Subscriber access provided by CMU Libraries - http://library.cmich.edu
Article
Development and validation of a high throughput UHPLCMS approach for screening of oxylipins and their precursors Arnaud M. Wolfer, Mathieu Gaudin, Simon D Taylor-Robinson, Elaine Holmes, and Jeremy Kirk Nicholson Anal. Chem., Just Accepted Manuscript • DOI: 10.1021/acs.analchem.5b02794 • Publication Date (Web): 26 Oct 2015 Downloaded from http://pubs.acs.org on November 4, 2015
Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.
Page 1 of 11
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
Development and validation of a high throughput UHPLC-MS approach for screening of oxylipins and their precursors Arnaud M. Wolfer,† Mathieu Gaudin,†,∥ Simon D. Taylor-Robinson,§ Elaine Holmes,†,ǂ Jeremy K. Nicholson*,†,ǂ †
Division of Computational and Systems Medicine, Department of Surgery and Cancer, Imperial College London, Sir Alexander Fleming Building, Exhibition Road, London SW7 2AZ, United Kingdom § Department of Medicine, Imperial College London, St Mary’s Campus, London, United Kingdom ǂ MRC-NIHR National Phenome Centre, Imperial College London, IRDB Building, Hammersmith Hospital, London W12 0NN, United Kingdom ABSTRACT: Lipid mediators, highly bioactive compounds synthesized from polyunsaturated fatty acids (PUFAs), have a fundamental role in the initiation and signaling of the inflammatory response. Although extensively studied in isolation, only a limited number of analytical methods offer a comprehensive coverage of the oxylipin synthetic cascade applicable to a wide range of human biofluids. We report the development of an ultra-high-performance liquid chromatography-electrospray ionization triple quadrupole mass spectrometry (UHPLC-MS) assay to quantify oxylipins and their PUFA precursors in 100 µL human serum, plasma, urine and cell culture supernatant. A single 15 minute UHPLC run enables the quantification of 43 oxylipins and 5 PUFAs, covering pro and anti-inflammatory lipid mediators synthetized across the cyclooxygenase (COX), lipoxygenase (LOX) and cytochrome P450 (CYP450) pathways. The method was validated in multiple biofluid matrices (serum, plasma, urine and cell supernatant) and suppliers, ensuring its suitability for large scale metabonomic studies. The approach is accurate, precise and reproducible (RSD 5 and satisfying precision and accuracy criteria as spiked in extracted biofluid, are prepared in triplicate from a described below. homogenous biofluid aliquot. Accuracy and Precision. Accuracy and Precision of the Matrix Effects. The matrix effect pertains to the interference method were determined by spiking quality control (QC) on ionization efficiency exerted by other compounds present (MeOH/H2O 1:1) at 4 analytes concentrations (0.5, 5, 50, 250 ng/mL for oxylipins, 5, 50, 500, 2500 ng/mL for PUFA). in the biological matrix. The matrix effect was assessed on IS by comparing response obtained for IS spiked in calibration This ensures each analyte is represented by a minimum of 3 matrix (MeOH/H2O 1:1) with the response of IS spiked in concentrations (low, medium and high) despite the variety of biological matrix after SPE extraction. The presence of a malinear ranges. Accuracy is defined as the closeness of mean tric effect for a given analyte must be evaluated, but does not test results obtained over the 6 replicates to the expected necessarily indicate that the method may not be valid if it is (nominal) concentration, the measured concentration in perconsistent and reproducible (low RSD). For serum, plasma and centage of expected value is reported. The method is accurate urine 6 different sources have been employed, as well as 2 if the concentration measured is 85-115% of the expected valsources for cell culture supernatant. For each source, 3 QC ue (80-120% for the LLOQ). Precision is defined as the closeconcentrations (0.5, 5, 250 ng/mL for oxylipins, 5, 50, ness of the 6 individual measures when the procedure is ap2500 ng/mL for PUFA) of IS spiked in calibration matrix plied repeatedly to multiple aliquot of a replicate homogenous (MeOH/H2O 1:1) and 3 QC concentrations of IS spiked in volume of QC, the relative standard deviation (RSD) of the extracted biofluid, are prepared in triplicate from a homogemeasurements is reported. The method is precise if the RSD is nous biofluid aliquot. inferior to 15% (20% for the LLOQ). Intra-day accuracy and precision were established by 6 replicates measurements of Analyte Stability. The stability of analytes was assessed by each QC using the same calibration curve. Inter-day accuracy triplicate measurement using the assay of QC samples spiked and precision were assessed by 6 measurements of each QC at 4 analytes concentrations (0.5, 5, 50, 250 ng/mL for oxover 6 days (6 calibration curves). ylipins, 5, 50, 500, 2500 ng/mL for PUFA) stored in long-term storage (-80 °C) for 20 weeks, which underwent 3 freeze-thaw Analytical Recovery. Recovery pertains to the extraction efcycles. The percent difference between the mean measured ficiency of SPE for each analyte in each biological matrix.
