Introduction

Eicosanoids are potent lipid mediators derived from arachidonic acid (AA), an essential component of cellular membrane lipids, via the cytochrome P450 (CYP), lipoxygenase (LOX), and cyclooxygenase (COX) pathways but also non-enzymatically by interaction with reactive oxygen species (ROS) [1–4] (Fig. 1). The different autacoids/autacoid groups, like the CYP-mediated epoxyeicosatrienoic acids (EETs) and 20-hydroxyeicosatetraenoic acid (20-HETE), the soluble epoxide hydrolase (sEH)-mediated dihydroxyeicosatrienoic acids (DHETs), the LOX-mediated leukotrienes (LTs) and lipoxins (LXs), the COX-mediated prostaglandins (PGs), thromboxanes (TXs), and 12-heptadecatrienoic acid (12-HHT), the cis–trans conjugated hydroxyeicosatetraenoic acids (HETEs) mediated by CYP, LOX, or COX, and the isoprostanes (iPTs) generated by the impact of ROS on AA and therefore used as reliable in-vivo indicators of ROS levels and oxidative stress, have a broad array of biological functions, from proinflammatory and vasoconstrictive to anti-inflammatory and vasodilatative [1–3, 5, 6].

Fig. 1
figure 1

Eicosanoid formation via the three main pathways of the AA cascade—CYP, LOX, and COX—and non-enzymatically by interaction of AA with ROS. AA = arachidonic acid; CYP = cytochrome P450; LOX = lipoxygenase; COX = cyclooxygenase; sEH = soluble epoxide hydrolase; ROS = reactive oxygen species; EETs = epoxyeicosatrienoic acids (42 = 14,15-EET, 43 = 11,12-EET, 44 = 8,9-EET, 45 = 5,6-EET); DHETs = dihydroxyeicosatrienoic acids (36 = 14,15-DHET, 37 = 11,12-DHET, 38 = 8,9-DHET, 40 = 5,6-DHET, 41 = 5,6-DHET δ lactone); 39 = 20-HETE = ω–oxygenase-mediated 20-hydroxyeicosatetraenoic acid; HETEs = hydroxyeicosatetraenoic acids (16 = 15-HETE, 17 = 11-HETE, 18 = 8-HETE, 19 = 12-HETE, 20 = 9-HETE, 21 = 5-HETE); LXs = lipoxins (2 = LXA4, 1 = LXB4); LTs = leukotrienes (11 = LTB4, 7 = LTC4, 6 = LTD4, 8 = LTE4); ω–ox–LTs = ω–oxygenase-mediated LT metabolites (5 = 20-HO-LTB4, 4 = 20-COOH-LTB4); TXs = thromboxanes (TXA2, 29 = TXB2, 26 = 2,3-dinor-TXB2, 24 = 11-dehydro-TXB2); PGs = prostaglandins (PGI2, 28 = 6-keto-PGF1α, 27 = 2,3-dinor-6-keto-PGF1α, 23 = 6-keto-PGE1, 33 = PGE2, 22 = 19(R)-hydroxy-PGE2, 34 = PGD2, 35 = PGJ2, 14 = Δ12-PGJ2, 3 = 15-deoxy-Δ12,14-PGJ2, 31 = PGF2α, 32 = 13,14-dihydro-15-keto-PGF2α); 15 = 12-HHT = 12-hydroxyheptadecatrienoic acid; iPTs = isoprostanes (30 = 8-iso-PGF2α, 25 = 2,3-dinor-8-iso-PGF2α). Eicosanoid numbering is the same as in Fig. 2

Full size image

Among all eicosanoids, the CYP-derived 20-HETE and EETs, and the sEH-derived DHETs with reportedly contradictory vasoregulative functions are of particular interest. These mediators are described as both vasodilators and vasoconstrictors acting by themselves, via metabolism by COX, or by inducing the expression of COX and thus the enhanced production of prostanoids [6–9]. Although EETs/DHETs were proposed over a decade ago as chemical correlates of the vasodilatative endothelium-derived hyperpolarizing factor (EDHF) in the cardiovascular compartment, and discussed since then [10–14], there still is no agreement about the nature of EDHF despite sustained efforts concerning characterization and identification. In the lung, in contrast to the other compartments, the role of EETs is poorly investigated, but first reports describe them as both vasoconstrictors and vasodilators [6, 7]. Interestingly, 20-HETE is a vasodilator in the lung but a vasoconstrictor in all other compartments and antagonizes EDHF-mediated relaxation [6, 9]. Intriguingly, CYP enzymes may control vascular homeostasis by simultaneous production of EETs, 20-HETE, and ROS [15]. ROS may be involved in vasoregulation directly [16] or by generating iPTs, considered as not just markers of oxidative stress but also as vasoconstrictors [3] and even as candidates for EDHF [17].

This brief synopsis suggests that comprehension of eicosanoid physiology is quite difficult when analyzing only some autacoids without considering all the other relevant mediators and the actual oxidative stress status. Thus, the question arises: What is the relative, i.e., actual contribution of 20-HETE, EETs, and DHETs in the context of all vasoactive representatives of the AA cascade and the iPTs to the observed physiological effect? To answer this question, a global profiling technique is required to enable direct comparison of all eicosanoid generation profiles as a prerequisite for comprehensive and differentiated understanding of the physiological role of 20-HETE, EETs, and DHETs in the overall mediator context, including possible interrelations with other autacoids, under conditions of variable ROS concentrations.

The aim of this work was to provide a set of bioanalytical information suitable for elaboration, as a basis for further investigations, of a first solid hypothesis regarding the role of free (i.e., not esterified into cellular membrane phospholipids) EETs, DHETs, and 20-HETE in epithelial/endothelial homeostasis in general and their contribution to EDHF-mediated responses and to hypoxic pulmonary vasoconstriction (HPV) in particular by use of a comprehensive method which enables accurate correlation between generation and bioactivity profiles of all the biologically relevant eicosanoids at a glance. High-performance liquid chromatographic (HPLC) separation followed by online spectrophotometric and/or gentle electrospray ionization mass-spectrometric (MS) detection is best suited for the analysis of a broad array of these unstable compounds [18–23]. The disadvantages of other techniques, for example gas chromatography–MS or HPLC with fluorescence detection reside in incomplete separation at high temperatures of the thermolabile autacoids or high background interference, respectively, and their need for sample derivatization [24, 25]. As free eicosanoids generally occur in biological sources at very low concentrations, the above mentioned aim can be achieved only by use of a method which is not merely comprehensive but also highly sensitive. Quantitative sensitivity can be improved dramatically by the use of capillary columns with small inner diameters (i.d. <0.5 mm) instead of large i.d. columns (microbore, narrowbore, or conventional analytical columns with i.d. = 1–4.6 mm), employed by previous methods, because of two major sensitivity-increasing factors:

  1. 1.

    peak efficiency, i.e., chromatographic signal intensity, is inversely proportional to column i.d. and thus very much higher for capillary columns than for large-i.d. columns, and

  2. 2.

    substance transfer from the capillary column to the mass spectrometer is virtually complete, because of the typically very low column effluent flow rates in the low microliter range in contrast with only partial analyte transfer because of the high flow rates typical of large-i.d. columns.

We therefore developed a technique combining:

  1. 1.

    source-specific sample extraction,

  2. 2.

    rugged isocratic and high-sensitivity capillary liquid chromatographic (CapLC) separation, and

  3. 3.

    reliable dual online photodiode array (PDAD) and electrospray ionization (ESI) tandem mass-spectrometric (MS2) detection in the negative-ion mode employing an ion-trap mass spectrometer for simultaneous identification and quantitation of all relevant CYP, sEH, LOX, and COX-derived eicosanoids (EETs, DHETs, 20-HETE, LTs, LXs, DiHETEs, PGs, TXs, 12-HHT, HETEs) and free iPTs within one chromatographic run.

