 ± 4.0	10 ± 3	1.0E+00	2.5E-02	1.1E-02	4.4E-02
		Sum (n = 13)	1.6 ± 0.5	8.3 ± 6.0	37 ± 29	175 ± 82	287 ± 156	225 ± 74	473 ± 164	243 ± 78	4.2E-02	7.6E-03	4.0E-01	1.7E-02
12/15-LO	LA	9-HODE	2.3 ± 0.5	LOD	2.2 ± 0.6	1.2 ± 1.2b	2.3 ± 0.7	2.2 ± 1.2b	2.4 ± 0.6	2.6 ± 1.3b	1.76E-04	6.49E-02	7.91E-01	7.39E-01
		13-HODE	2.4 ± 0.6	ND/LOD	2.2 ± 0.6	0.66 ± 0.63b	2.4 ± 0.7	0.94 ± 0.96b	2.2 ± 0.6	1.2 ± 1.3b	1.94E-04	1.04E-03	1.34E-02	1.18E-01
		9-KODE	1.1 ± 0.6	ND/LOD	1.0 ± 0.4	0.50 ± 0.54b	1.0 ± 0.4	0.69 ± 0.79b	0.94 ± 0.29	0.86 ± 0.94b	4.85E-03	8.71E-02	3.76E-01	8.64E-01
	AA	12-HETE	1.7 ± 1.1	ND/LOD	1.5 ± 1.0b	ND/LOD	1.5 ± 1.0b	ND/LOD	1.6 ± 1.0b	ND/LOD	2.55E-02	1.94E-02	1.69E-02	2.27E-02
		15-HETE	0.055 ± 0.69b	ND	0.097 ± 0.12b	0.51 ± 0.61b	0.36 ± 0.30b	0.76 ± 0.83b	1.2 ± 0.5	0.56 ± 0.62b	1.75E-01	1.56E-01	3.08E-01	9.82E-02
CYP	LA	9(10)-EpOME	3.2 ± 1.3	ND/LOD	3.5 ± 1.3	ND/LOD	3.6 ± 1.7	LOD	3.0 ± 1.2	LOD	1.81E-03	1.46E-03	5.06E-03	2.87E-03
		12(13)-EpOME	3.1 ± 1.2	ND/LOD	3.3 ± 1.3	ND/LOD	3.5 ± 1.7	ND/LOD	2.6 ± 0.9	ND/LOD	2.05E-03	2.12E-03	5.67E-03	1.01E-03
	AA	5(6)-EpETrE	0.15 ± 0.09b	ND	0.26 ± 0.19b	ND	0.61 ± 0.24	ND	0.81 ± 0.22	ND	1.38E-02	2.22E-02	1.75E-03	3.22E-04
		11(12)EpETrE	0.037 ± 0.047b	ND	0.13 ± 0.13b	ND	0.37 ± 0.15	ND	0.38 ± 0.08	ND	2.24E-01	6.92E-02	1.79E-03	8.57E-05
8-LO	DGLA	8-HETrE	ND/LOD	ND	0.40 ± 0.33b	1.9 ± 2.1b	2.8 ± 2.2	1.8 ± 1.8b	4.5 ± 2.4	1.8 ± 0.3b	3.63E-01	1.46E-01	3.79E-01	5.31E-02
	AA	8-HETE	ND/LOD	ND/LOD	ND/LOD	LOD	0.88 ± 0.90b	0.40 ± 0.57b	1.9 ± 1.4b	0.3 ± 0.4b	5.99E-01	1.75E-01	2.94E-01	4.03E-02
		11-HETE	0.22 ± 0.11	LOD	0.23 ± 0.12	2.2 ± 0.8	0.59 ± 0.16	2.8 ± 0.9	0.93 ± 0.33	3.2 ± 0.8	1.01E-02	1.37E-03	1.78E-03	5.82E-05

DGLA, dihomo-γ-linolenic acid; ND, not detected; PUFA, polyunsaturated fatty acid.

^aThe parent fatty acid for the biosynthesis of the shown lipid mediators.