ACS Paragon Plus Environment
Page 5 of 11
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
concentration and the expected values as well as the RSD of the measurements were evaluated. Results are satisfactory if the RSD of measurement is inferior to 15% (20% for LLOQ) and the mean measurement at 85-115% from expected values (80-120% for LLOQ). Sample Collection and Analysis. Oxylipin profiling was performed on serum, EDTA-plasma and urine collected from 21 healthy individuals. The study protocols were approved by an independent ethics committee (reference 09/H0712/82) and informed consent was obtained from each individual. Samples were collected for twelve individuals after a 12 hours fasting period (Age 30.6 [24-58], BMI 21.9 [15.6-28.6], 11 female / 1 male), and 2 hours after a fat rich meal for the remaining nine (Age 31.1 [25-54], BMI 28.1 [21-38.9], 2 female / 7 male). Serum, EDTA-plasma and urine samples were sub-aliquoted in 1.5 mL tubes and stored at -80 °C until extraction and analysis. RESULTS AND DISCUSSION Method Development. The development of a highthroughput method for the quantification of a panel of oxylipins faced four main challenges. Firstly, the high structural similarity of oxylipins and high number of isomeric analytes imposes the use tandem mass spectrometry for the selection of specific MRM transitions as well as liquid chromatography to separate analytes in the time domain depending on their physiochemical properties. Secondly, the physiologically low endogenous concentrations of oxylipins demands optimization of every step to guarantee the best limits of quantification possible. Post-column infusion of formaldehyde has for example been selected to promote negative ESI ionization and increase sensitivity32, this approach resulted in an increase in sensitivity by a factor two to three for the most hydrophilic analytes. A third challenge is the complexity of the biological matrices hosting oxylipins and the interaction of oxylipins with the other entities present (matrix effects). SPE reduces background noise and matrix interferences by extracting the majority of endogenous impurities (helping with sensitivity) while also dissociating oxylipins from circulatory proteins they are bound at physiological pH. Finally, in order to enable higher throughput, the gradient must be optimized for a short chromatographic run without sacrificing separation of critical transitions (when two analytes display the same fragmentation) (Figure 2). This strategy combined with the use of 96-well plate SPE, which facilitate preparation, enables the acquisition of 96 samples in 24 hours. The deployment of the method on a large scale is further helped by the use of a 1 mm column which significantly reduces quantities of solvent and sample required compared to a 2.1mm column without impact on chromatographic separation. Quantification of each oxylipin was obtained by the backcalculation of the ratio of an analyte to its deuterated internal standard. IS are employed to correct variation in extraction efficiency during SPE, instrument variability, ionization efficiency or to account for changes in volumes for each sample and each analyte. These stable structurally similar isotopelabelled IS present 4 to 11 deuterium atoms, making them distinguishable from the endogenous oxylipins by MS while presenting similar properties in terms of extraction, recovery and elution to the unlabeled analyte. As the availability of deuterated oxylipins is limited, when a deuterated analogue was not available, the closest surrogate was selected. We se-