Application of this CapLC–PDAD–ESI–MS2 method to hypoxic lung homogenate demonstrated its robustness, reliability, and high sensitivity in comprehensive “eicoprofiling” down to the femtogram range.

Experimental

Standards, chemicals, reagents, and solvents

The epoxyeicosatrienoic acids 14,15-EET, 11,12-EET, 8,9-EET, and 5,6-EET, and the dihydroxyeicosatrienoic acids 14,15-DHET, 11,12-DHET, 8,9-DHET, 5,6-DHET, and 5,6-DHET δ-lactone were supplied by Cascade Biochem (Reading, UK), Cayman Chemical (Ann Arbor, MI, USA), and Biomol (Hamburg, Germany). 12-HHT, 20-HETE, the hydroxyeicosatetraenoic acids 15-HETE, 11-HETE, 8-HETE, 12-HETE, 9-HETE, and 5-HETE, the leukotrienes 20-COOH-LTB4, 20-HO-LTB4, LTC4, LTE4, 6t-LTB4, 6t,12e-LTB4, LTB4, and LTD4, the lipoxins LXA4 and LXB4, the dihydroxyeicosatetraenoic acids 5(S),6(R)-DiHETE and 5(S),6(S)-DiHETE, and the non-deuterated internal standard (IS) prostaglandin B1 (IS = PGB1) were purchased from Biomol and Cayman. The prostaglandins 6-keto-PGF1α, 6-keto-PGE1, 2,3-dinor-6-keto-PGF1α, PGD2, PGJ2, Δ12-PGJ2, 15-deoxy-Δ12,14-PGJ2, PGE2, 19(R)-hydroxy-PGE2, PGF2α, and 13,14-dihydro-15-keto-PGF2α, the isoprostanes 8-iso-PGF2α and 2,3-dinor-8-iso-PGF2α, the thromboxanes TXB2, 11-dehydro-TXB2, and 2,3-dinor-TXB2, and stearidonic acid (SDA) were obtained from Cayman. The deuterated internal standards [2H8]14,15-EET, [2H8]11,12-EET, [2H8]8,9-EET, [2H8]5,6-EET-Me, [2H8]15S-HETE, [2H3]LTE4, and [2H3]2,3-dinor-6-keto-PGF1α, were purchased from Biomol. [2H6]20-HETE, [2H8]5S-HETE, [2H8]12S-HETE, [2H4]6-keto-PGF1α, [2H4]PGD2, [2H4]15-deoxy-Δ12,14-PGJ2, [2H4]PGE2, [2H4]PGF2α, [2H4]8-iso-PGF2α, [2H4]TXB2, and [2H4]11-dehydro-TXB2 were obtained from Cayman, and [2H4]LTB4 was obtained from both Biomol and Cayman.

Butylated hydroxytoluene (BHT), triethylamine (TEA), and diethylamine (DEA) were obtained from Sigma (Munich, Germany). Ethylenediaminetetraacetic acid disodium salt dihydrate (EDTA Na2.2H2O), tetrasodium salt tetrahydrate (EDTA Na4.4H2O), and formic acid (FA) were purchased from E. Merck (Darmstadt, Germany). Bovine serum albumin (BSA) and Coomassie Brilliant Blue G-250 dye reagent were from Bio-Rad Laboratories (Munich, Germany).

Krebs–Henseleit-buffer supplemented with 5% hydroxyethyl starch (KHB) was obtained from Serag Wiessner (Naila, Germany). Water was purified with a Milli-Q Synthesis A10 system (Millipore, Eschborn, Germany). Methanol was from Burdick and Jackson (Muskegon, MI, USA), acetonitrile was purchased from J. T. Baker (Deventer, Netherlands), isopropanol was obtained from Fluka (Buchs, Switzerland), and ethanol was from Merck. Table 1 shows the solvent mixtures used for solid-phase extraction (SPE).

Table 1 Composition (% v/v) of the solvent mixtures
Full size table

SPE and CapLC–PDAD–ESI–MS2 procedures and equipment

Eicosanoids were extracted from lungs by SPE and subjected to CapLC–PDAD–ESI–MS2. Oxidative loss was prevented by addition of the antioxidant BHT from the very beginning of sample extraction, prior to any manipulation and throughout the whole analytical procedure.

SPE

Optimal purification and concentration of eicosanoids was achieved by an extraction procedure specifically developed for rabbit lung homogenate. Lungs were subjected immediately to homogenization and extraction or stored at −85°C for up to 24 h until further processing. For assessment of recoveries and for quality control, 0.5-g samples of biological matrix (homogenate from blood-free rinsed lungs) and 0.5 mL KHB alone were spiked with known amounts of eicosanoid standards, constant amounts of internal standards (PGB1 and/or deuterated eicosanoids), SDA, and BHT. Genuine samples were processed in the same manner, except for supplementation with eicosanoid standards. In brief, 0.5 g tissue from hypoxic or normoxic rabbit lungs was supplemented with 50 μL 15 mmol L−1 BHT in methanol, 50 μL FA, 5 μL of solution E (Table 1), internal standards, and SDA, and subjected to mechanical homogenization in a chilled 2-mL borosilicate glass tube for 10 min at 1,500 rpm employing a Potter S homogenizer with PTFE plunger mounted on a stainless steel shaft, all from Sartorius (Göttingen, Germany). The homogenate samples were diluted with 0.2 mL KHB. Aliquots thereof (0.5-mL) were supplemented with BHT (1.5 mmol L−1, final concentration) and 0.7 mL solvent mixture F, then vortex mixed for 10 min and centrifuged for 30 min at 21250 g and 4°C. Prior to SPE, 0.6 mL supernatant was diluted with 4.4 mL of KHB and applied to the conditioned extraction cartridges filled with end-capped octadecylsilyl (ODS) silica (Chromabond; C18ec, sorbent mass, 200 mg; reservoir volume, 3 mL) mounted in a Chromabond vacuum manifold, all from Macherey–Nagel (Düren, Germany), without vacuum suction. Elimination of interfering substances was achieved by two-step elution of the columns with solvent mixtures optimized for this purpose (Table 1 and electronic supplementary material Table S1). After the final elution step, the collected samples were dried gently under a stream of N2 and subjected immediately to CapLC–PDAD–ESI–MS2 analysis or stored at −85°C for up to 24 h until further processing. Throughout the whole sample-handling procedure, beginning with homogenization and during the extraction process, all tissues, samples, solvents, and eluents were kept at low temperatures.

CapLC–PDAD–ESI–MS2

The CapLC–PDAD–ESI–MS2 system used for eicosanoid analysis consisted of two main parts:

  1. 1.

    A 1100 Series capillary LC unit (Agilent Technologies Deutschland, Waldbronn, Germany) comprising a binary micro pump, a micro autosampler thermostatted at 4°C and equipped with an 8-μL sample loop, and a photodiode-array detector (PDAD) with a nano flow cell (500 nL volume, 10 mm path length). ChemStation Rev. A.09.03 software was employed for device control, raw data acquisition, integration, spectral analysis, and processing.

  2. 2.

    An Esquire 3000+ ion trap mass spectrometer (Bruker Daltonik, Bremen, Germany) equipped with an ESI source was connected to the PDAD of the CapLC unit by means of a 75-cm length of a 25-μm inner diameter (i.d.) fused silica tubing. The peak efficiency of the capillary column and thus overall quantitative sensitivity was improved by avoidance of dead volume in the fluidic system of the CapLC–PDAD–ESI–MS2 set-up by use of 25 μm instead of 50 μm i.d. fused silica tubing, beginning with the injection valve of the micro autosampler. Instrument control and MS2 data processing was performed with the Esquire v 5.1 (Bruker) software fully integrated with the ChemStation Rev. A.09.03 (Agilent) software.

