Professional Documents
Culture Documents
Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma
a r t i c l e i n f o a b s t r a c t
Article history: Over the past decade, the number of known eicosanoids has expanded immensely and we have now
Received 1 May 2014 developed an ultra-performance liquid chromatography–electrospray ionization triple quadrupole mass
Received in revised form 25 June 2014 spectrometric (UPLC-QTRAP/MS/MS) method to monitor and quantify numerous eicosanoids. The UPLC-
Accepted 6 July 2014
QTRAP/MS/MS approach utilizes scheduled multiple reaction monitoring (MRM) to optimize sensitivity,
Available online 12 July 2014
number of metabolites that can be analyzed and the time requirement of the analysis. A total of 184
eicosanoids including 26 deuterated internal standards can be separated and monitored in a single 5 min
Keywords:
UPLC run. To demonstrate a practical application, human plasma samples were analyzed following solid-
Eicosanoids
Ultra-performance liquid chromatography
phase extraction (SPE) and the recovery rate and matrix effects were determined for the 26 deuterated
Tandem mass spectrometry internal standards added to the plasma. The method was validated and shown to be sensitive with the
Plasma limit of quantitation at pg levels for most compounds, accurate with recovery rates of 70–120%, and
Lipidomics precise with a CV < 30 for all compounds. Also, the method showed a linear response over a range spanning
several orders of magnitude. In a QC human plasma sample, we identified and rigorously quantified over
120 eicosanoids.
© 2014 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.chroma.2014.07.006
0021-9673/© 2014 Elsevier B.V. All rights reserved.
Y. Wang et al. / J. Chromatogr. A 1359 (2014) 60–69 61
This complete enzymatic system produces hundreds of (ACN), methanol (MeOH), and water were obtained from Fisher
eicosanoids derived from AA and related polyunsaturated fatty Scientific. Isopropanol (IPA) was purchased from Sigma–Aldrich.
acids with very similar structures, chemistries, and physical Formic acid (FA) was obtained from EMD Technologies. Dulbecco’s
properties [7], which makes the analysis of eicosanoids a chal- phosphate-buffered saline (DPBS) was obtained from Corning Life
lenging task, especially in biological samples. The concentration Science.
of eicosanoids in plasma or serum is the lowest among all endoge-
nous lipid metabolites [8,9]. However, under certain conditions,
the plasma level of eicosanoids may change considerably and thus, 2.2. Sample preparation
eicosanoids may serve as a useful readout reflecting disease pro-
gression [10]. As a result, current research is focused on developing 2.2.1. Preparation of primary standard and internal standard
fast, sensitive, and reliable methods that accurately profile and solutions
quantify eicosanoid biomarkers [11,12]. For the preparation of calibration curves, stock solutions were
In the past, eicosanoids were mainly analyzed by enzyme-linked prepared in ethanol that contained all eicosanoid standards, each
immunosorbent assays (EIA) [13,14], gas chromatography–mass at a concentration of 0.25 ng/L. Working standard solutions for
spectrometry (GC–MS) [15], and liquid chromatography–mass all eicosanoids were prepared by serial dilution of the stock solu-
spectrometry (LC–MS) [16]. The drawback of EIA is a lack of tions to create the necessary concentrations. A solution containing
specificity and the ability to determine multiple analytes in 26 internal (deuterated) eicosanoid standards was prepared at
a single set of analyses. GC–MS provides greater sensitivity 0.01 ng/L in ethanol. All solutions were stored at −80 ◦ C until
and selectivity for eicosanoid analysis, but requires chemical use.
derivatization steps that limit its application. The rapid progress
of liquid-chromatography–electrospray ionization tandem mass 2.2.2. Extraction of metabolites from plasma and tissue
spectrometry (LC–ESI-MS/MS) has facilitated the use of this Aliquots of 20 L control plasma (Human Source Plasma,
technology for accurate monitoring of eicosanoid metabolites Gemini Bio-Products) were diluted to 1 mL with phosphate salt
in biological samples [7]. Previous reports include liquid–liquid buffer spiked with 100 L of the internal standard solution. The
extraction for the determination of PGE2 and LTB4 in plasma using eicosanoids were extracted using Strata-X reversed-phase SPE
LC–MS [17], the analysis of a limited number of PGs and LTs in columns (8B-S100-UBJ, Phenomenex). Columns were washed with
cell culture media by LC–MS [18], an on-line two-dimensional 3 mL of MeOH and then equilibrated with 3 mL of H2 O. After loading
reversed-phase LC–MS for the simultaneous determination of the sample, the columns were washed with 10% MeOH to remove
PGE2 , PGF2a , and 13,4-dihydro-15-keto PGF2a [19], and an LC–MS impurities, and the metabolites were then eluted with 1 mL of
method for the simultaneous determination of 23 eicosanoids MeOH and stored at −80 ◦ C to prevent metabolite degradation.