^bSamples detected at a mixture of quantitative levels, LOD and/or ND levels (for details, see supplementary Table VI). For statistical purposes LOD were calculated as one third of the LOQ and ND as one tenth of LOQ levels (supplementary Table V).

COX products.

The sum of COX-regulated products secreted from the MLMCs was consistently ∼10-fold higher relative to the CTLMCs at all time points following activation (Table 1, Fig. 3). The three most abundant COX products [PGD₂, TXB₂, and 12-hydroxy-heptadecatrienoic acid (HHTrE)] together comprised ∼95% of the total amount of COX products in both cell types at 8 and 15 min. However, the portion of PGD₂ in the CTLMCs was significantly larger compared with the MLMCs (15 min: 64 ± 6% versus 24 ± 3%; P = 6.0E⁻⁰⁸), whereas the portions of TXB₂ (15 min: 8 ± 2% versus 29 ± 1; P = 6.5E⁻¹⁰) and 12-HHTrE (22 ± 6% versus 42 ± 2%; P = 1.8 E⁻⁰⁵) were significantly larger in the MLMCs (Fig. 2B). Moreover, the ratio of TXAS-dependent compounds (TXB₂ and 12-HHTrE) compared with compounds formed by PGD synthases (PGD₁ and PGD₂) was inverse between the two cell types (MLMCs ∼2.5 versus CTLMCs ∼0.5).

5-LO products.

The sum of 5-LO-regulated compounds was significantly higher in the MLMCs at control levels (P = 0.04), at 2 min (P = 0.008), not significant at 8 min (P = 0.4), and significantly higher in the CTLMCs at 15 min (P = 0.02), (Table 1, Fig. 3), demonstrating a difference in the rate of 5-LO-derived oxylipin production in the two cell types. At 2 min, 5-hydroxy-eicosatrienoic acid (HETrE), LTB₄, 6-trans-LTB₄, LTC₄, 5-KETE, 5,6-dihydroxy-eicosatetraenoic acid (DiHETE), LTB₅, and 5-hydroxy-eicosapentaenoic acid (HEPE) were significantly higher in the MLMCs. The same compounds were significantly higher at 15 min, except for LTC₄ and 5-KETE, which along with LTD₄ and 5-HETE were significantly higher in the CTLMCs. In relative proportion, ∼90% of the 5-LO products in the CTLMCs were distributed over four compounds at 15 min: LTC₄ (35 ± 7%), 5-HETE (47 ± 5%), LTB₄ (4 ± 1%), and 6-trans-LTB₄ (5 ± 1%) (Fig. 2C). A similar distribution was observed at 8 min. The pattern in the MLMCs at 15 min showed ∼90% of the 5-LO products distributed over five compounds: LTC₄ (32 ± 13%), 5-HETE (12 ± 3%), LTB₄ (9 ± 1%), 6-trans-LTB₄ (24 ± 6%), and 5-HETrE (11 ± 4%). Similar relations were observed at 2 min and 8 min.

12/15-LO, 8-LO, and CYP products.

At 15 min following stimulation, the oxylipins found in quantitative amounts from the 12/15-LO, 8-LO, and CYP pathways were minor relative to 5-LO and COX products; CTLMCs (1.6 ± 0.5%, 1.2 ± 0.3%, and 1.3 ± 0.4%, respectively) and MLMCs (1.4 ± 0.9%, 0.5%±0.3%, and 0.1 ± 0.04%, respectively) (Fig. 2A). The majority of these compounds were detected in significantly higher levels in the CTLMCs, but the measured concentrations were close to the LOD and also detected in significantly higher concentrations in the controls (Table 1). The main exception was the 8-LO product 8-HETE, which was observed in significantly higher concentrations in the CTLMCs 15 min following activation (Table 1).