The ion trap used had the capability to perform fragmentation beginning with MSn stage 2 (n = 2, MS2) through MSn stage 11 (n = 11, MS11). In the present method we employed only MS2 fragmentation. N2 was used as nebulizing and drying gas, and He was used as collision gas. The ESI source conditions nebulizing gas pressure, drying gas flow, and drying gas temperature were set to 11.0 psi, 6.0 L min−1, and 300°C, respectively. The sampling, transfer, and focusing voltages (capillary, skimmer, octapoles, lens) were optimized individually for each analyte by parameter ramping during direct infusion (DI) of analyte standards using a syringe pump (Cole-Parmer Instrument Company, Vernon Hills, IL, USA) at a rate of 2.5 μL min−1 (1–5 ng analyte min−1). The capillary voltage was set to 3600–3800 V. Isolation width and collision energy in terms of fragmentation amplitude were individually optimized for each compound and set to 0.7–1.5 Da and 0.48–0.78 V, respectively. Mass cutoff was set to 27% of the m/z value of the precursor molecular ion. In order to ensure optimal loading of the trap even under conditions of varying ion beam currents due to eluting LC peaks typical for online LC–MS2 coupling, the accumulation time was automatically adjusted by the ion charge control (ICC) parameter of the instrument. As a result, not only were the quantitation sensitivity and accuracy of the ion trap considerably improved but also the dynamic range of the instrument was greatly extended.

Separation of autacoids was performed on capillary columns (250 mm × 150 μm = L × i.d. and 250 mm × 300 μm = L × i.d.) filled with a high-purity spherical polymer-coated silica gel stationary phase of extremely high carbon content (Grom-Sil 80 ODS-7 pH with 22% C18; particle size, 4 μm; pore diameter, 80 Å) from Grom Analytik + HPLC (Rottenburg-Hailfingen, Germany). Prior to injection of 8-μL aliquots into the CapLC–PDAD–ESI–MS2 system, biological or quality control samples dried after SPE were redissolved in 20 μL solvent mixture D containing 30 μmol L−1 BHT, vortex mixed, supplemented with 20 μL water, vortex mixed again, sonicated, and centrifuged briefly. Alternatively, 20-μL aliquots of non-extracted eicosanoid standard mixtures for calibration, freshly prepared prior to analysis from the stock solution by serial dilution with solvent mixture D (Table 1) containing 30 μmol L−1 BHT and constant amounts of IS and SDA as reference substance, were supplemented with 20 μL water, vortex mixed, sonicated, and centrifuged. Thus, maximal mass sensitivity of capillary separation was achieved by large-volume injection (8 μL instead of typical 0.2 μL) employing “on-column focusing” [26, 27], i.e., enrichment of the analytes at the top of the capillary column before penetrating the column bed by keeping the solvent strength of the sample weaker than that of the mobile phase (i.e., by diluting all samples 1:1 with water prior to injection). This compression effect not only counteracts the classical band broadening effect but, in addition, enables a considerable gain in peak efficiency under large-volume injection conditions. As mobile phase a mixture of water–acetonitrile–methanol–FA (130:240:80:0.22, v/v/v/v) was used (pH 4.30, adjusted with DEA). The desired selectivity for the separation of the protophilic peptido-LTs (LTC4, LTD4, LTE4) could sensibly be chosen by up or down regulation of pH in steps of 0.01. Flow rate was 2.7 μL min−1 (for i.d. = 300 μm) or 0.95 μL min−1 (for i.d. = 150 μm) and column oven temperature was 33°C. The eluting compounds were subjected to dual online spectrophotometric and tandem mass-spectrometric analysis by employing the consecutively interfaced PDAD and ion trap. The PDAD provided full UV spectra (190 to 400 nm) of eluting analytes, enabled the definition of up to five detection wavelengths, and allowed checking for peak purity and subtraction of possible coeluting material. The ion trap was operated in the negative-ion mode with multiple reaction monitoring (MRM) for simultaneous acquisition of MS2 traces of up to ten coeluting non-isomeric eicosanoid precursor ions, thus enabling accurate assessment of analytes even under conditions of insufficient chromatographic resolution.

For chromatographic characterization of eicosanoids, the elution of each analyte relative to the retention time (RT) of the non-deuterated IS PGB1 and SDA as second reference substance was calculated by use of the following equation for the relative retention index (RRI):

$$ {\text{RRI}} = \frac{{{\text{RT}}{\left( {{\text{SDA}}} \right)} - {\text{RT}}{\left( {{\text{analyte}}} \right)}}} {{{\text{RT}}{\left( {{\text{SDA}}} \right)} - {\text{RT}}{\left( {{\text{PGB}}_{{\text{1}}} } \right)}}} $$
(1)

Mediator identification was performed by use of RRI and both from full UV spectra of the peaks detected on the optical traces and from MS2 spectra of the peaks detected on the MS2 traces of the analyte precursor ions. MS2 analyses of the molecular carboxylate ions of the eicosanoids were performed both offline by DI of single-analyte standards and online by monitoring the MS2 spectra of autacoid standards injected on to the interfaced CapLC unit.

Quantitation was performed by IS method using standard mixtures which contained:

  1. 1.

    PGB1 as a common IS when UV and MS2 or UV detection alone was employed, or

  2. 2.

    deuterated eicosanoid standards and PGB1 as IS when MS2 detection alone was used.

Recoveries were calculated by comparison of the quantities of standards chromatographed after sample extraction with those of standards injected directly in the CapLC–PDAD–ESI–MS2 system employing IS method. Measurements were performed for each eicosanoid at six concentrations (n = 6) and were corrected for endogenous autacoids by subtracting the amounts of eicosanoids found in non-spiked biological matrix from the overall quantities found in the same biological matrix spiked with eicosanoid standards.

Total protein assay

Eicosanoid generation in the lung was expressed as nanograms or picograms of analyte per milligram of lung protein per 240 min. The protein concentration in lung homogenate was determined by an assay based on the Bradford dye-binding procedure [28], with BSA as the protein standard and KHB as the blank. The absorbance change of the Coomassie brilliant blue G-250 dye reagent at 595 nm in response to protein concentration was measured by use of a Biomate 5 spectrophotometer (Thermo Spectronic, Rochester, New York, USA).

Isolated perfused and hypoxically ventilated rabbit lung

As previously described [29], lungs were excised from anaesthetized rabbits, rinsed with KHB for washout of blood, placed in a temperature-equilibrated housing chamber thermostatted at 38.5°C, perfused at 150 mL min−1 and ventilated with a mixture of 5.3% CO2, 21.0% O2, balance N2 (tidal volume, 30 mL; frequency, 30 strokes min−1, positive end-expiratory pressure, 1 cm H2O). Pulmonary arterial pressure and lung weight were recorded continuously. For hypoxic maneuvers, a gas mixing chamber (KM 60–3/6MESO; Witt, Witten, Germany) was used, this enabled gradual changes in O2 content from 21% v/v (pO2 ≈160 mmHg, baseline conditions) to 3% v/v (pO2 ≈23 mmHg, hypoxic conditions), 5.3% v/v CO2, with the balance N2. After an initial steady-state period of 30 min, time was set to zero and lungs were ventilated with either normoxic or hypoxic gas for 240 min, then homogenized and employed for mediator analysis and protein assay.

Statistics

Analytical data and pulmonary arterial pressure rise (ΔPAP) are expressed as mean ± SEM. Statistical evaluation was performed using one-way analysis of variance (ANOVA). Values of P < 0.05 for eicosanoid profiles and P < 0.01 for ΔPAP were considered statistically significant.

Results and discussion

Injection of 8 μL of a standard mixture containing the investigated 44 eicosanoids on to an isocratically eluted capillary column, resulted in chromatographic separation within 50 min followed by dual online PDA (Fig. 2) and MS2 (Fig. 3) detection of the eluting analytes.