[20]. We previously developed a widely used and comprehensive Prior to analysis, the eluant was dried under vacuum and redis-
high-throughput lipidomic analysis of eicosanoids using LC–MS to solved in 50 L of the UPLC solvent A (water/acetonitrile/acetic acid
monitor 141 unique species in a single 22 min analysis [21] and (60:40:0.02; v/v/v)) for UPLC/MS/MS analysis.
applied it to investigate the human plasma lipidome [8,10]. For the analysis of tissue samples, the wet weight is accurately
Subsequently, a UPLC–MS platform that enables profiling of 122 determined and at least 2 mg of tissue (wet weight) is homogenized
eicosanoids from human whole blood [22] was developed using an in 10% methanol. The eicosanoids are extracted from homogenates
LC–MS/MS approach similar to our previous protocol [7,21]. More by SPE following the identical purification protocol as for the
recently published protocols include a targeted HPLC–MS/MS anal- plasma samples.
ysis platform for 100 oxylipins and 36 oxylipins was detected from
250 L human plasma in 26 min [23], an LC–MS/MS method for
rapid and concomitant quantification of 26 PUFA metabolites from 2.3. UPLC–MS/MS
Caco-2 cells [24], an LC–MS/MS for the simultaneous analysis of
arachidonic acid and 32 related metabolites in 1 mL human plasma An Acquity UPLC system (Waters Corp.) was used. Reversed-
[25], an online HPLC–MS/MS analyzed more than 20 different oxi- phase separation was performed on an Acquity UPLC BEH shield
dized fatty acids and their precursors from 200 L plasma or urine RP18 column (2.1 × 100 mm; 1.7 m; Waters). The mobile phase
[26], a fast LC–MS/MS method including on-line SPE and LIT frag- consisted of (A) ACN/water/acetic acid (60/40/0.02, v/v) and (B)
ment confirmation for the profiling of 7 PUFAs and 94 oxidized ACN/IPA (50/50, v/v). Gradient elution was carried out for 5 min
metabolites from 200 L plasma [27]. However, these methods all at a flow rate of 0.5 mL/min. Gradient conditions were as follows:
need much larger biological samples and/or longer analysis times to 0–4.0 min, 0.1–55% B; 4.0–4.5 min, 55–99% B; 4.5–5.0 min, 99% B;
generally detect and quantify fewer lipid species. These constraints A 10 L aliquot of each sample was injected onto the column. The
render the application of these methods to large studies and with column temperature was kept at 40 ◦ C. All samples were kept at
often limited material impractical. 4 ◦ C throughout the analysis.
In order to address the need for large-scale high-throughput Mass spectrometry was performed on an ABI/Sciex 6500 QTRAP
analysis of eicosanoids in small quantities of human plasma and tis- hybrid, triple quadrupole, linear ion trap mass spectrometer
sue samples, we now report the development of a method utilizing equipped with a Turbo V ion source. Eicosanoids were detected
an ultra-high-performance liquid chromatography-QTRAP MS/MS in negative electrospray ion mode. Curtain gas (CUR), nebulizer gas
(UPLC–MS/MS) for monitoring 184 eicosanoids in a 5 min run. The (GS1), and turbo-gas (GS2) were set at 10 psi, 30 psi, and 30 psi,
new methodology is validated by identification and quantitation of respectively. The electrospray voltage was −4.5 kV, and the turbo
eicosanoids in 20 L human plasma. ion spray source temperature was 525 ◦ C.
Eicosanoids were analyzed using scheduled multiple reaction
2. Experimental monitoring (MRM). Mass spectrometer parameters including the
declustering potentials and collision energies were optimized for
2.1. Reagents each analyte. Nitrogen was employed as the collision gas. Data
acquisitions were performed using Analyst 1.6.2 software (Applied
All eicosanoids and deuterated internal standards were pur- Biosystems). Multiquant software (Applied Biosystems) was used
chased from Cayman Chemical. Optima LC–MS grade acetonitrile to quantify all metabolites.