Integrated multivariate modeling

Multivariate statistical modeling of the oxylipin concentrations at control and following stimulation was performed to investigate overall trends in the data. In the scores plot of the PCA model (R²X = 0.78, Q² = 0.72; constructed with two principal components), distinct clustering was observed of CTLMCs and MLMCs, both for controls and following stimulation (Fig. 4A). In contrast to the MLMCs, which grouped in a single large cluster following stimulation, CTLMCs separated into four distinct clusters (i.e., control, 2 min, 8 min, and 15 min). From the plot, it is evident that component 1 (R²X = 0.52, axis t[1]) primarily separated controls and stimulated cells, whereas component 2 (R²X = 0.26, axis t[2]) was driven by phenotype. The loadings plot in Fig. 4B shows that 5-LO and COX products were driving the model following stimulation in both cell types. CYP, 12/15-LO, and 8-LO products correlated with the CTLMCs, compared with TXAS and PGES products, which correlated with MLMCs.

Fig. 4.

PCA analysis of oxylipin levels and distribution in mast cells (R²X = 0.78, Q² = 0.72). Component 1 (R²X = 0.52, axis t[1]) primarily separated controls and stimulated cells, whereas component 2 (R²X = 0.26, axis t[2]) was driven by phenotype. (A) Scores plot. CTLMCs are shown as circles and MLMCs as squares. Measurements are colored as follows: Control (light yellow), 2 min (yellow), 8 min (orange), and 15 min (red). (B) Loadings plot. Oxylipins are colored based upon enzyme biosynthetic pathway: COX (gold), 5-LO (red), 8-LO (light blue), 12/15-LO (black), and CYP (dark blue). Oxylipin nomenclature is as described in the text and in supplementary Table I.

To investigate differences between CTLMCs and MLMCs at 2 and 15 min, two OPLS-DA models (Fig. 5A, B) were constructed, and the p(corr) values were plotted in a shared and unique structures (SUS) plot (Fig. 5C) (35, 36). The models, both constructed with one predictive and one orthogonal vector, were very strong: 2 min (R²Y = 0.99, Q² = 0.97) and 15 min (R²Y = 0.99, Q² = 0.98). The SUS plot showed an overall positive correlation of the variables in the two models (R² = 0.57), with the majority of COX and 5-LO products correlating with the MLMCs at both time points following stimulation. However, the 5-LO-derived compounds 5-HETE, 5-KETE, LTC₄, and LTD₄, as well as 8-LO-regulated 8-HETE and 8-HETrE and 12/15-LO-produced 15-HETE, evidenced a shift from correlating with the MLMCs at 2 min to correlating with the CTLMCs at 15 min. The responses of LTB₄ and 5,6-DiHETE were notable, shifting from strongly correlating with the MLMCs at 2 min [p(corr): 0.81 and 0.88] to essentially no correlation with the MLMCs at 15 min [p(corr): 0.13 and −0.14]. These results support those in Fig. 3A, with the 5-LO products shifting from higher concentrations in the MLMCs at 2 min to higher concentrations in the CTLMCs at 15 min. This trend was also observed for the main 8-LO products. Furthermore, the model illustrates that the CYP products correlate with the CTLMCs, and the COX products correlate with the MLMCs.

Fig. 5.

(A) Loading column plot of the OPLS model at 2 min. Mediators correlating with 95% confidence are shown. (B) Loading column plot of the OPLS model at 15 min. Mediators correlating with 95% confidence are shown. (C) SUS plot correlating the OPLS models of i) CTLMCs versus MLMCs at 2 min (X-axis) and ii) CTLMCs versus MLMCs at 15 min (Y-axis). Positive correlations are indicative of correlation with the MLMCs. Oxylipins are colored based upon enzyme biosynthetic pathway, i.e., COX (gold), 5-LO (red), 8-LO (green), 12/15-LO (black), and CYP (blue). Because LTE₄ was not detected in either MLMCs or CTLMCs at 2 min, it was excluded from the 2 min model and thus not included in the SUS plot.