Fig. 2
figure 2

CapLC–PDAD traces of eicosanoid standards. 8 μL of a mixture of AA-derived CYP, LOX, and COX metabolites and iPTs were injected on to a capillary column and eluted isocratically. Online PDAD of eluting eicosanoids enabled simultaneous detection at 303 nm (a), 270 nm (b), 240 nm (c), 216 nm (not shown), and 204 nm (d) and identification by spectrum plot on the peak maximum (e), and peak purity control assisted quantitation. Peak designation: 1 = LXB4; 2 = LXA4; 3 = 15-deoxy-Δ12,14-PGJ2; 4 = 20-COOH-LTB4; 5 = 20-HO-LTB4; 6 = LTD4; 7 = LTC4; 8 = LTE4; 9 = PGB1 = IS; 10 = 6t-LTB4 + 6t,12e-LTB4; 11 = LTB4; 12 = 5(S),6(R)-DiHETE; 13 = 5(S),6(S)-DiHETE; 14 = Δ12-PGJ2; 15 = 12-HHT; 16 = 15-HETE; 17 = 11-HETE; 18 = 8-HETE; 19 = 12-HETE; 20 = 9-HETE; 21 = 5-HETE; 22 = 19(R)-hydroxy-PGE2; 23 = 6-keto-PGE1; 24 = 11-dehydro-TXB2; 25 = 2,3-dinor-8-iso-PGF2α; 26 = 2,3-dinor-TXB2; 27 = 2,3-dinor-6-keto-PGF1α; 28 = 6-keto-PGF1α; 29 = TXB2; 30 = 8-iso-PGF2α; 31 = PGF2α; 32 = 13,14-dihydro-15-keto-PGF2α; 33 = PGE2; 34 = PGD2; 35 = PGJ2; 36 = 14,15-DHET; 37 = 11,12-DHET; 38 = 8,9-DHET; 39 = 20-HETE; 40 = 5,6-DHET; 41 = 5,6-DHET δ lactone; 42 = 14,15-EET; 43 = 11,12-EET; 44 = 8,9-EET; 45 = 5,6-EET; 46 = SDA; 47 = BHT. Single asterisk: PGJ2 (peak 35) was the sole analyte detected at 216 nm. This optical trace is not displayed for reasons of space. Double asterisk: The individual UV spectral maxima were: 270 nm for Peaks 4, 5, 10, and 11 (20-COOH-LTB4, 20-HO-LTB4, 6t-LTB4 + 6t,12e-LTB4, and LTB4, respectively), 275 nm for Peaks 12 and 13 (5(S),6(R)-DiHETE and 5(S),6(S)-DiHETE, respectively), and 280 nm for Peaks 6, 7, and 8 (LTD4, LTC4, and LTE4, respectively)

Full size image
Fig. 3
figure 3

CapLC–ESI–EIC MS2 traces of eicosanoid standards. Serial interfacing of the PDAD with an ion-trap mass spectrometer enabled online MS2 detection of eluting compounds in the negative-ion mode employing MRM. Mediators were identified and quantified by employing EIC MS2 traces defined by summing the intensities of selected substance-specific transitions of the precursor molecular anions. Peak designation is the same as in Fig. 2. Asterisk: The complete EIC trace definitions are: (j) EIC 317; 221; 209; 203; 197; 195; 185; 181; 177; 145; 129; 123 MS2 (335), (l) EIC 335; 317; 309; 299; 291; 281; 273; 263; 255; 247; 235; 223; 219; 217; 209; 195; 193; 191; 171; 165; 111 MS2 (353), (m) EIC 301; 291; 289; 275; 273; 257; 245; 221; 219; 208; 205; 203; 195; 191; 179; 175; 167; 163; 155; 151; 149; 139; 135; 129; 127; 123; 115; 113; 99 MS2 (319), (n) EIC 319; 301; 275; 257; 237; 225; 219; 207; 197; 191; 185; 179; 167; 163; 155; 145; 141; 129; 127; 123; 115; 111 MS2 (337)

Full size image

CapLC–PDAD analysis

The number and position of maxima in the UV spectra of eicosanoids depend on the number and configuration of double bonds in their molecules. These spectrophotometric properties were used both for defining the detection wavelengths of analytes with four (303 nm), three (270 nm), two (240 nm and 216 nm) conjugated double bonds and of substances with isolated double bonds (204 nm), and for UV-spectrometric identification of eluting compounds by employing online UV spectra monitored by the PDAD (Fig. 2). The peak numbering employed in Fig. 2 is used throughout this work and was guided by the well-tried practice of consecutive peak integration on the below presented five UV traces (303 nm, 270 nm, 240 nm, 216 nm, and 204 nm) following the elution order within these optical chromatograms rather than the absolute elution order, which cannot be clearly established because of coelution of autacoids belonging to the same or different UV chromatograms. As specified below, the (partly) coeluting peaks on the same UV traces, for example the peaks 22–34, 4 and 5, and 18 and 19, could be individually assessed by mass-spectrometric detection and were numbered according to their elution order within the individual substance-specific MS2 traces.

Quantitation of the eluting eicosanoids was performed by integrating the peak areas on the optical traces at 303 nm (LXs, 15-deoxy-Δ12,14-PGJ2) (Fig. 2a), 270 nm (LTs, DiHETEs, PGB1 = IS) (Fig. 2b), 240 nm (Δ12-PGJ2, 12-HHT, HETEs) (Fig. 2c), 216 nm (PGJ2) and 204 nm (19(R)-hydroxy-PGE2, 6-keto-PGE1, 11-dehydro-TXB2, 2,3-dinor-8-iso-PGF2α, 2,3-dinor-TXB2, 2,3-dinor-6-keto-PGF1α, 6-keto-PGF1α, TXB2, 8-iso-PGF2α, PGF2α, 13,14-dihydro-15-keto-PGF2α, PGE2, PGD2, 20-HETE, DHETs, EETs, SDA) (Fig. 2d). The optical traces displayed baseline separation for the majority of the eluting compounds, except for peaks 22–34 (22 = 19(R)-hydroxy-PGE2, 23 = 6-keto-PGE1, 24 = 11-dehydro-TXB2, 25 = 2,3-dinor-8-iso-PGF2α, 26 = 2,3-dinor-TXB2, 27 = 2,3-dinor-6-keto-PGF1α, 28 = 6-keto-PGF1α, 29 = TXB2, 30 = 8-iso-PGF2α, 31 = PGF2α, 32 = 13,14-dihydro-15-keto-PGF2α, 33 = PGE2, and 34 = PGD2) at 204 nm, the ω-oxidation products 4 + 5 = 20-COOH-LTB4 + 20-HO-LTB4 at 270 nm and the regioisomers 18 + 19 = 8-HETE + 12-HETE at 240 nm. Individual assessment of these analytes could be readily performed by detection on substance-specific MS2 traces (Fig. 3g,k,l,s,t,u,v,w,x for peaks 22–34, Fig. 3b for 4 = 20-COOH-LTB4, Fig. 3i for 5 = 20-HO-LTB4, Fig. 3q for 18 = 8-HETE, and Fig. 3r for 19 = 12-HETE).

Remarkably, many mediators, for example those with four (LXs, 3 = 15-deoxy-Δ12,14-PGJ2), three (LTs, DiHETEs, 9 = PGB1 = IS, 14 = Δ12-PGJ2), and two (15 = 12-HHT, HETEs, 35 = PGJ2) conjugated double bonds, display very characteristic UV spectra with partly multiple maxima at higher, and thus more specific, wavelengths (peaks 1–21 and 35 in Fig. 2e). These UV spectra were particularly suitable for enhancing the reliability of eicosanoid assessment. Thus, identification of several autacoids with “poor” or identical MS2 spectra is greatly improved or enabled by online monitoring of UV spectra. For instance, the identity of 3 = 15-deoxy-Δ12,14-PGJ2 with only one signal in its MS2 spectrum could be confirmed by its characteristic UV spectrum with a maximum at 307 nm (Fig. 2a and e). The individual assessment of the nearly coeluting isomers 14 = Δ12-PGJ2 and 35 = PGJ2 with identical MS2 spectra was possible only by UV detection, because of their characteristic and distinct UV spectra with maxima at 248 nm and 216 nm, respectively (Fig. 2c,d, and e). Moreover, non-ionic analytes, for example 41 = 5,6-DHET δ lactone (Fig. 2d and e), the availability of the non-ionic antioxidant 47 = BHT (eluting as last peak of the chromatogram and thus not interfering with any eicosanoid peak; Fig. 2d and electronic supplementary material Fig. S6d) as an effective protective agent at the end of the entire analytical procedure, and possible overloading and thus malfunction of the high-sensitivity capillary column by large amounts of polar non-ionic matrix interferences could readily be assessed by PDAD but not by ESI–MS2.