62 Y. Wang et al. / J. Chromatogr. A 1359 (2014) 60–69
2.4. Analytical validation It is assumed that fatty acids and their oxidized derivatives are
either part of circulating lipoprotein particles and/or bound to albu-
2.4.1. Linearity and LOD and LOQ min [29]. Protein will denature with MeOH as eluent of SPE and
A typical standard curve was prepared by adding 1 ng of each three-dimensional structure of protein will be destroyed. So the
internal (deuterated) eicosanoid standard to the following amounts elution steps may dissociate the attachment of fatty acid and pro-
of eicosanoid (non-deuterated) primary standards: 0.005, 0.015, tein.
0.025, 0.035, 0.05, 0.15, 0.25, 0.35, 0.5, 1.5, 2.5, 3.5, and 5.0 ng. Linear A crucial aspect of our method is the inclusion of 26 deuter-
regression analysis of the response ratios (peak area of analyte/peak ated internal standards. All samples are spiked with a mixture of
area of internal standard) obtained from the calibration curve was deuterated internal standards prior to lipid extraction. An internal
used to calculate the corresponding eicosanoid amounts in sam- standard is used to correct for variations in extraction efficiency, for
ples. The limit of detection (LOD) was defined as the concentration monitoring the chromatographic response, and for normalization
that resulted in a peak with a signal-to-noise ratio (S/N) greater purpose which allows for accurate quantification. This approach,
than 3:1 (3 S/N) and the limit of quantitation (LOQ) was defined as also known as the isotope dilution method, works on the princi-
(S/N) greater than 7:1 (7 S/N). ple that internal standards have different precursor and/or product
ion masses but are chemically and structurally related to the tar-
2.4.2. Recovery rate and matrix effect get analyte. This is ideally achieved by using a deuterated analog of
Recovery rates were determined by spiking internal standards the analyte. Many of these deuterated eicosanoid analogs are com-
into control plasma samples and comparing the peak areas after mercially available but the inventory does not cover all eicosanoids
extraction with the corresponding peak areas of pure internal included in our comprehensive profile. In instances where a deuter-
standard solutions. The matrix effect was assessed by spiking inter- ated analog was not available, we carefully selected a deuterated
nal standards into plasma extracts and comparing the peak areas surrogate to cover a related analyte. For example, (d4) 15d PGJ2
with those of the deuterated standards in the standard solution. All was employed as the internal standard not only for 15d-PGJ2 , but
determinations were performed in triplicate. also for PGJ2 , and 15d-PGD2 , for which no deuterated standards
are currently available. In total, we use a cocktail of 26 deuterated
internal standards to aid the quantification of the 158 eicosanoids
2.4.3. Accuracy and precision
in our profile. A complete list of all deuterated standards and their
Accuracy and precision of the method was determined by spik-
analyte assignments is shown in Table 1.
ing blank plasma quality control (QC) samples spiked at three levels
A targeted approach was used to identify and quantify
of eicosanoids: low (LQC) 0.15 ng, medium (MQC) 1.5 ng, and high
lipids using mass spectrometry (MS) coupled with ultra-high-
QC (HQC) 5 ng. Intra-batch (same sample processed three times)
performance liquid chromatography (UPLC–MS). A targeted MRM
and inter-batch (three different batches) accuracy and precision
approach provided for higher sensitivity than an unbiased full-scan
were determined by analyzing five QC samples covering the entire
MS analysis. UPLC also provided enhanced chromatographic reso-
calibration range. The precision of the quantitation was expressed
lution, sensitivity, reproducibility, and fast separation. Additionally,
as percent coefficient of variance (CV%), calculated by dividing the
a QTRAP 6500 was used in the present study. The advantages of
standard deviation by the mean and then multiplied by 100.
the QTRAP 6500 are scan speeds of up to 20,000 Da/s for opti-
These QC determinations included the known fortified level
mized UPLC strategies, a 20-fold increase in the detector dynamic
added to the control plasma plus the endogenous amount
range, and LOQ improvement of up to 5-fold. In summary, the
of analytes. The endogenous amounts of analytes in plasma
UPLC-QTRAP 6500 system provides a sensitive, accurate, and fast
were determined in five replicate measurements. The accu-
separation and monitoring platform.