DISCUSSION

From the results presented herein, it is evident that the two main mast cell phenotypes, CTLMC and MLMC, have different oxylipin profiles, both regarding the expression of the necessary enzymes for their generation and in activated pathways following stimulation. Both univariate and multivariate analyses found that the COX- and 5-LO-regulated lipid cascades are the major pathways activated following stimulation. However, the phenotypes evidenced major differences in the distribution of 5-LO and COX products at 15 min after stimulation (CTLMCs, 90% and 6%, versus MLMCs, 58% and 40%, respectively). Generally, gene expression data did not correlate closely with lipid mediator production, with significantly higher expression of 5-LO and FLAP in the MLMCs, and COX1 and COX2 levels significantly higher in the CTLMCs. However, the oxylipin profiles evidenced improved correlation with gene expression when focused on the enzymes downstream in the biosynthetic cascades [i.e., COX (PTGDS2 and TXAS were significantly higher in the MLMCs) and 5-LO (LTA4H was significantly higher in the CTLMCs, whereas LTC4S exhibited no difference between phenotypes)]. However, it should be noted that expression values were not measured in the cells following stimulation, and thus, correlating gene expression with activated cells should be performed with caution.

The observed differences in the oxylipins produced downstream of the COX and 5-LO cascades may be of biological significance. PGD₂, which is produced by PGDS, has previously been reported as the primary COX product in mast cells (20). Effects induced by PGD₂ are dependent on two main receptors, the PGD₂ receptor1 (DP1) (22) and the chemo-attractant homologous receptor expressed on Th2 cells (CRTH2) (37). The receptors can regulate inflammatory cell migration, control cytokine production, and mediate lipid synthesis, and there is also speculation that they may cross-talk during inflammatory events (38–40). It is notable that CRTH2 also can be activated by 11-dehydro-TXB₂ (41), a metabolite of TXB₂ that is formed via the COX/TXAS pathway (Fig. 1) (42). The levels of PGD₂ in the MLMCs were significantly higher at 15 min (P = 0.0003) compared with the CTLMCs (Fig. 3). TXB₂ is a stable, nonenzymatically produced product of TXA₂ that is rapidly formed via hydrolysis (42). 12-HHTrE is formed concurrently with TXA₂ from PGH₂ (43) and has been shown to have chemotactic activity in leukocytes (44), as well as serving as a substrate for NAD+-dependent 15-hydroxyprostaglandin dehydrogenase (45). Additionally, elevated levels of TXB₂ and 12-HETrE indicate extensive TXAS activity with high production and abundance of TXA₂, an important mediator possessing prothrombotic properties (46). TXA₂ has also been implicated in the development of bronchial hyper-responsiveness (47). It remains to be established whether the observed differences, although statistically significant, in COX-derived products have physiological relevance. The distribution of COX-generated compounds in MLMCs showed that TXAS-derived TXB₂ and 12-HHTrE were significantly larger in composition (%) relative to the CTLMCs, which produced PGD₂ as the main COX product (Fig. 2). It should be noted that both PGD₂ and PGE₂ will be further metabolized via 15-hydroxy PGDH and 15-oxo-PG Δ13-reductase to their corresponding 13,14-dihydro-15-oxo-metabolites (13,14-dihydro-15-keto PGD₂ and 13,14-dihydro-15-keto PGE₂) (48, 49). The half-lives of these compounds are relatively short and are further metabolized to give 11β-hydroxy compounds that have also undergone β-oxidation of one or both side chains. Accordingly, results regarding the relative contributions of the LO and COX pathways should be tempered with the knowledge that downstream metabolites were not assayed.

The response curve of the CTLMCs was linear over the measured time course, whereas the MLMC curve evidenced essentially saturated levels at 2 min. Interesting were the observed shifts in 5-LO products, which were significantly higher in the CTLMCs at 15 min (P = 0.02) compared with 2 min [higher in the MLMCs (P = 0.008)] and 8 min [not significant (P = 0.4)]. According to the SUS plot in Fig. 4, this shift is most prominent for the cys-LTs, as well as for 5-HETE and 5-KETE. Other compounds, such as LTB₄ and 5,6-DiHETE, also evidenced this shift from strongly correlating with MLMCs at 2 min to essentially no correlation at 15 min. It is unclear whether these shifts are further accentuated with increased time. The ratio is constant between 5-LO and COX products at 8 min and 15 min (8 min, 88 versus 6%; 15 min, 90 versus 6%, respectively), suggesting that they are increasing at the same relative rate. Considering the major role of LTB₄ and cys-LTs in mast cell biology and inflammatory regulation (6, 7, 10), the phenotype-specific differences in concentration and relative abundance of LTC₄ versus LTB₄ will most likely have biological implications. For example, numerous studies have established the impact of LTs in recruitment of T-cells and dendritic cells and in development of allergic airway disease (11, 12, 50). Data also indicate that LTC₄ and LTD₄ are key players in fibrosis and vascular injury via the CysLT2 receptor, and they have been implicated in the pathogenesis of human abdominal aortic aneurysm (17, 18, 51). Accordingly, the observed phenotype-specific differences in LT production may have ramifications for the etiology of pathological processes in both the onset and resolution of disease.