DI-ESI-MS2 and CapLC-ESI-MS2 analysis

The mass-spectrometric raw data acquired with the online interfaced ion trap operated in the negative-ion mode using MRM in the MS2 stage were used for definition of substance-specific extracted ion chromatograms (EIC) for the precursor molecular anions of the eluting eicosanoids. EIC MS2 traces were defined by summing the intensities of several selected substance-specific signals (precursor molecular ion → product/fragment ion transitions) from the MS2 spectra of the precursor molecular ions. The EIC trace definitions used in this work were the same as those employed by the Esquire software and by ion trap users [30]. For instance, EIC 420; 351; 333; 322; 317; 120 MS2 (438) means that this trace was constructed by summing the intensities of the fragment ions at m/z 420, 351, 333, 322, 317, and 120 resulting from MS2 transitions, i.e., MSn fragmentation stage 2 (n = 2, MS2), of the molecular ion at m/z 438 (corresponding to peak 8 = LTE4; Fig. 3e). The selection of product ions in the EIC MS2 trace definition of each eicosanoid was optimized for both maximal EIC signal intensity and substance specificity, thus enabling highly sensitive and individual analysis of all mediators, even of coeluting or chromatographically poorly resolved isomeric autacoids in the presence of unknown interfering isomeric compounds from the biological matrix. Figure 3 displays an overview of all EIC MS2 traces defined according to the encountered chromatographic situations and grouped as follows:

  1. 1.

    individual assessment of non-isomeric compounds (Fig. 3a,b,c,d,e,f,g,h);

  2. 2.

    individual assessment of isomeric non-coeluting mediators (Fig. 3i,j,k,l,m,n); individual assessment of 18 = 8-HETE and 19 = 12-HETE was enabled by the group 3 EIC MS2 trace definitions displayed in Fig. 3q and r, respectively;

  3. 3.

    individual assessment of (nearly) coeluting isomers: (Fig. 3o,p,q,r,s,t,u,v,w);

  4. 4.

    assessment of (nearly) coeluting isomeric autacoids together: (Fig. 3x,y).

Quantification of the eluting eicosanoids was performed by integrating the peak areas on the above listed EIC MS2 traces. Analyte identity and peak purity were checked by comparing the online monitored compound spectrum with the MS2 spectrum of the respective eicosanoid standard obtained offline by DI and/or online by LC–MS2 coupling and stored in the MSn library module of the Esquire v 5.1 software.

Identity and purity of analyte peaks were triple checked, i.e., chromatographically, by isocratic RRI, and by dual UV-spectrometric and MS2 analyses, as demonstrated in Figs. 2, 3, 4 and in electronic supplementary material Figs. S1, S2, S3, S4).

Fig. 4
figure 4

Exemplary MS2 spectra of eicosanoid standards. Online MS2 spectra of eluting 20 = 9-HETE (a), 21 = 5-HETE (b), 42 = 14,15-EET (c), 43 = 11,12-EET (d), 44 = 8,9-EET (e), and 45 = 5,6-EET (f). Eicosanoid numbering is the same as in Fig. 2. Further spectra are provided in the electronic supplementary material

Full size image

The molecular carboxylate ions of the eicosanoids (designated a) yielded non-characteristic fragments due to neutral loss of CO2 (b = a − CO2 = a − 44), H2O (a − H2O = a − 18, a − 2H2O = a − 36, a − 3H2O = a − 54), loss of both CO2 and H2O (a − CO2 − H2O = b − H2O = b − 18, a − CO2 − 2H2O = b − 2H2O = b − 36, a − CO2 − 3H2O = b − 3H2O = b − 54) but also characteristic fragments (designated c, d, e, f etc.) because of ring structures or the positions of double bonds and/or functional groups (hydroxyl, epoxide, cysteinyl), as listed in Table 2.

Table 2 Fragmentation patterns of eicosanoids: m/z values of noncharacteristic fragments resulting from MS2 analysis of the precursor molecular carboxylate anions (a) due to neutral loss of CO2 (b = a − CO2), loss of H2O (a − H2O, a − 2H2O, a − 3H2O), or loss of CO2 and H2O (b − H2O, b − 2H2O, b − 3H2O), and m/z values of characteristic fragment ions resulting from MS2 analysis of the precursor molecular ions due to ring structures or the position of double bonds and functional groups (hydroxyl, epoxide, cysteinyl)
Full size table

MS2 of eicosanoids

The MS2 spectra of 12-HHT (a at m/z 279), the DHETs ( a at m/z 337), and 20-HETE (a at m/z 319) were similar to those described elsewhere [21, 23, 30]. The spectra of 15 = 12-HHT, 36 = 14,15-DHET, 37 = 11,12-DHET, 38 = 8,9-DHET, 39 = 20-HETE, and 40 = 5,6-DHET are displayed in electronic supplementary material Figs. S1a,b,c,d,e, and f, respectively. We found that the signal at m/z 273 in the MS2 spectrum of 20-HETE was presumably formed by loss of carbon dioxide followed by loss of two hydrogen atoms (d = b − 2H) as suggested by the analogous signal at m/z 279 (273 + 6) generated by MS2 analysis of the carboxylate ion (m/z 325) of [2H6]20-HETE with the deuterium atoms at positions 16, 16′, 17, 17′, 18, and 18′.

The MS2 spectra of the LXs LXB4 (a at m/z 351) and LXA4 (a at m/z 351), the LTs 20-COOH-LTB4 (a at m/z 365), 20-HO-LTB4 (a at m/z 351), LTD4 (a at m/z 495), LTC4 (a at m/z 624), LTE4 (a at m/z 438), 6-trans-LTB4, 6-trans,12-epi-LTB4 and LTB4, and of 5(S),6(R)-DiHETE and 5(S),6(S)-DiHETE (a at m/z 335, for LTB4 isomers and DiHETEs) were similar to those reported elsewhere [31–39]. The spectra of 1 = LXB4, 2 = LXA4, 4 = 20-COOH-LTB4, 5 = 20-HO-LTB4, 6 = LTD4, 7 = LTC4, and 8 = LTE4, 11 = LTB4, are shown in electronic supplementary material Figs. S2a,b,c,d,e,f, and S3a,b, respectively.

The MS2 spectra of the molecular ions (a at m/z 319) of the HETEs showed signals similar to those described previously [21, 22, 30, 32, 34, 35, 37, 40–43]. The spectra yielded by MS2 analysis of 16 = 15-HETE, 17 = 11-HETE, 18 = 8-HETE, 19 = 12-HETE, and 20 = 9-HETE, 21 = 5-HETE are presented in electronic supplementary material Fig. S3c,d,e,f and Fig. 4a,b, respectively.

Fragmentation of the molecular ions of the EETs (a at m/z 319) yielded MS2 spectra with signals similar to those reported elsewhere [21, 23, 30, 40, 44]. The spectra of 42 = 14,15-EET, 43 = 11,12-EET, 44 = 8,9-EET, and 45 = 5,6-EET are displayed in Fig. 4c,d,e, and f, respectively.