racy of the analytic method was denoted by the relative error
All eicosanoids are detected in the negative ion mode, taking
(RE%), calculated as percent of the mean deviation from the
advantage of their conserved terminal carboxyl moiety. Fig. 1 shows
known amount plus endogenous amount, RE% = [(amount found –
a representative extracted ion chromatogram for the separation of
(known amount + endogenous amount) × 100/(known amount +
all 184 eicosanoids including the 26 internal standards, which are
endogenous amount)].
monitored in a single 5 min LC–MS/MS analytical run. The opti-
mal declustering potential (DP) and the collision energy (CE) has
2.4.4. Stability been determined for each MRM pair (Table 2). These values were
To determine the stability of the processed samples, they were optimized by directly infusing commercial standards into the mass
kept at 4 ◦ C in the autosampler and injected three times at 0, 4, and spectrometer.
8 h with three levels of quality control samples, respectively. The The MS/MS spectra of most eicosanoids show multiple product
peak area of the analytes at the initial point (0 h) was used as the ions. The various precursor and product ions are listed in Table 2
reference to determine the relative stability at subsequent points. with the product ions for MRM detection underlined. The precur-
sor and product ions for the MRM detection were selected to yield
3. Results and discussion the best discrimination from other co-eluting eicosanoids and to
yield the highest signal intensity. By balancing LC retention time
3.1. Method development and product ion selection, we were able to successfully distinguish
all eicosanoids included in the profile and to achieve highest sensi-
SPE techniques were used to extract eicosanoid metabolites tivity possible. For example, owing to a similar structure, (d4) PGE2
from plasma, which is more suitable for processing a large num- and PGE2 have the same LC retention time of 1.3 min, but the pre-
ber of samples than liquid/liquid extraction (LLE) techniques [28]. cursor and product ions of (d4) PGE2 and PGE2 are different with
The extraction efficiency of LLE is generally higher than SPE, but this transitions of 355→275 and 351→271, respectively. In the other
method also extracts many endogenous impurities that can affect case, the precursor and product ions of PGE2 and PGD2 are the same,
the separation and quantitation of target analytes. The ability of SPE 351→271, but the retention time of PGE2 and PGD2 are different.
to eliminate impurities is better than that of LLE, which improves Peak selectivity was demonstrated by comparison with indi-
the detection of eicosanoids in biological matrices, especially when vidual MRM transitions and the retention time of each analyte, as
present at low levels. Moreover, internal standards will compensate shown in Table 2. Scheduled MRM is an improvement over tradi-
and neutralize any potential losses during sample processing. tional MRM allowing for better data collection and more analytes
Y. Wang et al. / J. Chromatogr. A 1359 (2014) 60–69 63
Table 1
Internal standard assigned for analytes.
No. Internal standard Analytes assigned No. Internal standard Analytes assigned
to be monitored in a single analytical experiment. We found a 30 s main precursor for a wide spectrum of unique eicosanoids pro-
retention time window for each MRM pair sufficient to allow for duced by COX, LOX, and CYP. Biochemical characterizations of EPA
potential small shifts in retention time. In addition to providing and DHA have generally suggested that these fatty acids are less
good sensitivity, MRM approaches are highly selective, reducing the prone to metabolism by eicosanoid pathway enzymes [30]. In the
need for extensive sample cleanup. As a proof of principle, control present studies, we probed for 88 eicosanoids derived from AA, 22
plasma was analyzed and we found no discernable distortions. from DHA, 17 from EPA, 13 from DGLA, and 18 from other fatty
acids, that were identified as outlined in Fig. 2. The solid black
lines and circles indicate the number of metabolites from the var-
3.2. Different precursor fatty acids for eicosanoids ious PUFAs. The comprehensive and simultaneous analysis of all
eicosanoids is important because eicosanoids derived from dif-
The key precursor polyunsaturated fatty acid (PUFA) for ferent PUFA sources may have different physiological effects. For
eicosanoids is arachidonic acid (AA, 20:4, n-6). Other eicosanoids example, in neural trauma and neurodegenerative diseases, there
and related compounds are formed from eicosapentaenoic acid is a dramatic rise in the levels of AA-derived eicosanoids and in con-
(EPA, 20:5, n-3), docosahexaenoic acid (DHA, 22:6, n-3), and trast, DHA-derived metabolytes can prevent neuroinflammation
dihomo-␥-linolenic acid (DGLA, 20:3, n-6). Arachidonic acid is the [31].