The 8-LO products 8-HETE and 8-HETrE also evidenced a shift similar to 5-HETE, 5-KETE, and the cys-LTs in the SUS plot. In mice, these compounds are regulated via the 8-LO pathway (52), but in humans, these compounds are produced nonenzymatically (9). It has previously been shown that A23187 treatment of dorsal skin on NMRI mice causes increased levels of 8-HETE (53). It should also be noted that all of the HETEs can potentially be produced via free radical oxidation. The stereochemistry of the hydroxyl moiety can indicate synthetic route (S for LO activity versus R for monoxygenase activity); however, it is now well established that lipoxygenases can also remove a pro-R hydrogen from the substrate with antarafacial insertion of oxygen to generate products with R chirality (9). Accordingly, the enantiomeric excess is required to determine synthetic source, with a racemic mixture indicating free radical oxidation and predominantly R or S indicating a biosynthetic source. Because chiral chemistry was not employed in this study, results should be tempered with the understanding that the synthetic source of the HETEs is unclear.

Human mast cells have previously been shown to produce products of the 12/15-LO pathway, including 15-HETE, 15-KETE, and eoxins (23, 24). The primary 12/15-LO products are 12- and 15-HETE, of which 15-HETE was predominantly observed in both MLMCs and CTLMCs. (Table 1). The downstream oxidation products (12- and 15-KETE) were not observed, but this is potentially due to the high LOD levels for these compounds (∼10 nM). It has previously been shown that 15-KETE is the primary product in human cord blood mast cells, suggesting that there could be species-specific differences in mast cell eicosanoid levels (23). We also observed 5,15-DiHETE and LXA₄, which are produced by 12/15-LO following 5-LO action (7, 54, 55). LXA₄, which was primarily detected in the MLMCs (supplementary Fig. IV), is involved in the resolution phase of inflammation (7, 54, 56). As such, it is of particular interest to study its potential effects in the mast cell and its surrounding microenvironment. If lipoxin levels are also specific for the mast cell phenotype-specific human equivalents, it could have implications for disease. Additionally, some other AA- and LA-mediated 12/15-LO products were found, particularly in the CTLMCs (Figs. 4 and 5, Table 1). However, the amounts were detected in quantities close to the LOQ. The lack of eoxin detection may also indicate species differences between human and murine mast cells.

Of the characterized compounds in this study, the majority were derived from AA (n = 20). Additionally a number of LA (n = 5), dihomo-γ-linolenic acid (DGLA; n = 5), and eicosapentaenoic acid (EPA; n = 2) compounds were found in quantitative amounts in both phenotypes, demonstrating that murine mast cells have the capacity to produce these mediators if the substrate is present. As such, the ratio of AA versus EPA or DGLA products in the phenotypes could potentially be altered depending on source of nutrition and microenvironment. Notable also is that the DGLA-derived PGE₁ and PGD₁ (COX) followed the same response curve as PGE₂ and PGD₂ but that the 5-LO products of DGLA (LTB₃ and 5-HETrE) and EPA (LTB₅ and 5-HEPE) were still found in significantly lower amounts at 15 min in the CTLMCs.

This study is the first to provide detailed information on lipid mediator production in mast cells. The results provide evidence that the oxylipin profiles of murine mast cells are phenotype specific, with additional differences in biosynthesis response curves following stimulation with ionophore. Further mechanistic studies are required to investigate whether the trends observed in murine mast cells are also found in the human phenotypes. The disparate oxylipin profiles in murine mast cells may affect the inflammatory cycle from the activation to resolution phase, as well as the overall activity and regulation of the phenotypes, in a time-dependent manner, with implications for disease pathology.