The MS2 spectra of the prostanoids 6-keto-PGF1α (a at m/z 369), 2,3-dinor-6-keto-PGF1α (a at m/z 341), 6-keto-PGE1 (a at m/z 367), TXB2 (a at m/z 369), 11-dehydro-TXB2 (a at m/z 367), 2,3-dinor-TXB2 (a at m/z 341), PGE2 (a at m/z 351), 19(R)-hydroxy-PGE2 (a at m/z 367), PGD2 (a at m/z 351), PGJ2 (a at m/z 333), Δ12-PGJ2 (a at m/z 333), 15-deoxy-Δ12,14-PGJ2 (a at m/z 315), PGF2α (a at m/z 353), 13,14-dihydro-15-keto-PGF2α (a at m/z 353), were similar to those described elsewhere [4, 37, 45, 46]. The spectra of 28 = 6-keto-PGF1α, 29 = TXB2, 33 = PGE2, 35 = PGJ2, and 34 = PGD2 are illustrated in electronic supplementary material Figs. S4a,b,d,e, and f, respectively.

Comparative MS2 analysis of 33 = PGE2 and 34 = PGD2 was of particular importance for simultaneous and individual assessment of both compounds. Offline MS2 analysis of PGD2 (electronic supplementary material Fig. S4f ) yielded characteristic fragments due to cleavage between C14 and C15 (loss of hexanal) with gain of a hydrogen and charge retention on the carboxyl terminus (c at m/z 251), prior loss of water from the cyclopentane ring followed by cleavage between C14 and C15 with gain of a hydrogen and charge retention on the carboxyl terminus (d at m/z 233), and an ion of unknown origin (e at m/z 189). Comparison of the MS2 spectrum of PGE2 (electronic supplementary material Fig. S4d) with that of PGD2 showed no difference with the exception of the fragment ion at m/z 251, which is specific for PGD2 but is lacking in the MS2 spectrum of PGE2. This may be accounted for by facilitation of the prior loss of water from the cyclopentane ring due to the presence of the hydroxyl at C11 in PGE2, which leads to the formation of a double bond conjugated with the double bond at C13. This conjugation is not achieved for PGD2, because C11 is occupied by a ketone, thus enhancing the probability of formation of fragment c at m/z 251 by direct loss of hexanal without prior loss of water. The signal at m/z 251 is therefore rather characteristic of PGD2 and may serve as a basis for individual identification and quantitation of PGD2 and PGE2.

PGE2 was quantified by subtracting the amount of PGD2 obtained by integration of peak 34 on the EIC 251 MS2 (351) trace (Fig. 3w) from the total quantity of PGE2 and PGD2 taken together obtained by integration of peaks 33 + 34 on the EIC 251; 233; 189 MS2 (351) trace (Fig. 3x). This is possible only in the absence of 1 = LXB4 and 2 = LXA4, because the signals at m/z 251, 233, and 189 are also present in the spectra of the nearly coeluting LXs (Table 2, and compare electronic supplementary material Figs. S4d and f for the spectra of PGE2 and PGD2, respectively, with electronic supplementary material Figs. S2a and b for the MS2 spectra of LXB4 and LXA4, respectively). Conversely, quantitation of LXB4 and LXA4 by integration of peak 1 on the EIC 221; 163; 129 MS2 (351) trace (Fig. 3o) and peak 2 on the EIC 217; 145; 135 MS2 (351) trace (Fig. 3p), respectively, is easily possible, even in the presence of PGE2 and PGD2, because the characteristic signals enumerated in the EIC definitions of the two lipoxins are not shared by the two prostaglandins. As the LXs mostly occur, if at all, in very small amounts, the assessment of PGE2 and PGD2 is usually unproblematic.

The offline MS2 spectrum of 14 = Δ12-PGJ2 was identical with that of its nearly coeluting isomer 35 = PGJ2 (electronic supplementary material Fig. S4e; see also Table 2 for MS2 signal comparison). Therefore, only MS2 identification and quantitation of PGJ2 and Δ12-PGJ2 taken together (peak 14 + 35 = Δ12-PGJ2 + PGJ2 in Fig. 3y), rather than individually was possible. Fortunately, in contrast to MS2 detection, individual identification and quantitation of these two isomers could easily be performed by PDAD because of their characteristic and different UV spectra, as depicted in Fig. 2c–e (see the section CapLC-PDAD analysis, above).

The MS2 spectrum (not shown) of 3 = 15-deoxy-Δ12,14-PGJ2 displayed no characteristic signals and only one non-characteristic fragment at m/z 271 (Table 2). Fortunately, because of its specific UV spectrum (peak 3 in Fig. 2a,e), identification and quantitation reliability could be improved by additional complementary PDAD (see the section CapLC-PDAD analysis).

The MS2 spectra yielded by fragmentation of the molecular ions of the isoprostanes 30 = 8-iso-PGF2α (a at m/z 353) (electronic supplementary material Fig. S4c) and its metabolite 25 = 2,3-dinor-8-iso-PGF2α (a at m/z 325) (not shown) were similar to those reported [4, 45].

Assessment of calibration and recovery

Calibration of the UV and MS signals performed by the IS method, employing PGB1 as a common IS for all eicosanoids, yielded linear regression functions for all AA metabolites (exemplified for MS2 signal calibration in electronic supplementary material Fig. S5).

The described techniques of sample extraction followed by CapLC with dual online PDA and MS2 detection were applied to the biological matrix (homogenized samples of blood-free rinsed rabbit lung) and to KHB, both spiked with eicosanoid standards and with deuterated eicosanoids and/or PGB1 as IS, as specified in the Experimental section, yielding linear recovery plots with recoveries >75% (electronic supplementary material Table S2).

Comparison of the quantitation results obtained by adding PGB1 and deuterated eicosanoids as IS before extraction, prior to any manipulation, to both biological matrix (homogenate from blood-free rinsed lungs) and KHB, revealed no significant differences (RSD <16%), demonstrating that PGB1 and the deuterated eicosanoids were equally suitable as IS. Moreover, the ratios between the recovery of PGB1 and the recoveries of all eicosanoids remained unchanged during sample preparation.

It is worthy of note that comparison of the MS2 spectra of eicosanoid standards, including PGB1 and deuterated eicosanoids, mixed with the biological matrix and subjected to SPE (not shown) and of the MS2 spectra of endogenous autacoids from lung homogenate (electronic supplementary material Figs. S8 and S9) with those of the corresponding non-extracted standards (Fig. 4 and electronic supplementary material Figs. S1–S4) exhibited no matrix-related signal-suppression effects. Signal suppression was also checked for all analytes, including internal standards, by comparison of the signals generated by known amounts of standards diluted with eluate from the final step of SPE of the biological matrix or KHB sample extraction and the same standard amounts diluted with the mobile phase or water. The observed lack of MS2 signal suppression was accomplished by elimination of matrix interferences employing source-specific sample extraction and by the use of a mobile phase containing few and volatile ionic ingredients at very low concentrations. These findings correlated well with the fact that comparison of recovery assessment with PGB1 and deuterated eicosanoids as IS, employing biological samples as matrix, revealed no significant differences.

As a conjugated triene with high extinction coefficient, PGB1 was best suited as IS not only for MS2 detection but also for sensitive PDAD quantitation. Thus, we ultimately decided on the less expensive and simpler use of the most suitable PGB1 as a common IS for both quantitation of all analytes and, together with SDA as a reference substance for the calculation of isocratic RRI using Eq. 1. Addition of PGB1 as an IS to biological samples prior to any manipulation enabled correction for loss of substance occurring during sample handling and enabled accurate measurement of the original amounts of eicosanoids in the biological source.