Fig. 1. Extracted ion chromatograms of (A) a mixture of 184 eicosanoid standards and (B) a magnified view of the chromatograms of HETE metabolites.
64 Y. Wang et al. / J. Chromatogr. A 1359 (2014) 60–69
Table 2
Optimized MRM pairs and parameters for eicosanoids.
No. Common name Retention time (min) m/z DP (V) CE (V) LOQ (pg)
Precursor Product
Table 2 (Continued)
No. Common name Retention time (min) m/z DP (V) CE (V) LOQ (pg)
Precursor Product
Table 2 (Continued)
No. Common name Retention time (min) m/z DP (V) CE (V) LOQ (pg)
Precursor Product
present in all human plasma at various levels. Since a deuter- selective extraction method or a chromatographic separation can
ated internal standard is either an analogous lipid metabolite or be developed.
a molecule with similar chemical characteristics (the chemical Absolute recovery was determined by spiking the internal stan-
structure of the eicosanoids can be found in Refs. [2,24,26]), both dards into plasma samples before exaction, as shown in Table 3.
lipid metabolites and internal standards will have similar ion sup- Recovery of 21 of the internal standards was above 70%. The lower
pression and extraction efficiencies. Thus, internal standards were recovery for (d4) PGF2a, (d8) 5-HETE, and (d8) arachidonic acid may
used to evaluate the matrix effect and the recovery rate of the be caused by the matrix effect noted above. The lower recovery for
method. (d4) 14d PGJ2 and (d4) Resolvin E1 may be due to non-specific bind-
The matrix effect was determined by spiking the internal ing by the SPE absorbent. The error margin for the matrix effect and
standard mixture into a blank plasma sample after exaction. The absolute recovery rate was ±10%.
results are shown in Table 3. Except for (d4) PGF2a, (d8) 5-HETE,
and (d8) arachidonic acid, the intensities of the other 23 internal 3.3.2. Linear and lower limit of quantitation
standards are all above 80%, indicating that matrix effects and ion The limit of quantitation for each analyte was defined as the
suppression are minimal for most analytes. For the three excep- lowest signal obtained with S/N ≥ 7. Based on these criteria, the
tions, it is possible that co-eluting matrix components may reduce LOQ was less than 10 pg for 86% analytes, as shown in Table 2. The
the ion intensity of these analytes. Phospholipids, and, in particular, differences in LOQs were most likely due to differences in ionization
phosphatidylcholines and lysophosphatidylcholines, represent the efficiencies and matrix effects.
major class of endogenous compounds causing significant matrix Eicosanoid quantitation was performed by the stable isotope
effects, which will suppress the signal of other co-eluting lipids dilution method. For each eicosanoid to be quantified, an internal
[32]. So the removal of phospholipids is essential for analysis standard was selected and a linear standard curve was generated
of minor lipid mediators. Phospholipids contain one or two long where the ratio of analyte standard peak area to internal standard
hydrophobic fatty acid ester chains. Thus, the absorption of phos- peak area was plotted versus the amount of analyte standard. The
pholipids on the C18 PSE column is much stronger than that of method offers good linearity for all analytes, the R2 value is above
eicosanoids. If using C18 material, phospholipids generally may be 0.97, and the typical standard eicosanoid curves are shown in Fig. 3.
eluted with strong eluant, such as isopropanol, etc. In our experi-
mental conditions, phospholipids can generally not be eluted from 3.3.3. Accuracy and precision
SPE columns and do not interfere in subsequent LC–MS/MS analy- The results of intra- and inter-day precision and accuracy testing
sis. However, lysophospholipids may be co-purified by SPE together using QC standard samples prepared in plasma as a blank matrix
with eicosanoids but are likely separated by subsequent LC. Never- are summarized in Table 3. Human plasma contains a number of
theless, part of the matrix effect may be caused by lysophospholipid endogenous eicosanoids and we found that we could identify and
contaminants. To minimize the matrix effect, a more efficient and quantify 60 endogenous metabolites in 20 L of standard plasma.