lected 7 deuterated IS to help the quantification of 48 oxylipins and PUFA (Table S-2). Mass Spectrometry Optimization. Negative ionization mode was selected due to the oxylipins’ carboxyl moiety and MS parameters were optimized in order to achieve the highest sensitivity. The most sensitive transition and MRM window which provides sufficient discrimination from co-eluting oxylipins was selected for each analyte and is listed in Table S-6. Compounds and their deuterated equivalent were evaluated for cross-talk and carryover and rejected from the oxylipin panel when results were not satisfactory. Cross-talk between IS and analytes is deemed satisfactory if the response is inferior to 20% of the LLOQ concentration (5% between analytes) and carryover is acceptable if the response in a blank is inferior to 20% of the LLOQ response. For example, PGF2α-d4 was considered during method development for use as an IS (transition m/z 357 -> 197) but was rejected as the cross-talk with PGF2α (transition m/z 353 ->193) was too high for reliable quantification. Method Validation. The targeted profiling assay was validated for linearity (Table S-3), intra and inter-day precision and accuracy over 6 replicates on 6 consecutive days (Table S7), representing the chromatographic system and detector performance. Matrix effects (Table S-9)and analyte recovery (Table S-8) were studied in triplicate over 6 different sources of each human matrix (serum, plasma and urine) as well as 2 sources of cell culture supernatant, representing the interindividual as well as inter-matrix variability on the assay. Finally long-term storage stability was determined (Table S-10). As oxylipins and PUFA are present in all human matrices we employed at various levels, no “blank” matrices were available and different validation matrices had to be used for different experiments.33–35 Due to endogenous levels, the calibration matrix selected is identical to the solvent employed for reconstitution of the dried extracts after SPE (MeOH/H2O 1:1). For the recovery experiment standards were added to an unknown endogenous analyte concentration which is present in both pre and post spiked samples. Matrix effect could only be determined on deuterated analytes as post-spiked samples would have contained an unknown endogenous oxylipin concentration incompatible with this experiment. Selectivity. Proof of the provenance and purity of chemical standards was provided by the standards suppliers. During method development, peak selectivity was demonstrated by injection of individual compounds in calibration matrix. The product ions yielding the best discrimination from co-eluting oxylipins and the highest signal intensity were selected. Compounds and their deuterated equivalent were evaluated for cross-talk and carryover and rejected from the oxylipin panel when results were not satisfactory. In our method, two isobaric compounds quantified using the same transition (11(R)-HETE and 11,12-EET: m/z 319 -> 167) were successfully separated without overlapping MRM windows (RT = 7.90-8.60 min and RT = 9.40-9.80 min respectively ) (Figure S-1). LLOQ and Linear Range. The linearity of the method was established for each analyte on 6 calibration curves prepared on six consecutive days following the procedure described in the method section. Due to the wide variety of linear ranges, 17 calibration points were used to ensure that 6 concentrations are available for each analyte. The ratio of analyte area to its IS area was plotted against nominal concentration, fit was linear and no weighting factor was applied.