With 300 μm i.d. columns, the limits of quantification (LOQ) for MS2 detection defined as the on-column analyte amount at a signal-to-noise (S/N) ratio greater than ten were: 100 fg for LTs, LXs, PGs, TXs, and iPTs, 300 fg for DHETs, 12-HHT, and DiHETEs, 1 pg for 20-HETE and cis-trans conjugated HETEs, and 2 pg for EETs. With 150 μm i.d. columns, the LOQ for MS2 detection were a factor of 4 lower. Overall, the LOQ for PDAD were higher than for MS2 detection as follows: by a factor of 15 for EETs, DHETs, PGs, TXs, iPTs, and 20-HETE, a factor of 5 for 12-HHT, Δ12-PGJ2, and cis-trans conjugated HETEs, and a factor of 3 for LTs, DiHETEs, LXs, and 15-deoxy-Δ12,14-PGJ2. RSD ranged from 6 to 16% (n = 6).

The actual recovery of each compound was checked with quality control samples freshly prepared prior to SPE of a sample batch from standard stock solutions diluted to the final sample volume with biological matrix and/or KHB, and processed with each set of samples under the same conditions. Moreover, to assess possible non-enzymatic hydrolytic turnover of EETs into the corresponding DHETs during the extraction procedure, quality-control samples containing solely the four EET regioisomers were included in each sample sequence. We found that after storage of dried samples for 24 h at −85°C under an inert N2 atmosphere the percentages of hydrolyzed EETs (as measured by formation of the corresponding DHET regioisomers and, in the case of 5,6-EET, of 5,6-DHET δ lactone also) were much smaller than the RSD of quantification (6–16%) and thus negligible: 0.77% (14,15-EET), 0.52% (11,12-EET), 0.54% (8,9-EET), 2.31% (5,6-EET). Consequently, it can be assumed that the measured EET and DHET amounts reflect the in vivo status quo.

The storage time for biological material and samples was limited to a secure maximum of 24 hours at −85°C with the aim of reducing loss of the thermolabile and reactive EETs (especially the unstable 5,6-EET), which were the focus of this study, to a minimum. These storage conditions were estimated on the basis of maintaining negligibly low hydrolysis rates of EETs to the corresponding DHETs, as shown above. A clear tendency to enhanced hydrolytic turnover was observed after several days of storage. However, we found that within a storage period of a week at −85°C no significant (RSD <16%) decrease of the recoveries of the other eicosanoids occurred. Thus, if the target analytes do not include the EETs, the storage time may be extended to a couple of days.

Features of the developed CapLC–PDAD–ESI–MS2 technique

In the past three decades HPLC has developed into a separation technique that provides the highest sensitivity by reducing column i.d. to capillary dimensions [26, 47, 48], thus becoming the method of choice for eicosanoid analysis. The use of PDAD is favored by the fact that, as specified above, many mediators carry strong chromophores, e.g., conjugated double bonds, generating specific UV spectra [18–20]. However, at low analyte concentrations spectrophotometric detection may be difficult, especially for autacoids with weak chromophores, e.g., isolated double bonds. As the assessment of several mediators is enabled and/or greatly improved by PDAD, dual optical and mass-spectrometric detection proves to be a reliable and sensitive solution for all situations.

The present method employed source-specific SPE with antioxidative protection of analytes and robust isocratic elution of the highly efficient capillary column with a mobile phase of low optical absorption and ionic background noise, followed by online dual PDA and MS2 detection enabling reliable and highly sensitive assessment of all relevant mediators of the AA cascade and iPTs from complex biological material. The ion-trap technology employed for MS2 detection is well suited to lipid mediator analysis [21, 30, 38, 49] and equal in terms of mass accuracy and quantitative sensitivity with triple quadrupole and time-of-flight mass spectrometry [50].

The main characteristics of this technique are:

  1. 1.

    comprehensive and rapid analysis of all representative mediators of the AA cascade and iPTs in one chromatographic run of 50 min;

  2. 2.

    analytical reliability, accuracy and ruggedness;

  3. 3.

    maximal overall quantitative sensitivity.

Comprehensive and high-throughput CapLC-based analysis of eicosanoids enabled estimation of the relative and thus actual vasoregulative contribution of the free CYP/sEH-mediated 20-HETE, EETs/DHETs in the context of all the other relevant vasoactive mediators of the AA cascade including the cis–trans conjugated HETEs, the LOX-mediated LXs and LTs, the COX-mediated TXs, PGs, and 12-HHT, and the free iPTs as both vasoactive mediators and in-vivo markers of ROS-induced oxidative stress/lipid peroxidation by direct comparison of all relevant autacoid generation profiles.

Comprehensive mediator assessment, i.e., analyzing both short-lived primary autacoids and their longer-lived metabolites or degradation products, enables:

  1. 1.

    biochemically accurate mediator profiling by summing the amounts of related metabolites or employing selected representatives. The groups of metabolically related eicosanoids are (generation order; first compound in each group is the precursor): 30 = 8-iso-PGF2α, 25 = 2,3-dinor-8-iso-PGF2α; PGI2, 28 = 6-keto-PGF1α, 27 = 2,3-dinor-6-keto-PGF1α, 23 = 6-keto-PGE1; TXA2, 29 = TXB2, 26 = 2,3-dinor-TXB2, 24 = 11-dehydro-TXB2; the EETs and the corresponding DHETs; 11 = LTB4, 5 = 20-HO-LTB4, 4 = 20-COOH-LTB4; 7 = LTC4, 6 = LTD4, 8 = LTE4; 33 = PGE2, 22 = 19(R)-hydroxy-PGE2; 34 = PGD2, 35 = PGJ2, 14 = Δ12-PGJ2, 3 = 15-deoxy-Δ12,14-PGJ2; 31 = PGF2α, 32 = 13,14-dihydro-15-keto-PGF2α (see also Fig. 1).

  2. 2.

    quantitation of individual factors of interest, for example 22 = 19(R)-hydroxy-PGE2, 35 = PGJ2, 14 = Δ12-PGJ2, and 3 = 15-deoxy-Δ12,14-PGJ2, which have quite specific biological actions, i.e., different physiological functions compared with the respective precursor substance, and can thus not be considered as mere metabolites but rather as physiologically distinct mediators.

In view of the interrelations between COX and CYP [6–8], use of 15 = 12-HHT as an indicator of both TXA2 synthase [51] and COX [52] activity and as a marker of prostacyclin (PGI2) generation/PGI2-dependent relaxation [53] is of special informative importance.

Analytical reliability, accuracy, and ruggedness resulted from:

  1. 1.

    optimal chromatographic resolution of all eicosanoids due to column choice;

  2. 2.

    separation of analytes with reliable selectivity and optimal peak density distribution in a minimal runtime by isocratic elution with a mobile phase adapted to the column used;

  3. 3.

    quantification of eicosanoids by a simple and reliable IS method;

  4. 4.

    robust isocratic elution ensuring, in contrast to gradient elution, maximal reproducibility of retention times and thus, in combination with RRI as selectivity parameter, accurate chromatographic characterization and identification of analytes; followed by

  5. 5.

    dual consecutive online PDA and MS2 detection providing a wealth of complementary structural information contained in UV and MS2 spectra and thus enabling both maximal analytical reliability for all eicosanoids and simultaneous analysis of a broader array of compounds than with ESI–MS2 detection alone.

Maximal overall quantitative sensitivity with LOQ for ESI–MS2 detection reaching down to the femtogram range, a key advantage of the presented method, was achieved by:

  1. 1.

    dramatic chromatographic signal gain by use of capillary columns with high peak efficiency instead of large-i.d. columns with low signal yield employed by previous methods;

  2. 2.

    as a result of the inherently low capillary flow rates (0.8–3 μL min−1), virtually complete substance transfer to the mass spectrometer, in contrast to only partial substance transfer at high flow rates (250–1000 μL min−1) typical of large-i.d. columns employed by previous methods;

  3. 3.

    development of a mobile phase with both low ionic background and high optical transparency, due to the use of volatile ionic ingredients at very low concentrations and solvents with low optical absorption, thus increasing the S/N ratio for both MS2 and PDA detection;

  4. 4.

    increasing the amount of substance on the capillary column by large-volume injection; and

  5. 5.

    increasing the S/N ratio by source-specific sample extraction.