68 Y. Wang et al. / J. Chromatogr. A 1359 (2014) 60–69
J.W. Newman, R.J. Vreeken, Quantitative profiling of oxylipins through com- [29] C. Phillips, D. Owens, P. Collins, G.H. Tomkin, Low density lipoprotein non-
prehensive LC–MS/MS analysis: application in cardiac surgery, Anal. Bioanal. esterified fatty acids and lipoprotein lipase in diabetes, Atherosclerosis 181
Chem. 404 (2012) 1413–1426. (2005) 109–114.
[24] P.L. Faouder, V. Baillif, I. Spreadburyd, J.P. Mottae, P. Roussete, G. Chênec, C. [30] M. Wada, C.J. DeLong, Y.H. Hong, C.J. Rieke, I. Song, R.S. Sidhu, C. Yuan,
Guignéc, F. Tercéa, S. Vannerd, N. Vergnollee, J. Bertrand-Michel, M. Dubour- M. Warnock, A.H. Schmaier, C. Yokoyama, C.E.M. Smyth, S.J. Wilson, G.A.
deauc, N. Cenace, LC–MS/MS method for rapid and concomitant quantification FitzGerald, R.M. Garavito, D.X. Sui, J.W. Regan, W.L. Smith, Enzymes and
of pro-inflammatory and pro-resolving polyunsaturated fatty acid metabolites, receptors of prostaglandin pathways with arachidonic acid-derived versus
J. Chromatogr. B 932 (2013) 123–133. eicosapentaenoic acid-derived substrates and products, J. Biol. Chem. 282
[25] D.D. Shindea, K.B. Kima, K.S. Oha, N. Abdallaa, K.H. Liub, S.K. Baec, J.H. Shona, (2007) 22254–22266.
H.S. Kima, D.H. Kima, J.G. Shina, LC–MS/MS for the simultaneous analysis of [31] D. Tassoni, G. Kaur, R.S. Weisinger, A.J. Sinclair, The role of eicosanoids in the
arachidonic acid and 32 related metabolites in human plasma: basal plasma brain, Asia Pac. J. Clin. Nutr. 17 (2008) 220–228.
concentrations and aspirin-induced changes of eicosanoids, J. Chromatogr. B [32] O.A. Ismaiel, M.S. Halquist, M.Y. Elmamly, A. Shalaby, H.T. Karnes,
911 (2012) 113–121. Monitoring phospholipids for assessment of matrix effects in a liquid
[26] B.S. Levison, R.L. Zhang, Z.N. Wang, X.M. Fu, J.A. DiDonato, S.L. Hazen, Quantifi- chromatography–tandem mass spectrometry method for hydrocodone and
cation of fatty acid oxidation products using online high-performance liquid pseudoephedrine in human plasma, J. Chromatogr. B 859 (2007) 84–93.
chromatography and tandem mass spectrometry, Free Radical Biol. Med. 59 [33] W. Jian, R. Edom, N. Weng, P. Zannikos, Z. Zhang, Z.H. Wang, Validation
(2013) 2–13. and application of an LC–MS/MS method for quantitation of three fatty acid
[27] L. Kortz, J.L. Dorow, S. Becker, J. Thiery, U. Ceglareka, Fast liquid ethanolamides as biomarkers for fatty acid hydrolase inhibition in human
chromatography–quadrupole linear ion trap-mass spectrometry analysis of plasma, J. Chromatogr. B 878 (2010) 1687–1699.
polyunsaturated fatty acids and eicosanoids in human plasma, J. Chromatogr. [34] R. Dams, M.A. Huestis, W.E. Lambert, C.M. Murphy, Matrix effect in bio-analysis
B 927 (2013) 209–2139. of illicit drugs with LC–MS/MS: influence of ionization type, sample prepara-
[28] M.Y. Golovko, E.J. Murphy, An improved LC–MS/MS procedure for brain tion, and biofluid, J. Am. Soc. Mass Spectrom. 14 (2003) 1290–1294.
prostanoid analysis using brain fixation with head-focused microwave [35] O. Quehenberger, A. Armando, D. Dumlao, D.L. Stephens, E.A. Dennis,
irradiation and liquid-liquid extraction, J. Lipid Res. 49 (2008) 893– Lipidomics analysis of essential fatty acids in macrophages, Prostaglandins
902. Leukot. Essent. Fatty Acids 79 (2008) 123–129.