ACS Paragon Plus Environment
Analytical Chemistry
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Table 1: LLOQ (* loaded on the column) and linear range for each analyte. Compound name
C18:2 (LA) C20:3 (DGLA)
LLOQ*
Linear range
(pg)
(µg/mL)
1.3*10
2
1.3*10
2
(µM)
-2
8.9*10-2 - 35
-2
8.1*10-2 - 32
-3
2.5*10 - 10 2.5*10 - 10
C20:4 (AA)
25
5.0*10 - 5.0
1.6*10-2 - 16
C20:5 (EPA)
13
2.5*10-3 - 5.0
8.2*10-3 - 16
C22:6 (DHA)
13
9(S)-HODE
2.5*10
13(S)-HODE
0.5
Tetranor-PGDM Tetranor-PGEM
-3
7.6*10-3 - 7.6
-5
5.0*10 - 1.0
1.7*10-4 - 3.3
1.0*10-4 - 1.0
3.4*10-4 - 3.3
2.5*10 - 2.5 -1
-4
1.5*10-3 - 1.5
-3
7.6*10-3 - 1.5
-3
5.0*10 - 0.5
2.5 13
2.5*10 - 0.5
Tetranor-PGFM
25
5.0*10 - 1.0
1.5*10-2 - 3.0
12(S)-HEPE
1.3
2.5*10-4 - 0.5
7.8*10-4 - 1.5
15(S)-HEPE
1.3
-4
7.8*10-4 - 3.1
-4
2.5*10 - 1.0
5,6-EET
13
2.5*10 - 1.0
7.8*10-4 - 3.1
8,9-EET
2.5
5.0*10-4 - 1.0
1.5*10-3 - 3.1
11,12-EET
0.5
14,15-EET
5.0
5(S)-HETE
0.5
8(S)-HETE 11(R)-HETE
5.0*10 2.5*10
-1
0.5
15(S)-HETE
1.3
3.1*10-4 - 3.1
-3
1.0*10 - 1.0
3.1*10-3 - 3.1
1.0*10-4 - 1.0
3.1*10-4 - 3.1
1.0*10 - 1.0
-2
12(R)-HETE
-4
-5
3.0*10-5 - 1.5
-5
1.6*10-4 - 1.5
-4
1.0*10 - 1.0
3.1*10-4 - 3.1
2.5*10-4 - 1.0
7.8*10-4 - 3.1
1.0*10 - 0.5 5.0*10 - 0.5
2.5*10
-1
5,6-DHET
5.0*10
-2
1.0*10 - 1.0
3.0*10-5 - 2.9
8,9-DHET
5.0*10-2
1.0*10-5 - 1.0
3.0*10-5 - 2.9
16(R)-HETE
5.0*10
-2
14,15-DHET
5.0*10
-2
5-oxo-ETE
1.3
11,12-DHET
12-oxo-ETE 14-HDoHE
1.3
-5
1.6*10-4 - 3.1
-5
5.0*10 - 1.0
-5
3.0*10-5 - 2.9
-5
1.0*10 - 1.0
3.0*10-5 - 2.9
2.5*10-4 - 1.0
7.8*10-4 - 3.1
1.0*10 - 1.0
-4
7.8*10-4 - 1.5
-4
2.9*10-4 - 2.9
-4
2.5*10 - 0.5
0.5
1.0*10 - 1.0
17(S)-HDoHE
2.5
5.0*10 - 1.0
1.4*10-3 - 2.9
10(S),17(S)-DiHDoHE
2.5*10-1
5.0*10-5 - 0.5
1.4*10-4 - 1.3
LTB4
1.3
12-oxo-LTB4
2.5
LTC4
0.5
LTD4
2.5*10
LTE4
0.5
PGD2
1.3
PGE2 Lipoxin A4
2.5*10
2.5e*10 - 1.0 -4
-1
-1
1.3
Lipoxin B4
5.0
6-keto-PGF1α
2.5
PGF2α
4
2.5*10
8-iso-PGF2α
0.5
15dPGJ2
2.5
5.0*10 - 1.0
1.5*10-3 - 2.9
1.0*10-4 - 1.0
1.6*10-4 - 1.6
-5
1.0*10-4 - 1.0
-4
1.0*10 - 1.0
2.3*10-4 - 2.2
2.5*10-4 - 1.0
7.1*10-4 - 2.8
5.0*10 - 0.5
-5
1.4*10-4 - 1.4
-4
7.1*10-4 - 2.8
-3
1.0*10 - 1.0
2.8*10-3 - 2.8
5.0*10-4 - 0.5
1.3*10-3 - 1.3
5.0*10 - 0.5 2.5*10 - 1.0
-1
7.4*10-4 - 2.9
-5
5.0*10 - 0.25
1.4*10-4 - 7.0*10-1
-4
1.0*10 - 0.25
2.8*10-4 - 7.0*10-1
5.0*10-4 - 0.5
1.5*10-3 - 1.5
2.5*10
-1
-5
5.0*10 - 0.25
1.3*10-4 - 6.7*10-1
11-dehydro TxB2
2.5*10
-1
-5
5.0*10 - 0.25
1.4*10-4 - 6.7*10-1
Resolvin D1
5.0*10-2
1.0*10-5 - 1.0
3.0*10-5 - 2.6
TxB2
Resolvin D2
0.5
-4
1.0*10 - 1.0
2.7*10-4 - 2.6
Page 6 of 11
The goodness of fit R2 was superior to 0.998 for all metabolites but tetranor-PGFM (R2=0.9966) confirming the linearity of the assay in the selected calibration range (Table S-3). LLOQ was determined as the lowest amount of standard necessary to produce a S/N> 5 while being part of the linear range of the calibration curve (back-calculated residual < 20%). The obtained LLOQ ranged from 0.05 to 12.5 pg for oxylipins and 12.5 to 125 pg for PUFA. Our approach (48 oxylipins and precursors in 13 minutes) is to our knowledge one of the most sensitive reported (Table 1). The complexity of oxylipin quantification imposes a necessity to balance the number of analytes to quantify against the sample size, sensitivity achieved, as well as the throughput. Depending on the end use of the method, different approaches have been taken, favouring for example