It is worthy of note that, because of its high quantitative sensitivity, this CapLC method enabled assessment of free EETs, DHETs, 20-HETE, and iPTs, i.e., active mediators not esterified into cellular membrane phospholipids, typically occurring at much lower concentrations in biological sources compared with the corresponding autacoids incorporated into phospholipid pools.

Eicosanoid profiling of lungs by global CapLC–PDAD–ESI–MS2 analysis

Application of the method to lung homogenate demonstrated that ex-vivo perfused rabbit lungs ventilated under normoxic or hypoxic conditions generated virtually all relevant eicosanoids of the AA cascade including the CYP-mediated free EETs and 20-HETE, the sEH-mediated free DHETs, the LOX-mediated LTs, LXs, and DiHETEs, the COX-mediated TXs, PGs, and 12-HHT, the cis–trans conjugated HETEs, and free iPTs, as shown in electronic supplementary material Fig. S6 (optical traces and UV spectra), Fig. S7 (MS2 traces), and Figs. S8 and S9 (representative MS2 spectra).

A representative profile of eicosanoid formation in hypoxically vs. normoxically ventilated lungs is presented in Fig. 5. The autacoids are grouped according to the three enzymatic generation pathways of the AA cascade—CYP, LOX, and COX. The different eicosanoid classes are represented by selected autacoids as typical examples: EETs/DHETs by their 14,15 and 11,12-regioisomers (42 = 14,15-EET, 36 = 14,15-DHET, 43 = 11,12-EET, 37 = 11,12-DHET), LTs by 8 = LTE4, cis–trans conjugated HETEs by 21 = 5-HETE, PGs by 31 = PGF2α, 33 = PGE2, 34 = PGD2, and PGI2, and TXs by TXA2. The iPTs 30 = 8-iso-PGF2α and 25 = 2,3-dinor-8-iso-PGF2α are displayed in the left inset. 1 = LXB4 and 2 = LXA4 were only occasionally found at the detection limit and thus did not disturb assessment of PGE2 and PGD2.

Fig. 5
figure 5

Global profiling of eicosanoid generation (“eicoprofiling”) in isolated perfused rabbit lungs ventilated for 4 h under hypoxic or normoxic control conditions—exemplary formation profiles of selected mediators from homogenate. The three main enzymatic pathways of the AA cascade are displayed and represented by selected autacoids: the CYP pathway by 39 = 20-HETE, 42 = 14,15-EET/ 36 = 14,15-DHET, and 43 = 11,12-EET/37 = 11,12-DHET, the LOX pathway by 21 = 5-HETE and 8 = LTE4, and the COX pathway by 15 = 12-HHT, 31 = PGF2α, 34 = PGD2, 33 = PGE2, TXA2, and prostacyclin (PGI2). The unstable prostanoids PGI2 and TXA2 were measured as their longer-lived metabolites 28 = 6-keto-PGF1α, 27 = 2,3-dinor-6-keto-PGF1α, 23 = 6-keto-PGE1, and 29 = TXB2, 26 = 2,3-dinor-TXB2, 24 = 11-dehydro-TXB2, respectively. The quantitation of 31 = PGF2α includes its metabolite 32 = 13,14-dihydro-15-keto-PGF2α. The left inset displays the iPT (30 = 8-iso-PGF2α and its metabolite 25 = 2,3-dinor-8-iso-PGF2α) concentration as an in-vivo indicator of the overall ROS level (displayed in percent on the columns; the normoxic ROS level was set to 100%) in the homogenate of hypoxic and normoxic lungs. Eicosanoid numbering is the same as in Fig. 2. Asterisk: P < 0.05 versus control (normoxia). The right inset shows the pulmonary arterial pressure rise (ΔPAP) in hypoxic lungs as a measure of hypoxic pulmonary vasoconstriction (HPV). Double asterisks: P < 0.01 vs control (normoxia). Mean ± SEM; n = 6

Full size image

Although the CYP-derived EETs are well-known as vasodilatory mediators and putative chemical correlates of EDHF in the cardiovascular compartment [11–14], little is yet known about their vasoregulative role in the lung. However, as previously described [54], these oxiranes were the dominant eicosanoids in the human lung vasculature under conditions of increased pulmonary arterial pressure. Interestingly, application of this technique to isolated perfused lungs revealed substantially enhanced generation of free EETs and their hydration products, the free DHETs, under conditions of hypoxic pulmonary vasoconstriction (HPV, as assessed by ΔPAP) as compared to normoxic control ventilation conditions (Fig. 5).

The hypothetical relative and thus actual contribution of the vasoconstrictive and vasodilatative eicosanoids to HPV can be assessed by direct comparison of the generation profiles of all relevant AA-derived mediators under hypoxic vs. normoxic conditions:

  1. 1.

    Intriguingly, in contrast with the oxiranes, the level of the strong general vasoconstrictor TXA2 decreased considerably in hypoxic lungs. The levels of the vasoconstrictive free iPTs also decreased, indicating general attenuation of ROS-induced oxidative stress due to decreasing oxygen concentration [3] under hypoxic vs. normoxic ventilation. Moreover, the overall reduced ROS level in hypoxic vs. normoxic ventilated lungs, as assessed by the in-vivo generation of iPTs well matched previous findings obtained by electron spin resonance measurements [55], thus demonstrating the reliability of both methods. Furthermore, the production of the vasoconstrictive LTE4, representative of the LTs, increased, but at very low, negligible concentrations. Consequently, 5-HETE and PGF2α were the sole vasoconstrictor eicosanoids displaying increasing levels in response to hypoxia and thus contributory to HPV.

  2. 2.

    Remarkably, the levels of the pulmonary vasodilator 39 = 20-HETE [6] and the potent general vasodilator PGI2 remained unchanged, while the two vasodilators PGD2 and PGE2 displayed opposite generation profiles—increased production of the former, indicating a possible counteraction of HPV, and decreased formation of the latter, suggesting an indirect contribution to HPV. Thus virtually no overall indirect contribution of the vasodilatative mediators of the AA cascade to HPV was observed.

The unchanged level of 15 = 12-HHT clearly correlated with the likewise constant PGI2 level, thus confirming its above mentioned role as a prostacyclin inducer and indicator. As a marker of COX activity, 12-HHT also roughly reflects the overall production of TXs and PGs taken together under hypoxic vs. normoxic conditions: increasing amounts of PGF2α and PGD2 virtually compensated by decreasing amounts of PGE2 and TXA2.

Overall, direct comparison of the formation profiles of all relevant vasoactive autacoids has led us to hypothesize that EETs/DHETs may make a substantial contribution to the prolonged HPV response. This presumption correlates well with first reports on vasoconstrictive effects of oxiranes in the pulmonary vasculature [7], on attenuation of HPV by CYP inhibitors [6], and on the hypoxia-induced increase in CYP expression and EET production paralleled by oxirane-elicited HPV, hypertension, and vascular remodeling in mouse lungs [56]. Naturally, the mechanisms of regiospecific and/or enantiospecific EET activity have to be elucidated by further investigations.

Conclusions

The combined method of source-specific SPE followed by isocratic CapLC–PDAD–ESI–MS2 enables simultaneous and expeditive, robust, reliable, and highly sensitive analysis of all biological relevant mediators of the AA cascade, and free iPTs, in one chromatographic run of 50 min, thus enabling rapid, comprehensive, and accurate correlations between the biosynthesis at typically very low concentrations and the bioactivity of eicosanoids against the background of varying ROS levels by direct comparison of all autacoid generation profiles. To the best of our knowledge, this is the first method suited to accomplishing this task. Application of this approach to homogenates of hypoxically ventilated rabbit lungs revealed at a glance the hypothetical contribution of free CYP/sEH-derived EETs/DHETs to HPV/hypoxic pulmonary hypertension, thus demonstrating the suitability of the method for global “eicoprofiling”.