throughput at the expense of sensitivity and number of analytes (26 oxylipins in 8.5 minutes25), sensitivity and number of matrices over throughput (39 oxylipins in 21 minutes23), or throughput and number of analytes but limiting the applicability to a single validated matrix and comparatively low number of oxylipins quantified at their endogenous level in biological samples (158 oxylipins and PUFA in 5 minutes26). In the context of an assay designed for the profiling of multiple human biofluids in large scale metabonomic studies, we favoured high-throughput, applicability to a complementary range of matrices, while still offering sensitivities sufficient to quantify the highest number of oxylipins at their endogenous level, and covering the biosynthetic cascade of a wide range of PUFA. We obtained LLOQ inferior to 0.5 pg for more than 50% of analytes and lower than 5 pg for 90% of them, making our assay one of the most sensitive reported for a wide range of analytes. This helped us achieve our goal, illustrated by the high number of panel analytes quantified in unspiked biological samples. Accuracy and Precision. Intra and inter-day precision and accuracy experiments were calculated from QC standards prepared in the calibration matrix (MeOH/H2O 1:1) due to the lack of “blank” matrix available. The intra-batch and interbatch accuracy and precision have been established using 6 replicates and at a minimum of 3 QC concentrations (4 QC for a majority of analytes) (Table S-7). Accuracy was within 15% (20% for LLOQ) from the nominal concentration in 99% of samples, in all cases accuracy was under 30% from nominal concentration. Precision was within 15% (20% for LLOQ) relative standard deviation in 98% of samples, in all cases the RSD was under 30%. We can therefore conclude than our method is reproducible, accurate and precise across the range of concentrations used on QC samples. Analytical Recovery. Recovery was assessed by comparing the analyte response obtained for samples spiked with analyte before extraction to samples spiked after extraction. For serum and plasma six different sources have been employed, as well as two sources for cell culture supernatant. As urine samples are not subject to SPE extraction, recovery is non-existent. For each source, all analytes and deuterated IS were tested in triplicate at two QC concentrations (Table S-8). Recoveries were different for a same analyte in different biological matrices, however as this is also the case for the IS (selected for their similar physico-chemical properties), the constant ratio analyte/IS correct for matrix specific extraction yields. They range from 16% to 90% in serum, 9%-90% in plasma and 26%-93% in cell culture supernatant.
ACS Paragon Plus Environment
Page 7 of 11
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
Analytical Chemistry
For a given matrix, recoveries varied between oxylipin subfamilies but were generally identical inside these structural sub-families. For the vast majority of analytes, recoveries were similar at both concentrations. Notable exceptions in serum and plasma are DGLA, tetranor-PGEM and tetranor-PGFM. Variable recoveries for the tetranors may be due to matrix effects noted for this sub-family as discussed in the next section; however the relative standard deviation across the six matrix source is moderate (RSD